Masterson, Robert - Nuclear Reactor Thermal Hydraulics - An Introduction To Nuclear Heat Transfer and Fluid Flow-CRC Press (2020)
Masterson, Robert - Nuclear Reactor Thermal Hydraulics - An Introduction To Nuclear Heat Transfer and Fluid Flow-CRC Press (2020)
Masterson, Robert - Nuclear Reactor Thermal Hydraulics - An Introduction To Nuclear Heat Transfer and Fluid Flow-CRC Press (2020)
Relationship between Number of Fissions per Second and Thermal Energy Production
1 Watt Thermal Energy = 1 Joule/sec (Thermal) = 3.12 x 1010 fissions/second
Energy conversion using E = mc2: 1 AMU of mass m = 931.494 MeV/c2
Units of Pressure
Metric pressures 1 Pascal = 1 Pa = 1 N/m2
Atmospheric pressure 1 atm = 14.696 PSI = 101.325 kPa = 1.01325 bar
Stefan-Boltzmann Constant
Viscosity
Dynamic viscosity 1 kg/m-s = 1 N–s/m2 = 2419.1 lb/ft-hr
Kinematic viscosity 1 m2/s = 104 stoke = 10.764 ft2/s
Nuclear Reactor Thermal Hydraulics
An Introduction to Nuclear Heat Transfer
and Fluid Flow
Nuclear Reactor Thermal Hydraulics
An Introduction to Nuclear Heat Transfer
and Fluid Flow
by
Robert E. Masterson, PhD (ScD in Nuclear Science and Engineering)
Massachusetts Institute of Technology
Cambridge, Massachusetts
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to publish reliable
data and information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their
use. The authors and publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to
copyright holders if permission to publish in this form has not been obtained. If any copyright material has not been acknowledged, please write
and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any
electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any
information s torage or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.com/)
or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit
organization that provides licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the
CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and
explanation without intent to infringe.
v
vi Contents
3.7 The Advanced BWR and the Simplified BWR 78 5.2 Measuring Nuclear Energy 121
3.8 The Simplified BWR 79 5.3 Examining the Fission Process 122
3.9 Characteristics of BWRs 80 5.4 Cross Sections and Reaction Rates 122
3.10 Core Characteristics and Design Parameters 80 5.5 Converting the Kinetic Energy of
3.11 The Void Fraction and the Nuclear Particles into Thermal Energy 125
Quality in a BWR Core 81 5.6 Energy Produced by the Fission Process 126
3.12 Finding the Density from the Void Fraction 81 5.7 Estimating the Total Energy Release 127
3.13 The Relationship between the 5.8 Other Types of Nuclear
Void Fraction and the Quality 83 Particles and Radiation 128
3.14 BWR Flow Regimes 85 5.9 Gamma Rays and Their Properties 129
3.15 Operating Restrictions for a Typical BWR 88 5.10 Radioactive Decay Heat 129
3.16 BWR Operating Maps 88 5.11 Attenuation Coefficients for Different
3.17 Lessons That Can Be Learned Types of Nuclear Radiation 130
from a Reactor Operating Map 89 5.12 Energy of Fission Neutrons 131
Bibliography 91 5.13 The Maxwell–Boltzmann
Books and Textbooks 91 Probability Distribution 134
Web References 92 5.14 Thermal Energy Produced
Questions for the Student 92 in Nuclear Fuel Rods 135
Exercises for the Student 93 5.15 The Fuel Rod Cladding 137
5.16 Nuclear Fuel Assemblies 139
5.17 PWR, BWR, and LMFBR Fuel Assemblies 139
Chapter 4 5.18 The Number of Fuel Rods in a Fuel Assembly 141
Fast Reactors, Gas Reactors, and Military Reactors 95 5.19 Fuel Rods in Square Fuel Assemblies 141
5.20 Fuel Rods in Hexagonal Fuel Assemblies 141
4.1 An Introduction to Other Reactor Types 95
5.21 CANDU Reactor Fuel Assemblies 143
4.2 Advantages of Fast Reactors 95
5.22 Where Nuclear Energy Is
4.3 Fast Reactor Coolants 95
Produced in the Core 144
4.4 Advanced Gas Reactors 96
5.23 Ways of Measuring the Core Power 145
4.5 Liquid Metal Fast Breeder Reactors 97
5.24 Heat Generation in Nuclear Fuel Assemblies 147
4.6 Fuel Assemblies for LMFBRs 99
5.25 The Volumetric Power Densities
4.7 Temperature Profiles in an LMFBR Core 102
for Different Reactor Types 149
4.8 Other Reactor Concepts 102
5.26 Power Densities in Reactor Fuel Assemblies 150
4.9 Military Reactors and the MPR 104
5.27 Power Profiles in Reactor Fuel Assemblies 150
4.10 High-Temperature Gas Reactors 106
5.28 Using Burnable Poisons to Flatten the
4.11 HTGR Fuel 106
Power and Temperature Profile 152
4.12 Comparing the Designs of Gas Reactors 107
5.29 Axial Power Shapes and Power Peaks 153
4.13 Gas Reactor Cores and Design Parameters 110
5.30 Using Axial Zoning to Reduce Power Peaks 154
4.14 LMFBR Thermal Cycle Performance 111
5.31 Finding the Power Profiles in Simple Reactors 155
4.15 Characteristics of LMFBR Cores 112
5.32 Neutron Reflectors 155
4.16 Comparing the Power Densities
5.33 Axial Power Profiles in Uniform,
in Different Reactor Cores 113
Unreflected Cores 156
4.17 The NSSS and the Containment
5.34 Radial Power Profiles in
Building for Fast Reactors 114
Uniform, Unreflected Cores 157
4.18 Theoretical Thermal Efficiencies 115
5.35 Cylindrical Power Profiles
4.19 Earliest Fast Reactors 116
and Bessel Functions 158
Bibliography 117
5.36 Extrapolated Power Profiles 158
Books and Textbooks 117
5.37 Global Power Profiles in
Web References 117
Different Types of Cores 161
Questions for the Student 117
5.38 Core Peak-to-Average Power Ratios
Exercises for the Student 119
for Uniform, Unreflected Cores 162
5.39 Power Profiles for Heterogeneous Cores 162
Chapter 5 5.40 Flattening the Power Profile 162
5.41 How Control Rods Affect the
Thermal Energy Production in Nuclear Power Plants 121
Core-Wide Power Profile 163
5.1 Heat Production in Nuclear Power Plants 121 5.42 Power Fluctuations in Steady-State Cores 165
Contents vii
14.18 The Lagrangian and Eulerian 15.11 The Fluid Conservation Equations
Views of Fluid Mechanics 532 in Different Coordinate Systems 563
14.19 Finding the Acceleration of an Eulerian Fluid 533 15.12 The Conservation Equations for
14.20 Understanding the Differences between One-Dimensional Flows 564
Advection, Diffusion, and Convection 533 15.13 Thermal-Hydraulic Correlations
14.21 Finding the Acceleration of a Fluid and Their Uses 566
in an Eulerian Reference Frame 534 15.14 The Continuity Equation 567
14.22 Finding the Advective Acceleration 15.15 Working with the Continuity Equation 569
in an Eulerian Reference Frame 536 15.16 The Energy Equation 570
14.23 Some Observations Regarding 15.17 Working with the Fluid Energy Equation 572
the Advective Derivative 536 15.18 The Momentum Equations 573
14.24 The Material Derivative in 15.19 The Differential Momentum Equations 575
Eulerian Fluid Mechanics 537 15.20 Understanding the Advection of Momentum 577
14.25 Heat and Energy Transfer in 15.21 The Time-Dependent Momentum Equations 578
Reactor Fuel Assemblies 538 15.22 Finding the Pressure Field
14.26 Connecting the Lagrangian and the from the Velocity Field 580
Eulerian Descriptions of Fluid Mechanics 538 15.23 The Navier–Stokes Equations
14.27 Rotational Flows 539 in Cylindrical Coordinates 581
14.28 The Vorticity of a Rotational Flow 542 15.24 Comparing the Behavior of Newtonian
14.29 Fluid Deformation and Volumetric Strain 545 and Non-Newtonian Fluids 583
14.30 Visualizing the Flow of Fluid 15.25 Treatment of Fluid Friction in
through a Reactor Piping System 546 a Compressible Fluid 583
14.31 Path Lines and Streamlines 546 15.26 Cauchy’s Equations for a Compressible Fluid 584
14.32 Definition of a Path Line 546 15.27 Simplifying Cauchy’s Equations to Handle
14.33 Definition of a Streak Line 547 Incompressible Newtonian Fluids 585
14.34 Definition of a Streamline 548 15.28 A Practical Approach to Fluid
14.35 Streamlines and Stream Tubes 550 Friction in Nuclear Power Plants 586
14.36 Methods for Observing Path 15.29 Viscous Effects and the No-Slip
Lines and Streak Lines 550 Boundary Condition 588
14.37 Reactor Accidents and Particle 15.30 Euler’s Equation and Its Origins 588
Entrainment Theory 551 15.31 Boundary-Layer Approximations 589
Bibliography 551 15.32 Converting Euler’s Equation
Books and Textbooks 551 into Bernoulli’s Equation 591
Questions for the Student 552 15.33 Comparing Various Forms of
Exercises for the Student 553 the Conservation Equations 592
15.34 Adding Turbulence to the
Momentum Equations 592
Chapter 15 15.35 Velocity Fluctuations in Turbulent Flows 592
15.36 The Reynolds Decomposition 594
The Conservation Equations of Fluid Mechanics 555
15.37 Solving the Navier–Stokes
15.1 The Material Derivative 555 Equations for Turbulent Flows 594
15.2 The Number of Conservation 15.38 Modeling the Effects of Turbulence 595
Equations for Single-Phase Flow 556 15.39 Approximate Solutions to the
15.3 Simplifications to the Fluid Navier–Stokes Equations 595
Conservation Equations 556 15.40 The Origin of the Navier–
15.4 Lateral Flow and Cross-Flow 557 Stokes Energy Equation 596
15.5 Fluid Mechanical Modeling 15.41 Fluid Boundary Conditions 596
with a CFD Program 557 15.42 The No-Slip Boundary Condition 597
15.6 Common Simplifications to Computational 15.43 The Interface Boundary Condition 597
Fluid Dynamics Models for Nuclear Reactors 557 15.44 The Free Surface Boundary Condition 598
15.7 Types of Fluid Flow 557 15.45 Other Types of Boundary Conditions 598
15.8 Fluid Viscosity and Friction 558 15.46 Pressure Boundary Conditions 598
15.9 The Navier–Stokes Equations 561 15.47 Flow Boundary Conditions 599
15.10 Singular Solutions to the 15.48 Boundary-Layer Flow 600
Navier–Stokes Equations 561 15.49 Boundary-Layer Solutions 601
Contents xiii
Chapter 16
Single-Phase Flow in Nuclear Power Plants 609 Chapter 17
Laminar and Turbulent Flows with Friction 647
16.1 Single-Phase Flow 609
16.2 Using Gases as Reactor Coolants 610 17.1 Reactor Coolants and Their Viscosities 647
16.3 Single-Phase Flow without Friction 611 17.2 Viscosity and Fluid Friction 647
16.4 Compressible and Incompressible Fluids 611 17.3 Laminar and Turbulent Flows 649
16.5 The Importance of Fluid Friction 611 17.4 Characteristics of Laminar Flows 649
16.6 Bernoulli’s Principle and Bernoulli’s Equation 612 17.5 Characteristics of Turbulent Flows 650
16.7 The Importance of Bernoulli’s Equation 17.6 How Viscosity Affects Turbulence 651
to Nuclear Science and Engineering 612 17.7 Fluid Flow in Pipes, Reactor
16.8 Applications of Bernoulli’s Equation to Coolant Channels, and Tubes 651
Pipes and Other Simple Structures 613 17.8 An Introduction to the Reynolds Number 652
16.9 Applying Bernoulli’s Equation 17.9 The Temperature Dependence
to a Flow Blockage 616 of the Reynolds Number 653
16.10 The Effect of Directional Changes in the 17.10 The Reynolds Number for Open
Flow Field on the Coolant Pressure 617 Surfaces and Closed Surfaces 653
16.11 The Effect of Elevation on 17.11 The Critical Reynolds Number 654
the Coolant Pressure 620 17.12 Turbulent Flow in Pipes, Reactor
16.12 The Effects of Wall Friction Coolant Channels, and Tubes 655
on the Local Pressure 621 17.13 Reynolds Numbers for Reactor
16.13 The Equivalent Hydraulic Fuel Assemblies 656
Diameter and Its Applications 622 17.14 Boundary Layer Development 657
16.14 Calculating the Total Pressure 17.15 The Entrance Length and Entrance Effects 657
Drop in a Coolant Channel 625 17.16 Coolant Velocity Profiles for
16.15 Understanding Flow and Pressure Changes 627 Laminar and Turbulent Flows 659
16.16 Finding the Flow Rates and the 17.17 Estimating the Wall Shear Stress 661
Pressures in Nozzles and Diffusers 628 17.18 The Darcy Equation for the
16.17 Different Types of Pressures 630 Frictional Pressure Drop 662
16.17.1 The Static Pressure 630 17.19 Deriving the Darcy Equation
16.17.2 The Dynamic Pressure 630 for Laminar Flow 662
16.17.3 The Hydrostatic Pressure 630 17.20 Deriving the Darcy Equation
16.18 The Time-Dependent Bernoulli Equation 632 for Turbulent Flow 665
16.19 Estimating the Start-Up Time of a Reactor 17.21 Boundary-Layer Thicknesses 666
Coolant Pump from Bernoulli’s Equation 633 17.22 Applying What We Have Just Learned
16.20 Solving Bernoulli’s Equation with to a Reactor Coolant Channel 667
Loss Coefficients and Friction 17.23 Applying the Darcy Equation to
Factors in a Viscous Flow Loop 636 Coolant Channels with Different
16.21 Some Practical Applications Cross-Sectional Areas 667
of Bernoulli’s Equation 637 17.24 Turbulent Friction Factors 669
16.22 Applying Bernoulli’s Equation 17.25 The Blasius, McAdams, and Petukhov
to Laminar Flows 637 Correlations for Turbulent Flow 670
16.22.1 Applying Bernoulli’s Equation 17.26 Some Observations about the Moody Charts 671
to Steady-State Flows 637 17.27 The Pressure Drop and the Reynolds
16.22.2 Incompressible Flows 637 Number for Circular Pipes 672
xiv Contents
19.12 Corrections to the Pumping 20.6 Forced Convection and Natural Convection 782
Power for Pump Efficiency 747 20.7 Factors That Affect the Convective
19.13 Additional Observations Heat Transfer Coefficient 783
about Pumping Power 748 20.8 Other Methods to Enhance the
19.14 Finding the Pumping Power for the Convective Heat Transfer Coefficient 785
Primary Side of a PWR Steam Generator 748 20.9 An Introduction to the Prandtl Number 785
19.15 Types of Reactor Coolant Pumps 750 20.10 The Kinematic Viscosity and
19.16 Coolant Pump Reliability 750 the Thermal Diffusivity 786
19.17 Jet Pumps and Their Characteristics 751 20.11 Temperature Dependence of
19.18 Pump Start-Up Times 751 the Prandtl Number 787
19.19 Estimating the Start-Up Time for a 20.12 A Physical Interpretation of
RCP from Bernoulli’s Equation 752 the Prandtl Number 787
19.20 Solving the Momentum Equations 20.13 The Reynolds Analogy between
with Loss Coefficients and Friction Heat and Momentum Transfer 788
in a Viscous Coolant Loop 756 20.14 The Velocity Boundary Layer 789
19.21 Pump Performance Parameters 757 20.15 The Thermal Boundary Layer 789
19.22 Performance Curves for Pumps 758 20.16 The Nusselt Number 790
19.23 Matching the Required Head 20.17 Average Subchannel Nusselt Numbers 792
and the Available Head 759 20.18 The Nusselt Number for Laminar Flows 794
19.24 Connecting Reactor Coolant 20.19 Finding the Nusselt Number for
Pumps in Parallel 759 a Constant Axial Heat Flux 794
19.25 Centrifugal Coolant Pumps 761 20.20 Laminar Nusselt Numbers in
19.26 Finding the Torque on the Pump Shaft 764 Reactor Fuel Assemblies 796
19.27 Analogies between Turbines, 20.21 The Entrance Length for Laminar Flows 797
Compressors, and Pumps 764 20.22 Understanding the Nusselt Number and Heat
19.28 Other Types of Power Turbines 765 Transfer Coefficient for Turbulent Flows 798
19.29 Steam Turbines and Their 20.23 Parameters That Affect the
Mechanical Efficiency 767 Turbulent Nusselt Number 799
19.30 Ways to Define Turbine Efficiency 768 20.24 The Mass Flow Rate and the
19.31 Two-Stage Turbines 768 Heat Transfer Coefficient 800
19.32 Steam Turbine Blade Design 769 20.25 Fine-Tuning the Convective
19.33 Impulse Blades in HP Turbines 770 Heat Transfer Coefficient 800
19.34 Reaction Blades in LP Turbines 770 20.26 Boundary Layer Effects 801
19.35 Differences between Impulse 20.27 Liquid Metal Behavior 801
Turbines and Reaction Turbines 771 20.28 Constants Used in Liquid Metal
19.36 Preventing Cavitation in Heat Transfer Correlations 802
Reactor Coolant Pumps 771 20.29 The Reynolds Analogy for
Bibliography 773 Nuclear Heat Transfer 802
Books and Textbooks 773 20.30 Frictional Forces on Reactor Fuel Rods 804
Questions for the Student 773 20.31 Nusselt Numbers and Heat Transfer
Exercises for the Student 774 Coefficients for PWR Fuel Assemblies 804
20.32 Nusselt Numbers and Heat Transfer
Coefficients for LMFBR Fuel Assemblies 805
Chapter 20 20.33 Geometric Correction Factors
Fundamentals of Single-Phase Heat Transfer in for Reactor Fuel Assemblies 808
Nuclear Power Plants 777 20.34 Turbulent Heat Transfer at the
Entrance to the Core 810
20.1 An Introduction to Single-Phase
20.35 Steam Generator Heat Transfer 812
Nuclear Heat Transfer 777
20.36 Typical Steam Generator Heat Transfer Rates 813
20.2 Newton’s Law and the Bulk
Bibliography 814
Fluid Temperature 777
Books and Textbooks 814
20.3 Laminar Heat Transfer 779
Unit Conversion Factors 815
20.4 Turbulent Heat Transfer 780
Questions for the Student 815
20.5 Conduction and Convection 781
Exercises for the Student 816
xvi Contents
Chapter 26 Chapter 27
Core Temperature Fields 1001 Nuclear Hot Channel Factors, the Critical Heat
26.1 Thermal Behavior of the Fuel
Flux, and the DNBR 1039
and the Cladding 1001 27.1 The Boiling Crisis in Water Cooled Reactors 1039
26.2 Mechanistic Thermal Design 1001 27.2 Dryout of the Fuel Rods in PWRs 1039
26.3 A Coolant Energy Balance 1004 27.3 Dryout of the Fuel Rods in BWRs 1039
26.4 Axial Coolant Temperature in the 27.4 Coolant and Cladding Temperatures
Hottest Coolant Channel 1005 in Typical PWR Cores 1039
26.5 Corrections for Changes in the 27.5 The Critical Heat Flux and the Boiling Crisis 1041
Radial Power Distribution 1006 27.5.1 Representative Values for the
26.6 Cladding Temperatures for the Hot Channel 1007 CHF in Modern PWRs 1042
26.7 Finding the Location of the Maximum 27.5.2 Dependence of the CHF on
Cladding Temperature 1008 the Exit Flow Quality 1043
26.8 The Coolant and Cladding Temperatures 27.5.3 Dependence of the CHF on
when the Axial Heat Flux Is Uniform 1008 the Mass Flow Rate 1043
26.9 Temperature Drop from the 27.5.4 Dependence of the CHF
Cladding to the Coolant 1009 on Grid Spacers 1043
26.10 Finding the Temperature Profiles for the Fuel 1010 27.5.5 Dependence of the CHF on
26.10.1 Approach #1: Start with the the System Pressure 1044
Coolant Temperature to Find 27.6 Variation of the CHF and the
the Fuel Temperature 1010 DNBR with Position 1045
26.10.2 Approach #2: Start with the 27.7 Some Boiling Crisis Terminology 1046
Cladding Temperature to Find 27.8 CHF Correlations 1047
the Fuel Temperature 1014 27.9 The Tong Correlation 1047
26.11 Finding the Locations of the Maximum 27.10 The Bernath Correlation 1048
Fuel and Cladding Temperatures 1015 27.11 The Groeneveld Correlation 1049
26.12 Finding the Fuel, the Cladding, and 27.12 The Bowring Correlation 1050
Coolant Temperatures in Real Cores 1016 27.13 The Westinghouse W-3 Correlation 1051
26.13 Coolant Temperatures in a Boiling Core 1017 27.14 The W-3 Shape Correction Factor 1052
26.14 Fuel and Cladding Temperatures 27.15 Understanding Film Dryout in BWRs 1054
in a Boiling Core 1019 27.16 CHF Correlations for BWRs 1055
26.15 Comparing the Behavior of 27.17 The Critical Power Ratio and Its Uses 1058
PWR and BWR Cores 1020 27.18 The CISE-4 Correlation for the CFQ 1059
26.16 Accounting for Statistical Uncertainties 27.19 The GEXL Correlation 1059
in the Core Temperature Profiles 1021 27.20 The Hench–Gillis Correlation 1060
26.17 SPDs and Engineering Uncertainty Factors 1022 27.21 Thermal Design Limits and
26.18 Reactor Nodal and Subchannel Analysis 1024 Regulatory Requirements 1063
26.19 Using Super Cells for Reactor 27.22 PWR DNBR Limits 1064
Thermal-Hydraulic Analysis 1025 27.23 BWR DNBR Limits 1065
26.20 Conservatism in Subchannel Models 1026 27.24 Comparison of Critical Quality
26.21 Applying the Homogenous and CHF Correlations 1066
Equilibrium Model Equations to 27.25 Other Thermal Limitations on Core Design 1067
Reactor Subchannel Design 1031 27.26 Nuclear Hot Channel Factors 1068
26.22 Constructing Three-Dimensional 27.27 Engineering Uncertainty Factors and Their
Core Temperature Profiles 1033 Effect on the Nuclear Hot Channel Factor 1070
Bibliography 1033 Bibliography 1071
Books and Textbooks 1033 Books and Textbooks 1071
Web References 1034 Web References 1071
Additional References 1034 Additional References 1071
Questions for the Student 1036 Questions for the Student 1071
Exercises for the Student 1037 Exercises for the Student 1073
xx Contents
32.6 Common Shapes and Sizes for 33.6 Design Limits for the Fuel 1235
Containment Buildings 1205 33.7 Design Limits for the Cladding 1236
32.7 BWR Containment Buildings 1205 33.8 DNB Limits 1236
32.8 Mark I Containment 1206 33.9 Common Design Assumptions 1236
32.9 Mark II Containment 1207 33.9.1 Design Assumption I 1236
32.10 Mark III Containment 1207 33.9.2 Design Assumption II 1236
32.11 Suppression Pool and Other 33.9.3 Design Assumption III 1237
Containment Systems 1208 33.10 Effects of Rod Bowing and Core
32.12 Understanding the Nuclear Flow on DNBR Estimates 1237
Steam Supply System 1209 33.11 Effects of Flow Blockages
32.13 Electrical Generators 1211 on DNBR Estimates 1238
32.14 Condensers 1213 33.12 More Categories of DBAs 1240
32.15 Feedwater Heaters 1214 33.12.1 Reactivity Initiated Accidents 1240
32.16 Cooling Towers 1215 33.12.2 Loss-of-Flow Transients 1240
32.17 Spent Fuel Pools and Their Function 1216 33.12.3 Loss of Coolant Accidents 1241
32.18 NSSS for Westinghouse PWRs 1217 33.12.4 Accidents Caused by
32.19 Principles of Radiation Protection Equipment Failure 1241
Applied to Nuclear Power Plants 1218 33.12.5 Fuel-Handling Accidents 1241
32.20 Radiation Barrier Analysis 33.12.6 Other Types of Reactor Accidents 1241
and Its Implications 1219 33.13 Additional Safety Guidelines 1241
32.21 Treating the Fuel as a Radiation Barrier 1219 33.14 Chances of an Accident Actually Occurring 1242
32.22 Treating the Cladding as a 33.15 Probabilistic Risk Assessment
Radiation Barrier 1220 and Accident Event Trees 1242
32.23 Treating the Reactor Cooling 33.16 Reactor Safety Systems 1243
System as a Radiation Barrier 1220 33.17 Computer Programs for Analyzing
32.24 Treating the Reactor Pressure Reactor Accidents 1243
Vessel as a Radiation Barrier 1220 33.18 Probabilistic Risk Analysis (or PRA) Codes 1245
32.25 Treating the Containment Building 33.19 Fuel Behavior Codes 1245
as a Radiation Barrier 1221 33.20 Reactor Kinetics Codes 1245
32.26 The Final Line of Defense: The Site 33.21 Reactor Thermal-Hydraulics Codes 1245
Location and Exclusion Zone 1222 33.22 Severe Accident Analysis Codes 1246
32.27 Preventing the Buildup of Hydrogen Gas 1222 33.23 DBA Codes 1246
32.28 Pressure, Temperature, and Energy 33.24 Health Effects/Dose Calculation Codes 1247
Relationships in Containment 33.25 Radionuclide Transport Codes (for License
Building Design 1224 Termination and Decommissioning) 1247
Bibliography 1226 33.26 Additional Information Regarding Reactor
Books and Textbooks 1226 Accident Analysis and Safety Codes 1247
Web References 1226 Bibliography 1248
Additional References 1226 Books and Textbooks 1248
Questions for the Student 1226 Web Reference 1248
Exercises for the Student 1228 Questions for the Student 1248
Exercises for the Student 1250
Chapter 33
Thermal Design Limits, Operating Limits, and Chapter 34
Safety Limits 1229 Response of a Containment Building to a Reactor
LOCA 1251
33.1 Thermal Design Limits 1229
33.2 Design Limits, Safety Limits, 34.1 Containment Buildings and Their Function 1251
and Operational Limits 1229 34.2 Containment Building Behavior
33.3 American Nuclear Society during a LOCA 1252
Classifications of Reactor Accidents 1231 34.3 LOCA Heat Sources and Their Time Lines 1252
33.4 The NRC’s Classification of 34.4 Heat Sources Preceding a Reactor LOCA 1252
Reactor Accidents 1233 34.5 Heat Sources during a Reactor LOCA 1253
33.5 Design Limits for the Core 1234 34.6 Heat Sources after a Reactor LOCA 1253
Contents xxiii
xxv
An Overview of the Book
Practically speaking, this book can be subdivided into seven distinct sections or “parts”. Each section can then be subdivided into a
number of related chapters.
Content of Part I
In part I, which encompasses Chapters 1 to 5, all modern fission reactor types are discussed. This includes thermal water reactors
and fast reactors. Specifically, it includes pressurized water reactors (PWRs), boiling water reactors (BWRs), the CANada Deuterium
Uranium (CANDU) pressurized heavy water reactor, liquid metal fast breeder reactors (LMFBRs), gas reactors, mobile power reac-
tors, and military reactors. The materials used to build a practical fission reactor are also discussed. Within the framework of these
chapters, the architecture of a modern nuclear power plant is discussed. Various components of a plant including the core and the
nuclear steam supply system (or NSSS) are discussed. Great attention is paid to how the heat produced in these reactors is converted
into useful work and electric power. The reactor piping system, the heat exchangers, the steam generators, the coolant pumps, and
other important reactor components are thoroughly discussed. Thermodynamic efficiencies and operating characteristics are pre-
sented for each reactor type. Chapter 5 then discusses how thermal energy is produced from the nuclear fission process. It discusses
all common types of nuclear fuels and examines the relationship between the neutronic design of the core and its structural and
thermal-hydraulic components. A wide range of practical problems is further included to illustrate the thermal-hydraulic, structural,
and material compromises that must be made in the design of real reactors. All five chapters are essentially self-contained, and they
are written to set the stage for later chapters in which the thermal design of the core, the NSSS, and the plant as a whole are discussed.
Content of Part II
Following this brief but non-presumptive introduction to nuclear reactors and nuclear power plant design, Chapters 6–9 introduce
the reader to part II of the book. Here the reader is introduced to basic thermodynamic principles which are required to understand
the practical aspects of nuclear heat transfer and fluid flow. These concepts and principles include the laws of thermodynamics and
heat transfer, thermodynamic properties and equations of state, and thermodynamic state diagrams. The relationships between the
thermodynamic variables required to perform an energy balance on each reactor component are discussed. After introducing these
principles, the reader is then exposed to the Steam Tables for the first time. The steam tables are then used to illustrate the state of the
coolant (in this case water and steam) as it enters and leaves the core. This subsequently leads to a discussion of the NSSS and reactor
thermal cycles. Reactor thermal cycles are compared for all major reactor types, and methods are provided for calculating a plant’s
thermodynamic efficiency. Carnot heat engines and similar theoretical constructs are introduced. The reader is then introduced to
reactor heat exchangers and steam generators. Parallel flow and counter flow heat exchangers are discussed. Expressions are derived
for optimizing their heat transfer rates. This leads to a discussion of common techniques to improve the efficiency of reactor thermal
cycles such as reheat and regeneration. Steam turbines, steam generators, feedwater heaters, condensers, and reactor cooling towers
and cooling ponds are all examined and discussed. This is done within the context of a practical engineering framework that allows
the reader to calculate the thermodynamic efficiencies of various reactor components. This discussion is then concluded with a com-
parison of hybrid power plant thermal cycles.
xxvii
xxviii An Overview of the Book
to the steady-state heat conduction equation are provided for cylindrical fuel rods with a central void. The effect of burnup on the
temperature profiles is discussed. Fuel rod temperature limits and regulatory requirements are discussed. Temperature profiles are
found in objects such as radiation shields with exponential heat sources. The reader is then introduced to time-dependent heat con-
duction equation and the Heisler charts. General solutions to the time-dependent heat conduction equation are presented for common
reactor components. For more complex geometries, a lumped parameter approach is used. As far as the publisher is aware, this is the
only detailed discussion of time-dependent nuclear heat transfer in a mainstream nuclear engineering textbook.
Content of Part IV
Part IV of the book, which includes Chapters 13 through 19, is then used to introduce the reader to the subject of reactor fluid flow.
Following a brief but non-presumptive discussion of fluid mechanics, Chapters 13, 14, and 15 discuss the differences between fluid
statics and fluid dynamics. Chapter 13 discusses compressible and incompressible flows, laminar and turbulent flows, and other basic
fluid mechanical concepts Viscous and inviscid flows, as well as the kinematic and dynamic viscosities are discussed. These concepts
are then used to introduce the reader to the Reynolds number, the Prandtl number, and the Mach number. A brief introduction is also
provided to the concept of surface tension. Fluid behavior is discussed in both PWRs and BWRs. The reader is then introduced to the
Lagrangian and Eulerian views of fluid mechanics, which inexorably lead to the material derivative. Finally, path lines, streamlines,
and streak lines are discussed. Chapter 15 then uses these concepts to derive the fluid conservation equations for mass, energy, and
momentum from first principles. This subsequently leads to a discussion of lateral flow and cross flow in reactor fuel assemblies.
The Navier–Stokes equations are derived from first principles and introduced to the reader for the first time. These equations are
then used to derive Cauchy’s equation, Bernoulli’s equation, and Euler’s equation. Newtonian and non-Newtonian fluid mechanics is
discussed. Finally, various boundary conditions are applied to the conservation equations in reactor fuel assemblies. These include
pressure boundary conditions, flow boundary conditions, and combined boundary conditions. The conservation equations are then
cast into a more tractable form using a lumped parameter approach. Chapter 16 discusses the dynamics of single-phase flow in nuclear
power plants. Bernoulli’s equation is applied to the reactor piping system. The acceleration pressure drop, the gravitational pressure
drop, and frictional pressure drops are introduced and compared. Fluid mechanical concepts are then applied to nozzles, valves, dif-
fusers, and other components of the reactor piping system. Reactor coolants are classified and discussed in Chapter 17. This leads to
a colorful discussion of laminar and turbulent flows. Entrance effects and boundary layer development are discussed in reactor fuel
assemblies. The Darcy equation is introduced for the first time. This leads to a discussion of turbulent friction factors and Moody
diagrams. The Blasius, McAdams, and Petukhov correlations are applied to turbulent flows in reactor piping systems. The concepts of
the equivalent diameter and the equivalent annulus are discussed. The flow of coolants through PWR and LMFBR fuel assemblies is
discussed in considerable detail. In Chapter 18, isothermal and non-isothermal flows are discussed as well as the effects of tempera-
ture and surface roughness on the single-phase friction factor. This leads to a robust discussion of grid spacers, wire wrap spacers,
orifice plates, and grid spacer loss coefficients. Two- and three-dimensional core flows are discussed. This then provides a convenient
introduction to inter-assembly flow, intra-assembly flow, and cross flow. The components of the single-phase pressure drop are quanti-
fied for typical reactor cores. Finally, drag forces are calculated for nuclear fuel rods. This inexorably leads to a discussion of reactor
coolant pumps and power turbines. Chapter 19 provides an interesting comparison of all major coolants used in nuclear power plants
today. These coolants include light water, heavy water, industrial gases, and liquid metals. The thermophysical properties of these
coolants are then used to lay the groundwork for a comprehensive discussion of reactor pumping requirements (i.e., pump work). This
discussion is used to estimate the pumping power of a reactor coolant pump and the heat removal rate from the core. The design of
reactor coolant pumps is discussed in considerable detail. Both centrifugal pumps and jet pumps are discussed. This ultimately leads
to a discussion of pump capacity, best efficiency points, and pump performance curves. Finally, Chapter 19 concludes with a com-
prehensive discussion of the steam turbines used in nuclear power plants. This includes single-stage and two-stage turbines and their
blade configurations. An expression is derived for estimating the enthalpy drop across a set of impulse turbine blades.
Content of Part V
Chapters 20 through 26 constitute part V of the book. These chapters are used to introduce the reader to the subject of nuclear heat
transfer. Chapters 20 through 22 discuss single-phase nuclear heat transfer, while Chapters 23 through 25 discuss two-phase nuclear
heat transfer. These chapters present the most comprehensive introduction to nuclear heat transfer that has ever been attempted
in a mainstream nuclear engineering textbook. The discussion is broad, engaging, and deep. Chapter 20 uses a well-structured
pedagogical approach to introduce the reader to single-phase nuclear heat transfer. Both natural convection and forced convection
are discussed. Thermal boundary layers, velocity boundary layers, and the Reynolds analogy are explored. The Nusselt number is
defined, and its relationship to the convective heat transfer coefficient is quantified. Actual heat transfer coefficients and Nusselt
numbers are calculated for the reactor piping system and reactor fuel assemblies. The Nusselt numbers and heat transfer coefficients
An Overview of the Book xxix
for metallic coolants are also found. Nusselt numbers and heat transfer coefficients are calculated for the subchannels in PWR and
LMFBR fuel assemblies. Steam generator heat transfer coefficients are also discussed. Chapter 21 presents the most commonly
used correlations for single-phase nuclear heat transfer. These correlations include the Dittus–Boelter correlation and the Sieder–
Tate correlation. These correlations are then applied to reactor fuel assemblies and other plant components. Geometrical correction
factors are presented so that the correlations for circular flow channels to be applied to subchannels that are not circular in shape.
These correction factors are based on the Weisman, Markoczy, and Presser correlations. The Lyon–Martinelli and Seban correla-
tions are introduced for liquid metal heat transfer. Single-phase heat transfer coefficients are discussed for common reactor types.
Chapter 22 then extends this discussion to natural convection, where power to the coolant pumps is lost. In this context, passive heat
removal systems are discussed. Natural convection is analyzed in spent fuel pools. The Grashof number and the Rayleigh number
are defined, and these parameters are then used to predict the heat transfer coefficient in steam generators and PWR fuel assemblies
when on-site power is lost.
After completing a comprehensive discussion of single-phase heat transfer, Chapter 23 presents a robust introduction to two-phase
flow. Using this framework, the processes of evaporation, boiling, and condensation are discussed. Here, the reader receives a com-
prehensive introduction to the topic of bubble formation and growth. This discussion is then used to introduce the reader to the sub-
jects of pool boiling and flow boiling in nuclear power plants. Within this context, two-phase flow and the emulation of steam–water
mixtures are discussed. The quality and the void fraction are derived for two-phase mixtures using the assumption of thermodynamic
equilibrium. Expressions involving the slip ratio are then created to relate the equilibrium quality to the equilibrium void fraction.
When the slip ratio cannot be used to establish a definitive relationship between the quality and the void fraction, a drift flux model
with a local drift velocity is used. Conventional correlations for the void fraction are then compared to drift flux correlations. These
correlations are then used to introduce the reader to the flow regimes in a boiling water reactor core. This leads to a comprehensive
discussion of nucleate boiling, bulk boiling, slug flow, churn flow, annular flow, and mist flow. Flow regime maps are used to illustrate
the conditions under which these various types of flow occur. For light water reactors, this then leads to a discussion of the boiling
and non-boiling heights. Two-phase pressure drops are discussed for homogeneous mixtures and nonhomogeneous ones. This then
leads to a discussion of the two-phase multiplier and the Martinelli–Nelson correlation. The predictions of homogenous equilibrium
model (HEM) are compared to those of separated flow models. The Friedel correlation is introduced to find the two-phase multiplier
in BWR cores. The acceleration pressure drop is calculated using a variety of techniques for two-phase mixtures. The single- and
two-phase pressure changes due to sudden area reductions and expansions are discussed. These subjects then lead to a comprehensive
discussion of orifice flow, the vena contracta, and single- and two-phase orifice flow at subsonic speeds. Pressure losses in PWR and
BWR fuel assemblies are calculated with a knowledge of these effects.
Chapter 24 provides a comprehensive introduction to the subject of two-phase heat transfer. It begins with a discussion of pool
boiling and gradually proceeds to a discussion of flow boiling. The shape of the pool boiling curve is presented and explained. Each
section of the curve is analyzed for common reactor coolants such as water. The curve is segmented into the liquid convection region,
the nucleate boiling region, the critical heat flux region, the transition boiling region, the film boiling region, the mist flow region,
and the superheated vapor region (when applicable). Correlations for the convective heat transfer coefficient are presented for each
regime. For pool boiling, the Fishenden and Saunders correlation, the Rohsenow correlation, and the Berenson correlation are used.
Specialized correlations such as the Kutateladze correlation and the Zuber correlation are used to predict the maximum and the mini-
mum heat flux at strategic points (the critical heat flux point and the Leidenfrost point) along the pool boiling curve. The Leidenfrost
effect is discussed. The effects of surface roughness on the boiling heat transfer coefficient are explored. This is followed by an exten-
sive discussion of pool boiling in nuclear heat exchangers and spent fuel pools. Boiling is discussed over hot vertical and horizontal
tubes. The discussion is extended to include large tube banks and rod bundles. The dependence of the boiling heat transfer coefficient
on the ambient pressure is explored using the Mostinski equation and the Principle of Corresponding States. These concepts are then
applied to reactor components using the critical pressure ratio and the Cooper correlation. After completing a discussion of pool boil-
ing, the discussion is then redirected to the subject of flow boiling. The boiling heat transfer coefficient for forced convection is then
discussed as a function of the equilibrium quality. This subsequently leads to a discussion of the specific enthalpy, the equilibrium
quality, and the void fraction profiles in boiling reactor cores. Chapter 24 concludes with a discussion of fuel rod dryout which allows
the concepts of the DNB and film dryout to be introduced in a non-presumptive way. It also sets the stage for the flow boiling correla-
tions which are presented in Chapter 25.
In Chapter 25, many specific correlations are presented for the two-phase heat transfer coefficient. As the text reveals, these
correlations may simultaneously contain convective, vapor, and radiative components. Heat transfer correlations are presented for
forced convection which can then lead to nucleate boiling, bulk boiling, transition boiling, film boiling, and eventually film dryout.
Correlations discussed include the Bernath correlation, the Klimenko correlation, the Kandlikar correlation, the Chen correlation,
and the Groeneveld correlation. Other subjects of interest include the Chen correlation’s enhancement and suppression factors, the
Kandlikar correlation’s fluid-dependent parameter, and the Groeneveld correlation’s geometric correction factors. Considerable effort
is expended to discuss the advantages of each correlation as well as their ranges of applicability. In addition to equilibrium models,
xxx An Overview of the Book
non-equilibrium models are used to predict the onset of nucleate boiling point (the ONB point), the NVG (net vapor generation)
or bubble detachment point, and the point of zero equilibrium quality. The Griffith correlation, the Browning correlation, and the
Saha–Zuber correlation are used to determine the NVG point. The dependence of the equilibrium quality on the axial mass flux is
discussed. The void fraction is estimated using equilibrium and non-equilibrium models. Chapter 26 then extends these approaches
to predict the core temperature profiles in PWRs and BWRs. Nuclear hot channel factors, engineering uncertainty factors, and manu-
facturing uncertainty factors are discussed. Fuel pin, cladding, and coolant temperature profiles are presented in both two and three
dimensions. The effects of turbulent mixing on the core temperature profiles are discussed. The reader is introduced to the process
of mechanistic thermal design.
Content of Part VI
This part of the book, which includes Chapters 27 through 31, then introduces the reader to the subject of reactor safety. It introduces
the reader to the critical heat flux (the CHF), the departure from nucleate boiling ratio (DNBR), and the critical power ratio (CPR).
In Chapter 27, the reader is first introduced to the DNBR and its physical significance. Popular correlations are presented for calcu-
lating the critical heat flux in reactor cores. These correlations include the Tong correlation, the Bernath correlation, the Bowring
correlation, the Westinghouse W-3 correlation, and the Groeneveld correlation. The shape correction factor in the W-3 correlation is
discussed. The Groeneveld correlation, which is implemented as a lookup table, is also used in many famous reactor safety analysis
programs. For BWRs, the CISE-4 correlation, the GE or GEXL correlation, and the Hench–Gillis correlation are used to predict
the critical flow quality. Various regulatory requirements are introduced, including the value of the DNBR that is used in the United
States and most of the Western world to license commercial PWRs and BWRs. Nuclear hot channel factors, engineering uncertainty
factors, and the critical power ratio (CPR) are discussed within the context of this perspective. Chapter 28 introduces the reader to
the concept of thermal design limits, operating limits, and safety limits. Using these concepts different types of reactor accidents are
classified using the IAEA, the ANS, and NRC scales of accident severity. Design limits are discussed for both the fuel and the clad-
ding. Chapter 28 also derives Smoluchowski’s particle transport equation which describes the entrainment and transport of radioac-
tive particles during small, medium, and large break LOCAs. The wedge model of atmospheric dispersion is then used predict the
paths of radioactive clouds and plumes that escape from the containment building entirely. The findings of the WASH-1400 report
are presented and discussed.
Chapter 29 then discusses equilibrium and non-equilibrium flows, critical flow, and choke flow. The discussion begins with sub-
sonic single-and two-phase flow, and then progresses to single-and two-phase orifice flow. The effect of the back pressure on the mass
flow rate is discussed. Critical two-phase flow is analyzed for mixtures which may have multiple sound speeds. Critical flow rates are
established for both “long” and “short” reactor coolant channels. HEM and slip equilibrium models are developed for two-phase criti-
cal flow. These models include Richter’s model and Fauske’s model. Four, five, and six equation models are presented for dealing with
separated and segmented flows where thermal equilibrium has not yet been achieved. The discussion of these models is robust and
comprehensive. Finally, the HEM equations are applied to reactor subchannel design. The resulting set of equations is then applied
to popular reactor analysis codes such as RELAP, TRACE, COBRA, and VIPER. The computational assumptions adopted by these
codes are then compared to those of conventional computational fluid dynamics codes (CDF codes) such as ANSYS and FLUENT.
Chapter 30 introduces the reader to the subject of hypothetical reactor accidents (or HCDAs) including DBAs and LOCAs.
Design limits are discussed for LOCAs within a comprehensive conceptual framework. Several types of LOCAs are discussed
including small-break LOCAs, medium-break LOCAs, and large-break LOCAs. The WASH-1600 report is discussed and its con-
clusions are presented. The expected outcomes of common reactor LOCAs are discussed. The blowdown and refill stages of typical
reactor LOCAs are explored. The evolution and timing of typical LOCAs are discussed. This includes an analysis of both large- and
medium-break LOCAs. Reactor safety systems to minimize the consequences of reactor LOCAs are discussed for both PWRs and
BWRs. Regulatory limits regarding the fuel temperature and the cladding temperature are discussed. The exothermic zirconium–
water reaction is formally introduced to the reader, and hydrogen generation rates during reactions of this type are calculated. The
reader is then introduced to the Baker–Just equation which governs the rate of fuel rod failures during hypothetical reactor accidents.
Important temperature thresholds during common reactor LOCAs are discussed. Finally, Chapter 31 concludes with a discussion
of density waves and other hydrodynamic instabilities. These instabilities may be of either the local or global variety. Static insta-
bilities such as Ledinegg instabilities are discussed first and then dynamic instabilities such as density wave oscillations are added
to the discussion. Hydrodynamic instabilities are discussed in BWR cores. These instabilities are generalized to include coupled
power-flow oscillations and instabilities introduced by possible control system errors. The nature and origin of density waves are
discussed. This eventually leads to a discussion of stability maps based on the phase change number, the subcooling number, and
Ishii’s stability criterion. Flow oscillations are discussed during reactor Start-up and shutdown. The reader is introduced to operating
maps and the void coefficient of reactivity. Regulatory requirements are then discussed for mitigating the effects of these hydrody-
namic instabilities.
An Overview of the Book xxxi
and it will be corrected in the next edition. Finally, we hope that you will enjoy the warm and inviting style, and we hope that it will
inspire you to embrace the renaissance in the nuclear power industry that is occurring in the world today. We hope that you will enjoy
reading this book, and that you will find it to be an interesting and rewarding experience.
Sincerely,
Dr. Bob Masterson. ScD (MIT)
Cambridge, Massachusetts
July 4, 2019
Further comments and suggestions regarding this book can be addressed to the author at his global email address, which is
[email protected].
Author
Robert E. Masterson holds an MS and PhD (ScD) in nuclear science and engineering from the Massachusetts
Institute of Technology, Cambridge, Massachusetts, and a BS in p hysics from the University of Notre
Dame, Notre Dame, Indiana. The author has published extensively in the Transactions of the ANS and is a
member of the American Nuclear Society.
Dr. Masterson has served over the years as affiliate professor of nuclear science and e ngineering at
the Virginia Polytechnic Institute, Blacksburg, Virginia, and has an extensive background in the areas of
reactor thermal-hydraulic analysis, heat transfer, and fluid flow. The author has over 20 years of experience
in the field of nuclear science and engineering and worked for Westinghouse Nuclear Energy Systems at
the Hanford site. He is president of a major technology consulting firm, and was an early contributor to
the COBRA series of reactor analysis codes. He is also the author of the two popular nuclear engineering
textbooks
1. Nuclear Engineering Fundamentals: A Practical Perspective.
2. An Introduction to Nuclear Reactor Physics.
which are also available from CRC Press. In graduate school at MIT, he was the officemate of Mujid Kazimi and a student of Neil
Todreas, the co-authors of the popular reactor thermal-hydraulics book Nuclear Systems, which is also available from CRC Press.
xxxiii
1
Nuclear Power in the World Today
1.1 Types of Reactors and Their Characteristics
There are five basic types of reactors in relatively widespread use in the world today. These reactors use different combinations of
coolants and different combinations of uranium and plutonium to generate the electrical power the world needs. The ability of a reac-
tor to sustain a nuclear chain reaction over a long period of time is dependent upon the type of coolant, the composition of the core,
and the enrichment of the fuel that is used. All reactors must be able to produce more neutrons than they consume for the fission chain
reaction to continue over time. Control rods are then used to absorb the excess neutrons and keep the chain reaction under control.
Conceptually, it is possible to subdivide the commercial nuclear power reactors in the world today into five broad categories:
1. Pressurized water reactors (PWRs)
2. Boiling water reactors (BWRs)
3. Heavy water reactors (HWRs)
4. Gas-cooled reactors (GCRs)
5. Liquid metal fast breeder reactors (LMFBRs)
PWRs and BWRs use light water (i.e., ordinary water) to cool their cores, whereas HWRs, like the CANada Deuterium Uranium
(CANDU) reactor, use heavy water in which the hydrogen atoms that are bound to the oxygen atom have an extra neutron in their
nucleus. Heavy water absorbs fewer neutrons than light water does, and because of this, reactors that are cooled with heavy water do
not need enriched uranium to be able to operate successfully. However, they can also use enriched uranium if the need arises. All
commercial power reactors have a certain number of design features in common, including the need for fuel rods, control rods, and
some type of liquid or gaseous coolant to cool the core. With the exception of GCRs (such as the advanced gas reactor [AGR] and the
high-temperature, gas-cooled reactor [HTGR]), which can sometimes use the Brayton cycle in their primary loop, all other reactors
use the Rankine thermal cycle to generate steam that is sent to the power turbines. PWRs, HWRs, and LMFBRs have two coolant
loops (a primary loop and a secondary loop), whereas BWRs and GCRs have only one coolant loop. As we will mention in Chapter 4,
some LMFBRs also have a third coolant loop (called an intermediate loop) to isolate the core from the steam generators (SGs). This
prevents the liquid metal from interacting with the water and introducing radioactivity into the steam that drives the power turbines
in the event of a pipe rupture.
1.2
Number of Power Reactors around the World
The number of power reactors in the world today has increased steadily since nuclear power was first introduced in the 1950s.
Figure 1.1 shows how the number of power reactors has changed between the time that they were first introduced and the present day.
Today, there are approximately 436 nuclear power plants in operation in 31 countries. As we mentioned previously, these nuclear
power plants are usually implemented using a base-loading model, since the fuel is a very small part of the total cost of producing
electric power from these plants. These 436 nuclear power plants have an average output of about 374 GW, and there are an additional
65 nuclear power plants under construction with a projected capacity of about 65 GW. The majority of these new plants are being built
in China, India, and Russia (see Figure 1.2; Table 1.1).
By far the most common type of commercial power reactor in use in the world today is the PWR, which is manufactured and
designed by Westinghouse, AREVA, and a number of other reactor vendors. Table 1.2 shows the breakdown of the current installed base
of commercial power plants by reactor type. Notice that PWRs represent about two-thirds of the world’s total installed base of nuclear
generating capacity at this time. BWRs, which are manufactured and designed by the General Electric Company (GE), are the next
most common type of reactor in the world. The percentage of electrical power generated in each country by these power plants is also
shown in Figure 1.3. Notice that nuclear power generates about 20% of the total electrical power output of the United States, whereas
it generates almost 80% of the electrical power produced in France. Countries that do not have a great deal of coal and oil tend to rely
1
2 Nuclear Power in the World Today
400
Power (Giga-watts)
300
y
y d capacit
cit Realize
pa 200
d ca
lle
sta
In 100
0
Chernobyl Fukushima 500
Three mile
Island
400
power reactors
ors
Number of
act n 300
a l re uctio Active reactors
ion str
dit on 200
Ad der c
un 100
0
1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015
FIGURE 1.1 The number of nuclear power plants operating in the world as a function of time. (Data supplied by the World
Nuclear Association—see www.world-nuclear.org/.)
China
Russia
India
United States
South Korea
Japan
Taiwan
Pakistan
Slovenia
Ukraine
Argentina
Germany
Brazil
Finland
France
UAE
0 5 10 15 20 25 30
Number of reactors
FIGURE 1.2 The number of nuclear reactors under construction, worldwide, by country.
(Courtesy of the IAEA. Date: January 2013.)
1.2 Number of Power Reactors around the World 3
TABLE 1.1
Number of Nuclear Power Plants in the World Today (Either in Operation or under Construction)
TABLE 1.2
Number of Commercial Power Reactors in the World Today as a Function of Their Design Type (PWR, BWR, AGR, LMFBR, etc.) as of
January 2010
Percent of World
Reactor Type Main Countries Number GW(e) Fuel Coolant Moderator Total (%)
PWR Unites States, 271 270.4 Enriched UO2 Water Water
France, Japan,
Russia, China
BWR United States, 84 81.2 Enriched UO2 Water Water
Japan, Sweden
Pressurized HWR Canada 48 27.1 Natural UO2 Heavy Heavy
“CANDU” (PHWR) water water
GCR (AGR and Magnox) United Kingdom 17 9.6 Natural U (metal), CO2 Graphite
enriched UO2
Light water graphite reactor Russia 11 + 4 10.4 Enriched UO2 Water Graphite
(RBMK and EGP)
Fast neutron reactor Russia 1 0.6 PuO2 and UO2 Liquid None
(LMFBR) sodium
Total number of reactors in 436 399.3 100
operation
Source: Courtesy of the World Nuclear Association.
more heavily on nuclear power than other countries where these conventional resources are readily available (Figure 1.4;
Table 1.2). Because of this, Saudi Arabia and most Arabic countries do not generate any of their electrical power using
nuclear power plants, whereas countries like France and Switzerland rely on them quite heavily. However, Saudi Arabia
and the United Arab Emirates (UAE) have both expressed an interest in building a large number of nuclear power plants
in the future. More information about the role of nuclear power in the world today can be found at the following URL:
http://www.euronuclear.org/info/encyclopedia/n/nuclear-power-plant-world-wide.htm
1.3
Power Reactor Architectures
The basic architecture of each type of reactor is shown in Figure 1.5. In most cases, the core can be thought of as a
nuclear furnace that generates the heat that is supplied to the nuclear steam supply system (NSSS). The neutronic char-
acteristics of the core are very type specific, and they also affect how easy (or hard) a reactor is to control. Generally
speaking, water reactors are easier to control than LMFBRs because they have a “softer” neutron energy spectrum.
GCRs tend to fall somewhere between these two extremes. Nearly all of these reactor types (except LMFBRs) rely on
the coolant that flows past the fuel rods to slow down the neutrons that are needed to sustain the nuclear chain reaction.
In these reactors, the coolant is also known as “the moderator” (because it moderates or reduces the speed of the fis-
sion neutrons). The moderator is the first way that the nuclear chain reaction is controlled. If the moderator were to be
1.3 Power Reactor Architectures 5
France
Belgium
Slovakia
Ukraine
Hungary
Slovenia
Switzerland
Sweden
South Korea
Armenia
Czech Republic
Bulgaria
Finland
Spain
United States
Taiwan
Romania
Japan
Germany
United Kingdom
Russia
Canada
South Africa
Argentina
Pakistan
India
Mexico
Netherlands
Brazil
China
0 20 40 60 80
Percentage
FIGURE 1.3 The share of power in each country generated by nuclear power plants. (Courtesy of the IAEA. Date: January 2013.)
suddenly removed from the core of a thermal water reactor, the fission process would shut down, and the power genera-
tion rate would be suddenly and substantially reduced. (The only power that would be produced would then be the power
caused by the radioactive decay heat.) However, the same is not true for LMFBRs because the loss of liquid sodium
(a weak neutron moderator) can lead to very large reactivity insertions. (This is discussed in more detail in Chapter 4.)
Hence, the type of coolant used, its flow rate, its temperature, and its density all have a great deal to do with how the
core behaves. The enrichment of the fuel and the size of the fuel rods determine how hot the fuel rods can get and how
much heat must be removed from the core. The flow rate, the temperature, and the pressure of the coolant then determine
the thermodynamic efficiency of the plant. In PWRs, HWRs, AGRs, and LMFBRs, the hot coolant leaves the core and
transfers its heat to an intermediate device known as a steam generator. The SG then transfers this heat to the secondary
loop where it causes the water in the secondary loop to boil and to turn into relatively dry steam. The steam then turns
the power turbines to generate the electricity that is produced. The primary loops and the secondary loops in these plants
are completely separate from one another.
The NSSS in all five of these reactor types uses a Rankine cycle to recondense the wet steam after it leaves the power
turbines. Waste heat is rejected from the plant by passing it through a device called a condenser. The waste heat then
makes its way to a cooling pond, a lake, a river, or a cooling tower. This is essentially how a power plant thermal cycle
works. (We will have more to say about power plant thermal cycles in Chapter 9.) The cooler water (or liquid sodium)
is then returned to the core to be reheated again. Both the primary and secondary loops have different flow rates, and
the flow rates must be balanced so that the secondary loop removes exactly the same amount of heat delivered by the
primary loop. The core deposits its heat in the primary loop, and the process continues until the reactor is shut down.
BWRs use exactly the same process except that the secondary loop is removed from the plant entirely. In BWRs, the
steam generated in the core is sent directly to the power turbines, and so BWRs do not require SGs. (However, they still
use a steam dryer that is located at the top of the pressure vessel.) Eliminating the secondary loop helps to reduce the cost
of building the plant because it is no longer necessary to have two separate loops. This also eliminates the need to build
6 Nuclear Power in the World Today
United States
France
Japan
Russia
South Korea
India
Canada
China
United Kingdom
Ukraine
Sweden
Germany
Spain
Belgium
Taiwan
Czech Republic
Switzerland
Finland
Hungary
Slovakia
Pakistan
Argentina
Brazil
Bulgaria
Mexico
Romania
South Africa
Armenia
Netherlands
Slovenia
Iran
0 20 40 60 80 100 120
Number of reactors
FIGURE 1.4 The number of nuclear reactors in operation, worldwide, by country. (Courtesy of the IAEA. Date: January 2013.)
two or three SGs, which can be very large, complex, and expensive devices. BWRs are also based on a Rankine cycle
because they use liquid water as the coolant. AGRs use a gaseous coolant, and so they can theoretically use another plant
thermal cycle known as the Brayton cycle. The major difference between these two cycles is that the working coolant
in a Rankine cycle boils (i.e., it undergoes a change of phase), whereas in a Brayton cycle, it is a simple gas that does
not. However, most AGRs use the Rankine cycle just like thermal water reactors do.
1.4
Power Reactors and Their Design Features
Each of these five reactor types has a slightly different type of fuel assembly, and the number of fuel assemblies can
vary quite a bit depending upon the reactor and its power rating. Their size, design, and power density have also varied
over time as new designs have been introduced to replace older ones that have become obsolete. In this section, we
would like to compare and contrast the individual fuel assemblies that are used in each reactor type. The design of a fuel
assembly and the number of fuel assemblies in the core have a big impact on how the core behaves (from a neutronic,
thermal-hydraulic, and control perspective). In general, reactor vendors have never standardized on a “standard fuel
assembly,” and so each vendor tends to use its own fuel assembly with its own proprietary features and fuel. However,
the designs usually have a great deal in common. The similarities and differences in their basic design features are
described in Table 1.3.
1.5
Schematic of a Nuclear Power Plant
Sometimes, it is easier to visualize what a nuclear power plant looks like by referring to the diagrams shown in Figure 1.6.
These diagrams are representative of the size and complexity of most large nuclear power plants today. Notice that the
reactor is only one of a large number of other components within the plant, and some of these components include the
1.5 Schematic of a Nuclear Power Plant 7
1. PWR
2. BWR
3. CANDU
4. AGR
5. LMFBR
FIGURE 1.5 The basic architectures of different reactor types: PWR, BWR, CANDU, Canadian HWR, LMFBR, and AGR.
piping system, the SGs, the power turbines, the spent fuel pool, the containment building, and the condenser. Aside from
the nuclear reactor and the containment building, most nuclear reactors are remarkably similar in design to large coal-
fired power plants. Of course, the power source is different, but after the coolant is heated, it must pass through almost
the same set of components that a coal-fired power plant would use. A very crude diagram of these basic components
is presented in Figure 1.5. In future sections of this chapter, we will attempt to discuss each of these components in
more detail. It is usually appropriate at this time to introduce several quantities that can be used to describe the perfor-
mance of a specific reactor type. Each quantity is described in Table 1.4. Here, the formal definition of each quantity is
given as well as the units in which each of these quantities can be measured. Here, we have decided to use the metric
units system wherever possible. However, it is interesting to notice that a large number of these quantities are, in fact,
dimensionless numbers. Hence, they can be used to describe the relative performance of all types of reactors on an
apples-to-apples basis.
The design of a nuclear power plant is a monumental task that involves implementing literally hundreds or even
thousands of individual design trade-offs. While there is no single power plant design that optimizes all of these trade-
offs, most nuclear power plants are designed to achieve a reasonable compromise between cost, safety, size, longevity,
efficiency, and thermal-hydraulic performance. In recent years, there has also been a trend toward reducing the total
number of components (i.e., pipes, valves) that an individual reactor will use. This can be achieved through better overall
systems design. Finally, there has been a trend recently to move to more modular reactor designs in an attempt to save
money and to improve system reliability. Modular designs also have the advantage that a specific class of nuclear power
plants can be scaled to a much larger size by simply using a larger number of the same identical components (say one of
the SGs in the NSSS of a PWR). In this way, every nuclear power plant can use at least some of the same components
that smaller versions of the same power plant use (Figure 1.7). PWR modern nuclear power plants also use at least some
passive safety systems, and the purpose of these safety systems is to limit the amount of operator intervention that is
required in the event of a reactor accident. One of the other design goals of these passive safety systems is to limit the
number of “surprises” that can occur in the unlikely event of a serious reactor accident. These safety systems can react
much faster than human beings can, and by design, they are intended to limit the scope of a reactor accident as soon as
8 Nuclear Power in the World Today
TABLE 1.3
Design Parameters of a Modern Nuclear Power Plant (by Reactor Size or Type)
CANDU PWR
Feature PWR BWR (HWR) AGR LMFBR
Coolant (primary loop) Light water Light water Heavy water Carbon dioxide Liquid sodium
Pressure (primary 15.50 (2,250) 7.15 (1,040) 10 (1,450) 4.25 (615) ~0.1 (20)
loop) (MPa/PSI)
Temperature (primary 330°C 293°C 310°C 640°C 540°C
loop)a
Coolant (secondary Light water N/A Light water Light water Light water
loop)
Pressure (secondary 6.9 (1,000) N/A 4.7 (680) 17.0 (2,450) 17.5 (2,500)
loop) (MPa/PSI)
Temperature 285°C N/A 260°C 545°C 490°C
(secondary loop)a
Type of fuel Enriched uranium Enriched uranium Natural uranium Enriched uranium Uranium and
plutonium
Average enrichment 4%–5% 3%–4% 0.71% ~3% ~15%−20% Pu
Fuel assembly shape Square Square Concentric circles Concentric circles Hexagonal
of rods of rods
Cladding Zircaloy-4 Zircaloy-2 Zircaloy-4 Stainless steel Stainless steel
Moderator H2O H2O D2O Graphite None
Classification Thermal converter Thermal converter Thermal converter Thermal converter Fast breeder
Thermal efficiency 34% 32% 29% 42% 40%
SGs Yes No Yes Yes Yes
SG type U tube N/A U tube Helical coil Helical coil
Core power density 105 52 12 2.7 ~400
(kW/L)
Average burnup 50,000 50,000 7,500 20,000 110,000
(MWD/ton)
Source: Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
a Temperatures shown are the maximum average loop temperatures.
it begins. We will discuss several of these safety systems as we proceed further into this chapter. Plant-specific safety
systems are discussed in additional detail in Chapters 2–4 and Chapter 32.
1.6
Coolants Used in Nuclear Power Plants
Many factors go into the selection of a reactor coolant, and some of the most important of these factors include cost,
availability, neutronic performance, and thermal-hydraulic performance. As Table 1.5 shows, nuclear reactors use a
wide variety of coolants, but not all coolants are suitable for all reactor types. Reactor coolants can be further classified
into three broad categories:
1. Fluids
2. Liquid metals
3. Gases
By far the most common reactor coolant is ordinary water (H2O). Sometimes, ordinary water is also called light water.
PWRs and BWRs use light water to cool the core, and the only difference is that BWRs allow the coolant to boil,
whereas PWRs do not. CANDU reactors use heavy water in their primary loop and light water in their secondary
loop, but they are also moderated by another form of water called heavy water. Heavy water has an additional neutron
attached to the hydrogen nuclei in the water molecule. Sometimes, a hydrogen nucleus with this extra neutron is called
deuterium. CANDU reactors are sometimes called HWRs. Light water is used primarily because of its availability and
also because its thermodynamic properties are well known and well understood. The only problem with light water is
that it cannot be used in fast reactors because it contains a high concentration of hydrogen. Light water can also be overly
corrosive to the reactor piping system at times.
1.6 Coolants Used in Nuclear Power Plants 9
1
2
4
5
7 3
(a)
(b)
FIGURE 1.6 (a) A schematic of a modern nuclear power plant. (Picture provided by AREVA.) Numbers 1 through 9 in the
figure refer to the following: 1, the containment building; 2, the steel liner; 3, the reactor core; 4, the SGs; 5, the pressurizer; 6, the
RCPs; 7, the fuel storage area; 8, the building housing the steam turbines and the electrical generators; and 9, the auxiliary storage
building. (b) Another schematic view of a modern nuclear power plant. (Picture provided by Westinghouse.)
Water works very well in thermal reactors because the coolant also serves as the moderator. However, it does not
work well in fast reactors because it contains too many hydrogen atoms (two hydrogen atoms for each water molecule)
and these hydrogen atoms cause the neutrons in a fast reactor to slow down too quickly. Consequently, fast reactors must
use liquid metal coolants (such as liquid sodium, potassium, lead, and sometimes m ercury) instead of water to remove
the heat from the core. Liquid metals have excellent heat conduction and retention capabilities, and they can more easily
remove the large amounts of heat that the fast reactors produce. (Lead is also used as a coolant in military reactors and in
naval propulsion systems.) Finally, particle economy is also a major factor when deciding the best type of coolant to use.
CANDU reactors use natural uranium fuel, and they cannot use moderators such as light water that have high particle
absorption cross sections because there are simply not enough neutrons to go around. However, the neutron absorption
10 Nuclear Power in the World Today
TABLE 1.4
Useful Terms Used to Describe a Nuclear Power Plant
cross sections for heavy water are about 500 times lower than they are for light water, and so heavy water is an ideal
moderating material to use in these reactors in spite of its limited availability and cost.
1.7
Types of Nuclear Fuel
A nuclear fuel is a material that causes a nuclear reactor to create useful heat and power. Only a couple of elements
can be used as commercial nuclear fuels, and some variations of the same element (called isotopes) turn out to be bet-
ter nuclear fuels than others. Nuclear fuels are usually discussed in great detail in reactor physics books*, but we will
briefly touch on their properties here, because a basic understanding of nuclear fuels is required to fully appreciate the
subject of nuclear heat transfer. To create useful heat and power, a nuclear fuel has to be “burned.” Nuclear fuel is con-
sumed when a uranium or plutonium nucleus splits apart by absorbing a low-energy neutron. Only certain varieties of
uranium and plutonium, called fissile isotopes, have this important property. Examples of these isotopes include U-233,
U-235, Pu-239, and Pu-241. Other varieties of uranium and plutonium, including U-238, Pu-240, and Pu-242, can also
absorb a low-energy neutron, but they will not split apart or fission when they do so. Instead, they are converted into
Steamline
Walls made of Containment
concrete and cooling system
steel
3–5 ft thick
(1–1.5 m)
3 Steam
4
generator
Reactor Control
vessel rods
Turbine
generator
Heater
Condenser
Condensate
pumps 2
Coolant loop
Core
Feed 1
pumps
FIGURE 1.7 A typical PWR. Here, 1 denotes the core, 2 denotes the primary loop, 3 denotes the SGs, and 4 denotes the secondary
loop. (Courtesy of the U.S. NRC.)
TABLE 1.5
Characteristics of Some Common Reactor Coolants
other nuclear materials, which may then undergo radioactive decay. Incidentally, this is how most of the lead in the world
today, which is an excellent nuclear shielding material, was initially created. Isotopes of uranium and plutonium with
even numbers of protons and neutrons are called fertile materials.
The number immediately following the prefixes U and Pu refers to an element’s atomic number, that is, the total
number of protons and neutrons in the nucleus of that element. Hence in this case, U-235 is a fissile material, whereas
U-238 is a fertile one. The properties of these fertile and fissile materials are given in Table 1.6. Notice that an element
is defined by the total number of protons in the nucleus, while different isotopes of the element are defined by the total
number of neutrons in the nucleus. This distinction is explored in more detail in many reactor physics books.
12 Nuclear Power in the World Today
TABLE 1.6
Nuclear Fuels and Their Properties
1.8
P roperties of Nuclear Fuel
Uranium is nearly a perfect nuclear fuel because it is relatively abundant in nature, and in fact, it is more plentiful on
Earth than either silver or gold (see Table 1.7). However, this is not true in the universe as a whole, where the amount of
silver and gold far exceeds the amount of naturally occurring uranium. Plutonium does not exist naturally in significant
quantities, but it can be created by exposing Uranium-238 to a source of low-energy neutrons. Sometimes, this process is
known as breeding (Figure 1.8). The uranium in the Earth’s crust consists of mostly Uranium-238 (about 99.3% by weight)
and about 0.71% by weight U-235. This type of uranium is commonly referred to as natural uranium. Unfortunately, for
reasons that were explained in our companion book*, natural uranium is not a particularly good nuclear fuel (except in
CANDU reactors) because the coolant or moderator absorbs too many neutrons to keep a reactor critical. Because of this,
there are not enough neutrons left to sustain a nuclear chain reaction. To compensate for this neutron deficiency, the con-
centration of Uranium-235 in light water reactors is artificially increased to between 2.5% and 5% using a process known
as uranium enrichment. The enriched fuel is then known as enriched uranium. The process of enriching natural uranium
is not particularly complicated, but it does take some time to explain. For this reason, the interested reader should consult
a reactor physics book or a nuclear fuel cycle book such as Cochran and Tsoulfanidis (see Glasstone and Sesonske, 1981
TABLE 1.7
Natural Abundance of Various Nuclear Fuels in the Earth’s Crust
(a) (b)
FIGURE 1.8 Two common nuclear fuels, uranium (a) and plutonium (b), in their metallic form. (Courtesy of the U.S. NRC.)
1.9 Reactor Pressure Vessels and Their Properties 13
Centrifugal Laser
Gaseous diffusion
separation enrichment
at the end of the chapter) to learn more about how the process works. The original process for enriching natural uranium
was called the gaseous diffusion process, but unfortunately, it turned out to be very energy intensive and consumed large
amounts of electric power. Recently, the gaseous diffusion process has been replaced by the centrifugal separation pro-
cess and the laser enrichment process. The centrifugal enrichment process is the dominant uranium enrichment process
in the world today. It is so popular because it uses only about 2% as much electricity as the gaseous diffusion process does
and also has a much smaller physical footprint. The laser enrichment process, which uses industrial lasers, uses even less
electricity, but it has only been recently deployed on a commercial scale (Figure 1.9). The enriched uranium is then mixed
with two oxygen atoms to form a ceramic compound known as uranium dioxide or UO2. Uranium dioxide is the material
that is most commonly used in nuclear fuel rods today. The properties of uranium dioxide are discussed in Chapter 11.
Uranium dioxide is an excellent compound to use as a nuclear fuel.
1.9
Reactor Pressure Vessels and Their Properties
For a variety of reasons, the core of a nuclear reactor must be surrounded by a pressure vessel. A pressure vessel is a
large stainless steel or concrete container that holds the fuel assemblies, the control rods, and the coolant in place. The
pressure vessel serves as a radiation shield to keep the radiation that is produced in the core from leaking out into the
rest of the power plant. It also serves to pressurize the reactor coolant in PWRs and BWRs, and this pressurization
process tends to increase the thermal efficiency of the plant. Reactor pressure vessels can come in many different sizes
and shapes, but the ones used in PWRs and BWRs are primarily cylindrical in shape. Pressure vessels are sized to
the thermal output of the core, and so some pressure vessels can be as much as 20 m high and 20 m wide. However, on
average, they are much smaller than the maximum sizes quoted here. The sizes of the pressure vessels that are used in
commercial power reactors are shown in Figure 1.10 and Table 1.8. Generally speaking, PWRs have much smaller pres-
sure vessels than BWRs do (~13 m high and ~5 m wide vs. ~21 m high and ~6 m wide for a reactor having the same power
output). In other words, they have about half of the overall volume in spite of the fact that the average power density is
twice as high (~105 vs. ~52 kW/L). They are also about 50% thicker to account for the increased system pressure (~2,250
vs. ~1,050 PSI). The materials used in reactor pressure vessels vary slightly depending upon the design of the core as a
14 Nuclear Power in the World Today
FIGURE 1.10 The thickness of different reactor pressure vessels. (From Todreas, N. and Kazimi, M.S., Nuclear Systems, Vol. 1,
CRC Press, Boca Raton, FL (2008).)
TABLE 1.8
Reactor Pressure Vessel Sizes
b Pool-type LMFBR.
whole. PWRs and BWRs use low-carbon steel with a stainless steel liner to increase the lifetime of the pressure vessel
and to keep it from becoming brittle as it is exposed to high levels of radiation over prolonged periods of time (typically
40 or 50 years). AGRs, on the other hand, use pressure vessels made up of prestressed concrete with a steel liner because
they are so large.
Finally, fast reactors cooled with liquid sodium use relatively thin stainless steel pressure vessels because the operat-
ing pressure is so low (about 15 PSI). The thickness of these pressure vessels is dictated primarily by the type of coolant
and the power density of the core. No two pressure vessels are exactly the same, although their construction has become
more standardized in recent years as there has been a trend toward more modular designs. The thickness of several pres-
sure vessels is shown in Table 1.9. In general, for the same material, the thickness increases as the pressure difference
between the inside and the outside of the pressure vessel is increased. Interestingly enough, the relationship between the
system pressure and the thickness is not a linear one. The American Society of Mechanical Engineers (the ASME) and
the American Nuclear Society (the ANS) first established standards for the construction of reactor pressure vessels in the
1970s. Since that time, these standards have undergone several major revisions. Today, these standards can be found in
Section III of the ASME Boiler and Pressure Vessel Code. The reader can order copies of these standards directly from
the ANS or the ASME. Today, all reactor pressure vessels are manufactured according to these standards.
1.11 Characteristics of Reactor Fuel Assemblies 15
TABLE 1.9
Thickness and Design Pressure of Some Common Reactor Pressure Vessels
FIGURE 1.11 PWR fuel assemblies and the fuel rods from which they are made.
1.10
Characteristics of Reactor Cores
Reactor cores are constructed from arrays of fuel rods grouped together into larger structures known as nuclear fuel
assemblies. A nuclear fuel assembly can have between 30 and 300 fuel rods, and the number can vary greatly depend-
ing upon how a reactor is designed. The exact arrangement of the fuel rods depends upon whether the fuel assembly is
square, rectangular, or hexagonal in shape. We discussed the arrangement of these rods in Chapter 11. A picture of a fuel
assembly taken from a typical PWR is shown in Figure 1.11. The internal structure of a BWR pressure vessel is shown
in Figure 1.12. For reasons that we will subsequently discuss, BWR and LMFBR fuel assemblies have slightly different
sizes and shapes. When it comes to light water reactors, BWRs have the largest cores, whereas PWRs have the smallest
ones. Gas reactor cores are much larger because their average power density is so low (<10 kW/L). The parameters of
these cores are compared in Table 1.10. Reactor cores are constructed from large arrays of fuel rods and fuel assemblies.
Fuel rods and fuel assemblies are discussed in Chapter 11.
1.11
Characteristics of Reactor Fuel Assemblies
Reactor fuel assemblies come in many different sizes and shapes, and there is no such thing as a “standard” fuel assem-
bly. However, they do have many design features in common. Large PWR cores can have anywhere between 175 and
225 fuel assemblies, whereas a large BWR core can have anywhere between 700 and 800 fuel assemblies. Light water
reactor fuel assemblies today are designed to have an average burnup of about 50,000 megawatt days (MWD)/metric
ton. This means that the fuel rods normally stay in the core for 3 or 4 years. During this time, they continue to generate
heat and power, and the heat generated is proportional to the number of fuel assemblies and the average power density
of the core. We will describe how the power densities are defined in the next section.
16 Nuclear Power in the World Today
FIGURE 1.12 A view of the internal structure of a BWR pressure vessel. (Courtesy of TEPCO—see
http://cryptome.org/eyeball/ daiichi-npp15/daiichi-photos15.htm.)
TABLE 1.10
Design Parameters for Some Common Reactor Cores
Large LMFBRs, on the other hand, can have about 600 fuel assemblies (which are normally hexagonal in shape),
but in a fast reactor, some of these fuel assemblies are used to create the core and other fuel assemblies, which sur-
round them, are used to create the blanket, where new nuclear fuel is bred. The core contains a high concentration of
plutonium, and the blanket contains either U-238, depleted uranium, or thorium fuel rods. The power density in the
core can be extremely high. As a matter of fact, it can be five to ten times higher than the power density of an aver-
age light water reactor core. The relationship between the core and the blanket is shown in Figure 1.13. Sometimes,
LMFBRs are designed to employ an axial blanket as well as a radial one. Hence, the upper and lower 10%–20% of
each fuel assembly contains fertile material (U-238 or Thorium-232), which is used to create new reactor fuel. Fast
reactor fuel assemblies are designed to have an average burnup of over 100,000 MWD/metric ton. This is at least
twice the burnup of an average light water reactor fuel assembly. It is not uncommon for the uranium atoms in an
entire PWR core to weigh between 100 and 150 metric tons, and this is also true of the fuel assemblies in a BWR
core as well. PWR, BWR, and LMFBR fuel assemblies are about 4 m long, and the active fuel height is about 3.5 m.
CANDU and gas reactor fuel assemblies are not as long as PWR or BWR fuel assemblies, and large numbers of them
are normally stacked vertically or horizontally to create the core. A CANDU reactor core is about 6 m long, and gas
reactor cores tend to be 8–10 m high. LMFBR cores are about the same size as PWR cores. The dimensions of these
cores are compared in Table 1.10.
The pitch of the fuel rods (the center-to-center spacing between the fuel rods) can also vary from one type of reactor
to the next. The pitch of the fuel rods in a PWR fuel assembly is about 12.6 mm, whereas the pitch of the fuel rods in a
BWR fuel assembly is about 14.37 mm. The pitch of the fuel rods in a fast reactor fuel assembly (which is hexagonal in
shape) is about 9.8 mm, which is considerably smaller than either of these numbers. CANDU reactor fuel assemblies,
1.12 Other Important Reactor Properties (Power Density and Thermal Efficiency) 17
Axial blanket
Axial blanket
FIGURE 1.13 The core and blanket regions of a plutonium-fueled fast breeder reactor.
(http://www.ati.ac.at/fileadmin/files/research_areas/ssnm/nmkt/11_LMFBR.pdf.)
FIGURE 1.14 A reactor core and pressure vessel as seen previously. (From Google Images.)
which are sometimes referred to as fuel bundles, have a typical pitch of about 14.6 mm. Finally, gas reactors can have a
fuel rod pitch of 23–37 mm. Not surprisingly, these reactor fuel assemblies also have the lowest average power density
(2.5–8.5 kW/L). A picture of a complete reactor core, which is made up of a number of individual fuel assemblies, is
shown in Figure 1.14. We will discuss the structure of several reactor cores in more detail in Chapter 11. A reactor lattice
having a square pitch is shown in Figure 1.15.
1.12
O ther Important Reactor Properties (Power Density and Thermal Efficiency)
Two other extremely important properties of a nuclear power plant are the power density of the core and the thermal
efficiency of the plant. Both of these parameters are a function of the design of the core and the NSSS. The NSSS will
be discussed again in Chapter 8. In this section, we would like to discuss these two parameters and show how they are
related. The power output of a core P is then a function of its size and its power density.
18 Nuclear Power in the World Today
FIGURE 1.15 The pitch of a reactor lattice is the center-to-center spacing between adjacent fuel rods. The pitch of the fuel rods in
a PWR lattice is about 12.60 mm, and in a BWR, it is about 14.37 mm. (From http://montecarlo.vtt.fi/development.htm.)
1.13
T he Power Density
The power density of a reactor core is defined as the power P produced in the core divided by the volume of the core VCORE
in which the power is produced. The volume in this case refers to the active core volume. The heat Q that is generated is then
equated to the thermal power output PTHERMAL. The power densities of several reactor cores are compared in Figure 1.16.
We will have more to say about this later in the section. Hence, a common definition of the core power density is
Power Density
Q P
PD = = THERMAL . (1.1)
VCORE VCORE
It is easy to see that reducing the volume of the core and keeping the power output constant increases the power density,
while increasing the volume of the core and keeping the power output constant decreases it. In general, reactor design-
ers try to keep the core power density as high as possible because this reduces the size of the core and it also helps to
reduce the cost of constructing the plant. Normally, the power density is limited by the rate that heat can be removed
from the fuel rods. Cores that have high power densities tend to have thin fuel rods (because it is easier to remove heat
350
300 LMFBR
Power density (kW/L)
250
200
150
PWR
100
BWR
50
CANDU
0
FIGURE 1.16 A comparison of the power densities of modern PWR, BWR, and LMFBR cores. The differences in the power den-
sities are attributable primarily to differences in the enrichment of the fuel. Note that LMFBRs use a hexagonal lattice, while PWRs
and BWRs use a square one.
1.14 Thermal Efficiency 19
from a thin fuel rod than a thicker one). Consequently, a PWR will have thinner fuel rods than a BWR, and a BWR will
have thinner fuel rods than a GCR.
If the type of coolant and the system pressure are fixed, the diameter of a fuel rod is normally inversely proportional
to the power density of the core. If the power density gets too high (say beyond 120 kW/L), a conventional coolant (such
as liquid water) may have to be replaced by a liquid metal to remove the heat. Hence, the power density, the system pres-
sure, and the type of coolant are intimately related to one another. This means that all three variables must be taken into
account when designing a core to meet a particular need. The power densities of some common reactor cores are shown
in Figure 1.16. Notice that the power densities can range from about 3 kW/L in a GCR to about 300 kW/L in an LMFBR.
In some Russian fast breeder reactors, power densities as high as 500 kW/L have been reported. By any standard, this
is a very high power density.
1.14
T hermal Efficiency
Once heat is generated in the core, it is the responsibility of the NSSS to remove this heat and convert it into electric
power. In most reactors, this is done by converting the heat into steam, which is then fed to what is known as a steam
turbine or a power turbine. An electric generator is then attached to the shaft of the steam turbine, and turning the arms
of this generator causes electric power to be produced. On a very high level, this process is illustrated in Figure 1.17. The
thermal efficiency of a power plant η is defined as the ratio of the electric power produced by the plant to the amount of
thermal energy produced in the core. If the amount of heat produced in the core is Q and the output of electrical power
from the plant is P, then the definition of the thermal efficiency is
η = TE = Thermal Efficiency
P(MWE)
η= . (1.2)
Q(MWT)
This is the definition that is used in most reactor thermal-hydraulic books. Notice that the electrical power output P is
measured in megawatts electric (MWE) and the thermal power output is measured in megawatts thermal (MWT). The
reader should not confuse these two terms as we proceed further into the text. The thermal efficiency is a dimensionless
number. The actual thermal efficiency is defined by the type of plant thermal cycle that is used. Typical thermal efficien-
cies range from about 32% to 48%. Most nuclear power plants use a Rankine thermal cycle to convert the heat generated
in the core into useful power. However, GCRs can theoretically use a Brayton thermal cycle to achieve the same effect.
We will discuss both of these cycles in more detail in Chapters 8 and 9. However, the larger temperature difference ΔT
there is between the temperature of the coolant in the core and the heat sink to which this thermal energy is released,
the greater the thermal efficiency of the power plant will be. This is the same regardless of whether we are dealing with
a nuclear power plant or a coal-fired power plant. Hence, many of the components of the NSSS that are used in fossil-
fired power plants are also used in nuclear ones.
Steam
Generator
turbine
Power line Electricity produced Rotating shaft Energy extracted E
The concept of the thermal efficiency is discussed in great detail in many classical heat transfer books. However, in
the field of nuclear science and engineering, most textbooks do not do justice to this subject even though it is such an
important aspect of the design of a nuclear power plant. In this textbook, we will devote more time to this important
subject when we discuss the NSSS in Chapter 8. This discussion will then be followed by a comprehensive discussion
of reactor thermal cycles in Chapter 9. Hopefully, these discussions will demonstrate why optimizing the thermal effi-
ciency is so important to the economic attractiveness of a nuclear power plant. Notice that gas reactors and LMFBRs
tend to have the highest thermal efficiencies (in the range of 40%–48%). Conventional PWRs and BWRs tend to have a
thermal efficiency of about 33%, although this number can vary somewhat from one power plant to the next. Sometimes,
additional “tricks” such as reheat and regeneration can be used to extract more efficiency from a reactor thermal cycle.
We will explore how these processes work in more detail in Chapter 9.
The number of loops in the NSSS also has some effect on the thermal efficiency. Generally speaking, PWRs have
two loops (a primary loop and a secondary loop), and the coolant in the primary loop is a liquid, whereas the coolant in
the secondary loop is a vapor. BWRs have only one loop (a primary loop) because the coolant is intentionally allowed
to boil. Finally, LMFBRs can have as many as three separate loops to separate the heat-generating loop in the core from
the steam-generating loop, which is connected to the power turbines. The thermal efficiencies of several different types
of reactors are shown in Table 1.11. The NSSS for a commercial PWR is illustrated in Figure 1.18. Note that within a
TABLE 1.11
Thermal Efficiencies of Common Commercial Power Reactors in the World Today
Pressurizer Steam
generator
Reactor coolant
pump (RCP)
Piping
system
Pressure
vessel
FIGURE 1.18 The RCPs, the SGs, the pressurizer, the reactor vessel, and the primary coolant loops in a large PWR. (Courtesy of
Westinghouse—see http://me1065.wikidot.com/nuclear-pressurized-water-reactor-pwr.)
1.15 Control Rods and Their Function 21
given loop (primary, secondary, or tertiary), additional subloops (called coolant legs) are provided to carry the energy
transferred from the fuel rods to the coolant. For example, in a large PWR, the coolant flow through the primary loop
is handled by three or four additional subloops. Each of these subloops may then have a separate reactor coolant pump
(RCP) and steam generator (SG) attached to them. Hence, a 1,000 MWE PWR can have as many as four separate SGs
and RCPs. BWRs also have multiple subloops within the primary coolant loop. The subloop that receives the hot coolant
from the core is called the hot leg, and the subloop that feeds the cold coolant back into the core is called the cold leg.
Because all of the components of a nuclear power plant must be redundant, commercial power reactors must always have
at least two hot and cold legs. The most severe reactor accident (see Chapter 30) is normally assumed to occur when there
is a massive failure of one of the coolant pipes in the cold leg feeding the coolant back into the core. When modeling this
type of accident, a double-ended guillotine pipe break of the main coolant pipe is normally considered to be the most
limiting event. This pipe break then leads to what is called a Loss of Coolant Accident or LOCA.
1.15
Control Rods and Their Function
In nuclear reactors, control rods (see Figures 1.19 and 1.20) have four primary functions. First, they can be used to either
raise or lower the power level by changing the value of the reactivity ρ, which is defined as
(K effective − 1)
ρ= , (1.3)
K effective
where Keffective is the effective multiplication factor for the core. Changing the reactivity also changes the reactor period,
and this needs to be taken into account when a control rod is moved. Second, when put within the context of an overall
plan, control rods can help to shape the flux, eliminate hot spots, flux peaks, and promote a better burnup of the fuel.
Third, they can help to keep a reactor critical by compensating for changes in the reactivity due to the introduction of
burnable poisons, soluble poisons, and the burnup and depletion of the fuel. Finally, they can be used as an emergency
shutdown system to quickly and reliably shut down the core in the event of an unforeseen problem or a safety issue*.
Withdrawing a control rod from the core will tend to increase the power level since fewer thermal neutrons will be
absorbed. Conversely, moving a control rod further into the core will have exactly the opposite effect because more
neutrons will be absorbed. Control rods are usually designed so that each incremental movement in the position of a
control rod will correspond to a known reactivity change (+ or −). For example, if a control rod has a total of 20 possible
vertical positions (see Figure 1.19), each position will change the reactivity by 1/20th of this amount. The actual amount
that the reactivity is changed depends on how the core is designed. However, in most reactors, an incremental change in
the position of a control rod bank will change the reactivity by 1 cent to one-half of 1 cent.
(a) (b)
FIGURE 1.19 (a) A group of control rods (shown sideways) and a control rod drive assembly with graduations on the rod. Each
graduation is ~0.1 cent of reactivity. (b) A fuel assembly with the control rods partially inserted into the core. (Pictures provided by
the author and Westinghouse Nuclear Energy Systems.)
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
22 Nuclear Power in the World Today
(a) (b)
FIGURE 1.20 The fuel assemblies and control rods in a BWR (a) and in a PWR (b) directly compared.
(Pictures provided by the author.)
A fully withdrawn rod will have no net contribution to the reactivity. However, each time a rod is inserted into the core
by 1/20th of its length, the reactivity will be reduced by ½ or 1 cent. Thus, a fully inserted control rod will be able to
reduce the total reactivity by a total of 10–20 cents, while a rod that is inserted halfway will be able to change the reac-
tivity by only half of this amount. According to the U.S. Nuclear Regulatory Commission (or NRC), the formal defini-
tion of a control rod is as follows.
Hence, control rods are a very important component of any reactor core. Without them, it would be virtually impossible
to control the power level when quick changes to the neutron population are required in response to sudden temperature
or reactivity changes in the core. Control rods can also come in many different sizes and shapes, and they are generally
optimized to meet the needs of a particular design. BWRs have cruciform-shaped control rods, while most other types
of reactors (such as PWRs and LMFBRs) have cylindrical-shaped ones. Some research reactors also use plate-type or
rectangular control rods, but these control rods are generally not as common as those used in other designs, and they are
usually not as efficient at removing the heat.
1.16
Comparing PWR Control Rods and BWR Control Rods
For a number of reasons, BWRs and PWRs have taken a slightly different evolutionary path when it has come to the
design of their control rods. During normal operation, the moderator in a BWR is intended to boil throughout much of
the core, while the water in a PWR is only designed to undergo nucleate boiling near the top of the core (see Chapters 2
and 3). Because of this, the coolant exiting a BWR must employ a steam dryer to remove the excess moisture from the
steam at the top of the pressure vessel head, and because of space limitations, the presence of the steam dryer makes it
very difficult to mechanically insert the control rods into the core from above.
Consequently, BWRs have tended to adopt a design where the control rods are inserted into the core from below,
while the control rods in a PWR are inserted into the core from above. In addition, because the control rods are
inserted into the core from opposite directions, the shapes of the control rods are very different in these designs. For
example, PWRs use finger-shaped control rods that are similar in shape to the fuel rods themselves, while BWRs use
1.16 Comparing PWR Control Rods and BWR Control Rods 23
cruciform-shaped control rods that are much larger and more rigid than their PWR counterparts. Both designs are
shown in Figure 1.20. Here the control rods are shown in red while, the fuel rods are shown in dark gray or light blue.
Because the control rods in a BWR are so much larger than the control rods in a PWR, they have a much greater average
control rod worth as well. In a PWR, each cluster of control rods is inserted into the core from above using a mechani-
cal drive system. The drive system is attached to the pressure vessel head with a large number of electromagnets. In
the event a reactor needs to be shut down quickly, the power to the electromagnets is cut by pushing a scram button on
the control panel, and the rods fall into the core under the force of gravity alone. This is an example of what is called
an emergency reactor shutdown or “scram.” The design of the control rod linkages and the electromagnets that allow
the control rods to fall into the core go back to the early days of the U.S. Nuclear Submarine Program, and the designs
that are used today are not much different from the designs that were used 50 years ago. The control rod scram system
in a PWR is an example of a passive safety system because once the scram button is pushed (see Figure 1.21), the rods
fall into the core with no further human intervention (the scram button cuts off the electricity to the upper part of the
pressure vessel, and this “pulls the plug” on the electromagnets; see Figure 1.22). On the other hand, BWRs use a high-
pressure hydraulic system to insert the control rods into the core from below. As we mentioned previously, these rods
are generally much larger and heavier than their PWR counterparts, and in most cases, their reactivity is larger as well.
The hydraulic pressure exerted on the rods from below provides the necessary lifting force required to insert the rods
into the core. Internally, a cruciform-shaped rod consists of either of the following:
☉☉ A solid neutron-absorbing material in metallic form
☉☉ A hollow chamber into which many hollow tubes filled with B4C powder can be placed
The design of these rods has evolved over the years, and the materials that are used in most control rods today
(e.g., hafnium and iridium) are considerably different than those that were used several decades ago. In a reactor phys-
ics calculation, it is usually possible to model the rods in a BWR as sheets of black neutron-absorbing material and to
calculate their worth by treating them as a planar sheet of absorbing material located in an otherwise homogeneous
region of the core. It can be shown that a typical control rod in this case will have a total rod worth of about 0.2 β,
or W = 0.0013/0.0065 ≈ 20 cents per rod. A 1,000 MW (3,000 MWT) BWR will have about 150 of these rods. This
puts the total control rod worth for a large BWR core at about $25.00 on a system-wide basis. By comparison, a PWR
will have about 40% as many control rods (~60 clusters vs. 150), and their aggregate rod worth will be about $11.00.
(a)
(b) (c)
FIGURE 1.21 Three examples of the scram button on a reactor control panel. (a) A picture of the scram button from the
control room of the nuclear cargo ship, the USS Savannah, which was the world’s first nuclear-powered commercial ship;
(b) a picture of the scram button at the experimental breeder reactor I; and (c) a picture of an operator attempting to press the
scram button. (Pictures provided by Wikipedia.)
24 Nuclear Power in the World Today
PWR BWR
Core
Core
(a) (b)
FIGURE 1.22 Control rods are inserted into the core in opposite directions in a PWR (a) and a BWR (b). This has been part of
their standard design for about 50 years. In PWRs, the control rods are also held in place by electromagnets, which can be disabled
by simply cutting the power.
The average rod worth is still about 0.2 β or W = 0.0013/0.0065 ≈ 20 cents per cluster. Each cluster has about 20 finger-
shaped control rods, so an individual finger-shaped control rod in a PWR will have an average worth of 20 cents/20 or
about 1 cent per rod. The total rod worth is only about 40% as much in a PWR as it is in a BWR because of the use of
the chemical shim which is sometimes referred to as soluble boron or Solbor. Furthermore, only about one-fourth of
the fuel assemblies in a modern PWR will actually be equipped with a control rod. There are still more subtle differ-
ences between these competing approaches due to differences in the choice of control rod materials as well.
1.17
Use of the Scram Button and the Word SCRAM
Just after the U.S. Nuclear Navy was founded, the word scram was invented to describe a condition (see Masterson,
2017a) where a reactor onboard a nuclear aircraft carrier or submarine had to be shut down immediately - normally in
response to a system malfunction. At the time, the word scram happened to be an abbreviation for a “safety control rod
axe man.” However, today it is intended to indicate the SCRAM or emergency shutdown of any type of commercial or
military reactor, and it is usually initiated by the reactor operator by pushing the scram button, which is located on the
reactor control panel. In commercial power reactors, this emergency shutdown process is also called a “reactor trip,”
but the exact usage of this term varies from one type of reactor to the next. In most PWRs and BWRs, a reactor SCRAM
is now part of the emergency shutdown procedure that all reactors are required to undergo during a routine safety test.
However, when they are ready to be refueled, a more gradual rod insertion process is required. When a reactor needs to
be refueled, there is generally not enough excess reactivity for it to be able to be operated any longer, and the core may
also be affected by the buildup of fission product poisons such as Xenon-135 and Samarium-149 in the fuel. These fis-
sion product poisons have very large thermal absorption cross sections (their values are provided in Masterson, 2017b),
and they can absorb enough thermal neutrons to cause the effective multiplication factor of the core to fall below 1.0.
When this happens, the core can become subcritical even when all of the control rods are withdrawn. Some of the fuel
assemblies then have to be replaced before the reactor can be put into service again.
1.20 SGs AND THEIR USES 25
1.18
Maintaining the Criticality of the Core over Time
During operation of a commercial power reactor, the amount of fuel contained in the core decreases monotonically with
time if the power level P remains the same. If the reactor is designed to operate for a long period of time without being
refueled, then additional fuel in excess of what is needed for exact criticality must be added to the core before it is started
up. The positive reactivity generated by this excess fuel must then be balanced with negative reactivity introduced by other
sources. These sources can include moveable control rods and burnable poisons, and in the case of PWRs, this neutron-
absorbing capacity is augmented further by introducing a strong neutron-absorbing material, or poison, into the coolant in
the form of boric acid, or chemical shim. Sometimes this boric acid is also called soluble boron or Solbor. A cold, clean
core always has more fuel than it needs to achieve a critical mass. In a commercial power reactor, it is not uncommon for
the reactor to contain more than 30% more fuel than it needs to become critical. Without these other neutron poisons
to compensate for it, a typical (unpoisoned) Keffective for a PWR is about 1.40 at start-up, and the corresponding Keffective
for a BWR at start-up is about 1.33. This corresponds to an excess reactivity at start-up of about $45.00 for a PWR and
about $38.00 for a BWR. By design, the positive reactivity produced by this excess fuel is always balanced by the negative
reactivity from the reactor’s control rods, or in the case of a PWR, from both the control rods and the boric acid dissolved
in the coolant. The control rods are inserted into the core just far enough to keep the reactor in a subcritical state until a
decision is made to start it up. In PWRs, this excess reactivity is also held in check by the boron dissolved in the coolant.
Once a reactor is started up, the control rods are gradually withdrawn to compensate for the reduction in the reactivity due
to the buildup of xenon, and the increase in the operating temperature of the fuel. This process of withdrawing the control
rods continues for several hours until it has reached its desired operating state. In PWRs, the boron concentration in the
coolant is also reduced to achieve a similar effect. Once the reactor has reached a steady state, the amount of fuel in the
core starts to monotonically decrease as a function of time due to its increased burnup as a result of the increased neutron
flux. So although a reactor may initially start its life with much more fuel than it really needs to stay critical, this excess
fuel is what allows it to operate for long periods of time after initial criticality has been achieved. We will have more to
say about this process in our companion book*, when the effects of burnup and depletion are discussed in more detail.
1.19
Electrical Generating Systems in a Nuclear Power Plant
Basically, a nuclear power plant can be thought of as a large nuclear furnace that generates heat by splitting apart atoms
of uranium or plutonium instead of burning fossil fuels like coal and natural gas. The heat generated in this process is
eventually converted to steam, and the steam drives a power turbine to produce the electricity that is generated by the plant.
In this section, we would like to show exactly how this process works and how the thermal energy produced by the plant
(and measured in MWT) can be converted to electricity in MWE. Some reactors use coolants other than water (such as
carbon dioxide and liquid metals) to cool the core, but with the exception of GCRs, which can sometimes use the Brayton
thermal cycle, all other common reactor types use a Rankine thermal cycle to turn the electrical generators and to produce
useful power from the heat that is generated in the core. The Rankine thermal cycle assumes that the working fluid (which
in most cases is water) can undergo a change of phase, whereas the Brayton cycle, which GCRs use, assumes that it cannot.
1.20
SGs and Their Uses
With the exception of BWRs, all modern nuclear reactors use a device called a steam generator (or SG) to convert
the hot coolant produced by the core into steam to turn the power turbines. Steam generators are large and relatively
complex devices that take the hot working fluid produced in the core and cooler water from another reactor-cooling loop
(called the secondary loop) and convert it into steam. Most steam generators have no moving parts. However, designing
one to function properly can be an extremely difficult task because it involves a very detailed knowledge of heat transfer,
fluid flow, water chemistry, and nuclear materials. A cutaway view of a typical SG is shown in Figure 2.2 of Chapter 2.
Today steam generaotrs tend to come in two basic designs: the first design is called a once-through SG, and the other is
called a UTUBE SG. OTSGs can be 60–80 ft high (20–25 m), and UTUBE SGs are about one-third of this height. In a PWR,
hot water typically enters a steam generator with a pressure between 2,200 and 2,250 PSI, and at a temperature of about
325°C (620°F). In this state, it is said to be a saturated liquid. In other words, it contains as much thermal energy as it can
without boiling. The cooling water from the secondary side enters the SG at a pressure of about 900 PSI and at a temperature
of about 220°C (430°F). The steam on the secondary side emerges at a temperature of about 275°C (540°F). In this state, it
is said to be completely saturated, and it has an exit quality of 1.0. There are no water droplets remaining in the steam, and
if the steam is heated even further (to increase the thermodynamic efficiency), it is then said to be superheated.†
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
† Thus, superheated steam can have more energy than saturated steam does.
26 Nuclear Power in the World Today
TABLE 1.12
Temperatures and the Pressures of the Coolant Entering and Leaving the Primary and Secondary Sides of the SGs in a
Typical PWR
Side of the NSSS Pressure (MPa) (Absolute) Inlet Temperature (°C) Outlet Temperature (°C)
Primary side (tube side) 15.5 325 (liquid water) 275 (liquid water)
Secondary side (shell side) 6.2 220 (liquid water) 275 (saturated steam)
Note: The numbers can vary somewhat depending on the actual design.
Steam turbines prefer superheated steam, because it contains no water droplets and it is highly energetic. Hence, these turbines
can run for long periods of time without the impellers being damaged by these droplets as the steam pressure falls. The tubes
inside SGs are typically made of noncorrosive metals or exotic alloys of stainless steel. These high-performance alloys and
superalloys have been used in SGs for many years, and they are specifically designed not to rust or corrode. Some commonly
used alloys of this type include type 316 stainless steel, Alloy 400, Alloy 600MA (which is mill-annealed), Alloy 600TT
(which is thermally treated), Alloy 690TT, and Alloy 800Mod. References to other alloys can also be found in the literature.
The steam generators in a “typical” PWR in the United States have the operating parameters shown in Table 1.12. Steam gen-
erators made by other companies (such as AREVA) have slightly different sets of operating parameters than those shown here.
1.21
Steam Turbines
Once the steam is generated by a steam generator, it is then passed through the secondary loop into a device known as a steam
turbine. In a power plant, steam turbines are also called power turbines. In today’s power reactors, steam turbines are nor-
mally located outside the containment building. When the steam enters the power turbine, it causes the blades of the turbine
to spin, and this kinetic energy is eventually transferred to a rotating shaft. (The actual rate of rotation depends upon how fast
the shaft of an individual turbine is designed to turn.) We will have more to say about this in Section 1.23. Figure 1.23 shows
the inside of a modern steam turbine. The modern turbine was invented by an Englishman by the name of Charles Parsons.
1.22
SG and Steam Turbine Pairing
In PWRs, the steam generators and the power turbines tend to come in matched pairs. There are a number of design
reasons for this that are generally related to the specifics of the plant thermal cycle. However, almost all PWRs have this
same basic design. The major difference between two PWRs with different power levels is that they may have different
numbers of SGs or power turbines attached to the core. Generally speaking, another SG and/or turbine set is added to
the plant for each additional 500, 750 or 1,000 MW of thermal power that is generated. This way both the SGs and the
steam turbines can be designed as “matched pairs” to optimize the amount of electrical power that is produced. This
also helps reactor vendors to design plants that are more modular, and hence more reliable. If a higher power output is
required, one simply adds a larger core and another SG/steam turbine pair set to the secondary loop. This design concept
forms the basis for what we call advanced modular reactors today. Hence, a modern nuclear power plant that generates
1,000 MW of electrical power (about 3,000 MWT) will typically have three or four large SGs inside of the contain-
ment building. (The steam turbines are located outside of the containment building.) However, each of these SG/steam
(a) (b)
FIGURE 1.23 A large steam turbine in a nuclear power plant (a) and another steam turbine being repaired (b).
1.23 Electrical Generators in Nuclear Power Plants 27
turbine pairs is essentially an identical clone of the other, and so if we can understand how one of them works, then we
can understand how all of them work. We would now like to discuss the electrical generators that are attached to the
shafts of these steam turbines.
1.23
Electrical Generators in Nuclear Power Plants
An electrical generator is a device that takes the mechanical energy from a rotating shaft and converts it into electric
power. In nuclear power plants, this conversion is achieved by attaching the electrical generator to a steam turbine.
The modern-day electrical generator was first proposed by an Englishman by the name of Michael Faraday in 1861.
An electrical generator can be found at the heart of every nuclear power plant in the world today. An electrical gen-
erator is simply a large rotating device that converts mechanical power from a rotating shaft into electrical power. It
does so by creating some relative motion between a stationary magnetic field (called a stator) and a rotating conductor
(called a rotor). In an electrical generator, the shaft usually rotates and the magnetic coils that are used to generate
the electricity are usually fixed. The fixed coils are called the stator, and the moving shaft is called the rotor. The
faster the rotor turns, the more power the generator produces. It should come as no surprise that most of the AC power
in the modern industrial world is produced in this way. Nuclear power plants typically use e lectrical generators hav-
ing three- or four-pole rotors. These rotors are used because the temperatures and pressures in the NSSS are more
suitable for three- and four-pole generators than those having just one or two. The electrical output from the poles is
combined together to produce an alternating current (AC). Figure 1.24 shows the process that is involved to generate
the electrical current that reaches your home. The number of poles is determined by the rate that the shaft turns the
generator. The shaft of an electrical generator in a nuclear power plant normally rotates at about 1,800 revolutions
per minute (RPM). So, the output of three poles is combined together to produce the alternating current that is gener-
ated. By contrast, a hydroelectric generator rotates at a speed of about 100 RPM. In these generators, 72 poles are used
to obtain the same effect. Of course, this number varies slightly depending upon how the power plant is designed.
More information about these design trade-offs can be found at the following URL:
http://www.asope.org/pdfs/AC_Electrical_Generators_ ASOPE.pdf
AC power having three phases is called three-phase power. It is generally preferable to other forms of AC power because
it uses a less electrical conducting material to transmit the same electric power than equivalent single-phase or two-
phase systems with an identical voltage. For this reason and others, three phase alternating current was first patented
by Nikola Tesla in 1887 and 1888, and it subsequently came into widespread use in the early stages of the nineteenth
century. In a three-pole generator, the poles are arranged in such a way that the individual currents from each pole vary
sinusoidally at the same frequency but with a different phase. The peaks and troughs of their respective waveforms are
intentionally offset by one-third of a cycle to provide three complementary ACs with a phase separation of 120°. The
electrical power that is produced in this way is called three-phase power.
Field
excitation
FIGURE 1.24 A single-phase generator (a) can produce single-phase AC power, while a three-phase g enerator (b) can produce
three-phase AC power. The current is then summed in the three-phase generator to generate a three-phase AC (see later).
28 Nuclear Power in the World Today
Three-phase electric power is the most common form of electrical power in the world today, and most of the electrical
power in the world is generated in this way. In the United States and a few other countries, three-phase power is gener-
ally not allowed to enter a person’s home. Even in the areas where it does, it is typically split out at the main distribution
box, and most residential appliances are powered by the electric current that has only one phase. However, three-phase
current is still used in most of North America to power electric stoves, electric clothes dryers, and some types of dish-
washers. So if your electricity is generated by a nuclear power plant, then this is the form that it will eventually arrive at
when it reaches your home (see Figure 1.25). Figure 1.26 shows a large electrical generator that is used in a hydroelectric
dam (on the left) and another generator that is used in a nuclear power plant (on the right). A typical generator can be
between 5 and 10 m long, and it can weigh several thousand tons.
The exact size and shape depends on the rating of the electric generator and how many generators are needed to
generate the total electrical output of the plant. However, nuclear power plants having between 8 and 16 large electri-
cal generators are fairly common. Each generator is then tasked with the job of generating between 20 and 100 MW of
electrical power. For a variety of reasons, electric generators above 100 MWE are not very commonly used in commer-
cial applications. If you have ever been to one of the world’s larger dams, such as the Hoover Dam in North America,
or the Three Gorges Dam in China, you can begin to appreciate the size of these monumental devices. So in summary,
a nuclear reactor creates heat in the core to heat the coolant, the SGs convert this heat into high-pressure steam, and
the power turbines use this steam to turn the rotating shafts of large devices called electrical generators to convert the
rotational energy of the shafts into electric power. The movement of the rotors past the stators inside of the generators
then generates an electric current, which is then sent to the electrical power grid. An electric meter (see Figure 1.27)
is then used to measure the amount of electricity that is used in your home. Normally this electricity is measured in
kilowatt hours (or kWh).
+
180° 540°
0
360° 720°
FIGURE 1.25 How the voltages are summed together to form the three-phase current that enters your home.
(a) (b)
FIGURE 1.26 A large electrical generator used in a dam (a) and (b) a similar generator used in a nuclear power plant. Note the
immense scale and size that these electrical generators can have. (Pictures provided by Wikipedia.)
1.24 Common Measurements of Electrical Power Production 29
FIGURE 1.27 Two pictures of electric meters showing the usage of electricity in kilowatt hours.
1.24
Common Measurements of Electrical Power Production
The electrical generators in a nuclear power plant produce electricity in exactly the same way that the generators in a
coal-fired power plant do. The only significant difference is that the electrical generators in a coal-fired plant turn at a
slightly higher speed because the steam reaching the steam turbines from the boilers is hotter (i.e., it has a higher energy
content or specific enthalpy h). The AC produced by the electrical generators is then sent to the electrical power grid,
where it ultimately makes its way to your home. Every joule of electrical energy produced per second is then equivalent
to 1 W of electrical power.
Electric utilities normally measure the amount of energy that reaches your home in a slightly different unit of energy
called the kilowatt hour. A kilowatt hour is the amount of energy consumed by multiplying the number of watts of elec-
tric power used per hour times the number of hours. In other words, it is equivalent to 1,000 J of electricity being used
per second times the number of seconds in an hour (3,600 s). For this reason, 1 kWh is equivalent to 1,000 watt hours or
3.6 MJ of electricity. Because the thermal efficiency of most commercial nuclear power plants is about 33%, the reac-
tor core in one of these plants must generate approximately three times as many joules of thermal energy to produce
1 kWh. In other words, 1 kWh requires a nuclear power plant to produce approximately 3 × 3.6 = 10.8 MJ of thermal
power. The electric utility then bills you for this power based on the number of kilowatt hours that you consume (see
Figure 1.28). In Canada, China, India, and the United States, one kilowatt hour of electricity costs about 12 cents (the
actual range is between 6 and 15 cents). Hence, every time the core of a nuclear reactor produces 10.8 MJ of thermal
power, an electric utility charges you 12 cents (in USD) for this electrical power. It is helpful to keep these numbers in
mind the next time you receive your electric bill. Fortunately, nuclear power plants can produce electricity much more
cheaply than other types of power plants can, and this is one of the reasons why there are so many nuclear reactors in
operation in the world today.
FIGURE 1.28 The Empire State Building in New York City is a famous tourist attraction. The annual electric bill for the Empire
State Building is about $100 million—a truly astronomical amount of money, even by today’s standards, that the tenants of the
building are glad to pay!
1.25
Nikola Tesla and Thomas Edison and Their Contributions
to the Field of Nuclear Power
The exact way that a nuclear power plant interfaces to the electrical distribution system (or the electrical power grid)
in a particular country is a subject of some importance to the nuclear power industry as a whole. However, the reason
why a nuclear power plant (or a coal-fired power plant for that matter) interfaces to the power grid in the way it does
is the result of a number of interesting discoveries that occurred in the 1800s and the early 1900s. Almost all of these
discoveries were connected in some way to the motion of electrical charges through thin wires and transmission lines.
So in this section, we would like to take a moment to explore the history of the electrical power industry and to show
how the use of nuclear power became related to it. After Maxwell’s equations of electricity and magnetism were first
discovered in the 1800s (see Chapter 2 of our companion book*), people tried to use them to understand how they could
provide the power that was needed to sustain the industrial revolution. This led to the rise of famous entrepreneurs and
industrialists such as Andrew Carnegie, the Mellons, JP Morgan, the Rockefellers, and the Vanderbilts. A picture of JP
Morgan together with his friend and partner Thomas Edison can be seen in Figure 1.29. Most of the industrial world that
we live in today owes its existence to the efforts of about a dozen of these great men. However, the actual scientists and
engineers who made all of this possible generally did not become rich (or necessarily famous) as the result of their work.
In this section, we would like to discuss the efforts of two of these great men (Thomas Edison and Nikola Tesla) who
created much of the electrical infrastructure of the modern world that we live in today. If one looks into their efforts
objectively, many of the key components of nuclear power plants (and coal-fired power plants for that matter) were
invented by these men. Nikola Tesla (see Figures 1.30 and 1.31) then became indirectly responsible for the success of
George Westinghouse and the Westinghouse Electric Corporation. Westinghouse Electric is now the parent company of
the Westinghouse Nuclear Corporation, which dominates the world market for PWRs. So, we would now like to take a
moment to show you how important the flow of electricity and charged particles through wires can be. What we will dis-
cuss here is generally not discussed in other nuclear engineering textbooks. As a matter of fact, it is not discussed in any
nuclear engineering book that either the author or the publisher is aware of. Thus, the reader is often left to wonder how
the business of moving electricity through wires evolved and the role that coal-fired plants, and then eventually nuclear
plants, played in making the business a commercial success. Hence, to begin we would like to present a brief historical
overview of how the electrical distribution system in the industrial world evolved, and why nuclear power plants (and
coal-fired power plants) attempt to interface to it in essentially the same way. Since the system is based on the principle
of moving large numbers of charged particles (e.g., electrons) through thin wires and transmission lines, this is a subject
that all nuclear and electrical engineers need to understand.
1.25 TESLA, EDISON, AND THEIR CONTRIBUTIONS 31
(a) (b)
FIGURE 1.29 A picture of Thomas Edison (a) with his mentor and financier, industrialist JP Morgan (b). (From Google Images.)
(a) (b)
FIGURE 1.30 A picture of George Westinghouse (a), who pioneered the transmission of electrical power in the United States,
together with his friend and business partner Nikola Tesla (b). He invented the air brake and eventually made air brakes standard
equipment on all American trains. He developed the rotary steam engine and later applied the same principle to develop a water
meter. The same year he invented a device for placing derailed freight cars back on their tracks. Eventually, he received more than
100 patents and other awards and developed the rotary fan, the dishwasher, the refrigerator, and many similar devices. He founded
the Westinghouse Electric Corporation, which is now the largest supplier of military and commercial nuclear reactors in the world
today. (From Google Images.)
FIGURE 1.31 A picture of Nikola Tesla and one of his Tesla Coils in action. (Courtesy of Google Images.)
32 Nuclear Power in the World Today
At the start of the industrial revolution, oil discovered and refined by John D. Rockefeller (who ultimately became the
richest man who ever lived with a total net worth of about two-thirds of a trillion dollars) was the fuel that was used to
power the machines that were invented to automate the work of millions of individual people. However, transporting
this oil to remote locations eventually became a problem, and so another approach had to be found to provide a remote
power source. It was widely known at the time that an electric current could be made to flow through a wire if there
was a difference in the voltage between the ends of the wire. However, a great debate arose as to exactly what type of
electric current should flow through these wires over long distances: alternating current (or AC), which was invented
and patented by Nikola Tesla, or direct current (or DC), which was invented and patented by Thomas Edison.
In 1887, Tesla showed that three-phase alternating current is generally preferable to other forms of AC power
because it uses less electrical conducting material to transmit the same electric power than equivalent single-phase or
two-phase systems having an identical voltage. It also has lower transmission line losses than DC, or direct current
does. Hence, electrical generators were designed to take advantage of this fact, and this resulted in the electrical genera-
tors that are used in nuclear power plants today. Tesla also went on to invent the fluorescent light bulb, the Tesla induction
motor, and the Tesla coil, and he developed the AC electrical supply system that included a motor and transformer and
three-phase electricity. According to http://inventors.about.com, Tesla is now credited with the invention of the modern
radio as well. Moreover, the Tesla coil, invented in 1891, is still used in radio and television sets and other types of elec-
tronic equipment. Nikola Tesla was also Thomas Edison’s chief competitor at the end of the nineteenth century. In all,
Nikola Tesla was responsible for more than 100 patents and he invented countless other unpatented devices. His chief
rival, Thomas Edison, is best known for his invention of the phonograph, the telegraph, and the modern electric light
bulb. Thomas Edison still holds the world record with a total of 1,093 patents and other inventions. The town of Edison,
New Jersey, is also named in his honor. Very few men have ever changed the industrial world as much as these two men
did. Moreover they did so without the help of the government, without any government handouts, and without the help of
any social or academic institution. Simply put, they were just great entrepreneurs and capitalists whose collective vision
and drive laid the groundwork for much of the industrial world that exists today.
1.26
T he Relationship between Nikola Tesla and George Westinghouse
Then in 1885, George Westinghouse, founder of the Westinghouse Electric Company, bought the patent rights to Tesla’s
system of dynamos, transformers, and motors. According to historians, Westinghouse used Tesla’s AC system to light
the World’s Exposition in Chicago in 1893. Eventually, Tesla’s designs began to show up in coal-fired power plants. And
finally in the early 1950s, the derivatives of his designs began to appear in nuclear power plants. So Tesla, Westinghouse,
and Edison are probably more directly responsible than any other men who have ever lived for the ultimate rise of
the electrical power industry in the United States and, then over a period of time, the electrification of the world. The
Westinghouse Electric Corporation is now the largest supplier of commercial nuclear power plants in the world today.
The “back ends” of most nuclear power plants today continue to use the same equipment that Tesla and Edison designed
over 100 years ago. However, the equipment has been updated several times since then, and so it is considerably more
BIBLIOGRAPHY 33
advanced. Nevertheless, their same basic designs are still in use today. Eventually, the “front ends” of most modern
nuclear power plants were invented about 100 years later when Robert Oppenheimer, Enrico Fermi, and a couple of other
great men (including Richard Feynman and John Von Neumann) unlocked the secrets of the atom. We will delve more
into the contributions of these great men in our companion book.*
1.27
Reviewing What We Have Just Learned
At this time, we would like to take a moment to review what we have learned in this chapter, because in Chapters 2, 3,
4, and 5, we would like to discuss each specific type of reactor introduced in this chapter in more detail. Basically, there
are five types of commercial power reactors in widespread use in the world today. They are as follows:
1. PWRs
2. BWRs
3. HWRs (which are another variation of the PWR)
4. GCRs
5. LMFBRs
Counting the reactors currently under construction today, there will probably be over 500 commercial nuclear power
plants in operation by the year 2020. PWRs are the most common type of reactor in use today (accounting for about two-
thirds of the world market share), and BWRs are the second most common commercial reactor type. On average, these reac-
tors supply about 20% of the world’s electrical power needs. In some countries such as France, this number is considerably
higher (about 80%). In the foreseeable future, it is likely that the number of commercial nuclear power plants in the world
will continue to increase, and within 50 to 60 years, there will probably be 1,000 commercial nuclear reactors in operation
worldwide. In general, these power plants have an extraordinary safety record, and only three major reactor accidents have
occurred since the nuclear power industry was born in the early 1950s (TMI, Chernobyl, and Fukushima). However, the
accident at Fukushima was actually caused by a natural disaster beyond the stated design limits of the plant, and so it could
have been avoided if the large earthquake off the coast of Japan did not occur*. So, in spite of these three unfortunate inci-
dents, nuclear power will continue to play a major role in meeting the world’s energy needs, and as oil, coal, natural gas, and
other natural resources continue to decline. Consequently, the number of nuclear power plants in the world will continue to
increase for the foreseeable future. In future chapters, we will attempt to demonstrate how these plants behave and operate on
a day-to-day basis. We will then attempt to explore their neutronic and thermal-hydraulic performance as a function of time.
Bibliography
Books and Textbooks
Cochran, R.G. and Toulfanidis, N. The Nuclear Fuel Cycle: Analysis and Management, American Nuclear Society, La Grange Park,
IL (1999).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley & Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., New York,
NY (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. An Introduction to Nuclear Reactor Physics, First edition, CRC Press, Boca Raton, FL (2017a).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, First edition, CRC Press, Boca Raton, FL (2017b).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Waltar, A.E. and Reynolds, A.B. Fast Breeder Reactors, Pergamon Press, New York, NY (1981).
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
34 Nuclear Power in the World Today
Web References
http://inventors.about.com.
http://www.westinghousenuclear.com/.
http://en.wikipedia.org/wiki/Nikola_Tesla.
http://en.wikipedia.org/wiki/Steam_turbine.
http://en.wikipedia.org/wiki/Thomas_Edison.
http://en.wikipedia.org/wiki/Electric_generator.
http://www.asope.org/pdfs/AC_Electrical_Generators_ASOPE.pdf.
http://www.britannica.com/EBchecked/topic/641020/George-Westinghouse.
http://www.ge-energy.com/products_and_services/products/steam_turbines/.
http://www.westinghousenuclear.com/Our_Company/history/george_westinghouse.shtm.
33. Which reactor of the following four reactor types has the highest core power density, and which reactor has the
lowest: PWRs, BWRs, CANDU reactors, and LMFBRs?
34. What type of fuel does a commercial PWR use?
35. Which reactor uses heavy water and natural uranium fuel?
36. Which reactor has three separate coolant loops, and why does it have this many loops?
37. In the nuclear power business, what does the term nuclear steam supply system refer to?
38. Who is the primary manufacturer of BWRs in the world?
39. What is the thermal efficiency of a typical PWR?
40. Approximately what percentage of the electric power in the United States is generated by nuclear power?
41. What is the shape of the control rods in a PWR, and what is the shape of the control rods in a BWR?
42. Which reactor has a higher thermal efficiency: a PWR, a BWR, or an LMFBR?
43. In the nuclear power business, what is the word SCRAM an abbreviation for?
44. Suppose that 1 g of uranium is completely converted into energy in a nuclear power plant. Assuming that the plant
thermal cycle is 33% efficient, how many kilowatt hours of electricity does this produce?
45. At the average cost of electricity in the world today, approximately how many dollars of electricity does this 1 g of
burned uranium produce?
46. Assume that 1 kg of uranium ore costs $100 and that it contains 0.711% U-235. Assuming that all of the U-235 is
ultimately recoverable, how many kilograms of natural uranium does it take to produce 1 g of U-235?
47. Using the information presented in questions 44, 45, and 46, estimate the fuel cost per kilowatt hour for a typical
nuclear power plant in the world today (excluding enrichment and fabrication costs).
48. What percentage of the commercial nuclear reactors in the world today are PWRs?
49. Where is the fuel “bred” in a fast breeder reactor?
50. What industrial country has the highest percentage of its electrical power generated by nuclear power plants?
51. What is the average thermal efficiency of a Canadian CANDU reactor?
52. Write a mathematical expression for the definition of the power density for a nuclear reactor core.
53. What is the average burnup of a fast reactor fuel assembly?
54. Who was the richest man who ever lived?
55. Which has a higher average burnup: a fast reactor fuel assembly or a thermal reactor fuel assembly?
56. Nuclear power plants are a major competitor to coal-fired power plants in the world today. Is the energy content of the
cooling water in the core of a nuclear reactor higher or lower than that in the boilers of a modern coal-fired power plant?
57. What is the typical outlet temperature of the water from a commercial PWR core, and what is the typical outlet
temperature from the steam from one of its SGs?
58. If you were put in charge of building a nuclear power plant today, what type of nuclear power plant would you build
and why?
Exercise 1.2
Suppose that a research reactor core 1 m on a side has a thermal power output of 100 MW. If the reactor core contains
16 fuel assemblies in the shape of a cube, what do you expect the average power output and the average power density
of an individual fuel assembly to be?
Exercise 1.3
A plutonium-fueled fast breeder reactor has an electrical power output of 500 MW. If the core is cooled with liquid
sodium, what does the total volume of the core have to be, and what is its thermal power output?
Exercise 1.4
A thermal water reactor is cooled with light water where the coolant enters the core with an average energy content or
enthalpy of 1.3 MJ per kilogram and exits the core with an average energy content or enthalpy of 1.5 MJ/kg. If the reactor
36 Nuclear Power in the World Today
is a PWR with an electrical power output of 1,000 MWE, what must the average flow rate through the core be to remove
the heat generated by the nuclear chain reaction?
Exercise 1.5
In Section 1.24, we learned that a well-designed thermal water reactor can produce retail electrical power at a cost to
the consumer of between 8 and 12 cents/kWh. Using the average of these values, calculate the amount of revenue that
a 1,000 MWE nuclear power plant can produce by operating at full power for a single calendar day. How much revenue
is lost when it must be shut down for refueling every 18 months? Assume that the average downtime in this case is 30
calendar days.
Exercise 1.6
A CANDU reactor with a thermal efficiency of 29% is being compared to a Westinghouse PWR with a thermal effi-
ciency of 34%. If the fuel cost for the CANDU reactor is 10% less than the fuel cost for the Westinghouse PWR (because
the fuel for the CANDU reactor does not have to be enriched), which power plant will generate electricity more cheaply
if the capital cost for the two power plants is exactly the same? Assume that the capacity factor for both power plants is
the same as well.
Exercise 1.7
Reactor A operates at a capacity factor of 87% and reactor B operates at a capacity factor of 92%. If reactors A and B
have exactly the same fuel cycle cost, and the thermal efficiency of reactor A is 32%, while the thermal efficiency of
reactor B is 34%, which reactor generates nuclear power more cheaply? If the cost of the electric power being produced
by reactor A is 10/kWh, what is the cost of the electric power being produced by reactor B?
Exercise 1.8
Earlier in the chapter, we learned that the core exit temperature of an LMFBR is considerably higher than it is for a
PWR. Using representative values for these two numbers, and assuming that the heat in both cases is rejected from
the condenser at an average temperature of 150°C, how do the maximum theoretical thermal efficiencies of these two
reactors compare?
Exercise 1.9
Suppose that the wall thickness of a reactor pressure vessel increases as the square root of the maximum system p ressure.
Using this fact alone, estimate the ratio of the thickness of the pressure vessel in a PWR to the thickness of a pressure
vessel in a BWR. Assume that the reactor pressure vessels are made up of the same material in each case.
Exercise 1.10
A wise man comes to you and claims that it is possible to design a new type of reactor that converts half of all the
heat that is produced in the core into electric power. Is this a realistic claim to make in light of what you have learned
so far in this chapter? If so, what type of reactor would this be most likely to be? What type of reactor thermal cycle
would it use?
Exercise 1.11
Suppose that you are asked to reduce the power density in a reactor core by a factor of two. If the flow rate remains
the same, and ordinary water is used to cool the core, approximately how much can the system pressure be reduced
in the primary coolant loop before the water begins to boil? Assume that the initial pressure in the primary loop is
2,000 PSI.
Exercise 1.12
Suppose that two reactors have fuel rods that are exactly 1 cm in diameter. Reactor A has a square lattice with a pitch
of 12 cm, and reactor B has a square lattice with a pitch of 15 cm. If the enrichment of the fuel rods is exactly the same,
what is the ratio of the power density of the core in reactor A to the power density of the core in reactor B?
2
The Pressurized Water Reactor
2.1 P ressurized Water Reactors
Pressurized water reactors (PWRs) are the most common type of nuclear reactors in the world today. They account for about
two-thirds (67%) of the world’s nuclear generating capacity, and the primary manufacturers of PWRs in the Western world are
Westinghouse and AREVA. Their designs are slightly different from one another, although they have many things in common.
A PWR core uses light water (normal tap water that is highly distilled) for both the moderator and the coolant, and it is kept under
considerable pressure (about 15.5 MPa or 2,250 PSI) to prevent it from boiling. It enters the core at about 290°C and leaves the core
at about 325°C. It then flows to the steam generators (or SGs) where some of the heat is removed, and it is then pumped back into the
core from below. Having the water flow over the fuel rods increases its average temperature by about 35°C. Its density also decreases
as it is heated, and there can be anywhere between two and four steam generators connected to the primary loop. A picture of a typical
PWR is shown in Figure 2.5, and the core is shown in Figure 2.1.
37
38 The Pressurized Water Reactor
Control rod
drive mechanism
Reactor vessel
head
(a)
FIGURE 2.1 (a) A picture of a reactor core from a Westinghouse PWR. (Courtesy of the Westinghouse Nuclear Corporation.)
(b) The flow path of a fluid particle through the reactor pressure vessel and a cross-sectional view of the pressure vessel from above.
The flow path is shown in red on the left. (These pictures were drawn by the author.)
2.4 Flow Paths through the Pressure Vessel and the Core 39
TABLE 2.1
Typical Pressure Vessel Design Parameters for a Westinghouse PWR
TABLE 2.2
Common Materials Used in the Construction of PWR Pressure Vessels and Their Components (Flanges, Rings, Nozzles, and so
on) in the United States
2.4 F low Paths through the Pressure Vessel and the Core
In virtually all PWRs, the coolant enters the pressure vessel through a series of inlet nozzles and exits the core through a
separate set of outlet nozzles. The average inlet temperature is about 290°C, and the average outlet temperature is about
325°C. However, these numbers can vary depending on the number of loops required to remove the heat. Each loop has
40 The Pressurized Water Reactor
TABLE 2.3
Typical Core Design Parameters for Two-Loop, Three-Loop, and Four-Loop Westinghouse PWRs
a separate inlet nozzle and a separate outlet nozzle. After the coolant enters the pressure vessel, it is directed downward
between the core barrel or core shroud and the pressure vessel wall, and then enters the core from below. The flow is
distributed between the individual fuel assemblies using what are called orifice plates. The heated coolant emerges into
the outlet plenum and then exits the core. The actual flow path of a fluid particle through the pressure vessel is shown in
Figure 2.1(b). About 2% of the flow bypasses the fuel assemblies entirely and travels between the outer edge of the fuel
assemblies and the core shroud. This flow is sometimes called core bypass flow.
Instrumentation channels enter the core from below, and the control rods enter the core from above. The control
rod drive mechanisms (CRDMs) are welded to the top of the pressure vessel and are therefore an integral part of the
pressure vessel head. Since the pressure in the reactor pressure vessel is maintained at approximately 2,250 PSI
(15.5 MPa), the water does not boil—at least to any great extent. The steam for the power turbines is produced by
another device called the steam generator or SG (see Section 2.5). High-temperature heated water enters the steam
generators from below and passes upward through several thousand pressure tubes that are shaped in the form of an
inverted “U.” It is not uncommon for large U-tube steam generators to have between 20,000 and 30,000 of these
tubes. The outer surfaces of these tubes are surrounded by lower pressure and lower temperature feedwater returning
from the power turbines. This cooler water absorbs heat from the hot water inside of the tubes, and this causes the
cooler water on the outside of the tubes to boil and produce steam. This boiling occurs in a lower section of the steam
generator called the evaporator. The wet steam produced in the evaporator rises and enters a second section of the
steam generator called the steam drum. Here, the wet steam is dried using moisture separators before being sent onto
the steam turbines. The design of these steam generators is discussed in Section 2.5.
in the primary loop, and it is designed to evaporate into saturated steam at a temperature of about 275°C (530°F) or into
superheated steam at higher temperatures. It is then fed to the steam turbines, condensed, and returned to the steam
generators again. The secondary-side pressure is about 900 PSI (6.2 MPa or 60 atm), and the coolant on the primary side,
after it has lost some of its energy, is returned to the reactor vessel to be heated again. Steam generators come in two pri-
mary configurations: one configuration is called a once-through steam generator (OTSG), and the other configuration is
called a U-tube steam generator. In an OTSG, the pipes are straight and narrow, and in a U-tube SG, they are bent in the
form of a “U.” U-tube SGs are smaller than OTSGs, and their primary advantage is their smaller size and their slightly
lower weight. However, all steam generators tend to be very heavy devices that require a great deal of care to assemble
and operate properly. It is essential to maintain the correct water chemistry inside of the tube bank or the tubes inside of
the steam generator will fail or corrode.
The steam produced by most steam generators is slightly superheated to optimize the thermodynamic efficiency of
the plant. Pictures of a vertical U-tube SG are shown in Figure 2.2. Most steam generators used in nuclear power plants
are 25–35 m long. They can be mounted in either a vertical or a horizontal position, although a vertical orientation is nor-
mally preferred to a horizontal orientation to conserve space inside of the containment building. Both Westinghouse and
AREVA PWRs use steam generators that are mounted vertically rather than horizontally. However, in Russian PWRs
(also called VVERs), the steam generators can be either horizontal or vertical. An example of a horizontal steam gen-
erator is shown in Figure 2.3. The flow rate through these steam generators is carefully controlled to make the steam as
dry as possible. For example, Westinghouse steam generators use integrated moisture separators with very precise flow
Model 54F
Westinghouse
steam generator Steam outlet to
turbine generator
Secondary moisture
separator
Secondary side manway
Instrument connection
Upper shell
Primary moisture
separator swirl vanes
Instrument connection
Transition cone
Tube bundle
Channel head
Support foot
Primary coolant inlet Primary side manways
(outlet rotated out-of-view)
FIGURE 2.2 A cross-sectional view of a U-tube steam generator from a Westinghouse PWR. (From Westinghouse Nuclear.)
42 The Pressurized Water Reactor
control to produce superheated steam with an exit quality of 100%. Thus, the steam that is sent to the power turbines is
completely dry, and this helps improve the thermal efficiency of the power plant. Modern nuclear plants normally sup-
port between 10 °C and 20 °C of superheat. Large PWRs can have as many as four steam generators, whereas smaller
PWRs can have either two or three. Each steam generator is paired with a separate leg within the primary coolant loop.
Hence, a PWR with four steam generators will have four legs within the primary loop. An example of this configura-
tion is shown in Figure 2.5. Normally, this configuration is the type of configuration one would expect to see in a 1,000
MWE (3,300 MWT) Western PWR.
(a) (b)
FIGURE 2.4 A rendering of a U-tube steam generator (a) and the exposed tube bank from a U-tube steam generator (b). (Pictures pro-
vided by the Westinghouse Nuclear Corporation and Wikipedia—see http://me1065.wikidot.com/nuclear-pressurized-water-reactor-pwr.)
2.7 Reactor Coolant Pumps 43
TABLE 2.4
Design Data for the Westinghouse Model F steam generator, Which Is Used Extensively in Modern Nuclear
Power Plants
section contains moisture separation equipment. The steam drum is normally located at the top of the steam generator.
The specific design parameters for a Westinghouse steam generator are shown in Table 2.4. Water from the primary
loop flows in the opposite direction of steam produced in the secondary loop. This particular flow configuration results
in what is known as a counterflow steam generator because the flow streams are aligned in opposite directions.* Steam
generators used in nuclear power plants are designed and manufactured in accordance with Sections 2, 3, and 9 of the
ASME Boiler and Pressure Vessel Code. The vertical U-tube steam generator with an integrated steam drum was first
used by Westinghouse in a commercial nuclear power plant in 1960. Since that time, this design has undergone several
changes and major design revisions. The design changes have normally been evolutionary rather than revolutionary. The
model F steam generator has been the standard steam generator in Westinghouse PWRs since 1976. However, in the
AP-1000, another steam generator (which was originally designed by Combustion Engineering) is now used.
* In some steam generators, the fluids from the primary and secondary loops flow in the same direction. In this case, the SG is called
a parallel flow steam generator, but it is not as efficient at extracting energy from the primary loop as a counterflow steam generator.
44 The Pressurized Water Reactor
Pressurizer Steam
generator
Reactor coolant
pump (RCP)
Piping
system
Pressure
vessel
FIGURE 2.5 The RCPs, the SGs, the pressurizer, the reactor vessel, and the primary coolant loops in a large PWR. (Courtesy of
Westinghouse—see http://me1065.wikidot.com/nuclear-pressurized-water-reactor-pwr.)
FIGURE 2.6 The cross-sectional view of an RCP from a Westinghouse PWR. (Picture provided by the Westinghouse Nuclear
Corporation.)
2.8 The Pressurizer 45
TABLE 2.5
Design Parameters for a Westinghouse RCP
In modern nuclear power plants, the induction motor and pump assembly can be removed from the casing for inspection
and maintenance without having to separate any of the coolant pipes from the casing. The parts of the coolant pump
that come in contact with the coolant are made from stainless steel, and the bearings are the only major exception to this
rule. A large RCP normally contains three radial bearings that are made up of specially designed alloys. Two of these
are located in the motor, and the third is used to support the shaft. In most RCPs, the motor is a single-speed, air-cooled,
three-phase induction motor. The operating speed of the motor is about 1,200 RPM. The average horsepower of a RCP
is about 7,000 HP, which is about the same as the power output of 30 to 40 passenger cars. CANada Deuterium Uranium
(CANDU) reactors, which are another example of a PWR, use heavy water rather than light water in the primary loop
to cool the core. Heavy water allows these reactors to operate using uranium dioxide fuel derived from natural uranium
alone. Because of this, CANDU reactors use one type of RCP to pump the heavy water and another type of RCP to pump
the light water. Light water is always used to cool the secondary loop. However, the designs of these coolant pumps are
otherwise very similar.
2.8 T he Pressurizer
Another extremely important component of a PWR is an interesting device called a pressurizer. The pressurizer is used
by the reactor operators to maintain the pressure level in the primary loop and to prevent the coolant in the core from
inadvertently boiling. A picture of a typical pressurizer is shown in Figure 2.7. Pressurizers are massive devices that can
be between 8 m and 12 m high. They are essentially huge surge tanks that maintain the system pressure level at about
2,250 PSI (15.5 MPa) in the primary loop. The top of the pressurizer is filled with an inert gas, and the bottom of the
pressurizer is filled with ordinary water. An electrical heater at the bottom of the pressurizer heats the water and causes
it to expand. This thermal expansion increases the pressure level in the tank, and this causes the pressure in the primary
loop to increase. The opposite process is used to reduce the system pressure when the reactor becomes too hot.
46 The Pressurized Water Reactor
FIGURE 2.7 A cross-sectional view of a pressurizer used in a Westinghouse PWR. (Courtesy of the Westinghouse
Nuclear Corporation.)
A spray nozzle at the top of the pressurizer adds cold water to it from the cold leg, and this condenses the gas in the
upper part of the tank, which causes the pressure level in the pressurizer (and the primary loop) to fall. In the event of
a rapid power surge, the pressure level in the primary loop can spike, and the pressurizer responds by opening a safety
relief valve on the top of the pressurizer. In this way, the system pressure can be carefully controlled and/or monitored.
In other words, the primary function of the pressurizer is to limit system pressure changes in the primary loop during
planned or unplanned transients. During steady-state operation, about 60% of the pressurizer volume is occupied by
water and the remaining 40% is occupied by steam. The water in the pressurizer is normally kept at an operating tem-
perature of about 345°C, which gives a subcooling margin (the difference between the pressurizer temperature and the
highest temperature in the reactor core) of about 15°C. Thermal transients in the core can result in large swings in the
volume of the water in the pressurizer, and so the pressurizer must be designed to withstand these transients without
uncovering the electrical heaters or emptying the pressurizer entirely.
A pressurizer can also serve as a pressure compensation device when the demand for electricity increases and
decreases. For example, if the demand for electricity is temporarily reduced at night, this will cause a temporary
increase in the reactor coolant temperature. The coolant in the primary loop will expand, and this will cause the water
level in the pressurizer to rise. This increase in the water level compresses the steam and activates the pressure relief
valves at the top of the pressurizer so that cold water can be sprayed onto the steam to condense it. This quenching
action reduces the pressure in the primary loop and limits the system pressure increase. Conversely, an increase in the
plant electrical load, which normally occurs during the day, will temporarily decrease the average coolant temperature,
and the coolant volume in the primary loop will be reduced as well. Coolant then flows from the pressurizer into the
primary loop to give the cooling system the additional water it needs. Water in the pressurizer then flashes to steam
to limit the pressure reduction. This cycle may repeat itself several times during the course of a day while the electri-
cal load fluctuates up and down. The design parameters for a Westinghouse pressurizer are shown in Table 2.6. The
short-term control of a PWR is usually accomplished by moving control rods that are inserted up and down into the
core from above. The control rods are usually made up of boron carbide or some comparable material that is able to
absorb a large number of thermal neutrons. (Indium, cadmium, and hafnium are some of the others.) Control rods are
discussed in more detail in Chapter 5. However, PWRs are primarily controlled by using a chemical shim system (see
the discussion regarding this in Chapter 1 as well). In this type of system, boric acid is dissolved in the coolant (at a
concentration of about 1,500 parts per m illion [PPM]), and the Boron-10 in the boric acid is then used to absorb the
additional neutrons.
2.9 FUEL ASSEMBLIES FOR PWRs 47
TABLE 2.6
Design Parameters for the Pressurizers in a Westinghouse PWR
The fuel in a PWR can also have burnable poisons such as gadolinium mixed into the fuel rods to reduce the amount
of radial power peaking and to eliminate the need for additional chemical shim over the lifetime of the core. A picture
of how the concentration of the boron in the chemical shim is varied as a function of time in a typical PWR is shown
in Figure 2.27. When the concentration of soluble boron falls to 0, the fuel is completely burned, and some of the fuel
assemblies in the core have to be replaced. Normally, this happens at time intervals between 12 and 18 months. Then,
one quarter to one-third of the fuel assemblies in the core need to be replaced at the same time. When the size of a PWR
core becomes important, the core can be made smaller by increasing the average enrichment of the fuel. Normally,
nuclear submarines, aircraft carriers, and nuclear-powered rockets use very small and highly enriched reactor cores to
generate the power required for the propulsion systems. In the case of a nuclear submarine, the core of a reactor is only
allowed to be a meter or two tall, and to make this possible, the amount of U-235 in the fuel must be between 90% and
95% (by weight). Unfortunately, highly enriched uranium can be very expensive, and because of this, it is not economi-
cally feasible to use uranium of this type in a commercial power plant. However, its applications in military reactors and
in space probes have continued to grow.
A typical 1,000 MWE PWR can contain between 180 and 200 fuel assemblies, and each fuel assembly can contain between
250 and 350 fuel rods. This means that a 1,000 MWE PWR can have between 40,000 and 50,000 fuel rods, and between
15 and 18 million fuel pellets. Once the fuel assemblies have been loaded into the core, they may stay in the core for several
years depending upon the length of the refueling cycle. During refueling, which typically occurs at 12, 18, or 24 month
intervals, some of the used fuel—usually one-third of the core—is removed to storage, while the remaining fuel is put in
another location in the core to be burned again. This helps to flatten the radial power profile and to extract more energy from
the remaining unburned fuel. Normally, an individual fuel assembly generates between 15 and 17 MW of thermal power.
When the thermal efficiency of the power plant is taken into account, this means that each fuel assembly in the core
is responsible for generating about 5 MW of electrical power. This number declines somewhat as the fuel continues to
be burned. Flattening the radial power profile helps a PWR to operate more efficiently because it minimizes the release
of radioactive fission gases from the fuel. Russian PWRs are called VVERs, and their fuel assemblies are arranged in
hexagonal arrays (see Figure 2.10). Otherwise, overall their length and structure are very similar to those of a Western
PWR. The fuel assembly of a modern PWR is shown in Figures 2.8 and 2.9. In the next section, we would like to discuss
how a Russian VVER reactor is different than a Westinghouse or AREVA PWR.
Fuel rod
Control rod guide tube
FIGURE 2.8 A 17 × 17 fuel assembly used in a modern Westinghouse PWR. The blue water holes are reserved for reactor
instrumentation and control rods. (Picture drawn by the author.)
Rod cluster
control assembly
Top nozzle
Bottom
nozzle Fuel rod Plug
Grid assembly
Control rod
guide thimble Pellet
tube Fuel tube
Fuel rod
Fuel rod
FIGURE 2.10 Pictures of some hexagonal fuel assemblies from a Russian VVER reactor. (Courtesy of the World Nuclear
Association.)
50 The Pressurized Water Reactor
TABLE 2.7
Design Parameters for a VVER-1000 Reactor
from the fuel rods as they were in earlier designs (Table 2.7). Fuel rods in VVERs are about the same size as those in a
Western PWR. A fuel rod is about 9 mm in diameter, and the fuel pellets have a diameter of about 7.5 mm. The cladding
is about 0.65 mm thick. The burnup is slightly lower than that in a conventional PWR, although the difference between
the two designs has tended to narrow over time. The steam generators are also OTSGs having long straight banks of rods
that deliver the heat from the primary side to the secondary side. However, in the case of a VVER, the steam generators
are positioned horizontally rather than vertically. This is the primary difference between how the nuclear steam supply
system (the NSSS) differs between a VVER and most Western PWRs. Long-term control of the plant is maintained
by varying the boric acid concentration in the coolant. In Russia, the primary alternative to the VVER-1000 is the
RBMK BWR, whose picture is shown in Figure 2.11 and whose design parameters are shown in Table 2.8. The reactor
at Chernobyl was an example of an RBMK with a positive temperature coefficient. We will discuss the RBMK in more
detail in Chapter 3.
FIGURE 2.11 The exposed core of a large Russian RMBK. (Courtesy of Google Images.)
2.11 Canadian Pressurized Heavy Water Reactors (CANDU Reactors) 51
TABLE 2.8
Representative RMBK 1500 Design Parameters
absorbs fewer neutrons than light water does, and as a result of this, CANDU reactors can run on natural uranium fuel.
In other words, they do not require a uranium enrichment plant in order to operate. Sometimes CANDU reactors are
also called pressurized heavy water reactors (PHWRs). The name CANDU is an acronym for the CANada Deuterium
Uranium reactor. They currently account for about 6% of the world’s installed nuclear generating capacity. CANDU
reactors use horizontal pressure tubes in which heavy water moderates and cools the fuel. They can be run on either
natural (unenriched) uranium or slightly enriched uranium that is converted to UO2 fuel pellets. The fuel pellets are also
clad with some form of zirconium, which is usually an alloy called Zircaloy.
CANDU reactors were originally developed at the end of the Second World War, and they are based on a design that
was originally conceived by Nazi Germany. Even though heavy water does not absorb as many neutrons as light water
does, the heavy water is not as good at moderating neutrons as ordinary water is. The heavy water used in these reactors
has the symbol D2O, where D is another form of ordinary hydrogen called deuterium. Deuterium is chemically similar to
elemental hydrogen, except that it has another neutron attached to the nucleus. A picture of a deuterium atom is shown in
Figure 2.12. Pictures of heavy water molecules that contain two deuterium atoms can be found in Chapter 19. Neutrons
lose less energy per collision in collisions with molecules of heavy water than they do in collisions with molecules of
ordinary (or light) water, and because of this, they require more collisions and travel further before they reach the ther-
mal energy range (see Chapter 5). This means that the cores of PHWRs must be considerably larger than the cores of
conventional PWRs for the same power rating. Consuquently, it is simply too expensive to fabricate a PHWR pressure
vessel to operate at the pressures one would normally expect find in a conventional PWR (2,200 PSI or ~15 MPa). The
Canadian government overcame this problem by putting the fuel inside of pressure tubes instead. The D2O flows over
these pressure tubes but is not allowed to boil. The reactor vessel consists of a large cylindrical tank called a calandria
that is oriented horizontally rather than vertically. A picture of a calandria is shown in Figure 2.13. The calandria is also
filled with D2O, but the D2O in this case is very close to atmospheric pressure. So the D2O flowing through the pressure
tubes is pressurized, while the D2O in the reactor vessel is not. D2O enters these tubes at a temperature of about 266°C
and leaves these tubes at a temperature of about 310°C. It is then sent to an steam generator where steam is generated in
the secondary loop. Just like in Westinghouse PWRs, the steam is usually generated with a small amount of superheat.
The secondary coolant loop is filled with normal H2O. A cross-sectional view of a horizontal pressure tube is shown
+ +
+
– –
Hydrogen Deuterium Tritium
FIGURE 2.12 The hydrogen atom (left) compared to the deuterium and tritium atoms. The protons are colored in yellow, and the
neutrons are colored in red.
52 The Pressurized Water Reactor
FIGURE 2.13 The calandria makes up the core of a typical HWR. In this case, the calandria is the reactor vessel for a CANDU
HWR. (From candu.org.)
FIGURE 2.14 The small gaps between the fuel rods in a CANDU reactor fuel bundle carry heated water that is sent to the power
turbines. These small coolant channels are sometimes referred to as “pressure tubes.” (From Canadian Manufacturing.)
in Figure 2.14, and a picture of a complete fuel bundle is shown in Figure 2.15. A picture of the NSSS for a CANDU
PHWR is presented in Chapter 32 (also see Figure 2.17). Unfortunately, because of the low power density, the D2O in a
CANDU reactor core cannot be raised to the same pressures and temperatures as it can in an ordinary pressured water
reactor. Because of this, the thermal efficiency of most CANDUs is limited to between 28% and 30%. This is about 5
points lower than it is for a typical PWR. This also means that the steam produced in a CANDU reactor possesses less
superheat than it does in a conventional PWR. Finally, because the steam enters the power turbines at a lower pressure
(about 4.9 MPa), the blades of the turbines are designed differently than they are in conventional PWRs to extract the
maximum amount of work from the incoming steam.
FIGURE 2.15 Some fuel bundles from a typical Canadian CANDU reactor. (Picture provided by Wikipedia.)
normal operation because the thermal- and radiation-induced swelling is so small. Unlike a PWR or a BWR, the fuel
bundles are loaded into horizontal channels or pressure tubes, which extend along the entire length of the reactor
pressure vessel. In a CANDU reactor, the reactor pressure vessel is sometimes referred to the calandria. The fuel
can also be replaced while the reactor is operating at full power. Online refueling in CANDU reactors is performed
using a refueling machine such as the one shown in Figure 2.16. About 12 bundles are loaded into each fuel chan-
nel depending upon the model. A 1,000 MWE CANDU reactor contains about 600 fuel channels containing about
7,200 fuel bundles and about 6.3 million fuel pellets. The online refueling process is fully automated, and new fuel is
inserted into a channel at one end and used fuel is collected at the other. This feature means that a CANDU reactor
is inherently flexible with its fuel requirements and can run on different types of fuels depending upon the resources
that are available (Table 2.9). Each type of fuel can have a different residence time, and the fuels that can be used
range from natural uranium- to slightly enriched uranium- to plutonium-bearing fuels and thorium-based fuels. The
control requirements vary somewhat as the fuel is changed, but overall, a CANDU reactor has one of the world’s most
flexible fuel cycles (see Chapter 10). CANDU reactors have a great deal of appeal in developing countries because
one does not need a uranium enrichment plant to provide fuel for them. The overall size of a CANDU reactor, and
hence the amount of deuterium required, can be reduced substantially by using slightly enriched uranium rather than
natural uranium fuel. An average enrichment of about 1.2% appears to achieve the best balance between the extra cost
Pressure tubes
Plunger
Refueling machine
Shielding
FIGURE 2.16 A cross-sectional view of a refueling machine used to refuel a Canadian CANDU reactor while it is still operating.
(Picture provided by the author.)
54 The Pressurized Water Reactor
TABLE 2.9
Some Typical CANDU Reactor Design Parameters
of enriching the uranium and the money that can be saved by using a smaller pressure vessel. The CANDU reactor
is a remarkable achievement considering that it can run on natural uranium alone. This uranium is again mixed with
oxygen to produce a uranium dioxide or UO2 fuel pellet. UO2 is a complex ceramic material with a very high melting
point (2,865°C). In fact, its melting point is similar to the melting point of the heat shield and the heat tiles used on the
U.S. Space Shuttle (see Chapter 5). The fuel pellets in a CANDU reactor rarely melt (even during severe transients)
because the power density in the core is so low.
Light
Note: Fuel rods are water
in pressure tubes Hot steam
Control rods
Steam
RCP
generator
Calandria
Pressure
tubes
Concrete
shielding
FIGURE 2.17 The NSSS for the primary loop of a CANDU PHWR. (Picture provided by the author.)
2.14 Temperature Profiles in a PWR Core 55
TABLE 2.10
Comparison of the Power Plant Thermal Cycles for Several Types of PWRs
Type of PWR U.S. PWRa EPR U.S. APWR AP-1000 CANDU PHWR
Manufacturer Westinghouse AREVA Mitsubishi Westinghouse Atomic Energy of Canada
Plant thermal cycle Rankine Rankine Rankine Rankine Rankine
Number of coolant loops 2 2 2 2 2
Coolant in the primary loop H 2O H2O H2O H2O D2O
Coolant in the secondary loop H 2O H2O H2O H2O H2O
Thermal efficiency (%) 33.7 34.4 38.0 34.0 29.3
Number of legs in the primary loop 4 4 4 2 2
Number of pumps in the primary loop 4 4 4 4 2
Number of SGs 4 4 4 2 2
Types of SGs U-tube U-tube U-tube U-tube U-tube
Manufacturer Westinghouse AREVA Mitsubishi Westinghouse Atomic Energy of Canada
Coolant Fuel
temperature temperature
Core
centerline
Power shape
–H/2
TIN T
0
FIGURE 2.18 The axial temperature profiles in the core of a PWR with a co-sinusoidal axial power shape.
PMAX
z
Power
H PMAX
z
Power
FIGURE 2.19 An illustration of how moving the control rods can cause the flux shape to change in a PWR. In a PWR, the control
rods are inserted from above.
Figure 2.20—and because of this, the bulk temperature of the coolant never exceeds the boiling point (which is some-
times called the saturation temperature or TSAT). This ensures that a liquid film always remains on the surface of the
fuel rods to provide the cooling they require.
PWR core
Exit
Nucleate boiling
Saturated water
Coolant channel
Subcooled water
Inlet
Coolant flow
FIGURE 2.20 An illustration of where bubbles may form in the hot coolant channel of a PWR.
in 1979, many nuclear power plants had to be retrofitted to upgrade their safety systems and control systems to prevent
a similar accident from happening again. In some cases, plants had to be retrofitted differently because of design dif-
ferences between them. This had the effect of increasing the cost per kilowatt hour of electricity produced, and these
costs were ultimately passed on to the consumer. The accident at Chernobyl in 1986 also reinforced the need to have
standardized and well-designed safety systems. In an effort to make the cost of generating electric power from nuclear
plants more competitive, reactor vendors explored different ways to simplify the design of their existing plants and to
make the safety systems inside of them simpler and more reliable. This resulted in a decision to reduce the number of
components in the NSSS and to provide additional sources of cooling water in the event of an accident. This has also
reduced the number of custom parts that a particular reactor can have, as well as the total number of parts in many cases.
One other important practice that resulted from the process was the decision to use standard parts and design principles
wherever possible. For example, instead of offering reactors with an infinite range of electrical outputs, only two or three
output levels (small, medium, and large) are usually offered to electric utilities today. (The Russian VVER reactor is also
offered in only a couple of standard sizes). This is achieved by using only two, three, or four identical steam generators
in commercial PWRs and two or three identical recirculation pumps in commercial BWRs. This then results in a very
similar set of power plants having electrical outputs of 300, 600, 900, and 1,200 MWE. For example, the Westinghouse
AP-600, an advanced and relatively modular PWR, shares many of these design features. Another example of a reactor
of this type is the GE’s advanced BWR, which we will discuss in Chapter 3.
power to cool the plant. These systems are also designed so that they do not require operator intervention or human
decision-making to operate effectively. In the future, these systems will will probably be installed on all nuclear power
plants. Hence, they are designed to be fail-safe or nearly so. In addition to the core, safety systems can also be designed to
passively remove the heat generated during a reactor accident from the containment b uilding. This involves implementing
what is called a passive containment cooling system (or PCCS). Other passive systems can also be used to limit the peak
pressure in the containment building to about 50 PSI during a large break LOCA (see the discussion of LOCAs in Chapter
30 and Chapter 34). These passive safety systems will then result in safer and more reliable power plants in the future. The
Westinghouse AP-600 and AP-1000 and the GE simplified BWR (SBWR) attempt to implement these systems in different
ways. Next, we would like to compare and contrast these systems to those that are used in older power plants.
Outside cooling
air intake
Reactor components
(core, pressure vessel,
Foundation steam generators)
bundles, and hence, they do not require a large, robust pressure vessel to house the entire core. The core inlet tempera-
tures for conventional PWRs average from about 280°C to about 295°C, and the core outlet temperatures can vary from
about 320°C to about 330°C. The average coolant temperature rise is between 30°C and 35°C. This temperature rise
is essential for optimizing the thermal efficiency of the plant. However, there are usually much greater differences in
the temperatures and the pressures in the secondary loop because different steam generators, condensers, and steam
turbines are used. In particular, the Westinghouse AP-1000 has a much lower secondary-side coolant pressure than any
of the other designs. However, the thermal efficiencies of all of these PWRs (with the exception of the CANDU PHWR)
are very similar. The average thermal efficiency tends to average between 33% and 34%. In Chapter 9, we will learn that
several “tricks” are required to achieve this level of thermal efficiency.
TABLE 2.11
PWR Fuel Pin, Fuel Rod, and Fuel Assembly Design Parameters
Fuel Pins
Geometry Cylindrical Cylindrical Cylindrical Cylindrical
Diameter (mm) 8.19 8.19 8.19 8.19
Composition UO2 UO2 UO2 UO2
Enrichment (avg.)b (%) 3.5 4–5 4–5 4.8
Fuel Rods
Geometry Cylindrical Cylindrical Cylindrical Cylindrical
Outer diameter (mm) 9.50 9.50 9.50 9.50
Active fuel height (m) 3.67 4.20 4.27 4.27
Total rod length (m) 3.88 4.55 4.65 4.80
Cladding thickness (mm) 0.57 0.57 0.56 0.57
Cladding material Zircaloy-4 M5 Zircaloy-4 Zircaloy-4
Fuel Assemblies
Fuel rods per side 17 × 17 17 × 17 17 × 17 17 × 17
Fuel rod layout Square array Square array Square array Square array
Fuel rod pitch (mm) 12.60 12.60 12.60 12.60
Fuel rod locations 289 289 289 264
Number of fuel rods 264 265 264 264
Total weight (kg) ~640 ~760 ~760 ~785
Source: Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL.
a Example: Seabrook.
One of the most noteworthy features of the EPR is that it is currently the most powerful production PWR in the world.
It can be provided by AREVA with an electrical output as high as 1,650 MWE. This is approximately 40% higher than
any of its direct competitors in the Western world today. Notice that AREVA’s EPR, the APWR, and the AP-1000 all use
uranium dioxide fuel pellets, which, except for the enrichment, are virtually identical. They all have an outer diameter of
8.19 mm, and they are all stacked in vertical arrays in cylindrical tubes. The cladding has essentially the same thickness
(about 0.57 mm), and the fuel rods all have a diameter of 9.5 mm. This means that the fuel to cladding gap in all of these
designs is 0.09 mm (which is about 1% of the total diameter of the rods). Notice that the material used for the cladding
in the EPR is a zirconium alloy called M5, while all of the other designs use Zircaloy-4.
For 17 × 17 fuel assemblies, the fuel rod pitch (P) is also the same (about 12.6 mm). Therefore, the gap between the
fuel rods where the coolant can flow is 12.6 − 9.5 = 3.1 mm, and this happens to be almost exactly one-third of the thick-
ness of the rods. The total weight of a fuel assembly is between 760 kg and 790 kg. The enrichment of the fuel in a cold,
clean core is between 4.5% and 5.0%. (The design goal for the AP-1000 is 4.8% for an equilibrium core.) Notice that the
designs of the cores have converged more than they have for the secondary loops. The containment buildings also differ
from one another, but we will discuss these separately in Chapter 32.
2.19 U
nderstanding How the Passive Cooling System
Works in the Westinghouse AP-1000
One of the unique features of the AP-1000 is that it is able to safely shut down even if all of the power to the coolant
pumps is lost. In other words, a reactor accident similar to the one that occurred in Fukushima, Japan, could never
occur with the AP-1000—even if all of auxiliary power was lost. In this section, we would like to briefly illustrate how
the AP-1000 is able to function under these conditions. Some of the design features of the AP-1000 are also shared with
the GE SBWR, which we will discuss in Chapter 3, and with AREVA’s EPR, which competes directly with the AP-1000.
Now let us return to our discussion of how the passive safety systems in the AP-1000 work. The flow of coolant through
the AP-1000 is shown in Figure 2.22. Now suppose that the connection between the plant and the power grid is lost at
2.19 HOW PASSIVE COOLING SYSTEM WORKS 61
Steam
ΞΞΞΞΞΞΞΞΞ
generator Feedwater line
IRWST
Valve (closed)
CMT
(1 of 2)
FIGURE 2.22 The flow of coolant through the Westinghouse AP-1000 when the reactor is operating normally. Notice that the
valve to the IRWST is closed at this time.
t = 0. Then the plant has no way to transmit the electrical power it is generating, and it must be shut down to prevent
its electrical generating components from overloading. This shutdown is accomplished by scramming the reactor, and
inserting the control rods back into the core. Within a minute, all modern reactors (even primitive ones) are designed
to use the standby diesel generators that are present on the site to make up for this lost power. Thus, in most scenarios,
the diesel generators automatically go online and restore the lost electrical power to the plant (so that the coolant pumps
can continue to function). However, if the diesel generators do not start (or if they are destroyed by an earthquake or a
tsunami), the coolant pumps that feed the cooling water to the core and the steam generators cannot operate. The control
system senses this problem and immediately inserts the control rods into the core (causing the reactor to scram). This
terminates the fission process and shuts the reactor down. Unfortunately, as we will discover in Chapter 5, the reactor
core continues to produce large amounts of decay heat, and this decay heat must be removed by allowing the coolant to
flow over the fuel rods. If there is no electrical power available to run the cooling pumps, the water begins to stagnate in
the core, and the core flow goes from a state of forced convection to a state of natural circulation. (The water also begins
to heat up because there is no way to remove the decay heat.) In general, the pumps require a lot of power to operate
(say 5–10 MW per pump), and this amount of power is simply not available at the reactor site after the core has been
scrammed. At the same time, the core flow begins to stagnate and the coolant pumps are also shut off to the spent fuel
pool (because they also have no power). This means that cold water is no longer pumped to the pool, and the spent fuel
in the pool continues to transfer decay heat to the water in the pool. This causes the water in the pool to heat up, and the
temperature in the pool rises as well. In most conventional designs, this would result in some very severe cooling prob-
lems. However, in the AP-1000, this additional decay heat (which can be up to 7% of the total steady-state power—see
Chapter 5) causes a separate set of passive safety systems to come online. As soon as the power to the plant is lost, the
water level in the steam generators begins to drop (because the station blackout has caused the feedwater pumps to stop).
After about 2 minutes, the steam generator water level falls to the point that it activates what is called the passive core
cooling system (or the PCCS). The PCCS consists of a heat exchanger that is located inside a large water storage tank
immediately above the reactor pressure vessel. This water storage tank is sometimes referred to as the IRWST. Because
of a difference in elevation and a difference in density between the cold water in the tank and the warm water in the core,
the core flow automatically begins to circulate between the two systems. Heat is transferred to the storage tank, and heat
is removed from the core. This process (which is shown in Figure 2.23) continues under the influence of gravity alone
until the decay heat begins to subside. After 2 hours, the decay heat subsides to about 1% of its initial value and the time
62 The Pressurized Water Reactor
PRHR
heat exchanger Vent Steam line
Pressurizer
Steam
ΞΞΞΞΞΞΞΞΞ
generator Feedwater line
IRWST
Valve (open)
CMT
(1 of 2)
FIGURE 2.23 The flow of coolant through the Westinghouse AP-1000 when the power to the pumps is lost. Notice that the valve
to the IRWST is opened and that the decay heat is then sent to the IRWST. The core continues to be cooled passively without the
need for electric power. The control rods have been inserted into the core.
0 1 minute 2 minutes 30 minutes 3 hours 5 hours 6 hours ~7 hours >7 hours 36 hours 72 hours 7 days
Pumps lose power Flow begins Steam cools, hits the Diesel power
Water in the
(plant blacks out) to the IRWST steel liner, and is restarted
IRWST boils
starts to recondense
FIGURE 2.24 The timeline for an AP-1000 reactor accident involving the loss of on-site power.
line for this is shown in Figure 2.24. After about 3 hours, the cooling water in the spent fuel pool begins to boil as well
(although this boiling is restricted to just the hotter fuel assemblies in the pool). At this point, the decay heat from the
spent fuel pool is transferred from the water to the steam. This causes some of the water in the pool to evaporate. Any
evaporated water is replaced with water from a water storage tank that is located in the adjacent cask wash down pit.
This cold water is sent to the spent fuel pool under the action of gravity alone. Therefore, a separate coolant pump is not
required to replace the water in the pool.
After about 5 hours, the PRHRS has transferred enough decay heat from the reactor core to the IRWST so that
the water inside of the tank begins to boil. This causes steam to be produced inside of the containment building. The
pressure inside of the containment building begins to rise, and the water in the IRWST continues to boil. The resulting
steam starts to condense on the steel liner in the containment building walls. (The condensed steam then flows back
2.19 HOW PASSIVE COOLING SYSTEM WORKS 63
FIGURE 2.25 A picture of the PCCS in the AP-1000 while it is operating. (Picture provided by Westinghouse Nuclear.)
into the IRWST.) After about 6 hours, the control system decides to cool the containment building using the PCCS (see
Figure 2.25). It opens up another set of valves to begin the flow of cooling water from the PCCS. The water for this
system comes from a large water storage tank that is located near the roof of the containment building, and cold water
from this tank is allowed to run over the outside of the steel liner. This cools the top and the sides of the steel liner and
helps to remove the waste heat. The process, which is illustrated in Figure 2.21, continues for some time. The steam
that is generated in the IRWST is transferred to the steel liner inside of the containment building through a set of pipes
at the top of the tank. The water flowing over the outside of the steel liner removes this decay heat through the process
of evaporation. The air flowing between the steel liner and the outside of the containment building naturally cools the
heated steel liner. In other words, the air surrounding the containment building ultimately carries away the heat. (This
is the same process that is used in natural draft reactor cooling towers, which we will discuss in Chapter 8.) Then after
about 7 hours, enough heat is removed from the steel liner so that some of the steam begins to condense back into water
inside of the containment building. This water is redirected back into the IWRST for continued use in removing decay
heat from the core. This process of heat removal continues until the core has completely cooled. After about 36 hours,
the temperatures in the core and the containment building finally stabilize, and the reactor has been successfully shut
down. The decay heat removal process does not require an external power source. At this point, the reactor’s decay heat
is slightly more than one-half of 1% of its full power output.
The ancillary diesel generators are assumed to be restarted about 3 days after the initial blackout to provide power
for postaccident monitoring. Water makeup pumps are usually employed to transfer water from the ancillary water
storage tank to the passive containment cooling water storage tank to continue cooling the containment building. These
pumps also transfer additional water to the spent fuel pool. In most blackout scenarios, external power is assumed to be
completely restored to the plant within seven calendar days. This can occur by “rebooting” the power grid, or by restart-
ing the ancillary diesel generators. At this point in time, additional water is transferred to the ancillary water storage
tank from other sources that are normally assumed to exist on the plant site. After 7 days, the reactor’s decay heat falls
to about one-third of 1% of its full power output (see Chapter 5), and the accident is assumed to be officially over. For a
64 The Pressurized Water Reactor
FIGURE 2.26 The control room of a VVER-1000. This particular picture is from the NPP Balakovo VVER-1000. As discussed
earlier, VVER-1000 PWR is the main competitor to Russia’s BWR—the RBMK. The VVER-1000 does not have a positive power
coefficient as the RBMK does. It also does not generally come with a PCCS. (Courtesy of Wikipedia.)
1,000 MWE PWR, this corresponds to about 3,300 MWT × 0.003 = 10 MW of thermal energy produced every second.
(This is enough energy to heat a reasonably large cooling pond.)
A more comprehensive discussion of the AP-1000 and its passive safety system systems can be found at the
following URL:
http://ap1000.westinghousenuclear.com/station_blackout_home/passivecorecooling.html which is maintained by
the Westinghouse Nuclear corporation. The reader is encouraged to visit this site to view a short movie of how the
AP-1000’s passive safety systems work. The cooling of the fuel assemblies in the spent fuel pool is also discussed.
Eventually, all reactors will be equipped with some sort of passive heat removal system. Many articles that describe
how these systems work can be found on the Internet. The reader can find some excellent articles on these systems by
simply using the Google Search Engine. The Russian alternative to the Westinghouse AP-1000, which is also shown in
Figure 2.26, is an older design that is not always equipped with a PCCS.
* See “An Introduction to Nuclear Reactor Physics” by R.E. Masterson, CRC Press (2017).
2.20 BORIC ACID IN PWRs 65
1500
1000
1 year
500
0
0 100 200 300 400
Operating time (days)
FIGURE 2.27 A typical soluble boron curve for a PWR versus operating time. Start-up occurs at t = 0.
The boric acid used to create this Solbor can exist in the form of colorless crystals or a white powder which easily
dissolves in ordinary water. Solbor has the chemical formula H3BO3, which is also sometimes written as B(OH)3. The
Boron in Equation 2.1 can have several different isotopes, and the two most prevalent of these are Boron-10 and Boron-
11. About 20% of natural boron is Boron-10, and most of the remaining Boron is Boron-11. Boron-10 has a much larger
neutron absorption cross section than Boron-11, and it is much better at absorbing low-energy neutrons. Hence, the
Boron-10 helps to absorb many of the low energy neutrons in the core, and this provides an additional way to reduce
the power level when the demand for electricity becomes lower. Ideally, a reactor operator attempts to add just enough
boric acid to the coolant to keep the core critical at all times. Then as the nuclear fuel is burned, the concentration of
boron can be reduced proportionally to maintain a critical system. Changes to the boric acid concentration can be used
to effectively regulate the rate at which the fission process occurs. The control rods can then be reserved for their tra-
ditional functions of start-up, shutdown, and emergency core control. In addition to boric acid, soluble poisons such as
gadolinium nitrate (Gd(NO3)3) and sodium polyborate can be directly injected into the coolant system if an emergency
situation occurs. These neutron poisons are very effective in absorbing neutrons when a reactor needs to be shut down
quickly. They can be injected automatically into the coolant during an emergency, or they can be injected manually if
the emergency injection system fails.
When a PWR is first started up, it is a standard operating procedure (or SOP) to maximize the concentration of the
boric acid in the coolant to compensate for the excess reactivity that the fresh fuel may have. Then as the fuel burns up
and begins to become depleted, the excess reactivity is reduced, and the concentration of soluble boron required can be
reduced in direct proportion as well. In a modern PWR, the moderator temperature coefficient (which is discussed in
our companion book) can become POSITIVE when the boric acid concentration becomes too high. For this reason,
the boric acid concentration in a commercial PWR is normally limited to about 1,500 PPM for a fresh core. Then as
the fuel is burned, the soluble boron concentration falls, and the moderator temperature coefficient becomes nega-
tive again. If the power level remains the same, the Solbor concentration will decrease almost linearly with burnup to
keep the reactor critical, (see Figure 2.27) until it becomes time for the fuel to be replaced. The only exception to this
rule is when the boric acid concentration must be decreased rather quickly after Start-up to compensate for the buildup
of xenon in the core, which can reduce the reactivity by a couple of dollars by absorbing so many thermal neutrons.
However, this behavior only persists for about 2 days when a PWR is operating at full power. After that, the xenon will
build up to its equilibrium concentration, X∞, and the concentration of soluble boron does not have to be reduced any
further to compensate for the reactivity loss caused by the neutron-absorbing potential of the xenon (see Figure 2.27).
A typical cycle for the concentration of boric acid in a commercial PWR is shown in Figure 2.27 as a function of time
for a reactor that has been operating at a constant power level P. Notice that once the xenon and samarium c oncentrations
have built up to their equilibrium values, the boric acid concentration only needs to be changed to compensate for the
depletion and the burnup of the fuel. If the power level is held constant, then the burnup of the fuel varies linearly with
time, and the boric acid concentration falls inversely with time as well. If the power level is doubled, then twice as much
xenon will be produced, and the amount of boron in the coolant must also be reduced to half in the first 48 hours to
66 The Pressurized Water Reactor
compensate for the large number of neutrons that the xenon absorbs. Eventually, as things return to normal, the Solbor
concentration again returns to its long-term linear trend.
Bibliography
Books and Textbooks
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley & Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc. (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. An Introduction to Nuclear Reactor Physics, First edition, CRC Press, Boca Raton, FL (2017).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2012).
Tong, L.S. and Weisman, J. Thermal Analysis of Pressurized Water Reactors, Third edition, American Nuclear Society, La Grange
Park, IL (1996).
Tong, L.S. Boiling crisis and the critical heat flux, US Atomic Energy Commission Report TID-25887 (1972).
Web References
http://en.wikipedia.org/wiki/Pressurized_water_reactor.
http://www-pub.iaea.org/MTCD/publications/PDF/te_1120_prn.pdf.
http://ap1000.westinghousenuclear.com/station_blackout_home/passivecorecooling.html.
http://www.nrc.gov/reactors/pwrs.htmlhttp://en.wikipedia.org/wiki/List_of_PWR_reactors.
http://uk.areva.com/EN/home-668/how-does-a-pwr-work-description-and-explanation--areva-uk.html.
http://us.areva.com/EN/home-933/us-epr-reactor-generation-iii-nuclear-reactor-solution-for-united-states.html.
Exercise 2.2
Suppose one of the coolant pumps in the hot leg of a PWR is able to pump 6,000 kg of water per second through the core.
What is the velocity of the water in m/s if one of the pipes in the hot leg has a cross-sectional area of 1 m2?
Exercise 2.3
The velocity of the water flowing over the fuel rods in a modern PWR is about 5 m/s. How long does it take for a water
molecule to completely flow through the core?
Exercise 2.4
The coolant in one of the hot legs of a modern PWR has a mass flow rate about 12.5 times higher than the mass flow rate
in the steam generator on the secondary side. If the mass flow rate in the hot leg is 5,000 kg/s, what are the mass flow
rate and the average enthalpy change in the secondary loop? What do you think accounts for such a large change in the
coolant enthalpy in this case?
Exercise 2.5
The enthalpy of vaporization of the water in the secondary loop of a PWR that operates at 1,000 PSI (6.9 MPa) is about
1,512 kJ/kg. If the mass flow rate on the secondary side of one of the steam generators is 480 kg/s, what is the energy
transfer rate between the primary side of the steam generator and the secondary side?
Exercise 2.6
A PWR built in the 1970s has a thermal efficiency of 32%. The reactor vendor proposes to raise the thermal efficiency
of the PWR to 34% by modifying the heat and fluid flows in the secondary loop. We will find that this is possible using
a combination of two techniques called reheat and regeneration that are then discussed in Chapter 9. If the thermal effi-
ciency is increased by 2%, and the previous electrical power output of the plant was 1,000 MWE, what do you expect
the electrical power output to be after the required modifications are made to the secondary loop? Is the thermal power
output of the reactor core changed in this process?
Exercise 2.7
Suppose that the cost of the modifications to the secondary loop in Exercise 2.6 was $100 million. If the revenue from
the nuclear power plant is increased by half a million dollars a day when the plant is operating normally, and the average
time between refuelings is 18 months, how long will it take the utility to recover its original investment in retrofitting
the power plant? Assume that the modifications can be done while the power plant is shut down for its normal refueling
cycle. Do you think that this last assumption is a realistic one?
Exercise 2.8
In a PWR core, one in every four fuel assemblies has a control rod in it, and in these special fuel assemblies, about one
in every ten rods is a control rod. If the MFR in the core (the ratio of the volume of the moderator to the volume of the
fuel) is 4:1, estimate how many control rods there are in a large PWR core.
Exercise 2.9
Suppose that someone comes to you and tells you that the pitch of the fuel rods in a PWR core can be reduced from
12.6 to 12 mm, and that the additional heat that is generated in the core can be removed by simply pumping the coolant
through the core more quickly. How much is the volumetric power density increased in this case, and how much faster
must the coolant be pumped through the core to remove the additional heat? Assume that the outer diameter of the fuel
rods is 9.6 mm.
Exercise 2.10
In recent years, most PWRs have gone to an 18-month or a 2-year refueling cycle by increasing the average enrichment
of the fuel assemblies in the core. For each of these refueling cycles, what do you expect the initial soluble boron con-
centration in the coolant to be? Assume the reactivity is a linear function of the enrichment.
3
The Boiling Water Reactor
3.1 Boiling Water Reactors
Boiling water reactors (BWRs) are the second most popular type of reactor in the world today. A BWR is fundamentally different
than a pressurized water reactor (PWR) because the coolant in the core is intentionally designed to boil. BWRs account for nearly
22% of the world’s currently installed nuclear generating capacity. The primary manufacturer of BWRs today is the General Electric
Company (GE), which is based in the United States. Just as in a PWR, the water in a BWR is of the ordinary or “light” variety, and it
enters the core from below. However, it is intentionally designed to boil because the operating pressure is about half as high as it is in
a PWR (about 7.15 MPa or 1,050 PSI vs. 15.5 MPa or 2,250 PSI for a PWR). Consequently, its saturation temperature is lower (about
286°C) and the pressure vessel does not have to be quite as thick (see Chapter 1). Water enters the core at about 275°C and leaves the
core at about 286°C or 545°F. By this time, about 15% of the water (by weight) has turned into steam. The percentage of the mass of
the water that has turned to steam is sometimes called the quality of the water–steam mixture. The steam is dried using a set of steam
separators near the top of the pressure vessel, and then, it is sent to the power turbines.
When BWRs were first considered as an alternative to PWRs in the early 1950s, it was believed that hydrodynamic instabilities
would develop inside of the core if the water was allowed to boil. This would then lead to neutronic instabilities and power oscillations
that would make BWRs difficult to control. Fortunately, a number of experiments that were conducted in the early 1950s (including
the famous BORAX experiments) showed that these hydrodynamic instabilities only developed when the water was allowed to boil
at very low pressures (less than about 800 PSI or 5.5 MPa). At higher pressures, the boiling became stable, and the reactor was much
easier to control. Hence, the design pressure for modern BWRs was set at about 1,050 PSI (7.15 MPa), and this has been their normal
operating pressure ever since. The BWR has now attained about one-quarter of the total world market for commercial nuclear power
plants. Other types of reactors (such as the Canadian CANada Deuterium Uranium [CANDU] reactors, the British advanced gas
reactors [AGRs], and the liquid metal fast breeder reactor [LMFBR]) make up the remaining 10% of the world’s generating capacity.
There are several reasons why BWRs have continued to remain popular over the years. The first and most obvious one is that the
water flowing through the core is converted to steam inside of the reactor pressure vessel, and so a secondary loop with a number
of expensive steam generators is not required for a BWR. This saves a great deal of money and also the expense of building the sec-
ondary loop’s piping. BWRs employ a version of the Rankine thermal cycle called a direct cycle. There is no additional equipment
(except for the steam separators at the top of the pressure vessel) to power the steam turbines. Another interesting feature of BWRs is
the energy content of the steam. When water is turned to steam, more heat can be absorbed by the working fluid, and this means that
for a given power level, a BWR requires a lower flow rate than a PWR does to remove the heat. Because of this, less water must be
pumped through the core each second for the same power rating. This increases the thermal efficiency of the plant because the pumps
in a BWR require less power to move the water than the pumps in a comparable-sized PWR.
Unfortunately, because the water can become radioactive as it passes through the core, and it eventually passes through all of the
electricity-producing components in the plant (the turbines, the pumps, the condenser, and the pipes), more shielding is required in a BWR
than in a comparable-sized PWR (where only the primary loop must be heavily shielded). Because the operating pressure in a BWR is
lower (about 7.15 MPa or 1,050 PSI), the pressure vessel does not have to be as thick, and the power density required for the coolant to boil
(in W/m3) is only about half of what it is in a PWR (see Figure 3.20). This means that the overall dimensions of the reactor pressure vessel
for a BWR must be larger than they are for a PWR for the same power output (see Figure 3.1). As far as the cost of the pressure vessel is
concerned, these two effects tend to cancel and the costs of a BWR pressure vessel and a PWR pressure vessel are essentially the same.
A cross-sectional view of the pressure vessel used in a typical BWR is shown in Figure 3.1. The coolant flows through the pres-
sure vessel along the directions indicated by the arrows in Figure 3.3. The driving force for the coolant is the recirculation pumps.
There are typically several of these pumps in a commercial BWR. The coolant flows through an annular region along the outside of
the pressure vessel known as the downcomer. It then flows into the lower part of the reactor vessel, which is called the lower plenum,
where it subsequently flows upward through the core. Between 0.5 m and 1 m into the core, it starts to boil, and when it exits the top
of the core, the coolant has absorbed enough energy to convert between 15% and 17% of its resulting mass into steam. The steam at
69
70 The Boiling Water Reactor
Steam dryer
Steam outlet nozzle
Support flange
Steam separator
Reactor pressure vessel
Feedwater sparger
Feedwater inlet nozzle
Pump impeller
FIGURE 3.1 A cross-sectional view of a modern BWR core and pressure vessel. (Courtesy of GE.)
this point is very “wet,” and so it must pass through a set of steam separators to remove most of the moisture. The steam
then flows through a steam drier, which removes the remaining water and reduces its moisture content to about 1%. The
steam then exits the reactor vessel and is sent to the power turbines. The steam turns the blades of the power turbines,
and this is how the electricity is generated within the plant. Further moisture separation occurs before the steam enters
the steam turbines, and so the steam that impinges on the blades of the turbines is essentially “dry.”
(a)
(b)
FIGURE 3.2 A modern BWR control room (a) and a BWR control room simulator (b). (Both pictures are public domain pictures
taken from Google Images.)
The next generation of BWRs is expected to be called the BWR-7. The BWR-7 employs several passive safety systems
that earlier designs did not have, and the containment building is also considerably stronger. It resembles the contain-
ment buildings used in PWRs (see Chapter 32) in many respects. Some of the passive safety systems used in the BWR-7
are also similar to those used in the Westinghouse AP-1000, which we discussed in Chapter 2. The water level in the
pressure vessel comes up to about the middle of the steam separators. Above this, the pressure vessel is filled with
steam, and below this, the pressure vessel is filled with water. The top of the pressure vessel in a BWR is also called
the steam dome. The steam lines exit the top of the steam dome and sends the heated steam to the power turbines. A
BWR produces saturated steam at about 286°C and at about 1,050 PSI (7.15 MPa). This gives a modern BWR a thermal
efficiency between 33% and 34%. However, because of the lower operating pressure and the lower power density, the
enrichment of the fuel in the core can also be lower.
The fuel rods are more or less identical, except that BWRs use Zircaloy-2 instead of Zircaloy-4 for the cladding,
and the outer diameter of the fuel rods is about 11.2 mm, compared to about 9.5 mm for a PWR. The fuel pellets are
somewhat larger, and the cladding is somewhat thicker as well. One final difference between the design of a BWR and
a PWR is that the control rods in a BWR enter the core from below rather than from above. The reason for this is that
the steam driers and the moisture separators at the top of the pressure vessel simply take up too much space to allow
the control rod drive mechanisms to be placed there. Hence, a BWR must be scrammed from below rather than from
above (see Figure 3.10). The power level can also be reduced by simply reducing the amount of electrical power to the
pumps. This allows more voids (or bubbles) to form in the coolant, and this reduces the neutron production rate. The
reduction in the power level is then governed by the void coefficient of the reactivity, which is discussed in detail in our
companion book.*
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, First edition, CRC Press (2017).
72 The Boiling Water Reactor
Feedwater
Steam separators
Annulus
Chimney
Saturated steam
Fuel assemblies
Saturated water
Subcooled water
Core
Lower orifice
FIGURE 3.3 The global water and steam profile in a typical BWR. (Picture provided by the author.)
The water in a BWR begins to develop vapor bubbles (e.g., voids) about one-fifth of the way up the core, and by the time
it reaches the core midplane, it begins to boil vigorously (see Figures 3.3 and 3.4). This type of boiling is sometimes
called bulk boiling because the boiling becomes relatively pronounced. The exact point at which this boiling occurs
depends on the power distribution in the core. For a co-sinusoidal-shaped power distribution, the coolant starts to boil
about 1 m from the core inlet (where z ≈ 30 in.). Because a PWR operates at a much higher pressure than a BWR, the
bubbles in a PWR only form in the hot channel about 1/2 m from the top of the core. Otherwise, the water remains
a single-phase liquid under most conditions. The bubble profiles in each case are shown in Figure 3.4. In a BWR, the
average quality at the core outlet (the percentage of the mass of the water that is converted into vapor) is about 15% and
the average void fraction (the percentage of the volume of the coolant channels that is occupied by the vapor phase) is
about 80% at the same location (see Figure 3.12). By comparison, the quality in the hot channel of a PWR rarely exceeds
1% and the void fraction rarely exceeds 5% at the same axial elevation.
Core exit
Core inlet
FIGURE 3.4 The bubble profiles in the core of a modern BWR and a modern PWR.
Fuel rods
Control rod
Instrumentation
channel
Fuel assembly
“can”
FIGURE 3.5 Four 8 × 8 BWR fuel assemblies surrounding a cruciform-shaped control rod. The colors inside of the fuel
rods correspond to different average enrichment levels. Red, brown, and purple are higher enrichment levels than yellow.
The blue hole in the center of each fuel assembly is an instrumentation channel. (Picture drawn by the author.)
mechanisms. In these reactors, the control rod drive mechanisms are replaced by steam separators and steam
driers instead.
☉☉ Third, each BWR fuel assembly is enclosed in a square Zircaloy or stainless steel “can” that directs the flow of
coolant water through the fuel assembly during the time it moves through the core. As the water moves up through
the core, it begins to boil about 1 m from the inlet of the fuel assembly, although bulk boiling does not occur until
the water reaches approximately the core midplane. BWR fuel assemblies contain larger water channels than
PWR fuel assemblies do, and these channels are designed to provide appropriate neutron moderation as well.
74 The Boiling Water Reactor
In some designs, the Zircaloy tubes are replaced with water channels to increase the amount of moderator in the central
region of the fuel assembly. Different enrichment levels are then used in the fuel rods to create a uniform radial power
profile. Normally, lower enrichments are used in the outer fuel rods, and higher enrichments are used in the inner ones.
A BWR fuel assembly is designed to operate with an exit flow quality of 15%–17%, and so less neutron moderation
will occur at the top of the core than at the bottom. (The moderating power of the coolant is directly proportional to the
density of the coolant at a particular spatial location.)
* See Nuclear Engineering Fundamentals: A Practical Perspective by R.E. Masterson, First edition, CRC Press (2017).
3.5 RUSSIAN BWRs (RBMKs) 75
BWR/6 fuel
assemblies
and control
rod module
improve the safety during certain types of transients. As opposed to a CANDU reactor (see Chapter 2), a reactor new
fuel assembly can stay in the core for periods of up to about 6 years before it has to be replaced. This is said to improve
the long-term availability of the plant.
The higher enrichment (2.8% vs. 0.7% for a CANDU) is one of the primary reasons why the fuel assemblies can
remain in the core for so long. All RBMK reactors now use recycled uranium from the Russian VVER. As with other
reactor designs that employ pressure tubes, the RBMK reactor is capable of online refueling. A picture of a fuel assembly
from a large RBMK reactor (e.g., the RBMK-1500) is shown in Figure 3.8. Some of the design parameters of the RBMK-
1500 are also shown in Table 3.1. An RBMK-1500 produces 4,800 MW of thermal energy and about 1,500 MWE. The
overall thermal efficiency is 31.25% (efficiency = 1,500/4,800). This is slightly lower than the thermal efficiency of a
Western BWR. The steam pressure (about 7 MPa), and the core inlet and outlet temperatures (260°C and 284°C) of an
RBMK-1500 are similar to those of a Western BWR. The largest RBMKs are the 1,000 MWE RBMK-1000 and the
1,500 MWE RBMK-1500. Both were designed by the USSR’s Ministry of Nuclear Power.
FIGURE 3.7 The exposed core of a large RBMK. (Courtesy of the World Nuclear Association.)
Fuel rods
Steam drums
(to steam
separation)
Graphite
moderator
Fuel elements
Water
Water Pressure tubes
FIGURE 3.8 The RBMK reactor is a boiling water, graphite-moderated thermal water reactor, which uses pressure tubes to
cool the core. The fuel is contained in about 1,700 pressure tubes that are mounted vertically in blocks of graphite. Cooling
water passes through the tubes and is heated until it boils. The steam is then routed to a number of steam turbines, where
electrical power is produced. The original RBMKs did not meet international safety standards and deficiencies were known
to exist in the instrumentation and control systems, the fire protection system, and the emergency core cooling system. The
original design also did not meet international standards for the containment building. The RBMK reactor shown here is
similar to the one that exploded at Chernobyl. Other large RBMK reactors can be found in Leningrad, Ignalina, Kursk,
and Smolensk. (Courtesy of the World Nuclear Association.)
TABLE 3.1
RBMK 1500 Design Parameters
Average thermal power 4,800 MWT
Average electrical output 1,600 MWE
Average enrichment of the fuel 2.0%–2.2%
Type of coolant Light water
Core inlet temperature 260°C
Core outlet temperature 285°C
Water pressure in the core 7.0 MPa
Core height 7.0 m
Core diameter 11.8 m
Source: Lamarsh, J.R. Introduction to Nuclear Reactor Theory,
Second printing, Addison-Wesley Publishing Company,
Inc., Reading, MA (1972).
0°
Fuel bundle
Peripheral bundle
Control bundle
Control blade
270° 90°
Pressure vessel
Core shroud
180°
FIGURE 3.9 The layout of a typical BWR core. Note: Each small circle in the figure corresponds to an 8 × 8 fuel assembly.
A large BWR core may have as many as 800 of these fuel assemblies.
turned to steam is sometimes called the quality of the water–steam mixture. The steam is then dried using a set of steam
separators near the top of the pressure vessel, and it is then sent to the power turbines. The exact point at which the
coolant begins to boil depends on the axial power distribution in the core. For a co-sinusoidal-shaped power distribution,
the coolant begins to boil about 1 m from the core inlet (at z ≈ 30 in.). From this point upward, the coolant temperature
stays at the saturation temperature (the boiling point) and does not rise further. For a BWR during normal operation,
the saturation temperature is about 286°C (or 545°F). The fuel and cladding temperatures follow a similar trend to those
found in a PWR, except that the coolant temperature stays constant (after it starts to boil), and because of this, the fuel
and cladding temperatures are limited to much lower values. The power density in the core of a BWR is also about half
of the power density in a PWR core. These lower power densities and coolant pressures cause the boiling regimes (nucle-
ate boiling) versus bulk boiling and film boiling to be different around the fuel rods as well. The differences in these
boiling regimes are discussed in many nuclear thermal-hydraulics books (see, for example, Todreas and Kazimi, 2012),
and we will discuss these differences at great length in Chapters 23 and 27.
78 The Boiling Water Reactor
TABLE 3.2
BWR Design Parameters
φ(z)
Top of core H
H
0.5 H
Bottom of core 0
z=0
z = 0.4 H
z = 0.6 H
Control
rods
FIGURE 3.10 The effect of the control rods in a BWR on the shape of the neutron flux (and hence the axial power profile).
Compared to a PWR, the shape of the axial power profile is different because the control rods are inserted into the core from below
rather than from above.
FIGURE 3.11 A picture of the SBWR. The SBWR is designed to safely cool itself with no AC power or operation action for more
than 7 days, and using natural circulation, it also has 25% fewer coolant pumps and mechanical drive systems than previous versions
of the BWR. (Courtesy of General Electric.)
In an ABWR, the control rods are also adjusted automatically and in real time to more easily control the power level
during normal day-to-day operation. The steam separators at the top of the pressure vessel are redesigned to send
somewhat drier steam to the steam turbines after it leaves the core. However, the ABWR does not implement passive
safety systems to the same extent that the newer and more advanced designs, such as the simplified BWR (SBWR), do
(see Figure 3.11).
TABLE 3.3
BWR Fuel Design Parameters
Common BWR Fuel Pin, Fuel Rod, and Fuel Assembly Design Parameters
Type of BWR U.S. BWRa ABWR SBWR
Manufacturer General Electric General Electric General Electric
Fuel Pins
Geometry Cylindrical Cylindrical Cylindrical
Diameter 9.60 mm 8.76 mm 8.76 mm
Composition UO2 UO2 UO2
Enrichment (avg.) 3.5% 4%–5% 4%–5%
Fuel Rods
Geometry Cylindrical Cylindrical Cylindrical
Outer diameter 11.20 mm 10.26 mm 10.26 mm
Active fuel height 3.58 m 3.70 m 3.05 m
Total rod length 4.09 m 4.47 m 3.79 m
Cladding thickness 0.71 mm 0.66 mm 0.66 mm
Cladding material Zircaloy-2 Zircaloy-2 Zircaloy-2
Fuel Assemblies
Fuel rods per side 9×9 10 × 10 10 × 10
Fuel rod layout Square array Square array Square array
Fuel rod pitch 14.37 mm 12.95 mm 12.95 mm
Fuel rod locations 81 100 100
Number of fuel rods 74 92 92
Total weight ~273 kg ~183 kg (UO2 only) ~144 kg (UO2 only)
a Example: New Mexico Power-2 (a BWR-5).
3.12 Finding the Density from the Void Fraction 81
thickness (about 0.71 mm for a conventional BWR and about 0.66 mm for the ABWR and the SBWR), and the fuel rods
have diameters between 11.20 mm (for a conventional BWR) and 10.26 mm (for the ABWR and the SBWR). This means
that the fuel to cladding gap in all of these designs is 0.09 mm (which is about 1% of the total diameter of the rods).
Notice that the material used for the cladding is Zircaloy-2, which is better suited for use in BWRs than Zircaloy-4 is.
Standard BWRs have 9 × 9 fuel assemblies with an average pitch of 14.35 mm, whereas the ABWR and the SBWR have
10 × 10 fuel assemblies with an average pitch of 12.95 mm. Therefore, the gap between the fuel rods where the coolant
can flow is 14.35 mm − 11.20 mm = 3.15 mm for a standard BWR, and this happens to be about 28% of the thickness of
the rods. The total weight of a conventional BWR fuel assembly is about 275 kg. This is about 40% of the weight of a
PWR fuel assembly. BWR fuel assemblies weigh less than PWR fuel assemblies do because they have fewer numbers of
fuel rods. The enrichment of the fuel in a cold, clean core is less than it is in a PWR.
In conventional BWRs, the average enrichment of the fuel in an equilibrium core is between 3.0% and 3.5%. In
the ABWR and the SBWR, the average enrichments are considerably higher (4%–5%) and mirror those of PWRs in
many respects. The fuel pellets in a BWR are also longer than they are in a PWR. These design differences are high-
lighted in Table 3.3. Notice that the design of most BWR cores is nearly the same, and the only major difference is
the number of fuel rods and the enrichment of the fuel. The containment buildings also differ from one another, and
these differences will be discussed separately in Chapter 32. BWRs have three types of containment buildings: the
Mark-I, the Mark-II, and the Mark-III containments. The Mark-III is the newest model, and it has the most similari-
ties to the containment buildings that are used for PWRs. All BWR fuel assemblies are surrounded by a metal “can”
that is sometimes referred to as a “sheath” or a “shroud.” The shroud helps to stabilize the hydrodynamic behavior of
the core and prevents voids that form in one fuel assembly from propagating into another. Unfortunately, the shroud
absorbs some of the thermal neutrons that would otherwise be available for the fuel, and so the number of thermal
neutrons that are available in a BWR is slightly less than it is in a PWR. Due to differences in the enrichment of the
fuel and the neutron flux, the average power density (in kW/L) is about 50% lower. The power density of a modern
BWR is compared to that of a modern PWR in Figure 3.20, which can be found in Section 3.16. The power density
in a BWR fuel assembly averages about 50 kW/L today. This number will probably increase as more modern designs
come online.
100
Atmospheric pressure
80 3.6 MPa
7.1 MPa
40 14.2 MPa
0
0 2 4 6 8 10 12 14 16 18 20
Mixture quality, x (%)
FIGURE 3.12 How the void fraction and the quality are related when the liquid and the vapor are moving at the same speed.
(From El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).)
0.8 0.8
Void fraction, α
Quality, x
0.4 0.4
0.0 0.0
FIGURE 3.13 The value of the void fraction α and the quality x in a typical BWR as a function of axial elevation. Note: The
values may differ somewhat from one fuel assembly to the next. (From El-Wakil, M.M. Nuclear Heat Transport, American Nuclear
Society, La Grange Park, IL (1981).)
two-phase mixture is determined by the percentage of the mixture that is liquid and the percentage of the mixture that
is vapor. Using the following definition for void fraction α
TABLE 3.4
The Mixture Density as a Function of the Void Fraction in a BWR Fuel Assembly
Void Fraction (α) Liquid Density (g/cm3) Vapor Density (g/cm3) Mixture Density (g/cm3)
0 0.740 0.036 0.740
0.5 0.740 0.036 0.388
0.8 0.740 0.036 0.177
1 0.740 0.036 0.036
It is easy to see that the density of the steam–water mixture ρ(z) at any elevation in a BWR core can be found from the
following equation:
where
ρl is the density of the liquid water.
ρv is the density of the vapor (which in this case happens to be the steam).
α(z) is the void fraction at that specific location.
Representative values of the mixture density are shown in Table 3.4. Of course, this equation only applies to a fluid field
where the liquid and vapor phases are moving at the same speed (i.e., it can be used when vl ≈ vv). This is equivalent
to assuming that the liquid and vapor phases have no slip. Normally, the density of the liquid and vapor phases at the
saturation point in a BWR core is a function of the system pressure p. In a BWR fuel assembly, this system pressure
happens to be 1,050 PSI or 7.15 MPa. At this pressure, the density of the water is ρl = 0.74 g/cm3, and the density of the
vapor (i.e., the steam) is ρv = 0.036 g/cm3. In other words, if we know the void fraction α(z), we can find the density of the
two-phase mixture ρ from Equation 3.2 at any axial elevation z (assuming thermodynamic equilibrium). For example,
at the exit of the core (z = H), the void fraction happens to be about 80% (see Figure 3.13), so the density of the mixture
at this specific elevation happens to be
Applying the values for ρl and ρv mentioned previously, the density of the two-phase mixture at this point is ρ = 0.8 ×
0.036 + 0.2 × 0.74 = 0.177 g/cm3. Conversely, at the core midplane (z = H/2), where the void fraction is about 65%, the
average density of the two-phase mixture is
or ρ = 0.65 × 0.036 + 0.35 × 0.74 = 0.282 g/cm3. The coolant densities between the core exit and the core midpoint fall
somewhere between these two values. However, the relationship between the void fraction and the axial elevation is not
always a linear one, as the reader can clearly see in Figure 3.13. Once we know the void fraction, we can also determine
the flow quality x in a relatively simple way. We would now like to discuss how the value of the flow quality in a BWR
coolant channel can be found.
then the values of the void fraction and the quality are related to each other by the following equation:
1
α= (3.7)
( )( )
1 + (1 − x ) x ρv ρ1
where
ρl is the density of the liquid phase.
ρv is the density of the vapor phase.
This equation can be used when the slip ratio S between the liquid and vapor phases is 1.0 (i.e., when they are moving
at the same speed). The slip ratio S is formally defined by
where
vl is the velocity of the liquid.
vv is the velocity of the vapor (or in the case of a BWR, the velocity of the steam).
Normally, the slip ratio has a value greater than unity because the vapor (i.e., the steam) always moves faster than the
liquid phase does. The slip ratio in a BWR fuel assembly usually varies from 1.0 to about 4.0. However, in newer designs,
it can be higher). When the slip ratio is greater than 1.0 (which is usually true near the top of the core), the equation
relating the void fraction to the quality is
1
α(x) = (3.9)
( )( )
1 + (1 − x ) x ρv ρ1 S
1
x(α ) = (3.10)
( )( )( )
1 + (1 − α) α ⋅ ρ1 ρv ⋅ 1 S
In other words, Equation 3.10 can be used to determine the flow quality as a function of the void fraction as well. Notice
that the fluid quality always increases as the slip ratio increases. Hence, a slip ratio of 2.0 will always give a higher
quality at a specific location in a BWR core than a slip ratio of 1.0. The value of S also varies from one flow regime to
the next, and in a BWR, several different flow regimes may coexist at the same time at different axial elevations in the
same reactor coolant channel. (The exact flow regimes and their location depend on the flow rate and the shape of the
axial power profile.) The dependence of the quality on the void fraction is shown in Table 3.5 for some representative
values of the slip ratio S. The degree of sensitivity becomes greater as the void fraction is increased. Notice that the void
fraction, the slip ratio, and the quality are all dimensionless numbers. In general, these numbers must be used to make
an “apples-to-apples” comparison between the behaviors of the phases in a multiphase flow field.
TABLE 3.5
The Dependence of the Flow Quality on the Slip Ratio in a BWR Core
Note: At 7.15 MPa, the water in the core has a density of ρl = 0.74 g/cm3 and the
steam in the core has a density of ρv = 0.036 g/cm3.
volume of the moderator in a BWR fuel assembly divided by the volume of the fuel. Using the design parameters given
in Table 3.3, find the MFR for a modern 9 × 9 BWR fuel assembly.
Solution It is easy to show that the MFR for a square reactor lattice is given by MFR = (4/π)(P/D)2 − 1, where P is the
pitch of the lattice and D is the diameter of the fuel pins. Since P = 14.37 mm and D = 9.6 mm for a typical U.S. BWR,
P/D = 1.49 and we find that MFR ≈ 1.83. In a commercial PWR, this number is closer to 2.0 and because there is no
boiling, the core is somewhat easier to control. [Ans.]
Top of core
Annular flow
with entrainment
Forced convective
heat transfer
through liquid film
Annular flow
Bubbly flow
Subcooled boiling
Single-phase Convective
liquid heat transfer
to liquid
Bottom of core
FIGURE 3.14 An example of some of the common flow regimes in a typical BWR fuel assembly. Note that the flow regimes
depend on both the mass flow rate and the shape of the axial heat flux. (Adapted from a picture in http://me1065.wikidot.com/
fuel-assemblies-in-nuclear-reactors.)
A typical temperature profile in a BWR fuel rod is shown in Figure 3.15, and examples of how the temperature of
the coolant and the void fraction change as a function of axial elevation are shown in Figures 3.16 and 3.17. In practice,
the temperature profile has both radial and axial components. However, most of the time, the radial temperature profile
changes much more rapidly than the axial one. This effectively limits the value of the void fraction and the quality at
the top of the core to one that prevents the fuel and the cladding from ever melting. A similar but related process occurs
in PWRs, but this process is known as nucleate boiling. The dryout of the liquid film surrounding a PWR fuel rod is
then caused by a departure from nucleate boiling, or a DNB condition. There are a number of ways in which the DNB
can be predicted in a PWR and where the dryout of the liquid film can be predicted in a BWR. Each of these techniques
is discussed in considerable detail in Chapter 27, where the departure from nucleate boiling ratio (the DNBR) and the
critical power ratio (the CPR) are defined. The DNBR is used to predict the location of dryout in PWRs and the CPR is
used to predict the location of dryout in BWRs.
3000°C
Fuel melts (2800°C)
2500°C
500°C
Coolant boils (286°C)
0°C
Coolant flow
Fuel Gap
Cladding
FIGURE 3.15 The radial temperature profiles in a BWR fuel rod. Note that the maximum fuel pin centerline temperature is about
1,700°C and the surface temperature of the cladding is about 286°C under normal operating conditions.
TSAT = 286°C
Voids
Heat
Boiling
Heat
No
boiling
TIN >> 278°C
Coolant flow
FIGURE 3.16 An illustration of how the temperature of the coolant and the void fraction changes in a BWR coolant channel.
88 The Boiling Water Reactor
TSAT = 286°C
T
HB αV
α
HNB αT
TIN ≈ 278°C
FIGURE 3.17 An illustration of the boiling and nonboiling heights in a BWR coolant channel. Before the water starts to boil
(in the lower 25% of the core), the moderator coefficient of the reactivity αM can be used to find the reactivity change, and in the
upper 75% of the core, the void coefficient of the reactivity αV must be used to find the reactivity change.
120
110
Legend (100% of power)
100
A typical BWR 30
operating map 20 Pump cavitation
10
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Percent of rated core flow
FIGURE 3.18 A typical BWR operating map. The regions enclosed by the red lines (regions 1 and 2) indicate the power and
flow ratios that are not allowed during normal operating conditions. (From Instability in BWR NPPs—F. Maggini 2004.
Also see http://blog.metasd.com/2011/03/page/2/.)
where a BWR operates stably and reliably. All reactor operators are required to operate the plant within the limits
defined by these operating maps. Outside of these limits, the thermal-hydraulic behavior of the core becomes less
stable. However, it may still be safe in spite of the thermal-hydraulic i nstabilities that may develop when it is operated
outside of the envelope defined by the operating map. Sometimes, these thermal-hydraulic instabilities can take the
form of flow oscillations and density waves. These instabilities were first observed in BWRs in the early 1980s. Since
then, the control systems of many BWRs have been redesigned to prevent them from occurring when the core flow rate
becomes low. A very interesting discussion of these hydrodynamic instabilities is presented in Chapter 31.
FIGURE 3.19 An operating map taken from the Taiwan Power Company’s ABWR Lungmen Project. The red region in the picture
(region 3) is a region in which large hydrodynamic instabilities may occur. When the reactor is operated in region 2, the reverse
problem occurs because the steam is too wet to be completely dried by the steam separators, and this may damage the turbine
blades. (From http://www.microsimtech.com/ Startup/ABWR.html.)
to take this into account when modifying the flow rate to prepare for refueling. This corresponds to the blue line shown
in Figure 3.18. On the other hand, the flow to the core can also be turned off very quickly. (One simply has to turn off
the coolant pumps for this to happen.) When this occurs, the rated core flow can decline very quickly, and within a few
minutes, the bulk coolant flow can become essentially stagnant. This corresponds to the yellow and brown lines shown
in Figure 3.18. If one is crazy enough to turn off the coolant pumps without scramming the core, the reactor will end up
in the state represented by the yellow line in Figure 3.18. This operating state would be a major disaster in the making.
However, if the reactor has already been shut down (see the blue line in Figure 3.18), then a similar scenario can occur—
but it just takes more time for it to develop. If the flow rate is reduced too quickly, the reactor coolant temperature may
rise above 286°C (~547°F), and the liquid film may start to evaporate from the surface of the fuel rods. This corresponds
to the red line shown in the figure. Here, the fuel rods can melt and a major accident can occur. However, as discussed
previously, it takes a while for enough decay heat to be produced by the core for the reactor to reach this point. (If the
operator is able to restart the coolant pumps, this problem can be avoided.)
So the accident at Fukushima, Japan, corresponded to a path on the operating map where the power-to-flow ratio
first followed the blue line and then after an hour or so, began to follow the brown line. (The result of these actions was
then the red line). So in essence, the boiling reduced the water level in the pressure vessel and this made the cooling less
efficient. When the fuel rods overheated, the zirconium cladding reacted with the water to form hydrogen gas, which
exploded when vented into the reactor containment building. This reaction is called the zirconium–water reaction, and
in this book, we will discuss it in Chapters 30 and 33. The equation for the zirconium–water reaction is
In general, this is an exothermal reaction that produces much more energy than it consumes. At cladding temperatures
above 1,200°C (~2,200°F), the reaction becomes so violent that it becomes hard to stop. At these temperatures, the
energy output of the reaction is about 580 kJ/mol of zirconium metal (6.3 MJ/kg). This reaction was ultimately the cause
of the containment building failures that were observed in reactors 1 and 3 at the Fukushima site. The hydrogen ignited,
Bibliography 91
200
150
100
BWR
50
FIGURE 3.20 A comparison of the power densities in a modern BWR and PWR core.
and this is what caused the explosions in the containment buildings to occur. (See Chapter 32 for some pictures of what
the reactors looked like after the roofs were blown off of the containment buildings.)
So in conclusion, reactor operating maps can be very useful when it comes to visualizing what combinations of power
levels and flow rates are acceptable to use in BWRs and PWRs. In BWRs, these maps are generally more complicated
because the core flow can become unstable when the flow rate is low (say, 20%–40% of its normal value), and the power-
to-flow ratio is greater than 1.0. For this reason, the operators of BWRs typically have well-defined start-up plans to keep
the reactors from falling into one of these zones. (For example, region 3 in Figure 3.19.) Similar plans are also used to
shut a reactor down. Some of these plans are also described at the following URL:
http://www.microsimtech.com/Startup/ABWR.html
Normally, there are about a dozen common operating scenarios (start-up, shutdown, pump trip, LOCA, etc.) that every
reactor control system and every reactor operator are expected to anticipate. We will discuss some of these scenarios in
more detail as we proceed further into the book. Thus, a great deal of physics is contained in a BWR or PWR operating
map, and reactor operators must understand the implications of these maps in order to operate their reactors safely and
efficiently. Their respective power densities are also compared in Figure 3.20.
Bibliography
Books and Textbooks
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Lahey, R.T. and Moody, F.J. The Thermal Hydraulics of a Boiling Water Reactor, Second edition, American Nuclear Society,
La Grange Park, IL (1993).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. An Introduction to Nuclear Reactor Physics, First edition, CRC Press, Boca Raton, FL (2017).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2012).
Tong, L.S. Boiling Heat Transfer and Two Phase Flow, Taylor & Francis, Bristol, U.K. (1997).
92 The Boiling Water Reactor
Web References
http://blog.metasd.com/2011/03/page/2/.
http://www.nrc.gov/reactors/bwrs.html.
http://www.microsimtech.com/Startup/ABWR.html.
http://en.wikipedia.org/wiki/Boiling_water_reactor.
http://www.nrc.gov/reading-rm/basic-ref/teachers/03.pdf.
http://me1065.wikidot.com/fuel-assemblies-in-nuclear-reactors.
http://en.wikipedia.org/wiki/Slip_ratio_(gas%E2%80%93liquid_flow).
http://ocw.mit.edu/courses/nuclear-engineering/22-06-engineering-of-nuclear-systems-fall-2010/lectures-and-readings/
MIT22_06F10_lec06b.pdf.
41. Name a burnable poison that can be used in a BWR fuel assembly.
4 2. What is the shape of a control rod in a commercial BWR core?
43. Why do not BWRs use soluble boron or solbor to help control the reactivity swing of the core?
44. What is the purpose of the downcomer in a BWR?
45. In a major reactor accident in a BWR, how long does it take for the decay heat in the core to fall to one half of 1 %
of its full power value?
46. The reactor at Chernobyl was an example of what type of BWR?
47. What happens to the power level in a BWR when the coolant begins to boil?
48. What is the Russian version of the General Electric BWR called?
49. What is the shape of a fuel assembly in a Russian BWR?
50. In the Western world, how does the cladding used in a BWR differ from the cladding used in a PWR?
51. From what direction do the control rods enter the core of an American BWR?
52. What neutron-absorbing material can be found in the control rods of most BWRs?
53. Compare the pitch of the fuel rods in a standard BWR to pitch of the fuel rods in the ABWR or the SBWR.
54. What is the average diameter of a fuel rod in a BWR today, and what type of material is normally used for the cladding?
55. Write down the equation for the zirconium–water reaction that can occur when the power-to-flow ratio in a BWR
becomes too high.
56. Approximately how much energy is released per mole of zirconium in the zirconium–water reaction?
57. What are the three distinct types of boiling that are allowed to occur in a BWR?
58. At what temperature in a thermal water reactor does the reaction of the cladding with the coolant (which is also
called the zirconium–water reaction) become autocatalytic?
59. What is the purpose of a BWR operating map?
60. Why is the fuel pin centerline temperature limited to a maximum value of about 1,750°C in a BWR d uring normal
operating conditions?
61. How many separate batches of fuel are present in the core of a BWR at any given time?
62. Name a BWR that is moderated with carbon-12 atoms (i.e., conventional graphite).
63. What is the thermal efficiency of a modern BWR?
64. What happens to the waste heat in a BWR that is not used to produce electric power?
65. What type of BWR was damaged by the Tsunami at Fukushima, Japan?
66. If you decided to buy a BWR today, what safety features would you insist upon as the purchaser or the reactor operator?
Exercise 3.2
In Exercise 3.1, the reader was asked to find the void fraction of the coolant at the top of a BWR fuel assembly (i.e., the
volume of the coolant α that was converted into steam). Now suppose that we would like to know what percentage of
the total mass of the coolant is converted into steam at this point. What is this percentage called, and what is its value
at the top of the core (z = H)?
Exercise 3.3
In Exercise 3.2, it was assumed that the liquid and vapor phases were moving through a BWR fuel assembly at exactly
the same speed. In reality, the slip ratio in a BWR core is closer to 2.0; that is, the vapor moves at twice the speed of the
liquid. In this case, what is the exit quality at the top of the core?
Exercise 3.4
Today, most BWRs use 8 × 8 fuel assemblies where the fuel rods have an outer diameter of 11.20 mm and a rod pitch of
14.37 mm. Suppose that the outer diameter of the fuel rods is shrunk to 10.26 mm and the rod pitch is reduced to 12.95
so that a fuel assembly can now accommodate 100 fuel rods in a 10 × 10 array. If the enrichment of the new fuel rods
remains the same, how does this affect the overall power density of the core?
94 The Boiling Water Reactor
Exercise 3.5
Today, BWRs with 64 fuel rods per fuel assembly have an average power density of about 55 kW/L. Using the param-
eters supplied in Exercise 3.4, calculate how much the power density would change if the outer diameter of the fuel rods
was shrunk to 10.26 mm and the rod pitch was reduced to 12.95. How much must the flow rate be changed so that the
power-to-flow ratio in the core remains the same? Assume that the enrichment of the fuel rods remains the same.
Exercise 3.6
The average thermal efficiency of a modern BWR is about 32%. However, the ABWR and the SBWR both have thermal
efficiencies of 34.4%. Assuming that the steam handling components in the primary loop remain the same, how do you
think this increase in efficiency is achieved? Is the steam that reaches the power turbines wetter or drier than before?
Exercise 3.7
In a modern BWR, the average core flow rate is about 14,000 kg/s, which is about 70% of the flow rate in comparably
rated PWR (see Chapter 2). How much must the energy content of the steam be increased to achieve this effect, and what
does this say about the amount of energy required to convert some of the reactor coolant to steam?
Exercise 3.8
The zirconium–water reaction can be a serious concern if the cladding temperature in a thermal water reactor rises
above 1,200°C. What temperature must the cladding be kept at in a BWR to prevent the zirconium–water reaction from
ever occurring?
Exercise 3.9
During a severe reactor accident, assume that the cladding in the upper 25% of a BWR core dries out and that the clad-
ding temperature rises above 1,200°C. In a large BWR, the total amount of zirconium surrounding the fuel rods can
weigh 40,000 kg, and the total amount of zirconium in the channel boxes can weigh an additional 25,000 kg. How much
energy is released by the zirconium–water reaction if 50% of the zirconium in the upper 25% of the core is combined
with oxygen as a result of this reaction?
Extra credit: Refer to a more advanced textbook such as Todreas and Kazimi (2012) to determine the amount of
hydrogen gas that is released in this reaction. What is the molecular structure of the hydrogen gas that is released?
Exercise 3.10
Suppose that you would like to increase the volumetric power density of a BWR core so that it is the same as the volu-
metric power density of a PWR core. Assuming that the system pressure level can be adjusted so that three flow regimes
as discussed earlier in the chapter—subcooled boiling, bulk boiling, and annular or film boiling—do not change, how
much greater must the flow rate through the reactor coolant pumps be to accommodate this increase in the volumetric
heat generation rate if the current flow rate is 12,000 kg/s?
Exercise 3.11
One of the primary reasons why a BWR core behaves differently than a PWR core is that the coolant is designed to boil.
The bulk boiling of the coolant removes a great deal of the heat that is produced by the nuclear fuel rods by converting
the cooling water to steam. Using a publically available version of the steam tables, estimate the amount of energy it
takes to convert 1 kg of subcooled water in a BWR fuel assembly to superheated steam at a system pressure of 1,050
PSI (7.15 MPa).
Exercise 3.12
Suppose that the void fraction 1 m below the top of the hottest fuel assembly in a BWR core is 75%. What is the density
of the two-phase mixture at this location in the core? What is the value of the flow quality in this fuel assembly if the slip
ratio at this axial elevation is 1.5? Why does the steam move faster than the liquid at this particular location?
4
Fast Reactors, Gas Reactors,
and Military Reactors
4.1 A n Introduction to Other Reactor Types
Gas reactors, military reactors, and liquid metal fast breeder reactors (LMFBRs) make up most of the rest of the power reactor
installed base in the world today. In this chapter, we would like to discuss each of these designs and their relevance to the field of
nuclear power as a whole. Research reactors, such as the General Atomics TRIGA reactor, are discussed in our companion book,
where we will also consider the effects of temperature feedback on their behavior.*
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, First edition, CRC Press (2017).
95
96 Fast Reactors, Gas Reactors, and Military Reactors
Hot sodium also reacts violently with water, and this requires the sodium in the primary loop to be physically isolated
from the steam producing components of the NSSS in the secondary loop. In the early days of the nuclear industry,
these features of liquid sodium caused operational problems with early fast reactors such as those used in the USS
Seawolf (USSN-575) and also the Japanese Monju reactor. However, more modern fast reactors, such as Experimental
Breeder Reactor number two (EBR-II) and the French Superphenix, have been operated safely for over 30 years with no
sodium-related problems. Because fission neutrons are not slowed down to thermal energies in fast reactors, almost all
fast reactors use some type of liquid metal for their coolant. The early Clementine reactor* at Los Alamos used liquid
mercury for its coolant, but liquid mercury was found to absorb more fast neutrons than liquid sodium did, and so it did
not have very good neutron economy. In addition, the mercury in a fast reactor can become more radioactive than the
liquid sodium can, and it also emits more gamma rays. For these reasons, liquid mercury is not used as a coolant in fast
reactors today. Some fast reactors have also used molten lead as their coolant, and lead has become popular in some
types of naval propulsion systems. However, liquid sodium (Na-23) is still considered to be the best coolant to use in a
commercial fast reactor.
Some research and test reactors also use another metallic coolant called sodium–potassium (NaK) because it has a
relatively low melting point. NaK has been used as a coolant in experimental fast reactors. Unlike commercial nuclear
power plants, test reactors and research reactors have to be frequently shut down and refueled. Using conventional
coolants such as liquid lead or liquid sodium would require continually heating them to maintain them in a liquid state.
The use of NaK overcomes this problem and makes them easier to refuel. Liquid NaK, which has a potassium content
between 40% and 90%, continues to be a liquid at even room temperatures (~30°C), and so it is very attractive to use
when the cost of a reheating system becomes an issue. The Soviet RORSAT radar satellites used liquid NaK as their
coolant, and in addition to having a wide liquid temperature range, NaK has a very low vapor pressure, which is very
important in the vacuum of space. The final annoying feature of liquid sodium is that when it absorbs neutrons, and
even fast neutrons, it becomes a copious beta and gamma emitter called Na-24. which has a half-life of 15 hours. Na-24
is highly radioactive, and because it reacts so violently with water, it is usually not advisable to allow this isotope to be
pumped directly from the core to the steam generators. Hence, fast reactors cooled with liquid sodium use two sodium
loops - a primary sodium loop containing this radioactive sodium and an intermediate loop containing non-radioactive
sodium which carries the heat from the primary loop to the steam generators through an intermediate heat exchanger (or
IHX). The IHX is then where the actual heat transfer between the two sodium streams occurs. Since the sodium in the
primary loop is highly radioactive, all components in the primary loop must be heavily shielded. Also, because sodium
reacts violently with the water vapor in ordinary air, most of the components in the primary loop must be immersed in
an atmosphere of nitrogen or another inert gas, with which the sodium does not react. For similar reasons, empty regions
above the sodium in the reactor pressure vessel and the coolant pumps must be filled with argon or another gas which is
chemically inert. A combination of these practices has led to many successful fast reactors with few, if any, operational
problems. In almost all of these reactors, liquid sodium is used as the coolant of choice.
FIGURE 4.1 A cross-sectional view of a typical AGR fuel assembly. (Obtained from Wikipedia.)
* See “Nuclear Engineering Fundamentals: A Practical Perspective” by R.E. Masterson, CRC Press (2017).
98 Fast Reactors, Gas Reactors, and Military Reactors
FIGURE 4.2 A picture of sodium metal in its solid and liquid states. The melting point of sodium metal at atmospheric pressure is
98°C, which turns out to be less than the boiling point of ordinary tap water. (Courtesy of Wikipedia.)
example, sodium is highly reactive chemically because it is an alkaline metal. Liquid sodium reacts violently with mate-
rials like water and catches on fire when it is exposed to the oxygen in ordinary air. For this reason, the liquid sodium in
the primary loop, which is used to conduct the heat, must be kept physically separated from the water in the secondary
loop at all times. The spaces around the coolant pipes must also be kept free of ordinary air in case a pipe leak develops.
If a leak develops, this would result in what is known as a sodium fire. Finally, liquid sodium has a tendency to absorb
excess neutrons (even fast neutrons), and this can cause the sodium to become temporarily radioactive by turning into
Sodium-24, which is a beta emitter with a half-life of about 15 hours. Therefore, the liquid sodium that passes through
the core will eventually become radioactive. This means that LMFBRs that employ the Rankine cycle, with a secondary
loop containing water and steam to drive the power turbines, must have a primary loop, which carries the radioactive liq-
uid sodium, which is completely isolated from the secondary loop, which contains a water–steam mixture. The easiest
way to do this is to separate the primary side from the secondary side by employing an additional sodium loop between
the core and the steam generators. This additional loop is sometimes called an intermediate loop.
The primary loop carries the radioactive sodium, while the intermediate loop contains sodium that is not radioac-
tive. If a sodium leak happens to develop within one of the steam generators, the water–steam mixture that drives the
power turbines will not become radioactive. This requires an intermediate heat exchanger (or IHX) to transfer the heat
from the first sodium loop to the second sodium loop. The exact manner in which the intermediate loop is deployed
can result in two different configurations of LMFBRs. The first configuration is called a loop-type LMFBR, and the
second configuration is called a pool-type LMFBR. We would now like to discuss the features of each of these designs
separately. A schematic representation of a loop-type LMFBR is shown on the right-hand side of Figure 4.3, and a
schematic representation of a pool-type LMFBR is shown on the left. In the loop-type LMFBR, the core, the blanket,
and the control rods are located in a reactor pressure vessel that is not unlike that of a conventional light water reactor.
However, the pressure vessel wall can be much thinner than it is in a BWR or a PWR because it does not have to operate
at such high pressures. In fact, the cores of most LMFBRs are operated close to atmospheric pressure (see Table 4.5). If
any liquid sodium leaks out of the primary or secondary loops, it can then be collected into what is called a safety tank.
In a loop-type design, the IHX and all other components of the nuclear steam supply system (NSSS) are located outside
of the reactor pressure vessel. A standard 1,000 MWE LMFBR has three or four primary loops, and each primary loop
is mated to its own IHX and steam–water loop. The primary side is heavily shielded because the liquid sodium flow-
ing through the core (which has been partially converted to Na-24) is highly radioactive, whereas the shielding on the
secondary side can be kept to a much more reasonable level.
A pool-type LMFBR, which is pictured on the left-hand side of Figure 4.3, has an entirely different architecture
than a loop-type LMFBR because all of the critical components of the NSSS, including the primary loop, the second-
ary sodium loop, the IHXs, and even the coolant pumps, are located together with the core and the blanket within the
reactor pressure vessel. Only the steam generators and the power turbines are located outside of the reactor pressure
vessel. Intuitively, the loop-type LMFBR is a much simpler design because it is easier to service and all of the important
components of the NSSS are separate from one another and can be serviced independently. However, the primary loops
require more shielding than the loops in the pool-type design, and because of this, plants that use a loop-type design
must be relatively large and well constructed. A pool-type LMFBR has a different set of design trade-offs that must be
weighed against the fact that almost all of the critical components are immersed within a single pool of liquid sodium.
In a pool-type LMFBR, no radioactivity leaves the reactor vessel, and because of this, it is the only component of the
4.6 FUEL ASSEMBLIES FOR LMFBRs 99
FIGURE 4.3 An illustration of the difference between a pool-type and a loop-type LMFBR. (Courtesy of Wikipedia.)
FIGURE 4.4 A picture of the Superphenix—the largest pool-type LMFBR in the world. (Courtesy of the IAEA.)
plant that must be heavily shielded. Furthermore, because the pressure vessel is so large, at least part of the pressure ves-
sel can be located underground, and this reduces the capital investment in a pool-type LMFBR because only the upper
part of the reactor pressure vessel needs to be heavily shielded. Consequently, a pool-type LMFBR is much smaller and
more compact than a loop-type LMFBR of the same power rating. Two of the largest commercially operating LMFBRs
are the French Superphenix reactor (see Figure 4.4) and the Japanese Monju reactor. The Monju reactor is a loop-type
LMFBR, and the Superphenix is a pool-type LMFBR. The reliability of both of these reactors has been excellent.
P P
D D
FIGURE 4.5 How the reactor lattices and the unit cells compare in an LMFBR (a) and a light water reactor (b).
Each of the fuel assemblies in a LMFBR is surrounded by a hexagonal sheath or “can.” This can is normally made from
stainless steel. The purpose of the can is to prevent a shock wave from propagating through the core if the liquid sodium
were to become hot enough to boil. However, in a typical LMFBR, the liquid sodium in the primary loop is always
kept at least 150°C below its boiling point. During normal operation, this operating margin can be even greater (about
300°C) - see Table 4.5. Most fuel assemblies in the core are between 10 and 15 cm wide and between 3 and 4 m long.
Here, the fissionable fuel and the fertile breeding material are cast into long fuel pins that consist of either Uranium-238
or a Plutonium-239/Uranium-238 mixture (UO2 or PuO2). A fuel assembly in the center of the core contains fuel pins
near the center and blanket pins at either end. A fuel assembly near the edge of the core contains only blanket fuel pins
(which are usually made from U-238O2). Then, these fuel assemblies are assembled into large hexagonal arrays; the net
effect is to produce a central power-generating region with a radial and axial blanket on each side. Although the fuel
assemblies are hexagonal, the overall geometry begins to resemble that of a right circular cylinder. Liquid sodium enters
the core through a number of small holes near the bottom of each fuel assembly, and as it passes upward along the fuel
pins, it begins to heat up and eventually exits from the top of the core. The overall temperature rise through the core
is between 150°C and 200°C. (In pool-type LMFBRs, the temperature increase has tended to remain near the top of
this range). The control rods used in LMFBRs are usually small stainless steel tubes filled with boron carbide rods to
absorb the fast neutrons, although other materials are sometimes used. One of the challenges of building a fast reactor
is providing enough control rods to be able to absorb the fast neutrons that are produced in the core. Because the neu-
tron absorption rates are lower than they are at thermal energies, and because the number of delayed neutrons that are
produced is also much smaller, fast reactors are generally harder to control than thermal water reactors are. However, it
is still possible to build them and operate them safely. Only in this case, the control systems must be much more sophis-
ticated. The margin for error is also considerably less in the event a reactor operator attempts to do something stupid.
Because the temperature of the liquid sodium in the core is so high, the steam generated in the secondary or tertiary
loops is delivered to the steam turbines as dry steam which can also be superheated. The normal temperature at the
entrance to the steam turbines is about 500°C, and the average pressure is between 16 and 18 MPa. A large flow of steam
with this energy content means that the thermal efficiency of a commercial LMFBR can approach 40%. Hence, LMFBRs
are extremely efficient electricity producers because about 40% of the thermal energy produced in the core can be con-
verted into electric power. Water reactors, on the other hand, can only achieve thermal efficiencies of about 35%. The
fuel rods in LMFBRs are stainless steel tubes about 7–8 mm in diameter, and the average diameter of a fuel pin is about
6.6 mm. The fuel pellets inside of the pins contain a mixture of uranium and plutonium dioxide (UO2 and PuO2), and the
amount of plutonium in the fuel pins is equal to between 15% and 35% of the total uranium content (by weight). The fuel
pins are much closer together than they are in water reactors (about 17%), and they are kept apart by spacers, or in some
cases, helically wound wire around each fuel rod. The helically wound wire is sometimes called a wire wrap. Using a
helical wire in a fuel assembly has the advantage that it increases the amount of turbulent mixing, and therefore, it helps
to increase the heat transfer rate between the fuel rods and the coolant. It can also provide additional structural stability
and integrity in some cases. The fuel rods in the radial blanket are generally much larger than they are in the core. These
rods contain UO2, and almost all of the uranium in the blanket is U-238. The average diameter of a fuel pin in the blanket
is about 15 mm. This is about twice as large as a typical fuel pin in the core. The layout of the fuel pins in the core and
blanket regions is shown in Figure 4.6. Notice that the diameter of the fuel rods in the axial blanket (above and below
4.6 FUEL ASSEMBLIES FOR LMFBRs 101
Axial blanket
Axial blanket
FIGURE 4.6 The core and blanket regions of a plutonium-fueled fast breeder reactor.
(From http://www.ati.ac.at/fileadmin/files/research_areas/ssnm/nmkt/11_LMFBR.pdf.)
the core) is the same as that in the core, but in the radial blanket, the rods are much larger. The larger diameters can be
tolerated because they require less cooling than the fuel pins in the core do.
Although almost all commercial LMFBRs use the uranium–plutonium fuel cycle today, it is also possible to
operate them using the Thorium-232/Uranium-233 fuel cycle. This fuel cycle has the advantage that there is at least
three times more thorium in the Earth’s crust than uranium. In this fuel cycle, the center of the core contains U-233
instead of Pu-239, and in the blanket, thorium takes the place of natural or depleted uranium. The thorium in this
case consists of a thorium dioxide fuel pellet (ThO2). Unfortunately, substituting a Th-232/U-233 mixture for a
Pu-239/U-238 mixture has a devastating effect on the breeding ratio. A LMFBR fueled with U-233 and Thorium-232
effectively does not breed. (However, it does not consume any U-233 either.) The breeding ratios for both fuel cycles
are compared in Table 4.1. Notice that the thorium cycle reduces the breeding ratio from about 1.28 to about 1.04. It
also increases the doubling time from about 16 to 112 years. Finally, we would like to mention that the pitch-to-diameter
(P/D) ratio of the fuel rods in a LMFBR is considerably smaller than it is in a conventional light water reactor. This is
entirely a function of the fact that a moderator is not required to slow down the neutrons in a LMFBR. The majority of
the neutrons are absorbed long before they reach the thermal energy range, and because of this, the fast neutron energy
spectrum ends at effectively 10 keV. A fuel rod in a standard PWR fuel assembly has an outer diameter of about 9.5 mm
TABLE 4.1
Breeding Ratios for Fast and Thermal Reactors for Different Types of Hypothetical Fuel Cycles
and a cladding thickness of about 0.57 mm. The fuel rod pitch is 1.26 cm. This gives rise to a P/D ratio of about 1.326. In
a commercial BWR, a fuel rod has an outer diameter of about 10.5 mm and a cladding thickness of about 0.85 mm. The
fuel rod pitch is 1.62 cm. This gives rise to a P/D ratio of about 1.3. By comparison, a fuel rod in a commercial LMFBR
has an outer diameter of about 7 mm and a cladding thickness of about 0.44 mm. The fuel rod pitch is about 0.82 cm.
Hence, the P/D ratio in a LMFBR is about 1.17. The fact that the fuel rods are much closer together means that the power
density is also increased by a comparable amount.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, First edition, CRC Press (2017).
TABLE 4.2
Design Parameters of LMFBRs That Have Been Built in Several Countries
Reactor Name Phoenix SNR 300 Monju PFR CRBR BN-350 BN-600 Superphenix CDFR1 BN-1600
Country of operation France Germany Japan United Kingdom United States Russia Russia France Germany Russia
Electrical power (MWE) 250 312 300 250 350 350 600 1,200 1,300 1,600
4.8 Other Reactor Concepts
Thermal power (MWT) 568 762 714 612 975 1,000 1,470 3,000 3,420 4,200
Thermal efficiency (%) 440 409 420 409 359 35 408 400 380 381
Reactor type Pool Loop Loop Pool Loop Loop Pool Pool Loop Pool
Primary coolant loops 3 3 3 3 3 6 3 4 4 4
Secondary coolant loops 3 3 3 3 3 6 3 4 4 4
IHX 6 9 6 3 3 12 9 8 8 8
Secondary
radiation shield
Steam
Radiation
shield
Turbine
Condenser
Reactor
Steam generator
Primary
coolant pump
Feed pump
FIGURE 4.7 Military reactors in nuclear submarines and aircraft carriers tend to be quite small. They are almost all small,
high-powered PWRs with some exceptions. (Picture supplied by the U.S. Navy.)
* See Nuclear Engineering Fundamentals: A Practical Perspective by R.E. Masterson, CRC Press (2017).
4.9 Military Reactors and the MPR 105
FIGURE 4.8 The USS Nautilus, which is similar to the submarine shown here, was the first nuclear-powered submarine to
sail under the north pole. It used a military-type PWR with an average initial enrichment of about 90%. High enrichments are
commonly used in military power reactors to extend the refueling intervals to between 30 and 40 years. A larger version of the
same PWR is used in nuclear aircraft carriers as well.
FIGURE 4.9 Aircraft carriers in the U.S. Navy use nuclear reactors, and in particular, MPRs for their propulsion systems.
The propulsion system for a MPR is shown in Figure 4.7. The heat produced in the core creates steam that is used to
propel the ship. The propulsion system shaft is connected to an electrical generator that also provides electricity for
the crew. More steam increases the speed of the ship and less steam causes it to slow down. Nuclear powered aircraft
carriers and submarines can achieve top speeds as high as 30 knots (56 km/hr or 35 mph) using the steam produced by
MPRs. These ships use between two and four propeller shafts to achieve these speeds. Carriers can have between 2 and
4MPRs while nuclear submarines usually have just one. Recently, there has been an attempt to standardize the design of
these MPRs to improve their reliability and reduce their cost.
Rankine cycle
AGR
Cooler H2O
Q – waste heat
CO2 (Liquid)
HTGR
Brayton cycle
AGR
HTGR
FIGURE 4.10 A comparison of a Rankine thermal cycle and a Brayton thermal cycle for AGRs and HTGRs.
Charge face
Reactor core
Fuel element channel
Heat exchanger
Concrete
Steel
Cold gas
BISO particle
Two-layer coating
Fuel
TRISO particle
Three-layer coatin
Fuel
(a) (b)
FIGURE 4.12 The refueling floor for the Ft. St Vrain nuclear reactor, which was decommissioned in 1989 (a), and the spherical
fuel particles with a graphite coating that were used in the reactor (b). (Picture provided by Wikipedia.)
Notice that the coolant pressures are very similar (about 4.25 MPa for the AGR vs. 5.0 MPa for the HTGR). The inlet
temperatures and the outlet temperatures are higher for the HTGR than they are for the AGR, and in the case of the
HTGR, the outlet temperature can sometimes exceed 740°C (compared to an outlet temperature of about 325°C for a
conventional PWR). In the secondary loop, the pressure of the coolant is much higher than it is in a PWR (about 17 MPa),
and this is one of the reasons why the thermal efficiency of these gas reactors is so high (the higher pressure increases
the energy content of the steam). The average temperature drop across the secondary loop can be between 340°C and
380°C. Hence, gas reactors are highly efficient devices when measured by water reactor standards.
4.12 Comparing the Designs of Gas Reactors 109
FIGURE 4.13 The honeycomb in a beehive is similar in shape to the arrangement of the graphite blocks in an HTGR.
TABLE 4.3
A Summary of How the Thermal Cycles are implemented in the AGR and the HTGR
A Comparison of the Temperatures and the Pressures in Several Different Gas Reactors
Manufacturer National Nuclear General Atomics
TABLE 4.4
Common Gas Reactor Fuel Pin, Fuel Rod, and Fuel Assembly Design Parameters
Fuel Pins
Geometry Cylindrical Cylindrical
Diameter 14.50 mm 15.70 mm
Composition UO2 Compacted UO2 spheres
Enrichment (avg.) ~2.5% 90%–95%
Fuel Rods
Geometry Cylindrical N/A
Outer diameter 15.30 mm N/A
Active fuel height 987 mm 742 mm
Total rod/fuel height 1.04 m 793 mm
Cladding thickness 0.37 mm No cladding
Cladding material Stainless steel N/A
Fuel Assemblies
Moderator Graphite Graphite
Fuel rods per side N/A N/A
Fuel rod layout Concentric circles of rods Cylindrical fuel compacts
with a graphite sleeve within a graphite block
Fuel rod pitch 37 mm 23 mm
Fuel rod locations 36 132
Number of fuel rods ~190 132
Fuel assembly geometry Hexagonal block Hexagonal block
Source: Todreas and Kazimi (2012).
4.14 LMFBR Thermal Cycle Performance 111
Steam
generator
Turbine Generator
Cold plenum
Hot plenum Electrical
Control power
rods Heat
exchanger
Condenser
Heat sink
Pump
Primary
Secondary
sodium
sodium
(hot)
Pump
Pump
Core
Primary
sodium
(cold)
A Comparison of the Temperatures and the Pressures in Different Loops of a loop-type LMFBR
Manufacturer Novatome
1.94 m. Hence, the core is relatively compact by water reactor standards. The rod pitch is 9.8 mm in the core and 17.0 mm
in the radial blanket.
A typical fuel assembly in the core has 271 fuel rods, and in the radial blanket, a typical fuel assembly has 91 fuel
rods. The fuel assemblies are always hexagonal in shape, and they are always surrounded by a “can.” (The can limits
the propagation of a shock wave if the sodium happens to boil.) However, in these designs, the liquid sodium is kept at
least 300°C below its boiling point (~880°C). Some common core design parameters are shown in Table 4.6. Notice that
there is very little difference in the core design parameters between a pool-type LMFBR and a loop-type LMFBR. The
primary difference is in the placement of the loops, the pumps, and the heat exchangers in the rest of the plant.
TABLE 4.6
Common LMFBR Fuel Pin, Fuel Rod, and Fuel Assembly Design Parameters
Fuel Pins
Geometry Cylindrical
Diameter 7.14 mm (core)
Composition PuO2 and UO2
Enrichment (avg.) ~20%
Diameter 14.45 mm (blanket)
Composition Depleted uranium dioxide
Enrichment (avg.) ~1%
Fuel Rods
Geometry Cylindrical
Outer diameter 8.50 mm (core)
Active fuel height 2.70 m (core)
Geometry Cylindrical
Outer diameter 15.8 mm (blanket)
Active fuel height 1.94 m (blanket)
Cladding thickness 0.56 mm
Cladding material Stainless steel
Fuel Assemblies
Moderator None
Fuel rods per side N/A
Fuel rod layout Hexagonal rod array (core)
Fuel rod pitch 9.8 mm (core)
Fuel rod locations 271 (core)
Number of fuel rods ~271 (core)
Fuel assembly geometry Hexagonal array surrounded by a “can”
Fuel rods per side N/A
Fuel rod layout Hexagonal rod array (blanket)
Fuel rod pitch 17.0 mm (blanket)
Fuel rod locations 91 (blanket)
Number of fuel rods ~91 (blanket)
Fuel assembly geometry Hexagonal array surrounded by a “can”
Source: Todreas and Kazimi (2012).
114 Fast Reactors, Gas Reactors, and Military Reactors
300 LMFBR
250
150
PWR
100
BWR
50
CANDU
0
FIGURE 4.15 A comparison of the power densities in modern PWR, BWR, and LMFBR cores. The differences in the power
generation rate are attributable primarily to differences in the enrichment of the fuel. Note that LMFBRs use a hexagonal lattice,
whereas PWRs and BWRs use a square one.
enrichment, the choice of the coolant, and the choice of the moderator all affect the power densities that are achieved in
practice. In Figure 4.15, the power density is simply the total heat produced in a fuel assembly per second (Q), divided
by the volume of the fuel assembly V in which the heat is produced:
Military reactors also have much higher power densities than commercial PWRs, and they are designed to be in service
for as long as 40 to 50 years before they have to be refueled.
Solution From Table 4.6, we know that LMFBR fuel pins have a radius of about 3.6 mm and a cladding thickness of
about 0.55 mm. In classical heat transfer books (see Cengel and Boles, 2008; El-Wakil, 1981), it can be shown that the
nuclear heat flux q″ is related to the cladding temperature drop ΔT by the following equation:
where Δr is the thickness of the cladding and R f is the radius of the pin. (This equation assumes that the cladding is rela-
tively thin.) Therefore, if we know the value of the heat flux q″, the equation for the temperature drop across the cladding
is ΔT = Δrq″/2kRf. Plugging in the appropriate values for these quantities, we find that the temperature drop across the
cladding is ΔT = 0.55 × 50/(2 × 0.163 × 3.6) = 23.4°C.
Thus, the temperature on the inner surface of the cladding is 700°C + 23.4°C = 723.4°C. Obviously, since this is well
below the melting point of stainless steel (~1,400°C for SS 304), the cladding will not melt under these conditions. [Ans.]
1 − TOUT
ηMAX = (4.2)
TIN
where
TIN is the absolute temperature at which heat is added to the coolant.
TOUT is the absolute temperature at which heat is rejected from the coolant.
This thermal efficiency is known as the Carnot efficiency, and it is the maximum thermal efficiency that any reversible
heat engine can achieve. The Carnot efficiency must be measured using the absolute temperature scale, where all tem-
peratures are measured in degrees Kelvin (°K). Thus, to compute the Carnot efficiency, TOUT and TIN must be provided
in °K (rather than in °C). It also follows from Equation 4.2 that the entropy S = dq/T of any irreversible heat engine (such
as a nuclear power plant) must always increase as heat flows from the high-temperature reservoir to the low-temperature
reservoir.
Thus, the thermal efficiency of a nuclear power plant must be always less than 100% because TOUT and TIN are
defined to be positive numbers. In fact, as we discussed earlier in the chapter, it is very difficult to achieve a thermal
efficiency greater than 50% (unless the coolant exits the core at an extremely high temperature, which can cause the
cladding to melt). Sometimes, Equation 4.2 is also written as
1 − Q OUT
ηMAX = (4.3)
Q IN
where
QIN is the amount of heat added to the high-temperature coolant.
QOUT is the amount of heat rejected from the low-temperature coolant.
The difference between these two numbers is then the amount of useful work W that the power plant (or any other heat
engine) can produce:
W = Q IN − Q OUT (4.4)
Normally, the values of QIN and QOUT are inferred directly from the specific enthalpy because the working fluid in the
secondary loop (unless it is a gas) will undergo a phase change as it is cooled and as energy is extracted from it by the
steam turbines. The following problem shows how the maximum thermal efficiency can be found.
boiling point, which is 880°C at atmospheric pressure. If heat is to be rejected from the working fluid in the tertiary loop
at a temperature of 237.0°C, what is the theoretical thermal efficiency for this proposed design?
Solution From Section 4.18, the maximum thermal efficiency the plant can achieve is given by the equation η = 1 −
TOUT/TIN which represents the Carnot efficiency. The temperatures in this equation must be provided in degrees Kelvin.
Since the temperature of the liquid sodium at the core exit must be kept 150°C below its boiling point, it must exit the
core at a temperature no greater than 880°C − 150°C = 730°C. This sets the value of TIN at 730°C. The values of TIN
and TOUT in degrees Kelvin are therefore 1,003°K and 510°K. The theoretical thermal efficiency of this new LMFBR
can therefore be no greater than η = 1 − 510/1,003 = 0.492 = 49.2%. Needless to say, this is still a very good number by
nuclear standards. [Ans.]
In practice, the actual thermodynamic efficiency is always less than the Carnot efficiency because heat can leak
out of the piping system, and because of frictional losses in the other components (steam generators, pipes, heat
exchangers, etc.) of the NSSS. Some of this waste heat can also be recycled and fed back directly into the core to
increase the inlet temperature of the coolant and thus the value of T IN. In a nuclear power plant, this process is
called regeneration, and if enough waste heat can be redirected back into the core, the actual thermal efficiency
can be raised by 1% or 2%. However, additional components must be added to the NSSS (such as feedwater heaters)
so that that process is implemented correctly. Finally, it is possible to increase the thermal efficiency even further
by redirecting some of the high-pressure steam to the low-pressure steam turbines to boost the energy content of
the spent steam by reheating it. This process is called reheating the steam, and if both reheat and regeneration are
used, it is possible to increase the overall thermodynamic efficiency by 2% or 3%. We will have more to say about
how these processes work when we discuss reactor thermal cycles in Chapter 9. The NSSS is described in additional
detail in Chapter 8.
Administration. The EBR-I was built in the state of Idaho, and the power it produced was mostly sold to customers in
the Pacific Northwest.
Bibliography
Books and Textbooks
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, McGraw-Hill, New York (2008).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley & Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. An Introduction to Nuclear Reactor Physics, First edition, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, First edition, CRC Press, Boca Raton, FL (2017).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2012).
Waltar, A.E. and Reynolds, A.B. Fast Breeder Reactors, Pergamon Press (1981), New York, NY.
Web References
http://en.wikipedia.org/wiki/Thermal_efficiency.
http://en.wikipedia.org/wiki/Carnot_heat_engine.
http://en.wikipedia.org/wiki/Fast-neutron_reactor.
http://hyperphysics.phy-astr.gsu.edu/hbase/thermo/carnot.html.
http://www.ati.ac.at/fileadmin/files/research_areas/ssnm/nmkt/11_LMFBR.pdf.
http://www.iaea.org/Publications/Magazines/Bulletin/Bull206/20604782938.pdf.
22. The Ft. St Vrain power plant, which is located near Denver, Colorado, is an example of what type of reactor type?
23. Why is stainless steel such a popular cladding material for fast reactors?
24. In what type of reactor can a U-tube steam generator normally be found?
25. Helical coil steam generators are normally found in what type of reactors?
26. What type of reactor can be designed to run on natural uranium?
27. Why are the fuel assemblies in an LMFBR surrounded by a metal stainless steel “sheath” or “can”?
28. How far below the boiling point of liquid sodium does the coolant in the core of an LMFBR normally operate?
29. What is the refueling cycle of a typical military reactor?
30. Who is the designer of the American HTGR?
31. When it comes to designing an MPR, the number of neutrons absorbed by the coolant should be as small as p ossible.
Given a choice between light water and heavy water, what type of coolant would you be most likely to use if the cost
was not an issue?
32. Which type of reactor uses a BISO or TRISO fuel particle?
33. Why is NaK sometimes used as a coolant in fast research reactors?
34. When the liquid sodium coolant used in an LMFBR is exposed to ordinary air, what does it do?
35. What two early fast reactors had problems with small sodium fires?
36. How much more U-235 is there on a per volume basis in a fuel rod of a nuclear submarine than there is in a fuel rod
in a commercial PWR?
37. What is one major difference between the fuel rods in the radial blanket and the axial blanket of an LMFBR?
38. Fill in the following sentence with the appropriate word or phrase: Most of the neutrons in a fast reactor have
energies between keV and MeV. Fast reactors rarely have neutrons with kinetic energies below ________.
39. What is the average amount of PuO2 (Plutonium-239 by weight) in the fuel rods of a typical LMFBR?
4 0. Name two reactors that use graphite as a moderating material—one cooled by water and the other cooled by
gas.
41. What type of reactor thermal cycle is used for both pool-type and loop-type LMFBRs?
42. What are three major differences between a military reactor and a commercial power reaction?
43. How often does the nuclear reactor in a nuclear aircraft carrier have to be refueled?
44. Where was the first MPR invented?
45. What fissile material was used in the first atomic bomb, which was dropped on Hiroshima, Japan, and what fissile
material was used in the second atomic bomb, which was dropped on Nagasaki, Japan?
46. MPRs are typically lighter and smaller than commercial power reactors. What material would you use for the reflec-
tor of an MPR to reduce the neutron leakage rate from the core?
47. Why is the thorium fuel cycle not commonly used for fast reactors?
48. What is the average diameter of an LMFBR fuel rod, and what is the average pitch?
49. Is the thermal conductivity of liquid sodium higher or lower than that of ordinary water?
50. Assume that you are asked to design an MPR with a 20-year refueling cycle. What should be the average enrichment
of the fuel rods in the core?
51. Why is the average power density of a PWR higher than the average power density of a BWR?
52. Name four liquid metals that can be used to cool a fast reactor.
53. What material was used to cool the first experimental fast breeder reactor?
54. At one point in time, liquid mercury was considered as a possible coolant in an LMFBR. Why do you think that
liquid sodium was ultimately selected for use in fast reactors instead of liquid mercury?
55. Why is the critical mass of a fast reactor generally greater than that of a thermal reactor?
56. In addition to producing extra nuclear fuel, what can fast breeder reactors be used for?
57. How many separate coolant loops does a loop-type LMFBR have?
58. What was the code name for the Trinity test device, which eventually became the world’s first atomic bomb?
59. Why are the power densities in the cores of military reactors much higher than they are in the cores of commercial
power reactors?
60. How many fuel pellets are there in an AGR fuel assembly?
61. What is the average enrichment of the fuel rods in the core of a nuclear submarine?
62. In an LMFBR, are the fuel rods in the radial blanket the same width, smaller than, or larger than the fuel rods in the
core?
63. What coolant did the early Clementine fast reactor, which was built in Los Alamos, New Mexico, use?
64. Name two reasons why liquid sodium is preferable to liquid mercury as a coolant in a fast reactor.
65. What is the average fissile content of the fuel rods in a commercial fast reactor?
66. What is the average melting point of NaK, which is sometimes used as a coolant in fast research reactors?
Exercises for the Student 119
Exercise 4.2
Through better design practices, the breeding ratio of a fast reactor where the fuel is to be replaced every 4 years can be
increased from 1.1 to 1.2. By how many years does this improved design reduce the doubling time?
Exercise 4.3
Coolant leaves the condenser of an AGR at 200°C and is heated in the core to a temperature of 638.0°C. Assuming that
there are no other thermal losses or inefficiencies in the piping system or the compressors, what is the theoretical thermal
efficiency of this AGR?
Exercise 4.4
Suppose that an LMFBR core contains 50 fuel assemblies with an average thermal output of 50 MW per assembly. If the
thermal efficiency of the power plant is 40%, how much electric power does this LMFBR generate?
Exercise 4.5
LMFBRs are designed to run much hotter than thermal water reactors are. Suppose that heat is added to the working
fluid in the primary loop of an LMFBR at a temperature of 545°C and rejected from the working fluid in the tertiary
loop at a temperature of 237.0°C. What is the theoretical thermal efficiency η of the LMFBR that operates at these
temperatures?
Exercise 4.6
The energy content (or the enthalpy) of liquid sodium entering the core of an LMFBR at 395.0°C is 600 kJ/kg, and the
energy content (or enthalpy) of liquid sodium leaving the core of an LMFBR at 545.0°C is 790 kJ/kg. If the mass flow
rate through the core is 15,700 kg/s, what is the thermal power output of this LMFBR?
Exercise 4.7
Suppose that the LMFBR in Exercise 4.6 has an actual thermal efficiency of 40%. What is the electrical power output of
this power plant, and if it is a loop-type LMFBR, how many different cooling loops does it have?
Exercise 4.8
If the LMFBR in Exercise 4.7 can produce electrical power at a cost of 6 cents/kWh, and sell it to its retail customers
at an average price of 10 cents/kWh, how much profit (excluding taxes and regulatory fees) does this LMFBR make in
a 24-h period?
Exercise 4.9
Suppose that the cladding temperature around the fuel rods in an LMFBR rises to about 900°C. What h appens to the
coolant passing over the fuel rods in this case, and how does this scenario make it more likely that a sodium-cooled fast
breeder reactor should contain a stainless steel “can” around each fuel assembly?
Exercise 4.10
In Section 4.15, we mentioned that the entropy S of the coolant always increases if we add heat to it and then reject heat
from it at a lower operating temperature. Suppose that a reactor core injects an amount of heat dq into the surrounding
coolant at a temperature of T1 and that the same amount of heat is removed from the coolant at a temperature of T2,
where T2 < T1. Show that the entropy of the coolant is always increased in this case.
5
Thermal Energy Production
in Nuclear Power Plants
5.1 Heat Production in Nuclear Power Plants
Heat is produced in nuclear power plants by the process of nuclear fission, which occurs in what is known as the reactor core (see
Chapter 1) when the nucleus of a uranium or plutonium atom splits apart. This thermal energy is eventually distributed through the
core by energetic fission products and by alpha, beta, and gamma rays that are produced in pairs when the atomic nucleus splits apart.
Most of this heat is deposited in the fuel, but a few percent of it is deposited in the coolant and the cladding. Gamma rays also deposit
some of this energy in the pressure vessel wall. Alpha particles are ionized helium nuclei with their electrons removed, beta rays are
high-energy electrons or positrons, and gamma rays are high-energy photons. (Gamma ray energies are discussed in Section 5.7.) In
addition to these particles, between two and four fission neutrons are produced when an atomic nucleus splits apart. Sometimes these
fission neutrons are called prompt neutrons because they are produced immediately after the atomic nucleus fissions. The kinetic
energy carried away by these neutrons is eventually converted into heat, and this increases the temperature of the core. Coolant is then
pumped through the core to remove this heat, and the heated coolant is eventually sent to the nuclear steam supply system (NSSS) (see
Chapter 8) to generate high-pressure steam. Practically speaking, trillions of fission reactions are required per second to generate
meaningful amounts of electric power. In this chapter, we would like to explore the details of this process of energy production.
When a uranium or plutonium nucleus absorbs a passing neutron, there is a reasonable probability it will split apart. If it does so,
approximately 200 MeV of recoverable kinetic energy is produced. Most of this energy is carried away by the fission products and by
any alpha, beta, or gamma rays that are produced. In addition, fission neutrons carry away between 1% and 2% of the kinetic energy
that is released. A single fission event creates about 2 × 108 × 1.6 × 10 −19 = 3.2 × 10 −11 J of kinetic energy, and about 30 billion fissions
per second are required to create 1 W of thermal power.
Equation 5.2 implies that if a nuclear power plant is 33% efficient, approximately 100 billion-billion fissions per second are required
to produce 3,000 MW of thermal power and 1,000 MW of electrical energy. Hence, nuclear energy production depends on the con-
tinuous destruction of literally trillions of uranium or plutonium nuclei per second.
121
122 Thermal Energy Production in Nuclear Power Plants
FIGURE 5.1 An illustration of the difference between fission and fusion. Left: Uranium-235 combines with a neutron to
form an unstable intermediate nucleus, which quickly splits into two fission products (Barium-144 and Krypton-89) plus three
neutrons. Right: A deuterium and a tritium nucleus combine together through nuclear fusion to form a helium nucleus plus a
neutron. About 20 MeV of kinetic energy is released in a fusion reaction, and about 200 MeV of kinetic energy is released in a
comparable fission reaction.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
5.4 Cross Sections and Reaction Rates 123
FIGURE 5.2 (a) An example of what happens after the capture of a thermal neutron by a U-235 nucleus. About 86% of the time,
the nucleus will split apart into two fission products, and release two to three neutrons in the process. About 14% of the time, the
nucleus will absorb a neutron, emit a gamma ray, and then becomes a stable version of U-236. (b) A barn, which is the size of a
simple atomic nucleus, has a cross-sectional area of 1 × 10 −24 cm 2.
p arameter called a reaction cross section. Cross sections are normally measured in barns, where 1 barn = 1 × 10 −24 cm2.
Thus, 1 barn is equivalent to a cross-sectional area of 1 × 10 −24 cm2, and this area happens to be the cross-sectional
area of a simple atomic nucleus (see Figure 5.2b). Cross sections can be defined for fission, absorption, capture, and
scattering. They can also be defined for other nuclear particles such as electrons and photons. Cross section values are
material dependent, and they also depend on the energy of particles involved. Furthermore, fission and capture cross
sections usually increase as the kinetic energy of the incoming neutron decreases. We would next like to explore the
values of these cross sections in more detail. For radiative capture, the probability of a neutron being captured is equal
to the ratio of the neutron capture cross section σc to the neutron capture and fission cross sections σc and σf:
Probability of capturing a thermal neutron and releasing a gamma ray = σc /(σc + σf) ≅ 15% (for U-235)
The probability of a U-235 nucleus capturing a neutron and causing it to split apart is
Probability of capturing a thermal neutron and splitting apart = σf /(σc + σf) ≅ 85% (for U-235)
124 Thermal Energy Production in Nuclear Power Plants
FIGURE 5.3 The distribution of the masses of the fission fragments created from the thermal fission of Uranium-233,
Uranium-235, and Plutonium-239. (Picture provided by Wikipedia.)
FIGURE 5.4 The kinetic energy spectrum of the fission neutrons that are emitted from the thermal fission of U-235. In nuclear
science and engineering, this curve is also called the Watt curve.
Here, the word thermal is used to refer to a low energy or “thermal neutron,” which has achieved thermal equilibrium
with the other materials in the core. Sometimes the sum of the fission and capture cross sections is called the absorption
cross section σa:
Definition of the absorption cross section: σa = σc + σf
Thus, an absorption cross section measures the tendency of an atomic nucleus to absorb a neutron and subsequently
fission or not fission. (In the latter case, a gamma ray is released instead.) The probability of releasing a gamma ray or
causing the nucleus to fission can be represented by
Probability of releasing a gamma ray = σc /σa
Probability of causing a fission reaction = σf /σa
5.5 Converting the Kinetic Energy of Nuclear Particles into Thermal Energy 125
In addition, neutrons can scatter off an atomic nucleus without being absorbed. This process is known as elastic
scattering, and the elastic scattering probability can be found by taking the value of the scattering cross section σs and
dividing it by the total cross section σt
where the total cross section is defined to be the sum of the individual cross sections: σt = σs + σc + σf for all possible
reactions including capture, scattering, and fission:
Finally, because neutron cross sections are energy dependent, their values depend on the kinetic energy E of the incom-
ing neutron. Hence when referring to a particular cross section, the values for σt, σs, σc, and σf are generally written as
σt(E), σs(E) σc(E), and σf(E) when E is measured in eV. Normally, elastic scattering cross sections are small compared
to fission and capture cross sections for most nuclear fuels. However, they are larger for carbon, light water, and heavy
water. These materials, which are sometimes used to slow down fission neutrons, are also called neutron moderators.
In many reactors, the neutron moderator also serves as the coolant.
5.5 Converting the Kinetic Energy of Nuclear Particles into Thermal Energy
Once an atomic nucleus splits apart, the particles that are produced move away from the nucleus at very high
speeds. In general, these speeds are much higher than any man-made object can travel. Moreover, when the nucleus
fissions, the mass of the outgoing particles is always less than the mass of the nucleus itself. The mass difference
Δm between the mass of the original nucleus and the mass of the by-products is directly proportional to the kinetic
energy of the by-products. The kinetic energy of the by-products, including the fission products and the fission
neutrons, is then
KE = ∆mc 2 (5.3a)
where c is the speed of light. Equation 5.3a is simply an alternative form of Einstein’s equation of special relativity.
which forms the basis of the nuclear power industry in the world today. Einstein’s equation uses the concept of mass and
energy equivalence to convert matter into energy (and energy into matter). Because the speed of light is approximately
300,000,000 m/s, Einstein’s equation implies that a small reduction in the total mass Δm of the reactants can result in a
large amount of energy being produced. This concept also applies to the process of nuclear fusion, which is illustrated
in Figure 5.5. The kinetic energy of the particles produced by the fission process can then be written as
∑ ( m n v n 2 ) (5.4)
N
KE = 1
2
n =1
where N is the number of particles that are produced and vn is their velocity. This kinetic energy is equivalent to the
thermal energy ΔE that is released when mass is converted into energy. Thus, the kinetic energy of all of the particles
is given by
∑ ( m n v n 2 ) = ∆mc 2 (5.5)
N
KE = 1
2
n =1
126 Thermal Energy Production in Nuclear Power Plants
FIGURE 5.5 Einstein’s great equation E = mc2 implies that the mass m and the energy E of an object are interchangeable. It
explains how nuclear energy is produced in the interior of stars by the process of nuclear fusion, and how atomic bombs can be
produced by the process of nuclear fission. In both of these processes, a small amount of mass is converted into a large amount of
energy. The constant of proportionality between the amount of mass that is lost and the amount of energy that is produced is c2,
which is the square of the speed of light. The speed of light (about 300,000,000 m/s or 186,000 miles/s) is a very large number. In
the picture above, mass is converted into energy in the sun’s interior by thermonuclear fusion. The corona that surrounds the sun is
a glowing plasma of gas with an average temperature of several million degrees. The large plume of gas on the left is a solar flare.
Occasionally, the Earth is also hit by solar flares. (Picture provided by NASA.)
and this is the energy that eventually shows up in the core in the form of heat Q. Moreover, the first law of thermody-
namics (which we will discuss in Chapter 6) implies that the heat produced must always be conserved. We can express
this fact by writing
∑ ( m n v n 2 ) = ∆mc 2 (5.6)
N
Q = KE = 1
2
n =1
When particles such as photons are also produced, an additional term must be added to Equation 5.6 to account for the
presence of these particles. This can be done by extending the summation to read
∑ ( mn vn 2) + ∑
N M
Q = KE = 1
2 hfm = ∆mc 2 (5.7)
n =1 m =1
where M is the number of photons, h is Planck’s constant (a fundamental constant of nature), and f is their frequency
of vibration (in cycles per second). The value of Planck’s constant is approximately 6.6 × 10 −34 kg m2/s or 6.6260695 ×
10 −34 J s = 4.1356675 × 10 −15 eV s. Thus, a photon such as a 1 MeV gamma ray has a vibrational frequency of f = E/h =
1 × 10 6 eV/4.1356675 × 10 −15 eV s ≅ 2.42 × 1020 Hz. In other words, the kinetic energy is directly proportional to its
frequency.
where the variable Q represents the amount of kinetic energy that is released. In the reactions illustrated by Equation 5.8,
the value of Q varies between 190 and 220 MeV and the superscripts ′, ″, ‴, ″″, etc., are used to indicate that the same
amount of kinetic energy is NOT released in all of the reactions. In fact, the value of Q depends on the fission products
that are produced as well as the number of fission neutrons that escape from the reaction. The average amount of kinetic
energy that is released from common nuclear fuels is shown in Tables 5.1 and 5.2. Here, it is assumed that this kinetic
energy is eventually converted into thermal energy or heat. Although between 210 and 220 MeV of kinetic energy is
actually produced when a uranium or plutonium nucleus splits apart, only about 200 MeV of this energy is recoverable.
The rest, which is unrecoverable energy, is carried away by strange ghost-like particles called neutrinos. The properties
of neutrinos are explored in our companion book.*
TABLE 5.1
The Energy Released from the Thermal Fission of Uranium-235
Energy Released from the Fission of U-235 Average Energy Released (MeV)
Instantaneously Released Energy
Kinetic energy of fission fragments 169.10
Kinetic energy of prompt neutrons 4.8
Energy carried by prompt γ-rays 7.0
Energy from Decaying Fission Products at a Later Time
Energy of β-particles 6.5
Energy of delayed γ-rays 6.3
Energy released when those prompt neutrons that do not (re)produce fission are captured 6.8
Energy converted into heat in an operating thermal nuclear reactor 202.5
Energy of antineutrinos (unrecoverable energy) 6.8
Sum of all sources 209.3
* See Nuclear Engineering Fundamentals: A Practical Perspective, by R.E. Masterson, CRC Press (2017).
128 Thermal Energy Production in Nuclear Power Plants
TABLE 5.2
The Energy Released from the Thermal Fission of Plutonium-239
Energy Released from the Fission of Pu-239 Average Energy Released (MeV)
Instantaneously Released Energy
Kinetic energy of fission fragments 175.8
Kinetic energy of prompt neutrons 5.9
Energy carried by prompt γ-rays 7.8
Energy from Decaying Fission Products at a Later Time
Energy of β-particles 5.3
Energy of delayed γ-rays 5.2
Energy released when those prompt neutrons that do not (re)produce fission are captured 7.1
Energy converted into heat in an operating thermal nuclear reactor 207.1
Energy of antineutrinos (unrecoverable) 7.1
Sum of all sources 214.2
In this reaction, Cesium-137 transforms itself into Barium-137, and thus, beta decay can be used to convert one
element into another. The electrons that are released by beta decay are then called beta particles (or beta rays) to
distinguish them from ordinary electrons that are located in the electron cloud. In general, the energy of a beta
particle is much higher than that of an orbital electron. Extremely heavy atoms can also become energetically
unstable and spontaneously emit ionized helium nuclei to return to a stable state. These ionized helium nuclei are
sometimes called alpha particles. Polonium is an example of a heavy element that emits large numbers of alpha
particles. Another element that emits alpha particles is americium, which is sometimes used in household smoke
detectors. Alpha decay therefore requires two protons and two neutrons to be released from an atomic nucleus,
which is shown as follows:
Normally, alpha decay is limited to very heavy elements having atomic weights of 110 or more. Alpha decay does NOT
occur in lighter elements because they are energetically stable. The alpha particles that are released do not travel as far
as the beta rays or gamma rays do. Their kinetic energies range between 2 and 6 MeV, and they are almost always depos-
ited in the fuel rods where they are born. Alpha decay is a by-product of the process of neutron capture. The actinides,
which are heavy isotopes created by the process of neutron capture, frequently emit alpha and beta particles. Americium,
which is used in household smoke detectors, is an example of an actinide that emits alpha particles. Other radioactive
isotopes of neptunium, berkelium, and californium also fall into this category.
FIGURE 5.6 Comparing X-rays and gamma rays based on their kinetic energy alone. Gamma rays have much higher kinetic
energies than dental X-rays, microwaves, and visible light waves do. Gamma rays can have kinetic energies between 100 KeV and 100
MeV, and at the higher end of this energy range, their wavelength is about 1 × 10 −10 m, which is about the size of an atomic nucleus.
130 Thermal Energy Production in Nuclear Power Plants
Wigner–Way equation. We will discuss the implications of this equation later in this chapter. Finally, the quantity Q
of decay heat that is produced is equal to the difference between the masses of the individual atoms or particles before
they decay and the masses of the individual atoms or particles after they decay. From Einstein’s equation E = mc2,
the magnitude of the decay heat is then
Q= (∑ m before − ∑ m ) c (5.11)
after
2
where c is the speed of light. In the SI system of units, the speed of light is about 300,000,000 km/s. Hence, Einstein’s
equation implies that a very small amount of matter can be converted into a very large amount of heat.
where r is the distance (as the crow flies) from the point of particle emission. This relationship is different from the
energy attenuation rate predicted by the inverse square law in a vacuum, which is as follows:
where no exponential attenuation occurs. The value of λ that appears in Equation 5.12 determines the rate of
energy deposition from the particle. Normally, it is both particle and material dependent. Here, the Greek symbol λ
(lambda) is called the attenuation coefficient. The attenuation coefficient has the units of inverse length, and for a given
substance, it may also be energy dependent. That is, λj = λj(i, E), where i is the type of particle (α, β, or γ) and j is the
material type. Thus,
As Equation 5.14 shows, alpha particles are attenuated more quickly than beta particles, and beta particles are a ttenuated
more quickly than gamma rays (see Figure 5.7). Representative values of their attenuation coefficients are shown in
Table 5.3. Notice that the attenuation coefficients have the units of inverse length. Inside a reactor core, the attenuation
coefficient is normally quoted in inverse centimeters (cm−1) although inverse meters (m−1) are sometimes used. The ele-
ment lead (symbol Pb), which is also used as a shielding material, has a much larger attenuation coefficient than other
materials such as aluminum, zirconium, or stainless steel. Reinforced concrete also has a large attenuation coefficient
because it is so dense. Thermal energy is deposited at a faster rate by alpha particles and beta particles than it is by
gamma rays. There are several ways in which their kinetic energy is transferred to ordinary matter. For photons, this
occurs through three distinct processes called Compton scattering, photoelectric effect, and pair production. These
processes are discussed in our companion book*, and so we will not elaborate upon them here. However, these processes
5.12 Energy of Fission Neutrons 131
FIGURE 5.7 Different shielding materials and how they can be used to attenuate alpha, beta, and gamma rays.
TABLE 5.3
The Linear Attenuation Coefficients for Common Shielding Materials That Can Be Found in Nuclear
Power Plant (cm−1)
The Linear Attenuation Coefficients for X-Rays and Gamma Rays for Common Shielding Materials (cm−1)
Photon Energy (MeV) Water Aluminum Lead Steel Concrete
0.1 0.025 0.1007 58.6809 1.7707 0.0978
0.2 0.300 0.0742 9.2773 0.3848 0.0679
0.5 0.033 0.0286 0.0994 0.2314 0.0396
1.0 0.031 0.0772 1.1232 0.2054 0.0930
2.0 0.026 0.0626 0.3311 0.1818 0.0562
5.0 0.020 0.0518 0.3978 0.1786 0.0456
10.0 0.016 0.0490 0.5085 0.1968 0.0416
Source: Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering Prentice Hall, Upper Saddle
River, NJ (2001).
Note: The attenuation coefficients for alpha particles are about 1,000 times as large, and the attenuation coef-
ficients for beta particles are between 10 and 100 times as large. This implies that alpha particles and
beta particles are attenuated much more rapidly than gamma rays are.
play an important role in the field of radiation protection because they determine how quickly an X-ray or gamma ray
is absorbed.
they then transform themselves into what are known as thermal neutrons. If the velocity of a neutron is measured in
m/s, and its kinetic energy is measured in MeV, then its energy E can be found from its velocity v using the following
equation:
which the reader may recognize as the classical expression for the kinetic energy of a free particle. Inverting this
e quation, we also see that its velocity v is related to its kinetic energy E by
when E is measured in MeV. The velocity is shown as a function of the kinetic energy in Figure 5.8. In nuclear science
and engineering, the process of a fission neutron slowing down from fast energies to thermal energies can be described
by an elegant mathematical framework that has become known as neutron slowing down theory. For many neutrons,
between 20 and 1,000 collisions are required for a neutron to lose enough kinetic energy to enter the thermal energy
range, where E ≤ 1 eV. Normally, light elements such as hydrogen do a better job of slowing down neutrons than heavy
elements do. Various equations are available to describe their energy loss as a function of the number of elastic collisions
in which they participate. For head-on elastic collisions, the energy loss per collision is given by
FIGURE 5.8 The velocity of a neutron in a reactor core as a function of its kinetic energy.
5.12 Energy of Fission Neutrons 133
where E is the initial kinetic energy of the neutron, E′ is the final kinetic energy, α = (A − 1)2/(A + 1)2, and A is the atomic
mass (i.e., the number of protons and neutrons in the nucleus with which the neutron collides).
For glancing elastic collisions, the corresponding equation for the neutron energy loss is
which is a function of the neutron scattering angle θ (see Figure 5.9). Again, α = (A − 1)2/(A + 1)2, and A is the mass of
the target nucleus in AMU. The larger the scattering angle θ, the larger the energy loss becomes. The energy loss is
greatest when θ = 180°, and in this case, ΔE reduces to ΔE = (1 − α)E, which of course is Equation 5.18. Thus, head-on
collisions can be thought of as collisions where the neutron recoils in exactly the opposite direction from which it came.
It can also be seen from Figure 5.9 that θ = θ1 + θ2 is the total scattering angle between the outgoing neutron and the
nucleus after the collision. Example 5.2 compares the energy loss of a neutron in a single head-on elastic collision with
a hydrogen atom (where the scattering angle is 180°) and in a glancing collision with a carbon atom (where the average
scattering angle is 86°). Clearly, hydrogen is more effective than other elements are at slowing down high-speed neutrons
to low-speed ones.
FIGURE 5.9 A picture illustrating the elastic scattering of a neutron with an atomic nucleus. In ahead-on collision, the scattering
angle θ1 of the neutron is 180° because the neutron recoils in exactly the opposite direction from which it came. In a glancing colli-
sion, the neutron can have any scattering angle between 0° and 180°. The larger the scattering angle, the greater the energy loss.
134 Thermal Energy Production in Nuclear Power Plants
E = 3/2 kT (5.20)
where k is Boltzmann’s constant having the value of k = 1.38 × 10 −23 J/°K or 8.62 × 10 −5 eV/°K. Here, T is expressed in
°K. The most probable kinetic energy E of a particle in the distribution is then given by
E = 1/2 kT (5.21)
which is exactly one-third of its average kinetic energy. Hence, as the temperature of the core increases, the kinetic
energy of the neutrons in the core increases as well. According to Equations 5.20 and 5.21, both the average kinetic
energy and the most probable kinetic energy are directly proportional to the absolute temperature. Note that the aver-
age kinetic energy of a particle does NOT depend on the particle’s mass—or in fact, any of its other properties. It only
depends on the absolute temperature of the particle. At thermal energies, the most probable speed of a neutron is about
2,200 m/s, or about 1.4 miles/s. This is faster than any man-made vehicle except the NASA space plane, whose picture
is shown in Figure 5.13.
FIGURE 5.10 The percentage of the kinetic energy of a neutron that is converted into thermal energy as a function of the neutron
scattering angle for several important elements such as hydrogen, carbon, and iron.
5.14 Thermal Energy Produced in Nuclear Fuel Rods 135
FIGURE 5.11 The shape of a Maxwell–Boltzmann distribution as a function of a particle’s average temperature T.
FIGURE 5.12 James Maxwell and Ludwig Boltzmann—the inventors of the Maxwell–Boltzmann particle energy distribution.
FIGURE 5.13 A picture of the NASA space plane, which has about the same speed as a thermal neutron in a reactor core.
(Pictures provided by NASA—see nasa.gov.)
FIGURE 5.14 A picture of a nuclear fuel rod from a PWR and its internal structure.
although these numbers have tended to increase over time as uranium has become more expensive and fuel assemblies
have been allowed to remain in the core longer. The size of a fuel pellet is dictated by the rate at which heat must be
removed from the fuel rod. The fuel pellets are then inserted into the rods, which are stacked to the correct height,
welded shut, filled with a pressurized gas, and sealed. Not all of the fuel rods or fuel pellets have the same size, and their
actual dimensions can vary considerably from one reactor type to the next. In fact, they can even vary from one PWR
or BWR to the next. Fuel pellets used in PWRs and BWRs have an average diameter of about 0.9 cm. PWR fuel pellets
are slightly smaller than BWR fuel pellets because of the respective differences in the power densities between the two
designs. The sizes of these pellets are compared in Table 5.4. In general, the higher the power density of the core is, the
smaller the fuel pellet must be to safely remove the heat.
5.15 The Fuel Rod Cladding 137
TABLE 5.4
The Properties of Nuclear Fuel Pellets Used in Different Types of Reactors
The parameters shown in Table 5.4 can be considered to be “typical” in the sense that they represent the industry
average over many different designs. In the former Soviet Union, the dimensions of the fuel pellets were slightly different
than those shown here.
FIGURE 5.15 A cross-sectional view of two different types of nuclear fuel rods: a solid fuel rod on the left and an annular fuel rod
on the right.
138 Thermal Energy Production in Nuclear Power Plants
Unfortunately, the zirconium used in the cladding absorbed too many neutrons because it contained the element
hafnium—which is a strong thermal neutron absorber. These cladding materials absorbed more neutrons than was
ideal from a design perspective, and so as more advanced alloys such a Zircaloy-2 or Zircaloy-4 were developed. In
later designs, it then became possible to shrink the thickness of the cladding to improve neutron economy as well.
Thinner cladding means that fewer neutrons are absorbed. The latest generation of fuel rods now have cladding
with an average thickness of about two-thirds of a m illimeter, and the gap between the cladding and the fuel now
has an average thickness of about 1/10th of a millimeter (0.10 mm). Consequently, the total distance between the
surface of the fuel and the surface of the cladding is now about 0.70 mm (0.60 + 0.10 mm) in most light water reac-
tors. BWRs manufactured primarily by the General Electric Corporation tend to have the highest average cladding
thickness (0.66–0.88 mm), whereas CANada Deuterium Uranium (CANDU) reactors and advanced gas reactors
(AGRs) tend to have the lowest (about 0.40 cm). PWRs and liquid metal fast breeder reactors (LMFBRs) with an
average thickness of about 0.57 mm tend to fall somewhere between these two extremes. The thickness of the clad-
ding used in each reactor type is shown in Table 5.5. Notice that the thickness of the other major cladding material
(stainless steel) has remained about the same—even though it absorbs more neutrons and can have a lower melting
point. In the design and manufacture of nuclear fuel rods, the goal is to create as many fuel rods as possible that
are exactly the same as each other to that they can be put together into larger groups of rods called fuel assemblies.
Sometimes a group of these rods is also called a rod bundle. We will discuss how a fuel assembly works in more
detail in Section 5.16. Like classical part suppliers, reactor vendors will supply nuclear fuel and fuel rods to a utility
in any form that is required.
Cladding materials used today have a very high thermal conductivity, and as a rule of thumb, they also have
good resistance to the radiation produced by the fuel. Zircaloy is a much better material than stainless steel is in
this regard. It has a very low absorption cross section for thermal neutrons, and because of this, it does not absorb
as many neutrons. There are two versions of Zircaloy in widespread use today: Zircaloy-2, which is used primarily
in BWRs, and Zircaloy-4, which is used primarily in PWRs. The properties of these two alloys are compared in
Table 5.6. Notice that Zircaloy-4 has a slightly more iron and chromium than Zircaloy-2. It also has much less nickel.
The amount of tin and zirconium in both of these alloys is about the same. Most reactors use cladding that is between
0.40 and 0.70 mm thick. An excellent article discussing the physical properties of these two alloys is available at the
following URL:
http://web.ornl.gov/info/reports/1962/3445605716311.pdf.
The melting points of Zircaloy-2 and Zircaloy-4 are essentially the same. The gap between the fuel pellets and the inner
surface of the cladding is then filled with pressurized helium to improve the flow of heat across the gap. The internal
TABLE 5.5
The Cladding Materials and Thicknesses Used in Different Reactor Designs
TABLE 5.6
The Composition of Various Alloys of Zircaloy, Which Is a Common Cladding Material Used for Nuclear Fuel Rods
pressure of the helium inside of the fuel rod is normally set equal to the ambient pressure of the reactor environment
in which the fuel rods will be deployed. In the case of a BWR, this turns out to be about 1,050 PSIA, and in the case of
a PWR, it turns out to be about 2,250 PSIA. Although the size of the gap can vary somewhat from one reactor design
to the next, most reactors have a fuel–cladding gap of about 0.10 mm. (The industry average gap in modern PWRs and
BWRs today tends to be about 0.09 mm.) In reactor design, the gap is about one-quarter to one-third of the thickness of
the cladding. The smaller the gap, the easier it is to remove heat from the fuel when the reactor is operating. However, if
the gap is too small, the chance of the cladding being breached may be increased because of the mechanical interaction
between the cladding and the fuel. Over time, the design of most fuel rods has been changed to minimize these compet-
ing effects as the burnup of the fuel is increased. Sometimes the gap is called the fuel–cladding gap.
FIGURE 5.16 A cross-sectional view of a modern PWR fuel assembly manufactured by Westinghouse. Today most PWR fuel
assemblies have either 17 or 19 rods per side.
140 Thermal Energy Production in Nuclear Power Plants
the rod to hold the fuel pellets in position, and then, the fuel pellets are placed on top of it. The fuel rods are designed
so that an additional space, or fission gas plenum, about 2 ft in length, is provided at both the top and the bottom of the
rods to allow the gaseous fission products to accumulate. A picture of a fuel rod used in a PWR is shown in Figure 5.14.
The fuel pin springs hold the fuel pellets in contact with each other, and serve to reduce the axial stress on the rod.
Notice that there is one spring, and one plenum or spacer on each end of the rod (top and bottom) for fission gases to
accumulate. The rod is then evacuated and filled with a conductive gas such as helium. This gas helps to improve the
heat transfer coefficient across the fuel–cladding gap, and helps to prevent corrosion in the event any H2O is produced.
In PWRs, the fuel rods are loaded with fuel pellets to produce an active core height of between 3 and 4 m. In PWRs,
they are then pressurized to about 2,200 PSI (1.55 × 107 Pa), and in BWRs, they are pressurized to about 1,000 PSI
(about 6.9 × 106 Pa). The fuel rods are able to withstand these pressure levels with little difficulty. The rod ends are then
welded shut, and the fuel rods are placed in a vacuum chamber to test for leaks. If a rod successfully passes this “leak
test,” then it is etched in a nitric–hydrofluoric acid bath and cleaned with high-pressure steam. This produces an outer
surface film that is corrosion resistant, and has a very clean finish. The fuel rods are then ready to be assembled into fuel
assemblies. Interspersed with the fuel rods are control rods, which are usually made up of cadmium or boron, to absorb
excess neutrons that are generated by the fuel rods, and to control the speed and strength of the nuclear chain reaction.
A considerable amount of effort goes into the design of a nuclear fuel assembly, and a typical fuel assembly can weigh
several tons. There are about as many different fuel assembly designs as there are types of commercial airplanes in
operation today.
The major difference between these designs is in the spacing of the fuel rods and the location of the control rods.
There are also differences due to the overall geometry (hexagonal, square, or circular), and whether or not the fuel
assemblies are surrounded by a thin tube of metal called a “can” or a shroud. Figure 5.16 shows the cross sectional
view of a modern PWR fuel assembly. There are many variations on this particular design, including the number of fuel
rods, and the location and number of the control rods. Normally, PWRs have fuel rods arranged in rectangular arrays,
with 15 × 15 or 17 × 17 rods per side. A good example of a fuel assembly of this type ready to be loaded into a PWR
core is shown in Figure 5.17. In another class of reactors, called LMFBRs, the fuel assemblies are hexagonal rather than
rectangular in shape (see Figure 5.18). We will not discuss these in detail here except to say that it is theoretically pos-
sible to get a higher power output out of a hexagonal-shaped assembly than a rectangular one because of the way that
the rods are packed together. This makes fuel assemblies with hexagonal arrays a desirable thing to have in high power
applications. BWR fuel assemblies are similar to PWR fuel assemblies except that they are smaller, and they have “cans”
around them to separate them from one another from a thermal-hydraulic and neutronic perspective. The number of fuel
pins per assembly is also considerably smaller, and the surrounding cans are usually made up of stainless steel.
FIGURE 5.17 A picture of a modern PWR fuel assembly ready to be loaded into a reactor core. (Courtesy of the U.S. NRC.)
5.20 Fuel Rods in Hexagonal Fuel Assemblies 141
TABLE 5.7
Typical Fuel Assembly Design Parameters by Reactor Classification
Reactor Fuel Assembly Fuel Rods per Fuel Assembly Fuel Assembly
Type Manufacturer Characteristics Assembly Geometry Orientation
BWR General Electric 8–9 rods per side 64–81a Square Vertical
PWR Westinghouse 13–17 rods per side 169–289a Square Vertical
CANDU Atomic Energy of Canada 5–7 concentric 28–45a Cylindrical Horizontal
rings
AGR National Nuclear Concentric rings 190a Cylindrical Vertical
LMFRB Westinghouse 7–9 rods per side 169–271a Hexagonal Vertical
a Typical range—exact numbers can vary from those shown.
FIGURE 5.19 A modern BWR fuel assembly with its surrounding “can” and no control blade.
TABLE 5.8
The Number of Fuel Rods That Are Theoretically Possible in a Square Fuel Assembly
Rows of Rods Interior Rods Side Rods Corner Rods Total Rods
1 1 0 0 1
2 0 0 4 4
3 1 4 4 9
4 4 8 4 16
5 9 12 4 25
6 16 16 4 36
7 25 20 4 49
8 36 24 4 64
9 49 28 4 81
N (N − 2)2 for N > 1 4 (N − 2) for N > 1 4 for N > 1 N2
TABLE 5.9
The Number of Fuel Rods That Are Theoretically Possible in a Hexagonal Fuel Assembly as a Function of the “Ring Count”
Rings of Rods Interior Rods Side Rods Corner Rods Total Rods
1 7 0 0 7
2 7 6 6 19
3 19 12 6 37
4 37 18 6 61
5 61 24 6 91
6 91 30 6 127
7 127 36 6 169
8 169 42 6 217
9 217 48 6 271
n ∑n 6n + 1 for n > 1 6 (Nrings − 1). for n > 1 6 for n > 1 N = ∑n 6n + 1
assembly can contain, we simply find the total number of “rings” n that are used in the fuel assembly. For each ring
number, we multiply by 6, and then add 1 to the total. The correct expression to calculate the total number of rods is
then N RODS = ∑ n 6n + 1, where n is the ring number and N is the total number of fuel rods. The exact numbers of rods
are shown in Table 5.9. By inspection, the total number of “corner” fuel rods is always equal to 6, and the number of
side rods is six times the number of rings – 6, or 6(Nrings − 1). The specific formulas for each type of rod are also shown.
Table 5.8 shows the total number of fuel rods that are theoretically possible in this type of fuel assembly as a function of
the number of rings. We can summarize what we have just said with the following expression:
5.21 CANDU Reactor Fuel Assemblies 143
∑
N
N RODS = 6n + 1 (5.22)
n =1
where N is the total number of rings. For reasons that we will discuss later, it turns out that a hexagonal fuel assembly
will have about 16% more fuel rods for the same volume of space than a square fuel assembly will. The total number of
fuel rods for 8 and 9 rings is 217 and 271, respectively. By comparison, a square fuel assembly 17 rods wide and 19 rods
wide will have 289 and 361 fuel rods, respectively. However, its size is also much greater.
FIGURE 5.20 One of the fuel bundles from a CANDU pressurized heavy water reactor on the left and the calandria in which they
are placed on the right. The pressure tubes operate at about two-thirds the pressure of a conventional PWR or about 10 MPa.
144 Thermal Energy Production in Nuclear Power Plants
remarkable achievement because it can run on natural uranium alone. The uranium is again mixed with oxygen to pro-
duce a uranium dioxide or UO2 fuel pellet. UO2 is a complex ceramic material with a very high melting point (~2,865°C).
In fact, its melting point is similar to the melting point of the heat shield used on the U.S. Space Shuttle. The fuel pellets
in a CANDU reactor rarely melt because the power density in the core is so low. The volumetric power density is com-
pared to that of other common reactor types in Figure 5.26.
where ϕ is the neutron flux, D is the neutron diffusion coefficient, ∑f is the macroscopic fission cross section, and ∑a
is the macroscopic absorption cross section. The neutron flux and the diffusion coefficient are defined in many reactor
physics books, and in most cases, they have the units of neutrons/cm2/s and cm, respectively (other types of units are
sometimes used). The solutions to Equation 5.23a are then used to determine the global power distribution. The power
generation rate at a function of position is
FIGURE 5.21 The heat generation rate in the core is not uniform in either the radial or the axial directions. Normally, more heat
is produced in fuel assemblies with a higher enrichment than in fuel assemblies with a lower enrichment. Examples of how this heat
generation rate varies in the radial direction are shown.
5.23 Ways of Measuring the Core Power 145
FIGURE 5.22 The power profile in a homogeneous reactor core that can be derived from the fast and thermal neutron flux.
where ∑f is the macroscopic fission cross section having the units of cm−1, and qFISSION is the number of joules pro-
duced per fission (~3.2 × 10 −11 joules). When ∑f, ∑a, and D are independent of position, then it is possible to solve
Equation 5.23a in simple geometries such as rectangles, cylinders, and spheres. A core in which ∑f, ∑a, and D are
independent of position is called a homogeneous core. A core that is surrounded by a vacuum is called a bare core, and
a core that is surrounded by water, steel, or some other r eflective material is called a reflected core. In general, the power
profile is flatter in reflected cores than it is in bare cores, and this is because the reflective material reduces the neutron
leakage rate by reflecting some of the outgoing neutrons back into the core. An example of what the power shape looks
like with and without a reflector is shown in Figure 5.22. In homogeneous cores, the power profile is directly propor-
tional to the shape of the neutron flux ϕ.
In Equation 5.24, the heat generation rate dQ/dt is a function of position. However, during normal reactor operation,
it may also be a function of time. Normally, the heat generation rate is quoted in three different ways (see Table 5.10):
☉☉ The volumetric heat generation rate q‴(x, y, z) = P(x, y, z)/Volume
☉☉ The surface heat flux q″(x, y, z) = P(x, y, z)/Surface area
☉☉ The linear heat generation rate or linear power q′(z) = P(x, y, z)/Length
From Equation 5.24, the total core power is then seen to be
P=
∫ dQ(x, y, z)/ dt dV (5.25)
v
where the integral is performed over the entire volume of the core. There are two other ways in which the core power
density can be defined. They are
1. The core volumetric power density Q ′′′
2. The core specific power density Q ′′′
F
146 Thermal Energy Production in Nuclear Power Plants
FIGURE 5.23 A control volume used to perform an energy balance in a reactor fuel assembly.
TABLE 5.10
The Symbols for the Volumetric Power Density, the Surface Heat Flux, and
the Linear Power Density
where VCORE is the total volume of the core and MFUEL is the mass of the fuel in the core. Hence, the volumetric core
power density is based on the volume of the core and the specific core power density is based on the mass of the fuel in
the core. The units for each of these parameters are shown in Table 5.11. The volumetric core power density is normally
expressed in kW/L.
TABLE 5.11
The Symbols and Units of Measurement for the Core Power Density
and the Core Specific Power Density
TABLE 5.12
A Comparison of the Power Generating Parameters for a
Westinghouse PWR and a General Electric BWR-5
FIGURE 5.24 A picture comparing the cores of a commercial PWR and BWR. For the same thermal power output, a BWR has
about four times as many fuel assemblies as a PWR does. Hence, the average thermal power output per assembly is about ¼ as great.
A PWR fuel assembly has an average thermal power output of about 15 MW, and a BWR fuel assembly has an average thermal
power output of about 4 MW. However, these numbers can vary depending upon the actual design.
148 Thermal Energy Production in Nuclear Power Plants
FIGURE 5.25 The radial power profile in the core of a BWR-6 at the BOL.
The power generation parameters are compared in Table 5.12. All reactor fuel assemblies do not have the same amount
of fissionable material in them because different enrichments must be used to flatten the radial power profile. In general,
fuel assemblies have three or four different enrichments, and these enrichments are determined by what is known as the
core loading pattern or the fuel management scheme. The goal of nuclear fuel management is to maximize the amount
of energy that can be extracted from the fuel and to keep the average fuel temperature as high as possible. Thus, power
generation rates can vary considerably from one fuel assembly to the next. Also, some fuel assemblies will generate
more power than others at certain points in their life cycle. Most of the time, a fuel assembly is capable of generating the
most heat when it is first inserted into the core. An example of the radial power profile in a large BWR-6 core is shown in
Figure 5.25. Commercial PWRs have similar power profiles, and the shape of these power profiles also tends to change
as a function of burnup. Normally, the power differences are the greatest when the fuel is fresh, and these differences
tend to diminish as the fuel is burned because the “hotter” fuel assemblies tend to burn more quickly than the cooler
ones. As long as a core is relatively large and there is not a lot of neutron leakage, it is possible to assume that the number
of neutrons leaking into a fuel assembly is about the same as the number of neutrons leaking out of a fuel assembly.
In this case, a fuel assembly can be considered to be isolated from the other fuel assemblies in its immediate vicinity.
This is normally a good assumption except for the fuel assemblies that are found along the periphery. The power levels
are then determined primarily by the local U-235 concentration. The local U-235 concentration is directly proportional
to the enrichment of the fuel, which is defined by
Note that for very small enrichments (typically less than a few percent), Equation 5.28 reduces to
Fuel assemblies with higher enrichments tend to run hotter and generate more power than fuel assemblies with lower
enrichments. In thermal water reactors (excluding the CANDU PHWR), the average enrichment of a fresh fuel assembly
is between 3% and 5%. However, military reactors, such as those used in submarines and nuclear aircraft carriers (see
Chapter 4), can have average enrichments as high as 95%. Because of their higher enrichments, these reactors only have
to be refueled once every 30 or 40 years.*
FIGURE 5.26 Comparison of the volumetric power densities in modern PWR, BWR, and LMFBR cores. The differences in the
power generation rate are attributable primarily to differences in the enrichment of the fuel and the geometrical arrangement of
the core. Note that LMFBRs use a hexagonal lattice, while PWRs and BWRs use a square one.
where R is the radius of the rods. Equation 5.29 can also be written as
′′′
Q SQUARE = q′ P2
where q′ = πR2 q′′′. In other words, the volumetric power density is inversely proportional to the square of the pitch
and directly proportional to q′ for a cylindrical fuel rod. For a hexagonal lattice, the corresponding equation for the
volumetric power density is
′′′ = q ′
Q HEX 3/2P 2 = q ′ 0.866 P 2 (5.31)
Equation 5.31 shows that the volumetric power density is about 15.5% higher for a hexagonal lattice than it is for a
′′′ Q SQUARE
square lattice with the same fuel rod pitch because Q HEX ′′′ = 1.0/0.866 ≅ 1.155. A square lattice and a triangular
(or hexagonal) lattice are shown in Figure 5.27. Square and hexagonal lattices are the two most common configurations
of fuel rods in commercial power reactors, although research reactors can also have additional configurations, such as
those shown in Figure 5.27.
FIGURE 5.27 Examples of how the fuel rods are arranged in testing and research reactors. In these reactors, the fuel assemblies
are not necessarily square or hexagonal in shape. (Courtesy of Wikipedia.)
FIGURE 5.28 The radial power profile in a typical PWR fuel assembly as a function of burnup. When the fuel is fresh, the radial
power peaks are higher and the peak-to-average power ratio is between 1.2 and 1.3. When the fuel is burned, the power peaks become
lower, and at EOL, the peak-to-average power ratio is closer to 1.1. The fuel assemblies in this case are approximately 20 cm across.
152 Thermal Energy Production in Nuclear Power Plants
5.28 Using Burnable Poisons to Flatten the Power and Temperature Profile
Burnable poisons are materials used by reactor vendors to suppress local power peaks when the fuel rods are first
inserted into the core (see Figure 5.29). Then, as the fuel is burned, these burnable poisons are designed to “burn away”
so that they do not continue to suppress the heat generation rate. A burnable poison suppresses the heat generation rate
by absorbing neutrons that would otherwise be used to split uranium or plutonium atoms apart. Normally, burnable
poisons have much larger thermal cross sections than fast ones. Consequently, they are particularly adept at absorbing
low-energy neutrons in the thermal energy range.
Some of the most common burnable poisons in use today include boron, gadolinium, and cadmium. More recently,
the element hafnium (Hf) with an atomic number of 72 has been used. A partial list of burnable poisons that can be
mixed with UO2 or PuO2 is shown in Table 5.13. Other burnable poisons such as indium have also come into vogue in
recent years. Normally, burnable poisons are used in fuel rods that are located next to control rods and instrumentation
or water channels because that is where the radial power peaks are highest. Almost all PWRs and BWRs use some
type of burnable poison to flatten the power profile. By absorbing excess neutrons, burnable poisons are also used to
control the reactivity swing of the core. Burnable poisons, and their role in core power shaping and reactivity control,
are discussed extensively in our companion books.* Essentially, they can be thought of as a poor man’s control rod.
Finally, burnable poisons can be used to improve the economics of many nuclear power plants by reducing the number
of control rods that are needed.
FIGURE 5.29 Burnable poisons and their effect on the power generation rate in a fuel rod as a function of time. To illustrate this
point, the number of free neutrons in the vicinity of the fuel rod is assumed to be constant in this case.
TABLE 5.13
A Partial List of Burnable Poisons Used in Nuclear Reactors Today
Burnable Poison Chemical Symbol Isotope Used Initial Thermal Absorption Cross Section (Barns)
Cadmium Cd N/A 2,424
Boron B Boron-10 3,837
Gadolinium Gd N/A 49,000
Hafnium Hf N/A 102
Normally, fuel assemblies are configured to minimize the value of P(x, y, z) over the expected life of the core. When
the axial enrichment is uniform, and when no control rods or axial reflectors are used, the axial power profile becomes
co-sinusoidal in shape. The power generation rate is highest at the core midplane and lowest at the top and the bottom
of the core. The axial power profile is then shown on the left-hand side of Figure 5.30. The axial power profile can be
represented by
where
H is the active core height.
z is measured from the core midplane; that is, z = 0 at H/2.
P(z) is measured in W/cm3 or kW/m3.
FIGURE 5.30 On the left: The axial power profile in a single nuclear fuel rod with a uniform enrichment when the influence of
control rods and the axial reflector can be ignored. In this case, the axial power profile at BOL is co-sinusoidal in shape. At EOL,
the axial power profile tends to be much flatter because the fuel burns faster at the core centerline than it does at the edges.
154 Thermal Energy Production in Nuclear Power Plants
For a homogenous core, the power profile in the radial direction is normally independent of the power profile in the axial
direction. Thus, if one were to normalize the axial power shape so that the average value of P(z) is 1.0, then PMAX would
be approximately π/2 = 1.57 at the core midplane and zero at the periphery. The total power peaking factor then becomes
equal to the product of the radial and axial power peaking factors:
Using representative values for PRADIAL MAX and PAXIAL MAX, the peak-to-average power ratio when the fuel is fresh (at
BOL) is then
PMAX BOL = PRADIAL MAXBOL × PAXIAL MAXBOL ≅ 1.8 × 1.6 ≅ 2.9 (5.35)
when PRADIAL MAXBOL ≅ 1.8. Thus, the hottest fuel rod may be several times hotter than the coldest fuel rod. Finally, the
value of P(z) at the core midplane is roughly 40% higher than it is one-quarter of the way up the core. This can be seen
by simply comparing the values of cos(π z/H) at z = 0 and z = H/4 (which are 1.0 and 0.707, respectively). Obviously,
these peak-to-average power ratios are not ideal because they do not allow reactor designers to extract the maximum
amount of energy from the fuel. To reduce the size of these power peaks, axial zoning schemes are sometimes used. The
efficacy of these schemes is discussed in Section 5.28.
FIGURE 5.31 An example of some axial zoning schemes that are used to flatten the core power shape.
5.32 Neutron Reflectors 155
conditions, is about 1,600°C for PWRs and BWRs at the fuel centerline. The potential benefits of axial zoning can be
increased even further by introducing burnable poisons directly into the fuel. These burnable poisons disappear as the
fuel is burned, but they do so in such a way that allows as much power as possible to be extracted from an individual fuel
assembly before it has to be replaced. This requires a choice to be made between a number of competing factors such
as the length of the refueling cycle, the average enrichment of the fuel, and the manner in which the fuel assemblies are
deployed. A different burnable poison strategy may also be required if the fuel is loaded to minimize the neutron leak-
age rate. Just as in the case of a radial zoning scheme, fuel pellets with lower average enrichments are placed near the
center of the fuel rods, and the fuel pellets with the highest average enrichment are placed near the ends. With a typical
12 ft (4 m) fuel rod, this allows the 600–800 fuel pellets within the rod to be stacked in a number of different configu-
rations. Several axial zoning schemes are compared in Figure 5.31. With the correct combinations of radial and axial
zoning, it is then possible to achieve a more acceptable peak-to-average power ratio. In practice, peak-to-average powers
ratio close to 1.6 can be obtained in the axial direction over the lifetime of the core. This requires a very sophisticated
operating plan, and in particular, great care must be taken to ensure that the control rods are inserted and withdrawn in
such a way that they do not introduce any unnecessary peaks or dips in the global power distribution. Over much shorter
periods of time, their individual movements may cause additional power spikes to occur, but these will be averaged out
over time if the fuel management strategy and the operating plan supporting it are well designed.
FIGURE 5.32 An example of the fuel assembly layout in a light water reactor (left) and in a breeder reactor (right). Breeder
reactors use hexagonal fuel assemblies to increase the power density in the core, and so these assemblies tend to be deployed in
rings about a central fuel assembly. The rods in a hexagonal fuel assembly follow a similar trend.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
156 Thermal Energy Production in Nuclear Power Plants
TABLE 5.14
Power Peaking Factors for Bare (i.e., Unreflected) Reactor Cores Having Different Sizes and Shapes
Reactor Shape Dimensions Volume (V) Maximum Flux (ϕmax) Power Ratioa (Peak/Average)
Infinite slab Thickness Lx Lx 1.58 P/(c∑f V) 1.57
3-Dimensional rectangle Lx, Ly, Lz Lx × Ly × Lz 3.88 P/(c∑f V) 3.88
Finite cylinder R, Lz πR2 × Lz 3.63 P/(c∑f V) 3.64
Infinite cylinder R πR2 0.74 P/(c∑f R2) 2.32
Simple sphere R 1.33 πR3 0.78 P/(c∑f R3) 3.29
a Peak-to-average power ratio.
Here, c is a simple constant determined by the boundary conditions.
the fuel. Reflectors consist of light neutron-reflecting materials such as water and graphite that reflect outward-directed
neutrons back into the core. Commercial nuclear power plants also use reflectors to improve neutron economy. Reflectors
flatten the power profile by reducing neutron leakage from the core, and in doing so, they also create cores with lower peak-
to-average power ratios.* Commercial power reactors use radial reflectors between 9 and 12 in. thick (about 20 cm), although
these numbers can vary slightly from one type of reactor to the next. Reflectors also reduce the amount of fuel required for
a reactor to become critical. The resulting fuel savings is sometimes called the reflector savings. Finally, reflectors can keep
certain types of harmful radiation, including gamma rays and high-energy neutrons, from escaping the core and damaging
the pressure vessel wall. Thus, reflectors actually help to prolong the life of the pressure vessel.
when the origin is located at the core centerline (i.e., at the core midplane where z = 0). The volumetric power density
Q‴ is then
Q ′′′(z) = Q ′′′
MAX ⋅ cos(π z/H) (5.37)
where H is the active core height, P = Q‴ = Q/V, the top of the core is located at z = H/2, and the bottom of the core
is located at z = −H/2. The maximum axial power PMAX is reached at the center of the core (where z = 0), and the
power density Q‴(z) then falls to zero at the edges. The axial power profile for a uniform, unreflected core is shown in
Figure 5.33. Note that the shape of the power profile is a function of the boundary conditions that are used at the edges.
In reality, the neutron flux does not fall to zero exactly at the edge of the core—where the fuel ends. Instead, it falls to
zero a short distance δ beyond the top or the bottom of the core that is called the extrapolation distance δ. Under these
conditions, Equation 5.36 can still be used to find the power profile provided that the value of H is replaced by H′ = H + δ,
where δ is the extrapolation distance beyond the core boundary. For bare homogeneous cores, the extrapolation distance
is normally the same in the radial and axial directions. Relatively accurate expressions for the extrapolation distance can
be derived from the neutron transport equation, which is discussed in our companion book.* In general, the extrapolation
distance is one to two neutron mean free paths (2–4 cm) for a fuel assembly in a thermal water reactor. The extrapola-
tion distance in fast reactors is greater. Adding a neutron reflector then results in the power profile shown in Figure 5.33.
Hence, a reflector flattens the radial and axial power profiles and reduces the peak-to-average fuel pin temperatures.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
5.34 Radial Power Profiles in Uniform, Unreflected Cores 157
FIGURE 5.33 The axial power profile in a homogeneous cylindrical core using the extrapolated boundary condition.
where R is the core radius, Lx is the width of the core, Ly is the depth of the core, and PMAX is the maximum core power
level in the direction where the power profile is to be found. To use these equations, the center of the core must be located
at (0, 0, 0) for rectangular cores, at z = 0 and r = 0 for cylindrical cores, and at r = 0 for spherical ones.
158 Thermal Energy Production in Nuclear Power Plants
TABLE 5.15
The Value of a Bessel function of the First Kind as
a Function of the Argument xo
xo Jo(xo) xo Jo(xo)
0.0 1 2.1 0.166607
0.1 0.997502 2.2 0.110362
0.2 0.990025 2.3 0.05554
0.3 0.977626 2.4 0.002508
0.4 0.960398 2.5 −0.04838
0.5 0.93847 2.6 −0.0968
0.6 0.912005 2.7 −0.14245
0.7 0.881201 2.8 −0.18504
0.8 0.846287 2.9 −0.22431
0.9 0.807524 3.0 −0.26005
1.0 0.765198 3.1 −0.29206
1.1 0.719622 3.2 −0.32019
1.2 0.671133 3.3 −0.3443
1.3 0.620086 3.4 −0.3643
1.4 0.566855 3.5 −0.38013
1.5 0.511828 3.6 −0.39177
1.6 0.455402 3.7 −0.39923
1.7 0.397985 3.8 −0.40256
1.8 0.339986 3.9 −0.40183
1.9 0.281819 4.0 −0.39715
2.0 0.223891
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
5.36 Extrapolated Power Profiles 159
FIGURE 5.34 The shape of various Bessel functions as a function of the argument xo. Note that the first-order Bessel function, Jo,
is shown in red. Its first root is at xo = 2.405.
FIGURE 5.35 Several simple geometries can be used for a reactor core, including rectangular, cylindrical, and spherical ones.
FIGURE 5.36 The power shape in a bare rectangular reactor. The peak-to-average power ratio in a bare reactor that is the same
size on all three sides is about 3.8–1.
FIGURE 5.37 The radial power profile in a bare spherical reactor. The peak-to-average power ratio in this case is about 3.3–1. In
other words, much more power is generated near the center of the core than it is around the periphery.
it is easy to arrive at the following expressions for the core-wide power profiles. Note that in the case of a spherical core,
there is no value for P(θ, φ) as long as the core is spatially homogeneous. Then, the power shape P(r) is only a function
of r. The three-dimensional power profiles that are used to find the fuel rod temperatures are then
The Three-Dimensional Power Profile in a Homogeneous Spherical Core:
1.
Of course, we have neglected the dependence of P on the polar and azimuthal angles θ and φ in Equation 5.41,
but as long as the core is uniform, the power profile is independent of these variables and is only a function of
the distance r from the center of the core. At r = 0, the value of sin (πr/R)/r is 1.0. This can be verified by doing a
Taylor Series expansion of the function sin (πr/R) in Equation 5.41.
The Three-Dimensional Power Profile in a Homogeneous Cylindrical Core:
2.
This also leads us to the conclusion that the three-dimensional power profile P(r, z) in a critical cylindrical core
of radius R and height H is
where PMAX is the highest power level in the core (located at r = 0 and z = 0) and Jo is a Bessel function of the
first kind.
The Three-Dimensional Power Profile in a Homogeneous Rectangular Core:
3.
For rectangular cores, the three-dimensional power profile P(x, y, z) is given by
where
(
P(x) = PMAX x cos π x/L x )
P(y) = PMAX y cos ( π y/L ) (5.43)
y
( ) ( ) ( )
PTOTAL (x, y, z) = PMAX cos π x/L x cos π y/L y cos π z/H (5.44)
Q ′′′
i (z)=Fi Q ′′′(z) (5.45)
where Q‴(z) is the axial power profile. In general, the value of Fi can be greater than 1.0, less than 1.0, or equal to 1.0. If it
is greater than 1.0, the fuel assembly is a hotter than an average fuel assembly, and if it is less than 1.0, the fuel assembly
is a colder than an average fuel assembly. Normally, the radial power peaking factors follow a statistical distribution
of values such as the one shown in Figure 5.38. This distribution changes with time, and it also depends on the core
loading scheme. The power profile for a commercial reactor core is shown in Figure 5.25 (which was presented earlier).
The “hot” fuel assemblies are shown in red, and the cooler fuel assemblies are shown in yellow. Normally, the fuel pin
temperatures are proportional to the local volumetric power density.
FIGURE 5.38 Reactor fuel assemblies have a distribution of temperatures depending upon where they are located in the core and
what their average enrichment is. Normally, hotter assemblies have higher enrichments than colder ones.
5.41 How Control Rods Affect the Core-Wide Power Profile 163
FIGURE 5.39 By nonuniformly distributing the fuel from one side of a reactor to another, one can reduce the peak-to-average
power ratio and the peak fuel pin temperatures. Each zone in the reactor has a different enrichment, with the darker zones being
more highly enriched than the lighter ones. The water reflector is shown in dark blue at the edges.
temperatures) as uniform as possible in both the radial and axial directions. In general, this prolongs the life of the fuel
rods as well. In practice, there are several ways in which the core power profile can be made more uniform:
☉☉ First, the core can be surrounded by what is known as a neutron reflector (see Section 5.30). In reflected cores,
the neutron flux does not fall off as dramatically at the edge, and because of this, less fuel is required to achieve
a critical mass. (The reflector may also make it possible to build a more compact core.)
☉☉ Second, the enrichment of the fuel can be varied in both the radial and axial directions. This can be done by
reducing the U-235 concentration in the areas of high flux and by increasing the U-235 concentration in the areas
of low flux. This allows most fuel assemblies to be operated at essentially the same temperature, and therefore, the
fuel can be burned at a more uniform rate. The U-235 concentration can then be varied from one fuel assembly to
the next. This effect is illustrated in Figure 5.39.
☉☉ Finally, radial power peaks within fuel assemblies can be reduced by using what are called burnable poisons.
These materials are normally used wherever there is a large thermal flux peak. To be completely effective,
burnable poisons must be distributed in such a way that the product of the neutron flux ϕ(x, y, z) and the
macroscopic fission cross section ∑f(x, y, z) is as uniform as possible. The local power generation rate is then
given by P(x, y, z) = qFISSION ⋅ ∑f(x, y, z)·ϕ(x, y, z), where qFISSION ~ 3.2 × 10 −11 joules/fission).
Finally, more uniform fuel pin temperatures can be achieved by placing lower enriched fuel near the center of the core
(where the neutron leakage rate is low) and higher enriched fuel along the periphery (where the neutron leakage rate is
high). This optimizes the amount of energy that can be extracted from the fuel, and it also allows more fuel assemblies
to remain in the core for longer periods of time.
FIGURE 5.40 The fuel assemblies and control rods in a BWR (left) and in a PWR (right) directly compared.
Generally speaking, BWR control rods are larger and thicker than their PWR counterparts. However, most PWRs have more
control rods than BWRs do. Normally, the control rods in PWRs look like the fuel rods (i.e., they are cylindrical in shape).
However, instead of being stationary, they are designed to move up and down in what are known as control rod guide tubes.
When a control rod is withdrawn from a guide tube, the tube is then filled with water. Both the radial and axial temperature
profiles are affected by how the control rods are moved. Normally, control rods depress the power level in their immediate
vicinity, which happens to be a distance of several neutron mean free paths (about a couple of centimeters). At the BOL (see
Figure 5.41), the excess reactivity is very high, and the control rods must be fully inserted into the core in order to drive the
net reactivity to zero and keep the reactor under control (see the top of the picture to see what the power profile looks like in
this case). Then as the fuel is burned, this excess reactivity is lost, and the only way that the reactor can continue to operate is
to withdraw the control rods from the core (refer to the bottom of the picture to see what the axial power profile looks like in
this case). In other words, the axial power profile in a real PWR peaks near the bottom of the core at BOL, and then as the
control rods are withdrawn, the power peak moves much closer to the top of the core as EOL is reached. Over time, more
fission reactions occur in the hottest fuel assemblies and the peak-to-average power ratio becomes progressively smaller as the
fuel in these hot assemblies is depleted. Eventually, the control rods have to be completely withdrawn from the core to keep the
core in a critical state. Then, the plant has to be shut down for refueling. Another way to demonstrate this effect is to consider
what happens to a bare cylindrical core when we move the control rods up and down very quickly, but when we rely on another
common reactivity-control mechanism (like soluble boron) to keep the reactivity at zero. In this case, the solbor concentration
changes much more quickly than the fuel can be burned, and the number of Uranium-235 atoms can be considered to be con-
stant while the solbor concentration changes. Then, the shape of the axial power profile resembles that shown in Figure 5.42.
In PWRs, solbor is used in addition to control rods to keep the reactor in a critical state. Solbor is a form of soluble
boron that is frequently represented by the chemical symbol B(OH)3 or H3[BO3] when it is in solution. The solbor
concentration is relatively uniform in a PWR core, and because of this, it cannot be used to perform power shaping
functions.* Solbor is not used in BWRs because it plates out on the surface of the fuel rods when the coolant boils.
Thus, BWRs rely on control rods to both shape the power and control the reactivity, while PWRs use solbor for gross
reactivity control. Initially, the flux shape (and hence the axial power profile) is asymmetrical and peaks very close to the
bottom of the core. Then, when the control rods are fully withdrawn, it becomes much more symmetrical and resembles
the cosinusoidal flux shape (and axial power profile) that we discussed previously. The rest of the time, it tends to have
a shape that falls somewhere between these two extremes. It is then the responsibility of the reactor designer to dem-
onstrate to the NRC that the maximum temperature of the fuel and the temperature of the cladding never exceed their
recommended design limits. Typically, this analysis has to be performed at the BOL and at the EOL since these define
the most limiting cases.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
5.42 Power Fluctuations in Steady-State Cores 165
FIGURE 5.41 The effect of the control rods on the axial power profile in a PWR as the fuel is burned. The top figure shows the
power shape at the BOL, and the bottom figure shows the power shape at the EOL. Notice that the maximum volumetric power level
P is higher at the BOL than it is at the EOL even when the control rods are fully inserted into the core during start-up.
P =
∫ P(t) dt ∫ dt (5.47)
166 Thermal Energy Production in Nuclear Power Plants
FIGURE 5.42 The effect of control rod position on the flux shape ϕ(z) in a bare cylindrical core where the enrichment is constant.
and the integrations are performed over a period of several hours or several minutes. Here again, the value of P is
defined by
P=
∫ V
Q ′′′(x, y, z) dV (5.48)
The deviation of the instantaneous power P from the average power P is typically a few tenths of 1%, and these devia-
tions in the power level are typically oscillatory in nature (see Figure 5.43). While this oscillatory behavior occurs in
both PWRs and BWRs, it is generally more pronounced in BWRs because the coolant boils. The fuel pin temperatures
then oscillate slightly in response to small changes in the flow rate and the neutron production rate.*
FIGURE 5.43 The normal range for the reactivity ρ and the power P during steady-state conditions.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
5.46 Decay Heat from Beta Decay 167
TABLE 5.16
The Percentage of the Recoverable Energy Deposited in Each of the Major Components of a
Reactor Core
TABLE 5.17
The Actinides and Their Half-Lives
where n, p, e−, and νe are the symbols for the neutron, the proton, the electron, and the neutrino, respectively. Normally,
neutrinos do not interact with other nuclear materials, and they therefore deposit their kinetic energy somewhere outside
of the plant.† Some fission products can continue to emit beta rays for weeks or even months after the fission process
stops.
FIGURE 5.45 The blue glow from Cherenkov radiation in a spent fuel pool.
responsible for the blue glow that is frequently observed in reactor cores and spent fuel pools after a reactor has been shut
down for refueling. This light blue glow is sometimes called Cherenkov radiation, and the reason for its characteristic blue
color is discussed in our companion book.* A picture of Cherenkov radiation being emitted from a spent fuel pool is shown
in Figure 5.45. Cherenkov radiation is produced by photons that always accompany the emission of beta rays from the fuel.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
170 Thermal Energy Production in Nuclear Power Plants
FIGURE 5.46 The decay heat and radioactivity produced by high-level nuclear waste, such as the spent fuel assemblies that are
stored in a spent fuel pool. (Data provided by the IAEA.)
TABLE 5.18
Percent of the Total Reactor Power Po Released by Radioactive
Decay following Shutdowna
fuel to remain in the core for between 7 and 10 days before it is moved to a fuel storage tank. This leaves about 3 weeks
for new fuel to be loaded into the core, and for some of the partially burned fuel assemblies to be “shuffled around” or
relocated to other locations so that their burnup can continue. There are between 100 and 200 radioactive decay chains
that must be taken into account for all of the decay heat and radiation that are produced. Each decay chain contributes to
5.50 Decay Heat Removal after Shutdown 171
the total amount of heat produced in the fuel. This time-dependent heat generation is in turn a function of when the beta
and gamma rays are emitted from the isotopes in each of the individual decay chains. When the nuclear chain reaction
stops, the residual decay heat that is generated by the fuel is approximately equal to 7% of the total thermal output of
the core when it is operating at rated power. This percentage is approximately the same for all thermal water reactors,
although it varies slightly depending upon whether the reactor is fueled with U-233, U-235, or Pu-239. Needless to say,
this decay heat must be taken very seriously. For example, a 1,000 MWE thermal water reactor with a thermal output
of 3,300 MW produces about 230 MWT of decay heat when the core is initially shut down.
Even though this decay heat falls rapidly over time, the initial decay heat is still great enough to heat several thou-
sand homes. If the coolant flow to the core were to be lost during the first few days after shutdown, this excess decay
heat might be great enough to cause some of the fuel rods in the core to melt, and in an extreme case, a partial or total
meltdown of the core could occur. Table 5.18 shows the decay heat generated by a reactor that has been operating at full
power PMAX for several weeks before shutdown.
where
☉☉ Po is the reactor power before shutdown.
☉☉ to is the time of operation (in s) before shutdown.
☉☉ t is the time (in s) elapsed since shutdown
☉☉ Decay heat (t) is the power generated due to the decay of fission products in the core in the form of beta and
gamma rays.
The predictions of the Wigner–Way equation are shown graphically in Figure 5.47. Notice that it takes about 2 × 104 s
(approximately 6 hours) for the energy generated by the decay heat to fall to 1% of the steady-state core power level (Po).
It also takes about 3 × 107 s (about 1 year) for the power level to drop to two-tenths of 1% of its original value. So for all
intents and purposes, it takes at least 10 days for the power to fall to half of 1% of the original steady-state power level—
at which point the spent fuel assemblies are usually removed from the core. The overall trend is shown in Figure 5.48.
Notice that the amount of decay heat produced after a reactor is shut down does not fall exponentially with time—at
least not initially. Instead, it falls less than exponentially for the first 12 hours after shutdown because of the buildup of
some of the radioactive decay chains after the fission process ceases. After that, the slope of the power versus time curve
becomes relatively constant on a log–log plot (e.g., a straight line), and therefore, the falloff rate is essentially exponen-
tial (to the power of t−0.2) after the first 12 hours has passed.
It is also important to realize that if to is large (say 5 × 107 s), then the second term (to + t)−0.2 has very little effect on
the final outcome, and the drop-off in the power level will simply follow the first term, which is 0.0666Po/t0.2. Of course,
there is a great deal of physics involved in these simple equations, but the bottom line is that some form of active cooling
must be provided to prevent the reactor fuel from overheating for a substantial period of time after shutdown occurs.
Normally, the Wigner–Way equation is appropriate to use when a reactor is first shut down for refueling. Moreover, it
is still reasonably accurate even after the fuel has been removed from the core for the first year or two. However, after
about 500 years, the actinides (see Table 5.17) begin to generate more decay heat than the fission products do, and the
decay heat begins to fall off as roughly the inverse 0.75 power of the elapsed time rather than the inverse 0.20 power.
The decay heat generation rate in this case is shown in Figure 5.46.
172 Thermal Energy Production in Nuclear Power Plants
FIGURE 5.47 A graphic representation of the decay heat produced by the nuclear fuel after a reactor is shut down. During the first
few years, most of the decay heat is generated by the fission products. Then after about 500 years, the actinides and transuranics
generate more decay heat. However, this long-term decay heat is not modeled particularly well by the Wigner–Way equation.
This long-term decay heat falls as the inverse three-quarter power of the time, that is, P ~ t−0.75, while the Wigner–Way equation
predicts that the decay heat for shorter periods of time falls as P ~ t−0.20.
FIGURE 5.48 A typical timeline for a reactor operating and refueling cycle.
The Temporal Dependence of the Decay Heat over very Large Time Frames
Decay heat (t)~ t −0.75 (for times greater than 500 years) (5.52a)
Decay heat (t)~ t −0.20 (for times less than 100 years) (5.52b)
5.51 ANS Standards Governing Decay Heat 173
This temporal dependence is driven by the actinide concentration in the fuel rods. Hence, the magnitude of the decay
heat depends on how long the fuel has been removed from the core.
{ }
FU-239 (t, T) = E U-239 R 1 − e −λ1T e −λ1 t (5.53)
and
{
FNp-239 (t, T) = E Np-239 E Np-239 R λ1 ( λ1 − λ 2 ) (1 − e −λ T ) e −λ t − λ 2 ( λ1 − λ 2 ) (1 − e −λ T ) e −λ T } (5.54)
2 2 1 1
Here, EU239 and ENp239 are the average recoverable energy from the decay of U239 and Np239 and are equal to 0.474 and
0.419 MeV, respectively; λ1 and λ2 are the decay constants (4.91 × 10 −4 and 3.41 × 10 −6 s−1); R is the conversion factor. It
represents the number of U239 atoms produced per total fission events, evaluated at the reactor composition at the time
of shutdown. When attempting to understand the total decay heat generated from all of these energy sources changes
with time, it can be helpful to refer to the sequence of events shown in Figure 5.47. These events often parallel the way
that a commercial LWR is operated. There are usually three distinct regions in each operating cycle that have to be
taken into account in these calculations. We will assume a 1-year operating cycle in the discussion that follows. Other
common operating cycles, depending on the design of the reactor and the enrichment of the uranium fuel, can be either
TABLE 5.19
Decay Power after Shutdown as a Fraction of a Reactor’s Initial Power
Level as Predicted by the Wigner–Way Equation. In This Case, Three
Different Operating Times Are Shown
18 or 24 months. For most of the year (usually 330 days), a reactor is designed to operate at full power, which we will
call Po. Then, for the reasons discussed above, it is usually powered down to 80%–90% of normal power for a period
of a week or two. This helps to lower the amount fission products available to decay in the fuel rods (see Figure 5.48).
Finally, in the last 2 weeks of the operating year, it is usually powered down (in a linear manner), until it is finally shut
down for maintenance and refueling. The time that it is operating at full power corresponds to the time called to in the
Wigner–Way equation, which in this example is approximately 330 days. The amount of time that the reactor operates
at reduced power is called Δt, and this is an input into the second term in the equation. For a given value of Δt, the
value of F can be easily determined. Finally, the point in time where the reactor is shut down corresponds to t = 0 in all
of the aforementioned equations. After 1 h, t = 3.6 × 103 s, after 1 day, t = 8.64 × 104 s, after 1 month, t = 2.6 × 106 s, and
after 1 year, t = 3.16 × 107 s. After 1 year, the decay heat generation declines to 0.19% of the steady-state power Po. In a
2,000 MW power plant, the total power production after a year, if the fuel was undisturbed, would be about 0.0019 ×
2,000 MW = 3,800 KW. Normally, this would be enough energy to heat a large building.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Crowe, C. and Schwartzkopf, J., et al. Multiphase Flows with Droplets and Particles, CRC Press, Boca Raton, FL (2012).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
Eisberg, R. Fundamentals of Modern Physics, John Wiley and Sons, New York (1961).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Feynman, R.P., Leighton, R. and Sands, M. The Feynman Lectures on Physics, Narosa Publishing House, New Delhi (2011).
Glasstone, S. and Eedlund, M. C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Kaplan, I. Nuclear Physics Addison-Wesley, Reading, MA (1965).
Khartchenko, V. Advanced Energy Systems, Second edition, CRC Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc. New York,
NY (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi (2000).
Resnick, R. and Halliday, D. Fundamentals of Physics, John Wiley and Sons, New York (1967).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, CRC Press, Boca Raton, FL (2014)
Web References
http://www.1728.org/freqwave.htm.
http://en.wikipedia.org/wiki/Neutrino.
http://www.crystalinks.com/phila.html.
http://en.wikipedia.org/wiki/Antimatter.
http://home.web.cern.ch/topics/antimatter.
http://plato.stanford.edu/entries/equivME/.
http://en.wikipedia.org/wiki/Maxwell’s_equations.
https://www.youtube.com/watch?v=202fU9qIVK4.
https://www.youtube.com/watch?v=ChjyCR8V2Bg.
http://en.wikipedia.org/wiki/Lorentz_transformation.
Questions for the Student 175
http://en.wikipedia.org/wiki/Philadelphia_Experiment.
http://hyperphysics.phy-astr.gsu.edu/hbase/electric/maxeq.html.
http://hyperphysics.phy-astr.gsu.edu/hbase/nuclear/nucuni.html.
http://hyperphysics.phy-astr.gsu.edu/hbase/particles/neutrino.html.
http://en.wikipedia.org/wiki/Mass%E2%80%93energy_equivalence.
http://www.mathsisfun.com/algebra/quadratic-equation-derivation.html.
http://www.natureworldnews.com/articles/5968/20140210/mass-neutrinos-accurately-calculated-first-time-physicists-report.htm.
Exercise 5.2
A very high-speed proton having a kinetic energy of 10 MeV can be created in a particle accelerator. What is the velocity
of the proton inside of the accelerator?
Exercise 5.3
Part 1: What is the minimum frequency that a gamma ray emitted from the nucleus of a radioactive atom must have to
create an electron–positron pair in a nuclear reaction such as “pair production”?
Part 2: What is the wavelength of the gamma ray in this case?
Part 3: If the electron is emitted from the reaction at an angle of 30° in a standard Cartesian coordinate system,
what angle is the positron emitted at, and what are the x and y components of the momentum of the positron in this
case?
Part 4: Show that spin and charge are conserved in this reaction.
Exercise 5.4
Suppose that the gamma ray is emitted from the nucleus of a Carbon-12 atom with a frequency of 5 × 1020 Hz
Part 1: What is its wavelength, and what is its energy?
Part 2: What is the value of its linear momentum as well?
Part 3: What is the ratio of the frequency of this gamma ray to the frequency of a photon produced by a GE “soft white”
light bulb in your home?
Exercise 5.5
A Carbon-12 nucleus has a number of internal energy levels that correspond to different orbital configurations of the
protons and neutrons inside of it. These energy levels are shown in Figure 6.22 in Chapter 6. Suppose that a Carbon-12
nucleus in an excited state emits a gamma ray from the highest energy level shown, and then transitions itself to its most
stable state, which corresponds to the lowest energy level shown.
Part 1: What is the energy of the gamma ray that is emitted in this process?
Part 2: Suppose that the gamma ray then collides with the nucleus of a hydrogen atom and is subsequently absorbed.
What velocity does the hydrogen nucleus recoil away from this reaction? You may assume in this case that the collision
with the hydrogen atom is an elastic one.
Part 3: Suppose that the hydrogen nucleus subsequently decides to emit another gamma ray with 10% of the energy of
the original gamma ray. What are the frequency and the wavelength of the emitted gamma ray in this case?
Exercise 5.6
Derive an expression for the momentum of a photon which is based on its frequency of vibration f.
Exercise 5.7
Suppose that an electron and a positron happen to collide with each other in a nuclear reactor core and that the velocity
of each particle is relatively small compared to the speed of light? What are the energy and the frequency of the photon
that is created in the process?
178 Thermal Energy Production in Nuclear Power Plants
Exercise 5.8
In simple terms, explain why the photon has no rest mass.
Exercise 5.9
A small nuclear reactor produces 500,000 MeV of thermal energy each second. How many neutrons per second does it
have to convert entirely into energy to do this?
Exercise 5.10
A fission neutron is emitted from a Plutonium-239 nucleus with an initial kinetic energy of 10 MeV. What is its speed in
m/s immediately after it is born?
Exercise 5.11
Suppose that 500 billion uranium atoms are split apart every second in a fuel rod in a low-power research reactor.
If 200 MeV of kinetic energy is released in each fission reaction, how many Watts of power does the fuel rod produce
every second?
Exercise 5.12
Suppose that a particle is traveling through empty space at 30,000 m/s. What is the percentage difference between the
rest mass of the particle and its relativistic mass?
Exercise 5.13
The International Space Station orbits the Earth at a speed of about 11,000 m/s. A cesium clock to be installed on the
space station has a frequency of 9,192,631,770 cycles per second on the Earth. What will the frequency of the same clock
be aboard the International Space Station?
Exercise 5.14
What is the ratio of the speed of light in empty space to the speed of light in air? If the refractive index of a diamond is
2.417, what is the speed of light in a diamond?
Exercise 5.15
A gamma ray produced in a nuclear reactor has a frequency of 1 × 1020 cycles per second. What is its wavelength, and
what is its energy?
6
The Laws of Thermodynamics
6.1 A n Introduction to the Laws of Thermodynamics
Nuclear heat transfer and fluid flow are based on a set of empirical laws that have become known as the three laws of thermodynamics.
These laws are actually a set of “empirical principles” that were first proposed in the 1800s—about 150 years before the first nuclear
power plants were built. However, they have so much importance to the field of nuclear science and engineering that no discussion
of reactor thermal-hydraulics would be complete without them. In this c hapter, we would like to briefly discuss these laws and their
importance to the nuclear industry as a whole (See Figure 6.1 and Table 6.1).
FIGURE 6.1 Converting thermal energy into mechanical energy creates work and the rate at which work is performed is called power. In a
nuclear power plant, water is heated to create steam and the steam is used to turn the blades of a steam turbine. The steam turbine is connected to
an electrical generator that then produces electric power. All commercial nuclear power plants produce electricity in this way.
179
180 The Laws of Thermodynamics
TABLE 6.1
A Summary of the Three Laws of Thermodynamics and the Zeroth Law, or Nernst Postulate, which was not included in the
original set
Moreover, it follows immediately from this definition that any change in the total energy of a system (including a nuclear
one) must be equal to the difference between the total energy entering the system and the total energy leaving the sys-
tem. This applies to any process in which a nuclear reactor can participate. Since work is a form of mechanical energy
that can be converted into thermal energy and thermal energy can be converted into work, the first law of thermodynam-
ics can also be written as
Initial system energy + Total energy entering the system − Total energy leaving the system = Final system energy
or more concisely,
Equation 6.1 is called an energy balance, and it applies to any nuclear system as well as to thermal and electromechanical
ones. In practice, it can be applied to any form of energy transfer including internal energy transfer, kinetic energy
transfer, potential energy transfer, heat transfer, work transfer, and even mass transfer. Hence, the first law of thermo-
dynamics is the fundamental relationship that links the energy produced in a nuclear power plant to the amount of
useful work it can perform. In the next section, we would like to briefly discuss the ways in which this energy transfer
can occur.
This energy may also consist of work. If the final energy is less than the initial energy, then the system has lost energy,
and if the final energy is greater than the initial energy, then the system has gained energy. It is also possible that the
final energy is equal to the initial energy in which case the overall energy does not change. However, in this case, the
total energy may be deployed differently than the initial energy. For example, some of it may be converted into work.
The energy of a system can consist of kinetic energy, potential energy, work energy, and internal energy. In the absence
of electromagnetic fields and surface tension effects, the change in the total energy of a system can be expressed as the
sum of the changes in its internal, kinetic, and potential energy:
plus whatever work is added or subtracted from it. Here, the individual components are given by
182 The Laws of Thermodynamics
∆KE = 1 2 ( v 2 2 − v1 2 ) (6.4)
∆PE = mg ( z 2 − z1 )
∆WORK = WORK IN − WORK OUT
The values of u1 and u2 represent the specific internal energies of the system, which can be determined from material
property tables or thermodynamic relationships. For water–steam mixtures, these property tables are called the Steam
Tables. The Steam Tables are discussed in Chapter 7.
Q=
∫ dQ/ dt dt (6.5)
when the heat production rate is time dependent. The absolute temperature is a measurement of the internal energy
U of a material. The internal energy per unit mass is called the specific internal energy, and it is given the symbol u.
Consequently, the specific internal energy is defined as u = U/m.
6.5 Generalized Energy Transfer 183
where θ is the total energy per unit mass removed from or added to the system and the subscripts IN and OUT refer to
the system’s inflows and outflows. This equation can be used to perform an energy balance for any nuclear power plant
(or for any fossil-fueled power plant for that matter). The work required to push a unit mass of fluid into or out of a system
is called the flow work or flow energy, and it is sometimes written as
WFLOW = Pυ
where P is the system pressure, υ is the specific volume, υ′ = vA is the volume of fluid flowing perpendicular to a cross-
sectional area A per unit time, and v is the average fluid velocity normal to A. Sometimes υ′ is called the volumetric flow
rate, and it can also be expressed as
υ′ = vA = m
/ρ
where m = ρvA is the mass flow rate across one of the system boundaries. It must be noted that these energy flows must
be computed for every inlet and exit the system possesses. For any system containing inflows and outflows, it is some-
times convenient to combine the flow energy and the internal energy together into another quantity called the specific
enthalpy h. Then, the total energy of the fluid per unit mass θ crossing one of its boundaries is
θ = h + KE + PE = h + v 2 2 + gz
where z is the axial elevation of the boundary. When changes in the kinetic and potential energy of the fluid stream are
negligible, then the amount of energy transport due to the movement of this fluidic mass is EMASS = mh and the rate of
energy transport is E ′MASS = mh.
The total energy transported by a flowing fluid of mass m with uniform properties is
184 The Laws of Thermodynamics
This rate equation applies to any component of a nuclear power plant as well as to conventional power plants. We would
now like to give an example of how this energy balance works.
where q″ is a quantity called the nuclear heat flux (the heat flow rate per unit area Q/A) and h is a quantity called the
nuclear heat transfer coefficient. Different heat transfer coefficients are used for single-phase and two-phase flows.
Here, TFUEL corresponds to the high-temperature region where the heat is being produced (THIGH), and TCOOLANT cor-
responds to the low-temperature region (TLOW) where the heat is flowing to. In principle, all nuclear reactors can be
described in this way. In light water reactors, the heat that is generated by the fuel rods is then converted into steam, and
this steam is used to drive the blades of a steam turbine. This is the primary process by which electricity is produced in
nuclear power plants. Finally, it may be helpful to understand how the first law of thermodynamics fits into this picture.
On a fundamental level, nuclear energy is produced because of a process called nuclear fission (see Chapter 5) where the
nuclei of very heavy atoms are split apart. The average kinetic energy KE of the nuclear particles that are produced in
this process can be found from the equation
∑ ( m n v n 2 ) N (6.9)
N
KE = 1
2
n=1
where N is the number of particles produced and vn is the velocity of particle n. This kinetic energy KE is also equal to
the energy E that is released when a small amount of the mass Δm is converted into a large amount of energy. According
to Einstein’s equation E = mc2, the energy E that is released is also equal to E = Δmc2, where c is constant having the
value of 3.3 × 109 m/s that we will refer to as the speed of light. Thus, the kinetic energy of all of these particles can be
expressed as
∑ ( m n v n 2 ) = ∆mc 2 (6.10)
N
KE = 1
2
n=1
and this is the thermal energy that is eventually produced in the core. So the first law of thermodynamics, which is essen-
tially another statement of the conservation of energy, says that this kinetic energy must show up in the form of heat and
that the heat that is produced must always be conserved. We can also express this fact by writing
6.8 The Second Law of Thermodynamics 185
∑ ( m n v n 2 ) = ∆mc 2 (6.11)
N
Q = KE = 1
2
n=1
In other words, the first law of thermodynamics is nothing more than a statement of the conservation of mass-energy.
Note that when the thermodynamic efficiency is measured in this way, the efficiency is a dimensionless number. For
modern nuclear power plants, the thermodynamic efficiency can range from about 32% to 48%. The exact efficiency
depends on the temperature at which heat is produced in the core, and the temperature at which it is discharged to the
environment. In general, these temperatures must be measured in degrees Kelvin because the internal energy U of the
fluid that participates in the heat rejection process is also measured in K. We will have more to say about several alterna-
tive temperature scales that the nuclear industry uses at a later time. However, the absolute temperature must be used to
find the thermodynamic efficiency because the internal energy of the working fluid is expressed in this way.
where Q is the amount of heat that flows, A is the area of the surface through which it flows, T is the temperature of the
material, grad is the gradient operator (the gradient of the temperature), and k is an empirically determined constant
called the thermal conductivity of the material. According to this definition, Q has the units of J/s or W, T has the units
of °C or °K, and k has the units of BTU/h/ft °F in the English unit system or W/m °K in the SI unit system. In general,
the thermal conductivity k depends on both the composition of a material and its temperature, and for a particular mate-
rial, the value of k will usually change as the temperature changes; for example, k = k(T). It is important to point out that
both the first and second laws are required to arrive at this simple equation. Moreover, Equation 6.13 forms the basis of
the field of conductive heat transfer, which we will discuss in Chapter 10.
186 The Laws of Thermodynamics
In other words, if we want to force heat to flow in the opposite direction than it naturally does, we have to expend
additional energy (in the form of work) to do so. Hence, the second law establishes the direction of energy trans-
port and postulates that the flow of heat or molecular disorder will always be from a region of high temperature to
a region of low temperature (i.e., along with the negative temperature gradient). This general concept is illustrated
in Figure 6.4.
The second law of thermodynamics also leads us to conclude that the flow of heat can be thought of as the flow of
molecular disorder from a high-temperature region to a low-temperature region, and furthermore, that the amount of
heat ΔQ that flows is proportional to the absolute temperature T multiplied by the change in the molecular disorder
that we call the entropy S. Hence, the relationship between the change in the molecular disorder ΔS and the amount of
heat ΔQ that flows is
∆Q = T ⋅ ∆S (6.14)
where T is the absolute temperature of the material (in °K). Using this simple relationship, it can be seen that the change
in the entropy S of a system, which is a measure of this molecular disorder, can be expressed as
∆S = ∆Q/T (6.15)
FIGURE 6.4 According to the second law of thermodynamics, heat always flows from a region of high temperature to a region of
low temperature. The greater the temperature difference between the regions, or the greater the flow of heat, the greater the entropy
increase will be.
6.10 The Clausius Inequality 187
and that the entropy S has the units of kilojoules (kJ) per degree Kelvin (kJ/°K). Notice that the entropy change that
occurs as we move from an initial state 1 to a final state 2 is
2
∆S = S2 − S1 =
∫1
dQ/T (6.16)
Most of the time, nuclear engineers are interested in the change in the entropy of a system rather than the absolute value
of the entropy itself. Absolute values of the entropy are determined by the third law of t hermodynamics, which we will
discuss in Section 6.12. Therefore, it is possible to assign a zero value to the entropy at some arbitrarily selected initial
state 1, and the entropy at the final state 2 can be determined by evaluating the integral above. However, to be even more
precise, one needs to know the relationship between Q and T for a particular process to perform this integration, and
often times, this relationship is not readily available. So for the majority of the cases that are encountered in real life, we
have to rely on tabulated values for the entropy. For fluids such as water, these values can be found in what are known as
the Steam Tables. We will have more to say about the steam tables in Chapter 7.
Another consequence of the integral in Equation 6.16 is that the entropy change ΔS between two arbitrary
thermodynamic states is the same no matter what the path, reversible or irreversible, between these two states may
be. Hence, the entropy, like all other physical properties of a substance, has fixed values at fixed states. So even if we
do not understand how we got from state 1 to state 2, we know from Equation 6.16 what the entropy change ΔS will be.
where
2
∆S = S2 − S1 =
∫1
dQ/T (6.18)
This expression is called the Clausius inequality, and it has profound implications on the process of nuclear heat trans-
fer and fluid flow. In other words, the entropy of an isolated system will always increase or stay the same, it but never
FIGURE 6.5 The entropy change ΔS between two states (1 and 2) is the same regardless of the path that is chosen.
188 The Laws of Thermodynamics
decreases over time. This is a reminder of the fact that the molecular disorder of a system always increases over time
and that the entropy of an isolated system will continue to increase until the entropy reaches a maximum value SMAX.
At this point, the system is said to be in thermodynamic equilibrium with its environment, and the molecular disor-
der cannot increase any further. The energy distribution of the particles when this occurs is given by what is known
as a Maxwell–Boltzmann probability distribution, and in a Maxwell–Boltzmann probability distribution, the average
energy E of each particle is only a function of the absolute temperature T (which is assumed to be uniform everywhere).
In a Maxwell–Boltzmann distribution, the average energy of a particle is E = 3/2 kT and the most probable energy is
E = ½ kT, where k is Boltzmann’s constant.
Note that the average energy is greater than the most probable energy. Furthermore, the total energy when a state of
thermal equilibrium is reached is independent of the particle mass. A picture of a Maxwell–Boltzmann probability
distribution is shown in Figure 6.6. Even the energies of the neutrons in a reactor core are determined by this probability
distribution. Thus, the Maxwell–Boltzmann distribution is a thermal equilibrium distribution, and it cannot be used if
the particles it represents are NOT in thermal equilibrium with each other.
FIGURE 6.6 A picture of the Maxwell–Boltzmann probability distribution showing how the energies and speeds of particles
distribute themselves when they come into thermal equilibrium with their environment.
6.12 The Third Law of Thermodynamics 189
This principle follows directly from the Clausius inequality. It allows the heat in the core to flow from a region of
high temperature to a region of low temperature, and it also forces certain chemical reactions to come to a halt before
reaching completion. Entropy is conserved during ideal reversible processes, but it is NOT conserved in any actual
process. It always increases in reactors and is NOT a conserved property in these devices. The performance of energy-
producing devices is degraded as the system entropy increases, and the amount of entropy generation is a measure of
the irreversibilities that these processes possess. Example Problem 6.3 illustrates that entropy generation always occurs
when heat flows from a region of high temperature to a region of low temperature irrespective of the rate at which the
transfer occurs. In other words, entropy generation is the expected result of any heat transfer process. This pertains to
nuclear reactors as well as to coal-fired power plants.
In addition to the second law of thermodynamics, there is a third law of thermodynamics that is related to the process of
entropy generation and measurement. This law is called the third law of thermodynamics, and we would like to discuss
the basis for the third law in the next section.
FIGURE 6.7 A picture of Walther Nernst (left) and Rudolph Clausius (right)—two of the founders of the modern field of
thermodynamics.
190 The Laws of Thermodynamics
FIGURE 6.8 A picture illustrating the third law of thermodynamics for a pure substance or crystalline material. The third law of
thermodynamics is also referred to Nernst’s theorem or Nernst’s postulate.
observed that a material becomes more ordered on a molecular level when its temperature is reduced and that it can be
thought of as an ideal crystal that becomes more ordered the closer to absolute zero it becomes (see Figure 6.8).
Conceptually, this is no different than turning a cup of liquid water into a block of ice. As the temperature is reduced,
the water reverts to a crystalline form, and the ice certainly has more molecular order than the liquid water does. By the
same token, liquid water certainly has more molecular order than high-temperature steam. The third law of thermody-
namics is simply an observation of this fact. When statistical mechanics was first developed in the early 1900s, the third
law of thermodynamics, like many other laws, changed from being a fundamental law (which had to be inferred from
an experiment) to a derived law (which was derived from even more fundamental principles). Today, the third law can
be expressed in terms of the number of ground states that a crystalline material has as it approaches absolute zero (0°K).
So in a statistical sense, the entropy can be expressed as
S = k B ln N (6.22)
where k B is Boltzmann’s constant and N is the number of ground states that a material can possess. Boltzmann’s con-
stant has a value of 1.3806488 × 10 −23 J/°K in the SI unit system. Thus if a crystal possesses only one ground state as
T → 0°K, then ln 1 = 0, and the entropy S will be zero as well. The third law of thermodynamics does not have much
practical application to nuclear systems because nuclear power plants operate at much higher temperatures. However, it
is important to be aware of its implications, and because it is discussed frequently in more advanced textbooks.
FIGURE 6.9 A visual depiction of the zeroth or fourth law of thermodynamics and its implications.
observation that if two bodies are in thermal equilibrium with a third body, then they must be in thermal equilibrium
with each other (see Figure 6.9). This condition of mutual thermal equilibrium is a statement of another law of thermo-
dynamics that is called the zeroth law of thermodynamics.
The zeroth law cannot be derived from any of the other laws, and though it was first formulated by R.H. Fowler years
after the other three laws were known, it was called the zeroth law because it should have preceded the development
of the first and second laws. In some textbooks, the zeroth law is referred to as the fourth law of thermodynamics. The
primary importance of the zeroth law is that it validates the use of temperature as a measure of the molecular disorder
of an object. For example, if we replace the third body in the zeroth law with a thermometer, then the zeroth law can be
restated in the following form:
While this statement may seem self-evident or even lame, it justifies the use of temperature as a way to measure the
molecular disorder of a system. In particular, it validates the fact that the absolute temperature (which is used in the
statement of the third law) is the correct parameter to use to define the minimum entropy of a system. Of course, heat
and temperature are completely different from each other because heat is the amount of molecular vibrational energy
a system possesses while temperature is a way to measure it, but the zeroth law then allows us to relate the temperature
of an object to its thermodynamic equilibrium and therefore its average temperature T. When an object such as a reactor
coolant comes into thermal equilibrium with its environment, some atoms in the coolant may gain energy, whereas other
atoms in the coolant may lose energy. However, in an average sense, the energy of the coolant will be unchanged (unless
the temperature of the coolant is also changed). This is as much as thermodynamics can tell us about how heat and energy
behave in a nuclear power plant. The one thing that thermodynamics cannot tell us is how fast these processes occur.
TABLE 6.2
The Fundamental Differences between Thermodynamics and Heat Transfer
thermodynamics are summarized in Table 6.2. Notice that thermodynamics applies to a system in equilibrium while
heat transfer is a more dynamic and time-dependent process. This does not mean that heat transfer only considers time-
dependent heat flows. As a matter of fact, it considers both time-dependent and steady-state ones. It can also be used
to determine the distribution of temperatures within an object which may then be used to determine the rate at which
energy is transferred through the object. In general, conductive heat transfer is based on another conservation equation
that is called the general heat conduction equation. We will introduce the reader to the general heat conduction equation
in Chapters 10 and 11.
When calculating the thermal equilibrium of a pure substance, only the absolute temperature is required. However,
when calculating the heat transfer rate, a temperature difference is required.
°K = °C + 273 (6.23)
and to convert from degrees Fahrenheit to degrees centigrade, the additional relationship is used, which is given by
It then follows that °K = (5/9)(°F − 32) + 273 and °F = 9/5°C + 32. The entropy of a material, which is a measure of its
molecular disorder, is always lowest at a temperature of absolute zero (0°K) because the vibrational energy is minimized
194 The Laws of Thermodynamics
FIGURE 6.10 The three primary temperature scales in use in the world today and their respective ranges. Depending upon the
country in which one is located, the nuclear power industry uses all three of these scales.
when a temperature of absolute zero is reached. In Section 6.12, we mentioned that this is simply a statement of the third
law of thermodynamics. Reactor cores operate in different temperature regimes depending on how they are designed. In
general, liquid metal fast breeder reactors (LMFBRs) have higher coolant temperatures than pressurized water reactors
(PWRs), and PWRs have higher coolant temperatures than boiling water reactor (BWRs). The coolant temperatures in
gas reactors are even higher than those in LMFBRs. Table 6.3 compares the inlet and outlet temperatures for each reac-
tor type. In g eneral, the temperatures shown in Table 6.3 are industry averages, and higher or lower t emperatures may
be encountered in practice.
The thermodynamic efficiency of a nuclear power plant depends on the absolute temperature at which this heat
is rejected to the environment from the core. Typical thermodynamic efficiencies range from about 28% to 48%
(see Table 6.4). In particular, if the temperature at which the heat is rejected from the plant (to a lake, a river, or a cool-
ing tower) is fixed by the environment in which the plant is located, then the thermal efficiency is highest for reactors
having higher coolant temperatures. The temperature of the coolant exiting the core is determined by the power density
of the core and the melting point of the fuel. (It may also be affected by the melting point of the cladding.) In general,
the fuel and cladding must operate well below their respective melting points in order for the release of radiation to be
minimized during normal operation. Moreover, the fuel and cladding temperatures generally place an upper limit on
the thermodynamic efficiency of the plant. Because of this, coal-fired power plants have higher operating temperatures
than nuclear power plants do, and because they have higher operating temperatures, they also have higher thermody-
namic efficiencies. This may come as a surprise to someone who is not familiar with the field of nuclear power.
TABLE 6.3
The Inlet and Outlet Temperatures of Some Common Power Reactor Cores
The Inlet and Outlet Temperatures of Some Common Power Reactor Cores
Type of Reactor Inlet Temperature (°C) Outlet Temperature (°C)
CANDU 270 310
BWR 275 286
PWR 290 330
LMFBR 390 550
AGR 290 640
HTGR 320 740
Source: Todreas and Kazimi (2008).
Note: The numbers shown are representative. The actual values may vary from those shown.
6.22 The Definition of Energy 195
TABLE 6.4
The Thermal Efficiencies of Common Commercial Power Reactors in the World Today
Because the first law of thermodynamics requires heat energy to be conserved, the heat that flows into a system will
have a positive (+) value, and the heat that flows out of a system will have a negative (−) value. Heat energy is generally
given the symbol q (or capital Q), and it is usually measured in J or BTUs. Hence, we can define the signs that must be
associated with the flow of heat in the following way:
and these are the definitions that we will use throughout this book. As we will see, both heat and work are path func-
tions, which imply that their final values are a function of the path that a particular process or s ystem takes. These paths
can be cyclical or noncyclical—depending on the system and the specific type of process involved. In nuclear power
plants, heat is produced in the core and it exchanged between other components to generate electric power. Waste heat
is rejected to the environment, and the remaining heat is then recycled and sent back to the core. Because some of this
thermal energy is lost, no commercial power plant can be 100% efficient.
E = KE + PE (6.25)
However, this definition is not adequate for nuclear systems because it neglects the internal heat energy that an object
can possess. Hence, a more appropriate definition of the total energy of a material object is
196 The Laws of Thermodynamics
where U is an additional quantity called “the internal energy” of the object. In most cases, the internal energy U is a
function of three separate variables that we would now like to discuss. These include
1. The temperature and the pressure of a material, as well as something called the quality x. Hence, we may write
the internal energy per unit mass as u = u(P, T, x), where we will present a formal definition of the quality in
Section 6.31. The total energy of the object is then U = uM where M is its mass. Its internal energy per unit mass
is then called its specific internal energy.
2. Alternatively, we can express the specific internal energy u as a function of the pressure and the s pecific entropy.
Hence, we may write u = u(P, s), where the specific entropy s is a function of P, T, and x: s = s(P, T, x). We pre-
sented a formal definition of the entropy in a previous section. The total entropy of the object is then S = sM where
M is its mass.
The reader should bear in mind that these definitions are equivalent as long as we are careful to define the quality and
the entropy in an appropriate way. In thermodynamic terms, the energy of a system can also be considered to be the
capacity of a physical system to perform work. Hence, if we can convert some of this energy into “useful” energy, then
we can perform external work with it. Moreover, energy can also be changed from one form to another as long as the
total energy of a system is conserved. For example, we can convert some potential energy into kinetic energy when
we drop a tennis ball to the ground. We can also convert some potential energy into kinetic energy when we throw
the same tennis ball over a wall. In more familiar terms, suppose that two billiard balls collide with one another. The
kinetic energy that they possess may be transferred from one billiard ball to the next. If one of the billiard balls comes
to rest, the resulting kinetic energy must become sound and perhaps a bit of heat at the point of collision. However,
no matter what form the energy of the two billiard balls ultimately takes, it must always be conserved. The SI unit of
energy is the joule (J), which is also equal to 1 Newton-meter (N m). The joule is also the SI unit for another impor-
tant quantity that we call work. We would now like to discuss how heat energy and work are related to one another.
Example Problem 6.6 (see Figure 6.11) illustrates how the potential energy and the kinetic energy of bowling ball can
be exchanged as the ball rolls up and down a hill. There are many similar examples of this type of energy transfer in
everyday life.
FIGURE 6.11 A bowling ball can be used to demonstrate the conservation of energy including the conversion of potential energy
into kinetic energy and heat energy or friction.
6.23 The Definition of Work 197
Solution This problem illustrates the conservation of energy. The potential energy of the ball is converted into kinetic
energy, and its kinetic energy is converted into potential energy. However, if the ball does not lose any of this energy to
friction, then we may write
KE1 + PE1 = KE 2 + PE 2
The potential energy of the ball is given by PE = mgz, and its kinetic energy is given by ½ mv2. Here, v is its velocity,
g is the acceleration of gravity, and z is its distance from the bottom of the bowl. The conservation of energy therefore
requires that
1
2 mv1 2 + mgz1 = 1 2 mv 2 2 + mgz 2
where z1 = h. The maximum velocity of the ball is reached when z2 = 0 because at this point, all of its potential energy
is converted into the kinetic energy. Its kinetic energy is then KE = mgh = 1 2 mv 2 2 and the velocity of the ball is
v2 = √(2gh). In other words, when frictional effects can be ignored, the sum of the potential and the kinetic energies of
the ball remains constant. If we need to account for the effects of friction between the bottom of the ball and the bowl
and between the moving ball and the surrounding air, then the equation for the conservation of energy becomes
1
2 mv1 2 + mgz1 = 1 2 mv 2 2 + mgz 2 + WFRICTION
where W FRICTION is the frictional work performed by the ball. In other words, friction causes some of the kinetic energy
of the ball to be converted into thermal energy or heat. This observation is another statement of the first law of thermo-
dynamics. [Ans.]
where d is the distance traveled. In elementary calculus, it is also defined as “the integral of the force over a distance of
travel.” So if the force varies as a function of distance, then a more appropriate definition for the work is
W=
∫ F(r) r dr (6.28)
where r is the distance. According to this definition, an object has to move for mechanical work to be performed (see
Figure 6.12).
FIGURE 6.12 Work is a derived quantity that appears at the boundary of a thermodynamic system when a force F moves through a
distance d. It only appears at the boundary of a system when a system undergoes a change of state. If work is done by the system on
its environment, the work is said to be a positive number, and if work is done on the system by its environment, the work is said to be
a negative number.
F = ( P1 − P2 ) A or F = ∆P ⋅ A (6.29)
W = ∆P ⋅ A ⋅ d (6.30)
where d is the distance the piston moves. If work is done by the system on its surroundings, the work output of the system
is positive (+). If the surroundings do an equivalent work on the system, then the work output is negative (−). Looked
upon in this way, either of these scenarios can only occur if the internal energy U of the system increases or decreases
by the same amount. Hence, the relationship between the internal energy of a system and the work that it performs can
be expressed as
∆W + ∆U = 0 (6.31)
and this is another way of saying that the energy of a system (plus the work that it performs on its surroundings) is always
conserved. Finally, it follows from Equation 6.31 that the external work W a system can perform is
W = −∆U (6.32)
where ΔU is the change in the internal energy of the system. Unfortunately, many books tend to o bfuscate this fact.
However, at the end of the day, work can only be done if the internal energy U changes by a c omparable and opposite
amount.
where v is the velocity. Notice that because the force and the velocity are vectors, the power P is also a vector quantity.
Hence, energy is expressed in J or kJ, and power is expressed in W or kW. Other definitions relating energy, work, and
power follow this same basic rule.
According to this definition, the pressure in a PWR pressure vessel (about 2,250 PSI) is equivalent to
2,250 PSI × 6,802.72 Pa/PSI ≈ 1.5 × 107 Pa = 15.5 MPa. In light water reactors, the system pressure is used to
establish the boiling point of the coolant. When we measure the pressure of a system, we must be careful to distin-
guish between the relative pressure and the absolute pressure. Pressure gauges, vacuum gauges, tire pressure gauges,
and manometers are all designed to measure the pressure relative to the atmospheric pressure that we experience
at sea level. Hence, to get the absolute pressure, we must add the atmospheric pressure to the gauge pressure.
Therefore,
and the atmospheric pressure is either 14.7 PSI or 100,000 Pa (100 kPa). Using this same concept, a vacuum is defined
as the absence of pressure at a particular point in space or time. So a perfect vacuum can be obtained when the abso-
lute pressure is equal to 0, and when this occurs, the molecular momentum that is transferred to the instrument that is
measuring this pressure must be equal to 0. Relative and absolute values of the pressure are compared in Figure 6.13.
The relative pressure depends on the environment in which the pressure is being measured.
TABLE 6.5
Different Ways of Measuring the Absolute Pressure and the Relationships between Them
TABLE 6.6
The Coolant Pressures in Some Common Commercial Power Reactors in the World Today
independent variables. In addition, these state diagrams are important because they represent the thermodynamic states
that a reactor coolant can possess.
f(P, V, T) = 0 (6.36)
FIGURE 6.14 In open and closed thermal cycles, the path between the start point and the end point (point 1 and point 2)
d etermines the area under the curve. When drawing a P–V diagram for a closed system, the area under each curve is equal2 to the
in this case. ∫
heat Q that is added to or removed from the system. The net work W that is performed is then equal to the integral W = P dV
1
202 The Laws of Thermodynamics
The form of the state equation depends on the properties of a particular material. Any variable in this e quation can be
expressed as a function of the other two. Hence, the state of a system is completely specified by any two of the three:
P and V, P and T, or V and T. For example, if we were able to measure the temperature, the pressure, and the entropy of
a substance, we would be able to deduce its internal energy U from the information provided. Conversely, if we know
the enthalpy of a material when it is a liquid h L, the enthalpy of a material when it is a vapor hV, and what fraction of the
material x is a liquid or a gas, then we can deduce the total enthalpy h that the mixture will contain:
h = x h v + (1 − x)h l (6.37)
For two-phase mixtures, the mass fraction x is also called the quality. The quality will be formally defined in another
section. Throughout this book, we will rely on these state variables to tell us how much energy a system has, or how
much work it is capable of performing. We have discussed the temperature, the pressure, and the internal energy up to
this time, but we have not discussed the relationship of these variables to the enthalpy. We would now like to demon-
strate how this functional relationship can be defined.
U = U(T)
(6.38)
(when no external work is performed)
provided that the material does not perform any external work. Here, the value of T is the absolute temperature of the
material, and the absolute temperature of the material is measured in degrees Kelvin. Relative changes in the internal
energy are then found by evaluating the internal energies at different temperatures (say T1 and T2), which correspond to
different thermodynamic states. The change in the internal energy between these two states is then
∆U12 = U 2 − U1 = U ( T2 ) − U ( T1 ) (6.39)
This method of defining the internal energy is appropriate for most solids (such as iron and steel), but it does not work
well for liquids and gases because they can also expand and contract. When they do so, either they can have work per-
formed on them or they can perform work on their environment. A common example of this work is the expansion and
contraction of the gas in an internal combustion engine. When this gas is initially compressed and mixed with gasoline,
mechanical energy is added to it, but when it ignites and explodes, the resulting chemical reaction adds additional
thermal energy to the air–gas mixture. The resulting expansion causes the total energy to be greater than the internal
energy alone and another thermodynamic variable must be used to account for the external work that is performed.
In thermodynamics, this additional variable is called the enthalpy H. The enthalpy H is a measure of the total energy
content E of a system. In particular, it is used to define the total energy content of a fluid, which can consist of its
internal energy U, plus the amount of external work W that it is capable of performing on its environment. So, the
formal definition of the enthalpy is
H = U + W (6.40)
6.31 Finding the Enthalpy When a Material Changes Phase 203
where the external work it can do is equal to the product of the pressure of the fluid P and its volume V. Hence, an alter-
native definition of the fluid enthalpy is
H = U + PV (6.41)
where in the case of a fluid, PV is called the flow work. Now in general, the internal energy U is also a function of the
fluid temperature, and so the enthalpy can also be written as
H = U(T) + PV (6.42)
and this is the expression for it that is most commonly used. This particular definition implies that any change in the
enthalpy is given by
dH = dU + P dV + V dP (6.43)
or ∆H = ∆U + P ∆V + V ∆P (6.44)
In other words, the enthalpy of a fluid can change if either its pressure or its volume changes. These effects are very
important when attempting to calculate the energy of the coolant in a reactor core. Hence, if the coolant pressure
changes, or the coolant begins to expand or contract, the enthalpy change (and hence the total energy change ΔE) is
different than just the change in the internal energy alone. In practice, the value of H is measured in J or kJ. Sometimes
it is more convenient to express it as a thermodynamic variable that is mass-based. This can be done by dividing both
sides of Equation 6.41 by the fluid mass M. The corresponding equation for the fluid enthalpy is
where h = H/M is called the specific enthalpy of the fluid, u = U/M is called the specific internal energy, and υ = V/M
is called the specific volume. Here, the values of h, u, and υ have the units of J/kg, J/kg, and m 3/kg, respectively. The
values of these properties for some common reactor coolants can be found in the steam tables. The steam tables for
water, steam, and other fluids are presented in Appendices F, G, and H. (An abbreviated version of them can be found
in Tables 6.7 and 6.8.) For saturated water, the fluid properties at the saturation point are usually presented as a func-
tion of the saturation temperature TSAT and then as a function of the saturation pressure PSAT. The first table, called
a saturated temperature table, is used when the saturation temperature is known and the other properties must be
deduced from it. The second table, appropriately called a saturated pressure table, is used when the saturation pres-
sure is known and the other properties must be deduced from it. Because of the way the specific enthalpy is defined,
the specific enthalpy is always greater than the s pecific internal energy. The difference between these values is then
equivalent to the flow work Pυ.
h v = h l + h lv (6.46)
204 The Laws of Thermodynamics
TABLE 6.7
An Excerpt from the Steam Tables Showing the Properties of Saturated Water and Steam at Various Temperatures and Pressures
TABLE 6.8
An Excerpt from the Steam Tables Showing the Properties of Saturated Water and Steam at Various Temperatures and Pressures
where the value of the enthalpy in each case has the units of J/kg (or kJ/kg). A related quantity called the heat of vapor-
ization, which we will discuss in Section 6.41, can be used to find the boiling point of a fluid if its temperature and its
pressure are known. In most cases, the heat of vaporization is expressed in J/mol or kJ/kilomol. It is also possible to
express the total enthalpy h of a liquid–vapor mixture as
h = x h v + (1 − x)h l (6.47)
where the symbol x refers to the aforementioned mass fraction or quality. We will have more to say about how the qual-
ity is defined in Chapter 7. However, assume for the moment that a quality of zero (x = 0) corresponds to a fluid that is
all liquid, and a quality of one (x = 1.0) corresponds to a fluid that is all vapor. A quality between these two limits, such
as 0.5, means that a two-phase mixture contains the same amount of liquid and gas (by weight). So to be specific, the
enthalpy of a fluid is a measure of the total energy that it carries. If it loses some of this energy, then another substance
must gain exactly the same amount of energy, or the energy must be transformed into useful work W. We will discuss
how to formally describe this process in Chapter 7.
∆H ~ ∆Q (6.48)
This means that when a fluid reaches its saturation temperature TSAT, all of the energy that goes into it goes into produc-
ing a phase change. The steam tables are a convenient way to describe the change in the internal energy that occurs. In
Section 6.31, we discussed the enthalpy of vaporization hlv, which is the energy required to convert a known amount
of liquid into a vapor at a specific pressure P. In the steam tables, this number is added to the enthalpy of the liquid to
calculate the enthalpy of the vapor. Thus, the enthalpy of the vapor hv must be equal to the enthalpy of the liquid hl, plus
the enthalpy of vaporization hlv:
h v = h l + h lv (6.49)
Now if a fluid does not completely vaporize, then the energy of the mixture h must be equal to the energy of the liquid
plus the energy of the vapor. However, each of these may have a different mass fraction x, and if the mass fraction of
the vapor is x, then the mass fraction of the liquid will be 1 − x. When this occurs, the total enthalpy of the fluid is just
the sum of the enthalpies of each phase, with the appropriate mass fractions applied to each. In thermodynamic terms,
we can then write
h = x h v + (1 − x)h l (6.50)
where x is the mixture quality that was previously discussed. Hence, when the quality is 50%, half of the two-phase
mixture will consist of liquid, and the other half will consist of vapor. The enthalpy of the m ixture h is then halfway
between hl and hl + hlv, and this value can be easily found with the help of the steam tables. The following example
illustrates how the enthalpy of a water–steam mixture is found under these c onditions. The reactor core is assumed to
operate at 1,050 PSI (or 7.15 MPa).
* Note that in some textbooks, hv is replaced by the symbol hg, hl is replaced by the symbol hf, and hlv is replaced by the symbol hfg.
In any event, these symbols refer to exactly the same thing.
208 The Laws of Thermodynamics
FIGURE 6.15 The coolant temperature profile in a typical BWR core. Once the saturation temperature TSAT is reached, the cool-
ant temperature stays the same until all of the water in the coolant channel is converted into a steam. As the coolant flows upward
through the channel, it undergoes a number of physical changes that lead to different flow regimes and flow patterns.
FIGURE 6.16 A coffee pot of boiling water, and the critical point at which the water and the vapor become indistinguishable from
one another.
increase any further, and adding additional energy to the fluid will just cause some additional liquid to be converted
into vapor. Finally, when all of the liquid is completely vaporized, the temperature of the vapor starts to increase again.
If the temperature increases beyond the saturation temperature of the fluid, the resulting mixture is said to be super-
heated. The phenomenon of superheat is important in nuclear power plants because it allows the nuclear steam supply
system to be more efficient. In a nuclear power plant, superheat is used to create additional power by creating what is
known as superheated steam. In Westinghouse and AREVA PWRs, the water from the core passes through a steam
6.35 The Clausius–Clapeyron Equation 209
generator, which creates superheated steam that is sent to a steam turbine. This superheated steam turns the blades of the
turbine at very high pressure. Since the turbine is connected to an electrical generator, this causes the shaft of the genera-
tor to turn and electrical power to be produced. We will explore this process in more detail in Chapter 9. Normally, steam
generators in nuclear power plants are designed to produce steam with between 5°C and 20°C of superheat. Steam that
is superheated is always more energetic than saturated steam.
where
TBOIL = the boiling point (°K)
R = the ideal gas constant (which has a value of 8.314 J/°K mol)
P = the current pressure (atm)
T = the current temperature (°K)
ΔhVAP = the heat of vaporization of the liquid (J/mol)
ln = the natural logarithm to the base e
The Clausius–Clapeyron equation is a very simple equation, but it has many important applications. Specifically, it
states that the boiling point increases with increasing pressure up to a certain point called the critical point, where the
vapor phase can no longer be distinguished from the liquid phase. The critical point is located on top of the vapor dome
in Figure 6.16. In other words, at the critical point, both phases reach a state where their properties become identical.
The boiling point cannot be increased beyond the critical point. Likewise, the boiling point decreases with decreasing
pressure until the triple point is reached. The boiling point cannot be reduced below the triple point. The Clausius–
Clapeyron equation can be used to show that the boiling point of water on top of Mount Everest (which is 8,848 m or
29,029 ft high) is about 69°C (or approximately 156.2°F). The pressure on top of Mount Everest is about 260 mbar (or
26.000 kPa). In general, the boiling point of water decreases about 1°C for every 285 m of elevation, or 1°F every 500 ft.
210 The Laws of Thermodynamics
In a nuclear reactor, exactly the opposite effect occurs. For example, in a PWR (which operates at a system pressure of
about 2,250 PSI), the water does not boil until it reaches a temperature of about 330°C (about 625°F), and in a BWR,
which operates at a pressure of about 1,050 PSI, the corresponding boiling point is about 285°C (543°F). Exercise 6.1
helps to demonstrate another important application of the Clausius–Clapeyron equation. It shows the effect that increas-
ing the pressure can have on the boiling point of seawater. Although this example is somewhat fictitious in nature, it
serves to illustrate how important the ambient pressure can be.
6.36 T
he Relationship between the Pressure and the Boiling
Point of Water in Light Water Reactors
In general, the temperature at which a reactor coolant begins to boil is determined by its design pressure. For liquid
water, the transition to boiling takes place in a number of distinct steps that are shown in Figure 6.17. Prior to boiling
beginning, water exists as a pure liquid, which is sometimes called a compressed liquid or subcooled liquid. In this
state, its internal energy is much lower than the energy required for bubbles to form. Then, as more heat is added
from the fuel rods, its temperature begins to rise, and this temperature rise continues until it is about to vaporize, at
which point the water is said to be a saturated liquid. Once the water begins to boil, it remains as a two-phase mixture
until all of the water is converted into steam. During this process, the temperature of the liquid and the vapor phases
FIGURE 6.17 A T–V diagram showing how water behaves as it is heated at constant pressure. In this process, it transforms itself
from a compressed or subcooled liquid (point 1) to a saturated liquid (point 2). Then, it transforms itself from a saturated liquid–
vapor mixture (point 3), to a saturated vapor (point 4), and finally into a superheated vapor (point 5). The temperature at which boil-
ing begins is called the saturation temperature, and the pressure at which boiling begins is called the saturation pressure. Nuclear
reactors typically use very high pressures and temperatures to embed as much energy as possible into the coolant and increase their
thermal efficiency.
6.38 Pressure–Temperature Diagrams and Phase Diagrams 211
TABLE 6.9
Boiling Point of Water at Different Ambient Pressures
remains the same. That is, until the water is completely vaporized, the phases remain in thermodynamic equilibrium
where T LIQUID = T VAPOR = TSAT and additional heat goes into creating additional steam. Then when all the liquid has
been boiled away, the resulting mixture becomes what is known as a superheated vapor or superheated steam. Any
additional heat that is added to the steam causes its temperature to rise again. Both of these effects are illustrated in
Figure 6.17. The boiling point is then shown for different design pressures in Table 6.9. In general, reactor operating
temperatures and pressures are chosen to embed as much energy into the coolant as possible without melting the fuel.
Keeping the fuel below the melting point also means that nuclear reactors operate at lower temperatures than coal-
fired power plants do.
FIGURE 6.18 Some examples of useful thermodynamic state diagrams for H2O. (Pictures provided by Wikipedia.)
FIGURE 6.19 A three-dimensional phase surface from which PV, PT, and VT diagrams can be constructed. (Picture provided by
Wikipedia.)
Each of these regions has a well-defined boundary for a given temperature and pressure. A pressure–temperature dia-
gram for a pure substance is shown in Figure 6.20, and pressure–temperature diagrams for water and carbon dioxide
are shown in Figure 6.21. Most materials have phase diagrams that look similar to this. Every phase diagram also has
a line of demarcation between the liquid and vapor phases that is called a vaporization line. This line is the line of
demarcation between a liquid and a gas. In addition, all phase diagrams also have a triple point. A triple point is a P–T
combination where all three of the phases (solid, liquid, and vapor) coexist in thermodynamic equilibrium at the same
time. For example, the triple point of water is reached at a temperature of 0.01°C (273.16°K) and a pressure of 611.73 Pa,
6.1173 mbar, 0.0883 PSI, or 0.0060373 atm. The triple point for other materials can occur at d ifferent pressures and tem-
peratures. Finally, the vaporization line will eventually end at what is known as the critical point. Here, the liquid phase
and vapor phases can no longer be distinguished from one another. At pressures above the critical point, the liquid can
no longer convert itself into a vapor. Instead, the liquid will convert itself into a very dense fluid.
6.39 Temperature–Volume Diagrams 213
FIGURE 6.20 An example of a P–T phase diagram for a pure substance. A P–T diagram of this type can also be applied to
common reactor coolants such as water.
FIGURE 6.21 On the left—a P–T phase diagram for water, and on the right—a P–T phase diagram for carbon dioxide. Although
carbon dioxide is normally considered to be a gas, it can be converted to a liquid or a solid by changing its temperature or pressure.
(Pictures provided by Wikipedia.)
the liquid region, becomes flat as we pass through the vapor dome, and rises again when we leave the right-hand side
of the dome. Each progressively higher value of the system pressure corresponds to a different horizontal line that is
closer to the critical point. On the left-hand side of the vapor dome, the liquid is called a subcooled liquid because it
has not yet begun to boil. On the right-hand side of the vapor dome, it is called a superheated vapor because it has
no more liquid left in it. The left-hand side of the vapor dome is called the saturated liquid line, and the right-hand
side of the vapor dome is called saturated vapor line. At a given pressure halfway between the left-hand side of the
dome and the right-hand side of the dome, a fluid will be half liquid and half vapor. At a pressure above the critical
point, a fluid will become a supercritical fluid. The temperature–volume diagram for water is shown in Figure 6.22.
Other liquids have similar T–V diagrams.
Heat flows from regions of high temperature to regions of low temperature, and this has the effect of always increasing
the entropy of a system. In reactor thermal-hydraulics, the entropy is given the symbol S. The entropy has the units of
energy divided by temperature (E/T), and so it has the units of joules per degree Kelvin (J/°K) in the SI unit system. The
term “entropy” was originally coined by Rudolf Clausius in 1865, and it is based on the Greek word εντροπία [entropía],
which stands for a turning toward. (Rudolf Clausius’s picture is also shown in Figure 6.7.) Eventually, everything in the
universe is consumed by the molecular disorder this entropy represents.
In classical thermodynamics, the concept of entropy is defined by the second law of thermodynamics, which
was introduced to the reader earlier in the chapter. The second law states that the entropy of a closed system will
always increase or remain the same. In other words, it is a measure of the tendency of a system to proceed in a
certain direction toward a particular state. It implies that heat always flows from a region of high temperature to
a region of low temperature. This process reduces the molecular order of a system, and therefore, entropy is an
expression of the molecular disorder or randomness it has. Entropy is an important state variable because it can be
used to calculate the internal energy U of a material. It can also be used to determine how much of this internal
energy U can be converted into useful work.
The thermal cycle of a Carnot heat engine is illustrated in Figure 6.24. Here, the temperatures of both reservoirs must
be expressed in degrees Kelvin. One of the interesting consequences of a Carnot engine is that the quantity QHIGH/
THIGH = QLOW/TLOW must always be conserved. Thus, the ratio of the heat in each of these reservoirs to the absolute
FIGURE 6.23 A picture of William Thompson (aka Lord Kelvin) on the left, who is credited with proposing the absolute
temperature scale, and Frenchman Sadi Carnot on the right, who is credited with proposing the Carnot heat engine.
(Pictures provided by Wikipedia.)
216 The Laws of Thermodynamics
FIGURE 6.24 A heat engine called the Carnot heat engine sets a theoretical limit on the maximum thermal efficiency that any
heat-generating device can attain. It also sets the maximum thermal efficiency that a nuclear power plant or a coal-fired power plant
can attain.
temperature of each of these reservoirs can be thought of as a conserved quantity that Rudolf Clausius subsequently
interpreted as the entropy S. This also implies that if the entropy S of reservoir 1 is S1, and the entropy of reservoir 2
is S2, then in an ideal heat engine, S1 = S2, and in real heat engines, S1 ≠ S2. This has enormous practical significance
because it enables an engineer to express the amount of heat that can be transferred from a high-temperature region to
a low-temperature region as dQ = TdS, or more precisely, ΔQ = TΔS at each combination of temperature and entropy
that a system possesses. So if there is a closed system where the temperature and the entropy initially start out at T1 and
S1, go through a number of intermediate steps where the values of the temperature and the entropy are T2, S2, T3, S3, …
and TN, SN, etc., and return to the initial state T1 and S1, then the total work W that the system can do is equal to the area
inside of this curve on a T–S diagram.
The thermodynamic efficiency is therefore a direct measure of how efficient this energy conversion process can be. In
most nuclear power plants, the thermal efficiency turns out to be a number between 32% and 42%. In a modern PWR, it
averages 34% to 35%. In addition, there are a number of “tricks” that can be used to increase the value of the numerator
in Equation 6.53 because the denominator is fixed by the thermal output of the power plant. We will present some of
these tricks in Chapter 7.
FIGURE 6.25 An example of a T–S diagram for a typical thermal cycle. On a T–S diagram, the red area inside of the curve repre-
sents the total work that a thermodynamic system can do. This work is equal to the difference between the heat that is added to the
system QIN and the heat that is removed from the system QOUT. In other words, W = QIN − QOUT.
calculate the actual thermodynamic efficiency. In this case, THIGH = 1,273°K and TLOW = 573°K. The thermal efficiency
of the heat engine is then η = 1 − TLOW/THIGH = 1 − (573/1,273) = 0.55 = 55%. [Ans.]
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc. New York,
NY (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi (2000).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
218 The Laws of Thermodynamics
4 4. What dependence does the critical point have on the system pressure?
45. What does the third law of thermodynamics say?
46. What is the difference between power and work?
47. Write a mathematical definition for the quantity called “flow work”.
48. What is the difference between the specific internal energy of a fluid and its specific enthalpy?
49. What is the difference between the enthalpy of a fluid and its specific enthalpy?
50. What is the purpose of the steam tables?
51. What thermodynamic variables do the steam tables attempt to quantify?
52. Fill in the following sentence with the appropriate word or phrase: The entropy of a material is a measure of its
molecular ________.
53. What is the definition of the thermal efficiency of a nuclear power plant? Is this definition any different than the
definition of the thermal efficiency for a coal-fired power plant?
54. Name at least three founders of the modern science of thermodynamics.
Exercise 6.2
A Carnot heat engine takes heat from a high-temperature reservoir at 1,000°C and delivers it to a low-temperature res-
ervoir at 100°C. What is the maximum theoretical efficiency of the engine?
Exercise 6.3
The thermal efficiency of a nuclear power plant is 36%. If the plant generates 3,000 MW of thermal power, how much
electrical power does it produce? If 7% of this electrical power is lost in the transmission lines between the plant and a
nearby city, how much power eventually reaches the city?
Exercise 6.4
Using the Clausius–Clapeyron equation, estimate the boiling point of water on the top of Mt. Everest at an elevation of
8,800 m.
Exercise 6.5
On a P–V diagram, at what temperature and pressure does the triple point for water occur? What is the relationship
between the phases at the triple point?
Exercise 6.6
What is the difference between the specific enthalpy and the specific internal energy of water at a temperature of 300°C
and a pressure of 15.5 MPa (2,250 PSI)? Which of these two fluid properties is greater?
Exercise 6.7
Fluid is pushed in and out of a reactor coolant pipe at a pressure of 500 kPa. If the specific volume of the fluid is 10 m3/kg,
how much flow work is performed by pushing the fluid through the pipe?
Exercise 6.8
The bathroom shower in your home is designed to receive water from the hot water line at a temperature of 140°F and
cold water from the cold water line at a temperature of 50°F. What does the ratio of the mass flow rates from the two
lines have to be so that the average temperature of the water exiting the shower head is 110°F?
220 The Laws of Thermodynamics
Exercise 6.9
Compressors are used in many industrial applications to increase the pressure and temperature of a working fluid.
Suppose that air at 100 kPa and 10°C is steadily compressed by an industrial compressor to 600 kPa and 130°C. The
mass flow rate of the air is 0.1 kg/s, and a heat loss of 16 kJ/kg occurs during the compression process. Assuming that the
changes in the kinetic energy and the potential energy of the air are negligible, how much energy is required to power
the compressor?
Exercise 6.10
In Chapter 19, we will learn that steam turbines perform the opposite function of reactor coolant pumps. Suppose that
the power output of a steam turbine is 5 MW. The steam entering the turbine has a specific enthalpy of 3,250 kJ/kg, and
the saturated liquid–vapor mixture leaving the turbine has a specific enthalpy of 2,361 kJ/kg. If the turbine is 100%
efficient, what is the mass flow rate of the steam through the turbine?
Exercise 6.11
Heat flows from a high-temperature reservoir to a low-temperature reservoir at the rate of 1,000 J/s. If the temperature
difference between the reservoirs is 100°K, how much is the entropy of the system changed?
Exercise 6.12
Suppose that the power output of the turbine in Problem 6.10 is increased to 10 MW and the efficiency of the turbine is
decreased to 90%. What is the mass flow rate of the steam through the turbine? What is the approximate value of the
equilibrium quality at the turbine’s exit?
7
Thermodynamic Properties
and Equations of State
7.1
T hermodynamic Properties and Nuclear Power Plants
Nuclear reactors rely on the laws of thermodynamics to function properly, and nuclear power plants rely on these same laws to convert
thermal energy into electric power. (see Figure 7.1). Many of the thermodynamic properties used in the design of nuclear power plants
can be deduced from what are known as thermodynamic state variables. Examples of these state variables are the temperature, the pres-
sure, the volume, and the entropy. Sometimes the internal energy and the enthalpy are also considered to be state variables, but they are
usually considered to be thermodynamic properties because they can be derived from other state variables. The purpose of this chapter
is to discuss the thermodynamic properties of reactor coolants and show how their thermodynamic properties are related to these state
variables. In particular, we would like to discuss the difference between a reactor coolant when it is a liquid and when it is a vapor. The
thermodynamic properties of water and steam are tabulated in a set of famous thermodynamic property tables called the steam tables.
The steam tables are presented in Appendix G. Normally, the steam tables are applied to reactors where hot water and steam are mixed
together to generate electric power. State variables such as the temperature and the pressure can also be used to determine the thermo-
dynamic efficiency of a nuclear power plant. As we pointed out in Chapter 6, any increase in the thermodynamic efficiency allows more
work to be performed from the same amount of heat. In other words, a reactor can be thought of as large thermonuclear furnace in
which copious amounts of thermal energy are produced. The temperature and the pressure of the core and the thermodynamic proper-
ties of the coolant determine how efficiently this heat can be converted into electric energy. The most common reactor coolant is water
FIGURE 7.1 The laws of thermodynamics and thermodynamic property tables form the basis of the nuclear power industry. Heat in the core
produces steam, and this steam is used to turn the blades of a power turbine. This produces electrical energy that is carried by transmission lines to
a nearby city or town. The same process is also used by coal-fired power plants.
221
222 Thermodynamic Properties and Equations of State
(which can be either light or heavy), although other coolants such as liquid metals, hydrogen, helium, and carbon dioxide
can also be used (see Chapter 1). Now let us discuss how these coolants are used in nuclear and coal-fired power plants.
7.2
Comparisons to Coal-Fired Power Plants
There are many similarities between coal-fired power plants and nuclear ones. In particular, the pressures and the
temperatures in coal-fired power plants are similar to those in nuclear ones. If a coal-fired power plant is cooled by water,
then the same thermodynamic properties that are used for the coolant in a coal-fired power plant can be used for the cool-
ant in a nuclear one. Generally speaking, coal-fired power plants operate at higher temperatures and pressures than nuclear
ones. This is because the temperatures in a nuclear plant cannot exceed the melting point of the fuel, while in a coal-fired
power plant, the energy production process relies upon burning, melting, or vaporizing the fuel. In other words, the water
in a boiler of a coal-fired power plant can actually contain more energy than the same water in the core of a nuclear plant.
Because of this, the thermodynamic efficiency of most fossil-fired power plants is higher than the thermodynamic effi-
ciency of nuclear ones. We will have more to say about this in Chapter 9 when we discuss reactor thermal cycles.
7.3
T hermodynamic State Variables
Reactor coolants require two independent state variables to specify their thermophysical properties. If these state
variables are NOT independent, then an additional state variable such as the quality may have to be used. This is the
result of an important thermodynamic principle called the state postulate, which we will discuss in Section 7.14. For
simple substances, the temperature T, the pressure P, and the volume V are related to each other by what is called an
Equation of State. An example of an equation of state (or EOS) is
In this equation, any one of the state variables can be expressed as a function of the other two. For example, the state of a
substance can be determined from a combination of the pressure and the volume (P and V), the pressure and the temperature
(P and T), or the temperature and the volume (T and V). Thus the following combinations of state variables can be used to
specify the state:
In some state equations, the entropy S or the particle count N are also used to specify the state.
7.4
T hermodynamic Equations of State
Most thermodynamic properties that are used in reactor design cannot be measured directly. However, their values can
be deduced from their thermodynamic state. For example, the enthalpy S and the internal energy U can be arbitrarily
set to zero at a known state which is sometimes called the reference state. In classical thermodynamics, this reference
state corresponds to an absolute temperature of zero. Changes in their values can then be determined based on how far
they deviate from this reference state. Some state equations are relatively simple, while other state equations can be very
complex. For most solids and liquids, their thermodynamic properties can be deduced from just two independent state
variables. Thus variables such as the pressure and the temperature can be used to calculate the internal energy or the
enthalpy. Moreover, the temperature and the entropy can be used to calculate the specific volume (which is defined as
the inverse of the density). In these equations, the entropy is represented by the symbol S, the enthalpy by the symbol H,
and the internal energy by the symbol U. Usually these properties are also tabulated on a per unit mass basis by dividing
the entropy, enthalpy, and internal energy by the mass of the substance for which these properties are found. This leads
to what are called the specific entropy,* the specific enthalpy, and the specific internal energy, which are defined by
Specific entropy s = S/M (J/kg °K)
Specific enthalpy h = H/M (J/kg)
Specific internal energy u = U/M (J/kg)
* Sometimes the degree symbol ° is suppressed when quoting the absolute temperature in SI units, but we will not adhere to this
convention in this book to avoid confusing the absolute temperature with Boltzmann’s constant and the thermal conductivity.
7.5 The Equation of State for the Ideal Gas Law 223
The specific volume υ = V/M (in m3/kg) is determined in a similar way. Notice that the density ρ then becomes the inverse
of the specific volume υ since υ = 1/ρ and this requires the density of a substance to be expressed in kg/m3 or g/cm3.
7.5
T he Equation of State for the Ideal Gas Law
One of the most famous equations of state used in nuclear science and engineering is the ideal gas law, which states that
In this equation, P is the absolute pressure of the gas, V is the volume of the gas, T is the absolute temperature of the gas
(in °K), and Ru is the universal gas constant, which is independent of the type of gas (i.e., hydrogen, helium, or carbon
dioxide) that is used. The gas constant R is then related to the universal gas constant Ru by
R = R u M (7.4)
where M is its molar mass. In chemical engineering, the molar mass is also called its molecular weight. The gas con-
stant R which appears in Equation 7.4 is expressed in kJ/(kg °K) or kPa m3/(kg °K), and the gas constants for common
industrial gases are shown in Table 7.1. The universal gas constant Ru is the same for all of these gases and it has the
value of
P1V1 T1 = P2 V2 T2 (7.6)
TABLE 7.1
The Gas Constants for Common Industrial Gases That Can Be Used as Reactor Coolants and Their Behavior at the Critical Point
(see Figure 7.4)
Gas Constants, Molar Masses, and Other Fluid Properties at the Critical Point
Used as a
Gas Constant Molar Mass Temperature Pressure Volume Reactor
Gas Symbol (kJ/kg °K) (kg/kmol) (°K) (MPa) (m3/kmol) Coolant
Air ATM (varies) 0.287 28.97 132.5 3.77 0.0883 No
Carbon dioxide CO2 0.189 44.01 304.2 7.39 0.0943 Yes
Helium He 2.077 4.003 5.3 0.23 0.0578 Yes
Hydrogen H2 4.124 2.016 33.3 1.30 0.0649 Yes
Oxygen O2 0.260 31.999 154.8 5.08 0.0780 No
Water vapor H2O 0.462 18.015 647.1 22.06 0.0560 Yes
224 Thermodynamic Properties and Equations of State
FIGURE 7.2 A picture of Englishman Robert Boyle and his gas law, which states that the product of pressure and the volume is a
constant for a given mass of confined gas as long as its temperature is constant.
where the subscript 1 refers to state 1 and the subscript 2 refers to state 2. Notice that the ideal gas law uses just three
independent state variables: T, V, and P to specify the state of any gas. If we know the values of any two of these state
variables we can then find the third. The ideal gas law also allows us to find another important thermophysical property
called the specific volume of the gas. The specific volume is defined by
where M is the molar mass of the gas. With this definition, the ideal gas law can then be written as
where R = Ru/M, and υ is the specific volume of the gas. In this equation, the specific volume has the units of m3/kg or
cm3/g, and the density ρ = 1/υ, which is the inverse of the specific volume, has the units of kg/m3 or g/cm3. Equation 7.8
was first proposed by Englishman Robert Boyle, whose picture is shown in Figure 7.2. Under certain conditions, gaseous
reactor coolants (and sometimes even low pressure steam) can be assumed to behave as ideal gases. We will have more
to say about this in the next section.
7.6
Treating Steam as an Ideal Gas
One of the questions that often arises in the study of reactor thermal cycles is whether the steam produced in a nuclear
power plant can be treated as an ideal gas. The answer to this question depends on the pressure of the steam. At pressures
below about 10 KPa (~1.5 PSI), water vapor, regardless of its temperature, behaves like an ideal gas and the ideal gas law
can be used to represent its state. However, at higher pressures (>1,000 PSI), such those that are encountered in pressur-
ized water reactors (PWRs) and boiling water reactors (BWRs), the ideal gas law produces large errors (on the order of
100%), and these errors become even larger in the vicinity of the critical point and near the saturated vapor line. For this
reason, the ideal gas law CANNOT be used in thermal water reactors or in any reactor where steam is produced and
cycled at very high pressures. However, it can be used to study the thermal performance of gas reactors, which we will
have more to say about in Chapter 8.
The pressures and temperatures where steam can be treated as an ideal gas are shown in Figure 7.3. The green region
corresponds to the pressures and temperatures where deviations from the ideal gas law are small (in other words, less
than 1%). Thus Figure 7.3 shows that at 260°C and 5 MPa (which are close to the operating range of a commercial BWR)
the percent deviation from the ideal gas law is about 25%. Similarly, at 10 MPa and 310°C (which are close to the operat-
ing range of a commercial PWR), the percent error is about 50%. Hence the ideal gas law is not appropriate to use for
water vapor or steam under these conditions.
7.8 Other Equations of State 225
FIGURE 7.3 The percentage error 100 × (υACTUAL − υIDEAL)/υACTUAL involved in approximating steam as an ideal gas in a nuclear
power plant is given by the compressibility factor Z. The region in green indicates the regions where steam can be treated as an
ideal gas and the overall error using the ideal gas law as the equation of state is less than 1%.
7.7
Measuring the Deviation of Steam from an Ideal Gas
The deviation of a vapor or gas from the ideal gas law can be expressed in terms of a parameter called the error correction
factor Z. This factor is sometimes called the compressibility factor for the gas, and it is defined by
where υ is the specific volume. Using this correction factor, the “ideal” gas law can then be written as
Pυ = ZRT (7.10)
The compressibility factor is therefore equivalent to
Z = υ ACTUAL υ IDEAL (7.11)
where υIDEAL = RT/P. When a gas obeys the ideal gas law, it is then easy to see that Z = 1. However, real gases can have
values of Z greater than 1.0 or less than 1.0. The value of Z is important when studying the behavior of gas-cooled
reactors, which we will discuss in Chapter 9.
7.8
O ther Equations of State
In addition to the ideal gas law, many other equations of state have been proposed over the years. These equations of
state are more accurate than the ideal gas law over a wider range of conditions; however, they are also more complex. For
monatomic and bi-atomic gases, these equations include the van der Waals equation of state, the Beattie–Bridgeman
equation of state (which was first proposed in 1928) and the Benedict–Webb–Rubin equation of state (which was first
proposed in the early 1940s). The ubiquitous van der Waals equation of state was proposed by Johannes Diderik van der
Waals in 1873. These equations predict the behavior of most gaseous substances better than the ideal gas law, and they
also have a much wider range of applicability. They do so by incorporating additional terms into the conventional form
of the ideal gas law to account for the molecular attraction that the ideal gas law is unable to do.
For example, the van der Waals equation of state improves the accuracy of the ideal gas law by introducing two
additional terms to account for the molecular attraction between the molecules and the volume actually occupied by the
molecules. The most commonly accepted form of the van der Waals equation of state is
FIGURE 7.4 The location of the critical point for a pure material can be found at the top of the vapor dome on a P–V diagram.
The values of TCR and PCR are the critical temperatures and pressures at this point. The critical point is the end point of the phase
equilibrium curve. The most well-known example is the liquid–vapor critical point under which both liquid and vapor can coexist
for the critical isotherm.
and TCR and PCR are the temperature and pressure of the gas at the critical point shown in Figure 7.4. In other words,
the values of A and B are based on the critical point behavior alone. Normally, the values of A and B are tabulated in
what is called a lookup table. Lookup tables typically contain the values of A and B for between 50 and several hundred
pure substances. The Beattie–Bridgeman equation of state and the Benedict–Webb–Rubin equation of state use the
same approach as the van der Waals equation except that they have five and eight experimentally determined constants,
respectively. Other material scientists and engineers have proposed expressing the equation of state for pure substances
by a similar expression of the form
where υ is the specific volume. Equation 7.14 is sometimes called the virial equation of state and the corresponding coef-
ficients A(T), B(T), C(T), etc. are called the virial temperature coefficients. These temperature coefficients are functions
of the temperature alone, and in some cases, they can be deduced from the principles of statistical mechanics. The van
der Waals equation of state and the other equations of state can be used for the gaseous phase of most materials, but they
should NOT be used for the liquid phase or for a liquid–vapor mixture. Different equations of state should be used for
these substances. As far as vapors and gases are concerned, the Benedict–Webb–Rubin equation is the most accurate of
the three. However, it requires eight empirical coefficients rather than just one or two to achieve this additional accuracy.
7.9
I nferring the Behavior of Reactor Coolants from Other State Variables
A person looking at the property tables for the first time may notice that every table contains a thermodynamic property
called the enthalpy H. The enthalpy is different than the internal energy U of a substance but these two fundamental
properties are also related. The enthalpy is a property that is unique to the behavior of fluids, while the internal energy
can be used to describe the behavior of either solids or liquids that do not perform external work. (such as uranium, steel,
or nuclear fuel). The internal energy can be thought of as a measure of the vibrational energy of the atoms and mol-
ecules in a material. As the absolute temperature increases, the internal energy (which is due to this molecular agitation)
7.10 Defining the Enthalpy and Specific Enthalpy of a Fluid 227
increases as well. The use of the enthalpy as a fluidic property was first proposed by German scientist Richard Mollier
in the early 1900s. Mollier was one of the first engineers to recognize the importance of enthalpy in the design of power
producing devices such as steam turbines. The properties of steam can be represented in either tabular or graphical
form, but in graphical form, they give rise to what are called the Mollier diagrams. Examples of some Mollier diagrams
are presented in Section 7.30. Conversely, the enthalpy H can be thought of as the total energy content of a fluid, which
is different than its internal energy U. The enthalpy is always greater than the internal energy, and on a per unit mass
basis (see Chapter 6), its value is offset by an amount equal to υP, where υ is the specific volume of the fluid and P is its
absolute pressure. Because of this, we can show that
Next, we would like to discuss how these important properties are related.
7.10
D efining the Enthalpy and Specific Enthalpy of a Fluid
Thermodynamically speaking, the enthalpy H is a measure of the total energy content of a fluid. This energy can be
subdivided into the internal energy U plus the amount of external work W that a fluid is capable of performing. So the
formal definition of the enthalpy is
where the external work W is equal to the product of the pressure of the fluid P and its volume V. Sometimes this external
work is also called PV work or flow work and these distinctions are presented in Figure 7.5. Hence, an alternative defini-
tion of the enthalpy, which is specific to fluids, is
In Equation 7.17, both the internal energy U and the enthalpy H have the units of joules. Now in general, the internal
energy U is also a function of the temperature and the pressure, and so the enthalpy is sometimes written as
H = U(P,T) + PV (7.18)
However, in engineering applications, it is usually more convenient to express the values of the internal energy and the
enthalpy on a per unit mass basis. If the internal energy is U, the enthalpy is H, and the mass of a fluid containing these
properties is M, then the internal energy per unit mass u and the enthalpy per unit mass h are given by
FIGURE 7.5 The product of pressure and volume always has the units of energy or work. Hence, the integral of the pressure times
the volume can be used to find the amount of work that is performed.
228 Thermodynamic Properties and Equations of State
In this case, u and h are called the specific internal energy and the specific enthalpy of the fluid. Accordingly, the
specific internal energy u and the specific enthalpy h have the units of J/kg (or kJ/kg). Thus, a slightly more practical
definition of the enthalpy is
Since V/M is the same as the specific volume υ, we can also write Equation 7.20 as
This is then the definition the majority of the nuclear industry uses. Hence the specific internal energy is given by
u = h − υP (7.21b)
and therefore, the specific internal energy is always less than the specific enthalpy by an amount equal to υP. For PWRs,
this difference is illustrated in Problem 7.1. Hence, if we know the mass flow rate through the core in kg/s, then multiply-
ing the mass flow rate m = dm/dt by the specific enthalpy h allows us to calculate the number of watts of thermal energy
that are being removed from the core:
An Expression for the Amount of Thermal Energy the Coolant Can Remove from a Reactor Core
Thermal energy removed = dm/dt h = mh
(7.22)
Watts of thermal energy removed = kg/s × J/kg = J/s
This expression forms the basis of the energy balance that is performed in most nuclear power plants.
7.11
Finding the Enthalpy of a Two-Phase Mixture
When water in a reactor core boils (i.e., when it transforms itself from a liquid to a liquid and a vapor), one can measure
the specific enthalpy of the liquid phase hl and the specific enthalpy of the vapor phase hv independently. For water-steam
mixtures these values are tabulated separately in what are called the steam tables. A complete list of these properties for
saturated water and steam are provided in Appendix G. In addition to these properties, the enthalpy of vaporization hlv,
which represents the amount of energy required to convert 1 kg of the liquid phase to the vapor phase is also presented
as a function of the absolute pressure P. (see Figure 7.6). Thus, the enthalpy of the vapor hv must be equal to the enthalpy
of the liquid hl plus the enthalpy of vaporization hlv:
h v = h l + h lv (7.23)
7.12 Phase Changes in Reactor Coolants 229
FIGURE 7.6 A phase change diagram showing the transition of water from a solid to a liquid to a gas.
where the values of h in each case have the units of J/kg (or kJ/kg). It is also possible to express the total enthalpy h of
the liquid–vapor mixture as
h = xh v + (1 – x)h l (7.24)
where x is another thermodynamic variable called the quality. When we rewrite Equation 7.24 in a slightly different
form we see that
h = h l + x ( h v – h l ) = h l + xh lv (7.25)
h lv = h v − h l (KJ/kg) (7.26)
Thus, hlv represents the difference between the specific enthalpy of the vapor phase and the specific enthalpy of the
liquid phase. Because hv > hl, it is always a positive number with the units of J/kg or kJ/kg. Moreover, as the steam tables
show, it is also pressure dependent, and it tends to fall as the absolute pressure rises. In the secondary loop of a PWR
at 8 MPa, the values for hl, hlv, and hv are 1,317.3, 1,441.4, and 2,758.7 respectively. We will have more to say about the
effect of P on hlv in Section 7.19.
7.12
Phase Changes in Reactor Coolants
When a reactor coolant starts to boil, it undergoes what is called a phase change. When this occurs, a fraction of
the liquid phase is converted into the vapor phase. The exact fraction that is converted into a vapor depends on
how much energy the coolant absorbs. Phase changes occur in nuclear power plants all of the time, and they can
occur in either the primary or the secondary loops - depending on the design of the plant. The temperature and the
pressure at which this phase change occurs are called the saturation temperature and the saturation pressure, and a
fluid that has absorbed just enough energy to boil is called a saturated liquid. Before it reaches this point, it is then
referred to as a subcooled liquid. Compressing a subcooled liquid will increase its boiling point, and this is why
reactor coolants (such as pressurized water) boil at much higher temperatures than they do at atmospheric pressure
(100 kPa or 1 ATM).
After boiling begins, the temperature remains constant until the coolant is completely vaporized. Adding more
energy to the liquid phase simply converts it into more vapor. Eventually, a point is reached where all of the liquid has
230 Thermodynamic Properties and Equations of State
been vaporized, and this type of vapor is called a saturated vapor. Two phase mixtures between a saturated liquid and
a saturated vapor are called a saturated liquid–vapor mixture because the liquid and vapor phases coexist in thermal
equilibrium, although they do not necessarily exist in equal amounts. Once all that remains is saturated steam, adding
more energy to this steam will only increase its ambient temperature. Once the temperature of this steam exceeds its
saturation temperature, it then becomes known as superheated steam. Superheated steam has more energy than ordinary
steam, and the hotter this steam gets, the more efficient the plant thermal cycle will be.
7.13
P roperty Diagrams
This phase change process for a liquid–vapor mixture is illustrated in Figure 7.6. If the pressure is increased or
decreased, only the path the mixture takes through this diagram is changed. In property diagrams, the temperature or
the pressure is normally plotted on the vertical axis, and another independent variable (such as the specific volume, the
entropy, or the enthalpy) is plotted on the horizontal axis. Depending on the physical process that is being modeled, the
resulting phase diagram is called a P–V diagram, a T–V diagram, a P–T diagram, a T–h diagram, or a T–S diagram.
All of these diagrams represent various aspects of reactor thermodynamic behavior. In applications involving nuclear
power plants, most of these phase diagrams contain a liquid region, a vapor region, and a region where the resulting
mixture is partially liquid and partially vapor. Later we will learn that T–S diagrams are particularly helpful because
the area encapsulated by a reactor coolant as it transverses a T–S diagram is equivalent to the amount of external work
it performs.
The region between the saturated liquid line (on the left) and the saturated vapor line (on the right) is called the
vapor dome. If the coolant represented by this dome is ordinary water, then the vapor dome is called a steam dome.
The region immediately to the left of the vapor dome (region 1 in Figure 7.7) is called the subcooled liquid region,
and the region immediately to the right of the vapor dome (region 3 in Figure 7.7) is called the superheated vapor
region. The region under the vapor dome (region 2) is called the two-phase mixture region. Any coolant that under-
goes a phase change will pass through all three of these regions. At the top of the vapor dome, the liquid and vapor
phases blend together, and when this occurs, it becomes impossible to distinguish them from each other. On T–V,
P–T, or T–S diagrams, the point at which they become indistinguishable is called the critical point. The triple point
for water at 273.16°K (0°C) occurs at 611.2 Pa. In classical thermodynamics, triple point becomes the basis for the
definition of the Kelvin.
7.14
Using Property Diagrams to Describe Phase Changes
As we first pointed out in Chapter 6, phase diagrams can be uniquely defined in terms of a number of independent ther-
modynamic variables called state variables. Examples of these state variables are the temperature T, the pressure P, the
volume V, and the entropy S. Once enough of these state variables are specified, the values of the other thermodynamic
properties are uniquely defined. The number of state variables required to fix the state of a system is determined by an
empirical principle that has become known as the state postulate. The state postulate says that
State variables are considered to be independent of each other if one variable can be changed while the other is held
fixed. For example, the state of a single-phase liquid can be fixed by the temperature and the specific volume because
they are independent variables. The temperature and the pressure can also be used to fix the thermodynamic state of a
single-phase liquid. However, the temperature and the pressure cannot be used to determine the state of a two-phase
7.14 Using Property Diagrams to Describe Phase Changes 231
FIGURE 7.7 A phase surface showing the phases of a pure substance in three-dimensional space and their relationship to the
temperature, the volume, and the pressure.
mixture because they become dependent when a phase change occurs. Then, a third state variable such as the quality is
required to unambiguously fix the state.
If we plot these variables for ordinary water in three dimensional space, we create what is called a phase surface.
A phase diagram is just a two-dimensional slice of a phase surface. A classic example of a phase surface is shown in
Figure 7.7. This particular surface is called a P–V–T surface. The point at which all three phases co-exist in thermo-
dynamic equilibrium is called the triple point. It is sometimes convenient to plot a combination of two of these three
thermodynamic variables graphically along the x- and y-axes. An example of such a plot would be the pressure and the
volume. Another example would be the temperature and the entropy. The first combination is called a P–V diagram,
and the second combination is called a T–S diagram. Both types of diagrams can be extremely helpful when attempting
to visualize the state of a thermodynamic system when a third thermodynamic variable such as the temperature or the
pressure is fixed. The points lying on a curve of equal temperature are called an isotherm, and the points lying on a curve
of equal pressure are called an isobar. Examples of a P–V diagram and a T–V diagram for water are shown in Figure 7.8.
Many other diagrams of this type can also be drawn. As we mentioned previously, P–V diagrams and T–V diagrams
FIGURE 7.8 A P–V diagram and a T–V diagram for ordinary water.
232 Thermodynamic Properties and Equations of State
are two-dimensional slices of more general three-dimensional diagrams called phase surfaces. In other words, phase
surfaces are just different presentations or graphical representations of the same interrelated data set.
7.15
P ressure–Temperature Diagrams
In classical thermodynamics, pressure–temperature diagrams are used to show the different molecular states a mate-
rial can possess, and P–T diagrams are used to study materials that have solid, liquid, and vapor phases. They are also
helpful in understanding the temperature at which a material will freeze, the temperature at which it will melt, and
the conditions under which it will transform itself into a vapor. P–T diagrams usually consist of three distinct regions
called the solid region, the liquid region, and the vapor region. Each region has a well-defined boundary for a specific
temperature and pressure. For instance, in the phase diagram for water shown in Figure 7.9 water vapor occurs at differ-
ent pressure and temperature combinations than ordinary water, and ice occurs at still others. Most common materials
have a phase diagrams that are similar to this. Each phase diagram also has a line of demarcation between the liquid
and vapor phases that is called a vaporization line. This line represents the specific combinations of temperature and
pressure at which a phase transition occurs.
7.16
Temperature–Volume Diagrams
There is another way to disect a P–V–T diagram, which involves slicing it along the T- and V-axes. This gives rise to
what is called a T–V diagram. Temperature volume diagram help to visualize how a working fluid can be used to extract
useful work from a power source (such as a reactor core). A T–V diagram has a solid dome-shaped region through the
center of it called a vapor dome. Under the vapor dome, the fluid is a two-phase mixture. To the left of it, it is a liquid,
and to the right of it, it is a vapor. The top of the vapor dome ends at what is called the critical point. One can then plot
a line through the vapor dome that corresponds to a state of constant pressure. This line is sometimes called an isobar.
This isobar starts in the liquid region, becomes flat as one crosses the vapor dome, and rises again after one leaves the
right‑hand side of the dome. Each progressively higher value of P corresponds to a different horizontal line that is closer
to the critical point. On the left‑hand side of the dome, the liquid is a subcooled liquid because it has not yet begun to
boil, and on the right hand side, it is called a superheated vapor because it has no liquid left in it. The left‑hand side of
the dome is called the saturated liquid line, and the right‑hand side of the dome is called saturated vapor line. At any
pressure halfway between the left‑hand side of the dome and the right‑hand side of the dome, the fluid will be a two-
phase mixture. At a pressure above the critical point, the water becomes a supercritical fluid. A temperature–volume
diagram for water is shown in Figure 7.10. Other liquids have T–V diagrams with similar shapes. For several reasons,
T–V diagrams are used more frequently in nuclear science and engineering than P–T diagrams.
FIGURE 7.9 A P–T diagram for high-pressure water. Notice that ice has several different phases at very high pressures.
(Figure provided by MIT.)
7.17 Pressure–Volume Diagrams 233
7.17
P ressure–Volume Diagrams
In reactor work, another way to look at the thermodynamic performance of a power plant is a P–V (or pressure–volume)
diagram. A P–V diagram looks very similar to a T–V diagram. The primary difference is that the vertical axis now rep-
resents the pressure rather than the temperature. For water, a P–V diagram still contains a vapor dome However, instead
of constant pressure lines (isobars), the lines across the dome are constant temperature lines (isotherms). A second
significant difference is that the constant temperature lines start out much higher in the liquid part of the diagram, and
end much lower in the vapor part of the diagram. In a T–V diagram, the lines run in the opposite way. The temperature
increases in the vertical direction—just like the pressure in a T–V diagram does. Thus both diagrams have many
common characteristics. The temperature–volume diagram for water is shown in Figure 7.11.
FIGURE 7.11 The T–V diagram for water. The pressures are shown in MPa, and the void fraction is shown in black. (From http://
thermodynamik.hs-zigr.de/fpc/range_of_validity/range_of_validity_water.htm.)
234 Thermodynamic Properties and Equations of State
7.18
D efining the Saturation Temperature and Pressure
Irrespective of how phase diagrams are drawn, liquids have specific pressure and temperature combinations where
they start to boil. When the pressure is fixed, the temperature at which a coolant begins to boil is called its saturation
temperature. For ordinary water, this occurs at atmospheric pressure (14.7 PSI or 100 kPa) at a temperature of 212°F or
100°C. In other words, the saturation temperature is another term for the boiling point. Liquid in a vacuum has a lower
boiling point than the same liquid at atmospheric pressure. In other words, the boiling point is a function of the pressure.
Moreover, for the same pressure, different liquids boil at different temperatures. This is due to the molecular structure of
the liquid, as well as its surface tension. (The effects of surface tension are discussed in Chapter 23.) Saturated liquids
contain as much thermal energy as possible without boiling. Thus when a fluid reaches the point where it cannot hold
any more energy without undergoing a phase change, it is said to be completely saturated. Liquids can also be converted
into vapors at temperatures far below their saturation temperatures through evaporation. When liquids evaporate, some
of their thermal energy is spontaneously dissipated in the form of vapor molecules. Evaporation occurs close to the
surface of an object where the molecules are not subject to the same molecular forces as the molecules further inside the
material. Hence evaporation can occur at the surface of both solids and liquids. Normally it is more closely associated
with liquids (see Chapter 23).
7.19
T he Clausius–Clapeyron Equation
As we first mentioned in Chapter 6, the boiling point of a reactor coolant is a function of its pressure. If the heat of
vaporization and the vapor pressure are known at a specific temperature, its boiling point can be determined from what
is known as the Clausius–Clapeyron equation. In essence, the Clausius–Clapeyron equation states that the boiling point
TBOIL is related to the heat of vaporization ΔhVAP in the following way:
where
TBOIL = the boiling point (°K)
R = the ideal gas constant (which has a value of 8.314 J/°kmol)
P = the current pressure (atm)
T = the current temperature (°K)
ΔhVAP = the heat of vaporization of the liquid (J/mol) = hv − hl
ln = the natural logarithm to the base e
Specifically, Equation 7.27 states that the boiling point increases with increasing pressure up to a certain point called
the critical point, in which case the vapor phase can no longer be distinguished from the liquid phase. In other words,
the vapor phase and the liquid phase eventually reach a temperature and a pressure where their properties become
indistinguishable. This point is called the critical point and the critical point for a water-steam mixture occurs at 374 °C
and 22 MPa (see Figure 7.12). The boiling point cannot be increased beyond the critical point. Likewise, the boiling
point decreases with decreasing pressure until the triple point is reached. Then, the b oiling point cannot be reduced
below the triple point.
7.20
D etermining the Boiling Point from Clausius–Clapeyron Equation
When a phase change occurs, the pressure and the temperature can no longer be treated as independent variables.
Instead, they become dependent on each other and the temperature at the saturation point TSAT becomes a function of
the saturation pressure PSAT; that is, TSAT = f(PSAT). The Clausius–Clapeyron equation can be used to determine the
shape of the vapor pressure saturation curve under these conditions. Various applications exist to calculate the saturation
temperature from the vapor pressure saturation curve. On the Internet, these applications can be found at
☉☉ http://www.wikicalculator.com/formula_ calculator/Molar-enthalpy-of-vaporization-(The-Clausius-
Clapeyron-equation)-438.htm
☉☉ http://calistry.org/calculate/clausiusClapeyronCalculator
7.22 The Quality of a Two-Phase Mixture 235
FIGURE 7.12 Two diagrams showing the critical point where water and steam become indistinguishable from one another.
Example Problem 7.3 helps to illustrate another important feature of the Clausius–Clapeyron equation. It shows how
increasing the atmospheric pressure changes the boiling point of seawater. Although this example is rather abstract, it
serves to illustrate how the ambient pressure affects the boiling point.
7.21
P roperties of Liquid–Vapor Mixtures
Now, suppose we would like to calculate the physical properties of water after it changes phase. After a phase change
occurs, some of the coolant becomes a saturated liquid and some of the coolant becomes a saturated vapor. If the phase
change occurs slowly enough, both the liquid and vapor phases will have the same temperature, and then they are said
to be in thermal equilibrium with each another. However, their phase velocities may still be different, and the vapor may
move at a different speed than the liquid. We will have more to say about this in Chapter 23.
7.22
T he Quality of a Two-Phase Mixture
The properties of two-phase mixtures can be found using another thermodynamic property called the quality of the
mixture (see Section 7.11). In a thermodynamic sense the quality represents the ratio of the mass of the vapor to the
mass of the entire mixture; that is
since MMIXTURE = MVAPOR + MLIQUID. Hence the quality is a dimensionless number between 0 and 1.0; that is, 0 <= ×
<= 1.0. When a two phase mixture consists entirely of saturated liquid, the quality is 0, and when the mixture consists
entirely of saturated vapor, the quality is 1.0. If half of the mass of the mixture is liquid and the other half of the mass of
the mixture is vapor, then the quality is 0.5. The definition of the quality is depicted graphically in Figure 7.13.
236 Thermodynamic Properties and Equations of State
FIGURE 7.13 A diagram showing the definition of the quality and how the quality increases between a saturated liquid
(on the left) and a saturated vapor (on the right).
7.23
Using the Quality to Define the Properties of a Two-Phase Mixture
For two-phase mixtures, the specific volume may not be initially known. However, if the masses of the liquid and vapor
phases are known, then the total mass of the mixture is
M MIX = M LIQUID + M VAPOR = M L + M V (7.30)
Then, the physical properties of the mixture become the average properties of the mixture. For example, in the case of
the specific volume, the average specific volume becomes
υ MIX = (1 – x) υ L + xυ V (7.31)
where υL and υV are the specific volumes of the liquid and vapor phases, respectively. If the flow is homogeneous and
both phases move at the same speed, then the mixture density can also be written as
since υ = 1/ρ. Notice that the quality in this case can be thought of as the mass fraction of the vapor phase. Similarly, the
specific internal energy and the specific enthalpy of the mixture can be defined as
where, of course, x is the aforementioned mixture quality. Finally, it is easy to see that we can rewrite these equations as
u MIX = u L + x ( u V − u L ) (7.35)
h MIX = h l + x ( h v − h l ) (7.36)
where hvl = hv − h l is the enthalpy of vaporization. Thus the enthalpy of vaporization represents the amount of
thermal energy required to convert a u nit mass of saturated liquid into a unit mass of saturated vapor. Hence it
is a tabulated quantity, and for different temperatures and pressures, its values can be found in Appendix G. In a
7.24 Relationships between the Quality and the Void Fraction 237
FIGURE 7.14 A T‑H diagram for water. The pressure is in MPa, and the void fraction is shown in black. (From http://thermody-
namik.hs-zigr.de/fpc/range_of_validity/range_of_validity_water.htm.)
BWR core, its value is approximately 1500 kJ/kg. Sometimes it is easier to discuss what we have just said in terms
of a T–h diagram, such as the one shown in Figure 7.14. Inside of the vapor dome the mixture specific volume and
the mixture specific enthalpy can be found by combining the relative amounts of each phase between the right hand
side and the left hand side of the dome.
7.24
Relationships between the Quality and the Void Fraction
The properties of two-phase mixtures we have discussed so far are mass based. That is, they depend on the relative
mass fractions of the liquid and vapor phases—which are in turn represented by the quality x. However, sometimes the
aggregate properties of two-phase mixtures can be expressed more conveniently in terms of their volume fractions. This
can be done using another thermodynamic property called the void fraction. For two-phase mixtures where both phases
are moving at the same speed (and generally in the same direction), the void fraction α is simply the ratio of the volume
of the vapor to the volume of the vapor–liquid mixture; that is,
where VV is the volume occupied by the vapor phase and VL is the volume occupied by the liquid phase. Note that the
void fraction is a dimensionless number, and just like the quality, it has a value between zero and unity. Pure liquids
have a void fraction of zero, and pure vapors has a void fraction of 1.0. Hence, the void fraction is a convenient way to
calculate the density of a two-phase mixture. From Equation 7.37, the average density of the mixture is then
where ρL and ρV are the densities of the liquid and vapor phases, respectively. In reactor work ρV is usually much smaller
than ρL and so the average density of most two-phase mixtures can be written as
ρMIX ≅ (1 − α)ρL (7.39)
238 Thermodynamic Properties and Equations of State
Hence, ρMIX decreases as α increases. For most applications the value of α is considerably larger than the value of x.
Using the definitions presented above and in Appendix G, it can then be shown that the value of the void fraction is
related to the value of the equilibrium quality by
Equation 7.40 an then be used to find the quality from the void fraction for an equilibrium mixture. Conversely, we can
also solve Equation 7.40 for the void fraction as a function of the quality:
The reader can then see that when x = 0, α =0, and when x = 1, α =1. For reactor work ρl /ρV can be a very large num-
ber, and thus very small qualities can lead to very large void fractions. (The exact values are presented in Chapter 23.)
Normally, Equations 7.40 and 7.41 are used in the design of BWRs where the coolant in the upper half of the core boils.
At pressures of 7.25 MPa (~ 1050 PSI) which are representative of modern BWR cores, an equilibrium quality of 10%
corresponds to a void fraction of about 70%. (The exact relationship is presented in Figure 7.15). In general, the void frac-
tion increases much more rapidly than the quality, and this relationship is not a linear one. Finally, if the liquid and vapor
phases happen to move at different speeds, an additional correction factor must be added to Equations 7.40 and 7.41 to
account for this difference in phase velocity. This “correction factor” is called the slip ratio, and as we shall see shortly,
it represents the ratio of the velocity of the vapor phase to the velo0city of the liquid phase; that is S = slip ratio = vv/vl.
Normally, the slip ratio in a BWR core has a value between 1.0 and 4.0, although during some transients, it can become as
large as 8. Thus if the slip ratio is unity (S = 1), both phases move at the same speed, and Equations 7.40 and 7.41 represent
exact relationships between the quality and the void fraction for a two-phase mixture in thermodynamic equilibrium.
FIGURE 7.15 A picture showing how the void fraction and the quality are related to each other when the liquid and vapor phases
are moving at the same speed and in the same direction. These conditions correspond to what is known as a slip ratio of 1. The value
of the void fraction α and the quality x in a typical BWR core are shown as a function of axial elevation on the left. In practice, the
values shown here may differ slightly from one BWR fuel assembly to the next.
7.26 The Steam Tables 239
7.25
Finding the Entropy of a Two-Phase Mixture
Finding the entropy of a two-phase mixture is similar to finding the value of other thermodynamic properties. The
entropy S is also tabulated in fluid property tables, and for saturated water and steam, these values are provided in
Appendix G. Normally, the entropy used in this case is the specific entropy s, which is the absolute value of the entropy
per unit mass
If sL is the entropy of the saturated liquid and sV is the entropy of the saturated vapor, then the entropy of the mixture is
also given by
s = (1 – x)s L + x s V (7.43)
where x is the mixture quality. Notice that Equation 7.43 can also be written as
s = s L + x s LV (7.44)
where sLV is the entropy of vaporization. In general, the value of sLV is pressure and temperature dependent. A T–S
diagram that shows the specific entropy as a function of the absolute temperature is presented in Figure 7.16.
7.26
T he Steam Tables
In reactor work, property tables are available for many reactor coolants—and even for many materials that are not reac-
tor coolants. However, none of these property tables has received more notoriety than the property tables for water and
steam because almost all nuclear power plants use some version the Rankine thermal cycle, which we will discuss in
Chapter 9. Historically, these tables have become known as the steam tables because they contain the physical properties
of both saturated water and steam. Copies of the steam tables are available from many reputable sources and they can
also be obtained in digital form from the ANS or the ASME. In recent years, property tables for many reactor coolants
have also appeared on the Internet. An excellent example of a site where the physical properties of water and steam can
be found is http://thermodynamik.hszg.de/fpc/index.php.
FIGURE 7.16 A T–S diagram for water showing the characteristic vapor dome. (Diagram provided by Wikipedia.)
240 Thermodynamic Properties and Equations of State
FIGURE 7.17 The range of temperatures and pressures where the thermodynamic properties of water, steam, and water–steam
mixtures can be obtained at http://thermodynamik.hszg.de/fpc/index.php. The fluid properties that are available at this site include
the saturation temperature and the saturation pressure, the fluid quality, the internal energy, the density, the specific volume, the
enthalpy, and the specific entropy.
This site contains numerous applications that can also be accessed using an i-Phone, an i-Pad, and most Android
tablets and devices. The physical properties for water and steam this site presents reflect the recommendations of the
International Association for the Properties of Water and Steam (the IAPWS). Properties for water and steam are avail-
able from 0°C to 800°C at pressures between 0.0006 and 100 MPa and temperatures up to about 2,000°C. The water
and steam physical property data encompassed by the IAPWS recommendations is shown in Figure 7.17. In most cases,
the property tables are presented for each physical state that a specific fluid can attain.
7.27
T hermodynamic Properties of Water and Steam
The property data in the steam tables is presented in such a way that it allows us to make an important distinction
between the thermo physical states of water and steam. Liquid water that is below the saturation temperature is
referred to subcooled water, and a separate table is sometimes used to present its thermodynamic properties. In this
book, this data is presented in Appendix F. Once this water reaches its saturation temperature, it is then called satu-
rated water and its properties are presented in Appendix G. When additional energy is added to this saturated water,
it transforms itself into steam and the fraction of its mass that becomes steam is called its static quality. Steam having
a quality greater than 0 and less than 1.0 is called wet steam or saturated steam, and its thermodynamic properties are
presented as a function of the pressure in Table 7.3. When all of the remaining water is converted into steam and no
liquid water remains, the quality of the mixture becomes equal to 1.0 and the resulting steam is called dry steam. The
thermodynamic properties of dry steam are presented as a function of the system pressure in Appendix H. Finally,
when the temperature of the dry steam increases beyond the saturation temperature, the steam is called superheated
steam, and the properties of superheated steam are presented as a function of pressure in Appendix H. This implies
that five property tables are required to describe the thermo-physical properties of water and steam. These property
tables include
1. A subcooled property table (for subcooled water)
2. A saturated property table (for saturated water)
3. A wet steam property table (for the saturated water–steam mixture)
4. A dry steam property table (for saturated steam)
5. A superheated steam property table (for superheated steam)
The reference conditions under which each of these tables is valid are summarized in Table 7.2. The steam tables for wet
steam where 0 < x < 1.0 are shown in Table 7.3. The steam tables for subcooled water, saturated water, saturated steam,
and superheated steam are presented in Appendices F, G, and H.
7.27 Thermodynamic Properties of Water and Steam 241
It is important to keep in mind that significant differences can exist between the thermodynamic properties of sub-
cooled water, saturated water, wet steam, dry steam, and superheated steam. For example, at 7 MPa, subcooled water at
200°C has a specific enthalpy of 855 kJ/kg, while saturated water has a specific enthalpy of 1,267 kJ/kg. Wet steam with
a quality of 50% has a specific enthalpy of 2,020 kJ/kg, and dry steam has a specific enthalpy of 2,772 kJ/kg. Finally, the
same steam with 50°C of superheat has a specific enthalpy of 2,969 kJ/kg. The reader should bear in mind that numbers
are also pressure dependent. The specific volume, the density, and the specific entropy behave in similar ways. The con-
ventions used to distinguish between these different forms of water and steam are summarized in Table 7.2. In addition
TABLE 7.2
The Definitions for Various Types of Steam and Water and the Conditions under Which They Exist
Type of Coolant Reference Conditions Location of Selected Physical Properties
Subcooled water T < TSAT; x = 0 Appendix F
Saturated water T = TSAT; x = 0 Appendix G
Wet steam T = TSAT; 0 < x < 1.0 Table S.3
Dry steam T = TSAT; x = 1.0 Appendix H
Superheated steam T > TSAT; x = 1.0 Appendix H
TABLE 7.3
The Thermodynamic Properties of Wet Steam
Atmospheric Pressure (~0.1 MPa)
Saturation Temperature = Specific Volume Density Specific Enthalpy Specific Entropy
100°C υ (m3/kg) ρ (kg/m3) h (kJ/kg) s (kJ/kg °C)
Quality = 5% 0.085692 11.6697 530.312 1.60537
Quality = 10% 0.170341 5.87057 643.188 1.90818
Quality = 25% 0.424288 2.35689 981.815 2.81662
Quality = 50% 0.847533 1.1799 1,546.19 4.33068
Quality = 75% 1.27078 0.78692 2,110.57 5.84474
Quality = 90% 1.52472 0.655856 2,449.2 6.75318
2 MPa
Saturation Temperature = Specific Volume Density Specific Enthalpy Specific Entropy
212.38°C υ (m3/kg) ρ (kg/m3) h (kJ/kg) s (kJ/kg °C)
Quality = 5% 0.0060969 164.016 1,003.11 2.64164
Quality = 10% 0.0110171 90.7676 1,097.6 2.83624
Quality = 25% 0.0257777 38.7932 1,381.06 3.42006
Quality = 50% 0.0503787 19.8496 1,853.5 4.3931
Quality = 75% 0.0749797 13.3369 2,325.95 5.36613
Quality = 90% 0.0897387 11.1435 2,609.39 5.94992
4 MPa
Saturation Temperature = Specific Volume Density Specific Enthalpy Specific Entropy
250.35°C υ (m3/kg) ρ (kg/m3) h (kJ/kg) s (kJ/kg °C)
Quality = 5% 0.00367878 271.829 1,173.1 2.96031
Quality = 10% 0.00610499 163.801 1,258.78 3.12396
Quality = 25% 0.0133836 74.7183 1,515.8 3.61492
Quality = 50% 0.0255146 39.1932 1,944.16 4.43319
Quality = 75% 0.0376457 26.5635 2,372.53 5.25145
Quality = 90% 0.0449227 22.2605 2,629.52 5.74234
(Continued )
242 Thermodynamic Properties and Equations of State
to being sensitive to the system pressure, they are also a function of the mixture quality and the saturation temperature.
Several graphical depictions of the information presented in the steam tables are presented in Figures 7.18 and 7.19.
FIGURE 7.18 A graphical depiction of the information contained in the steam tables for ordinary water.
FIGURE 7.19 Another graphical depiction of the information presented in the steam tables. In this case, a T–S diagram is shown.
7.28
T he Behavior of Air–Water Mixtures
In addition to water–steam mixtures, it is also possible to have water vapor or steam mixed together with air in the
reactor containment building. When this occurs, other effects such as vapor pressure and phase equilibrium must be
taken into account to deduce the properties of the air-water-steam mixture. In this section, we would like to briefly
246 Thermodynamic Properties and Equations of State
discuss how this can be done. If a container such as a containment building contains more than one gaseous substance,
the pressure that the mixture exerts on the walls is the sum of the pressures of the individual substances. This observa-
tion is known as Dalton’s law of partial pressures, and it was first proposed by Englishman John Dalton in 1801 (see
Figure 7.20). If there are N separate substances in the mixture, then the law of partial pressures says that
PTOTAL = P1 + P2 + P3 + + PN (7.45)
where PN is the partial pressure of the Nth component. The molecules of ordinary air can be shown to obey this equation
as well. Only in this case, the air consists of a mixture of dry air (which contains no humidity) and water vapor, which
provides its moisture content. The atmospheric pressure of moist air is the sum of the pressure of the dry air and the
pressure of the water vapor, which is called its vapor pressure. The total pressure of the mixture is then
Normally the vapor pressure is small compared to the total air pressure (less than 3%), but the amount of water vapor in
ordinary air has a major impact on other important processes such as drying and evaporation. For a given temperature
and pressure, air can hold only a certain amount of moisture, and the ratio of the amount of moisture it contains to the
maximum amount of moisture it can hold is called its relative humidity Φ.
Defined in this way the relative humidity will have a value between 0 and 100. Hence, the relative humidity is normally
measured as a percentage of the maximum amount of moisture the air can hold. The vapor pressure of saturated air
is equal to the vapor pressure of water at the saturation temperature. For example, for saturated air at 25°C, the vapor
pressure is 3.17 kPa. The amount of moisture the air can hold is then completely specified by the relative humidity and
the air temperature, and the vapor pressure is related to the relative humidity Φ by
FIGURE 7.20 John Dalton and a statement of his law of partial pressures. (The figure on the left was provided by the University of
Texas, and the figure on the right is from estvideo.net.)
7.31 Thermodynamic Cycles and Path Functions 247
where PSAT is the saturation pressure of water at the ambient air temperature. For example, the vapor pressure of moist
air at 25°C with a relative humidity of 50% is PVAPOR = Φ·PSAT = 0.50 × 3.17 kPa = 1.58 kPa. This is the “humidity” that
we feel on an average day. The total pressure of an air–water vapor mixture is then
However, the amount of moisture the air can hold increases with increasing temperature because the saturation pres-
sure PSAT increases with this temperature as well. If the temperature inside a containment building drops, the amount
of water the air can hold will drop as well and the remaining water may condense out of the air in the form of a liquid
film on the containment building walls. The reason why this process is important in the study of nuclear engineering
is that it sometimes comes into play following a reactor loss-of-coolant accident (or LOCA). As the air–steam mixture
cools down, water condenses out of the mixture and builds up on the containment building floor. We will examine the
mechanics of this process when we discuss the physics of LOCAs in Chapter 34.
7.29
Using the Steam Tables to Determine the Properties of Pressurized Water
Pressurized water is used to cool most reactors. In fact, it is the only coolant used in commercial PWRs and BWRs.
Depending on how the NSSS is designed, this water can have an operating pressure between 1,000 PSI (~6.9 MPa) and
2,250 PSI (~15.5 MPa). After it passes through the steam turbines, the pressure falls to about 1 PSI (6.9 kPa). In this sec-
tion, we would like to illustrate how the steam tables can be used to determine the energy content of the water when this
occurs. From our previous discussion, we know that this requires us to specify both the temperature and the pressure.
With this knowledge, we can always fix the thermodynamic state, and this allows us to determine its energy content and
its enthalpy.
7.30
Using the Steam Tables to Calculate the Enthalpy
Normally, the steam tables provide explicit values of the specific enthalpy for reactor coolants such as water. The spe-
cific enthalpy is always greater than the specific internal energy by an amount equal to υP. Hence, h = u + υP. However,
if the pressure is constant, the coolant is incompressible, and it does not boil, then dP = 0 and dυ = 0, and the enthalpy
change dh is the same as the internal energy change du:
Relating the Enthalpy Change to the Internal Energy Change for an Incompressible Fluid
dh = du (when P is constant) (7.50)
This equation applies to reactor coolants such as liquid sodium and ordinary water provided that they do not boil. Hence,
in PWRs (except near the top of the core) and in liquid metal fast breeder reactors (LMFBRs), the enthalpy change in
the primary loop dh is the same as the internal energy change du. The enthalpy in a BWR must be found using a slightly
different approach because the coolant is intentionally allowed to boil. The specific enthalpy change then requires one
to know the quality as one moves upward through the core. Normally, the quality at the core exit is about 15%, and it
is 0 (or even slightly negative) at the core inlet. Mollier diagrams that can be used to model this behavior are shown in
Figures 7.21 and 7.22.
7.31
T hermodynamic Cycles and Path Functions
Since reactors are designed to be closed thermal systems (with the exception of the heat that is dumped into a lake,
a river, or a pond), the fluid flows around the NSSS in closed loops or paths. Next, we would like to introduce the
concept of a reactor thermal cycle and a path function, which are closely related to the thermodynamic behavior of
these loops. In classical thermodynamics, there are certain quantities that cannot be defined by just a single point
on a curve—like the value of the pressure, the temperature, or the specific volume. Instead, they must be defined
by the area under a curve, and for a specific process, the area of the curve is defined by the path that the process
takes. Quantities that are measured in this way are called path functions, and they are completely different than the
pressure, the temperature, or the volume, which are called point f unctions. Common examples of path functions
are the heat Q and the work W. Most path functions can be described by a series of processes whose start and end
points known (and are generally are the same. Then the processes form what is called a closed cycle. (A piston
moving up and down in an internal combustion engine is another example of a closed cycle). The energy generated
in the thermal cycle of a nuclear power plant is another. Thermal cycles can be represented by closed curves on
248 Thermodynamic Properties and Equations of State
FIGURE 7.22 Another Mollier diagram in which the enthalpy is plotted as a function of the pressure.
pressure–volume diagrams or temperature–entropy diagrams (refer to Figure 7.23). When this occurs the area inside
of the path is a statement of either the heat a system acquires or rejects, or the work that a system can perform. In a
closed system consisting of several distinct states (say state 1, state 2, state 3, and state 4), the path around the sys-
tem will always return to its initial state. Hence, if one starts out at state 1, then a thermodynamic cycle for a closed
system will also return to state 1. There are many examples of these thermal cycles in practice. One example is the
Rankine thermal cycle, and another example is the Brayton thermal cycle. We will discuss both of these thermal
cycles in Chapter 9.
7.32 The Relationship between the Path a System Takes, and Its Heat and Work Output 249
FIGURE 7.23 A diagram illustrating the differences between a path function and a point function. The value of a path function
depends on the path followed by a process as well as the end states, while the value of a point function depends on the state of a
system only, and it does NOT how the system reaches that state.
7.32
T he Relationship between the Path a System Takes, and Its Heat and Work Output
To illustrate how path functions apply to nuclear power plants, suppose that we would like to move between two states on
a P–V diagram. We can move from an initial pressure and volume P1 and V1 to a final pressure and volume P2 and V2 in
many different ways. In fact, there are an infinite number of ways that we can get from state 1 (P1 V1) to state 2 (P2 V2).
The amount of work it takes to move from state 1 to state 2 depends on the path we take. Similarly, on a T–S diagram,
there are an infinite number of ways to move from state 1, defined by T1 and S1, to state 2, defined by T2 and S2. In this
case as well, the amount of work it takes to move from state 1 to state 2 depends on the path. Thermodynamically, this
is a consequence of the fact that the amount of heat (or work) expended in moving from state 1 to state 2 depends on the
product of P and V, or the product of S and T. Hence, the integral of these two quantities, taken between points 1 and 2,
determines the amount of work required to get from state 1 to state 2. However, if a thermodynamic cycle is a closed
cycle, and we want to get back to state 1 again, we can take a different path to get from state 2 to state 1 than we did to
get from state 1 to state 2. The difference is the area under the curve from state 1 to state 2, and the area under the curve
from state 2 to state 1 is then the work that must be expended to return to the initial state. In other words, the work to
get from point 1 to point 2 and then return to point 1 is
V2
Work12 =
∫ V1
P dV (7.52)
or
If the temperature along the paths is different in each case, then the work to take the entire round trip is
or
Now suppose that we would like to apply the same logic to the power generating components of a nuclear power plant.
In this case, the area between the two curves is the useful work that can be extracted from the plant thermal cycle, and
the larger we can make this area, the more efficient the plant thermal cycle will be. This simple observation also leads
us to conclude that heat and work are both path functions and thermodynamic properties like P, T, and V are point
functions. Thus, properties are point functions that depend on the state of a system only, and not how the system reaches
that state; however, path functions can have different values depending upon how the state is reached. For example,
consider the two p rocesses—process A and process B, which are shown in Figure 7.23. Since Figure 7.23 is a P–V
2
diagram, the volume change ΔV between states 1 and 2 is given by ∆V =
∫
1
= V2 – V1 , and the volume change during
process A or process B is always equal to the volume at state 2 minus the volume at state 1, regardless of the path that
is taken. However, the total work done in process A is different than the total work done in process B because the area
under the curves, which is given by W = ∫P dV, is not the same. To be specific, the work done in process A is less than
the work done in process B. Hence, Work A < Work B. Hence, we can summarize the difference between a path function
and a point function as follows:
Every thermal process in a nuclear power plant must obey these basic rules.
7.33
D eriving Work from a State Diagram
In addition to T–S diagrams, a lot of useful information concerning heat engines and thermal cycles can be gleaned
P–V diagrams. A P–V diagram is a pressure–volume diagram in which the pressure of a material (like water) is plotted
on one axis and the volume is plotted on another. The area under the curve in a P–V diagram then corresponds to the
useful work that a system can perform. Sometimes the work is called PV work or flow work.
We can illustrate this concept by considering the various paths that a thermodynamic process can follow. If the
process is a closed process (i.e., the start points and the end points are the same), then the path the process takes makes
a great deal of difference when it comes to the amount of useful work that can be produced. The heat QIN that must be
added to a fluid to move from point 1 to point 2 is the area under the curve between point 1 and point 2 (Path A). The
heat QOUT that must be removed from the fluid to return from point 2 to point 1 is the area under the curve between
7.34 Finding the Thermal Efficiency from a T–S Diagram 251
point 2 and point 1 (Path B). The difference between the heat that is added QIN and the heat that is removed QOUT is then
the useful work W, which is given by
If we divide the useful work by the energy we have added to the fluid QIN, then we arrive at the thermal efficiency for
the particular path we have taken:
The astute reader may notice that the right‑hand side of Equation 7.61 can also be written as
η = 1 – Q OUT Q IN (7.62)
As discussed previously in Chapter 6, one can see that this is just the classical definition of the thermal efficiency. If we
keep QIN fixed, then the only way to increase the thermal efficiency is to reduce the value of QOUT relative to QIN (i.e.,
the area under the curve between paths A and B). Hence, optimizing the area under the curve enclosed by paths A and
B optimizes the thermal efficiency because it also maximizes the numerator.
7.34
Finding the Thermal Efficiency from a T–S Diagram
T–S diagrams (refer to Figure 7.24) as well as P–V diagrams (refer to Figure 7.25) perform a very interesting role when it
comes to analyzing reactor thermal cycles. They generally make the analysis of reactor thermal cycles easier, and in some
cases, we can infer the efficiency of a reactor thermal cycle by simply finding the area under the curves when the path func-
tions are known. Of course, if we change the start points and the end points, the area under the curve will change and the
thermal efficiency of the plant thermal cycle will change as well. In the discussion that follows, we would next like to illus-
trate how state diagrams can be used to estimate the efficiency of a power plant thermal cycle. Let us begin by illustrating
how the concept of entropy can be used in conjunction with a T–S diagram.
As we mentioned earlier the entropy of a material is a measure of its molecular disorder. A material whose lattice
is highly structured has a low specific entropy, and a material whose lattice is highly disorganized has a high specific
entropy. Heat flows from a region of high temperature to a region of low temperature, and this always has the effect of
FIGURE 7.24 A T–S diagram for water that can be used to find the work performed in an NSSS.
252 Thermodynamic Properties and Equations of State
FIGURE 7.25 A P–V diagram for water showing various isotherms or lines of constant temperature.
increasing the amount of molecular disorder (and hence the value of the specific entropy). The entropy is usually given
the symbol S and is defined in the following way:
where Q is the amount of heat a thermal reservoir at a temperature of T degrees Kelvin contains. If a fraction of this heat
ΔQ flows from a high-temperature reservoir having temperature THIGH to a low-temperature reservoir having tempera-
ture TLOW, then the entropy change as the result of this process is
because the high-temperature reservoir loses the heat and the low-temperature reservoir gains the heat. We can then
write the entropy change for this process as
( )
∆S = ∆Q × 1 TLOW − 1 THIGH (7.65)
Because TLOW is always smaller than THIGH, it follows from Equation 7.65 that the net entropy is always increased as a result
of this heat flow. It also follows that the amount of heat dQ that flows along a reversible path on a T–S diagram is given by
dQ = T dS (7.66)
when the entropy change is small. The total heat transferred between states 1 and 2 is then
2
Q12 =
∫
1
T dS (7.67)
In other words, the value of Q12 corresponds to the area under the path function between points 1 and 2 that a process
takes. If the path function is used to represent a closed thermal cycle, then the total heat transferred between states 2
and 1 on the return path is
1
Q 21 =
∫ T dS (7.68)
2
7.34 Finding the Thermal Efficiency from a T–S Diagram 253
The difference between Q12 and Q21 then represents the net work that the system performs:
Many reactor thermal cycles can be analyzed in this way. In Equation 7.65 the entropy has the units of energy divided by
the absolute temperature, and so it has the units of J/°K (or kJ/°K) in the SI unit system. The specific entropy s, or entropy
per unit mass, is then s = S/M, where M is the mass of the system for which the entropy is being found. Historically the
term entropy was originally proposed by Rudolf Clausius in 1865, and it was based on the Greek word εντροπία [entro-
pía], which stands for a turning toward. The entropy also appears in the study of Carnot thermal cycles, which were first
proposed by Sadi Carnot around 1825 (see Figure 7.26). Carnot’s work attracted little attention when he was alive, but it
was later used by Rudolf Clausius and Lord Kelvin to formalize the second law of thermodynamics (see Chapter 6) and
also to precisely define the entropy as a thermodynamic state variable.
In classical thermodynamics, the concept of entropy is represented by the second law of thermodynamics (see
Chapter 6) which states that the entropy of a closed system will always increase or remain the same. In other words, it is
a measure of the tendency of a system to proceed in a certain direction toward a particular end state. It implies that heat
always flows from a region of high temperature to a region of low temperature. This process reduces the state of order of
the system, and therefore, entropy is an expression of molecular disorder or randomness that the new system possesses.
Entropy is an important state variable because it can be used to calculate the internal energy U of a material, and it can
also be used to determine how much this internal energy changes. The concept of entropy originally arose from Rudolf
Clausius’s study of the Carnot thermal cycle, which is the thermal cycle for an ideal heat engine. In a Carnot cycle, heat
is absorbed from a “hot” reservoir at a high temperature THIGH, and it is given up isothermally to a “cold” reservoir at a
low temperature TLOW. In this process, work can be performed when there is a temperature difference ΔT between the
hot reservoir and the cold one. The amount of this work is a function of the temperature difference and the amount of
heat absorbed. The work is also proportional to the difference between the heat absorbed at the hot reservoir and the heat
rejected at the cold one. This implies that the thermal efficiency η of a Carnot engine is given by η = 1 − (TLOW/THIGH),
and that the Carnot engine is the most efficient of all possible heat engines.
Here, the temperatures of both reservoirs must be measured in absolute degrees (°K). One of the interesting conse-
quences of a Carnot heat engine is that the quantity QHIGH/THIGH = QLOW/TLOW is always be conserved. Thus, the ratio of
the heat in each of these reservoirs to the temperature of each of these reservoirs was the quantity that Rudolf Clausius
ultimately interpreted to be the entropy. This also means that if the entropy of reservoir 1 is S1, and the entropy of res-
ervoir 2 is S2, then in an ideal heat engine, S1 = S2, and in a real heat engine, S1 ≠ S2. This observation has enormous
practical implications because it allows us to express the amount of heat that can be transferred from a high-temperature
region to a low-temperature region as dq = TdS, or more precisely, as Δq = TΔS at each combination of temperature and
entropy that a system possesses. Hence, if we have a closed system where the temperature and the entropy initially start
out at T1 and S1, go through a number of intermediate steps where the value of the temperature and the entropy are T2, S2,
FIGURE 7.26 A picture of James Prescott Joule (left) for which the joule is named and Sadi Carnot (right) who proposed the
thermal efficiency of a Carnot heat engine. An energy generation rate of 1 J/s is the same as 1 W. Note that joules can also be used to
represent the production of thermal or electrical power.
254 Thermodynamic Properties and Equations of State
T3, S3, … and TN, SN, etc., and return to the initial state T1 and S1, then the total work W that the system can do is equal
to the area inside of this curve on a T–S diagram. An example of such a curve is shown in Figure 7.24. Note that all
nuclear power plants (and coal-fired power plants for that matter) can use these T–S diagrams to calculate their thermal
efficiency. Only in this case, the thermal efficiency of a power plant η is defined as the electrical energy that the power
plant produces divided by the total heat that it produces. In other words, the thermal efficiency is given by
The thermal efficiency of modern nuclear power plants ranges from 32% to about 38%. In a modern PWR, it is about
34%. In addition, there are a number of “tricks” that can be used to increase the value of the numerator in Equation 7.71
because the denominator is fixed by the thermal output of the plant. We will learn more about these tricks and their
applications to nuclear systems in Chapter 9. However, some of the more commonly used ones are called reheat and
regeneration. (Not surprisingly, reheat and regeneration are sometimes called R & R.) An important implication of
Equation 7.71 is that if a nuclear power plant has a thermal output of 3,300 MWT, and a thermal efficiency of 33%, then
its electrical output will be
or MWE = 0.33 × 3,000 MWT ≈ 1,000 MWE. Example Problem 7.7 helps to illustrate this point.
7.35
D efining the Energy Storage Capacity of a Material
The ability of a material to store heat is related to another important thermodynamic property called the specific heat. In
general, the specific heat is measured for different materials having the same amount of mass. In other words, the spe-
cific heat is defined on a per unit mass basis. Normally, the specific heat is measured in joules per kilogram per degree
Kelvin (J/kg/°K). In nuclear applications, it is also measured in kilojoules per kilogram per degree centigrade. Thus the
specific heat is the amount of energy required to raise the absolute temperature of a material by 1°K:
For example, it takes about 4,180 J of thermal energy to raise the temperature of 1 kg of liquid water by 1°K, while it
takes about one-quarter as much thermal energy (approximately 1,275 J) to raise the temperature of 1 kg of liquid sodium
(which is a common coolant in LMFBRs) by the same amount. Consequently, it is desirable to have a single thermo-
dynamic property that enables us to compare the energy storage capacities of different materials. Using this approach,
we can then compare the amount of heat a nuclear fuel rod, a reactor coolant, or even the pressure vessel can absorb.
7.37
Specific Heat at Constant Volume
The specific heat at constant volume Cv is defined as the amount of energy required to raise the temperature of a unit
mass of a substance by one degree while the volume of the substance remains unchanged (V = constant). This tempera-
ture increase can be measured in °F, °C, or °K. In the SI unit system, the temperature change is expressed in °K. Because
the volume of a material is not allowed to change in this process, the substance to which heat is being added cannot do
any external work on its environment. Because of this, the value of Cv is sometimes used to represent the internal energy
u. Both Cv and u are usually associated with solids (which only expand or contract slightly) rather than liquids or gases
(which can expand or contract a great deal more). Liquids and gases, which can do a considerable amount of external
work when their volume changes, are described by another thermodynamic property called the specific heat at constant
pressure Cp. Not surprisingly, the specific heat at constant pressure is more closely related to the specific enthalpy h than
it is to the specific internal energy u.
7.38
Specific Heat at Constant Pressure
The specific heat at constant pressure Cp is defined as the amount of heat required to raise the temperature of a unit mass
of a substance by 1° while the pressure of the substance is held constant. The specific heat at constant pressure is always
greater than the specific heat at constant volume because external work can be done by a substance while it expands. If
the volume change at constant pressure is ΔV, the external work that is done by the substance on its environment is then
W = P∆V (7.73)
where this work is measured in J or kJ. In classical thermodynamics books, Equation 7.73 is also called the PV work or
the flow work. The value of Cp is more closely related to liquids and gases than it is to solids. Consequently, when we
compare the two definitions, we find that
The exact difference between the values of Cp and Cv depends on the substance in question, and whether it can be treated
as compressible or incompressible. Next we would like to explore this distinction between Cp and Cv for substances that
are commonly used in the construction of nuclear power plants.
7.39
T he Specific Heat for a Pure Substance
Values for Cp and Cv can be defined for any conceivable substance, including solids, liquids, and gases. Because Cv
represents the energy storage capacity of a material, and Cp represents its ability to store energy and perform useful
work, it is reasonable to assume that changes in the internal energy are related to changes in the absolute temperature by
and changes in the enthalpy are related to changes in the absolute temperature by
In differential form, this leads us to the following expressions for du/dT and dh/dT
In practice, Cp and Cv can then be defined by partial derivatives, which are given by the following equations:
Cp (∂ u/ ∂T)p (7.78b)
The specific heats of water and other substances which use these definitions are shown in Figure 7.27. Physical inter-
pretations for Cp and Cv are provided in Figure 7.28. Notice that the definition of the specific heat at constant volume
involves a change in its specific internal energy u, while the definition of the specific heat at constant pressure involves
a change in its specific enthalpy h. This is because the change in the enthalpy at constant pressure is a result of a change
in the net volume of the material, and this is due to the fact that the material can perform PV work. This distinction
between Cp and Cv is often ignored in most elementary text books. However, it is important in the design of gas reactors,
and so we must be aware of this distinction in order to accurately calculate the efficiency of a gas reactor thermal cycle.
The changes in the specific internal energy and the specific enthalpy of a pure substance can then be written as
FIGURE 7.27 The specific heat of water compared to the specific heats of other substances (left) and the specific heat of air as a
function of the ambient temperature. In general, the specific heats of a gas are temperature dependent.
TABLE 7.4
The Specific Heats for Some Common Reactor Coolants (in J/kg °K)
As long as a material does not boil or change phase, energy changes in nuclear systems can be correlated with
temperature increases and decreases in this way. Example Problem 7.8 illustrates how these equations can be applied
to both liquid and solid substances.
When attempting to understand the thermal-hydraulic performance of reactors, normally the specific heats that are
most relevant to their operation are the specific heats of uranium, iron, steel, and reactor coolants such as water, sodium,
carbon dioxide, and helium. The specific heats for liquid water and for water vapor are shown in Table 7.4. In general,
water has the highest specific heat of any common reactor coolant. Thus, it makes an excellent coolant because its
energy storage capacity is so high.
7.40
T he Specific Heat for an Ideal Gas
Now, let us attempt to deduce the specific heats for an ideal gas from the definitions we have just presented. The specific
heats in this case can be found by applying these definitions to the ideal gas law. The specific heat at constant volume
can be found by taking the derivative of the internal energy with respect to the temperature Cv = (∂u/∂T)v, and the
specific heat at constant pressure can be found by taking the derivative of the enthalpy with respect to the temperature
Cp = (∂h/∂T)p. The most convenient form of the ideal gas law to employ for these calculations is
Pυ = RT (7.81)
where υ is the specific volume and R = Ru/M is the molar gas constant, which differs from the universal or ideal gas
constant Ru because it depends on the molecular weight M. From the definition of the specific enthalpy we presented
previously, we know that
h = u + Pυ (7.82)
However, because we are dealing with an ideal gas, we can replace Pυ with RT. This leads us to conclude that
h = u + Pυ = u + RT (7.83)
258 Thermodynamic Properties and Equations of State
FIGURE 7.29 The specific heats of some common gases that can be used as reactor coolants. Notice that carbon dioxide is the
only industrial gas that has a higher specific heat than ordinary water.
We also know that the definition of the specific heat at constant pressure is Cp = dh/dT. Since
Cp = dh/dT = Cv + R (7.85)
or
Thus, the difference between Cp and Cv for an ideal gas is equal to the molar gas constant R where the value of R
depends on the type of gas. The values of the specific heats for several ideal gases are shown in Figure 7.29. Thus, any
change in the internal energy or enthalpy of an ideal gas can be deduced from the equations
where Cp = Cv + R and ΔT = (T2 − T1). Notice that the value of R differs from one gas to the next because it depends
on the molecular weight. Sometimes the ratio of Cp to Cv is called the specific heat ratio γ of the gas. The specific heat
ratio is defined as
γ = Cp Cv (7.88)
and sometimes this ratio appears in the study of gas-cooled reactors. The specific heat ratio is also temperature dependent, but
for many applications, this variation is small, and the temperature dependence of Cp and Cv can be ignored. For monatomic
gases, γ has a value of about 1.667. For diatomic gases such as helium and carbon dioxide, it has a value of about 1.4 at 20°C.
7.41
O ther Ways to Find the Specific Heat
Now, let us consider some alternative ways to determine the values for the specific heats. Earlier, we learned that the
specific heats for an ideal gas were related to each other by
When we evaluated these derivatives using the fact that PV = RuT and ρ = P/RuT, we also discovered that
Cp = Cv + R (7.91)
Sometimes this relationship is referred to as Mayer’s relationship. For a simple monatomic gas such as helium where
the volume remains fixed, the heat capacity Cv is given by
where n is the number of moles of the gas. Notice that the number of moles n is related to the total number of molecules
N in the gas, and the two of them are related by the following equation:
n = N A o (7.93)
where Ao is Avogadro’s number, which has a value of 6.022 × 1023 molecules/mol. It is also possible to show that
where k is Boltzmann’s constant. The heat capacity Cp of a monatomic gas at constant pressure is therefore
Cp = Cv + R = 5/2 R (7.95)
where the value of R depends on the type of the gas. This equation is commonly used for gas reactors where the pressure
drop of the gas is relatively low as it flows across the core.
Notice that these alternative formulations of the ideal gas law use Boltzmann’s constant k and Avogadro’s number Ao.
While these representations are rarely used in reactor thermal analysis and in nuclear heat transfer, they appear fre-
quently in the field of nuclear chemistry. Now, let us examine the relationships between these parameters. Earlier we
found that Boltzmann’s constant is equal to the universal gas constant Ru divided by Avogadro’s number:
Using Boltzmann’s constant, the ideal gas law can then be written as
PV = N kT (7.97)
where N is the number of molecules in the gas. For n = 1 mol, N is equal to the number of particles in 1 mol, which
corresponds to Avogadro’s number Ao. The product of Boltzmann’s constant and the absolute temperature (kT) then
acquires the units of energy. When it comes to a gas, this energy is normally measured in joules. Thus, the left‑hand
side of Equation 7.97, which is equal to PV, can be loosely interpreted as the flow energy or flow work that the gas is
capable of performing. Practically speaking, Boltzmann’s constant is the fundamental link between the microscopic
world of nuclear particles and the macroscopic world of gases and fluids. It relates the particle energy at a molecu-
lar level to the absolute temperature of a particle. The values of Avogadro’s number, the universal gas constant, and
Boltzmann’s constant are presented in Table 7.5. The values of Boltzmann’s constant are also shown for several differ-
ent unit systems in Table 7.6.
TABLE 7.5
The Values of Avogadro’s Number, the Universal Gas Constant, and Boltzmann’s Constant Which
Frequently Appear in the Study of Nuclear Engineering
TABLE 7.6
The Values of Boltzmann’s Constant for Different Unit Systems
Thus, the ideal gas law can be used to define the energy of the gas when the product of Boltzmann’s constant and the
absolute temperature are interpreted in this way.
7.42
Tabulated Values for the Specific Heat
The values of the specific heats for most nuclear materials are tabulated in property tables. In reactor heat transfer and
fluid flow, the most important specific heats are those for liquid water, water vapor, liquid sodium, carbon dioxide, and
helium. These coolants are the four primary coolants that are used in commercial power reactors today. Some experi-
mental test reactors use heavy oils or liquid mercury as their coolants, but these coolants are relatively rare compared to
those used in commercial nuclear plants. The values of these specific heats are presented in Table 7.4. Notice that carbon
dioxide has approximately the same specific heat (about 1,240 J/kg at 1,000°K) as liquid sodium does. This is one of the
reasons it is so attractive to use as a coolant in gas reactors. It is also much cheaper than liquid sodium, and this helps
to explain one of the reasons why it is so popular in some nuclear applications.
7.43
Applying the Specific Heats to Solids and Incompressible Liquids
There are a number of other observations we can make about the relationship between the specific heats at constant
volume and constant temperature. When it comes to substances such as solids and incompressible liquids, the volume
change Δυ is so small that the external work done by the substance on its environment (the PV work in the expression for
the enthalpy) can be neglected in most calculations of the enthalpy. When this approximation is valid, the specific heat at
constant volume has exactly the same definition as the specific heat at constant pressure, and we are led to conclude that
If water is considered to be an incompressible fluid, then the values of Cp and Cv also become the same. However, when
enough heat is added to the water for it to become a vapor, they are different and we must use the original definition*
7.44
Applying Specific Heats to Reactor Coolant Pumps
There are some additional stipulations to what we have said if a reactor coolant happens to be water, and the water does
not boil or change phase. Suppose we start with the expression for the fluid enthalpy h = u + Pυ and assume that the
specific volume υ is constant. Then, the total differential of the specific enthalpy is
If we are dealing with a simple solid, then the value of υ ΔP is insignificant and the enthalpy change becomes
However, if we are dealing with a reactor coolant pump (RCP) (where the temperature is assumed to be a constant), then
ΔT = 0 and we obtain
This last equation has some practical importance because it implies that the work it takes to pump an incompressible
fluid (such as water) through a coolant pump is equal to the specific volume of the fluid times the pressure head ΔP across
the pump. Moreover, because the density ρ is inversely proportional to the specific volume, the work required to do this is
Of course, this assumes that the pump that pumps the fluid is 100% efficient. In practice, RCPs are about 85% efficient,
and so the actual energy expended by the pump is about 1.17 times this value. We can also use Equation 7.104 to find the
pumping power of the pump. In this case, we must simply multiply ΔP/ρ by the mass flow rate m = dm/dt, which gives the
pumping power (in J/s or Watts) as
If the pump is not 100% efficient, then we can measure the actual efficiency and modify Equation 7.105 to express this
fact. In this case, the pumping power for a real RCP is as follows:
where η is the pump efficiency. Notice that all we needed to derive this expression was the definition of the enthalpy
for an incompressible fluid. We will have more to say about how RCPs work and how they behave in Chapter 19. In the
meantime, we can apply Equation 7.106 to find the amount of power it takes to run an RCP. The amount of power it takes
to run a large reactor coolant pump can sometimes exceed several MW (see Problem 7.9).
7.45
Calculating the Heat Transfer Rate Using the Specific Heats
The equations we have just presented can be used to provide some practical examples of how the specific heats can
be used. The values of Cv and Cp relate changes in the temperature of a substance to changes in its internal energy or
enthalpy. The governing equations that determine this behavior are
Equations 7.107a,b can be used as long as a phase change does NOT occur. Since these equations measure the energy
change per unit of mass, the change in the specific internal energy or enthalpy is expressed in kJ/kg °K. When a reactor
coolant flows over a nuclear fuel rod that is hotter than the coolant itself, a temperature difference ΔT exists between
the coolant and the surface of the fuel rod. We can use Equations 7.107a,b to estimate how much energy is transferred
to the coolant in this case. If we know the mass flow rate m = dm/dt over the surface of the rod, the amount of energy
transferred from the rod to the coolant per second is then
Q ′ = dH/dt = m
Cp ∆T (kJ/s) (7.108)
262 Thermodynamic Properties and Equations of State
FIGURE 7.30 Different materials store different amounts of thermal energy, and they also store this thermal energy at different
rates. Water has a higher heat capacity than other liquids, and it can store about 30 times more heat as a comparable amount of gold.
Water also takes about 30 times longer to heat up than gold, and this is one of the reasons why it is such a good reactor coolant.
where ΔT = TROD − TFLUID. This equation is the basis of most convective heat transfer calculations in nuclear power
plants. The amount of heat transferred to the coolant from the rod in a time interval Δt is then
Q = ∆H = ( mC
p ∆T ) ∆t (kJ) (7.109)
where Δt = t2 − t1. Of course, this relationship assumes that the temperature difference ΔT between the fuel rod and the
coolant is constant. If the coolant is an incompressible liquid and not a gas, then C = Cp = Cv. For most reactor coolants
such as water and liquid sodium, this allows us to write
where m = dm/dt. Note that Equation 7.110 requires us to know the value of ΔT as well as the mass flow rate m.
Equation 7.110 is valid whenever the coolant does not boil. Example Problem 7.11 illustrates a simple application of
this equation to a reactor pressure vessel. Other observations regarding the specific heats are provided in Figure 7.30.
7.46
D etermining the Coolant Temperature Profiles in a PWR
In PWRs, the coolant is specifically designed not to boil except at the top of the hottest fuel assemblies. Because of this,
Equation 7.110 can be used to find the coolant temperature profiles along the length of each rod if the value of Q′(z) is
known (see Figure 7.31). Normally, the value of Q′ is independent of the coolant temperature and can be treated as an
independent variable when calculating the coolant temperature profile. This leads us to conclude that
∆T = Q ′(z)/mC
(7.111)
In general, Q′(z) is a function of axial position, but when it is constant, ΔT = TFINAL − TINITIAL = Q′/mC,
and the value
of TFINAL at the top of a fuel assembly is simply the value of TINITIAL at the bottom of the fuel assembly plus the energy
which is shown in the following equation:
generation term Q′/mC,
FIGURE 7.31 An example of how the coolant temperature profile in a PWR fuel assembly behaves (see the right‑hand side) when
Q′ is independent of axial position. Normally, Q′ has an axial dependence, and this results in the coolant temperature profile shown
on the left. This shape is discussed in more detail in Chapter 26.
It then becomes a trivial exercise to find the value of TFINAL if the values of Q′ and m are known. For example, if
Q′ = 15 MW thermal, C = 4.65 kJ/(kg °K), m = 100 kg/s, and TINITIAL (at the core inlet) = 300°C, then TFINAL = TINITIAL + Q′/
= 300°C + 15,000,000/(4,650 × 100) = 300°C + 32°C = 332°C. Of course, this example is an idealized one because the
mC
value of Q′ does not change with axial position, but it does illustrate the general approach that most nuclear engineers use.
7.47
Performing an Energy Balance Using the Enthalpy in a Reactor Heat Exchanger
In addition to using Equation 7.112 and the specific heats to determine how much thermal energy a reactor coolant
can absorb, a similar analysis can be performed using the enthalpy difference between two points in a reactor cooling
system. Normally, enthalpy differences are used instead of temperature differences when the coolant changes phase
because there is no longer a direct relationship between the temperature change of the coolant and its energy content
(h or u). Places where this can occur include the steam generator tubes or the condenser in a PWR (refer to Figure 7.32)
FIGURE 7.32 The energy Q added to the coolant from the core can be found from a knowledge of the temperature difference
ΔT = TROD − TFLUID between the coolant and the surface of the fuel rods. In the absence of this information, the energy transferred
= dm/dt, and multiplying it by the enthalpy difference Δh = h2 − h1.
to the coolant can still be found by taking the mass flow rate m
Most reactor coolant systems in which the coolant boils are analyzed in this way. The aforementioned example shows that an energy
balance is performed in a reactor heat exchanger. This would normally correspond to a component of the NSSS called the condenser.
264 Thermodynamic Properties and Equations of State
and also around the fuel rods in the core of a BWR. Enthalpy changes can also occur between the high-pressure and
low-pressure stages of a steam turbine where the quality becomes progressively lower as the steam is depressurized.
In any event, the enthalpy change of a reactor coolant can be used to determine the heat transfer rate or the amount
of work performed. For example, suppose that the enthalpies at points 1 and 2 in a reactor coolant system are known.
The specific enthalpy at point 1 is h1, and the specific enthalpy at point 2 is h2. Moreover, suppose for the moment that
h2 > h1. If the mass flow rate between points 1 and 2 is m, then the amount of heat Q′ that is added to the coolant per
second is
In BWR cores, Q′ corresponds to the heat generation rate between points 1 and 2 in the fuel rods. A similar approach
can be applied to other reactor components as well. Examples of these components include the core, the fuel assemblies,
the steam generators, the steam turbines, and the condensers. In general, enthalpy differences rather than temperature
differences are employed whenever a two-phase mixture is present. Hence, all steam-generating components rely on
enthalpy differences rather than temperature differences when calculating the thermal energy flows. We will examine
how a reactor is cooled in more detail when we discuss the components of the NSSS in Chapter 8. Normally, the values
of the specific enthalpy and the mass flow rate are sufficient to perform an energy balance on any component of the
cooling system when heat is added to or subtracted from it. From a practical perspective, the thermal efficiency of the
NSSS can also be calculated in this way.
7.48
More Accurate Methods for Finding the Boiling Point of a Reactor Coolant
Earlier we mentioned that the Clausius–Clapeyron equation is sometimes used to find the boiling point of reactor cool-
ants such as water and liquid sodium as a function of their ambient pressure. Unfortunately, several assumptions used
to derive this equation come into question at very high pressures and temperatures (or those very close to the critical
point), and when this occurs, the Clausius–Clapeyron equation begins to introduce errors into the vapor pressure curve.
For most applications involving nuclear power plants, nuclear engineers require more reliable estimates of the boiling
point because reactors can operate at very high temperatures and pressures.
An equation such as the Antoine equation provides a good alternative to the Clausius–Clapeyron equation under
these conditions. It does so by calculating the boiling point using a simple three-parameter curve fit to experimental
vapor pressures over a specific temperature range. Normally, these temperatures are subdivided into two or three spe-
cific temperature ranges to provide the best fit to the experimental data available. For common reactor coolants such
as water, these temperature ranges are normally 1°C–100°C and 100°C–374°C. The Antoine equation for the vapor
pressure can then be written as
where γ = A − (B/(C + T)) and A, B, and C are called the Antoine coefficients. In general, these coefficients are tempera-
ture dependent. Sublimations and vaporizations of the same material use separate Antoine coefficients, as do compo-
nents in material mixtures. The Antoine equation is accurate to within a few percent for almost any substance when the
correct coefficients are used. Antoine coefficients for many common materials can be found in Lange’s Handbook of
Chemistry (12th ed., McGraw-Hill, New York, 1979), and they are also available on the Internet from NIST’s Chemistry
WebBook. We will have more to say about the use of these coefficients in Chapter 13.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Questions for the Student 265
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi (2000).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Web References
http://www.wikicalculator.com/formula_calculator/Molar-enthalpy-of-vaporization-(The-Clausius-Clapeyron-equation)-438.htm.
http://calistry.org/calculate/clausiusClapeyronCalculator.
http://engr.bd.psu.edu/davej/classes/thermo/chapter2.html.
http://thermodynamik.hszg.de/fpc/index.php.
http://webbook.nist.gov.
Exercise 7.2
5,000 kg of saturated steam enters a reactor containment building during a LOCA. The temperature of the steam is
200°C, and the pressure of the steam is 1,555 kPa. 10,000 kg of cold water at atmospheric pressure and 20°C is then
sprayed on the steam from above to condense it. If 100% of the steam is condensed, what is the final enthalpy of the
resulting mixture?
Exercise 7.3
Derive an expression for the specific enthalpy and the specific internal energy of the coolant leaving the core of a BWR
if the equilibrium quality is 15%. If the saturation temperature is 288°C and the exit pressure is 7.25 MPA (1,050 PSI),
calculate the values for each of these properties assuming that the mixture is a homogeneous one.
Exercises for the Student 267
Exercise 7.4
Armed with a knowledge of just the specific heats for subcooled water at 80°C and 1 atm, determine the difference
between the values of its specific enthalpy and the specific internal energy. Which one of these properties is usually the
larger one?
Exercise 7.5
Derive an expression for the density and the specific volume of the coolant leaving the core of a BWR if the equilib-
rium quality is 14%. If the saturation temperature is 288°C and the exit pressure is 7.25 MPA (1,050 PSI), calculate
the values for each of these properties assuming that the mixture is a homogeneous one.
Exercise 7.6
In a thermal mixing tank, 500 kg of saturated steam at 100 kPa is mixed with 10,000 kg of subcooled water at 20°C.
What are the final temperature and density of the resulting mixture?
Exercise 7.7
Water enters a PWR steam generator as a subcooled liquid with a temperature of 230°C and leaves the steam gen-
erator as a superheated vapor with a temperature of 280°C. What is the enthalpy increase of the coolant across the
steam generator tube bank? Assume that the pressure of the water on the shell side of the steam generator is 7 MPa
(1,015 PSI).
Exercise 7.8
In Exercise 7.7, at what temperature does the water first start to boil and what amount of superheat does it possess when
it leaves the tube bank?
Exercise 7.9
The specific enthalpy is often used to determine the rate at which thermal energy is added to or removed from a reactor
coolant. Suppose that the specific enthalpy of the coolant is 1,000 kg/kJ when it is a saturated liquid and the enthalpy
of vaporization is 2,000 kJ/kg. If the quality of the coolant reaches 50% (x = 0.50), what is the specific enthalpy of the
saturated mixture?
Exercise 7.10
When calculating thermodynamic properties beneath the steam dome on a P–V diagram, the quality is used much more
frequently than the void fraction to calculate the properties of a two-phase mixture. Why is this the case?
Exercise 7.11
Calculate the specific volume of water and the density of water at the critical point. On a P–V diagram, at which tem-
perature and pressure does the triple point occur?
Exercise 7.12
0.05 kg of water at 20°C fills a container having a cubic volume of 1 m3. What is the specific volume of the water in the
container, and what is its average density?
Exercise 7.13
Occasionally, the equilibrium quality of a reactor coolant must be deduced from a knowledge of the specific enthalpies
alone. Write an equation for the equilibrium quality of a reactor coolant as a function of the liquid enthalpy and the
enthalpy of vaporization. Suppose that a coolant such as water enters a PWR core from below at a temperature of 280°C
and a pressure of 15.5 MPa. What is the equilibrium quality of the coolant at the core entrance?
268 Thermodynamic Properties and Equations of State
Exercise 7.14
Using the equation m MIXυMIX = mlυl + mvυv for the conservation of mass for a two-phase mixture, show that the specific
volume of the two-phase mixture can be written as υMIX = (1 − x) υl + x υv, where x is the quality of the mixture. Also,
show that x = (υMIX − υl)/υlv. What is the value of υlv called?
Exercise 7.15
Using the equation derived in Exercise 7.14, write an expression for the density of a two-phase mixture as a function of
the equilibrium quality.
Exercise 7.16
The density of a two-phase mixture can also be expressed as a function of the void fraction using the r elationship ρMIX =
(1 − α) ρl + α ρv. Using this expression and the expression for the mixture density as a function of the equilibrium quality
in Exercise 7.15, derive an equation relating the values of the void fraction and the quality for a homogeneous mixture.
For saturated water at 300°C, what is the value of the void fraction if the equilibrium quality is 20%?
Exercise 7.17
A container of water molecules in an unknown state is maintained at a pressure of 0.5 MPa. If the specific enthalpy of
the molecules in the container is 2,890 kJ/kg, is the water in a liquid or vapor state? What is the temperature of the water
molecules?
Exercise 7.18
A two-phase mixture exits the core of a commercial BWR with a specific enthalpy of 1,500 kJ/kg. If the p ressure at this
location is 7.25 MPa (1,050 PSI), what is the equilibrium quality of the mixture?
8
The Nuclear Steam Supply System
and Reactor Heat Exchangers
8.1 T he Nuclear Steam Supply System
All commercial nuclear power plants use a nuclear steam supply system (NSSS) (see Figure 8.1) to convert the heat produced in the core
into useful work and power. The exact way in which this is done can vary slightly from one reactor type to the next (i.e., from a boiling
water reactor [BWR] to a pressurized water reactor [PWR] to a liquid metal fast breeder reactor [LMFBR]), but each power plant uses
the same basic thermodynamic principles to generate electric power. In this chapter, we would like to explore how this process works,
and we would like to discuss how the NSSS functions thermodynamically. We would then like to discuss power plant thermal cycles,
which determine the efficiency with which the thermal energy produced in the core is converted into useful work and power.
Practically speaking, the NSSS takes a certain amount of heat Q, which is produced in the core, and uses water or some other
working fluid (which serves as the coolant) to create steam, which drives the blades of a power turbine. The rotation of these blades
produces the electric power that the plant supplies. The thermal cycle for a nuclear power plant is no different than the thermal cycle
for most coal-fired power plants or even those that burn natural gas. These plants behave like thermonuclear heat engines where
heat is produced at a high temperature in the core and is rejected at a lower temperature to the environment. Because there are some
inevitable losses in this process, both nuclear and coal-fired power plants have lower thermodynamic efficiencies than Carnot heat
engines, which we first introduced to the reader in Chapter 7. However, they can approach the theoretical efficiency of a Carnot heat
engine under certain conditions (see Figure 8.2). Nuclear engineers may be surprised to learn that nuclear power plants use similar
FIGURE 8.1 The NSSS for a four-loop PWR. Only one loop is shown. Notice that the heat content of the coolant divides itself rather neatly into
three distinct regions: one for the core, one for the SGs, and one for the condensers. (Courtesy of Westinghouse.)
269
270 The Nuclear Steam Supply System and Reactor Heat Exchangers
FIGURE 8.2 The thermal cycle of a nuclear power plant is similar to that of a Carnot heat engine where heat is produced in the
core at temperature THIGH and heat is rejected from the plant by the NSSS at temperature T LOW. The theoretical efficiency of a Carnot
heat engine is given by η = 1 − T LOW/THIGH. Normally, a nuclear power plant has a lower thermal efficiency than the efficiency
of this engine due to irreversible losses in the NSSS. Carnot heat engines can be used to describe the behavior of both coal-fired
and nuclear power plants. The Carnot heat engine was first proposed by Sadi Carnot in the late 1820s. At the time, steam engines
were able to achieve a thermal efficiency of about 3%. Today, most nuclear power plants are able to achieve a thermal efficiency of
between 30% and 40%. (Pictures provided by the University of North Carolina and www.mpoweruk.com.)
thermal cycles to those employed in fossil-fired power plants. However, fossil-fired power plants are designed to operate
at higher temperatures and pressures than nuclear power plants. For this reason, their thermodynamic efficiencies are
generally higher than those of nuclear ones. Some of the reasons for these efficiency differences will be discussed in
this chapter. Others will be deferred until Chapter 9.
To begin our discussion, we would like to present an overview of the NSSS and the components it contains. We will
then discuss the temperatures and pressures at which these components operate. In general, most commercial power
reactors use water as their coolant, although this does not always have to be the case. For example, gas reactors are
cooled with industrial gases such as carbon dioxide, hydrogen, and helium, and fast reactors are cooled with liquid met-
als such as sodium, mercury, and potassium. In water-cooled reactors, the water is converted into high-pressure (HP)
steam, which is then used to spin the blades of a steam turbine. In other words, most nuclear power plants implement
some form of the Rankine thermal cycle which we discussed in Chapter 7. Gas-cooled reactors are capable of using
another thermal cycle called the Brayton cycle, but as of the writing of this book, very few gas reactors using the Brayton
thermal cycle have been built. There are a number of practical reasons for this which we will subsequently explain. In
the next section, we would like to present an overview of the NSSS and explore its basic function.
FIGURE 8.3 The NSSS in a modern BWR. (Courtesy of the U.S. NRC.)
All reactors with the possible exception of gas reactors use the Rankine thermal cycle to convert heat into electrical
power. The fraction of the heat that is converted into electrical power is known as the thermal efficiency of the plant.
In general, between 30% and 35% of the heat produced in light water reactors (LWRs) is converted into electric power.
(The rest of the heat is discarded to the environment.) Thus, the thermal efficiency of a water reactor is in the range of
30%–35%. However, some reactors can have thermal efficiencies higher or lower than this. The thermal efficiency of
an LMFBR is significantly higher (about 40%) because the liquid metal in the primary loop can operate at much higher
temperatures. Gas reactors, which can use the Brayton thermal cycle, have an even higher thermal efficiency than the
Rankine cycle. Their thermal efficiencies can be as high as 38%–42% under certain conditions. However, for a variety
of reasons, most gas reactors use the Rankine thermal cycle today.
FIGURE 8.5 BWRs use one primary loop, while PWRs use two and LMFBRs use three. PWRs need two loops because they use
an SG, and LMFBRs need three loops because an intermediate sodium loop is required to keep the coolant in the tertiary loop from
becoming radioactive.
8.3 The Components of the NSSS 273
FIGURE 8.6 (a) A picture of an SG used in a commercial Westinghouse PWR. (Source: Westinghouse Nuclear.) (b) Military
r eactors in nuclear submarines and aircraft carriers tend to be quite small. They are almost all small, high-powered PWRs with
some exceptions. They are also surrounded by primary and secondary radiation shields.
depleted (i.e., it loses its pressure), and when it exits the turbines entirely, it can sometimes fall below atmospheric
pressure. Allowing the steam to fall below atmospheric pressure after it leaves the turbines enables a reactor designer to
increase the thermodynamic efficiency of the plant. When the steam enters the turbines, it is designed to be as dry as
possible in order to not damage the blades. The drier the steam (and the lower the moisture content), the faster the blades
can turn. The moisture content of the steam can be reduced by superheating the steam or increasing its temperature
beyond the saturation temperature TSAT. Once the pressurized steam hits the blades, it begins to lose energy, and it starts
to recondense. The pressure drops, and water droplets begin to appear in the flow. At this point, the steam becomes so
274 The Nuclear Steam Supply System and Reactor Heat Exchangers
wet that it can no longer be used to drive the turbine blades again. However, even this steam can be recycled if its specific
enthalpy can be increased. The process of increasing its energy content is called reheating the steam, and both nuclear
and coal-fired power plants implement some form of reheating to allow the hot steam to be reused again. As we will sub-
sequently see, reheating the steam before it reenters the steam turbines also increases the thermal efficiency of the plant.
where P and Q are typically expressed in kW or kJ/s. P is usually the number of megawatts of electrical energy pro-
duced, and Q is typically the thermal output of the power plant in megawatts thermal (MWT). Hence, P is usually fixed
by the design of the plant, and Q is normally fixed by how hot the fuel rods can become before they reach their design
limits. The aggregate efficiency can be increased even more by using some additional techniques to recycle the remain-
ing waste heat through the NSSS. This has the effect of increasing the value of P, but not the value of Q. The two most
common techniques for doing this are called reheat and regeneration (R & R). Exactly how these techniques are imple-
mented is discussed in many heat transfer books.
Reheat and regeneration are also used to improve the efficiency of coal-fired power plants. The thermal efficiencies
of common reactor types are shown in Table 8.1. Note that it is very difficult to achieve an aggregate thermal efficiency
in excess of 40%. In practice, the NSSS can be subdivided into a number of separate components or subsystems. In
general, the NSSS consists of six primary components. These components include
1. The reactor core and the pressure vessel
2. The SGs (in a PWR)
3. The steam turbines
4. The electrical generators
TABLE 8.1
The Thermal Efficiencies of Different Types of Nuclear Power Plants
FIGURE 8.7 The layout of the hot and cold legs in the primary loop of a modern PWR.
FIGURE 8.8 The cross-sectional view of a typical RCP from a Westinghouse PWR. (Picture courtesy of Westinghouse.)
SGs, while American PWR manufacturers by Westinghouse and AREVA use vertical ones. In Russian VVER reactors,
the tubes inside of the SGs are mounted horizontally. An example of a horizontal SG is shown in Figure 8.10. The SG
in this picture is too small to be used in a commercial nuclear power plant. However, it does illustrate the concepts
involved. Vertical SGs in American PWRs use a feedwater ring on the outer edge of the SG to feed colder water from
the condenser down into the SG to be heated. The water is then directed downward along the outer annulus of the SG
8.7 TYPES OF REACTOR SGs 277
TABLE 8.2
Design Parameters for a Typical Westinghouse RCP
between its casing and an inner support plate known as the shroud. This flow channel where the water flows downward
within the SG is sometimes called the downcomer. It then flows through an orifice plate and is directed upward to flow
along the SG tubes where the water begins to pick up the heat. This increases its temperature until boiling occurs and
the water is converted into steam.
FIGURE 8.9 An example of a once-through vertical SG from a Westinghouse PWR. (Picture provided by Westinghouse Nuclear.)
TABLE 8.3
Design Data for the Westinghouse Model F SG, Which Is Used Extensively in Nuclear Power Plants
FIGURE 8.11 (a) The flow of fluid through a simple U-tube SG of the shell-and–tube design. This generator is s imilar to the design
of SGs that are used in nuclear power plants. (b) A comparison of the flow directions over the tubes in parallel-flow and counterflow
heat exchangers.
cylindrical chamber that is surrounded by a shroud. As the hot water from the primary loop passes through these
tubes, colder water is circulated over these tubes from the secondary side. Heat flows from the hot fluid in each tube
through the wall of the tube and into the cooler water flowing over it. This flow of heat heats the colder water and
removes heat from the hotter water. The flow of heat is proportional to the surface area A of these tubes, the number
of tubes in the SG, and the temperature difference ΔT between the inlet and the outlet. The following example helps
to illustrate these points.
8.10 SGs AND HEAT EXCHANGERS 281
The greater the surface area A of the tubes is, the easier it is to transfer heat from the primary side to the second-
ary side. The water on the secondary side is intentionally kept at a lower temperature and pressure than the water on
the primary side. This means that heat will always flow from the primary side to the secondary side. It does not take
a rocket scientist to realize how important this flow of heat can be. In most designs, water enters the tube nest from
the primary side at a temperature of about 330°C and a pressure of about 2,200 PSI (15.5 MPa). Cooler water enters
the SG from the secondary or shell side at a temperature of about 220°C and a pressure of about 900 PSI (6.9 MPa).
As this water flows over the hot tubes from the primary side, heat begins to flow into the cooler water. This causes
the water to boil and to eventually become pure steam. The steam generated is then sent on to the power turbines,
which use the steam to turn the blades of the turbines. This, in turn, produces electrical energy. In most SGs, the water
from the secondary side is allowed to completely vaporize before it reaches the end of the tube nest. To increase the
energy content of the steam even further, the steam exiting the SGs is designed to be slightly superheated. Today, most
reactor SGs are designed to superheat the outgoing steam by between 5°C and 20°C. Higher levels of superheat are
sometimes possible, but to attain these, the tube bank must either be made longer, the surface area of the tube bank
must be increased, or the pressure in the primary loop must be raised. There are generally space, size, or mechanical
constraints to keep this from occurring.
Similarly, under steady-state conditions, heat must always flow from the hot side to the cold side. If we do an energy
balance at any point along the surface of a tube, it must be true that
where c is the specific heat of fluid that the SG uses. In this case, we will assume that the fluid on both sides of a tube is
the same. Hence, before the fluid boils, the values of c will normally cancel, and we will end up with
This equation can be solved for the average axial temperature of the tube wall T WALL (z) and from this, we can deduce the
axial temperature profiles of the fluid in the hot and cold sides. The situation becomes slightly more complex when the
coolant on the low-temperature side begins to boil because we can no longer assume that dq(z) = m COLD c dT. However,
when we redo the analysis, it turns out that a counterflow SG will still produce more useful steam for the same mass flow
rate and geometry than a parallel-flow SG will. Because of this, counterflow or U-tube SGs are used in most nuclear
power plants today.
Solution The temperature drop across the SG is ΔT = TIN − TOUT = 30°C. A single tube has a surface area A = πDL =
0.94 m2 and an average tube transfers 100 kW of thermal energy between the primary and secondary sides. The equa-
tion governing the rate of heat transfer is Q′ = hAΔT. Since ΔT = 30°C, A = 0.94 m2, and Q′ = 100,000 W, the value of
h must be h = Q′/AΔT = 100,000/(0.94 × 30) = 3,546 W/m2 °C. Hence, the water leaving the SG is 30°C cooler than the
water entering the SG. [Ans.]
FIGURE 8.12 Sir Charles Parsons (left)—the inventor of the modern steam turbine shown on the right.
8.13 SG and Steam Turbine Pairings 283
FIGURE 8.13 The inside of a large steam turbine used in a nuclear power plant.
Steam enters the turbine from the top and initially hits the small blades that are closest to the shaft. The steam then hits
an intermediate set of blades, which causes the shaft to turn faster, and then it finally hits a much larger set of blades that
are attached to the rotor. Thus the momentum from the HP steam is transferred to the turbine blades to make the shaft
spin. The steam at the inlet to the turbine is always at the highest possible pressure (about 900 PSI), and the steam at
the exit of the turbine (where it has expended most of its energy) is at a much lower pressure (usually less than 1 ATM
or 14.7 PSI). However, it is also important to realize that as the superheated steam passes through the turbine, it loses
energy and pressure as it goes. This reduced pressure causes more water droplets to form in the steam. Hence, the steam
actually gets wetter the further it goes into the turbine. Eventually, a point is reached where it becomes too wet to con-
tinue without damaging the turbine blades. When this occurs, it must be collected and sent to another component of the
NSSS called the reheater for reheating. It can then be recycled and sent to a LP turbine to generate even more electricity.
We will have more to say about how this process works in the next section.
PWRs have adopted this “pairing principle.” The major difference between PWRs with different power levels is that
they may have different numbers of SGs attached to the core. Generally speaking, another SG and/or turbine combi-
nation must be added to a plant for each additional 500, 750 or 1,000 MW of thermal power it produces. This way
both the SGs and steam turbines can be designed as matched pairs to optimize the amount of electrical power that is
produced. This also helps reactor vendors to design plants that are more modular and less expensive. If a higher power
output is required, one simply adds more fuel assemblies to the core and an additional SG and steam turbine pair to
the secondary loop. In most designs, the SGs and the steam turbines are located inside of the containment building. In
some reactors that were built before 1980, the steam turbines were located outside of the primary containment in what
was called the auxiliary building.
FIGURE 8.14 A modern electric generator driven by a steam turbine in a nuclear power plant.
8.15 How an Electric Generator Works 285
simply a large rotating device that converts mechanical power from a rotating shaft into electrical power. It does so by
creating some relative motion between a stationary magnetic field and a conductor. In an electrical generator, the shaft
rotates and the magnetic coils that are used to generate the electricity are usually fixed. However, this does not always
have to be the case. The fixed coils are called the stator, and the moving shaft is called the rotor. The faster the rotor
turns with respect to the stator, the more electric power is produced. It should come as no surprise that most of the AC
power in the modern industrial world is generated in this way.
Nuclear power plants typically use electrical generators with three or four pole rotors. These rotors are used because
the temperatures and pressures in the NSSS are more suitable to three and four pole generators than those with just one
or two. The electrical output from the poles is combined together to produce an alternating current. Figure 8.16 shows
the actual process that is used. The number of poles is determined by the rate that the shaft turns the generator. The
shaft of an electrical generator in a nuclear power plant rotates at about 1,800 RPM. So the output of the three poles is
combined together to produce the alternating current that is generated. In a hydroelectric generator that rotates at a speed
FIGURE 8.15 A picture of British scientist and engineer Michael Faraday, who developed the modern-day electric generator.
His first electric generator, called the Faraday disk, is shown on the right. The horseshoe-shaped magnet created a magnetic field
through the disk. When the disk was turned, this induced an electric current radially outward from the center toward the rim. The
current flowed out through the sliding spring contact, through the external circuit, and back into the center of the disk through the
axle. (Pictures provided by Wikipedia.)
FIGURE 8.16 A single-phase generator (on the left) can produce single-phase AC power, while a three-phase generator
(on the right) can produce three-phase AC power. The current is then summed together in the three-phase generator to
generate a three-phase power.
286 The Nuclear Steam Supply System and Reactor Heat Exchangers
of 100 RPM, approximately 72 poles are used. Of course, this number varies slightly depending upon how the plant is
designed. More information about this can be found at the following URL:
http://www.asope.org/pdfs/AC_Electrical_Generators_ ASOPE.pdf
AC power having three phases is generally preferable to other forms of AC power because it uses less electrical conduct-
ing material to transmit the same electric power than equivalent single-phase or two-phase systems with an identical
voltage. For this reason and others, it was first patented by Nikola Tesla in 1887 and 1888, and it subsequently came into
widespread use in the early part of the nineteenth century. In a three-pole generator, the poles are arranged in such a way
that the individual currents from each pole vary sinusoidally at the same frequency but with a different phase (see Figure
8.17). The peaks and troughs of their respective waveforms are intentionally offset by one-third of a cycle to provide
three complementary ACs with a phase separation of 120°. The electrical power that is produced in this way is called
three-phase power. Three-phase electric power is the most common form of electrical power in the world today. Most
of the electrical power in the world is generated in this way. Three-phase AC power has a number of characteristics that
make it very desirable from a transmission perspective.
☉☉ First, in a three-phase system, the phase currents tend to cancel each other out. This makes it possible to eliminate
or reduce the size of the neutral conducting wire because all the phase conductors carry the same amount of
electric current and so, for a balanced load, they can all have the same size.
☉☉ Second, the power transferred to a linear balanced load will always be the same. This helps to reduce generator
and motor vibrations, and this can be a great practical advantage in certain types of applications.
☉☉ Finally, three-phase systems can produce a magnetic field that rotates in a specified direction (clockwise or
counterclockwise). This can help to simplify the design of an electric motor.
Thus, the combination of these three factors, is why a nuclear power plant always produces three-phase power. Three
is also the lowest phase order that is able to achieve all three of these goals. Each phase is typically represented by a
separate color that can vary from country to country. In the United States, black, red, and blue are used for the “live
wires” and white or gray is used for the “neutral wire.” A green or a bare copper wire is used for the ground. The AC
is typically produced with a frequency of 50 or 60 Hz. In Europe, 50 Hz is the preferred frequency for the current. In
the United States, the preferred frequency is 60 Hz. In the majority of the world, the power that a nuclear power plant
produces actually enters homes and factories in this way. In the United States and a few other countries, three-phase
power is generally not allowed to enter one’s home. Even in the areas where it does, it is typically split out at the main
distribution box, and most residential appliances are powered by the electric current that has only one phase. However,
three-phase currents are still used in most of North America to power electric stoves, electric clothes dryers, and some
types of dishwashers. So if your electricity is generated by a nuclear power plant, then this is the form that it will gener-
ally arrive at in your home. Figure 8.18 shows a large electrical generator that is used in a hydroelectric dam and another
design that is used in a nuclear power plant. A typical generator can be between 5 and 10 m long, and it can weigh several
thousand tons. The exact size depends on the rating of the generator, and how many generators are needed to generate
the total electrical output of the power plant. However, power plants having between 8 and 16 large generators are fairly
FIGURE 8.17 Three-phase current produced by a nuclear power plant. Here, the phases are 120 degrees out of phase.
8.16 Condensers and Other Heat Rejection Devices 287
FIGURE 8.18 A large electrical generator used in a dam (left) and a similar generator used in a nuclear power plant. Note the
immense scale and size that these electrical generators can have.
common. Each generator is then tasked with the job of generating between 20 and 100 MW of electrical power. For a
variety of reasons, electric generators above 100 MWE are not very commonly used. The shaft of the rotor in nuclear
power plants is rotated by steam pressure, and the shaft of the rotor in hydroelectric dams is rotated by water pressure.
In nuclear power plants, the design of the steam turbines influences how fast the shaft can rotate.
It is also possible to bypass the condenser directly and dump the spent steam back into the atmosphere or into another
“heat sink” outside of the plant. However, when this is done, the cooling water cannot be recycled or reused again.
288 The Nuclear Steam Supply System and Reactor Heat Exchangers
FIGURE 8.19 The anatomy of a condenser in a nuclear power plant. The picture shown here represents a simplified
view of the condenser.
Most of the time, the cooler water from the condensers is simply pumped back into the plant, where it is recycled. This
also allows any radioactivity the cooling water contains to be kept within the plant itself. An NSSS that implements this
feature is called a closed loop design. After the condensed water leaves the condenser, it is pumped through a water
purifier called a demineralizer before it is sent back to the steam-producing side of the NSSS. This removes any impuri-
ties that may have been added to the water by the previous stages. Sometimes the condensed water that enters the water
purifier is called the condensate.
8.17 T he Demineralizer
The water that leaves the condenser is then pumped through a device called a demineralizer. The demineralizer is
used to purify the water before sending it back to the core or the SGs. The demineralizer is a passive system that has
no moving parts. The purpose of a demineralizer is to remove impurities from the cooling water that would otherwise
cause the components of the NSSS to corrode. In the next chapter, we will discuss the effects that this corrosion can
have on a SG. If not removed, these impurities can considerably shorten the useful life of many of the components of
the NSSS. The cost of corrosion and radioactive contamination caused by poor water quality in nuclear power plant can
be enormous. Demineralizers are an attempt to control or completely eliminate this problem. Next, we will attempt to
address the mechanics of how demineralizers work. The chemical theory behind their operation can be found in the
References section at the end of the chapter. Basically, demineralizers are ion exchangers whose sole purpose is to hold
ion-exchange resins and transport water through them. Ion exchangers can be classified into two basic categories: single-
bed ion exchangers and mixed-bed ion exchangers. Both of these ion exchangers are housed in cylindrical tanks with
connections at the top for water inlet and resin addition, and connections at the bottom of the tanks for the water outlet.
The resins that are used can usually be changed through a connection at the bottom of the tank. The resin beads are kept
in the demineralizer by upper and lower retention plates, which are essentially fine screens with a mesh size smaller
than the resin beads themselves. The feedwater from the condensers enters the demineralizer at a set flow rate and flows
down through the resin beads, where the flow path causes physical filtering to occur as well as chemical ion exchange.
Normally, the pH of the water is also controlled in this way. The purified water then flows out the bottom of the tank.
A schematic of a single-bed demineralizer is shown in Figure 8.20. Several competing designs with different geo-
metrical configurations have been proposed over the years. However, the one shown in this figure is normally the
8.19 Types of Reactor Cooling Towers 289
simplest and easiest to implement. The ion-exchange process causes a number of positive (or negative) ions from the
resins to bond with the impurities in the cooling water. This tends to capture most of the suspended minerals, and deposit
them in the demineralizer. The chemistry of how this is done is a fascinating subject that we will not have time to cover
here. However, a brief discussion of these chemical processes can be found at the following URL:
http://nuclearpowertraining.tpub.com/
Other good discussions of this subject are available on the Internet.
FIGURE 8.21 A picture of a feedwater heater for a nuclear power plant in the United States.
of surface area is required for every 1,000 m2 that a comparable cooling pond or a lake would use. Cooling towers can
minimize the thermal pollution to lakes and rivers surrounding a power plant, and in addition, they can also allow most
of the c ooling water to be recycled.
The most common cooling towers that are used by nuclear power plants are called induced draft cooling towers. A
picture of an induced draft cooling tower is shown in Figure 8.22. In this design, the water is cooled by direct contact
with the air. This cooling effect is provided primarily by the exchange of latent heat between the water and the air,
which raises the temperature of the air. The heat transferred from the water to the air is then dissipated directly to the
atmosphere. This is the cause of the mist or steam that comes out of the top of the cooling towers. In an induced draft
cooling tower, the water from the condenser is dispersed throughout the cooling tower using a spray header. The water
is directed down over baffles that are designed to maximize the contact between water and air. The air is then drawn
over these baffles with a large number of large circulating fans. These fans cause evaporation to occur, and this causes
the water to be cooled. This system works well when the air in the vicinity of the power plant is cool and dry. However,
it does not work as well when the air is hot and humid. Hence, a safety factor must be built into the design of these
cooling towers so that they can still function on hot, humid days. A picture of a typical cooling tower is shown on the
left-hand side of Figure 8.23. Notice how small the reactor containment building is in comparison to the tower. This is
a picture of a natural draft cooling tower that is used at a European power plant. The cooling tower is approximately
400 ft (120 m) high.
FIGURE 8.22 A cross-sectional view of an induced draft cooling tower that is used in a nuclear power plant. Induced draft cooling
towers come in two varieties: a counterflow design on the left and a cross-flow design on the right.
8.20 Heat Transfer through a Reactor Cooling Tower 291
FIGURE 8.23 Some examples of reactor cooling towers. In general, reactors are very small in relation to their cooling towers.
FIGURE 8.24 A picture of the Empire State Building on the right and some reactor cooling towers on the left. In this picture, the
Empire State Building and the cooling towers are drawn to approximately the same scale.
A natural draft cooling tower differs from an induced draft cooling tower because it contains no moving parts.
Instead, the differential pressure difference between the cold air on the outside of the tower and the hot humid air on
the inside of the tower is used as the driving force to cool the steam. No circulating fans are required in a design like
this. Whether natural draft or induced draft towers are used depends on the climate that a nuclear power plant is to be
operated in. In colder climates, natural draft cooling towers tend to be more popular than induced draft cooling towers.
Figure 8.23 shows the Watts Bar Nuclear Plant in Knoxville, Tennessee. This particular power plant has two cooling
towers, and they are again fed with water from the Tennessee River (near the top of the picture). The Watts Bar nuclear
power plant is operated by the Tennessee Valley Authority, and it is used for electric power generation and tritium
production. The cooling towers at the Watts Bar power plant are 506 ft (150 m) high and have a diameter at the base of
405 ft (about 120 m). These two pictures illustrate how large a reactor cooling tower can be. As a point of comparison,
the Empire State Building, which is one of the highest buildings in North America, is only about 1,473 ft (449 m) or three
times as high. However, the Empire State Building is much narrower at its base (see Figure 8.24).
evaporation of some of the water passing through the air. Evaporation occurs as the result of a temperature difference
between the air and the water. The greater this temperature difference and the greater the surface area through which
the air and the water interact, the greater the amount of evaporation will be. To remove as much heat as possible from
the water, the water is sprayed into the surrounding air in the form of a light rain. An upward mass of cooler air is then
blown through the rain with a large cooling fan. The heat gained by the air must equal the heat lost by the water. Using
the thermodynamic properties we introduced in Chapter 7, the heat removed from the water is
or
where TIN is the temperature of the water entering the cooling tower, TOUT is the temperature of the water leaving the
cooling tower, m WATER is the mass flow rate of the water, and CP is its specific heat. Within the opposing airstream, the
rate of heat transfer from the water to the air is given by
AIR ( h 2 − h1 ) (8.6)
Q ′AIR = m
where
AIR is the mass flow rate of dry air through the tower (kg/s).
☉☉ m
☉☉ h1 is the enthalpy of the dry air entering the tower (kJ/kg).
☉☉ h2 is the enthalpy of the moist air exiting dry air (kJ/kg).
Since some of the water evaporates as it moves through the dry air, a small amount of it goes into increasing the humid-
ity of the air leaving the tower. The loss rate L′ of the water due to this evaporation can be found by examining the
relative humidity of the air entering and leaving the cooling tower. Since the rate of evaporation must equal the rate of
change in the humidity ratio (or the absolute humidity) of the airstream, the rate of heat transferred to the air as the result
of this change in the humidity is simply
AIR ( H 2 − H1 ) h 2 (8.7)
Q ′AIR = m
where
☉☉ H1 is the humidity ratio of the air entering the tower (kg of water per kg of dry air).
☉☉ H2 is the humidity ratio of the air leaving the tower (kg of water per kg of dry air).
☉☉ h2 is an expression of the specific enthalpy of the water at the cold water temperature (kJ/kg). (The enthalpy of
water is assumed to be zero at 0°C.)
8.21 Types of Heat Exchangers 293
Including the loss of heat through the evaporation process, the total heat balance between the air and the water can be
expressed as
AIR dh = L ′dT + m
m AIR dH h 2 (8.8)
L′ = m ( )
AIR dh/dT − h 2 dH/dT * (8.9)
This is a differential equation relating the mass flow rate of the air to the loss rate of the water for a given temperature
difference dT. Normally, about 1% of the water in the tower is evaporated for each 5°C temperature difference between
the dry air and the hot water arriving from the condensers. This cooling water must be made up by providing additional
cooling water to the condensers from an external source. Sometimes this additional water is also called makeup water. In
addition, some excellent references describing the design of reactor cooling towers can be found at the following URLs:
http://web.iitd.ac.in/~pmvs/courses/mel709/mel709-41.ppt#508,2,Natural Draught
Cooling Tower http://deltacooling.com/resources/principles-of-cooling-towers/
https://www.youtube.com/watch?v=norF9nLvzXQ
http://www.cti.org/whatis/coolingtowerdetail.shtml
The reader is encouraged to consult these references in the event additional information regarding the design or sizing
of a cooling tower is required. A complete derivation of Equation 8.9 can be found in A Comprehensive Approach to
the Analysis of Cooling Tower Performance by D.R. Baker and H.A. Shryock, printed in the August 1961 issue of the
Journal of Heat Transfer and available from Marley Cooling Technologies.
FIGURE 8.26 An annular tube used to transfer heat from a hot fluid (in red) to a cooler fluid (in blue).
294 The Nuclear Steam Supply System and Reactor Heat Exchangers
Heat is transferred from the hotter fluid to the colder fluid by heat transfer through the tube wall. Heat exchangers
normally do not perform external work, and there is usually a negligible kinetic energy or potential energy difference
between the two flow streams. The mass flow rates are adjusted between the streams to transfer as much thermal
energy as possible. For example, if the mass flow rate and specific enthalpy of the first stream are m 1 and h1 and the
mass flow rate and the specific enthalpy of the second stream are m 2 and h 2, then the mass flow rates are usually
adjusted so that
1 ∆h1 = m
m 2 ∆h 2 (8.10)
This equation is an expression of a steady-state energy balance between the streams. Perhaps the most common types
of heat exchangers in nuclear power plants are shell-and-tube-type heat exchangers. Nuclear SGs and condensers are
examples of this design. A schematic of a shell-and-tube-type heat exchanger is shown in Figure 8.27. Shell-and-tube-
type heat exchangers contain a large number of tubes (up to several thousand) that are packed into a shell with their axes
parallel to the axis of the shell. Heat transfer occurs when one fluid flows through the tubes inside of the shell while the
other fluid flows over the tubes within the shell. Baffles are strategically placed at convenient locations inside the shell to
force the shell-side fluid to flow along the axis of the shell and to increase the heat transfer rate with the tubes. Shell-and-
tube heat exchangers can further be classified as straight-through and U-tube heat exchangers depending upon whether
the tubes are aligned in a straight line or bent into a “U.” Within the shell, the fluid can also pass over the tubes multiple
times. A heat exchanger in which the fluid passes over the tubes only once is called a one-pass heat exchanger, and a heat
exchanger in which the fluid passes over the tubes twice is called a two-pass heat exchanger. In many designs, the fluid
can pass over the tubes three or four times. This is an example of what is called a multipass heat exchanger. Normally,
the more passes that are made, the more efficient the heat exchanger becomes. Heat exchangers can be further classified
into parallel-flow heat exchangers and counterflow heat exchangers. In parallel-flow heat exchangers, both the hot and
cold fluids enter the heat exchanger at the same end and move in the same direction. In counterflow heat exchangers,
the hot and cold fluids enter the heat exchanger at opposite ends and flow in opposite directions. The flow direction then
affects how quickly the heat can be removed from the hot fluid and transferred to the cold fluid. The temperature profiles
for each type of heat exchanger are shown in Figure 8.28. In most cases, a counterflow heat exchanger is more efficient
at removing heat from the hot fluid than a parallel-flow heat exchanger. When a heat exchanger uses a large number
of tubes that are spaced very closely together, a bundle of these tubes is sometimes called a tube nest (see Figure 8.29).
FIGURE 8.27 A shell-and-tube heat exchanger with baffles to force the fluid along the primary axis of the shell.
8.22 Heat Exchanger Design 295
FIGURE 8.28 The temperature profiles in parallel-flow and counterflow heat exchangers.
FIGURE 8.30 The direction of the flow fields and the behavior of the coolant temperatures in parallel-flow and
counterflow heat exchangers.
because fluids with different temperatures are flowing over the inner and outer surfaces of the tubes, we must also calcu-
late the values of THOT (z) and TCOLD (z) at each axial location. In the example that follows, we will assume that all of the
tubes are oriented parallel to one another and that they are oriented in an upright position. In other words, the bottom of
a tube is located at z = 0, and the top of a tube is located at z = L. The length of a tube is simply L. We will also assume
that the tubes containing the hot fluid are circular in shape, although this does not always have to be the case. Circular
tubes are easier to fabricate, and in most cases, an ordinary steel pipe meets most design requirements.
Reactor heat exchangers take this process one step further because we can no longer ignore the fact that at least
one side of the tubes may be covered with steam. To cut down on the buildup of crud or minerals on the tube surface
(see Section 8.27), the tubes are generally made from exotic alloys of stainless steel, and water chemistry is carefully
controlled to keep the buildup of the crud to a minimum. However, whenever boiling occurs, there will always be some
crud deposited on the surfaces of the tubes, and this will cause the heat transfer rate to decrease over time. However, if
we are careful about how we analyze the tubes, we can account for the effects of this crud buildup in a very simple way.
Now let us turn our attention to the problem at hand, and show how the heat flow between the hot and cold fluids can be
found. Consider the simple circular tube shown in Figure 8.31, where there is a hot fluid flowing inside of the tube in one
direction and there is a cold fluid flowing over it in another direction. This tube is just one of a large number of identical
FIGURE 8.31 A tube in a counterflow heat exchanger where the temperature changes with axial position.
8.23 Finding the Heat Transfer Rate through a Heat Exchanger Tube 297
tubes that a heat exchanger can have. In practice, a “typical” heat exchanger can have anywhere from 1,000 to 10,000 of
these tubes. (Reactor SGs can have even more.) So if we can predict the heat flow rate Q′ out of one tube, then we can
′
multiply by the number of tubes N in the heat exchanger to get the total heat flow rate Q TOTAL . Thus,
Q ′TOTAL = ∑ Q′ (8.11)
i
i
8.23 Finding the Heat Transfer Rate through a Heat Exchanger Tube
It is possible to predict what the axial temperature profile in a heat exchanger looks like by performing an energy bal-
ance between the hot and cold streams. No matter what the length of the channels is, the heat that is lost from the hot
channel is given by
q HOT = m ( )
HOT c HOT THOT IN − THOT OUT (8.12)
q COLD = m ( )
COLD c COLD TCOLD OUT − TCOLD IN (8.13)
Normally, the specific heats are evaluated at constant pressure. Thus, the symbol c is really equivalent to the spe-
cific heat at constant pressure (Cp). In parallel-flow heat exchangers, the fluid flow in both of the channels is in the
same direction, and in counterflow heat exchangers, it is in the opposite direction. To calculate the temperature
profiles, we have to add a coordinate system to indicate which way the fluid is flowing. The easiest way to do this
is to use the coordinate system shown in Figure 8.32. The diagram at the top of the figure shows the direction of
the flows in a parallel-flow heat exchanger, and the diagram at the bottom shows the direction of the flows in a
counterflow heat exchanger. Other than the difference in the direction of the flow on the cold side, the two heat
exchangers are identical. Now from our previous discussion, we know that the total amount of heat lost from the
hot channel is given by
q HOT = m ( )
HOT c HOT THOT IN − THOT OUT (8.14)
and that the total amount of heat gained by the cold channel is given by
q COLD = m ( )
COLD c COLD TCOLD OUT − TCOLD OUT (8.15)
However, these expressions do not tell us what the heat flow q(z) is as a function of axial position z. To find this, we need
to calculate the heat flow dq(z) across a small cross-sectional area dA between the hot and cold channels.
298 The Nuclear Steam Supply System and Reactor Heat Exchangers
FIGURE 8.32 A fluid energy balance for a parallel-flow and a counterflow heat exchanger.
If the wetted perimeter of the tubes is P, then the value of this differential cross-sectional area must be
dA = Pdx. (8.16)
Obviously, if all of the tubes are circular tubes, then the value of dA is
A = 2πR dz (8.17)
where R is the radius of the tubes. So if this is true, and we are considering only single-phase flow with uniform fluid
properties, then the amount of heat flowing out of the hot channel is
dq(z)= − (mc)
HOT dTHOT (z) (8.18)
dq(z) = (mc)
COLD dTCOLD (z) (8.19)
where the minus sign shows that heat is leaving the hot channel and the plus sign means that heat is being added to the
cold channel. Hence, using only a simple energy balance, we can write
Now, suppose that we subtract the value of dTHOT (z) from the value of dTCOLD (z) in the aforementioned expressions.
After doing so, we find that
(
dTHOT (z) − dTCOLD (z) = − dq(z) 1 (mc) ) (
COLD (8.21)
HOT + 1 (mc)
)
This expression applies to all axial locations from z = 0 to z = L. Notice that the sign of m will be positive (+) for
the cold channel in a parallel-flow heat exchanger, and it will be negative (−) for the cold channel in a counterflow
heat exchanger. With the exception of this sign difference, the energy balance that we have just written applies to both
parallel and counterflow heat exchangers. We also know from our discussion of convective heat transfer in Chapter 10
that
dq(z) = h ( THOT (z) − dTCOLD (z) ) dA = − h ( THOT (z) − dTCOLD (z) ) Pdz (8.23)
where P is the wetted perimeter. Now, suppose that we substitute Equation 8.23 into Equation 8.21. After some
rearrangement, we obtain
( dTHOT (z) − dTCOLD (z)) ( THOT (z) − dTCOLD (z)) = − dz hP {(mc) COLD } (8.24)
HOT − (mc)
(
ln THOT out − TCOLD out ) (T
HOT in )
− TCOLD in = − hA 1 (mc){
HOT + 1 (mc)
COLD (8.25) }
where A = PL, and L = ZOUT − ZIN. Finally, from our previous discussion, we know that
(
HOT = q HOT THOT IN − THOT OUT (8.26)
(mc) )
and
(
COLD = q COLD TCOLD OUT − TCOLD IN (8.27)
(mc) )
and we also know that q HOT must be equal to qCOLD everywhere along the channel, so we can set
c)HOT and (m
If we substitute these values for (m c)COLD back into Equation 8.25, we soon discover that
( ) (
q = hA THOT OUT − TCOLD OUT − THOT IN − TCOLD IN )
(8.29a)
(
ln THOT OUT − TCOLD OUT ) (T HOT IN − TCOLD IN )
or
q = hA ∆TAVG (8.29b)
300 The Nuclear Steam Supply System and Reactor Heat Exchangers
Here, the average temperature difference ΔTAVG is given by a rather unusual expression of the form
( ) ( ) (
∆TAVG = THOT OUT − TCOLD OUT − THOT IN − TCOLD IN ln THOT OUT − TCOLD OUT ) (T HOT IN )
− TCOLD IN (8.30)
In the next section, we would like to examine the implications of this unusual expression.
( )
∆TAVG = ( ∆TMAX − ∆TMIN ) ln ∆TMAX ∆TMIN (8.31)
In practice, this interesting equation is also called the log-mean temperature difference (or LMTD). In plain English, the
LMTD is the temperature difference at one end of the heat exchanger minus the temperature difference at the other end
of the heat exchanger, divided by the natural logarithm of the ratio of these temperature differences.
The LMTD is the fundamental driving force that allows us to find the heat flow rate between a hot channel and a cold
one. It allows us to calculate the heat flow rate q for a single channel i. To calculate the heat flow rate for a heat exchanger
consisting of N identical but separate channels (or tubes), then we simply multiply Equation 8.29b by the value of N:
For a typical heat exchanger in a nuclear power plant, a common value for N can be between 1,000 and 10,000. For
very large heat exchangers, N can be as large as 25,000 or 30,000. Hence, the only other parameters we need to know
to calculate the thermal capacity of a single-phase heat exchanger are the surface area A of the tubes and the overall
heat transfer coefficient h. If we know the perimeter of the tubes P, it is easy to find the surface area since A = PL,
and L is the length of the tube bank. However, the heat transfer coefficient h is more difficult to obtain because it
consists of a number of terms including the thermal resistance of the tube and the convective heat transfer coefficients
on each side of the tube. From our discussion in Chapter 10, we will see that the total heat transfer coefficient for a
single tube is given by
( ) ( )
h = 1 1 h TUBE IN A IN + R + 1 h TUBE OUT A OUT (8.33)
where R is the thermal resistance of the tube wall. For a cylindrical tube having an inner radius r1 and an outer
radius r2, R = ln (r2/r1)/2πkL where k is the thermal conductivity of the tube. If the tube is a flat plate of thickness Δx,
then R = Δx/kA where Δx is the distance over which the temperature gradient occurs The same logic can be extended
to multilayer tubes having different material layers as well. For these tubes, we simply adjust the value of R to account
for these additional layers (see Chapter 10).
In practice, we can sometimes get away with the first assumption, but the second and third assumptions can be much
more limiting. Obviously, if the fluid on either side of the tubes boils, then we would have to redo our derivation using
an entirely difference energy balance. This is exactly what happens in a reactor SG where the primary fluid may be
slightly superheated and the secondary fluid is almost always highly superheated. Finally, the third assumption can be
equally important because of entrance effects, fluid viscosity changes, and thermal conductivity changes. For this rea-
son, numerical methods are normally used to correct for these shortcomings. However, if the two fluids do not change
phase, and the tubes are long enough, then using the LMTD method can lead to reliable results. Example 8.5 serves to
illustrate a problem where this is true.
Hence, we have employed LMTD to calculate the amount of heat transferred through a single-phase heat exchanger
tube. This approach allowed us to calculate the heat transfer rate when the surface of the tube was clean and also when
it was dirty. If we know the total amount of heat that needs to be transferred to the secondary side, we simply have to
divide QTOTAL by the heat that a single tube is able to transfer to find the number of tubes N that are needed to do the
job. If the inlet and exit temperatures are known, and the surface area A of the tubes is known, then this is a relatively
simple thing to do. However, as soon as one (or both) of the fluid streams start to boil, then this becomes a more difficult
problem. Most of the time, studies of this type are performed with the help of a computer program or a computer code.
When the fluid in the tubes begins to boil, we can no longer use the LMTD to find the total heat flow rate, and we must
resort to a computer simulation of the entire system. Normally, a computer program like RELAP is used to perform these
calculations. However, starting with RELAP-5, Mod 2, the source code is no longer available to tweak the empirical
correlations that some SGs and heat exchangers use. Then, other computer programs such as TRACE or FLUENT must
be used when the mass flow rates are low or when there is a violent mixing of the cold water and the high-temperature
steam in the downcomer. This problem will be explored in more detail in Chapter 23.
FIGURE 8.33 The temperature profiles for the hot and cold streams in the parallel-flow heat exchanger
discussed in Example Problem 8.5.
302 The Nuclear Steam Supply System and Reactor Heat Exchangers
((
d ( THOT − TCOLD ) ( THOT − TCOLD ) = − 1 (mc) ) (
HOT ± 1 (mc) ))
COLD hP dz (8.34)
where we implicitly assumed that the sign of the last term was positive. However, if the cold fluid flows in the opposite
direction compared to the hot fluid, then the sign of this second term must be negative. So a more general approach
would have been to write Equation 8.34 as
where
((
ξ = 1 (mc) ) (
HOT ± 1 (mc)
COLD (8.36)))
When we use this alternative expression for ξ, we find that ξ is smaller because 1/(m
c)HOT and 1/(m
c)COLD are no longer
a dditive. We can represent this by writing
( )
THOT OUT − TCOLD OUT = THOT IN − TCOLD IN e − ξhA (8.37)
So when the fluids flow in the same direction, ξ has a larger value than when the fluids flow in the opposite direction. If
you think about it, this implies that the value of THOT OUT − TCOLD OUT will be smaller when the fluids flow in the opposite
direction, and it will be larger when the fluids flow in the same direction. If we interpret the temperature difference
between THOT OUT and TCOLD OUT as a measure of the relative efficiency of the heat exchanger, then a counterflow heat
exchanger will clearly be more efficient than a parallel-flow heat exchanger. This is the reason why almost all heat
exchangers (e.g., condensers, SGs, and reheaters) used in nuclear power plants are of the counterflow type. They are
simply better at pushing around the same amount of heat if the geometry of the heat exchanger remains the same!
FIGURE 8.34 The temperature profiles for the hot and cold streams in the counterflow heat exchanger discussed in
Example Problem 8.6.
where h is the total heat transfer coefficient for the tube, and the LMTD is given by
The value of h is normally found from Equation 8.34, or from Equation 8.46 (see Section 8.29), when the tubes become
fouled. The heat flow rate through a heat exchanger consisting of N identical tubes is then
where the LMTD can be interpreted as the average temperature difference ΔTAVG along the tube (see Equation 8.31).
The use of the LMTD for multitube heat exchangers is illustrated in Example 8.7. Reactor heat exchangers can have
thousands of these tubes, and in very large plants, the value of N can approach 20,000 or 30,000. The fluid flow in these
heat exchangers is not only parallel to the tube nest, but in some cases, it is also perpendicular to it. This perpendicular
flow is sometimes called cross-flow, and heat exchangers that take advantage of this flow are called cross-flow heat
exchangers. In the next section, we would like to demonstrate how the LMTD can be applied to these designs. In some
applications, cross-flow heat exchangers are more economical when it comes to space utilization than concentric tube
heat exchangers. In addition, the fluid velocities are usually at right angles to the tube bank.
coefficient for a representative tube is 100 W/m2 °C, what is the value of the LMTD for a tube, and what is the capacity
of the entire heat exchanger?
Solution In this heat exchanger, the surface area of a single tube is A = πDL = 6.28 cm × 300 cm = 1,884 cm2 =
0.1884 m2 , and the surface area of all the tubes is 1,000 × 0.1884 = 188.4 m2. The value of the LMTD is LMTD =
(40°C − 20°C)/ln(40/20) = 28.9°C. The capacity of the entire heat exchanger is QHX = Nh Ao LMTD = 1,000 × 100 ×
0.1884 × 28.9 = 544,475 W. [Ans.]
where F is a new parameter called the cross-flow correction factor, which typically has a value between 0.5 and 1.0.
There are many different designs in which cross-flow exists, and the amount of cross-flow can vary from 0% to 100%.
The size of F is geometry dependent, and it also depends on the size of the tubes and the tube-to-tube spacing. The
reader is referred to El-Wakil (1981) for more information concerning this relationship. Hence, a value for F of 0.8 would
result in a heat transfer rate of approximately 20% less than the heat transfer rate for a concentric tube heat exchanger.
Automobile radiators are a good example of cross-flow heat exchangers. Shell-and-tube heat exchangers can also exhibit
a significant amount of cross-flow.
( ) ( ) ( )
h TOTAL = 1/R = 1 ( R1 + R 2 + R 3 ) = 1 1 h i A i + ln r2 r1 /2πkL + 1 h o A o (8.42)
where A is the surface area of the tube, the subscripts “i” and “o” refer to the inner and outer tube walls, and R is
the equivalent thermal resistance. Equation 8.42 is based on the assumption that the walls of the tubes are clean
and smooth. It is also assumed that no mineral deposits (which are sometimes referred to as crud) could build up on
the surface of the walls. Unfortunately, this is almost never the case. In practice, the chemicals in the coolant plate
out on the surface of the tubes if a heat exchanger is operated for a long period of time. This happens even more
quickly when the coolant is allowed to boil. A heat exchanger tube containing some representative crud deposits
is shown in Figure 8.35. It should therefore come as no surprise that this can become a significant issue in reactor
SGs and condensers, because SGs produce a lot of steam, and condensers convert a lot of it back to colder water.
This builds up a layer of crud on each side of the tube wall, and the buildup of this crud can adversely affect the
heat transfer rate over a period of time. This reduction in the heat transfer rate must be taken into account when an
SG is designed. Otherwise, the steam generator will not be able to produce the amount of steam that is desired. We
can estimate the effect that this crud has on the overall heat transfer coefficient by adding a thin conductive layer
of thickness t to each side of the tube wall. In this case, the thermal resistance has four terms in it instead of three,
and we can illustrate this by writing
R = R1 + R 2 + R 3 + R CRUD (8.43)
where now RCRUD is the additional thermal resistance the crud layers provide. The total thermal resistance is then
R = R1 + R 2 + R 3 + R CRUD (8.44)
8.30 Tube Fouling Factors 305
FIGURE 8.35 A fouled heat exchanger tube (on the left) and fouled and clean tube bundles (on the right).
or
Thus it is easy to see that any value of RCRUD (except zero) will increase the value of the denominator and reduce the
value of the heat transfer coefficient hTOTAL. Moreover, if we completely eliminate the final term, then the total heat
transfer coefficient becomes the heat transfer coefficient that we would use if there was no crud buildup. For this reason,
the water supplied to a SG must be very pure, free of particulate matter, and any unwanted chemicals. When the water
begins to boil, these chemicals can plate out and deposit themselves on the surface of the tubes in the tube nest. This can
either clog up the tubes or cause them to fail entirely. Even if they do not fail or clog up, a great deal of undesired cor-
rosion can occur. To prevent this from happening, most reactors employ a WPS (or Water Purification System) to make
sure that the water flowing through the plant is as clean as possible and that its pH is maintained at the appropriate levels.
In PWRs, soluble boron (in the form of boric acid) is also dissolved in the coolant on the primary side to help control
the reactor power level and the neutronic characteristics of the core. So water chemistry can be extremely important to
monitor and control on both the primary and secondary sides of an SG. If the water chemistry is not carefully controlled
within very strict limits, then the fuel rods in the core or the tubes in the SGs will begin to corrode and may even fail.
Many problems with SGs over the years have been attributed to the buildup of this crud.
( ) ( ) ( )
h TOTAL =1 1 h i A i + ln r2 r1 /2πkL+ 1 h o A o (8.47)
where Ao and Ai are the areas of the outer and inner surfaces, and ho and hi are the heat transfer coefficients at those loca-
tions. This is the equation for the heat transfer coefficient of a clean tube surface. In this equation, the wall resistance
of the tube, RTUBE, is written as
( )
R TUBE = ln r2 r1 2πkL (8.48)
306 The Nuclear Steam Supply System and Reactor Heat Exchangers
where k is the thermal conductivity of the material from which the tube is made. Then, the overall heat flow Q rate must
be calculated using the outer surface of the tube, and the surface area in this case must be
A o = 2πro L (8.49)
where L is the length of the tube. When a heat exchanger surface becomes fouled with dirt and debris, an additional
thermal resistance term R FOUL must be added to the energy balance to account for these effects. In this case, the heat
transfer coefficient when the tube is fouled is given by
Notice that when fouling occurs, RFOUL will always have a finite value, and the value of the overall heat transfer coef-
ficient in this case will always be decreased. Normally, the value of the fouling factor is measured experimentally and
entered into a database that the designer of the heat exchanger uses. Some typical values of the fouling factors for some
representative coolants are shown in Table 8.4. Others can be found in El-Wakil (1981) at the end of the chapter.
However, in the case of a SG, things are not so clear-cut because the fouling factor can change with time. Initially, the
fouling factor starts out at a value very close to 0, and the value of R FOUL in Table 8.4 for clean or distilled water can be
used. However, the longer the SG continues to operate, the larger the value of R FOUL will become. Usually, an experiment
is performed to measure the size of R FOUL as a function of time. Then, the value of the fouling factor can be expressed as
( )
R FOUL (t) = R FOUL 0 1 + At + Bt 2 (8.51)
where R FOUL 0 is the initial fouling factor (at t = 0) and A and B are empirically determined constants called fouling coef-
ficients. These coefficients are used to account for the buildup of the crud in the tubes as the SG continues to operate. If
B = 0, then this equation is said to be a first-order equation for the fouling factor, and if B ≠ 0, then this equation is said
to be a second-order equation for the fouling factor. Hence, this approach can be adapted to find the total fouling factor
R FOUL (t) as a function of operating time. Every reactor SG supplier has its own database in which the values of these
fouling factors are stored. However, this data is usually proprietary, and SG vendors have gone to great lengths to keep
it private. Thus, this data is usually not available in the open literature. However, if one is willing to look hard enough,
some publicly assessable tables for the values of these fouling factors do exist. Many of these tables having representa-
tive values of R FOUL can be found on the Internet. Example 8.7 illustrates the effect these fouling factors can have on the
overall heat transfer coefficient.
TABLE 8.4
Tube Fouling Factors or Thermal Resistances for Different Types of Fluids
Fluid or Coolant Thermal Resistance RFOUL (h °F ft2/BTU) Thermal Resistance RFOUL (m °K/W)
Seawater 0.0005 0.000072
Well water 0.001 0.000144
Hard water 0.003 0.000432
Mississippi River water 0.003 0.000432
Chicago sanitary canal water 0.008 0.001152
Distilled water 0.0005 0.00072
Exhaust steam 0.001 0.000144
Source: Wolf, H. Heat Transfer, Harper & Row, New York (1983).
8.32 Fluid Properties and Their Effect on Thermal Efficiency 307
Solution Before the tube becomes fouled, its thermal resistance per unit length is RTUBE = ln(r2/r1)/2πk, and after it
becomes fouled, its thermal resistance per unit length is RTUBE = R FOUL + ln(r2/r1)/2πk. For a stainless steel tube with
a wall thickness of 4 mm, r1 = 11 mm and r2 = 15 mm. The value of ln(r2/r1)/2πk is 0.3/100π ≅ 0.00095 m °C/W, and
this is the thermal resistance before it becomes fouled. The tube fouling factor for distilled water (from Table 8.4)
is 0.00072 m °C/W. The thermal resistance of the tube after it becomes fouled is RTUBE = 0.00095 + 0.00072 =
0.00167 m °C/W. Hence, the fouling of the tube wall increases its thermal resistance by about 75%! [Ans.]
FIGURE 8.36 In a large PWR SG, baffles and support plates are used to reduce the vibration in the tube nest and prevent the
tubes from wearing prematurely. In this picture, the baffles are wrapped around the tube bank and the orifice plate is used to
help regulate the flow.
308 The Nuclear Steam Supply System and Reactor Heat Exchangers
TABLE 8.5
The Specific Volume, the Density, the Specific Enthalpy, and the Specific Entropy of Water and Steam in the
Secondary Loop of the NSSS (System Pressure: 7 MPa)
pressure in the secondary loop (where the SGs are located), and the second pressure corresponds to the pressure in
the primary loop (where the core is located). Notice that the values for the specific volume, the density, the specific
entropy, and the specific enthalpy are much different on the primary side than they are on the secondary side. In
Chapter 9, we will use these thermodynamic properties to estimate the thermal efficiency of a commercial PWR. This
requires a specific knowledge of the temperature and the pressure of the reactor coolant as it enters and leaves each
component of the NSSS. In general, the thermodynamic efficiency increases as the temperature and the pressure rise.
A similar procedure can be used for BWRs, but in this case the physical properties must be evaluated at 7.5 MPa or
1,050 PSI. Again, these properties are presented for different operating conditions in Appendix G. Now, let us turn
out attention to the subject of reactor thermal cycles.
Bibliography 309
TABLE 8.6
The Specific Volume, the Density, the Specific Enthalpy, and the Specific Entropy of Water and Steam in the
Primary Loop of the NSSS (System Pressure: 15.5 MPa)
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc. (1972).
310 The Nuclear Steam Supply System and Reactor Heat Exchangers
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi (2000).
Thomas, L. “Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Wolf, H. Heat Transfer, Harper & Row, New York (1983).
Web References
http://web.iitd.ac.in/~pmvs/courses/mel709/mel709-41.ppt#508,2,Natural Draught Cooling Tower.
http://www.peacesoftware.de/einigewerte/wasser_dampf_e.html.
http://deltacooling.com/resources/principles-of-cooling-towers/.
https://www.youtube.com/watch?v=norF9nLvzXQ.
http://www.nrc.gov/reading-rm/basic-ref/teachers/04.pdf.
http://naabzist.com/pdf/cooling%20tower%20desingn.pdf.
http://www.cti.org/whatis/coolingtowerdetail.shtml.
http://nuclearpowertraining.tpub.com/.
Other Reference
Baker, D.R. and Shryock, H.A. A Comprehensive Approach to the Analysis of Cooling Tower Performance, August 1961 issue of the
Journal of Heat Transfer and available from Marley Cooling Technologies.
Exercise 8.2
Water enters the core of a commercial BWR as a subcooled liquid with a temperature of 270°C and leaves the core as a
saturated liquid with a temperature of 288°C. What is the enthalpy of the coolant at the inlet and the outlet of the core?
Assume that the pressure of the water is 7.25 MPa (1,050 PSI) and the exit quality is 15% (x = 0.15).
Exercise 8.3
Water enters the core of a PWR as a subcooled liquid with a temperature of 285°C and leaves the core as a subcooled
liquid with a temperature of 320°C. What is the enthalpy rise of the coolant through the core? What is the equilibrium
quality of the water at the inlet of the core and at the outlet of the core?
Exercise 8.4
A 1,000 MWE PWR generates 3,300 MW of thermal energy. Water enters the core of a PWR as a subcooled liquid with
a temperature of 275°C. The mass flow rate through the core is 20,000 kg/s. What is the temperature of the water leav-
ing the core? What is its equilibrium quality? Assume that the core is maintained at a pressure of 15.5 MPa (2,250 PSI).
Exercise 8.5
Water enters the core of a PWR as a subcooled liquid with a temperature of 280°C and leaves the core as a subcooled
liquid with a temperature of 310°C. What is the enthalpy rise of the coolant through the core? At what temperature will
the water first begin to boil? Assume that the core is maintained at a pressure of 13.8 MPa (2,000 PSI).
Exercise 8.6
Water enters the core of a BWR as a subcooled liquid with a temperature of 272°C and leaves the core as a saturated
liquid with a temperature of 288°C. At what temperature will the coolant in the core first begin to boil? Assume that the
core is maintained at a pressure of 7.25 MPa (1,050 PSI).
Exercise 8.7
Derive an expression for the specific enthalpy and the specific internal energy of the coolant leaving the core of a BWR
if the equilibrium quality is 15%. If the saturation temperature is 288°C and the exit pressure is 7.25 MPA (1,050 PSI),
calculate the values for each of these properties assuming that the mixture is a homogeneous one.
312 The Nuclear Steam Supply System and Reactor Heat Exchangers
Exercise 8.8
Water enters a PWR SG as a subcooled liquid with a temperature of 230°C and leaves the SG as a superheated vapor
with a temperature of 280°C. What is the enthalpy increase of the coolant across the SG tube bank? Assume that the
pressure of the water on the shell side of the SG is 7 MPa (1,015 PSI).
Exercise 8.9
In Exercise 8.8, at what temperature does the water first start to boil and what amount of superheat does it possess when
it exits the tube bank?
Exercise 8.10
Derive an expression for the density and the specific volume of the water leaving a BWR core if the e quilibrium quality
is 14%. If the core is maintained at 7.25 MPA (1,050 PSI), at what temperature will the water first start to boil? Calculate
the values for each of these properties at the core exit assuming that the mixture is a homogeneous one.
Exercise 8.11
Water enters a PWR SG as a subcooled liquid with a temperature of 225°C and leaves the SG as a superheated vapor
with a temperature of 270°C. What is the enthalpy of the coolant at the entrance and at the exit of the SG? Assume that
the pressure of the water on the secondary (or shell) side of the SG is 5.75 MPa (~835 PSI).
Exercise 8.12
HP water flows through a reactor core at the rate of 20,000 kg/s. If the average rise in the temperature of the water
through the core is 30°C and the water enters the core at 290°C, what is the heat transfer rate between the fuel rods and
the coolant? Assume that water in the pressure vessel is maintained at 2,250 PSI (15.5 MPa).
Exercise 8.13
A heat exchanger contains 1,000 parallel tubes that are used to transfer heat between a high-temperature fluid and a
low-temperature fluid. The surface area of the tubes is 1,000 m2. Hot water enters the heat exchanger on the primary
side at 250°C and leaves at 100°C, while cold water enters the heat exchanger on the secondary side at 20°C and leaves
at 80°C. If the heat exchanger is a counterflow heat exchanger and heat transfer coefficient for an average tube is
100 W/m2 °C, what is the value of the LMTD and what is the capacity of the entire heat exchanger?
Exercise 8.14
A cooling tower located in a Scandinavian country receives hot water from a geothermal power plant at 100°C and
returns cold water to the plant at 30°C. The pressure of the water is 100 kPa. If the mass flow rate of the water through
the tower is 1,000 kg/s, how much thermal energy does the tower discharge into the air in 30 s?
Exercise 8.15
A horizontal tube passes through a heat exchanger where water enters the tube at 20°C and leaves the tube at 40°C.
Saturated steam is then passed over the tube at 80°C where it is condensed. The heat transfer coefficient on the outer
surface of the tube is 10,750 W/m2 °C, and the heat transfer coefficient on the inner surface of the tube is 8,775 W/m2 °C.
The tube is made of brass with a thermal conductivity of k = 110 W/m °C. The tube has an ID of 1.34 cm and an outer
diameter of 1.59 cm. Calculate the total thermal resistance of the tube.
Exercise 8.16
Using the LMTD, calculate the heat flow rate through the tube in the previous problem (Exercise 8.15). Assume that the
tube is 1.58 m long. If a heat exchanger contained 1,000 of these tubes, and they were arranged in such a way that the
heat transfer rates were identical, how much steam would 1,000 of these tubes condense in 1 s? Assume that a latent heat
of vaporization is 2,308 kJ/kg.
9
Reactor Thermal Cycles
9.1 T he Purpose of Reactor Thermal Cycles
Thermal energy is circulated through a nuclear power plant by the coolant, which uses this energy to produce useful work and power.
The heat produced in the core is transported through the nuclear steam supply system (NSSS) by reactor coolant pumps in what is
called a reactor thermal cycle (see Chapter 8). The NSSS converts the thermal energy produced in the core into high-pressure steam,
which in turn drives the blades of a steam turbine. The blades of the turbine are connected to the shaft of an electric generator, which
in turn produces electric power. After leaving the turbines, the steam is converted into liquid and the process is repeated again. The
waste heat is removed from the steam using a device called a condenser. As long as the coolant is recirculated in the plant, a thermal
cycle of this type is called a closed thermal cycle. Closed thermal cycles are by far the most common thermal cycles in use today.
Coal-fired power plants as well as nuclear ones use these thermal cycles. Thermal cycles can even be used to extract energy from the
ocean (see Figure 9.1).
Most reactor thermal cycles use variations of a famous thermodynamic cycle called the Rankine cycle, which we will discuss in
this chapter. Virtually, all electric power in the world today is produced using this cycle. The thermodynamic efficiency η of a nuclear
power plant is simply the ratio of the amount of electrical energy the plant produces to the amount of thermal energy it produces. For
most reactors, the aggregate thermal efficiency is about 35%. The thermal efficiency of a reactor thermal cycle can be determined
by means of an energy balance. This energy is represented by a thermodynamic property of the coolant called the enthalpy (see
Chapter 7). The enthalpy H is measured in J, and the enthalpy per unit mass is measured in J/g or kJ/kg. The enthalpy per unit mass
is called the specific enthalpy (also see Chapter 7). The specific enthalpy h is a function of the type of coolant that is used, and in
most cases, this coolant is water. The specific enthalpy is a function of temperature and pressure (T and P), but it is also a function of
FIGURE 9.1 Thermal cycles in which a working fluid such as freon is circulated in a loop can also be used to extract energy from the ocean.
(Design concept provided by the ASME.)
313
314 Reactor Thermal Cycles
the temperature and the entropy (T and S). If the coolant is a two-phase mixture, then the specific enthalpy of the liquid
phase is hl and the specific enthalpy of the vapor phase is hv. The specific enthalpy of the two-phase mixture is then
where x is another thermodynamic property called the quality of the mixture. The quality is simply the fraction of the
mass of the two-phase mixture that is vapor. Hence, the quality is defined as
x = M V ( M V + M L ) (9.3)
where ML and MV are the masses of the liquid and the vapor, respectively. This definition is important because it enables
us to find the specific enthalpy of any two-phase mixture. The values of the specific enthalpy are also shown in the steam
tables (see Appendix G). Thus, if we also know the mass flow rate m of the coolant that possesses this specific enthalpy,
then the total amount of energy E carried by the coolant per second as it moves through the plant is
where m is the mass flow rate in kg/s and h is the specific enthalpy in J/kg (or kJ/kg). Multiplying these variables together
then gives the energy flow rate E′. The energy flow rate is normally measured in J/s or W. It may also be measured in
kJ/s or kW. In large power plants, the energy flow is expressed in megawatts (MW).
FIGURE 9.2 The inventors of the two most common thermal cycles used in the world today: William Rankine (on the left) and
George Brayton (on the right). Among other things, William Rankine made many other contributions to the understanding of the
thermal cycle of the modern steam engine. (Pictures provided by Wikipedia.)
FIGURE 9.3 The four steps in the classic implementation of an ideal Rankine cycle. (Picture provided by Wikipedia.)
316 Reactor Thermal Cycles
FIGURE 9.4 A picture showing how the Rankine cycle is implemented for a PWR.
FIGURE 9.5 A picture showing how the Rankine cycle is implemented for a BWR.
thermodynamic underpinning of most coal-fired power plants since 1900. In the Rankine cycle, the working fluid consists
of a combination of water and steam. Water is converted into steam, which turns the blades of a power turbine, and the
steam is then converted back to water, which is re-heated and converted into steam again. Hence, the Rankine cycle can
be thought of as a closed loop cycle because water (the working fluid) continues to be recirculated through the plant. The
Rankine cycle consists of four distinct processes or steps (see Figure 9.3), and each step in the cycle can be linked to a
specific point on a temperature–entropy diagram (a T–S diagram) that engineers use to describe the state of the fluid as it
moves through the cycle. From the perspective of the NSSS, these four steps are as follows:
☉☉ Step 1: The working fluid is pumped from a low-pressure state to a higher pressure one. As the fluid is a liquid at
this stage in the process, the pump requires little energy to move the fluid.
9.3 The Steps in the Rankine Thermal Cycle 317
☉☉ Step 2: The high-pressure liquid enters a boiler (or reactor core) where it is heated at constant pressure by an
external heat source to become a dry saturated vapor. The energy required to do this can be easily calculated from
Mollier diagram (i.e., an h-s chart) or from an enthalpy–temperature–pressure diagram, which in tabular form is
represented by a steam table.
☉☉ Step 3: The dry saturated vapor expands through a turbine, which generates power in the process. This decreases
the temperature and pressure of the steam, and some condensation may occur. The energy change of the steam
can be easily calculated using an enthalpy–entropy chart or the steam tables.
☉☉ Step 4: The wet steam then enters a device called a condenser where it is condensed at a constant temperature to
become a saturated liquid. The saturated liquid is then returned to Step 1.
This process can be depicted visually by a T–S diagram such as the one shown in Figure 9.6. In this type of diagram,
the area under the curve (which in this case connects points 1, 2, 3, and 4) is the useful work W the working fluid can
perform. These points can also be correlated to specific values of the specific enthalpy h, and if we know the values of
the enthalpy h1, h2, h3, and h4 at points 1, 2, 3, and 4, and we know the mass flow rates at these points in the piping sys-
tem, then we can calculate the amount of work W (e.g., the amount of electrical power) that the system can perform. Of
course, there are small mechanical inefficiencies and other irreversibilities in the steam turbines, the coolant pumps, and
the piping system, but they are generally small enough to neglect in our initial analysis. We can then equate the useful
work W to the amount of electrical power P that is produced. Hence, the thermal efficiency in this case is
where Q is the thermal output in MW, and W is the work that the fluid performs (e.g., the area under the curve in a T–S
diagram). In many textbooks, the thermal efficiency is given the symbol η (eta). Note that the thermal efficiency is a
dimensionless number when measured in this way. Now when it comes to a nuclear power plant, let us consider some
of the temperatures and pressures that a typical Rankine cycle uses. In a PWR (which operates at about 2,250 PSI), the
coolant in the primary loop enters the core at a temperature of about 560°F (293°C) and exits the core at a temperature of
about 615°F (324°C). So at a pressure of 2,250 PSI, its temperature rises only about 55°F (~30°C). By design, the coolant
in the core is never allowed to boil (except occasionally near the top of the core in the hottest fuel assemblies), because
this boiling can cause the rods to dry out and it can also cause the rods to fail.
FIGURE 9.6 A T–S diagram used to depict an ideal Rankine cycle. The four basic steps in the Rankine cycle are represented
by indices 1–4.
318 Reactor Thermal Cycles
However, because there is no steam generator in a BWR, the opposite problem occurs, and the water in the core is
allowed to boil by design. Normally, this boiling starts about one-quarter of the way up the core. However, the boiling
must occur in such a way that the liquid film does not leave the surface of the fuel rods, because this can cause the rods
to heat up prematurely and it can also cause the rods to fail. So in either case, the Rankine cycle in a Light Water Reactor
(or LWR) must be implemented in such a way that the fuel rods never “dry out,” and this places some constraints on what
the thermal efficiency of the plant can be. In American BWRs, the water enters the core at a pressure of about 1,050 PSI
and at a temperature of about 525°F (274°C) and leaves the core at a temperature of about 550°F (288°C). So the coolant
in a BWR has a lower temperature and pressure than it does in a PWR, and this behavior is primarily intended to make
sure that the water boils without removing the liquid film from the fuel rods.
where WTURBINE is the useful work (or electricity) produced by the turbine, WPUMP is the amount of work required to
power the pump, and MWT is the number of megawatts of thermal energy produced in the core. In practice, the amount
of work required to power the pump is usually small compared to the amount of electricity generated by the turbine, so
for all practical purposes we can write
where MWT is the thermal output of the reactor in MW. This expression is valid as long as the pump work is small and
there are no irreversible losses in the piping system. We can then derive an explicit expression for the value of WTURBINE
based on the thermodynamic properties of the steam flowing into and out of it. Now let us do this using the simple
Rankine cycle in Figure 9.6 as our guide.
WTURBINE = m
3h3 − m
4 h 4 (9.9)
3 and m
Unless the steam is “reheated” before it leaves the turbine, m 4 are the same, and we can set them equal into the
at the turbine inlet. Then, the expression for the useful work performed by the turbine becomes
mass flow rate m
( h 3 − h 4 ) (9.10)
WTURBINE = m
9.5 An Energy Balance through the NSSS 319
is the inlet flow rate (in kg/s). The corresponding expression for the thermal efficiency becomes
where m
( h 3 − h 4 ) MWT (9.11)
η= m
Now, let us apply this expression to the schematic of the PWR shown in Figure 9.7. For a typical PWR, the enthalpies
at points 1 through 4 are
The Enthalpies in the Secondary Loop of a PWR based on a Simple Rankine Cycle
h1 = 188.5 kJ/kg (T = 45°C, P = 10 kPa)
These numbers are based on the steam leaving the steam generators at a pressure of about 7 MPa and at a tempera-
ture of about 295°C. Since the saturation temperature at this point is about 285°C, a typical PWR steam generator is
designed to support about 10°C of superheat. The steam leaving the turbines at point 4 is relatively wet and has an exit
quality of about 60% and a saturation temperature of about 200°C. (At higher exit temperatures, the quality is less.)
When it leaves the condenser, it becomes a saturated liquid again, and its saturation temperature is determined by the
saturation pressure. Normally, the pressure inside the condenser is kept at about 10 kPa or 1.5 PSI. This results in a
saturation temperature of about 45°C. Thus, lowering the condenser pressure lowers the saturation temperature and
reduces the specific enthalpy at point 1. The specific enthalpy is then increased by the feedwater pumps as they pump
the saturated water back to point 2. Because the pressure is higher when it leaves the pumps (about 7 MPa or 1,000
PSI), the enthalpy increases because of the pressure change. In other words, the pumps transfer some of their mechani-
cal energy to the water to increase its specific enthalpy. The energy to do so is provided by the electrical generators
attached to the steam turbines. Now let us calculate the thermal efficiency of the plant in this case. The thermal output
in MWT is
( h 3 − h 2 ) (9.13)
MWT = Q IN = m
FIGURE 9.7 A picture depicting the energy flows in a Rankine thermal cycle in a typical PWR. Indices 1–10 indicate specific
points in the cycle where the fluid enthalpy or entropy must be found in order to calculate the thermal efficiency η. Since a PWR
consists of two coolant loops, additional indices are required to describe the entire cycle.
320 Reactor Thermal Cycles
( h 4 − h1 ) (9.14)
Q OUT = m
The Thermal Efficiency of a Typical PWR found from the Heat Input and the Heat Output
η = 1 − (QOUT/QIN) = 1 − (h4 − h1)/(h3 − h2) = 1 − 1,824/2,620 = 30.4% (9.15)
Alternatively, the thermal efficiency can be found from the work equation
η = W Q IN (9.16)
where W is the net work the thermal cycle performs. For this very simple Rankine cycle, W = WTURBINE (OUT)
− W PUMP (IN) where
( h 3 − h 4 ) (9.17)
WTURBINE (OUT) = m
( h 2 − h1 ) (9.18)
WPUMP (IN) = m
The Thermal Efficiency of a Typical PWR found from the Work Equation
η = ((h3 − h4) − (h2 − h1))/(h3 − h2) = (803 − 6)/2,620 = 30.4% (9.19)
In other words, Equations 9.15 and 9.19 lead to exactly the same result! Hence, a representative PWR using an ideal
Rankine cycle converts about 30.4% of the heat it sends to the steam generators into useful work. Note that this is the
simplest possible Rankine cycle that a nuclear power plant can employ. However, it is not always the most efficient one
because the waste heat that leaves the steam turbines cannot be recycled and used again. If we do not recycle the waste
heat, this would result in an aggregate efficiency between 30% and 31%. However, because nuclear power plants are so
expensive to build, most utilities prefer to operate them at an efficiency of between 33% and 35%, and this requires using
a modified Rankine cycle to recycle some of the waste heat to make the numerator in Equation 9.16 higher. We would
now like to discuss how this can be done.
ηCARNOT = 1 − (Tmin/Tmax) = 1 − (45 + 273)°K/(295 + 273)°K = 0.44 = 44% (9.20)
This difference in thermal efficiency is due to irreversible losses in the pumps and the power turbines as well as fric-
tional losses as the water and the steam cycle through the secondary loop. Specifically, the irreversibility in this case
results in a thermal efficiency reduction of (44% − 30.4%)/44% ≅ 31%. Losses of this type are common in any practical
Rankine cycle that is used in a nuclear power plant.
Each of these processes attacks the problem of improving the thermal efficiency in a slightly different way, and so
we would like to discuss them separately to illustrate how they can be used. In general, all three of these methods are
implemented in commercial nuclear power plants to improve their thermodynamic performance. When all three of these
methods are implemented simultaneously, they can increase the thermal efficiency of a typical plant by between 3% and
5%. Although this may not seem like much, this efficiency increase can reduce the cost of the electricity that is generated
by between 10% and 20%. In other words, the financial payback is so large that all three of these methods are routinely
implemented in modern nuclear power plants. Now let us discuss how each of these methods is implemented in the NSSS.
( h 3 − h 4 ) MWT (9.21)
η= m
superheat increases the numerator by increasing the value of h3 but it does not increase the denominator. (The value of
h4 remains unchanged.)
322 Reactor Thermal Cycles
FIGURE 9.8 (a) Superheating the steam increases the area under the curve in a T–S diagram for an ideal Rankine cycle. The red
area represents the additional work that can be performed. (b) A simplified diagram of the reheat process in a light water reactor.
The reheater for reheating the steam is shown in red.
TABLE 9.1
In BWRs, the Saturation Temperature of Water Is Reached at a Pressure of about 7.15 MPa, and in the Primary Loop of a PWR, It
Is Reached at a Pressure of Approximately 15.5 MPa. The Secondary Loops of PWRs Operate at Pressures between 6 and 7 MPa
Saturation Temperature (°C) Pressure (MPa) Saturation Temperature (°C) Pressure (MPa)
100 0.10142 280 6.41646
200 1.55467 290 7.44164
210 1.90739 300 8.58771
220 2.31929 310 9.86475
230 2.79679 320 11.2839
240 3.34665 330 12.8575
250 3.97594 340 14.6002
260 4.69207 350 16.5292
270 5.50284 360 18.6664
heat to this steam before it is discarded, then it will have some additional energy that we will be able to extract. This is
the fundamental premise that both reheat and regeneration use. However, each process differs slightly in the way that it
attempts to reuse this waste heat. Consequently, we would like to show what each of these strategies implies.
FIGURE 9.9 To achieve the maximum possible efficiency when reheating the steam, one usually sets the reheat pressure equal to
one-quarter of the maximum system pressure within the steam-generating loop. In other words, for the reheat to be as effective as
possible in a nuclear power plant, Preheat ≅ ¼PMax. The red area above shows the additional work that the reheat process allows the
reactor to perform.
324 Reactor Thermal Cycles
is known as the “reheater” using hotter steam from another power source. In BWRs, this hotter steam comes directly
from the reactor core. In PWRs, it comes from the steam generators. In any event, the reheated steam is then sent on to
one or more low-pressure turbines. Because these turbines operate at lower pressures than the high-pressure turbine, the
reheated steam may even be superheated at this point. In a modern PWR, the reheating process creates about 50°C of
superheat before the steam enters the low-pressure turbine. Hence, the electrical output of the plant is actually increased
because we are getting power from two turbines rather than just one. This process is illustrated in Figure 9.8b. The
reheating process is illustrated in red. In practice, moisture separation and reheating are usually performed in the same
physical unit, and there is one unit for each low-pressure turbine. The increase in the thermal efficiency is then given by
where ΔMWE is the increase in electrical output that can be attributed to the “kick” provided by the additional low-
pressure turbines. To be more specific, ΔMWE consists of a separate piece for each turbine, and after the primary
turbine, we can write this as
∆MWE = ∑ ∆MWE
n
n = ∆MWE1 + ∆MWE 2 + ∆MWE 3 + + ∆MWE N (9.23)
where the summation is to be performed over all the secondary turbines from 1 to N.
Notice that the thermal efficiency increases because the numerator in Equation 9.22 is increased, while the denomi-
nator (which is the total thermal output of the reactor) remains unchanged. From the low-pressure turbines, the spent
steam is then sent to a condenser, where the waste heat is removed, and it is converted back into liquid water. Sometimes
this water is saturated, and sometimes it is not. The exact thermodynamic state depends on the design of the condenser.
It also depends on whether or not the condensed water is to be reheated again. Ordinarily, reheating the spent steam adds
another 2% to the thermal efficiency of a typical nuclear power plant. Since this can mean about a 2% increase in its
electrical output, the cost of the additional moisture separators and reheaters will quickly pay for itself. Not surprisingly,
reheating is used in virtually all nuclear power plants and coal-fired power plants in operation today. It simply provides a
great deal of additional “bang for the buck.” The process of reheat can be understood thermodynamically by considering
the two images shown in Figure 9.9. The T–S diagram for the reheat process is shown on the right. The Rankine cycle
illustrated here is different than an ideal Rankine cycle because the expansion of the steam takes place in two stages
rather than just one. In the first stage (which involves the high-pressure turbine), the hot steam is expanded isentropically
to an intermediate pressure, and it is then sent back to the reheaters where it is reheated at constant pressure (usually to
the inlet temperature of the high-pressure turbine). It is then expanded isentropically through the low-pressure turbine
before being sent on to the condenser. Thus, the total heat input and the total turbine work output when a reheat stage is
added to the Rankine cycle are as follows:
and
Obviously, reheating the steam increases the thermal efficiency because more work can be performed by a two-stage
turbine for the same energy input (h3). Incorporating a single reheating stage between the turbines can increase the
overall efficiency by about 2% by increasing the average temperature at which heat is transferred to the steam. The aver-
age temperature during the reheating process can be increased even further by increasing the number of expansion and
reheating stages and by adding more stages to the turbine drive shaft. However, a point is eventually reached (normally
at about three stages) where the resulting gain in thermodynamic efficiency is too small to justify the additional cost
and complexity. When this occurs, the expansion and reheating cycle approaches an isothermal process at the maximum
temperature shown in Figure 9.10.
In most nuclear power plants, this point occurs when more than two reheating stages are used. The theoretical
increase in the thermal efficiency from adding a second reheating stage is about half as great as the increase in the
9.10 Method 3: Regenerating the Cooling Water 325
FIGURE 9.10 The average temperature at which heat is transferred to the steam during reheating increases as the number of reheat
stages increases. The thermal efficiency improvement when multiple stages are used is equal to the thermal efficiency increase for a
single stage multiplied by (1/2)N where N is the n umber of reheat stages.
thermal efficiency from a single reheating stage alone. For this reason, very few power plants are built with more than
two reheating cycles. The thermal efficiency therefore increases as
where Δη(N−1)REHEAT is the efficiency improvement due to the previous reheating stage and N is the number of stages that
are used. Reheat was first implemented in coal-fired power plants in 1925, but it was abandoned in the 1930s because
there were too many operational problems associated with its implementation. However, single-stage reheat was reintro-
duced into the same plants in the late 1940s as higher pressure boilers became more reliable and double-stage reheat was
introduced in the early 1950s. Reheat temperatures are normally very close or equal to the high-pressure turbine inlet
temperature, and the optimum reheat pressure is about one-quarter of the maximum cycle pressure. For example, if the
maximum pressure in the secondary loop of a PWR is 7 MPa, the optimum reheat pressure is about 1.75 MPa. This can
be seen in the reheat cycle shown in Figure 9.9. The reheat pressure ratio applies to some nuclear power plants but not to
others. In modern plants, it can sometimes be closer to 20%.
FIGURE 9.11 An example of how the process of regeneration is implemented within the NSSS. The feedwater heaters that
recirculate regenerated coolant water to the core are shown in red.
the condensers is relatively cold, and if we can add additional waste heat to it before it is sent back to the steam genera-
tors (or to the core in the case of a BWR), then it will have a higher “energy content” before it goes back to the steam-
producing components. Consequently, less additional heat is needed to be able to superheat it again. The net effect of
recycling this waste heat is that the thermal efficiency of the power plant will increase. The regeneration process in
a nuclear power plant is implemented by extracting or “bleeding” steam from the steam turbines between the high-
pressure and low-pressure stages, and in some cases, after the steam exits the secondary stage. Some of this depleted
steam is then used to heat the feedwater. This regeneration process not only improves the thermodynamic efficiency of
the power plant but also helps to deaerate the feedwater by removing any air that may leak into the feedwater from the
condenser. This helps to prevent the tubes in the steam generators from corroding, and it also helps to control the flow
rate of the steam exiting the low-pressure turbines. Regeneration has been used in virtually all nuclear power plants
since they were first introduced in the 1950s. In nuclear power plants, regeneration is normally implemented using a type
of feedwater heater called a closed feedwater heater.
In closed feedwater heaters, heat is transferred from the extracted steam to the feedwater without any direct mixing
of the hot and cold streams. The two streams can therefore be maintained at different pressures thermodynamically
since they do not mix. In coal-fired power plants, other feedwater heaters called open feedwater heaters are sometimes
used. An open feedwater heater is basically a large mixing chamber, where the steam extracted from the turbines mixes
with the water returning from the condensers. In a good design, the water leaves the mixing chamber as a saturated
liquid at the pressure of the heater. The regenerated water is then sent back to a steam generator or the reactor core to
be converted into steam again. The implications of this process are shown for a power plant with one feedwater stage in
the T–S diagram in Figure 9.12a. Under ideal conditions, the feedwater is heated to the exit temperature of the extracted
steam, and then returned to the core or the steam generators. However, in real power plants, the feedwater leaves the
feedwater heater below the temperature of the steam because a temperature difference of at least a few degrees is
required inside of the feedwater heater to maintain the heat transfer rate. The condensed steam is then either pumped
into the feedwater system or routed to another heater or to one of the condensers through a device called a trap. The trap
traps the remaining vapor but allows the residual liquid to be throttled to a lower pressure region. The energy content of
the steam remains constant throughout this throttling process.
FIGURE 9.12 (a) An example of regeneration in a nuclear power plant. On the top left: a power plant running an ideal regenera-
tive Rankine cycle with one open feedwater heater. On the top right: a T–S diagram for an ideal regenerative Rankine cycle with
one open feedwater heater. On the bottom left: a power plant running an ideal regenerative Rankine cycle with one closed feedwater
heater. On the bottom right: a T–S diagram for an ideal regenerative Rankine cycle with one closed feedwater heater. (Pictures
provided by https://ecourses.ou.edu/.) (b) The temperatures and pressures at each step in the regeneration process for a coal-fired
power plant (on the left) and for a nuclear power plant (on the right). Notice that coal-fired power plants operate at higher boiler
temperatures and pressures than nuclear power plants do and are therefore more efficient. However, they also produce considerably
more emissions.
328 Reactor Thermal Cycles
low-pressure turbine and eventually reaches the condenser at a lower pressure and enthalpy that we will call state 7.
This steam transfers its energy to the condenser and leaves the condenser as a saturated liquid and at the condenser
pressure that we will call state 1. The condensed water which is then referred to as the feedwater passes through a pump
where it is compressed to the feedwater heater pressure (state 2). Here, it is routed to one of the feedwater heaters, where
it receives energy from the steam extracted from the power turbines. Enough steam is extracted from the turbines so that
the feedwater leaves the feedwater heater as a saturated liquid and at the heater pressure, which corresponds to state 3.
A second pump raises the pressure of the water to state 4, which is the pressure where it enters the steam generators or
is sent back into the core.
The thermodynamic cycle is completed by adding heat to the water directly from the core or through the steam gen-
erators so that the steam enters the high-pressure turbine at a pressure and an enthalpy, which corresponds to state 5. In
other words, the feedwater “climbs” the left-hand side of the vapor dome shown in Figure 9.12a, and because its energy
content is higher, the area under the thermal cycle curve is higher as well. This implies that the thermal efficiency is
also greater. The efficiency of the Rankine cycle can be increased even further by adding additional feedwater heaters
to the flow. Figure 9.12b shows the pressures at each step in this process for a coal-fired power plant on the left and for
a nuclear power plant on the right. In the case of a coal-fired power plant, the temperatures and pressures at state 5 are
higher because the boiler operates at a much higher temperature. Thus, the steam at this point has more internal energy.
For example, in a BWR, the system pressure is about 7 MPa and the saturation temperature is about 286°C, while in the
secondary loop of a PWR, the system pressure is about 7 MPa and the superheat temperature is about 292°C because
the steam generators are designed to superheat the steam. The pressure at point 5 in a coal-fired power plant is about
15 MPa, while the temperature is about 600°C. Now let us compare the enthalpies at each step in this process. For a
coal-fired power plant, the enthalpies at points 1 through 7 are
The Enthalpies for a Coal-Fired Power Plant with an Ideal Regenerative Rankine Cycle
h1 = 191.8 kJ/kg
h 2 = 193.0 kJ/kg
h 3 = 798.3 kJ/kg
h 4 = 814.0 kJ/kg (9.27)
h 5 = 3,583.3 kJ/kg
h 6 = 2,860.2 kJ/kg
h 7 = 2,115.3 kJ/kg
These enthalpies assume the use of a single open feedwater heater. If a BWR or PWR is designed with the same
f eedwater heater, the corresponding enthalpies would be
h 2 = 193.0 kJ/kg
(
h 5 = 2,803.7 kJ/kg T = 292°C, P = 7 MPa )
h 6 = 2,243.2 kJ/kg
h1 = 191.8 kJ/kg
h 2 = 193.0 kJ/kg
h 3 = 798.3 kJ/kg
(9.29)
h 4 = 814.0 kJ/kg
h 5 = 2,773.5 kJ/kg
h 6 = 2,220.2 kJ/kg
h 7 = 1,500.0 kJ/kg
Note that the values for h5 and h7 are different for both reactor types because the average pressures are about half as high.
Now let us compare the thermal efficiencies for each design. In the case of the coal-fired power plant,
where f is the fraction of the steam extracted from the steam turbine to power the feedwater heaters (typically 0.23% or
23%). The thermal efficiency of the coal-fired power plant is then
Now let us turn our attention to some representative nuclear power plants, which in this case are an American PWR and
a BWR. In the case of the PWR, h5 = 2,803.7 kJ/kg, and in the case of the BWR, h5 = 2,773.5 kJ/kg. The values of QIN are
η(PWR) = (QIN − QOUT)/QIN = 1 − (QOUT/QIN) ≅ 34.7%
(9.37)
η(BWR) = (QIN − QOUT)/QIN = 1 − (QOUT/QIN) ≅ 33.2%
Thus, the superheat in the PWR results in a slightly higher thermal efficiency (34.7% vs. 33.2%) that we can obtain in the
BWR. As we mentioned previously, this is due to the fact that the steam generators are designed to provide an additional
5°C–10°C of superheat. Again, the design parameters used in this exercise can vary somewhat from one plant to the
next. However, they are representative of the conditions that are actually encountered in practice.
330 Reactor Thermal Cycles
the feedwater heaters affect the numerator and not the denominator. The numerator is affected because the water that
enters the core (or the steam generators) has a higher energy content, and so it requires less additional energy to be
FIGURE 9.13 An ideal regenerative Rankine cycle with multiple feedwater heaters. In this case, four closed feedwater heaters are
used to reheat the feedwater.
9.14 Reducing the Condenser Temperature and Pressure 331
FIGURE 9.14 Feedwater heaters are extensively used to preheat the cooling water delivered to a reactor core. Preheating the feed-
water reduces the irreversibilities in the steam generation process and therefore improves the thermodynamic efficiency. They also
help to reduce plant-operating costs. Feedwater heaters are typically shell and tube designs where the feedwater passes through the
tubes and is heated by steam extracted from the power turbines. Feedwater heaters are designed with one to three zones to accom-
plish this task. (The picture was provided by Indiana Engineering Industries—see http://ieipl.co.in/product&service.html.)
FIGURE 9.15 A cross-sectional view of a two-zone feedwater heater. The steam from the power turbines enters the heater from above
where it reheats the feedwater in the condensing and subcooling zones. An energy balance can be used to determine the amount of
reheating provided. The feedwater heater above is a two-zone reheater. (Picture provided by Korea Energy—see www.korenergy.co.kr.)
superheated. Thus, the regeneration process results in a positive increase in the net work performed and nothing more.
However, a 2% efficiency gain is still highly desirable because it can increase the electrical output of the plant by about
2%. Hence, the regeneration process (and the feedwater heaters that are needed to implement it) is cost-effective enough
to use in almost every nuclear power plant in the world today.
rejection occurs. Since steam exists in the condenser as a saturated mixture at the saturation temperature corresponding
to the local pressure, lowering the pressure in the condenser has the effect of lowering the temperature of the steam and
therefore the temperature at which heat is rejected from the working fluid. This process can be used effectively as long as
the saturation temperature inside the condenser is higher than the saturation pressure and temperature of the heat sink into
which the heat is being dumped. For example, suppose that the condenser is to be cooled by drawing cold water from a
nearby lake or river where the average temperature of the water is 15°C. Normally, a temperature difference of about 10°C
is required between this fluid and the water–steam mixture inside of the condenser for the heat transfer process to be effec-
tive. The steam temperature inside of the condenser must therefore be above 25°C. Therefore, the condenser pressure must
not fall below 3.2 kPa (0.46 PSI), which is the saturation pressure at an average temperature of 25°C. One of the biggest
problems with lowering the condenser pressure is that it increases the possibility of air leaking into the condenser. More
importantly, it increases the moisture content of the steam exiting the last stage of the power turbines. This decreases the
turbine efficiency and can damage the turbine blades, which normally spin at very high speeds. Fortunately, this problem
can be corrected by superheating the steam before it reaches the blades of the high-pressure stage. The superheated steam
is drier than ordinary steam, and it causes less erosion as it passes through to the low-pressure turbine.
In most nuclear power plants, the temperature to which the steam can be heated is limited by metallurgical consider-
ations or by the temperature at which the fuel rods can be operated in the core. With the steam turbines in service today,
the highest steam temperatures that can be tolerated in the high-pressure stage are about 625°C (~1160°F). Above this
temperature and pressure (approximately 1,000 PSI or 7 MPa), most materials begin to deform or fatigue and the service
life of the turbine blades is reduced. Future generations of nuclear power plants will probably use more exotic materials
in an attempt to eliminate this problem. Ceramic compounds similar to those used in nuclear fuel rods and on the U.S.
Space Shuttle hold a great deal of promise in this regard.
9.16 Balancing the Energy Flow between the Primary and Secondary Loops in a PWR
In PWRs, LMFBRs, and in most gas reactors it is important to know how the mass flow rates in the primary loop com-
pare to the mass flow rates in the secondary loop. In this section, we would like to show how these flow rates can be
determined. Normally this is done by performing an energy balance on the flow streams. Consider for the moment the
picture of the NSSS shown in Figure 9.17. Assume for the moment that the total power generated in the core is given by
9.16 Balancing the Energy Flow between the Primary and Secondary Loops in a PWR 333
FIGURE 9.16 On the left: some condenser internals and on the right: how reducing the pressure inside of the condenser can increase
the thermal efficiency by increasing the net work. Here, an ideal Rankine cycle is shown. (Condenser picture provided by Wikipedia.)
TABLE 9.2
The Thermal Efficiencies of Various Types of Nuclear Power Plants
Reactor Coolant Used in the Coolant Inlet Coolant Outlet Coolant Temperature Overall Thermal
Type Primary Loop Temperature (°C) Temperature (°C) Difference (°C) Efficiency η (%)
CANDU Heavy water 270 310 40 29
BWR Light water 278 293 15 32
PWR Light water 300 330 30 34
LMFBR Liquid sodium 380 540 160 40
AGR Carbon dioxide 290 640 350 42
Source: Todreas and Kazimi.
FIGURE 9.17 An overview of the two primary loops in the NSSS of a PWR. In PWRs, the primary and secondary loops are
isolated from each other by high-pressure tubes through which the heat is transferred.
the symbol P (not to be confused with the system pressure). Also assume that the mass flow rates are measured in kg/s
and that the specific enthalpies are measured in J/kg or kJ/kg. The primary coolant enters the core at point 7 with a mass
flow rate and enthalpy of m 7 and h7, and leaves the core at point 8 with a mass flow rate and an enthalpy of m
8 and h9. In
most cases, the mass flow rate entering and leaving the core is the same, so
7 =m
m 8 ≡m
P (9.39)
334 Reactor Thermal Cycles
P refers to the mass flow rate through the primary loop. From a simple energy balance, we can see that the total
where m
power produced by the core is then
P ( h8 − h 7 ) (9.40)
P=m
After the primary coolant leaves the core, it moves on to the steam generators if the reactor happens to be a PWR. The
primary coolant enters the steam generators at point 9 with a mass flow rate and a specific enthalpy of m 9 and h9, and
leaves the steam generators at point 10 with a mass flow rate and a specific enthalpy of m 10 and h10. Similarly, the coolant
from the secondary side enters the steam generators at point 2 with a mass flow rate and a specific enthalpy of m 2 and
h2, and it leaves the steam generators at point 2 with a mass flow rate and a specific enthalpy of m 2 and h2. In most cases,
the mass flow rate entering and leaving the steam generators on each side of the flow stream is the same, so we can set
9 =m
m 10 = m
P (9.41)
on the secondary side. Here, m S is the mass flow rate in the secondary loop. Under steady-state conditions, the mass flow
rates on the primary and the secondary sides must be adjusted so that the total heat lost from the fluid on the primary
side must be equal to the total heat gained by the fluid on the secondary side. To make this happen, we must require that
P ( h 9 − h10 ) = m
m S′ ( h 3 − h 2 ) (9.44)
and this is the condition that the steam generator must support during steady-state operation. So if the enthalpy of the
fluid entering and leaving the primary side is known, and the enthalpy of the fluid entering and leaving the secondary
side is known, then the ratio of the flow rates in the primary and the secondary loops must be
The Ratio of the Mass Flow Rates between the Primary and Secondary Loops
S = ( h 3 − h 2 ) ( h 9 − h10 ) (9.45)
P m
m
the steam flows directly through the turbines, and no steam is diverted to other subsystems (such as the r eheaters), then
the total energy converted to mechanical work by the turbines is
S ( h 3 − h 4 ) (9.46)
W=m
where W is the mechanical work in J/s or kJ/s (W or kW). Equation 9.46 assumes that the energy lost in the turbine
due to friction is small. In most cases, this is a good assumption. When energy is extracted from the steam, the specific
enthalpy of the steam will drop below its saturation value hv. When this happens, it becomes a liquid–vapor mixture
whose specific enthalpy is given by
h = xh v + (1 − x)h l (9.47)
or equivalently
h = h l + xh lv (9.48)
where x is the quality of the steam, hlv is the enthalpy of vaporization, and when the mixture is in thermal equilibrium,
0 < x ≤ 1.0.
m S ( h 4 − h1 ) ( h 6 − h 5 ) (9.49)
CON = m
It is also possible to bypass the condenser directly and dump the spent steam back into the atmosphere or into another
“heat sink” outside the plant. However, when this is done, the spent steam cannot be recycled and reused again. However,
most of the time, the condensed steam is pumped back into the plant, where the entire process is repeated again.
A NSSS that treats the coolant in this way is called a closed loop design. We will compare this to an open-loop design
in Section 9.20. After the condensed steam leaves the condenser, it is pumped through a water purifier or demineralizer
before it is sent back to the steam-producing side of the NSSS. This removes any impurities that may have been added to
FIGURE 9.18 Two pictures of condensers that may be found in nuclear power plants. (Images provided by Wikipedia.)
336 Reactor Thermal Cycles
the water by the previous stages. Sometimes the condensed water that enters the water purifier is called the condensate.
Condensers are heat exchangers that are similar in many respects to feedwater heaters. However, because their operat-
ing pressures are lower, condensers are not always implemented the same way. They are essentially surface condensers
in which a large number of tubes containing cold water are allowed to pass through a stream of depleted steam (see
Figure 9.19). The steam is converted from its gaseous state to its liquid state at pressures below atmospheric pressure. As
we discussed previously, the pressure inside the condenser is kept as low as possible to increase the aggregate thermal
efficiency. However, it cannot be lower than the saturation pressure and temperature of the heat sink into which the heat
is being dumped. Otherwise, there will be no aggregate thermal efficiency improvement.
where W PUMP is the amount of electrical energy it takes to keep the pumps running. Moreover, WPUMP is the amount of
energy it takes to keep ALL of the pumps running—and not just one or two. So realistically, W PUMP represents the total
electrical energy expended by all of the coolant pumps in the plant:
kg/s
In Chapter 19, we will derive an explicit expression for the pumping power needed to sustain a mass flow rate of m
across a pressure drop of ΔP Pa or ΔP PSI. The pumping power P for an ideal coolant pump is then given by
where m is the mass flow rate (in kg/s), ΔP is the pressure drop (in Pa), and ρ is the average density of the fluid in kg/m3.
Here, ΔP is the sum of the individual pressure drops through each loop, and the total pressure drop ΔP consists of the
sum of the pressure drops from many different components:
Sometimes the pressure drop in Equation 9.52 is also expressed in terms of another important variable called the pres-
sure head of the pump. The pressure head h (not to be confused with the specific enthalpy) is the height of a stationary
column of water that produces the value of ΔP which appears in Equation 9.52. Thus, the pressure drop in Equation 9.52
is given by ΔP = ρgh, where ρ = 1,000 kg/m3, g = 9.81 m/s2, and h is the height of the water column in m that corresponds
to this drop. Unfortunately, a reactor coolant pump is never 100% efficient, and in practice, the efficiency η of an average
coolant pump turns out to be about 85%. So the equation for the pumping power that should be used instead is
where η is the pump efficiency (which is typically about 0.85). Clearly, one can see that if the efficiency is lower, more
power will be required to pump the coolant through the plant because the value of WPUMP will be higher. In practice, it
is usually convenient to have different expressions for the pumping power for the primary and secondary loops. So it is
often better to write
where the superscripts P and S refer to the primary and secondary loops, respectively. Moreover, if the pumps have
d ifferent efficiencies, then the total pumping power W PUMP will be
WPUMP = ∑ m ∆P
n
n n ρn ηn (9.56)
where the summation must be performed over all of the pumps from n = 1 to N. In reactor work, the value of ΔPn is flow
regime dependent, and it also depends on the Reynolds number and how much the bulk temperature of the fluid departs
from the wall temperature. Finally it depends on whether the flow is single phase or two phase. For the same mass flow
rate, the pressure drop ΔPn, and hence the pump work W PUMP n, will be greater for two-phase mixtures than it will be for
single-phase liquids. Similarly, the steam turbines are not 100% efficient, and a certain amount of energy will be lost
when the turbine spins due to mechanical losses in the drive shaft. So although the turbines generate copious amounts
of electrical power, they do not generate as much as a simple enthalpy balance across their inlets and outlets would sug-
gest. Thus there are always certain frictional and mechanical losses associated with keeping the turbines running, and
the amount of work required to overcome these losses can be represented by the symbol W TURBINE_LOSS. Obviously, when
there is more than one turbine, the total amount of energy that is dissipated to keep them running is
and this expression must be added to the equation for the plant efficiency as well. When we do so, we find that the final
expression for a plant’s thermal efficiency is
and this is the expression that most reactor vendors use for their preliminary design work. The primary reason why this
expression is so important is that the coolant pumps in the primary loop of a large PWR or a BWR can consume a very
large amount of electrical power. For example in large nuclear power plants, they can consume about 6 MW of electrical
power per pump. So if a 1,000 MWE (3,000 MWT) light water reactor has three or four of these pumps in its primary loop,
then as much as 20 MW of electrical power, or about 2% of the electrical output of the plant (and about 6% of its thermal
output), is needed just to keep the pumps running! Variations in plant design can also affect the aggregate thermal efficiency,
but in the absence of any additional information, Equation 9.58 is a good equation to use to estimate the overall efficiency.
338 Reactor Thermal Cycles
Reason 1
First, the atmospheric pressure would be the lowest pressure to which the steam could be ejected. When a condenser
is used, the system becomes a closed system and the condenser pressure (to which the turbine exhausts its steam) can
be maintained at values well below atmospheric pressure. This resulting pressure difference means that additional
work can be done when the steam expands. So the turbine can do more work, and the thermal efficiency of the plant
will be higher.
Reason 2
The second reason why a nuclear power plant recycles the condensed steam is that the water must be kept pure at all
times. As a matter of fact, all nuclear power plants are equipped with water purifiers. Failure to maintain this high
level of purity would mean that the dissolved solids in the water would be left behind in the core or steam generator
when the water is vaporized. This would cause the dissolved solids to accumulate on the surface of the fuel rods.
The heat transfer coefficient would decrease, and it would be more difficult to remove heat from the core. Eventually,
the heat transfer surfaces in the core would become fouled or the recirculation tubes in the steam generator would
become plugged. This could lead to a catastrophic failure of some of the components in the primary or secondary
coolant loops. The reactor would then have to be shut down to replace these failed components. Finally, the cost of
all this additional purification equipment means that any water used in the system must be kept very pure at all times.
As a practical matter, it becomes necessary to reuse this purified water to keep the operating cost of the plant down.
Reason 3
A third reason why a nuclear power plant recycles the waste steam is that the steam can become radioactive, and this
recycling creates a natural barrier to prevent radioactive materials from being released into the environment. Any
radioactive materials that become lodged in the primary or secondary loops stay in the primary or secondary loops
for a long time. In an open-cycle power plant, there would be simply no way to prevent this from occurring. Thus,
recycling the steam actually reduces the risk that any unintended radioactive materials will be released from the plant.
Reason 4
Finally, and perhaps most importantly from an environmental perspective, the amount of steam a nuclear power plant
produces can be enormous. Even if it were not radioactive, discharging this much steam into the atmosphere would
9.21 Removing the Waste Heat from a Nuclear Power Plant 339
mean that it could severely affect the local humidity for miles around. In fact, it would turn the area immediately
surrounding the plant into a very humid place. Unfortunately, condensing the steam directly in the condenser has its
own set of problems because the waste heat still has to go somewhere. In most plants, the waste heat is discarded to
a river or the ocean. As long as the river is large enough, the water temperature will not rise beyond an acceptable
level. And clearly, the waste heat from the power plant will not cause the ocean to boil. So this is the environmental
trade-off that most nuclear power plants tend to make. It is not a perfect solution, but it is an acceptable one. Now, we
would like to quantify exactly how much heat we are talking about.
( )
∆T = 2 × 10 9 1 × 10 7 × 4,186/60 = 2 × 10 9 0.6975 × 10 9 = 2.87°C (or about 5.16°F )
Practically speaking, the actual flow rate through the condenser is limited to about 20% of the average flow rate of a
river like the Hudson. This then translates to a temperature rise of about 2.87/0.2 = 14.35°C (about 25.8°F) at the exit
of the condenser. However, this temperature rise is also limited because the river water temperature cannot exceed the
condensation temperature of the steam, which is about 37.78°C (about 100°F). Now this estimate also assumes that the
entire flow of the Hudson River is diverted through the condenser. So if we had to add 14.35°C of temperature to each
kilogram of the water, the river temperature must be less than 37.78°C − 14.35°C = 23.45°C (or about 74.17°F) to make
FIGURE 9.20 The Hudson River in upper New York State (shown above) which flows past the borough of Manhattan in New York
City. The Indian Point nuclear power plant (shown below) uses the Hudson River to absorb its waste heat.
340 Reactor Thermal Cycles
this thermal cycle work. Fortunately, the thermal cycle mathematics work out well in this case because the temperature
of the Hudson River in Upstate New York never rises above 23.45°C (or about 74.17°F) in the summer. Otherwise, the
power output of the plant would have to be reduced because we would have nowhere else for the waste heat to go! This
example presents a stunning illustration of how much waste heat a nuclear power plant can generate. For this reason,
most nuclear power plants are located close to larger bodies of water (like the Mississippi River or the ocean). In some
cases, a very large lake (like one of the Great Lakes) can also be used to dissipate this waste heat.
FIGURE 9.21 On the left, an illustration of the deviation of a nuclear power plant vapor cycle from an ideal Rankine cycle due to
irreversibilities in the plant. On the right, the effect of pump and turbine irreversibilities on the work that can be performed.
9.24 The Four Steps in an Ideal Brayton Cycle 341
TABLE 9.3
The Sources of Irreversible Losses in the Major Components
of a Commercial Nuclear Power Plant
FIGURE 9.22 George Brayton (left) and John Barber (right) who were responsible for inventing and perfecting the modern-day
Brayton cycle. (Pictures provided by Wikipedia.)
* In thermodynamics, an adiabatic process is one that occurs without a gain or loss of heat. For a process to be adiabatic, there can be
no heat transfer Q into or out of the system, and for these steps, Q = 0.
342 Reactor Thermal Cycles
FIGURE 9.23 The four steps in an ideal Brayton cycle. (See http://en.wikipedia.org/wiki/Brayton_cycle.)
these steps in an ideal Brayton cycle are assumed to be both reversible and isentropic. However, the compressor work is
also a larger fraction of the turbine work than the pump work would be in a comparable Rankine cycle. Several excellent
articles which discuss the Brayton cycle can be found on the Internet. The Brayton cycle is also applicable to gas-fired
power plants. Hence, an ideal Brayton cycle has exactly the same number of steps as an ideal Rankine cycle. Each of
these four steps can be described as follows:
Step 1:
In Step 1, energy is added to the working fluid by sending it through a compressor. In a Rankine cycle, the equivalent
of the compressor is a pump. However, because there is no phase change in a Brayton cycle, the gas temperature
constantly changes as we proceed through the cycle. The energy increase is given by ΔE = m cP (T2 − T1), and this
energy increase is the same as the compressor work WC (e.g., ΔE = WC).
Step 2:
In Step 2, the gas receives heat from an external heat source. This causes the temperature of the working fluid to rise.
In an ideal cycle, the pressure is held constant. In a practical cycle, it is not. In a nuclear power plant, the heat source
is the reactor core. The amount of heat received from the core is QC, and the total temperature change of the fluid is
ΔT = (T3 − T2) = QC/m cP.
Step 3:
In Step 3, the working fluid passes through a turbine, where its pressure falls, and its temperature is also decreased.
The kinetic energy of the fluid is converted into mechanical work or power that turns the blades of the power turbine.
In an ideal Brayton cycle, this process is assumed to be reversible. In a practical cycle, it is not. The temperature
change of the fluid is (T4 − T3), and the work done by the turbine is WT = m cP (T3 − T4). The work done by the turbine
is a positive number and it is equal to the energy loss of the fluid which is a negative number.
Step 4:
In Step 4 thermal energy is exchanged between the working fluid and what is called a “heat sink.” A heat exchanger is
used to facilitate this process. In an ideal Brayton cycle, the pressure of the working fluid is held constant. In a practical
cycle, it is not. The temperature drop across the heat exchanger is given by T4 − T1, and the heat removed by the heat
exchanger is given by QHE = m cP (T4 − T1), where m
is the mass flow rate of the gas. If we sum up the temperature losses
and the temperature gains around the loop (from steps 1 to 4), the net temperature change is 0. However, a lot of energy
is obviously exchanged in the process. Heat is exchanged between the reactor core and the coolant, and between the
turbine, the heat exchanger, and the external environment. Heat energy is also added to the system by the compressor.
However, the total energy flow is always balanced (e.g., the amount of energy input is always offset by the amount of
energy output), and this is why the working fluid always returns to its initial state (T1 and P1) at point 1. We would now
like to demonstrate how the thermal efficiency of this process can be calculated. For an ideal Brayton cycle which circu-
lates an ideal gas, the thermodynamic states in T–V and T–P space are related to each other by the following expressions:
9.24 The Four Steps in an Ideal Brayton Cycle 343
where γ (gamma) is a constant called the specific heat ratio that can be derived from the ideal gas law
γ = Cp Cv (9.60)
and the working fluid is assumed to be an ideal gas. The enthalpy for an ideal gas is also a function of the system
temperature
h = Cp T (9.61)
because the specific heat Cp is constant. When we calculate the thermal efficiency in the same way as we did for a
Rankine cycle, for example, TE = η = Net work/MWT, we arrive at the following equation for the thermal efficiency η
of an ideal Brayton cycle:
where CR is called the compression ratio. In the four-step process we have just described, the compression ratio is
given as
where the pressures P1, P2, P3, and P4 are the gas pressures at each step in Pa (or PSI). When the cycle is an ideal one,
it can further be shown that the optimum compression ratio (for the maximum work output from the cycle) is given by
( )
γ /2(γ −1)
CR OPTIMUM = T3 T1 (9.64)
where T1 and T3 are the temperatures at points 1 and 3 (the inlet and the outlet). Unfortunately, Brayton cycles are not
always ideal in practice, and there are frictional losses and compression losses as the gas is processed. Then, the actual
temperature–entropy diagram more closely resembles the one shown in Figure 9.24. To find the actual work performed
by the turbine (vs. the ideal work), we must find the actual work performed by the compressor versus the ideal work.
When we do so without regeneration, we end up with a thermal efficiency of approximately 31%. Clearly, this is not
something to get excited about. Furthermore, when we repeat the same analysis accounting for all forms of frictional
losses (including fluid friction and duct pressure losses), the thermal efficiency falls to about 25% (Ouch!). So it would
seem at first glance that a Brayton cycle is not nearly efficient as a Rankine cycle.
Fortunately, there are a number of things that can be done to increase the aggregate efficiency. We can use regenera-
tion in the same way as we did for the Rankine cycle in our previous examples. However, the results turn out to be much
more satisfying in this case. When we go through the required math (see Todreas and Kazimi (2008) to understand what
is involved), the thermal efficiency of the Brayton cycle increases to about 36%. This is in the normal efficiency range
that is reported for gas-cooled reactors in the Western world (35%–38%). The fluid states for the ideal Brayton cycle
shown in Figure 9.23 with CO2 as the working fluid are summarized in Table 9.4 (which is also adopted from Todreas
and Kazimi). Notice that both the temperature and the pressure of CO2 vary by about a factor of 3 as one moves through
the various stages in the cycle. The values of the specific heats required to compute the thermal efficiency are also shown
in Table 9.5.
344 Reactor Thermal Cycles
FIGURE 9.24 The four steps in an actual Brayton cycle with pump and turbine inefficiencies.
TABLE 9.4
State Conditions for an Ideal Brayton Cycle Implemented for a Gas-Cooled Reactor. In This Example, Carbon dioxide Is Assumed
to Be the Working Fluid
TABLE 9.5
The Specific Heats for Various Industrial Gases That Can Be Used as Coolants in Gas Reactors
FIGURE 9.25 A hybrid thermal cycle for a gas-cooled reactor. Note that similar cycles have been implemented for gas-turbine
power plants, and some very modern power plants using this cycle have achieved aggregate thermal efficiencies in excess of 60%.
cycle is shown in Figure 9.25. The Brayton thermal cycle “tops” the Rankine thermal cycle in the T– S diagram for the
combined cycle. The combined area under the two curves is always greater than that of either thermal cycle alone. Thus
hybrid thermal cycles can be extremely attractive if they are implemented properly.
FIGURE 9.26 A picture of the Superphenix in Creys-Malville, France. The Superphenix is one of the best known examples of a
pool-type LMFBR. (Picture provided by the IAEA.)
346 Reactor Thermal Cycles
On the secondary side, the inlet temperature to the steam generators is about 237°C and the outlet temperature
from the steam generators is about 487°C. Needless to say, this results in a huge improvement in the thermal efficiency
when the primary loop is operated close to these temperatures. Part of this efficiency improvement is due to the higher
volumetric power density of the core (see Chapter 5), and the rest of it is due to the superior heat retention and heat con-
ducting capacity of a metallic coolant. The properties of the liquid sodium are shown in Table 9.6. The thermodynamic
properties of the fluids that comprise the thermal cycle are shown in Table 9.7. Notice that there is still plenty of “head-
room” in most LMFBRs to increase the thermal efficiency even more because the liquid sodium that flows through the
core is still about 300°C below its boiling point. Today, almost all LMFBRs use Sodium-23 (Na-23) as their metallic
coolant. Sodium in this form is a shiny liquid metal that resembles the element mercury in many respects. A picture of
sodium in its solid and liquid forms is shown in Figure 9.27. It is soft, silver-white, and highly reactive. Its only stable iso-
tope is Na-23. At sea level, metallic sodium becomes a liquid at 98°C, which turns out to be slightly less than the boiling
point of ordinary water. Liquid sodium is also attractive as a reactor coolant because it has a high thermal conductivity
(~0.542 W/cm °K), which means that it has excellent heat transfer characteristics.
Hence a LMFBR cooled with liquid sodium can be operated at a very high volumetric power d ensity compared to
conventional thermal water reactors. Furthermore, liquid sodium has a very high boiling point (about 882°C at 1 atm),
so the reactor coolant loops can be operated at very high temperatures near atmospheric pressure w ithout causing the
sodium to boil. This also means that the reactor pressure vessel wall does not have to be very thick. In fact, it only has
to be about 20% as thick as the wall in a PWR or a BWR. Over time, some of the conservatism associated with the
design of LMFBRs will probably be reduced, and thermal efficiencies as high as 46% may become more common-place.
Commercial LWRs do not have thermal efficiencies that even remotely approach these values today. Notice that most
TABLE 9.6
The Thermophysical Properties of the Liquid Sodium Coolant in a Fast Reactor
TABLE 9.7
The Design Parameters for a Pool-Type LMFBR
FIGURE 9.27 A picture of sodium metal in its solid and liquid state. The melting point of sodium metal at atmospheric pressure is
98°C, which turns out to be less than the boiling point of ordinary tap water. (Courtesy of Wikipedia.)
FIGURE 9.28 The thermal cycles for both pool-type and loop-type LMFBRs. (Picture provided by Wikipedia.)
LMFBRs have an Intermediate Heat Exchanger or IHX to provide an additional layer of isolation between the metallic
sodium in the primary loop and the pressurized water in the secondary loop. Although these IHXs increase the overall
cost of the plant slightly, it makes it relatively easy to service one of the IHXs if a leak in the primary loop develops. There
are about 20 other LMFBRs in operation in the world today. All of them have between one and three loops (except for
the Dounreay LMFBR in the United Kingdom and the Shevchenko LMFBR in Kazakhstan, which has six). There are
also about half a dozen “pool-type” LMFBRs in operation in today. In these designs, the core is completely immersed
in a large “pool” of liquid sodium, and there is therefore no need for any additional piping. The Superphenix in Creys-
Malville, France, is a good example of a pool-type design. The thermophysical properties of the primary, the secondary,
and the tertiary loops in a three-loop LMFBR are shown in Table 9.7. The values shown in this table are industry aver-
ages, and the actual values may be higher or lower. The thermal cycles for both pool-type and loop-type LMFBRs are
shown in Figure 9.28. At this time, both implementations appear to be extremely popular.
9.27 Moving On
Now that we have concluded our discussion of reactor thermal cycles, we would like to turn our attention to the fuel rods
and the coolant. In Chapters 10 and 11, this will lead to a discussion of conductive heat transfer, and then to a discussion of
convective heat transfer. These two modes of heat transfer are the primary modes of heat transfer in nuclear power plants
today—although radiative heat transfer can also play an important role under certain conditions. Chapter 10 provides
348 Reactor Thermal Cycles
a much needed introduction to this subject matter for anyone who has not been exposed to the thermal sciences before.
Additional articles regarding conductive and convective heat transfer can also be found in the References at the end of
Chapter 10 and on the Internet. Only a rudimentary knowledge of fluid mechanics is required to begin our discussion of
convective heat transfer. We will start with single-phase heat transfer and then proceed to two-phase heat transfer. Both
forced convection and natural circulation will then be discussed.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Crowe, C. and Schwartzkopf, J., et al. Multiphase Flows with Droplets and Particles, CRC Press (2012).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M. C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, Florida (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc. New York,
NY (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi (2000).
Reactor Theory (Nuclear Parameters) DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, Florida (2014).
Exercise 9.2
A geothermal power plant in Iceland takes hot water from deep within the Earth and uses it to produce electric power.
The water enters the plant at 150°C and 500 kPa and leaves the plant at 40°C and 100 kPa. If the efficiency of the plant
350 Reactor Thermal Cycles
is 70% of the efficiency of a Carnot heat engine, what is its actual t hermal efficiency? How does this compare to the
thermal efficiency of the nuclear power plant in Exercise 9.1?
Exercise 9.3
A hydrothermal power plant takes hot water from the surface of the Gulf of Mexico at 40°C and runs it through a heat
exchanger where it is returned to the deep ocean at 10°C. The energy extracted from the water is used to generate electric
power. Assume that the energy transfer occurs using a refrigerant such as freon. By treating this plant as a Carnot heat
engine, estimate its maximum thermal efficiency if all of its internal processes are isentropic.
Exercise 9.4
Water enters the core of a PWR as a subcooled liquid with a temperature of 290°C and leaves the core as a subcooled
liquid with a temperature of 320°C. What is the increase in the specific entropy of the water if its pressure is 15.5 MPa
(2,250 PSI)? If water flows through the core at a rate of 10,000 kg/s, how much does the flow of water through the core
increase the total entropy of the world in 1 h?
Exercise 9.5
A nuclear power plant uses three reheating stages to improve its thermal efficiency. If the reheating stages are nearly
identical and the first reheating stage increases the thermal efficiency by 1.5%, by how much do the second and third
reheating stages improve its aggregate thermal efficiency? Is the use of a third reheating stage justified economically
under these conditions?
Exercise 9.6
Steam flows through a high-pressure turbine in a nuclear power plant, which can be considered to be adiabatic. Steam
enters the inlet at 10 MPa, 450°C, and 80 m/s, and steam leaves the outlet at 10 kPa, with a 92% quality, at a velocity of
50 m/s. The average mass flow rate is 12 kg/s. Determine the inlet area of the turbine in m2, its power output, and the
change in the kinetic energy of the steam. What is the mass flux at the inlet?
Exercise 9.7
A well-insulated pressure relief valve is used to throttle hot steam from 8 MPa and 500°C to 6 MPa. What is the
temperature of the steam after it passes through the valve?
Exercise 9.8
Hot water at 40°C enters a well-insulated mixing chamber with a mass flow rate of 0.5 kg/s where it is subsequently
mixed with a cold water stream at 20°C. If the desired temperature of the mixture when it leaves the chamber is 42°C,
determine the mass flow rate of the cold water stream. Assume that both streams are maintained at a pressure of
0.25 MPa.
Exercise 9.9
Gas turbines have many practical applications in power generation today. Suppose an adiabatic gas turbine in a gas-
cooled reactor expands air at 1 MPa and 500°C to 0.1 MPa and 150°C. Air enters the turbine through an opening with a
cross-sectional area of 0.2 m2 at an average velocity of 40 m/s. The air leaves the turbine through a square duct 1 m on a
side. Determine the mass flow rate of the air through the turbine and the power the turbine produces.
Exercise 9.10
Steam enters the condenser of a nuclear power plant at 20 kPa with an equilibrium quality of 95%. The mass flow rate
of the steam through the condenser is 20,000 kg/h. The steam is to be cooled by water from the Colorado River by
circulating the water through tubes within the condenser. To prevent thermal pollution, the river water is not allowed to
experience a temperature rise of more than 10°C. The steam is designed to leave the condenser as a saturated liquid with
a pressure of 20 kPa. Determine the mass flow rate of the Colorado River water through the condenser.
10
The Laws of Nuclear Heat Transfer
10.1 A n Introduction to Nuclear Heat Transfer
Heat transfer in nuclear power plants is the combination of three distinct processes called conduction, convection, and radiation. First,
it occurs in nuclear fuel rods through the process of conductive heat transfer. Secondly, it occurs between the coolant and a hotter or
colder surface by what is called convective heat transfer. Third, it occurs when radioactive particles (which may also include particles
of light) transfer their kinetic energy to the materials through which they are passing. This third form of nuclear heat transfer is called
radiative heat transfer. Normally, conductive and convective heat transfer are more important than radiative heat transfer in reactor
design, but this does not apply to all reactor components. In this chapter we would like to discuss each of these forms of heat transfer
and we would like to discuss when they become important to the operation of a nuclear power plant. Each of these forms of heat
transfer can be described by a different set of physical laws:
1. Conductive heat transfer can be described by what is called Fourier’s law of conduction
2. Convective heat transfer can be described by what is called Newton’s law of convection
3. Finally, radiative heat transfer can be described by what is called the Stefan - Boltzmann law
Each of these forms of heat transfer are compared in Figures 10.1, 10.2, 10.3, and 10.4. These laws can be applied to nuclear fuel rods,
the reactor pressure vessel, and even to radiation shields. In the sections that follow, no previous knowledge of nuclear heat transfer
and fluid flow is required although some prior exposure to the concepts of conductive and convective heat transfer is assumed.
FIGURE 10.1 Two examples of the process of convective heat transfer. The figure on the left shows the thermal convection of magna in the
Earth’s mantle. Colors closer to red are hot areas, and colors closer to blue are cold areas. A hot, less-dense lower boundary layer sends plumes of
hot material upward, and likewise, cold material from the top moves downward. The figure on the right shows the convection of heat away from a
hot, flat surface. Again, the colder temperatures are indicated in blue. (wiki.palabos.org and Wikipedia.)
351
352 The Laws of Nuclear Heat Transfer
FIGURE 10.2 Two examples of the process of radiative heat transfer. This figure on the left shows the Earth’s long-wave thermal
radiation intensity, from clouds, atmosphere, and surface. The figure on the right shows a red-hot iron object, transferring heat to the
surrounding environment primarily through thermal radiation. (Wikipedia.)
FIGURE 10.3 Two examples of conductive heat transfer where heat flows from a region of high temperature to a region of low tem-
perature. In this case, the rate of heat flow is proportional to the temperature gradient dT/dr. (www.phys.org and universetoday.com.)
FIGURE 10.4 Two examples of problems where conduction, convection, and radiative heat transfer processes occur at the same
time. (www.nc-climate.ncsu.edu and schoolworkhelper.net.)
10.2 Comparing Conduction, Convection, and Radiation 353
1.
In reactor work, conductive heat transfer refers to the transfer of thermal energy from one region of space to
another as the result of a temperature difference between the regions. Conductive heat transfer does not require
any direct transfer of mass between the regions through which the heat flows. Hence, it applies primarily to solid
materials, and it can be thought of as the molecular diffusion of heat through a stationary object (such as a nuclear
fuel rod). As we shall see shortly, it can be described by a second order differential equation called the heat con-
duction equation. The heat conduction equation (see Chapter 11) is a classical form of what is called a particle
diffusion equation.
2.
Convective heat transfer is the transfer of heat to (or from) a fluid as a result of the macroscopic motion of the
molecules of the fluid. In other words, in addition to a temperature difference, convection requires some mass
transfer to occur. Convection is only possible if the motion of the fluid results in a net transfer of heat. A moving
fluid with a temperature gradient will always transfer some heat from one part of the fluid to another. If the fluid
is in motion, this heat transfer will involve some form of convection. Sometimes convection is further subdivided
into diffusion and advection. Advection requires the bulk motion of a fluid, while diffusion does not. The diffu-
sion of thermal energy is a molecular process that is similar to the conduction of heat through solids. However,
when it applies to convection, the solid object is replaced by a liquid or a gas.
Definition of Convection
Convection = Advection + Diffusion (10.1)
3.
Radiative heat transfer is the transfer of thermal energy through space or time by the motion of nuclear particles
such as alpha rays, beta rays, and gamma rays. These nuclear particles are not necessarily indigenous to the
material medium through which they are passing. Radiative heat transfer is different than other forms of heat
transfer because it does not require a conductive material to be present. All material bodies radiate some thermal
energy (as long as their temperature is above absolute zero), and so radiative heat transfer will occur if a hot object
emits more heat than it receives, or if a cold body receives more heat than it emits. In reactor work, radiative heat
transfer can be subdivided into one class of heat transfer caused by the emission of photons and another class of
heat transfer caused by the emission of other nuclear particles (such as electrons and alpha rays). Electrons are
heavier than photons, and their range is typically more limited. Their energy deposition in a nuclear power plant
is governed by what is called an exponential attenuation law. This exponential attenuation law is manifested by
a position dependent heat source q′′′(r) that appears as a source term in the general heat conduction equation (see
Chapter 11). Terms of this type frequently appear in the calculation of temperature profiles for radiation shields.
The boundary conditions applied to the heat conduction equation then determine exactly where the temperature
peaks inside of these shields.
In nuclear power plants, electromagnetic radiation is carried through space by nuclear particles called photons, but in
nuclear fuel assemblies, it can also be carried through the fuel and the coolant by electrons, protons, neutrons, and alpha
particles. (Alpha particles are helium nuclei with their electrons removed.) Each of these forms of radiation has a different
penetrating power and photons (which are usually manifested by energetic gamma rays) can easily travel through the core
and the pressure vessel wall without being absorbed. Hence the kinetic energy they carry can be deposited far away from
the nuclear fuel rods where they originate. The wavelength of this radiation can be longer or shorter than that of visible
light. Gamma rays have much shorter wavelengths than visible light (see Figure 10.5) and microwaves (which are used to
heat food in microwave ovens) have much longer ones. The total amount of radiative energy a photon can carry is given by
where f is its frequency of vibration (in cycles per second) and h is a fundamental constant of nature called Planck’s
constant. Planck’s constant has the units of energy-seconds, and its value is 10.62606957 × 10 −34 m2 kg/s. In other words,
the faster a photon vibrates the more kinetic energy it is capable of carrying. The highest frequency photon that can
be produced in a nuclear power plant is called a gamma ray, and gamma rays have an average frequency of about 1 ×
1020 Hz (see Figure 10.5). At this frequency, their kinetic energy can be as great as 20 MeV, and at these energies, they
can be harmful to living tissue. Gamma rays are created inside the nucleus of an atom, and they can easily pass through
several layers of concrete or steel. X-rays, on the other hand, are produced by transitions between energetic electrons
in the electron cloud of atoms, and they have much lower energies than gamma rays do (see Chapter 5). They are not
354 The Laws of Nuclear Heat Transfer
FIGURE 10.5 The electromagnetic energy spectrum of light and the frequencies of gamma rays that can often be produced in a
reactor core.
produced in the nucleus itself. Steel, concrete, and lead are commonly used to shield humans from the harmful effects of
X-rays and gamma rays. Large amounts of water are also very effective at shielding reactor personnel from high-energy
neutrons, which are also produced in large numbers in the core. The properties of these nuclear particles and their inter-
action with matter is discussed in our companion books*.
FIGURE 10.6 A picture of Frenchman Joseph Fourier—the inventor of Fourier’s law of heat conduction. Joseph Fourier was a French
mathematician, physicist, and philosopher who is best known for his work on Fourier series and their applications to problems of time-
dependent heat transfer and mechanical vibrations. The Fourier transform and Fourier’s law of heat conduction are also named in his
honor, and he is also generally credited with the discovery of the greenhouse effect. (Picture of Joseph Fourier provided by Wikipedia.)
* See Nuclear Engineering Fundamentals: A Practical Perspective, by Robert E. Masterson, CRC Press (2017) and An Introduction
to Nuclear Reactor Physics, by Robert E. Masterson, CRC Press (2017).
Both of these books are available directly from CRC Press and they can also be purchased on Amazon.com.
10.4 The Assumptions Used in Fourier’s Law of Conduction 355
Fourier’s law states that the rate of heat transfer q through a material object is equal to the negative gradient of the
temperature profile through the object, multiplied by the area A through which the heat is flowing. Fourier’s law of
conduction has two basic forms: a differential form and an integral form. The differential form of Fourier’s law is
where ∇ is the gradient operator. Here, q″ is the heat flowing through a unit area of the material, and it is defined by
q″ = q/A. In some textbooks, q″ is referred to as the heat flux, and in reactor work, it is referred to as the nuclear heat
flux. In the SI system, it has the units of kW/m2. However, in nuclear science and engineering, it is also expressed in W/
cm2—particularly when it pertains to nuclear fuel rods and radiation shields. The constant of proportionality between
the heat flux and negative temperature gradient is called the thermal conductivity k. The thermal conductivity is mate-
rial dependent and it has the units of W/m-°K. In nuclear applications, the thermal conductivity is also expressed in
W/cm-°C. Sometimes the degree symbol (°) is suppressed when quoting the temperature in SI units, but we will not
adhere to this convention here. In many applications the thermal conductivity is constant, but in nuclear applications
involving uranium or plutonium fuel, this is not always the case. Fourier’s law in integral form is
∂q/ ∂ t = − k
∫ ∇Τ dA (10.5)
S
Here the temperature gradient ∇T is always normal (i.e., perpendicular) to the surface S through which the heat
flows, and Fourier’s law can also be used to derive the general heat conduction equation, which we will discuss in
Chapter 11.
Assumptions:
☉☉ The calculation of the heat flux takes place under steady-state conditions.
☉☉ The temperature gradient is constant, and the temperature profile is linear.
☉☉ The heat flow is unidirectional.
☉☉ There are no internal heat sources in the material through which the heat is flowing.
☉☉ The material through which the heat is flowing is homogeneous and isotropic.
☉☉ The value of the thermal conductivity k is the same in all directions.
In anisotropic materials, the thermal conductivity typically varies with direction. In other words, the value of k can be
different in the x direction than it is in the y or z directions. Mathematically, we can express this fact by writing the
thermal conductivity as a tensor. That is,
k xx k xy k xz
k ij = k yx k yy k yz (10.6)
k zx k zy k zz
356 The Laws of Nuclear Heat Transfer
where the subscripts x, y, and z refer to the thermal conductivity along the x, y, and z directions. In nonuniform materials,
the thermal conductivity can also vary with spatial location and temperature. In most applications of conductive heat
transfer in reactor design, the thermal conductivity is assumed to be independent of position, or isotropic. However, it
can still be temperature dependent - as Todreas and Kazimi (2008) clearly indicates.
Sometimes the temperature of the surrounding environment is also called the ambient temperature. Newton’s law of
cooling was originally proposed by Isaac Newton in the early 1700s, and it can also be used to predict how the tempera-
ture of the coffee in your coffee cup behaves as it is heated or cooled (see Figure 10.7). Newton’s law of cooling can also
be written in integral form as
FIGURE 10.7 Newton’s law of cooling applies to a hot object which is attempting to cool off. Over a period of time, the
temperature of the object drops exponentially and eventually, and the object comes into thermal equilibrium with its environment.
(Picture of Newton provided by Wikipedia.)
10.6 The Thermal Conductivity and Wiedemann–Franz Law 357
where T(t) is the temperature of the object at time t, Te is the ambient temperature of the environment (which is normally
assumed to be constant), T0 is the initial temperature of the object, and α is a constant of proportionality that depends on
the material properties of the object. We can also see from Equation 10.8 that the temperature of an object falls exponen-
tially when it is exposed to an environment that has a lower ambient temperature Te. A curve illustrating Newton’s law
of cooling is shown on the right-hand side of Figure 10.7. Notice that the temperature of the object eventually declines to
the same temperature as its environment. However, the rate at which this occurs depends on the value of α. In conductive
heat transfer value of α is called the thermal diffusivity of the material.
Hence the thermal diffusivity can be thought of as the ratio of the heat conducted by a material to the heat it stores. The
inverse of the thermal diffusivity is then the time constant of the material, and the time constant is represented by the
symbol λ. Hence, the time constant is given by λ= 1/α where λ = ρc/k.
Newton’s law of cooling assumes that the environment in which the object is placed is large enough that its temperature
does not change as the object cools down. Fortunately, this turns out to be a good assumption most of the time. More
information regarding Newton’s law can be found at
http://www.ugrad.math.ubc.ca/coursedoc/math100/notes/diffeqs/cool.html
Now let us turn our attention to how the thermal conductivity that appears in Equation 10.10 is defined.
FIGURE 10.8 The process of heat conduction in solids is based on the temperature difference between a region of high temperature
and a region of low temperature. Under these conditions, the amount of heat flow q is proportional to the thermal conductivity k.
FIGURE 10.9 The thermal conductivity of ordinary materials can span about five orders of magnitude—from high values for good
conductors to low values for good insulators. Note: Due to space limitations, ranges are not always drawn to scale.
where T is the temperature of the material. The proportionality of k/σ with temperature was first discovered by Ludvig
Lorenz in 1872. Hence, the Wiedemann–Franz law exists because electrons in metals can carry both heat and electric
charge. In this law, k is the thermal conductivity, σ is the electrical conductivity, and L is an empirical constant called
the Lorenz number. The thermal conductivity can also depend on the following factors:
☉☉ The moisture content
☉☉ The material structure
☉☉ The pressure and the density
The thermal conductivity for most common materials (including some nuclear ones) is shown in Table 10.1. Metals have
a thermal conductivity between 200 and 400 W/m °K, while alloys of stainless steel have a thermal conductivity between
20 and 120 W/m-°K. Common liquids like water have values between 0.2 and 1.0 W/m-°K, and finally, building materi-
als, insulators, and most industrial gases have values between 0.004 and 3.0 W/m-°K. In many materials, the thermal
10.6 The Thermal Conductivity and Wiedemann–Franz Law 359
TABLE 10.1
The Thermal Conductivities of Some Common Materials at Standard Temperature and Pressure
Material Thermal Conductivity (W/m °K) Material Thermal Conductivity (W/m °K)
Silver 400–420 Asbestos (sheets) 0.16–0.18
Copper 380–390 Ash 0.11–0.13
Aluminum 220–230 Cork or felt 0.05–0.10
Steel alloys 20–120 Saw dust 0.06–0.08
Iron 55–65 Glass wool 0.02–0.04
Concrete 1.10–1.30 Water (liquid) 0.20–1.0
Glass (typical) 0.70–0.80 Freon (air conditioners) ~ 0.008
Note: Most of the values shown are also temperature dependent.
FIGURE 10.10 The temperature profile in a planar wall depends upon the behavior of the thermal conductivity. If the thermal
conductivity does not depend on temperature, the temperature profile will be a linear one under steady-state conditions. However,
if the value of β is greater than unity or less than unity, then the temperature profile will deviate from a linear one. The same thing
happens in nuclear fuel rods under certain conditions.
conductivity is also temperature dependent, and the dependence of the thermal conductivity k on the temperature is
basically a linear one. We can express this fact by writing
where ko is the thermal conductivity at 0°C (273°K), and β is the temperature coefficient of the thermal conductivity.
When defined in this way, β has the units of inverse temperature. β is usually a positive number for nonmetals and
insulating materials, and it is a negative number for metallic conductors (with aluminum and nonferrous alloys being
the most notable exceptions). In solids and liquids, the thermal conductivity is only weakly dependent on the ambient
pressure P, and in gases, the value of k does not depend on the pressure (at least when it is close to atmospheric pressure).
The temperature profile in solids (see Figure 10.10) then depends on the value of β.
TABLE 10.2
The Specific Heats of Some Common Materials at 20°C
FIGURE 10.11 The heat capacities of UO2 and PuO2—two materials that are commonly used in nuclear fuel rods. (Adapted from a
study performed by the Oak Ridge National Laboratory. http://katedra-reaktoru.cz/dokumenty/TER/109264.pdf.)
FIGURE 10.12 A picture of the heat tiles on the U.S. Space Shuttle. (Courtesy of NASA.)
Here T WALL is the temperature at the outer surface of the object, which is also called the temperature of the wall and
TFLUID is the “average” temperature of the fluid passing over the wall. The average temperature of the fluid is also called
its bulk temperature. In Equation 10.15, q is the heat flow rate (in W/s or kW/s), A is the cross-sectional area over which
this flow occurs (in cm2), and h is a constant of proportionality called the convective heat transfer coefficient. In the SI
system of units, the convective heat transfer coefficient is expressed in W/m-°K, and in most nuclear applications, it is
also expressed in W/cm-°C. In other words, the value of h depends upon both the thermal conductivity of the fluid and
the speed that the fluid is flowing past the wall. In Chapter 20, we will find that we can express h as the product of two
additional parameters called the Reynolds number Re and the Prandtl number Pr. The Reynolds number is a function of
the way we choose to operate a reactor, and the Prandtl number is a function of the type of coolant the reactor uses. In
other words, we can express the convective heat transfer coefficient as
where k is the thermal conductivity of the fluid, the Reynolds number Re is a function of the average velocity of fluid
V, and the Prandtl number Pr is a function of its material properties (such as its heat capacity c and internal viscosity
μ). There are other dimensionless numbers that appear in the study of reactor thermal-hydraulics (a complete list is
provided in Appendix D). For example, one of these numbers is the Stanton number St. The Stanton number appears in
the Reynolds analogy for heat and momentum transfer (see Chapter 20) and also appears in some correlations for the
friction factors for turbulent flow. The Stanton number is defined as the Nusselt number Nu divided by the product of the
Reynolds number and the Prandtl number. Like the Reynolds number and the Prandtl number, it is also a dimensionless
number and it is defined by the following equation:
The Stanton number is the ratio of the heat transferred to a reactor coolant to its specific heat (i.e., its heat capacity).
It was named for Thomas Stanton (1865–1931), and it is sometimes used to characterize nuclear heat transfer in forced
convective flows. However, in forced convective heat transfer in reactor coolant channels, the Reynolds number and the
Prandtl number are more commonly used (see Chapter 20).
where C is an empirical constant that depends on the geometry of the flow channel. The value of C depends on whether
the lattice is square or hexagonal, and it also depends on what the pitch-to-diameter ratio (P/D) of the fuel rods are.
Finally the value of h depends on whether the flow is laminar or turbulent and whether the flow is single phase or two
phase. Now let us make some additional observations about the value of h.
Observation 1:
In general, turbulent flows always have higher heat transfer coefficients than laminar flows. This is because the
fluid is more thoroughly mixed when the flow is turbulent, and this helps to enhance the heat transfer rate at the wall
surface.
Observation 2:
As long as the liquid film on the wall surface does not dry out, two-phase flow will generally have higher heat trans-
fer coefficients than single-phase flow. Again, bubbles near the wall surface tend to enhance the mixing process, and
10.10 Radiative Heat Transfer and the Greenhouse Effect 363
TABLE 10.3
The Convective Heat Transfer Coefficients for Some Common Fluids and Reactor Coolants
Typical Heat Transfer Coefficients for Various Reactor Coolants (W/m2 °K)
Fluid Natural Convection Forced Convection (in Pipes) Two-Phase Mixture (Flow Boiling) Condensation
Gases 5–30 5–600 — —
Water 50–1,000 250–12,000 2,500–50,000 5,000–100,000
Sodium — 2,500–25,000 — —
Source: Todreas and Kazimi (2008).
this causes the heat transfer coefficient to increase as long as the bubbles do not aggregate into much larger ones.
Table 10.3 shows the values of the convective heat transfer coefficient for some common fluids including as water and
air. The value of h for water is much greater than it is for air, and boiling this water causes its heat transfer coefficient
to increase even further. Table 10.3 also shows that the best way to remove heat from a reactor core is to cool it with
water and that we can remove 5–10 times more heat when the flow is turbulent than when the flow is laminar. Finally,
as long as the boiling does not become too severe, it is actually better to use boiling water instead of liquid water
because the heat transfer coefficient is about an order of magnitude higher. Hence, from the perspective of reactor
heat removal, it is about 100 times more efficient to make the water boil and to make the flow turbulent than it is to
use liquid water and keep the flow laminar.
For this reason and others, most commercial power reactors are designed to take advantage of at least some localized
boiling near the top of the core. In pressurized water reactors (PWRs), this boiling is called nucleate boiling, and
in boiling water reactors (BWRs), it is called bulk boiling. In American-made BWRs, bulk boiling begins about 1
meter from the bottom of the core, and by the time the top of the core is reached, the equilibrium quality can exceed
15%. (In legacy designs, this value may be lower). Finally, a large Reynolds number almost always results in a greater
heat transfer rate than a small Reynolds number. The Reynolds number in a modern PWR fuel assembly is between
100,000 and 600,000 when the flow is forced. Of course, Reynolds numbers of this size can only be obtained when
the reactor cooling pumps are running. If the pumps are turned off for any reason, then the reactor core may shut down
rapidly or scram. Under these conditions, the Reynolds number can easily fall below 10,000, and depending on the
design of the core, it can even fall below 2,000. Then the flow transitions itself from a state of forced convection to a
state of natural circulation and the convective heat transfer coefficient falls by a factor of 20 or more. This reduction in
the heat transfer coefficient may cause the core to overheat, and in some cases, it may even cause the fuel rods to fail.
For this reason, modern power reactors are designed with very elaborate safety systems to keep the coolant flowing
at all times. We would now like to discuss how nuclear radiation produced by the fuel rods affects the temperature of
the core. To do this, we would like to discuss a related problem in radiative heat transfer that has become known as the
“greenhouse effect”.
produced primarily by burning fossil fuels and from other industrial activities such as cement production and the cutting
down of the rain forests in the tropics. Measurements of the CO2 level in the upper atmosphere taken from the Mauna
Loa observatory in the state of Hawaii (see Figure 10.13) have shown that the concentrations of CO2 have increased from
about 313 PPM in 1960 to about 389 PPM in 2010. These levels exceed the levels that have been extracted from ice cores
during previous periods of climate change. They suggest that the average temperature of the Earth increases in almost
direct proportion to the amount of carbon dioxide in the Earth’s atmosphere. Hence, nuclear power plants may actually
help to reverse this trend because they produce no greenhouse gases while they generate electric power. Physically, the
greenhouse effect is an example of what is called radiative heat transfer (see Figure 10.14 and Section 10.11). Radiative
heat transfer is caused when wavelengths of visible light from the sun pass through a transparent medium such as the
Earth’s atmosphere and are absorbed, but longer wavelengths of the infrared reradiation from the heated objects are
unable to pass through the atmosphere and are reflected back toward the Earth. The trapping of this long wavelength
light leads to a general heating of the Earth’s surface and it is believed that this effect is responsible for greenhouse effect
FIGURE 10.13 The average CO2 levels in the Earth’s upper atmosphere on the top of Mauna Loa in the state of Hawaii at
approximately 13,000 ft or 3,800 m above sea level. (Data provided by Wikipedia.)
FIGURE 10.14 Two illustrations of the greenhouse effect and its effects on the Earth’s climate. Note that nuclear power plants do
not emit greenhouse gases and do not contribute directly to the warming of the planet as a whole. (Courtesy of Wikipedia.)
10.11 The Stefan–Boltzmann Law 365
on Venus, which caused it to become a dead planet. A common example of this effect is the heating of an automobile
when sunlight passes through the windshield and the excess heat that it generates cannot be dissipated by the convection
of the air in the passenger compartment alone. In the Earth’s atmosphere, carbon dioxide gas strongly absorbs infrared
radiation and it does not allow it to escape into space. A number of interesting discussions of the greenhouse effect are
available on the Internet. Two of the more interesting articles can be found at
http://en.wikipedia.org/wiki/Greenhouse_effect
http://hyperphysics.phy-astr.gsu.edu/hbase/thermo/grnhse.html
However, since plants on the Earth’s surface also absorb carbon dioxide and emit large amounts of oxygen, biological
feedback loops may actually reduce the amount of global warming that is predicted to occur without their contributions.
Finally, some scientists also believe that global warming is partially due to natural causes and that the Earth is between
two ice ages—the last of which occurred about 10,000 years ago. At the end of this ice age, the receding glaciers in
North America created the Great Lakes and changed the course of the Columbia River in Eastern Washington State to
create the Grand Coulee, which can be seen in Figure 10.15. At the base of the Grand Coulee, they also created one of
the largest waterfalls in the history of the Earth. The remains of this gigantic waterfall (which is shown on the left-hand
side of Figure 10.15) can still be seen in Eastern Washington today at the Dry Falls National Monument. This waterfall
was approximately 1,000 times larger than Niagara Falls is today, and its floodwaters created one of the greatest floods
in the history of the Earth. More information about Dry Falls, the Grand Coulee, and this great flood can be found at
https://en.wikipedia.org/wiki/Grand_Coulee
https://en.wikipedia.org/wiki/Dry_Falls
Other articles can be found on the Internet that show how climate change has occurred over geologic periods of time.
FIGURE 10.15 On the right: a picture of the Grand Coulee—one of the great glacial canyons of the world which was created at the
end of the last ice age by running water. On the left: the remains of the waterfall at the base of the Grand Coulee, which is known as
Dry Falls today. At its peak, the flow of water over Dry Falls was over 10 times greater than the combined flow of all of the rivers of
the world and over 1,000 times that of modern-day Niagara Falls. (Pictures of Dry Falls provided by Dr. Bob Masterson and pictures
of the Grand Coulee provided by Wikipedia.)
366 The Laws of Nuclear Heat Transfer
photons in the form of electromagnetic waves. This photonic energy is concentrated primarily in the infrared region of
the electromagnetic energy spectrum (see Figure 10.5) at frequencies between 0.003 × 1014 and 4 × 1014 Hz. In this case,
the infrared energy is transmitted in the form of heat. These photons are similar to the microwaves that are used to cook
food in microwaves ovens. However, in a reactor core, they tend to “cook” the materials in the core. The relationship
between the absolute temperature of these materials and the amount of thermal energy they radiate is governed by an
empirical relationship called the Stefan–Boltzmann law. This law states that
where
Q is the heat radiated by the object (in W).
ε is the emissivity of the object (normally 1 for an ideal radiator).
σ is Stefan’s constant, which has a value of 5.6703 × 10 −8 W/m2-°K4.
A is the radiating area (m2).
T is the temperature of the radiator (in °K).
Ta is the ambient temperature of the environment (in °K) in which the radiator resides.
In other words, the Stefan–Boltzmann law predicts that the amount of thermal energy that is radiated to other parts
of the core is proportional to the absolute temperature to the fourth power, and if this temperature is increased by a
factor of 2, while the temperature of the environment remains constant, the heat energy that is radiated by a hot object
increases by a factor of 16. Equation 10.19 also implies that a good absorber of electromagnetic radiation will also be a
good emitter. For example, blackbody radiation is emitted by dark objects that absorb virtually all of the radiation that
is incident upon them, and they then reradiate this radiation with an energy distribution that follows a blackbody curve.
An example of a blackbody curve is shown in Figure 10.16. Notice that the emitted energy spectrum of the blackbody
depends on only its absolute temperature. The prediction of the shape of this curve was one of the great triumphs of
modern quantum theory because it implied that more modes of photonic energy transfer were possible as the tempera-
ture of a blackbody increased. More information about this important equation can be found at the following URLs:
http://hyperphysics.phy-astr.gsu.edu/hbase/thermo/stefan.html#c2
http://en.wikipedia.org/wiki/Black-body_radiation
When a reactor is operating normally, about 10% of the thermal energy produced in the core can be attributed to nuclear
r adiation (in the form of alpha, beta, and gamma rays) that originate in the fuel rods. The remaining heat is produced by
FIGURE 10.16 A blackbody curve for thermal radiation emitted from a hot object.
10.12 ENERGY DEPOSITION DUE TO RADIATION IN A REACTOR CORE 367
fission neutrons (usually 2 to 4 per fission) and two fission products of almost equal mass that are also produced by the process
of nuclear fission. The creation rates for these particles and their kinetic energies are discussed in our companion books*.
Q(r) = Q o e − λ r r (10.20)
FIGURE 10.17 A cross-sectional view of a PWR core with a pressure vessel and the water gap between the fuel assemblies. Note
that the enrichments of the fuel assemblies (when they are initially loaded into the core) are shown in different colors.
* See Nuclear Engineering Fundamentals: A Practical Perspective, by Robert E. Masterson, CRC Press (2017) and An Introduction
to Nuclear Reactor Physics, by Robert E. Masterson, CRC Press (2017).
† See, for example, our companion book An Introduction to Nuclear Reactor Physics, by Robert Masterson, for more information
about the relationship between the neutron density and the neutron flux.
368 The Laws of Nuclear Heat Transfer
While a reactor is operating normally, about 5% of the energy produced in the core is distributed in this way. The remaining
95% is produced directly from the fission of uranium and plutonium, and most of this is distributed locally by the fission
products that are produced when the uranium and plutonium nuclei split apart. These fission products are produced with
the distribution of masses shown in Figure 10.18. Their masses are measured in atomic mass units or AMUs. The kinetic
energies of the fission products are then
Here, the index 1 refers to the lighter fission product and the index 2 refers to the heavier fission product. In a fission pro-
cess that releases 200 MeV of kinetic energy per fission, about 170 MeV is carried off by the fission products themselves.
(The remaining is carried off by the fission neutrons, the gamma rays, and other nuclear particles.) Because of their
relative mass difference, the lighter fission product receives a kinetic energy of about 100 MeV and the heavier fission
product receives a kinetic energy of about 70 MeV. As soon as they leave the nucleus, they rip through the surround-
ing materials in their way, and they take some of the electrons from the parent nucleus with them. Then, over a period
of time, they pick up additional electrons, but they never become electrically neutral atoms again. Thus, more kinetic
energy is always carried away by the lighter fission product than by the heavier one. On average, between two and three
fission neutrons are produced per fission. The number of neutrons released per fission is given the Greek symbol ν (nu).
These fission neutrons have longer mean free paths than the fission products do (see Table 10.4), and the kinetic energy
they carry can be distributed up to a meter away from their initial point of origin.
FIGURE 10.18 The distribution of fission products produced from the thermal fission of U-235 as a function of their initial kinetic
energy E.
TABLE 10.4
The Range of Fission Products in Different Nuclear Materials
Then, the heat deposited by fission neutrons is distributed through the core according to the equation
Q(r) = Q o e − ∑ r (10.22)
where Σ is called the total material cross section. (This cross section was first introduced to the reader in Chapter 5.)
In Chapter 5 we also learned that the total cross section of any material is energy dependent, and that its value typically
ranges from 1.0 to 0.01 cm−1. Its value becomes larger as the kinetic energy of the fission neutrons becomes lower. The
value of Σ also depends on the materials through which the fission neutrons are traveling. When the fission neutrons are
traveling through a large number of uniformly mixed materials, the value of Σ becomes
∑ = ∑1 + ∑1 + ∑1 + + ∑ N (10.23)
where N is the total number of materials present. Thus, the total energy deposition of the fission neutrons becomes
Q(r) = Q o e ∑1 + Q o e ∑ 2 + Q o e ∑ 3 + Q o e ∑ 4 + + Q o e ∑ N (10.24)
− r − r − r − r − r
when there are N materials in the core. Finally, it can be shown that the inverse of Σ (or 1/Σ for scattering reactions) is
equivalent to another important parameter called the neutron diffusion coefficient D. The diffusion coefficient has the
units of length (cm), and it always appears in one form or another in the neutron diffusion equation, which can be used
to predict the distribution of the neutrons in a reactor core.
Equation 10.25 is the foundation of modern reactor physics, and it can be used to determine the distribution of fission
neutrons anywhere in the core, including where their concentrations (and densities) are highest and where their concen-
trations and densities are lowest. The value of D is normally expressed in centimeters rather than meters, and physically,
it represents how far an average neutron travels before it interacts with an atomic nucleus. Not surprisingly, the diffusion
coefficient is also energy dependent, and its value is about 10 times larger for fast neutrons than it is for thermal neu-
trons. The neutron flux ϕ that appears in Equation 10.25 can be thought of as the number of neutrons passing through
a control surface having a cross-sectional area of 1 cm2 in 1 s. Thus, the units of the neutron flux are neutrons/cm2 s. In
commercial power reactors, ϕ has a value of about 1 × 1015 neutrons/cm2 s. The term “D∇ 2ϕ” that appears in the neutron
diffusion equation represents the local neutron leakage rate, and the term “S” represents an external source of neutrons
of strength neutrons/cm3 s that is sometimes added to the core. Normally, this term is independent of the neutron density
because the production of fission neutrons is governed by the term “νΣf ϕ”. Again, the Greek symbol ν refers to the
number of neutrons produced per fission (which is usually between two and four). Now, let us turn our attention to the
heat deposited in the core and in nearby radiation shields by gamma rays.
q ′′′(d) = q o e −λ d d (10.26)
where d is the distance from the point where the gamma ray originates and λ is a constant that is material dependent.
Sometimes λ is called the attenuation coefficient of the material, and it has the units of inverse length. Usually heavier
370 The Laws of Nuclear Heat Transfer
FIGURE 10.19 A radiation shield with an exponentially varying heat source. The shield is surrounded by air or water on each side.
The front surface corresponds to location 1 and the rear surface corresponds to location 2.
materials have larger values of λ than lighter materials do. The exact reasons for this are discussed in our companion
book.* When gamma rays encounter the pressure vessel wall or a radiation shield in close proximity to it, they deposit
their thermal energy according to Equation 10.26. For most radiation shields, this results in the spatially dependent heat
generation rate shown in Figure 10.19. Notice that the shield is not necessarily hottest where the gamma ray concentra-
tion is the highest. In fact, gamma rays usually deposit their energy in radiation shields in such a way that the tempera-
ture near the center of the shield is higher than the temperature at the edges. The actual temperature profiles then depend
on the boundary conditions that are used. In conductive heat transfer, these boundary conditions are implemented by
specifying the exterior surface temperatures at locations 1 and 2 (see Figure 10.19).
* See, for example, Nuclear Engineering Fundamentals: A Practical Perspective, by R.E. Masterson, CRC Press, Boca Raton, FL (2017).
10.15 Cooling Spent Nuclear Fuel 371
TABLE 10.5
The Percentage of Recoverable and Unrecoverable Energy Produced by the Fission Process in a Thermal Water Reactor
Reactor Component(s) Total Energy Deposited (%) Recoverable Energy Deposited (%)
Reactor fuel rods (including the cladding) 91.4 96.2
Moderator and coolant 2.5 2.6
Other core structures 1.1 1.2
Neutrinos 5.0 None
Total 100% Total 100%
Source: Todreas and Kazimi (2008).
that the spent fuel rods no longer have to be cooled by water and they can be cooled by air instead. This reduction in
their thermal output means that they can be placed in large outdoor storage casks such as the ones shown in Figure 10.20
or they can be shipped to a reprocessing facility to be recycled. Under these conditions there is still some radiative
heat transfer, but the temperature at the surface of the fuel storage casks never rises much above the temperature of the
surrounding air. The average surface temperature of a spent fuel storage cask is about 30°C. In other words, the cask
is cooled by allowing it to come into direct physical contact with the surrounding air. A spent fuel storage cask on its
way to a disposal site is shown in Figure 10.21. The heat transfer at the surface of the cask is due primarily to natural
convection, while inside the cask it is due primarily to conduction and radiation. The radiation remains in the fuel and
FIGURE 10.20 A picture showing two ways of storing nuclear waste in the United States. The site on the left is a low-level waste
storage pit in the state of Nevada, and the site on the right is a dry cask storage facility at the Connecticut Yankee Nuclear Power
Plant. Dry cask storage facilities are used for storing spent nuclear fuel rods that are normally considered to be high-level nuclear
wastes. (Pictures provided by the U.S. Department of Energy.)
FIGURE 10.21 On the left, a truck transporting high-level nuclear waste to a waste storage plant. On the right, a thermal image of some
spent fuel storage casks being transported on a railroad car in the United States. (Pictures provided by the U.S. Department of Energy.)
372 The Laws of Nuclear Heat Transfer
does not escape from the cask. However, it interacts with the walls of the cask and heats them over time. This heat is
then transferred from the walls of the cask to the air by natural convection. The heat transfer rate is proportional to the
surface area of the cask, and it is also proportional to the temperature difference between the surface of the cask and the
surrounding air. When a cask is ready to be transported to a disposal site, this temperature difference is usually about
10°C; that is, TWALL − TAIR = 10°C.
10.16 T
hermal Expansion and Temperature-Induced
Stresses in Reactor Components
When fuel rods and other reactor components experience severe radial or axial temperature gradients, they can expand
or contract at different rates, and this nonuniform thermal expansion can sometimes give rise to mechanical stresses
within the fuel and along the surface of the rods. These stresses are usually much greater in the radial direction than they
are in the axial direction. Thus, it is possible that some of these rods can fail from mechanically induced stresses if these
temperature gradients are not properly controlled. These temperature-induced stresses can sometimes translate into
physical deformities in the fuel rods themselves. This can cause the fuel rods to fail, and once this happens, they can fail
at virtually any location in the core. Our companion book discusses the actual failure rates that are observed in nuclear
power plants today*. These failure rates vary slightly from PWRs to BWRs, and both have tended to decline as improve-
ments have been made to the design of the fuel assemblies.* The number of rod failures in commercial power reactors
has declined in recent years due to better engineering practices and thermal-physical optimization. The percentage of
the rods that fail today averages between about 1 in 100,000 and 1 in 1,000,000. Again, these statistics are presented
for both PWRs and BWRs in our companion book*. Finally, tube failure rates in PWR steam generators are much lower
than they used to be, but they can still occur due to unwanted thermal stresses, crud buildup, and unnecessary vibration
of the tube nest. Some of these issues are discussed in Chapter 8.
Q = − kA ∆T ∆x (10.27)
where ΔT/Δx = dT/dx is the temperature gradient in the direction of the heat flow Q. Ohm’s law, which is equivalent to
Fourier’s law in the electrical world, states that
I = ∆V/R (10.28)
* See Nuclear Engineering Fundamentals: A Practical Perspective, by R.E. Masterson, CRC Press (2017).
10.19 The Thermal Resistance and Its Applications 373
where I is the current flow in response to a voltage difference between two points of ΔV, and R is the electrical resistance.
By comparing these two equations, we see that
This analogy implies that the thermal resistance to the flow of heat can be written as R = Δx/kA. Thus, in one dimension
where Δx is the distance over which the temperature difference ΔT exists. A more general definition of the thermal
resistance is then
where Q is the total heat flow rate. Hence, a large thermal resistance R results in a smaller heat flow than a small thermal
resistance for the same relative temperature difference ΔT. In the SI system of units, the units for thermal resistance
are °K/W, and in the Celsius system, they are °C/W.
Q = (1/R) ( T1 − T2 ) (10.32)
Q = hA ( TWALL − T∞ ) (10.33)
374 The Laws of Nuclear Heat Transfer
FIGURE 10.22 Under steady-state conditions, the temperature profile in a planar wall is a straight line and the heat flow q is
analogous to the current flow I in an electrical circuit.
FIGURE 10.23 The thermal resistance for the steady-state flow of heat through a planar wall requires the presence of a convective
resistance on each side of the wall. Away from the surface of the wall, the fluid reaches an asymptotic or bulk fluid temperature T∞.
where T WALL is the wall temperature at the pellet’s surface. If this equation is written as
Notice that when h becomes large, the convective resistance at the surface of the pellet becomes small and R → 0. In
fact, for very small values of R, TWALL ≈ T∞ and the temperature difference between the pellet and the coolant becomes
10.20 The Thermal Resistance of an Object in a Radiation Field 375
negligible. In reactor fuel assemblies, this happens all of the time because the value of h can be very large. In fact, it is
not unusual for the temperature at the surface of the pellet to be only a few °C above the average coolant temperature.
where
P or Q is the net power radiated by the object (in W)
ε is the emissivity of the object (which is equal to 1.0 for an ideal radiator).
σ is Stefan’s constant, which has a value of 5.6703 × 10 −8 W/m2 °K4.
A is the radiating area (in m2).
T is the absolute temperature of the radiator (in °K).
Te is the absolute temperature of the environment (in °K).
In other words, when a significant difference exists between the absolute temperature of an object and its environment,
photons will radiate from the object and they will carry thermal energy from a region of high temperature to a region
of low temperature. Normally these regions are not in direct contact with each other. The heat flow rate can then be
written as
( )
Q rad = εσA T 4 − Te 4 = h rad A ( T − Te ) (10.36)
where
( )
h rad = εσ T 2 + Te 2 ( T + Te ) (10.37)
FIGURE 10.24 In reactors, surface heat transfer from a solid object can involve both convection and radiation. In the figure above,
both processes occur simultaneously and the total heat flow is the sum of the individual heat flows.
376 The Laws of Nuclear Heat Transfer
Now, suppose that we would like to combine the processes of conduction and convection together to predict the flow
of heat through a planar wall of thickness L, area A, and thermal conductivity k. The wall is exposed to moving fluids
on each side with heat transfer coefficients h1 and h2, and bulk fluid temperatures T1∞ and T2∞. Figure 10.23 shows the
orientation of the wall. The thermal resistances of the fluid-wall combination are Rconv1, RWALL, and Rconv2, respectively.
If T1∞ > T2∞, the temperature profiles will resemble those shown in Figure 10.23. Notice that the thermal resistance can
be written as the sum of three separate terms
The equation relating the temperature difference to the heat flow rate is then
where
q = h1 A ( T∞1 − T1 ) = kA ( T1 − T2 ) L = h 2 A ( T2 − T2 ∞ ) (10.42)
or alternatively,
( ) ( ) ( )
q = ( T∞1 − T1 ) 1 h1A = ( T1 − T2 ) L kA = ( T2 − T2 ∞ ) 1 h 2 A (10.43)
( ) ( ( )) ( )
RTOTAL = Rconv 1 + RWALL + Rconv 2 = 1 h1A + 1 L kA + 1 h 2 A (10.46)
This result can be generalized to any steady-state heat conduction problem when the flow of heat is serial. For serial
heat conduction, the equivalent thermal resistance can be expressed as the sum of the thermal resistances for each of
the individual regions:
where N is the number of regions. Equation 10.47 can then be used to find the thermal resistance for any serial heat con-
duction problem. Example 10.4 illustrates how to use this equation when the heat flows through multiple material regions.
10.22 Composite Thermal Resistances 377
( ) ( ( )) ( ) (
RTOTAL = Rconv 1 + RWALL 1 + RWALL 2 + Rconv 2 = 1 h1A + 1 L1 k1A + 1 L 2 k 2 A + 1 h 2 A (10.49) )
FIGURE 10.25 In a multilayer planar wall with two material regions, the thermal resistance consists of four separate terms: two
for the regions inside of the wall and two for the external surfaces. The total resistance of the wall is the sum of the four thermal
resistances. Cylindrical and spherical objects can be treated in a similar way—except that the thermal resistances are different.
378 The Laws of Nuclear Heat Transfer
In other words, when heat flows along the same path and through multiple regions having different values for k and L,
the total thermal resistance RTOTAL becomes the sum of the individual thermal resistances along the direction of flow.
Hence, a two-layer object can be analyzed in the same way as a homogeneous object except that an additional thermal
resistance must be added for each layer. This result can be extended to planar objects having more than two layers by
simply adding more thermal resistances. This approach can be applied any object where the heat flow is constant and
there is no time-varying internal heat generation. When there is time-dependent internal heat generation, a slightly dif-
ferent approach must be used. This approach will be discussed in Chapter 12.
h c = 1 Rc (10.50)
This expression can then be used with Newton’s law of convection to obtain
( )
Q = h c A∆Tinterface = A∆Tinterface 1 Rc (10.51)
Equation 10.51 can then be solved the temperature drop or the heat transfer rate across the interface. Thus the underly-
ing physics is lumped into the value of Rc. Most experimentally determined values for Rc fall between 0.000005 and
0.0005 m2-°C/W, and the materials used in nuclear power plants are no exception to this rule. The temperature profile
across the interface (see Figure 10.26) is also different than it is when the materials are in perfect contact. Hence, the
temperature profiles are different when there is perfect and imperfect contact.
FIGURE 10.26 When two different materials touch, the interface between them is never in perfect contact. This creates a tempera-
ture drop ΔT across the interface that does not exist when the contact is ideal. The interface typically contains pockets of gas or air
that contribute to this effect.
10.25 Heat Conduction in Cylindrical and Spherical Objects 379
10.24 T he Relationship between the Thermal Resistance and the Nuclear Heat Flux
In the study of nuclear fuel rods, the heat flux q″ from the surface of a rod is defined as q″ = q/A, and the linear heat
generation rate q′ from the same fuel rod is defined as q′ = q/L, where L is the rod’s length. If Fourier’s law is written as
q′′ = q A = − k ∆T ∆x (10.52)
or
then the definition of the thermal resistance is somewhat different than the definitions given previously. Specifically, the
thermal resistance becomes
when referring to the linear heat generation rate q′. In nuclear engineering, these definitions are sometimes used inter-
changeably so the reader should be aware of this distinction exists because it is possible for the thermal resistance to
have several different sets of units for each unit system in which it is quoted. The same conclusion applies to cylindrical
fuel rods which we will now discuss.
Q = − kA dT/dr (10.56)
is no longer constant. In fact, it varies with distance in the radial direction and it is a function of the local radius r. Hence
we must write the surface area for heat transfer as A = 2πrL, and this surface area varies in the direction of heat flow.
To illustrate this point, suppose that we apply Fourier’s law to an infinitely long cylinder having an inner radius of r1
and an outer radius of r2. Further, suppose that T = T1 at r = r1 and T = T2 at r = r2. Then, integrating Fourier’s equation
between r = r1 and r = r2 gives
∫ Q A dr = − ∫ k dT (10.57)
or
( )
Q = 2πLk ln r2 r1 ( T1 − T2 ) (10.58)
The thermal resistance through a cylindrical object having a uniform value of k is then
FIGURE 10.27 The thermal resistance for a cylinder or a sphere is defined in the same way as it is for a planar wall, but the values
of Rcylinder and Rsphere are different. In this case, a cylindrical (or spherical) shell is subjected to convection from both sides.
where r1 and r2 are the inner and outer radii, respectively. For a spherical object, the surface area is also a function of
the radius, but in this case, A = 4πr2. Using Fourier’s law and performing the same integrations again result in a slightly
different expression for the thermal resistance:
These two expressions can be used to find the heat flow rate through a cylindrical or spherical object if the tempera-
tures at points r1 and r2 are known and the objects have a uniform material composition. Alternatively, they can be used
to determine one of the temperatures if the other temperature is known and the heat flow rate Q is also known. Now
consider the one-dimensional steady-state flow of heat through a cylindrical or spherical object that is exposed to a
fluid flowing over each side. The temperatures of the fluid are T1∞ and T2∞, and their respective convective heat transfer
coefficients are h1 and h2. A problem of this type is depicted in Figure 10.27. The thermal resistance to the flow of heat
consists of one conductive resistance and two convective resistances. Moreover, because there is only one path through
which the heat can flow, these resistances can be considered to be in series. Hence for steady-state conditions, the heat
flow rate can be written as
where
Notice that the value of A must be the surface area of the object at the radius r where the convection occurs. This
approach can also be extended to multilayer cylinders and spheres. However, for these objects, an additional term must
be added to the thermal resistance for each additional layer that is present. Normally, the thermal resistance can be
applied to other problems provided that the proper conductive resistances and surface areas are used for the convective
terms.
10.26 Finding the Thermal Resistance of the Cladding and the Coolant 381
10.26 Finding the Thermal Resistance of the Cladding and the Coolant
Next let us turn our attention to a situation where fluid is flowing over an object having surface area A. In this case,
Newton’s law of convection requires that
Q = hA∆T (10.64)
where h is the convective heat transfer coefficient, and ΔT is the temperature difference ΔT = TSURFACE − TCOOLANT
between surface of the object and the coolant. In this case, TCOOLANT is to be interpreted as the average temperature or
the bulk temperature of the coolant. Now, let us write Newton’s law and Ohm’s law side by side again. When we do so,
we see that
These expressions imply that the thermal resistance must be written as R = 1/hA when a fluid flows over a heated surface.
Thus, we have “derived” two very general expressions for the value of the thermal resistance R in a one dimension, and
we did not have to be particularly clever to do so. All we had to do is to compare the appropriate terms in Equation 10.65.
Now let us turn our attention to how this can be done for a cylindrical fuel rod. This rod can be a nuclear fuel rod, and
the rod can also be either solid or hollow. In cylindrical coordinates, the thermal resistances become
( )
R = ln R OUT R IN 2 πk L For an annular cylinder (with an inner and outer surface) (10.66b)
where ROUT is the outer radius of the cylinder, R IN is the inner radius of the cylinder, and L is the cylinder’s length. In the
case of a fuel rod where the cladding is very thin, we can also write
( ) ( )
ln R OUT R IN = ln 1 + ∆R CLAD R CLAD (10.67)
where ΔRCLAD is the thickness of the cladding and RCLAD is its inner radius. If ΔRCLAD ≪ RCLAD, then we obtain
( )
ln 1 + ∆R CLAD R CLAD ≈ ∆R CLAD R CLAD (10.68)
Equation 10.68 is called the thin cladding approximation. In this case, the thermal resistance R reduces to
( )( )
R = 1 2 πkL ∆R CLAD R CLAD (10.69)
where again RCLAD is the inner radius of the rod. Similarly, if we repeat this process for a cylinder with axial coolant
flow, we find that R = 1/hA, where A = πR2L in this case and R is the cylinder’s outer radius. The results of this work
are summarized in Table 10.6. It does not take much imagination to see that the thermal resistance can be used to solve a
wide variety of serial or parallel heat conduction problems. In the next section, we would like to illustrate how the ther-
mal resistance can be applied to more realistic nuclear fuel rods. Solid and annular fuel rods are shown in Figure 10.28.
Solution Using Equation 10.60, the rate at which heat leaks from the tank is Q = (T2 − T1)/[(r2 − r1)/(4πkr1r2)]. From the
information given, we see that r2 = 2/2 = 1 m and r1 = r2 − 0.10 = 0.90 m. Plugging in the appropriate values, we see that the
rate of heat leakage is Q = (200)/[(0.10)/(4π × 0.08 × 0.90)] = 2,210 W = 2.21 kW. This is the leakage rate per second. In
1 h (or 3,600 s), 2.21 × 3,600 = 7,956 kJ = 7.96 MJ of thermal energy is required to keep the sodium at its recommended
temperature. [Ans.]
R = R1 + R2 (10.70)
and if we have three regions in series, then the thermal resistance becomes
R = R1 + R2 + R3 (10.71)
TABLE 10.6
The Thermal Resistances for Various Geometric Shapes
FIGURE 10.28 Reactor fuel rods can be either solid or annular in shape. Most fuel rods are solid, but in some cases, they are annu-
lar because this is the most efficient way to keep the fuel temperatures low.
10.27 Treating a Nuclear Fuel Rod as a Serial Heat Conduction Problem 383
∑
N
RTOTAL = Rn (10.72)
1
where N is the number of regions through which the heat is flowing. Suppose that we consider the fuel rod shown in
Figure 10.29 where we have a long cylindrical rod with fluid flowing over its outer surface. The temperature of the
fluid is T∞, and the fuel rod consists of a central pellet of uranium or plutonium dioxide (with a small void in it) sur-
rounded by a protective layer of cladding. The temperature on the inner surface of the fuel is T1, and the temperature
on the outer surface is T 2. The temperature on the outer surface of the cladding is T 3. The thermal resistances of the
individual regions are
( )
R1 = ln r2 r1 2 πLk fuel (for the fuel) (10.73)
( )
R2 = ln r3 r2 2 πLk clad (for the cladding) (10.74)
Q = ( T1 − T∞ ) R (10.76)
where
( ) ( )
R = R1 + R2 + R3 = ln r2 r1 2 πLk fuel + ln r3 r2 2 πLk clad + 1 hA (10.77)
Thus, the thermal resistance has allowed us to take a complicated problem, and reduce it to a very simple one. If the
coolant temperature is known, we can then solve Equation 10.76 to find the temperature at the center of the fuel pin.
Conversely, if the temperature at the center of the fuel pin is known, we can use Equation 10.76 to determine the heat
flow rate Q if the coolant temperature is known. Many related problems in nuclear heat transfer can be solved in this
way. Thus, we only need the correct values of the thermal resistance for each material region. Later on, we will find
that the thermal resistance of the fuel changes with burnup because the thermal conductivity of the fuel is burnup
dependent. For the moment, this is all we would like to say about serial nuclear heat transfer.
FIGURE 10.29 The temperature drop across a long cylindrical fuel rod with central void can be found by calculating the thermal
resistances through the fuel, the cladding, and over the cladding surface and then summing them together to find the total thermal
resistance R.
384 The Laws of Nuclear Heat Transfer
FIGURE 10.30 A picture illustrating the difference between serial heat conduction and parallel heat conduction in a
multi-region material.
FIGURE 10.31 The total thermal resistance R is different in a serial heat conduction problem than it is in a parallel one.
The equations above show how the value of R can be found in each case.
10.29 Problems with both Serial and Parallel Heat Transfer 385
We can calculate the thermal resistance of these two parallel conductors in the same way. Only in this case, R1 = Δx/k1A
and R2 = Δx/k2A. Hence the thermal resistance R of the parallel combination is
( )(
R = ∆x k1A ∆x k 2 A ) ( ∆x k A + ∆x k A ) (10.79)
1 2
or
( )(
R = 1 k1 1 k 2 ) (1 k 1 )
+ 1 k 2 (10.80)
Q = ( T1 − T2 ) R = ∆T/R (10.81)
where R = (1/k1)(1/k2)/(1/k1 + 1/k2). This result can obviously be extended to problems with more than two parallel
regions by realizing that
1 R = 1 R1 + 1 R2 +1 R3 + + 1 RN (10.82)
For a problem consisting of three parallel material regions, the thermal resistance
1 R = 1 R1 + 1 R2 + 1 R3 (10.83)
becomes
R = R1 R2 R3 ( R1 R2 + R2 R3 + R1 R3 ) (10.84)
which is also the same result as one would obtain for three parallel resistors.
R = R1 + R2 + R34 + R5 + R6 (10.85)
386 The Laws of Nuclear Heat Transfer
FIGURE 10.32 A conductive heat transfer problem involving both serial and parallel heat transfer processes.
RTOTAL = R1 + R2 + R3 R4 ( R3 + R4 ) + R5 + R6 (10.86)
and the rate of heat transfer across this combination of materials becomes
( )
Q = ( T1 − T2 ) RTOTAL ⇒ Q = ( T1 − T2 ) R1 + R2 + R3 R4 ( R3 + R4 ) + R5 + R6 (10.87)
The heat flow in most composite materials, including complex nuclear fuel rods, can be handled in this way.
where TFUEL is the maximum temperature of the fuel, TCLAD is the temperature on the outer surface of the cladding, and
RROD is the thermal resistance for the rod. Because the rod is completely solid, the maximum fuel temperature is reached
at the center of the rod where r = 0. Heat will also leave the outer surface of the cladding by thermal convection, and
this thermal convection occurs because the heat is transferred to the coolant. The convective heat transfer coefficient
h is a function of the type of coolant (water, a liquid metal, or a gas) and how fast it flows past the rod. Again, because
this is a serial conduction problem, the total thermal resistance is simply the sum of the individual thermal resistances.
For this particular rod, we have
( ( ) )
RROD = RFUEL + RCLAD + RSURFACE = 1 4 πk F + 1 2πk C ⋅ ln R CLAD ( R CLAD − ∆R CLAD ) + 1 2 πhR CLAD (10.89)
10.31 Heat Conduction in Fuel Rods with a Fuel–Cladding Gap 387
Normally the cladding is much thinner than the outer radius of the rod (ΔRCLAD ≪ RCLAD), and we can also write the
term ln (RCLAD/RCLAD − ΔRCLAD)) as
( ) ( )
ln ( R CLADI + ∆R CLAD ) R CLADI = ln 1 + ∆R CLAD R CLADI ≈ ∆R CLAD R CLAD (10.90)
( ) ( )
RROD = 1 4πk F + 1 2πk C ∆R CLAD R CLAD + 1 2πh 1 R CLAD (10.91)
Notice that when there is no cladding, or when the thermal conductivity is high and the cladding is very thin,
( )
RROD ≈ 1 4 πk F + 1 2πh R FUEL (10.92)
where R FUEL is the radius of the fuel pin. Once Q is known, we can determine the temperature Tj at any intermediate
point j by using the relationship
Q = ( Ti − Tj ) RTOTAL,i-j (10.93)
across any region or regions where Ti is the known temperature at point i and RTOTAL, i-j is the total thermal resistance
between points i and j. Hence, the ratio ΔT/R must be equal to the heat flow Q across any layer, and this ratio remains
constant for any one-dimensional, steady-state heat conduction problem. This same concept can be applied to annular
fuel rods and to fuel rods that have a fuel–cladding gap. The only difference is that the number of terms in Equation
10.89 will increase as the number of regions increases. In general, there will be one more term for each layer in the fuel
rod or each additional type of thermal resistance that is present.
This definition allows us to calculate the temperature drop ΔTGAP across the gap if all of the relevant physical parameters
are known. In general, the value of hGAP depends on the average thickness of the gap (which can change as a function
of time), and it also depends on the burnup of the fuel, the contact pressure between the fuel and the cladding, and the
thermal conductivity of the fission gases. It typically ranges between 0.5 and 1.2 W/cm2 °C and it is about 0.65 W/cm2 °C
when the fuel is fresh. (see Figure 10.33). However, we also know that for a cylindrical rod
FIGURE 10.33 The gap heat transfer coefficient in a thermal water reactor has a value between 0.7 and 1.1, and its value increases
as the burnup increases. This increase in the value of hGAP is caused by the thermal expansion of the fuel.
and
( )
q′ = 1 2πR F h GAP ∆TGAP (10.98)
This then leads us to conclude that the thermal resistance of the gap is
when the heat transfer rate q′ is expressed in W/cm. Thus, the thermal resistance of the gap declines as the gap heat
t ransfer coefficient increases. For a solid fuel pin surrounded by such a gap, the total thermal resistance of the rod
becomes
where R ROD = R FUEL + RGAP + RCLAD + RSURFACE. For a desired linear power density of q′, the rod temperature drop
will be
Because q′ = 2πR FUEL q″, an alternative expression for the rod temperature drop which involves the surface heat flux
q″ is
( )
∆TROD = q′′ × R FUEL 1 2k F + 1 h GAP R FUEL + 1 k C ∆R CLAD R CLAD + 1 hR CLAD (10.102)
10.33 FINDING THE HEAT TRANSFER RATE 389
This eliminates the π’s entirely from the equation for a cylindrical fuel rod. However, it also introduces a scaling factor
R FUEL into our equations, which is a function of the radius of the fuel. Notice that the fuel–cladding gap simply adds
another term to the equation for the thermal resistance. The temperature drop across the gap is on the order of 200°C
for a linear power density of about 500 W/cm. In Chapter 11, we will compare this to the temperature drops across other
parts of the same nuclear fuel rod.
( )
RROD = RSURFACE 1 + RFUEL + RSURFACE 2 = 1 2 πh1R FUEL 1 + 1 2 πk F ⋅ ln R FUEL 2 R FUEL 1 + 1 2 πh 2 R FUEL 2 (10.103)
and if the fuel rod is covered on both sides with a thin layer of cladding.
( )( )
=1 ( 2 πh1R CLAD1 ) + 1 2 πk C ⋅ ∆R CLAD1 R CLAD1 + 1 4 πk F
( ) ( )
+ 1 2 πk C ⋅ ln ∆R CLAD2 R CLAD2 + 1 2 πh 2 R CLAD2 (10.104)
Clearly, this approach can be extended to fuel rods with even more regions if we want. In the next chapter, we will
show how this can be used to deal with restructured fuel rods at high burnups. At high burnups, the thermo-physical
properties of the fuel change, and the fuel develops an additional columnar and equiaxed grain structure. This creates
additional annular regions or zones, and each of these regions can have a different physical density, thermal resistance,
and thermal conductivity than a fresh fuel rod.
10.33 F
inding the Heat Transfer Rate When the Thermal
Conductivity Is a Function of Temperature
In practice, the thermal conductivity of most materials is not constant over their entire operating range. Thus, it either
rises or falls as the temperature changes. For most materials, this dependence is almost a linear one, which implies that
we can write
where
ko is the thermal conductivity at 0°C and
β is the temperature coefficient of the conductivity, which usually measured in /°C. Here the value of β is normally
positive for nonmetals (magnesite bricks being the primary exception) and negative for metals (aluminum and
certain nonferrous alloys again being the primary exceptions).
When the effect of this variable thermal conductivity is taken into account, Fourier’s equation in one dimension
Q = −kA dT/dx can be written as
L T2
q′′
∫ 0
dx = − k o
∫ T1
(1 + βT ) dT (10.108)
or
or
or
where TAVG = (T1 + T2)/2 and k AVG = ko (1 + βTAVG) is now called the mean thermal conductivity of the material. Most
problems in conductive heat transfer where the thermal conductivity changes with temperature can be handled in this
way. When this same logic is applied to an annular fuel pin of length L such as the one shown in Figure 10.28, we obtain
Q = − k o (1 + βT ) 2πr dT dr (10.112)
where r1 is the radius of the central void and r2 is the outer radius of the pin (excluding the cladding). After integrating
both sides, Equation 10.112 becomes
( )
Q = k AVG 2πL ( T1 − T2 ) ln r2 r1 (10.113)
where again TAVG = (T1 + T2)/2 and k AVG = ko (1 + βTAVG). Equation 10.113 then defines the thermal resistance
( )
RANNULAR = ln r2 r1 2 πL k AVG (when the thermal conductivity is temperature dependent) (10.114)
for an annular fuel pin when the thermal conductivity is temperature dependent. Now, let us summarize what we have
just learned.
T2
k AVG =
∫ T1
k(T)dT ( T2 − T1 ) (10.115)
This relationship assumes that the rate of heat transfer through a material with a value of k AVG is the same as the rate of
heat transfer through the same material with a temperature-dependent thermal conductivity k(T). Hence, when k(T) = k,
Equation 10.105 reduces to k AVG = k, which is the expected result for a material with a constant thermal conductivity.
10.34 Summarizing Our Findings 391
FIGURE 10.34 The thermal conductivity of many materials is temperature dependent. This includes liquids, metals, and gases.
For plate-type and cylindrical fuel rods, the steady-state heat transfer rate can be found by replacing the constant t hermal
conductivity with the expression for the average conductivity given by Equation 10.115. When we do so, we find that
T2
Q PLATE = − k AVG A ( T2 − T1 ) L = (A/ L)
∫ T1
k(T) dT (for a plate) (10.116)
T2
( ) (
Q CYLINDER = −2πk AVG L ( T2 − T1 ) ln r2 r1 = 2πL ln r2 r1 ⋅ ) ∫ T1
k(T) dT (for a cylinder) (10.117)
where r2 and r1 are the outer and inner radii of the cylinder, respectively. Sometimes the variation in the thermal conduc-
tivity is written as a linear relationship of the form
where ko is the thermal conductivity at a previously defined reference temperature. The parameter β is then called the
temperature coefficient of the thermal conductivity. Normally, this coefficient is a simple constant that is much less than
unity. The average value of the thermal conductivity between T1 and T2 can then be found using
T2
k AVG =
∫ T1
( )
k o (1 + βT)dT ( T2 − T1 ) = k o 1 + β ( T1 + T2 ) 2 (10.119)
or
where TAVG = (T1 + T2)/2. The value of β can be either greater than 0 or less than 0. When β is equal to zero, the thermal
conductivity no longer has any temperature dependence. The effect of β on the temperature profile through a planar wall
is shown in Figure 10.10. Hence values of β < 0 tend to depress the temperature profile in the rod.
temperature ranges: 500°C–1,000°C, 1,000°C–1,500°C, and 1,500°C–2,000°C. Over which temperature range is the
value of β the least?
Solution Using the graph shown at https://www.nuclear-power.net/wp-content/uploads/2017/10/Thermal-Conductivity-
Uranium-Dioxide-chart.png, the values for β are approximately −0.002/°C from 500°C to 1,000°C, −0.001/°C from
1,000°C to 1,500°C, and essentially 0.000/°C from 1,500°C to 2,000°C. Clearly, the value of β is least above a tempera-
ture of 1,500°C where the central void develops. [Ans.]
10.35 Typical Values of the Nuclear Heat Flux and the Fuel Rod Temperature
The temperature profile in a nuclear fuel rod depends on its geometry as well as its material composition. We will pres-
ent some representative temperature profiles for these rods in Chapter 11. However, for a typical PWR fuel rod, we can
use the equations we have just derived to estimate the size of the temperature drops across the fuel, the cladding, the
fuel–cladding gap, and the outer surface of the cladding. Normally, PWR fuel rods operate at a centerline temperature
of about 1,600°C. The temperature drop across the fuel at a linear power density of 500 W/cm is about 1,000°C. The
temperature drop across the fuel–cladding gap is about 150°C, and the temperature drop across the cladding alone is
about 75°C. Finally, the temperature drop between the surface of the cladding and the coolant is about 10°C. This results
in a total temperature drop between the center of the fuel rod and the coolant of about 1,300°C.
Thus, the bulk temperature of the coolant temperature tends to average about 300°C between the top and the bottom
of the core. Some representative temperature profiles are shown in Figure 10.35. Of course, these profiles are specific to
the cylindrical fuel rods used in PWRs. Plate-type fuel rods used in research reactors and cylindrical fuel rods used in
BWRs have slightly different temperature profiles. Also, the temperature profiles in BWR fuel rods are similar to those
shown in Figure 10.35 except that they are flatter because the linear power density is about half as high (about 250 W/cm)
FIGURE 10.35 The fuel rod temperatures as a function of the linear power density for a UO2 fuel pellet with fresh fuel.
In Chapter 11, we will find that these values change as the fuel is burned.
10.38 Problems with Conduction, Convection, and Radiative Heat Transfer 393
and the diameter of the rods is about 20% greater. In Chapter 11, we will derive the shape of the actual radial and axial
temperature profiles under these conditions.
FIGURE 10.36 Normally, three types of subject matter are associated with reactor thermal-hydraulics: reactor thermodynamics, reactor materials, and reactor heat transfer. Reactor heat
transfer is subsequently subdivided into conductive, convective, and radiative heat transfer.
The Laws of Nuclear Heat Transfer
Questions for the Student 395
FIGURE 10.37 Some interesting problems in everyday life where interesting combinations of conductive, convective, and radiative
heat transfer processes can occur at the same time. (Pictures provided by the Ohio State University.)
everyday problems where all three forms of heat transfer occur simultaneously. The closer one comes to a hot object, the
more likely it is that a significant amount of radiative heat transfer will occur.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Crowe, C. and Schwartzkopf, J., et al. Multiphase Flows with Droplets and Particles, CRC Press, Boca Raton, FL (2012).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc. New York,
NY (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi (2000).
Reactor Theory (Nuclear Parameters) DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
53. Which heat transfer process is believed to be responsible for the greenhouse effect?
54. According to Newton’s law of cooling, how does the temperature of a hot object behave when it is dropped into a
tank of cold water?
55. Which forms of heat transfer are responsible for the steam rising from a cup of hot coffee?
56. In nuclear power plants, how is the Reynolds number defined?
57. In a radiation shield, the radiation attenuates exponentially, and in doing so, it deposits heat in the shield as it moves.
In a planar shield in a nuclear power plant, at what location is the internal temperature normally the highest?
58. Fill in the following sentence with the appropriate word or phrase: In conductive heat transfer, heat flows from a
region of _____ temperature to a region of _____ temperature. The magnitude of this flow is proportional to the
negative _______ of the temperature. The constant of proportionality between the temperature _______ and the rate
of heat flow is the ________.
Exercise 10.2
A 2 cm-thick reactor coolant pipe has hot water flowing through it at 250°C. If the temperature on the inside surface of
the pipe is 250°C and the pipe is made up of stainless steel, what is the temperature on the outside surface of the pipe?
Assume a heat flux of 1,000 W/m2 °C.
Exercise 10.3
Suppose that the pipe in Exercise 10.2 is surrounded by a layer of air maintained at 50°C. What is the rate at which
heat flows from a 1 m-long section of the pipe if the convective heat transfer coefficient at the pipe’s outer surface
is 10 W/cm °C?
Exercise 10.4
A planar radiation shield with a linear absorption coefficient μ = 0.52/cm is bombarded by 2 MeV gamma rays imping-
ing on its outer surface at a rate of 1 × 106 γ-rays/cm2 s. Write down an equation for the heat q‴(x) generated in the shield
assuming that the shield is made up of lead of 10 cm thickness. Write down the equation that must be solved to find the
temperature profile in the shield if the temperatures on the inner and outer surfaces of the shield are known.
Exercise 10.5
Using the information provided in Exercise 10.4, write down a general solution for the temperature profile in the shield.
If the thickness of the shield is w, and the boundary conditions on the outer and inner surfaces of the shield are T(0) = T1
and T(w) = T2, derive an expression for the temperature inside of the shield. Exactly where does the temperature peak
in the shield?
Exercise 10.6
Suppose that the pipe in Exercise 10.3 is positioned horizontally rather than vertically. If the average velocity of the air
flowing over the pipe is 0.1 m/s, and the average temperature of the air is 50°C, what is the Reynolds number of the air
flowing over the pipe? Is the flow of air laminar or turbulent?
Exercise 10.7
The core of a fast reactor is cylindrical in shape, having a radius of 40 cm and a height of 80 cm. A reactor physics calculation
shows that the power density in the core can be represented approximately by q′′′ ( r,z ) = q′′′ MAX 1 − ( r/50 ) cos(πz/100)
2
398 The Laws of Nuclear Heat Transfer
where q′′′
MAX is the maximum power density at the core midplane and r and z are the respective distances in centimeters
from the axis and the core midplane. What is the peak-to-average power ratio for this core? Calculate the peak-to-
average power ratios in the radial and axial directions. In what direction is the peak-to-average power ratio the greatest?
Exercise 10.8
A fuel pellet weighing 10 g and consisting of 50% uranium dioxide and 50% plutonium dioxide is stress-tested by heat-
ing it to 2,000°C and then immersing it in a bath of cold water at 10°C. Approximately how long will it take for the
temperature of the pellet to drop to 100°C?
Exercise 10.9
A spherical tank is used to store liquid sodium (a common reactor coolant) at a temperature of 400°C. The tank is 1.4 m
in diameter and 9 cm of thickness. Find the rate of heat leakage from the tank if the temperature difference between the
inner and outer surfaces is 220°C. Assume that the thermal conductivity of the tank wall is 0.083 W/m °C.
Exercise 10.10
An aluminum pipe carries steam at 110°C to a condenser where it is to be converted back to water. The pipe has an inner
diameter (ID) of 10 cm and an outer diameter (OD) of 12 cm. The pipe has a thermal conductivity of 185 W/m °C. The
pipe is located in a room where the ambient air temperature is 30°C and the convective heat transfer coefficient between
the pipe and the air is 15 W/m2 °C. If the pipe is 10 m long, determine the heat transfer rate between the pipe and the air.
Exercise 10.11
To reduce the heat loss between the pipe and the air in Exercise 10.10, the pipe is covered by a 5 cm-thick layer of insu-
lation, which has a thermal conductivity of 0.20 W/m °C. Determine the heat transfer rate between the pipe and the air
after the insulation is installed.
Exercise 10.12
A radiation shield made from bricks and concrete has an inner surface temperature of 60°C and an outer surface tem-
perature of 35°C. The shield has a surface area of 10 m2. Calculate the rate of heat transfer through the shield if it is 22 cm
thick. Assume that the thermal conductivity of the shield is 0.50 W/m °C.
Exercise 10.13
A horizontal tube passes through a heat exchanger where water enters the tube at 20°C and leaves the tube at 40°C.
Saturated steam is then passed over the tube at 80°C where it is condensed. The heat transfer coefficient on the outer
surface of the tube is 10,750 W/m2 °C, and the heat transfer coefficient on the inner surface of the tube is 8,775 W/m2 °C.
The tube is made up of brass and has a thermal conductivity of k = 110 W/m °C. The tube has an ID of 1.34 cm and an
OD of 1.59 cm. Calculate the thermal resistance of the entire tube.
11
Heat Removal from Nuclear Fuel Rods
11.1
T he Flow of Heat through Nuclear Materials
Now that we have discussed the laws of thermodynamics and heat transfer in Chapters 6 and 10, we would like to turn our attention
to the application of these laws to nuclear fuel rods and to the reactor coolant that removes the heat from these rods. In conductive
heat transfer, heat flows from one material to another based on a temperature difference between the materials. Heat can also flow
based on internal temperature differences within a material. In other words, conductive heat transfer can be thought of as a molecu-
lar diffusion process where there is no macroscopic movement of the material through which the heat is flowing. Thermal energy
is transferred from a hotter region to a cooler one because the hotter region has more random molecular motion than the cooler one.
In other words, this vibrational energy is transferred from a region of high temperature to a region of low temperature, and we can
interpret this energy transfer as the flow of heat. If we were to examine the process of conduction more closely, we would find that
the atoms in the fuel rods vibrate mindlessly about their equilibrium positions in the lattice, and as they vibrate, they tend to transfer
some of this vibrational energy to the atoms that are closest to them. Sometimes these atoms are called their nearest neighbors.
When the conduction of heat occurs in a fuel rod, thermal energy is transferred from the high temperature regions of the rod to the
low temperature regions of the rod. This thermal energy transfer can be described by a diffusion equation that is called the heat
conduction equation. This transfer of heat also requires the existence of a temperature gradient.
Convection is different than conduction because it requires the fluid (or coolant) that receives the heat to be at least partially
in motion. In other words, in convective systems, we can no longer assume that the material that is losing or gaining energy is at
rest. In nuclear power plants, this material is called the reactor coolant, and in boiling water reactors (BWRs) and pressurized
water reactors (PWRs), it is light water or a mixture of light water and steam. However, fast reactors use l iquid metals to cool
the core, and gas reactors are normally cooled with carbon dioxide or other high temperature gases. This convection of heat can
occur with either single phase or two phase fluids. Single phase heat transfer is easier to understand than two phase heat transfer
and the correlations used to describe the convective heat transfer process are simpler (see Chapters 20 through 25). When the
coolant boils, the underlying processes governing the convective heat transfer rate can be remarkably complex. These processes
can then be described by what is called two-phase nuclear heat transfer (see Chapters 23–25). Finally, in radiative heat transfer,
particles of light called photons, in addition to alpha and beta rays, can deposit some of the energy produced by the fuel rods in
other regions of the core. This energy deposition process can be described by either the Stefan–Boltzmann law (see Chapter 10)
or by adding an energy deposition term to the heat conduction equation. We will examine both of these approaches later in the
chapter. Thus, conduction and convection are the primary forms of heat transfer in nuclear power plants, although radiative heat
transfer, which involves high energy photons called gamma rays, must also be taken into account in the design of the pressure
vessel and radiation shields.
11.2
Fourier’s Law and Its Application to Nuclear Systems
The underlying law that describes the flow of heat between two materials (or between different regions of the same material) is
Fourier’s law of conduction, which we introduced in Chapter 10. Fourier’s law was initially proposed by French mathematician
Joseph Fourier in 1820 when he hypothesized that heat not only flowed from a region of high temperature to a region of low tempera-
ture, but that the rate at which it flowed could be described by the equation
q = − kA dT/dx (11.1)
where the heat transfer rate was proportional to the negative gradient of the local temperature; that is,
q = − kA grad T = − kA ∇T (11.2)
399
400 Heat Removal from Nuclear Fuel Rods
The direction of this flow is shown in Figure 11.1. If we divide both sides of Equation 11.2 by A (the area through which
the heat flows) the heat flow rate can also be expressed in terms of the nuclear heat flux q″ = q/A.
q ′′ = − k grad T = − k ∇T (11.3)
In Equation 11.3,
☉☉ q is the amount of heat that flows.
☉☉ A is the area through which the heat flows.
☉☉ T is the temperature of the material containing the heat.
☉☉ grad or ∇ is the gradient operator.
☉☉ q″ = q/A is the heat flux or the heat flow per unit area.
☉☉ k is a constant of proportionality called the thermal conductivity.
According to these definitions, q″ has the units of BTU/h/ft2 or W/m2, T has the units of °F, °C, or °K, and k has the
units of BTU/h/ft °F in the English unit system or W/m °K in the SI unit system. In nuclear applications, the thermal
conductivity depends on both the composition of a material and its temperature. Hence for a given material, the value
of k is temperature dependent for example, k = k(T). The values of k are tabulated for most common materials in a
predefined format, and we will discuss the values of k for many of these materials later in the chapter. A material with
a very low value of k is referred to as an insulator, and a material with a very high value of k is referred to as a conduc-
tor. Most common materials have values for the thermal conductivity that make them neither insulators nor conductors
(See Chapter 10).
However, the heat flow rate q and the heat flux q″ are vector quantities because they have both magnitude and
d irection, while the temperature is a scalar quantity with only magnitude but no direction. In three dimensions, the
d irection and magnitude of the heat flux q″ can be written as the sum of the heat flux along each coordinate direc-
tion, multiplied by the appropriate unit vectors (i, j, or k). Hence, the heat flux q″ in a simple isotropic medium can
be written as
where q ′′,
x q ′′,
y and q z′′ are the individual heat flows in the x, y, and z directions. It then follows that the magnitude of the
total heat flux is
q ′′ = ( q′′
x
2 2
)
+ q ′′y + q z′′2 (11.5)
FIGURE 11.1 Heat always flows in the direction of decreasing temperature, and because of this, we can write Fourier’s law in the
x direction as q ′′x = − k dT/dx. A line of constant temperature perpendicular to the t emperature gradient is then called an isotherm.
11.4 Deriving the Heat Conduction Equation 401
FIGURE 11.2 A cube with heat flowing through it in response to a temperature difference.
and that the temperature difference is the driving force that causes this flow of heat to occur. To be more specific, the
heat flux along any direction for an isotropic medium is proportional to the negative gradient of the temperature along
that direction. That is to say, the heat flux in the x, y, and z directions is given by
and these are the fundamental principles upon which conductive heat transfer in nuclear engineering is based.
11.3
T he Equations of Heat Conduction for a Nuclear System
The flow of thermal energy into or out of a region of space generally defines how hot or cold the region will
become. Because of this, it is possible to derive an analytical expression for the temperature of the region by
simply measuring all the heat energy that flows into the region, and then subtracting all the heat energy that flows
out of the region. This energy balance, in conjunction with Fourier’s law, can then be used to derive what is called
the t ime-dependent heat conduction equation. This equation can then be used to predict the temperature profile
within that region of space. Since the process of conductive heat transfer is so important to the study of nuclear systems,
we would like to take a moment to derive the heat conduction equation for an isotropic medium. It is then possible to
generalize our derivation to other problems where the medium is no longer spatially homogeneous or isotropic. The heat
conduction equation can be solved directly for simple geometric shapes like cylinders, rectangles, and spheres. However,
for more complex geometric shapes, it is usually solved numerically.
11.4
D eriving the Heat Conduction Equation
In order to gain some insight into how the process of conductive heat transfer works, consider for the moment the volume
of space V shown in Figure 11.2. This volume consists of a cube having dimensions Δx, Δy, and Δz. The total volume
of the cube is V = ΔxΔyΔz, and this volume is centered about a point in space which is located at (x, y, z). Sometimes,
this hypothetical volume is also called a control volume. Each of the six faces of the cube have a surface areas of ΔxΔy,
ΔxΔz, or ΔyΔz. (See Figure 11.2). Thermal energy can travel into and out of each of the six faces at different rates. On
some faces, the heat flux q″ will be positive, or into the cube. On other faces, the heat flux q″ will be negative, or out of
the cube. The total heat transfer rate across any face of the cube is then equal to the product of the heat flux, q″ across
402 Heat Removal from Nuclear Fuel Rods
that face, and the area of the face itself (which can be either ΔxΔy, ΔxΔz, or ΔyΔz). In other words, the flow of heat
moving across each face is q″ ΔA, where ΔA = ΔxΔy, ΔxΔz, or ΔyΔz, and q″ is the component of the vector normal to
the surface. We can then write an energy balance for the control volume as
This is a simple expression of the conservation of thermal energy within the control volume. Now, let us call the internal
energy of the material inside the control volume U, and assume for the moment that its internal energy can be written as
the product of the density ρ of the material inside the cube and another important parameter called the specific heat c.
Hence, we will assume that we can write the internal energy of the material in the cube as
U = ρcT (11.7)
Moreover, in a conducting material that does not expand or contract very much, the internal energy U of the material is
directly proportional to its absolute temperature T. This turns out to be appropriate for solids because it distinguishes the
way that solids behave from the way that liquids and gases behave. Of course, this specific linear relationship can change
slightly from one type of solid to the next (because the values of ρ and c can be temperature dependent), but in general, this
relationship is valid for any material that does not expand or contract appreciably. In Equation 11.7, the density ρ has the
units of g/cm3, the specific heat c has the units of J/g °K, and the temperature of the material T is usually expressed in °K.
Referring to Figure 11.2 again, we can write our expression of the conservation of energy for the control volume as
where Ax is the area of the face in the x direction, Ay is the area of the face in the y direction, and Az is the area
of the face in the z direction. Here, ∆q ′′x is the difference in the heat flux between point x1 = x + Δx/2, and point
x2 = x − Δx/2. Similarly, ∆q ′′y is the difference in the heat flux between point y1 = y + Δy/2 and point y2 = y − Δy/2, and
∆q z′′ is the difference in the heat flux between point z1 = z + Δz/2 and point z2 = z − Δz/2. It is also worth noting that
Ax = ΔyΔz, Ay = ΔxΔz, and Az = ΔxΔy. Since Ax = ΔyΔz, Ay = ΔxΔz, and Az = ΔxΔy and V = ΔxΔyΔz, we can rewrite
Equation 11.8 as
∆ρc/∆t = ∆q x ∆x + ∆q y ∆y + ∆q z ∆z (11.10)
Upon realizing that dρc/dt = Δρc/Δt, dqx/dx = Δqx/Δx, dqy/dy = Δqy/Δy, and dqz/dz = Δqz/Δz, Equation 11.10 can be
converted into its differential equivalent
d(ρcT)/dt = dq x dx + dq y dy + dq z dz (11.11)
∂(ρcT)/ ∂ t = ∂q / ∂x + ∂q / ∂ y + ∂q / ∂z (11.12)
or
where ∇ is the gradient operator we previously discussed. The equation we have just derived can be applied to any
material that can absorb thermal energy, and it can be looked upon as a fundamental expression of the conservation of
energy—where the energy we are referring to in this case is heat. We can extend this equation to other geometries, such
as cylindrical or spherical ones, by simply substituting the appropriate form of the gradient operator for the Cartesian
version that we have already used.
if we know the value of ∇T. Moreover, since the heat flux q″ is a vector, we can substitute the values for q directly into
time-dependent conservation equation we have just derived
( ) ( ) ( )
d(ρcT)/dt = d/dx k dT/dx + d/dy k dT/dy + d/dz k dT/dz + q ′′′(t) (11.17)
Equation 11.17 is the time-dependent heat conduction equation in its most general form. Note in particular that the
thermal conductivity k in this equation can be a function of the temperature T, and that the values of ρ and c can also be
temperature dependent as well. Finally, in time-dependent problems, the volumetric heat generation rate q‴ can also be
time dependent. So to be completely accurate, we should write Equation 11.17 as
( ) (
d ( ρ( T)c( T)T ) dt = d/dx k(T) dT/dx + d/dy k(T) dT/dy )
(11.18)
( )
+ d/dz k(T) dT/dz + q′′′(t)
and this is the form of the heat conduction equation that many reactor thermal-hydraulic analysis codes attempt to solve.
Finally, it is worth mentioning that if the conditions specified in Section 10.4 are satisfied—namely that we are d ealing
with a molecular diffusion process where Fourier’s law applies, and the internal energy U of a material is proportional to its
average temperature T, then this is the most straightforward way to express the conservation of thermal energy in the pres-
ence of a temperature gradient. To derive this equation we have assumed that the heat flux q″ is proportional to the negative
gradient of the temperature profile, and the temperature gradient is the driving force that allows thermal energy to flow.
Again, the difference between conduction and convection in this case is that in pure conduction, the material through which
the heat flows does not move, and because of this, we do not have to account for the flow of mass in the energy balance.
However, if the thermal conductivity k, the specific heat c, and the density ρ are all constants that are position independent,
then we can simplify the time-dependent heat conduction equation a bit, and in three dimensions, it becomes
or more generally,
∇ = ∂ / ∂x + ∂ / ∂ y + ∂ / ∂z (11.21)
Assuming for the moment that k, ρ, and c are constant and independent of position, we can write the heat conduction
equation in its most general form as
where we have taken the liberty to introduce a new symbol, called the Laplacian operator ∇2. The Laplacian operator
appears in virtually every diffusion equation in the fields of mathematics, chemistry, and physics. It is a shorthand nota-
tion for the second derivative of whatever vector or scalar quantity it operates on. The exact version of the Laplacian
operator varies from one type of coordinate system to the next. For example, we have derived it here for a Cartesian
coordinate system. In the case of a cylindrical or a spherical coordinate system, the Laplacian will have a completely
different form. The exact expressions for the Laplacian and the gradient operator are given in Table 11.1 for an arbitrary
function F. We will use this table occasionally to show how the heat flux behaves in geometries other than Cartesian
ones (see for example, Chapter 5). Depending on the coordinate system used, F can be a function of x, y, z, or r, θ, z, or
r, θ, φ; that is, F (x, y, z) or F (r, θ, z) or F (r, θ, φ). Sometimes it is convenient to write the time-dependent heat conduc-
tion equation in an alternative form by introducing another parameter called the thermal diffusivity α. In general, the
thermal diffusivity is defined as
α = k / ρc (11.23)
and it has the units of cm2/s or m2/h. Hence, a small value for k implies a small value for α, and a large value for k (which
is true in the case of a good conductor) implies a large value for α. The thermal diffusivity also appears in the study of
convective heat transfer, and we will introduce it to the reader again in Chapter 17. The time-dependent heat conduction
equation in three dimensions then becomes
(ρc k ) ∂T ∂t = ∂ T ∂x
2 2
+ ∂ 2 T ∂ y 2 + ∂ 2 T ∂z 2 + q ′′′ k (11.25)
The reader should recognize this as the standard form of the time-dependent heat conduction equation that appears in
many heat transfer books. This latter equation forms the basis of most of the problems that we will encounter in the field
TABLE 11.1
Expressions for the Gradient and the Laplacian Operators in Various Coordinate Systems
of nuclear heat transfer, and its solutions have been the subject of a great deal of research over the years. In particular,
we can apply this equation to nuclear fuel rods where q‴ ≠ 0, and to the cladding and structural materials surrounding
the rods, where to a first approximation, it can be assumed that q‴ = 0.
11.6
T he Steady-State Heat Conduction Equation
In problems where the temperature does not change as a function of time, we can set ∂T/∂t = 0, and the time-dependent
heat conduction equation becomes
This equation is called the steady state heat conduction equation. In some textbooks, it is also called Poisson’s
e quation. Notice that this form of the heat conduction equation is a second-order differential equation that depends on
the volumetric heat generation rate q‴ and on the thermal conductivity k. In general, the volumetric heat generation rate
is position dependent, and so q‴(x, y, z) does not have to be the same everywhere. For example, this is true for a radiation
shield (refer to the discussion in Chapter 10) where the actual form of q‴ is q‴(x) = (A/k) e−Bx, where A and B are simple
constants. In a region where there are no heat sources, such as in the cladding surrounding a fuel rod, then q‴ = 0, and
the steady-state heat conduction equation becomes
This form of the heat conduction equation is also called Laplace’s equation. Many problems in steady-state conductive
heat transfer can be solved using it. The only dependent variable in this case is the material temperature T(x, y, z).
11.7
T he Form of the Heat Conduction Equation in Different Coordinate Systems
Before attempting to solve Equation 11.27, we would like to write it explicitly in a number of coordinate systems to
understand its structure. In Cartesian coordinates, we have already done so, but for the sake of completeness, we will
rewrite it here
∂ 2 T ∂x 2 + ∂ 2 T ∂ y 2 + ∂ 2 T ∂z 2 = − q ′′′ k (11.29)
( )
∇ 2 = ∂ 2 ∂r 2 + (1/r) ∂ / ∂r + 1 r 2 ∂ 2 ∂θ2 + ∂ 2 ∂z 2 (11.30)
( )
∂ 2 T ∂r 2 + (1/r) ∂T/ ∂r + 1 r 2 ∂ 2 T ∂θ2 + ∂ 2 T ∂z 2 = − q ′′′ k (11.31)
where θ is the azimuthal angle. In one-dimensional problems, where a fuel rod is very long, and there is no azimuthal
variation in its material temperature, the heat conduction equation reduces to
since dT/dz = 0 and dT/dθ = 0. The cylindrical coordinate system that can be used in this case is shown in Figure 11.3.
Finally, in spherical coordinates (see Figure 11.4), the values for x, y, and z become
FIGURE 11.3 A cylindrical coordinate system that can be used to find the temperature profile in a cylindrical fuel rod.
FIGURE 11.4 A spherical coordinate system that can be used to find the temperature profile in a spherical fuel pellet.
( ) ( ) ( )
∇ 2 = ∂ 2 ∂r 2 + (2 / r) ∂ / ∂r + 1 r 2 sin 2 θ ∂ 2 ∂ψ 2 + 1 r 2 sin θ ∂ ∂θ sin θ + ∂ ∂θ (11.34)
and for the one-dimensional case where the problem is completely symmetric, dT/dθ = 0 and dT/dφ = 0. Then, the
Laplacian operator reduces to
This final form of the heat conduction equation is helpful for finding the temperature profile in spherical fuel pellets such
as those used in pebble bed reactors, and in advanced gas reactors (AGRs), where spheres of uranium dioxide have a
conductive carbide or graphite coating. Finally, if the heat generation rate q‴ is a linear function of the fuel temperature
T, we can write
q ′′′ = q o T (11.37)
∇ 2 T + B2 T = S(r) (11.38)
where B2 = q o′′′ T/k and S(s) = −q‴(r)/k. This latter form of the heat conduction equation is called Helmholtz’s equation,
in honor of Ludwick Helmholtz, who first explored its properties in the mid-1800s.
11.8
Some Interesting Reactor Physics Analogies
Anyone who has taken a reactor physics class knows that the Helmholtz equation possesses the same canonical form as
the steady-state neutron diffusion equation, which forms the basis for much of the nuclear power industry in the world
today. This equation is sometimes written as
or
Here ϕ is the neutron flux and J is the neutron current. The neutron flux is a scalar quantity, and the neutron current is a
vector one. In the neutron diffusion equation, BM 2 is a constant called the material buckling, and its value is determined
by the macroscopic fission and absorption cross sections of the fuel (see Chapter 5). Neutron-producing regions have a
positive value for the material buckling, and neutron-absorbing regions have a negative one. Normally, the neutron flux
ϕ is measured in the units of neutrons/cm2/s. In addition, the neutron current J can also be derived from Fick’s law of
diffusion, which is sometimes written as
J = − D∇
∇φ (11.41)
Notice that the heat flux q″, which is the mathetical equivalent of the neutron current in this case, can be found from
Fourier’s law
which is the thermal equivalent of Fick’s law. Thus, the diffusion coefficient D in the neutron diffusion equation plays
a similar role to what the thermal conductivity plays in the heat conduction equation. Moreover, the source term S(r)
which appears in Equations 11.39 and 11.40 is sometimes called an external neutron source. Thus it serves a similar
function to the term S(t) = −q‴(t)/k, that appears in Helmholtz’s equation. Only in the case of the neutron diffusion
equation, it generates neutrons instead of producing heat.
11.9
O ther Ways to Classify the Heat Conduction Equations
The equations of conductive heat transfer can also be classified based on the way they are used. From the simplest
problems to the most complicated ones, they can then be classified into four basic categories which are defined as
☉☉ Steady-state conduction with no heat generation q‴
☉☉ Steady-state conduction with volumetric heat generation q‴
☉☉ Time dependent conduction with no heat generation q‴
☉☉ Time dependent conduction with volumetric heat generation q‴
408 Heat Removal from Nuclear Fuel Rods
TABLE 11.2
Various Forms of the Heat Conduction Equation and Their Equivalent Equations in Mathematics and Reactor Physics
d 2 T dr 2 + (1/r)dT/dr + d 2 T dz 2 = 0
d 2 T dr 2 + (1/r)dT/dr + d 2 T dz 2 + q ′′′ k = 0
(11.43)
d 2 T dr 2 + (1/r)dT/dr + d 2 T dz 2 = (1/α)dT/dt
and in spherical coordinates with spherical symmetry the corresponding equations are
d 2 T dr 2 + (2/ r)dT/dr = 0
How let us see how these equations can be applied to nuclear fuel rods such as the ones shown in Figure 11.5.
11.10
A Brief Review of Nuclear Fuel Rods and Their Properties
As we first discussed in Chapter 5, most nuclear fuel rods are cylindrical in shape and they are surrounded by a thin
layer of conductive material called the cladding. The cladding serves to isolate the fuel from the coolant. In BWRs and
PWRs, the fuel rods are between 4 and 5 m long, and their average diameter is about 1 cm. This includes the diameter
of the fuel pin, the cladding surrounding the fuel pin, and a small conductive gap between the fuel and the cladding.
Normally, the cladding is made from an alloy of zirconium called Zircaloy-2 or Zircaloy-4. In some cases, it can also
be made of alloys of stainless steel.
11.11 Fuel Rod Thermal Properties 409
FIGURE 11.5 In a reactor lattice with the same boundary conditions, the flux shapes and the temperature profiles in the fuel
rods are similar. This is due to the fact that the neutron diffusion equation and the steady-state heat conduction equation are both
examples of second-order partial differential equations with constant coefficients.
The volumetric heat generation rate q‴ is a function of the amount of fuel a fuel assembly contains, and generally
speaking, the more U-235 or Pu-239 that is present (or the higher the enrichment of the fuel), the higher the volumet-
ric heat generation rate will be. Normally, U-235 and U-238 atoms are combined with two oxygen atoms to form a
compound called uranium dioxide (or UO2). UO2 is a complex ceramic material with a very high melting point (about
2,800°C), and in some cases, it can withstand similar temperatures to the heat tiles that are used on the U.S. Space
Shuttle (see Chapter 10). Sometimes plutonium dioxide PuO2 (a by-product of the fission process) is also mixed with
uranium dioxide to form what is called metal oxide (or MOX fuel). The melting point of PuO2 is about 2,400°C, and its
mechanical properties are very similar to those of UO2. A considerable amount of heat can be generated by even a single
nuclear fuel rod, and volumetric power densities as high as q‴ = 110 kilowatts per liter (kW/L) are common in modern
PWRs. On the other hand, liquid metal fast breeder reactors (LMFBRs) can have even higher power densities (in the
range of q‴ = 300 kW/L – 500 kW/L), and so metallic coolants (such as liquid sodium and liquid mercury) are used to
cool them because they are more efficient at removing this additional heat. Using the heat conduction equation we can
then predict the shape of the temperature profile T(x, y, z) in a nuclear fuel rod if the value of q′′′(x, y, z) is known.
11.11
Fuel Rod Thermal Properties
The physical properties of the materials used in nuclear fuel rods are presented in Table 11.3. These values represent
the average values of their thermophysical properties, and some of these properties are also temperature dependent. For
example, the thermal conductivity of UO2 can vary by as much as a factor of 5 -6 depending on the temperature of the
fuel and its burnup. In general, the thermal conductivity of UO2 decreases with increasing burnup and so it becomes
harder to remove heat from the fuel the longer a reactor is operating. Fortunately, the gap thermal conductivity increases
over time, and this usually compensates for the reduced thermal conductivity of the fuel. A number of correlations have
been developed over the years to determine the effect of temperature on the thermophysical properties of the fuel and
the cladding. For example, in BWRs and PWRs, the thermal conductivity of the cladding is temperature dependent, and
this temperature dependence can be represented by the equations:
TABLE 11.3
Thermal Properties of Common Nuclear Materials
Similar correlations are available for the thermal conductivity of UO2 and PuO2. The thermal conductivity of uranium
dioxide and plutonium dioxide can be represented by
For solid UO2: k(T′) = 1/(7.5408 + 17.692 T′ + 3.6142 T′ 2) + 64 (T′−2.50) e−16.35/T′ (W/cm °K)
For solid PuO2: k(T') = 0.0844 − 0.07445 Т′ + 0.02236 T′ 2 (W/cm °K)
(From http://www-pub.iaea.org/MTCD/publications/PDF/IAEA-THPH_web.pdf)
when the theoretical density is 95%. Again, T′ = T/1,000 and T is the absolute temperature in °K. The first correlation
can be used for fuel pin temperatures from 300°K to 3,120°K. At the upper end of this range, the fuel begins to melt
and its thermophysical properties change. Nevertheless, because of its high melting point, its dimensional stability, and
its ability to retain fission gases, uranium dioxide is by far the world’s most popular nuclear fuel. The specific heat of
UO2 also has some temperature dependence, and its value can vary from about 300 J/kg at 500°C to about 500 J/kg at
2,500°C. The specific heat of uranium dioxide can be determined from
For solid UO2: c(T′) = A(52.1743 + 87.951 T′ − 84.2411 T′2 + 31.542 T′3 − 2.6334 T′4 + 0.7139/T′2)
(From http://www-pub.iaea.org/MTCD/publications/PDF/IAEA-THPH_web.pdf)
where again, T′ = T/1,000, and T is measured in °K. When A = 0.369, this correlation provides the specific heat in
J/kg °K, and when A = 1.0, it provides the specific heat in J/mol °K. Finally, the densities of UO2, PuO2, zirconium,
Zircaloy-2, and Zircaloy-4 are nearly independent of their operating temperatures until they reach their melting points.
Their respective melting points are shown in Table 11.4. Notice that UO2 has the highest melting point of any nuclear
fuel, and this is one of the reasons why it is so popular in light water reactors today.
TABLE 11.4
The Melting Points of the Most Common Materials Used in Nuclear Fuel Rods
11.12
T he Power Output of the Fuel and the Core
There are several ways to measure the power output of the core. First, the power output can be expressed in terms of
a core average power density in which the power produced is measured in kW/L. This method of power measurement
can be used to determine how much power is produced in the core as a whole. It does NOT tell us how much power is
produced in an individual fuel rod. The core power density is defined
Again, this definition is similar to the one used for the core power density. Sometimes it is more convenient to multiply
q ′′′
ROD by the surface area A of the rod. This allows us to express the power produced in terms of another quantity called
linear heat generation rate of the rod, which is given by
Here, the linear heat generation rate of the rod, q ′ROD , is measured in kW/m or W/cm. In reactor work, the units of W/
cm are more commonly used. The power densities for several different reactor types are shown in Table 11.5. A typical
PWR or BWR fuel rod can produce between 50,000 and 75,000 J of thermal energy per second. (Other rods can produce
even more.) Because their designs are very similar, the linear heat generation rates for PWR and BWR fuel rods are
about the same.
TABLE 11.5
The Typical Power Output for Different Types of Reactor Cores and Fuel Rods
11.13
Comparing the Temperature Drops in BWR and PWR Fuel Rods
Modern BWR fuel rods are similar to PWR fuel rods except that they are slightly larger and their power densities are
slightly lower. We can use this information to calculate the temperature drop across the fuel and the cladding if we know
the appropriate values for q′. Recalling from our earlier discussion, the temperature drop ΔT across the rod is propor-
tional to its thermal resistance R ROD of the rod. Hence we can write
where RFUEL and RCLAD are the thermal resistances of the fuel and the cladding, respectively. The total temperature drop
across the rod is then
where RROD is the total thermal resistance of the rod. From Table 11.5, we see that q ′PWR ≈ 179W/cm and
q ′PWR MAX ≈ 446 W/cm . Similarly, for a BWR, q ′BWR ≈ 176 W/cm , and q ′BWR MAX ≈ 472 W/cm . Finally, the dimensions for
both types of rods can be found in Table 11.6. Now for a cylindrical rod, the thermal resistance of the fuel and that of
the cladding are given by
( )
RFUEL = 1 ( 4πk F ) = 1/(12.56 × 0.026) = 3.06 cm °C/ W (11.51)
RCLAD PWR = ∆R c 2πR F k c = 0.58/(6.28 × 4.10 × 0.141) = 0.16 (cm °C/ W) (11.52)
RCLAD BWR = ∆R c 2πR F k c = 0.70/(6.28 × 4.38 × 0.141) = 0.18 (cm °C/ W) (11.53)
Thus, the temperature drops across the fuel pellets in a PWR and a BWR are
∆TFUEL PWR = q ′PWR RFUEL = 179 × 3.06 = 548°C (average rod) and 1,365°C (hottest rod) (11.54)
∆TFUEL BWR = q ′BWR RFUEL = 176 × 3.06 = 539°C (average rod) and 1,444°C (hottest rod) (11.55)
TABLE 11.6
Fuel Rod Sizes in Several Common Reactor Types
∆TCLAD PWR = q ′PWR RCLAD = 179 × 0.16 = 29°C (average rod) and 71°C (hottest rod) (11.56)
∆TCLAD BWR = q ′BWR RCLAD = 176 × 0.18 = 31.7°C (average rod) and 85°C (hottest rod) (11.57)
The total fuel rod temperature drops are therefore the sum of the individual temperature drops, which is given by the
following equations:
∆TROD = ∆TFUEL + ∆TCLAD = 577°C (average rod) and 1,436°C (hottest rod) (for a PWR) (11.58)
∆TROD = ∆TFUEL + ∆TCLAD = 571°C (average rod) and 1,529°C (hottest rod) (for a BWR) (11.59)
Thus, the temperature drops across the fuel rods in both reactors are about the same. Of course, we have neglected to
include the fuel–cladding gap in our calculations. The temperature drop across the gap is discussed in Section 11.32.
11.14
Boundary Conditions Used by the Heat Conduction Equation
Boundary conditions are auxiliary conditions that are required to obtain an actual solution to the heat conduction
equation. In general, these boundary conditions consist of
1.
Steady-state boundary conditions, which are used to specify the conditions for steady-state operation (i.e., when
the heat flux q″ or the temperatures on the boundaries of the rod are not changing with time).
2.
Transient boundary conditions that are needed when either the heat flux q″ or the temperatures on the boundaries
of the rod are time dependent.
Both types of boundary conditions present unique challenges to the developer of a nuclear power plant, and it is impor-
tant to get the boundary conditions correct so that the temperature profiles and the heat fluxes that are deduced from
them are also correct. This is probably the primary source of manual error in the design of nuclear systems today. Other
sources of error are introduced by uncertainties in the few group cross sections and in the manufacturing process.
However, these errors are normally accounted for using what are called nuclear hot channel factors. The use of these
factors is described in Lamarsh and Baratta (2001); Todreas and Kazimi (2008).
In addition to steady-state and transient boundary conditions, there are also physical boundary conditions that need
to be applied to the heat conduction equation when we attempt to find a solution. For example, if we would like to find
the temperature distribution in a single homogeneous region, or in a set of regions that are separately homogeneous, but
have different material properties (such as the density, the thermal conductivity, and the specific heat) then we need to
specify some sort of boundary condition at the interfaces, or the boundaries between each of these material regions, and
some type of physical boundary condition is obviously required to do this. In steady-state heat conduction problems,
there is a choice of two physical boundary conditions that can be used at each material boundary or material interface.
The first one is called a temperature boundary condition because it requires us to specify the temperatures at each of
these interfaces. The second one is called a heat flux boundary condition because it requires us to specify the heat fluxes
across each of these interfaces.
Finally, a third type of boundary condition can be used when a problem exhibits some form of physical symmetry.
This latter boundary condition is called a symmetry boundary condition, and it can be used when the temperature profile
must be symmetric about a given point in a fuel rod (such as along the centerline). Then the radial temperature profile is
actually flat at the centerline (at r = 0), and so at this point, we can require that dT/dr = 0 (see Figure 11.6)—and this is
the symmetry boundary condition that we seek. In addition, this boundary condition can be used when a fuel rod devel-
ops a central void after the rod has been irradiated for some time. After the void develops, it is appropriate to assume
that the heat flux q″ = −k dT/dr at the inner surface of the void is 0, and hence, the radial temperature gradient dT/dr must
be 0 at r = RV as well. This is illustrated on the left-hand side (LHS) of Figure 11.6.
11.15
I mplementing Different Types of Boundary Conditions
Now let us illustrate how these boundary conditions can be implemented in a real reactor core, In practice, the easiest
boundary condition to use is the temperature boundary condition. In this case, we simply specify the temperatures T1
and T2 at the boundaries of each material region or zone (say at r = r1 and r = r2), and the solution of the steady-state heat
conduction equation is then completely specified. Alternatively, we can specify the temperature at one interface, and the
414 Heat Removal from Nuclear Fuel Rods
FIGURE 11.6 An illustration of how temperature and heat flux boundary conditions can be applied to a nuclear fuel rod.
temperature gradient (which is a symmetry boundary condition) at the other if the physical layout of the fuel rod permits
us to do so. In this case, the first boundary condition might be dT/dr = 0 at r = r1, and the second boundary condition
might be T = T2 at r = r2. The final solution is then completely determined by these boundary conditions because we
end up with two equations (from the boundary conditions) and two unknowns (from integrating the heat conduction
equation). The resulting set of algebraic equations can then be solved for the temperatures at the boundaries. Finally, in
a problem where there are a number of different material regions that are connected to each another, we can implement
a third type of boundary condition that is called the heat flux boundary condition. In this case, the value of the heat
flux across the interfaces is used to connect each adjacent set of material regions together. For example, at the interface
1 between material regions 1 and 2, and interface 2 between two other material regions (say 2 and 3), we can require
that q1′′ = q ′′2 at the first interface and q ′′2 = q 3′′ at the second interface. Hence, the actual heat flux boundary conditions
(see Figure 11.7) become
FIGURE 11.7 A fuel rod consisting of three material regions. In this case, region 3 may be the cladding.
11.16 SOLUTIONS TO THE STEADY-STATE HEAT CONDUCTION EQUATION FOR PLATE-TYPE FUEL RODS 415
and
However, it is important to keep in mind that it does not necessarily follow that the temperature profiles across each
side of the interfaces must be exactly the same. In fact, they generally are not. All that we can say is that the heat flux
q″ that flows across each of the material boundaries must be continuous because the conservation of energy. This
also assumes that we are dealing with a steady state problem in which the heat conduction equation can be solved
analytically. Consequently, the temperature gradients dT/dr1 and dT/dr2 on each side of material interface 1 can be
d ramatically different than one another—particularly if there is a large difference in the thermal conductivities k1
and k 2 across the boundary. You can convince yourself of this by solving Equation 11.60 for dT/dr1 as a function of
dT/dr2. The result is
( )
dT dr1 = k 2 k1 dT dr2 (at r = interface 1) (11.62)
So if the thermal conductivity k1 of material 1 is ten times less than the thermal conductivity of material 2 (k2), then
k2/k1 = 10 and dT/dr1 = 10 dT/dr2. Although this situation is an extreme one, similar situations can occur in practice
where k2/k1 ≈ 2, and under these conditions the temperature gradients on each side of a material interface can be
completely different. Thus, the reader must bear this in mind whenever there are two or more material regions with a
common interface across which heat can flow. As we shall soon see, nuclear fuel rods that have been irradiated for some
time are a classic example of when a heat flux or temperature boundary condition can be used.
11.16
Solutions to the Steady-State Heat Conduction Equation for Plate-Type Fuel Rods
We would now like to turn our attention how a temperature or heat flux boundary condition can be applied to a
nuclear fuel rod. To begin our discussion, suppose that we would like to apply these boundary conditions to the
simplest possible fuel rod that we can imagine—a plate-type fuel rod with no cladding. An example of such a “fuel
rod” is shown in Figure 11.8. The fuel inside of the plate is assumed to be of thickness 2W, and it is also assumed to
be uniformly distributed. Hence, the heat generated within the fuel is generated uniformly at the rate of q‴ BTU/h
ft 3 or q‴ kW/L.
FIGURE 11.8 An example of a plate-type fuel rod without the surrounding cladding. If q‴ is independent of position, then the
temperature profile between the center of the rod and the edges will be quadratic in shape.
416 Heat Removal from Nuclear Fuel Rods
11.17
T he Temperature Profile of a Fuel Pin in a Plate-Type Fuel Rod
When the temperature profile T(x) has reached its steady-state value throughout the fuel element, the time-dependent
terms in the heat conduction equation become negligible. The one-dimensional heat conduction equation that must be
solved in the fuel is then
d 2 T dx 2 = − q ′′′ k f (11.63)
The informed reader may recognize this as the thermal-hydraulic equivalent of Poisson’s equation. Here, k F is the
thermal conductivity of the fuel. Because Equation 11.63 is a second-order differential equation, we need to specify two
boundary conditions to obtain a unique solution. For nuclear fuel rods, the following boundary conditions can be used:
where TMAX is the maximum fuel temperature (which must occur at the centerline in this case), and
This second boundary condition can be deduced from the symmetry of the problem, since there can be no heat flow
q ′′ = − k F dT/dx (11.66)
at the center of the fuel rod in this case. Integrating Equation 11.63 once gives
where A and B are the constants of integration. Setting T = TMAX at x = 0 gives B = T MAX, and because dT/dx = 0 at
x = 0, it immediately follows that A = 0. Hence, the specific solution for the temperature profile within the fuel is
and a picture of this profile is shown in Figure 11.9. Notice that the fuel temperature has a maximum value of TMAX
directly at the center of the rod (where x = 0) and that the fuel temperature falls off quadratically with increasing
distance from the center. At the surface of the fuel (where x = W), the temperature of the fuel is
Now let us see if we can use these equations to determine the total amount of heat that flowing out of the rod. The reader
may recall from Chapter 10 that the total heat production rate is equal to q‴ multiplied by the volume V of the fuel,
which in this vase is V = 2WA, where A is the area of one surface of the fuel. The heat flowing out of one face of the
fuel is therefore
q = q ′′′ WA (11.71)
which is half of the total heat flow rate. A similar result can be obtained directly from Fourier’s law. In this case, the heat
flow per unit area (or the heat flux q″) is given by
q ′′ = − k F dT dx (11.72)
and upon taking the derivative of T at x = W, we find that q″ = q‴ W. The total amount heat flowing out of one side of
the fuel is then
FIGURE 11.9 An example of a plate-type fuel rod covered with a conductive cladding.
which of course is exactly what we would expect. Occasionally, it is convenient to eliminate q‴ from this expression
entirely and write q as a function of the temperature profile that we have just derived:
Solving for q‴ at x = W, we find that q‴ = (TMAX − TSURFACE)2k f/W2. If we substitute this into the equation for the total
heat flow, we find that
where RF = W/2A k F is called the thermal resistance of the fuel. It is interesting to see that these results can be derived
from a simple energy balance on the fuel or from a simple application of Fourier’s law to the fuel rod. The temperature
drop across the fuel (from its centerline at x = 0 to its surface at x = W) is then
and if we know the heat generation rate q = q‴V, all that is needed to find the temperature drop is its thermal resistance RF.
11.18
T he Temperature Profile for the Cladding in a Plate-Type Fuel Rod
Now, let us turn our attention to the temperature profile in the cladding. It should be apparent from our earlier discussion
that the temperature profile must be given by a modified form of Poisson’s equation. In this case, this equation is called
CLAD = 0 . Actually, a small amount of heat is generated in the cladding from
Laplace’s equation. In Laplace’s equation, q ′′′
beta rays and γ-rays, but this amount of heat is generally small compared to the amount of heat generated in the fuel
418 Heat Removal from Nuclear Fuel Rods
d 2 T dx 2 = 0 (11.78)
and all we must now do is to integrate this equation twice, and apply the appropriate boundary conditions. It does not
take a rocket scientist to see that the most appropriate boundary conditions to use are
where TCLAD SURFACE is the outer surface temperature of the cladding, and t is its thickness. Integrating Laplace’s
e quation twice gives
T(x) = Ax + B (11.80)
where A and B are again the constants of integration. Both A and B can be found from the aforementioned boundary
conditions. After applying these boundary conditions, we find that the final expression for the temperature profile in the
cladding is
Notice that the outer surface of the cladding is reached when x = W + t. When this occurs, the temperature profile
reduces to T(W + t) = TCLAD SURFACE. The other intriguing thing about Equation 11.81 is that the temperature profile
in the cladding falls off linearly with increasing distance from the fuel, while in the case of the fuel, the temperature
profile falls off quadratically with the distance from the centerline. In fact, we can clearly see that the presence of the
heat production term q‴ makes a big difference in the form that the final solution takes. The temperature profile for a
plate-type fuel rod is shown in Figure 11.9.
11.19
T he Thermal Resistance of a Plate-Type Fuel Rod
Another important parameter of conductive heat transfer can be derived from a knowledge of the temperature profile.
In Chapter 10, we referred to this parameter as the thermal resistance. Essentially, the thermal resistance can be thought
of as a way to calculate the temperature drop across an object from a rudimentary knowledge of the heat flow q into or
out of an object and its overall geometry. For nuclear fuel rods, the thermal resistance allows us to find the tempera-
ture drop between the center of the rod and its surface without having to solve the heat conduction equation explicitly.
We would now like to demonstrate how the thermal resistance R can be defined for a plate-type fuel rod. To begin our
discussion, consider the equation we just derived for the temperature profile in the cladding:
( )
T(x) = TFUEL SURFACE − TFUEL SURFACE − TCLAD SURFACE (x − W)/ t (11.82)
where t is the thickness of the cladding. Since there are no heat sources within the cladding itself, all of the heat pro-
duced by the fuel must ultimately flow out of the cladding. This rate of heat flow is usually found by multiplying the
surface heat flux q″ by the surface area of the cladding. When we apply Fourier’s law to Equation 11.82, we find that the
total heat flow rate is given by
( )
q = q ′′A = k C A TFUEL SURFACE − TCLAD SURFACE t (11.83)
where kC is the thermal conductivity of the cladding, and t is its thickness (in cm). We can also write this as
But we know from our previous solution to the heat conduction equation for the fuel that
So we can eliminate TFUEL SURFACE from these equations entirely, and express the heat flow rate q in terms of the tem-
perature difference between the center of the fuel and the outer surface of the cladding. When we do so, we find that
(
q = TFUEL MAX − TCLAD SURFACE ) R (11.86)
where now the thermal resistance R is given by
The temperature drop across the fuel and the cladding is therefore
(
∆TROD = ∆TFUEL + ∆TCLAD = q ( RF + RC ) = (q/A) W 2k F + t k C )
(11.88)
(
= q′′ W 2k F + t k C )
Thus, we have found that the thermal resistance of a plate-type fuel rod with two regions can be written as the sum of
two terms—the thermal resistance for the fuel and the thermal resistance for the cladding. Moreover, we have found that
each of these terms consists of the width of each material region (W or t) divided by its effective thermal conductivity
and multiplied by its surface area A. So when the heat produced in the rod passes through a succession of two materials
in series, the total thermal resistance becomes the sum of the thermal resistances for each material. This turns out to be
a very important observation because it means that any geometrical shape where the heat flows through a sequence of
well-defined material regions can be modeled as a serial heat conduction problem where the total resistance is the sum
of the thermal resistances of each of the individual material regions. Then, instead of having to solve the heat conduc-
tion explicitly, we can “stitch together” a number of additional material regions in series if the regions have different
material properties. For example, if the fuel pin consists of two different types of fuel having thermal conductivities k F1
and k F2, and their widths are W1 and W2, the solution to the heat conduction equation for a three-zone fuel rod becomes
(
q = TFUEL MAX − TCLAD SURFACE ) R (11.89)
where the total internal resistance is given by
(
R = (1/A) W1 2k F1 + W2 2k F 2 + t k C (11.90))
and of course, the third zone is the cladding. Hence, the number of material zones can be increased without limit, and
as long as we know what their geometrical and material properties are, calculating the total heat flow rate q becomes a
trivial exercise.
Finally, it should be obvious that if we were to add additional material regions to this problem, then the total thermal
resistance would always increase because each region would have a different value for W or k. In other words, the heat
flow rate q will always be reduced when we add more regions to a serial conductor because the value of R will always
increase for the same temperature drop ΔT between the center of the fuel element and its surface. Now let us turn our
attention to how this process works for a cylindrical fuel rod.
11.20
Solutions to the Steady-State Heat Conduction Equation for Cylindrical Fuel Rods
In most power reactors, the fuel rods are cylindrical rather than rectangular in shape. There are many reasons for this,
but the two principal reasons are that the pipes used to make the cladding are easy to manufacture and their thermo-
physical properties are well known and well understood. Moreover, the fuel pellets are relatively to cast into simple
420 Heat Removal from Nuclear Fuel Rods
shapes (such as cylinders), and because these pellets small diameters (D < 1 cm), we do not need exotic materials (aside
from UO2 or PuO2) to be able to fabricate them. Consequently, let us see how the temperature profiles can be found in
these rods. To do so we have to solve one version of the heat conduction equation
in the fuel and a simpler version of the heat conduction equation (which is similar to Laplace’s equation - see Table 11.2)
C = 0, and two the equations we need to solve are
in the cladding. If there is no heat generated in the cladding, then q ′′′
To simplify our discussion, we will temporarily ignore the conductive gap between the fuel and the cladding and
we have also made the assumption that the heat generation rate in the fuel is spatially uniform. In Section 11.31, we
will present another equation to estimate the temperature drop across the fuel-cladding gap. In Equations 11.92(a) and
11.92(b) the thermal conductivity is again assumed to be independent of temperature.
11.21
T he Temperature Profile of the Fuel in a Cylindrical Fuel Rod
Just as in a plate-type fuel rod, we will assume that our cylindrical fuel rod consists of uranium dioxide (UO2) fuel
p ellets surrounded by a thin layer of cladding. The cladding is normally made from an alloy of zirconium or s tainless
steel, and the most common alloys of zirconium and stainless steel used in reactors today are Zircaloy-2, Zircaloy-4,
SS-304, and SS-316 stainless steel. Cylindrical fuel rods in commercial power reactors are almost always 4 -5 m long.
The steady-state heat conduction equation that needs to be solved in the fuel is
where ∇2 is our old friend, the Laplacian operator. However, in cylindrical coordinates, the appropriate form of the
Laplacian operator becomes
when the temperature is a function of r and z. Equation 11.93 assumes that thermal energy is produced at a constant rate
q‴ in the fuel and that very little (if any) additional heat is produced in the cladding. Moreover, if the fuel rod happens
to be very long and narrow, (which is a common characteristic of most modern nuclear fuel rods), then we can neglect
the variation in the temperature in the axial direction, and we can set ∂ 2T/∂z2 = 0. Finally, if the uranium dioxide in
the radial direction is distributed uniformly across the rod, then k F becomes a simple constant (if it is not temperature
dependent), and the heat conduction equation becomes
Again, this is a second-order differential equation with constant coefficients. If r = 0 corresponds to the center of the
fuel, the appropriate boundary conditions to use for the rod are
In order to solve Equation 11.93, we will also adopt the convention that the outer radius of the fuel is R FUEL and the outer
radius of the cladding is RCLAD. Finally, if the cladding is ΔRCLAD mm thick, then the relationship between the radius of
the fuel and the radius of the cladding is
and the symbols q ′′FUEL and q ′′CLAD will be used to refer to the respective heat fluxes wherever they need to be found.
Integrating the heat conduction equation once gives
where A and B are constants of integration to be determined from the boundary conditions. Using the boundary condi-
tions we discussed previously, it is easy to see that the symmetry condition dT/dr = 0 at the centerline of the fuel (r = 0)
implies that A = 0, and the second boundary condition (T = TMAX at r = 0) implies that B = TMAX. So the exact solution
for the fuel temperature profile is
and it is easy to see that the temperature profile in the fuel declines quadratically with increasing distance from the
center of the rod.
11.22
T he Temperature Profile of the Cladding in a Cylindrical Fuel Rod
If no heat is generated in the cladding, then q ′′C = 0, and the heat conduction equation to be solved for the cladding is
Obviously, the term (1/r) and the value of kC cancel from the LHS, and so we are left with
r dT/dr = A (11.104)
T(r) = A ln r + B (11.105)
( )
(1/r) d/dr k C r dT dr = 0 (11.106)
the general solution for the temperature profile in the cladding must be
T(r) = A ln r + B (11.107)
Now, consider the boundary conditions that need to be applied to the cladding. For the cladding, the most logical
boundary conditions to use are
422 Heat Removal from Nuclear Fuel Rods
If we substitute these boundary conditions into Equation 11.107, the equations for A and B become
( )
A = ( TCLAD − TFUEL ) ln R CLAD R FUEL (11.110a)
( )
B = TFUEL − ( TCLAD − TFUEL ) ln ( R FUEL ) ln R CLAD R FUEL (11.110b)
Substituting these values back into Equation 11.107 and rearranging slightly, we find that the radial temperature profile
in the cladding is
where TFUEL is the temperature at the outer surface of the fuel and TCLAD is the temperature at the outer surface of the
cladding. We also know from our previous definitions that
and
( ) ( )
ln R CLAD R FUEL = ln 1 + ∆R CLAD R FUEL (11.113)
Now, when the cladding is very thin, the value of the natural logarithm becomes
( )
ln 1 + ∆R CLAD R FUEL ≈ ∆R CLAD R FUEL (11.114)
An Alternative Equation for the Temperature Profile when the Cladding Is Thin
( )
T(r) = TFUEL − R FUEL ( TFUEL − TCLAD ) ∆R CLAD ln r R FUEL (11.115)
In other words, Equation 11.115 shows that the radial temperature profile in the cladding falls off logarithmically with
increasing distance from the fuel. Moreover, we find that when r = RCLAD, T = TCLAD and when r = R FUEL, T = TFUEL.
The radial temperature profile is shown in Figure 11.10. Again, this profile is for a cylindrical fuel rod rather than for a
rectangular one.
11.23
T he Thermal Resistance of a Cylindrical Fuel Rod
Now, let us see if we can derive an expression for the thermal resistance R of the fuel and the cladding. Obviously, the
geometry we are dealing with is somewhat more complex, and because of this, we should not expect the results to be
exactly the same as they were for a plate-type fuel rod. However, we can draw upon our previous experience to find the
value of R we need. From Fourier’s law, the total heat q flowing through the cladding is
FIGURE 11.10 A cylindrical fuel rod shown from above and from the side with and without a central void caused by the
restructuring of the fuel. Note that the temperature profile is quadratic in the fuel and logarithmic in the cladding.
and H is the height of the fuel rod in the core. The value of dT/dr in Fourier’s law is
so the equation for the heat flow q at the surface of the cladding (where r = RCLAD) is
where ΔRCLAD is the thickness of the cladding. The thermal resistance of the cladding RC is therefore
since q′ = q/H. Notice that since RCLAD ≈ R FUEL, this can also be written as
and a practical expression for the thermal resistance of the cladding when the cladding is thin compared to the fuel is
Now, let us turn our attention to the thermal resistance of the fuel RF. The rate at which heat flows from the fuel is given
by q = VFUEL q ′′′ = π R FUEL 2Hq ′′′. Solving for q‴ and introducing this into the equation for the temperature profile in the
fuel below
we obtain
where TMAX is the maximum temperature of the fuel, which usually occurs at the fuel centerline. When r = R FUEL, this
expression becomes
and since q′ = q/H, we arrive at a remarkably simple expression for a cylindrical fuel pellet, which is as follows:
and we have made absolutely no restrictive assumptions in arriving at this result. Because heat flows out of the fuel first,
and then out of the cladding, the heat flow in this case is serial, and the thermal resistance R of the entire fuel rod then
becomes the sum of the thermal resistance of the fuel plus the thermal resistance of the cladding:
The Total Thermal Resistance for the Fuel Rod (Fuel + Cladding)
R = RF + RC -> (11.129a)
( )
R = 1 ( 4 πk f ) + 1 ( 2 πk c ) ∆R CLAD R FUEL (11.129b)
Finally, in terms of the overall temperature difference TMAX − TCLAD, the heat flowing out of the rod is
and this can be verified directly by combining the heat conduction equations for the fuel and the cladding together.
11.24
Comparing the Thermal Resistance Terms in a Plate-Type
Fuel Rod and in a Cylindrical Fuel Rod
Armed with the equations we have just derived, we can compare the size of the thermal resistance for both types of fuel
rods. The results of this exercise are shown in Table 11.4. Notice that the thermal resistance for a plate-type fuel rod
11.25 Understanding the Physical Restructuring of Nuclear Fuel 425
TABLE 11.7
A Comparison of the Values of the Thermal Resistance for a Cylindrical and a Planar Geometry
always contains a width divided by an area. Moreover, the thermal resistance of a cylindrical rod is always equal to unity
divided by the thermal conductivity of the fuel k F or the thermal conductivity of the cladding kC. In fact, the only real
difference between these two expressions is that the thermal resistance of the fuel is divided by an additional factor of 2,
but this factor is due to the fact that only one-half of the fuel was considered in our analysis (because we assumed that
the fuel rods were symmetric about the centerline). Hence, the thickness-to-area ratio and the difference in the thermal
conductivities of each region are the only things that separate one solution from the other. As expected, the thermal
resistance has the inverse dimensions of the thermal conductivity (see Table 11.7).
11.25
Understanding the Physical Restructuring of Nuclear Fuel
As nuclear fuel is burned, a fresh uranium dioxide fuel pellet will undergo a number of structural changes if the fuel
temperature is high enough and if the fuel is irradiated for a long enough period of time. Most of the time, this will cause
the fuel pellet to develop what is called a central void or hole at the center of the rod. An example of such a void is shown
in Figure 11.13. The size of this void depends on the temperature of the rod, and this void develops because the density
of the fuel tends to increase over time. At the same time the density increases, the thermal conductivity of the fuel k F
decreases, and the combination of these two factors makes it harder to extract heat from the fuel. The temperature at
which the central void starts to form is sometimes called the sintering temperature.
The shrinkage of the fuel near the centerline may be accompanied by a loss of porosity as well. In light water reac-
tors, this reaction will occur when the fuel reaches a temperature of between 1,600°C and 1,800°C. The resulting region
is sometimes called the equiaxed grain region. If the temperature of the fuel becomes even higher (which is common
in fast reactors), then the fuel may restructure itself into three distinct regions rather than just two. This third region is
sometimes called the columnar grain region. The columnar grain region begins to appear when the fuel temperature
rises above 1,800°C. Another interesting feature of this restructuring is that the theoretical density of the fuel increases
from 90% to 95% to between 97% and 99%. In other words, the pores disappear. In the equiaxed grain region, the theo-
retical density increases to 97%–98%, and in the columnar grain region, the theoretical density increases to between
98% and 99%. Most of this physical restructuring occurs within the first few days of full-power operation. After this
time, additional structural changes occur much more slowly and are much less pronounced (see Figure 11.11).
The fact that the fuel shrinks rather than expands is counterintuitive because most materials tend to expand thermally
as their temperature is raised. The thermal conductivity of the fuel k F is also different after the fuel restructures itself
than before it does. In general, the only way to determine the temperature profile T(r) in a fuel rod after it has been
FIGURE 11.11 A UO2 fuel pin with two and three different material zones caused by the restructuring of UO2.
426 Heat Removal from Nuclear Fuel Rods
restructured is to solve the steady-state heat conduction equation in three separate regions or zones instead of just one.
In the discussion that follows, we will refer to these zones as
1. The columnar zone
2. The equiaxed zone
3. The normal (or unrestructured) zone
Mixed oxide or MOX fuels as they are called, consist of a mixture of uranium dioxide (UO2) and plutonium dioxide
(PuO2). These fuels also tend to shrink and develop a central void after they are irradiated. Now, let us discuss how the
temperature profile changes after this void forms.
11.26
Solutions to the Steady-State Heat Conduction Equation for
a Cylindrical Fuel Rod with a Central Hole or Void
In order to determine the radial temperature profile T(r) in a fuel rod after a central void has formed, we must first
modify the steady-state heat conduction equation to account for the presence of the void. Assuming for the moment that
the thermal conductivity of the fuel remains constant, the heat conduction equation we need to solve is
This is again a second-order differential equation with constant coefficients. However, the boundary conditions must
be changed because the temperature gradient dT/dr will only be equal to 0 at the inner boundary of the central void.
Suppose that this boundary is located at r = RV where the subscript V is now used to refer to the void. Then, the boundary
conditions we need to use to find the temperature profile are
where A and B are the constants of integration. However, when we apply these boundary conditions to the central void,
we find that the values of A and B become
A = q ′′′ R V 2 2k F (11.134a)
( )
T(r) = TMAX − q ′′′r 2 4k F 1 − R V 2 r 2 + q ′′′ R V 2 2k F ln r R V (11.135)
In other words, the temperature drop across the fuel is similar to that for a solid pellet, except for the presence of the
( ) ( )
additional term q ′′′R V 2 2k f ln r R V , which evaluates to zero when r = RV. Since q′ = q ′′′π r 2 − R V 2 for a cylindrical
rod with a central void, an alternative expression for the radial temperature profile is
This solution can be used as long as there is a single average value for kF. Thus the heat transfer coefficient across an annu-
lar pin with a central void is greater than the heat transfer coefficient across a solid pin with the same outer radius. Taking
the heat transfer coefficient for the solid pin to be hSOLID = 4πkF, and the heat transfer coefficient for the annular pin to be
( ) ( )
h ANNULAR = 4 πk F 1 − 2R V 2 r 2 − R V 2 ln r R V (11.137)
we see that
Now, suppose that the radius of the central void is about 10% of the radius of the pin when the fuel rod has been in the
( )
core for some time. Then, (R F/RV) ≅ 10 and 2R V 2 R F 2 − R V 2 ≅ 0.02. Moreover, at the surface of the fuel, (R F/RV) ≅ 10
and ln (R F/RV) ≅ 2.3. The ratio of the heat transfer coefficients for the annular and solid pins is therefore
In other words, the heat transfer coefficient for the annular pin is about 5% greater than the heat transfer coefficient for
the solid pin for the same value of k F and the same outer diameter. If the void grows to 15% of the diameter, then the
heat transfer coefficient becomes about 10% higher. In other words, the heat transfer coefficient for an annular pellet is
between 5% and 10% greater than it is for a solid pellet for the same value of kF. The exact values for these two coef-
ficients are compared when the central void grows to 15% of the pellet radius in Example Problem 11.2. Clearly, this is
an important effect that needs to be taken into account when the fuel is burned.
The calculations we have just performed also imply that the average fuel pin temperature T for the annular pellet
is lower than the average temperature for the solid pellet. However, if the thermal conductivity in each zone becomes
sufficiently different so that we can no longer use a volume-averaged value for k F, then the situation becomes even more
complicated because we must now solve the steady-state heat conduction equations in all three zones:
ZONE 3: k F3 ∇ 2 T + q′′′
3 = 0 (11.140c)
This can become quite messy in practice, and we must use the heat fluxes at the boundaries to join the three s olutions
together. But rather than boring you with all of the details of how this can be done, we will simply illustrate what the
temperature profile looks like in this case. This temperature profile is shown in Figure 11.12. Clearly, the temperature
profile is no longer a simple quadratic function or r. In fact, solving the steady state heat conduction equations with three
material zones results in a fuel pin temperature T = T MAX at r = RV, which is about 20°C lower than with two material
zones.
FIGURE 11.12 The radial temperature profile in a reactor fuel pellet that has been restructured into three separate zones.
11.27
How the Thermal Conductivity of Uranium Dioxide
Changes with Temperature and Burnup
Now, let us discuss how the thermal conductivity of uranium dioxide changes with time. Figure 11.13 shows how the
thermal conductivity of UO2 (in W/cm °C) is affected by the burnup. This figure shows that the thermal conductiv-
ity falls by as much as 50% as the porosity of the fuel changes and fission gases begin to accumulate in the surviving
pores. Eventually, the thermal conductivity of UO2 converges to an asymptotic value of about 0.02 W/cm °C at very high
temperatures and high burnups. This is the standard value of the thermal conductivity that is used in most Final Safety
Analysis Reports (FSARs). The volume VVOID of the central void depends on the operating temperature of the fuel and
the amount of radiation it receives. For PWRs and BWRs, a nuclear fuel supplier will generally warrant the thermal
and structural performance of the fuel up to a burnup of about 40,000 MWD/ton. This practice depends on the fuel sup-
plier, and in addition, different fuel suppliers may use different experimental data to justify their claims. If the thermal
conductivity falls, the fuel will run hotter for the same volumetric power density q‴, and great care must be taken to
ensure that the fuel will not melt. In general, each reactor operator is required to present an analysis to the Nuclear
Regulatory Commission (or NRC) in the plant FSAR that demonstrates that this is true. Of course, these conclusions
require considerable knowledge of how the thermal conductivity changes as a function of burnup and as a function of
time. In general, there is also no consensus as to when the equiaxed and columnar grain regions will begin to appear in
slightly irradiated fuel. The sintering temperatures and the pellet densities that several fuel suppliers use are shown in
Table 11.8. The spread in this data is probably attributable to slight differences in the design of the fuel pins themselves.
FIGURE 11.13 An example of how the thermal conductivity of a UO2 fuel pin changes as a function of the burnup. (Adapted from
Todreas and Kazimi—Nuclear Systems, Second edition.)
TABLE 11.8
The Fuel Transition Temperatures and Densities Used by Different Nuclear Fuel Suppliers
Solution From Table 11.8, the fuel in the equiaxed grain zone will shrink by 1%–2% as the fuel is irradiated, and the
fuel in the columnar grain zone will shrink by 3%–5% as the fuel is irradiated. Hence, the fuel pellet stack will shrink
to 0.98–0.99 of its original size in the equiaxed region and to 0.95–0.97 of its original size in the columnar region. If the
fuel pellet stack is 3 m long, the total shrinkage will be 0.02 × 3 m = 6 cm in the equiaxed zone and 0.05 × 3 m = 15 cm
in the columnar zone. This compaction may put additional stress on the cladding that can sometimes cause the pin to
fail prematurely. [Ans.]
11.28
T he Effects of Operation and Irradiation on a Nuclear Fuel Rod
After a nuclear fuel rod has been in operation for several weeks, its material structure changes and the fuel begins to
expand and contract (i.e., it shrinks or swells). After only a couple of days, small cracks begin to develop in the fuel pel-
lets, and over longer periods of time (say weeks and sometimes months), the fuel will also densify (become structurally
denser) as its internal molecular structure begins to change. Normally, there are three distinct phases that a light water
reactor fuel rod goes through as it is burned. These phases are outlined below. Fission gases are also produced in the
fuel while this restructuring occurs.
FIGURE 11.14 A picture showing how the fuel in a typical nuclear fuel rod restructures itself as it is burned. Pictures provided by
the US Department of Energy.
TABLE 11.9
The Physical Properties of Uranium Dioxide (UO2) Fuel
shown in the middle of the figure. The fuel in this state is typically called partially burned fuel. Its thermal con-
ductivity and its ability to transfer heat is also reduced at this time (see Table 11.9).
11.29
Fission Gas Release in a Nuclear Fuel Rod
Usually fission gases that are produced when UO2 or PuO2 is burned are contained within the cladding as long as the fuel
rods continue to maintain their structural integrity. Of course, their structural integrity is a function of their operating envi-
ronment as well as their burnup. When nuclear fuel is first being designed, designers attempt to estimate the number of fission
gases that will be released, and also the space that is required within the fuel to accommodate these fission gases. Normally
fission gases are designed to accumulate in a region immediately above and below the fuel pellets called the fuel rod plenum.
When uranium dioxide fuel is used, some of these fission gases are released at very low temperatures. However, as the
temperature rises, the fuel pellets may undergo a number of structural changes, and additional fission gases may be
released. Generally speaking, the higher the temperature is that the fuel operates at, the greater the probability is that
more fission gases will be released. If the burnup of the fuel is high enough (say greater than about 40,000 MWD/ton),
TABLE 11.10
The Percentage p of the Fission Gases Released from a Uranium Dioxide (UO2) Fuel Pellet at Various Operating Temperatures
during a Fast Reactor Transient
the fission gases trapped in the grain boundaries can suddenly be released during high-speed transients, and this may
cause the rods to overpressurize or break. Most of the time, reactor designers attempt to avoid this problem by simply
building a fuel rod with a larger fission gas plenum, or by increasing the thickness of the cladding that surrounds the
rod. However, increasing the thickness of the cladding causes additional neutrons to be absorbed, and this is usually
undesirable because it reduces the neutronic performance of the core.
The percentage of fission gases that are released depends on the temperature of the fuel. However, for burnups below
about 40,000 MWD/ton, several empirical correlations have been developed in which the percentage of the fission
gas released can be correlated to the fuel pin temperature. An example of a table that summarizes the results of these
correlations is shown in Table 11.10. This table is based on a large amount of empirical data, and it cannot be deduced
from first principles. However, it shows that when the fuel pin temperature exceeds 1,500°C, the fission gas release rate
increases dramatically. Hence, in order to minimize the release of these fission gases, most reactor designers try to keep
the majority of the fuel rods below 1,600°C. Other tables to calculate the percentage of these fission gases released are
available, and most of these tables are based on a more mechanistically correct approach. However, most reactor design-
ers still prefer to use empirical tables when calculating the fission gas release rate.
11.30
Accounting for Density Changes in the Fuel
As we mentioned earlier, the density of the fuel changes over time when it is irradiated. If a hole or central void develops
in a pin, it does so because the density of the fuel increases—and it does not take very much of a density change (see
Example Problem 11.2) for the hole to get large. A secondary effect of this densification is that the thermal conductiv-
ity also changes, and in general, it will decrease over time. Hence, it becomes progressively harder to remove the heat
q = q‴V from the fuel as the fuel is burned. However, there is another important effect that a change in the density can
have. Since the amount of heat produced in a fuel rod is generally a constant if the neutron flux ϕ and the enrichment are
fixed, shrinking the fuel will cause the value of q‴ to increase. This can be easily seen by remembering that q‴ = q/V.
So if q is fixed by the neutronic properties of the core, reducing the volume of the fuel V (because it can shrink) will
cause the value of q‴ to increase. For cylindrical fuel rods, we can estimate the importance of this effect by perform-
ing a mass balance on the volume of the fuel. Since the mass of the fuel must always be conserved, it follows that if the
initial density of the fuel pellet is ρ, and its final density is ρ′, then the total mass of the fuel in the pellet will be given by
M = ρV (11.141)
M = ρπ R FUEL 2H (11.142)
where H is the active core height. Now, consider what happens after the fuel is irradiated for some time at an elevated
temperature above its sintering temperature. A central void will develop in the middle of the fuel, and the void will have
a radius RV. Since the mass of the fuel must still be conserved, we must require that
( )
ρπ R FUEL 2 H = ρ′ π R FUEL 2 − π R V 2 H (11.143)
432 Heat Removal from Nuclear Fuel Rods
( )
ρ′ = ρR FUEL 2 R FUEL 2 − R V 2 (11.144)
since the H’s cancel. Now, since q‴ is proportional to the density of the fuel, the new value of q‴ in the fuel will be equal
( )
to the old value or q‴, multiplied by the factor R FUEL 2 R FUEL 2 − R V 2 . Hence,
NEW = q ′′′
q′′′ OLD ⋅ R FUEL
2
( )
R FUEL 2 − R V 2 (11.145)
and this is the value of q‴ that we must use in the heat conduction equation after a reactor is started up. Finally, because
the value of k F will decrease as the fuel is burned, the value of k F that we must use in the heat conduction equation is
really k F (BU), (where BU is the burnup) and we must look up the value of k F (BU) from a table, and insert new values
of k F into the heat conduction equation as the fuel is burned. So in reality, the analysis we must perform depends in part
on the temperatures that the fuel is to be operated at. If the sintering temperature of the fuel (~1,600°C) is exceeded,
which it usually is, the analysis of the thermal performance of a fuel pin becomes more complicated than if it does not.
The same can also be said about the influence of the fuel–cladding gap. Not surprisingly, this is the next subject that we
would like to discuss.
11.31
Adding a Gap to a Fuel Rod
In almost all commercial power reactors, a small “gap” is intentionally left between the fuel and the cladding to allow
for the thermal expansion and contraction of the fuel. This gap must always be taken into account when calculating how
a fuel rod behaves. As a rule of thumb, the thickness of the gap before the fuel is burned is about one-fifth of the total
thickness of the cladding, or about 0.1 mm. This number can vary slightly from one type of reactor to the next, but for
a modern light water reactor, this is a typical value. The actual cladding and gap thicknesses for several reactor types
are shown in Table 11.5. In most textbooks, this physical gap is referred to as the fuel–cladding gap. Initially, this gap is
filled with pressurized helium gas to improve the flow of heat across the gap. The helium has a thermal conductivity of
about 0.002 W/cm °C. The thermal conductivity of the helium is less than the thermal conductivity of the fuel, and the
thermal conductivity of the fuel is less than the thermal conductivity of the cladding.
Comparing the Thermal Conductivity of the Fuel, the Cladding, and the Gap
k GAP < k FUEL < k CLAD (11.146)
In a fresh fuel rod, kCLAD is about six times higher than k FUEL and k FUEL is about ten times higher than kGAP. In an irradi-
ated rod, kCLAD is about 7 times higher than k FUEL and k FUEL is about 20 times higher than kGAP because the fission gases
that are released have an even a lower thermal conductivity than the helium does. However, other conductive gases such
as nitrogen and carbon dioxide can also be used. Table 11.11 compares the thermal conductivities of these gases. The
internal pressure of the gas inside of the fuel rod is generally set equal to the pressure of the reactor environment in to
which the fuel rod will be placed. For example, in the case of a BWR, this turns out to be about 1,000 PSIA (~7 MPa),
and in a modern PWR, this turns out to be about 2,200 PSIA (~15.5 MPa). Hence, there will always be some type of
“gas” in the fuel–cladding gap. Then as the fuel is burned, up to 30 additional gases will be released into this gap, and
the average thermal conductivity kg of gap will change over time.
TABLE 11.11
The Thermal Conductivity of Some Fission Gases and Fill Gases in Nuclear Fuel Rods
TABLE 11.12
The Cladding Materials Used in Different Reactor Designs
In earlier sections, we ignored the temperature drop ΔTGAP across the fuel–cladding gap when d iscussing the fuel rod
temperature profile. In general, calculating this temperature drop can be a very messy process because the thermal
conductivity kg and the thickness of the gap tG can change as the fuel is burned. Unfortunately, both of these numbers
can sometimes change wildly as the fuel is burned, and they can change differently from one fuel rod to another. They
can also change axially from the bottom to the top of the fuel. This means that we cannot use a mechanistic model to
predict the temperature drop between the fuel and the cladding, and in most cases, we cannot solve Laplace’s equation
In Section 11.32, we will discuss how this can be done. We will do so using the properties listed in Table 11.12.
11.32
Finding the Gap Temperature Drop
After the fuel rods have been in the core for some time, the gap heat transfer coefficient changes because the fuel rods
undergo a number of structural and physical changes as the fuel is burned. In fresh fuel rods, the gap between the fuel
and the cladding is approximately one-tenth of a millimeter (0.10 mm). The gap is filled with helium or another pres-
surized gas to improve the flow of heat across the gap. However, as soon as the fuel begins to burn, it starts to expand
thermally, and as its burnup increases, it begins to restructure. This brings more of the fuel into direct contact with the
cladding, and the thermal expansion rate increases even more. At the same time, additional fission products are pro-
duced in the fuel and its density increases as well. Some of these fission products are fission gases that become trapped
in small gaps, grain boundaries, and cracks. Eventually, the accumulation of these effects results in the gap heat transfer
coefficient increasing to at least twice its original value. This reduces the temperature drop across the gap and makes
it easier to remove heat from the fuel. However, at the same time, the thermal conductivity of the uranium dioxide fuel
falls. To a certain extent, these two factors tend to offset each other. Computer codes such as FRAPCON are used to
model these effects when computing the temperature profiles in actual fuel rods. Small changes take place in the thermal
conductivity of the cladding as well, but they are generally less important than those that occur in the fuel. Now, let us
attempt to find the value of ΔTGAP when we take all of these factors into consideration. To do so, recall that heat flux
across the gap is defined by
or to be more explicit
where ΔTGAP is the temperature drop across the gap, and hGAP is an empirically determined “constant” called the gap
heat transfer coefficient. When defined in this way, the gap heat transfer coefficient has the units of kW/m2 °K in the
SI unit system and W/cm2 °C for most reactor work. In practice, the value of hGAP depends on many factors including
the burnup of the fuel, the materials used, and the contact pressure between the cladding and the fuel pellet. Using
Equation 11.149, it is easy to see that temperature drop is given by
and for a cylindrical fuel rod where we can write q″ = q′/2πRGAP, the expression for the temperature drop across the gap
becomes
The gap heat transfer coefficient hGAP used in the aforementioned examples has a value between 0.6 and 1.2 W/cm2 °C (6 and
12 kW/m2 °C), and its value normally INCREASES as the fuel is burned. The dependence of the gap heat transfer coef-
ficient on the burnup of the fuel is shown in Figure 11.15. In a new fuel assembly where the fuel is fresh, most reactor
vendors assume a value for hGAP of about 6 kW/m2 °C. This results in a temperature drop between the fuel pin outer
surface and the cladding inner surface of about 250°C, which is expressed as
However, when the burnup of a fuel pin exceeds about 15,000 MWD/ton, the value of hGAP DOUBLES and the gap
temperature drop falls to about 125°C (about half of its initial value). Hence, after about a year of full-power operation,
and the temperature drop between the fuel pin outer surface and the cladding inner surface is considerably less. Clearly,
the value of hGAP depends on the value of q′, and this value in turn depends on the values of q″ or q‴. Today, most light
FIGURE 11.15 The gap heat transfer coefficient in a thermal water reactor has a value between 0.7 and 1.1, and its value increases
as the burnup increases. This increase in the value of hGAP is caused by the thermal e xpansion of the fuel.
11.33 Finding the Cladding Temperature Drop 435
TABLE 11.13
The Temperature Drop across the Gap ΔTGAP for Different Values of q′ and hGAP for Both BWR and PWR Fuel Rods at
Different Burnups
Note that the gap temperature drop is affected by the burnup of the fuel. Here, it is assumed that RGAP ≈ RFUEL.
water reactor fuel assemblies operate with a maximum linear power density between 400 and 500 W/cm (40 and 50 kW/m),
and the linear power density is limited to about 600 W/cm (60 kW/m) to keep the fuel from melting. Hence, ΔTGAP can
have a range of values that depend on both the burnup of the fuel and q′. For cylindrical PWR fuel pins with an outer
diameter of 8.2 mm (0.82 cm) and cylindrical BWR fuel pins with an outer diameter of 9.6 mm (0.96 cm), the gap tem-
perature drop is shown in Table 11.13 for representative values of q′. Notice that for very hot fuel rods with large values
of q′, the gap temperature drop ΔTGAP can be even greater (see Figure 11.18). However, in cylindrical fuel rods, the value
of ΔTGAP decreases linearly with the fuel pin diameter. Hence, in Figure 11.18, ΔTGAP is simply represented as a straight
line between the fuel and the cladding.
11.33
Finding the Cladding Temperature Drop
In general, the temperature drop across the cladding is smaller than the temperature drop across the gap, and it is always
smaller than the temperature drop across the fuel.
Comparing the Temperature Drops across the Fuel, the Cladding, and the Gap
∆TCLAD < ∆TGAP < ∆TFUEL (11.153)
The temperature drop across the cladding is the temperature drop between its inner surface located at RCLAD IN and its
outer surface located at RCLAD OUT. When calculating this temperature drop, the radius of the inner surface of the c ladding
can be assumed to be the same as the radius of the outer surface of the fuel because the gap is so thin. The temperature
drop across the cladding is therefore
(
∆TCLAD = q ′ 2πk C ln R C R F )
(11.154)
(for a cylindrical rod)
where RC refers to the outer radius of the cladding and R F refers to the outer radius of the fuel. The cladding temperature
drop can also be written as
(
∆TCLAD = q ′ 2πk C ln ( R F + TC ) R F )
(11.155)
(for a cylindrical rod)
where TC is the thickness of the cladding. Most of the time, the thickness of the cladding is small compared to the
diameter of the fuel and the value of ln((R F + TC)/R F) can be approximated by
The cladding temperature drop for a cylindrical fuel rod can then be written as
In practice, this simplification is sometimes called the thin cladding approximation. Notice from Equation 11.157 that
the cladding temperature drop is directly proportional to the power density q′ and its only d ependence on the geometry
is on the ratio of TC to R F. Thus, if the thickness of the cladding is fixed, the temperature drop decreases as the radius
increases. Now, consider the cladding temperature drop for some representative PWR and BWR fuel rods. The thermal
conductivity of the cladding kC is about 0.141 W/cm °C for a PWR and 0.170 W/cm °C for a BWR. The actual values are
shown in Table 11.14. From Equation 11.157, the temperature drop for a typical BWR fuel rod is
( )
∆TCLAD (BWR) = q ′ 2πk C TC R F = 400 (6.28 × 0.170) ⋅ (0.75/4.38) = 374.67 × 0.171 = 64.2°C
( )
∆TCLAD (PWR) = q ′ 2πk C TC R F = 400 (6.28 × 0.141)(0.57/4.1) = 451.7 × 0.139 = 62.8°C
when q′ = 400 W/cm. Values of the cladding temperature drop for other values of the linear power density are shown
in Table 11.15. Normally, the temperature drop across the cladding is about one-fourth of the temperature drop across
the gap when the fuel is fresh and about half of the temperature drop across the gap when the fuel has been in the core
for about 1 year.
TABLE 11.14
A Summary of the Thermal Conductivities for the Fuel Rod Cladding in Various Reactor Types
TABLE 11.15
The Temperature Drop across the Cladding ΔTCLAD for Different Values of q′ for Both BWR and PWR Fuel Pins
11.34
Finding the Temperature Drop across a Fresh Fuel Pin
The temperature profile changes as the fuel is burned. When the fuel is first inserted into the core, the radial temperature
profile is parabolic
for a cylindrical rod, where k F is the average thermal conductivity of the fuel. In terms of the linear power density
q′ = πrF2 q‴, the temperature drop can be written as
where TMAX is the fuel pin temperature at the centerline. Since the thermal conductivity is temperature dependent, the
average thermal conductivity k F must be found from
kF =
∫ k (T)dT (T
T
F MAX − TF ) (11.160)
where the limits of integration are from T = TF to T = TMAX (see Table 11.15). In other words, the fuel pin temperature
drop is directly proportional to the linear power density q′, but for cylindrical rods, it does NOT depend on the radius
of the fuel. The thermal conductivity for a fresh (i.e., unirradiated) ceramic fuel pellet made from UO2 is typically
about 0.024 W/cm °C. This is rather low by nuclear standards, and in general, it is about a factor of 6 lower than the
cladding. Hence, this lower thermal c onductivity leads to a rather large temperature drop in the radial direction. Using
representative values for q′ of about 400 W/cm, and assuming that k F ≅ 0.024 W/cm °C, the temperature drop across a
fresh, unirradiated UO2 fuel pellet is
Notice that this is several times larger than the temperature drop across the gap or the cladding. Of course, this only
applies to a fresh fuel pin and NOT an irradiated or restructured one. We would now like to consider what happens to
the temperature drop when the fuel is burned.
11.35
Finding the Temperature Drop across an Irradiated Fuel Pin
After the fuel has been irradiated for a while, it begins to crack and restructure. Near the center of the rod, it also begins to
densify. Fission gases start to fill in the cracks, and the average thermal conductivity actually falls across the entire region
occupied by the fuel. The extent of this reduction is a function of temperature and the burnup, but at normal burnups and
operating temperatures, the value of k F can fall by between 10% and 20%. In other words, it is less for an irradiated fuel
pellet than it is for a fresh one. Starting in the 1950s, a number of experiments were performed, most notably at The Idaho
National Engineering Laboratory (INEL) to correlate the reduction in the effective thermal c onductivity k ′F to changes in
the density, the temperature, and the burnup of the fuel. In most experiments, the new thermal conductivity k′ could be
correlated to the old (or unirradiated) thermal conductivity using an empirical relationship of the form
( )
k ′F = k F 1 − A + Bx + Cx 2 (11.161)
where A, B, and C were empirically determined constants. These constants were different for UO2 than they were for
PuO2, and the value of x was found to be a function of the burnup, the density, and the temperature. The temperature
profile across a restructured fuel pin could then be written as
( ) ( )
T(r) = TMAX − q ′ 4 πk F 1 − 2R V 2 r 2 − R V 2 ⋅ ln r R V
(11.162)
(for an irradiated fuel pin with a central void)
438 Heat Removal from Nuclear Fuel Rods
and this equation is valid as long as we can assume a single average value for k F. Hence, the temperature profile is no
longer a simple quadratic function of r; specifically, it falls off logarithmically from the center of the fuel to the edge.
Moreover, the maximum fuel pin temperature TMAX occurs at r = RV and when r = RV, dT/dr = 0. Equation 11.162 also
implies that the thermal resistance for the fuel in a restructured fuel rod with a central void is
( ) ( )
RRESTRUCTURED = 1 4 πk F 1 − 2R V 2 r 2 − R V 2 ln r R V (11.163)
which is different than the standard result for a solid rod, which is given by the following expression:
RSOLID = 1 4 πk F (11.164)
( ) (
The thermal resistance is therefore lower for the restructured pin because the value of 1 − 2R V 2 r 2 − R V 2 ln r R V is )
always less than one. This additional term results in the modified temperature profile shown in Figure 11.16. Notice that
the temperature drop across the pin can be as much as 10% lower for the same value of k F. This means that as long as
k F does not change, a restructured pin will run about 10% cooler than an unrestructured one with the same outer dimen-
sions. Now, consider the effect of k F on the overall temperature drop. Since the thermal conductivity falls as the fuel is
burned, the thermal resistance in the same fuel pellet can actually become higher by a comparable amount. If the linear
power density is unchanged, TMAX now occurs at r = RV rather than at r = 0. Using a value for q′ of about 400 W/cm, and
assuming that k F is 10% lower, or k F ≅ 0.0215 W/cm °C, the temperature drop after the pin has been irradiated in the
core for some time is
( ) ( )
∆TFUEL = q ′ 4 πk F 1 − 2R V 2 r 2 − R V 2 ln r R V (11.165)
FIGURE 11.16 For the same linear power rating q′, a restructured irradiated pin will have a fuel temperature of about 10% lower
than a fresh unrestructured pin. Generally speaking, local temperature peaks in nuclear fuel rods become less pronounced as the
fuel is burned.
11.37 Comparing the Average Fuel Pin Temperatures in Solid and Annular Pins 439
Now, suppose that the radius of the central void is about 15% of the outer radius of the pin when the restructuring is
complete. Then, (R F/RV) ≅ 6.67 and ln (R F/RV) ≅ 1.9. Using a new thermal conductivity of k F ≅ 0.0215 W/cm °C to
account for the additional cracking, densification, and restructuring, we find that
( ) ( )
∆TFUEL = q ′ 4 πk F 1 − 2R V 2 r 2 − R V 2 ln r R V (11.166)
∆TFUEL = 400 ( 4 π 0.0215) × [1 − (0.046)(1.9)] ≅ 1350°C (when the fuel has been burned)
In other words, the temperature drop across the fuel is slightly higher for an irradiated pin (1,350°C vs. 1,325°C) when
the linear power density is 400 W/cm and when the change in the thermal conductivity of the fuel is taken into account!
Moreover, since the value of ΔTGAP falls by about 125°C when the fuel reaches a burnup of 15,000 MWD/ton, a reactor
operator can run the restructured pin about 10% hotter than the fresh pin and still maintain the same surface temperature
at the outer surface of the cladding! Simply put, changes in the thermal conductance of the gap allow a reactor opera-
tor to increase the linear power density by about 10% and still keep the maximum fuel pin temperature T MAX below
1,600°C, which is its normal design limit for steady-state operation. Conversely, if the maximum fuel pin temperature
is required to stay below 1,600°C, we can use some of the equations we have just presented to “reverse e ngineer” the
maximum value of q′ that enables this to happen. Example 11.3 shows how this can be done. In most power reactors,
reactor designers attempt to limit the maximum fuel pin temperature to about 1,600°C during steady-state operation
because this temperature allows most of the radioactive fission products (and the fission product gases) to remain trapped
in the fuel by the UO2 ceramic matrix. This design limit is sometimes violated for brief periods of time, but normally
these violations occur during transients that are very short in duration, and do not allow additional fission gases to be
released. Thus, a temperature limit of 1,600°C is normally associated with 100% full-power operation—at least in most
thermal water reactors.
11.36
Fuel Rod Temperature Limits and Regulatory Requirements
In practice, commercial power reactors are temporarily pushed beyond this design limit for economic reasons or when
overpower transients unintentionally occur. Most overpower transients raise the average power level by an additional
10% or 20%. Thus, for reasonably short periods of time (up to perhaps a day or two), the value of q′ can reach 110%–
120% of its steady-state value. Thus, the fuel pin temperatures can reach 1,800°C or even 1,900°C for very short periods
of time. Even then, the cladding temperatures in a PWR or a BWR cannot be allowed to exceed 1,200°C and no transient
can be intentionally introduced where the fuel in the rods starts to melt. Fresh UO2 melts at about 2,840°C, but its actual
melting point is affected by its oxygen-to-metal ratio and by its plutonium content, which increases as the fuel is burned.
For this reason, most reactor vendors assume that the melting of the fuel occurs at 2,600°C when performing an accident
simulation. In these simulations, it is important to look at the entire temperature profile holistically (both radially and
axially) to see how it behaves. As we mentioned previously, the temperature drops across the individual components of
the rod are directly proportional to the value of q′. Hence, the value of q′ becomes the single most important parameter
in determining how the fuel temperature, the gap temperature, and the cladding temperature behave over very short
periods of time.
11.37
Comparing the Average Fuel Pin Temperatures in Solid and Annular Pins
When the thermal conductivity is the same, an “apples-to-apples” comparison can be made between the average fuel tem-
perature in a solid fuel pellet and the average fuel temperature in an annular fuel pellet having the same outer diameter.
In annular pellets, a small hole is intentionally drilled into the center of the pellet to reduce the thermal stresses and to
limit the temperature that the fuel pin can attain. Cylindrical fuel rods having solid and annular pellets are compared in
Figure 11.17. Starting from the surface of the fuel and working inward, the fuel pin temperature at any radial position is
( )
T(r) = TFUEL + q ′′′ R F 2 4k F 1 − r 2 R F 2 (11.167)
where TFUEL is the temperature at the outer surface. Using the fact that q ′′′ ( π R F 2 ) = q ′ for a cylindrical pin, we can also
write the radial temperature profile as
( )
T(r) = TFUEL + q ′ 4 πk F 1 − r 2 R V 2 (11.168)
440 Heat Removal from Nuclear Fuel Rods
FIGURE 11.17 Examples of conventional nuclear fuel rods using both solid and annular fuel pellets. Notice that annular pellets
reduce the maximum temperature TMAX that the fuel rod can attain for the same linear power rating. In this example, the fuel rod is
only cooled on the outer surface.
where q′ is the linear power density, measured in W/cm or KW/m. The corresponding temperature profile for an annular
pellet with an inner radius of RV is
( ) (
T(r) = TFUEL + q ′′′ R F 2 4k F 1 − r 2 R F 2 + q ′′′ R F 2 2k F ln r R V (11.169) )
where the subscript V again refers to the radius of the void. Since the area of the annular pellet is π r 2 − R V 2 and ( )
( )
q ′ = q ′′′π r 2 − R V 2 , the temperature profile becomes
( ( )) (
T(r) = TFUEL + q ′ 4k F 1 − 2R V 2 r 2 − R V 2 ln r R V (11.170)
)
Now, suppose that we would like to find the average fuel pin temperatures for both types of fuel rods. For a solid pellet,
the average fuel pin temperature is
RF
T SOLID =
∫ 0
2πrT(r) dr π R F 2 (11.171)
or
In other words, for a solid pellet, the average temperature is the mean of the fuel pin centerline and surface tempera-
tures. Now, let us repeat the same exercise for an annular pellet. The average temperature of the annular pellet is
RF RF
T ANNULAR =
∫ 0
2πrT ( r ) dr
∫ π (r
0
2
)
− R F 2 dr (11.173)
11.38 An Overview of the Temperature Drop across a Nuclear Fuel Rod 441
( )) (
RF RF
T ANNULAR =
∫ 0 (
2πrTFUEL + q ′ 4k F 1 − 2R V 2 r 2 − R V 2 ln r R V dr
) ∫ π (r
0
2
)
− R F 2 dr (11.174)
or
Because the term in the braces is always less than one, the average temperature of the annular fuel pellet is less than
the average temperature of the solid fuel pellet. Now, suppose that we would like to compare the average temperatures
for both types of pellets. The ratio of the temperature changes as we move inward from the outer surface of the fuel is
∆T ANNULAR
∆T SOLID ( ( ))
2
(
= 1 – 2R V 2 r 2 − R V 2 − 4R V 4 r 2 − R V 2 ln R V r (11.176)
) ( )
Now, let us evaluate these ratios for voids that are 10% and 20% of the outer fuel pin radius, that is, when r = 10RV and
r = 5RV. When RV = 10% of the outer radius
∆T ANNULAR
∆T SOLID ( ) (
= (1 – 2R V 2 100R V 2 − R V 2 − 4R V 4 100R V 2 − R V 2 ln 1 10
2
) ( )
= 1 − 0.0202 + 0.0004 × 2.3 = 1 − 0.0202 + 0.0009 ≅ 0.98
and when RV = 20% of the outer radius
∆T ANNULAR
∆T SOLID ( ( )) (
= 1 − 2R V 2 25R V 2 − R V 2 − 4R V 4 25R V 2 − R V 2 ln 1 5
2
) ( )
= 1 − 0.083 + 0.007 × 1.6 = 1 − 0.083 + 0.0112 ≅ 0.93
Thus, the average temperatures of the annular pellets are 2% and 7% lower than the average temperatures of a solid fuel
pellet with the same outer radius. In other words, this example demonstrates that annular pellets can sometimes run
cooler than solid pellets do.
11.38
A n Overview of the Temperature Drop across a Nuclear Fuel Rod
Now, let us see if we can use the thermal resistance to calculate the temperature drop across an entire fuel rod that
consists of the fuel pin, the cladding, and the gap. From our previous discussion the thermal resistance of the gap is
and in a serial heat conduction problem (like the one we have with a cylindrical fuel rod), we can calculate the total
thermal resistance of the rod as
( )
RROD = 1 4 πk F + 1 2 πk C ∆R C R F + 1 2 πR GAP h GAP (11.178b)
Now, let us examine the size of each of these temperature drops—ΔTFUEL, ΔTCLAD, and ΔTGAP. To be s pecific, suppose
for the moment that the fuel has a linear power generation rate of q′ ≈ 350 W/cm. This power generation rate is some-
where between the power generation rates of an “average” rod and a “hot” rod in a PWR. The total temperature drop
ΔTROD can be written as
where
( )
∆TFUEL = q ′RFUEL = q ′ 1 4 πk F (11.180a)
( )
∆TGAP = q ′RGAP = q ′ 1 2 πR GAP h GAP (11.180b)
Using representative values for the fuel, the cladding, and the gap, we find that
( ) ( )
RFUEL = 1 4 πk F = 1 12.56 × 0.026 = 3.06°C/W
Thus, we have found what each of the contributions to the total temperature drop can be. In general, the greatest tempera-
ture drop occurs across the fuel, and the next greatest temperature drop occurs across the fuel–cladding gap. Moreover, the
gap temperature drop is about twice the cladding temperature drop. The total temperature drop is shown in Figure 11.19
as a function of the linear heat generation rate q′. Notice that a linear heat generation rate of ~350 W/cm produces a rod
temperature drop of 1,070°C + 135°C + 66°C ≈ 1,250°C. For a PWR coolant temperature of 310°C, this results in a maxi-
mum fuel pin temperature of about 1,560°C. In practice, these numbers depend on the burnup, and as a rule of thumb,
the total temperature drop across a PWR fuel rod can be anywhere from 800°C to about 1,300°C. The actual temperature
profile in the fuel rod then looks like the temperature profile shown in Figure 11.20. Again, the temperature drop across
the fuel is parabolic for a cylindrical rod, logarithmic across the cladding, and linear across the fuel–cladding gap.
11.39
Finding the Operating Temperature of a Typical Nuclear Fuel Rod
In reality, the power shape in a commercial power reactor is never completely uniform. In fact, the power profile can
vary both radially and axially. On average, a well-designed PWR has a peak-to-average power ratio of about 1.3 in the
radial direction and about 1.6 in the axial direction. Thus, if the hottest rod at the core centerline has a maximum fuel
temperature of about 1,600°C, an average fuel rod at the same axial elevation will operate at a peak fuel temperature of
about 1,230°C = 1,600°C/1.3 (see Figure 11.18). The average temperature of the fuel within a pin is then the arithmetic
11.40 EFFECTS OF GAP CLOSURE AND PARALLEL CONDUCTION COEFFICIENT 443
FIGURE 11.18 The individual temperature drops across a nuclear fuel rod as a function of the linear heat generation rate q′.
To find the peak fuel temperature, we would have to add ~310°C to the total rod temperature drop in a PWR.
average of the fuel surface temperature TSURFACE = TFUEL and the maximum temperature TMAX, which is given by the
following equation:
To compute the average temperature, let us first try to estimate the fuel surface temperature. In a commercial PWR, the
coolant at the core centerline has an average temperature of about 300°C. The temperature increases about 75°C as we
move inward across the cladding and 250°C as we move inward across the gap. Thus, the fuel pin surface temperature
for a fresh pin is about TSURFACE = 300°C + 250°C + 75°C ≅ 625°C. The average fuel pin temperature is therefore
As the fuel is burned, the temperature drop across the fuel stays nearly the same but the temperature across the gap falls
by about 125°C because the gap heat transfer coefficient doubles. Then, the average temperature of the fuel is
Thus, the change in the gap heat transfer coefficient reduces the average operating temperature of the fuel by about
60°C. Interestingly enough, the average operating temperature is about one-third or 33% of fuel melting temperature if
a melting point of 2,600°C is assumed. Hence, the fuel pin temperatures in most power reactors are kept well below the
melting point of the fuel (see Figure 11.19).
11.40
T he Effects of Gap Closure and Parallel Conduction
on the Gap Heat Transfer Coefficient
In the real world, calculating the gap heat transfer coefficient (hGAP) can be a very messy business. When the fuel has
been irradiated for some time, it begins to expand, swell, or crack. In addition, because the fuel pellets in the core oper-
ate at different temperatures, they can expand or contract at different rates. In Chapter 10, we attempted to illustrate
some of the ways in which these fuel pellets could become distorted as they burn. We have illustrated this behavior again
in Figure 11.22. Typically, large numbers of fuel pellets are stacked on top of each other to form a “column of fuel”
that is loaded into a fuel rod. Sometimes the height of this column is referred to as the active fuel height. Normally, the
active fuel height is between 3 and 5 m. Within the active fuel height, the enrichment of the fuel is varied in the axial
444 Heat Removal from Nuclear Fuel Rods
FIGURE 11.19 The fuel rod temperatures as a function of the linear power density for a UO2 fuel p ellet with fresh fuel. Irradiation
and cracking then increase ΔT FUEL slightly, but ΔTROD decreases with burnup because ΔTGAP decreases.
direction to reduce axial power peaks. There are between 400 and 800 fuel pellets in a typical fuel rod, and in a well-
designed core, there can be 20 or 30 different types of fuel pellets when different combinations of burnable poisons and
enrichments are used. For more information regarding these pellets, the reader should consult our companion books*
(See Figure 11.20).
Now, consider how we might go about estimating the gap heat transfer coefficient under these conditions. If we take
a look at a large number of fuel rods closely, we find that most fuel rods resemble those shown in Figure 11.23. There are
some places where the fuel touches the cladding, and there are other places where it does not. Because there are several
paths for the heat to flow between the fuel and the cladding, the heat transfer process across the gap is actually a parallel
heat transfer process. If we were to redraw Figure 11.23 somewhat differently, we would arrive at the picture shown in
Figure 11.24. In this picture, some of the heat flows from the fuel to the cladding directly, and the rest of the heat flows
from the fuel to a layer of fission gas and then into the cladding. In other words, there are two parallel heat conduction
paths between the fuel and the cladding. The amount of heat that flows through each of these paths can vary as the fuel
is burned, but there are always two paths through which the heat can flow:
If we were able to take a large enough statistical sample of these rods, and dissect them under a m icroscope, we would
find that the average thickness of the restructured “gap” would be a parameter that we call δ. Hence, the gap heat transfer
* See
(1) An Introduction to Nuclear Reactor Physics, by Robert Masterson, CRC Press, Boca Raton, FL (2017).
(2) Nuclear Engineering Fundamentals: A Practical Perspective, by Robert Masterson, CRC Press, Boca Raton, FL (2017).
11.40 EFFECTS OF GAP CLOSURE AND PARALLEL CONDUCTION 445
process is really equivalent to the electrical circuit shown in Figure 11.25 where there are two thermal resistances in
parallel. Suppose that the first resistance R1 is the thermal resistance between the fuel and the cladding in the vicinity
of the gap and the second resistance R2 is the thermal resistance between the fuel and the gases in the gap. The total
thermal resistance of the gap is then
1 RGAP = 1 R1 + 1 R2 (11.182)
because the heat conduction is parallel (see Table 11.16). Equation 11.182 can also be written as
RGAP = R1 R2 ( R1 + R2 ) (11.183)
TABLE 11.16
The Typical Temperature Drops and Fuel Temperatures in Fresh and Irradiated Fuel Rods that are Used in Light Water Reactors
(or LWRs)
Rod Temperature or Temperature Drop (°C) Fresh and Unirradiated Cracked and Irradiated
Fuel temperature drop ΔTFUEL ≈ 1,400°C ΔTFUEL ≈ 1,470°Cb
Gap temperature drop ΔTGAP ≈ 250°C ΔTGAP ≈ 125°C
Clad temperature drop ΔTCLAD ≈ 85°C ΔTCLAD ≈ 85°C
Rod temperature drop ΔTROD ≈ 1,735°C ΔTROD ≈ 1,680°C
Maximum fuel temperaturea TMAX ≈ 2,065°C TMAX ≈ 2,010°C
Source: Todreas and Kazimi.
a These numbers assume a coolant temperature of 330°C and a heat generation rate of about 450 W/cm. Note that an irradiated fuel
rod has a lower temperature drop than a fresh fuel rod because the gap temperature drop decreases as the fuel is burned.
b However, the temperature drop across the fuel increases due to the lower thermal conductivity of the fuel with burnup.
446 Heat Removal from Nuclear Fuel Rods
In addition, we learned earlier that the heat flow can be modeled as an electrical circuit where the temperature drop
ΔTGAP can be correlated with the heat flow rate q′ and the thermal resistance R, which is given by the following
equation:
In other words, if we can find the correct value for RGAP, we can calculate what the heat transfer coefficient hGAP in the
gap should be. Obviously, this estimate will only be approximately correct, but we would like to go through the process
to illustrate the procedure involved. Then, for the moment, we will assume that
RGAP = R1 R2 ( R1 + R2 ) (11.186)
Fortunately, we have already derived the appropriate expressions for R1 and R2 for a cylindrical fuel rod. The appropriate
expressions to use are
where k F is the thermal conductivity of the fuel, and hGAS is the heat transfer coefficient of the gas in the gap. The value of
R1 represents the thermal contact resistance of the fuel in the gap after the thermal expansion and cracking are complete.
If the thermal conductivity kGAS of the gas filling the gap is known, and the average thickness δ of the gap is known, then
we can also write the thermal resistance of the gas as
( )
R2 = RGAS = 1 2 π k GAS δ (11.188)
Finally, suppose that a fraction of the gap consists of fuel touching the cladding and the remaining fraction consists of
fission gas touching the cladding. Moreover, suppose that we call the first fraction f FUEL and the second f raction fGAS. If
H is the active height of the fuel rod where this conduction can occur, then we must conclude that
So in principle, the thermal resistance of the gap can be expressed as a combination of R1 and R2, adjusted for the afore-
mentioned values of f FUEL and fGAS:
RGAP = R1 R2 ( R1 + R2 ) (11.190)
or
( ) (
RGAP = fFUEL 4πk F fGAS 2 πk GAS δ ) (( f
FUEL ) ( ))
4 πk F + fGAS 2 πk GAS δ (11.191)
The value of kGAS for a filler gas like helium is about 0.002 W/cm °C, and its value increases with about the 0.80th power
of the absolute temperature T; that is,
where different gases have different values for C, and T is measured in °K. However, in practice, one finds that most
other fission gases have lower values for the thermal conductivity than helium does, and the average value of kGAS actu-
ally falls as the fuel is burned (see Figures 11.21 and 11.22).
In particular, xenon, iodine, and krypton have an average thermal conductivity of about 0.00013 W/cm °C.
Unfortunately, this value does not account for surface-related effects (such as the contact pressure P or the average gap
width δ) that also need to be included in the calculation. Hence, the expression we have just derived only predicts some
of the things that actually occur in the gap. Fortunately, there is another expression that can be used to find the “real”
gap heat transfer coefficient without an explicit knowledge of kGAS. This involves an equation called the Calza-Bini
equation, which was first developed in Europe in 1975. The Calza-Bini equation states that
11.40 EFFECTS OF GAP CLOSURE AND PARALLEL CONDUCTION 447
FIGURE 11.21 A fuel pellet with cupped or dished edges is used to prevent radial and axial stresses and to prevent the pellet
from cracking.
FIGURE 11.22 Different types of bowing in an irradiated fuel pellet caused by high burnup and uneven heat generation.
(
h GAP = Bk FUEL k CLAD ( k FUEL + k CLAD ) ⋅ P √ δ ) ( in BTU ft 2
)
h °F (11.193)
where B is an empirical constant, P is the contact pressure between the cladding and the fuel in PSI (where they meet),
and δ is the average thickness of the gap. The gap heat transfer coefficient described by the Calza-Bini equation is
usually expressed in BTU/ft2 h °F and then converted to W/cm2 s °C (see Figure 11.23).
The value of the constant B is B = 2/√10M, where M is called the Meyers hardness number of the softer material.
For a nuclear fuel rod, this number is about 1.3 × 105 PSI for stainless steel and about 1.4 × 105 PSI for Zircaloy-2 or
Zircaloy-4. Normally, the gap heat transfer coefficient increases as the power density increases because there will be
more physical contact between the fuel and the cladding. One final point that must be kept in mind is that the overall
gap heat transfer coefficient hGAP tends to increase as the fuel is burned. This is because the fuel expands as it is burned
448 Heat Removal from Nuclear Fuel Rods
FIGURE 11.23 The fuel–cladding gap in a nuclear fuel rod is not constant, and its width varies as a function of both radial and
axial positions.
and this thermal expansion tends to increase the fraction of the fuel that is in direct contact with the cladding. Thus,
although the conductivity of the filler gas in the gap becomes lower over time (due to fission gas contamination), the
overall gap conductivity becomes higher because HGAS decreases and H F increases. When licensing commercial reac-
tor fuel, the gap heat transfer coefficient is determined using a computer program called FRAPCON. FRAPCON is a
very popular computer program for analyzing the thermal performance of nuclear fuel rods, and how their performance
changes as a function of burnup. Many reactor vendors and nuclear fuel suppliers use a version of FRAPCON called
the FRAPCON-3 to calculate the steady state, thermal–mechanical performance of UO2 and PuO2 fuel rods at different
power levels and burnups. FRAPCON can also be used to calculate the fission gas release rate, the fission gas tempera-
tures and pressures in the fission gas plenum, and the mechanical stresses in many d ifferent types of nuclear fuel rods
as a function of time. These mechanical stresses can ultimately result in the structural failure of a fuel rod if they are
not carefully controlled. These stresses can be estimated using Hooke’s law, which is discussed in great detail in many
material science classes. To learn more about FRAPCON and its applications, the interested reader should refer to the
following URLs:
http://www.nrc.gov/reading-rm/doc-collections/nuregs/contract/cr7022/
http://www.nrc.gov/reading-rm/doc-collections/nuregs/contract/cr7022/v1/cr7022v1.pdf
The U.S. NRC provides the first URL to anyone who is interested in using FRAPCON, and the second URL provides a
very detailed description of how the FRAPCON computer program works. Several other computer programs have been
written over the years to perform similar functions, and the FRAPCON user’s manual discusses the pros and cons of
these competing computer programs in some detail.
11.41
T he Advantages of Annular Fuel Rods
Over the past few years, annular fuel rods have come back into vogue again. The amount of heat that a fuel rod can
safely produce is ultimately a function of the amount of heat that can be removed from the rod. (Otherwise, a point
would be reached where the fuel would melt.) So once the thermal performance of the fuel, the cladding, and the gap
have all been optimized, the only practical ways to remove more heat from the rod is to
1. Increase the surface area of the rod.
2. Use a coolant with a higher thermal conductivity.
3. Increase the flow rate of the coolant over the rod in the hope that the convective heat transfer coefficient will be
increased as well.
11.41 The Advantages of Annular Fuel Rods 449
However, all three of these schemes can become self-limiting in practice because the surface area is fixed by the outer
diameter of the rod. Fortunately, we can solve most of these problems by embracing an alternative fuel rod design
known as an annular fuel rod. In an annular fuel rod, the center of the rod is hollow, so coolant can also flow through
a small hole in the center as well as over the outer surface. A typical rod of this type is shown in Figure 11.26. This
design allows more heat to be removed from the rod than we can remove from a similar cylindrical rod having the
same outer diameter. Two different rods having the same diameter outer D are shown in Figure 11.27. The rod on the
left is a conventional fuel rod, and the rod on the right is an annular fuel rod. Now, suppose that the standard fuel rod
has an outer diameter DOUT and that the annular fuel rod also has an inner diameter DIN. Then, we can solve the steady-
state heat conduction equation in both fuel rods and compare the results. To simplify our analysis, suppose that we can
replace the inner and outer diameters by two different radii called ROUT and R IN. We will also neglect the presence of
the cladding, the gap, and the coolant initially because we already know how they affect the fuel rod temperature drop.
So, the problem we would like to solve is shown in Figure 11.26. The relevant form of the heat conduction equation
we must solve is
(1/r)d/dr(rdT/dr) = − q ′′′ k f
(11.194)
(for a cylindrical rod)
Again, this is a second-order differential equation with constant coefficients. However, the boundary conditions are
different for the cylindrical rod because we can no longer use the symmetry condition that dT/dr = 0 at r = 0. Instead,
we must use the fact that there can be slightly different temperatures TIN and TOUT at r = R IN and r = ROUT. Then, the
boundary conditions that must be applied to the annular rod are
where A and B are constants of integration. However, when we apply these boundary conditions to the inner and outer
surfaces, we find that
These are two linear equations with two unknowns. When we solve for A, we find that
( )
A = TIN + q ′′′ R IN 2 4k F − B + ln R IN (11.198)
When we substitute this into the second equation, the value for B becomes
( )( )
B = TOUT + q ′′′ R OUT 2 4k F + TIN + q ′′′ R IN 2 4k F − B ln R OUT ln R IN (11.199a)
or
( ) ( ( ))
B = TOUT + q ′′′ R OUT 2 4k F + TIN + q ′′′ R IN 2 4k F 1 + ln R OUT ln R IN (11.199b)
{ ( ) ( (
A = TIN + q ′′′ R IN 2 4k f − TOUT + q ′′′ R OUT 2 4k f + TIN + q ′′′ R IN 2 4k f 1 + ln R OUT ln R IN ))} ln R IN (11.200)
450 Heat Removal from Nuclear Fuel Rods
with
TIN + q′′′ R IN 2 4k F
A = ln R IN (11.201)
( ) ( (
− TOUT + q′′′ R OUT 2 4k f + TIN + q′′′ R IN 2 4k F 1 + ln R OUT ln R IN
))
and
( ) ( (
B = TOUT + q′′′ R OUT 2 4k F + TIN + q′′′ R IN 2 4k F 1 + ln R OUT ln R IN ))
and this solution is valid as long as we can assume a single average value for k F. Again refer to Figures 11.24 and 11.25 to
clarify the assumptions we have made. Now, suppose that we can set TIN and TOUT to the same average surface tempera-
ture that we will call TSURFACE, or TS for short. Then, the values for A and B become
{ ( ) (
A = TS + q ′′′ R IN 2 4k f − 2TS + q ′′′ 4k f R OUT 2 + R IN 2 1 + ( ln ( R OUT − R IN )) )} ln R IN (11.202a)
FIGURE 11.24 The flow of heat through a nuclear fuel rod can be treated as a parallel conduction problem with two regions if we
neglect the gap and with three regions if we include the gap.
FIGURE 11.25 The electrical analog of a parallel heat conduction problem with two thermal resistances.
11.41 The Advantages of Annular Fuel Rods 451
and
( )
B = 2TS + q ′′′ 4k f R OUT 2 + R IN 2 (1 + ln ( R OUT − R IN )) (11.202b)
While these expressions are rather messy, at least their final structure is now apparent. Now, let us see what the tem-
perature profile looks like for an annular fuel rod (again refer to Figure 11.26). The radial temperature profile is shown
in Figure 11.27 for a typical commercial PWR. For a linear power output of 250 W/cm, the temperature drop across the
rod is about 800°C for a solid fuel rod and about 600° for an annular one. So the presence of the natural log (ln) in the
temperature profile tends to offset the value of −q‴ r2/4k F, and the net effect is that the temperature profile now peaks
somewhere close to the center of the fuel pellet—although it is offset slightly to the outside. Its exact location can be
found by setting dT/dr = 0 and solving for the value of r. We will leave this as an exercise for the student. An even more
important conclusion that can be drawn from these results is that an annular fuel pin runs about 200°C (about 20%
FIGURE 11.26 An annular fuel rod can have a significantly higher power density than a cylindrical fuel rod under certain conditions.
FIGURE 11.27 A diagram showing the difference between the temperature profiles in a solid fuel rod and an a nnular fuel rod
containing UO2. As a rule of thumb, the annular rod runs about 200°C cooler than a solid fuel rod having the same outer diameter.
This is because it has two surfaces to remove the heat from the fuel.
452 Heat Removal from Nuclear Fuel Rods
cooler) than a solid fuel pin does with the same material properties and the same effective radius. This can be easily veri-
fied by comparing the values of TMAX in each case. This behavior can be an enormous operational benefit during certain
types of reactor transients because it shows that the total power density q‴ can be increased by about 20%, and that under
most conditions, the fuel pin will still not melt! In other words, we can increase the total volumetric power density of
the core by about 20%, or we can make the core about 20% smaller for the same power output. This means that we can
reduce the cost of building a nuclear plant, or we can get more power out of the fuel the plant will use. In either case, an
annular fuel rod allows us to obtain a “win-win scenario” in which the plant is either cheaper to operate or safer to use.
(R ) ( )
2
r= OUT
2
− R IN 2 ln R OUT R IN
r = R IN (( R OUT R IN )
2
−1 ) (
ln R OUT R IN )
2
Finally, does this equation imply that the maximum fuel temperature TMAX is reached closer to the inner or the outer
surface of the fuel in this case?
11.42
Fuel Rods with Axially Dependent Heat Generation Rates
In real reactors, the axial heat generation rate q‴(z) is also a function of core height. As a matter of fact, in most reactors,
it can be anything but uniform. So even if we neglect the conduction of heat in the axial direction (e.g., by neglecting the
term d2T/dz2 in the heat conduction equation), the temperature profile in a nuclear fuel rod can change dramatically as
the axial elevation is changed. In Chapter 5, we found that the axial heat generation rate q‴ in a nuclear fuel rod could
be approximated by the equation
q ′′′(z) = q ′′′
MAX cos(πz /H) (11.203)
where H is the core height (see Figure 11.28). Here, the value of z is usually measured from the midpoint of the core, and
q ′′′
MAX is the maximum power generation rate at the core mid-plane (z = 0). Again, this neglects the effects that the control
rods and enrichment variations can have on the axial power shape. Generally speaking, the radial temperature profile
across a fuel rod T(r) is much steeper than the axial temperature profile T(z) (see Figure 11.27).
FIGURE 11.28 Reactor temperature profiles as a function of axial elevation—assuming no bulk boiling in the core.
11.43 Temperature Profiles in Objects with Exponential Heat Sources 453
As a result of this, most of the heat flows directly to the surface of the rod, and almost no heat flows directly in the axial
direction. This is why most analyses of the radial temperature profile neglect the flow of heat in the axial direction.
A nonuniform power distribution can give rise to a nonuniform temperature distribution along the entire length of
the rod. This temperature distribution will mirror the shape of q‴(z) exactly if there is no coolant flowing over the rod.
So the fuel centerline temperature will have an axial temperature profile that looks like
and the cladding will have a similar shape, but the average cladding temperature will be lower:
See Figure 11.28 to see how both of these shapes compare. Because the heat flow rate q‴(z) in the radial direc-
tion must be constant for a given value of z, the heat flux through the fuel and the cladding must be related in the
following way:
( )
q ′′CLAD (z) = q ′′FUEL (z) R FUEL R CLAD (11.206)
where RCLAD = R FUEL + t, and t is the thickness of the cladding. However, as soon as the coolant begins to flow, some of
this heat is transported away by the coolant, and the axial temperature distribution for both the fuel and the cladding
will be offset from the simple symmetrical distributions that we have been describing. We will have more to say about
this in Chapter 25. Until then, it may be constructive to compare the values of the fuel temperatures and the cladding
temperatures at the core centerline (z = 0). The following example problem helps to illustrate these differences.
q′ ≈ 575 W/cm. The temperature drop across the fuel rod is therefore ΔT ROD = ΔT FUEL + ΔT CLAD. Hence, ΔT ROD = q′
(R FUEL + RCLAD) = q′ × (1/4πk F + (1/2πkC) (TCLAD/R FUEL)) = 575 × (3.06 + 0.165) ≈ 1,855°C. The outer surface tempera-
ture of the cladding is then 2,160°C − 1,855°C = 305°C. This is a reasonable number for a PWR with a high volumetric
power density (refer again to Figure 11.28). [Ans.]
Finally, as soon as the coolant moves past the rod, the axial temperature profile of the fuel TFUEL(z) begins to change.
In particular, it begins to deviate from the axial power profile q‴(z), and when the axial power profile is co-sinusoidal
in shape, the peak fuel pin temperature TFUEL MAX(z) occurs about three-fourths of the way between the bottom and the
top of the core. This can be clearly seen in Figures 11.28 and 11.29. Moreover, the cladding temperature, which is more
tightly coupled to the coolant temperature, peaks close to the top of the core where the coolant temperature is high-
est. This occurs in all power reactors where the coolant does not boil and the coolant flow is vertical. We will derive
explicit expressions for the axial temperature profiles for the fuel and the cladding in Chapter 25. In BWRs, the radial
fuel pin temperature profiles are similar to those in PWRs, but the axial temperature profiles are NOT. This is because
as soon as the coolant begins to boil (about one-quarter of the way up the core), its bulk temperature TBULK remains at
the saturation temperature TSAT until all of the water has been converted into steam. This results in the rod and coolant
temperature profiles shown in Figure 11.29. It also results in different inlet and outlet temperatures to optimize the plant
thermal cycle (see Chapter 9).
11.43
Temperature Profiles in Objects with Exponential Heat Sources
Radiation that is produced in the fuel rods can sometimes make its way to the reactor pressure vessel, or even into the
materials surrounding the pressure vessel. When this occurs, thermal energy can be deposited in the surrounding mate-
rials according to a more or less exponential distribution that is proportional to the intensity Io of the particular type of
454 Heat Removal from Nuclear Fuel Rods
FIGURE 11.29 A comparison of fuel temperatures and coolant temperatures in commercial PWRs and BWRs.
radiation streaming through the core. If the surrounding material is very thick, and the radiation source is very far away,
the heat generation rate will be proportional to an equation of the form
where both Io and μ are constants that are material dependent. For X-rays and γ-rays, μ is called the linear absorption
coefficient, and it has the units of inverse cm (or cm−1). If we substitute this value for q‴ into the steady-state heat conduc-
tion equation, the heat conduction equation in one dimension becomes
d 2 T dx 2 = − I o ke −µx (11.208)
Again, we must be very far from the source of the radiation for this expression to be valid. For the problem shown in
Figure 11.30, the general solution to the heat conduction equation is
( )
T(r) = A + Bx − I o kµ 2 e −µx (11.209)
and the validity of this solution can be found by simply substituting the general solution back into the heat conduction
equation. Here, A and B are constants of integration that are determined by the boundary conditions. Now, suppose
that we have a thermocouple so that we can measure the temperatures on both sides of the radiation shield shown in
Figure 11.30. Assume that these temperatures are T1 at x = 0 and T2 at x = R. Then, the final expression for the tempera-
ture profile is
( )( ( ))
T(r) = T1 + ( T2 − T1 ) x R + I o kµ 2 1 − e −µx − x R 1 − e −µR (11.210)
When this temperature profile is plotted for certain values of R, Io, k, and μ, it can be shown that the temperature near
the center of the material can sometimes rise to a maximum value that is higher than the values of either T1 or T2 if
certain conditions are met.
11.44 Future Trends in Fuel Rod Design and Conductive Heat Transfer 455
FIGURE 11.30 A radiation shield with an exponentially varying heat source. The shield is surrounded by air or water on each side.
In other words, the radiation source can deposit enough energy in a radiation shield for the temperature near the center of
the shield to become higher than the temperatures at either the front or the back! Equation 11.210 is the starting point for
calculating the heat removal rate from pressure vessels and radiation shields in which the incident radiation is absorbed
exponentially. It can also be used to calculate the thermal stresses that are induced in these materials when a nonuniform
temperature profile exists in the radial direction.
11.44
Future Trends in Fuel Rod Design and Conductive Heat Transfer
To increase the average operating temperature without melting the fuel, a great deal of research has gone into alter-
native fuel rod designs over the years. The goal of this research has been to either increase the core power density
q‴ or reduce the maximum fuel pin temperature T MAX for a given power density. Most attempts at increasing the
power density (e.g., how much energy can be extracted from a given volume of the fuel) have involved making
the fuel pins smaller to increase the surface area-to-volume (S/V) ratio so that more moderator or coolant can come
into contact with the hot fuel. This has several practical advantages because more thermal neutrons can leak into the
fuel, and therefore, more power can be extracted from the same volume of fuel. However, this approach eventually
runs into problems because the fuel pins can only be made so small before they begin to lose their structural
integrity.
In recent years, several alternative approaches have been used to increase the power density of the core. For PWRs
and LMFBRs, the most popular approach involves using fuel rods that are annular in shape to extract the heat more
quickly and to expose more of the fuel to the thermal neutron flux. An example of such a fuel rod is shown in Figure 11.27.
The open channel down the center of the rod allows the moderator to flow along the inside wall as well as along the
outside one. This increases the surface area of the fuel rod by roughly 50%, and because the heat can be removed more
easily and more thermal neutrons can leak into the fuel, the amount of power that can be extracted from the fuel without
melting the fuel rod is also about 20% higher. Annular fuel rods will probably be approved for use in American PWRs
in the next couple of years. Several advanced fuel rod designs have also been proposed for BWRs as well. One promis-
ing design uses a cruciform-shaped cross section that is also twisted in the axial direction. A picture of these unusual-
shaped fuel rods is shown in Figure 11.31. After the cladding is fabricated, it is then filled with matching cross-shaped
fuel pellets. This new design, although it looks strange, and is somewhat unconventional, is better than a normal BWR
fuel rod in many respects.
456 Heat Removal from Nuclear Fuel Rods
FIGURE 11.31 On the left—a fuel rod temperature profile generated by a popular conductive heat transfer program, and on the
right—the cladding used in an advanced BWR fuel rod design. (Courtesy of the Department of Nuclear Engineering at MIT.)
First, the cruciform shape of the rod increases the S/V ratio, and the twisted surface causes the water flowing by it
to become more turbulent. This ensures that fresh, cool liquid is constantly brought to the hot surface. When these
strange looking rods are assembled into a fuel assembly, the twisted rods touch each other at regular intervals, so they
support one another without the need for additional grid spacers—the h orizontal plates that conventional fuel rods
pass through every few feet to keep them stable and separated. This reduces the amount of power required to pump
the cooling water through a fuel assembly made up of these rods. In addition, the open channel ensures that the water
and steam can circulate freely around the rods. The increased turbulence also prevents the formation of hot spots on
the surface of the rods.
Simulations of the neutronic and thermal-hydraulic performance of these rods with sophisticated computer pro-
grams like RELAP and TRACE have shown that the combined effect of these proposed improvements can allow a
BWR to operate at power densities that are about 25% higher than they are today. Moreover, during normal operation,
the highest centerline temperature for the fuel is about 960°C, or nearly 225°C lower than the centerline temperature
of an average cylindrical BWR fuel rod. Because both an annual fuel rod design and this unusual design can allow a
reactor to operate at a much higher power density, they have the potential to reduce the cost of electrical power produc-
tion by several percent. This is obviously very attractive to utilities that have existing nuclear power plants and would
like to reduce their future operating costs. In Chapter 12, we will expand upon our initial discussion of this subject
to show how the fuel rod temperatures change when the volumetric heat generation rate q‴ becomes time dependent.
This will lead us to a discussion of the Heisler charts and other tools that can be used to find the time-dependent tem-
perature profiles.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Questions for the Student 457
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York, (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc. New York,
NY (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi (2000).
Reactor Theory (Nuclear Parameters) DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
Web References
http://www.nrc.gov/reading-rm/doc-collections/nuregs/contract/cr7022/v1/cr7022v1.pdf.
http://wwwme.nchu.edu.tw/Enter/html/lab/lab516/Heat%20Transfer/chapter_4.pdf.
http://www.nrc.gov/reading-rm/doc-collections/nuregs/contract/cr7022/.
http://tutorial.math.lamar.edu/Classes/DE/SeparationofVariables.aspx.
17. Write down the general heat conduction equation that can be solved to find the temperature profiles in a nuclear fuel
rod.
18. How is Fourier’s law of conduction related to this equation?
19. How does the thermal conductivity in this equation change as the fuel is burned?
20. Write down an equation that can be used to find the temperature profile in a radiation shield.
21. Write down an expression for the thermal resistance of the fuel for a cylindrical fuel rod. Does the t hermal resis-
tance of the fuel depend in any way on the size of the rod?
22. Write down an equation to find the temperature profile in an object with an exponential heat source.
23. What types of boundary conditions should be applied to this equation?
24. For a nuclear fuel rod, how is the thermal diffusivity defined? In the SI unit system, what units does the thermal
diffusivity have?
25. In Fourier’s law, how is the thermal conductivity defined?
26. In the general heat conduction equation, where does the specific heat appear? In thermal heat conduction problems,
what is the purpose of the specific heat?
27. What are representative values for the thermal conductivity and the specific heat of a uranium dioxide fuel pin?
28. What is the melting point of uranium dioxide in this pin?
29. Do materials that retain heat for a long period of time have higher or lower thermal diffusivities than materials that
do not?
30. Write an expression for Fourier’s law of conduction in both differential and integral forms.
31. What is the difference between the specific heat of a material and its heat capacity?
32. Is the thermal conductivity of zirconium and uranium dioxide temperature dependent?
33. What nuclear particles are responsible for most of the radiative heat transfer that occurs in nuclear power plants?
34. Write an expression for the thermal resistance of a cylindrical fuel rod.
35. Write an expression for the thermal resistance of a plate-type fuel rod.
36. How are the thermal resistances summed across a rod in this case?
37. How is the aggregate thermal resistance defined in a serial heat transfer problem?
38. How is the aggregate thermal resistance defined in a parallel heat transfer problem?
39. Using the thermal resistance for the fuel, the cladding, and the fuel–cladding gap, write an expression for the tem-
perature drop across a nuclear fuel rod.
40. Modify this expression to account for the convective resistance of the outer surface of the cladding.
41. Using these expressions, estimate the temperatures on the inner and outer surfaces of the cladding.
42. Write an expression for the temperature drop between the outer surface of the cladding and the coolant.
43. How does this temperature drop change as the mass flow rate of the coolant increases?
44. What is the purpose of the Calza-Bini equation? To what types of nuclear fuel rods can this equation be applied?
45. What is the primary advantage of an annular fuel rod over a solid one?
46. In what reactor types are annular fuel rods the most appealing to use?
47. Fill in the following sentence with the appropriate word or phrase: An annular fuel pin runs about
degrees cooler than a solid fuel pin does with the same material properties and the same effective radius.
48. At what temperatures do radioactive fission gases begin to be released in large amounts from a uranium dioxide fuel
pin?
49. How does the rate of release of these fission gases affect the temperatures at which fuel pins are operated in PWRs
and BWRs?
50. What is the highest temperature that the fuel is allowed to reach in a thermal water reactor?
51. What is the highest temperature that the cladding is allowed to reach in a thermal water reactor?
52. What is the regulatory purpose for setting these particular limits on the fuel and cladding temperature?
53. When a fuel pin is burned for a while, it restructures itself into three different regions or zones. What are the names
of these regions or zones, and what are their primary physical characteristics?
54. Is the fuel in any of these regions or zones denser than it was before the fuel pin was put into operation?
55. After the densification of a fuel pin occurs, a central void develops at the center of the pin. How large is the volume
of this void compared to the rest of the pin?
56. What boundary conditions must be applied to the pin after it develops this central void in order to find its new
temperature profile?
57. In such a pin, is the temperature gradient higher in the radial or axial directions?
58. How is the conduction of heat in the axial direction taken into account in this case?
59. What is the purpose of the computer program called TRACE?
60. What is the purpose of the FRAPCON computer program?
Exercises for the Student 459
Exercise 11.2
Two identical fuel pins having an outer diameter of 1 cm are being designed for use in a thermal water reactor. One fuel
pin is made up of uranium dioxide, and the other fuel pin is made up of plutonium dioxide. Assuming that the fuel pins
are fresh and the linear power density is 400 W/cm, what is the temperature drop across the pins if the pins are solid and
cylindrical in shape?
Exercise 11.3
In a typical PWR fuel assembly, how large is a PWR fuel rod? What are the typical dimensions for the fuel and the
cladding? If the linear power density is 200 W/cm, what is the temperature drop across the fuel and the cladding? For a
cylindrical fuel rod, does the temperature drop depend in any way on the size of the rod? Assume that the fuel is made
up of uranium dioxide and the cladding is made up of Zircaloy-4.
Exercise 11.4
In a typical BWR fuel assembly, how large is a BWR fuel rod? What are the typical dimensions for the fuel and the
cladding? If the linear power density is 200 W/cm, what is the temperature drop across the fuel and the cladding?
Assume that the fuel is made up of uranium dioxide and the cladding is made up of Zircaloy-2. What is the difference
between Zircaloy-2 and Zircaloy-4?
Exercise 11.5
In a typical LMFBR fuel assembly, how large is an LMFBR fuel rod? What are the typical dimensions for the fuel
and the cladding? How do they compare to those of a PWR or BWR? If the linear power density is 500 W/cm, what is
the temperature drop across the fuel and the cladding? Assume that the fuel is made up of plutonium dioxide and the
cladding is made up of stainless steel.
Exercise 11.6
Compare the average power output of a PWR fuel rod to that of a BWR fuel rod. Which fuel rod has the highest power
output, and which one fuel rod has the lowest? How many fuel rods does a typical PWR fuel assembly have, and how
many fuel rods does a typical BWR fuel assembly have? Approximately how much power does a PWR fuel assembly
generate, and how much power does a typical BWR fuel assembly generate? What is responsible for the difference,
if any, in their total power output?
Exercise 11.7
Write an equation for the temperature drop across the cladding of a solid, cylindrical fuel rod. In a commercial PWR
or BWR, what is the approximate size of this temperature drop? Upon what factors does the temperature drop depend?
Is it larger or smaller than the temperature drop across the fuel–cladding gap?
Exercise 11.8
Using an equation, explain how the temperature drop can be found across a fuel pin when the thermal conductivity
is temperature dependent. What is the average thermal conductivity of uranium dioxide and plutonium dioxide in a
thermal water reactor? Does the thermal conductivity increase or decrease as the temperature is raised?
Exercise 11.9
Write an equation that relates the linear power output to the volumetric power output for a cylindrical fuel rod. If the
volumetric power output and the physical dimensions of the rod are known, derive an equation for the heat flux at the
surface of the rod. For a typical PWR or BWR fuel rod, what is an approximate value for the surface heat flux?
460 Heat Removal from Nuclear Fuel Rods
Exercise 11.10
The fuel–cladding gap can be used to calculate the heat flow out of the fuel by defining a gap conductance hGAP which
appears in the equation
q ′′ = h GAP ∆T
where q″ is the heat flux and ΔT is the temperature drop across the gap. Show that the thermal resistance of the gap is
given by
R GAP = 1 h GAP A
where A is the area of the fuel. What is a representative value for hGAP for fresh PWR or BWR fuel?
Exercise 11.11
Derive an expression for the average temperature of a solid fuel pin and another expression for the average temperature
of an annular fuel pin. Express the ratio of these two temperatures as a function of the radius RV of the central void. In
general, does an annular fuel pellet operate at the higher average temperature or a lower average temperature than a solid
fuel pellet? What advantages and disadvantages does an annular fuel pin have?
Exercise 11.12
Using Fourier’s law, show that the total heat flowing from the cladding to the coolant is given by
q = 2πR c H c k C dT dr
where Hc is the height of the cladding surrounding the rod and Rc is its external radius.
Exercise 11.13
Using the expression in Exercise 11.12, show that the thermal resistance of the cladding is
( )
R c = ln 1 + t R FUEL 2πH c k C
where t is the thickness of the cladding and R FUEL is the radius of the fuel. Show that when t ≪ R FUEL, the thermal
resistance of the cladding becomes Rc = t/2πR FUELHckC. What is the approximate error introduced into the value of the
thermal resistance by using this approximation for a PWR fuel rod? In practice, this approximation is called the thin
cladding approximation.
Exercise 11.14
Write an equation for the temperature drop across the fuel–cladding gap. How does the size of this temperature drop
compare to the temperature drop across the fuel or the cladding? Does the size of this temperature drop increase or
decrease as the fuel is burned? What are the factors responsible for this difference?
12
Time-Dependent Nuclear Heat Transfer
12.1 Time-Dependent Heat Transfer in Nuclear Power Plants
In this chapter, we would like to discuss the subject of time-dependent nuclear heat transfer. For many reactor components, the time-
dependent temperature profiles can be found by solving a simplified version of the general heat conduction equation called Fourier’s
heat conduction equation, which is shown below:
Fourier’s equation can be used when the volumetric heat generation rate is so low that q‴ ≈ 0. Although this is normally not the
case for nuclear fuel rods, solutions to Fourier’s equation can provide valuable insight into what the time-dependent temperature
profiles look like in other reactor components such as the pressure vessel, the control rods, and the steam generators. Fourier’s
equation conserves the flow of heat (or more precisely the flow of thermal energy) through a material object, and it relies on the
fact that the heat flow rate is proportional to the negative gradient of the local temperature. The heat flow rate at a particular point
is then given by
where A is the surface area of the object, k is the thermal conductivity in W/cm °C, and dT/dr is the radial temperature gradient.
Equation 12.2(a) is sometimes called Fourier’s heat flow equation. If we divide both sides by A (see Chapter 10), we then obtain the
more familiar equation for the nuclear heat flux:
In the SI unit system, the heat flow rate q is measured in J/s or W and the heat flux is measured in J/cm2–s or W/cm2. In reactor work,
the total heat flow rate is usually measured in kW or MW. This is because the heat flow rate through most reactor components is
very high.
12.2
Boundary Conditions and Fourier’s Equation
Solutions to Fourier’s equation (Equation 12.1) require boundary conditions to be specified on the surface of the object where the
temperature profiles are to be found. The total heat generation rate Q is then the volumetric heat generation rate q‴ multiplied by the
volume V of the object, which is given by the following equation:
Fourier’s equation can be used to find the time-dependent temperature profiles in reactor components with relatively simple shapes,
but it is nearly impossible to solve analytically when these shapes become large or complex. Under these conditions, time-dependent
heat transfer calculations are performed by assuming that each reactor component can be approximated by a large monolithic object
with a uniform internal temperature. Under these conditions, we can neglect the temperature gradients at the object’s surface, and
we can predict how the average temperature of the object behaves. This is done by performing a time dependent energy balance on
the entire object.
461
462 Time-Dependent Nuclear Heat Transfer
12.3
A Lumped Parameter Approach to Nuclear Heat Transfer
In reactor work, the average temperature of a reactor component can be found using what is called a lumped param-
eter approach. In a lumped parameter approach, an energy balance is performed on the entire component, and, the
rate of change of the internal energy of the component dU/dt is equated to the rate of heat flow across the compo-
nent’s surfaces. This then allows an expression to be derived for the average temperature T or bulk temperature
of the component. The time-dependent behavior of almost all large reactor components can be found in this way.
12.4
Numerical Solutions to the General Heat Conduction Equation
Finally, when more precise estimates of the time-dependent temperature profiles are required, the general heat conduc-
tion equation (see Chapter 11) must be solved for the temperature profiles instead. The general heat conduction equation
is similar to Fourier’s equation except that an additional term must be added to account for the volumetric heat genera-
tion rate q'''. The general heat conduction equation can then be written as:
where the value of q''' can be both spatially and temporally dependent; that is, q‴ = q‴(r, t). Normally, the volumetric heat
generation rate q‴(r, t) is measured in W/cm3 or kW/m3. In reactor work, the general heat conduction equation must be
solved numerically because exact analytical solutions do not exist. Fortunately, a great number of computer programs have
been written to solve Equation 12.4, and most of these programs are capable of handling complex geometric shapes.
12.5
Time-Dependent Heat Transfer in Nuclear Fuel Rods
In many textbooks, the subject of time-dependent nuclear heat transfer is not discussed at all because solutions to the
general heat conduction equation do not exist in closed form. However, there are a few special situations where approxi-
mate solutions can be found under ideal conditions. Normally, these situations involve a rapid increase or decrease in the
power level and a uniform volumetric heat generation rate q‴. A scenario such as this is shown in Figure 12.1.
FIGURE 12.1 An illustration of how quickly the reactor power level can rise and fall with a rapid reactivity reduction or a rapid
reactivity insertion. A rapid stepwise increase in the reactivity will lead to what is called the “prompt jump,” and a rapid stepwise
reduction in the reactivity will lead to what is called the “prompt drop.” Both of these scenarios are discussed in our companion
book (see Masterson 2017). The reactivity can be changed by simply inserting or removing control rods from the core.
12.6 The Temperature Profile during a Rapid Power Increase 463
12.6
T he Temperature Profile during a Rapid Power Increase
During very short high-powered transients, static solutions to the heat conduction equation can be used to estimate how
the time dependent fuel pin temperature TFUEL(r, t) behaves. These transients require the power level to increase so rapidly
that there is not enough time for the temperature gradients in the fuel to change. When this occurs, the local temperature
profile T(r, t) becomes relatively independent of the thermal conductivity of the fuel, the density of the fuel, the heat capac-
ity of the fuel, and the convective heat transfer coefficient h. Hence, for very short periods of time, these parameters have
virtually no effect on how hot the fuel can get because the temperature gradients change much more slowly than the rate
at which heat is added to the component. To illustrate what happens next, suppose that the temperature of a cylindrical
fuel pellet before a transient begins is given by
where Rf and Rc represent the thermal resistance of the fuel and the cladding and q‴ is the volumetric heat generation
rate. During very short transients, the maximum fuel pin temperature TMAX will be directly proportional to q‴ (t). When
this occurs, we can write
which implies that the maximum fuel temperature is directly proportional to q‴(t) when the cladding temperature is constant.
Thus if q‴(t) happens to be increasing exponentially with time, then TMAX(t) will increase exponentially with time as well:
o e ( Rf + Rc ) VFUEL (12.7)
λt
TMAX (t) = TCLAD + q′′′
Again, these equations can only be used when the local temperature gradients change more slowly than the volumetric
heat generation rate q'''(t) so the local temperature profiles increase or decrease uniformly by the same amount. The
rate of increase in the fuel pin temperature is then given by the reactor time constant λ, which is used to measure
how quickly the reactor power level rises or falls. Normally, the reactor power level has to double every few seconds
for Equation 12.7 to be valid. The shape of the temperature profile during such a transient is shown in Figure 12.2.
FIGURE 12.2 An illustration of how the reactor temperature profile can change during a very fast transient where
the power level can rise or fall exponentially with time. Here the symbol K refers to the effective multiplication factor,
which is discussed in Masterson (2017).
464 Time-Dependent Nuclear Heat Transfer
Again that this approach can only be used for very fast transients where the temperature gradients do not change their
shape, but instead simply rise or fall uniformly with time. As soon as the shape of these temperature gradients change,
a different approach is required, and normally a computer program is required to determine how hot the fuel will get.
12.7
Fuel Pin Power Decreases
A similar situation occurs when the nuclear chain reaction is temporarily suspended by inserting large numbers
of control rods into the core. These rods absorb excess thermal neutrons and shut the fission process down rather
abruptly. Then, instead of the reactor power level suddenly increasing with time, it suddenly decreases with time, and
if most of the control rods are used, the reactor power falls very rapidly. It may decrease slowly if only one control
rod is inserted into the core, but it may decrease rapidly when the entire core is scrammed (see Chapter 5). The first
scenario is called a planned shutdown, and the second scenario is called an emergency shutdown. In an emergency
shutdown, the power level suddenly decreases in the stepwise manner shown on the left-hand side of Figure 12.1. (The
stepwise increase caused by moving the control rods in the opposite direction is illustrated on the right-hand side.)
The magnitude of this instantaneous power change can be found by using what are called the prompt jump and the
prompt drop approximations, which are discussed in our companion book*. However, after the power level in a fuel
rod suddenly falls to the level predicted by the prompt drop approximation, it then continues to decrease at an entirely
different rate determined by the decay heat produced in the fuel. The decay heat does not stop after a reactor is shut
down. Instead, it continues to be produced in large quantities for at least a week or two after the initial shutdown.
Most of this decay heat is due to radioactive beta decay, and immediately after shutdown at t = to, the amount of decay
heat produced is given by the Wigner–Way equation, which is expressed by the following equation:
Wigner–Way Equation
P(t) = 0.066Po t −0.20 − ( t o + t )
−0.20
(12.8)
where
☉☉ Po is the reactor power before shutdown.
☉☉ to is the time of operation (in s) before shutdown.
☉☉ t is the time (in s) elapsed since shutdown.
☉☉ P(t) is the power generated due to the decay of fission products in the core in the form of beta and gamma rays.
The amount of decay heat produced depends on the power level Po before shutdown. Normally, the Wigner–Way equa-
tion can be used when a reactor is first shutdown, and it is still reasonably accurate for the first year or two after shut-
down, which implies that q‴(t) is proportional to the inverse one-fifth power of the elapsed time when the fission process
ceases. Hence, immediately after shutdown, we can write
where A is a constant on the order of 0.07. In this case, the decay heat can be removed from the fuel rods more slowly,
and conduction and convection do affect the value of the fuel temperature as a function of time. These effects can be
described by the general heat conduction equation, which we will discuss next.
* See for example, our companion book An Introduction to Nuclear Reactor Physics, by Robert Masterson (2017) for more informa-
tion about the relationship between the neutron flux and the reactor power.
12.8 Understanding the Time-Dependent Heat Transfer Equation 465
12.8
Understanding the Time-Dependent Heat Transfer Equation
When the temperature of a fuel rod becomes time dependent, the temperature profile can be found by solving the general
heat conduction equation, which is sometimes called the general transient heat conduction equation. In three dimen-
sions, this equation can be written as
(ρc/k) ∂T/ ∂ t = ∇ 2 T + q′′′ k (12.10)
or
(1/α) ∂T/ ∂ t = ∇ 2 T + q′′′ k (12.11)
where ∇2 is the Laplacian operator and α is a constant called the thermal diffusivity whose value is given by α = k/ρc.
The reader may recall that the Laplacian operator in cylindrical coordinates can be written as
∇ 2 T(r) = ∂ 2 T/ ∂r 2 + (1/r) ∂T/ ∂r + ∂ 2 T/ ∂z 2 (12.12)
∇ 2 T(x, y, z) = ∂ 2 T/ ∂x 2 + ∂ 2 T/ ∂ y 2 + ∂ 2 T/ ∂z 2 (12.13)
Thus, the general heat conduction equations for these specific geometries are
(1/α) ∂T/dt = ∂ 2 T/ ∂r 2 + (1/ r) ∂T/ ∂r + ∂ 2 T/ ∂z 2 + q′′′(r, z) k
(12.14)
(for Cylindrical coordinates)
and
(1/α) ∂T/ ∂ t = ∂ 2 T/ ∂ x2 + ∂ 2 T/ ∂ y 2 + ∂ 2 T/ ∂z 2 + q′′′(x, y, z) k
(12.15)
(for Cartesian coordinates)
where c is the specific heat of the material in which the time-dependent temperature profiles are to be found. (For con-
venience, we have suppressed the subscript P because the temperature change is assumed to occur at constant pressure.)
The thermal diffusivity α which appears in these equations represents the ratio of the thermal conductivity to the heat
capacity ρc. Thus, it measures the ability of a material to conduct heat relative to its ability to store heat and retain it.
In the SI system of units, the thermal diffusivity is given in m2/s. In time-dependent heat transfer problems, materials with
large values of α respond quickly to temperature changes, while materials with small values of α respond more slowly and
will therefore take longer to reach a condition of thermal equilibrium. The value of α for a uranium dioxide fuel pellet is
0.0000072 m2/s. The values of α for other common nuclear fuels and cladding materials are shown in Table 12.1. In gen-
eral, most nuclear fuels have relatively low values for α. This means that they tend to retain heat rather than dissipate it.
TABLE 12.1
The Values of ρ, k, c, and α for UO2, PuO2, SS, and Zirconium—Some Common Nuclear Materials
12.9
Fourier’s Equation for Conductive Heat Transfer
In most cases, the time-dependent heat conduction equation cannot be solved in closed form. However, approximate
solutions can be obtained for a simpler version of the heat conduction equation called Fourier’s equation. Fourier’s
equation is identical to the general heat conduction equation (Equation 12.4) except that in Fourier’s equation, q‴ = 0:
We would now like to present solutions to Fourier’s equation for a long cylindrical fuel rod with no internal heat genera-
tion and for a long plate-type fuel rod with no internal heat generation. For a cylindrical fuel rod where we can neglect
the variation of the temperature in the axial direction, Fourier’s equation becomes
and for a plate-type fuel rod that is also very long relative to its thickness,
(1/α) ∂T/ ∂ t = ∂ 2 T/ ∂x 2
(12.19)
(For a plate-type fuel rod)
In this case, the plate-type fuel rod is narrowest in the x direction (see Figure 12.3).
FIGURE 12.3 Normally, plate-type fuel rods on the right and cylindrical fuel rods on the left are many times longer than their
overall width. In practice, this means that the temperature derivative ∂ 2T/∂z2 in the axial direction can be neglected when calculating
their time-dependent temperature profiles.
12.10 Solutions to Fourier’s Equation for Some Simple Reactor Geometries 467
12.10
Solutions to Fourier’s Equation for Some Simple Reactor Geometries
To solve Fourier’s equation in these geometries, we would like to temporarily ignore the cladding surrounding the fuel.
Moreover, we would like to assume that the fuel rods are solid. If the shape of the fuel pin temperature profile does not
change as a function of time, then we can solve Fourier’s equation using a mathematical “trick” called the separation of vari-
ables method. More information regarding how the separation of variables method works can be found at the following URL:
http://tutorial.math.lamar.edu/Classes/DE/SeparationofVariables.aspx
and in many other locations on the Internet. This particular method assumes that the spatial temperature shape T(r) does
not change as a function of time, but simply rises or falls in a nuclear fuel rod in a uniform way, which can be represented
by a time-dependent term T(t). This means that if we already know what that the temperature profile looks like at t = 0,
for example, T(r), then we can find T(r, t) by simply multiplying this shape by the time-dependent part, which is given by
T(t). In layman’s terms, this is equivalent to writing
Clearly, this is just an approximation to the real solution because the fuel rods continue to generate decay heat for a
very long period of time, but if q‴ is very small, then this turns out to be a reasonable thing to do. This also implies
that if we know the spatial temperature profile T(r) before a reactor is shut down, then we can find T(r, t) by simply
multiplying this known temperature shape by T(t). This physical decoupling of the spatial and temporal terms means
that we can replace Fourier’s equation, which is a very difficult equation to solve, with two simpler differential equa-
tions that can be solved with the knowledge we already possess. To illustrate how this is done, suppose that we substi-
tute T(r, t) = T(r) T(t) into Fourier’s equation and rearrange the remaining terms. The result of this substitution is then
where we have put all of the spatially dependent terms on the left-hand side and all of the temporally dependent terms
on the right-hand side. By now, it should be apparent that at any given point in time, both sides of Equation 12.21 must
be equal to Co, which is just an arbitrary constant. For the sake of argument, suppose that this value is a constant and
that it can have any value we want. In any event, as long as this constant has some finite value, we can replace the time-
dependent heat transfer equation by the following two differential equations:
where we have replaced the value of Co with −λ2 for the sake of convenience. This means that instead of having to
solve one extremely difficult equation (Equation 12.17) for T(r, t), we can now solve two simpler differential equations
(Equations 12.22)(a) and 12.2(b) for T(r) and T(t). In Cartesian coordinates, the general solutions to these equations are
In Equation 12.24(a), Jo is a Bessel function of the first kind, whose characteristic shape is shown in red in Figure 12.4,
and C1, C2, and C3 are arbitrary constants of integration whose values are determined by the boundary conditions.
468 Time-Dependent Nuclear Heat Transfer
FIGURE 12.4 The shape of various Bessel functions as a function of the parameter ro. Note that the first-order Bessel function, Jo,
is shown in red. Its first root is at ro = 2.405. (Picture inspired by Wikipedia.)
The previous URL explains how the values of C1, C2, and C3 can be found. The Bessel functions that appear in Equation
12.24(a) are oscillating and slightly damped functions of the argument ro (not to be confused with the value of r). Bessel
functions such as this were first proposed by German mathematician Friedrich Wilhelm Bessel in the 1700’s, and their
values are tabulated as a function of the argument ro in many technical books.* A more extensive tabulation of these
Bessel functions can be found in the References at the end of the chapter, and in particular Masterson (2017). Table 12.2
shows the values of Jo for various values of the argument ro for a Bessel function of the first kind, and Figure 12.4 shows
the shape of the same Bessel function for values of ro from 0 to 10. In this figure, Jo is shown in red. At ro = 0, Jo = 1, and
at ro = 2.405 (which in our case corresponds to the edge of the cylinder), Jo = 0. These functions then enable us to take
our previous expressions for the time-dependent temperature profiles and write them as
( )
2
For a cylindrical object: T(r, t) = T(r)T(t) = C4 Jo λ r/R e − λ t (12.25b)
where T(r) and T(x) are the aforementioned steady-state solutions and C4 = C1 × C3. Hence, the time-dependent solu-
tions for each geometric shape are just the steady-state solutions T(r) or T(x), multiplied by a time-dependent term T(t),
which depends on the value of λ. For the case of a long solid cylinder, this gives
( )
2
T(r, t) = T(r)T(t) = C4 Jo λ r/R e − λ t
(for a cylinder) (12.26)
where T(x) and T(r) are the steady-state solutions required by the boundary conditions and we are neglecting the tem-
perature variation in the y and z directions. Notice that the value of Jo(λr/R) = 0 when λr/R = 2.405, and R is the outer
radius of the rod. Hence, the term λ(r/R) corresponds to the argument ro in this case; that is, ro = λ(r/R).
* Bessel functions were originally used by Bessel to help solve Kepler’s three laws of planetary motion, but they have been applied to
many applications since then, including the study of vibrating membranes on the surface of a drum and cylindrical heat flow. You
will see them referenced whenever the Laplacian operator in a cylindrical coordinate system is used.
12.11 The Biot Number and the Fourier Number 469
TABLE 12.2
The Shape of Various Bessel Functions as a Function of the Parameter xo
12.11
T he Biot Number and the Fourier Number
Now, let us turn our attention to how the value of λ is found. In general, the size of λ determines how quickly the tempera-
ture changes with time. Clearly, large values of λ will cause the temperature to fall off faster than smaller ones. For plate-
type objects and cylindrical objects, it can be shown that the values for λ are a function of a dimensionless parameter called
the Biot number Bi. The Biot number appears frequently in the study of time-dependent nuclear heat transfer, and it can be
thought of as the ratio of the average temperature drop in a homogeneous object to the temperature difference between the
surface of the object and its environment. The Biot number can therefore be written as
where h is the convective heat transfer coefficient at the object’s surface and L is a characteristic dimension of the object.
For a plate-type fuel rod, L is the width W of the rod, and for a cylindrical fuel rod, L is the rod diameter D.
In other words, the Biot number plays an important role in problems that involve some form of convective heat
transfer. Finally, the Biot number is related to another important dimensionless number called the Fourier number
Fo. The Fourier number can be thought of as a measurement of dimensionless time, and it is given by
where L is again the characteristic dimension of an object. In practice, the Fourier number is sometimes given the
symbol τ.
470 Time-Dependent Nuclear Heat Transfer
12.12
T he Time Constant for Nuclear Heat Transfer
For time-dependent heat transfer problems where the shape of the temperature gradients is not initially known, the value
of λ can be set equal to the product of the Biot number and the Fourier number; that is
where A is the surface area of the component and V is its internal volume. This value for λ is normally derived from an energy
balance between an object and its environment. Notice that the value for λ gets smaller as the value of ρc gets larger because
an object is able to retain heat for a longer period of time. Conversely, if the convective heat transfer coefficient at the surface
of the object is large, the value of λ will also be large and the temperature of the object will fall very rapidly. Thus, the ratio
of h to ρc as well as the surface-to-volume ratio A/V determines how quickly heat can be removed from a reactor component.
An alternative way to find an appropriate value for λ is to consider the explicit shape of the temperature gradients in
the objects we have just discussed. Using Fourier’s equation as our guide, we find that values for λ will be different for
each object because the shape of the temperature gradients is different. For rectangular and cylindrical objects, values
of λ correspond to the roots of the following equations for the Biot number:
In Equation 12.32, J1 is called a Bessel function of the second kind, and it is shown in black in Figure 12.4. Normally,
each of these equations has N roots and these roots correspond to the N eigenfunctions and eigenvalues required for
these equations to balance. However, only the first root (which is given by n = 1) is required to obtain an approximate
solution because the higher order roots corresponding to n > 1 tend to be larger and simply decay away. Thus, when
t becomes large, we are left with a only single value for λ1, and this value determines the asymptotic temperature of
the object. The values for λ1 are given for plates and cylinders in Table 12.3. The resulting equations can then be solved
either graphically or numerically. Graphical solutions are provided by the Heisler charts (see Figure 12.5), which can
be thought of as a graphical analysis tool for the evaluation of heat transfer in thermal engineering. Since M.P. Heisler’s
original charts that were published in 1947, additional charts called Grober charts have become available showing
Q/QMAX as a function of h2αt/k2. These charts first appeared in the literature in the early 1960s. Many time-dependent
heat transfer problems in nuclear science and engineering can be solved with the aid of these charts.
TABLE 12.3
The Time Constants and Biot Numbers for Some Simple Geometrical Shapes
Biot Number Flat Plate Time Long Cylinder Biot Number Flat Plate Time Long Cylinder
(Bi = hL/k) Constant λ1 Time Constant λ1 (Bi = hL/k) Constant λ1 Time Constant λ1
0.01 0.100 0.141 2.00 1.077 1.600
0.02 0.141 0.199 3.00 1.193 1.789
0.04 0.199 0.281 4.00 1.265 1.908
0.06 0.243 0.344 5.00 1.314 1.990
0.08 0.279 0.396 6.00 1.350 2.049
0.10 0.311 0.442 7.00 1.376 2.094
0.20 0.433 0.617 8.00 1.398 2.129
0.30 0.522 0.747 9.00 1.415 2.157
0.40 0.593 0.852 10.00 1.429 2.180
0.50 0.653 0.941 20.00 1.496 2.288
0.60 0.705 1.018 30.00 1.520 2.326
0.70 0.751 1.087 40.00 1.533 2.346
0.80 0.791 1.149 50.00 1.540 2.357
0.90 0.827 1.205 100.00 1.555 2.381
1.00 0.860 1.256 ∞ 1.571 2.405
12.13 Higher Order Solutions to Fourier’s Equation 471
FIGURE 12.5 An example of one of the Heisler charts, which were originally proposed by M.P. Heisler in 1947. This particular
chart shows the time-dependent temperature profiles at various locations within a long homogeneous cylinder.
12.13
H igher Order Solutions to Fourier’s Equation
In Cartesian geometries, any solution to Fourier’s equation can be represented by an infinite series of the form
∑ A n cos ( λ n x ) e λ n (12.33)
2
− τ
T(x, t) =
n
where the summation in Equation 12.33 is to be performed from n = 1 to ∞. Equation 12.33 implies that there are an
infinite number of solutions that look like T(x, t) = A cos ( λx ) e −λ τ and that the solution to any linear time-dependent
2
heat conduction problem can be written as the sum of n of these. In other words, although we implied earlier that the
time-dependent temperature profiles for rectangular and cylindrical objects could be written as
( )
2
T(r, t) = T(r)T(t) = A Jo λ r/R e − λ τ (for long cylinders) (12.35)
∑A cos ( λ n x ) e λ n (12.36)
2
− τ
T(x, t) = n
n
∑A Jo ( λ n r/R) e λ n (12.37)
2
− τ
T(r, t) = n
n
where the index n corresponded to the specific eigenvalues in the expansion. Equations 12.36 and 12.37 can initially
seem rather abstract because it is not immediately obvious what value of n should be chosen to truncate the series
expansions. However, we know intuitively that the more values for n that are used, the more accurate the final solution
will become. Fortunately, it can be shown that Equations 12.36 and 12.37 converge very rapidly to the exact solution as
472 Time-Dependent Nuclear Heat Transfer
t becomes large, and in the limit where Fo > 0.2, keeping the first term and neglecting the remaining terms results in an
average error of about 1.5%. For this reason, it is usually appropriate to use just the first eigenvalue and the first eigen-
function in the Fourier series expansion to obtain a good approximation to the time-dependent temperature profiles.
Fourier’s number Fo declines as the inverse square of an object’s diameter for a cylindrical object, and so when the object
becomes thin, it is almost always the case that Fo > 0.2.
12.14
Time-Dependent Temperature Profiles in Large Reactor Components
Other reactor components including the pressure vessel, the core support plate, and the core barrel are so large that their
average time-dependent temperatures can also be found from a simple energy balance. This approach works best when a
reactor component has a large surface-to-volume ratio (S/V) and a relatively high thermal conductivity k. Then the inter-
nal resistance to heat flow (L/kA) is usually small compared to the convective resistance (1/hA) at the object’s surface.
Then, the temperature gradients at the surface of the object do not affect the average temperature very much, and each
component can be considered to be a single monolithic object with a uniform temperature T. The average temperature
can then be found by performing an energy balance on the object as a whole. This energy balance provides a simple
and convenient way to determine the thermal time constant for the object, and this time constant can be used in turn to
determine how quickly the object heats up or cools down. Consider the hypothetical object shown in Figure 12.6 whose
initial temperature is Ti everywhere, and which is suddenly cooled by a fluid having a different ambient temperature Ta.
The transient temperature response of the object can be found by equating the rate of change of its internal energy dU/dt
to the heat transfer rate at the surface. The heat flow rate between the object and its environment is then
Q = − hA ( T − Ta ) (12.38)
where h is the convective heat transfer coefficient at the object’s surface, A is the surface area of the object, T is the
average temperature of the object, and Ta is the surrounding temperature of the environment, which is sometimes called
the ambient temperature. Remembering that U = ρVcT, the heat that leaves or enters the object is
We can then solve the remaining equation for T(t) if Ti and Ta are known. The result is
( T − Ta ) ( Ti − Ta ) = e − (hA/ρVc)t (12.41)
FIGURE 12.6 An energy balance for a monolithic object used in a lumped parameter approach.
12.14 Time-Dependent Temperature Profiles in Large Reactor Components 473
( T − Ta ) ( Ti − Ta ) = e − λt (12.42)
where λ = hA/ρVc. In practice, T − Ta and Ti − Ta are sometimes written as θ = T − Ta and θi = Ti − Ta in which case
Equation 12.42 becomes
Equation 12.43 can then be used to find the time t required for the object to reach an average temperature T. Alternatively,
it can be used to estimate the temperature the object will reach after some time t. The quantity hA/ρVc that appears in
Equation 12.43 has the units of inverse time and is sometimes called the thermal decay constant λ. The inverse quantity
τ = 1/λ = ρVc/hA has the dimensions of time and is called the thermal time constant τ. Hence, the relationship between
τ and λ is τ = 1/λ, and the value of τ determines how quickly an object will respond to an environmental temperature
that is different than the temperature of the object itself. Notice that the average temperature of the object increases or
decreases exponentially with time when the ambient temperature Ta of its environment is fixed. This is an example of
what is called Newtonian heating and cooling. There are also a number of interesting analogies between the thermal
time constant τ and the time constant governing the flow of current through an electrical circuit, which in electrical
engineering is called the RC time constant. This can be seen by writing
where R is the resistance to convective heat transfer and C is the lumped thermal capacitance of the object in question
(see Figure 12.6). Thus, an increase in the value of R or C will cause the object to respond more slowly to changes in its
environmental temperature, and this will increase the time required for T to approach Ta. This behavior is similar to the
voltage drop that occurs when a capacitor is discharged through a resistor in a RC circuit (see Figure 12.8). The thermal
and electrical analogies that this equation implies are shown in Figure 12.8. The total energy transfer Q that occurs from
the time the ambient temperature is specified until some later time t can be found from
t
Q = hA
∫ θ( t ) dt (12.45)
0
− (hA/ρVc)t
where θ( t ) = T(t) − Ta = θi e . Integrating Equation 12.45, it is easy to see that
Hence as t → ∞, Q → ρVc θi. Thus, the residual heat in the object is eventually discharged into the coolant. Conceptually
this is no different than what happens when the reactor coolant pumps are kept running after the core is shutdown,
and the reactor coolant stays at approximately the same temperature for several days. The average time-dependent
temperatures of many large nuclear power plant components can be modeled in this way. Finally, using the definitions
for the Biot number and the Fourier number that were presented previously, we also see that
( T − Ta ) ( Ti − Ta ) = θ θi = e − Bi ⋅F (12.47)
o
where Fo = αt/L2. Thus, the time-dependent temperatures for these components can be written as a function of the Biot
number divided by the Fourier number Fo. The values of the characteristic length L which the Fourier number uses are
shown in Table 12.4 for some common geometric shapes. As a rule of thumb, the characteristic length is defined as the
volume of the object V divided by its surface area A, and this is given by the following expression:
Thus, the Fourier number signifies how quickly changes in the ambient temperature can penetrate a large monolithic
component of a power plant and cause its average temperature to change. A graphical representation of these tem-
perature changes is shown in Figure 12.7. All large reactor components that obey Newton’s laws of heating and cool-
ing will experience a similar exponential variation in the value of their average temperature with time. For simple
474 Time-Dependent Nuclear Heat Transfer
TABLE 12.4
The Characteristic Length Scales Used to Find the Fourier Number and the Biot Number for a Plate, a Cylinder, a Sphere, and a Cube
FIGURE 12.7 Some typical heating and cooling curves caused by the exponential heating and cooling of a large monolithic
object used in a lumped parameter approach.
FIGURE 12.8 The thermal analogy to an electrical circuit when a lumped parameter approach is used to estimate the time-
dependent temperature of a large monolithic object. Notice that the thermal resistance of the object is given by R = 1/hA and its
thermal capacitance is given by C = ρcV.
12.14 Time-Dependent Temperature Profiles in Large Reactor Components 475
FIGURE 12.9 The Newtonian time constants for some common geometrical shapes. Here, the decay constant is given by λ = Bi Fo,
and it typically assumes a range of values of less than 10 for most geometric shapes.
geometric shapes such as plates, cylinders, spheres, and cubes, this approach results in errors of less than 5% when
Bi is small and especially when Bi < 0.1. The incremental temperature change θ/θi = (T − Ta)/(Ti − Ta) is then plotted
in Figure 12.9 for different values of λ. Notice that cubes and spheres have the most rapid temperature losses because
they have the largest surface-to-volume ratios and infinite cylinders and infinite plates have the slowest. Of course, we
would have expected this behavior based on just physical arguments alone. In conclusion, using the geometry of an
object and its thermal decay constant allows us to predict the value of T − Ti at any future time if the value of Ta − Ti
at a previous time is known. Example Problem 12.2 demonstrates how T(t) can be found from a knowledge of just Ta,
Ti, and λ.
τ = 103.7, so T = 2,000 + (100 − 2,000) 1.031 × 0.000000001 = 1,999.99°C. Thus, the plates leave the oven at essentially
the same temperature as the air in the oven because the Fourier number is so large.
Approach 2: We can also solve this problem using a lumped parameter approach because the Biot number is less than
1.0. The value of λ determined from a lumped parameter approach is then λ = Bi·Fo = hA/ρVcp = 2Ah/ρ2LAcp = h/ρLcp =
476 Time-Dependent Nuclear Heat Transfer
100/(10,970 × 0.005 × 328) = 0.0055/s. The temperature of the plate at t = 30 min = 1,800 s is determined from T = T∞ +
(Ti – T∞) e−λt = 1,000 + (100 − 1,000) e−9.9 = 1,000 + (100 − 1,000) 0.000045 = 1,999.96°C, which is practically identical
to the result obtained using the Heisler charts. Thus, we can use a lumped parameter method with confidence when the
Biot number is sufficiently small. [Ans.]
12.15
Exact and Approximate Solutions to Fourier’s Equation
Exact solutions to Fourier’s equation are available for simple geometric shapes, and approximate solutions are available
for several others. For reactor work, we learned earlier that
∑A cos ( λ n x ) e λ n
2
− τ
T(x, t) = n (for plates) (12.48a)
n
∑A Jo ( λ n r/R) e λ n
2
− τ
T(r, t) = n (for cylinders) (12.48b)
n
∑A sin ( λ n r/R) e λ n
2
− τ
T(r, t) = n (for spheres) (12.48c)
n
are the exact solutions to Fourier’s equation for objects that are spatially uniform. These solutions do not require a
lumped parameter approach to determine the time-dependent temperature profiles within these objects.
However, if the shapes are symmetrical and the boundary conditions are uniform, the temperature gradients at
the center of each shape will be zero, and we can deduce from this that dT/dx = 0 and dT/dr = 0 at x = 0 and r = 0.
Equations 12.48(a), 12.48(b), and 12.48(c) then require different values for λ n and A n for each value of n. The first term
in the solutions (corresponding to n = 1) represents the largest term in the expansions because the other terms, which
contain e −λ 2 , e −λ3 , e −λ 4 , and e −λ N , die away more rapidly over time. Specifically, the eigenvalues in the solution set are
ordered such that λ1 < λ2 < λ 3 … < λ N, where λ1 is the smallest or fundamental eigenvalue of the set. When τ = αt/L2
> 0.2, the higher order terms in the solution become very small and only the first term retains its o riginal value. Thus,
the time-dependent temperature profiles in large, uniform plates, cylinders, and spheres can be approximated by the
equations
( ( ) ) (T − T ) = A cos ( λ1x/L) e λ1
2
− τ
Flat plate or wall : θ( x, t ) = T x, t − Ta i a 1 (τ > 0.2) (12.49a)
( ( ) ) (T − T ) = A Jo ( λ1r/R ) e λ1
2
− τ
Long cylinder : θ( r, t ) = T r, t − Ta i a 1 (τ > 0.2) (12.49b)
where the values for λ1 and A1 are shown in Table 12.6. Notice that the values for A1 and λ1 are only a function for the
Biot number Bi = hL/k in this case, and they do NOT depend on the product of the Biot number and the Fourier number
when τ = Fo > 0.2.
12.16
Finding the Biot Number
In order to use Table 12.6, the Biot numbers in the first column must be evaluated properly. The Biot number for a
plate-type object of thickness 2L is Bi = hL/k, and for a cylindrical object or a sphere of radius R, the characteris-
tic length used to define the Biot number is R. Hence, the Biot number for both cylinders and spheres of radius R is
Bi = hR/k. Once the Biot number is known, and the appropriate values for λ1 and A1 can be found from Table 12.6.
Then the previous equations (Equations 12.49(a), 12.49(b), and 12.49(c)) can be used to find the temperature at any point
inside of a homogeneous plate, cylinder, or sphere. The Biot numbers for these shapes are given in Table 12.5. These Biot
numbers can be applied to any reactor component where there is no significant internal heat generation due to nuclear
fission, radioactivity, or gamma ray absorption. When the value of q‴ becomes significant, other approaches (which are
usually numerical in nature) must be used instead.
12.17 Estimating the Centerline Temperatures in Homogeneous Objects 477
TABLE 12.5
The Biot Number as a Function of the Characteristic Dimension for Some Simple Geometric Shapes
Shape Width (W) Characteristic Dimension Used to Find the Biot Number Value of the Biot Number (Bi)
Flat plate 2L L/2 (semi-thickness) Bi = hL/k
Cylinder R R Bi = hR/k
Sphere R R Bi = hR/k
TABLE 12.6
The Parameters Used to Find the Time-Dependent Temperature Profiles for a Flat Plate, a Cylinder, and a Sphere
12.17
Estimating the Centerline Temperatures in Homogeneous Objects
The centerline temperatures for homogeneous objects with minimal internal heat generation can be found from
Equations 12.49(a), 12.49(b), or 12.49(c) when the value for Bi is known. The values of A1 and λ1 can then be found from
Table 12.6. For plate-type objects and a spherical objects, we know that cos(x) = 1.0 when x = 0 and sin(r)/r = 1.0 when
478 Time-Dependent Nuclear Heat Transfer
r = 0. The time-dependent temperatures at the center of these objects are then given by TCENTER = T(0, t), and the values
of θ(x, t) at the center of these objects become
( ( ) ) (T − T ) = A e λ
2
− 1 τ
Center of a flat plate or wall : θCENTER (x, t) = T 0, t − Ta i a 1 (τ > 0.2) (12.50a)
( ( ) ) (T − T ) = A e λ
2
− 1 τ
Center of a long cylindrical object: θCENTER (r, t) = T 0, t − Ta i a 1 (τ > 0.2) (12.50b)
( ( ) ) (T − T ) = A e λ
2
− 1 τ
Center of a spherical object : θCENTER (r, t) = T 0, t − Ta i a 1 (τ > 0.2) (12.50c)
Here Ti is the initial temperature at the center of each object and the ambient temperature of the coolant Ta surrounding
each object assumed to remain unchanged during the cooling process. Sometimes Table 12.6 is called a transient tem-
perature table because the solutions in this table are based on what is called the one-term approximation. This table has
the advantage that it is easy to read than a Heisler chart. More advanced heat transfer books discuss the derivation of the
Heisler charts and discuss the higher order terms (n > 1) that are required to obtain a more accurate solution. The initial
temperatures Ti at the start of each transient are assumed to be the centerline temperatures when the equations above are
used. Normally, there are three Heisler charts for each geometric shape. The first chart is used to determine the tempera-
ture at the center of each object as a function of time t. The second chart is used to determine the temperature at other
locations within the object in terms of the centerline temperature TCENTER. The third chart can then be used to determine
the amount of heat Q transferred from the object to its surroundings up to a specific time t when Ta is assumed to be
constant. Examples of these charts are presented in many heat transfer books. They may also be found on the Internet at
https://en.wikipedia.org/wiki/Heisler_chart
Small values for the Biot number imply that the internal resistance to the conduction of heat is small relative to the
convective resistance at an object’s surface. Hence, for small Biot numbers, the temperature profile within an object
becomes uniform very quickly, and a lumped parameter approach can be used to determine the time-dependent tem-
perature profile. Conversely, when the Biot number becomes very large, the temperature gradients inside of the object
change more gradually, and a lumped parameter approach cannot be used to determine the object’s internal temperature
profile. Usually, Biot numbers between 0.1 and 1.0 are used to determine the line of demarcation between these differ-
ent approaches. Hence the Bi number can be used to establish the following guidelines for when a lumped parameter
approach can be used and when it cannot:
If Bi < 1.0, Equation 12.30 can also be used to determine the heat conduction time constant
λ = Bi ⋅ Fo = hA/ρVc (12.51)
12.19 Transient Heat Conduction for Multidimensional Shapes 479
that appears in the equations for the time dependent temperature profiles (Equations 12.49(a), 12.49(b), and 12.49(c)),
and if Bi > 1.0, the time constant λ must be found directly from the Heisler charts. Now let us examine the role the
Fourier number τ plays in the Heisler charts. When Bi > 1.0, the time-dependent temperature profiles consist of an infi-
nite number of terms such as those shown below
∑A cos(λ n x)e λ n
2
− τ
T(x, t) = n (for uniform plates) (12.52a)
n
∑A Jo ( λ n r/R) e λ n
2
− τ
T(r, t) = n (for uniform cylinders) (12.52b)
n
∑A sin ( λ n r/R) e λ n
2
− τ
T(r, t) = n (for uniform spheres) (12.52c)
n
However, the first terms in each equation are usually the dominant ones, and these terms correspond to setting n = 1.
Using only the first terms, Equations 12.52(a), 12.52(b), and 12.52(c)) become
T(r, t) = T∞ + ( Ti − T∞ ) A1 Jo ( λ1r/R) e λ1
2
− τ
Long cylinder: (τ > 0.2) (12.53b)
( ) (
T(r, t) = T∞ + ( Ti − T∞ ) A1 λ1 r/R sin λ1 r/R e λ1 )
2
− τ
Spherical object: (τ > 0.2) (12.53c)
where τ = αt/L2. Now, we mentioned previously that the Fourier number τ could be defined by
Thus, the Fourier number measures the amount of heat conducted through an object relative to the amount of heat
stored in the object. Consequently, a large value for the Fourier number indicates that heat propagates relatively quickly
through an object. Now the terms in Equations 12.52(a), 12.52(b), and 12.52(c) converge rapidly when t is large, but when
τ > 0.2, keeping the first term and neglecting all of the remaining terms results in a maximum error of less than 2%.
Thus, when τ > 0.2, the Heisler charts provide excellent predictions of the time-dependent temperature profiles as long
as a object is spatially uniform; that is, spatially homogeneous.
12.19
Transient Heat Conduction for Multidimensional Shapes
Earlier we mentioned that Fourier’s equation
could be thought of as a second order linear differential equation. Because it is linear in space and time, we can find
solutions to Equation 12.54 in one dimension, and then superimpose these solutions on the solutions from other spatial
dimensions. For example, when we apply this approach to a long rectangular bar with a different length in the x and y
directions, the temperature profile T(x, y) can be written as the product of the temperature profiles in the x and y direc-
tions, which are T(x) and T(y):
In cylindrical coordinate systems, a similar equation can be written for cylindrical rods if the radius R and the height
H are known:
These rods are quite similar to the control rods used in pressurized water reactors (or PWRs). Both types of rods are
shown in Figure 12.10. However, instead of separating the solutions to Fourier’s equation into separate spatial and tem-
poral components, we can separate them into separate spatial components for each coordinate direction. For example,
in the case of a sphere or a cylinder, we can write
because each spatial component is independent of the other. From the Heisler charts, we also see that
θ( r, z, t ) = θ( r, t )θ(z, t )
(12.59)
(
θ( r, z, t ) = T(r, t) − Ta ) ( T − T ) ( T (z, t ) − T )
i a a ( Ti − Ta ) (for a solid cylinder)
Finally, for a long rectangular bar with no internal heat generation, we can write
θ( x, y, z, t ) = θ( x, t )θ( y, t )θ(z, t )
( ( ) ) (T − T )(T ( y, t ) − T ) ( T − T )( T (z, t ) − T ) ( T − T )
θ( x, y, z, t ) = T x, t − Ta i a a i a a i a (for a rectangular bar) (12.60)
Obviously, the temperature profiles in other two and three dimensional reactor components can be found in a
similar way.
FIGURE 12.10 Time-dependent temperature profiles in finite and infinite cylinders, and bars can be determined analytically when
the material composition is uniform. The Heisler charts can be applied to these objects when their length is much greater than their
radius or their width and when the ambient temperature Ta of the s urrounding fluid is assumed to be constant.
Bibliography 481
λ1 = 0.28, and τ = 2.5, so T = 1,000 + (100 − 1,000) 1.01 × 0.785 = 286.5°C. Therefore, the centerline temperature of the
rod will drop from 1,000°C to 286.5°C in 30 minutes. [Ans.]
12.20
Numerical Alternatives for Finding the Time-Dependent Temperature Profiles
In reality, most reactor components are too complicated to model using the techniques we have just described. For most
of these components, the time-dependent temperature profiles must be found numerically. In other words, the tem-
perature profiles must be found by solving the general heat conduction equation with a digital computer. A numerical
approach has the advantage that it can be applied to any reactor component having a finite value for q‴. In other words,
the internal heat generation rate q‴(r, t) within the component can be a function of both position and time. A numerical
solution to the heat conduction equation requires converting it into a set of algebraic equations called finite difference
equations. The exact procedures for doing this are discussed in many heat transfer books. A computer program that
solves these equations has the advantage that different boundary conditions and convective heat transfer coefficients
can be applied to different parts of the same object. Moreover, the heat transfer coefficients on one side of the pressure
vessel wall can be completely different than the convective heat transfer coefficients on the other side. This also allows
us to model reactor geometries where
1. The material composition of each component is no longer uniform.
2. There are internal heat sources within the component due to nuclear fission or gamma ray absorption.
3. The internal heat generation rate and the convective heat removal rate are time dependent.
Hence, virtually, all reactor components can be modeled in this way today.
0.02. From Table 12.6, the thermal time constant for the pellet is λ1 = 0.759 and A1 = 1.06. After 30 m inutes
(1,800 s), the Fourier number is τ = αt/R2 = 0.00000072 × 1,800/(0.005)2 = 51.8. At the center of the pellet,
T = T + ( T − T ) A e λ1 = 2,000 + 1.06(30 − 2,000)e −30 = 1, 999.999°C . Hence, the air has hardly cooled the pellet
2
− τ
∞ i ∞ 1
in 30 minutes. [Ans.]
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
482 Time-Dependent Nuclear Heat Transfer
Glasstone, S. and Eedlund, M. C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, CRC Press, Boca Raton, FL (2014).
Lamarsh J. R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., New York,
NY (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi (2000).
Reactor Theory (Nuclear Parameters) DOE-HDBK-1019/2–93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, CRC Press, Boca Raton, FL (2014).
Web References
http://www.iaea.org/inis/collection/NCLCollectionStore/_Public/34/065/34065217.pdf.
http://wwwme.nchu.edu.tw/Enter/html/lab/lab516/Heat%20Transfer/chapter_4.pdf.
http://tutorial.math.lamar.edu/Classes/DE/SeparationofVariables.aspx.
20. Under what conditions can the Heisler charts be used to find the time-dependent temperature profiles in a large
reactor component?
21. How is the thermal time constant for the Heisler charts defined? Is it defined differently for a cylindrical object than
it is for an object having the shape of a sphere or a plate?
22. When using the Heisler charts, what does a Fourier number greater than 0.2 signify? How many terms must be used
to approximate the time-dependent temperature profile in a reactor component when τ > 0.2? What is the largest
error that can occur in estimates of the time-dependent temperature profiles under these conditions?
23. Under what conditions should the time-dependent temperature profiles in a reactor component be found numeri-
cally rather than analytically? What boundary conditions should be applied to a component to obtain a physically
meaningful solution?
24. In addition to the Heisler charts, other charts are sometimes used to obtain solutions to time-dependent reactor heat
transfer problems. What are the names of these charts, and how do they differ from the Heisler charts?
25. A nuclear fuel rod is 5 m long and 1 cm in diameter. What is the thermal time constant for this fuel rod if it is made
up of uranium dioxide (UO2)?
26. How is the Biot number defined, and what is its physical significance?
27. What is the purpose of the one-term approximation? When this particular approximation is used, what is the maxi-
mum error that can be expected in the values of the time-dependent temperature profiles d etermined using the
Heisler charts?
28. In time-dependent nuclear heat transfer, what does the term “lumped parameter approach” refer to? When can it be
applied to a reactor component?
29. Beneath what specific range of Biot numbers can the time-dependent temperatures in a reactor component be found
using a lumped parameter approach? Above this Biot number, how are the temperature profiles determined for
simple geometric shapes?
30. Compare the time-dependent temperature profiles for a hot object cooling off in a large tank of cold water to the
time-dependent temperatures given by Newton’s law of cooling, which is presented in Chapter 10. What are the
similarities and differences between the governing equations in each case?
Exercise 12.2
An egg is to be cooked in a pot of boiling water. Assume that the egg is spherical in shape with a radius of 2.5 cm and
has a convective heat transfer coefficient of 1,200 W/m2. Assume that the thermal conductivity of the interior of the egg
is 0.6 W/m °C. Can a lumped parameter approach be used to find the temperature at the center of the egg? Determine
how long it will take for the temperature at the center to reach 70°C. Can you take a short walk before the egg reaches
its target temperature?
Exercise 12.3
In a factory making specialized components for nuclear heat exchangers, large brass plates of 10 cm thickness that are
initially at a uniform temperature of 20°C are heated by passing them through an oven that is maintained at 500°C.
The plates remain in the oven for 10 min. Taking the combined convection and radiation heat transfer coefficient to be
100 W/m2 °C, determine the surface temperature of the plates when they come out of the oven. What is the Biot number
and the Fourier number for the plates? Compare the final temperatures calculated using the Heisler charts and a lumped
parameter approach.
Exercise 12.4
A long cylindrical shaft used in the construction of reactor coolant pipes is made from stainless steel 304. The shaft has a
diameter of 20 cm and comes out of an oven at a uniform temperature of 600°C. The shaft is then allowed to cool slowly
484 Time-Dependent Nuclear Heat Transfer
in a holding chamber at 100°C with an average heat transfer coefficient of 80 W/m2 °C. Determine the temperature of the
center of the shaft 1 h after the start of the cooling process.
Exercise 12.5
Consider a sphere and a cylinder of equal volume made up of stainless steel. Both the sphere and the cylinder are ini-
tially at the same temperature and are exposed to convection in the same environment. Which one do you think will
cool faster and why?
Exercise 12.6
A flat metal plate has a characteristic thickness of 4 cm, a thermal conductivity of 100 W/m °C, and a convective heat
transfer coefficient of 120 W/m2 °C. What is the Biot number for this plate, and can a lumped parameter approach be used
to find this plate’s time-dependent temperature profile?
Exercise 12.7
A uranium dioxide fuel pellet used in a research reactor is 10 cm long, 1 cm wide, and cylindrical in shape. The fuel pel-
let is sintered in an oven at 2,000°C to improve its dimensional stability. If the fuel pellet comes out of the oven with a
uniform temperature of 1,800°C, can the Heisler charts be used to determine the centerline temperature of the fuel pellet
as a function of time? If so, suppose that the pellet is exposed to air at 30°C until it is completely cooled. What is the
centerline temperature of the pellet 30 min after it is removed from the oven? Assume that the heat transfer coefficient
of the surrounding air is 100 W/m2 °C.
Exercise 12.8
Stainless steel bearings used in the construction of a crane for a nuclear power plant have an outer diameter of 1.2 cm,
and after finishing, are to be quenched in water. The balls leave an oven at a uniform temperature of 900°C and are
exposed to air at 30°C for a while before they are dropped into the water. If the temperature of the balls is not to fall
below 850°C before quenching, and the heat transfer coefficient of the air is 125 W/m2 °C, determine how long they can
be exposed to the air before being dropped into the water.
Exercise 12.9
A square copper plate measuring 400 mm on a side and 5 mm thickness has a uniform temperature of 250°C. Its
temperature is suddenly lowered to 30°C. Calculate the time required for the plate to reach an average temperature
of 90°C. Assume that the convective heat transfer coefficient at the surface of the plate is 90 W/m 2 °C, and the prop-
erties of the plate are ρ = 9,000 kg/m 3, cp = 380 J/kg °C, and k = 370 W/m °C.
Exercise 12.10
A long cylindrical bar used in the construction of a reactor containment building has an outer radius of 75 mm coming
out of a finishing oven with a temperature of 875°C. It is then cooled by quenching it in a tank of water at 38°C. The
convective heat transfer coefficient between the bar and the water is 175 W/m2 °C. Determine (1) the time it takes for
the center of the bar to reach 116°C, (2) the surface temperature of the bar when its center temperature is 116°C, and (3)
the value of the temperature gradient on the outer surface of the bar at the same point in time. Assume that the bar has
a thermal conductivity of k = 17.5 W/m °C and a thermal diffusivity of α = 0.0185m2/h.
13
Nuclear Reactor Fluid Mechanics
13.1 Characterizing Reactor Fluid Flow
A reactor can be thought of as a nuclear furnace that transfers the heat it produces to the coolant that flows through the core. In most
reactors, the coolant is a simple liquid (like ordinary water), although some reactors such as gas reactors are cooled by carbon dioxide
or helium and other reactors such as fast reactors are cooled by liquid metals. Because the coolant is always isolated from the outside
world and it is never allowed to flow freely outside the confines of a duct, a pipe, or the pressure vessel, the flow of fluid through the
reactor core and the nuclear steam supply system (NSSS) is a classic example of what is called an internal or enclosed flow. In other
words, the flow of fluid within the core, within the pressure vessel, and within the plant as a whole is completely bounded by solid
surfaces on all sides. This means that we must treat these flows differently than we would treat the flow of water in a river or a stream
or the flow of air over an aircraft wing. These latter flows are examples of open flows, and there is a great deal of difference between
the behavior of open flows and the behavior of enclosed flows.
FIGURE 13.1 A figure showing the flow of fluid over a flat horizontal plate. Here, regions of viscous flow exist close to the plate where the velocity is
changing and regions of inviscid flow appear to exist further away. Outside the boundary layer, the flow is essentially undisturbed because the velocity
gradients are so small. Hence if either the viscosity μ or the velocity gradient du/dy is zero, the shear stress τ will be zero as well.
485
486 Nuclear Reactor Fluid Mechanics
The internal resistance to the flow can be quantified by a fluid property called the dynamic viscosity μ. The viscos-
ity is a measure of the internal stickiness of a fluid and its resistance to forward motion. Normally, the viscosity is
quoted in the units of kg/m s or Pa s, where 1 Pa = 1 N/m2, although other units of measurement are sometimes used.
For example, in nuclear science and engineering, the most commonly used units for the viscosity are the poise, which is
equivalent to 0.1 Pa s, and the centipoise, which is equal to one hundredth of a poise, or 0.001 Pa s. These units for the
viscosity are summarized in Table 13.1. In the English unit system, the primary unit for the viscosity is lbf/s ft2. A large
number of viscosity conversion tools are available on the Internet to convert between one unit system and another. Some
of the most popular conversion tools can be found at the following websites:
http://www.unitconversion.org/unit_converter/viscosity-dynamic.html
http://www.gordonengland.co.uk/conversion/dynamic_viscosity.htm
Sometimes the viscosity μ is also referred to as the absolute or dynamic viscosity because it is associated with fluids in
motion. The viscosity can be thought of as the internal friction that arises whenever two molecules of the fluid collide
with each other as they move. A similar effect can also occur in gaseous coolants. However, because the molecules are
closer together in a liquid than they are in a gas, the viscosity of most liquids is greater than it is for most gases. This
difference in viscosity affects the detailed structure of the flow field as well.
No ordinary fluid has an absolute viscosity of zero, and flows in which the viscosity becomes important are called
viscous flows. In many problems of practical importance to reactor design, there are regions away from any solid surface
where viscous effects are small compared to gravitational or inertial ones. In these regions, viscosity has little effect
relative to other forces, and flows in which the effects of friction can be ignored are called inviscid flows. Neglecting the
effects of friction can greatly simplify the analysis of certain problems without a corresponding decrease in the accuracy.
However, these problems normally require the fluid to have a very low viscosity or to be very far away from the surface
of a solid object. Another commonly used measurement for the viscosity is the kinematic viscosity. The kinematic viscos-
ity ν is defined as the dynamic or absolute viscosity μ divided by the density ρ, which is given in the following equation:
The kinematic viscosity also appears in the definition of the Reynolds number, which is another important parameter of
fluid mechanics that is used to determine whether the flow is laminar or turbulent. The Reynolds number can be thought
of as the ratio of the inertial forces in a fluid to the viscous forces:
In practice, the kinematic viscosity ν is measured in m2/s and/or the stoke, where 1 stoke = 1 cm2/s = 0.0001 m2/s. Notice
that the kinematic viscosity is independent of the mass of the fluid for which it is being measured. For most reactor cool-
ants, the kinematic viscosity depends on both the temperature and the pressure, although its dependence on the pressure
can be rather weak. For water and metallic coolants, the dynamic and kinematic viscosities are practically independent
of the pressure, and coolants that possess this behavior are said to be incompressible. The dynamic viscosity μ for
TABLE 13.1
Common Measurements of the Absolute or Dynamic Viscosity
Most gases follow this law at low and intermediate pressures. Finally, the kinetic theory of gases predicts that the viscos-
ity of most gases is directly proportional to the square root of the absolute temperature. Because of this, the viscosity of
the coolant in a gas-cooled reactor (GCR), whether it is hydrogen, helium, or carbon dioxide, is subject to the following
relationship:
where T is the temperature in degrees Kelvin. However, for diatomic and polyatomic gases, some correction factors need
to be applied to Equation 13.3 to account for the true dependence of the viscosity on the absolute temperature.
The viscosity of liquid and gaseous reactor coolants is compared in Figure 13.2. Notice that they behave differently
as the absolute temperature is raised. This is due to a difference in the strength of their molecular bonds. Finally, for
most liquids, the absolute viscosity can be approximated by
where T is the absolute temperature and A, B, and C are experimentally determined constants. For a s ubstance like
water, the values of these constants are A = 2.414 × 10 −5 N s/m2, B = 247.8°K, and C = 140°K. This expression results in
a percentage error for the viscosity of water of less than 2.5% between 0°C and 370°C, which is an unusually wide range.
FIGURE 13.2 The absolute viscosity μ of most liquids and gases is temperature dependent. In liquids, the viscosity decreases, and
in gases, the viscosity increases as the temperature increases.
488 Nuclear Reactor Fluid Mechanics
where c is the speed of sound in the material that comprises the flow. In ordinary air at room temperature and at sea
level, the speed of sound is about 346 m/s (1,135 ft/s). Hence, airplanes traveling faster than this can c reate a shock wave
or sonic boom. A flow is called a subsonic flow when Ma < 1, a sonic flow when Ma = 1, and a supersonic flow when
Ma > 1. Except when a LOCA occurs, the flow in a reactor core is generally considered to be s ubsonic. As a matter of
fact, the velocity of the coolant is usually limited to about 10 m/s (which translates to a Mach number of about 0.03 Ma).
Normally, a gas flow can be treated as incompressible if the change in its density is less than about 5%. For ordinary
air, this corresponds to a Mach number of about 0.3. Consequently, the compressibility of air and other gases can be
neglected for speeds below about 100 m/s.
FIGURE 13.3 A picture illustrating the difference between a laminar flow, a turbulent flow, and a transitional flow. The numbers
quoted here are for enclosed flows in reactor fuel assemblies. For different applications, the transition points may be different.
field is intentionally designed to be as turbulent as possible, because for a given flow rate m = ρVA this is the best way to
maximize the rate at which heat can be removed from the core. When the flow is laminar, the motion of a fluid is highly
ordered (i.e., it has a relatively low entropy), and the fluid particles move past each other in smooth layers that can easily
be seen by the naked eye. The flow of most liquids and gases including air, water, oils, and liquid metals at low velocities
is predominately laminar. The fluid particles move past each other in orderly layers that resemble the laminates used to
cover some types of woods, metals, and plastics. See Figure 13.3 for an illustration of how laminar flows behave.
When the velocity of a fluid becomes high enough, the motion of the fluid begins to become disordered, and the flow
field that was initially laminar begins to develop small eddies and vortices in it. These turbulent eddies and vortices
give rise to random velocity fluctuations within the flow itself. A flow field that contains these velocity fluctuations is
then said to be turbulent. The flow of low-viscosity coolants such as carbon dioxide and air will eventually become tur-
bulent if the average velocity of the flow field becomes high enough. In a reactor core, there is a big difference between
the amount of energy required to pump a laminar fluid and the amount of energy required to pump a turbulent one.
We will examine the reasons for this in Chapter 19. A flow that alternates between being laminar and being turbulent
is sometimes called a transitional flow. The difference between a laminar flow regime, a transitional flow regime,
and a turbulent flow regime is shown in Figure 13.3. Normally, a dimensionless quantity called the Reynolds number
Re is used to determine when a flow becomes laminar, transitional, or turbulent. We will have more to say about the
Reynolds number and how it can be used in Chapter 15. The Reynolds number can also be used to distinguish between
laminar and turbulent behavior in external and enclosed or internal flows. For enclosed flows, the flow in reactor fuel
assemblies remains laminar for Reynolds numbers below 2,300, becomes partially laminar and partially turbulent for
Reynolds numbers between 2,300 and 10,000, and becomes fully turbulent for Reynolds numbers above 10,000. The
flow for Reynolds numbers between 2,300 and 10,000 is sometimes called a transitional flow (see Figure 13.3). The flow
of a fluid over a stationary plate or an airplane wing is an example of an external or open flow. In general, the Reynolds
number is the key parameter that is used to distinguish between laminar and turbulent flows.
FIGURE 13.4 An example of the circulatory patterns between fuel rods during natural convective flow.
or over the object to be cooled. In natural convective flows, fluid motion is due solely to natural processes such as
buoyancy-induced effects caused by differences in the fluid density. These flows usually manifest themselves by the
rise of warmer plumes of fluid and the fall of colder plumes of fluid, which creates a circulatory pattern in a clockwise
or counterclockwise direction. Some of the flow loops that can develop are shown in Figure 13.4. The coolant in these
loops continues to circulate as long as the warmer fluid is less dense than the colder fluid. This density difference sets
up a pressure differential that continues to drive the motion of the fluid particles until the temperatures of all of the fluid
particles become the same. Then, the flow stops and the velocity of the fluid falls to zero. In reactors, natural convective
flows are also referred to as natural circulation flows.
FIGURE 13.5 A steady flow is one in which the velocities of the fluid particles and their paths do not change as a function of time.
In the picture above, a fluid flows past a stationary sphere and the flow rate is constant. Since the trajectories and the velocities of the
fluid particles are constant in time, this is an example of a steady flow field. (Simulation performed by FLUENT.)
homogeneous everywhere; that is to say, it is uniform over a large region of space. In fact, the term “uniform” implies no
change at all as a function of position over a specified region of space. The opposite of a “steady” flow is an “unsteady”
one. An unsteady flow implies that the flow field is changing as a function of position and time, while in a steady flow
(see Figure 13.6), the velocity of the fluid particles at any point in space remains constant in time. This is somewhat dif-
ferent than a “transient flow,” which is sometimes used to describe a flow field that is still developing. Reactors can have
both developing flow fields and fully developed ones. The term periodic flow refers to a special type of unsteady flow
that oscillates about an average value or mean. Certain types of reactor transients allow this to occur.
Many components of the NSSS, including the turbines, steam generators, condensers, and reheaters, are designed to
operate for long periods of time with no appreciable changes to the flow field. Small changes from these average values
FIGURE 13.6 Flows in a nuclear reactor can be either steady or unsteady. In steady flows, the velocity remains the same at every
point in time, while in unsteady flows, the velocity can change as a function of time.
492 Nuclear Reactor Fluid Mechanics
can occur on an hourly, weekly, or daily basis, but over time, the fluid flow field can be assumed to operate about an
average set of values or “mean.” Thus, these components of the NSSS are considered to be steady-flow devices. Within
each device, the fluid properties can change from point to point, but at any fixed point, they remain essentially the same.
Within the core, most fuel assemblies are also designed to operate around a set of steady-state flow conditions that can
be approximated with a steady-state flow field. Thus, the volume of the fluid, the mass of the fluid, and the energy con-
tent of a steady-state flow remain relatively constant during the normal operating cycle of the plant. Any deviation from
this set of operating parameters must be taken into account in the design of the component itself. Because of this, one
of the most important jobs of a nuclear engineer is to determine if it is acceptable to study the time-averaged behavior
of a NSSS component, or whether a more detailed analysis of the component involving the use of unsteady flows is
required. Normally, the U.S. Nuclear Regulatory Commission (or NRC) specifies the minimum and maximum operat-
ing limits that each component in the NSSS must be designed to support. These operating limits may also account for
flow-induced vibrations, the possibility of small pipe leaks, the propagation of pressure waves and shock waves and their
transient response as the fluid temperature is periodically raised and lowered in response to power changes in the core.
Most of the examples that we will present in this book deal with the behavior of a power reactor when the flow rates
and the flow fields are assumed to be reasonably steady or time-averaged. However, in the unlikely event of a reactor
accident, the behavior of unsteady flows can be come very important. In particular, the effects of compressibility can
dominate the behavior of the flow.
Solving problems of this type requires a completely different mindset than studying the behavior of steady flows
because one can no longer simply time-average uncertainties in the expected values of the flow field away. Instead,
we must be able to mechanistically predict how each of these components behave when a transient or reactor acci-
dent occurs. This requires a detailed understanding of computer simulations and computational methods because a
computer program must now be written to predict the behavior of the flow. In reactor cores, this can be a daunting task,
but several computer programs have been written over the years to make an analysis of this type possible—at least for
the current generation of nuclear power plants. Computer programs of this type include COBRA, VIPER, RELAP,
RETRAN, and TRACE. The author was one of the early contributors to the COBRA series of codes, whose descen-
dants are now used by companies such as Westinghouse to find the flow and temperature fields in reactor cores.* We
will have more to say about these computer programs in Chapters 26 and 30. Note in particular that practical solutions
to most fluid-related problems in reactor engineering require at least an elementary knowledge of the Navier–Stokes
equations, which we will discuss in Chapter 15. The fluid field must also be coupled to the heat transfer field, and both
of these fields may change as a function of time. Transients of this type frequently occur during the operation of most
nuclear power plants.
* See Improved Multidimensional Numerical Methods for the Steady State and Transient Thermal-hydraulic Analysis of Fuel Pin
Bundles and Nuclear Reactor Cores by Robert Edward Masterson, PhD Thesis, Massachusetts Institute of Technology (1977).
13.11 Classifying Reactor Coolant Flows 493
FIGURE 13.7 For horizontal pipe flow, the radial velocity profiles are different at the pipe entrance than they are when the flow
becomes fully developed. In the fully developed region, the radial velocity profile is parabolic when the flow is laminar. In most hor-
izontal pipes, the flow becomes fully developed at a distance between 5 and 10 diameters from the entrance to the pipe—depending
upon whether the flow is laminar or turbulent.
For a Reynolds number of 20, the entry length is about the same size as the pipe diameter. However, it increases linearly
with the average velocity, and in the limiting case where Re = 2,300 for laminar flows, the entry length is about 115
D. When the flow becomes turbulent, turbulent mixing completely dominates the effects of molecular diffusion. The
hydrodynamic entry length is then given by Bhatti and Shah correlation, which is as follows:
Thus, the mass flow rate m = ρVA through the pipe as well as the fluid velocity V and the Reynolds number Re deter-
mines when the flow becomes fully developed and when it is not. When the flow is still developing, the heat transfer
coefficient is different near the entrance to the pipe than it is near the center, and we must take this difference into
account when trying to calculate the heat transfer rate. Normally, a nuclear engineer must apply a “correction factor” to
the heat transfer coefficient at the entrance of the pipe to obtain the correct fluid temperature profile as a function of axial
position. Here, the axial position is assumed to be measured from the start of the pipe where z = 0.
FIGURE 13.8 Single-phase flow in reactor coolant channels can take a number of different forms. The picture above shows the
ways in which the flow fields in a nuclear reactor can be classified.
stress τ is a linear function of the velocity gradient du/dy. The constant of proportionality between the shear stress and
the velocity gradient is the fluid viscosity μ, which was also called the dynamic viscosity earlier in the chapter. Hence,
Newtonian coolants are defined in the following way:
where μ is the dynamic viscosity and du/dy is called the local shear velocity gradient. Water, metallic coolants, and most
gases behave in this way. In general, the entrance length is much shorter when the flow becomes turbulent, and its depen-
dence on the Reynolds number is weaker. Hence, most entrance effects become negligible beyond a length of about 10 D,
and the hydrodynamic entrance length is then Le ≅ 10 D. Precise values for the frictional losses in the entrance region can
be found on the internet. However, in reactors, most pipes have lengths many times longer than the diameter, and thus,
the flow through these pipes can be assumed to be fully developed over the entire length of the pipe. In shorter pipes, this
simple approach underpredicts the wall shear stress and therefore the size of the friction factor.
and drawbacks. Hence, reactor designers rely extensively on experimental data and detailed numerical simulations to
try to reduce the conservatism in their one-dimensional models. Until the computers used to do this became powerful
enough, it was simply not practical to try to solve the fluid mechanics equations on this scale.
FIGURE 13.9 Reactor piping systems and other reactor components can be analyzed using what are known as control volumes.
A control volume is a region of space bounded by a control surface.
496 Nuclear Reactor Fluid Mechanics
FIGURE 13.10 Any physical system through which a fluid flows can be modeled as a control volume with different inputs and
outputs.
FIGURE 13.11 A reactor steam turbines can be modeled as a set of interconnected components where each component is rep-
resented by a single node or a set of nodes. The nodes are connected together by what are known as junctions or interfaces. In the
example above, ten nodes and nine interfaces are used to model a two-stage steam turbine.
13.16 The State Postulate 497
FIGURE 13.12 The flow channels in most reactor fuel assemblies are square or triangular in shape. On the left, the flow channels
are shown from a light water reactor fuel assembly, and on the right, they are shown from a fast reactor fuel assembly. In each type
of fuel assembly, there can be center channels, side channels, and corner channels. The flow channels are also called subchannels.
more to say about how lumped parameter approaches are implemented in Chapter 15. Normally, each region of space
over which this spatial averaging is done corresponds to one of the control volumes we previously discussed. Control
volumes are also one of the fundamental constructs of Eulerian fluid mechanics.
then fixes the thermodynamic state (see Chapter 7). There are several properties that are particularly important in the
study of fluids. The first one is the density ρ, and the second one is its specific volume υ. The specific volume is the
inverse of the density. Hence, the specific volume can be found by simply taking the reciprocal of the density or υ =
1/ρ. For example, if the volume of the system is V and its mass is m, its specific volume is υ = V/m. Hence, the specific
volume is an intensive property of a fluid. The density, which is the inverse of the specific volume, is defined by ρ = m/V.
Normally, these two properties are used interchangeably when determining the state, but the density appears explicitly
in the Navier–Stokes equations, while the specific volume does not.
Sometimes the relative density is also called the specific gravity. Notice that the specific gravity is dimensionless and
the specific gravity is equal to the density quoted in kg/m3. For single-phase applications, we can usually treat water as
an incompressible fluid because its density will only change by a couple of percent between the top and the bottom of
the core. Values of the density are presented for many different operating conditions in the appendices. For everyday
temperatures, the density of water at atmospheric p ressure is presented in Table 13.2.
TABLE 13.2
The Density of Liquid Water as a Function of Its Temperature in Different Unit Systems
The values given here represent the density of ordinary water at atmospheric pressure (14.7 PSI).
Note: The values below 0°C refer to supercooled water.
13.20 Partial Pressures and the Vapor Pressure 499
to boil is called its saturation temperature TSAT. As the pressure of the liquid is raised, the saturation temperature also
increases. Conversely, if we reduce the system pressure, we will find that boiling will begin at a lower temperature.
Alternatively, when the temperature is fixed, the pressure at which a pure substance changes phase is called the satura-
tion pressure PSAT. The vapor pressure PVAPOR of a pure substance is defined as the pressure exerted by its vapor phase
when it is in thermal equilibrium with its liquid phase at a given temperature T.
In other words, both phases have to be in thermodynamic equilibrium for the concept of the vapor pressure to be thermo-
dynamically meaningful. Other definitions for the vapor pressure are also possible, but this implies that phase equilib-
rium is achieved at different temperatures for each phase. The vapor pressure is also different than the system pressure
that is sometimes encountered when using the ideal gas law. To illustrate this d ifference, we would like to first discuss
the ideal gas law and then show how the concept of partial pressures is related to it. Notice that liquid water tends to get
less dense as it is either heated or cooled relative to its maximum density at 4°C.
where T is the absolute temperature, p is the absolute pressure, ρ is the density of the gas, and R is the gas constant. As
we discussed previously, the value of R is different from one type of gas to the next, and its value may be found from
R = R u M (13.11)
where Ru is the universal gas constant whose value is 8.314 kJ/kmol °K, and M is its molar mass, which is also called
its molecular weight. The values for M and Ru can be found for many pure substances in the References section at the
end of the chapter as well as on the Internet. The pressure p that is used in the ideal gas law is the pressure pMIX of the
gaseous mixture to which the ideal gas law applies. In other words, the ideal gas law can be applied to situations where
more than one ideal gas is present at the same time. However, when this occurs, it is assumed that the gases coexist at
the same temperature; that is, they have reached a state of thermal equilibrium.
The vapor pressure pVAPOR is usually just one of these components. This distinction is particularly i mportant when ana-
lyzing the behavior of a reactor containment building during reactor accidents or LOCAs. During these accidents, the
steam that flashes from the reactor pressure vessel mixes with other gases (normally air) in the containment building to
form a complex heated mixture. In atmospheric air, the air we breathe is a mixture of dry air (with no water vapor in
it) and a small amount of water vapor. The atmospheric pressure of ordinary air is then given by the sum of the partial
pressures of the dry air and the water vapor. So using the aforementioned equation, we see that
500 Nuclear Reactor Fluid Mechanics
This implies that if the temperature of both gases is the same, the atmospheric pressure increases as the humidity gets
higher. Normally, the partial pressure exerted by the water vapor is only a small fraction (less than 3%) of the total pressure
of wet air. However, if a bucket of water is placed in open air, it will continue to evaporate until all of the water in the bucket
is gone. This is because there is a fundamental difference between the vapor pressure pVAPOR and the partial pressure that
goes into the calculation of the mixture pressure pMIX in the ideal gas law. This distinction can be seen in Table 13.3. For
example, the vapor pressure of ordinary water at 25°C is 3.17 kPa (refer to Table 13.4 for some actual numbers). The water in
the bucket will always evaporate until either there is enough water in the bucket to establish phase equilibrium with the sur-
rounding environment or the evaporation will stop when the partial pressure of the water vapor reaches 3.17 kPa at which
point phase equilibrium with the liquid phase is established. The reason why the vapor pressure is so important is that it
explains why water boils at a higher temperature as the ambient pressure is increased. Vapor pressure always increases with
increasing temperature because there is more molecular motion (and hence greater evaporation) at the liquid–vapor inter-
face. In general, this effect cannot be ignored when hot steam is mixed with the dry air in a reactor containment building.
The other reason why the vapor pressure is so important is that if the system pressure in any part of the NSSS
becomes too low, the liquid water in the pipes will flash to steam or form bubbles at locations that can d amage the blades
of a coolant pump or the tips of its impellers. While this is normally not a concern when a reactor is operating normally,
suppose that we consider high-pressure water at 300°C. The relationship between the vapor pressure and the tempera-
ture of this water is given by the Antoine equation. The Antoine equation says that the vapor pressure of ordinary water
is given by
TABLE 13.3
Saturation (or Vapor) Pressures for Water at Different Temperatures
Temperature (°C) Saturation Pressure (kPa) Temperature (°C) Saturation Pressure (kPa)
−10 0.260 30 4.25
−5 0.403 40 7.38
0 (freezing point) 0.611 50 12.35
5 0.873 100 (boiling point) 101.3 (atmospheric pressure)a
10 1.228 150 475.8
15 1.710 200 1,554
20 2.338 250 3,973
25 3.168 300 8,581
a 1 atm.
TABLE 13.4
The Vapor Pressure of Water between the Freezing Point and Room Temperature
FIGURE 13.13 The Antoine chart for the vapor pressure of water as a function of the absolute temperature.
and the temperature T is measured in °C. Here, the values of A, B, and C for water are
A picture of the vapor pressure curve for ordinary water is shown in Figure 13.13. The vapor pressure of water at 250°C is
30,050 mm of mercury or 4.0 MPa (580 PSI). Thus, if the pressure in the hot leg or the cold leg of the NSSS falls below 580
PSI (about one-fourth of the standard operating pressure), vapor bubbles will begin to form in the flow stream and damage
the tips of the impellers of the reactor coolant pumps. Obviously, this is a condition we would like to avoid by keeping the
operating pressure as high as possible. If the pressure in another part of the hot leg or cold leg is higher than 580 PSI, the
vapor bubbles that were generated in the flow stream can also collapse back into the surrounding fluid, and this can produce
shock waves in the high-pressure zones. These pressure waves can generate unwanted vibrations in some parts of the NSSS,
loud noises in others, and even damage some of the surrounding coolant pipes and piping equipment. Hence, after a lot of
energy is added to a reactor coolant such as water, great care must be taken to remove it before reducing the ambient pres-
sure too quickly. This is just one example of how unwanted pressure changes can cause unwanted bubbles to form and col-
lapse in the flow stream. In municipal power plants, this is also a problem even when the power source is not a nuclear one.
different between compressible and incompressible ones. One way to quantify this effect is to define a coefficient of
compressibility κ and a coefficient of expansion β. When a fluid expands or contracts, it acts like an elastic solid, and to a
certain extent, it can be modeled in an analogous way as a deformable object. In the study of solids, the rate of deforma-
tion is defined by a quantity called Young’s modulus of elasticity. All solids, except very hard and brittle ones, exhibit
some tendency to deform themselves when they are subjected to enough pressure. Young’s modulus is a measure of the
amount of deformation that can occur. In fluids, it is possible to define a coefficient of compressibility κ that performs an
analogous function to what Young’s modulus does in the world of solids, but is specific to the properties of a particular
fluid. The coefficient of compressibility is defined as
where υ is the specific volume of the fluid. Thus, the compressibility is a measure of the relative volume change of a fluid
or solid in response to a pressure change Δp, where V is the volume and p is the pressure. This concept is illustrated in
Figure 13.14 where it is applied to an internal combustion engine. In general, gases are more compressible than liquids
are. In classical physics, the speed of sound c is also related to the isentropic compressibility βS:
c =1 ρβS (13.16a)
Therefore, the coefficient of compressibility must have the units of pressure, and in practice, it is generally expressed
in atm, Pa, or PSI. Because of the way it is defined, the compressibility coefficient represents the fractional change in
the pressure that is required to produce a corresponding change in the density ρ while the temperature of a fluid is held
constant. This implies that the coefficient of compressibility for an incompressible fluid must be infinite because in this
case both Δρ and Δυ are zero! In other words, when κ is large, this indicates that a large change in the system pressure
is required to produce a small change in the specific volume Δυ. For example, at atmospheric pressure, the pressure of
liquid water must be increased to about 210 atm (~3,000 PSI) to compress it by 1%. This corresponds to a coefficient of
compressibility of about 21,000 atm. Gases, on the other hand, tend to be more compressible, and the compressibility
coefficient is very small compared to that of liquids (about 1 atm). For example, normal air at atmospheric pressure has
a value of κ = 1 atm, and so for air at a pressure of 1,000 atm, κ = 1,000 atm, and a decrease of 1% in its volume requires
FIGURE 13.14 A graphical representation of compressibility. As we pointed out earlier, compressibility can be applied to liquids,
gases, and even solids. For liquids and gases, compressibility is defined by the coefficient of compressibility κ. For solids, it is
defined by Young’s modulus.
13.23 Understanding Water Hammers 503
an increase of 10 atm in its pressure. For an ideal gas having an absolute temperature T, it can be shown that the relation-
ship between a change in its density Δρ and a change in the pressure Δp is
where Δρ is the density change and Δp is the pressure change. This relationship can be useful when attempting to under-
stand how much pressure is required to compress a gas. Finally, for some applications, it is convenient to define the inverse
of the coefficient of compressibility, which is called the isothermal compressibility α. This coefficient appears in many
nuclear engineering textbooks in a discussion of the Brayton thermal cycle. The isothermal compressibility is defined as
and it has the units of inverse pressure. Thus, the isothermal compressibility represents the fractional change in the
volume or the density that occurs as the result of a corresponding change in the pressure Δp.
where p is the absolute pressure of the gas, T is the absolute temperature, and V is its molar volume. This means that
the compressibility factors for actual gases can be greater than unity, less than unity, or equal to unity. When the com-
pressibility of a real gas is the same as that of an ideal gas, the compressibility factor is unity. The deviation of the
compressibility factor from unity tends to become more significant as one approaches the critical point, where the pres-
sure is high or the temperature is low. In this case, a general compressibility chart or an alternative equation of state must
be used to obtain the appropriate compressible behavior. A generalized compressibility chart is shown in Figure 13.15
for air. Here, the reduced temperature and pressure are defined by
where TCRITICAL and PCRITICAL are to be evaluated at the critical point. The compressibility factor for air is shown for
temperatures between 250°K and 1,000°K and pressures between 0 and 500 bars (0–7,250 PSI). Similar compressibility
charts are available for coolants such as hydrogen, helium, carbon dioxide, and steam. As we mentioned in Chapter 7,
steam cannot be treated as an ideal gas at pressures above 10 KPa (~1.5 PSI).
FIGURE 13.15 The compressibility factor for ordinary air for various temperatures and pressures. Note that for most temperatures
and pressures, the compressibility factor is positive, indicating a significant departure from ideal gas behavior. (Picture provided by
Wikipedia.)
FIGURE 13.16 A water hammer can be created when a valve closes abruptly. This causes the fluid to be compressed locally, and
to relieve this tension, it generates a shock wave that propagates in the opposite direction. (Picture on the right provided by fine-
homebuilding.com.)
These waves can also cause the pipes to vibrate as they move. The physics of the water hammer is illustrated in Figure
13.16. Obviously, great care must be taken to avoid creating a water hammer in a reactor coolant system because of the
potential damage it can inflict on the pipes.
would like to introduce another useful parameter called the volumetric coefficient of e xpansion. In general, the den-
sity of a liquid is more affected by its temperature T than by its pressure p. This v ariation of density with temperature
explains why the coolant in a reactor core will continue to circulate long after the coolant pumps have been turned off,
and it also explains why hot air rises and cold air sinks. Fluids such as liquids and gases both exhibit a similar tendency
to expand and contract, but the expansion is generally more pronounced in gases than it is in liquids. This effect can
be quantified using another parameter called the volumetric coefficient of expansion β. This coefficient represents
the variation of the volume with fluid temperature. Normally, the volumetric coefficient of expansion is measured at
a constant pressure. In other words, the pressure is held constant as the temperature of the fluid is raised or lowered.
Any changes in the volume are then recorded, and this is how the value of β is measured. The volumetric coefficient
of expansion is defined by
and it can be seen to have the units of inverse degrees. The value of β is normally given in °K−1. It can also be expressed
in terms of finite changes to the temperature and the specific volume as
when the pressure is held constant. Notice that large values for β imply that a fluid will have a large density change as
the temperature is changed. The magnitude of this change is of course given by Δρ = 1/Δυ where υ is its specific volume.
If the fluid happens to be an ideal gas, the equation for the expansion coefficient is even simpler. In this case, the value
of β becomes
where T is the absolute temperature. This expression can be derived rather easily from the ideal gas law. A graph of
the expansion coefficient for water is shown in Figure 13.17. The expansion coefficients for other common fluids can be
found on the Internet.
FIGURE 13.17 The volumetric expansion coefficient for water (β) at atmospheric pressure.
506 Nuclear Reactor Fluid Mechanics
(
β = ( ρ∞ − ρ) ρ ) (T
∞ − T ) (13.24)
where the values of T and ρ refer to the temperature and the density of the fluid next to the wall. We can also write this
expression as
ρ∞ − ρ = βρ( T − T∞ ) (13.25)
or
∆ρ ρ = β ∆T (13.26)
and this equation establishes a relationship between the respective densities in each case. Here, ρ∞ is the density of the
fluid at effectively an infinite distance from the wall surface (where the fluid is assumed to be undisturbed). In Chapter 22,
we will learn that the convective currents that we observe during natural circulation are induced by buoyancy-related
effects. In other words, the force of gravity exerts a slightly different force on a cold fluid than it does on a hot one.
Because of these slight density differences, the hotter parts of the fluid tend to rise and the colder parts of the fluid tend
to sink. The larger the temperature difference ΔT = (T − T∞) is between the fluid at the wall and the main body of fluid
surrounding the wall, the greater this density difference Δρ = ρ∞ − ρ will be. We can also relate the change in specific
volume Δυ to a combination of changes in the temperature and the pressure. Since υ = υ(p, T), we can differentiate the
state equation for υ to find how υ changes as a function of temperature and pressure. Taking the partial derivative of υ
in each case gives
d υ = (β dT − α dp)υ (13.28)
or
d υ/ υ = β dT − α dp (13.29)
If the values of dT and dp are finite, then the aforementioned expression becomes
and this is how changes in the volume and the density are calculated in practice. Note in particular that Δυ/υ and Δρ/ρ
are now the percentage changes in the specific volume and the density that can be attributed to temperature and pres-
sure changes in the fluid as a whole. In reactors, the pressure is normally fixed in the primary loop and the temperature
is varied to produce the changes in the coolant density which are observed. Since the colder fluid enters the core in
most reactor designs from below (see Figure 13.18), the coolant at the bottom of the core is colder and denser than the
13.26 Viscosities of Common Gases and the Sutherland Correlation 507
FIGURE 13.18 In PWRs, BWRs, fast reactors, and gas reactors, the coolant enters the core from below and exits the core from
above. This flow pattern allows the heated coolant to rise even when the pumps are shut off. This core flow can be modeled using
imaginary control volumes with junctions and interfaces. The sole exception to this flow pattern is the Canadian CANada Deuterium
Uranium (CANDU) reactor where the coolant flows through the core through pressure tubes in the horizontal direction. In each
case, the heated coolant is sent to a power turbine where it is used to spin the blades and generate electric power. Normally, the
working fluid that reaches the turbine blades is high-pressure gas or steam. (Images provided by world-nuclear.org.)
coolant at the top. The coolant temperature then gradually increases between the bottom and the top of the core. The
These
rate of change of the temperature dT/dz depends on the shape of the axial power profile and the mass flow rate m.
parameters can then be used to determine how hot the fuel rods become.
Sutherland Correlation
1
( )
µ GAS = A T 2 1 + Β Τ (13.31)
where A and B are empirically determined constants and T is the absolute temperature of the gas. Notice that we
only have to measure the viscosity at two different temperatures to determine the values of these coefficients. The
other thing that the Sutherland correlation does is to modify the equation μ GAS = C √T we wrote previously because
we can now write
1
µ GAS = C(T)T 2 (13.32)
508 Nuclear Reactor Fluid Mechanics
FIGURE 13.19 The kinematic viscosity of some common reactor coolants (water and helium) compared to conventional motor oil.
where C(T) = A/(1 + B/T). For air at atmospheric pressures, the values of A and B are A = 1.46 × 10 −6 kg/(m s °K½) and
B = 110.4°K. The values of A and B for other gases can be found at the following URL:
http://en.wikipedia.org/wiki/Viscosity
and it is usually measured in units of reciprocal poise, that is, cm s/g. The fluidity is seldom used in nuclear science and
engineering, but references to it do appear from time to time.
The kinematic viscosity is usually represented by the Greek letter ν (nu). The kinematic viscosity turns out to be a very
convenient parameter when calculating the Reynolds number Re, which expresses the ratio of the inertial forces of a
fluid to its viscous ones:
where L is the length scale of the system for which the Reynolds number is to be found and v is the velocity of the cool-
ant. In the case of a circular pipe, the value of L is typically the diameter of the pipe. Then, the relationship between the
Reynolds number and the kinematic viscosity becomes
Re = vD/ν (13.36)
where v is the average velocity of the fluid and D is the diameter of the pipe. The kinematic viscosity also appears in
some definitions of the Nusselt number which we will discuss in Chapter 20. The variation of the kinematic viscosity for
some common materials with temperature is shown in Figure 13.19. For example, the viscosity of the motor oil in your
car falls as the temperature is raised. The viscosities shown here are evaluated at atmospheric pressure (1 atm), and their
values may increase or decrease as the temperature is changed.
FIGURE 13.20 In any fluid, the attractive forces applied to a submerged molecule by the surrounding molecules balance each
other out, and there is no net electrostatic force exerted on the submerged molecule although there is a gravitational one. But the
attractive forces acting on a surface molecule are not symmetrical, and this creates a net attractive force on the surface molecule that
tends to pull the molecule toward the interior of the liquid. This force imbalance creates the phenomenon of surface tension, and
value of the surface tension depends on the strength of the molecular bonds.
glass surface. If you fill the same container with mercury, you will notice that the mercury curves downward when it
touches the glass surface. This effect is usually described by saying that the water “wets” the glass surface because
it sticks to it, while the liquid mercury does not. The strength of this wetting process is determined by the contact
angle θ between the fluid and the surface. A liquid is said to wet the surface when θ < 90° and a liquid is said to not
wet the surface when θ > 90°. Here, the wetting angle θ is defined as the angle that the tangent to the surface of the
liquid makes with the solid surface. The contact angle for both wetting and non-wetting fluids is shown in Figure
13.21. The degree of wetting is also a function of the strength of the molecular bonds (see Figure 13.20).
FIGURE 13.21 The degree of wetting of a surface is a function of the strength of the molecular bonds. The strength of these bonds
is the difference between a liquid that wets a surface well and one that does not. A reactor coolant such as liquid sodium has a high
degree of wetting and therefore adheres to a fuel rod very well. Liquid water does not adhere to a fuel rod as well as liquid metals
do. Chemicals called surfactants can be added to liquids to increase their ability to adhere to a hot surface. Different liquids can wet
a surface in different ways. The contact angle θ between the liquid and the surface is a function of the surface tension σ. The higher
the surface tension, the larger the contact angle will be. (Picture provided by biolinscientific.com.)
BIBLIOGRAPHY 511
The Pressure Difference across a Spherical Bubble as a Function of the Surface Tension
∆p = 4σ SURFACE R (13.37)
The Pressure Difference across a Hemispherical Bubble as a Function of the Surface Tension
∆p = 2σ SURFACE R (13.38)
for hemispherical ones (i.e., for ones that stick directly to the surface). Here, the value of Δp is given by Δp = pIN − pOUT
where pIN is the internal pressure and pOUT is the external pressure of the bubble. Notice that the internal pressure must
always be greater than the external pressure for a bubble to form. These equations can also be used to understand the
process of bubble formation in reactor fuel assemblies when nucleate boiling occurs. We will examine nucleate boiling
and how it affects the nuclear heat transfer rate in Chapter 23. Finally, surface tension can be reduced by adding impuri-
ties to the wetting fluid. In fact, it is possible to design certain chemicals called surfactants to decrease the surface ten-
sion of a particular liquid by weakening the molecular bonds between the adjacent molecules. This causes fluids such as
water to better adhere to certain surfaces. Sometimes additional chemicals are added to the reactor coolant to improve
the heat transfer rate by increasing the number of nucleation sites or by decreasing the wetting angle of the coolant.
Normally, the water chemistry of the cooling water must be very carefully controlled to prevent unwanted impurities
from depositing on the surfaces of the fuel rods and fouling the tubes in the steam generators of PWRs. Some impuri-
ties in the water system can be a good thing, but too many impurities can be disastrous especially if they plate out on
the surfaces of the fuel rods and cause the rods to prematurely fail by reducing the overall heat transfer rate. Normally,
distilled water is used in reactor coolant systems to minimize the number of unnecessary minerals in the coolant. In
many countries, coal-fired power plants have adopted a similar practice as well.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M. C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, CRC Press, Boca Raton, FL (2014).
Lamarsh J. R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., New York,
NY (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi (2000).
Reactor Theory (Nuclear Parameters) DOE-HDBK-1019/2–93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, CRC Press, Boca Raton, FL (2014).
512 Nuclear Reactor Fluid Mechanics
Web References
http://www.unitconversion.org/unit_converter/viscosity-dynamic.html.
http://www.gordonengland.co.uk/conversion/dynamic_viscosity.htm.
http://en.wikipedia.org/wiki/Viscosity.
1. What is the difference between an enclosed flow and a free or open flow?
2. How is a Newtonian fluid defined?
3. In the definition of a Newtonian fluid, what is the constant of proportionality between the stress and the strain?
4. What is the difference between a flow with friction and an inviscid one?
5. What is the difference between the dynamic viscosity and the kinematic viscosity?
6. Does the absolute viscosity of most liquids increase or decrease with temperature?
7. How is the Mach number defined?
8. What is the difference between a sonic and a subsonic flow?
9. In ordinary air, at what speeds can flow be considered to be subsonic?
10. What is the difference between laminar and turbulent flows?
11. At what Reynolds number does an enclosed flow normally become turbulent?
12. What is the difference between a transitional flow and a turbulent flow?
13. What is the difference between a compressible and an incompressible flow?
14. At what Reynolds number does an open flow normally become turbulent?
15. In reactor fluid mechanics, is the pressure considered to be a scalar or a vector?
16. What is the name of the thin layer of fluid next to the surface of a nuclear fuel rod where most of the heat transfer
in a nuclear reactor occurs?
17. How is the Reynolds number defined?
18. What is the difference between a forced and unforced flow?
19. What is the difference between a steady and unsteady flow?
20. Under what conditions does natural convection occur in reactor fuel assemblies?
21. What is a typical Reynolds number in a PWR fuel assembly?
22. What is a typical Reynolds number in a BWR fuel assembly?
23. What is a typical Reynolds number in a liquid metal fast breeder reactor (LMFBR) fuel assembly?
24. At what velocity does the coolant flow through the core of each of these different reactor types?
25. Why is the flow normally directed upward through a nuclear reactor core?
26. Are there any commercial nuclear reactors in which the flow through the core is not vertical? If so, what are the
names of these reactors?
27. In reactor fluid mechanics, what is the state postulate?
28. Does the absolute viscosity of most gases increase or decrease with temperature?
29. What is the definition of the vapor pressure?
30. What is the difference between the partial pressure and the vapor pressure?
31. What is a streamline and under what conditions do streamlines occur?
32. What is the purpose of the Sutherland correlation?
33. What is surface tension, and how is it defined?
34. Why is the value of the surface tension important at the surface of a nuclear fuel rod?
35. Under what conditions can a bubble form on the surface of a nuclear fuel rod?
36. What is the condition for a bubble to remain stable after it leaves the surface of a nuclear fuel rod?
37. What is the difference between surface tension and surface energy?
38. How is the wetting angle defined in reactor fluid mechanics?
39. What is the condition for a bubble to form on the surface of a nuclear fuel rod?
40. What is the definition of the fluidity?
41. What is the difference between the viscosity of a fluid and its fluidity?
42. What is the definition of a surfactant, and can surfactants be used in nuclear power plants?
43. How is the volumetric expansion coefficient of a fluid defined?
44. What is a water hammer, and how does one form?
Exercises for the Student 513
45. How can a water hammer be detected, and how can a water hammer be prevented from forming?
46. Write an equation for the volumetric expansion coefficient for an ideal gas.
47. What is the definition of the compressibility factor for a nonideal gas?
48. What is the difference between an intensive property of a fluid and an extensive one?
49. Write down at least one form of the ideal gas law.
50. What is the difference between the gas constant and the universal gas constant?
51. What is the relationship between Boltzmann’s constant, Avogadro’s number, and the universal gas constant?
52. How can the universal gas constant be converted to the gas constant for a common gas such as air?
53. How many unique gases does a sample of ordinary air contain?
54. What is the speed of sound in air at sea level?
55. What is the purpose of Antoine equation? What is its range of applicability?
56. What are several common units for the absolute (or dynamic) viscosity? What particular unit is used in most reactor
work?
57. What are several units that are used for the kinematic viscosity?
58. In what units is the surface tension usually measured?
59. How are the partial pressures of the individual gases within ordinary air defined?
60. What is the difference between the density and the specific gravity of a reactor coolant?
61. Name two reactor coolants that are used in water reactors and two reactor coolants that are used in gas reactors.
62. What are the densities and the viscosities of these coolants during normal reactor operation?
63. Why is water chemistry in nuclear reactors so important?
64. Name at least one metallic coolant that is used in LMFBRs.
65. Is the surface tension of a metallic coolant greater than, less than, or the same as that of a nonmetallic one?
66. How does surface tension affect the rate of bubble formation on the surface of a nuclear fuel rod?
Exercise 13.2
Water flows through a reactor core at approximately 6 m/s. At what fraction of the speed of sound does the water flow?
Exercise 13.3
Using the ideal gas law (P = ρRT) and the relationship c2 = γ(∂P/∂ρ)T where γ = cp/cv, show that the speed of sound in the
gas is given by c = √γRT. Is the gas constant in this case the universal gas constant or the molar gas constant?
Exercise 13.4
Water flows through a horizontal reactor coolant pipe in which changes in elevation can be neglected. When the flow
at any point along the pipe is the same and we can write m
is steady, the mass flow rate m = ρVA = constant. Using the
chain rule and dividing by the mass flow rate, show that this leads to the useful relationship
Exercise 13.5
Neglecting elevation changes, the conservation of energy for a steady flow with no heat addition or work extraction can
be written as
ρh + ρV 2 2 = constant
or
514 Nuclear Reactor Fluid Mechanics
h + V 2 2 = constant
dP/ρ + V dV = 0
Exercise 13.6
Combining the equations from Exercises 13.4 and 13.5, show that this leads to
(
dA/A = dP/ ρ 1 V 2 − dρ dP )
Then, using the relationship c2 = (∂P/∂ρ)S show that
(
dA/A = dP ρV 2 1 − Ma 2 )
where Ma is the Mach number. This is an important relationship for isentropic flow in reactor coolant pipes since it
describes the variation of the pressure with the flow area.
Exercise 13.7
Using the equation derived in Exercise 13.6, show that for subsonic flow (Ma < 1), dA and dP must have the same
sign, and thus, the pressure of the coolant must increase as the flow area increases and the pressure of the coolant must
decrease as the flow area decreases. Also show that for supersonic flow (Ma > 1), the opposite effect occurs and the
pressure of a fluid must increase as the flow area decreases.
Exercise 13.8
Another important relationship can be derived from the equation in Exercise 13.6
(
dA/A = dP/ ρV 2 1 − Ma 2 )
by substituting ρV = −dP/dV into the equation above. The result is
(
dA/A = − dV/V 1 − Ma 2 )
This equation governs the design of nozzles and diffusers in nuclear power plants (see Chapter 29). Noting that A and V
are positive quantities, it also implies that
Show how these equations can be used to predict the flow speed out of a converging nozzle if the inlet speed and the
area change are known. Why can the flow speed within the nozzle never exceed the sonic v elocity in this case? When
the sonic velocity is reached, the flow leaving the nozzle is then said to be choked (see Chapter 29). At which point does
this choking occur? In what other applications in nuclear science and engineering is choked flow important? Hint: Refer
to Chapter 30.
14
Fluid Statics and Fluid Dynamics
14.1 Static Behavior of Fluids
Before we discuss how fluids behave in a reactor when the coolant pumps are turned on, we would first like to discuss how fluids
behave when the coolant pumps are turned off. In a fresh core where there is no decay heat, the coolant can be considered to be
stationary. The subject of how fluids behave when they are stationary is called fluid statics, and the study of how fluids behave when
they are in motion is called fluid dynamics. Normally, the static behavior of fluids is easier to understand than the dynamic behavior
because we do not have to track the motion of the fluid through the core. When a fluid is at rest, the forces acting on the fluid are bal-
anced, and the pressure on a fluid molecule is the same in all directions. Classically, the pressure p is defined as the force that a fluid
exerts on a unit surface normal or perpendicular to the applied force.
In the SI system of units, the pressure is measured in N/m2 (Pa), and in the British system of units, it is measured in lb/in.2 (PSI). The
pressure exerted by a fluid on a solid surface is shown in Figure 14.1. Hydrostatics is the study of the forces that a fluid exerts on a
submerged object such as a dam or a nuclear fuel rod.
The hydrostatic force that has the greatest effect on the behavior of static fluids is gravity. Once the size of the gravitational accel-
eration is specified, the pressure at every point in a fluid becomes a function of depth. The product of the depth, the density, and the
gravitational constant also give rise to buoyancy-induced forces, which are important during natural convection.
FIGURE 14.1 The pressure at any point in a fluid is the same in all directions as long as a fluid particle is at rest. Furthermore, the pressure
exerted by a fluid on the surface of a wall always acts perpendicular to the surface of the wall.
515
516 Fluid Statics and Fluid Dynamics
14.2
P ressure at a Point
Before discussing how the pressure of a fluid varies with depth, we would first like to say more about how the pressure
is defined. Fluid pressure is normally measured in pascals, where 1 pascal (1 Pa) is equivalent to 1 N/m2:
1 Pa = 1 N m 2 (14.2)
In most nuclear applications, the pascal is too small to be of any practical use, and so pressure in reactors is measured
in kilopascals where 1 kPa = 103 Pa or megapascals, where 1 MPa = 106 Pa. The coolant pressure in the primary loop
of most light water reactors varies from about 7 MPa (or 1,050 PSI) in boiling water reactors (BWRs) to about 15.5 MPa
(or 2,250 PSI) in pressurized water reactors (PWRs). Fluid pressure can be thought of as a compressive force that is
exerted on the fluid in all directions. Superficially, this might lead one to believe that the pressure is a vector because
it has both magnitude and direction. However, the pressure exerted on any point in a stationary fluid is the same in all
directions, and therefore, the pressure in a stationary fluid is a scalar and NOT a vector.
14.3
P ressure in Horizontal Planes
A consequence of what we have just said is that the pressure of a stationary fluid does not change in the horizontal
direction. In other words, it stays the same in the horizontal plane because otherwise there would be a force imbalance
that would set the fluid in motion. This behavior has many practical applications because it implies that as long two fuel
assemblies are not isolated from each other (i.e., they are not surrounded by separate sheaths or “cans”), the fluid pres-
sure in each fuel assembly will tend to equalize itself horizontally. The pressure at each axial elevation is then
where g is the acceleration of gravity and h is the depth of the fluid beneath the surface. Thus, the variation in the pres-
sure in a reactor fuel assembly is no different than the variation in the water pressure in a lake (see Figure 14.2). The
deeper the water is, the greater the pressure will be. The gravitational acceleration g used in Equation 14.3 has a value
of 9.807 m/s2 at sea level, and its value decreases slightly with increasing elevation. For example, on the top of Mount
FIGURE 14.2 The pressure P of a static fluid increases linearly with depth if the fluid is incompressible. The magnitude of the
pressure change is illustrated in the figure above.
14.3 Pressure in Horizontal Planes 517
Everest, which is 8,800 m high, its value is about 9.764 m/s2, which is about four-tenths of one percent less (0.434%) than
it is at sea level. Hence, the value of g can be assumed to be constant no matter where a nuclear power plant is built. If
a fluid does not have a well-defined surface, then Equation 14.2 can be interpreted as the pressure below a certain point
in space which can exist anywhere in the fluid. Thus if the ambient pressure at a specific point in space is pAMBIENT, then
the pressure at a depth d beneath this point is
Normally, pAMBIENT is taken to be the reactor operating pressure, although this does not always have to be the case. If
we do a force balance on a small element of a stationary fluid, it can be shown that the variation in its pressure with
elevation is given by
where the negative sign signifies that we have taken the positive z direction to be upward so that dP is negative when
dz is positive. Hence, in reactors when the fluid is static, the pressure decreases with increasing elevation. When the
variation in the density with the elevation is known, the pressure difference between two arbitrary points 1 and 2 can be
determined by integrating Equation 14.5. The result is
since g can be considered to be constant. Equation 14.6 also applies to compressible fluids as well as incompressible ones
(see Figure 14.3). Thus if the pressure at point 1 is known, the pressure at point 2 is
2
p2 = p1 − g
∫
1
ρ(z) dz (14.7)
FIGURE 14.3 In a reactor core, the pressure field attempts to equalize itself in the horizontal direction—just like it does in the
water behind a dam. In other words, at a given depth below the top of the core, the lateral pressure is essentially the same. This is
true as long as the coolant channels are simply connected and there are no “cans” or other obstacles in the way. In PWR cores, this
is almost always the case.
518 Fluid Statics and Fluid Dynamics
14.4
Pascal’s Law of Pressure
Because of what we have just said, it should come as no surprise that the pressure in a fluid at rest is independent of the
size or the shape of the container in which it is placed. Thus, in a reactor core with no heat addition, the pressure only
changes with elevation, but it does not change horizontally as long as it is at rest. In the 1850s, a Dutch mathematician
by the name of Simon Stevin asserted that all fluids behave in a similar manner as long as they are directly connected. In
other words, as long as the points where the pressure p is measured belong to the same fluid and are at the same height,
the pressure will be the same. Stevin’s principle is illustrated in Figure 14.4. A direct consequence of his principle is that
any additional pressure applied to a fluid in a closed container increases the pressure throughout the container by the
same amount. In other words, it does not matter what the shape of the container is or where it is located. For example,
suppose that a reactor operator increases the pressure in a reactor pressure vessel by 100 PSI using a pressurizer. The
pressure of the water after the pressure increase is related to the pressure of the water before the pressure increase by
where p′ is the new pressure and Δp is the size of the pressure increase.
The mathematical principle embodied in Equation 14.8 is called Pascal’s law of fluid pressure, and it plays an
important role in such diverse applications as hydraulic lifts and antilock brakes. Refer to Figure 14.5 for more informa-
tion concerning its origins. For example, since the pressure of a connected fluid can be increased by the same amount
anywhere, we can take two hydraulic cylinders with different cross-sectional areas A1 and A2, and if the fluid beneath
them is at the same pressure p, then the force exerted on one cylinder is given by F1 = p1A1 and the force exerted on the
second cylinder is given by F2 = p2A2. Since p1 = p2, the ratio of the forces exerted on the cylinders is
p1 = p2 or F1 A1 = F2 A 2 (14.9)
( )
F2 = F1 × A 2 A1 (14.10)
FIGURE 14.4 The pressure beneath a dam is the same at every point in a horizontal plane as long as all of the points are simply
connected within a given volume of fluid. Thus, the pressure at points A, B, C, D, E, F, and G is the same, while the pressure at
points H and I is not. This is true regardless of the shape of the container in which the fluid is placed. However, this does not apply to
two different fluids such as water and oil that are not simply connected. In this figure, these points correspond to locations H and I.
14.5 Measuring the Pressure Level in a Tank, a Pipe, or a Reactor Pressure Vessel 519
FIGURE 14.5 French scientist Blaise Pascal first proposed Pascal’s law of fluids in the early 1800s. However, Pascal was also an
accomplished mathematician and philosopher, and made many additional contributions to the field of fluid mechanics. One of his
most famous quotes was “I am fascinated by the power of flies. They are so mighty they can win battles, paralyze our minds, and
eat up our bodies.” It is said that Napoleon Bonaparte uttered this phrase when he withdrew from the Battle of Waterloo in 1815.
(Pascal’s picture was provided by Wikipedia.)
Thus, if the pressure beneath both of the cylinders is the same (because they are at the same elevation), then exerting
a force F1 on cylinder 1 will cause a much greater force F2 to be exerted on cylinder 2 (as long as A2 > A1). The area
ratio A2/A1 in this case is referred to as the mechanical advantage, and many reactor components take advantage of this
principle as well.
(
F2 = F1 × A 2 A1 )
(14.11)
( )
where A 2 A1 = Mechanical advantage
Notice that the pistons that are used to perform this mechanical work do not have to be cylindrical for Pascal’s law to
apply. In fact, they can have any shape at all as long as the pressure beneath them is the same. These force ratios are
illustrated in Figure 14.6. Pascal’s law also applies to pressure measurement devices used in nuclear power plants.
14.5
Measuring the Pressure Level in a Tank, a Pipe, or a Reactor Pressure Vessel
For example, we can measure the pressure in a tank, a pipe, or even the reactor pressure vessel with a device called a
manometer. However, today, most manometers have been replaced by digital pressure sensors that rely on microproces-
sors to measure the ambient pressure. Sensors of this type are also used in modern cars to remotely sense the local tire
pressure. The system in which they are used is then called a TPMS (or Tire Pressure Monitoring System). The electronic
sensors used in reactors are similar, but are designed to be much more reliable and sophisticated. Manometers were
originally introduced in the 1600s to measure small and moderate pressure differences in pipes and tubes. A manometer
consists of a small glass or plastic tube bent into the shape of a “U,” and because of its shape (see Figure 14.7), this tube
is called a “U-tube.” The working part of the tube contains a liquid such as mercury, oil, water, or even alcohol. To keep
manometers from becoming too large, heavy fluids such as mercury are sometimes used. Consider for the moment the
simple manometer shown in Figure 14.7. This manometer is used to measure the pressure level in a tank of gas. Since
the gas fills the entire tank, and gases are not particularly dense, the pressure everywhere in the tank is approximately
the same. In other words, we can neglect the value of ρ GAS hg when calculating pressure differences between the top
and the bottom of the tank. Because of Pascal’s law, the pressure also does not vary in the horizontal direction. This
means that the pressure at point 1 must be the same as the pressure at point 2. Hence again, p1 = p2. Now suppose that the
differential column of fluid to the far right of the tank is exposed to the atmosphere. Then, the pressure at point 2 must
be the atmospheric pressure pATM plus the pressure due to the fluid in the tube. Thus, the pressure at point 2 is
FIGURE 14.6 An illustration of force ratios and the mechanical advantage which can be inferred from Pascal’s law.
FIGURE 14.7 A U-tube manometer used to measure the pressure in a tank of gas. In this case, the working fluid is mercury.
and of course, this is just the pressure of the gas in the tank (pTANK = p2). We can then write
Here, ρ is the density of the working fluid that is used. There are many different types of manometers, and their designs
depend on their application. However, in each case, Pascal’s law can be used to infer the value of the pressure.
or
∆p = (ρgz) BOTTOM – (ρgz)TOP (14.14b)
where the subscript TOP refers to the top of the core and the subscript BOTTOM refers to the bottom. If we can further
treat the coolant as being incompressible, then the density does not depend on elevation, and even if the temperature
of the fluid changes, the density will probably not. Then, the pressure drop due to elevation changes alone is given by
∆p = ρg∆z (14.15)
where Δz = zBOTTOM − z TOP is the elevation difference between the top and the bottom of the core. It is instructive to
examine the size of this pressure difference for a static reactor core filled with water. If the coolant is at room tempera-
ture, the density is approximately 1,000 kg/m3. Since a reactor fuel assembly is about 4 m high, and the local gravita-
tional acceleration is g = 9.81 m/s2, the total pressure difference between the top and the bottom of the core is
Since 1 kPa = 1,000 N/m2, the pressure head due to gravity alone is Δp = 39.24 kPa (~5.8 PSI) when the core is cold. In
other words, the pressure difference is about 10 kPa for each meter of axial elevation for a cold, clean core. When the
core heats up, the pressure of the coolant in a PWR is regulated to keep it from boiling. The standard operating pressure
in a PWR core is about 2,250 PSI (~15.5 MPa), and the coolant has an average temperature of about 310°C. In this case,
the density falls to about 700 kg/m3, and the pressure head due to the force of gravity alone is 27.5 kPa (~4 PSI). Here
again, we have used the fact that 1 kPa = 0.145 PSI. Again, these estimates are based on the assumption that the cool-
ant is at rest and the fuel assemblies are about 4 m long. Figure 14.8 depicts how the air pressure behaves as a function
of elevation in the Earth’s atmosphere. In this case, there is a significant difference between the pressure at the top of
Mt. Everest (the highest point on Earth) and the pressure at the surface of the Dead Sea (which is the lowest).
14.7
P ressures for Layered Fluids
Now, suppose that a number of fluids having different densities are stacked on top of each other. Over time, the densest
fluid settles to the bottom and the lightest fluid (fluid 1) rises to the top (refer to Figure 14.9). If there are three fluids, then
the pressure level at the bottom of the stack is
FIGURE 14.8 For a substance like air, the pressure is an exponential function of elevation because it is compressible. However, the
water in a reactor fuel assembly or behind a dam is incompressible, and so its pressure is a linear function of elevation as long as it is
not moving.
522 Fluid Statics and Fluid Dynamics
FIGURE 14.9 When several layers of fluid are stacked together, the pressure change across a single layer of fluid having density
ρ and height h is ΔP = ρgh. When different fluids having different densities are combined in layers (like layers of oil and water),
Equation 14.20 must be used to find the total pressure change.
The pressure level as a function of axial position from the top to the bottom of the stack is also shown. We can general-
ize what we have just said to a container with N different fluids, each of which has a different density ρN, by writing the
differential pressure Δp between the top and the bottom of the stack as
If all the fluids have the same density, then this expression reduces to
∆p = ρg ( h1 + h 2 + h 3 + + h N ) = ρgh (14.19)
which is, of course, what we would expect. The layering of fluids is very rare in nuclear power plants (except for water–
steam mixtures), and so one rarely encounters a situation where it is necessary to use Equation 14.17 in practice. Now,
let us turn our attention to how the pressure changes when a fluid flows through a horizontal pipe.
14.8
Measuring the Pressure Drop in a Horizontal Pipe with a Monometer
Manometers are used to measure the pressure drop in a horizontal pipe when there is fluid flowing through it, and they
can also be used to determine the effect of an area change on the pressure as one moves through the pipe. In this case,
the design of the manometer is slightly different than the ones we previously discussed; however, the principle that is
used to measure the pressure change is exactly the same. Suppose that we would like to use a manometer to measure the
pressure difference between points 1 and 2 in Figure 14.10. The fluid in this case can be either a liquid or a gas. However,
suppose for the sake of argument that it is a liquid with density ρ1. The density of the fluid in the manometer is ρ2, and
the height of the fluid is h. Then, the relationship between the pressure at point 1 and the pressure at point 2 is
p2 = p1 − ( ρ2 − ρ1 ) gh (14.20)
∆p = p1 – p2 = ( ρ2 − ρ1 ) gh (14.21)
Notice that points 1 and 2 can be separated by any distance at all. As long as there is enough space to accommodate
the physical dimensions of the manometer, the differential pressure drop along the pipe can be found in this way. In
practice, this pressure drop is due to wall friction, and we will have more to say about this when we study viscous fluids
14.8 Measuring the Pressure Drop in a Horizontal Pipe with a Monometer 523
FIGURE 14.10 A manometer used to measure the pressure drop in a horizontal pipe.
in Chapter 17. If the fluid flowing through the pipe is a gas instead of a liquid, we can then assume that ρ1 ≪ ρ2 and the
pressure drop Δp can be found from
∆p = p1 – p2 ≈ −ρ2 gh (14.22)
The barometric pressure of air is measured in a similar way. Only in this case, it is measured using an instrument called
a barometer. A barometer is constructed from a tube similar to the ones used in manometers except that it is perfectly
straight (see Figure 14.11). As we discussed earlier, air pressure is a function of elevation. It is 101.325 kPa at sea level
(1 atm = 101.325 kPa) and about 26.5 kPa at an elevation of 10,000 m. Hence, the atmospheric pressure on top of Mt.
Everest is about one-fourth of what it is at sea level. Conversely, the atmospheric pressure at the surface of the Dead Sea,
FIGURE 14.11 The density of air is almost negligible. However, a column of atmospheric air weighs enough to drive a column
of extremely dense mercury almost a meter up a glass vacuum tube. This is the basic principle used to create a barometer, which
measures atmospheric pressure.
524 Fluid Statics and Fluid Dynamics
FIGURE 14.12 In either SI or English units, the boiling point of water is affected by the elevation. It is 100°C at sea level, but at
8,800 m (at the top of Mt. Everest), it is only 70°C or 160°F. The boiling point of other liquids is affected in a similar way.
which is the lowest point on Earth at an elevation of 400 m below sea level, is 800 torr. The pressure at which water and
other reactor coolants boil is also affected by changes in the atmospheric pressure as well. The boiling point of water at
sea level is 212°F (100°C), but at 8,800 m (on the top of Mt. Everest), it is only 160°F (70°C). The variation in the boiling
point with pressure is shown in Figure 14.12. Normally, this is not important in reactors because the pressure level is
so high. However, it is important when attempting to understand the difference between a PWR and a BWR. We would
briefly like to describe this difference in the next section.
14.9
P ressure Levels in Reactor Cores
For cold clean cores, the gravitational pressure head is the primary component of the pressure drop. However, as soon
as the coolant pumps are turned on and the core begins to heat up, the gravitational pressure head, which is given by
ρgΔz, is no longer the largest component of the total pressure drop. The primary loops of American PWRs are designed
to operate at system pressures of about 2,250 PSI, and those of American BWRs are designed to operate at system pres-
sures of about 1,050 PSI. These pressures are selected to optimize the thermal efficiency of the plant. In PWRs, the
system pressure is made as high as possible in the primary loop to prevent the coolant from boiling. This is done using
a large pressure regulation tank called a pressurizer (see Chapter 2). Water from the core is converted into steam in
the secondary loop using a device called a steam generator. Normally, the pressure in the secondary loop is only about
40% of the pressure in the primary loop. BWRs do not have secondary loops because the coolant boils directly in the
core. The steam produced in the core is then sent to the steam turbines. Because of this, BWRs do NOT have steam
generators and the system pressure in the primary loop is closer to that of the secondary loop of a PWR. The core pres-
sure drop in a PWR averages 25 to 30 PSI, and in a BWR, it averages about 25 PSI. Hence, the static pressure head in
a PWR core (which is about 4 PSI) represents only 15% of the total pressure drop when the coolant pumps are running
at normal speed. The rest of the pressure drop can be attributed to frictional losses and the expansion and acceleration
of the coolant.
14.10
P ressure Equalization between Reactor Fuel Assemblies
As long as reactor fuel assemblies are directly connected—that is, as long as there are no physical barriers between
them—the coolant can flow between these assemblies to equalize the lateral pressure differences between them. Thus, as
far as the core is concerned, reactor fuel assemblies are also subject to Pascal’s law, which we introduced to the reader
in Section 14.4. Consider for a moment the pictures of the PWR and BWR cores shown in Figure 14.13. In each case,
the fluid enters the core from below and exits the core from above. As the coolant flows through the core, it is gradually
heated, but if the fuel assemblies do not have exactly the same radial and axial temperature profiles, then the coolant in
some parts of the core will become hotter or colder than the coolant in other parts of the core. This causes density differ-
ences to develop between subchannels or between the fuel assemblies themselves. These density differences then give rise
to lateral pressure differences at the same axial elevation. In an attempt to equalize these pressure differences, a small
14.10 Pressure Equalization between Reactor Fuel Assemblies 525
FIGURE 14.13 PWR and BWR cores have different numbers of fuel assemblies, and in general, pressure differences can develop
between individual coolant channels in both PWRs and BWRs. In PWRs, pressure differences can also develop between adjacent
fuel assemblies. This gives rise to lateral core flows or cross-flows. Normally, the coolant flows from hotter fuel assemblies to colder
ones in the radial direction.
amount of coolant may flow laterally between the subchannels or between the fuel assemblies themselves. This lateral
flow is called cross-flow, and in nuclear applications, it is given the symbol w. Under normal conditions, the lateral mass
flow rate is about 1% of the axial mass flow rate. It is large where the radial temperature gradients are large and small
where the radial temperature gradients are small. It is also affected by how the fuel assemblies are loaded into the core.
In PWRs, cross-flow can occur between subchannels within the same fuel assembly (see Figure 14.14), or it can
occur between adjacent fuel assemblies because PWRs are open cores. Between assemblies, this effect is illustrated in
Figure 14.15. Cross-flow of this type is called intra-assembly flow. Thus, once the coolant enters a flow channel, it can
flow into an adjacent flow channel as long as there are no obstructions to prevent the flow from occurring. Allowing
cross-flow to occur between these subchannels and fuel assemblies generally results in a more uniform radial tempera-
ture profile as long as the coolant does not boil. Hence, it is generally considered to be beneficial by reactor designers
because it helps to optimize the thermal performance of the core.
FIGURE 14.14 Cross-flow between adjacent subchannels can occur if the pressure in one reactor coolant channel is different than
the pressure in another. In general, the coolant flows from the hot subchannels to the colder ones. Cross-flow is caused by lateral
pressure differences induced by different power generation rates in the radial direction.
526 Fluid Statics and Fluid Dynamics
FIGURE 14.15 In PWRs, the pressure field in the horizontal direction is no different than it is in a lake (see Figure 14.4) as long
as the coolant pumps are turned off. Thus, the pressure attempts to equalize itself in a horizontal plane. When the coolant pumps
are turned on, the pressure field equalizes itself in a different way, but it does so by allowing coolant to flow laterally through the
core. In reactor thermal-hydraulics, this lateral or traverse coolant flow is called cross-flow. In the picture above, cross-flow occurs
between adjacent fuel assemblies. It can also occur across the tube banks in reactor heat exchangers (see Chapter 8).
In BWRs, the core can be made more neutronically and thermally stable by encasing the individual fuel assemblies in
what are called sheaths or “cans.” Because of the use of these “cans,” BWRs can be considered to be “closed cores”
where no intra-assembly cross-flow is allowed. The coolant can only flow between the coolant channels within an indi-
vidual fuel assembly. Thus, cross-flow can still occur in BWRs, but it only occurs between coolant channels within an
assembly and NOT between adjacent assemblies. Lateral flows are also more complicated to analyze when the coolant
boils because the liquid and vapor phases do not move at the same speed. In other words, a condition of slip can exist
between the two phases. We will have more to say about the origins of this slip in Chapter 23.
14.11
Cross-Flow in Rod Bundles
In “open” fuel assemblies and “open” cores, radial pressure gradients may cause the lateral flow rate to approach
1%–2% of the axial flow rate. For example, in PWR fuel assemblies, this causes the fluid from the hotter channels to
mix together with the fluid from the colder channels. This lateral flow of mass and energy helps flatten the temperature
profiles by redistributing the thermal energy produced by the fuel rods. Hence, when cross-flow occurs, the temperature
profiles are usually more uniform than when it does not.
14.12
P ressure Drops in Open and Closed Cores
Core-wide pressure profiles in open and closed cores are very similar, but between the inlet and outlet, the shape of the
pressure gradients can be different. Gravity and other forces attempt to equalize these pressure gradients, while differ-
ent radial and axial temperature profiles tend to accentuate them. In general, the pressure gradients in closed cores are
larger than they are in open cores—especially when a phase change occurs. We would now like to compare the behavior
of a PWR core to that of a BWR core. Our comparison will help to highlight some of the conceptual differences.
14.13
P ressure Behavior in PWRs
Suppose for a moment that we take a closer look at how the water molecules in a PWR behave when the coolant pumps
are running. Then, the water molecules are no longer at rest, and they try to distribute themselves radially so that
the molecular pressure within a horizontal plane remains the same. Within an individual fuel assembly, the coolant
will also accelerate or decelerate due to the presence of support structures called grid spacers or wire wraps. When the
14.13 PRESSURE BEHAVIOR IN PWRs 527
coolant encounters these support structures, small lateral pressure differences may temporarily develop between adja-
cent coolant channels at the same axial elevation. To equalize these lateral pressure differences, a small percentage of
the fluid may also flow laterally between the high-pressure coolant channels and the low-pressure coolant channels. This
is especially true when the coolant channels are next to each other and therefore hydraulically connected. An illustration
of a typical cross-flow pattern in a PWR core is shown in Figure 14.16. BWR cores do NOT have intra-assembly cross-
flow, as we have tried to indicate in Figure 14.17.
FIGURE 14.16 PWR and BWR cores support different types of cross-flow. In PWRs, cross-flow can occur between fuel assem-
blies, but in BWRs, it cannot. Cross-flow in BWRs is primarily confined to individual fuel assemblies because each of these fuel
assemblies is surrounded by what is known as a sheath or a “can.” The can is usually made up of zirconium or stainless steel.
FIGURE 14.17 In BWR cores, hydrodynamic forces cause the pressure drops to be the same between the bottom and the
top of each assembly, but within a given fuel assembly, they may be considerably different as a function of height because the
fuel assemblies are not simply connected. PWR cores behave differently than BWR cores because they are open and simply
connected. Because of this, the pressure at a specific axial location in a PWR core is essentially uniform in a horizontal plane.
The core-wide pressure drop in a PWR is between 26 and 28 PSI, and in a BWR core, it is between 24 and 26 PSI. The numbers
quoted represent industry averages, and they can vary somewhat from one design to the next.
528 Fluid Statics and Fluid Dynamics
14.14
P ressure Behavior in BWRs
Now, let us consider how the same water molecules behave in a BWR core. In BWRs, the fuel assemblies are surrounded
by thin cans or sheaths to keep the core neutronically and thermally stable. Thus, every BWR fuel assembly is isolated
hydraulically from every other BWR fuel assembly. However, they will still attempt to maintain the same pressure head
Δp = pIN − pOUT between the inlet and the outlet of the core, and in order to do this, the axial pressure gradients between
the inlet and the outlet may be different. Moreover, to maintain the same core pressure head, the axial mass flow rates
can be different from one assembly to the next. In fact, the frictional pressure drops within the individual assemblies can
become flow regime dependent; that is, they can be different for subcooled boiling than they are for bulk boiling or film
boiling. Moreover, the locations of these transition points can vary from one fuel assembly to the next. To maintain the
same pressure head across all assemblies, the inlet flow rates may also be different. For example, the flow rates may be
lower through hot fuel assemblies than they are through colder ones. At high power-to-flow ratios, this can sometimes
cause one flow regime to transform itself into another. In Chapter 31, we will discover that this leads to a condition called
a Ledinegg instability. Normally, these instabilities can be avoided by following a reactor operating plan. This requires
the reactor operators to operate the plant within an approved range of conditions defined by an operating map. Examples
of these operating maps were first provided in Chapter 3.
14.15
Typical Pressure Levels in the Nuclear Steam Supply System
Most of the time, the pressure levels in BWR cores are significantly lower than they are in PWR cores. In BWRs, the
average core pressure is about 1,050 PSI (7.24 MPa), and in PWR cores, the average core pressure is about 2,250 PSI
(15.51 MPa). There are a number of reasons for such a large difference. First, in PWRs, the coolant is designed not to
boil because of the higher pressures and the power densities involved. The coolant in the primary loop remains primar-
ily a subcooled liquid that enters the core at a temperature of about 290°C and leaves the core at a temperature of about
320°C. (The exact numbers can vary based on the reactor vendor.) For example in Westinghouse PWRs, and the AP1000
in particular, the average inlet and outlet temperatures are 280°C and 320°C, respectively. However, in advanced PWRs
manufactured by AREVA, the core inlet and outlet temperatures are 295°C and 330°C, respectively. Thermal energy
is extracted from the fluid in the primary loop using steam generators. The steam generators connect the primary and
secondary loops together, and the pressure in the secondary loop of each steam generator is much less than the pres-
sure in the primary loop. Thus, the coolant in the secondary loop is intended to boil. Steam generators are discussed in
Chapter 8. The lower pressure in the secondary loop is required to generate steam and also to feed steam to the other
components of the NSSS. These components include the steam turbines and the condensers. The pressure of the steam
leaving the steam generators in the secondary loop averages 6–7.6 MPa (870–1,100 PSI), and the temperature of the
coolant when it leaves the steam generators is 270°C–290°C.
The pressure of the steam drops slightly before it enters the power turbines because of piping losses and frictional
effects. It then returns to the steam generators as a subcooled liquid with a temperature between 225°C and 235°C. The
exact values are shown for different designs in Table 14.1. Each of the steam turbines in a nuclear plant has between
three and six substages. Each of these substages uses a different blade design to optimize the useful energy that can be
extracted from the steam. In BWRs, there is no need to use steam generators because the coolant is allowed to boil in
the core. This boiling eventually becomes bulk boiling and the steam that is produced in this way can be sent directly
to the steam turbines. However, it is normally dried by the steam separators (or steam driers) at the top of the pressure
vessel to reduce the moisture content of the steam as much as possible. This prevents water droplets from forming in
the steam. If the moisture content of the steam is too high, it may damage the blades of the turbines and reduce their
operating life. After accounting for pressure drops in the piping system, the steam enters the steam turbines at a pres-
sure of about 1,000 PSI and it eventually ends up exiting the turbines with a pressure of about 1 PSI. At this pressure,
its bulk temperature is about 30°C and its equilibrium quality is 10%–12%. After passing through the condensers, it is
then reheated and returned to the core. In two-stage turbines (see Chapter 9), about 30% of the power is extracted by
the high-pressure (HP) stage and the remaining 70% is extracted by the low-pressure (LP) stage. The exact ratios are
presented in Chapter 9. HP and LP turbines can also have separate substages in which the steam is extracted and can be
reheated several times.
Thus, the pressure difference Δp between the inlet and the outlet of these turbines is what forces the turbine blades
to turn. The spinning shaft is then connected to an electrical generator that generates the electricity the plant produces.
Upon leaving the turbines, the spent steam is sent to a device called a condenser to be recondensed. The external compo-
nents of the condenser (i.e., the components of the condenser that are not directly connected to the primary or secondary
loops) are maintained at an ambient pressure close to 1 atm or about 7% of atmospheric). This allows the most energy
to be extracted from the steam because heat is rejected from the thermal cycle at very low temperatures (see Chapter 9).
Normally the pressure drop across the core is limited to about 28 PSI (see Chapter 18), and it is distributed roughly
14.15 Typical Pressure Levels in the Nuclear Steam Supply System 529
TABLE 14.1
Representative Pressures and Temperatures in the Core and Steam Generators of Some Common
PWRs and BWRs Used for Commercial Power Generation
Primary Loop
Pressure (MPa) 15.5 15.5 15.5
Inlet temperature (°C) 281 295 289
Outlet temperature (°C) 321 329 324
Core flow rate (kg/s) 13,500 21,000 22,000
Secondary Loop
Pressure (MPa) 5.7 6.7 6.9
Inlet temperature (°C) 227 230 236
Outlet temperature (°C) 273 292 283
Source: Todreas and Kazimi (2008).
equally between frictional losses due to the coolant flowing over the fuel rods and frictional losses caused by the grid
spacers and the support plates. These overall pressure drops are summarized in Figure 14.18. The numbers quoted here
represent industry averages and the actual numbers may be higher or lower. In BWRs, the total core pressure drop is
about 25 PSI. Most of this pressure drop can be attributed to the bulk boiling of the coolant. By the time the coolant
reaches the core midpoint (see Figure 14.19), large bubbles develop in the flow, and the flow transitions from bubbly flow
to slug flow. The shear forces between these bubbles and the liquid film on the fuel rod surface are responsible for most
of the pressure drop that occurs in the upper half of the core. In PWRs, the coolant hardly boils at all. A small amount of
nucleate boiling is sometimes allowed to occur in the hottest fuel assemblies, but by design, this only occurs near the top
of the core. However, this boiling is usually negligible compared to the boiling that occurs in a BWR. The differences
in the thermal behavior of these cores are shown in Figure 14.20.
FIGURE 14.18 In a PWR, the pressure in the primary loop is about twice the pressure in the secondary loop. The pressure drop
across the primary loop is about 120 PSI (~0.8 MPa), and about half of this occurs across the steam generators. The pressure drop in
the secondary loop is about 900 PSI (~6 MPa), and most of this occurs across the steam turbines. Between points 3 and 4, the pres-
sure drops dramatically and the steam becomes very wet. The pressure in the condenser is normally less than 100 kPa because the
steam is so wet.
530 Fluid Statics and Fluid Dynamics
FIGURE 14.19 The coolant temperature profile in a typical BWR core. Once the saturation temperature TSAT is reached, the cool-
ant temperature stays the same until all of the water in the coolant channel is converted into a steam. As the coolant flows upward
through the channel, it undergoes a number of physical changes which lead to different flow regimes and flow patterns.
FIGURE 14.20 In BWRs, bulk boiling starts to occur close to the center of the core, but in PWRs, the coolant hardly boils at all
and a small amount of nucleate boiling is allowed to occur in the hottest fuel assembly at the top of the core.
14.17 Fluid Kinematics and Fluid Dynamics 531
14.16
F luid Dynamics—The Behavior of Fluids in Motion
Fluids in motion behave differently than fluids at rest. This is because the local pressure can be affected by the way in which a
fluid moves. Sometimes fluids in motion move at the same speed everywhere, and sometimes they do not. When the velocity
of a fluid changes, a force is applied to the fluid that causes it to accelerate or decelerate. This acceleration or deceleration can
result in a pressure change that is interpreted as a pressure drop or a pressure gain. The magnitude of this change can affect
the way the core of a reactor behaves. The motion of a fluid can be described by either Lagrangian or Eulerian fluid mechan-
ics (see Section 14.17), and each of these descriptions of fluid mechanics leads to an important parameter called the material
derivative. The material derivative is needed to understand the motion of fluids when area or direction changes occur.
14.17
F luid Kinematics and Fluid Dynamics
In classical fluid theory, fluid kinematics is the study of fluids in motion without necessarily considering the forces that
are responsible for their motion. Fluid dynamics also considers the forces that lead to this motion. When considering
the motion of fluids, the behavior that is perceived depends on how fast an observer is moving with respect to the fluid.
There are two basic ways to describe this behavior called the Lagrangian view of fluid mechanics, and the Eulerian view
of fluid mechanics. The Lagrangian description was first proposed by Joseph-Lewis LaGrange around 1750, and the
Eulerian description was proposed by Swiss mathematician Leonhard Euler about 10 years later. Pictures of LaGrange
and Euler are shown in Figure 14.21. In Eulerian fluid mechanics, a set of conservation equations is used to describe how
a fluid behaves when it moves past a stationary observer or object. A separate conservation equation is used for (1) the
conservation of mass, (2) the conservation of energy, and (3) the conservation of momentum. Each conservation equa-
tion is derived with the help of an imaginary region of space called a control volume. The control volume is stationary
with respect to the observer. The resulting set of equations is a system of hyperbolic equations that describes the flow.
In their simplest form, these equations are applicable to what are called adiabatic or inviscid flows. In reactor work, the
energy equation is modified to make them applicable to non-adiabatic flows.
Now let us turn our attention to Lagrangian view of fluid mechanics. In Lagrangian fluid mechanics, the trajectories
of individual fluid particles are tracked using Newton’s laws of motion. When the paths of enough of these particles
are found, the overall motion of the flow can be easily understood. Each particle is assigned a vector, which identifies
its velocity and its position. Thus the Lagrangian description (see Figure 14.22) has many similarities to a Monte Carlo
simulation of the nuclear particles in a reactor core*. Unfortunately, this description can easily become cumbersome
in practice especially when the trajectories of large numbers of fluid particles must be found. Hence in the limit where
FIGURE 14.21 Louis Lagrange (left) and Leonhard Euler (right) are some of the founders of the modern field of fluid mechanics.
(Pictures of LaGrange and Euler were provided by Wikipedia.)
* See for example, An Introduction to Nuclear Reactor Physics by Robert E. Masterson, CRC Press (2017).
532 Fluid Statics and Fluid Dynamics
FIGURE 14.22 In Lagrangian fluid mechanics large numbers of individual particles must be tracked which possess mass, energy,
momentum, and other important properties. Mathematical laws can be written to describe the behavior of each of these particles. The
resulting equations can then be solved to determine where each of these particles goes. By tracking the motions of enough of these
particles, one can then infer the overall behavior of the particles in the continuum (See http://slideplayer.com/slide/5755461/.)
the aggregate motion of these particles is used to represent the behavior of the continuum, the Lagrangian description
becomes equivalent to the Eulerian description. Unfortunately the paths of literally millions of fluid particles have to be
tracked to understand the intricacies of the flow.
14.18
T he Lagrangian and Eulerian Views of Fluid Mechanics
In reactor work, almost all reactor design is based on the Eulerian view of fluid mechanics. Eulerian fluid mechan-
ics uses the concept of a control volume or a unit cell to derive the equations that describe the motion of the fluid.
In Chapter 15, we will see that one form of these equations is the Navier-Stokes equations. An example of such a control
volume is shown in Figure 14.23. A control volume like this can have any size or shape, although it is usually easiest to
understand Eulerian fluid mechanics when the control volume is rectangular and has dimensions Δx, Δy, and Δz. In this
case, the space enclosed by the control volume is
∆V = ∆x∆y∆z (14.23)
If the control volume becomes infinitesimal, then Equation 14.23 becomes dV = dxdydz and we can use several field
variables to describe the state of the flow. There is then one “set” of field variables for each control volume. Examples of
these field variables are the pressure p, the density ρ, the specific enthalpy h, the velocity V, and the acceleration a. Some
FIGURE 14.23 A control volume can be used to illustrate the Eulerian description of fluid mechanics. In this way of looking at
fluids, variables such as the pressure and the density are defined within a control volume, and mass and energy are allowed to flow
across the surfaces of the control volume in response to pressure differences. Almost all reactor thermal-hydraulics is based on an
Eulerian description of fluid mechanics.
14.20 Understanding the Differences between Advection, Diffusion, and Convection 533
field variables (such as the temperature and the density) are normally located at the center of each cell, which others (such
as the fluid velocities) are located on the boundaries. However, other field variables can also be used. Most of the variables
in Eulerian fluid mechanics are used to represent scalar fields, while other variables are used to represent vector fields.
For example, the enthalpy h (x, y, z) is normally a scalar field, while the velocity and the acceleration are normally vector
fields. The difference between Lagrangian and Eulerian fluid mechanics can be most easily understood by considering the
flow of fluid relative to a stationary observer who is standing on the bank of a river and is watching the water flow by. In
Lagrangian fluid mechanics, the person throws a stick into the river and watches it flow downstream. Thus, its velocity and
its location as it is swept along with the water describe how it is viewed from a Lagrangian point of view. On the other hand,
Eulerian fluid mechanics is represented by someone standing in a shark cage in the middle of the river (thus in this example
we are assuming the river has sharks). The person watches the water flow into and out of the cage, and he or she measures
its temperature and pressure as it moves. However, as soon as the water leaves the cage, its properties and its direction are
no longer measured by the observer. From this perspective, the “cage” corresponds to the Eulerian control volume shown
in Figure 14.24. Thus, the Eulerian description of fluid mechanics is more appropriate for describing the motion of large
ensembles of particles, while the Lagrangian description is more appropriate for describing the motion of just a few.
14.19
Finding the Acceleration of an Eulerian Fluid
Now let us discuss how the acceleration of a fluid particle is described in the Eulerian frame of reference. If the fluid is
acted upon by an external force, it accelerates or decelerates in response to this force. A fluid can also accelerate or
decelerate if the channel through which it is flowing expands or contracts. Moreover, it can accelerate or decelerate if its
direction changes, or if its density changes due to a pressure or temperature change. Finally, the density can be changed
by adding energy to the fluid or taking energy away from it, in which case the acceleration or deceleration is propor-
tional to the density change. In the next few sections, we would like to use these concepts to introduce the reader to
Eulerian fluid mechanics, and our discussion will eventually lead to an important parameter called the material deriva-
tive. However, before discussing the material derivative, we would first like to outline some of the processes that are
responsible for its existence. In particular, we would like to discuss the relationship between advection, diffusion, and
convection in a reactor core. These processes are sometimes thought to be the same, when in fact, they are completely
different. Now let us explore some of the reasons for these differences.
FIGURE 14.24 In Eulerian fluid mechanics, individual fluid particles are not tracked as they are in the Lagrangian description. Instead,
a volume of space or a control volume is defined through which the particles flow. The average properties of the particles within this vol-
ume are calculated, and the control volume is considered to be fixed in space and time. Hence, Eulerian fluid mechanics does not follow
the paths of individual fluid particles as they move. Equations can be derived for each description of particle flow, and these equations can
be converted from one description to the other. In practice, this involves the use of the material derivative, which is discussed in the text.
534 Fluid Statics and Fluid Dynamics
Thus, convection is the sum of both advection and diffusion. Advection is sometimes confused with the process of diffusion,
but these processes refer to entirely different things. Diffusion does NOT require the bulk motion of a fluid, while advec-
tion does. In diffusion, mass and energy transport occurs at the molecular level and no bulk motion of a fluid is required.
Advection is also confused with the more encompassing process of convection. However, convection is normally defined
as advection plus diffusion (see Equation 14.24). Sometimes it is helpful to think about advection as a process that combines
the effects of both vector and scalar fields. In this context, the velocity field V can be thought of as a vector field in which
a scalar field such as the internal energy u or the specific enthalpy h is imbedded. These combined fields then define the
transport properties of the medium. A practical example of the difference between the processes of advection and diffusion
can be seen by observing the motion of a red dye that has been injected into a lake or river (see Figure 14.25). As the water
in the river flows downstream, the dye moves with it because it is “advected” by the current. If the same amount of dye is
added to a lake where the water does not move and V = 0, then the dye will simply diffuse outward from its injection point.
Since the water in the lake is stationary, there is no advection of the dye. However, in the river, the dye will be trans-
ported downstream and it will also spread out as it moves because of molecular diffusion. The sum of these two effects
then constitutes the convective behavior of the dye. Normally, advection is a much faster process than diffusion, and
because of this, diffusion can usually be neglected in reactor work. However, the rate that convection occurs is affected
by whether the flow is laminar or turbulent, and whether the flow is single phase or two phase. Sometimes one encoun-
ters a situation where solid particles are inter-mixed with the fluid and eventually become entrained inside it. When this
occurs, the particles can move at different speeds than the fluid and their collective behavior is described by what is called
a transport equation. The shear forces between the solid particles and the surrounding liquid then determine how fast the
particles move. Transport equations are important in the study of reactor accidents where radioactive particles and material
objects may become entrained in a water–steam mixture during the blowdown phase of a LOCA. The equations that gov-
ern this behavior are discussed in Chapter 28. When the flow is laminar, the drag forces the fluid exerts on these particles
can then be found using what is called Stokes Law.
14.21
Finding the Acceleration of a Fluid in an Eulerian Reference Frame
Just like ordinary material objects, fluid particles accelerate or decelerate in response to the external forces that are applied
to them. In Eulerian fluid mechanics, these forces are applied to the particles in a control volume. The application of these
forces to the fluid in the control volume results in fluid motion that can be expressed in terms of a parameter called the mate-
rial derivative. The material derivative can then be used to find the acceleration of the fluid particles within the control
volume. In order to understand the practical importance of the material derivative, consider a fluid particle having mass
MPARTICLE located inside of a fluidic pool. If we apply Newton’s second law to the particle, the force acting on the particle is
FIGURE 14.25 In Eulerian fluid mechanics, the distinction between advection and diffusion can be seen by observing the motion
of a colorful dye in a river or a lake. The dye moves with the current in the river because it is advected by it; however, even when
there is no current, the dye can still move due to molecular diffusion. In a lake, where there is very little current, the dye simply
spreads out in a circular pattern. Here, the dye is shown in red.
14.21 Finding the Acceleration of a Fluid in an Eulerian Reference Frame 535
where aPARTICLE is the particle’s acceleration. Since the acceleration is the first derivative of the velocity (when the mass
is held constant), we may also write the equation governing the motion of the particle as
However, the velocity of the particle VPARTICLE consists of an x component u, a y component v, and a z component w. In
three dimensions, this enables us to write
We can then find the value of the acceleration by taking the total derivative of the velocity vector with respect to x, y, z,
and t. Using the chain rule,* the total derivative of the velocity becomes
where the symbol ∂ refers to the partial derivatives of the individual components. Now, consider the partial derivatives
∂x/∂t, ∂y/∂t, and ∂z/∂t that appear in the expression for the particle’s acceleration (Equation 14.29). They are simply the
values for the particle’s velocity along the x, y, and z directions. In particular, we can write
y velocity: v = ∂ y/ ∂ t (14.30b)
since ∂t/∂t = 1. This allows us to write the total acceleration of the particle as
where ∂V/∂t is called the local acceleration and u ∂V/∂x + v ∂V/∂y + w ∂V/∂z is called the global acceleration. In vector
notation, the total acceleration can also be written as
where V again refers to the velocity vector. In other words, in Eulerian fluid mechanics, the acceleration vector consists
of a time-dependent term ∂V/∂t that only appears when the velocity is changing, and a space-dependent term (V ∇V)
that may be present even when the flow is constant. The space-dependent term is then called the advective derivative.
We would next like to explore the implications of the advective derivative.
* The chain rule is a method for finding the derivative of a function when it is contained within another function. For example, when
function g is contained within function f, the resulting function is called a composite function. The derivative of the composite
function is then ( f ⋅ g(x) )′ = g′(x)f ′ ( g(x) ) or dy/dx = (dy/du) × (du/dx).
536 Fluid Statics and Fluid Dynamics
14.22
Finding the Advective Acceleration in an Eulerian Reference Frame
From what we have just said, it is easy to see that the advective acceleration (V ∇V) that appears in Equation 14.31 can
exist for even steady-state flows; in other words, it can exist for flows that are time-invariant.
Thus, the advective acceleration applies to a fluid particle moving to a new location in the flow field, and at this location,
its velocity may be different than it was at its previous location. Now, let us demonstrate what the advective derivative
looks like under these conditions. In Cartesian coordinates, the gradient operator is
∇ = i ∂ / ∂x + j ∂ / ∂ y + k ∂ / ∂z (14.34)
The important thing to notice about this equation is that the directional components of the advective acceleration ax = u
∂u/∂x, ay = v ∂v/∂y, and az = w ∂w/∂z are based on the velocity changing as a function of position. They do not require
the flow pattern itself to change as a function of time.
14.23
Some Observations Regarding the Advective Derivative
Now, let us take a moment to make some additional comments about the behavior of the advective derivative. As
Equation 14.33 suggests, the advective acceleration (V ∇V) can cause the flow to accelerate even when the fluid velocity
is independent of time—as long as there is a velocity gradient somewhere in the flow. This implies that the advective
acceleration (V ∇V) that appears in Equation 14.33 accelerates the flow by “advecting” mass, energy, and momentum
from one point in space to another. The consequences of this are illustrated in Figure 14.26. Thus even steady flows can
be accelerated, and the magnitude of this acceleration can be described by the advective derivative. Now, let us discuss
how the motion of a fluid particle is affected by the sign of this derivative. If the advective terms are positive, the particle
accelerates, and if the advective terms are negative, the particle decelerates. This acceleration or deceleration then has
the following effect on the pressure:
If (V ∇V) > 0 → fluid is accelerating → the pressure is dropping
If (V ∇V) = 0 → fluid is unaffected → the pressure is constant
If (V ∇V) < 0 → fluid is decelerating → the pressure is rising
FIGURE 14.26 The material derivative D/Dt is composed of a local or time-dependent part and a convective or advective part. The
first part disappears for steady-state flows, but the advective part does not. The material derivative can be derived directly from the
chain rule. In this particular figure, a fluid particle is accelerating to the right as it moves.
14.24 The Material Derivative in Eulerian Fluid Mechanics 537
In other words, area changes can cause the coolant anywhere in a nuclear power plant to accelerate or decelerate, and
this will in turn produce a pressure change near the area change. Because of this, there are just two scenarios where the
advective derivative does not affect the flow
1.
First, it has no affect on the flow field when there are no velocity gradients in the flow. Then V is constant, and
strictly speaking, this can only be true for frictionless flows.
Secondly, this can occur when a fluid is at rest and V = 0. Then the flow can be described by the laws of fluid
2.
statics that we discussed in Chapter 13, and because it does not move, there is no advective acceleration.
14.24
T he Material Derivative in Eulerian Fluid Mechanics
Next, we would like to examine the relationship between the advective derivative and the material derivative, which also
contains the temporal derivative of the flow. In Eulerian fluid mechanics, the chain rule (refer to the discussion at the end
of Section 14.21) allows the total derivative to be written as
or
The first part (∂/∂t) is the temporal derivative or local derivative, and the second part is the spatial derivative or advec-
tive derivative. In fluid mechanics, the sum of these two parts is then called the material derivative.
Material derivative = Local derivative (or temporal derivative) + Advective derivative
Many authors assign a special notation D/Dt to the material derivative because it appears so frequently in the study of
Eulerian fluid mechanics:
Thus, by writing
it can be seen that the material derivative can be used to describe any scalar field φ(r, t) that has a spatial or temporal
dependence. Thus if the enthalpy is a scalar field and the velocity is a vector field, we can write
The material derivative can also be applied to other properties of the flow. For example, it can be applied to the tempera-
ture of the fluid, its chemical concentration C, and even to its pressure. For the pressure, it becomes
Then Equation 14.42 represents the rate of change of the pressure in a coolant channel because the fluid in the channel
is accelerating or decelerating. This pressure change consists of a local part (∂p/∂t) that applies to time-dependent
flows and a global or advective part (V ∇p) that applies to steady flows (which are also called temporally invariant).
538 Fluid Statics and Fluid Dynamics
Example 14.1 shows how the local acceleration can be deduced from changes in the velocity field for a two dimen-
sional flow. In this case, the flow is a steady flow.
14.25
Heat and Energy Transfer in Reactor Fuel Assemblies
The material derivative can also be used to deduce the flow of energy through a reactor coolant channel. In this case, the
energy of the fluid is represented by the specific enthalpy h, which is also a scalar field imbedded in the velocity field V:
Equation 14.43 must then be solved to find where the energy deposited in the coolant goes. Much of the study of reactor
thermal-hydraulics involves finding solutions to this equation. A simplified form of this equation can be used to perform
an energy balance on a reactor piping system. Here, the material derivative for enthalpy transport is
If additional energy is added to the coolant (by heat produced in the fuel rods), the enthalpy transport equation becomes
where Q is the rate of heat addition to the coolant. Normally the rate of heat addition is assumed to be position depen-
dent, and in reactor fuel assemblies, it is usually written as Q(z), where Q(z) = q″(z)A. Here q″(z) is the nuclear heat
flux and A is the surface area of the fuel rods through which the heat flows. As Equation 14.45 suggests, the rate of heat
addition can also be time dependent. This can become important in certain types of transients where the power level can
change by 10% or 20% in a relatively short period of time. The consequences of these transients must be analyzed in the
Final Safety Analysis Report or FSAR.
14.26
Connecting the Lagrangian and the Eulerian Descriptions of Fluid Mechanics
Finally, the material derivative D/DT can be thought of as the fundamental link between Lagrangian and Eulerian
fluid mechanics. When we apply the material derivative to a Lagrangian system, the spatially dependent advective terms
This equation then describes the acceleration of a fluid particle from a Lagrangian perspective. Similarly, if we know
the value of aPARTICLE in the Lagrangian representation, then we can find the acceleration of the particle in the Eulerian
representation by including the advective terms
14.27 Rotational Flows 539
FIGURE 14.27 The coolant in a reactor piping system passes through various valves, fittings, bends, elbows, tees, inlets and outlets, and
enlargements and contractions as it moves. These components interrupt the flow of the coolant and cause the pressure to change because
of flow separation and turbulent mixing. In a typical piping system with long pipes, these losses are small compared to the total pressure
drop across the pipes. The losses induced by valves, bends, elbows, area changes, and tees are normally expressed in terms of a loss coef-
ficient k. The loss coefficient can be found from the head loss (HL) using the equation k = HL/v2/(2g) where the head loss for a particular
component is given by HL = ΔPL/ρg. The value of ΔPL is found by measuring the additional pressure drop that the component creates.
u ⋅ ∂v / ∂x + v ⋅ ∂v / ∂ y + w ⋅ ∂v / ∂z (14.48a)
to the transport equation describing the advection of the flow. Normally, advection is only important when the fluid
velocity changes because of a change in the flow area or a change in its direction. Then the velocity gradient, ∇V
becomes finite, and the fluid can accelerate in a number of different ways. The total acceleration is then given by
Equation 14.31. This fluid behavior can occur when the coolant flows through a valve (see Figure 14.27), around a bend
in a pipe, or around a grid spacer in a fuel assembly (see Chapter 18). It can also occur in the vicinity of a flow blockage.
Figure 14.28 shows how the flow behaves under these conditions. Normally the flow accelerates as it moves around the
blockage, and this causes the local pressure to change.
14.27
Rotational Flows
Although rotational flows sometimes occur in nuclear power plants, their rotational momentum is generally considered
to be small compared to their linear momentum. Hence, rotational effects are normally ignored in the design of the
NSSS. However, in certain situations, such as when one is dealing with the blades of a steam turbine or the impellers
of a coolant pump, the rotational energy that is imparted to the coolant can become significant. Here we would like to
discuss the sources of this rotational motion, and we would also like to develop an equation to describe its magnitude
and direction. There are two ways that the rotation of a reactor coolant can be described. The first way which is called
circulation measures the macroscopic rotation of a large number of fluid elements across a defined region of space. The
second way which is called vorticity measures the microscopic rotation of a fluid element about a particular point inside
this region of space. These two quantities are related, but they are not the same because the circulation is a scalar while
the vorticity is a vector. Thus the circulation has only magnitude, while the vorticity has both magnitude and direction.
The circulation, represented by the symbol C, is the amount of force that pushes fluid elements along an enclosed bound-
ary or path. It is defined as the line integral along the contour of the component of the velocity vector V that is locally
tangent to the contour. Mathematically, it can be described by the equation
∫
C ≡ V × dl (14.48b)
where the line integral starts and ends at the same point. Thus it represents the sum of the velocities parallel to the sur-
face of an enclosed loop. A negative value for C means that the flow is clockwise and a positive value for C means that
it is counterclockwise. Naturally, a value of C = 0 means that there is no net rotation. In other words, Equation 14.48(b)
measures the macroscopic rate of rotation of the flow. If a circular ring of coolant with radius R rotates at angular veloc-
ity ω about the z axis, its velocity is given by V = ω x R, where R is the distance from the axis of rotation to the ring of
coolant. Thus the circulation around the ring is
540 Fluid Statics and Fluid Dynamics
FIGURE 14.28 In reactor fuel assemblies, grid spacers and flow blockages can also cause the coolant to accelerate and decelerate.
The picture above shows how the coolant accelerates and decelerates around a flow blockage. Reactor thermal-hydraulics codes and
computational fluid dynamics codes can be used to simulate the effect of these unexpected blockages on the fuel rod temperatures in
a reactor fuel assembly.
∫
C ≡ V × dl = 2ω π × R 2 = 2ω A (14.48c)
where A = π × R 2. In other words, the circulation is just 2 π times the angular momentum of the coolant around
its axis of rotation. Conversely, the circulation divided by the area enclosed by the circulating loop is just twice the
angular speed of rotation of the coolant ring (when the rotation is measured in rad/s). The vorticity vector φ then
measures both the magnitude of this rotation and its direction at a microscopic level. This local behavior can be
more readily understood using local derivatives of the velocity vector V rather than a line integral of the velocity
along an enclosed path.
Not surprisingly, the rotational energy imparted to a fluid is also an indirect measure of the efficiency of a cool-
ant pump. Before beginning a discussion of this subject, we would first like to introduce some additional terms to
the reader.
A rotational flow can be thought of as a volume of fluid rotating in a circular direction with an angular velocity ω.
The angular velocity measures the rate of rotation of the fluid. Some fluids are harder to set in motion than others, and
this is due to both their weight and their viscosity. A fluid with a higher viscosity is harder to set in motion than a fluid
with a lower viscosity because its internal friction is higher. Similarly, heavier fluids are harder to rotate than lighter
ones. Reactor coolant pumps (or RCPs) and other mechanical devices like blowers or compressors work by converting
the mechanical energy of a spinning shaft into fluid momentum and rotational energy. In rotating fluids, large elements
of the fluid can rotate or deform (see Figures 14.29(b) and 14.30). The rate of rotation of a hypothetical fluid element
about a fixed point in space can then be described by the equation
14.27 Rotational Flows 541
FIGURE 14.29 Fluid elements in a reactor core are subject to various types of translation, rotation, and deformation. Some of the
most common types of motion and deformation are illustrated above.
FIGURE 14.30 Fluid flow can give rise to rotation, deformation, and strain. Strain tensors can be used to quantify the amount of strain
that occurs in a compressible fluid. These tensors have the same mathematical properties as other tensors and can be used to calculate the
amount of volumetric dilation or strain. This picture illustrates a general two-dimensional flow field in which all possible types of trans-
lation, rotation, and deformation can occur. The entries in the strain tensor can be found from a specific knowledge of the velocity field.
when the fluid is Newtonian. In other words, Equation 14.49 can be used when the stress is proportional to the strain. In
three dimensions, Equation 14.49 must be replaced by another expression since the direction of rotation can be different
along each axis. Then the total rate of rotation is equal to an angular velocity vector ω, which in Cartesian coordinates
has separate components along the x, y, and z directions:
Each of these components is then equivalent to the difference between two velocity gradients which are different for
each direction. For the x direction, these gradients are dw/dy and dv/dz, for the y direction they are du/dz and dw/dx, and
for the z direction they are dv/dx and du/dy. In other words, the value of ω depends on the size of these gradients as well
as their sign. Normally, the rate of rotation ω is expressed in radians per second (rad/s). Notice that the components of
ω are based on an explicit knowledge of the velocity field. If the rotational rate at a particular point in the flow field is
finite, then the fluid that happens to occupy this point must rotate and the flow surrounding this point is said to be part
of a rotational flow. However, if the value of ω is negligibly small (or sometimes zero), then the fluid particles around
this point are not rotating, and the flow is said to be an irrotational flow. Normally, reactor flows are at least partially
rotational. Once a fluid is set in motion, it continues to rotate until some external force (such as friction) stops its rota-
tion. Hence, frictional forces eventually dissipate its rotational energy, and how long this takes depends on the rate of
rotation, the dimensions of the pipe, and the velocity at which the coolant is flowing through the pipe. It also depends on
the dynamic or kinematic viscosity and the local temperature. We will now discuss some convenient ways to measure
the rotational energy of a flow.
14.28
T he Vorticity of a Rotational Flow
The vorticity is a commonly accepted way to measure the rotational motion of the flow around a point. Values for the
vorticity can be defined for both laminar and turbulent flows, but the degree of vorticity is much higher for turbulent
flows than it is for laminar ones. If the vorticity at a particular point is greater than zero (or less than zero), then a fluid
particle that happens to occupy this point is rotating, and consequently, it possesses some angular momentum or spin.
The flow in the region of space surrounding this point is then said to be rotational. Conversely, if the vorticity is zero
(or very small), then the flow is said to be irrotational. In nuclear power plants, the coolant always has some vorticity
because it is constantly agitated (to increase the heat transfer rate), but the direction of this rotation (and hence the sign
of the vorticity) can vary from one point to the next. Thus, the vorticity can be considered to be a pseudo-vector that
describes the rotational energy of a spinning fluid. Most of the time, its value is both spatially and temporally dependent.
In most textbooks, the vorticity is represented by the vector φ (pronounced phi), but this symbol is not universally
accepted and other symbols such as χ (chi) or even Ω (omega) are also used to describe it. Like the angular rotation vector
ω, which is measured in radians per second, the value of the vorticity is measured in radians per second (rad/s). The rota-
tional energy is then proportional to the vorticity. Thus, the higher the vorticity, the higher the rotational energy will be.
The magnitude of the vorticity vector is highest in regions of the fluid where the velocity gradients are high, and lowest in
regions of the fluid where the velocity gradients are low. For example, close to the surface of a fuel rod, the vorticity is rela-
tively high. When the flow is laminar and the Reynolds number is low, the flow develops a boundary layer close to the fuel
rod surface. Within this boundary layer, the vorticity is high because the velocity gradients are high. Outside the boundary
layer, the flow becomes irrotational and the vorticity becomes low. Both of these situations can be seen in Figure 14.31. In
fact, far away from the boundary layer, the coolant may have no vorticity at all. This behavior is illustrated for a cylindrical
FIGURE 14.31 Vorticity is created in boundary layers when the fluid elements rotate. Velocity gradients cause the flow to rotate
within the boundary layer, but above the boundary layer, the flow becomes irrotational and the vorticity falls to zero.
14.28 The Vorticity of a Rotational Flow 543
FIGURE 14.32 Outside of a boundary layer, a fluid is generally free of vorticity. Vorticity arises when velocity gradients within
the boundary layer cause the fluid particles to rotate in a preferred direction and acquire spin. In some cases such as the one shown
above, the boundary layer can detach from an object although it can continue downstream of it. This is a common phenomenon
when dealing with cylinders and spheres.
object in Figure 14.32. In nuclear power plants, the same phenomenon also occurs in circular pipe flow. Notice that a fluid
does not necessarily have to be turbulent for there to be some vorticity. In fact, when the flow is laminar, the boundary
layer can have very steep velocity gradients, and since these velocity gradients are large, the fluid particles will rotate in
a direction defined by the gradient. Hence the vorticity can be significant close to the surface of a fuel rod - regardless of
whether the flow is laminar or turbulent. Now let us examine how the vorticity is defined. In vector calculus, the vorticity
vector φ is defined as the curl of the velocity vector
and hence the curl describes the differential infinitesimal rotation of the flow field in three dimensional Euclidian space.
In electricity and magnetism (or E&M), the curl of an electric field E also produces a time-dependent magnetic field B:
∇ × E = − ∂B / ∂t (14.52)
Consequently, the curl operator in Equation 14.51 defines the axis of rotation, and just like Equation 14.52, the direction
of the curl, which is specified by the right hand rule - see Figure 14.34, defines the direction of rotation. The magnitude
of the curl then becomes the magnitude of the rotation. Because the vorticity is always a vector, it will have different
values in the x, y, and z directions. In Cartesian coordinates, the components of the vorticity vector are
ϕ = ϕ x ⋅ i + ϕ y ⋅ j + ϕ z ⋅ k (14.53)
where
Notice that the magnitude of the vorticity depends on the size of these velocity gradients. As we mentioned earlier, the
direction or the sign of the vorticity φ is given by the right-hand rule for the cross-product. The magnitude of the vorticity
vector is then
544 Fluid Statics and Fluid Dynamics
ϕ= (ϕ x
2
)
+ ϕ y 2 + ϕ z 2 (14.55)
Thus, the vorticity is equal to twice the angular velocity of the fluid particle that is rotating (see Figure 14.33). The rate
of rotation vector ω and the vorticity vector φ are then related to each other by
FIGURE 14.33 For two-dimensional flows in the x-y plane, the vorticity vector always points in the +z or −z directions. In this
particular illustration, the fluid flow is counterclockwise so the vorticity vector φ points in the +z direction and is equal to twice the
angular velocity vector ω.
FIGURE 14.34 The direction of the vorticity is defined by the right-hand rule. The vorticity vector φ is equal to twice the angular
velocity vector ω for a rotating fluid particle.
14.29 Fluid Deformation and Volumetric Strain 545
In two dimensions, the resulting equations can be simplified a bit, and the only component of the vorticity vector lies in
the x-y plane. The value for this component is
14.29
F luid Deformation and Volumetric Strain
Just like solid objects, fluid elements can shrink, stretch, expand, or even become deformed as they move. The amount
of this expansion or contraction is one way to determine if a flow is compressible or incompressible. If a flow is
incompressible, the net volume of a fluid element will not change; in other words, if it stretches in one direction, it
must shrink by the same amount in another direction so that the mass of the element remains constant. However, if a
flow becomes compressible, a fluid element can increase or decrease in size because its density can change, but even
when it does so, its mass must remain the same. In Chapter 17, we will learn that there is a specific relationship for
Newtonian fluids between the stress τ (or force per unit area) exerted on a fluid particle and the amount of deformation
(or strain ε). This relationship is
The Relationship between the Stress and the Strain for a Newtonian Fluid
ε = τ /µ = du/dx (14.58)
In three dimensions, the strain ε can then be represented by a tensor because deforming a particle in one direction also
deforms it in several other directions. This leads to the following expressions for shear strain rate:
From the internal symmetry of the fluid, we know that εxy = εyx, εzy = εyz, and εyz = εzy. Then we can combine the linear
strain rate and the shear strain rate into single tensor that describes the deformation of a three-dimensional element of fluid:
ε xx ε xy ε xz
ε ij = ε yx ε yy ε yz (14.61)
ε zx ε zy ε zz
Not surprisingly, this tensor is called the strain rate tensor. Like other tensors, it obeys the laws of tensor mathematics that
govern the translation and transformation of objects. It can also be used to determine when a flow is compressible or incom-
pressible, and in the form presented here, it is symmetric. Now suppose that we define the volumetric strain rate ύ or the bulk
strain as the rate of change of the volume of a fluid element as it expands or contracts. This kinematic property of a fluid is
positive when the volume increases and negative when the volume decreases. It then turns out that the volumetric strain rate
ύ is the sum of the linear strain rates along each of the three mutually exclusive coordinate directions. In Cartesian space,
this allows us to write
Equation 14.62 is then called the volumetric strain rate equation and it sometimes appears in the study of computational
fluid mechanics. However, it rarely appears in reactor work, although it can still be used to demonstrate that the volumet-
ric strain rate is zero for an incompressible fluid. Example Problem 14.2 provides a simple application of this concept.
546 Fluid Statics and Fluid Dynamics
This also demonstrates that the flow is incompressible. We will examine the rotational aspects of this flow in the exer-
cises at the end of the chapter. [Ans.]
14.30
Visualizing the Flow of Fluid through a Reactor Piping System
Sometimes it is helpful visualize the flow of particles through a specific component of the NSSS such as a steam genera-
tor, a coolant pump, or even a reactor fuel assembly. Within these components, several methods can be used to visualize
the coolant flow. Two of the most popular of these methods involve the use of what are called path lines and streamlines.
Path lines and streamlines were first used to visualize open flows, such as the flow of air over an aircraft wing, but they
can also be used to visualize the behavior of enclosed flows, such as those which occur in reactor coolant pipes. Many
CFD programs now allow path lines and streamlines to be automatically displayed, and in recent times, this technology
has begun to be used by the nuclear industry. In the next few sections, we would like to discuss these visualization tools
as well as some of their similarities and differences. Normally path lines and streamlines are most appropriate to use for
laminar flows, while the behavior of turbulent flows can be most easily visualized using other techniques, such as those
described in Section 14.36.
14.31
Path Lines and Streamlines
Path lines are fundamentally different than streamlines although these two approaches to particle visualization have many
things in common. Both techniques can be used to understand the flow of fluid through a reactor coolant channel. If the
coolant were at rest, there would be no need to use path lines or streamlines. However, most CFD programs* use path lines
or streamlines to describe the trajectory of the particles they simulate. Now, let us begin our discussion with a brief introduc-
tion to path lines.
14.32
D efinition of a Path Line
Most of the time, the motion of fluid particles is best described graphically. In fluid mechanics, the definition of a path
line is
Path lines are based on the Lagrangian view of fluid mechanics. In this view, the aggregate behavior of the flow is defined
by the sum of the individual paths of all of the fluid particles. The trajectory of these particles then defines a path line.
Strictly speaking, Eulerian fluid mechanics does not require path lines to exist, but path lines are even used in Eulerian
CFD programs to visualize the flow of fluid between adjacent nodes or computational cells. Hence, a path line is the same
as the material position vector δ(x, y, z, t)PARTICLE traced out by a particle (or a set of particles) over a specified period of
time. Then, the start of the tracing defines the starting point of the path line δ(x, y, z)START and the end of the tracing defines
the end point δ(x, y, z)STOP of the path line. A representative path line is shown in Figure 14.35. Thus a path line can be
thought of as the Lagrangian equivalent of a streamline. The time-dependent particle position vector can then be written as
where x(t), y(t), and z(t) represent the instantaneous positions of the particle. If the components of this vector are known,
the particle position in the Lagrangian representation is simply
Thus, a particle’s path line is the trajectory the particle takes as it moves through space and time. In addition to path
lines, the behavior of fluid particles can also be described by streak lines and streamlines. In the next few sections, we
would like to discuss these alternative forms of visualization and show how they can be used.
14.33
D efinition of a Streak Line
A streak line differs from a path line because it represents the locus of the fluid particles that have sequentially
passed through a prescribed set of points in space and time. Normally, only one point is required to initiate a streak
line. Streak lines are probably the most natural way to visualize the flow of fluids over solid objects (such as nuclear
fuel rods). In other industrial applications, they are used to visualize the flow of water around the hull of a ship, or
the air flow over an aircraft wing. Streak lines are also used to visualize the flow of air around jet fighter wings and
over the heat tiles on the US space shuttle (see Figure 14.36). Streak lines are sometimes confused with path lines
and streamlines, but they are NOT the same thing. While all three lines are the same when a flow is steady, they
can be very different when a flow becomes unsteady or time dependent. The formal definition of a streak line is
Thus, path lines and streak lines have the shared attributes of history, time, and age. Thus, a path line is the time-
exposed flow path of an individual fluid particle over a specified period of time, while a streak line is an instantaneous
snapshot of a time-integrated flow pattern or set of particle trajectories. Streak lines can be created by injecting a
FIGURE 14.35 A path line is formed by following the actual path of a fluid particle between its starting point and its end point.
548 Fluid Statics and Fluid Dynamics
FIGURE 14.36 A picture of the streak lines created by a jet fighter plane. (Pictures provided by NASA—see Nasa.gov.)
colored dye into an airflow stream. A visual depiction of these streak lines can be seen in Figure 14.36. Many simula-
tions of jet fighter planes and unmanned space ships have been aided by the study of these lines. Streak lines are ideal
for visualizing the behavior of single phase liquids in nuclear power plants, and they can also be used to give reactor
designers a great deal of insight into what is happening in the core—particularly when the flow becomes two or three
dimensional. Now, let us turn our attention to a third form of visualization, which involves the use of streamlines.
14.34
D efinition of a Streamline
Streamlines are one of the most popular ways to visualize low speed flows. They appear in most derivations of Bernoulli’s
equation (see Chapter 16), and they can be deduced from the conservation of energy for an inviscid flow. Mathematically
speaking, a streamline (see Figure 14.38) is defined as
DEFINITION OF A STREAMLINE
“A streamline is a curve in the flow field that is everywhere tangent to the instantaneous local velocity
vector.”
Thus streamlines are snapshots of the instantaneous direction of the flow. Streamlines cannot be observed directly
except in steady flows where they are coincident with path lines and streak lines. However, it is possible to write an
equation for a streamline based on its mathematical definition alone. In the x-y plane, this equation becomes
In practice, this equation can be solved if analytical expressions are available for both u and v. However, most of the time,
these expressions are not readily available. Perhaps the easiest and most satisfying way to derive an analytical expression
for a streamline is to consider an infinitesimal arc of length dr = dx i + dy j + dz k, which represents the path of a stream-
line through three-dimensional space. Now, dr must be parallel to the local velocity vector V = u i + v j + w k because of
the way in which streamlines are defined. Using simple geometrical arguments, it can then be shown that the directional
components of dr must be directly proportional to those of the velocity vector V. Taking this ratio, we find that the equation
for a streamline becomes
where now dr is the magnitude of dr and V is the magnitude of the velocity vector: V = √(u2 + v2 + w2). A practical
application of this equation can be found in Figure 14.37, and its mathematical definition is presented in Figure 14.38. In
Figure 14.38, the final term in Equation 14.61 is ignored because the flow is primarily two dimensional.
Suppose that the x and y components of the velocity vector shown in Figure 14.38 are given by u = 0.5 + 0.8x and
v = 1.5 − 0.8y. The equation for the streamlines that develop in this flow pattern (from Equation 14.63) is then dy/dx = (0.5 −
0.8y)/(0.5 + 0.8x). This equation can then be written as:
∫ 1/ (1.5 – 0.8y) dy = ∫ 1/ (0.5 + 0.8x) dy (14.68)
FIGURE 14.37 Examples of streamlines from the automotive and aeronautical industries. Some NASCAR surface pressure
contours and streamlines are shown on the left. Airplane surface pressure contours, volume streamlines, and surface streamlines are
shown on the right. (From Wikipedia.com.)
and after some algebraic manipulation (which we leave to the reader), we arrive at the following expression for y as a
function of x along the streamlines:
Here C is a constant of integration that can be set to different values to allow us to plot individual streamlines. Several
of these streamlines are shown in Figure 14.38. Here, the velocity vectors are tangent to the streamlines, and the fluid
particles within the streamlines always stay within the streamlines. Notice that the speed of the fluid particles cannot be
determined directly from the streamlines alone.
1 =m
m 2 (14.70)
or
Thus, stream tubes can be thought of as visual representations of the continuity equation for incompressible fluids.
FIGURE 14.39 If an incompressible fluid flows through a set of parallel streamlines, then the collection of these lines is sometimes
referred to as a stream tube or flow tube. For steady flows, the flow of mass within the stream tube is always conserved. Thus, stream
tubes are a visual representation of the conservation of mass for steady incompressible flows. (Pictures provided by Wikipedia.)
Bibliography 551
FIGURE 14.40 An example of a wedge used to describe the flow of particulate matter in the wedge model
of atmospheric dispersion.
experimentally by injecting a colored dye known as a tracer into the flow stream and recording their motion with
a camera. Once the trajectories of these particles are known, the curves they follow define their path lines. Then
shortly before Y2K, a new technology called, particle image velocimetry (or PIV) emerged as a popular method
for determining the trajectories of large numbers of particles. PIV measures the velocity field over an entire flow
plane (normally, a plane like the x-y plane or the y-z plane). Different parallel planes, offset by fixed amounts, can
then be “stitched together” with a computer program to create a composite picture of the three-dimensional veloc-
ity field. With the help of these techniques three-dimensional velocity fields can now be mapped and measured on
the fly. Other techniques can also be used to find the particle trajectories in reactor fuel assemblies. Normally, the
particle velocities for the liquid and vapor phases are different, and this is a subject we will have more to say about
in Chapter 23 when we discuss what is called the slip ratio.
14.37
Reactor Accidents and Particle Entrainment Theory
Finally, particulate matter can sometimes become entrained in a moving fluid, and due to the viscosity of the fluid, this
particulate matter will move with the flow. During certain types of reactor accidents, fission products may be released
from some of the fuel rods, and some of these fission products (in the form of small- to medium-sized particles) may
become entrained in the flow. Eventually, these particles will be released into the containment building, and they have to
be tracked to determine where the radioactivity is high and where the radioactivity is low. This then leads to the subject
of particle entrainment theory and Stokes Law, which are critical to understanding the radiological consequences of a
LOCA. The diffusion of radioactive particles also plays a role in this behavior. Both of these subjects are discussed in
Chapter 28 when we also discuss the Wedge model of atmospheric dispersion. In this model, particulate matter flows into
the atmosphere in the form of radioactive plumes, and if the direction of the wind remains the same, they disperse into
the atmosphere in the form of a wedge containing several radio-nuclides. A mathematical representation of this wedge
is shown in Figure 14.40. Over time, the radio-nuclides in this wedge decay into simpler materials, and no longer remain
harmful. However, it may take several days or weeks for this to occur. Accordingly, in Chapters 15 and 28, we will turn
our attention to how the Navier–Stokes equations can be used to determine the magnitude and the direction of this par-
ticulate flow. In reactor work, the solutions to the Navier-Stokes equations are often simplified by employing empirical
correlations for the frictional terms. Additional assumptions are then made to account for the effects of turbulence on the
convective heat transfer coefficient. We will have more to say about these assumptions in Chapter 15 when we explore
implications of the Navier-Stokes energy equation.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
552 Fluid Statics and Fluid Dynamics
Glasstone, S. and Eedlund, M. C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, CRC Press, Boca Raton, FL (2014).
Lamarsh J. R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc.,
New York, NY (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi (2000).
Reactor Theory (Nuclear Parameters) DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, CRC Press, Boca Raton, FL (2014).
35. Which type of line represents the most natural way to visualize the flow of fluid over a solid object such as a nuclear
fuel rod?
36. What are the formal definitions of a path line and a streak line?
37. How is the rate of rotation of a fluid particle related to its vorticity?
38. How is the direction of the vorticity vector defined?
39. In fluid flow, the vorticity is considered to be a vector. What is its closest analog in the field of classical
mechanics?
40. When is the vorticity of a fluid the greatest?
41. How is the acceleration of a fluid found in a Lagrangian reference frame?
42. How is the acceleration of a fluid found in the Eulerian view of fluid mechanics?
43. How is the advective derivative defined in Eulerian fluid mechanics?
44. How is the material derivative related to the advective derivative?
45. What is the difference between advection and diffusion in a reactor fuel assembly?
46. Name at least two scalar fields that can be advected by a velocity field in Eulerian fluid mechanics.
Exercise 14.2
Water flows at 10 m/s through a reactor coolant pipe. If the diameter of the pipe gradually doubles over a distance of
several meters, but the elevation of the pipe does not change, what is the velocity of the coolant when it exits the pipe?
Exercise 14.3
Use the advective derivative to show that for a one-dimensional subsonic flow, the pressure decreases as the flow area
increases.
Exercise 14.4
Using the chain rule, show how the equation for a two-dimensional streamline dy/dx = v/u can be rewritten as dr/V =
dx/u = dy/v = dz/w in three dimensions. How is the magnitude V of the velocity vector V defined in this case?
Exercise 14.5
Translate the material derivative in an Eulerian reference frame to the material derivative in a Lagrangian reference
frame. What terms in the derivative are zero in this case?
Exercise 14.6
Write down the advective derivative for one-dimensional pipe flow in which the area changes. For incompressible and
isothermal flow, what is the relationship between the flow area and the velocity? Does the mass flow rate m′ change if
the area changes?
Exercise 14.7
Suppose that the two-dimensional velocity field for a steady-state, two-dimensional flow is given by V = (u, v) = (0.5 + 0.8x)
i + (1.5 – 0.8y) j where i and j are the unit vectors in the x and y directions. In this equation, the lengths are measured in meters,
the velocities are measured in m/s, and time is measured in seconds. What is the magnitude of the fluid velocity at x = 2 and
y = 3? What is the rate of rotation of the fluid particles in this flow? Is the flow rotational or irrotational? What is the value of
the vorticity? How can we determine if the flow is compressible or incompressible?
Exercise 14.8
What is the value of the advective acceleration in Exercise 14.7 at a location in the flow field where x = 2 m and y =2 m?
What is the direction of the advective acceleration at this point? Can this type of flow be expected to occur in a reactor
core? Would the flow be primarily two or three dimensional in this case?
15
The Conservation Equations
of Fluid Mechanics
15.1 T he Material Derivative
The behavior of reactor coolants can be described by a set of conservation equations called the Navier–Stokes equations. These
equations are a set of nonlinear partial differential equations based on Eulerian view of fluid mechanics, whose computational under-
pinnings were first espoused by Leonhard Euler around 1760. In Eulerian fluid mechanics, the material derivative
is used to find the rate of mass, energy, and momentum transfer between any two points in space and time. Material derivatives are
frequently applied to reactor components by subdividing these components into a number of control volumes or unit cells. The mate-
rial derivative then defines the rate of advection into and out of each cell. The material derivative also consists of time-dependent and
time-independent parts. The time-independent parts are then called steady-state parts. For any vector or scalar field φ, the material
derivative can be written as
where
∂ϕ / ∂ t = time-dependent part (15.2b)
u ∂ϕ / ∂ x + v ∂ϕ / ∂ y + w ∂ϕ / ∂ z = steady -state part (15.2c)
The time-dependent part is only required for time-dependent flows (i.e., those where the flow changes as a function of time), and the
steady-state part applies to steady-state flows where acceleration, deceleration, or advection occurs. The steady-state part is called the
advective derivative, and the advective derivative was derived in Chapter 14. It is given by the expression:
In the expression for the advective derivative, the variables u, v, and w refer to the components of the fluid velocity vector V along
the x, y, and z directions and the advected variable φ (which can be either a scalar or vector quantity) represents a property such as
ρ, h, or ρv. The energy flow through the core is proportional to the advected specific enthalpy h*, and the specific enthalpy is in turn
a function of T, p, and x:
( )
h = f T, p, x (15.4)
For single-phase liquids, this reduces to hSP = f(T, p) and the density is also a function of x, T, and p
555
556 The Conservation Equations of Fluid Mechanics
( )
ρ = f p, T, x (15.5)
In three dimensions, the Navier–Stokes equations can be very complex - especially in their original form, and in some
cases, they become so difficult to solve that they are replaced by simpler sets of differential equations that involve an
alternative treatment of fluid friction. Finally, time-averaged values for the fluid velocities ( u , v , and w ) are some-
times used to replace the fluctuating components u′, v′, and w′ of the velocity field. This reduces the complexity of the
energy and momentum equations, and it allows approximate solutions to be obtained under certain conditions.
TABLE 15.1
The Number of Equations Required to Model Fluid Flow in One, Two, and Three Dimensions for a Liquid, a Gas, or a
Homogeneous Two-Phase Mixture
Momentum Equations
Number of Dimensions Continuity Energy x Momentum y Momentum z Momentum
1 x x x
2 x x x x
3 x x x x x
15.7 Types of Fluid Flow 557
terms containing ρuv, ρuw, and ρvw are sometimes ignored when the values of v and w are small compared to the
value of u. In general, reactors are too complex to design with most CFD (or Computational Fluid Dynamics) programs
directly. Consequently, alternative approaches have been devised to replace “brute force” ones.
15.6 C
ommon Simplifications to Computational Fluid
Dynamics Models for Nuclear Reactors
Because analyzing a reactor fuel assembly is rarely a straightforward process, most reactor fuel assemblies are ana-
lyzed by “lumping” the underlying physical processes into empirical expressions called thermal-hydraulic correlations.
Normally, there are different types of correlations for account for fluid friction than there are to account for nuclear
heat transfer. Applying these correlations to large volumes of space is sometimes called a lumped parameter approach.
Other times, it is called an integral or integral-based approach. Within each region of space, all of the relevant physical
parameters may not be known with the same degree of precision. For example, many popular heat transfer correlations
have statistical errors between 2% and 5%. In reactor geometries, lumping the underlying physics into larger computa-
tional cells has the advantage of demystifying these processes as long as the regions of space over which this averaging
is done do not become too large. Normally, these regions of space are called control volumes. Almost all reactor design
is performed in this way. When using a control volume approach, empirical correlations are used for both the friction
factors and the heat transfer coefficients. These correlations are based on the local velocity of the fluid as well as the
roughness of the surface.
FIGURE 15.1 A wave breaking on a beach is a simple example of turbulent flow, and the flow over the hull of a nuclear submarine
can be either laminar and turbulent.
FIGURE 15.2 In laminar flow, the streamlines are well defined and continuous, while in turbulent flow, they are not. In enclosed
flows, such as those in reactor fuel assemblies, the transition between laminar flow and turbulent flow occurs at a Reynolds number
of about 2,300. The transition point is usually lower for open flows such as those for boat hulls and airplane wings.
We may also write the viscosity in terms of the viscous force per unit area that is exerted on the fluid, or the shear stress
τ as
µ = τ /∇V (15.7)
where τ = F/A, and the shear stress τ is the force per unit area tending to cause deformation of a material by slippage
along a plane or planes parallel to the imposed stress. Hence, the higher the viscosity of a fluid, the “stickier” the fluid
becomes, and the greater resistance it will have to internal flow. Molasses is an example of a fluid that has a very high
viscosity, and water is an example of a fluid that has a lower one. Viscosity also effects how much energy it takes to pump
a reactor coolant through the core. Perhaps the easiest way to visualize the amount of force it takes to move a viscous
fluid is to consider a small volume of fluid dV that is confined between two solid plates. An experiment that can be used
to measure this shear force is shown in Figure 15.3.
The lower plate is stationary, and the upper plate is moving in the x direction with velocity uo. Suppose that we would
like to know how much force F is required to keep the upper plate in motion if the bottom plate remains stationary.
Moreover, suppose that the plates are separated from each other by a distance L and that the lower plate is rigid and
cannot move. The internal friction of the fluid forces the fluid in contact with the lower plate to remain stationary, while
it also forces the fluid in contact with the upper plate to move with velocity uo. If we decided to run a number of experi-
ments to find how much force F was required to keep the upper plate in motion, we would find that the force is
☉☉ Proportional to the area A of the plates.
☉☉ Proportional to the velocity uo of the upper plate.
☉☉ Inversely proportional to the distance L between the plates.
Thus, we may write
F = µ Au o L (15.8)
where μ is a constant of proportionality called the dynamic or absolute viscosity. If the flow is laminar, and no turbulent
eddies are created by the motion of the upper plate, then the velocity of the fluid u between the plates can be shown to
vary linearly between z = 0 and z = L. Then, we may also write the force required to keep the upper plate in motion as
The viscous force per unit area, or the shear stress τ, is then written as
τ zx = µ ∂ u/ ∂z (15.10)
The subscript zx is used to indicate that the component of the force in the x direction is due to the velocity gradient in
the z direction. In other words, shear forces act in the opposite direction of fluid flow, and they are proportional to the
gradient of the fluid velocity du/dz perpendicular to the direction of flow. We can use this simple expression to find
the force that is required to accelerate the fluid contained within a volume of space having dimensions Δx, Δy, and Δz.
Sometimes this volume of space is also called a control volume. A control volume used to perform a force balance on a
small element of fluid is shown in Figure 15.4.
If the shear force is linear, and the shearing force on the upper surface is counteracted by an equal shearing force on
the lower surface, then all we need is a simple force balance to derive an expression for the amount of friction the fluid
possesses. Suppose that the shearing force at the center of V is τzx. Then, the stress at the upper surface may be written
The net force is the sum of viscous forces on the upper and lower faces of the element
[ τzx + ∂τzx/ ∂z∆z/ 2 ] ∆x∆y − [ τzx − ∂τzx/ ∂z∆x/ 2 ] ∆x∆y = ∂τzx/ ∂z∆x∆y∆z (15.13)
To obtain the force per unit volume, we must divide by the volume ∆V = ∆x∆y∆z of the element. The result is
F = ∂τ zx ∂z = ∂ / ∂z ( µ ∂ u/ ∂z ) (15.14)
If the dynamic viscosity µ is also a constant, this equation may be simplified to read
F = µ ∂ / ∂z ( µ ∂ u/ ∂z ) = µ ∂ 2 u/ ∂z 2 (15.15)
Thus, the shear force that the fluid exerts on itself as it moves is proportional to the second derivative of the velocity
perpendicular to the direction of flow. For three-dimensional problems, the viscous shear forces will have three separate
components, and if the viscosity is the same in each direction, the frictional force exerted by the fluid is
FIGURE 15.4 A force balance used to find the fluid shear stresses.
15.10 Singular Solutions to the Navier–Stokes Equations 561
Fx = µ ∂ 2 u ∂x 2 + ∂ 2 u ∂ y 2 + ∂ 2 u ∂z 2 = µ∇ 2 u (15.16)
Fy = µ ∂ 2 v ∂x 2 + ∂ 2 v ∂ y 2 + ∂ 2 v ∂z 2 = µ∇ 2 v (15.17)
Fz = µ ∂ 2 w ∂x 2 + ∂ 2 w ∂ y 2 + ∂ 2 w ∂z 2 = µ∇ 2 w (15.18)
where u, v, and w are the components of the velocity vector V in the x, y, and z directions. Notice that we must always
have a velocity gradient perpendicular to the direction of flow for a shear force to exist. Sometimes the dynamic viscosity
(which is volume based) is replaced by the kinematic viscosity ν (which is mass based). The relationship between these
two definitions of the viscosity is
where ρ is the density of the fluid measured in kg per cubic meter. Thus, in the SI unit system, the dynamic viscos-
ity has the units of kg/(m s) and the kinematic viscosity has the units of m 2/s. For STP (or Standard Temperature
and Pressure at sea level), the kinematic viscosity of water has a value of about 1 × 10 −6 m 2/s. By comparison, the
kinematic viscosity of water in a PWR core has a value of about 2.8 × 10 −7 m 2/s at 15 MPa. Hence, these equations
can be used to find the internal friction of the coolant in an operating reactor, or for that matter, any enclosed space
where the pressure and velocity gradient are known perpendicular to the direction of flow. The shear forces that the
walls exert on the fluid also cause a resistance to the flow, and this resistance must be overcome by the action of a
reactor coolant pump. This requires a force to be exerted on the fluid that is expressed in terms of a pressure differ-
ence between the bottom and the top of the core. Sometimes this pressure difference can be thought of as a pressure
gradient, which is the pressure drop across the core ΔP, divided by the core height H, or ∆P/∆x = ( PIN – POUT ) H.
However, the pressure gradient in the core is not completely uniform, and because of the internal structure and dif-
ferences in coolant temperature, the frictional pressure losses as the coolant moves can vary as a function of radial
position and height. See Figure 15.5 to see how the pressure gradient behaves. The fluid momentum equations that
describe this flow are based on the Navier–Stokes equations. In Sections 15.18 and 15.19, we would like to formally
derive these equations and show how they are used. However, before doing so, we would like to say a few words
about how these equations were initially discovered.
FIGURE 15.5 The coolant pressure p and the pressure gradient dp/dz as a function of axial elevation z. Notice that the two of them
are not the same thing, although on average, the average pressure gradient is equal to the total core pressure drop divided by the
total core height.
FIGURE 15.6 Frenchman Claude-Louis Navier (left) and British engineer George Gabriel Stokes (right), who developed the fric-
tional terms that now appear in the Navier–Stokes equations. (These are public domain pictures provided by Wikipedia.)
15.11 The Fluid Conservation Equations in Different Coordinate Systems 563
FIGURE 15.7 A unit cell with the pressure at the center and the velocities at the edges.
TABLE 15.2
The Seven Unsolved Problems of Modern Mathematics as Defined by the Clay
Mathematics Institute in Cambridge, Massachusetts
Institute in Cambridge, Massachusetts, has offered a prize of 1 million dollars ($ 1,000,000 USD) to anyone who is able
to prove this assertion or present a practical counterexample. This problem has been referred to as one of the seven great-
est unsolved problems of modern mathematics, and it is sometimes called the Navier–Stokes existence and smoothness
problem. See Table 15.2. More information about this problem can be found at the following Website
http://en.wikipedia.org/wiki/Clay_Mathematics_Institute
which is maintained by the Clay Mathematics Institute. Fortunately, in the history of the nuclear power industry,
engineers have never encountered a situation that has been found to possess a singularity. However, inside of a black
hole such as the one shown in Figure 15.8, the Navier–Stokes equations could become singular, and a unique physical
solution to these equations might not even exist.
FIGURE 15.8 A black hole is one of the few objects in nature inside of which a guaranteed solution to the Navier–Stokes equa-
tions might not exist. (Picture provided by NASA—see nasa.gov.)
the z direction. However, the momentum equations interact with each other in very complicated ways because changes
in the momentum along one coordinate direction can affect the momentum in a direction orthogonal to the original one.
This effect is called cross-coupling, and it arises physically because some of the momentum is transferred by the viscos-
ity to other directions that are perpendicular to the direction of the primary flow. For example, as long as a reactor cool-
ant has some friction, any momentum in the z direction will have some effect on the momentum in the x and y directions.
The only time this effect can be ignored is when the fluid has no friction (μ = 0) or when the Reynolds number
becomes infinite (Re → ∞). Then, a moving slug of fluid behaves in the same way as a solid object, and its trajectory can
be found using Newton’s laws of motion. This cross-coupling of the momentum equations also causes the Navier–Stokes
equations to become nonlinear, and this is one of the primary reasons why they are so difficult to solve. Sometimes the
transfer of mass and momentum between adjacent control volumes is called advection, and thus advection requires a
fluid to be in motion. The Navier–Stokes equations do not apply to just fluid flow in reactor fuel assemblies. They can
be used to model ocean currents and global weather patterns (see Figure 15.9), and they can also be used to perform a
magnetohydrodynamic simulation of the solar wind.
FIGURE 15.9 The Navier–Stokes equations can be used to model ocean currents and global weather pattern (left), and they can be
used to perform a magnetohydrodynamic simulation of the solar wind such as the simulation shown on the right. In this simula-
tion, additional terms must be added to the Navier–Stokes equations to account for electromagnetic effects. (Pictures provided by
quantamagazine.org and Wikipedia.)
FIGURE 15.10 In turbulent flow, turbulent energy is transferred from one point in space to another by turbulent eddies. The move-
ment of these eddies is statistical in nature, and as the Reynolds number increases, the number of turbulent eddies increases as well.
Furthermore, new eddies may not be created in the same way as old eddies. In practice, one attempts to employ a
statistical approach to determine where all of the eddies go. Even if two eddies are created at exactly the same point
in space and time, they may eventually end up taking different paths through the fluid because their interaction with
the fluid molecules is both chaotic and statistical. This also implies that most practical solutions to the Navier–Stokes
equations do not concern themselves with where individual eddies go. Instead, they try to predict the average motion of
the turbulent eddies over time. This process requires the velocity field v = i ⋅ v x + j ⋅ v y + k ⋅ v z to have an instantaneous
velocity v that can be expressed as the sum of its average value v plus a temporally dependent term that contains the
instantaneous turbulent fluctuations v′. We may then write the velocity at any point in time as
v = v + v ′ (15.20)
566 The Conservation Equations of Fluid Mechanics
For laminar flows. the temporally fluctuating component v′ becomes zero. and the instantaneous velocity v then becomes
equal to the average velocity v . This has the effect of making the equations much easier to solve.
FIGURE 15.11 The Navier–Stokes equations in differential form. (Figure provided by NASA—see nasa.gov.)
TABLE 15.3
Types of Thermal-Hydraulic Correlations Used in Nuclear Reactor Analysis
under certain conditions. The “best” correlations in a particular category generally have a lower margin of error than
the “good” ones, and they also have a wider range of applicability. Table 15.3 illustrates what some of these correla-
tions can be. In general, a “good” correlation is also a function of the flow geometry, and geometrical effects (such as
the pitch-to-diameter [P/D] ratio of a reactor lattice) can determine the “best” correlation(s) to use for a particular
problem. Some of these correlations are presented in Appendix I.
dM/dt = m
IN − m
OUT (15.21)
where the mass flow rate per second across each surface is m = ρuA, A is the area of the surface, and M = ρV is the
mass of the fluid inside the boundaries of V. If we substitute the appropriate expressions for M, m IN, and m
OUT into
Equation 15.21, then the expression for the conservation of mass becomes
dρ/dt = {(ρu) IN
− ( ρu )OUT } ∆x (15.24)
And as Δx → 0, the differential form of the continuity equation becomes
It is easy to see that we can generalize this approach to more than one dimension by adding the contributions from the
control surfaces in the y and z directions. When we do so, we find that the continuity equation becomes
FIGURE 15.12 The inflow or outflow of mass through the faces of a differential control volume. The red dots in the Figure indi-
cate the center of each face.
where u is the velocity of the fluid in the x direction, v is the velocity of the fluid in the y direction, and w is the veloc-
ity of the fluid in the z direction. This is the continuity equation for a simple homogeneous fluid. Using the divergence
operator ∇ ⋅ ( ρv ) = ∂( ρu ) / ∂x + ∂( ρv ) / ∂ y + ∂( ρw ) / ∂z , we can also write the continuity equation as
∂ρ/ ∂ t + ∇ ⋅ ( ρV ) = 0 (15.27)
and sometimes this is the form of the continuity equation that mathematicians prefer. Notice that for an incompressible
fluid, the density is constant, and the continuity equation becomes
In other words, starting with just a simple rectangular control volume, we were able to derive the continuity equation for
the flow of an incompressible fluid in three dimensions. It should be obvious from what we have just said that a similar
approach can be applied to a cylindrical or spherical coordinate system by simply changing the shape of V. Then in
cylindrical coordinates (r, z, θ), it becomes
∂ρ/ ∂ t + (1/r 2 ) ∂ / ∂r(r 2ρV ) + (1/r sin θ) ∂ / ∂θ(sin θ ⋅ ρV ) + (1/r sin ϕ) ∂ρV / ∂ϕ = 0 (15.31)
15.15 Working with the Continuity Equation 569
Finally, if each of these coordinate systems is angularly symmetric so that the material derivative has no dependence on
either θ or φ, then the appropriate forms of the continuity equation become
( ) ( )
Spherical coordinates: ∂ρ/ ∂ t + 1 r 2 ∂ / ∂r r 2 ρV = 0 (15.34)
where it must be remembered in the final two instances that V is the velocity vector V = i u + j v + k w replaced by its cylin-
drical or spherical equivalents. Here, the density of the fluid ρ can be constant, or it can be pressure and temperature depen-
dent. When it is a constant (i.e., independent of the pressure and the temperature), these equations reduce even further to
( ) ( )
Spherical coordinates: 1/r 2 ∂ / ∂r r 2 V = 0 (15.37)
These are called the continuity equations for an incompressible fluid. Notice that when a fluid is incompressible, the
density does not appear explicitly in the continuity equation.
or
Equation 15.40 then represents the conservation of mass for a unidirectional flow. Examples of flow channels that this
equation would apply to would include a round pipe and a square duct. For the duct, A IN = ∆x∆y IN and A OUT = ∆x∆y OUT ,
and for the pipe, AIN = π R IN 2 and A OUT = π R OUT 2. Example Problem 15.1 illustrates how this particularly simple form
of the continuity equation can be used to predict the velocity change when there is an area change in the flow channel.
Depending upon whether or not heat is added to the channel, the inlet and outlet densities may be the same or they may
be different. Thus the value of the densities can also affect the velocities.
Solution At 30°C and 1 atm, the density of the water is 995.65 kg/m3. The mass flow rate through the pipe is then
m = ρvA = 199 kg/s. When the pipe contracts, the flow area is cut in half, but the mass flow rate remains the same.
For this to happen, the velocity of the water must double. Hence, the velocity of the water after the contraction point
is 2 m/s. [Ans.]
For other flow channels, the areas that appear in Equation 15.38 must be the areas perpendicular to the direction of flow.
Hence, Equation 15.38 shows that the coolant can accelerate or decelerate if there is an area change. In a reactor fuel
assembly with a fuel rod pitch of P and a fuel rod diameter of D, the flow areas are
To a use Equation 15.41, one simply substitutes the appropriate value for the area into the continuity equation.
E = mh (15.42)
where h was its specific enthalpy and m was its mass. The specific enthalpy is in turn defined as the internal energy per
unit mass u plus the flow work pυ:
Here, both the internal energy u and the flow work pυ are expressed on a per mass basis and υ is the specific volume of
the fluid measured in m3/kg. In the SI unit system, the specific enthalpy is measured in J/kg or kJ/kg. So in principle, it
is easy to see that the equation for the conservation of energy can be written as
If we substitute the appropriate values for E, EIN, and EOUT into this equation, then the expression for the conservation
of energy becomes
Here, E IN = (ρuhA) IN and E OUT = (ρuhA)OUT, and we have tacitly assumed that we are not adding to or subtracting any
additional energy from the fluid within the region of space defined by V. Equation 15.45 is just an integral form of the
fluid energy equation because it states that the rate of change of energy within an arbitrary region of space is equal to the
energy that goes into the region, minus the energy that goes out of the region. To derive the differential energy equation,
we again need to give the control volume a defined shape. Again, consider a small rectangle having the dimensions Δx,
Δy, and Δz. The volume contained within this rectangle is V = ΔxΔyΔz. The surface area that the fluid must flow across
in the x direction is A = ΔyΔz, and the thickness of the rectangle in the x direction is Δx. Hence, the exact expression
for the conservation of energy (see Figure 15.13) becomes
FIGURE 15.13 The inflow or outflow of energy through the faces of a differential control volume. The red dots in the Figure
indicate the center of each face.
In other words, this is what the energy equation looks like for a unidirectional flow. It is also called the fluid energy
equation. In deriving this equation, we assumed that no additional energy was added to or subtracted from the fluid in
the form of heat. However, if heat energy is added to the fluid, then the energy equation must be written as
where Q′ is the heat generation (or loss) rate per second. In terms of the material derivative, Equation 15.49 can then be
ritten as
w
It is easy to see that we can generalize this process to more than one spatial dimension by adding the contributions from
the control surfaces in the y and z directions. When we do so, we find that the differential form of the energy equation
becomes
Of course, we have neglected the effects of elevation and shaft work so far. If we include the effects of shaft work so
that Equation 15.51 can be applied to a rotating shaft (such as a reactor coolant pump or steam turbine), then the fluid
energy equation becomes
where W′ is the rate at which external work is added to or removed from the flow. If we use the gradient operator ∇V,
we can also write the energy equation as
where, we have not added any kinetic energy or potential energy to the fluid because the size of the control volume is
very small. So again, starting with just a simple rectangular control volume, we have been able to derive the energy
equation for a three-dimensional flow. It should be apparent from what we have just said that we can do the same thing
in cylindrical or spherical coordinates as well. We just have to change the shape of the control volume and recalculate
the surface areas. After doing so, we find that the appropriate form of the energy equation to use for a cylindrical coor-
dinate system is
∂(ρh)/ ∂ t + (1/r ) ∂(rρhV )/ ∂r + (1/r ) ∂ρhV / ∂θ + ∂ρhV / ∂z = Q ′ + W ′ (for cylindrical coordinates) (15.54)
( ) ( )
∂(ρh)/ ∂ t + 1/r 2 ∂ / ∂r r 2ρhV + (1/r sin θ ) ∂ / ∂θ(sin θ ⋅ ρhV ) + (1/r sin ϕ) ∂ρhV / ∂ϕ = Q ′ + W ′ (for spherical coordinates)
(15.55)
Finally, if all the regions of space within the sphere and the cylinder are angularly symmetric, then the appropriate forms
of the energy equation to use become
where V is the velocity vector defined in Section 15.22 for cylindrical or spherical coordinates. For Cartesian coordi-
nates, it is simply V = i ⋅ u + j ⋅ v + k ⋅ w. These equations can be applied to any reactor component as long as care is taken
to use the correct values for Q′ and W′. The values for Q′ and W′ are positive for any component that adds energy to
the flow and negative for any component that removes it. When the flow becomes isothermal, both Q′ and W′ are zero.
For a steady flow where the mass flow rate through the pipe does not change with time, d (ρhV)/dt = 0. Hence, the energy
equation for the pipe reduces to
If the pipe is square, A IN = ∆x∆y IN and A OUT = ∆x∆y OUT , and if the pipe is round, A IN = π R IN 2 and A OUT = π R OUT 2.
Example Problem 15.2 illustrates how the energy equation can be used to find the specific enthalpy when heat is applied
15.18 The Momentum Equations 573
to the walls of the pipe. When heat is added to the fluid from the walls, we can set, Q = q″APIPE = q′L, where L is the
heated length of the pipe, and APIPE is the surface area through which the heat is flowing.
where V is the velocity vector. In Cartesian coordinates, we can also write the momentum as
p = i ⋅ mv x + j ⋅ mv y + k ⋅ mv z (15.62)
where mvx, mvy, and mvz are the individual components of the momentum along the x, y, and z directions and i, j, and k
are the appropriate unit vectors. It then follows from the Pythagorean theorem that the total momentum of a solid object is
Momentumi =
∫ ρv dV (15.64)
v
i
where the index i refers to the directional component (i = x, y, or z). The rate of change of the momentum is then
∂ / ∂t
∫ ρv dV (15.65)
v
i
and we can equate this to the force that the fluid experiences by writing
Fi = ∂ / ∂ t
∫ ρv dV (15.66)
v
i
574 The Conservation Equations of Fluid Mechanics
However, the momentum can also change as it flows across the control surfaces. If we interpret the momentum of the
fluid per unit volume as
where V is the volume and V is the velocity, then the rate that it must flow across an arbitrary surface surrounding the
control volume with surface area dA is
∫ A (ρv ) v dA (15.68)
i
We can convert this surface integral into a volume integral using Green’s theorem. When we do so, we obtain
Equating these two expressions to the applied force on the control volume, we find that
where Fi is the applied force along coordinate direction i. Notice that either i or j can be equal to x, y, or z. In other words,
Equation 15.70 represents three separate momentum equations in integral form. In terms of just the volume integrals,
a separate equation can be written for each coordinate direction. Thus, the momentum equation for the x direction is
∂ / ∂t
∫ ρv dV + ∫ ∇ (ρv v ) dV + ∫ ∇ (ρv v ) dV + ∫ ∇ (ρv v ) dV = F (15.71)
v
x
v
x x
v
x y
v
x z x
∂ / ∂t
∫ ρv
v
y dV +
∫ ∇ (ρv v ) dV + ∫ ∇ (ρv v ) dV + ∫ ∇ (ρv v ) dV = F (15.72)
v
y x
v
y y
v
y z y
∂ / ∂t
∫ ρv dV + ∫ ∇ (ρv v ) dV + ∫ ∇ (ρv v ) dV + ∫ ∇ (ρv v ) dV = F (15.73)
v
z
v
z x
v
z y
v
z z z
x direction: ∂ / ∂ t
∫ ρu dV + ∫ ∇(ρuu) dV + ∫ ∇(ρuv) dV + ∫ ∇(ρuw) dV = F (15.74)
v v v v
x
y direction: ∂ / ∂ t
∫ ρv dV + ∫ ∇(ρvu) dV + ∫ ∇(ρvv) dV + ∫ ∇(ρvw) dV = F (15.75)
v v v v
y
z direction: ∂ / ∂ t
∫ ρw dV + ∫ ∇(ρwu) dV + ∫ ∇(ρwv) dV + ∫ ∇(ρww) dV = F (15.76)
v v v v
z
Notice that there are nine advective derivatives in the momentum equations and that the components of the momentum
that are advected along the x, y, and z directions are
FIGURE 15.14 The inflow or outflow of momentum through the faces of a differential control volume. The red dots in the Figure
indicate the center of each face.
These terms are needed to conserve momentum because frictional forces can cause the momentum in one direction to
affect the momentum in other directions. These equations are then “integral equations” because they are based on
volume integrals. Next, we would like to derive the differential forms of these same equations. However, we also need to
replace Fi by the sum of at least two external forces—friction and gravity—to represent the behavior of the coolant in a
real reactor. When we do so, we obtain a separate differential equation for each coordinate direction. If we assume that
the friction along a specific coordinate direction i can be written as
fi = ∂τ ii ∂x + ∂τ ij ∂ y + ∂τ ik ∂z (15.80)
and the shear stress along a given coordinate direction i is proportional to the velocity gradient along that direction,
that is,
τ = µ ∇V (15.81)
then the frictional forces become easier to calculate. The momentum equations then reduce to the conventional forms
of the Navier–Stokes equations that are presented in many fluid mechanics books. The terms in these equations are
illustrated in Figure 15.14.
and apply it to the region of space shown in Figure 15.14. The force on the left-hand side is simply the pressure on each
face multiplied by the area of the face
F = d ( mv ) /dt = ( px – px + ∆x ) A (15.83)
576 The Conservation Equations of Fluid Mechanics
The force term on the left-hand side is simply the mass m times the time derivative of the velocity on each face
d ( mv ) /dt = m ( u x + ∆x − u x ) ∆t (15.84)
( px – px + ∆x ) A = m ( u x + ∆x − u x ) ∆t . (15.85)
Now, we know from our previous discussion of the continuity equation that the mass inside the control volume is
M = ρAΔx. If we substitute this into the proceeding equations, we get
( px – px + ∆x ) A = ρA ( u x + ∆x − u x ) ∆x/∆t. (15.86)
But Δx/Δt is equal to v, so ( px – px + ∆x ) A = ρA ( u x + ∆x − u x ) u. If we divide both sides by Δx, and cancel the areas, we
obtain
( px – px + ∆x ) ∆x = ρ( u x + ∆x − u x ) ∆x ⋅ u (15.87)
or
− ( px + ∆x – px ) ∆x = ρ( u x + ∆x − u x ) ∆x ⋅ u (15.88)
And finally, since ∆P = Px + ∆x – Px and ∆u = u x + ∆x − u x, Newton’s second law inexorably leads us to the following relation-
ship between the pressure gradient and the velocity gradient
This is the law of momentum conservation for a one-dimensional inviscid fluid. In differential form, it can be written as
If we were to repeat exactly the same exercise with a different force term F on each face, then there would also be a force
acting on the fluid within the control volume, and we could represent this force by
F= ∑f
i
i (15.91)
where F could be the force from many things (gravity, friction, electromagnetic effects, etc.). So a slightly more general
form of the one-dimensional momentum equation would be
where F could be any number of forces. In practice, the two most important forces are the internal friction, ffriction, and
the gravitational force, fgravity. So as long as the fluid does not contain any charged particles,
ffriction = − µd 2 u dx 2 (15.94)
where µ is the dynamic viscosity, and the gravitational force is simply fgravity = −ρg, so the total force acting on the fluid is
( )
F = ffriction + fgravity = − µd 2 u dx 2 + ρg (15.95)
15.20 Understanding the Advection of Momentum 577
The final form of the momentum equation along a single coordinate direction then becomes
This is the standard form of the momentum equation that appears in most fluid mechanics books. In more than one
dimension, it is easy to see that it generalizes to
( ( ) ( )
x direction: − ( dp/dx ) = ρ v x dv x dx + v y dv x dy + v z dv x dz ( ))
(
− µ d 2 v x dx 2 + d 2 v x dy 2 + d 2 v x dz 2 ) (15.97)
( ( ) ( )
y direction: − ( dp/dy ) = ρ v x dv y dx + v y dv y dy + v z dv y dz ( ))
(
− µ d 2 v y dx 2 + d 2 v y dy 2 + d 2 v y dz 2 ) (15.98)
( ( ) ( )
z direction: − ( dp/dz ) = ρ v x dv z dx + v y dv z dy + v z dv z dz ( ))
(
− µ d 2 v z dx 2 + d 2 v z dy 2 + d 2 v z dz 2 − ρg ) (15.99)
where z is assumed to be the vertical direction. Again making the substitutions u = vx, v = vy, and w = vz, we arrive at
the more familiar forms:
(
x direction: − ( dp/dx ) = ρ u ( du/dx ) + v ( du/dy ) + w ( du/dz ) )
(
− µ d 2 u dx 2 + d 2 u dy 2 + d 2 u dz 2 ) (15.100)
(
y direction: − ( dp/dy ) = ρ u ( dv/dx ) + v ( dv/dy ) + w ( dv/dz ) )
(
− µ d 2 v dx 2 + d 2 v dy 2 + d 2 v dz 2 ) (15.101)
(
z direction: − ( dp/dz ) = ρ u ( dw/dx ) + v ( dw/dy ) + w ( dw/dz ) )
(
− µ d 2 w dx 2 + d 2 w dy 2 + d 2 w dz 2 ) − ρg (15.102)
they contain
( )
x advection: ρ u ( du/dx ) + v ( du/dy ) + w ( du/dz ) (15.106)
578 The Conservation Equations of Fluid Mechanics
( )
y advection: ρ u ( dv/dx ) + v ( dv/dy ) + w ( dv/dz ) (15.107)
( )
z advection: ρ u ( dw/dx ) + v ( dw/dy ) + w ( dw/dz ) (15.108)
So one may immediately wonder why the results of these two derivations are different. Fortunately, these two deriva-
tions are equivalent because the total derivatives of the advective terms in Equations 15.103–15.105 can be written as
(
∂(ρuu)/ ∂x + ∂(ρuv)/ ∂ y + ∂(ρuw)/ ∂z = ρ u ( du/dx ) + v ( du/dy ) + w ( du/dz ) )
+ ρu ( du/dx ) + ( dv/dy ) + ( dw/dz ) (15.109)
(
∂(ρuv)/ ∂x + ∂(ρvv)/ ∂ y + ∂(ρvw)/ ∂z = ρ u ( dv/dx ) + v ( dv/dy ) + w ( dv/dz ) )
+ ρv ( du/dx ) + ( dv/dy ) + ( dw/dz ) (15.110)
(
∂(ρuw)/ ∂x + ∂(ρvw)/ ∂ y + ∂(ρww)/ ∂z = ρ u ( dw/dx ) + v ( dw/dy ) + w ( dw/dz ) )
+ ρw ( du/dx ) + ( dv/dy ) + ( dw/dz ) (15.111)
Collecting the derivatives into two logical groupings, we see that the second set of advective terms on the right-hand side
are zero because the continuity equation for an incompressible fluid is
In other words, both derivations lead to the same momentum equations, except that one set conserves momentum in
differential form and the other set conserves momentum in integral form. It must be pointed out that this equivalence
depends on the fluid being incompressible. If the fluid happened to be compressible, these formulations would NOT be
equivalent, and then the advective terms in Equations 15.102–15.104 should be used instead. These are then the momen-
tum equations that most reactor thermal-hydraulic analysis codes use.
( ( ) ( )
− ( dp/dx ) = ρdv x dt + ρ v x dv x dx + v y dv x dy + v z dv x dz ( ))
(
− µ d 2 v x dx 2 + d 2 v x dy 2 + d 2 v x dz 2 ) (15.113)
( ( ) ( )
− ( dp/dy ) = ρdv y /dt + ρ v x dv y dx + v y dv y dy + v z dv y dz( ))
(
− µ d 2 v y dx 2 + d 2 v y dy 2 + d 2 v y dz 2 ) (15.114)
( ( ) ( )
− ( dp/dz ) = ρdv z dt + ρ v x dv z dx + v y dv z dy + v z dv z dz ( ))
( )
− µ d 2 v z dx 2 + d 2 v z dy 2 + d 2 v z dz 2 − ρg (15.115)
15.21 The Time-Dependent Momentum Equations 579
These equations are applicable to any problem where the x, y, and z axes are orthogonal to each another. We can also
write them in operator form as
where p, V, and F are now vectors. The frictional forces are represented by the term µ∇2 V, and external forces (such as
gravity) are represented by the term F. We can simplify this notation even further by writing
where the quantity DV/Dt = ∂ V / ∂ t + V ⋅ ∇V is our old friend the material derivative. Hence, we can write the material
derivative of the velocity field as
DV /Dt = ∂ V / ∂ t + V ⋅ ∇V (15.118)
( ) ( ) ( ) ( )
V x, y, z, t = u x, y, z, t ⋅ i + v x, y, z, t ⋅ j + w x, y, z, t ⋅ k (15.119)
when the coordinates are Cartesian. In Equation 15.116, ∇ is the gradient operator, and for a vector quantity V, its formal
definition is
grad V = ∇V (15.120)
Similarly, in Equation 15.115, ∇2 is our old friend, the Laplacian operator, which is defined by
Once the momentum equations are written in this way, it is a relatively simple matter to extend the momentum equations
to other geometries (cylindrical or spherical). We simply find the appropriate expressions for the Laplacian and gradient
operators, and then substitute them into Equation 15.116. The appropriate forms of the momentum equations then follow
from this simple procedure. The exact expressions for the Laplacian operator and the gradient operator are shown in
Table 15.4 for an arbitrary function F (not to be confused with the force vector). Depending on the coordinate system
one chooses to use, F can be a function of x, y, z, or r, θ, z, or r, θ, φ: that is, F(x, y, z) or F(r, θ, z) or F(r, θ, φ). Now that
we have presented the mass, energy, and momentum equations, we next need to solve them to understand how a fluid
behaves. In general, this is not a trivial task. The main difficulty in obtaining a solution is that the advective terms are
nonlinear, and because of this, an exact solution may be difficult to obtain. Fortunately. there are a number of “tricks”
that can be used to make a solution possible. However, under certain conditions, it is still possible to obtain a solution.
TABLE 15.4
Expressions for the Gradient and the Laplacian Operators in Various Coordinate Systems
p ( x ) = ρ Au ( x ) + Bu ( x ) + C (15.123)
2
where C is a constant of integration and u is the velocity of the fluid in the axial direction. (The values of A and B are
determined by the velocity profiles.) This means that when the flow is incompressible and ρ is constant,
To obtain the correct value for C, we must measure (or otherwise obtain) the value of p somewhere in the flow field. In
other words, the pressure at the entrance to the fuel assembly can be used to find the value of C. In practice, this inlet
pressure, which is in turn set by the system pressure, is called a pressure boundary condition. The value of the fluid
velocity u(x) at the entrance to the fuel assembly then becomes the velocity boundary condition. In three dimensions,
implementing this procedure is not as simple as it first appears because an incompressible fluid with constant proper-
ties requires us to specify the three components of the velocity field (u, v, and w) and the pressure p to find the three-
dimensional pressure gradients. In other words, four equations (three momentum equations and the continuity equation)
are required to find four unknowns (u, v, w, and p). This approach can only be used when the density is constant and
temperature changes are small. When the temperature changes become large, the continuity and momentum equations
become more tightly coupled and the only way to represent this coupling is numerical. Then, the equation of state ρ(p,
h) must also be included in the solution set. So when large changes to the fluid temperature occur, we now have six equa-
tions to solve for six unknowns, and these equations and unknowns are
and
The temperature is usually deduced from the pressure and enthalpy: T = f(p, h) using a lookup table. If the system
pressure po is known at the entrance to the core, then the pressure gradients can be used to find the local value of p(x).
The values of p(x) and h(x) can then be used to find the value of ρ(x). In two and three dimensions, this process quickly
becomes time-consuming and exact solutions to the Navier–Stokes equations are difficult to obtain.
15.23 The Navier–Stokes Equations in Cylindrical Coordinates 581
where er, eθ, and ez are the corresponding unit vectors along the r, θ, and z directions. Notice how the unit vectors are
defined in this case (refer to Figure 15.16). If the flow is incompressible, the continuity equation becomes
( ( ) ( ) )
ρ ∂ u r ∂ t + u r ∂ u r ∂r + u θ r ∂ u r ∂θ − u θ 2 r + u z ∂ u r ∂z = − ∂p/ ∂r + ρgr
(15.128)
( ((
+ µ (1/r ) ∂ r ∂ u r ∂r )) ( ) ( )
∂ r – u r r 2 + 1/r 2 ∂ 2 u r ∂θ2 – 2/r 2 ∂ u θ ∂θ + ∂ 2 u r ∂z 2 )
FIGURE 15.15 In cylindrical coordinate systems, control volumes have different shapes than they do in a Cartesian one.
582 The Conservation Equations of Fluid Mechanics
FIGURE 15.16 Unit vectors er and e θ in cylindrical coordinates are coupled together. Movement in the θ d irection causes er to
change direction, and this leads to extra terms in the Navier–Stokes equations and d ifferent values for v r and vθ.
( ( ) ( ) )
ρ ∂ u θ ∂ t + u r ∂ u θ ∂r + u θ r ∂ u θ ∂θ − u r u θ r + u z ∂ u θ ∂z = − (1/r ) ∂p/ ∂θ + ρgθ
(15.129)
( ) ( ) ( )
+ µ (1/r ) ∂ r ( ∂ u θ / ∂r ) ∂r – u θ r 2 + 1/r 2 ∂ 2 u θ ∂θ2 – 2/r 2 ∂ u r ∂θ + ∂ 2 u θ ∂z 2
( ( ) )
ρ ∂ u z ∂ t + u r ∂ u z ∂r + u θ r ∂ u z ∂θ + u z ∂ u z ∂z = − ∂p/ ∂z + ρgz
(15.130)
(( )) ( )
+ µ (1/r ) ∂ r ∂ u z ∂r / ∂r + 1/r 2 ∂ 2 u z ∂θ2 + ∂ 2 u z ∂z 2
Notice that the value of the density does not appear explicitly in the continuity equation because the fluid is incom-
pressible. The extra terms in the radial and angular momentum equations appear because of the properties of the
cylindrical coordinate system. Namely, as we move along the angular direction, the unit vector eθ also changes direc-
tion. Thus, the r and θ components of the momentum equations are numerically coupled. This coupling is not present
in Cartesian coordinates (even when we rotate the axes), and so there are no extra terms that are needed to describe
this coupling. The components of the viscous shear stress tensor transform in a similar way. Hence if we define the
stress tensor as
τ rr τ r θ τ rz (15.131a)
τ ij = τ θ r τ θθ τ rz (15.131b)
τ zr τ z θ τ zz (15.131c)
15.25 Treatment of Fluid Friction in a Compressible Fluid 583
where
( ) ( )
τ rr = 2µ ∂ u r / ∂r; τ rθ = µ r ∂ u θ r / ∂r + (1/r ) ∂ u r ∂θ ; τ rz = ∂ u r ∂z + ∂ u z ∂r (15.132)
( )
τ θr = µ r ∂ u θ r / ∂r + (1/r ) ∂ u r ∂θ ; τ θθ = µ (1/r ) ∂ u θ ∂θ + u r r ; τ θz = µ (1/r ) ∂ u z ∂θ + ∂ u θ ∂z ; (15.133)
( )
τ zr = ∂ u r ∂z + ∂ u z ∂r ; τ zθ = µ (1/r ) ∂ u z ∂θ + ∂ u θ ∂z ; τzz = 2µ ∂ u z ∂z (15.134)
then the analogy is complete. Again, the viscous stress tensor is symmetric so that τrz = τzr and τθz = τzθ. Hence, in
cylindrical coordinates, there are six independent components of the stress tensor, and in practice, their values must be
found experimentally.
τ xx τ xy τ xz
τ yx τ yy τ yz (15.138)
τ zx τ zy τ zz
Here, each of the shear stresses τij is a number that must be determined by performing an experiment. The values of the
shear stresses are different for laminar flow than they are for turbulent flow, and for a Newtonian fluid where the shear
stress is proportional to the velocity gradient, the values of the coefficients in the stress tensor are
In general, the off-diagonal terms are symmetric about the diagonal, and because a Newtonian fluid is isotropic, τxy = τyx,
τzy = τyz, and so on. In long-hand notation, the nine viscous shear stresses are
Only simple derivatives of the velocity field are required to find these stresses.
( )
y component: ρ ∂v/ ∂ t + u ( ∂v/ ∂x ) + v ( ∂v/ ∂ y ) + w ( dv/dz )
= ρg y − ∂τ yx ∂x + ∂τ yy ∂ y + ∂τ yz ∂z (15.146)
(
z component: ρ ∂w/ ∂ t + u ( ∂w/ ∂x ) + v ( ∂w/ ∂ y ) + w ( dw/dz ) )
= ρgz − ∂τ zx ∂x + ∂τ zy ∂ y + ∂τ zz ∂z (15.147)
These momentum equations are called Cauchy’s equations in honor of the French engineer and mathematician Augustine
Louis de Cauchy, who first proposed them in the early 1800s. They can be used for compressible and incompressible
flows because they are not based on any specific assumptions regarding the compressibility. Here, D/Dt is the familiar
material derivative, which in Cartesian coordinates becomes
D/Dt = ∂ / ∂ t + u ∂ / ∂x + v ∂ / ∂ y + w ∂ / ∂z (15.148)
Notice that the derivative of the pressure does not appear directly in Cauchy’s equations, but it is imbedded in the
shear stresses. Thus, Cauchy’s equation can be thought of as a general differential equation for the conservation of
linear momentum for a homogeneous fluid. The shear stresses τij that appear in Cauchy’s equation consist of a static
pressure component Pii, which is due to the local hydrostatic pressure when the fluid is at rest and an additional dynamic
component σii that becomes relevant when the fluid is in motion. In tensor notation, these terms are
τ ij = σ ij + pii (15.149)
or to be more explicit
τ xx τ xy τ xz = p 0 0 + σ xx σ xy σ xz (15.150)
τ yx τ yy τ yz = 0 p 0 + σ xx σ xy σ xz (15.151)
τ zx τ zy τ zz = 0 0 p + σ zx σ zy σ zz (15.152)
15.27 Simplifying Cauchy’s Equations to Handle Incompressible Newtonian Fluids 585
In their original form, Cauchy’s equations could not be solved analytically, and in even a single coolant channel, a
numerical solution could take hours to obtain. The primary drawback to Cauchy’s equations is that the nine components
of the fluid stress tensor τij (which represent the frictional forces acting on the fluid in the x, y, and z directions) are
independent variables in addition to the velocity and the pressure. There are nine components to the stress tensor in three
dimensions, but only six of them are unique because the stress tensor is symmetric. Thus, in addition to the pressure and
the three values for the velocity, there are six additional unknowns for a total of 10. This assumes that the value of ρ can
be found from an ancillary equation of state
Of course, to obtain a unique solution, we need to come up with six more equations so that the number of equations
is equal to the number of unknowns. In practice, this problem can be avoided by expressing the nine components of
the stress vector in terms of another independent variable such as the velocity V. This can be done for incompressible
Newtonian fluids, and after doing so, there are now as many equations as there are unknowns. Normally, the flow of
coolant in the primary loop of a PWR can be modeled in this way. Now, let us turn our attention to how the stress
tensor [τ] can be simplified.
Fx = µ ∂ 2 u ∂x 2 + ∂ 2 u ∂ y 2 + ∂ 2 u ∂z 2 = µ∇ 2 u (15.154)
Fy = µ ∂ 2 v ∂x 2 + ∂ 2 v ∂ y 2 + ∂ 2 v ∂z 2 = µ∇ 2 v (15.155)
Fz = µ ∂ 2 w ∂x 2 + ∂ 2 w ∂ y 2 + ∂ 2 w ∂z 2 = µ∇ 2 w (15.156)
because there is a direct relationship between the stress and the strain. In vector notation, the net viscous force per unit
volume becomes
Ffriction = µ∇ 2 V (15.157)
( )
ρ Du/Dt = − ( dp/dx ) + µ d 2 u dx 2 + d 2 u dy 2 + d 2 u dz 2 (15.158)
( )
ρ Dv/Dt = − ( dp/dx ) + µ d 2 v dx 2 + d 2 v dy 2 + d 2 v dz 2 (15.159)
( )
ρ Dw/Dt = − ( dp/dx ) + µ d 2 w dx 2 + d 2 w dy 2 + d 2 w dz 2 + ρg (15.160)
∇ 2 = ∂ 2 ∂x 2 + ∂ 2 ∂ y 2 + ∂ 2 ∂z 2 (15.161)
586 The Conservation Equations of Fluid Mechanics
we can combine these components of the momentum together into a single vector equation in which the friction is a
function of the local velocity V = iu + jv + kw:
This equation is called the Navier–Stokes momentum equation for an incompressible fluid.
The frictional forces given by F = μ∇2V can then be replaced by the second derivatives of the velocity V along each
coordinate direction:
Normally, the buoyancy force ρg (which is the result of gravity alone) is assumed to act in only the z direction. The final
forms of the momentum equations with a reasonable representation of fluid friction are then
These equations are appropriate for any isothermal fluid where the temperature, the density, and the viscosity are con-
stant (or nearly so). For most applications, these equations are the cornerstone of modern fluid mechanics. In many text-
books, compressible fluids are defined as fluids where the fluid experiences a significant variation in its density—either
spatially or temporally—as the result of a pressure change. If the density changes by only a few percent, then a fluid is
generally incompressible, and if we can treat it as a simple constant (ρ = constant), then this simplification and the fact
that ∇V = 0 lead to the simpler theory of incompressible flow. Ordinarily, the effects of compressibility only become
important when the density variations are large enough to create a shock wave. This can sometimes occur in the reactor
pressure vessel during a loss-of-coolant accident (or LOCA). However, with this rare exception, we can treat the flow as
being incompressible as long as the Mach number is below 0.3 and no phase change occurs.
becomes multidimensional. This is because in three-dimensional space, the fluidic friction terms involve a large number
of spatial derivatives. One way to avoid this problem is to ignore the term μ∇2V entirely (i.e., by setting μ = 0), and when
we do so, we can combine the continuity and momentum equations together create what is called Bernoulli’s equation
(Bernoulli’s equation is discussed extensively in Chapter 16). Unfortunately, Bernoulli’s equation is only applicable to
inviscid and semi-inviscid flows, and most reactor coolants do not behave in this way. When the effects of friction can
no longer be ignored, it is better to use an alternative approach to estimate the pressure change. In this approach, one
tries to solve the momentum equations in a simple reactor geometry (such as a circular pipe) when the velocity profile
v(r) is fully developed. When we do so, the Laplacian operator becomes
( )
∇ 2 V = 1/r d/dr r dv/dr (15.169)
where r refers to the radial position of the fluid in the pipe (r = 0 being the centerline). Under these conditions, we find
that the solution to the one-dimensional momentum equation for steady-state laminar flow becomes
dp/dz = − 1 2 fρ V /D (15.170)
2
where D is the diameter of the circular pipe and V is the average velocity of the coolant flowing through the pipe.
In three dimensions, this is
V =
∫ vV dV ∫ v dV (15.171)
or more explicitly
2 2 2
V = √ vx + vy + v z (15.172)
where
Vx =
∫ vu dV ∫ v dV (for the x direction) (15.173)
Vv =
∫ vv dV ∫ v dV (for the y direction) (15.174)
Vz =
∫ vw dV ∫ v dV (for the z direction) (15.175)
So at least for laminar pipe flow, we can replace the value of μ∇2V with 1 2 fρ V /D. The value of the dynamic viscos-
2
ity μ in this case is embedded in the equation for the friction factor:
f = f ( ρ, µ, V ) (15.176)
The friction factor can also be expressed as a function of the local Reynolds number:
f = f ( Re ) (15.177)
If this approach is extended to turbulent flows (with a different value for f), we can replace the frictional force term
μ∇2V with an expression like 1 2 fρ V . This simple substitution eliminates the need to deal with the Laplacian opera-
2
tor ∇ directly, and the second derivatives of the velocity never enter into the momentum equations. Not surprisingly,
2
most reactor thermal-hydraulic analysis programs favor this approach because of its simplicity and engineering appeal.
For a circular pipe with a constant cross-sectional area A = πD2/4, the pressure difference between two points p1 and p2
separated by a distance L is then
588 The Conservation Equations of Fluid Mechanics
( p2 – p1 ) /L = − 1 2 fρ V 2 /D (15.178)
or
∆pFRICTION = − 1 2 fρ V
2
( L/D ) (15.179)
where D is the diameter of the pipe. This is the most common form of the frictional pressure drop that is used in the
study of reactor coolant channels. Hence, the frictional pressure drop ∆pFRICTION = − 1 2 fρ V (L/D) that appears
2
in many reactor thermal hydraulics codes really comes from a very simple approximation to the Navier–Stokes
equations.
then gives rise to the pressure gradients that appear in the momentum equations. The value of the friction factor f
determines the magnitude of this drag force based on the average fluid velocity V in the flow stream. For simple
geometries like flat plates and circular pipes, the momentum equations can be solved without turbulence to find the
value for f. (We will demonstrate how this can be done in Chapter 17.) Then, the shear stresses disappear entirely from
the momentum equations, and all we are left with is Ffriction = − 1 2 fρ V D for the frictional force. Notice that the
2
sign of this frictional force is negative because friction acts in the opposite direction of the flow. This simple-minded
approach has a lot of virtues and very few vices because it enables one to find the flow field in a number of intercon-
nected coolant channels at the same time. It is also handy when it comes to solving the fluid conservation equations in
a single flow channel with an area change. Normally, this approach would appall a CFD developer, but when we have
to find the flow field in hundreds or even thousands of interconnected coolant channels, it is a much more practical
way to proceed.
when the frictional forces are represented by the term μ∇2V and other external forces such as gravity are represented by F.
Hence, F was generally assumed to be equal to ρg. These equations could then be used in any reactor coolant as long
as the velocity V of the coolant was low compared to the speed of sound c. Specifically, we learned that a single-phase
isothermal fluid can be treated as being incompressible as long as its Mach number is less than 0.3. The criterion for
incompressibility is then
where c is the speed of sound in the fluid itself. As Equation 15.181 is written, it can be used for both liquids and gases.
Next, consider a reactor coolant with a very low value of the viscosity so that the effects of fluid friction can be ignored.
A fluid with these characteristics is said to be an inviscid fluid. When μ = 0, the term μ∇2V = μ (d2u/dx2 + d2v/dy2 + d2w/
dz2) vanishes, and the Navier–Stokes momentum equations become
15.31 Boundary-Layer Approximations 589
Euler’s Equation
ρ∂ v / ∂ t + ρV·∇V = −∇p + ρg
or (15.182)
FIGURE 15.17 A picture of Swiss Mathematician and Physicist Leonhard Euler, who was the first man to derive the modern form
of the momentum equation for an inviscid flow. He is also credited with Euler’s formula that links the five fundamental constants in
Mathematics, namely, e, the base of the natural logarithms, i, the square root of –1, Pi, the ratio of the circumference of a circle to its
diameter, 1 and 0, together. The formula is commonly written as eix = sin x + i cos x. (Pictures provided by Wikipedia.)
These interesting looking equations were first derived by Leonhard Euler in 1755 (see Figure 15.17), and not surpris-
ingly, they are called Euler’s equations. Notice that Euler’s equations do not contain any terms to account for the effects
of wall friction or even fluidic friction on the velocity. The dominant terms in Euler’s equation are then represented by
the material derivative D/Dt = ∂/∂t + V∇. When the flow field approaches a steady-state or equilibrium value, the veloc-
ity becomes independent of time (∂V/∂t = 0), and Euler’s equation reduces to
−∇p = ρV ⋅ ∇V + ρg (15.183)
This form of the momentum equations is general enough to predict the behavior of the flow as long as the effects of fric-
tion can be ignored. Practically speaking, this is good enough to obtain at least a preliminary estimate of how the flow
field behaves. However, in reactors, friction is generally important enough that its effects cannot be ignored. Then, the
2
full momentum equations with either Ffriction = μ (d2u/dx2 + d2v/dy2 + d2w/dz2) or Ffriction = ½ f ρ V D) must be used to
obtain the velocity and pressure fields. This creates a specific dependence between the pressure and the velocity which
all of the other conservation equations must respect.
earlier, there are at least two situations where the viscous terms in the Navier–Stokes equations can be ignored.
The first occurs in regions of space where the Reynolds number is very high (Re > 100,000) and where the viscous
forces are known to be small compared to the inertial forces. The second occurs in regions of space where the flow
becomes irrotational and the vorticity is small (see Chapter 14). In both cases, the removal of the viscous terms from
the Navier–Stokes equations then leads to Euler’s equations. While this greatly simplifies the process of obtaining a
solution, it also becomes difficult to specify a no-slip boundary condition at the fuel rod surface. A practical way to
apply this boundary condition was first proposed by Ludwig Prandtl in 1904 when he introduced what has become
known as the boundary-layer approximation. Prandtl’s idea was to subdivide the flow field into two separate regions
which are sometimes called zones: an outer region that is inviscid or irrotational and an inner region called the
boundary layer where viscous effects are important. In the outer region, we can then use the continuity equation
and Euler’s equation to obtain the velocity field, and Bernoulli’s equation to obtain the pressure field. Within the
boundary later, the boundary-layer equations can then be solved using the pressures and velocities obtained from
the outer region. Since the boundary layer becomes progressively thinner as the Reynolds number gets higher, the
boundary-layer approximation becomes more reliable at high Reynolds numbers. In reactor work, this is exactly
what happens when the Reynolds number exceeds 100,000. Normally, it is assumed that the pressure gradient
through the boundary layer is small. This then leads one to conclude that dP/dy ≈ 0. Hence, the pressure can vary
along a fuel rod in the x (or z) direction, but perpendicular to the surface of the rod, it does not. Next, because the
boundary layer becomes very thin when the Reynolds number is high, streamlines within the boundary layer have
negligible curvature when observed at the scale of the boundary-layer thickness δ. The momentum equation inside
the boundary layer then becomes
The last term in this equation is not negligible in the boundary layer because the derivative of the velocity gradient is
large enough to offset the typically small value of the kinematic viscosity ν. Finally, we know from our previous dis-
cussion that the pressure P across the boundary layer is the same as that outside the boundary layer. Hence, we can use
Bernoulli’s equation
to describe the flow outside the boundary layer. Differentiating this equation with respect to x, we get
where both P and U are functions of x only. Substituting Equation 15.186 into Equation 15.184, we obtain
and we have eliminated the pressure from the boundary-layer equations. Hence, the continuity and momentum equa-
tions that determine the velocity field within the boundary layer for steady-state, incompressible laminar flow are
∂ u/ ∂x + ∂v/ ∂ y = 0 ( continuity )
(15.188)
u(∂ u/ ∂x) + v(∂ u/ ∂ y) = UdU/dx + ν ∂ 2 u/ ∂ y 2 ( momentum )
These equations are called the boundary-layer equations, and they can be used to deduce the friction factor for
both laminar and turbulent flows. Notice that the momentum equation in the boundary layer (the second equation
of Equation 15.188) is parabolic, and thus, we only have to specify boundary conditions on three sides of the two-
dimensional flow domain because information is not passed in the downstream (or −x) direction. Conversely, the full
Navier–Stokes equations are elliptic, and this requires us to specify boundary conditions over the entire flow domain
because information is passed both upstream and downstream. This also makes the Navier–Stokes equations more
difficult to solve.
15.32 Converting Euler’s Equation into Bernoulli’s Equation 591
1. laminar
2. incompressible
3. frictionless
then we can convert Euler’s equation into another important equation of fluid mechanics called Bernoulli’s equation.
Bernoulli’s equation is one of the most famous equations in all of fluid mechanics, and it is also one of the easiest to
use. Bernoulli’s equation can be derived from the conservation of energy, but it can also be derived directly from the
momentum equations. In Chapter 16, we will explore the implications of this famous equation in more detail. In general,
Bernoulli’s equation can be used for problems where the flow is laminar, frictionless, and isothermal. Many reactor pip-
ing systems can be described using Bernoulli’s equation before the flow becomes turbulent. The easiest way to derive
Bernoulli’s equation from Euler’s equation is to consider a laminar streamline during steady-state, incompressible flow.
An example of such a streamline is shown in Figure 15.18. A streamline is a curve that is everywhere tangent to the local
instantaneous velocity vector. Hence, when streamlines appear in the flow, a fluid particle cannot cross a streamline.
Suppose that the streamline is three dimensional and suppose that we select our coordinate system so that one coordi-
nate n is normal to the streamline, and the other coordinate s is parallel to the streamline. Applying Euler’s equation to
the streamline in the s direction allows us to write
{ }
∂ / ∂s ρV 2 2 + p + ρgz = 0 (15.190)
provided that the density is constant. This then implies that anywhere along the streamline
FIGURE 15.18 Examples of streamlines around a spherical object and an aircraft wing. In fluid mechanics, a streamline is defined
as a curve that is everywhere tangent to the local instantaneous velocity vector. Streamlines are normally associated with laminar
flow. (This picture was provided by NASA—see NASA.gov.)
592 The Conservation Equations of Fluid Mechanics
This latter expression is just another form of Bernoulli’s equation, which we will discuss in more detail in Chapter 16.
It also shows that the kinetic energy plus the “pressure energy” plus the potential energy of a fluid particle within a
streamline is constant everywhere within the streamline. In other words, in the absence of friction, it is an expression of
the conservation of energy for the fluid particles within the streamline.
( ) ( )
ρdu/dt + ρ u ( du/dx ) + v ( du/dy ) + w ( du/dz ) = − ( dp/dx ) + µ d 2 u dx 2 + d 2 u dy 2 + d 2 u dz 2 (15.192)
( ) ( )
ρdv/dt + ρ u ( dv/dx ) + v ( dv/dy ) + w ( dv/dz ) = − ( dp/dy ) + µ d 2 v dx 2 + d 2 v dy 2 + d 2 v dz 2 (15.193)
( )
ρdv z /dt + ρ u ( dw/dx ) + v ( dw/dy ) + w ( dw/dz ) − ( dp/dz )
( )
= −µ d 2 w dx 2 + d 2 w dy 2 + d 2 w dz 2 − ρg (15.194)
where u = vx, v = vy, and w = vz. Now, suppose that we would like to understand how turbulence affects the form of these
equations for a homogeneous, incompressible fluid. To demonstrate the effect this has, it is sometimes helpful to write
these equations in a slightly different form than we have done so far. This involves writing the instantaneous velocity
vector along each coordinate direction as the sum of its average value 〈 V〉, which represents a time average, and a fluc-
tuating component V′, which accounts for the effects of turbulence. Now let us discuss exactly how this can be done.
TABLE 15.5
A Comparison between the Navier–Stokes Equations and Other Famous Equations of Fluid Mechanics
v = v + v ′ (15.195b)
w = w + w′ (15.195c)
∫
V = (1/T) v(t) dt (15.196)
and the limits of integration are from t = 0 to t = T. In this case, the value of V represents the temporal average of the
velocity field at a particular point in space instead of the spatial average of the velocity field at a particular point in time
(which was used in our previous definition for V ). An example of how the velocity profile can fluctuate along each
of these coordinate directions is shown in Figure 15.19. Thus, in turbulent flows, the value of V at a particular point in
the flow tends to exhibit a considerable amount of temporal noise. If we substitute these values for the velocity into the
momentum equations and average them over time, we find that
( ) (
ρd/dt + ρ ( d/dx ) + V ( d/dy ) + W ( d/dz ) = − ( dp/dx ) + µ d 2 dx 2 + d 2 dy 2 + d 2 dz 2 )
−ρ ( u′u′ + u′v′ + u′w′ ) ← New terms (15.197)
(
ρd V /dt + ρ ( d V /dx ) + V ( d V /dy ) + W ( d V /dz ) )
(
= − ( dp/dy ) + µ d 2 V dx 2 + d 2 V dy 2 + d 2 V dz 2 )
−ρ ( v′u′ + v′v′ + v′w′ ) ← New terms (15.198)
FIGURE 15.19 Turbulence is a statistical process where the generation and movement of turbulent eddies is entirely random
although the number of eddies and their expected size follow a statistical probability distribution. At a specific point in space, the
velocity in a turbulent flow will exhibit a large amount of statistical noise about an average value of the fluid velocity V . This aver-
age value is needed to define the Reynolds number and other important indicators of fluid behavior.
594 The Conservation Equations of Fluid Mechanics
( )
ρd V /dt + ρ ( d W /dx ) + V ( d W /dy ) + W ( d W /dz ) − ( dp/dz )
(
= −µ d 2 W dx 2 + d 2 W dy 2 + d 2 W dz 2 )
−ρ ( w′u′ + w′v′ + w′w′ ) ← New terms (15.199)
where we have temporarily dropped the gravitational term to demonstrate the symmetry.
u ′u ′ u ′v ′ u ′w ′
{t} = −ρ v ′u ′ v ′v ′ v ′w ′ (15.200)
w ′u ′ w ′v ′ w ′w ′
The value of {τ} is independent of the viscosity μ, and it depends on only the amount of turbulence in the flow.
It represents the amount of angular momentum that is “diffused” through the fluid when it becomes turbulent. If the
flow becomes laminar, then this term disappears, and the Navier–Stokes equations revert back to their original form
(because the fluid no longer spins at a microscopic level). The Reynolds stresses must be measured experimentally.
In reactor work, the Reynolds stresses help to redistribute the thermal energy produced in the core. This process is
sometimes called turbulent mixing, and it is an important effect because it helps to increase the rate of mass and energy
transfer. Usually, the amount of turbulent mixing is defined by a turbulent mixing coefficient CMIX that is a function of
the Reynolds number and the P/D ratio of the fuel rods.
The more turbulent the flow becomes, the greater the amount of turbulent mixing will be.
there are four continuity and momentum equations and four unknowns. These unknowns are the three components of
the velocity vector u, v, and w, and the pressure p. Here, it is assumed that the density can be deduced from the pressure.
and these are then the scales a CFD program must use to capture the underlying physical processes. Note that separate
length and timescales are required in this case. Hence, while practical methods for modeling turbulence in nuclear reac-
tors involve averaging the spatial and temporal components of the velocity field, the second key objective of turbulence
modeling is to accurately predict the Reynolds stresses for a variety of flow geometries—especially in coolant channels
where grid spacers and wire wraps are present. These additional structures make it difficult or even impossible to under-
stand the microscopic behavior of the flow without making a number of computational compromises.
where vz is the average axial velocity in the pipe, R is its radius, and μ is the dynamic viscosity. Later, we will find that
this leads to another important equation of fluid mechanics called Poiseuille’s equation, which is used to calculate the
mass flow rate through a circular pipe when the pressure gradient dp/dz is known. A similar exercise can be performed
for the same pipe when the flow becomes turbulent. In more than one dimension, we can guess the pressures, and
solve for the velocities, or we can guess the velocities and solve for the pressures. If the velocities we find by doing this
do not satisfy the continuity equation, then we can adjust the velocities and the pressures until they do. This is done
using what are called pressure correction algorithms or velocity correction algorithms. If we use these algorithms
correctly, the solution eventually converges. A variation on this approach is to assume that the mass flow rate m = ρvA
is proportional to the magnitude of the pressure gradient Δp/Δz = (p1 − p2)/Δz between two Eulerian cells. Thus, if we
know the pressures in the cells, the spacing Δz between the cells, (and therefore the pressure gradient), we can calcu-
late the mass flow rates and the direction the fluid is flowing. In this case, the coolant always flows from a region of
high pressure to a region of low pressure, and the flow rate is proportional to the pressure difference divided by the
center-to-center spacing between the cells:
Adding a constant of proportionality C that depends on the geometry of the flow channel then makes the equivalence
complete:
where C(A, μ) can also depend on the dynamic viscosity. The final factor that enters into the calculation of the pressures
and the velocities is the temperature and the enthalpy of the coolant. To estimate the effects these can have, we also need
to solve the fluid energy equation which we discussed previously.
At least one of these boundary conditions must be used to obtain a unique solution. Furthermore, different boundary
conditions will lead in different physical solutions. We would now like to discuss these boundary conditions and show
how they can be mixed or matched. Our discussion will be initially focused on single-phase flow. The boundary condi-
tions to be used for two-phase flows will be discussed at a later time.
In reactor piping systems and fuel assemblies, the wall does not move and so the surfaces are assumed to be station-
ary. Then, the velocity of the wall and the velocity of the fluid at the surface of the wall must be zero (VWALL = 0). The
no-slip boundary condition is illustrated in Figure 15.20. In other words, this boundary condition requires the coolant
to “stick” to the wall.
This boundary condition is illustrated in Figure 15.21. Normally the fluids are assumed to be at a similar temperature.
FIGURE 15.20 In the cylinders of internal combustion engines (left), a thin film of oil sticks to the surface of each cylinder.
This oil is sheared between the piston and the cylinder as the cylinder moves. However, at the surface of the cylinder, its velocity
is zero. In reactor fuel assemblies (see the diagram on the right), an analogous situation occurs because the coolant sticks to the
surface of the fuel rods. The liquid film on the rod surface may be only a few molecules thick, but where it touches the fuel rods, its
velocity is zero.
598 The Conservation Equations of Fluid Mechanics
FIGURE 15.21 At the interface between two fluids, the velocity of the two fluids must be equal. In addition, the shear stress τ
parallel to the interface must be the same in both fluids.
In order for the shear stresses to be the same, the aforementioned equation implies that the velocity gradient for air must
be 50 times greater than the velocity gradient for water at the point where the two fluids meet (∂u/∂yAIR ≫ ∂u/∂yWATER).
This implies that the water drags the air along with it much more easily than the air drags the water. The free surface
boundary condition is then
The free surface boundary condition: pLIQUID = pGAS and τ LIQUID ≅ 0 (15.211)
FIGURE 15.22 Because of the closed architecture of BWR cores, pressure boundary conditions are normally used at the inlet
and the outlet of each fuel assembly. In PWR cores, flow boundary conditions are normally used because the cores are more open.
A reference pressure is then specified at the core inlet. In practice, these boundary conditions are also applied to the individual
channels within an assembly.
between the inlet and the outlet or the fluid will have no effective viscosity. Conversely, the greater the value of Δp,
the greater the mass flow rate through the core will be. In American-made PWRs, the Reynolds number in a typical
fuel assembly is about 500,000 and the coolant velocity is 5–6 m/s. This in turn leads to a pressure drop of 26–28 PSI
(178–198 kPa) between the top and the bottom of the core and a mass flux of 3,500–4,000 kg/m2s. Again, these numbers
can be thought of as industry averages, and the actual values may be higher or lower than those quoted here.
p ( z ) = pREFERENCE − ∆p ( z ) (15.212)
where the pressure drops are measured upward from the bottom of the core (where z = 0) to the top (where z = H). Here,
Δp(z) is the fluid pressure drop as a function of axial elevation. If we want to do this for each individual fuel assembly,
then we can write
p(n, z) = pREFERENCE − ∆p(n, z) (15.213)
where n is the index of the assembly. The internal bookkeeping required to do this may differ slightly from one fluid
mechanics program to the next. However, this is essentially the approach that most CFD programs use. These flow
boundary conditions are illustrated in Figure 15.22. Notice that the coolant in a BWR behaves much differently than it
does in a PWR because large amounts of bulk boiling are allowed.
600 The Conservation Equations of Fluid Mechanics
where u represents the fluid velocity in the x direction and v represents the fluid velocity in the y direction. These equa-
tions assume that the density of the coolant in the boundary layer is uniform and that the pressure gradient ∂p/∂y as
we proceed through the boundary layer in the vertical direction (which is perpendicular to the surface of the fuel rods)
is small. They also assume that the effects of gravity can be ignored because the boundary layer is so thin. In three
dimensions, the Navier–Stokes equations then become elliptic equations, and this implies that the pressure and velocity
boundary conditions must be applied over the entire boundary of the flow domain. In other words, information concern-
ing the flow can be passed both upstream and downstream.
However, in one dimension, the x momentum equation is parabolic and this implies that the pressures and velocities
must be specified on three sides of the boundary layer. It also implies that we only need to specify the boundary condi-
tions upstream and on the top and the bottom of the flow—not downstream where the boundary layer is still developing.
This is particularly important because one can simply “march downstream” in the x direction, solving the boundary-
layer equations as we go. The general procedure for doing this is shown in Figure 15.23. Note that the velocity at the top
of the boundary layer, where the fluid velocity approaches the free stream velocity, is sometimes given the symbol U∞.
Of course, at the surface of the wall, both u and v are zero because viscous fluids cannot “slip” over the surface of an
object due to the no-slip boundary condition. Moreover, for simple problems, it turns out that the boundary-layer equa-
tions can be solved analytically if we are willing to make a few simplifying assumptions. When we do so, we find that
we can predict both the thickness of the boundary layer and the amount of drag it exerts on the underlying surfaces. The
process for doing this is shown in Figure 15.24. For plate-like fuel rods, the Reynolds number is defined by
FIGURE 15.23 Boundary-layer solutions in nuclear reactors are similar to those in other applications. The pressure may change
along a boundary layer (in the x direction), but the change in pressure across a boundary layer (in the y direction) is almost negli-
gible. The velocity in the y direction (v) is much smaller than the velocity in the x direction (u). Finally, since the boundary-layer
equations are parabolic, boundary conditions need to be specified on only three sides of the flow domain. Boundary conditions are
normally not specified on the upstream edge.
15.49 Boundary-Layer Solutions 601
Re = ρvx/µ (15.216)
Hence, the Reynolds number is measured from the point where the boundary layer begins (at x = 0). The location where
x = 0 is then the leading edge of the rod. This means that Equations 15.209 and 15.210 are appropriate to use for both
laminar and turbulent flows as long as a surface is relatively flat and smooth. A similar set of equations can be derived
for cylindrical fuel rods. These equations were presented in Section 15.23.
δ = 5.48x √ Re x (15.217)
where x was the distance (measured in cm or m) from the leading edge of the surface and Rex was the local Reynolds
number. For these problems, the local Reynolds number was defined by
Re x = ρvx/µ (15.218)
and thus its value would grow as x became larger. To obtain a solution, he further assumed that the velocity profile was
parabolic in the boundary layer and that it had a value of v = 0 at the surface of the plate and a value of v = U∞ far away
from the plate where the flow was essentially undisturbed (see Figure 15.25). Hence, it worked well for laminar flows
where the boundary conditions could be “clearly defined.” Moreover, he showed that the average drag coefficient over
the entire length L of the solid surface was given by
CD = 1.155 √ Re L (15.219)
where ReL was another version of the Reynolds number called the global Reynolds number or the plate Reynolds num-
ber. In turn, the plate Reynolds number was given by
Re L = ρU ∞ L µ (15.220)
where L was the length of the plate (in cm or m). Later Heinrich Blasius, a German physicist and fluid dynamicist,
extended Prandtl’s work to other classes of problems and showed that the exact values for the boundary-layer thickness
and the average drag coefficient were
when the same definitions for the Reynolds number were used. In the process, he also defined a local drag coefficient
CD(x)
CD ( x ) = 0.664 √ Re x (15.223)
which was defined as the ratio of the wall shear stress τo to ½ρU∞2 along the surface of the plate
CD ( x ) = τ o ( x ) 1 2 ρU ∞ 2 (15.224)
Notice that this definition for the drag coefficient is different than the definition for the average drag coefficient CD:
CD = FD 1 2 ρAU ∞ 2 (15.225)
602 The Conservation Equations of Fluid Mechanics
because the local drag coefficient depends on the value for x and
A is the area of the surface or plate.
U∞2 is the free stream velocity away from the boundary layer where the flow field is undisturbed.
Sometimes the local drag coefficient CD(x) represented by Equation 15.224 is called the coefficient of skin friction to
emphasize this distinction. Eventually, their work was extended to turbulent flows and to turbulent boundary layers such as
those that formed over the surface of nuclear fuel rods. For these solutions, the surfaces were assumed to be horizontal and
smooth, and the value for x was assumed to be 0 at the leading edge. Then according to the previous equations, the thick-
ness δ(s) of the boundary layer was zero at x = 0. The boundary layer in circular pipe flow behaves in a similar way when
“entrance effects” are taken into account. We will have more to say about the origin of these effects in Chapters 16 and 17.
If Re = ρU∞x/μ < 500,000—the boundary layer is laminar (and the velocity profile is parabolic)
If Re = ρU∞x/μ > 500,000—the boundary layer is turbulent on that portion of the surface (and the velocity profile
follows a logarithmic or power law)
Classical pipe flow and reactor fuel assembly flow, which use the alternative definition Re = ρUDe/μ for the Reynolds
number have the following definitions when the flow is developing:
If Re = ρUDe/μ <2,300—the flow is laminar (and the velocity profile is parabolic).
If Re = ρUDe/μ > 2,300 and Re = ρUDe/μ < 10,000—the flow is transitional (and the velocity profile is only p artially
parabolic).
If Re = ρUDe/μ > 10,000—the flow is turbulent (and the velocity profile follows a logarithmic or power law).
Next, let us attempt to highlight some of these differences with some practical examples (see Figure 15.25).
( ) ( ) ( )
u i, j, k = u i, j, k + u ′ i, j, k (15.226)
( ) ( ) ( )
v i, j, k = v i, j, k + v ′ i, j, k (15.227)
( ) ( ) ( )
w i, j, k = w i, j, k + w′ i, j, k (15.228)
where i is the x index, j is the y index, and k is the z index for each control volume or computational cell. For most
fuel assemblies, values for i and j are on the order of 250, and values for k are on the order of 10 to 20. In addition, a
large core can contain over 200 fuel assemblies (if it is a PWR) and over 800 fuel assemblies (if it is a BWR). When
the flow becomes time dependent, a temporal index also needs to be added to the velocity vectors for each time
step. Normally this temporal index is represented by the symbol n. Hence for just the x direction, the velocity vector
becomes u(i, j, k, n) = 〈u(i, j, k, n)〉 + u′(i, j, k, n). For any meaningful transient, the value of n can easily exceed 1,000
(or even 10,000). The nodalization scheme required to solve a problem of this type is illustrated in our companion
book*. Similar nodalization schemes are routinely employed by reactor designers—except that they are applied to
the whole core—and just not to a single fuel assembly. To reduce this computational cost, one usually ignores the
FIGURE 15.24 A procedure that is used to solve the Navier–Stokes equations inside a boundary layer in the x-y plane for a steady,
incompressible flow.
604 The Conservation Equations of Fluid Mechanics
fluctuating components of the velocity field, and exact solutions become restricted to problems involving only steady
(i.e., non-time-dependent) flows. Over the years, many attempts have been made to solve the fluid mechanics equations
more efficiently. (The author attempted to do this in his PhD thesis). One of the most popular approaches has been to
simplify the equations that must be solved by ignoring the values for u′v′, u′w′, and v′w′ entirely. In some cases, the
effects of this turbulence can then be simulated using turbulent mixing coefficients. The U.S. NRC uses a specially
designed computer program called TRACE to perform this function, and the U.S. Department of Energy (DOE) uses
similar tools such as RELAP, RETRAN, COBRA, and VIPER. We will have more to say about the phenomenologi-
cal and numerical models which these programs use in Chapter 26.* In reactor fuel assemblies, they attempt to solve
slightly different forms of the continuity, energy, and momentum equations based on what is called a “lumped param-
eter approach”
FIGURE 15.25 A picture showing the development of the laminar and turbulent boundary layers over a flat horizontal surface.
Notice that the transition point between the laminar and turbulent layers occurs at a Reynolds number of about 500,000 because the
flow is an open flow. The boundary layer gets progressively thicker from the start of the plate (where x = 0 and δ = 0) to the end of
the plate (where x = L).
* See: An Introduction to Nuclear Reactor Physics by Robert E. Masterson, CRC Press (2017).
Questions for the Student 605
assembly. Eventually, the equations we have derived must be extended to two-phase mixtures where the phases may
have different temperatures and may even move in different directions. We will encounter these equations again when
we delve into the subject of boiling heat transfer in Chapter 24. Then the convective heat transfer coefficients become
flow regime dependent and very sophisticated correlations must be used to find the correct heat transfer.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York, NY (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York, NY (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York, NY (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York, NY (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York, NY (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York, NY (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York, NY (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York, NY (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, CRC Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc. New York,
NY (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi, India (2000).
Reactor Theory (Nuclear Parameters) DOE-HDBK-1019/2-93, Washington, DC (1993).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, CRC Press, Boca Raton, FL (2014).
Web References
http://www.mit.edu/course/1/1.061/www/dream/SEVEN/SEVENTHEORY.PDF.
https://en.wikipedia.org/wiki/Navier%E2%80%93Stokes_equations.
http://en.wikipedia.org/wiki/Clay_Mathematics_Institute.
https://en.wikipedia.org/wiki/Reynolds_decomposition.
https://en.wikipedia.org/wiki/Turbulence.
14. Which famous scientist and mathematician was responsible for developing the Lagrangian representation of fluid
mechanics?
15. Which famous scientist and mathematician was responsible for developing the Eulerian representation of fluid
mechanics?
16. Which of these two representations relies on the concept of a stationary control volume to conserve mass, energy,
and momentum?
17. Write down Bernoulli’s equation in one dimension.
18. What is the difference between Euler’s equation and Bernoulli’s equation?
19. Under what conditions does Euler’s equation reduce to Bernoulli’s equation?
20. What is the difference between a compressible flow and an isothermal flow?
21. What is the difference between a compressible flow and an incompressible flow?
22. When does a compressible flow become an incompressible one?
23. Which type of flow more accurately describes the flow of fluid through a reactor coolant channel?
24. At what temperature and pressure does a single-phase flow become a two-phase flow?
25. Which equation, which is important in reactor thermal-hydraulics, was not included in the original f ormulation of
the Navier–Stokes equations?
26. How many Navier–Stokes equations are there in one dimension?
27. What is the primary difference between Cauchy’s equation and Euler’s equation?
28. Not including the equation of state, how many Navier–Stokes equations are required in two and three dimensions?
29. Write down the Navier–Stokes equations for a compressible fluid in one, two, and three dimensions in a Cartesian
coordinate system.
30. Write down the same equations in spherical and cylindrical coordinate systems.
31. How is the Laplacian operator defined in each of these coordinate systems?
32. At what Mach number does a compressible flow become an incompressible one?
33. Which specific terms in the Navier–Stokes equations are nonlinear in nature?
34. In the Navier–Stokes equations, what terms are used to describe fluidic friction?
35. How many of these terms are required in two and three dimensions?
36. For a Newtonian fluid, what relationship is assumed to exist between the stress and the strain?
37. In Newtonian fluid mechanics, what is the coefficient of proportionality between these two important quantities?
38. Write down the Navier–Stokes equations for a turbulent, compressible flow in two dimensions.
39. Are the resulting equations parabolic or elliptic in nature?
40. Can the Navier–Stokes equations be used to describe the boundary layer next to the surface of a nuclear fuel rod?
41. For highly turbulent flow, what is the approximate thickness of this fluid boundary layer?
42. For two-dimensional problems, at how many different physical locations in this boundary layer must a boundary
condition be specified?
43. What is the purpose of the Reynolds decomposition?
44. When a compressible flow becomes turbulent, how can the Reynolds decomposition be used to simplify the Navier–
Stokes equations?
45. How are the Reynolds stresses evaluated in this case?
46. At what Reynolds number does the transition between laminar flow and turbulent flow normally occur?
47. What is the Navier–Stokes existence and smoothness problem?
48. Which famous mathematical institution has offered a million dollar reward to solve this problem?
49. Can the Navier–Stokes equations be used to describe the flow of matter through a “black hole?”
50. In reactor fluid mechanics, what is the difference between convection, diffusion, and advection? Do the Navier–
Stokes equations describe all three of these physical processes?
51. What is turbulent mixing, and why is it important in reactor thermal-hydraulic analysis?
52. Does turbulent mixing require any net exchange of mass between adjacent coolant channels?
53. What is the turbulent mixing coefficient, and how is it defined?
54. Name at least three computer programs that are used by nuclear scientists and engineers to find the flow fields in
reactor coolant channels. Can any of these computer programs be used to model isothermal flows?
55. Name a famous CFD program that is frequently used in industrial applications but not nuclear ones.
56. Why are conventional CFD programs not used more often in reactor work?
57. Name a famous fluid dynamics program that the U.S. NRC uses to perform reactor thermal-hydraulic analysis.
58. Why is a lumped parameter approach preferable to “brute force approach” when solving the fluid mechanics equa-
tions in commercial nuclear reactors?
59. Name four types of boundary conditions that are commonly applied to the Navier–Stokes equations.
Exercises for the Student 607
6 0. Which one of these boundary conditions is most applicable to BWR fuel assemblies?
61. Which one of these boundary conditions is most appropriate to use for PWR cores?
62. Which fluid boundary condition normally requires the use of a “reference pressure” at the inlet to a reactor core?
63. Which one of these boundary conditions is used at the surface of a nuclear fuel rod? What constraint does this
boundary condition place on the fluid velocity at the surface of the rod?
64. In reactor fluid mechanics, what is this constraint called?
65. What is the relationship between the dynamic (or absolute) viscosity and the kinematic viscosity? In most reactor
flow problems, which one is larger?
66. Can a symmetry boundary condition be used to describe the flow of fluid through a reactor coolant channel?
67. Give an example of a free surface boundary condition, and show a boundary condition of this type can be
implemented.
68. For what types of flow(s) can approximate solutions to the Navier–Stokes equations be obtained?
69. What is the Kolmogorov scale, and why is it important in reactor fluid mechanics?
70. What time step size must be used to capture the physics of turbulent flow when the Kolmogorov scale is used?
71. What is the explicit form of the Reynolds stresses that are associated with the Reynolds decomposition?
72. Why is it important to use engineering judgment when applying CFD programs to nuclear power plants?
73. How many Navier–Stokes equations are required to describe the behavior of two-phase flows?
74. What are the three specific conditions under which Euler’s equation reduces to Bernoulli’s equation?
dp/dx = µd 2 u dy 2
For flow between two parallel plates having a separation of 2L, derive an expression for the velocity of the fluid in the
y direction. Using this velocity profile, derive an expression for the pressure drop −dp/dx as fluid flows along the plates.
Exercise 15.2
For fully developed, one-dimensional, steady, and incompressible flow through a circular pipe, show the Navier–Stokes
momentum equation becomes
If the inner radius of the pipe is R, derive an expression for the velocity of the fluid in the radial direction. Using this
velocity profile, derive an expression for the pressure drop −dp/dx as the fluid flows along the pipe.
Exercise 15.3
In regions of space where the viscous terms are negligible compared to the inertial terms, show that the Navier–Stokes
equations reduce to Euler’s equation
ρ∂ v / ∂ t + ρV∇V = −∇p + ρg
Show how Euler’s equation can be reduced to Bernoulli’s equation along a streamline. Under what three conditions is
this derivation possible?
Exercise 15.4
Write down the equations describing the flow of fluid in a static, incompressible laminar boundary layer over a nuclear
fuel rod. How many of these equations are there? Can these equations be used to find the laminar friction factor for the
boundary layer? How many boundary conditions must be used?
608 The Conservation Equations of Fluid Mechanics
Exercise 15.5
In the original derivation of the Navier–Stokes equations, an energy equation was not included. Write down an energy
equation that is appropriate to use for a reactor coolant channel. How is energy transferred to the fluid in this case?
Exercise 15.6
Write the components of the material derivative and the advective derivative in three dimensions. What parameters are
actually required to evaluate these derivatives? In a Lagrangian reference frame (see Chapter 14), what do the compo-
nents of the advective derivative become?
Exercise 15.7
For laminar flow around solid objects such as nuclear fuel rods, the pressure field can be found rather easily from the
velocity field. For a one-dimensional flow over a flat plate, what does the general solution to the pressure field p(x)
become?
Exercise 15.8
Using the Reynolds decomposition, show how the Reynolds stresses can be defined for turbulent flow. For what
type(s) of problems do the Reynolds stresses become symmetric? How can these stresses be evaluated analytically or
experimentally?
Exercise 15.9
Air flows over a plate in a reactor cooling tower which is 4 m long and 2 m wide. The velocity of the air is 5 m/s at 15°C.
If ρ = 1.2 kg/m3 and ν = 1.47 × 10 −5 m2/s, calculate (1) the length of the plate over which the boundary layer is laminar
and the thickness of the boundary layer at that point, (2) the shear stress at the location where the boundary layer ceases
to be laminar, and (3) the total force on both sides of the plate before the boundary layer becomes turbulent.
Exercise 15.10
The thickness of the boundary layer over a flat object depends on the shape of the velocity profile that is assumed within
the boundary layer. When the flow is laminar, a linear velocity profile results in the following expression for δ: δ =
3.46x/√Rex and a parabolic velocity profile results in the following expression for δ: δ = 5.48x/√Rex.* For a fluid such as
water flowing over a flat plate, by what percent thicker is the boundary layer when the velocity profile is parabolic than
when the velocity profile is linear?
FIGURE 16.1 The density of water at atmospheric pressure. In reality, reactors operate at much higher temperatures and pressures, and the density
of water generally decreases under these conditions. For example, in a PWR core at 15.5 MPa and 300°C, its average density is about 720 kg/m3.
609
610 Single-Phase Flow in Nuclear Power Plants
FIGURE 16.2 The density of liquid water at the saturation point as a function of the system pressure.
generator operating at 5.5 MPa and 270°C, its density is about 770 kg/m3. Its actual dependence is tabulated in various
fluid property tables and is also available on the Internet. A URL where its properties are tabulated as a function of
temperature and pressure can be found at
https://www.engineeringtoolbox.com/water-density-specific-weight-d_595.html
Many similar sites can be found with the Google search engine. In addition to commercial PWRs and BWRs, Canadian
CANada Deuterium Uranium (CANDU) reactors use another version of water called heavy water. Heavy water absorbs
fewer neutrons than light water does, and hence, its neutronic properties are considerably different. However, most of its
physical properties including its density, its viscosity, and its thermal conductivity are similar. The physical properties of
heavy water are compared to those of ordinary water in Table 16.1. Normally, heavy water is not toxic to human beings, and
when properly used, it allows a reactor to operate with less enriched uranium or on natural uranium alone.
TABLE 16.1
The Physical Properties of Heavy Water and Light Water, Which Are Two Popular
Reactor Coolants
function of their temperature and their pressure; that is, ρ = ρ(P, T). For monatomic gases, the density ρ, the temperature
T, and the pressure P are related to each other by the ideal gas law, which was introduced in Chapter 7. The ideal gas
law can be used to determine the behavior of a gas while it is being compressed, and the ideal gas law plays an impor-
tant role in determining the efficiency of power plant thermal cycles such as the Brayton thermal cycle (see Chapter 9).
Essentially, the ideal gas law is an equation of state for an ideal gas.
−(dP/dx) = − µ ⋅ d 2 v x dx 2 (16.1)
where μ is the dynamic viscosity measured in kg/m s. If we were to write this equation in a cylindrical coordinate system
(so that it is applicable to a problem like a horizontal circular pipe), the momentum equation we would have to solve
would be
If the fluid were frictionless, then µ = 0, and the momentum equation would become
dP/dx = 0 (16.3)
So when a fluid has no internal friction, the pressure will be the same everywhere (P = a constant), since dP/dx = 0. So in
the absence of gravity (which can be ignored for a horizontal pipe), friction is what causes the pressure drops in reactor
fuel assemblies to occur, and if there is no friction, it simply takes a constant force F = PA to move the fluid through the
system. Equation 16.3 is not a very realistic equation, but it indicates that friction can cause pressure changes to occur
in reactor piping systems and in the core.
coolant channels in a reactor core are no exception to this rule. Before we can discuss the concept of fluid friction, we would
first like to demonstrate how fluids behave when they are frictionless. This causes the momentum equations to become very
simple, and for frictionless flows, the continuity equation is the equation that has the greatest direct effect on how the flow
behaves. The continuity equation simply states that the flow of fluid into a control volume must be balanced by the flow of
fluid out of the control volume. In other words, the mass flow rate through a coolant channel or a coolant pipe must always
be conserved. Now let us turn out attention to a famous equation of classical fluid mechanics called Bernoulli’s equation.
Fluid friction can also be incorporated into Bernoulli’s equation under certain conditions. This allows Bernoulli’s equa-
tion to be used when there are valves, elbows, or tees in the reactor piping system and when there are frictional losses
associated with these devices.
FIGURE 16.3 Daniel Bernoulli and the famous equation that bears his name. (Courtesy of Wikipedia.)
16.8 Applications of Bernoulli’s Equation to Pipes and Other Simple Structures 613
incompressible. The one notable exception to this rule is CO2 which is used in gas reactors. However, for water reactors,
a great deal can be gleaned about their behavior from the simple application of Bernoulli’s principle. In its most basic
form, Bernoulli’s equation, which follows from Bernoulli’s principle, can be thought of as an energy conservation equa-
tion for inviscid flows. It states that the total energy E of a fluid (which consists of its kinetic energy plus its potential
energy plus its pressure energy) is always constant. We can express this fact by writing
Bernoulli’s Principle
Kinetic energy + Potential energy + Pressure energy = Constant (16.5)
A common form of Bernoulli’s equation that follows from this principle and is valid at an arbitrary point along a stream-
line where the gravitational acceleration g is a constant is
Bernoulli’s Equation
1
2 ρv 2 + ρgz + P = constant (16.6)
where
☉☉ v is the velocity of the fluid.
☉☉ P is the pressure of the fluid.
☉☉ g is the acceleration of gravity (normally 9.8 m/s2).
☉☉ z is the axial elevation of the fluid (in m or cm) in a flow channel or coolant pipe.
☉☉ ρ is the density of the fluid.
Bernoulli’s equation can be used to find how the pressure in a reactor coolant channel changes as a function of the
flow area A and the axial elevation z. However, it does NOT account for the effects of friction on the pressure drop.
Nevertheless, it is possible to include the effects of fluid friction if we can correlate the pressure drop ΔP with the aver-
age velocity v of the fluid. We will demonstrate how this can be done in Section 16.12.
FIGURE 16.4 The radial velocity profiles in a simple pipe for frictionless, laminar, and turbulent flows. The radial veloc-
ity profile (and hence the mass flow rate) in a reactor coolant channel depends on whether or not a fluid has friction. If a fluid is
frictionless, the radial velocity profile will be the same everywhere, but if a fluid has a finite viscosity (and therefore some internal
friction), the fluid velocity profile will be a function of radial position. In this case, some of the fluid molecules will cling to the
pipe wall, and this is referred to as a no-slip condition. The shape of the radial velocity profile will then depend upon whether the
flow is laminar or turbulent.
Any inviscid flow will have this velocity behavior, and moreover, the radial velocity profile will also be independent
of the mass flow rate m = dm/dt = ρvA. If we can ignore the effects of compressibility, then the density ρ will also be
constant, and thus, the velocity of the fluid for a given mass flow rate m will only be a function of the cross-sectional
area A of the pipe:
v = m/
ρA (16.8)
Equation 16.8 implies that the velocity of an inviscid flow will change when only the cross-sectional area A of the cool-
ant channel changes. Now, let us see if we can explore the implications of this equation for a horizontal pipe. When the
coolant expands or contracts (because it is heated or cooled or because of an area change), it will tend to accelerate or
decelerate as it travels through the pipe. This acceleration (or deceleration) will cause a change in the momentum of the
fluid, and this will create a net force. This force F is equal to the acceleration pressure drop ΔPACCELERATION times the
cross-sectional flow area A of the pipe.
This pressure drop is usually small for single-phase fluids, but it can become quite large for two-phase flows. If vIN is
the velocity of the fluid at the inlet to the pipe, and vOUT is the velocity of the fluid at the outlet of the pipe, then it follows
from Bernoulli’s equation that
F = A ⋅ ∆PACCELERATION (16.11)
Hence, even a minor density change (ρ OUT vs. ρIN) induced by a change in temperature can cause the pressure in the
pipe to change. However, a more common way for the pressure to change is if there is an area change in the direction
of the flow. Consider for the moment the pipe shown in Figure 16.5 which undergoes a sudden expansion or contraction.
The fluid entering the pipe is initially a single-phase liquid having no bubbles or voids. Now, we would like to calculate
16.8 Applications of Bernoulli’s Equation to Pipes and Other Simple Structures 615
FIGURE 16.5 Using a manometer, it is possible to measure the pressure drop in a pipe with an area change. The pressure is higher
when the fluid is moving slowly and lower when the fluid is moving faster. For a frictionless flow, h1 = h2 and h3 = h4, and for a flow
with friction h1 > h2 and h3 > h4.
the difference in the pressure between point 1 (where we know ρ1, A1, and v1) and point 2 (where we know ρ2, A2, and v2).
Then according to Bernoulli’s equation,
P1 + 1 2 ρ1 v1 2 = P2 + 1 2 ρ2 v 2 2 (16.12)
If the density is constant, then we can set ρ1 = ρ2 = ρ, and the pressure at point 1 is related to the pressure at point 2 by
P1 + 1 2 ρv1 2 = P2 + 1 2 ρv 2 2 (16.13)
1 = m
Moreover, we know from the continuity equation (see Chapter 15) that m 2 so
ρv1 A1 = ρv 2 A 2 (16.14)
This then allows us to establish a unique relationship between the upstream velocity v1 and the downstream
v elocity v2:
v 2 = v1 A1 A 2 (16.15)
( (
∆PACCELERATION = P2 − P1 = 1 2 ρv1 2 1 – A1 A 2 ) ) (16.16)
2
So when A1 ≠ A2, there will be a pressure change. Now, let us see if this causes the pressure at P2 to larger or smaller
than the pressure at P1. There are two cases to consider.
Case 1: If A1 > A2, then the fluid will flow faster at point 2 than at point 1, and the pressure will drop.
Case 2: If A2 > A1, then the fluid will flow slower at point 2 than at point 1, and the pressure will rise.
Thus whenever the fluid accelerates, the pressure will drop, and whenever the fluid decelerates, the pressure will rise.
It does not take much imagination to see that this is exactly what causes an airplane to fly! The air moves over the
top of the wings faster than it does over the bottom of the wings (see Figure 16.6), and this causes the pressure on the
616 Single-Phase Flow in Nuclear Power Plants
FIGURE 16.6 A fluent simulation of fluid flow over an aircraft wing. (Courtesy of NASA.)
top of the wings to be lower than the pressure on the bottom of the wings. When we multiply this pressure difference
ΔPACCELERATION times the area of the wing AWING, we get the lift force on the wing
So in addition to being useful in reactor work. Bernoulli’s equation can be used to explain why airplanes fly
Example Problem 16.1 helps to illustrate this point.
P1 + 1 2 ρ1 v1 2 = P2 + 1 2 ρ2 v 2 2
where P1 is the pressure before the area contraction and P 2 is the pressure after it. Hence,
( ) ( ) ( )
P2 = P1 + 1 2 ρ v1 2 − v 2 2v 2 2 = 690,000 + ( 960/2 ) 52 – 20 2 = 690,000 – 480 × 375 = 690,000 − 180,000 = 510,000
Pa = 510 kPa ≈ 74 PSI . Hence, the pressure is reduced by about 25% due to the area change alone. In practice, this pres-
sure drop is called the acceleration pressure drop. [Ans.]
FIGURE 16.7 The effect of a blockage in a reactor coolant channel on the flow in the axial and lateral directions.
the fluid will be the same in all channels. However, some coolant can flow between the central channel and the adja-
cent channels if there is a gap between the fuel rods. Before the blockage occurs, all three channels behave in exactly
the same way and the pressure drop is uniform (see Figure 16.7). However, as soon as the central channel becomes
blocked, the pressure downstream of the blockage falls and the pressure immediately in front of the blockage rises. Just
downstream of the blockage, water flows into the blocked channel from either side. This flow continues until the pres-
sure equalizes itself between all three channels in the lateral direction. (Again refer to Figure 16.7). Thus, the blockage
distorts the flow field, and Bernoulli’s equation can be used to predict how much fluid must flow from the unblocked
channels in to the blocked one to equalize the lateral pressure drop.
16.10 T he Effect of Directional Changes in the Flow Field on the Coolant Pressure
In addition to area changes caused by a blockage or the expansion or contraction of a pipe, there is another way that the
fluid can be accelerated or decelerated. In classical mechanics, the momentum of the fluid is a vector quantity that has
both magnitude and direction. So if the momentum of the fluid is given by
then changing the direction of the flow will cause its momentum to change (because the direction of v will change).
When this occurs the momentum change manifests itself as a pressure drop ΔP. In power plants, this occurs when the
coolant changes its direction as it goes around a bend in a pipe (which can also be an elbow or a tee). Figure 16.8 shows
when this can occur. When there is an elbow, the cross-sectional area of the flow channel does not change, but the
momentum certainly changes (because the pipe bends). So this directional change must create a force that acts on the
fluid, and in the process of accelerating or decelerating the fluid, the pressure must change. The size of this pressure drop
can be found by assuming that the pressure drop due to a change in direction or form alone is given by
FIGURE 16.8 When coolant pipes contain bends, elbows, or tees, the pressure falls when the flow direction changes. This is due to
a change in the linear momentum, and the pressure drop created by this directional change is proportional to the loss coefficient K.
The value of the loss coefficient is different for directional changes than it is for area changes, and it is also affected by the value of
the fluid viscosity.
where K is a dimensionless coefficient known as the loss coefficient or the Borda–Carnot coefficient. Sometimes it is
also called the k factor or the form factor. Thus, when the fluid is moving and there is a rapid directional change, the
total pressure drop consists of two terms: one due to an area change ΔPAREA and the other due to a direction change
ΔPDIRECTION. The total pressure drop is then the sum of these two terms:
where for steady flows, ΔPDIRECTION is due to the advective derivative (see Chapters 14 and 15). The value of
ΔPDIRECTION can be reduced slightly if the change in flow direction can be made more gradual. When this occurs,
the value of K will approach 0, and only the area change ΔPAREA will become relevant again. The pressure drop due
to viscous dissipation can be reduced even further if we can make the directional change as gradual as possible.
Then the flow will not separate, and this results in what is called a subsonic diffuser. Normally, the values for K are
measured experimentally and they are then tabulated for pipe fittings having different sizes and shapes. Using such
a table, we can then find the geometry we need, and look up the appropriate values for K. However, sometimes when
unique geometries are involved (like those in a reactor steam generator), the fluid can flow through a support plate
before it gets to the tube nest. Then there is usually no “official table” that has the appropriate value for K. We may
then have to adjust the value for K we find in the literature, or generate a new value from the experimentally measured
pressure drop ΔPDIRECTION. Because a change in direction is usually accompanied by a change in form, the K factor is
sometimes called a form factor. Thus a form factor can also be used to model flow blockages, pipe entrance and exit
effects, and comparable situations where the direction of the flow changes rapidly. Situations of this type are shown
in Figure 16.8. The loss coefficient at the inlet to a pipe is a function of the inlet geometry. It is much smaller for
smooth pipes with well-rounded inlets than it is at the inlet of pipes with sharp edges. A sharp-edged inlet can cause
about half of the velocity head to be lost as the fluid enters the pipe, and in turn, this can be attributed to the fact that
the sharp edges cause the flow to separate. Pipes with well-rounded inlets do not have this problem. Moreover, pipes
with sharp-edged inlets can act like flow constrictors. The velocity increases as the flow contracts and this causes the
pressure to fall. Thus, the area contraction increases the pressure drop at the inlet to the pipe, and this pressure head
reduction can be represented by an increase in the loss coefficient.
The total pressure drop ΔPTOTAL is then the algebraic sum of the individual pressure drops between the inlet and the
outlet. In most cases, coolant loops in the NSSS can be assumed to consist of a single flow path with many bends, turns,
and area changes. In these cases, the value of K is usually less than 1.0 for turbulent flows. However, it can become
much larger if the area changes discontinuously (such as when we have an abrupt area change where the ratio of A1 to
A2 becomes very, very large or very, very small). Values for the loss coefficient for common piping system geometries
are shown in Table 16.2. The head loss at the inlet of a pipe is almost negligible for well-rounded inlets (K ≈ 0.03 for
an edge radius to pipe diameter ratio >0.2), but it increases dramatically to about 0.50 for sharp-edged inlets. In other
16.10 The Effect of Directional Changes in the Flow Field on the Coolant Pressure 619
TABLE 16.2
The Loss Coefficients for Various Components of a Reactor Piping System When the Flow Is Turbulent
words, it can change by a factor of about 15 between a sharp-edged inlet and a well-rounded one. Thus, the reader
must always take the design of the edge into account when determining the size of the loss coefficient at the entrance.
Consequently, if we temporarily neglect the effects of gravity and wall friction, then the pressure drop will consist of
the sum of two terms
( (
∆PTOTAL = 1 2 ρv1 2 1 – A1 A 2 ) )+
2
1
2 Kρv 2 2 (16.22)
or
( )
∆PTOTAL = 1 2 ρv1 2 K + 1 – (β)2 (16.23)
where β = A1/A2 is sometimes called the expansion or contraction factor for the flow channel. Notice that when β = 1,
the area of the flow channel does not change, and the pressure drop between points 1 and 2 is
In this case, the pressure drop for a frictionless fluid is due to directional changes only.
For fluids with friction, the value of K will be different because directional changes will also cause frictional losses,
and this simple formula handles everything well but the effects of friction (and gravity). Now, let us see how we can add
friction and gravity to the expression for the pressure drop. We will then demonstrate how this approach can be extended
to geometries having more than one flow dimension.
620 Single-Phase Flow in Nuclear Power Plants
or more concisely,
∆PELEVATION = ρg ( z 2 − z1 ) = ρg ⋅ ∆z (16.26)
where Δz = z2 − z1 is the elevation change. In a reactor, this can be a significant component of the total core pressure
drop. For example, a column of liquid water about 1 meter high at room temperature will have a pressure difference
between the top and the bottom of about 1.42 PSI (9.8 kPa). So if a fuel assembly in a light water reactor (LWR) is 5 m
long, the effective pressure drop due to elevation alone (adjusted for temperature differences) would be about 7.1 PSI
(or ~46 kPa)*. This is about one-quarter of the total core pressure drop in a typical PWR core. The rest of this pressure
drop is due to friction, acceleration, deceleration, form losses, and other comparable effects. The density of water at
room temperature is about 1 g/cm3 (see Table 16.3), but at 315°C and 2,000 PSI, it is 0.69 g/cm3. Similarly, in a BWR
at about 270°C (540°F) and 1,000 PSI, it is about 0.75 g/cm3. In an LMFBR cooled with liquid sodium, this pressure
difference can be even greater because although liquid sodium is about 7% lighter than ordinary water at room tem-
perature, it is about 35% heavier at normal operating conditions. Hence, if a fast reactor fuel assembly was 5 m long,
the total pressure drop between the bottom of the core and the top of the core due to elevation effects alone would
be about 7 PSI (~50 kPa) at 350°C. Needless to say, it would take a very strong pump to pump a liquid metal coolant
through the core. Clearly, we cannot neglect the effects of an elevation change when calculating the pressure drops in
reactor fuel assemblies.
TABLE 16.3
The Physical Properties of Water as a Function of Temperature at Atmospheric Pressure
Temperature Density Pure Density Pure Density Tap Density Pure Specific Gravity Specific Gravity
(°C) Water (g/cm3) Water (kg/m3) Water (g/cm3) Water (lb/ft3) at 4°C at 60°F
0 (solid) 0.9150 916.0 — — 0.915 —
0 (liquid) 0.9999 999.9 0.99987 62.42 0.999 1.002
4 1.0000 1,000 0.99999 62.42 1.000 1.001
20 0.9982 998.2 0.99823 62.28 0.998 0.999
40 0.9922 992.2 0.99225 61.92 0.992 0.993
60 0.9832 983.2 0.98389 61.39 0.983 0.985
80 0.9718 971.8 0.97487 60.65 0.972 0.973
100 (gas) 0.0006 See steam tables — —
Source: From hypertextbook.com.
between the inlet and the outlet. However, to perform this integration, we must first know the shape of the velocity
profile v(r) in the radial direction. If the shape of this profile is unknown, then we can either estimate it or determine its
exact value by performing an experiment. In practice, there is always a frictional pressure drop ΔPFRICTION that appears
when a viscous fluid flows over a solid surface. Normally, this surface is called the “wall”. In a reactor core, this surface
corresponds to the surface of the fuel rods. In a pipe, it corresponds to the inner surface of the pipe. In any event, shear
stresses between the wall and the surrounding fluid distort the velocity profile, and the amount of distortion increases as
the flow rate increases. For most problems, it is possible to show that the value of the pressure drop due to friction alone
is proportional to the momentum of the fluid ρv2 multiplied by a constant that depends on the geometry of the coolant
channel. We can express this relationship by writing
∆PFRICTION = C ⋅ ρv 2 (16.28)
where C is the aforementioned constant. In general, C = f(L, D, ρ, μ). This description of friction works well for lami-
nar and turbulent flows, and it can even be adapted to geometries other than circular pipes where D is not uniform.
Obviously, if μ = 0, then C = 0, but this is almost never the case. The value of C is usually divided into a geometric
component, L/D, and a component that depends on the properties of the fluid and the rate at which it flows. These
components can then be combined together to create another parameter called the friction factor f. The friction factor
is different for laminar flow than it is for turbulent flow. Using Equation 16.28 as our guide, the frictional pressure drop
ΔPFRICTION can be written as
Obviously, the ordering of these terms depends on the particular problem one is trying to solve, and it can also depend
on the properties of the fluid itself. However, about 90% of the time, this order of magnitude analysis is indicative of the
size of the terms that cause the pressure levels in a reactor coolant channel to change. The reason why pressure changes
are so important is that the fluid always flows from a region of high pressure PHIGH to a region of low pressure PLOW (see
Figure 16.9). Therefore, if we can incrementally estimate the pressure drop as we move through the core, we can deter-
mine the direction that the coolant will flow at any given location. In addition, the rate that the fluid flows is proportional
to the pressure gradient between any two points. The larger the pressure gradient, the faster the flow will be. Finally, the
pressure drop through the core and through the piping system is needed to estimate the pumping power required to keep
the fluid flowing. This, in turn, enables us to understand how large the coolant pumps must be. The value of f is different
for laminar flows than it is for turbulent flows. Normally, it is a function of the velocity of the fluid v, the density of the
fluid ρ, and the viscosity of the fluid μ:
f = f(v, ρ, µ) (16.31)
For a variety of reasons, the value of f declines as the flow becomes more turbulent. However, the frictional pressure drop
ΔPFRICTION always increases as the velocity v increases because of the ρv2 term in the Darcy equation (Equation 16.29).
In Chapter 17 we will show that the value of f can be correlated to another important parameter of fluid mechanics called
the Reynolds number Re. For enclosed flows, the Reynolds number is defined as
Re = ρvD e /µ (16.32)
where De is the equivalent diameter of the flow channel. Thus, the Reynolds number represents the ratio of the momen-
tum (or inertia) of the fluid ρv to its internal viscosity μ.
FIGURE 16.9 The fluid in a reactor always flows from a region of high pressure to a region of low pressure, and the rate of flow is
proportional to the pressure gradient dP/dx. The higher the pressure gradient, the greater the flow rate will be.
channel that is not circular in shape, the concept of the equivalent diameter De must be used instead. Sometimes the
equivalent diameter is also called the effective diameter or the hydraulic diameter, but all three terms refer to essentially
the same thing. In fluid mechanics, the equivalent diameter is defined as
D e = 4A/PW (16.33)
where A is the cross-sectional area of the flow channel perpendicular to the direction of flow, and PW is its wetted perim-
eter. The wetted perimeter is defined as the perimeter of the flow channel, which includes all of the surfaces wetted by
the coolant. These surfaces can include the surfaces of nuclear fuel rods, control rods, or even the channel walls. In other
words, the equivalent diameter is the diameter that gives the same channel volume-to-heat transfer surface ratio in
both circular and noncircular channels. When used in this way, the equivalent diameter has the dimensions of length.
To illustrate this point, suppose we would like to calculate the equivalent diameter for a purely circular flow channel.
In this case,
A = πD 2 /4 and PW = πD (16.34)
In other words, when a channel is truly circular, then De = D, which is exactly the result we would expect. Now, suppose
that we would like to find the value of De for a square duct. Suppose that the sides of this duct have the dimension L.
Then according to Equation 16.33,
A = L2 and PW = 4L (16.36)
FIGURE 16.10 Values of the equivalent diameter for pipes with different sizes and shapes.
Heat transfer coefficients and friction factors for noncircular channels can be obtained from the equations we have
already presented by using the value of De in these equations. However, their accuracy is only acceptable when a
noncircular channel has about the same shape (or aspect ratio) as a circular one. The further a channel deviates from
being exactly circular in shape, the larger the error in the calculations will be. So while the equivalent diameter does
not strictly apply to every problem we will ever encounter in reactor work, it certainly applies to most problems as long
as we know what its limitations are. Examples of coolant channels where the equivalent diameter will give reasonably
good results include
☉☉ Elliptical channels
☉☉ Square channels
☉☉ Rectangular channels with small aspect ratios
☉☉ Hexagonal channels
☉☉ Triangular channels where all of the sides have approximately the same length.
Examples of coolant channels where the equivalent diameter does not give good results are
☉☉ Channels with large aspect ratios
☉☉ Channels that are highly elongated
☉☉ Channels that are not symmetrical.
If any of the final three conditions applies, then we can still use the equivalent diameter in our calculations, but the error
will be much greater. In some cases, it can be as great as 30%. Finally, the accuracy of our estimates will decline even
further if the flow is laminar (Re < 2,300), if the Prandtl number is low (which it can be for liquid metals), and if the
temperature gradient near the surface of the wall is much less than the velocity gradient (which is also true for certain
gases and liquid metals). In reactor coolant channels, one is typically confronted with geometries that resemble those
shown in Figure 16.11. In this case, the flow area is A = P2 − πD2/4, the wetted perimeter is PW = πD, and the equivalent
diameter can be shown to be
Equivalent Diameter for the Interior Coolant Channels in a Reactor Fuel Assembly
( )
D e = 4A/PW = 4 P 2 − πD 2 /4 πD = 4P 2 /πD − D (16.38)
16.14 Calculating the Total Pressure Drop in a Coolant Channel 625
FIGURE 16.11 The center, edge, and corner subchannels in LWR fuel assemblies (on the left) and in LMFBR fuel assemblies
(on the right).
where P is the pitch or the center-to-center spacing between adjacent fuel pins. Of course, Equation 16.38 only applies
to the interior channels in a square lattice. The corner channels and the edge channels will have different equivalent
diameters if the fuel assembly is surrounded by a can. The values for a square lattice are shown in Table 16.4. The values
for a triangular (or hexagonal) lattice are left as an exercise for the reader. The values for a few other common geometries
are also shown in the table. In reactor fuel assemblies, most correlations for the turbulent friction factor work best when
the pitch-to-diameter (P/D) ratio is 1.1 or more. In fact, Figure 16.12 shows that the best results can be obtained (at least
when the coolant is water) when the P/D ratio is close to 1.1. As the pitch gets larger, the flow channel is still close enough
to a circular one that the error never rises above about 10%. On the other hand, when the P/D ratio falls below 1.1, the
difference between a circular channel and a channel having another shape can become significant. In particular, very
tightly packed reactor fuel assemblies with a P/D ratio of 1.02 can have an average error of about 30% compared to the
identical friction factors for a circular pipe.
TABLE 16.4
The Equivalent Diameters for Different Flow Channels Encountered in Reactor Work
FIGURE 16.12 Turbulent friction factors in reactor fuel assemblies behave differently than they do in circular pipes. When the
standard definition of the equivalent diameter is applied to a subchannel, the Nusselt numbers and friction factors are different than
they are for circular pipes. Below P/D ratios of about 1.1, the differences based on the use of the equivalent diameter can be as great
as 30%. In Chapter 17, we will show how geometric correction factors can be used to compensate for these inconsistencies.
We can then write the total pressure drop ΔPTOTAL as the algebraic sum of these individual pressure drops:
We also know that explicit expressions exist for each of these terms. Thus, the total pressure drop can be written as
( )
∆PTOTAL = f·L D e 1 2 ρv 2 + K 1 2 ρv 2 + 1 – (β)2 1 2 ⋅ ρv 2 + ρg ⋅ ∆z (16.40)
FRICTION DIRECTION AREA ELEVATION
( )
∆PTOTAL = 1 2 ρv1 2 fL/D e + K + 1 – (β)2 + ρg ⋅ ∆z (16.41)
where β = A1/A2. We can use Equation 16.41 to find the single phase pressure drop in any reactor component. However,
it must be kept in mind that L, De, K, and β are the functions of the geometry, Δz is a function of the height, ρ is deter-
mined by the type of fluid, and f is a function of the Reynolds number (e.g., the operating conditions). Example Problem
16.6 helps to illustrate what their values are for a typical LWR. If the coolant boils, then most of the values will be
different than those presented here.
ΔPELEVATION = ρg Δz = 710 × 9.81 × 4 = 27,860 Pa = 27.8 kPa or 4 PSI. Hence, the frictional pressure drop at rated flow
is about twice the gravitational pressure drop. [Ans.]
∆P = p2 – p1 = 1 2 ρ ( v 2 2 – v1 2 ) + 1 2 K c ρv 2 2 (16.42)
∆P = p2 – p1 = 1 2 ρ ( v 2 2 – v1 2 ) − 1 2 K e ρv1 2 (16.43)
Here, subscript 1 refers to the upstream value, and the subscript 2 refers to the downstream value. The irrecoverable
pressure losses are determined by the values Kc and Ke. The values of these coefficients for some different configura-
tions are shown in Table 16.5. Generally speaking the loss coefficients have finite values even when the coolant is
frictionless. This is because the pressure drops that are represented by the loss coefficients are partially due to area and
FIGURE 16.13 In a reactor piping system with area changes, the pressure can increase or decrease. For viscous flows, the pressure
drops are not only a function of the area ratios but also the viscosity of the coolant. In practice, loss coefficients are used to find the
pressure changes due to area changes, and when the flow area is constant, the pressure drop is proportional to the friction factor f.
TABLE 16.5
Some Representative Loss Coefficients for Turbulent Flow in a Reactor Piping System
directional changes, and the remaining components of these drops are due to fluid friction. For example, suppose that
we would like to find the total pressure drop in a pipe that contains a number of elbows, tees, and valves. If we express
the total pressure drop as a function of the mass flow rate m = dm/dt = d(ρvA)/dt, it is easy to see that this leads to the
following equation for the pressure drop:
2 /2ρ ⋅ ( K1 + K 2 + K 3 + + K N ) + ρg ⋅ ∆z (16.44)
∆PTOTAL = m
Here, there may be N separate components in the pipe, and each of these components may have different loss coef-
ficients. However, when we add the effects of fluid friction back into Equation 16.44 (especially away from the compo-
nents themselves), the expression for the total pressure drop becomes
∆PTOTAL = m ( )
2 /2ρ ⋅ f L/D + K1 + K 2 + K 3 + + K N + ρg ⋅ ∆z (16.45)
where f is the friction factor, which it is different for laminar flows than it is for turbulent flows.
16.16 Finding the Flow Rates and the Pressures in Nozzles and Diffusers
Nozzles and diffusers can be found in almost all nuclear power plants. Sometimes they are found in the steam turbines and
other times they are found in the containment building or the cooling towers. Practically speaking, a nozzle is a device that
increases the velocity of the coolant by reducing its pressure. Conversely, a diffuser is a device that increases the pressure
of the coolant by slowing it down. Hence, nozzles and diffusers perform complimentary functions, and their designs are
determined by the applications they are intended to perform. Both devices use area changes to redirect the coolant and
accelerate or decelerate it. These area changes lead to pressure changes that cause the fluid properties at the inlet to be dif-
ferent than the fluid properties at the outlet. In practice, the flow of heat between the surface of a nozzle or a diffuser is small
compared to the kinetic energy of the flow. Hence, when performing an energy balance on the flow, one can normally set
Q′ ≅ 0 because the residency time of the coolant is too short for any significant amount of heat transfer to occur.
Nozzles and diffusers are usually not intended to perform external work, and changes to the potential energy of the
flow are relatively small because the inlet and the outlet are normally located at the same elevation. When this occurs,
we can set W′ ≅ 0 and ΔPE ≅ 0 when performing an energy balance on either device. Furthermore, because both devices
rely on large velocity changes to perform their intended functions, large changes in the kinetic energy of the flow can
occur between the inlet and the outlet. This means that the kinetic energy of the coolant (KE) is usually the most impor-
tant parameter in determining how nozzles and diffusers perform. In general, a energy balance can then be used to
find the temperature and pressure at the outlet if the temperature and pressure at the inlet are known. When it comes to
nozzles, the outlet pressure is always lower than the inlet pressure, and when it comes to diffusers, the outlet pressure is
always higher than the inlet pressure. Thus, nozzles accelerate the flow, while diffusers decelerate it (see Figure 16.14).
Hence, if the inlet mass flow rate m IN, the inlet velocity vIN, and the inlet specific enthalpy hIN are known, then their
values at the outlet are related to their values at the inlet by
( )
IN h IN + v IN 2 2 = m
m ( )
OUT h OUT + v OUT 2 2 (16.46)
when there is just a single inlet and outlet. If flow energy is dissipated due to the effects of friction then an additional
term Q ′FRICTION must be added to the right-hand side to ensure that the energy of the fluid is reduced. Then, the energy
equation for a nozzle or diffuser becomes
( )
IN h IN + v IN 2 2 = Q ′FRICTION + m
m ( )
OUT h OUT + v OUT 2 2 (16.47)
Equation 16.47 applies to steady flows in nozzles and diffusers. Now, let us show how the coolant velocities at the inlet
and the outlet are related. If the inlet mass flow rate is known, and there is just one inlet and one outlet, then the mass
flow rates and the velocities are related by
IN = m
m OUT (for the mass flow rate) (16.48)
or
FIGURE 16.14 Nozzles and diffusers perform opposite functions in nuclear power plants. Nozzles accelerate the fluid flowing
through them, and diffusers decelerate it.
Here, the subscripts IN and OUT denote the state of the fluid at the inlet and outlet, ρ is the density of the coolant, v is
the average velocity of the coolant in the direction of flow, and A is the cross-sectional area perpendicular (or normal)
to the flow. Now suppose we would like to apply these equations to one problem involving a nozzle and another problem
involving a diffuser. In the first problem, water is the working fluid and the specific enthalpy of the water is known.
In the second problem, the working fluid is air.
( )
IN h IN + v IN 2 2 = Q ′OUT + m
m (
OUT h OUT + v OUT 2 2 )
Plugging in the appropriate values for h and v, we find that hOUT = 3,150 kJ/kg. At 200 PSI (1.38 MPa), the temperature
of the steam exiting the nozzle (from the steam tables) is 350°C. Hence, the temperature of the steam drops by approxi-
mately 20°C as it passes through the nozzle. (The heat lost through the nozzle walls is too small to have any significant
effect on the outcome). In fact, this example also implies that if the exit pressure falls below the saturation pressure,
a nozzle can convert a single-phase liquid into a two phase mixture. Many industrial processes are designed to take
advantage of this fact. [Ans.]
Now let us turn our attention to diffusers. A diffuser is a device that slows down the fluid that passes through it and this
causes its kinetic energy to decrease. Thus, the pressure and the specific enthalpy must rise, and if no phase change
occurs, the temperature must increase as well. This enthalpy rise is due primarily to the conversion of kinetic energy into
internal energy. If the velocity differences between the inlet and the outlet are large enough, then this enthalpy change
can be significant. Example 16.8 shows how large this enthalpy change can be. In particular, it shows how a diffuser can
change the temperature of ordinary air between the inlet and the outlet. Again, the air temperature rises because the air
pressure at the exit increases.
630 Single-Phase Flow in Nuclear Power Plants
Thus, nozzles and diffusers perform complementary (but opposite roles) in nuclear power plants. They can change the
velocity of the coolant flowing through them, and the thermodynamic properties at the outlet will be different than
the thermodynamic properties at the inlet. Sometimes nozzles and diffusers are deployed in serial or parallel arrays to
perform more complex tasks.
where each term has the units of pressure. Therefore, each term represents a different type of pressure. We can then clas-
sify these pressures into three broad categories which are explained below. Each category is associated with a different
physical process or condition.
FIGURE 16.15 The relationship between the dynamic pressure, the static pressure, and the total pressure can be derived from the
fluid momentum equation. (Figure provided by NASA.)
into a reactor core (from above), the greater this hydrostatic pressure will be. The hydrostatic pressure is equal to
PHYDROSTATIC = ρgz, where z is the depth within the liquid. The only place where the hydrostatic pressure does not exist
is at the surface of the fluid (where z = 0), or in outer space, where g = 0. At sea level, the value of g is about 9.8 m/s2, and
this value of g is called the gravitational acceleration. The hydrostatic pressure is responsible for part of the pressure
head ΔP that a reactor coolant pump (RCP) must overcome to set the fluid in motion. The sum of the static, the dynamic,
and the hydrostatic pressure is then equal to the total pressure PTOTAL:
PTOTAL = PSTATIC + PDYNAMIC + PHYDROSTATIC = CONSTANT (16.51)
The sum of the static and dynamic pressures is also called the stagnation pressure PSTAGNATION.
PSTAGNATION = PSTATIC + PDYNAMIC = P + ½ρv2 (16.52)
The stagnation pressure can be thought of as the pressure at a point where a working fluid is brought to rest isentropi-
cally. The fluid velocity when the hydrostatic pressure can be ignored) is
( )
v = √ 2 ( PSTAGNATION – P ) ρ (16.53)
Thus, the stagnation pressure is usually associated with the flow of fluid through a horizontal pipe where the flow
encounters an obstruction that causes the fluid to stop at the tip of the obstruction. The point at which the fluid stops
is sometimes called the stagnation point. The streamline that extends from the original flow to the stagnation point
is called the stagnation streamline (see Figure 16.16). The stagnation pressure is also the pressure measured by a
FIGURE 16.16 In fluid mechanics, a stagnation streamline can develop when the velocity of a fluid particle moving along a
streamline falls to zero at a solid surface. In this example, the surface happens to be the tip of an aircraft wing.
632 Single-Phase Flow in Nuclear Power Plants
Pitot tube, which is used to measure the pressure of a stagnating flow. However, reactors use more sophisticated pressure
sensors because the temperatures and the pressures in the NSSS are so high.
∫ ∂v ∂t ⋅ dr + ∫ dP/ρ + (v 2
)
2 + gz = c(t) (16.55)
and this equation is referred to as the integral form of Bernoulli’s equation. Here, c(t) is a constant of integration, which
in some cases is time dependent. This allows Bernoulli’s equation to be written as
2 2
∫ 1
∂v ∂ t ⋅ d r +
∫
1
(
dP/ρ + v 2 2 + gz ) − (v
2
2
)
2 + gz 1 = c(t) (16.56)
where points 1 and 2 are two points in the flow field. This equation can be used to predict the time-dependent flow as
long as the pressure drop ΔP = P2 − P1 is known. It can also be applied to many components of the nuclear steam supply
system (NSSS) such as those shown in Figure 16.17 when the Reynolds numbers are low.
FIGURE 16.17 The high-level components of a simplified coolant loop in a PWR. (Taken from newenergyweek.com.)
16.19 ESTIMATING THE START-UP TIME OF A RCP FROM BERNOULLI’S EQUATION 633
16.19 Estimating the Start-Up Time of a Reactor Coolant Pump from Bernoulli’s Equation
Bernoulli’s equation can be used to infer some additional relationships between the pressure drop and the mass flow
rate for a time-dependent flow. Equation 16.56 can also be used to model the start-up and shutdown of a reactor coolant
pump (or RCP). To show how this is done, assume for the moment that the flow is incompressible, one dimensional, and
irrotational. For these flows, Bernoulli’s equation can be written as
2
∂v ∂ t ⋅ dr + ( P2 – P1 ) + ρg ( z 2 – z1 ) + ρ /2 ( v 2 2 – v1 2 ) = 0 (16.57)
ρ
∫ 1
This equation applies to any two points on a streamline. Suppose that we would like to determine the time depen-
dent mass flow rate m(t) = dm/dt through a piping system consisting of N different sections having a total pressure
drop ΔP = POUT − PIN between the first component and the last. When the flow is incompressible, Bernoulli’s equation
becomes
N
ρ
∫
1
( )
∂v ∂ t ⋅ d r + ( POUT – PIN ) + ρg ( z N – z1 ) + ρ /2 v N 2 – v1 2 = 0 (16.58)
The path integral in this equation can be evaluated by assuming that the flow rate for an incompressible fluid does not
change as long as each of the components has a single inlet and outlet. Then, the path integral can be evaluated to read
∑
N N N E
∫ ∫ ∫ ∫
N
ρ ∂ v ∂ t ⋅ d r = ∂ / ∂ t ρ v ⋅ dr = ∂ / ∂t
(m/A) ⋅ dL = dm/dt
⋅ dL/A = dm/dt
⋅ L i A i (16.59)
1 1 1 1 i =1
∑
N
Here, the dm/dt L i A i can be thought of an equivalent inertial length that depends on the system’s internal
i =1
ρ, and A, we find that Bernoulli’s integral
geometry. Hence, after expressing the time-dependent velocity in terms of m,
becomes
+ ( POUT – PIN ) + ρg ( z N – z1 ) + m
(L/A)dm/dt (
2 2ρ 1 A N 2 – 1 A1 2 = 0 (16.60) )
where
∑
N
L/A = L i A i (16.61)
i =1
Now, let us examine what this modified form of Bernoulli’s equation says. All of the terms (with the exception of
the first) represent the values at the end points of a piping system segment. Hence, if the value of the pressure drop
ΔP = POUT − PIN is known or can be specified in advance, then Equation 16.60 can be solved for m(t). Unfortunately,
this equation is still a nonlinear equation because the time-dependent mass flow rate appears as the square of m(t).
Fortunately we can use a mathematical “trick” called partial factoring to convert it into a more manageable linear
differential equation. Suppose that we model the NSSS using the five components shown in Figure 16.18. These five
components represent one leg of the primary loop of a commercial PWR.
Now suppose that a RCP is used to establish a steady time independent flow rate in this loop. The coolant pump in
this case then becomes the round green object in Figure 16.18. Next, suppose that we would like to determine the time
it takes the reactor coolant, which is initially at rest, to reach this steady-state mass flow rate. Since there is only a single
coolant pump in this case, and the inlet and the outlet are at approximately the same elevation, we can solve for the
time-dependent mass flow rate if the pressure head ΔPHEAD = PIN − POUT is known. The only parameter that we have not
accounted for is the inertia of the pump’s rotors and impellers. However, if we can temporarily neglect this inertia, the
following equation can be used to find the time-dependent mass flow rate
= ∆P − m
(L/A)dm/dt (
2 1 A N 2 – 1 A1 2 2ρ (16.62) )
Here, the inlet and the outlet of the pump are assumed to be at the same a little differently elevation so that ρgΔz ≈ 0.
2. However, we can still find a solution by rewriting it in a
Clearly, Equation 16.62 is nonlinear because of the term m(t)
634 Single-Phase Flow in Nuclear Power Plants
FIGURE 16.18 The parameters of some of the important components in the primary loop of a commercial PWR.
slightly different form. Suppose that we take the time-dependent Bernoulli equation that describes the mass flow rate
through the loop
= ∆P − m
(L/A)dm/dt ( )
2 1 A 5 2 – 1 A1 2 2ρ (16.63)
and rewrite it as
dt = (L/A)(1/∆P)dm/ ( )
2 C2 (16.64)
1– m
where
(
C 2 = 1 A 5 2 – 1 A1 2 ) ( 2ρ ⋅ ∆P ) (16.65)
Then, we can also write Equation 16.64 as
(A/L) ∆P dt = dm/[(1
⋅ (1 + mC)]
– mC) (16.66)
or
−m
where we have replaced the term dm/(1 2C2) by the sum of two partial fractions. Thus, we have replaced Equation 16.64
After integrating Equation 16.67, we find that
with a linear differential equation that can be solved for m.
Since the mass flow rate at time t = 0 is m = 0, the constant of integration is α = 0. The final solution is
[(1 + mC)/(1
] = e λt (16.69)
– mC)
16.19 ESTIMATING THE START-UP TIME OF A RCP FROM BERNOULLI’S EQUATION 635
or
where λ = (A/L) 2C ΔP. In other words, this equation tells us that m → 1/C as t → ∞. Using representative values to
approximate the NSSS components shown in Figure 16.19, we find that the steady-state mass flow rate is ~30,000 kg/s
when the pressure head ΔP is 85 m. To obtain the time constant for the coolant pump, we use Equation 16.70 and esti-
mate how long it takes for the mass flow rate to reach 99% of its steady-state value. Since
∑
5
(L/A)LOOP = (L/A)i ≈ 103/m (16.71)
i =1
and
λ ≈ 0.55/s (16.72)
the time it takes for the system to reach 99% of its maximum flow rate is
or
Again, these values are correct as long as the presence of frictional forces can be ignored in the pipes.
16.20 S
olving Bernoulli’s Equation with Loss Coefficients
and Friction Factors in a Viscous Flow Loop
Now let us see if we can solve Bernoulli’s equation again using a friction factor and some realistic loss coefficients to
account for the behavior of the flow as it moves from the reactor pressure vessel to the steam generator. When we include
these parameters in the fluid momentum equation we obtain
+ K R v R 2 2 + K SG v SG 2 2 +
(L/A)dm/dt ∑ f(L/D) ρv
i
i i
2
2+m (
2 2ρ ⋅ 1 A 5 2 – 1 A1 2 = ∆P (16.76) )
where K R and KSG are the loss coefficients for the pressure vessel and the steam generator, and vR and vSG are the veloci-
ties of the flow in the pressure vessel and the steam generator. Writing this equation in terms of the total mass flow rate
m = ρvA, and assuming that the friction factor f is independent of position, we obtain
+ m
(L/A)dm/dt (
)
2 /2ρ 1 A 5 2 – 1 A1 2 + K R A R 2 + K SG A SG 2 + ∑ f(L/D) (1 A ) = ∆P (16.77)
i
i i
2
In other words, the only difference is that the expression for C has been converted from
( )
C2 = 1 A 5 2 – 1 A1 2 (2ρ ⋅ DP) (for an inviscid flow) (16.78a)
to C2 = 1 A 5 2 – 1 A1 2 + K R A R 2 + K SG K SG 2 +
∑ f(L/D) (1 A ) (2ρ ⋅ DP)
i
i i
2
(for a viscid one) (16.78b)
From our previous discussion, we know that the time dependent mass flow rate is
where only the value of C has changed. Now, let us determine the revised value for C using the values shown in
Table 16.6. For a commercial PWR, the loss coefficient within the core can average about 20, and for a modern steam
generator, the loss coefficient can average about 50. Assuming the friction factor is about 0.015, and the pump head ΔP
is approximately 85 m, the only remaining parameter we need to provide is the value of ρ. The density of saturated water
at 300°C and 2,200 PSI is about 720 kg/m3, but to keep the math simple, we will assume that the reactor is just being
TABLE 16.6
Some Representative Parameters Needed to Find the Mass Flow Rate through the Primary Flow Loop of a Hypothetical PWR
When the Pump Head ΔP Is Known
started up, so the core is very cool. In this case, assume that we can set ρ = 1,000 kg/m3. After doing so, we find that the
value of C2 is
The steady-state mass flow rate is then m = 1/C ≈ 7,450 kg/s within a single leg of the primary loop. The velocity
of the coolant exiting the core is v3 = (m/ρ)(1/A 3) = 7.45/0.4 ≈ 18.6 m/s. This is about three times the velocity of the
coolant through a typical fuel assembly. Hence, the inertia in the coolant system affects the time constant for the
pump. Notice that the steady-state mass flow rate is close to that of a typical PWR coolant pump in the primary loop
(7,500–8,000 kg/s). If the flow areas happened to be larger, the time constants (which measure the amount of fluid
inertia) would be larger as well.
∫ dP/ρ + ∫ dV/ dt ⋅ ds + 1
2 v 2 + gz = constant (16.82)
For a streamline, Equation 16.82 reduces to Equation 16.6 when the flow becomes steady:
1
2 ρv 2 + ρgz + P = constant (along a streamline) (16.83)
vorticity (see Chapter 15), then the value of C may be the same for many streamlines. Then, Bernoulli’s equation can be
used across multiple streamlines as well.
FIGURE 16.20 The flow field in a centrifugal pump with an impeller. Bernoulli’s equation can be used in the regions of the
pump before the impeller breaks up the streamlines. After the streamlines are destroyed, an energy balance must be used to
find the pressure drop across the pump. (From forum.simscale.com.)
16.24 Poiseuille’s Equation 639
FIGURE 16.21 Poiseuille’s equation can be used to find the viscous resistance and the volumetric flow rate in a circular pipe of
radius r. The mass flow rate through the pipe is then equal to the volumetric flow rate F multiplied by the density of the coolant.
μ is the dynamic viscosity, R is the radius of the pipe, and L is the length of the pipe. Normally, the volumetric flow rate
F is expressed in cm3/s, m3/s, L/s, or gallons/s. The significance of Poiseuille’s equation is that it can also be written as
( ) ( )
F = πR 4 8µ ⋅ ( P1 – P2 ) L = πR 4 8µ ⋅ dP/dx (16.86)
In other words, for laminar flows in circular pipes, it relates the volumetric flow rate F to the pressure gradient dP/dx
across the pipe
Notice that the flow rate through the pipe is proportional to the fourth power of the pipe’s radius R. To find the mass flow
(in kg/s or lb/s) through a circular pipe having a constant cross-sectional area A = πR2, we simply multiply the
rate m
volumetric flow rate by the fluid density ρ. This gives
( )
= ρA 2 /8πµ ⋅ dP/dx (16.89)
m
where A is the cross-sectional area of the pipe. In other words, once the flow has become established, the mass flow rate
is proportional to the square of the cross-sectional area A. Thus, doubling the cross-sectional area increases the flow rate
by a factor of 4! This is because the radial velocity profile v(r) is parabolic in a circular pipe when the flow is laminar (see
Figure 16.22). For optimum results, Poiseuille’s equation requires a fluid to be incompressible and Newtonian, the flow to
be laminar, and the pipe to have a constant circular cross section that is substantially longer than its diameter so that the
flow can become fully developed. In Chapter 17, we will discuss how the mass flow rate and the pressure drop can be found
640 Single-Phase Flow in Nuclear Power Plants
FIGURE 16.22 A FLUENT simulation of Poiseuille flow in a circular pipe performed by the author.
when the flow becomes turbulent. We will find that this requires us to replace Poiseuille’s equation by the Darcy equation,
to dP/dx. In general, the values for f
which uses an empirical friction factor f instead of (ρA2/8πμ) to relate the value of m
are a function of the Reynolds number. Hence, a coolant like water will have a different friction factor for a given flow rate
than a metallic coolant does. We have more to say about the Darcy equation and its applications in Chapter 17.
Hence, the Joule–Thomson coefficient measures the rate of change in temperature with pressure during a constant
enthalpy process. Throttling valves are usually small and the flow through them is normally adiabatic because there is
neither sufficient time nor sufficient surface area for any effective heat transfer to occur. Since there is no external work
done by these devices and most potential energy changes are small, we can set q ≅ 0, w = 0, and Δ PE ≅ 0. Finally, even
though the exit velocity is often considerably higher than the inlet velocity, the change in kinetic energy is usually small
compared to the change in internal energy. This allows us to set Δ KE ≅ 0. Then, the fluid energy equation becomes
where h2 is the enthalpy downstream of the valve and h1 is the enthalpy upstream of it. In other words, the enthalpies
at the inlet and the outlet are the same and the valve is said to be isenthalpic. Then, we can express Equation 16.91 as
u1 + P1υ1 = u 2 + P2 υ 2 (16.92)
16.25 Throttling Valves 641
FIGURE 16.23 The dependence of the Joule–Thomson on the temperature and pressure of liquid water.
(https://www.lanl.gov/expertise/files/5638f02236e8e_ As_published_ JT_TiPM_ONLINE.pdf.)
or
Internal energy + Flow energy = constant (16.93)
Thus, the outcome of the throttling process depends on which of these two quantities changes the most. If the energy of the
flow increases when P2υ2 > P1υ1, the internal energy must decrease, and this is usually accompanied by a sharp drop in the
fluid temperature. Conversely, if P2υ2 < P1υ1, the internal energy of the fluid will increase and the temperature of the fluid will
also increase. Now, let us see how this affects the value of the Joule–Thomson coefficient (Equation 16.90). If the tempera-
ture increases while the flow is being throttled, the value of K is negative. If the temperature decreases during the same pro-
cess, the value of K is positive. Finally, if K = 0, the temperature remains the same irrespective of whether the downstream
pressure increases or decreases. Typical values for the Joule–Thomson coefficient can range from about 0.025°C/MPa for
gases like nitrogen to between −0.25 and + 0.25°C/MPa for liquids like water. The Joule–Thomson coefficient for water is
temperature dependent, but it is usually negative below 250°C. Its actual dependence is shown in Figure 16.23.
All common fluids have a point on the temperature–pressure curve called the inversion point where the Joule–
Thomson coefficient changes sign. The temperature at this point is called the Joule–Thomson inversion temperature,
and hence, the Joule–Thomson coefficient depends on the pressure of the coolant before the throttling process begins.
In general, it is advantageous in nuclear power plants for the Joule–Thomson coefficient to be positive because the tem-
perature decreases as the pressure decreases and a phase transition is less likely to occur. However, in gas reactors, an
interesting effect sometimes occurs when one attempts to throttle a valve. If the coolant can be described by the ideal gas
law, then the enthalpy becomes a function of the temperature alone; that is, h = h(T). Then, T2 = T1 because h2 = h1, and
the Joule–Thomson coefficient described by Equation 16.90 has a value of zero. Hence, throttling a valve cannot be used
to raise or lower the temperature of an ideal gas! A formal proof of this assertion is given in Example 16.11. The Joule–
Thomson coefficients for several gaseous coolants including H, He, and CO2 are provided in the following reference:
https://en.wikipedia.org/wiki/Joule%E2%80%93Thomson_effect
Other information concerning this subject can be found in classical heat transfer books and on the Internet.
dh = c P ⋅ dT + ( υ − T(dυ /dT)P ) ⋅ dP
642 Single-Phase Flow in Nuclear Power Plants
( ) ( ) ( )
K = − 1 c P ( υ − T(dυ /dT)P ) = − 1 c P (υ − TR/P) = − 1 c P (υ − υ) = 0 Q.E.D.
The astute reader should not find this result to be surprising since the enthalpy of an ideal gas depends on its temperature
alone. Therefore, a throttling process cannot be used to change the temperature of the gas because the enthalpy remains
constant. [Ans.]
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M. C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, CRC Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc. New York,
NY (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi (2000).
Reactor Theory (Nuclear Parameters) DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, CRC Press, Boca Raton, FL (2014).
Web References
https://www.engineeringtoolbox.com/water-density-specific-weight-d_595.html.
http://www.unit-conversion.info/pressure.html.
https://en.wikipedia.org/wiki/Joule%E2%80%93Thomson_effect.
10. Fill in the following sentence with the appropriate word or phrase: In pipe flow, Bernoulli’s equation is most accurate
when it is applied to pipes having ________. In this case, the value of C in Bernoulli’s equation ½ρv2 + ρgz + p = C
is _________.
11. In Bernoulli’s equation, the energy of a fluid consists of three essential components. What are the names of these
components, and in Bernoulli’s equation, what terms do they represent?
12. Fill in the following sentence with the appropriate word or phrase: Practically speaking, Bernoulli’s equation can
be considered to be an expression of the conservation of _______ along a streamline.
13. Can Newton’s three laws of motion be used to derive Bernoulli’s equation?
14. As the saturation pressure of water is raised, how does its density behave at the saturation point?
15. Fill in the following sentence with the appropriate word or phrase: A fluid that has no friction is said to be ________.
16. How is a form coefficient or loss coefficient defined?
17. When a loss coefficient is used to describe the pressure change caused by a form change or a change in the velocity
of a fluid, what is this loss coefficient called?
18. Suppose a compressible fluid is flowing through an expanding pipe. Neglecting frictional losses, is the pressure at
the end of the pipe higher than, lower than, or the same as the pressure at the beginning of the pipe?
19. Suppose that a compressible fluid is flowing through a contracting pipe. Neglecting frictional losses, is the pressure
at the end of the pipe higher than, lower than, or the same as the pressure at the beginning of the pipe?
20. What is the difference between a sonic flow and a subsonic one?
21. Write down Bernoulli’s equation for the motion of fluid along a streamline.
22. At what value of the Reynolds number do streamlines typically disappear?
23. Fill in the following sentence with the appropriate word or phrase: The ________ is the pressure that exists in a flow
field when the flow is not moving.
24. Fill in the following sentence with the appropriate word or phrase: The _________ is the pressure that exists in
a flow field when the flow is moving. It represents the pressure increase when a fluid in motion is brought to rest
isentropically.
25. Fill in the following sentence with the appropriate word or phrase: The _________ is the pressure that exists in a
flow field because of gravitational effects. It represents the pressure exerted on a fluid particle from all of the other
fluid particles above it.
26. How is the equivalent diameter defined for a reactor coolant channel?
27. How is the hydraulic diameter defined for a circular pipe?
28. Under what conditions can the equivalent diameter be used to obtain accurate estimates of the pressure drop in a
reactor fuel assembly?
29. Under what conditions will the use of the equivalent diameter lead to significant errors?
30. Write down the energy equation that can be used to predict the behavior of the fluid through a nozzle or a
diffuser.
31. Does the pressure increase or decrease when a fluid flows through a nozzle? What happens to the pressure when it
flows through a diffuser?
32. In reactor fuel assemblies, are most of the observed pressure changes due to (1) frictional effects, (2) gravitational
effects, (3) directional changes, or (4) the acceleration and deceleration of a fluid?
33. Is the frictional pressure drop when a coolant boils less than, equal to, or greater than the pressure drop when it does
not? Assume in both cases that the mass flow rate is the same.
34. How can Bernoulli’s equation be used to predict the start-up times for a RCP? What version of Bernoulli’s equation
must be used in this case to obtain physically meaningful results?
Exercise 16.2
A reactor coolant pipe is allowed to expand gradually. Coolant enters the pipe at a velocity of 1 m/s with a density of
960 kg/m3. If the diameter of the pipe is eventually increased by a factor of 2, and the pressure in the pipe before the area
increases is 100 PSI (690 kPa), what is the pressure in the pipe after the area is changed? Assume that the streamlines do
not collapse as the flow decelerates so Bernoulli’s equation can be used.
644 Single-Phase Flow in Nuclear Power Plants
Exercise 16.3
Write down Bernoulli’s equation for a time-dependent compressible flow. For a streamline, write an alternative form
of Bernoulli’s equation when the flow becomes steady. What types of problems can this equation be applied to in
reactor work?
Exercise 16.4
For a reactor coolant channel, write an equation for the total pressure drop that accounts for area changes, directional
changes, elevation changes, and frictional losses. When the flow is laminar, which one of these terms is the largest?
When the flow is turbulent, which of these terms is the largest?
Exercise 16.5
Area changes in reactor coolant pipes occur all of the time. When an area change is an abrupt one, write down an equa-
tion that can be used to find the pressure change in the vicinity of a sudden expansion or contraction. How are the loss
coefficients in an equation such as this found?
Exercise 16.6
Water flows through a reactor coolant pipe at a rate of 5 m/s. The pressure of the water is approximately 800 PSI (5.5 MPa),
and the density of the water is approximately 770 kg/m3. The pipe suddenly expands so that the flow area increases by a
factor of 2. Neglecting the effects of friction, what is the pressure increase due to the change in the pipe’s area?
Exercise 16.7
A reactor core is 3.5 m high. Calculate the pressure head (the gravitational pressure difference) between the bottom and
the top of the core due to the elevation difference. Assume that the temperature of the water is 300°C and the pressure
of the water is 2,250 PSI (15.5 MPa).
Exercise 16.8
Water enters the core of a commercial PWR at 2,250 PSI (15.5 MPa). The water at the center of the core has an aver-
age temperature of 310°C. If the core is 5 m high, and the water flows through the core at 5 m/s, calculate the frictional
pressure drop. Assume that the turbulent friction factor is 0.012 and the equivalent diameter of the subchannels through
which the water flow is De = 1.18 × 10 −2 m.
Exercise 16.9
Water flows through a reactor coolant pipe at a rate of 5 m/s. The pressure of the water is approximately 800 PSI
(5.5 MPa), and the density of the water is approximately 770 kg/m3. The pipe makes an abrupt 45° turn but remains
horizontal. What is the pressure drop due to the directional change in the pipe?
Exercise 16.10
Nozzles and diffusers are frequently used in nuclear power plants to control the speed at which the coolant flows. Using
Bernoulli’s equation and the equation m = ρVA for the mass flow rate through a horizontal pipe, show that the pressure
change dP is related to the area change dA by
dA/A = dP/ρV2 (1 − (V/c)2)
where c is the speed of sound of the fluid flowing through the pipe. For a subsonic flow, how does the pressure change
when the area changes?
Exercise 16.11
As we mentioned previously, nozzles and diffusers have many different applications in nuclear power plants. Suppose
that air is forced into a diffuser at a velocity of 200 m/s. At room temperature, what are the speed of sound and the Mach
number at the diffuser inlet? Is the flow at the inlet sonic or subsonic?
Exercises for the Student 645
Exercise 16.12
When water is pumped at very high speeds through long circular coolant pipes in a nuclear power plant, the Reynolds
number can approach 850,000. When this occurs, the velocity of the water is about 5 m/s and the turbulent friction
factor is about 0.012. Suppose that the water then flows through a PWR fuel assembly where the Reynolds number is
the same. If the fuel assembly has a P/D ratio of 1.33, and the rods are smooth, what is the turbulent friction factor for
the assembly?
Exercise 16.13
Derive an expression for the equivalent diameter De of a reactor coolant channel in the center of a hexagonal fuel assem-
bly as a function of its P/D ratio. As the P/D ratio increases, does the equation for the equivalent diameter become more
similar to or less similar to that for a circular pipe?
17
Laminar and Turbulent
Flows with Friction
17.1 Reactor Coolants and Their Viscosities
Reactors are machines that rely on fluidic friction to operate properly. Any fluid that has no internal friction is called an inviscid fluid,
and a fluid that has at least a small amount of friction is called a viscous fluid. However, all reactor coolants have at least a small
amount of friction, and this implies that the viscosity μ of these coolants must be taken into account when determining the charac-
teristics of the flow field. Hence, we can distinguish between a viscous fluid and an inviscous one based on the amount of viscosity
it possesses.
where μ is a quantity called the dynamic or absolute viscosity. In general, the dynamic viscosity is used to characterize the “ stickiness”
of a fluid, or its internal resistance to continuous flow. The dynamic viscosity is measured in the units of poise (Pi) or centipoise,
where 1 centipoise = 1/100th of a poise. In the SI unit system, 1 poise is equal to 0.1 kg/ms, and 1 centipoise is equal to 0.001 kg/
ms. The dynamic viscosity of water at 20°C and atmospheric pressure is 1 centipoise, while in a pressurized water reactor (PWR) at
2,250 PSI and 340°C, it is about 0.0007 centipoise. Hence, it is less in a PWR than it is in a glass of drinking water at atmospheric
pressure. Another related unit for the viscosity is the kinematic viscosity ν. The kinematic viscosity is defined by ν = μ/ρ, where ρ is
the fluid density. Thus, the kinematic viscosity represents the ratio of the dynamic viscosity to the fluid density. In the SI unit system,
the kinematic viscosity has the units of m2/s. In fluid mechanics, the kinematic viscosity is also called the momentum diffusivity. It
appears in the definition of the Reynolds number, which we will discuss shortly. The Reynolds number can be thought of as the ratio
of the inertial forces of a fluid to its internal viscous forces. The Reynolds number is usually given the symbol Re.
F = µAU/Y (17.2)
647
648 Laminar and Turbulent Flows with Friction
TABLE 17.1
The Densities and Viscosities of Some Common Reactor Coolants
Note: For water, the physical properties are evaluated at 315°C and 15.5 MPa, which are representative of the conditions in the
primary loop, and also at 275°C and 7 MPa, which are representative of the conditions in the secondary loop. Note that the
density falls in a PWR as the pressure increases. The physical properties for liquid sodium are evaluated at 535°C and
0.1 MPa. In this table, the properties of carbon dioxide and helium are evaluated at 800°C and 20 MPa.
FIGURE 17.1 A parallel plate experiment used to measure the viscosity μ of a reactor coolant. In this experiment, the upper plate
moves with a constant velocity u in the x direction, while the lower plate is stationary. Because the velocity gradient is linear in the y
direction, the wall shear stress is given by τ = F/A = μdu/dy and the fluid is assumed to be Newtonian.
The force per unit area F/A on the surface of each plate is then called the wall shear stress τ, and it is given by the
following equation:
Finally, because U/Y is simply the velocity gradient of the fluid between the upper and lower plates, it is also possible
to write
FIGURE 17.2 Two pictures illustrating the concept of viscosity. In the picture on the right, the viscosity increases from left
to right, making it harder and harder for the fluid to flow. Conceptually speaking, the viscosity can be thought of as the internal
resistance of a fluid to continuous flow. The friction of a fluid is related to this internal molecular resistance.
where du/dy is called the local shear velocity. Finally, the dynamic viscosity is related to the shear stress τ on the surface
of a fuel rod by
where du/dy is the velocity gradient perpendicular to the surface of the rod. Two pictures that illustrate how the viscosity
affects the behavior of a fluid are shown in Figure 17.2. Fluids in which the shear stress is a linear function of the veloc-
ity gradient at the wall surface are said to be Newtonian fluids, and almost all reactor coolants can be considered to be
Newtonian when thought of in this way. Thus, the total frictional force that a moving fluid exerts on a fuel rod is given by
where A is the contact area between the fluid and the rod. For a cylindrical fuel rod of length L and diameter D,
this contact area is A = πDL. The relationship between dμ/dy and τ for some common reactor coolants is shown in
Figure 17.3. Notice that the viscosity of liquid sodium (a common coolant in fast reactors) is much less than that of water
at atmospheric pressure, but it is much greater than that of saturated water at 2,000 PSI.
FIGURE 17.3 Newtonian fluids are fluids that exhibit a linear relationship between the velocity gradient and the shear stress.
Almost all reactor coolants are Newtonian in nature. The rate of deformation of a Newtonian fluid (the velocity gradient) is
proportional to the shear stress, and the constant of proportionality is the viscosity μ.
FIGURE 17.4 When a flow field is laminar, the streamlines flow over each other in orderly layers and do not intermix.
The Reynolds number for these flows is low. This is true for both waterfalls and aircraft wings. Laminar flows can also occur
in reactor coolant channels when the coolant pumps are turned off. However, in most cases, the coolant flow in a reactor core is
highly turbulent. The picture on the right is a numerical simulation performed by NASA.
FIGURE 17.5 Several examples of turbulent flow in everyday life. The image on the upper left is an example of stratified
t urbulence, which occurs when a lighter fluid flows over a heavier one at very high speeds. The turbulent wake created by a car
driving down the road is shown on the upper right. The picture on the bottom shows the flow of blood through an artery or a vein.
The vortices and turbulent eddies are clearly visible in blood flowing in a turbulent flow field when the Reynolds number is high.
(Pictures provided by NASA and Wikipedia.)
652 Laminar and Turbulent Flows with Friction
Some textbooks state that the flow can become turbulent when the Reynolds number reaches 2,100. However, in prac-
tice, the most widely accepted value of the Reynolds number for an enclosed flow to become at least partially turbulent
is 2,300. Hence, this is the value that we will use in this chapter.
The inertial force of a fluid is related to its momentum ρv, and its internal friction is related to its dynamic viscosity μ.
Thus, the Reynolds number can be expressed in terms of the ratio
where μ is the dynamic viscosity, v is the velocity of the fluid, and ρ is its density. In this definition, L is the charac-
teristic length of the physical system for which the Reynolds number is to be found. For open flows, such as the flow
of fluids over flat plates and aircraft wings, the value of L is usually taken to be the length of the wing or the length of
the plate. However, for enclosed flows, such as those which occur in reactor coolant pipes and reactor fuel assemblies,
FIGURE 17.6 A picture of Osborne Reynolds, who is generally considered to be the father of the modern-day Reynolds number.
(Picture of Reynolds provided by Wikipedia.)
17.10 The Reynolds Number for Open Surfaces and Closed Surfaces 653
L is usually taken to be the effective diameter or the hydraulic diameter of the pipe or coolant channel through which
the fluid flows. Hence, in the case of a circular pipe with a diameter D, the appropriate expression to use for the
Reynolds number is
Normally, this definition is used for reactor fuel assemblies as well. In these assemblies correct value to use for D is
slightly different.
Re = vD/ν (17.11)
where ν is a related parameter, not to be confused with the velocity v, called the kinematic viscosity. The kinematic
viscosity is defined by ν = μ/ρ, and it represents the ratio of the dynamic viscosity μ to the fluid density ρ. Sometimes the
kinematic viscosity is also called the momentum diffusivity. For common reactor coolants such as water, the dynamic
viscosity and the kinematic viscosity fall as the temperature rises, provided that the pressure remains the same. This
effect is shown in Table 17.2, and it is also shown graphically for water in Figure 17.7.
TABLE 17.2
The Behavior of the Dynamic and Kinematic Viscosity of Ordinary Water as a Function of Temperature
for Water Up to Its Boiling Point at Atmospheric Pressure
FIGURE 17.8 The flow of air and water over the Earth’s surface as well as the flow of fluid around an impeller are examples of
open flows. These are different than the flows in reactor coolant channels, which are generally considered to be enclosed flows.
(Pictures provided by NASA and Baylor University.)
where D is the hydraulic diameter or equivalent diameter of the channel. Example Problem 17.1 illustrates what this
velocity is for water in a PWR fuel assembly. For a circular pipe, D is equal to the diameter of the pipe. Normally, the
critical velocity vCRITICAL becomes lower as the temperature of the coolant is raised. The critical velocity vCRITICAL can
be expressed in terms of the temperature-dependent kinematic viscosity ν as
where ν(T) = μ(T)/ρ(T). Other single-phase reactor coolants behave in a similar way.
FIGURE 17.9 Transitional flows are defined as flows with both laminar and turbulent components. In transitional flows, the
boundary layer stays intact close to the surface of the wall and turbulent eddies do not reach the wall surface. When the flow
becomes fully turbulent, the turbulent eddies pass closer to the surface of the wall and only a thin laminar sublayer remains.
TABLE 17.3
The Transition Points between Laminar, Turbulent, and Transitional Flows in Open and Closed Systems
Type of Flow Closed System or Enclosed Flow Open System or Surface Flow
Laminar Re < 2,300 Re < 1,000a
Transitional 2,300 < Re < 10,000 1,000 < Re < 2,300
Turbulent Re > 10,000 Re > 2,300
a Depends on the geometry.
656 Laminar and Turbulent Flows with Friction
FIGURE 17.10 A typical BWR fuel assembly (on the left) and a typical PWR fuel assembly (on the right). In a BWR core, the fuel
assemblies are smaller and are surrounded by metal sheaths or “cans.” In BWRs, the control rods are also cruciform in shape, which
is surrounded by four separate BWR fuel assemblies.
17.15 The Entrance Length and Entrance Effects 657
FIGURE 17.11 A PWR core containing a lot of different fuel assemblies arranged in the shape of a rough circular cylinder with
different enrichments and burnable poison distributions.
TABLE 17.4
Some Typical Coolant Velocities and Reynolds Numbers in PWR and LMFBR Coolant Channels during Start-Up (Low
Power Operation), Full Power Operation (Forced Convection), and after About a Week after Shutdown (Natural Circulation)
The fluid’s velocity profile will change as the boundary layer develops. The extent of this effect can be seen in Figure 17.12.
FIGURE 17.12 When a fluid flows over a solid surface, the fluid interacts with the surface and creates a boundary layer. In prac-
tice, it takes a while for the boundary layer to develop. At the entrance of a reactor coolant channel, the boundary layer is not fully
developed. This region is called the entrance region to the channel. Then if the channel becomes long enough, the flow becomes
fully developed and the shape of the boundary layer no longer changes. In reactor fuel assemblies, it generally takes about half a
meter for the boundary layer to become fully developed.
called its entrance length (Le), and the entrance length for a circular pipe of diameter D can be determined from one of
the following equations:
Thus, the entrance length Le can be thought of as a simple multiple of the pipe diameter D where the flow becomes fully
developed. For a Reynolds number of 1,000, Equation 17.14 shows that the entrance length for a laminar flow is approxi-
mately six times the diameter D of a circular pipe, and for the same pipe with a Reynolds number of 10,000, the entrance
length is about 70 times the diameter of the pipe because the flow is turbulent. Once the entrance length has been reached,
the radial velocity profile does not change again unless heat is added to the pipe. It can also change at the exit of the pipe,
and there the velocity profile begins to become destabilized as the flow area is changed. The flow in a reactor coolant
channel tends to exhibit the same behavior. However, the size of the entrance length can be estimated by replacing the pipe
diameter D by the hydraulic or equivalent diameter DH or DE. (In practice, the values of DH and DE are used to refer to the
same thing.) Normally, the entrance length will extend about half a meter beyond the entrance to the assembly. However,
the axial power profile will also effect exactly where this point occurs.
Figure 17.13 shows the regions in a typical reactor core where the flow can be considered to be fully developed or
undeveloped. Normally, the radial velocity profile is relatively uniform at very high Reynolds numbers (Re > 100,000),
and so phase changes are normally required to alter the profile if there are no grid spacers or area changes in the c hannel.
This is also true in straight pipes that do not contain coolant pumps or impellers. In water reactors, the velocity of an
average water molecule moving through the core is about 5 m/s. In other parts of the NSSS, this number can be lower.
FIGURE 17.13 The regions of developed and undeveloped flow in a water reactor fuel assembly approximately 5 m long. Note that
it takes about half a meter before the flow field becomes fully developed at the entrance to the fuel assembly, and it takes about one-
fifth of a meter after the fluid passes the grid spacers before the flow field becomes fully developed again.
where the subscript 0 refers to the larger pipe and the subscripts 1 and 2 refer to the smaller pipes. Since v1 = v2 and
A1 = A2, A0v0 = 2A1v1. Finally, since A1 = A0/4, v1 = v2 = 2v0. Because the diameters are half as large and the velocities
are twice as high, the Reynolds numbers in the smaller pipes must be the same as the Reynolds number at the entrance
to the larger pipe. Thus the Reynolds number in the smaller pipes is 1000! As we mentioned earlier, this will continue
to be true for any flow that is adiabatic and incompressible. [Ans.]
Normally, the viscous sublayer is only a few molecules thick. However, the behavior of the flow in this layer is
very important when it comes to determining the rate at which heat can be transferred from the wall to the coolant.
The same effect occurs in reactor coolant channels that are not circular in shape. However, in this case, the turbu-
lent core extends over most of the coolant channel, and the fluid velocity only becomes low near the wall surface.
Hence, the friction factors exhibit a much lower sensitivity to the shape of the channel when the flow is turbulent.
Figure 17.14 shows the radial velocity profiles in a circular channel at Reynolds numbers of 1,000, 10,000, and
50,000. Notice that as the Reynolds number increases, the velocity profile in the center becomes flatter and flatter.
The same effect also occurs in reactor coolant channels. Notice that in the center of the channel, the radial velocity
FIGURE 17.14 The radial velocity profiles in a reactor coolant channel become flatter in the center and steeper at the wall as the
Reynolds number increases. When the flow field is laminar, the radial velocity profile is roughly parabolic in shape, but when the
flow becomes turbulent, it becomes more rectangular in shape. However, the velocity of the fluid is always zero at the surface of a
fuel rod because of the no-slip boundary condition.
FIGURE 17.15 The wall shear stresses and therefore the velocity gradients are much greater for turbulent flows than they are for
laminar flows even though the turbulent boundary layer may be thicker than the laminar boundary layer for the same value of the
free stream velocity u∞.
17.17 Estimating the Wall Shear Stress 661
profile is nearly flat as long as the Reynolds number is high enough. Of course, if there are grid spacers in the
coolant channel, they will temporarily disrupt the radial velocity profile before it returns to its previous shape. The
velocity gradients in a laminar and turbulent flow field are compared near the surface of the wall in Figure 17.15.
Example Problem 17.3
Suppose that water is flowing through a straight pipe in a reactor piping system and the Reynolds number in the pipe is
approximately 10,000. If the pipe has a diameter of 0.10 m, approximately how far from the entrance to the pipe will the
flow become fully developed?
Solution The flow in this case is clearly turbulent, so according to Equation 17.14, it will become fully developed when
Le = 0.073Re·D. Plugging in the appropriate numbers, we see that this occurs when Le = 0.073 × 10,000 × 0.10 = 73 m.
Since the pipe is probably as long as the containment building is high, the chances are good that the flow will never
become fully developed in this pipe. [Ans.]
where du/dy is the velocity gradient perpendicular to the rod surface. Thus, when the flow is laminar, the total frictional
force the fluid exerts on the rod is
where A is the contact area between the fluid and the rod. However, when the flow becomes turbulent, the radial velocity
profile is no longer independent of time, and it now consists of an average value u(r) which is time invariant plus a
fluctuating component u′ that is time dependent:
The total shear stress on the fuel rod then consists of a laminar part, which is given by
τ LAMINAR = µd u dy (17.18)
In other words, the total shear stress on a fuel rod when the flow becomes turbulent is
or
(17.20)
τ TOTAL = µd u dy + µdu′/dy
(
= µ d u dy + du′/dy )
* Most reactor coolants such as water, helium, carbon dioxide, and liquid sodium can be considered to be Newtonian as long as they
do not boil. In other words, for these coolants, the frictional forces exerted on a fuel rod are proportional to the velocity gradient at
the surface of the rod.
662 Laminar and Turbulent Flows with Friction
This final equation implies that the frictional pressure drop must always be greater for a turbulent flow than it is for a
laminar flow in the same coolant channel. Moreover, the shear force that a turbulent flow exerts on a nuclear fuel rod
will always be greater than the shear force that a laminar flow exerts on the same rod.
which is called the Darcy formula. The Darcy formula can be used to find the pressure drop between two points in a
reactor piping system separated by a distance L. It can also be used to find the pressure drop in a reactor fuel assembly
if the equivalent diameter De is properly defined. The terms in the Darcy formula are
☉☉ ΔP is the frictional pressure drop (in Pa or PSI).
☉☉ f is the friction factor (always dimensionless).
☉☉ ρ is the density of the fluid (in kg/m3 or g/cm3).
☉☉ V is the average velocity of the fluid flowing through the pipe.
☉☉ L is the length of the pipe or coolant channel where the pressure drop is to be found (in cm or m).
☉☉ DE is the diameter of the pipe (in cm or m).
When using the Darcy formula, a reactor coolant channel does not have to be circular, but to find the correct value for
ΔP, an appropriate value for the equivalent diameter DE must be used. As we just mentioned, the equivalent diameter
can be found from
where A is the cross-sectional flow area (in cm2 or m2), and PW is the wetted perimeter (in cm or m). The equivalent
diameter is measured in centimeters or meters. The constant of proportionality between the right- and left-hand sides of
Equation 17.22 is an important parameter called the Darcy–Weisbach friction factor f. For laminar flows (Re < 2,300), the
value of f can be measured experimentally or it can be derived from the Navier–Stokes equations. In the next s ection, we
would like to show how this can be done. For turbulent flows, the derivation becomes more complicated, and so one nor-
mally requires an experiment to find the correct value for f. In this case, the value for f also depends on the Reynolds number:
However, for laminar flows, the derivation is simple enough that we can present it in about a page. Not surprisingly, this
is what we would like to discuss next.
FIGURE 17.16 The friction factors are different for laminar and turbulent flows because the velocity profiles close to the wall are
different. In turbulent flow, the velocity profile in the center of a reactor coolant channel is generally flatter and more uniform than it
is for laminar flow. However, the velocity gradient is steeper closer to the surface of the wall.
the velocity field, which is called u′ in Equation 17.17, can be set to zero (u′ = 0). From our previous discussion, only the
laminar component of the velocity field, u(r) , then remains. Therefore, Equation 17.17 reduces to
The steady-state momentum equation that predicts the pressure drop in a circular pipe (see Chapter 15) is then
To find the pressure drop or the pressure gradient under these conditions, it is easiest to align the pipe horizontally (in
the x direction) so that the fluid velocity v(r) is a function of radial position r only. The value of r is 0 at the centerline
of the pipe, and the value of r is R at the pipe wall. The general layout of the problem in which we would like to find the
pressure gradients is shown in Figure 17.17. If we integrate Equation 17.26 twice, and apply the appropriate boundary
conditions (v = 0 at r = R and dv/dr = 0 at r = 0), we obtain
( )
v(r) = − (R 2 4µ) ⋅ dP/dx 1 – (r/R)2 (17.27)
In other words, when the flow becomes laminar and fully developed, the radial velocity profile becomes parabolic in
shape. The fluid velocity peaks at the center of the pipe, and it then falls to 0 at the wall surface (where r = R). The shape
of the velocity profile can be seen in Figure 17.17. Now suppose that we would like to use Equation 17.27 to derive the
aforementioned Darcy formula. We can do this by simply solving for the average velocity of the fluid flowing through
the pipe V = VAVG. We can do this by integrating the mass-weighted velocity profile from the center of the pipe to the
edge, and then dividing by the cross-sectional area (πR2).
V=
∫ ρ2πrv(r) dr ∫ ρ2πr dr (17.28)
Here, the integration is to be performed between the center of the channel (where r = 0) and the edge of the channel
(where r = R).
664 Laminar and Turbulent Flows with Friction
FIGURE 17.17 The fully developed radial velocity profile for laminar flow is parabolic in shape in a round pipe. Here, the coolant
velocity peaks at the center of the pipe and is zero when the fluid touches the wall.
When we integrate Equation 17.28, the denominator becomes πR2, and we discover that
V = R 2 /8µ(− dP/dx )
(17.29)
or dP/dx = −8µV/R 2
At this point, the reader may wonder why we went through all this work. However, if we can manipulate the expression
for dP/dx a little further, we can derive the Darcy equation from it. Suppose that we multiply both the numerator and the
denominator by ½ρV2 (the kinetic energy of the fluid in the pipe). The result is
The Pressure Gradient and the Friction Factor for Laminar Flow in a Circular Pipe
dP/dx = − f/D ⋅ 1 2 ρV 2
where (17.31)
f = 64/(ρVD/µ) = 64/Re
So in about a paragraph, we have just derived the famous Darcy equation for the pressure drop in a round, straight
pipe:
f = 64/Re (17.33)
and f is appropriately called the Darcy–Weisbach friction factor. Notice that “the definition” of the friction factor sim-
ply “fell out” of the momentum equation when we integrated it twice! So this is the definition of the friction factor that
17.20 Deriving the Darcy Equation for Turbulent Flow 665
should be used for fully developed laminar flows. Moreover, it turns out that we can use exactly the same expression for
fully developed turbulent flows. All we need to do is to develop a different expression for the friction factor when the
flow is turbulent. Not surprisingly, this is the next subject we would like to discuss.
where δ is the thickness of the boundary layer and r has been replaced by y. Similar expressions exist for the turbulent
velocity profile in round, straight pipes. Notice that u/U ≅ 1 when y > δ. In other words, beyond the boundary layer, the
velocity profile becomes uniform. Figure 17.18 shows the shape of the turbulent boundary layer in this case. The black
wavy lines are the instantaneous velocity profiles, and the thick blue line is the time-averaged value of these profiles.
Figure 17.19 also compares the laminar and turbulent velocity profiles normalized to the boundary layer thickness δ.
The free stream velocity beyond the boundary layer is represented by the symbol U∞. The turbulent friction factor f then
becomes a more complicated function of the Reynolds number Re. Once the time-averaged velocity profile is known, the
same procedure we used in Section 17.19 can be used to calculate the value of the friction factor for the turbulent layer.
For circular pipes, this calculation results in an expression of the form
f = A/Re b (17.34b)
FIGURE 17.18 The instantaneous behavior of the turbulent boundary layer as well as its time-averaged value. The black wavy
lines represent the instantaneous velocity profiles, while the thick blue line represents the time-averaged velocity profile. As one
approaches the edge of the boundary layer where y → δ, the velocity profile becomes flat and the local velocity gradient approaches
zero. Here, the fluid velocity becomes equal to the free stream velocity U∞.
666 Laminar and Turbulent Flows with Friction
FIGURE 17.19 The shape of the laminar and turbulent velocity profiles above a long flat plate. The velocity profiles are normalized
to the boundary layer thickness δ. Notice that the turbulent velocity gradient close to the wall is much steeper than the laminar
velocity gradient.
where A is an empirical constant and b is an exponent having a value between about 0.20 and 0.25. Later, we will learn
that several popular correlations for the turbulent friction factor take advantage of this fact. They include the Blasius
correlation and the McAdams correlation.
δ( x ) = 4.91x ( Re x )
0.5
(for laminar flow over a flat plate) (17.35a)
δ( x ) = 0.16x ( Re x )
1/ 7
(for turbulent flow over a flat plate) (17.35b)
Equation 17.35b can be deduced from the one seventh power law. One of the interesting things about these equations
is that they show the thickness of the turbulent boundary layer is generally greater than the thickness of the laminar
boundary layer (by about a factor of 4–5 for the flow of water over a smooth, flat plate). Thus, although the velocity
gradient is greater within the boundary layer when the flow is turbulent, the boundary layer is thicker at the same
location for turbulent flow than it is for laminar flow. A similar set of equations can be derived for flow in circular pipes.
When the flow is turbulent, the thickness of the boundary layer is given by
δ( x ) = 0.38x ( Re x )
0.20
(for turbulent flow through a circular pipe) (17.35c)
Thus, the thickness of the boundary layer (and the shape of the velocity gradients within the layer) determine the values of the
friction factors. We will have more to say about how the heat transfer coefficients behave under these conditions in Chapter 20.
17.23 APPLYING THE DARCY EQUATION TO COOLANT CHANNELS 667
This is an exact expression for f as long as Re < 2,300, and the pipe is circular. In fact, if we decide to interpret the value
of D as the equivalent diameter DE (in cm or m), then we can use the Darcy formula to find the pressure drop in a reactor
fuel assembly as well. We simply need to find De = 4A/PW, where A is the cross-sectional flow area (in cm2 or m2) and
PW is the wetted perimeter (in cm or m) of the region through which the fluid is flowing. This value of DE must also be
used in the expression for the Reynolds number Re:
Re = ρvD e /µ (17.37)
Thus, we can use the Darcy formula to predict the pressure drop in any kind of pipe (square, triangular, hexagonal, oval,
etc.) when the correct value of the equivalent diameter is used. To illustrate where this argument leads, suppose that we
would like to determine the friction factor for fully developed laminar flows in pipes and coolant channels having dif-
ferent shapes than circular ones. When the appropriate laminar velocity profiles are inserted into the momentum equa-
tions, one arrives at the laminar friction factors shown in Table 17.5. These friction factors are the correct factors to use
for fully developed laminar flows. Notice that the friction factors for rectangular and elliptical channels are generally
larger than they are for circular channels, and that the friction factors are smaller for triangular coolant channels
having the same Reynolds number. This is due entirely to differences in the shape of the velocity profile as we move
from the center of the channels to the edge. Moreover, the friction factor is also dependent on the aspect ratio (a/b) or on
the angle θ of an isosceles triangle for a triangular duct. In general, the laminar friction factors become larger as the
aspect ratio is increased. This implies that the further we depart from a circular geometry, the less likely it is we can use
the Darcy equation without applying some sort of geometrical correction factor. Again, the Reynolds numbers must be
found using the following definition for the hydraulic diameter DH, which is also shown in Equation 17.27:
D H = 4A/PW (17.38)
where A is the cross-sectional flow area of the flow channel (in cm 2), and PW is its wetted perimeter (in cm). Finally, the
reader should recall that the Darcy–Weisbach friction factor f should NOT be confused with the Fanning friction factor
f′ = f/4, which also appears in the literature from time to time. In this book, we will exclusively use the Darcy–Weisbach
friction factor when attempting to calculate the pressure drop. To avoid any possible confusion, the Fanning friction
factor, which is equal to the Darcy friction factor divided by 4, will never be used. Assuming the correct value for the
friction factor is available, then it becomes possible to apply the Darcy equation to composite pipes and composite flow
channels having several different diameters and pitch-to-diameter (P/D) ratios. For these channels, the geometry of one
section of the channel may be different than that of another.
17.23 A
pplying the Darcy Equation to Coolant Channels
with Different Cross-Sectional Areas
The pressure drop ΔP along a channel then becomes the algebraic sum of a number of different pressure drops if the
channel changes its size or its shape. If a pipe is made up of a number of different sections with different hydraulic
668 Laminar and Turbulent Flows with Friction
TABLE 17.5
Shape-Dependent Friction Factors for Laminar Flow
diameters, then the Reynolds number in each section will be different, and the friction factors will be different as well.
Hence, the value of ΔP as we move along the pipe must be the algebraic sum of the individual pressure drops:
∆P = ∑ n
∆Pn = ∆P1 + ∆P2 + ∆P3 + ∆Pn (n = 1,2,… , N − 1, N) (17.39)
Notice that there can be N individual sections in even a single pipe. The continuity equation ρvA = constant then allows
us to predict the average velocities v1 + v 2 + v 3 + v N in each of these sections. For a pipe with N individual sections,
the equation we must use to find the pressure drop is then
( ) (
∆PFRICTION = f1 L1 D e1 ρ1 v1 2 2 + f2 L 2 D e2 ρ2 v 2 2 2 )
(17.40)
( ) ( )
+ f3 L 3 D e3 ρ3 v 3 2 2 + + fN L N D eN ρN v N 2 2
where
f1 = 64/Re1
f2 = 64/Re 2
f3 = 64/Re3
⋅
fN = 64/Re N
Re n = ρn v n D e n µ n (17.41)
and
( )
∆Pn = fn L n D e n ρn v n 2 2 (17.42)
17.24 Turbulent Friction Factors 669
Thus, the Reynolds number changes as the hydraulic diameter changes. Later we will present an example of how
Equation 17.40 can be used to find the frictional pressure drop in a three-section horizontal pipe. In this problem, each
of the sections of the pipe has the same length (10 m), but the diameter becomes progressively smaller as one goes from
Section 1 to Section 3. Entrance and acceleration effects are ignored to keep the analysis simple. These effects will be
discussed in the next chapter where we will also add gravity to the momentum equations. Adding the effects of gravity
and acceleration allows us to consider vertical pipes and coolant channels with heat addition as well as horizontal iso-
thermal ones. Note in particular how the Reynolds number changes as we move along the pipe. The friction factors in
each section vary as well, and the value of ΔP changes axially as a result of this. However, the flow field never becomes
turbulent so the laminar friction factors continue to apply to each section. In the real world, care must be taken whenever
the flow regime changes because a different correlation for the friction factor may have to be applied to each region. In
the sections that follow, we would like to discuss the correlations that are appropriate to use for turbulent flow.
FIGURE 17.20 The Moody chart or Moody diagram is used to find the friction factors in reactor fuel assemblies for both laminar
and turbulent flows. It is a graph in nondimensional form that relates the Darcy–Weisbach friction factor, Reynolds number, and rela-
tive roughness for fully developed flow in a circular pipe. It can be used for working out pressure drop or flow rate down such a pipe.
670 Laminar and Turbulent Flows with Friction
a range of experimental conditions where the temperature differences between the coolant TCOOL and the wall T WALL
are very small. Typically, these temperature differences, which are also a function of axial position, can be represented
by ΔT(x) = TCOOL(x) − T WALL(x). In general, Moody charts give very good turbulent friction factors when the flow is
isothermal or nearly so. However, Moody’s original work did not account for changes in the viscosity of the fluid with
temperature, and so a Moody chart should be used with caution when temperature differences between the pipe wall and
the surrounding fluid become significant. In their original form, the Moody charts required an engineer to determine
the value of the single-phase friction factor directly from the chart. Sometimes this could become cumbersome, and so
the data was subsequently “fit” to a number of empirical correlations that plotted the friction factor as a function of the
Reynolds number Re. In the next section, we would like to discuss the form of some of these correlations.
TABLE 17.6
A Table Showing Some Common Single-Phase Friction Factor Correlations for Both Laminar and Turbulent Flows
Recommended Friction Factor Correlations for Single-Phase Flow (Laminar and Turbulent)
Flow Regime Recommended Correlation Expression for the Friction Factor (f) Range of Reynolds Numbers
Laminar Darcy equation f = 64/Re 0 < Re < 2,300
Turbulent Blasius correlation f = 0.316Re –0.25 4,000 < Re < 100,000
Turbulent McAdams correlation f = 0.184Re–0.20 30,000 < Re < 1,000,000
Turbulent Petukhov correlation f = 1.0/(1.82 log Re–1.64)2 3,000 < Re < 5,000,000
17.26 Some Observations about the Moody Charts 671
Each of these correlations can be used to predict the friction factor for a smooth pipe (i.e., one having very smooth sur-
faces). However, the friction factors in the Moody chart also depend on the surface roughness ε. In Chapter 18 we will
present several correlations that attempt to simulate the effect of surface roughness on the value of f. However, as soon
as one begins to deviate from a round, straight pipe, additional correction factors must be applied to obtain the correct
value for f. In general, f becomes a function of ε, Re, and the geometry G:
f = f(ε,RE,G) (17.46)
In reactor fuel assemblies, this geometrical dependence G is represented by the P/D ratio. Notice that the friction factor f is
lowest for very smooth surfaces, and it increases as the roughness of the surface increases. However, this relationship is by
no means a linear one. Example 17.6 compares the values of f for a smooth circular pipe with a Reynolds number of 100,000.
Notice that the value of f differs in this case by about 5%. At other Reynolds numbers, the standard deviation is different.
☉☉ Finally, the Moody charts were originally developed for circular pipes. As soon as we try to apply them (or
the Blasius, McAdams, and Petukhov correlations) to reactor coolant channels, the friction factors they predict
are different than the friction factors we measure; hence, a geometric correction factor must be applied to obtain
the correct results. In some cases, these correction factors can be as large as 10% or 20%. In other words, the tur-
bulent friction factor in a reactor coolant channel can be either higher or lower than what the Moody charts suggest.
17.27 T he Pressure Drop and the Reynolds Number for Circular Pipes
Based on the data presented in the Moody charts, it is tempting to conclude that the pressure drop ΔP decreases as
the Reynolds number increases. However, this never occurs because the product of f and v2 determines the size of ΔP,
and no matter what correlation we use for f, the value of v2 always increases faster than the friction factor decreases.
To understand this dependency, suppose that we write the Darcy equation as a function of the Reynolds number alone.
Since Re = ρvD/μ, we can write
1
2 ( )
ρv 2 = µ 2 2ρD 2 Re 2 (17.47)
where c is a simple constant that depends on only the values of D, μ, L, and ρ. The Blasius and McAdams correlations
tell us that f is proportional to Re−0.2 or Re−0.25. Thus, the total pressure drop is proportional to the Reynolds number to
the 1.8 or 1.75 power when the flow becomes turbulent:
This equation shows that if we double the Reynolds number, the core pressure drop increases by a factor of 3.36 to 3.48,
and if we triple the Reynolds number, the core pressure drop increases by a factor of 6.84 to 7.22. These numbers apply
to any coolant channels where the surface of the wall is smooth. However, when the flow becomes laminar, we learned
earlier that f becomes proportional to 1/Re, and then the relationship between the frictional pressure drop and the
Reynolds number is linear:
Hence, for laminar flows, the pressure drop is a linear function of the Reynolds number. Finally, we can conclude from
Equations 17.49 and 17.50 that the pressure drop always increases as the Reynolds number increases. This behavior is
shown in Figure 17.21. Notice that the rate of increase is less for laminar flows than it is for turbulent ones.
FIGURE 17.21 For single-phase laminar flows, the pressure drop per unit length, or axial pressure gradient dP/dz, is proportional
to the Reynolds number to the first power, and for turbulent flows, it is approximately proportional to the Reynolds number to
the second power. Thus within a given flow regime, the pressure gradient always increases with the coolant velocity and the
Reynolds number.
becomes rough. In practice, most reactor fuel assemblies operate at Reynolds numbers between 100,000 and 500,000.
However, these Reynolds numbers are based on a hydraulic diameter DH that is defined by
where A is the cross-sectional flow area of the flow channel (in cm 2), and PW is its wetted perimeter (in cm). Sometimes
the hydraulic diameter is also called the equivalent diameter DE. In the transition region between Re = 2,300 and
Re = 10,000, the friction factor can be determined graphically. It can also be found by averaging the results of the Darcy
equation with one of the other correlations we have also discussed.
D H = 4A/PW (17.52)
However, when we apply this definition to a real channel, we must replace the values of A and PW with expressions that
are a function of the P/D ratio of the lattice (where P is the pitch of the fuel rods and D is their diameter). For example, in
674 Laminar and Turbulent Flows with Friction
a square lattice where the diameter of the fuel rods is D and the pitch is P, the flow area is A = P2 − πD2/4 and the wetted
perimeter is PW = πD. This results in the following expression for the hydraulic diameter:
For a hexagonal lattice, the corresponding expression for the hydraulic diameter becomes
Light water reactor fuel assemblies tend to have P/D ratios close to 1.33. In fast reactors, the fuel assemblies have P/D
ratios closer to 1.15. In PWRs and BWRs, the fuel assemblies are square, and in LMFBRs, they are usually hexagonal.
The actual values for PWRs, BWRs, and LMFBRs are provided in Table 17.7, and pictures of square and hexagonal
fuel assemblies are provided in Figures 17.22 and 17.23. Usually, nuclear fuel rods are very smooth, and if we know the
P/D ratio of the lattice, we can find the values of the hydraulic diameters directly from the diameters of the fuel rods.
No other information is required to find the hydraulic diameters in this case.
TABLE 17.7
The Hydraulic Diameters, Flow Areas, and Wetted Perimeters for Reactor Coolant Channels with
Representative P/D Ratios
FIGURE 17.22 Two of the most popular reactor lattices used today: a square lattice on the left and a hexagonal lattice on the right.
The square lattice is used in PWRs and BWRs, and the hexagonal lattice is used in LMFBRs.
FIGURE 17.23 The center, edge, and corner subchannels in light water reactor fuel assemblies (on the left) and in LMFBR fuel
assemblies (on the right).
Re 2 = v 2 D 2 ν (17.58)
where ν = μ/ρ is the kinematic viscosity. We also know from the continuity equation that
ρv1 A1 = ρv 2 A 2 or v1 A1 = v 2 A 2 (17.59)
because the fluid is assumed to be incompressible. If the flow channel is circular in shape, then A1 = πD12/4 and
A 2 = πD22/4. It then follows from the Equations 17.57 to 17.59 that
( )( ) ( )( ) (
Re 2 Re1 = v 2 D 2 v1D1 = D 2 D1 v 2 v1 = D 2 D1 A1 A1 = D1 D 2 (17.60) )
676 Laminar and Turbulent Flows with Friction
FIGURE 17.24 As a coolant channel contracts, the coolant velocity increases faster than the hydraulic diameter decreases. Hence,
the Reynolds number becomes larger as the coolant channel contracts. This can cause the flow in a constricting coolant channel to
change from laminar to transitional to turbulent over a very short distance. The Reynolds number at which the flow becomes fully
turbulent is called the critical Reynolds number.
or
( )
Re 2 Re1 = D1 D 2 (17.61)
( )
Re 2 = D1 D 2 Re1 (17.62)
and the size of the Reynolds number at point 2 will either increase or decrease depending on the value of the ratio
(D1/D2). The size of the Reynolds number at point 2 then depends on the ratio of D2 to D1. This relationship can be used
for any incompressible fluid as long as the density does not materially change between points 1 and 2. However, if the
kinematic viscosity ν becomes temperature dependent, then we must also include the effect of this temperature change
on the values of ν1 and ν2.
PW = πD = 0.0298 m
D H = 4A/PW = 0.01178 m
Re = ρvD H µ (17.63)
the value of the Reynolds number for an individual subchannel is Re = 5 × 105 ≈ 550,000. Clearly, this indicates how
turbulent the flow field in a coolant channel can be.
PW = πD/2 = 13.6 mm
D H = 4A/PW = 3.96 mm
678 Laminar and Turbulent Flows with Friction
Now since the Reynolds number for a single subchannel is defined as
Re = ρvD H /µ (17.64)
the value of the Reynolds number for an individual subchannel is Re = (817.7 × 6 × 0.00396)/0.000228 ≈ 85,250. This is about
six times lower than the Reynolds number in a PWR fuel assembly. However, the heat transfer coefficient is about three times
higher because the thermal conductivity of liquid sodium is so high. (We will have more to say about this in Chapter 20).
17.33 Finding the Reynolds Number for the Pressure Tubes in a PWR Steam Generator
Finally, suppose that we would like to calculate the Reynolds number in one of the pressure tubes of a PWR steam
generator. Most steam generators used in PWRs today are U-tube steam generators where the tube bank is intentionally
bent in to the form of an inverted “U.” A cross-sectional view of one of these steam generators is shown in Figure 17.25.
A large PWR can have between 2 and 4 of these steam generators. The primary side of each steam generator in a four-
loop Westinghouse PWR has a flow rate of about 5,000 kg/s. (The actual flow rate can vary depending on the design.)
The steam generator tubes circulate hot pressurized water from the core. The coolant flows through approximately 4,000
tubes inside of the steam generator with an average diameter of 0.0222 m (about seven-eighths of an inch). The Reynolds
number inside each of these tubes is
Re = ρvD H /µ (17.65)
where DH is the hydraulic diameter inside of the tube. Because each of these tubes is circular in shape, the hydraulic
diameter is the same as the internal diameter D. At 315°C and 15.5 MPa, the density of the water entering the steam gen-
erator is ρ = 710.5 kg/m3 and the dynamic viscosity is μ = 8.56 × 10 −5 kg/ms. The velocity of the coolant flowing through
a typical pressure tube is about 4 m/s. The Reynolds number during full power operation is then
This implies that the flow through the tubes is highly turbulent but not as turbulent as the flow through the core
(where the Reynolds number can be up to six times higher).
FIGURE 17.25 A rendering of a U-tube steam generator (left), and the exposed tube bank from a U-tube steam generator (right).
(Pictures provided by Westinghouse Nuclear.)
17.34 Corrections to the Laminar Friction Factors for a Reactor Fuel Assembly 679
17.34 Corrections to the Laminar Friction Factors for a Reactor Fuel Assembly
Next, we would like to discuss the pressure drop through a PWR fuel assembly. To do so, we must first calculate the
pressure gradient dP/dx from the Darcy equation, and then multiply it by the length of the coolant channel L:
If the pressure gradient varies in the axial direction, then we must integrate this gradient between the bottom and the top
of the core to find the total pressure drop, which is given by the following equation:
∆PCORE =
∫ (dP/ dx) dx (17.67)
L
where L is the core length (also called the core height). To do this, we must know the correct friction factor to use. Clearly,
the laminar friction factor in a reactor fuel assembly depends on the diameter D of the fuel rods and on the pitch P of the
lattice. In the early 2000s, Todreas and Cheng developed a comprehensive model to account for these effects by replac-
ing the value of 64 in the Darcy–Weisbach friction factor with an empirical constant called C. The new laminar friction
factor is then given by
fLAMINAR = C/Re (17.68)
where C depends on the type of subchannel (corner, edge, or center). For a central subchannel, the appropriate expres-
sion to use for C is
C = a + bx + cx 2 (17.69)
where x = P/D − 1, and P is the pitch of the assembly. For the central subchannels, the values for a, b, and c are
Hexagonal Lattice
(For Laminar Flow)
1.0 ≤ Pitch/D ≤ 1.1
a = 26.00
b = 888.2
c = −333.4
(17.70)
1.1 ≤ Pitch/D ≤ 1.5
a = 62.97
b = 217.9
c = −190.2
Square Lattice
(For Laminar Flow)
1.0 ≤ Pitch/D ≤ 1.1
a = 26.37
b = 374.2
c = −493.9 (17.71)
1.1 ≤ Pitch/D ≤ 1.5
a = 35.55
b = 263.7
c = −190.2
680 Laminar and Turbulent Flows with Friction
TABLE 17.8
The Laminar Friction Factors for Fully Developed Flows in Reactor Fuel Assemblies with Different P/D Ratios
P/D Ratio Laminar Friction Factor (Square Lattice) Laminar Friction Factor (Hexagonal Lattice)
1.1 fLAMINAR = 59/Re fLAMINAR = 82/Re
1.2 fLAMINAR = 80/Re fLAMINAR = 99/Re
1.3 fLAMINAR = 98/Re fLAMINAR = 111/Re
1.4 fLAMINAR = 110/Re fLAMINAR = 119/Re
1.5 fLAMINAR = 120/Re fLAMINAR = 124/Re
These results are surprisingly accurate if one is willing to ignore the effects of the corner and edge subchannels. Next,
we would like to examine the value of C that appears in the expression for the laminar friction factor. Suppose that we
calculate this coefficient for a P/D ratio of 1.33. The laminar friction factor we can use for a fuel PWR assembly with
this P/D ratio is then
and the corresponding value for a hexagonal lattice with the same P/D ratio is
Table 17.8 presents the laminar friction factors for fully developed flows in square and hexagonal fuel assemblies with
P/D ratios between 1.1 and 1.5. Notice that the laminar friction factors tend to increase as the P/D ratio increases.
Moreover, they are consistently larger than the values one would expect for smooth circular pipes.
C = a + bx + cx 2 (17.75)
Hexagonal Lattice
(For Turbulent Flow)
1.0 ≤ Pitch/D ≤ 1.1
a = 0.09378
b = 1.398
c = −8.664
(17.76)
1.1 ≤ Pitch/D ≤ 1.5
a = 0.1458
b = 0.03632
c = −0.03333
17.36 The Rehme Correction for the Turbulent Friction Factor 681
TABLE 17.9
The Turbulent Friction Factors for Fully Developed Flows in Reactor Fuel Assemblies with Different P/D Ratios
P/D Ratio Turbulent Friction Factor (Square Lattice) Turbulent Friction Factor (Hexagonal Lattice)
1.1 fTURBULENT = 0.140/Re 0.18 fTURBULENT = 0.147/Re0.18
1.2 fTURBULENT = 0.148/Re 0.18 fTURBULENT = 0.152/Re0.18
1.3 fTURBULENT = 0.152/Re0.18 fTURBULENT = 0.154/Re0.18
1.4 fTURBULENT = 0.154/Re0.18 fTURBULENT = 0.155/Re0.18
1.5 fTURBULENT = 0.155/Re 0.18 fTURBULENT = 0.156/Re0.18
Square Lattice
(For Turbulent Flow)
1.0 ≤ Pitch/D ≤ 1.1
a = 0.09423
b = 0.5806
c = −1.239
(17.77)
1.1 ≤ Pitch/D ≤ 1.5
a = 0.1339
b = 0.09059
c = −0.09926
Now consider the values of C for a number of different P/D ratios. The friction factors for fully developed turbulent
flows are shown in Table 17.9. Again, it is worth mentioning that the turbulent friction factors tend to increase as the
P/D ratio increases. Moreover, they are consistently larger than the values one would expect for a smooth circular pipe.
Of course, the reader must keep in mind that these results only apply to the interior channels of a reactor fuel assembly
and that the fluid they are designed to be used with is turbulent single-phase water. To find the values of a, b, and c for
the corner channels and the edge channels, the reader should consult Todreas and Kazimi (2008). These correlations are
much more appropriate to use than the ones for circular pipes.
where f TURBULENT is the turbulent circular tube friction factor, and again, R = (P/D) − 1. The Rehme correlation gives a
different value for the turbulent friction factor than the Todreas and Cheng correlation does, but it is still widely used in
the nuclear business as a whole. The Rehme correlation was used in the design of some of the world’s first fast reactors.
Solution A PWR has a square lattice, and we know from the previous example that DH = 1.25 cm = 0.0125 m. The values
of the density and the viscosity of water at this temperature and pressure are ρ = 713.8 kg/m3 and μ = 0.000086 kg/ms. If
the water is moving at 5 m/s, the Reynolds number is Re = ρvA/μ = 713.8 kg/m3 × 5 m/s × 0.0125 m/0.000086 kg/ms =
517,750. Thus, the flow field is highly turbulent, and we should use the Todreas and Cheng correlation (Equation 17.63)
to find the turbulent friction factor. The value of C in this case is 0.1530, so the turbulent friction factor must be given by
f TURBULENT = 0.1530/Re0.17. If Re = 517,750, the value of the turbulent friction factor is f TURBULENT = 0.0143. [Ans.]
17.38 Comparing the Friction Factors for a Square Fuel Assembly and a Circular Pipe
Next, we would like to compare the friction factors for a PWR coolant channel to those for a smooth circular pipe. The
P/D ratio for most PWR fuel assemblies is about 1.33. According to Todreas and Cheng, the values of a, b, and c for
laminar flow are 35.55, 263.7, and −190.2, respectively, and those for turbulent flow are 0.1339, 0.09059, and −0.09926.
For this P/D ratio, the values of C to use for laminar and turbulent flows are then:
FIGURE 17.26 The behavior of the turbulent friction factors in a reactor fuel assembly compared to the behavior of those in a
round pipe. In fuel assemblies, the frictions are a function of the P/D ratio.
17.40 Finding the Pressure Drop in the Primary Side of a PWR Steam Generator 683
Notice that the laminar friction factor is about 60% higher than the Darcy equation predicts for a smooth circular
pipe, and for a Reynolds number of 500,000, the turbulent friction factor is 0.0143, which is about 7% higher than the
value given by the McAdams correlation, where f MCADAMS = 0.0133. Hence, both the laminar and turbulent friction
factors in PWR fuel assemblies are consistently higher than those for smooth circular pipes. This is one of the reasons
why the Moody charts should be used with caution in reactor work. Thus, this exercise demonstrates that the Moody
charts, which were presented earlier in the chapter, actually underestimate the turbulent friction factors by 3%–10%
in PWR fuel assemblies for turbulent isothermal flows. However, we will find in the next chapter that most of this
difference disappears when heat is added to the coolant channel. The flows in this case then become what are called
non-isothermal flows.
When v = 5 m/s, L = 5 m, DH = 0.0125 m, and ρ = 713.8 kg/m3, we find that the turbulent pressure drop is
∆PFRICTION = f ( L/D e ) ρv 2 2 = 0.0142 × (5/0.0125) × 713.8 × 30.25/2 = 61,755 Pa ≈ 62 kPa ≈ 9.0 PSI
Of course, we have ignored the fact that the corner and side channels have different friction factors than the interior ones,
and we have also ignored the presence of the grid spacers, which make the pressure drop larger. In most PWRs, we will
also learn that the total core pressure drop is about 28 PSI. This includes a gravitational pressure head of ΔPGRAVITY =
ρgH = 713.8 × 9.8 × 5 ≈ 35 kPa ≈ 5.1 PSI. Thus, roughly half the core pressure drop can be attributed to the grid spacers,
the upper and lower orifice plates, and the other support structures that are used to hold the fuel rods in place. The inlet
and outlet plenums also contribute to the pressure drop in this case.
17.40 Finding the Pressure Drop in the Primary Side of a PWR Steam Generator
Now suppose that we would like to find the pressure drop through the pressure tubes on the primary side of a PWR
steam generator. An example of a modern PWR steam generator is shown in Figure 17.27. The primary side of a steam
generator in a four-loop Westinghouse PWR has a standard flow rate of about 5,000 kg/s. The coolant flows through
approximately 4,000 pressure tubes inside of the steam generator with an average diameter of 0.0222 m (about seven-
eighths of an inch). The mass flow rate through one of these tubes is 1.25 kg/s, and the average length of these tubes
is about 17.5 m (≈55 ft). The Reynolds number inside one of these tubes is about 75,000. The turbulent friction factor
f TURBULENT inside an individual pressure tube can be found from the McAdams correlation:
which we discussed earlier. Since Re ≈ 75,000, the value of the turbulent friction factor is f TURBULENT ≈ 0.02. If the aver-
age density of the coolant is 710 kg/m3, the velocity of the coolant through a typical tube is about 4 m/s. From the Darcy
equation, the pressure drop across an individual tube with a diameter of 0.0222 m is
Thus, the frictional pressure drop through the pressure tubes in a PWR steam generator is only about 15% as great as the
pressure drop across a PWR fuel assembly. However, in both cases, the flow is designed to be as turbulent as possible
to increase the heat transfer rate. Notice again that these estimates require the flow field to be isothermal. In the next
chapter, we would like to remove this important restriction.
684 Laminar and Turbulent Flows with Friction
FIGURE 17.27 An example of a U-tube steam generator used in a modern PWR. (Picture provided by Westinghouse Nuclear.)
Normally, the corner channels and the edge channels have turbulent friction factors that are about 1% lower than they
are for the interior channels. This makes a great deal of sense because the fluid has a smaller surface area to cling to. For
example, for a P/D ratio of 1.3, the turbulent friction factors for the edge and corner channels are
which the reader can easily compare to the values shown in Table 17.9. Thus, averaging the friction factors is normally
not required for a PWR fuel assembly. However, in BWRs, where the fuel assemblies are surrounded by a “can,” the
friction factors for the interior, edge, and corner channel can be considerably different. In this case, a bundle-averaged
friction factor can make a great deal of practical sense. In the next chapter, we would like to extend this discussion to
include the effects of grid spacers, wire wraps, and orifice plates on the single-phase pressure drop. This will lead to a
Questions for the Student 685
discussion of what is called transverse flow or cross-flow between the reactor coolant channels. Cross-flow also occurs
in reactor heat exchangers (see Chapter 8).
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi, India (2000).
Reactor Theory (Nuclear Parameters), DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
Other References
Luxton, R.M., Bull, M.K. and Rajagopalan, S. “The thick turbulent boundary layer on a long fine cylinder in axial flow”,
The Aeronautical Journal, 88:186–199, 1984.
Tutty, O. “Flow along a long thin cylinder” Journal of Fluid Mechanics, 602:1–37, 2008, Cambridge University Press.
17. Does the boundary layer in a surface flow have the same thickness everywhere, or does its thickness change as one
moves along the surface of an object?
18. For the same fluid and the same hydraulic diameter, is the boundary layer thicker when the flow is laminar or when
the flow is turbulent?
19. What is the thickness of the boundary layer in a reactor fuel assembly when the local Reynolds number approaches
500,000?
20. How far from the entrance of the assembly does the flow become fully developed?
21. What is this distance from the entrance usually called?
22. What is the Reynolds number in a PWR fuel assembly when the coolant pumps are running normally?
23. What is the Reynolds number in an LMFBR fuel assembly when the coolant pumps are running normally?
24. How is the Reynolds number determined in a BWR fuel assembly after the coolant boils?
25. When the flow becomes turbulent and fully developed, what law of fluid mechanics can be used to determine its
shape next to the surface of a nuclear fuel rod? What is this law called?
26. For laminar and turbulent flows, what important information is required to deduce the friction factor from the
momentum equations?
27. What is the shape of the velocity profile in a circular pipe when the flow is fully developed and laminar?
28. What is the shape of the velocity profile in a circular pipe when the flow is fully developed and turbulent?
29. What set of charts or diagrams are used to determine the friction factor in a circular pipe for an isothermal flow?
30. Who was the inventor of these diagrams and what are they usually called?
31. Can these diagrams be used to find the friction factors for both laminar and turbulent flows?
32. Write down an explicit expression for the laminar friction factor after the flow becomes fully developed in a circular pipe.
33. How does the friction factor depend on the Reynolds number in this case?
34. When the flow in a circular pipe becomes fully developed and turbulent, how does the friction factor depend on the
Reynolds number?
35. How can the pressure gradient in a pipe or coolant channel be deduced from the Reynolds number?
36. What is the expression for the Darcy–Weisbach friction factor? Under what set of conditions can this expression be used?
37. What is a representative value for the turbulent friction factor in a reactor fuel assembly?
38. Name three correlations that can be used to determine the turbulent friction factor in a circular pipe.
39. How can these correlations be modified so that they can be used in reactor coolant channels as well?
40. What is a representative value for the Reynolds number in the pressure tubes of a PWR steam generator?
41. How is the equivalent diameter determined for a reactor coolant channel?
42. For a square reactor lattice, how is the equivalent diameter correlated with the P/D ratio of the fuel rods?
43. For fully developed flows, how is the entrance length correlated with the equivalent diameter and the Reynolds number?
44. For the same equivalent diameter D, is the entrance length longer for laminar flows or turbulent ones?
45. What is the Darcy equation and why is it important?
46. Can the Darcy equation be used to determine the frictional pressure drop in a reactor fuel assembly for both laminar
and turbulent flows?
47. What is the explicit form of the Darcy equation when the flow is primarily one-dimensional?
48. Make four important observations regarding the Moody charts.
49. Are the Moody charts suitable for multicomponent flows where a significant amount of slip may occur?
50. How can the friction factors for circular pipes be used to determine the friction factors for reactor fuel assemblies?
51. To provide this conversion, what types of correlations can be used? Name at least three of these correlations.
52. In general, are the friction factors in reactor fuel assemblies higher than, lower than, or the same as the friction
factors in circular pipes having the same equivalent diameter?
53. Above what P/D ratio do the friction factors become larger?
54. Below what P/D ratio do the friction factors become smaller?
55. When these friction factors are used, what is the approximate value of the pressure drop across the tubes on the
primary side of a PWR steam generator?
56. When the same friction factors are used, what is the approximate value of the frictional pressure drop through a
PWR fuel assembly?
57. Is the gravitational pressure drop less than, equal to, or greater than the frictional pressure drop?
58. For turbulent flow in reactor fuel assemblies, the friction factor becomes proportional to approximately what power
of the local Reynolds number?
59. Write down the explicit forms of the Blasius, the McAdams, and the Petukhov correlations when the flow becomes
fully developed and turbulent.
60. If you were to design a new reactor from scratch, which one of these correlations would you use?
Exercises for the Student 687
(
u(r) = 2V 1 – (r/R)2 )
where V is the average velocity of the flow, R is the pipe radius, and r is the distance measured from the center of the
pipe. Using this equation, calculate the velocity of the fluid at the center of the pipe and the velocity of the fluid at the
surface of the pipe wall. What is the boundary condition at the surface of the pipe wall called?
Exercise 17.2
For a Newtonian fluid, the velocity gradient at the surface of a solid object like a nuclear fuel rod affects the fric-
tional forces that are exerted on the object. Normally, the velocity gradients at the object’s surface are greater when
the flow is turbulent than when the flow is laminar. This is true even when the turbulent boundary layer is thicker
than the laminar boundary layer for the same value of the free stream velocity. Using the velocity profile presented
in Exercise 17.1, calculate the velocity gradient du/dr at the surface of the wall of a circular pipe. Using this velocity
gradient, derive an expression for the shear stress τ the fluid exerts on the wall. For water flowing through a pipe at
0.1 m/s and having an average temperature of 20°C, find the magnitude of the shear stress (in N/m 2) exerted on the
pipe wall.
Exercise 17.3
A 5 cm-diameter stainless steel pipe is placed horizontally in a reactor containment building. The pressure at the inlet
to the pipe is 100 PSI (~0.7 MPa). Water at 30° flows through the pipe at a rate of 0.1 m3/s. Determine the pressure drop,
the head loss, and the power required to pump the water steadily through a 20 m-long section of the pipe.
Exercise 17.4
A reactor coolant pipe experiences a gradual area reduction. Coolant enters the pipe at a velocity of 1 m/s with a density
of 960 kg/m3. If the diameter of the pipe is reduced by a factor of 2, and the pressure in the pipe before the area contracts
is 200 PSI (1,380 kPa), what is the pressure in the pipe after the area contracts? Assume the flow is laminar so Bernoulli’s
equation is applicable.
Exercise 17.5
A reactor coolant pipe carries subcooled water to the core. The pipe is 0.2 m in diameter, and the velocity of the water is
2 m/s. If its density is 720 kg/m3, what is the Reynolds number? Is the flow in the pipe laminar or turbulent?
Exercise 17.6
Suppose that a pipe in a reactor containment building has a radius of 0.10 m and that water is flowing through the pipe at
20°C. If the flow in the pipe becomes fully turbulent at a Reynolds number of 10,000. What will be the critical velocity
of the fluid vCRITICAL when this occurs?
Exercise 17.7
The entrance length of a pipe is defined as the length required for the flow to become fully developed. At this point,
the velocity profile in the pipe no longer changes, and the boundary layer becomes stable. Suppose that after the core
is shut down, hot water at 250°C and 800 PSI (5.5 MPa) flows through the pipe at 0.1 m/s. The pipe has a diameter
of 0.1 m. What distance into the pipe does the flow become fully developed? Is the flow at this point laminar or
turbulent?
Exercise 17.8
Water flows over a fuel rod in a PWR core at 5 m/s. Approximately 1 m from the start of the rod, the Reynolds number
is 600,000. What is the thickness δ of the momentum boundary layer at this point?
688 Laminar and Turbulent Flows with Friction
Exercise 17.9
The coolant flows through the core of a commercial PWR at speeds between 4 m/s and 8 m/s. The fuel rods are assumed
to be smooth and straight. At 6 m/s, the Reynolds number is 750,000. What is the value of the turbulent friction factor
at 6 m/s?
Exercise 17.10
The length of a coolant pipe in a reactor containment building is tripled, but the Reynolds number remains the same.
What happens to the pressure drop across the pipe?
Exercise 17.11
Water flows through a coolant pipe in the secondary loop of a PWR at 3 m/s. The pressure of the water is approxi-
mately 800 PSI (5.5 MPa), and the density of the water is approximately 770 kg/m3. The pipe is 0.1 m in diameter, and
the t urbulent friction factor is 0.014. What is the pressure drop in the pipe due to friction alone? Assume the pipe is
10 m long.
Exercise 17.12
The fuel rods in a PWR core have an outer diameter of 0.8 cm. If the P/D ratio is 1.3, what is the hydraulic diameter of
the surrounding coolant channels?
Exercise 17.13
Calculate the Reynolds number through a 17 × 17 PWR fuel assembly under the following conditions:
☉☉ Fuel rod diameter D = 9.5 mm
☉☉ Fuel rod pitch P = 12.6 mm
☉☉ Average coolant pressure = 15.5 MPa
☉☉ Average coolant temperature = 306°C–309°C
☉☉ Average coolant density ρ = 710.5 kg/m3
☉☉ Average coolant viscosity μ = 8.56 × 10 −5 kg/ms
☉☉ Average coolant mass flux = 4,500 kg/m2 s
What is the turbulent friction factor under these conditions?
Exercise 17.14
Calculate the Reynolds number through an LMFBR fuel assembly under the following conditions:
☉☉ Fuel rod diameter D = 8.5 mm
☉☉ Fuel rod pitch P = 9.8 mm
☉☉ Average coolant pressure = 0.1 MPa
☉☉ Average coolant temperature = 530°C–540°C
☉☉ Average coolant density ρ = 817.7 kg/m3
☉☉ Average coolant viscosity μ = 2.28 × 10 −4 kg/ms
☉☉ Average specific heat c = 1,254 J/kg °C
☉☉ Average coolant velocity = 6 m/s
☉☉ P/D ratio = 1.15
Assume the fuel assembly is cooled with liquid sodium. What is the turbulent friction factor based on these design
parameters?
18
Core and Fuel Assembly Fluid Flow
18.1
Friction Factors for Reactor Coolant Channels with Heat Addition
In reactor work, the turbulent friction factors for circular flow channels can be found from the Moody charts or from the Blasius,
McAdams, or Petukhov correlations as long as the temperature of the coolant is similar to the temperature of the wall. In other words,
the frictional pressure drop in a reactor coolant channel can be found from the Darcy equation
as long as the flow is isothermal and T WALL ≈ TBULK. However, for the same Reynolds number, a reactor coolant channel with a
smooth surface will have a higher value for f than the values predicted using any of these methods. The exact amount depends on
the geometry of the fuel assembly and its pitch-to-diameter ratio (P/D). In general, a hexagonal fuel assembly will have values for
the turbulent friction factor that are 3%–5% higher than those given by the Moody smooth surface lines, and square fuel assemblies
will have values that are between 5% and 10% higher. This applies to Reynolds numbers between 10,000 and 500,000. Of course, the
Moody charts were developed from isothermal flow experiments in circular pipes, and so it should not be surprising they pertain to
fully developed flows in cylindrical pipes that are unheated and uncooled.
18.2
Correction Factors for Isothermal and Non-Isothermal Flows
In a reactor fuel assembly, or for any fluid flowing inside an enclosed surface, an isothermal flow is defined as one where the wall
temperature and the fluid temperature are the same, and a non-isothermal flow is defined as one where heat is added to the fluid (or
taken away from it) because of a temperature difference between the fluid and the wall. Normally, the fluid temperature is represented
by the bulk temperature TBULK of the fluid, which is found by averaging the fluid temperature T(r) radially across a coolant channel
of radius R.
R R
TBULK (z) =
∫ 0
T(r, z)2πrρv(r, z) dr
∫ 0
2πrρv(r, z) dr (18.2)
If the flow is not isothermal, then a correction factor (CF) may need to be applied to get the correct results. In this case, the friction
factor must be multiplied by a correction factor to create a new friction factor f′, which is given by
f ′ = CF ⋅ f (18.3)
to account for the fact that the viscosity may be temperature dependent. For water, the value for the correction factor is
where ΔTFILM is the temperature drop across the film between the wall surface and the center of the channel.
Thus, whenever the wall temperature exceeds the bulk temperature, the friction factor is lower than the value predicted by the Moody
smooth surface lines. In pressurized water reactors (PWRs), this temperature difference can be between 10°C and 20°C.
689
690 Core and Fuel Assembly Fluid Flow
FIGURE 18.1 The temperature profiles for the fuel and the coolant in a PWR as a function of axial elevation—assuming no bulk
boiling of the coolant.
For a cosinusoidal axial heat flux, the shape of the fuel, the cladding, and fluid temperature profiles are shown in
Figure 18.1. Notice that the fuel and cladding temperatures peak about three quarters of the way between the bottom
and the top of the core, whereas the coolant temperature peaks at the top of the core where z = H. Thus, the film tem-
perature drop in a PWR is position dependent, and in some cases, the average temperature drop may exceed 20°C or
30°C. This thermal correction factor works for the non-isothermal flow of water in square fuel assemblies with smooth
or rough surfaces. Similar but smaller correction factors are also available for hexagonal fuel assemblies and for other
coolants—although we will not discuss them here.
where ν is the kinematic viscosity, and ν = μ/ρ. Then, we can express the temperature dependence of the Reynolds number by
For saturated water, it is possible to measure μ(T) and ρ(T) and to develop a reasonably accurate correlation for ν(T). A
quadratic curve fit to this data for water gives
where T is the temperature in degrees C, and 0 ≤ T ≤ 360°C. The water is still liquid at 360°C because the pressure in
a PWR is so high. Now let us assume that the temperature at the entrance to the core is 300°C, and the temperature at
the exit of the core is 335°C. According to the correlation for the kinematic viscosity we just presented, the ratio of the
Reynolds numbers at the exit and the entrance is
Re(335°C)/Re(300°C) = ν(335°C)/ν(300°C) = 1.18
18.4 Effect of Surface Roughness on the Friction Factor 691
Hence, the Reynolds number increases by 18% due to just the viscosity change alone. The same thing happens in a PWR
fuel assembly, except that the operating pressure is now 2,250 PSI (~15.5 MPa). Using the McAdams correlation where
the turbulent friction factor is proportional to one-fifth power of the Reynolds number,
it is easy to see that the turbulent friction factor between the top and the bottom of the core must increase by
(1.18)0.2 − (1.0)0.2 ≈ 3% due to just this temperature dependence. Other reactor coolants tend to exhibit a similar behavior
as the temperature is raised.
ε=
∫ε LOCAL (z) dz/∆z (18.10)
FIGURE 18.2 When fluid flows over a rough surface, the average surface roughness ε may be different than the local
surface roughness εLOCAL.
692 Core and Fuel Assembly Fluid Flow
Hence, the local roughness can be either higher or lower than the average roughness. The relative roughness can change
the value of the Darcy friction factor for certain flow regimes. If the flow is laminar, this effect is minimal because the
streamlines tend to flow around any obstacles in their path. The flow field does not separate from the surface, and this
lack of separation is shown in Figure 18.3. However, when the flow becomes turbulent, the friction factor increases as the
surface roughness increases because the additional “obstacles” along the surface cause more vortices to form and detach
from the surface. This increases the interfacial friction and therefore the size of the pressure drop in the predominant
direction of flow. Generally speaking, the rougher a surface is, the larger the Darcy friction factor f will be.
If we take the value of the friction factor f for a smooth fuel rod and ask what the friction factor will be for rough fuel
rod with the same Reynolds number, the new friction factor f ROUGH will be given by
where RCF is the roughness correction factor. In general, the RCF is not a simple function, and it depends on many
parameters including the Reynolds number. If the average surface roughness is ε, then increases in the friction factor can
be correlated to the ratio ε/D, which is called the relative roughness. In this case, the relative roughness is the dimen-
sionless quantity that is used in the Moody charts. For transitional flow and turbulent flow in circular pipes, the friction
factor can be correlated to the relative roughness and the Reynolds number by a relationship called the Colebrook equa-
tion, which is expressed as
where RR = ε/D is the relative roughness. Unfortunately, the Colebrook equation is an implicit equation for the fric-
tion factor because the friction factor can only be found by a tedious iterative process unless an equation solver (like
MATLAB or MATHAMATICA) is used. A somewhat simpler correlation for the friction factor that does not require
an iterative approach is
FIGURE 18.3 In laminar flow, increasing the roughness of the surface of a fuel rod does not appreciably affect the friction factor
because the flow field does not separate from the wall. However, as soon as the flow becomes turbulent, vortices start to spin off of
the wall, and these vortices cause the friction of the flow field to increase.
18.4 Effect of Surface Roughness on the Friction Factor 693
This explicit correlation is called the Haaland correlation. Here, one simply inserts the values of Re and RR into the
Haaland correlation to find the value of f. The Haaland correlation is slightly less accurate than the Colebrook equation,
and the average difference in the friction factor predicted by these two correlations is about 2%.
As one might expect, the friction factors are smallest for a smooth circular pipe, and they increase as the r elative sur-
face roughness increases. The friction factor as a function of the relative roughness is shown in Table 18.1 for a Reynolds
number of about 1 million, which is about twice the average Reynolds number for the coolant channels in a PWR fuel
assembly. Obviously, for a smooth pipe, the value of the relative roughness is zero (RR = 0). Another depiction of the
relationship between the friction factor, the relative roughness, and the Reynolds number is shown in Figure 18.4. Notice
that the relative roughness has almost no effect on the friction factor for laminar flows, but it does have a large effect on
the friction factor for turbulent flows. Specifically, the surface roughness can increase the turbulent friction factor by a
factor of five to seven. This effect is typically greatest at very high Reynolds numbers (Re > 1 million). Moreover, in most
pipes, the friction factor becomes independent of the Reynolds number for a given value of ε/D when Re > 1,000,000.
In nuclear reactors, surface roughness can be very important because intentionally increasing the roughness ratio can
increase the heat transfer coefficient by a factor of 4 or 5. However, this same effect can also be achieved with smooth
fuel rods by connecting them together with rod spacers with axial separations of between 0.5 and 1 m. The rod spacers
then agitate the flow field and increase the local heat transfer rate. We will have more to say about the design of these rod
spacers in another section. Sometimes rod spacers are also called grid spacers.
TABLE 18.1
The Turbulent Friction Factor for a Smooth Pipe for a Reynolds Number of 1,000,000
Relative Roughness of a Coolant Pipe (ε/D) Turbulent Friction Factor (from Colebrook’s Equation)
0.00000 (smooth) 0.0119
0.00001 0.0120
0.0001 0.0134
0.0005 0.0172
0.0010 0.0199
0.0050 0.0305
0.0100 0.0380
0.0500 0.0716
Source: Cengel.
Note that the friction factor is a minimum for a smooth pipe, and it increases with the relative roughness.
FIGURE 18.4 The friction factor as a function of the surface roughness (ε/D) for both laminar and turbulent flow.
694 Core and Fuel Assembly Fluid Flow
In addition to rod spacers, wire-wrapped fuel rods can also be used to increase the heat transfer coefficient and remove
energy more efficiently from the core. Wire-wrapped fuel rods are used primarily in fast reactors.
18.5 The Pressure Drop when the Flow Is Both Laminar and Turbulent
Sometimes a situation can develop where the flow in a reactor coolant channel is both laminar and t urbulent. For
example, suppose that the flow rate m = dm/dt = ρvA in the coolant channel is fixed and does not change with time. If
the area of the coolant channel becomes smaller as we move along the channel, a point is eventually reached where the
velocity becomes high enough for vortices and turbulent eddies to form. In the first part of the coolant channel, the flow
is laminar, in the next part of the channel, the flow is partially turbulent, and in the last part of the channel, the flow is
fully turbulent. In this case, we can divide the channel into N distinct regions and find a separate value of the friction
factor for each region. The physical problem we are trying to solve is illustrated in Figure 18.5. The total frictional pres-
sure drop ΔPFRICTION is then the algebraic sum of the individual pressure drops:
FIGURE 18.5 The single-phase flow field in a constricted channel where the mass flow rate at the entrance and the exit of the
channel are constant. Note that as the channel constricts, the Reynolds’s number increases, and the flow becomes turbulent.
This is because the velocity increases faster than the hydraulic diameter D decreases.
18.6 EFFECT OF EXTENDED SURFACES, RIBS, AND VANES 695
Notice that there can be N individual sections in even a simple pipe. The continuity equation ρvA = constant then allows
us to predict the average velocities v1 + v 2 + v 3 + v N in each of the individual sections. For a pipe with N individual
sections, the equation we must use to find the pressure drop is then
( ) ( ) ( ) ( )
∆PFRICTION = f1 L1 D e1 ρ1 v1 2 2 + f2 L 2 D e2 ρ2 v 2 2 2 + f3 L 3 D e3 ρ3 v 3 2 2 + + fN L N D e N ρN v N 2 2
(18.15)
where
or
Re n = ρn v n D e n µ n (18.19)
Now let us see if we can illustrate this point with a simple example.
In other words, the amount of turbulence can be controlled by simply changing the diameter D of the coolant channel
for the same value of the mass flow rate m = dm/dt = ρvA. As the channel becomes smaller, the ratio of the inertial
forces to the viscous forces in the fluid becomes greater, and the flow becomes more turbulent. The same thing happens
in a reactor coolant channel when we reduce the hydraulic diameter DH = 4A F/PW, where PW is the wetted perimeter. In
other words, reducing the rod pitch P will always cause the flow field to become more turbulent (for the same mass flow
rate), and this will cause the heat transfer coefficient to increase.
18.6 The Effect of Extended Surfaces, Ribs, and Vanes on the Friction Factor
In some reactor components, notably the pressure tubes in advanced heat exchangers, the surface area can be inten-
tionally augmented by machining ribs, vanes, or other protrusions into the surface of the tube. The purpose of these
protrusions is to increase the surface area, thereby increasing the heat transfer coefficient between the fluid and the
surface of the tube as much as possible. In other words, if we can disturb the conductive film covering the surface by
creating more turbulent mixing, then we can enhance the heat transfer through the surface sub-layer. Unfortunately,
this now increases the pressure drop along the surface of the tubes, and we must account for this effect in the cor-
relations that are used for the friction factors. In the late 1990s, Ravigururajan and Bergles were among the first to
systematically study this problem by proposing a set of experiments that could be used to determine the effect that
intentionally machining augmentations into a pipe could have on the pressure drop. In general, they found that the
roughness correction factor (or RCF) was not a simple function of the relative roughness ε/D. Instead, they found that
it depended on a number of factors including
696 Core and Fuel Assembly Fluid Flow
Hence, their correlation was based on a number of empirically determined constants that depended on the Reynolds
number (see Equation 18.21). Here, RR is the relative roughness of surface (RR = ε/D), P is the pitch or distance between
the ribs or protrusions, D is the diameter of the coolant channel, α is the angle of the helix (if there is one) that is
machined into the side of the tubes, β is the contact angle profile in degrees, and n is the number of sharp corners facing
the direction of the flow field (which happens to be two for both rectangular and triangular ribs). An example of this
surface type can be seen in Figure 18.6. Graphical definitions for n and β are also provided. The appropriate values to
use for a, b, c, and d are then
0.01 ≤ RR ≤ 0.2
FIGURE 18.6 This picture illustrates the various types of ribs and fins that can be machined into the side of a coolant pipe to
increase the heat transfer coefficient. The RCFs in this case can be found from the Ravigururajan and Bergles correlation, which is
given by Equation 18.20.
18.7 Loss Coefficients 697
The roughness correction factor works well for liquid water, and it also works for other fluids such as air, hydrogen gas,
and butyl alcohol. The total friction factor in this case is then the smooth Darcy friction factor f, multiplied by the RCF:
f ′ = f ⋅ RCF (18.22)
The RCF is good for Reynolds numbers between 3,000 and 500,000. Hence, it is ideally suited for heat exchangers and
reactor coolant channels where the Reynolds numbers are between 50,000 and 500,000. However, at least inside the
core, reactor vendors have tended to shy away from intentionally machining cavities into the surface of the fuel rods. The
surface stresses that these irregularities create are simply too large for the cladding to endure over the life of the fuel.
Consequently, grid spacers and wire-wrap spacers are used to increase the amount of turbulent mixing. We would like
to spend the next few sections describing the additional friction and turbulence that these spacers create. The best way
to illustrate the importance of this effect is to start with a bare rod bundle, and to then add grid spacers and wire-wrap
spacers to the bundle as we go.
FIGURE 18.7 The pressure in a reactor coolant channel depends on whether the channel is expanding or contracting. The pressure
difference between points 1 and 2 is given by P2 − P1 = K·½ρv12 where the loss c oefficient K = (1 − A1/A2)2 can be either positive or
negative depending on the values of A1 and A2.
698 Core and Fuel Assembly Fluid Flow
a form coefficient or a form factor because it indicates that the shape or form of the coolant channel is about to change.
Normally, the loss coefficient is correlated to the pressure change ΔP by
which we briefly discussed in Chapter 16. Thus, for the same mass flux G = ρv, larger values of the loss coefficient
lead to larger pressure changes, and smaller values of the loss coefficient lead to smaller ones. Before discussing
how reactor flow fields are affected by these structural changes, we would first like to derive an expression for how
the pressure changes when there is an area change in a coolant channel. For incompressible fluids, it is possible to
assume that ρ is constant, and this simplifies the momentum equations that govern the flow field. Consider the sud-
den change in the area of the flow channel shown in Figure 18.7. The pressure immediately before the area change
(at point 1) is P1, and the pressure immediately after the area change (at point 2) is P 2. When the wall friction is
small or can be neglected entirely, then the energy equation governing the pressure change is simply Bernoulli’s
equation:
ρv 2 2 2 + P2 = ρv1 2 2 + P1 (18.24)
ρ v 2 A 2 = ρv1 A1 (18.25)
( )
2
P2 − P1 = 1 − A1 A 2 ⋅ 1 2 ρv1 2 (18.26)
P2 − P1 = K ⋅ 1 2 ρv1 2 (18.27)
TABLE 18.2
The Loss Coefficients for Turbulent Flow through Pipes, Valves, and Other Components of a Nuclear Power Plant
Pipe Fittings
45° elbow 0.20–0.45 Inside pipe
90° elbow 0.35–0.90 Inside pipe
Standard tee 0.20–0.60 Downstream velocity v2
Valve—half open 13–450 Inside pipe
Valve—fully opened 0.15–15.0 Inside pipe
Source: Todreas and Kazimi (2008).
18.7 Loss Coefficients 699
where K = (1 − A1/A 2)2. Thus, the acceleration or deceleration of a fluid (which is represented by the area ratio A1/A 2)
causes a pressure change to occur even when there is no friction in the coolant channel! Sometimes the ratio A1/A 2
is given the symbol β which is called the expansion or contraction factor. In this case, the expression for the loss
coefficient becomes K = (1 − β2). Notice that the value of β can be greater than 1.0, unity, or less than 1.0. If β < 1,
then P 2 > P1 and the pressure increases in the coolant channel. On the other hand, if β > 1, then P 2 < P1 and the pres-
sure decreases in the coolant channel. Thus, it is only when β = 1 and A1 = A 2 that the pressure remains unchanged.
In other words, we have just shown that the form factor can be either a positive or a negative number depending
upon what is happening in the coolant channel, and that different form factors can cause the pressure to increase or
decrease. Normally, the values for K are tabulated for different geometric shapes such as elbows, tees, and valves,
and the values of these form factors are presented in Table 18.2. The overall pressure field in a contracting pipe as
the result of these form factors is shown in Figure 18.8. Sometimes engineers object to the equations we have just
presented because they ignore the effects of the Reynolds number on the loss coefficients. However, this objection can
be addressed by measuring the form factors experimentally and then presenting them to include friction in the value
of K. In general, this means that the form factors must be tabulated as a function of the Reynolds number. A diagram
that attempts to do this is shown in Figure 18.9. Many tables similar to this table are used to determine the pressure
changes in a reactor piping system due to bends, elbows, valves, couplings, and tees. Next, we would like to apply
these form factors to other objects that can be found in a reactor core. These objects include the grid spacers and the
orifice plates.
FIGURE 18.8 The pressure field in a contracting pipe. In an expanding pipe, exactly the opposite effect occurs, but there is a pres-
sure increase where the pipe widens and this pressure increase is due to deceleration and form effects.
700 Core and Fuel Assembly Fluid Flow
FIGURE 18.9 The loss coefficients (i.e., form factors) for expansion and contraction from a bank of parallel tubes connected to
a header which then feeds into a larger flow area or inlet/outlet plenum. The Reynolds number in this case is the Reynolds number
inside of an individual tube. (Kays and London, Compact Heat Exchangers, McGraw Hill.)
or
∆P = 1 2 ρv 2 = (1/2ρ)G 2 (18.29)
because the loss coefficient is normally close to 1.0. Normally, the pressure drop occurs within a few centimeters of
the interface between the pipe and the plenum. An analogous situation occurs when a number of pipes receive coolant
from a plenum pool. In this case, the fluid constricts as it enters the pipes, and the loss coefficient becomes a function of
the shape of the pipe at the entrance (see Table 18.2). For a well-rounded entrance, the loss coefficient is only 0.05, but
for a sharp entrance, it rises to about 0.5. A pipe with a slightly rounded entrance has a loss coefficient closer to 0.25.
Finally, if the pipe protrudes completely from the plenum, the loss coefficient is close to 1.0 again. Examples of both of
these situations are illustrated in Figure 18.10. In other words, the direction of the flow determines the size of the loss
coefficient at the mouth of the pipe. The flow into a plenum always has a larger loss coefficient than the flow out of a
plenum. The reader is asked to keep this in mind when visualizing the way in which the pressure changes in the reactor
18.9 The Effect of Grid Spacers on the Pressure Drop 701
FIGURE 18.10 In general, the loss coefficients for pipe flow into a plenum are different than the loss coefficients
for pipe flow out of a plenum.
pressure vessel and in the rest of the nuclear steam supply system. The example below illustrates the size of the pressure
drop near the exit to a pipe when the pipe dumps its contents into a large plenum pool.
where m = dm/dt = ρvA was the mass flow rate. For a very large area change (like a flow blockage), a “k factor” can have
a value between about 0.5 and 1.0. A similar k factor can be used to predict the pressure drop caused by the presence of
grid spacers or orifice plates. When the flow in a reactor coolant channel encounters a grid spacer, the flow area tempo-
rarily contracts because the grid spacer obstructs part of the channel. This causes the fluid velocity in the vicinity of the
grid spacer to increase, and so the pressure must fall along the grid spacer because of the acceleration change. It is not
uncommon for about half of the flow area to be obstructed by a properly designed grid spacer (see Figure 18.11). Most
fuel assemblies have grid spacers at five or six different locations along their length, and for structural reasons, they are
generally spaced about 1 m (~3 ft) apart. A picture of some PWR grid spacers is shown in Figure 18.12.
702 Core and Fuel Assembly Fluid Flow
FIGURE 18.11 In some PWRs, the grid spacers can obstruct up to half of the subchannel flow areas at different axial elevations.
This gives rise to frictional and form pressure losses as the coolant passes over and around the spacers.
FIGURE 18.12 A PWR fuel assembly showing the location of the grid spacers. (Pictures provided by the US NRC.)
Rod spacers or grid spacers are used to hold reactor fuel assemblies together, and they keep the fuel rods inside of the assem-
blies from vibrating or bowing as the coolant flows by. The flow of coolant in a reactor core can become very violent at times.
Sometimes the flow of the coolant over the fuel rods can be enough to lift an entire fuel assembly out of the core if it is not
restrained by springs. The grid spacers improve the structural integrity of the assembly, and so it is not practical to build fuel
assemblies without them. In PWRs, the pressure drop across the grid spacers is typically 20%–25% of the total pressure drop
through the core. The design of these grid spacers can be very complex because in addition to keeping the fuel rods in place,
they must not create any hot spots on the fuel rods that prevent the coolant from reaching the cladding (see Figure 18.13).
FIGURE 18.13 The picture on the right shows a grid spacer with control rod guide tubes, and the picture on the left shows a grid
spacer assembly surrounding a fully loaded 17 × 17 PWR fuel assembly. (Pictures provided by AREVA.)
So if we were to use the recommended equation to find the pressure drop caused by a grid spacer (ΔPGRID = KGRIDρv2/2),
then the value of KGRID must consist of the sum of three separate parts:
Ordinarily, the geometry term “KGEOMETRY” and the friction term “K ROD” within the spacer region are about the same
size, and they are about ten times larger than the grid friction term “K FRICTION”. For a typical turbulent flow pattern with
a Reynolds number between 50,000 and 100,000, the values of these coefficients are
Naturally, they are a function of the Reynolds number. Most papers list the value for KGRID at about 0.50. However,
this value can change quite a bit depending on how a particular spacer is designed. Now suppose that we would like
to find the pressure drop produced by a single grid spacer in a PWR fuel assembly. The fluid flows through the core
at a rate of about 6 m/s with an average density of ρ ≈ 714 kg/m 3. For a loss coefficient of 0.52, this gives a pressure
drop of
For a typical fuel assembly with six grid spacers, the total pressure drop is then 6 × 6.68 ≈ 40 kPa ≈ 6 PSI. Grid spacers
are also used in boiling water reactors (BWRs), and a similar methodology can be used to find the pressure drop through
these assemblies. However, two-phase pressure drops can be quite different than single-phase pressure drops, and they
can depend on many additional factors including the void fraction and the slip ratio. We will have more to say about the
size of these pressure drops in Chapter 23. In general, for the same flow rate, two-phase pressure drops are much larger
than single-phase pressure drops.
704 Core and Fuel Assembly Fluid Flow
where A is the subchannel area away from the wire-wrap, AW is the projected frontal area of the wire-wrap, and KWW is the
k factor or “drag coefficient” for the wrap. The k factor is a function of the Reynolds number, and in most correlations, it is
a function of the Reynolds number in a reactor coolant channel without a wire-wrap spacer. In most cases, the values of the
k factor are smaller than they would be for normal grid spacers. Also, it is easy to see that the k factors are generally lower
for a square lattice than they are for a hexagonal one. In some cases, they can be as much as 10%–20% lower. Lenticular
wires also have loss coefficients that are about 30% lower than circular wires. In addition to liquid metal fast breeder reac-
tors (LMFBRs), wire-wrap spacers are also used in CANada Deuterium Uranium (or CANDU) reactors (see Figure 18.15).
FIGURE 18.14 A picture of a wire-wrap spacer used in fast breeder reactor. In these reactors, the fuel assemblies are hexagonal in shape.
18.12 Loss Coefficients and the Reynolds Number 705
FIGURE 18.15 The wire-wrap spacers in a CANDU reactor fuel assembly (top right) and the wire-wrap spacers in a LMFBR fuel
assembly top (left). The picture on the right was taken from canteach.candu.org. Below is a picture of a hydrodynamic simulation
of the coolant pressure in a 217-pin wire-wrapped LMFBR fuel assembly with an axial coolant flow. Notice that the position of
the wire-wrap spacers causes the pressure field to corkscrew through the fuel assembly as one moves up through the core axially.
(Courtesy of Argonne National Laboratory.)
FIGURE 18.16 The drag coefficients for grid spacers and wire-wrap spacers are a function of the local Reynolds number.
In general, they become smaller as the Reynolds number is raised. (Todreas and Kazimi 2008.)
706 Core and Fuel Assembly Fluid Flow
component of the total loss coefficient, KGEOMETRY, is a function of the Reynolds number, and in general, it can decrease
by 40%–50% as the Reynolds number increases:
Obviously, the value of Re must be evaluated at the location where the obstruction occurs.
FIGURE 18.17 In the picture above, a PWR core support plate is shown on the left, and a BWR core support plate is shown on
the right. The picture below shows the side view of a grid spacer assembly for a PWR. In this view, the control rod guide tubes have
already been inserted in to the assembly. The pictures above were provided by Skoda. See www.skoda.com, and the picture below
was provided by Korea Electric Power.
18.15 Changing the Core Flow with an Orifice Plate 707
FIGURE 18.18 In the rendering of the nuclear steam generator on the right, the orifice plate can be clearly seen at the
top. The picture on the left is a cross-sectional view of a Westinghouse nuclear steam generator used in most PWRs.
(Westinghouse Nuclear and Modelbuilders.net.)
is
and the mass flow rate m
1 = A1ρ1 √ ( 2ρ∆P/k1 )ρ
m (Hole type 1) (18.36a)
2 = A 2ρ2 √ ( 2ρ∆P/k 2 )ρ
m (Hole type 2) (18.36b)
3 = A 3ρ3 √ ( 2ρ∆P/k 3 )ρ
m (Hole type 3) (18.36c)
Suppose that the density ρH of the holes across the plate varies in the radial direction, and the density of the holes is
defined as the number of holes NHOLES located in a unit area of the plate divided by the surface area APLATE of the plate:
The radially varying hole density is then ρH(r). This leads us to the conclusion that the axial mass flow rate as a function
of radial position across the plate is
= [m
m(r) 1ρH1 (r) + m 3ρH3 (r) ] A PLATE (18.38)
2 ρH2 (r) + m
where r corresponds to the plate centerline. In general, plates can be designed so that there is a different value of ρH(r)
for each hole type. Equation 18.38 can then be used to shape the mass flow rate as a function of the radial position r. In
reactor orifice plates, the hole density is normally the same for each type of hole. In this case, Equation 18.38 becomes
= [m
m 1+m 3 ] ρH (r)A PLATE (18.39)
2 +m
Any radially varying flow distribution can be created across an orifice plate in this way. Normally, reactors are designed
so that more fluid is forced through the center of the core (where the power density q‴ is higher) than around the edges
of the core (where the power density q‴ is lower). A small amount of fluid is also allowed to flow around the core shroud.
Thus, the orifice plate takes the coolant flow which enters the inlet plenum (and is normally uniform in the radial direc-
tion), and distributes it so that it cools the core as efficiently as possible. This redistribution of core flow is shown in
Figure 18.19. We can then calculate the pressure drop across a horizontal orifice plate alone from the equation:
where K PLATE is the loss coefficient for the plate. In commercial power reactors, the orifice plate has a loss coefficient of
about 1.0. At a coolant velocity of 6 m/s and a density of ρ ≈ 715 kg/m3, the orifice plate pressure drop is
A reactor orifice plate also provides structural support for the fuel assemblies in the core. These fuel assemblies can
weigh as much as 275 kg in a BWR and about 650 kg in a PWR. Since a large PWR core contains about 200 of these fuel
assemblies and a large BWR core contains about 800 of them, the total weight of the fuel assemblies can be between
130 and 220 metric tones.
FIGURE 18.19 The core flow patterns can be changed as a function of distance from the center of the core using an orifice
plate with properly designed holes.
this effect by considering two parallel fuel assemblies that have the same overall pressure drop between the top and the
bottom of the core. We know from our previous discussion that we can write the total core pressure drop as
where G is the mass flux (G = ρv), H is the core height, and the value of K can consist of many things. If we solve this
equation for the axial mass fluxes G1 and G2, and the pressure drop ΔPCORE between the top and the bottom of the core
is fixed, then it is easy to see that the flow rates in the fuel assemblies must be related by
if the frictional forces along the fuel assemblies can be neglected. The mass fluxes in channels 1 and 2 are then related
to each other by
( )( )
G1 2 = K 2 K1 ρ1 ρ2 G 2 2 (18.43)
or
( )( )
G1 = √ K 2 K1 ρ1 ρ2 G 2 (18.44)
Hence, we can adjust the axial mass flux by simply varying the values of K1 and K 2. Now suppose that the energy being
generated in fuel assembly 2 is 10% higher than the energy generated by fuel assembly 1. If they are to both operate at
approximately the same temperature, then G2 must be equal to 1.10G1. Neglecting the density differences between them,
the values of the loss coefficients must be adjusted so that
(K ) ( )( ) = (ρ )
2
ρ1 (1/1.1) (18.45a)
2
2 K1 = ρ2 ρ1 G1 G 2 2
or
K 2 = 0.82K1 (18.45b)
A similar procedure is used in a real reactor to keep the temperatures of the fuel rods as uniform as possible. Only in
this case, the value of the turbulent friction factor f must be included in the calculation.
710 Core and Fuel Assembly Fluid Flow
During normal operation, about 3% of the flow bypasses the core entirely, and so only 97% of the flow actually reaches
core. These core flow splits exist during both forced convection and natural circulation. However, the Reynolds number
always falls precipitously after the reactor coolant pumps are shut off, and so the flow can be either laminar or turbulent
after the power to the pumps is lost. (The exact flow regime that appears depends on the amount of heat that is generated
in the core at the time the flow pattern is measured.) We will discuss the implications of this in Chapter 22.
FIGURE 18.20 The fluid velocity profile in a BWR fuel assembly generated by the Fluent computer program. Notice that the
maximum coolant velocity is about 2.5 m/s in this simulation. (Argonne National Laboratory.)
18.20 Subchannel Cross-Flow 711
FIGURE 18.21 In a BWR, each fuel assembly is surrounded by a stainless steel “can” or shroud designed to prevent the voids in
one fuel assembly from propagating to another. The different colors in the picture represent different enrichments. In the nuclear
business, the U-235 content of a fuel rod is referred to as its enrichment. The control rods in BWRs are also called control blades
because of their cruciform shape. BWR control rods are much larger than their PWR counterparts.
of these assemblies is always the same. This implies that if there are N fuel assemblies in a BWR core, and the total core
pressure drop is ΔPCORE, then we can write
where i is the index of fuel assembly i, and N is typically a number on the order of 800. Thus, three-dimensional core flows
do NOT exist between fuel assemblies in BWRs because there is virtually no coolant flow between the assemblies in the
radial (or transverse) direction. However, the flow field within an individual fuel assembly can still be three-dimensional
if there are large radial power gradients. Finally, PWRs have exactly the opposite behavior because the fuel assemblies
are not hydraulically isolated from each other by cans or shrouds. Reactor cores that are designed in this way are called
open cores. Thus, the fuel assemblies in a PWR are connected hydraulically in the transverse direction, and for detailed
design calculations, it is no longer possible to assume that the fuel assembly flow fields are isolated from one another in
the transverse direction. In fact, it is often not possible to assume that the individual coolant channels within a specific fuel
assembly are hydraulically isolated from one another. We will have more to say about this in the next section.
18.19
Cross-Flow and Its Origins
In PWRs, a small fraction of the coolant can flow between adjacent coolant channels in the radial or t ransverse direc-
tion, and it can also flow between adjacent fuel assemblies under certain conditions. If there is a t ransverse pressure
difference ΔP = P2 − P1 between two coolant channels at the same axial e levation z, then some coolant will flow from
the channel having the highest pressure to the channel having the lowest pressure. This is a consequence of Pascal’s
law, which we first discussed in Chapter 14.
Finally, if PWR fuel assemblies happen to be stored in a spent fuel pool, a lot of different things can happen if the flow
of coolant to the pool is suddenly shut off. After a period of time, the water will heat up enough that it will begin to boil.
However, the boiling will start in the hottest assemblies, and then over time, it will propagate to the colder ones. The flow
field in this case becomes three-dimensional because there is no longer a “can” to keep the water from flowing between
the assemblies. These three-dimensional effects can become important when determining where the fuel rods will melt.
Next, we would like to examine how to model these three-dimensional effects when the coolant does NOT boil. Then, we
will discuss what happens to the flow field when boiling enters into the equation.
18.20
Subchannel Cross-Flow
In commercial power reactors, transverse flow (which is also called cross-flow) occurs all of the time because of differ-
ences in the density of the coolant between one flow channel to the next. These density differences develop when the
712 Core and Fuel Assembly Fluid Flow
FIGURE 18.22 Cross-flow between adjacent subchannels can occur if the temperature in one reactor coolant channel is differ-
ent than the temperature another. In general, the coolant flows from the hot subchannels to the colder ones. Cross-flow is caused by
lateral pressure differences induced by different power generation rates in a reactor’s fuel rods.
coolant is heated, and these density differences can also change as a function of time. The resulting lateral flow rate of
coolant in a reactor core is called cross-flow. Cross-flow was originally introduced into reactor thermal-hydraulics in
the early 1970s when computers became powerful enough to understand the consequences of its existence. Sometimes a
certain amount of cross-flow is intentionally designed into a fuel assembly by making subtle changes in the design of the
grid spacers or the orifice plates to shape the radial flow pattern. Cross-flow can also be created when the control rods are
moved into and out of a fuel assembly. Their vertical motion can lead to significant radial power fluctuations and power
peaks, and this in turn will also produce some measureable amounts of cross-flow. Ordinarily, the cross-flow w between
two adjacent coolant channels is modeled by assuming that the mass flux G in the transverse direction is directly pro-
portional to the average pressure difference between the two channels. We can express this in the form of an equation as
where K ij is the cross-flow coupling coefficient between channels i and j, and Aij is the area of the gap between channels
i and j. Cross-flow is usually measured in kilograms per second. Normally, the cross-flow rate is also a function of axial
elevation, and so strictly speaking, the cross-flow is given by
This implies that the value of the cross-flow depends on the axial elevation z and also that its direction can change back
and forth (from + to −) depending upon where one is in the core. Normally, the cross-flow coupling coefficients Kij are
measured experimentally as a function of the P/D ratio, and as a function of the average Reynolds number or void frac-
tion. Then, as the pressure drops are determined for the individual coolant channels from the Darcy equation, a different
pressure drop may occur in each channel at the same axial elevation. This causes the channel flow rates to change because
some fluid moves from the high-pressure channels to the lower pressure ones. The situation is depicted in Figure 18.22.
18.21
I ntra-Assembly Cross-Flow
Finally, in open cores, it is even possible for some coolant to flow between adjacent fuel assemblies at a given axial eleva-
tion. The flow in this case is called intra-assembly cross-flow. Intra-assembly cross-flow is illustrated in Figure 18.23.
The cross-flow rate is found in exactly the same way as it is found between individual coolant channels, except that the
18.22 Core Flow Patterns 713
FIGURE 18.23 When coolant flows between adjacent fuel assemblies in the transverse direction due to radial temperature
differences between the assemblies, this type of flow is called intra-assembly cross-flow. In reactors with open cores, the coolant
normally flows from a hotter assembly to a colder one.
cross-sectional areas are much larger, and the pressures Pi and Pj are now the average fuel assembly pressures at each
axial node (i, j, k). The coupling coefficients K ij may have to be modified to account for the fact that we are dealing
with larger cross-sectional areas. However, even though we are applying these coefficients to fuel assemblies rather
than individual coolant channels, the concept is exactly the same. Ordinarily, in water-cooled reactors, the cross-flow
coupling coefficients have values of about 0.02 kg/cm2 Pa s. However, these values can change if the P/D ratio changes,
and they can also change if we move from a water-based coolant to a liquid metal one. In some designs, their values can
approach 0.05 kg/cm2 Pa s.
18.22
Core Flow Patterns
Most reactors today (with the notable exception of Canadian CANDU reactors) allow cold coolant to enter the core from
below and exit the core from above. Reactors operate in this way because the heat generated by the core tends to expand
the coolant, and reduce its density. Density differences then cause the coolant to rise. Hence, when power to the coolant
pumps is lost, the core can continue to cool itself as the result of these temperature-induced density differences. The flow
patterns in this case are referred to as natural circulation flows. Various types of upflow are possible when these natural
circulation patterns develop. Some of the more common patterns are shown in Figure 18.24. In addition to this type of
up flow, the coolant will normally flow from the hotter fuel assemblies to the colder ones. This horizontal flow redistri-
bution is driven by lateral pressure differences induced by the thermal expansion of the coolant. Even within individual
fuel assemblies, the coolant will flow from the hotter coolant channels to the colder ones. In both cases, exactly the same
hydrodynamic principle applies.
714 Core and Fuel Assembly Fluid Flow
FIGURE 18.24 During both natural convection and forced convection, the primary direction of the flow through the core is
vertical. However, a small amount of the fluid may also flow horizontally rather than vertically due to lateral pressure and density
differences. These differences can occur during both laminar and turbulent flows, and the transverse flow in this case is generally
referred to as cross-flow.
Important Observation
Pressure differences induced by temperature and density fluctuations in the core allow these horizontal flow
patterns to develop. These horizontal flows are then referred to as cross-flows.
Normally, the amount of cross-flow is limited to only a few percent of the total vertical mass flow rate, for example,
We can estimate the magnitude and the direction of this cross-flow wij by applying the correlations that we already
discussed. We simply apply the formula
to every coolant channel, starting at the bottom of the core and proceeding to the top. To simplify our discussion, assume
for the moment that the pressure drop induced by the grid spacers is the same radially and that the gravitational pressure
drop is much less than the frictional pressure drop. Then, we can write
where ΔPij(z)CHANNEL is the horizontal pressure difference between coolant channels i and j at axial e levation z. If the
flow is turbulent, then we can apply the McAdams correlation f = 0.184/Re0.2 (see Chapter 17) to find the frictional pres-
sure drop, and the result is
( ) ( )
∆Pij (z) = ∆PI (z)FRICTION − ∆PJ (z)FRICTION = fi L D Ei ρi v i 2 2 − f j L D Ej ρ j v j 2 2 (18.52)
18.22 Core Flow Patterns 715
or
In practice, Equation 18.53 can be used to find the value of the pressure difference between two arbitrary channels i and j
(unless we shut off the coolant pumps and the flow becomes laminar). So according to this definition, ΔPij(z) can be either
positive (+) or negative (−), and its value can change as a function of axial elevation. The corresponding cross-flow is then
where K ij are the coupling coefficients that we introduced previously. If we substitute the McAdams correlation into this
equation for the cross-flow, the result is
( )
w ij = 0.092LK ij A ij (z) µ i 0.2ρi 0.8v i 1.8 D Ei 0.8 − µ j 0.2ρ j 0.8v j 1.8 D Ej 0.8 (18.55)
( )
w ij = 0.092LK ij A ij (z) µ i 0.2ρi 0.8v i 1.8 D Ei 0.8 − µ j 0.2ρ j 0.8v j 1.8 D Ej 0.8 (18.56)
and finally, if the equivalent diameters are the same, we arrive at the following expression for the amount of cross-flow:
Cross-Flow Correlation
( )
w ij = Cij µ i 0.2ρi 0.8v i 1.8 − µ j 0.2ρ j 0.8v j 1.8 (18.57)
where
( )
∆Pij (z) = 1 2 L fi ρi v i 2 D ei − f jρ j v j 2 D ej + ( ρi (z) − ρ j (z) ) g (18.58)
However, when we do so, the results are essentially the same. In the core, the greatest contributor to this pressure dif-
ference is simply the density difference between the two channels. So as the coolant expands, it forces the flow into
the cooler channels around it. This is a form of positive feedback because it means that when the cross-flow starts, it
can always get larger if the heat flux q″(z) stays the same. Fortunately, reactors are designed to account for this fact. As
ρ falls, the heat generation rate q‴ also falls because there are fewer thermal neutrons produced. This causes the flow
rates to eventually stabilize around an equilibrium configuration that we can measure. Figure 18.22 illustrates what the
cross-flow looks like when we have a “hot” coolant channel, surrounded by two cooler coolant channels. As the figure
shows, the cross-flow always flows from the hot channel to the cooler channels, and this happens to be the cross-flow
that we measure in practice. Needless to say, it is very important to understand the three-dimensional flow patterns in
each case, and the cross-flow provides us with a convenient tool to do so. In Exercise 18.5, the reader is asked to derive
the expression for the cross-flow when the effects of gravity are also taken into account. The lateral flow between hot
fuel assemblies and cold fuel assemblies also exhibits a similar trend.
716 Core and Fuel Assembly Fluid Flow
FIGURE 18.25 In BWRs, the fuel assemblies are surrounded by “sheaths” or “cans” to prevent vapor bubbles and other hydrody-
namic instabilities from propagating between two fuel assemblies in the transverse direction. Most PWRs do not have these cans,
but the coolant can still flow laterally from hot fuel assemblies to cold ones.
An Important Fact
Differences in the viscosity and the density between adjacent coolant channels always create the cross-
flows that we observe.
18.23
T he Effects of a Flow Blockage
Our previous discussion was based on the assumption that the flow rate through the core was essentially independent of
time. However, during some types of transients, the flow rate in certain parts of the core can change rather quickly, and in an
extreme case, the fuel rods may become overheated or fail. In this case, a flow blockage can develop in some of the reactor
coolant channels. There are several ways in which this can happen. Normally, some molten fuel or cladding can be ejected
into the coolant. When the fuel and cladding subsequently refreeze, they create what is called a flow blockage. Normally, a
flow blockage will obstruct part of a channel, but in most cases, it will not obstruct all of it. Assume that we can partition the
core in the vicinity of the blockage into a number of axial nodes like the ones shown in Figure 18.26. Now suppose that the
blockage develops at axial node number 7 (near the core centerline) and that it obstructs roughly half of one of the coolant
channels. Then, the coolant will slow down immediately in front of the blockage because there is an obstruction in its way.
This causes the coolant to redirect itself around the b lockage and cross-flow to increase in the lateral direction. The coolant
that is able to continue to flow past the blockage in the central channel then accelerates because there is a smaller flow area
through which to flow. According to Bernoulli’s principle (see Chapter 16), the pressure will drop behind the blockage and
increase in front of it. This will cause the axial flow to reverse direction immediately behind the blockage. This flow reversal
is shown between nodes 8 and 9, and in reactor thermal-hydraulics, this phenomenon is known as reverse flow. Thus, in
addition to upflow and cross-flow, the presence of the blockage can also create downflow. When people first discovered this
effect in the 1960s, they were concerned that a small flow blockage could be a very bad thing. However, if the size of the
blockage does not exceed about 50% of the area of the coolant channel, the flow recovers very quickly (within a couple of
centimeters), and no real harm is done. We would now like to turn our attention to how the pressure field in the core behaves.
18.24 Estimating the Total Core Pressure Drop 717
FIGURE 18.26 In the vicinity of a flow blockage, the flow may actually reverse, creating what is known as “downflow” or “reverse
flow.” In PWRs, this condition normally persists a few centimeters beyond the location of the blockage itself.
18.24
Estimating the Total Core Pressure Drop
In commercial power reactors, the reactor coolant usually enters the core from below and exits the core from above.
The pressure always falls as the elevation rises. (The sole exception to this rule is the Canadian CANDU reactor,
where the coolant enters and exits the core from the sides.) In PWRs, this pressure drop is quite small compared to
the total operating pressure of the plant (~30 PSI vs. 2,250 PSI), but it can have a major affect on the flow patterns
because the coolant flows from high-pressure regions to low-pressure ones. In BWRs, the core-wide pressure drop
is a larger percentage of the average operating pressure (about 2.5% of 1,050 or 25 PSI) than it is in a PWR, but this
is due to the fact that the coolant boils as it flows upward through the core. The principal causes for these core-wide
pressure drops are
Sometimes the rod pressure drop ΔPRODS is also subdivided into two parts—the single-phase pressure drop ΔPSP and
the two-phase pressure drop ΔPTP:
FIGURE 18.27 The core pressure profile in a typical PWR as a function of axial elevation. In a BWR, the core pressure profile is
different because the coolant begins to boil approximately one-fourth of the way into the core (see Chapter 23).
The two-phase pressure drop is much larger than the single-phase pressure drop for the same mass flow rate. In fact, in the
primary loop of a BWR, this difference can be as great as a factor of ten (see our discussion of this subject in Chapter 23).
In a boiling core, the two-phase pressure drop is then found by taking the single-phase pressure drop ΔPSP and multiply-
ing it by a two-phase multiplier MTP, having a value between unity and about 10. The two-phase pressure drop is then
where the single-phase pressure drop ΔPSP is evaluated using the properties of the mixture. The rod pressure drop
ΔPRODS tends to be the largest of the individual pressure drops, but it occurs gradually over the entire length of the fuel
rods. The grid pressure drop ΔPGRIDS is also fairly large, but its effects much more localized because the pressure drop is
driven by the acceleration of the coolant. The third and fourth system-wide pressure drops ΔPINLET and ΔPOUTLET occur
only at the top and the bottom of the core, and so they can be thought of as one-time perturbations to the flow field.
Gravity also changes the core pressure drop, but it is roughly the same magnitude as the pressure drop across the grids.
The gravitational pressure drop is simply
18.25
Comparing the Sizes of the Pressure Drops
Table 18.3 lists the individual components of the pressure drop as a function of reactor type. These c omponents assume
a coolant velocity of about 6 m/s for a PWR and about 2.5 m/s for a BWR. In a PWR, the largest contributors to
18.25 Comparing the Sizes of the Pressure Drops 719
TABLE 18.3
Representative Pressure Drops in Different Components of the NSSS (Here 1 PSI ~ 6.9 kPa)
the pressure drop are the frictional forces on the fuel rods and the grid spacers. The gravitational pressure head
ΔPGRAVITY = ρgH is also a significant contributor to the pressure drop in the axial direction. In general, the pressure drop
in a PWR core is about 10% larger than it is in a BWR core (~28 vs. ~26 PSI). In a PWR, the acceleration pressure drop
caused by the expansion of the coolant is very small, and it can usually be neglected except in the vicinity of the grid
spacers, where changes to the flow patterns can have some effect on the DNBR, or departure from nucleate boiling ratio.
In a BWR, the individual components of the core-wide pressure drop behave somewhat differently. The gravitational
pressure drop ΔPRODS = ρgH is only about half as large as it is in a PWR because the average void fraction in the core
is about 45%. In other words, about half of the coolant is converted into steam, and the gravitational pressure head is
due primarily to the remaining liquid. The other major difference between a BWR and a PWR is that the boiling of the
coolant in a BWR causes the flow field to accelerate much more rapidly in the upper half of the core, and this causes an
acceleration pressure drop that is on the order of 14–18 kPa. In fact, the acceleration pressure drop can be even larger
than the gravitational pressure drop. Table 18.4 summarizes the individual pressure drops by region instead of by source.
The largest pressure drops still occur across the fuel rods, and surface friction is primarily responsible for this. BWRs
are interesting in this respect because the frictional pressure drop in the lower half of the core is about one quarter of
what it is in a PWR because the flow rate is about half as great. In the upper half of the core, the two-phase pressure drop
is about four times the single-phase pressure drop, but because the flow rate is about half of what it is in a PWR, these
two effects tend to offset each other, and the pressure drop in the top half of the core is comparable to that in a PWR.
Therefore, on average, the total pressure drop due to friction in a BWR is about ½(H/2 × 1.0 + H/2 × 0.25)/H or about
TABLE 18.4
The Pressure Drops in Several Light Water Reactor Cores—as Defined by Region
Region of the Core PWR Pressure Drop (PSI) BWR Pressure Drop (PSI)
Core inlet 3–4 2–3
Grid spacers 6–8 6–8
Fuel rods 12–14 ~14
Core outlet 3–4 2–3
Total pressure drop ~24–32 PSI (28 Avg) ~24–28 PSI (26 Avg)
Note: The numbers shown here represent industry averages, and the actual numbers may be higher lower.
720 Core and Fuel Assembly Fluid Flow
FIGURE 18.28 In a PWR, the pressure in the primary loop is about twice the pressure in the secondary loop. The pressure drop
across the primary loop is about 120 PSI (~0.8 mPa), and about half of this occurs across the steam generators. The pressure drop
in the secondary loop is about 900 PSI (~6 mPa), and most of this occurs across the steam turbines. Between the entrance and the
exit to the turbines, the pressure drops dramatically, and the steam becomes very wet. The pressure in the condenser is usually held
below atmospheric pressure (100 kPa).
60% of what it is in a PWR having the same height. Here, H is defined as the total core height. Finally, we would like to
mention that these numbers can sometimes be deceptive because we have not included the frictional losses in the pres-
sure vessel, the downcomer, the piping system, the coolant pumps, and the steam generators. When we do so, we find
that the total pressure drop in the primary loop of a commercial PWR is about 120 PSI (during start-up) and about 85 PSI
at 315°C and 2,200 PSI. This corresponds to a pressure head through one of the reactor coolant pumps of about 85 m.
Hence, a pressure head of 85 m corresponds to a pressure drop of about 140 PSI at room temperature. Thus, only about
one-quarter of the total pressure drop in the primary loop occurs in the core. The rest occurs in the steam generators,
the coolant pumps, the piping system, and the rest of the primary loop. In most PWRs, there is also a similar pressure
drop through the secondary loop. The system-wide pressure drops in a commercial PWR are shown in Figure 18.28. The
pressure drops shown in the figure represent industry averages, and the actual values may be higher or lower.
18.26
L oss Coefficients in the Reactor Piping System
Previously, we discussed the loss coefficients for grid spacers, rod spacers, wire-wrap spacers, and other geometrical
obstructions associated with flow area changes in reactor fuel assemblies and orifice plates. However, for highly tur-
bulent flows (corresponding to those with very large values of the Reynolds number), many components of the reactor
piping system have predictable loss coefficients that can be found in the literature. A brief listing of these coefficients
18.27 Effect of Surface Roughness on the Drag Coefficient 721
TABLE 18.5
Loss Coefficients and Pressure Drops for Different Reactor Piping System Configurations
was presented in Table 18.2. However, a more comprehensive listing can be found in the Idelchik compilation, which is
not restricted to just nuclear power plants, and can be used for coal-fired power plants as well:
☉☉ The Handbook of Hydraulic Resistance, second edition, by I.E. Idelchik, from Hemisphere Press (1986), New York.
Other compilations of loss coefficients suitable for specialized valve and flow blockage work are shown in Table 18.5.
This table also presents a graphical description of the conditions under which these loss coefficients apply.
18.27
Effect of Surface Roughness on the Drag Coefficient
Normally, there is little motivation to increase the surface roughness of a fuel rod except to increase the heat transfer rate.
This is because thermal stresses increase dramatically when grooves are machined into the sides of the rod, and this may
cause the rod to fail earlier than it would if it were smooth. However, there are sometimes advantages to using “rough”
tubes in steam generators because this increases the heat transfer rate between the inside and the outside of the pressure
tubes. However, it also increases the drag forces on the tubes, and these drag forces may induce unwanted vibrations if
they are not properly controlled. Normally, the surface roughness for the tubes in a steam generator tube bank is expressed
722 Core and Fuel Assembly Fluid Flow
as the ratio of the height of the surface augmentation ε to the diameter of the rod D. This quantity is called the rough-
ness ratio (ε/D), and it appears in many reference tables as a dimensionless number. Not surprisingly, the drag coefficient
CDRAG is also a function of the surface roughness ratio. Consider, for example, the external flow of fluid over a cylindrical
tube of diameter D perpendicular to the axis of the tube. These flows are commonly found in reactor heat exchangers, such
as the condenser in the secondary loop of a PWR (see Chapter 8). Coolant flows perpendicular to the tube bank in these
devices are also called cross-flows. The characteristic length to be used in the calculation of the Reynolds number is the
external diameter D of the tubes. Thus, the Reynolds number for flow around an individual tube is defined by
Re = ρvD/µ (18.63)
where v is the uniform velocity of the fluid as it approaches the tube. The critical Reynolds number for flow around a
circular tube is about 200,000. Below this number, the flow remains laminar, and above it, the flow becomes turbulent.
The cross-flow over a cylindrical tube can have many complex flow patterns, and some of them are illustrated in Figure
18.29. The fluid approaching the cylinder encircles the cylinder and forms a boundary layer that wraps around the front
side of it. At low Reynolds numbers, this boundary layer stays intact and the drag coefficient remains low. However,
at high Reynolds numbers, the boundary layer breaks apart, and flow separation occurs. When the boundary layer is
destabilized, the drag coefficient actually decreases. Nevertheless, the drag force still continues to increase because it
is proportional to the square of the fluid velocity, and this increase in fluid velocity normally offsets the decrease in the
drag coefficient. Above a Reynolds number of about 1,000, the drag coefficient stays constant until the Reynolds number
exceeds 100,000. Then, it drops dramatically (usually at about Re = 200,000). Normally, this large reduction in the drag
coefficient is due to the boundary layer becoming turbulent. In this range, even the drag force FDRAG decreases as the
velocity increases, and in the case of an aircraft wing, this may even cause a flying object to fall. Because of this sudden
reduction in the drag coefficient when the flow becomes turbulent, surface roughness will not always increase the drag
force on a heat exchanger tube. For these reasons and others, care must be taken to keep the Reynolds number for the tube
bank either well below the critical Reynolds number of 200,000 or well above it. If the Reynolds number in a reactor heat
exchanger approaches this number, the drag coefficient can quickly change by a factor of 10 or more, and this can cause a
lot of unwanted vibration to develop. This is just one example of how important the Reynolds number can be in reducing
the vibration in a heat exchanger tube. For other geometric shapes, the drag force can be found from
FDRAG = CDRAG A ( 1
2 )
ρv 2 2 (18.64)
where the area A for a cylindrical tube is A = LD. Here, D is the apparent frontal “width” of the tube, and L is its length
or “height.” Later on, we will find that a similar expression can be used to find the drag force on an obstruction in a cool-
ant channel. In this case, A represents the frontal area that the obstruction presents to the flow field. Finally, consider
the effect of surface roughness on the drag coefficient for a spherical particle, such as a golf ball (see Figure 18.30). At
a Reynolds number of 100,000 where the flow is still laminar, CDRAG = 0.5 for a smooth sphere and CDRAG = 0.1 for a
FIGURE 18.29 The drag coefficient CDRAG for cross-flow perpendicular to the axis of a long smooth cylinder is a function of the
outer diameter of the cylinder, and the diameter is also used for finding the Reynolds number (see Equation 18.63).
18.28 FRICTIONAL FORCES ON NUCLEAR FUEL RODS 723
FIGURE 18.30 The drag coefficient for a sphere is also a function of the Reynolds number. However, it is strongly dependent on
the average surface roughness (ε/D).
sphere with a surface roughness ratio of (ε/D) = 0.0015. In other words, making the surface rougher reduces the drag
coefficient by a factor of 5 in this case! However, as soon as the flow becomes turbulent at Re = 200,000 the situation
changes. For a Reynolds number of 1,000,000 and the same surface roughness factor, the drag coefficient is now 0.4 for
the rough surface and 0.1 for the smooth one. In other words, making the surface rougher increases the drag coefficient
by a factor of 4 at Re = 1,000,000!
18.28
Frictional Forces on Nuclear Fuel Rods and Other Support Structures
When a reactor coolant moves past the surface of a pipe or a nuclear fuel rod, it exerts a drag force on the object. This
drag force F DRAG is exerted parallel to the direction of the flow and it is a function of the Reynolds number Re. The
amount of drag depends on the momentum of the fluid, its density ρ, and its dynamic viscosity μ. Using Newton’s laws
of motion, it can be shown that the magnitude of this force is given by
( )
FDRAG = CDRAG A ρv 2 2 (18.65)
where A is the surface area of the object the fluid touches, v is the average velocity of the fluid, and CDRAG is a
dimensionless parameter called the drag coefficient. Normally, the drag coefficient has a value less than 1.0. In some
textbooks, the drag coefficient is also called the friction coefficient, but the friction coefficient should not be confused
with the friction factor. The drag force is normally measured in Newton. For blunt objects such as flow blockages,
orifice plates, and grid spacers, the surface area A can be interpreted to be the area projected onto a plane normal to
the primary direction of flow. In other words, for flow blockages, it is the frontal area that obstructs the coolant flow.
Various empirical correlations have been developed for the drag coefficient CDRAG. These correlations are normally
a function of the shape of the object and the Reynolds number. The drag coefficient may also depend on the rough-
ness of the object’s surface. Fundamentally speaking, the drag force is the net force exerted by a moving fluid on
an object in the direction of flow due to the combined effects of pressure and wall shear. As shown in Figure 18.31,
the total drag force experienced by an object is equal to the sum of (1) the frictional drag force and (2) the pres-
sure drag force exerted on the object. In this case, the drag coefficient can be considered to consist of two separate
drag coefficients—one drag coefficient CFRICTION that is due to wall shear stress or friction, and the other coefficient
CPRESSURE that is due to the pressure drop created by the acceleration (or deceleration) of fluid around the object. The
total drag coefficient is then
FIGURE 18.31 The total drag force experienced by any object is the sum of the frictional drag force and the pressure drag force.
Normally, the sum of these two forces is minimized to obtain the best overall design.
FIGURE 18.32 The drag coefficient for a cylindrical object is a function of its aspect ratio. It is 1.0 for a round circular cylinder
and drops to about one-quarter of that value when the aspect ratio becomes large. CDRAG is based on the frontal area of the cylinder
where A = WD, D is its diameter, and W is its width.
Here, the value of CDRAG is also a function of the aspect ratio (see Figure 18.32). At low Reynolds numbers, most of CDRAG
is the result of frictional drag on the object. Since the frictional drag is proportional to the surface area A, objects with
large surface areas tend to have more drag than objects with smaller surface areas. The drag coefficient is independent
of the surface roughness for laminar flows, but it is a strong function of the surface roughness for turbulent flows. Not
surprisingly, rough surfaces tend to break apart the turbulent boundary layer, and the breakup of the boundary layer
transfers more momentum to the surface. The pressure drop which also contributes to the total drag is due to a pres-
sure difference between the front and the back of the object over which the fluid is flowing. Therefore, the pressure
18.28 FRICTIONAL FORCES ON NUCLEAR FUEL RODS 725
FIGURE 18.33 The frictional drag coefficient CFRICTION is a function of axial position for a plate-type fuel rod. In this case, the
pressure drag coefficient CPRESSURE is zero because the fluid is flowing parallel to the surface of the rod.
coefficient is usually largest for blunt objects, and it is normally smallest for streamlined objects such as nuclear fuel
rods and aircraft wings. The drag coefficients for laminar and turbulent flow over a plate-type fuel rod (see Figure 18.33)
are given by
where x is the distance from the leading edge of the plate, and Re(x) = ρvx/μ is the local Reynolds number at that location.
In general, the drag coefficient reaches its maximum value when the flow becomes fully turbulent, and it then
decreases as a function of x−0.2 in the flow direction. In reactor thermal-hydraulics, the average drag coefficient is some-
times more useful than the local drag coefficient because it determines the drag force the entire flow field exerts on an
object. For a plate-type fuel rod, the average drag coefficient can be found by integrating CFRICTION(x) over the length of
the rod and dividing by the overall length:
L
CFRICTION =
∫ 0
CFRICTION (x) dx/L (18.69)
since the pressure drag coefficient is zero in this case. If the flow is laminar, the integration is performed using Equation
18.67, and if the flow is turbulent, the integration is performed using Equation 18.68. The average drag coefficients are then
In many applications, a plate-type fuel rod is long enough for the flow to become turbulent, but not long enough to neglect
fluid friction in the laminar region. Then, the average drag coefficient can be found by performing the integration:
L
CFRICTION =
∫ 0
CFRICTION (x) dx/L (18.72)
726 Core and Fuel Assembly Fluid Flow
over both flow regimes (both laminar and turbulent). Thus, the integration must be performed in two parts: Part 1 in the
laminar region from x = 0 to x = xCR, and Part 2 in the turbulent region from x = xCR to x = L.
x CR L
CFRICTION = (1/L)
∫
0
CFRICTION (x) dx +
∫ x CR
CFRICTION (x) dx (18.73)
Here, xCR is the critical distance where the transition from laminar to turbulent flow occurs. Using Equations 18.67 and
18.68 for CFRICTION, we find that the average drag coefficient for the entire plate is
The first term represents the average drag coefficient for the turbulent zone, and the second term represents the aver-
age drag coefficient for the laminar zone. Hence, when the flow is part laminar and part turbulent, the sum of these
two terms must be used to obtain the total drag coefficient for the rod. For a plate-type fuel rod, the transition between
laminar and turbulent flow takes place at a Reynolds number of about 500,000 when the Reynolds number is defined by
Re(x) = ρvx/μ. Note that this definition is different than the conventional definition of the Reynolds number for a reactor
coolant channel because the hydraulic diameter D is replaced by x. Hence, one must be careful to use the correct defini-
tion for the Reynolds number in order to obtain the correct result. In Chapter 22, we will find that a similar definition for
the Reynolds number is required to find the convective heat transfer coefficient for natural convective flows. This then
leads the reader to another parameter called the Grashof number, which like the Reynolds number is another important
dimensionless number that appears in the field of fluid mechanics.
18.29
Frictional Drag versus Pressure Drag
The structural components inside a nuclear power plant are subject to both frictional drag and pressure drag. The fric-
tional drag force exerted by a moving fluid on a stationary object such as a nuclear fuel rod is given by
(
FFRICTION = CFRICTION A ρv 2 2 (18.75) )
FIGURE 18.34 The drag coefficients for parallel flow over plate-type fuel rods that have both smooth and rough surfaces.
18.30 The Drag Force on a Nuclear Fuel Rod 727
where CFRICTION is the drag coefficient due to the friction of the fluid against the surface of the rod. The values of CDRAG
for a typical fuel rod are shown in Figure 18.34. If there is a pressure difference between the front and the back of the
rod, the rod may also experience an additional drag force called a pressure drag force FPRESSURE. The pressure drag force
on the rod is the drag force due to a pressure difference between the front and the back of the rod. The pressure drag
force is defined by
( )
FPRESSURE = CPRESSURE A ρv 2 2 (18.76)
and its magnitude depends on another coefficient called the pressure drag coefficient CPRESSURE. For flow parallel to the
surface of a fuel rod, we can normally neglect the pressure drag coefficient because it is many times smaller than the
friction drag coefficient:
For flow parallel to a fuel rod or other surface: CFRICTION CPRESSURE (18.77)
However, for flow perpendicular to the surface of a fuel rod, the two coefficients can have similar m
agnitudes when the
Reynolds number is very high. In this case, we must write
For flow perpendicular to a fuel rod or other surface: CFRICTION ≈ CPRESSURE (18.78)
In other words, the total drag force F DRAG on the rod in this case is
( ) ( )
FDRAG = FFRICTION + FPRESSURE = CFRICTION A ρv 2 2 + CPRESSURE A ρv 2 2 (18.79)
and for fuel rods inside of fuel assemblies, we can normally neglect the value of CPRESSURE because it is so small.
Form drag or pressure drag becomes important when a moving fluid encounters a blunt object or an obstruction in
its path. The object causes the fluid to wrap around the object if it is to continue to flow. Two classic examples of
objects of this type are the grid spacers and support plates. For these objects, the value of CPRESSURE becomes sig-
nificant when the velocity of the fluid becomes too high for the fluid to follow the curvature of the surface. In this
case, the fluid begins to separate from the surface (instead of wrapping around it), and this flow separation creates a
low-pressure region to form behind the object. The pressure difference between the front and the back of the object
due to this flow separation can cause the pressure drag coefficient CPRESSURE to become very large. Thus, the pressure
drag becomes proportional to the frontal area of the object and the fluid pressure difference ΔP between the front
and the back of the object. For large objects like a steam generator orifice plate or the core support plate, the value
of CPRESSURE can become very large. Thus, the drag force F PRESSURE created by this pressure difference becomes very
large as well.
Normally, airplane wings and other streamlined objects do not have this problem because they are designed so that
the pressure drag coefficient CPRESSURE is as small as possible. In other words, streamlining effects the frictional and
pressure drag coefficients in opposite ways. Streamlining decreases the value of CPRESSURE by slowing the boundary
layer separation, but it increases the frictional drag by increasing the surface area. The variation of the frictional drag
coefficients, the pressure drag coefficients, and the total drag coefficients are shown in Figure 18.31 for an aircraft
wing. Therefore, minimizing the total drag on the wing requires optimizing these two competing effects. When the
total drag force on a structural component in a reactor becomes a concern, a similar process is followed to reduce the
drag force. In general, streamlining an object by adjusting its aspect ratio is the best strategy to apply. If the object is
elliptical in shape, the drag coefficient can be controlled by adjusting the aspect ratio. The variation of the drag coef-
ficient CDRAG = CFRICTION + CPRESSURE for a long elliptical cylinder is shown in Figure 18.32. Here, CDRAG is based on the
frontal area WD of the object, and W is its width. Notice that streamlining the object can reduce the total drag coef-
ficient CDRAG = CFRICTION + CPRESSURE by a factor of 5 or more. The pressure drop across the core support plate can also
be reduced by beveling or rounding the edges of the holes in the plate.
18.30
T he Drag Force on a Nuclear Fuel Rod
Now let us see if we can use these equations to determine the size of the drag force on a nuclear fuel rod. To perform this
exercise, we need to know the values of the drag coefficients for flow parallel to a nuclear fuel rod and perpendicular to
it. For flows perpendicular to the rod, the drag c oefficient changes with the Reynolds number, and it is also a function
728 Core and Fuel Assembly Fluid Flow
of the length-to-diameter ratio (L/D). For large values of (L/D), CDRAG is about 6 when Re = 1, and 2 when Re = 10. The
Reynolds number in this case is found using the diameter D of the fuel rod:
Re = ρvD/µ (18.80)
For larger values of the Reynolds numbers (Re > 1,000), the drag coefficient falls to about 1.0, and it stays at that value
until the Reynolds number becomes greater than 500,000. The drag coefficients for perpendicular flows, which are
sometimes called cross-flows, are shown in Figure 18.29. These drag coefficients can be used as long as the coolant
does not boil. Now let us calculate the drag force when the coolant is flowing parallel to the surface of the rod. In this
case, the drag force varies with axial position z, and the drag coefficient CDRAG is a function of axial position as well.
The reader may recall from our previous discussion that the frictional force F per unit area which the fluid exerts on the
fuel rod is called the wall shear stress τ = F/A. The shear stress depends on the velocity gradient dv/dy at the fuel rod’s
surface. Hence, the shear stress τ is given by
where A is the surface area of the rod. In 1984, a British engineer by the name of R.M. Luxton found that the average
shear stress along the surface of a long cylindrical fuel rod was given by
where L is the rod’s length. Notice that the drag force increases with the fluid velocity v or the mass flow rate m′. Since
Re = ρvD/μ, and A = πD, we can also write the total drag force as
At 2,250 PSI and T = 300°C, ρ ≈ 714 kg/m3, μ ≈ 0.000086 kg/ms. Since the diameter of a typical fuel rod is about 1 cm
(0.01 m) and the average length is about 5 m, a practical expression for the total drag force is
where v is the velocity measured in m/s. At 6 m/s, the drag force on a single fuel rod is about 85 N. Since a 17 × 17
PWR fuel assembly has almost 300 fuel rods, the upward force exerted on the fuel assembly by the coolant is at least
F DRAG = 300 × 85 = 25,000 N. This is a considerable amount of force for the coolant to exert on just 300 smooth fuel
rods. If we include the additional drag forces exerted on the grid spacers and the fittings at the inlet and the outlet, we
would find that the force of the coolant flowing through the core is sufficient to lift the entire fuel assembly out of its
mounting cradle! For this reason, reactor fuel assemblies are normally held in place with high-tension springs to keep
them from being lifted out of their cradles at very high flow rates.
18.31
T he Drag Force on a Plate-Type Fuel Rod
The drag coefficients for plate-type fuel rods such as those shown in Figure 18.35 are also available in tabular or graphi-
cal form. When the fluid flows parallel to the surface of a plate-type fuel rod, the frictional drag force exerted by the
fluid on the rod can be significant, but the pressure drag coefficient CPRESSURE is usually zero. When the Reynolds number
is defined using the length L of the rod, the flow becomes turbulent when the Reynolds number reaches approximately
Re = 5 × 105. When the flow is turbulent, the average drag coefficient on the entire rod is
where Re = ρvL/μ. This equation for the drag coefficient is based on the assumption that the surface of the rod is smooth.
When the surface is rough, the average drag coefficient for turbulent flow is
CDRAG = (1.89 − 1.62 log(RR))−2.5 (fully developed rough turbulent flow) (18.87)
where RR = ε/L is the relative roughness of the surface of the plate. Equation 18.87 is based on the assumption that the
Reynolds number is defined using the rod’s length L instead of its width. Hence, for the same coolant velocity, long fuel
rods will have higher Reynolds numbers than short fuel rods. Sometimes Equation 18.87 is called the Schlichting cor-
relation. The Schlichting correlation can be used for turbulent flow over a rough flat surface when Re > 106, and in par-
ticular, when the relative roughness exceeds 0.0001. The drag coefficients for parallel flow over both smooth and rough
plates are shown as a function of the Reynolds number in Figure 18.34 for both laminar and turbulent flow. Notice that
the drag coefficient increases as the rod becomes rougher in the turbulent flow regime. Moreover, as the plate becomes
very rough and the flow becomes highly turbulent, the drag coefficient becomes independent of the Reynolds number.
The region of the figure in which this occurs is called the fully rough region. Thus, Figure 18.34 can be thought of as the
flat-plate version of the Moody chart for turbulent flow in rough circular pipes. In any event, drag forces on the fuel rods
must be taken very seriously when the flow rate becomes high.
18.32
Noise and Vibration
The effects of friction must also be taken into account to ensure that the piping system is structurally sound. A final
example of this effect is the drag force exerted on a long elliptical cylinder of height D and width L. The flow field in
FIGURE 18.35 Plate-type fuel rods are still used in reactors today. Normally, these rods are reserved for low power research and
test reactors although this does not always have to be the case.
730 Core and Fuel Assembly Fluid Flow
this case is perpendicular to the axis of the cylinder. The drag coefficient is shown in Figure 18.32. Notice that when
the elliptical cylinder becomes a circular one, the drag coefficient becomes equal to 1.0. The results shown are for a
Reynolds number of 100,000. As L/D becomes smaller and the cylinder becomes more streamlined, the drag force
becomes smaller too. In some designs, streamlining the shape of cylindrical tubes by modifying their aspect ratio can
reduce the amount of noise and vibration that they can produce. Again, the effects of fluid friction play an important role
in determining the amount of vibration that can occur. These forces can also affect the behavior of pumps and turbines,
which we will discuss in the next chapter.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi, India (2000).
Reactor Theory (Nuclear Parameters), DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
Other References
Luxton, R.M., Bull, M.K. and Rajagopalan, S. “The thick turbulent boundary layer on a long fine cylinder in axial flow”,
The Aeronautical Journal, 88:186–199, 1984.
Tutty, O. “Flow along a long thin cylinder” Journal of Fluid Mechanics, 602:1–37, 2008, Cambridge University Press.
39. Can this cross-flow also occur between adjacent fuel assemblies? If so, in which direction does the cross-flow nor-
mally go—from a hot fuel assembly to a cold fuel assembly or from a cold fuel assembly to a hot fuel assembly?
How do you think this affects the peak fuel pin temperatures?
40. For flow across a circular tube, at what Reynolds number does the flow become turbulent? What is the value of this
Reynolds number called?
41. Does surface roughness have more effect on the drag coefficient across a nuclear fuel rod when the flow is laminar
or when the flow is turbulent? What is the physical reason for this difference?
42. What is the average value of the void fraction in a BWR core?
43. Is the gravitational pressure drop in a BWR core greater than, less than, or the same as it is in a PWR core?
4 4. If the friction factor is known for a smooth fuel rod, how does one determine the friction factor for a rough fuel
rod?
45. Normally, is there any incentive to roughen the surface of a fuel rod in a PWR or BWR core?
46. What are the two primary contributors to the pressure drop in a typical light water reactor core?
Exercise 18.2
Most PWR fuel assemblies have six to eight grid spacers placed at equal distances along its active length. If the form loss
coefficient for each spacer is 0.5, calculate the total pressure drop due to these spacers alone when the coolant is flowing
through the assembly at 6 m/s. Assume the pressure is 2,250 PSI (15.5 MPa), and the average temperature of the coolant
is 310°C. Why are the spacers spaced at equal distances from each other?
Exercise 18.3
The fuel rods in nuclear power plants are usually very smooth, and their relative roughness is close to zero. Suppose
the relative roughness is increased to 0.0050 to increase the heat transfer coefficient. How much is the turbulent friction
factor increased in this case? For the same mass flow rate and cross-sectional area, how much is the pressure drop along
a 5 m-long rod increased?
Exercise 18.4
Water flows through a straight pipe in a reactor piping system. The Reynolds number in the pipe is 1,500. By how
much do we have to decrease the diameter of the pipe for the flow to become turbulent if the mass flow rate is to
remain the same?
Exercise 18.5
A 4 m-long coolant channel in the core of a research reactor has an equivalent diameter of 1 cm. Water enters the chan-
nel from below at a rate of 50 kg/s and is 20°C subcooled. The velocity of the flow is 6 m/s. Ten MW of thermal energy
is added sinusoidally to the channel. The average pressure is 7 MPa. Neglecting the extrapolation lengths discussed in
Chapter 5, calculate the friction and acceleration pressure drops. Assume the fuel rods are completely smooth.
Exercise 18.6
Suppose the radial pressure difference ΔP(z) = (P1(z) − P2(z)) between coolant channels 1 and 2 in a PWR fuel assem-
bly at an axial elevation of z = 2 m is 0.02 kPa. If the cross-flow coupling coefficient between these coolant channels is
0.02 kg/cm2 Pa s, and the P/D ratio of the lattice is 1.3, what is the cross-flow rate across an axial segment 5 cm high?
Assume the diameter of the fuel rods in 1 cm.
Exercises for the Student 733
Exercise 18.7
Hot water flows from several coolant pipes at 2,250 PSI (15.5 MPa) and 290°C into a single outlet plenum at the bottom
of the reactor pressure vessel. The pipes have very smooth edges and are connected to the outlet plenum using a large
flat plate. If the velocity of the water inside of the pipes is 2 m/s, what will be the average pressure drop between the exit
of the pipes and the plenum?
Exercise 18.8
Calculate the drag force due to fluid friction alone on a PWR fuel assembly consisting of 225 individual fuel rods that
are 6 m long and have an outer diameter of 1 cm. If the fuel assembly weighs 750 kg, is the upward force strong enough
to lift the fuel assembly out of its cradle if the velocity of the coolant is 10 m/s?
Exercise 18.9
A 1 m-long coolant channel in a mobile power reactor has a cross-sectional area of 10 cm2. It receives heat uniformly
from the sides of the channel at a rate of 1 × 106 W/m2. The average channel pressure is 10 MPa. Water enters the channel
as a subcooled liquid at 1 m/s. The water is 20°C subcooled at the inlet, and the turbulent friction factor is 0.02. Calculate
the friction and acceleration pressure drops between the start and the end of the channel.
Exercise 18.10
Water flows over a steel plate in a reactor steam generator at the rate of 5 m/s. The plate is 5 m long and 0.75 m
wide. When the flow is turbulent, the drag coefficient for the water flowing over the plate is C D = 0.455/
(log10 ReL)2.58. Using this equation, calculate the Reynolds number and the total drag force on the plate. Assume
that ν = 0.011 × 10 −4 m 2/s.
19
Reactor Coolants, Coolant
Pumps, and Power Turbines
19.1
Reactor Coolants and Their Properties
Reactors use a wide variety of coolants, but not all coolants are suitable for all reactor types. Reactor coolants can be divided into
three broad categories:
1.
Aqueous coolants (such as light and heavy water)
2.
Metallic coolants (such as liquid sodium, potassium, and mercury)
3.
Gaseous coolants (such as hydrogen, helium, and carbon dioxide)
Each of these coolants has different heat transfer coefficients, Prandtl numbers, and Nusselt numbers, and within each category,
different coolants can have different neutronic characteristics; that is, they can affect the power producing capabilities of the core in
different ways. The most common reactor coolant is ordinary water (or H2O) which is sometimes called light water. Light water is
used in most of the world’s pressurized water reactors (PWRs) and boiling water reactors (BWRs) and its thermodynamic and physical
properties are well known and well understood. Ordinary water contains both heavy water and light water molecules, and these two
molecules have different atomic weights—18 and 20. Conventional PWRs and BWRs use this water to cool the core, and the only
difference is that BWRs allow the coolant to boil vigorously, whereas PWRs do not. Heavy water is a less common variety of ordinary
water that has an additional neutron attached to the hydrogen atoms in the water molecules (see Figure 19.1). These neutrons make
the water molecule heavier, and they also change its propensity to absorb thermal neutrons (i.e., its neutron absorption cross section).
In nature, there are approximately 6,420 molecules of light water for every molecule of heavy water.* Heavy water has an atomic
FIGURE 19.1 A picture illustrating the difference between a light water molecule (left) and a molecule of heavy water (right). The heavy water
molecule is found in ordinary water, and it has an additional neutron attached to the hydrogen nucleus. This additional neutron allows heavy water
to absorb fewer neutrons than ordinary water does, and this difference is important in the design of thermal water reactors. Normally, ordinary
water contains about 6,000 times more molecules of light water than heavy water.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
735
736 Reactor Coolants, Coolant Pumps, and Power Turbines
weight of 20 (16 + 2 + 2), and light water has an atomic weight of 18 (16 + 1 + 1). However, heavy water absorbs fewer
neutrons than light water does. In fact, a heavy water molecule is approximately 500 times less likely to absorb a ther-
mal neutron than a light water molecule is*. This means that reactors cooled with heavy water (such as the Canadian
CANada Deuterium Uranium reactor – see Chapter 2, require less enriched uranium (U-235) than other reactors do. In
fact, it is possible to run heavy water reactors on refined natural uranium alone. This feature of heavy water makes it
particularly attractive to governments who do possess their own uranium enrichment facilities.
Sometimes the hydrogen atoms in the heavy water molecule are called deuterium atoms. The deuterium atoms in the
water molecule are given the chemical symbol D2. In addition to deuterium, ordinary water may also contain some
tritium atoms, which contain two neutrons bound to each hydrogen atom instead of just one. (Tritium is an important
nuclear fuel in fusion reactors). The chemical symbol for tritium is D3. Sometimes tritium is also given the symbol T3.
The chemical symbol for heavy water is D2O.
19.2
Heavy Water Production
Producing heavy water from ordinary water is a difficult process that requires a fair amount of technical infrastructure.
The first successful separation of heavy water from ordinary water was achieved by Nazi Germany during the Second
World War. Heavy water is produced from ordinary water by by first purifying and then distilling ordinary water. This
process requires a great deal of electrical energy, so heavy water production facilities are normally located near hydro-
electric facilities where electric power is cheap and plentiful. Most of these facilities are located in Canada, Norway,
Sweden, and the United States.
19.3
Light and Heavy Water
The distillation process used to create heavy water has many different stages, and at each stage, the lighter water mol-
ecules are separated from the heavier ones by heating them and evaporating the mixture. The lighter molecules are
evaporated away, and over time, only the heavy water molecules remain. However, it takes several thousand stages to
create a small amount of pure heavy water. The most common process for creating heavy water is the Girdler sulfide
process, which is widely used in Canada, Norway, and the United States. As we mentioned earlier, heavy water or D2O
has a natural abundance of about one part in 6,420. The rest of the water in the Earth’s oceans, lakes, and rivers is H2O.
A vial of heavy water is shown in Figure 19.2. Normally, heavy water is not toxic to human beings. It is chemically
similar to ordinary water and looks about the same. Heavy water is about 10% heavier than light water, and its density is
1.10 g/cm3 at 4 °C and 0.99823 gm/cm3 (or 998.23 kg/m3) at 20 °C. The density of ordinary water is 1.00 g/cm3 at 4 °C and
0.99823 gm/cm3 (or 998.23 kg/m3) at 20 °C. Moreover, its thermal–physical properties are almost the same. The boiling
point for heavy water is about 1.3°C higher than it is for ordinary water. Thus, heavy water boils at 101.4°C, and light
water boils at 100°C. This difference in the boiling point is essential to the success of the heavy water production process.
19.4
Factors in Selecting a Reactor Coolant
Compared to heavy water, ordinary water is used in nuclear power plants because of its low cost and high availability,
and because its thermodynamic properties are well known and well understood. The only drawback to ordinary water is
that it cannot be used in fast reactors because it contains too much hydrogen. The hydrogen atoms in the water molecules
reduce the energy of the fission neutrons too quickly and affect the ability of the core to breed. Consequently, fast reac-
tors use metallic coolants such as sodium, potassium, and mercury instead. These coolants are excellent conductors of
heat, and they also have very high boiling points. Near atmospheric pressure, they can easily remove the additional heat
that that fast reactors produce. Hence, they are very attractive from both a thermodynamic and economic perspective.
Particle economy is also a major factor in deciding the best type of coolant to use. CANDU reactors are powered by
natural uranium and cannot use coolants that have high particle absorption rates because there are simply not enough
neutrons to keep the core critical. They are limited to using heavy water because the particle absorption rates are lower*.
Gas-cooled reactors use fuel with a higher enrichment, and they need gases in conjunction with a solid moderator to
function properly. Normally, this moderator is some form of crystallized carbon or graphite. However, they have the
19.4 Factors in Selecting a Reactor Coolant 737
FIGURE 19.2 A vial of heavy water produced by the Norsk hydroelectric plant in Norway. Heavy water must be separated from
light water by a distillation process. This distillation process takes advantage of the fact that boiling point of heavy water is about
1.5° higher than the boiling point of ordinary water. In small quantities, heavy water is not toxic to humans and tastes nearly the
same as natural water does. (Picture provided by Wikipedia.)
advantage that they use the Brayton thermal cycle instead of the Rankine thermal cycle (see Chapter 9), and if the tem-
perature of the gas exiting the core is high enough, the thermal efficiency can exceed that of a comparable water reactor
by several percent. Carbon dioxide (CO2) is the most popular coolant in gas reactors, although other coolants such as
hydrogen and helium can also been used. When selecting a reactor coolant, a number of competing factors must be taken
into account. These factors fall into three broad categories:
1. Economic
2. Physical
3. Nuclear
Now let us discuss each of these categories separately.
19.4.1
E conomic Factors
Economic factors include
19.4.2
Physical Factors
The physical characteristics of the coolant are also important when selecting the appropriate coolant to use. The most
important physical factors are
738 Reactor Coolants, Coolant Pumps, and Power Turbines
19.4.3
Nuclear Factors
1. The particle absorption rate of the coolant
2. The radiation stability of the coolant
3. The neutron economy of the coolant
4. The moderating power of the coolant, which is important in thermal reactors
Coolants with low neutron absorption rates (and therefore good particle economy) are preferable to those with poor particle
economy. In this respect, heavy water is a very attractive coolant. Needless to say, no single reactor coolant possesses an
optimum combination of all these characteristics. Hence, the selection of a coolant becomes an engineering compromise
between several different and often c ompeting requirements. The characteristics of the most common reactor coolants
are shown in Table 19.1. Clearly, water has a number of advantages over most of the other coolants. The thermophysical
properties of each coolant are shown in Table 19.2. The properties shown here are evaluated at typical operating conditions.
19.5
Reactor Coolants Used in Different Reactor Types
A suitable reactor coolant is one that has a low neutron absorption rate, a low pumping power, a high heat transfer
coefficient, good thermal stability, good natural availability, and low initial cost. For many reasons, fast reactors use
different coolants than thermal reactors do.
TABLE 19.1
A Comparison of the Characteristics of Some Common Reactor Coolants
TABLE 19.2
The Thermo-physical Properties of Some Common Reactor Coolants in Use Today
19.5.1
Water Reactor Coolants
Light water and heavy water are used primarily in thermal water reactors. Light water is the same as ordinary water,
but in nuclear applications, it is more highly purified (so that it does not corrode the reactor piping s ystem). PWRs and
BWRs use light water, whereas Canadian PHWRs (also called CANDU reactors) use heavy water. The physical proper-
ties of light water and heavy water are similar in many respects (except for their molecular weights), but their nuclear
properties are considerably different. Heavy water absorbs fewer neutrons than light water does. It also takes longer to
slow down a fission neutron from high energies to low energies (i.e., from fast to thermal). Hence, heavy water reactors
tend to be larger than light water reactors do.
19.5.2
G as Reactor Coolants
Gas reactors are cooled by industrial gases such as helium (He) and carbon dioxide (CO2). Most gas reactors use the
Rankine thermal cycle, but they can also use the Brayton cycle, which is discussed in Chapter 9. In England and most
of Western Europe, gas reactors are called advanced gas reactors or AGRs for short. The core inlet temperatures range
from 400°C to 500°C, and the core outlet temperatures range from 700°C to 850°C. Outside of the United States, most
gas reactors are cooled with carbon dioxide (CO2). In the United States, gas reactor technology has been centered around
the development of the high-temperature gas reactor or HTGR. It is a graphite-moderated thermal reactor. HTGRs are
cooled with helium, and in general, helium is an excellent reactor coolant. It is far more inert than CO2, and it does not
absorb as many thermal neutrons. Hence, it does not become nearly as radioactive. Typical HTGRs have a core inlet
temperature of about 750°C and a core outlet temperature of about 850°C. The primary provider of HTGRs in the
United States is the General Atomics Company, which is located in San Diego, California. However, in spite of its appar-
ent technical advantages, HTGRs evidently do not produce electricity as cheaply as PWRs do, and while much interest
in them remains, their worldwide market share continues to be relatively small.
19.5.3
Fast Reactor Coolants
Fast reactors are different than thermal reactors are because they designed to create heat based on the use of fast neu-
trons alone. Technically speaking, fast reactors do not need a moderator, and so the core and the blanket contain only
three types of materials: the fuel, the structure, and the coolant (to take away the heat). In the early days of the nuclear
industry, a number of coolants were proposed to cool fast r eactors. Then, over a period of time, it became apparent that
metallic coolants were the best coolants to use. These included sodium, potassium, and mercury. However, over a period
of time, liquid sodium became the coolant of choice. Today, virtually all fast reactors use liquid sodium to cool the core,
and these reactors are referred to as liquid metal fast breeder reactors or LMFBRs for short. Sometimes they are also
called sodium cooled fast reactors or SCFRs. Most fast reactors have three coolant loops—a primary loop, a secondary
loop, and a tertiary loop.
The primary loop pumps the coolant though the core. The secondary loop contains an intermediate heat exchanger
(or IHX) to isolate the radioactive sodium flowing through the core from the other components of the NSSS. Heat is
transferred from the secondary loop to an additional tertiary loop through a steam generator. The tertiary loop uses
water as its working fluid (See Figure 19.3). The high-temperature sodium converts the water into steam, and the
steam is used to drive the blades of a steam turbine. The steam turbine is connected to an electric generator which
then produces electric power. Metallic coolants have many positive attributes compared to other coolants, but they
also have a few negative ones. They are excellent heat conductors, and so LMFBRs that use them can be operated
at very high power densities. This in turn means that LMFBR cores can be relatively small compared to their water
reactor counterparts. Liquid sodium also has a very high boiling point (about 880°C at atmospheric pressure), and so
reactor coolant loops can be operated at very high temperatures and very low pressures (LPs). This means that large,
heavy, and expensive pressure vessels are not required to hold the coolant. Sodium is also not as corrosive as water,
and so components immersed in liquid sodium can look almost new after many years. The biggest drawback to liquid
sodium is that it is highly reactive chemically. It reacts violently with water, and catches fire when it comes in contact
with ordinary air. A fire of this type is called a sodium fire. Next, it has a very high melting point (about 100°C), and
so the entire cooling system must be kept heated at all times to prevent the sodium from solidifying. Finally, liquid
sodium absorbs fast neutrons more easily than water, and this can cause it to become radioactive. Therefore, almost
all LMFBRs use two sodium loops—a primary loop containing the radioactive sodium from the core and an inter-
mediate loop containing sodium that is not radioactive which can be passed to a steam generator. The intermediate
loop contains an IHX to transfer heat from the core to the power turbines. A picture of sodium in its liquid and solid
states is shown in Figure 19.4. Again, the melting point of sodium at atmospheric pressure is approximately 98°C.
740 Reactor Coolants, Coolant Pumps, and Power Turbines
FIGURE 19.3 Virtually all fast reactors cooled with liquid metals use an intermediate loop with an IHX to separate the radioactive
sodium in the core from the other components of the NSSS. While there is no thermal-hydraulic reason to use an IHX, there are a
number of safety-related reasons to do so.
19.6
Physical Properties of Reactor Coolants
The most important thermophysical properties of reactor coolants are
1. Their density ρ
2. Their viscosity μ
3. Their thermal conductivity k
4. Their specific heat c
Their boiling points, their freezing points (in the case of metallic coolants) and their operating pressures are also very
important. The properties of interest to most engineers are shown in Table 19.2. Clearly, water has a very desirable set
of characteristics when all of its thermophysical properties are taken into account. This, plus its widespread natural
FIGURE 19.4 A picture of sodium metal which is used in fast reactors in both its solid and liquid states. (Pictures provided by
Wikipedia.)
19.7 Reactor Coolants and Representative Flow Rates 741
abundance and availability, explains why it is so popular in nuclear reactors today. Its neutronic characteristics are
discussed in our companion book.*
19.7
Reactor Coolants and Representative Flow Rates
In most reactors, the coolant flows through the core at very high speeds, which can often approach 5–10 m/s. The flow
of the coolant can be expressed in terms of
1.
The mass flow rate m = ρvA (which is usually measured in kg/s)
The mass flux G = m
2. /A (which is usually measured in kg/m2 s)
Now let us take a look at the values of these parameters for typical PWRs, BWRs, and LMFBRs. The coolant is circu-
lated through the NSSS by reactor coolant pumps or RCPs. Normally, there is a separate coolant pump for each leg of
the coolant system (see Chapter 2). A cross-sectional view of one of these pumps is shown in Figure 19.5. LMFBRs and
gas reactors require different coolant velocities and mass flow rates than PWRs and BWRs. However, in each instance,
the flow is designed to be as turbulent as possible, and local Reynolds numbers between 50,000 and 100,000 are quite
common in most of the NSSS. In fact, the local Reynolds numbers can reach between 500,000 and 600,000 in com-
mercial PWR cores. These Reynolds numbers are usually associated with the individual subchannels inside of the fuel
assemblies (see Chapter 18). Now let us turn our attention to the specific mass flow rates and coolant velocities on a plant
by plant basis. These parameters are summarized in Table 19.3. In large PWRs, the individual coolant pumps have
mass flow rates between 5,000 kg/s and 7,500 kg/s. Three-loop PWRs normally use three 7,500 kg/s coolant pumps,
whereas four-loop PWRs (with the same thermal output) normally use four 5,000 kg/s pumps. Thus, the mass flow
rate through a 3,400 MW thermal PWR core is typically 18,000–22,000 kg/s. (Of course, the exact numbers depend
on the core power level.) The coolant velocity through the core is 5–7 m/s. Thus, it only takes 5–6 s for the coolant
FIGURE 19.5 The cross-sectional view of a RCP from a Westinghouse PWR. (Picture provided by Westinghouse Nuclear.)
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
742 Reactor Coolants, Coolant Pumps, and Power Turbines
TABLE 19.3
The Flow Rates Shown Here Are Typical for a 3,400 MWT Plant
flowing through the core to travel the length of an entire NFL football field (from one goal post to the other—see
Figure 19.6). By comparison, the fastest player in the NFL takes about 10 seconds to cover the same distance.
Clearly, a lot of coolant must be pumped through the core to remove such large amounts of heat.
For large BWRs with the same thermal output, the core mass flow rates are a bit lower (typically 12,000–15,000
kg/s), and the power densities are lower as well (~55 vs. 110 kW/L). The water in the upper half of the core is designed to
boil vigorously, and the exit quality can sometimes approach 15%. The average velocity of the coolant is about 2.5 m/s,
and the coolant above the core midplane is a water–steam mixture. The core average void fraction is about 40%. (It is
zero at the core entrance and 80%–90% at the core exit.) By comparison, 1,000 MWE LMFBRs have core mass flow
rates between 12,000 and 18,000 kg/s, and they can have between two and eight primary coolant pumps. (LMFBRs have
much greater variation in their design parameters than water reactors do.) Finally, a large HTGR can have flow rates of
about 350 kg/s (since the gas is not very dense). However, the pumping power requirements are much greater than those
for water reactors because the gas has to be recirculated through a compressor when the Brayton thermal cycle is used.
If a large HTGR uses this cycle, the pumps, compressors, and recirculators can consume up to one-third of the total
power output of the plant! Hence, although HTGRs can have greater thermal efficiencies than water reactors, more of
the power is required to keep the coolant flowing. The mass flow rates, the number of coolant pumps, and the thermal
efficiencies are compared in Table 19.3. Again, these numbers are industry averages, and some variation from the num-
bers quoted can be expected to occur in practice.
FIGURE 19.6 Comparing the velocity of the coolant in a PWR core to the speed of the fastest football player in the NFL.
19.8 Work Performed by a RCP (Pump Work) 743
19.8
Work Performed by a RCP (Pump Work)
The amount of energy required to set the coolant in motion can be expressed in terms of a parameter called the pump
work. The pump work is essentially the force acting on a volume of fluid times the distance over which the force is
applied. Hence, pump work, like conventional work, is expressed in units of force times distance (f × D). In the SI unit
system, the energy expenditure of a pump is expressed in Newton meters (N m) or joules (J), where
1 J = 1 N/m (19.1)
The amount of work performed per second is then called the pumping power of the pump. The pumping power is
, where W
represented by the symbol W = dW/dt. Pump work is measured in J/s or W. Because RCPs are so large, the
pumping power required to turn the impellers inside of the pumps is measured in megawatts rather than watts, where
1 MW = 1 × 106 W. In coolant pumps, the rotational motion of the blades sets the coolant in motion, and the work W
expended by the pump is equal to the change in the energy of the coolant:
Here, the change in coolant energy includes any changes to its potential energy, its kinetic energy, and its internal
energy. If the coolant is incompressible, then the pump work required to set the coolant in motion becomes
where h is the specific enthalpy of the coolant, and u is its specific internal energy. Moreover, we know from Chapter 16
that
E = ρh + 1 2 ρv 2 + ρgz (19.4)
where
h = u + PV (19.5)
These equations apply to frictionless fluids or those that are almost so. However, as soon as the effects of friction are
included, a more comprehensive definition of the pump work is
where WSHAFT is the shaft work, W PUMP is the useful output of the pump, and W FRICTION is the sum of the frictional losses
in the fluid and the pump’s internal components. The frictional work W FRICTION becomes heat, and this lost mechanical
energy manifests itself as an increase in the temperature of the coolant. We can estimate the size of the temperature
increase by assuming that the mechanical energy loss ELOST is the same as the frictional energy loss W FRICTION. We can
then write
or
where we have equated the actual work output of the pump WPUMP to the mechanical energy transferred to the coolant,
that is, W PUMP = WCOOLANT. To illustrate why this is important, suppose that 14 kW of mechanical power is supplied to
744 Reactor Coolants, Coolant Pumps, and Power Turbines
the shaft of a small RCP each second, but only 12 kW of this is imparted to the coolant in the form of mechanical energy.
The “lost” mechanical energy that shows up in the form of heat is then
E LOST = 14 − 12 kW = 2 kW (19.9)
If the coolant happens to be water or a metallic coolant, the temperature rise of the coolant due to this mechanical inef-
ficiency can be determined from
( u 2 − u1 ) = mc
E LOST = m ⋅ ∆T (19.11)
where c is the specific heat of the coolant. Suppose that the coolant is water and its mass flow rate is 100 kg/s. Then, the
rise in the temperature of the water as it flows through the pump is
∆T = E LOST mc
= 2 kW/(100 kg/s ⋅ 4.18 kJ/kg/ °C) = 0.0048°C (19.12)
Normally, this temperature change is small, and it can be neglected for most practical applications. However, it serves to
illustrate how large the frictional and thermal losses in a pump can be. For most coolant pumps found in nuclear power
plants about 85% of a pump’s work is used to set the coolant in motion and the remaining 15% is dissipated within the
pump and the coolant in the form of heat. This means that large RCPs have a mechanical efficiency of about 85%. In
other words, if 1 MW of electrical energy is required to turn the shaft of the pump, only 850,000 J of energy per second
is transferred to the coolant to make it move. The remaining 150,000 J/s is converted into heat and is essentially unre-
coverable. Thermodynamically, this is referred to as an irreversible pump energy loss.
19.9
F low Work
The work done by a coolant pump on a reactor coolant is an example of flow work. Flow work is defined as the work
required to set a coolant in motion. However, it has a more precise thermodynamic definition that we would next like to
discuss. When we neglect frictional losses within the pump, we can equate the shaft work WSHAFT to the increase in the
mechanical energy of the coolant WCOOLANT. This allows us to write
For coolant flowing through the pump, we can then use Reynolds’s transport theorem (see Chapter 15) to equate the
energy expended by the shaft to the change in the energy content of the fluid. Hence, for an element of fluid with volume
V, we can write
SHAFT = d/dt
Q IN + W ∫ CV
eρ dυ +
∫ CS
(Pυ + e)ρ( v ⋅ n ) dA (19.14)
where Pυ = P/ρ is the flow work required to move a unit mass of the fluid into and out of V. Equation 19.14 is not particu-
larly helpful because the second integral requires an explicit knowledge of the velocity profile on the surface of V, which
is also called the control surface. However, if we replace the integral with the average fluid velocity v and the average
mass flow rates m into and out of the control volume, we obtain
SHAFT = d/dt
Q IN + W ∫ CV
eρ d υ +
∑ OUT
υ + e) −
m(P ∑ IN
υ + e) (19.15)
m(P
19.9 Flow Work 745
where e = u + v2/2 + gz is the total energy content per unit mass. With a little more manipulation, we get
SHAFT = d/dt
W ∫ CV
eρ d υ +
∑ OUT
m (
Pυ + u + v 2 2 + gz − ) ∑ m (Pυ + u + v
IN
2
)
2 + gz − Q IN (19.16)
or
SHAFT = d/dt
W ∫ CV
eρ d υ +
∑ OUT
m (
h + v 2 2 + gz − ) ∑ m (h + v
IN
2
)
2 + gz − Q IN (19.17)
where h = u + Pυ. Equation 19.17 allows us to estimate the power required to convert a fluid with an initial value of h1,
v1, and z1 into a fluid with a final value of h2, v2, and z2. For steady-state flows, the total energy of the fluid within V does
not change with time, and thus, we can set
d/dt
∫ CV
eρ d υ = 0 (19.18)
Equation 19.17 then becomes
SHAFT =
W ∑ OUT
m (
h + v 2 2 + gz − ) ∑ m (h + v
IN
2
)
2 + gz − Q IN (19.19)
Thus, when the flow rate is known, we can find the shaft work WSHAFT required to set the coolant in motion. Notice that
the shaft work is directly proportional to the mass flow rate if the pump has a single inlet and outlet. Since h = u + p/ρ,
we can also write Equation 19.19 as
W
( 2
) (
SHAFT = P ρ + u + v 2 2 + gz − P ρ + u + v 2 2 + gz − Q IN (19.20)
1
)
where u is the coolant’s specific internal energy, v2/2 is its kinetic energy, gz is its gravitational potential energy, and P/ρ
is its flow energy. Here, the superscripts 1 and 2 refer to the inlet and outlet. Thus, we can write
( P ρg + v 2
) ( )
2g + z 1 + H PUMP = P ρg + v 2 2g + z 2 + H OUT + H LOSS (19.21)
and when there are no frictional losses through the pump, this reduces to
SHAFT mg
H PUMP = W (19.22)
where HPUMP is the useful head delivered by the pump, and HOUT is the extracted head removed from the fluid by the
pump. Since coolant pumps are not 100% efficient, we can also write W PUMP = ηW SHAFT, where η is the mechanical
efficiency of the pump. Thus, the pump head can be deduced from the pumping power using
Here, η is less than unity because no pump is 100% efficient. Usually, the pump head HPUMP is expressed in meters. Once
the pump head is known, the pressure difference across the pump is given by
For the RCPs in the primary loop of a 1,000 MWE PWR, the pump head is about 85 m. Thus, the pressure differential
across the pump is
19.10
E quations for the Pumping Power of a Pump
The pumping power W PUMP = dW/dt required to force-circulate the coolant through the core can also be derived from
the work equation of classical mechanics. The work required to move a volume of fluid a distance d is
where the force F is simply equal to the pressure P times the flow area A, and the distance d is simply equal to the veloc-
ity v times the time t. Thus, the work required to move the fluid through a specific component i in a given amount of
time t is
Wi = Pi A i v i t (19.27)
The pumping power W PUMP is then the work per second that must be expended to keep the fluid moving. Thus, the pump-
ing power for the ith component i in the primary or secondary loops is
PUMP =
W ∑ W (19.29)
i
i
or equivalently,
where the summation is performed over all of the components from i = 1 to i = N. This is the origin of the equation for
the pumping power that is used in most reactor work. The pumping power W is usually expressed in W (J/s) because it
is the first derivative of the work. In the next section, we would like to calculate the pumping power for a typical PWR.
The values of fi in the Darcy equation depend on the Reynolds number Rei in each component of the NSSS. Even though
the mass flow rate is fixed, there can be a different Reynolds number for each component of the piping system. Since
there can be a lot of different cross-sectional flow areas inside of the individual components, there can be many different
coolant velocities as well. So more than one friction factor correlation may be required to find all of the individual fric-
tion factors—f1, f2, f3, … f N. The friction factors for single-phase laminar and turbulent flow were discussed in Chapter
17. The friction factors for two-phase flow are discussed in a Chapter 23.
19.11
Calculating the Pumping Power for a Complete Loop
As we mentioned earlier, the coolant is force-circulated through the core with high-powered coolant pumps. These
pumps circulate the coolant at very high speeds (5–10 m/s), and the flow is almost always turbulent. Normally, this load
is shared by three or four pumps (see Figure 19.7). Once the pressure head Δp is known, the pumping power W PUMP
required to overcome frictional and form losses in the core, the piping system, the valves, and the heat exchangers is
where
☉☉ Δp is the pressure drop through the loop (in N/m2).
☉☉ v is the velocity of the coolant (in m/s).
☉☉ A is the cross-sectional area through the coolant must flow (in m2).
PUMP = ∆pm/
W ρ (19.32)
19.12 Corrections to the Pumping Power for Pump Efficiency 747
FIGURE 19.7 The RCPs, the steam generators, the pressurizer, the reactor vessel, and the primary coolant loops in a large PWR.
(Courtesy of Westinghouse—From http://me1065.wikidot.com/nuclear-pressurized-water-reactor-pwr.)
Now let us see if we can calculate the power required to keep one of these coolant pumps running. Suppose that
Δp = 0.65 MPa, m
= 7,000 kg/s, and ρ = 710 kg/m3 (at 2,250 PSI). The work required to keep the coolant flowing is then
PUMP = ∆pm/
W ρ = 6.5 × 10 5 kg/ms 2 × 7,000 kg/s ÷ 710 = 6.4 × 10 6 W = 6.4 MW
and this is the power required to run a single pump. The total power required to keep all of the pumps in the primary
loop running is then three or four times this number.
19.12
Corrections to the Pumping Power for Pump Efficiency
Earlier, we mentioned that no reactor coolant pump (or RCP) is 100% efficient, and a well-designed pump has a mechan-
ical efficiency between 80% and 90%. In other words, we need to account for frictional losses in the equation for the
pump work. When we do, we find that
PUMP =
W (∑ ∆p A v ) η = (∑ ∆p m ρ ) η (19.33)
i
i i i
i
i i i
748 Reactor Coolants, Coolant Pumps, and Power Turbines
where η is the actual efficiency of the pump. In nuclear power plants, the real pumping power is always greater than the
theoretical pumping power for an isentropic pump. This means that if an ideal pump requires 1 MW to pump the coolant
through a small part of the power plant, then a real pump will require about 1 MW/0.85 ≅ 1.2 MW to perform the same
function. Example 19.2 illustrates how important the efficiency of a reactor coolant pump can be.
19.13
Additional Observations about Pumping Power
Now let us make some additional observations concerning RCPs. In commercial reactors, the flow is highly t urbulent,
and the fuel rods inside the core are very smooth. Since smooth surfaces are used in most reactor work, then the friction
factor in most of the coolant loops will be given by
which the reader may recognize as the McAdams correlation. Alternatively, the Blasius correlation (see Chapter 17)
can be used
because the flow is highly turbulent. We can also write the McAdams correlation as
( )
0.2
f = 0.184 ρv D e µ (19.36)
where μ is the absolute viscosity of the coolant. If we plug this expression for the turbulent friction factor into the Darcy
equation and plug the Darcy equation into the previous formula for the pump work, we find that
( )( )
PUMP = ACL D e 1.2 ⋅ ρ0.8µ 0.2 ⋅ v 2.8 (19.37)
W
or
( )
PUMP = G ⋅ ρ0.8µ 0.2 ⋅ v 2.8 (19.38)
W
where G and C are constants, A is the area of the flow channel, and the other terms have their usual meanings. The
first term in Equation 19.41 depends on the channel geometry and is represented by the symbol G (not to be confused
with the mass flux); the second term (ρ 0.8μ 0.2) depends on the physical properties of the coolant; and the third term v2.8
which contains the velocity depends on the operating state of the reactor. So practically speaking, we would like to
pick a coolant where (ρ 0.8μ 0.2) is as small as possible to minimize the pump work. Thus, the primary reason why this
is so important is that RCPs can consume a lot of electrical power. In large PWRs, they can consume about 6 MW of
electrical power per pump. So if a 1,000 MWE (~3,300 MWT) PWR has three pumps in its primary loop, then these
pumps alone can consume as much as 20 MWE, or about 2% of the electrical output of the plant (which is equivalent
to about 6% of its thermal o utput). Thus, a typical pump consumes about 1 MW of electrical power for each 1,000
kg/s of coolant pumped.
19.14
Finding the Pumping Power for the Primary Side of a PWR Steam Generator
Earlier, we were able to calculate the electric power required to pump the coolant through the primary loop. Now sup-
pose that we would also like to calculate the electrical power required to pump the coolant through the primary side of a
19.14 FINDING THE PUMPING POWER 749
PWR steam generator. This is a common problem that every engineer must deal with in coal-fired power plants as well.
The location of the steam generator in a PWR is shown in Figure 19.8. The pumping power W in this case is
PUMP =
W (∑ ∆P A v ) η (19.39)
i
i i i
The primary side of a steam generator in a four-loop PWR has a rated flow rate of about 5,000 kg/s. The coolant flows
through approximately 4,000 tubes inside of the generator with an average diameter of 0.0222 m (about seven-eighths of
an inch). The average length of these tubes is about 19.5 m. The Reynolds number inside of these tubes is about 75,000.
The turbulent friction factor f TURBULENT inside of a tube can be found from the McAdams correlation:
which we discussed in Chapter 17. Since Re ≈ 75,000, the value of the turbulent friction factor is fTURBULENT ≈ 0.02. If
the average density of the coolant is 750 kg/m3, the velocity of the coolant through a typical tube is about 4 m/s. From
the Darcy equation, the pressure drop across an individual tube with a diameter of 0.0222 m is
A = πD 2 /4 ≈ 0.00039 m 2 (19.42)
Since v ≈ 4 m/s and η ≈ 0.85 (85%), the pumping power required to pump the coolant through a single tube is
Since a typical steam generator has about 4,000 tubes, the pumping power required to keep the coolant flowing through
the primary side is
Clearly, this is an enormous amount of power just to get the water to flow through the tube bank. For a four-loop
PWR with four steam generators, the steady-state pumping power for the primary side of the steam generators alone is
2.5 MW! A similar procedure can be used to find the pumping power for the secondary side.
FIGURE 19.8 A steam generator is located between the primary and secondary loops in a commercial PWR.
750 Reactor Coolants, Coolant Pumps, and Power Turbines
19.15
Types of Reactor Coolant Pumps
In PWRs, the coolant pumps are typically vertical, single-stage centrifugal pumps that can move large amounts
of coolant through the core. These pumps operate at temperatures between 530°F (276.7°C) and 590°F (310°C)
and pressures of approximately 2,250 PSI (15.5 MPa). A large pump can sustain flow rates of approximately
5,000–7,500 kg/s. (The flow rate depends on the pump’s design.) The coolant enters the pump from above (in the
same direction as the axis of the rotating shaft) and is discharged radially or t angentially along the outer radius of
the pump’s casing. A cross-sectional view of a centrifugal pump is shown in Figure 19.9. The fluid enters the pump
from above and is forced down a flow channel until it encounters a set of rotating blades or impellers. The coolant
flows over the blades, and the blades transfer additional momentum to the coolant. A typical coolant pump is 8–9
m high and has an outside diameter of about 2 m.
19.16
Coolant Pump Reliability
RCPs are one of the most reliable components of a nuclear power plant. They are also mechanically efficient, and
compared to the pumps used in other industries, they have very low levels of noise and vibration. A great deal of care
FIGURE 19.9 The cross-sectional view of a vertical centrifugal pump. In this design, the coolant enters the pump from above and
leaves the pump from below. Most RCPs are examples of vertical centrifugal pumps. (directindustry.fr.com.)
19.18 Pump Start-Up Times 751
TABLE 19.4
Design Specifications for a Typical Westinghouse RCP
goes into the design of a RCP. The pump bearings and seals are relatively easy to replace. The electric motors have an
average power output of about 7,000 HP. The Westinghouse 93A1 is probably the most widely used RCP in the world.
Its specifications are shown in Table 19.4. Most of the time, the 93A1 is paired to the model F or H Westinghouse steam
generators, which are vertical U-tube designs. The 93A1 has an excellent reliability record. In general, it is a very well-
designed pump.
19.17
Jet Pumps and Their Characteristics
BWRs use different pumps than PWRs do. The most common BWR coolant pump is called a jet pump. Jet pumps
are located inside the reactor pressure vessel, and they can almost never fail because they have no moving parts.
Commercial BWRs use anywhere from 8 to 24 of these pumps to circulate the coolant. (The exact number depends on
the power rating of the plant.) The location of these pumps is shown in Figure 19.11. Normally, jet pumps are mounted
to the downcomer between the core barrel and the reactor pressure vessel. They direct fluid downwards into the pres-
sure vessel which is then directed upwards into the core. Jet pumps use the Bernoulli effect to create suction inside
of the pump with a secondary flow stream created by the coolant pumps located outside of the pressure vessel. This
suction forces additional coolant through a sonic diffuser without the need for an impeller or rotating blade. A cross-
sectional view of a jet pump is shown in Figure 19.10. The external flow which generates the pressure head inside of
the pump is provided by a separate set of recirculation pumps which are located outside of the reactor pressure vessel.
Approximately one-third of the core flow through a BWR passes through these recirculation pumps.
19.18
P ump Start-Up Times
Using Bernoulli’s equation, it is possible to estimate how quickly a pump can be started up or shut down. The start-up
or shutdown time is a function of the mass flow rate. If we start with the time-dependent Bernoulli equation, the flow
between any two points along a streamline can be described by
2 2
∫1
∂ v ∂ t ⋅ dr +
∫
1
( ) ( )
dp/ρ + v 2 2 + gz 2 − v 2 2 + gz 1 = c(t) (19.45)
752 Reactor Coolants, Coolant Pumps, and Power Turbines
FIGURE 19.10 The cross-sectional view of a jet pump similar to the jet pumps used in BWRs.
FIGURE 19.11 The jet pump locations and other pump locations in a commercial BWR.
We can then use this equation to predict the flow as long as the pressure drop Δp = p2 − p1 between points 1 and 2 is
known.
19.19
Estimating the Start-Up Time for a RCP from Bernoulli’s Equation
In principle, Bernoulli’s equation can be used to model the start-up and shutdown of a RCP. This can be done by solving
the integral form of Bernoulli’s equation between the inlet and the outlet of the pump. To show how this is done, suppose
that the flow is incompressible, one-dimensional, and irrotational. Bernoulli’s equation can then be written as
2
∂ v ∂ t ⋅ dr + ( p2 − p1 ) + ρg ( z 2 − z1 ) + ρ/2 ( v 2 2 − v1 2 ) = 0 (19.46)
ρ
∫
1
This equation represents a path integral along two points on a streamline. In this case, we are assuming that the flow is
laminar because streamlines do not exist for turbulent flow. Suppose that we would like to determine the mass flow rate
m (t) = dm/dt through the pump when the piping system in the primary loop has N distinct sections, and each section has
19.19 ESTIMATING THE START-UP TIME FOR A RCP 753
a different length to area ratio Li/Ai, where i is the index of the section. If we know the pressure head Δp = pOUT − pIN
between the inlet and the outlet of the pump, Bernoulli’s equation for an incompressible and one-dimensional flow can
be written as
N
ρ
∫
1
( )
∂ v ∂ t ⋅ dr + ( pOUT − p IN ) + ρg ( z N − z1 ) + ρ/2 v N 2 − v1 2 = 0 (19.47)
Here, the integral can be evaluated rather easily because the flow rate for an incompressible fluid is not spatially depen-
dent. Then, the integral can be evaluated to read
N
∑L
N N N N
ρ
∫1
∂ v ∂ t ⋅ dr = ∂ / ∂ t ρ
∫1
v ⋅ dr = ∂ / ∂t
∫
1
(m/A) dL = dm/dt
∫
1
dL/A = dm/dt
i =1
i A i (19.48)
∑
N
Here, dm/dt L i A i is an equivalent inertial length whose value can be found if we know the system’s internal
i =1
+ ( pOUT − p IN ) + ρg ( z N − z1 ) + m
(L/A)dm/dt (
2 2ρ⋅ 1 A N 2 − 1 A1 2 = 0 (19.49) )
where
∑
N
L/A = L i A i (19.50)
i =1
Now let us see if we can understand what Equation 19.49 says. All of the terms in this equation (with the exception
of the first term) represent the value of quantities at the end points of a piping segment. Hence, if the value of the
pressure drop Δp = pOUT − pIN is known or specified in advance, we can determine the time-dependent mass flow
rate required to obtain this pressure drop. In principle, we simply have to solve a first-order differential equation for
the value of m (t). Unfortunately, this equation is nonlinear because the time-dependent mass flow rate appears as the
square of m (t). However, we can use a procedure called partial factoring to convert it into an equivalent set of linear
differential equations. Suppose that we model the flow using the five components shown in Figure 19.12. These five
components represent one leg of the primary coolant loop. Now suppose that we would like to use a RCP to estab-
lish a steady-state flow in this loop. This is equivalent to positioning the pump as the round green object shown in
Figure 19.12. Another picture of the primary loop is presented in Figure 19.13. Next, we would like to determine the
time it takes the reactor coolant, which is initially at rest, to reach its steady-state flow rate (or maximum flow rate)
after the coolant pump is turned on. Since we are dealing with only one coolant pump in this case, and the inlet and
the outlet of the pump are at essentially the same elevation, we can solve for the time-dependent flow rate once we
specify the value of the pump head ΔpHEAD = pIN − pOUT. The only thing that we have not accounted for is the inertia
of the rotors and impellers inside of the pump.
However, if we can temporarily neglect this inertia, then the following equation can be used to find the time-
dependent flow rates in our five-component system:
= ∆p − m
(L/A)dm/dt (
2 1 A N 2 − 1 A1 2 2ρ (19.51) )
Next, we can find a solution to this equation with a simple mathematical trick. Suppose that we consider only the end
points (1 and 5) of the flow loop shown in Figure 19.12. The time-dependent Bernoulli equation that describes the mass
flow rate through this loop is
= ∆p − m
(L/A)dm/dt (
2 1 A 5 2 − 1 A1 2 2ρ (19.52) )
Suppose that we rewrite this equation as
dt = (L/A)(1/∆p)dm (
1 − m 2 C2 (19.53) )
754 Reactor Coolants, Coolant Pumps, and Power Turbines
FIGURE 19.12 The parameters of some of the components in the primary loop of a PWR.
FIGURE 19.13 A depiction of the high-level components of a PWR coolant loop from the Modular Accident Analysis Program
(or MAPP) computer program, which is sometimes used to model severe reactor accidents. (fauske.com.)
(
where C2 = 1 A 5 2 − 1 A1 2 ) ( 2ρ⋅ ∆p). In principle, we can also write Equation 19.53 as
(A/L) ∆pdt = dm/[(1
− mC)
⋅ (1 + mC)]
(19.54)
or
19.19 ESTIMATING THE START-UP TIME FOR A RCP 755
/(1 − m
Here, we have replaced the term dm 2C2) by the sum of two partial fractions:
dm (
1− m )
2 C2 = 1 2 dm[1/(1
− mC)
+ 1/(1 + mC)]
(19.56)
Furthermore, we can assume that the inlet and the outlet of the pump are at located at approximately the same axial
elevation. This allows us to set ρg·Δz ≈ 0. After integrating Equation 19.55 once, we obtain
[(1 + mC)/(1
− mC)]
= e −λt (19.58)
or
where the value of λ is given by λ = (A/L)·2C·Δp. In other words, Equation 19.59 says that if we wait long enough for t →
→ 1/C. Now let us see if we can use this equation to determine the steady-state mass flow rate m
∞, then m that allows us
to balance the pressure drop Δp against the pump head. Using representative values for the parameters shown in Figure
19.13, we find that the steady-state mass flow rate is ~30,000 kg/s for a pressure head Δp of 85 m. Of course, this does
not account for the effects of fluid friction. To obtain the time constant for the coolant pump, we take Equation 19.59 and
determine how long it takes for the mass flow rate to reach 99% of its steady-state value. Since
∑
5
(L/A)LOOP = (L/A)i ≈ 103/m (19.60)
i =1
and
λ ≈ 0.55/s (19.61)
The time it takes for the system to reach 99% of its maximum flow rate is then
or
1.99 = 0.01e 0.55t (19.63)
Again, these values are approximate ones, but they are correct as long as the frictional forces in the piping system can be
ignored. Since Equation 19.59 is based on Bernoulli’s equation, it is only a precise description of the pump’s behavior
for laminar flow. For turbulent flow, the predictions of Equation 19.59 are just approximate ones.
19.20 S
olving the Momentum Equations with Loss Coefficients
and Friction in a Viscous Coolant Loop
Having just solved Bernoulli’s equation for the time-dependent mass flow rate in an inviscid flow loop, we can perform
a similar exercise by including fluid friction and some realistic loss coefficients into the simulation. When we include
these loss coefficients and a representative friction factor f, the fluid dynamics equation we need to solve is
+ K R v R 2 2 + K SG v SG 2 2 +
(L/A)dm/dt ∑ f (L/D) ρv
i
i i
2
2+m ( )
2 2ρ⋅ 1 A 5 2 − 1 A1 2 = ∆p (19.65)
where K R and KSG are the loss coefficients for the reactor and the steam generator, and vR and vSG are the velocities of the
flow in the reactor and the steam generator. Writing this equation in terms of the mass flow rate m = ρvA, and assuming
that the friction factor f is constant, the equation we have to solve for the time-dependent mass flow rate is
+ m
(L/A)dm/dt (
)
2 2ρ 1 A 5 2 − 1 A1 2 + K R A R 2 + K SG A SG 2 + ∑ f(L/D) (1 A ) = ∆p (19.66)
i
i i
2
In other words, this equation has exactly the same form as the previous differential equation we solved to find the time-
dependent mass flow rate for an inviscid flow loop. The only difference is the form of the inertial coefficient that we
happened to call C:
(
C 2 = 1 A 5 2 − 1 A1 2 ) ( 2ρ∆p) (for an inviscid flow field) (19.67)
C2 = 1 A 5 2 − 1 A1 2 + K R A R 2 + K SG A SG 2 +
∑ f (L/D) (1 A )
i
i i
2
( 2ρ∆p) (for a viscid one) (19.68)
Now let us plug some representative values for the loss coefficients into this equation. For a typical PWR, the loss coef-
ficient for the core can average about 20, and for the steam generators, it can average about 50. Assuming a friction factor
of 0.015, and a pump head Δp is approximately 85 m, the only other parameter we need to estimate is the value of ρ. The
density of saturated water at 300°C and 2,200 PSI is about 720 kg/m3, but to keep the math simple, we will assume the
reactor is just being started up, so the core is very cool. In this case, we can assume that ρ = 1,000 kg/m3. When we do
so, we find that the value of C2 is
The steady-state mass flow rate in a single leg of the primary loop is therefore m = 1/C ≈ 7,450 kg/s. The velocity of the
coolant exiting the core is v3 = (m /ρ)(1/A3) = 7.45/0.4 ≈ 19.6 m/s. This is about three times the velocity of the coolant
through a typical fuel assembly, but considering the simplicity of our model, we have developed a remarkably useful
way to simulate the time-dependent flow field. Hence, the inertia in the reactor coolant system affects the pump like
the time constant for a large capacitor. Eventually, the capacitor discharges, and the steady-state flow rate is determined
by the value of C. Note that the steady-state flow rate in this case is close to that of a typical PWR coolant pump in
the primary loop (7,500–8,000 kg/s) because the frictional pressure drop and the pressure drop due to the loss coef-
ficients are approximately twice the acceleration pressure drop. If the flow areas happened to be larger or the friction
factors happened to be larger, the time constants (which measure the amount of fluid inertia) would be larger as well.
In practice, the inertia of the rotors and the impellers in the pump has to be taken into account to predict the flow rate
more accurately.
19.21 Pump Performance Parameters 757
19.21
P ump Performance Parameters
For single-phase coolants such as water and liquid metals, the flow through most coolant pumps can be assumed to be
incompressible and isothermal. When this occurs, it is often more convenient to use the volumetric flow rate υ′ = dυ/dt
= dm/dt to measure the performance of a pump. In this context, the volumetric flow rate
rather than the mass flow rate m
is sometimes called the capacity of the pump
and the capacity is measured in cubic meters per second (m3/s). For example, if the capacity of a pump is 10 m3/s, the
mass flow rate is m = υ′ρ = 10,000 kg/s if the coolant is water at room temperature (where ρ ≈ 1,000 kg/m3), and the
= υ′ρ = 8,000 kg/s if the coolant is liquid sodium at 375°C. The properties of liquid sodium and water
mass flow rate is m
are presented in Table 19.2. The performance of a RCP can also be expressed in terms of its net head H, which, for an
incompressible fluid, is also called its Bernoulli head:
Here, the pump head and the Bernoulli head have the units of length, and when comparing different RCPs, it is often
specified as the equivalent height of the column of water that the pump is capable of supporting, even when the pump
is not pumping water. For a coolant pump in which the inlet and the outlet diameters are the same, and the inlet and the
outlet are at approximately the same elevation, the equation for the pump head becomes
In practice, the pump head is proportional to the power delivered to the fluid by the pump. To convert the pump head
to the pumping power, we must multiply the pump head by the mass flow rate m and the gravitational acceleration g.
Consequently, because the power is equal to the force times the distance (Power = Force × Distance), the pumping power
can be written as
Occasionally, the pumping power is also given the symbol W (or W PUMP). In practice, no reactor coolant pump is 100%
efficient. For this reason, some of the energy directed to the pump must be used to combat frictional losses in the fluid,
in the bearings, and other losses caused by flow separation and other forms of turbulent and viscous dissipation. This
means that the mechanical energy supplied to the shaft W SHAFT must always be larger than the actual pumping power W
PUMP which is used to measure the fluid displacement. The ratio of the pumping power to the shaft power is then called
the pump’s efficiency η:
In reactors, the efficiency η is a dimensionless number having a value of about 0.85 (85%). In practice, we can express the
shaft work W SHAFT performed by the pump as the product of the torque TSHAFT supplied to the shaft and the rotational
speed ω of the shaft expressed in radians per second:
when ω is measured in rad/s. These equations allow us to write the overall efficiency of the pump as the ratio of its useful
power to the supplied electric or mechanical power:
The efficiencies of all RCPs can be found in this way. Notice that the efficiency of the pump is NOT a simple constant.
It is a function of both the rotational speed of the shaft ω and the pump’s delivery rate υ’.
19.22
Performance Curves for Pumps
Normally, the efficiency of a pump is a function of its rotational speed, and its efficiency varies with the shaft speed
because the frictional losses sustained by the rotors and the impellers change with the rotational speed and the delivery
rate υ′. For this reason, most RCPs reach their maximum efficiency at a point between zero power and full power that is
called the best efficiency point or the BEP of the pump. The BEP then corresponds to the maximum value of H versus ω
on what is called a pump performance curve. A pump performance curve that applies to RCPs is shown in Figure 19.14.
All RCPs have a curve such as this, and normally, the flow rate (the steady-state operating point) is designed to coin-
cide with the value of the shaft rotational speed ω where the delivery rate υ′ is maximized. For PWRs, this happens to
FIGURE 19.14 The pump head, the base horsepower, and the pump efficiency as a function of the delivery rate υ′. The operating
point is defined by the intersection of the systems requirement curve and the pump performance curve (both shown in purple).
19.24 CONNECTING RCPs IN PARALLEL 759
correspond to a delivery rate υ’ between 6 and 8 m3/s. The maximum flow rate through a RCP is reached when there
is no resistance at either the inlet or the outlet. In other words, the volumetric flow rate υ’ is maximized when the pump
head H is zero (H = 0). This flow rate is called the pump’s free delivery rate, and under these conditions, there is no
apparent load on the pump. At this delivery point, the value of υ′ is large, but the value of H is zero. In other words, the
pump is doing no useful work to move the coolant through the core.
From Equation 19.79, it can also be seen that when the pump head H is zero, the net efficiency η is zero. Now let
us consider the opposite situation. When the outlet port is blocked, the pump head H is large but the delivery rate υ′ is
zero. This head is called the shutoff head because no coolant is allowed to flow through the pump. RCPs tend to operate
somewhere between these two states. Between shutoff and free delivery, the efficiency η increases for a while and then
decreases as a function of the flow rate υ′. The pump’s efficiency reaches its maximum value at the BEP which is shown
in Figure 19.14. Curves of H versus υ′ and η vs. υ′ are called the pump’s performance curves. Typical curves are provided
for different rotational speeds. For a RCP, the performance curves (and therefore the point of maximum efficiency) are
a function of its rotational speed.
19.23
Matching the Required Head and the Available Head
There can be many elevation changes and major and minor frictional losses in reactor piping systems. We know from
our previous discussion that the pressure head H = ΔP between the inlet and the outlet of a RCP is a function of the mass
flow rate through the pump. Specifically, it is proportional to the square of the coolant velocity v or the delivery rate υ′
and the gravitational pressure head ρg·Δz. In other words, the required head increases with the flow rate, whereas the
available head decreases with the flow rate. Once the required head HREQUIRED is set by the delivery rate υ′, the avail-
able head HAVAILABLE must be matched to the required head to determine what is called the operating point of the pump.
Correct operation then requires the required head to be matched to the available head:
In most reactors, the operating point of the pump is close to the BEP. This is important because RCPs require a couple
of percent of the total power output of the plant just to meet the required heat removal rate. The closer the BEP is to the
operating point, the more efficient the power plant will be. If the operating point is different than the BEP, the plant will
not operate at optimum efficiency. For this reason, RCPs have been redesigned over time to more closely align the BEP
with the operating point (which corresponds to a mass flow rate of 5,000–7,000 kg/s). Aligning the BEP point with the
operating point results in the lowest overall power production cost.
19.24
Connecting Reactor Coolant Pumps in Parallel
In commercial nuclear power plants, several coolant pumps are connected in parallel to achieve a higher mass flow rate
than is possible with a just single pump (see Figure 19.15). For example, in large PWRs, between two and four RCPs are
connected in parallel in the primary loop to remove heat from the core. A separate coolant pump is then connected to
each coolant leg in the primary loop, which is also attached to its own steam generator. Normally, this approach is used
because the RCPs are identical and have exactly the same capacities, expected heads, and BEP points. The combined
delivery rate is then equal to the sum of the delivery rates of the individual pumps. We can express this fact by writing
FIGURE 19.15 Identical coolant pumps with large electric motors being operated in parallel.
where N is the number of pumps. If the pumps are identical, their performance curves are additive (refer to Figure 19.16).
The pump heads are the same, and causing one pump to shut down will not necessarily cause a flow reversal in any of
the other pumps. This is normally how the coolant pumps in the primary loop are designed to operate. However, sup-
pose that a reactor happened to be designed with three different types of coolant pumps instead of just one. This would
mean that each pump would have a slightly different delivery rate υ′ and a slightly different pump head H than every
other pump. Then, the combined pump performance curve would also be an additive one, and it would resemble the
curve shown in Figure 19.17. However, if the pump heads and the pump capacities are different, then we must be careful
to operate the pumps properly to avoid a capacity problem or a flow reversal. For low flow rates and low values of the
pump head H, the capacity of the three pumps is the sum of the capacities of the individual pumps, and this is shown by
the purple line on the right-hand side of the figure. However, as soon as we exceed the rated head H1 for pump 1, pump
1 should be shut off and its flow should be blocked by closing a valve. Otherwise, its motor would be pushed beyond its
normal operating limit, and the pump shaft might be damaged.
FIGURE 19.16 Operating several RCPs in parallel has many practical advantages as long as the pumps are identical. The delivery
rate is simply the sum of the delivery rates of the individual pumps.
19.25 Centrifugal Coolant Pumps 761
FIGURE 19.17 Operating several RCPs in parallel must be done carefully if the pumps are not identical. To avoid pump damage,
flow reversals, or loss of the net head, any individual pump must be shut off as soon as the total head exceeds the individual shutoff
head. If all three coolant pumps were identical, this would not be a problem because the shutoff head of each pump would occur at
the same net head.
Moreover, the direction of the flow through the pump could reverse because pumps 2 and 3 have higher heads than
pump 1! Obviously, this would be a disaster in the making because it would redirect more hot fluid back into the core,
and this would keep it from reaching the steam generators to be cooled. If we shut down pump 1, pumps 2 and 3 would
then have to pick up the load, and their combined capacity would be υ′ = υ′2 + υ′3. However, above the shutoff head H2 for
pump 2, the pump could be damaged, and the flow could also reverse itself, so another valve would have to be actuated
to shut off the flow to pump 2. The combined capacity is then equal to that of pump 3 alone, or υ′ = υ′3. In other words,
this exercise illustrates the problems involved when designing the primary (or secondary) loop of a reactor cooling
system using different capacity pumps. Within a given loop, it is much simpler (and safer) to make all of the pumps the
same. Then, the pump heads and the delivery rates will be identical as well. There is also an economic incentive to using
identical pumps because the overall cost of the pumps will most likely be lower.
19.25
Centrifugal Coolant Pumps
Before moving on, we would first like to take a moment to illustrate how Bernoulli’s equation can be modified to
account for the rotation of the impellers and the shaft in a centrifugal pump. For a variety of reasons, most RCPs
are centrifugal ones. Centrifugal pumps are also used in vacuum cleaners, leaf blowers, electric furnaces, and even
reactor cooling towers. Other types of rotary pumps are called axial flow and mixed flow pumps. However, in most
industrial applications, centrifugal pumps scale better than axial pumps or mixed flow pumps. This is particularly
true at the flow rates required to pump the coolant through the core. In a centrifugal pump, the fluid enters the pump
axially (in the same direction as the axis of the rotating shaft) and is discharged radially or tangentially along the
outer radius of the pump casing. The construction of a typical centrifugal pump is shown in Figure 19.18. The fluid
enters the pump from above and is forced down a flow channel until it encounters a set of rotating blades. Sometimes
these blades are called impellers. The coolant flows over the blades, and the blades transfer additional momentum to
the coolant. The coolant may also acquire an additional radial velocity due to the centrifugal forces in the pump cas-
ing. If the flow is incompressible and the inlet and outlet diameters are the same, the average flow rate at the outlet is
identical to that at the inlet. However, the pressure increases from the inlet to the outlet, and the rotation of the fluid
is what creates the pressure head. The amount of momentum that is transferred to the fluid depends on the length of
the blades and the blade angle. Here, the blade angle is defined as the angle between the blade tips and the tangent
to the circle inscribed by the rotating blades. The blade angle is illustrated in Figure 19.19. Normally, the blade angle
762 Reactor Coolants, Coolant Pumps, and Power Turbines
FIGURE 19.18 A detailed schematic showing the internal structure of a large, centrifugal pump. The coolant enters the pump
along the primary axis (on the right), acquires mechanical energy from the rotating impellers or blades, and is ejected from the
pump at a 90° angle from the angle from which it entered. (Picture provided by www.Directindustry.com.)
in a RCP is between 30° and 45°. Depending on the flow rate required, some RCPs may have blade angles greater
than or less than this. At any point along the radius r of the impeller, an energy balance can be performed on the
fluid using a steady-state version of Bernoulli’s equation. This equation can be written as
( )
H = (1/2g) ( v 2 2 − v1 2 ) + ω 2 ( r2 2 − r1 2 ) − v 2 REL − v1 REL (19.82)
2 2
In other words, for an incompressible fluid with the same axial elevation, the pressure head H through a centrifugal
pump is proportional to the change in the kinetic energy of the coolant plus the rotor tip kinetic energy minus the change
in the relative kinetic energy of the fluid between the inlet and the outlet of the impeller. This is equivalent to saying that
Bernoulli’s equation at any radius r along the impeller is
Of course, this is equivalent to saying that the kinetic energy of the fluid plus the kinetic energy of the impeller blades is
conserved. Again ω is the rotation rate of the shaft in rad/s. If the speed of the shaft is known in revolutions per second
(or revolutions per minute), its rotational speed must be converted from RPM to radians per minute to use Equation 19.82
properly. Otherwise, substantial errors may occur. To convert from RPM (rotations per minute) to radians per minute,
one simply multiplies the number of rotations per minute by 2π. Hence,
To convert to radians per second, we must then divide this result by 60. This results in the following conversion factor:
FIGURE 19.19 The pressure profiles and blade angles for different types of centrifugal coolant pumps. Centrifugal pumps can
have backward-inclined blades, radial blades which are straight (like the spokes of a wheel), and forward-inclined blades. The blade
angle determines how much mechanical energy is imparted to the fluid by the impellers. Sometimes the blade angle of a pump is
called its attack angle. Note that pumps with forward and backward blade angles have different pressure heads and horsepower
requirements than pumps with radial blades.
where r1 is the radius at which the fluid first touches the blades, and r2 is the radius at which the fluid finally leaves the
blades. In this equation, v1 and v2 are the measured velocities of the blades at radii r1 and r2, and vREL is the r elative velocity
between the fluid and the blades. The importance of making the proper conversions is illustrated in Example 19.5. In
centrifugal pumps, the internal pressure profiles are found using ANSYS, FLUENT, or another similar CFD program
(see Figure 19.20).
FIGURE 19.20 A centrifugal pump and its internal pressure profile found by ANSYS.
PUMP = mgH
W = ρgυ′H (19.86)
Of course, this equation assumes that the coolant pump is 100% efficient. If the efficiency is less than this (say 85%),
then the power required to turn the shaft is
SHAFT = W
W PUMP η (19.87)
where η is the aggregate efficiency. Normally, the efficiency is a function of the delivery rate of υ′: η = f(υ′). We can also
express the shaft work W SHAFT performed by the pump as the product of the torque TSHAFT supplied to the shaft and the
rotational speed ω of the shaft expressed in radians per second:
SHAFT ω = W
TSHAFT = W PUMP ηω (19.89)
or
where H is the pump head. The torque T in this case has the units of N m. Using the parameters given above, it is easy
to show that Westinghouse 93A1 coolant pump has a rated horsepower of W PUMP ≈ 7,000 HP under normal operating
conditions.
19.27
A nalogies between Turbines, Compressors, and Pumps
In coal-fired, gas, hydroelectric, and nuclear power plants, the device that drives the electric generator is the turbine.
As fluid passes through a turbine, work is done against the turbine’s blades, which are attached to a shaft that turns
the electric generator. As a result, the shaft rotates, and the turbine produces work. Compressors are similar to pumps
because they are used to increase the energy content of a fluid. Both pumps and compressors supply energy to a fluid
19.28 Other Types of Power Turbines 765
using a rotating shaft, and work is supplied to the shaft from an external energy source. Pumps work very much like
compressors except that they handle liquids instead of gases. Turbines produce energy from a working fluid, while
pumps, compressors, and fans consume it.
This energy is normally produced or provided in the form of electric power. Steam turbines in nuclear power plants
are described by a similar set of equations to those used to describe pumps. In this section, we would like to briefly
discuss these turbines and show how fluid kinetic energy can be converted into rotational mechanical energy. We would
then like to present the equations that can be used to determine their thermodynamic efficiency. Broadly speaking, a
turbine is designed to extract energy from a working fluid, while a pump is designed to add it. In other words, a turbine
performs the INVERSE function of a pump. It then follows from this simple analogy that the efficiency of a power
turbine is the reciprocal of the efficiency of a pump. Normally, nuclear power plants contain four to six large steam
turbines deployed in parallel. (The actual number used depends on the power rating of the plant.) These turbines take
HP (or high pressure) steam from the core (in the case of a BWR) or from the steam generators (in the case of a PWR)
and convert this steam into mechanical energy by turning a rotating shaft. This shaft then turns the armatures inside of
an electrical generator. The rotation of these armatures is used to produce electric power. The rotation rate of the shaft
is used to determine the frequency of the alternating current in the country in which the reactor is located. Normally,
steam turbines and the electric generators are designed as matched pairs. By this, we mean that each steam generator is
paired to one or two identical steam turbines.
19.28
O ther Types of Power Turbines
There are several types of power-producing turbines in the world today. These include hydraulic turbines (which are
used in hydroelectric dams), steam turbines, and gas turbines. For a number of reasons, most nuclear power plants
use steam turbines. A large steam turbine normally contains multiple sections or stages, and each stage is designed
to handle the steam at a progressively lower pressure than the previous stage. A cross-sectional view of a reactor
steam turbine is shown in Figure 19.21. As the pressure of the steam falls, its water content increases, and more water
FIGURE 19.21 A large steam turbine used in a nuclear power plant. Note that the blades in the primary and secondary stages have
different sizes and blade angles. This is to done to compensate for the fact that the steam in the second stage is cooler and wetter
than the steam in the primary stage.
766 Reactor Coolants, Coolant Pumps, and Power Turbines
droplets condense out of the HP steam. These water droplets impinge on the turbine blades, and the steam becomes
“wet.” Normally, steam turbines in nuclear power plants are designed so that the water content in the secondary stage
does not exceed one percent. (Otherwise, the droplets may damage the turbine blades.) In PWRs, there is one steam
turbine for each leg or steam generator in the secondary loop. BWRs also have multiple steam turbines. Depending on
their design, there can be between two and four steam turbines per plant. Within a steam turbine, all of the stages are
connected to a single rotating shaft, which is in turn connected to what is called an electrical generator. The rotating
shaft turns what are called armatures inside of the generator, and the rotation of these armatures generates a magnetic
field that produces an electric current. The armatures are connected to a rotating turbine shaft such as the one shown in
Figure 19.22. Normally, steam turbines and electrical generators used in nuclear power plants are enormous machines
between 10 and 20 m long. A single steam turbine can weigh several thousand tons. Figure 19.23 shows an electrical
generator used in a nuclear power plant. Electric generators are slightly smaller than steam turbines, but they are still
very massive devices.
FIGURE 19.22 A steam turbine connected to an electric generator with a rotating shaft. (Picture provided by http://geothermal.
marin.org.)
FIGURE 19.23 A large electrical generator used in a dam (left), and a similar generator used in a nuclear power plant. Note the
immense scale and size that these electrical generators can have.
19.29 Steam Turbines and Their Mechanical Efficiency 767
19.29
Steam Turbines and Their Mechanical Efficiency
In both nuclear and coal-fired power plants, the working fluid in the secondary loop is normally steam, and so turbines
that convert steam energy into mechanical energy are called steam turbines. They do so by converting the kinetic energy
of the steam into the mechanical energy of a rotating shaft. Whereas the rotating part of a RCP is called the impeller, the
rotating part of a steam turbine is called the runner. In some turbines, the runner may consist of several separate parts
or stages. The first stage may be used to handle high pressure (or HP) steam, and the second stage may be used to handle
low pressure (or LP) steam. As we discussed earlier, almost all steam turbines used in nuclear power plants are multi-
stage turbines. These turbines can consist of several HP and LP stages. In other words, these turbines are optimized to
extract as much energy as possible from the steam by using different rotor/vane combinations. The blade angles can also
be different between the HP stages and the LP stages. The HP and the LP stages in a multistage turbine are illustrated in
Figure 19.24. In general, energy-producing devices like turbines have higher theoretical efficiencies than RCPs.
There are several reasons for this difference. First, RCPs operate at higher rotational speeds than steam turbines
and power turbines (a factor of five or six higher). Because of this, shear stresses and frictional losses are higher in
RCPs than they are in power turbines. Second, converting kinetic energy into flow energy is less efficient than convert-
ing flow energy into shaft energy because the frictional losses are higher. Third, power turbines (especially the steam
turbines used in nuclear power plants and the hydroturbines used in hydroelectric dams) are much larger than RCPs,
and viscous losses tend to become less important as the size of the turbine blades increases. Finally, power turbines
operate over a much narrower range of speeds than RCPs, and because of this, they can be designed to operate more
efficiently than RCPs. In fact, most power turbines are designed to operate at a constant speed. (The rated speed
depends on the size of the turbine and its intended use.) For example, in the United States, standard alternating current
(AC) is produced at 60 Hz (3,600 cycles per minute), so the power turbines which turn the electrical generators must
rotate at some fraction of this speed. In particular, steam turbines and hydroturbines rotate at speeds that are some
selected fraction of this number, namely, 7,200 RPM divided by the number of poles on the electric generator. If the
number of poles is an even number (say 2, 4, or 8), then the speed of the steam turbine must be 7,200/2 or 7,200/4 or
7,200/8 = 3,600 RPM, 1,800 RPM, or 900 RPM. Large turbines which are used in hydroelectric dams operate at even
lower speeds like 7,200/48 = 150 RPM or 7,200/60 = 120 RPM. As a rule of thumb, the larger the turbine, the slower it
turns. The pressure head H across a single stage turbine is defined as the pressure difference between the turbine inlet
pIN and the turbine outlet pOUT:
FIGURE 19.24 A cross-sectional view of a multistage turbine used in a cogeneration facility. (Picture provided by
Cogencanada.org.)
768 Reactor Coolants, Coolant Pumps, and Power Turbines
In principle, the pressure head across a power turbine is no different than the pressure head across a RCP. In the case of
a steam turbine, the steam enters the turbine as dry steam with a pressure of pIN and leaves the turbine as wet steam with
a pressure of pOUT. (The condensation of the steam then produces the energy reduction that is observed.)
19.30
Ways to Define Turbine Efficiency
SHAFT (the energy output)
By convention, the efficiency of a power turbine is defined as a ratio of the shaft horsepower W
to the power extracted by the fluid flowing through the turbine W FLUID = ρgυ′H.
where bhp is the base horsepower extracted by the turbine shaft. Thus, the turbine efficiency ηTURBINE is the reciprocal
of the pump efficiency ηPUMP because a turbine performs the opposite function of a pump. The relationship between the
two is shown in Figure 19.25. However, a well-designed hydroturbine or steam turbine can have an aggregate efficiency
between 90% and 95%, whereas a well-designed coolant pump can have an aggregate efficiency between 85% and 90%.
RCPs and steam turbines follow this same basic trend. In other words, none of these devices can have an aggregate
efficiency greater than 1.0, and we can never get more energy out of a working fluid than we put into it. Normally, a
Computational Fluid Dynamics (or CFD) program is used to optimize the design of steam turbines and RCPs. When it
comes to nuclear power plants, FLUENT is probably the most widely used computer program to perform this function.
19.31
Two-Stage Turbines
Nuclear power plants use steam turbines with multiple stages to optimize their efficiency. Their thermal efficiency is the
work that is performed by the turbines divided by the heat that is produced in the core:
In two-stage turbines, the steam first enters a HP stage, and then, the spent steam is diverted to a LP stage. When
implemented properly, a two-stage turbine allows more work to be performed than a single-stage turbine. Now let us
examine the thermodynamic state of the steam as it moves through a two-stage turbine. In a two-stage design, steam
enters the HP stage at about 285°C and 6.9 MPa. Here, the enthalpy of the steam is about 2,775 kJ/kg. After it leaves the
HP stage, its temperature drops to about 190°C, and its pressure drops to about 1.2 MPa. Its moisture content increases
FIGURE 19.25 On the left, the force of HP steam impinging on a turbine blade using FLUENT. On the right, how efficiencies are
computed for turbines and pumps. By definition, the efficiency must always be less than 1.0, and the efficiency of a turbine is the
reciprocal of the efficiency of a pump.
19.32 Steam Turbine Blade Design 769
FIGURE 19.26 Representative state changes in a typical two-stage PWR steam turbine. Energy in the form of external work is
extracted from the steam as it passes through each stage. Between the HP and LP stages, it is also reheated.
to between 12% and 15%. At this point, it is simply too wet to be useful unless additional moisture can be removed from
the steam. However, additional work can still be extracted from the steam if the water droplets can be removed and it
can be reheated again. Moisture is removed from the steam in the turbine using a device called a moisture separator.
The remaining steam is then reheated using another device called a reheater. In modern PWRs, the steam is reheated to
about 240°C, at which point it becomes superheated again. The amount of superheat is design dependent, but it can be
as great as 50°C. The superheated steam then enters the LP stage where more mechanical work is performed. It enters
the LP stage at about 1.2 MPa and 240°C and leaves the LP stage at about 0.01 MPa and 45°C. At this point, its mois-
ture content increases to about 50%, and its enthalpy falls to about 1,400 kJ/kg. It is then sent to the condenser where it
becomes liquid water again. Thus, about 80% of the pressure drop occurs across the HP stage, and about 20% of the
pressure drop occurs across the LP stage. The entire process is illustrated in Figure 19.26. Most of the remaining pres-
sure drop occurs as it passes through the steam generator.
19.32
Steam Turbine Blade Design
Reheating the steam between the HP and LP stages increases the amount of useful work that can be extracted from the
steam. However, to realize this efficiency gain, the design of the blades in the HP stage must be different than the design
of the blades in the LP stage. In steam turbine design, this requires adjusting the blade angle, the size of the blades, and
the way in which the steam is allowed to impinge upon the blades. Normally, steam turbine blades fall into two general
categories:
1. Impulse blades
2. Reaction blades
770 Reactor Coolants, Coolant Pumps, and Power Turbines
Impulse blades are used in the HP stage, and reaction blades are used in the LP stage. We would now like to discuss
some of the differences between the designs of these blades.
19.33
I mpulse Blades in HP Turbines
When steam first enters the HP stage, mechanical work is performed by allowing the steam to hit the blades. Newton’s
second law of motion describes the transfer of momentum from the steam to the blades. The steam is fired at the turbine
blades through small nozzles, which cause the blades to rotate. These blades are bucket-shaped blades to catch the
steam and redirect it at an angle or sometimes even back in the direction from which it came. The work performed by
the steam is then the applied force on the blades, multiplied by the distance through which the blades move:
Turbines used in hydroelectric dams also use impulse blades, but in these turbines, the working fluid is water. The
blades in an impulse turbine are shown in Figure 19.27. Most of the pressure drop across an impulse turbine occurs
in the nozzles that are used to spray the steam onto the blades. In these nozzles, the pressure head is converted into a
velocity head by accelerating the flow of the steam. The rate of momentum transfer between the redirected steam and
the impulse blades is given by
where
☉☉ Δh is the specific enthalpy drop of the steam across the blades.
☉☉ T is the entry temperature or the stagnation temperature.
☉☉ u is the peripheral velocity of the turbine’s rotor.
☉☉ Δv is the change in the whirl velocity of the blades.
Equation 19.95 can be deduced from Euler’s equation (see Chapter 15) or from the Navier–Stokes equations directly.
The overall efficiency of the HP stage can be estimated in this way.
19.34
Reaction Blades in LP Turbines
LP turbines are normally reaction turbines where the surface area between the steam and the blades is made as large
as possible. In reaction turbines, the blades sit in a much larger volume of moving fluid, and they turn as the fluid
FIGURE 19.27 The blades from an impulse turbine are shown above, and the blades from a reaction turbine are shown in
Figure 19.28. In multistage steam turbines, a combination of these two blade designs is used.
19.36 PREVENTING CAVITATION IN RCPs 771
flows past them. In the LP stage, the passing fluid happens to be steam. A reaction turbine does not change the direc-
tion of the steam as dramatically as an impulse turbine does. Instead, it simply spins the shaft as the steam pushes
past the blades. Wind turbines are perhaps the best known examples of reaction turbines. The turbochargers in your
car are yet another.
19.35
D ifferences between Impulse Turbines and Reaction Turbines
In impulse turbines, the hot steam hits the blades and bounces off of them. In reaction turbines, the steam flows over
the blades and around them. The blades in a reaction turbine simply “go with the flow” of the steam. A picture of the
steam flow is illustrated in Figure 19.28. In general, reaction turbine blades tend to be larger than impulse turbine
blades. This is necessary to extract the maximum amount of kinetic energy from the steam. In nuclear power plants,
the primary stage is usually an impulse turbine, and the secondary stage is usually a reaction turbine. The steam
turbines in virtually all nuclear power plants are designed in this way.
19.36
P reventing Cavitation in Reactor Coolant Pumps
The coolant flowing through a reactor coolant pump is usually a single-phase liquid. This liquid is pumped through the
primary and secondary loops at very high pressures, and most of the time, the liquid pressure exceeds the saturation
pressure. However, sometimes it is possible for a transient to occur where the sudden acceleration of the liquid causes
the local pressure to fall below the saturation pressure, that is,
If the coolant is water, the water flashes into steam, and vapor bubbles appear in the flow. These bubbles can be gener-
ated in vast quantities near the tips of the impellers or even at the pump inlet when the pressure suddenly falls. The vapor
bubbles, which are called cavitation bubbles because they form “cavities” in the surrounding liquid, collapse as they are
swept away from the LP regions. As they collapse back into the flow, they generate powerful and highly destructive pres-
sure waves that propagate through the entire piping system. This process, which sometimes leads to pump performance
problems, is called cavitation, and under certain conditions, it can erode or damage the impeller blades (see Figure
19.29). Hence, cavitation must be taken into account in the design and operation of hydraulic coolant pumps and tur-
bines. In nuclear power plants, cavitation can be avoided by keeping the system pressure above the saturation pressure.
In the primary loop, this task is performed by a device called the pressurizer. Hence, reactor vendors typically specify
a minimum pressure at which a reactor is expected to be operated. For the Westinghouse AP 1000 (see Appendix J),
FIGURE 19.28 The blades from a reaction turbine are different than those from an impulse turbine because they are larger and
spin more gradually. A reaction turbine blade is more like a propeller. The main difference is that there are more vanes attached to
the turbine shaft and often multiple sets of vanes and multiple stages are used. (Pictures provided by the US Department of Energy.)
772 Reactor Coolants, Coolant Pumps, and Power Turbines
FIGURE 19.29 On the left, an example of the blade damage that can be caused by cavitation in a hydraulic turbine; on the right,
the critical cavitation numbers below which cavitation can occur with filtered or unfiltered water.
this minimum pressure is 2% below the nominal system pressure (i.e., 2,200 PSI vs. 2,250 PSI). Hence, as long as this
minimum pressure level is maintained, cavitation does not occur. The presence of cavitation in the piping system is
sometimes accompanied by a characteristic tumbling sound. If this sound becomes audible, the easiest way to stop the
ensuing cavitation is to raise the system pressure level. Example 19.9 illustrates why this is important. In both pumps
and turbines, a parameter called the cavitation number (dimensionless) is used to determine whether or not cavitation
will occur. The cavitation number Ca is defined as
where
ρ is the density of the liquid (in kg/m3).
P is the local pressure (in Pa).
PSAT is the saturation pressure (in Pa).
v is the characteristic velocity of the flow (in m/s).
The cavitation number at which cavitation begins is called the critical cavitation number σ. Below the critical cavita-
tion number, cavitation occurs, and above it, it does not. For incompressible fluids with rigid boundaries, the cavita-
tion number is a function of only the mass flow rate and the pressure. Negative cavitation numbers always result in
cavitation, and slightly positive ones can also cause it to occur. Normally, cavitation will not occur in water above
a cavitation number of 0.3–0.5.* For tap water, the critical cavitation numbers are shown on the right hand side of
Figure 19.29. In general, the critical cavitation number increases as the Reynolds number increases, and it can be
reduced by filtering and degassing the water because this decreases the number of nucleation sites. Because reactors
use water that is very pure, the critical cavitation number in a RCP is close to 0.25. Above this number, the coolant
will not generally cavitate.
* See https://authors.library.caltech.edu/25019/1/chap5.htm.
Questions for the Student 773
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi, India (2000).
Reactor Theory (Nuclear Parameters), DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
22. Write an equation for the pumping power of a RCP. What variables are actually required to determine the pumping
power?
23. How does one correct the pumping power for a pump that is not 100% efficient?
24. Where does most of this loss in efficiency come from?
25. What is a jet pump and how does it work? Why do jet pumps have no moving parts?
26. In what type of reactor are jet pumps normally found? What are the capacities of these pumps?
27. How is shaft work related to the torque supplied to a pump shaft and its rotational speed?
28. If the torque is held constant and the rotational speed is doubled, what happens to the work required to turn the
shaft?
29. How is the torque on a pump shaft related to the pump head?
30. How is the pressure head defined for a hydraulic turbine?
31. What is the design capacity of a Westinghouse 93A1 coolant pump?
32. What are two ways in which the design capacity of a RCP can be measured?
33. What is a centrifugal coolant pump and how does it differ from a jet pump?
34. What are the most common types of RCPs in nuclear power plants today?
35. How large is a typical coolant pump in the primary loop of a PWR? How much power does a pump like this
consume?
36. In reactor steam turbines, there are impulse blades, and there are reaction blades. What is the difference between
these blades, and which blades are used in the HP stage? Which blades are used in the LP stage?
37. In multistage reactor steam turbines, how much of the pressure drop occurs across the HP stage and how much of
the pressure drop occurs across the LP stage?
38. Why do the high and LP stages require different types of blades?
39. Approximately what percentage of the energy is extracted from the steam in the high-pressure stage, and approxi-
mately what percentage of the energy is extracted from the steam in the LP stage?
40. Write an expression for the “blade work” in a coolant pump or a steam turbine. Why do RCPs and steam turbines
require such large numbers of blades?
41. How can a RCP “cavitate” and what can be done to prevent this from occurring?
42. What is the BEP of a pump and how can this point be determined?
43. Why do RCPs operate close to the BEP point?
44. What parameters are required to determine the delivery rate of a RCP?
45. Fill in the following sentence with the appropriate word or phrase: When the flow through a RCP is incompressible
and isothermal, it is often more convenient to express the pump’s capacity in terms of the ________ rather than the
actual mass flow rate.
46. In general, the efficiency of a RCP is NOT a simple constant. Instead, it is a function of both the rotational speed of
the shaft and what other important hydrodynamic variable?
47. Fill in the following sentence with the appropriate word or phrase: in large PWRs, RCPs can consume about ______
MW of electrical power per pump.
48. Under what conditions can Bernoulli’s equation be used to determine the time constant of a RCP?
49. When several coolant pumps are connected in parallel, why is it important for all of the coolant pumps to be identi-
cal and have exactly the same capacities and delivery rates?
50. Fill in the following sentence with the appropriate word or phrase: when several RCPs are connected in parallel, the
______ rate is equal to the sum of the ________ rates of the individual pumps.
Exercise 19.2
Suppose the efficiency of a reactor coolant pump (or RCP) in a commercial PWR can be increased from 85% to 90%
by adopting an innovative blade design. The PWR contains four of these coolant pumps, and before the design change
is made, each coolant pump consumes approximately 6 MW of electric power. With the old blade design, the plant puts
out 1,000 MW of electric power, and its aggregate thermal efficiency is 33%. After the design change is made, what will
the new aggregate thermal efficiency of the plant be?
Exercise 19.3
A RCP is required to pump 6,000 kg of coolant through the core of a commercial PWR each second. If the water pass-
ing through the pump has a temperature of 290°C and a pressure of 2,250 PSI (15.5 MPa), what is the power required to
power the pump? Assume the efficiency of the pump is 85%.
Exercise 19.4
One of the coolant pumps most commonly used in a Westinghouse PWR is Westinghouse model number 93A1 coolant
pump. This pump has a mass flow rate of 6,200 kg/s and an operating speed of approximately 1,200 RPM. Using the
data presented earlier in the chapter, estimate the pressure head for the pump. How is the pressure head for a pump such
as this defined?
Exercise 19.5
Find the amount of electrical energy required to pump pressurized water having a density of 720 kg/m3 through the cold
leg of a commercial PWR. Assume the pump installed in the cold leg is not 100% efficient and its average efficiency is
85%. Finally, assume the pump must support an average mass flow rate of 7,500 kg/s and that the pressure head the pump
is required to support is 600 kPa (87 PSI).
Exercise 19.6
A RCP in the secondary loop of a commercial PWR is required to pump HP water to the steam generators at a rate of
500 kg/s. The density of the water is approximately 770 kg/m3. Under these conditions, what is the delivery rate of the pump?
Exercise 19.7
Water, sodium, carbon dioxide, and helium are popular reactor coolants. Compare the pumping power requirements for
these coolants under the following conditions: air at 1 MPa and 1,000°C, carbon dioxide and helium under the same
conditions, sodium at 600°C and 150 kPa, water at 300°C and 15.5 MPa, and water at 250°C and 6 MPa. Assume a mass
flow rate of 10 kg/s through a 0.2 m diameter stainless steel pipe with a pressure head of 100 kPa.
Exercise 19.8
The impellers in a large centrifugal RCP are designed to turn at 1,200 RPM. What is the speed of the shaft in radians? If
the torque on the shaft is 5,000 N m, how much power is required to turn the shaft? Assuming the pump is 80% efficient,
how much electric power is required to operate the pump?
Exercise 19.9
Two identical RCPs are required to pump different coolants with the same mass flow rate through two identical coolant
pipes with the same pressure head. One coolant is liquid sodium at 500°C and 110 kPa (16 PSI), and the other coolant is
water at 250°C and 5.5 MPa (800 PSI). Which pump requires more electric power to perform this function? What is the
percentage difference in their relative pumping powers?
Exercise 19.10
The water distribution system serving a nuclear power plant is allowed to pump water with a temperature as high as
30°C to the plant. Determine the minimum pressure that must be maintained by the system to avoid cavitation when the
toilets in the plant’s restrooms are flushed.
20
Fundamentals of Single-Phase Heat
Transfer in Nuclear Power Plants
20.1 A n Introduction to Single-Phase Nuclear Heat Transfer
Reactors rely on coolants such as water, liquid metals, and industrial gases to help cool the core. During normal operation, these coolants
are force-circulated through the core by large reactor coolant pumps (or RCPs) to keep the convective heat transfer coefficient as high
as possible. For example, in pressurized water reactors (PWRs), these coolant pumps pump the coolant through the core at velocities
approaching 5 to 6 m/s. In fast reactors, the coolant velocities can be even higher, and in some cases, they can approach 10 m/s. The pur-
pose of this chapter is to provide an introduction to the subject of single-phase nuclear heat transfer (two phase heat transfer is then dis-
cussed in Chapter 24). Single-phase heat transfer can be thought of as the flow of heat between a solid surface S and a fluid that is flowing
over that surface. In single-phase flow, the fluid (which is normally a liquid) does not change phase. Hence, the primary goal of single-
phase nuclear heat transfer is to find the value of the convective heat transfer coefficient h that appears in Newton’s Law of Convection.
In Equation 20.1, TWALL is the temperature of the surface or the wall over which the fluid flows, and TBULK is the bulk or average
temperature of the fluid. In reactor work, TWALL can represent the surface temperature of the pressure vessel or the fuel rods, and most
of the time, the temperature of the wall is much greater than that of the coolant; that is, T WALL > TBULK. When applying Equation 20.1
to reactor fuel assemblies, the fluid is usually assumed to be in motion, and for the purposes of this chapter, the fluid can be either a
liquid or a gas. Then the heat flow rate q can be deduced directly from a knowledge of h, TBULK, and T WALL. However, if the coolant
reaches its boiling point, the heat transfer process becomes considerably more complex because the liquid phase is now mixed with
the vapor phase, and the bulk temperature of the fluid becomes equal to the saturation temperature TSAT:
TBULK = TSAT (when boiling occurs) (20.2)
The fluid temperature then rises no further until all of the liquid is converted into vapor. We can summarize these observations by
saying that
For single-phase flow: TBULK < TSAT (20.3)
The process of two-phase heat transfer, where TBULK = TSAT, is then discussed in Chapters 23, 24, and 25. Here, the coolant may also
have a number of additional flow regimes after it reaches the saturation temperature. These flow regimes can be flow rate, heat flux,
geometry, and pressure dependent. Thus several convective heat transfer coefficients may have to be used to describe the convective
heat transfer process instead of just one. These correlations are usually more complicated than those for single phase fluids, and in
general, they are flow regime dependent. Now let us turn our attention to how the bulk fluid temperature in Equation 20.1 is defined.
For a homogeneous two phase mixture where the phases are in thermal equilibrium, this temperature is simply TBULK = TSAT.
777
778 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
☉☉ T is the temperature of the surface of the object which transfers heat to the fluid (in °C).
☉☉ TBULK is the bulk fluid temperature (in °C) defined below.
☉☉ h is the convective heat transfer coefficient (in W/cm2 °C).
In some countries, the convective heat transfer coefficient is expressed in kW/m2 °K or BTU/h ft2 °F. The convective heat
transfer coefficient can be converted from W/cm2 °C to kW/m2 °K by multiplying by 10, and it can be converted from
kW/m2 °K to BTU/h ft2 °K by multiplying by 317. Thus, converting from W/cm2 °C to BTU/h ft2 °K requires multiplying
by 10 × 317 = 3,170, and converting from W/cm2 °C to BTU/h ft2 °F requires multiplying by 5/9 × 3,170 = 1,761.1. Some
additional unit conversions that are sometimes needed are shown below:
Sometimes Newton’s Law of Convection is written in a slightly different form involves using the surface heat flux
q″ = q/A instead of the total heat flow rate q. Then, we may write Newton’s Law of Convection (Equation 20.1) as
where A is the area of the surface through which the heat is flowing and q″ is the nuclear heat flux. In our earlier discus-
sions, we assumed that a value for h existed that allowed us to predict the heat flux q″ = q/A from a surface having a wall
temperature T WALL and a bulk fluid temperature TBULK. In order to do this, Newton’s Law of Convection required us to
define the bulk temperature or average temperature of the coolant as
R R
TBULK (z) =
∫ 0
T(r,z)2πrρv(r,z) dr
∫ 0
2πrρv(r,z) dr (20.6)
where this definition represents the velocity-weighted average of the fluid temperature profile, and R is the radius of the
coolant channel (if it happens to be circular). Without knowing the specific shape of the velocity profile v(r, z), we cannot
define the bulk fluid temperature precisely.
Hence, different velocity profiles can lead to different values of TBULK(z), and consequently, the velocity profile v(r, 0)
at the entrance to a pipe can lead to different values for TBULK(z) than when the flow is fully developed (see Figure 20.1)
and its shape is different. A similar conclusion applies to partially developed flows as well. For this reason, one must be
FIGURE 20.1 When fluid flows over a solid surface, the fluid interacts with the surface and creates a boundary layer. In practice,
it takes a while for the boundary layer to become fully developed. At the entrance of a reactor coolant channel, the boundary layer
is not fully developed. This region is called the entrance region to the channel. Then, if the channel becomes long enough, he flow
becomes fully developed and the shape of the boundary layer no longer changes. In reactor fuel assemblies, it generally takes about
1 m for the boundary layer to become fully developed.
20.3 Laminar Heat Transfer 779
careful to apply Newton’s Law of Convection to systems where the bulk fluid temperature is calculated according to
Equation 20.6 and the flow is fully developed. Otherwise, the heat transfer rate between the wall and the coolant may not
be the value we want to use. As far as reactors are concerned, Newton’s Law does not tell us anything about the direc-
tion of the heat flux q″ except that the heat must flow from a region of high temperature to a region of low temperature.
If TWALL > TBULK, the heat will flow from the surface of the wall to the coolant (which is cooler), and if TWALL < TBULK,
the heat will flow in the opposite direction. The single-phase heat transfer coefficient h then becomes the constant of
proportionality that relates the surface heat flux to the temperature difference ΔT between the wall and the coolant. By
convention, the flow of heat into the coolant causes the heat flux to have a positive value, and the flow of heat out of the
coolant causes the heat flux to have a negative value. This convention is commonly used in reactor work as well.
FIGURE 20.2 The flow over a simple object such as a flat plate can be both laminar and turbulent. For example, when one
examines the flow of fluid over a flat plate, the boundary layer next to the surface of the plate can contain both laminar and
turbulent regions. The flow in the velocity boundary layer starts out laminar, but if the plate is sufficiently long, the flow will
also become turbulent.
* See Kays and Crawford, Convective Heat and Mass Transfer, 2nd Edition, McGraw-Hill, New York, NY (1980).
780 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
FIGURE 20.3 The boundary layers for laminar and turbulent flow over a nuclear fuel rod. In reactors, most nuclear fuel rods
are oriented vertically rather than horizontally (the CANada Deuterium Uranium reactor is the sole exception), and the flow is
designed to be as turbulent as possible. The velocity of the fluid at the surface of a fuel rod is always zero because of the no-slip
boundary condition.
where C is a constant of proportionality that is geometry dependent, and k is the thermal conductivity of the coolant.
In other words, laminar heat transfer is a relatively simple process once the shape of the velocity profile in the bound-
ary layer is known. In circular pipes, this velocity profile is parabolic once the flow becomes fully developed, and this
profile can be used to derive exact expressions for the convective heat transfer coefficient in simple flow channels like
rectangular, triangular, and elliptical ones. In more complicated ones, the heat transfer coefficient may have to be deter-
mined experimentally.
Because of the effects of this turbulence, laminar flows have lower heat transfer coefficients than turbulent ones:
TABLE 20.1
The Convective Heat Transfer Coefficients for Even Common Reactor Coolants Can Vary Greatly Depending upon
Whether the Flow Is Natural or Forced
For water, h can vary by a factor of more than 100 depending on the size of the Reynolds number, and whether the flow
is laminar or turbulent (and single phase or two phase). Metallic coolants such as liquid sodium have higher heat trans-
fer coefficients than water, and water has a much higher heat transfer coefficient than gaseous coolants such as helium
and CO2. The heat transfer coefficients for several common reactor coolants are shown in Table 20.1. The values of the
convective heat transfer coefficients also depend on whether the flow is natural or forced.
The relative values of hCONDUCTION and hCONVECTION determine whether the correlations for the convective heat transfer
coefficients have one or two parts. See Figure 20.4. For water and many industrial gases, hCONDUCTION ≪ hCONVECTION, and
for metallic coolants like liquid sodium, hCONDUCTION ≥ hCONVECTION. Thus, water and gaseous coolants have convective
heat transfer coefficients that are primarily the result of convection
because their thermal conductivity is so low, while metallic coolants such as liquid sodium and mercury have convective
heat transfer coefficients that consist of two distinct parts:
because conduction and convection are equally important. For metallic coolants, hCONDUCTION can be a very large num-
ber. Its value depends heavily on the boundary conditions at the wall surface, and it also depends on the shape of the
coolant temperature profile. In fast reactor fuel assemblies cooled with liquid metals, the conductive component of the
heat transfer coefficient can sometimes become as large as the convective component. In fact, in Section 20.27, we will
compare the size of these components directly. Finally, for water and most industrial gases, hCONDUCTION ≪ hCONVECTION,
and so it is common practice (at least when the flow becomes turbulent) to neglect the effects of conduction entirely.
Then, this case, the convective heat transfer coefficient has just one component which can be written as
The value of hCONVECTION then depends on the value of the Reynolds number.
782 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
where the subscript SP refers to single phase, and the subscript TP refers to two phase. For gas-cooled reactors,
FIGURE 20.4 A coolant flowing over a stationary nuclear fuel rod comes to a complete stop at the surface of the rod. Heat transfer
from the rod surface to the surrounding coolant can occur by conduction or convection. Most reactor heat transfer is based on forced
convection, but under certain circumstances, natural convection or even conduction can occur. Conduction is particularly important
for metallic coolants because their thermal conductivity is so high.
20.7 Factors That Affect the Convective Heat Transfer Coefficient 783
The values of h for common liquid metals, gases, and single-phase and two-phase water are shown in Table 20.1. Under
some conditions, the reader can see that a variation in the value of h by as much as a factor of 10,000 is possible. For this
reason, finding the correct value for h is fundamental to understanding how quickly heat can be removed from the core.
Finally, the value of h can depend upon several factors including
☉☉ The Reynolds number
☉☉ The roughness of the heating or cooling/surface
☉☉ The geometry of the flow channel (round, rectangular, other)
☉☉ Coolant density changes
☉☉ The coolant thermal conductivity k
☉☉ The coolant viscosity μ
☉☉ The pitch-to-diameter ratio for a given reactor geometry (square, hexagonal, or triangular)
☉☉ The presence of wire-wrap spacers or support grids.
and whether or not other turbulent mixing devices (e.g., fins, vanes, orifice plates, or injectors) are used. Reactors take
advantage of several of these devices to optimize the size of the heat transfer coefficient. An extensive summary of these
coolants and their physical characteristics was presented in Chapter 19, where their virtues and vices were discussed in
great detail.
An Important Observation
The higher the Reynolds number, the larger the convective heat transfer coefficient will be.
In practice, this means that turbulent flows have higher heat transfer coefficients than laminar flows (hTURBULENT > h LAMI-
NAR). This specific functional dependence is shown in Figure 20.5. At low flow rates, the flow becomes laminar, and the
heat transfer coefficient is independent of the Reynolds number because no turbulent mixing can occur. That is to say,
when the flow becomes fully developed
where C is a constant that depends on the coolant and the geometry of the flow channel. However, as soon as the flow
becomes turbulent, the heat coefficient changes. It then becomes proportional to the Reynolds number raised to some
power of N
FIGURE 20.5 The behavior of the convective heat transfer coefficient in a reactor fuel assembly as a function of the Reynolds
number. For fully developed flows, the heat transfer coefficient increases as the 0.8 power of the Reynolds number for turbulent flow
and as the zeroth power of the Reynolds number for laminar flows. In other words, for fully developed laminar flows, the convective
heat transfer coefficient is independent of the Reynolds number.
Here C has a different value than it does for laminar flows, and for turbulent flows except those involving liquid met-
als, N ≈ 0.8. Now let us turn our attention to the thermal conductivity k, which appears in both Equation 20.17 and
20.18. Reactor coolants such as water have thermal conductivities of about 0.5 W/m °C at 315°C, and other coolants
such as liquid sodium have a thermal conductivity of about 62.6 W/m °C at 540°C. Finally, gases such as CO2 have a
thermal conductivity of about 0.0577 W/m °C at 500 °C. Normally, their thermal conductivities are quoted in W/m °C
(see Table 20.2). Thus, the heat transfer coefficients in reactor fuel assemblies are directly proportional to the value of
k and, for turbulent flows, to the Reynolds number raised to the four-fifth power. Hence the value of h can be thought
of as varying directly with k and with v0.8 for single phase fluids (expect liquid metals). Because of their high thermal
conductivity, metallic coolants have much higher convective heat transfer coefficients than other coolants do. (A more
extensive discussion of their properties is presented in Chapter 19). Because the thermal conductivity of these coolants
is so high (for example, a factor of almost 100 greater than that of water), an additional term must be added to the cor-
relations for their heat transfer coefficients to account for the effects of this higher thermal conductivity. Then a typical
correlation for the convective heat transfer coefficient becomes
h = (k/D) Nu (20.18b)
TABLE 20.2
The Physical Properties for Some Common Reactor Coolants
Note: For water, the physical properties are evaluated at 310°C and 15.5 MPa. The physical properties for liquid sodium are evalu-
ated at 535°C and 0.1 MPa. The properties of carbon dioxide and helium are evaluated at 800°C and 20 MPa.
20.9 An Introduction to the Prandtl Number 785
where
Nu = C(Re × Pr)N + D, (20.18c)
N is an exponent close to 0.8, C is about 0.025, and D is a dimensionless constant having a value of about 5 for a constant
wall temperature and 7 for a constant heat flux. A more extensive discussion of liquid metal behavior is presented in
Section 20.27. Thus liquid metals combine the effects of both conduction and convection in their expressions for the
convective heat transfer coefficient h (see Equation 20.12).
Prandtl Number
Pr = µc/k (dimensionless) (20.19)
786 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
FIGURE 20.6 A picture of German physicist Ludwig Prandtl (left) circa 1920 and a convective flow based on the use of the
Prandtl number (right) which he helped to invent. (Pictures provided by Wikipedia.)
The value for the Prandtl number can be found by inserting the appropriate values for μ, c, and k into the equation above.
Note in particular that the Prandtl number depends on only three physical properties of the coolant:
☉☉ The viscosity μ (in kg/ms)
☉☉ The specific heat c at constant pressure (in kJ/kg °C)
☉☉ The thermal conductivity k (in W/m °C)
Here, the specific heat is defined as the energy (in J or kJ) required to raise the temperature of a unit mass of fluid (which can
be expressed in grams or kilograms) by 1°C. Because these properties are temperature dependent, the Prandtl number is tem-
perature dependent as well. In reactor work, the Prandtl number is essentially fixed by the choice of the coolant. Sometimes it
is even tabulated as a physical property (see the discussion about this in the next section), and lists of temperature dependent
Prandtl numbers can be found in many heat transfer books. For water, these values can also be found in the Steam Tables. The
Steam Tables are then presented as separate temperature tables and pressure tables. Some nice examples of these tables can
be found in the references at the end of the chapter, and specifically, in Cengel and Cimbala (2006).
Kinematic Viscosity
ν = µ /ρ (dimensionless) (20.20)
Thermal Diffusivity
α = k/ρc (dimensionless) (20.21)
The kinematic viscosity ν measures the rate of momentum transfer between the molecules of a fluid, and it primarily
affects the shape of the velocity gradient. The thermal diffusivity is the ratio of heat transfer rate to the energy storage
rate, and it primarily affects the shape of the temperature gradient. Because the Prandtl number is defined as the ratio
of the kinematic viscosity to the thermal diffusivity
Pr = ν /α = µc/k (20.22)
the Prandtl number therefore relates the velocity gradient to the temperature gradient. Low values for the Prandtl num-
ber (Pr ≪ 1.0) imply that the velocity gradient is much larger than the temperature gradient, and this is generally true
for metallic coolants. Conversely, Prandtl numbers closer to 1.0 mean that the temperature gradients and the velocity
gradients are comparable in size, whereas Prandtl numbers much greater than 1.0 imply that the temperature gradient is
much steeper than the velocity gradient. These observations are summarized in Table 20.3.
20.12 A Physical Interpretation of the Prandtl Number 787
TABLE 20.3
Prandtl Numbers and Their Physical Interpretations for Some Popular Reactor Coolants
For fully developed turbulent flows, the radial heat flux q″ in a circular pipe is q″ = −k″dT/dr, where k″ = k + ρcεH is the
effective thermal conductivity, and εH is the turbulent diffusivity of heat. The turbulent diffusivity has the dimensions of
square meters per second. Thus, the effective thermal conductivity k″ can be thought of as the sum of two processes—
(1) molecular conduction, which is represented by the thermal conductivity k, and (2) turbulent conduction, which is
represented by the term “ρcεH.” Noting that the thermal diffusivity α can be written as α = k/ρc, we can also write the
effective thermal conductivity as
and so the temperature gradient dT/dr is controlled primarily by the value of α. Now let us see how these parameters
affect the velocity gradient. The second parameter in the Prandtl number is the kinematic viscosity ν. We can write
TABLE 20.4
Prandtl Numbers Are Generally Lower at Higher Temperatures Because the Viscosity of Most Fluids Falls More Quickly
than the Heat Capacity or Thermal Conductivity as the Fluid Temperature Is Increased
an analogous equation to describe the diffusion of momentum between the surface of the wall and the coolant. This
equation is
τ = µ ′′dv/dr (20.25)
where τ is the shear force (drag force per unit area), in N/m2, and μ″ is the effective turbulent viscosity of the fluid. Using
a similar dimensional analysis, it can be shown that the effective turbulent viscosity can also be written as the sum of
two terms:
µ ′′ = µ + ρε Μ (20.26)
The first is the kinematic viscosity μ and the second is the product of the density ρ and the turbulent diffusivity of
momentum εM. Thus, the effective turbulent viscosity μ″ is also the sum of two terms—the normal dynamic viscosity μ,
which is a tabulated quantity, and a turbulent viscosity ρεM, which is due to the turbulence of the fluid. Noting that the
kinematic viscosity ν is given by ν = μ/ρ, we can also write the effective turbulent viscosity as
and so the fluid velocity gradient dv/dr is controlled primarily by the value of μ″. Both the turbulent diffusivity for
momentum and the turbulent diffusivity for heat (εM and εH) have the dimensions of length2 per unit time, and there is a
good reason to believe that their values are about the same for turbulent flows. For laminar flows, their values are equal
to 0, and for highly turbulent ones, their values are much greater than either ν or α. For metallic coolants, the values of
α and εH are believed to be similar.
and
Substituting the appropriate values for μ″ and k″, and noting that (dT/dr)/(dv/dr) = dT/dv, we then find that
Equation 20.33 is called the Reynolds analogy between heat and momentum transfer. It states that the ratio of the
temperature gradient to the velocity gradient is proportional to the heat flux q″ divided by the wall shear stress τ. The
constant of proportionality is c−1 where c is the fluid’s specific heat. Equation 20.33 can then be used to evaluate the heat
transfer rates for highly turbulent flows. It is particularly helpful for coolants where the Prandtl number is close to 1.0.
20.15 The Thermal Boundary Layer 789
FIGURE 20.7 A picture slowing the development of the velocity boundary layer over a plate-type fuel rod. In this case, the thick-
ness of the boundary layer is represented by the symbol δV. The velocity boundary layer becomes thicker the further one moves
away from the leading edge of the rod. The thickness δV(x) at any point is defined as the distance from the surface where v = 0.99v∞
or 99% of the free stream velocity v∞.
790 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
FIGURE 20.8 The thermal boundary layer next to a nuclear fuel rod. In this case, the rod is hotter than the surrounding coolant so
TROD > T∞.
gradient dT/dy becomes zero. The region above the surface of the rod where the temperature gradient normal to the
surface of the rod is large enough to be important is called the thermal boundary layer. Like the velocity boundary
layer, the thermal boundary layer is defined as the distance δT from the surface of the rod where the temperature profile
normal to the surface no longer changes with increasing distance. The thickness of the thermal boundary layer is then
which is again the point where the fluid temperature reaches 99% of its free stream value. The size and shape of the
radial temperature gradient dT/dy in the thermal boundary layer dictates the conductive heat transfer rate at the rod’s
surface. Hence, whenever a reactor coolant flows past the rod, both the thermal and velocity boundary layers will
develop simultaneously, but depending on the thermal conductivity and the viscosity, they may develop at different rates.
Thus, the Prandtl number
Prandtl Number
Pr = Momentum diffusivity/Thermal diffusivity = Cp (µ /k) (dimensionless) (20.35)
attempts to represent this difference in their rate of development. Hence, it represents the relative thickness of the veloc-
ity boundary layer to the thermal boundary layer:
For this reason, the Prandtl number is a convenient way to determine how the physical properties of the coolant
affect the value of the convective heat transfer coefficient. Next, we would like to show exactly how the value of h can
be found.
FIGURE 20.9 A picture of Wilhelm Nusselt (left) and the campus at the Technical University of Munchen, in Munchen, Germany,
where Nusselt did most of his work. Nusselt invented the Nusselt number which is widely used in convective heat transfer today.
(Pictures provided by Wikipedia.)
Like the Reynolds number and the Prandtl number, the Nusselt number is dimensionless, and for specific reactor cool-
ants, it is a function of position for developing flows, and a simple constant for fully developed ones. Moreover, it is
independent of any unit system and it is scale invariant. A picture of Wilhelm Nusselt who first proposed it around 1900
is shown in Figure 20.9. For fluids in motion, the Nusselt number measures the ratio of the convective heat transfer rate
to the conductive heat transfer rate. For fully developed laminar flows in reactor coolant channels, Nu ≤ 8. However,
when the flow becomes turbulent, the values for Nu can be considerably higher.
Nusselt numbers normally fall into several well-defined categories depending on the reactor coolant that is used. First,
let us discuss how Nusselt numbers behave in water reactors. For low-speed turbulent flows, they can range from 10 to
500. Nusselt numbers for high-speed turbulent flows (such as those in PWR fuel assemblies) can have values between
500 and 2,000. In LMFBR cores, Nusselt numbers can range from 5 to about 15. (Some actual values are found in
Section 20.27). The Nusselt numbers are lower in this case because the thermal conductivity of the liquid sodium is so
high.
These Nusselt numbers are compared in Table 20.5. It can be seen that although the Nusselt numbers for liquid metals
are very low, this does NOT imply that the convective heat transfer coefficients for these metals are low. In fact, liquid
metals almost always have higher convective heat transfer coefficients than other coolants do. For Nusselt numbers
close to unity, conduction and convection are equally important but larger values of Nu imply that convection is more
TABLE 20.5
Typical Nusselt Numbers and Heat Transfer Coefficients for Water Reactors and LMFBRs
Type of Reactor Type of Flow Nusselt Number Heat Transfer Coefficient (W/m2 °C)
Water reactors Laminar 2–10 80–400
(single phase) Low-speed turbulent 10–500 400–20,000
High-speed turbulent 500–1,000 20,000–40,000
Fast reactors Laminar 4–5 60,000–80,000
(metallic coolants) Low-speed turbulent 5–7 80,000–100,000
High-speed turbulent 7–15 100,000–120,000
792 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
important than conduction. The Nusselt number is a function of both the Reynolds number and the Prandtl number. It is
also related to the convective heat transfer coefficient by the expression
where
☉☉ L is the characteristic length (in cm or m).
☉☉ k is the thermal conductivity of the fluid (in W/m °C).
☉☉ h is the convective heat transfer coefficient (in W/m2 °C).
Hence, the Nusselt number Nu is dimensionless, and the units of h are determined by the units of k and L. The convec-
tive heat transfer coefficient can then be found by solving Equation 20.39 for h:
This definition establishes a direct relationship between the value of h and the value of Nu. In fact, Equation 20.39
implies that the convective heat transfer coefficient is directly proportional to the value of Nu when the thermal con-
ductivity is constant. The selection of the characteristic length L depends on the geometry of the flow field. For a coolant
channel in a reactor fuel assembly, the characteristic length is the equivalent diameter De of the channel. In the case of
a horizontal bank of tubes, it is the outer diameter of the tubes, and in the case of natural convection, it is the length of
the vertical surface undergoing the convection; for a spherical object, it is the diameter of the sphere. For more complex
shapes, the length may be defined as the volume of the fluid body divided by the total surface area. Here, the thermal
conductivity is evaluated at the film temperature, which for engineering purposes is generally equal to the average of
the bulk fluid temperature and rod surface temperature. The characteristic length scales used to define the convective
heat transfer coefficient are summarized in Table 20.6. Other definitions are sometimes used for more intricate shapes.
we can write
h = (k/L)Nu (20.42)
TABLE 20.6
The Length Scales Used in the Definition of the Convective Heat Transfer Coefficient for Single-Phase Flows
where the value of h in this case is assumed to be the average value of the convective heat transfer coefficient along the
flow channel in question. Specifically, for a coolant channel of height H, the average value of h is
H H H
h = (k/L) ⋅
∫ 0
Nu(z) dz
∫ 0
dz = (k/L) ⋅
∫ 0
Nu(z) dz/H (20.43)
where H represents the subchannel height and L = DH. In most fuel assemblies, the coolant flow is vertical, and H can
be interpreted as the active core height. For forced convection, the average Nusselt number can then be expressed as
a function of two other dimensionless numbers—the Reynolds number and the Prandtl number, Nu = f(Re, Pr). For
natural convection, it is expressed as a function of the Grashof number Gr and the Prandtl number, Nu = f(Gr, Pr).
The product of the Grashof number and the Prandtl number is also called the Rayleigh number Ra. Hence, Nu = f(Ra).
The mass transfer analog of the Nusselt number is then the Sherwood number. Of course, all of these numbers are
dimensionless, and they are just different ways to define the convective heat transfer coefficient. For most reactor
coolants, the value of h is directly proportional to some power of the Nusselt number; that is, h ∝ Nu M, where N is
an empirically determined constant having a value close to 1.0. For highly turbulent flows, the Nusselt number is
usually written as the product of the Reynolds number and the Prandtl number—each raised to a different power.
Sometimes these powers can be the same, and sometimes they can be different. Hence, the average Nusselt number
can be written as follows:
where for most coolants, a ≠ b, C is a constant of proportionality that is fluid dependent, and D is another constant that
arises from the treatment of liquid metals. For liquid metals, D ≅ 5 to 7, and for water and gases, D ≅ 0. The advan-
tage to this approach is that it allows us to model virtually any coolant that a nuclear reactor can use. The values of
a, b, C, and D can also be different, but the underlying logic is the same. It is important to realize that Equation 20.44
requires the reactor coolant pumps to be turned on so that the convection is forced. If the pumps are turned off, and the
flow becomes natural, then the Nusselt number can no longer be expressed in the same way. Then, the Nusselt number
becomes constant, and C becomes a function of just the geometry of the flow channel:
For nonmetallic coolants, the most common practice is to set a = 0.8, b = 0.4, C = 0.023, and D = 0. This leads to what
is called the Dittus–Boelter correlation:
The Dittus–Boelter correlation: is one of the most famous correlations in all of nuclear heat transfer. We will have
more to say about its origins in Chapter 22.
Solution The Reynolds number for this coolant is Re = ρvD/μ = (1,000 × 5 × 0.01)/0.0001 = 500,000, and the Prandtl
number is Pr = μc/k = (0.0001 × 5,000)/0.50 = 1. Since the flow is turbulent, and the coolant is nonmetallic (due to its
high Prandtl number), we can calculate the Nusselt number from the Dittus–Boelter correlation:
The result is Nu = 0.023(500,000)0.8·10.4 = 0.023 × 36,240 × 1 = 833.5. The Nusselt number for water under similar
c onditions is between 500 and 1,000. Therefore, this can be a very good reactor coolant. [Ans.]
20.19 Finding the Nusselt Number for a Constant Axial Heat Flux
In general, both q and q″ are functions of axial position, but when the heat flux at the surface of a fuel rod is constant,
then q(z) = q, and it is possible to derive an exact expression for h. Using Equation 20.41 as our guide, the relationship
between the Nusselt number and the heat transfer coefficient is
For a circular pipe, the characteristic length L which appears in the definition of the Nusselt number is
L = D = 2R (20.48)
Then we have
However, when the flow becomes fully developed, h(z) becomes independent of position, and Nu(z) becomes indepen-
dent of position as well. Then we can write
Nu = h ⋅ (2R/k) (20.50)
To find h, all we need are the radial temperature and velocity profiles T(r) and v(r) inside the pipe. In Chapter 15, we
showed that we could write the radial velocity profile for fully developed flow in a circular pipe as
where v was the average velocity of the fluid in the pipe, and R was the pipe radius. From the energy equation, it is also
possible to show that
where dT/dz is the axial temperature gradient. If the temperature gradient along the surface of a fuel rod is constant,
then dT/dz will also be constant. Now suppose that we substitute this expression for q″ into Newton’s Law (Equation
20.47a). Since q = q″·πR2, and
that
TBULK = TROD − 11/48 ⋅ρvcR 2 /k ⋅ (dT/dz) (20.55)
Hence,
When we simplify this equation a bit, most of the terms cancel, and all we are left with is
In other words, we have just shown that the convective heat transfer coefficient in a long circular pipe where the flow is
laminar and fully developed is
( )
Nu = D e k h (20.59)
This is a remarkable result because all we needed to derive it was the energy equation, the velocity and temperature
profiles, and the definition of the bulk fluid temperature. A similar procedure can be applied to square pipes, r ectangular
pipes, triangular pipes, and long, wide parallel plates. The value of Nu turns out to be slightly different for these shapes,
but as long as the flow is laminar and fully developed, exact values for Nu can be found in this way. Table 20.7 shows the
Nusselt numbers for two different scenarios that often arise in reactor work:
1. Laminar flow through differently coolant channels shaped having a constant surface heat flux in the axial
direction (q″(z) = constant and Re < 2,300)
2. Laminar flow through differently coolant channels shaped having a constant fuel rod surface temperature in the
axial direction (T ROD(z) = constant and Re < 2,300)
In each case, the Nusselt number is higher when the heat flux is constant than when the wall temperature is con-
stant. For example, when a pipe is circular in shape, Nu = 4.36 when the heat flux is constant, and Nu = 3.66 when
796 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
TABLE 20.7
Some Representative Nusselt Numbers for Laminar Flow (Forced Convection)
Shape of Flow Channel Aspect Ratio (L/W) Nu (q″ = Constant) Nu (TROD = Constant)
Circular 1:1 4.36 3.66
Square 1:1 3.61 2.98
Rectangular 1.4:1 3.73 3.08
Rectangular 2:1 4.12 3.39
Rectangular 3:1 4.79 3.96
Rectangular 4:1 5.33 4.44
Rectangular 8:1 6.49 5.60
Isosceles triangle 1:1:1 3.00 2.35
Flat parallel plates (infinite) ∞ 8.24 7.54
From Kays and Crawford, Convective Heat and Mass Transfer (1980).
the wall temperature is constant. Moreover, fully developed laminar flows always have Nusselt numbers that are
independent of position. However, when the flow becomes turbulent and fully developed, the Nusselt number is NOT
constant because it depends on the values of both Re and Pr:
Nu = f(Re,Pr) (20.61)
Then it is not possible to develop such simple expressions for the values of h and Nu.
FIGURE 20.10 The Nusselt numbers for fully developed laminar flows in hexagonal fuel assemblies. In an LMFBR fuel assembly,
the P/D ratio is close to 1.15, and this implies that the Nusselt number is close to 7.5. In general, the axial Nusselt number is slightly
higher for a uniform heat flux than for a uniform wall temperature. (Todreas and Kazimi.)
20.21 The Entrance Length for Laminar Flows 797
shows how the laminar Nusselt number behaves for a hexagonal lattice as a function of the P/D ratio. In most fuel
assemblies, the P/D ratio is less than 1.35, and so laminar Nusselt numbers in these geometries rarely exceed 15. This
behavior can be seen in both PWR and LMFBR cores. However, when the coolant boils, the Nusselt numbers can no
longer be found using this approach, and so, they must be found experimentally.
Suppose that the Reynolds number is 200 and the Prandtl number is 1.0. Using these values, we find that
where De is the equivalent diameter which must be calculated from the equations presented in Chapter 17. Usually
the entrance length becomes longer as the flow becomes turbulent. Then when, the flow becomes fully developed, the
convective heat transfer coefficient approaches an asymptotic value which is illustrated in Figure 20.11. First, it reaches
this value about half a meter from the entrance to the core, and then, when the flow encounters a grid spacer, it also
takes several centimeters beyond the location of the spacer before the flow becomes fully developed again. As we will
then discuss in Appendix I and Chapter 27, any additional perturbations to the flow also affect the value of the critical
heat flux immediately downstream of the grid spacers because this agitation increases the turbulent mixing coefficient
and also causes the value of the Departure from Nucleate Boiling Ratio (or DNBR) to change. We will have more to say
about exactly how the critical heat flux is affected by these obstructions in Chapter 27.
798 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
FIGURE 20.11 In water reactors, the flow becomes fully developed, and the convective heat transfer coefficient approaches an
asymptotic value of about 0.5 m above the entrance to a fuel assembly. It also takes a small distance beyond the grid spacers for the
flow to become fully developed again. Here the areas shown in white correspond to those where the flow is fully developed, and the
areas in light blue correspond to those where the flow is still developing.
where k is the thermal conductivity of the coolant, and De is the equivalent diameter of the flow channel. This equation
can be thought of as an expression for the transfer of heat to a moving fluid, and we can establish a relationship between
the nuclear heat flux q’’ = q/A and the temperature gradient dT/dr in the boundary layer by realizing that
or
q = hA ( T − TBULK ) = ( kA ) ⋅ Nu ⋅ ( T − TBULK ) D e (20.65)
Thus, the Nusselt number can be thought of as the constant of proportionality that relates the temperature gradient dT/
dr to the heat flux q″. In other words, the Nusselt number can be used as a “proxy” (i.e., a nondimensional heat transfer
coefficient) that extends Fourier’s Law of Conduction to convective flows. To make this occur, we simply replace the
value of k (which appears in Fourier’s Law) with k′ = k Nu/2. Then the effects of both conduction and convection can
be lumped into a single value of k’, which depends on only the value of Nu! It turns out that the same procedure also
works for two-phase flows, and we will have more to say about this in Chapter 24. Now let us take a look at the Nusselt
number in a slightly different way.
where the value for D is zero for coolants other than liquid metals. For water, the heat transfer coefficient is then
given by
where a and b are the coefficients shown in Table 20.8. If a ≅ b, and we refer to the exponent they represent by the symbol
N, then we can write
In this expression,
☉☉ The first term “C(De)N−1” depends on the geometry of the core and it is called the geometry term.
☉☉ The second term “(k)1−NcN” depends on the physical properties of the coolant and it is called the property term.
☉☉ The third term “(ρv)N” depends on the coolant velocity and it is called the operational term.
These three terms then determine how the convective heat transfer coefficient behaves. The first term is geometry
dependent, and it depends on only the physical characteristics of the fuel assembly. The second term depends on the
TABLE 20.8
The Values of the Coefficients a, b, C, and D Used to Find the Nusselt Numbers for Popular Reactor Coolants
Type of Reactor Type of Flow Nusselt Number Coefficients and Their Values
Water reactors (single phase) Laminar Nu = C(Re)a × (Pr)b + D C = 1–10; D = 0; a = 0, b = 0
Turbulent Nu = C(Re)a × (Pr)b + D C ≈ 0.025; D = 0; a = 0.8, b = 0.4
Fast reactors (metallic coolants) Laminar Nu = C(Re) × (Pr) + D
a b C = 0; D = 5–7; a = 0, b = 0
Turbulent Nu = C(Re)a × (Pr)b + D C ≈ 0.025; D = 5–7; a = 0.8, b = 0.8
800 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
type of coolant that is used (because it contains the thermal conductivity), and the third term depends on how the reactor
is operated. The first two terms are essentially fixed by the design of the core, so the value of h is affected primarily by
how fast the coolant flows. If the coolant velocity is high, h will be high, and if the coolant velocity is low, h will be low.
Then we can write
Equation 20.72 also implies that the turbulent heat transfer coefficient is proportional to the mass flux G = m/A
= ρv:
Hence, to a first approximation, the “operational term” dominates thermal behavior of the core, and it is the only real
“lever” a reactor operator has (except for moving the control rods) to control the heat transfer rate. The density ρ of the
coolant in a BWR can also be controlled by varying the speed of the recirculation pumps, which then changes the value
of G as well.
where the value of N is N = 0.4 when the coolant is gaining energy, and N = 0.3 when the coolant is losing energy.
Because the operational term depends on the Reynolds number, the convective heat transfer coefficient increases with
the four-fifth power of the mass flux G:
This means that, if we wish to double the convective heat transfer coefficient when the flow becomes turbulent, we
must increase the mass flux by a factor of more than 2! Example Problem 20.5 calculates the change that is required.
(Also see Figure 20.12).
FIGURE 20.12 The convective heat transfer coefficient does not change while the flow is laminar because the boundary layer
remains intact. However, when the flow becomes turbulent, the heat transfer coefficient increases with the Reynolds number.
This is true for both water and liquid metals.
In the majority of these correlations, N has a value close to 0.8, C is about 0.025, and D is a dimensionless constant
having a value of about 5 for a constant wall temperature and 7 for a constant heat flux. In Equation 20.76, the symbol
Pe refers to another dimensionless number called the Peclet number. The Peclet number is defined as the product of the
Reynolds number and the Prandtl number:
Peclet Number
Pe = Re ⋅ Pr = ρcvD e /k (20.77)
Hence, for metallic coolants, the Peclet number has no direct dependence on the viscosity. Like the Reynolds num-
ber and the Prandtl number, the Peclet number is both dimensionless and scale invariant. It represents the ratio
of the bulk heat transfer coefficient to the conductive heat transfer coefficient. The Peclet number was named for
French physicist Jean Claude Eugène Péclet, who first proposed it in 1820. In most LMFBR fuel assemblies (see
Example 20.6 below), the Peclet number has a value between 500 and 2,000. If the Peclet number in these assemblies
were 1500, the convective component of the Nusselt number would be about 8.7. Since the conductive component,
which is represented by D, is usually 5 to 7, the convective component actually exceeds the conductive component
when the core is at full power. However, when the pump speed is reduced, the values of these two components are
assumed to be comparable.
FIGURE 20.13 At low flow rates, metallic coolants have completely different velocity and temperature profiles in the radial
d irection than water or gases. For fully developed flows, this means that the Nusselt numbers are c ompletely different when the
Reynolds number becomes low and that a metallic coolant will have a higher heat transfer coefficient even when the Reynolds
number falls to zero.
FIGURE 20.14 The variation of the convective heat transfer coefficient h(z) along the surface of a nuclear fuel rod. In this case,
the axial mass flow rate is very low (approximating natural circulation conditions), so the flow over the surface of the rod has both
laminar and turbulent components.
804 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
For the idealized case where the Prandtl number is unity (or very close to it), the equation for the Nusselt number
reduces to
Nu = C ⋅ Re (20.80)
However, using the Reynolds analogy between the velocity boundary layer and the thermal boundary layer (see Section
20.13), we find that the value of C for a flat surface can be written as
C = CD /2 (20.81)
where CD is now the fluid drag coefficient for the surface. In other words, the Nusselt becomes
Nu = CD /2 ⋅ (Re) (20.82)
For a flat plate, this is called the Reynolds analogy between heat and momentum transfer. Specifically, it shows that a
dynamic similarity exists between the velocity field and the temperature field when the flow becomes fully developed.
Hence by inverting Equation 20.83, we can deduce the drag coefficient from the heat transfer coefficient if the heat
transfer coefficient is known. This allows us to find the shear stress the fluid exerts on the plate.
( )
FROD = CD A ρv 2 2 (20.84)
where FROD is the drag force in Newton, A is the surface area of the rod, and v is the average fluid velocity in the coolant
channel. In a fuel assembly consisting of N individual rods, the drag force is then
( )
FASSEMBLY = NFROD = NCD A ρv 2 2 (20.85)
Notice that the surface area A for a cylindrical rod of diameter D and length L is given by
A = πDL (20.86)
Hence, the total drag force becomes greater as the length of the fuel assembly is increased.
20.31 Nusselt Numbers and Heat Transfer Coefficients for PWR Fuel Assemblies
Next, suppose we would like to calculate the Nusselt number for an individual subchannel in a PWR fuel assembly.
In Chapter 2, chapters, we learned that large PWR cores can have 200 fuel assemblies have an average mass flow rate
of about 20,000 kg/s. (Recall that this number is design dependent, but it is pretty representative of the actual mass flow
rates) If the core has 200 fuel assemblies, the average actual mass flow rate through a single assembly must be
20.32 Nusselt Numbers and Heat Transfer Coefficients for LMFBR Fuel Assemblies 805
Now if the fuel assembly has 17 × 17 = 289 fuel rods, control rods, and instrumentation channels, a single subchannel
of approximately
will have a mass flow rate m
where the hydraulic diameter is about 11.8 mm = 0.0118 m. For a P/D ratio of 1.33, the area of a typical subchannel is
approximately 8.8 × 10 −5 m2. This implies that the average subchannel mass flux G is G = m/A
Finally, the average density of the coolant at a temperature of 307.5°C and a pressure of 15.5 MPa is 710.5 kg/m 3.
The average viscosity of water under these conditions is 8.55 × 10 −5 Pa s. This leads to an average coolant velocity
v of v = G/ρ
The Prandtl number of liquid water under these conditions is 0.94, and the Reynolds number (see Chapter 16) is
about 500,000. Using the Dittus–Boelter correlation (see Section 20.24), the Nusselt number for an individual subchannel
is then
Of course, this assumes that coolant pumps are delivering the coolant at 100% of their rated flow rate. Finally,
we learned from Chapter 18 that the hydraulic diameter DH of a typical fuel assembly subchannel is about 1.18 ×
10 −2 m. If the thermal conductivity of liquid water under these conditions is 0.53 W/m °C, the average heat transfer
coefficient is
The dependence of the Nusselt number on the Reynolds number for a typical subchannel is shown in Figure 20.15.
20.32 Nusselt Numbers and Heat Transfer Coefficients for LMFBR Fuel Assemblies
Now let us perform exactly the same calculation for the subchannels in a LMFBR fuel assembly. If liquid sodium flows
through a LMFBR core with P = 9.8 mm, D = 8.5 mm, the equivalent diameter of a subchannel is
D e = (8/πD)
( )
3/4 P 2 − ( π /8)D 2 = 3.96 mm.
806 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
FIGURE 20.15 For a constant axial heat flux, the Nusselt number for water is only about 4 when the flow is laminar, but it can
become almost 1,000 when the flow in a reactor coolant channel becomes highly turbulent. However, this level of turbulence
requires a Reynolds number of about 500,000.
when the P/D ratio is 1.15. The mass flow rate through a large LMFBR core is about 16,000 kg/s, and this mass flow rate
is distributed between about 360 fuel assemblies in the core and an additional 240 breeding assemblies in the radial blan-
ket. Taking the total number of fuel assemblies to be about 600, the flow rate through an individual fuel assembly is about
A typical LMFBR fuel assembly has nearly 270 individual subchannels, and a single subchannel has a mass flow rate
of about
m
For a P/D ratio of 1.15, the area of a typical subchannel is approximately 1.32 × 10 −5 m2. This implies that the average
subchannel mass flux G is G = m/A
The average density of liquid sodium under these conditions is 817.7 kg/m3. This results in an average coolant velocity
v of v = G′/ρ
Finally, since the thermal conductivity of liquid sodium is about 62.6 W/m °C, the average heat transfer coefficient is
In other words, the convective heat transfer coefficient in a LMFBR coolant channel is about three times higher than
it is in a PWR coolant channel when the cores are operating normally. The Nusselt numbers and other relevant design
information is summarized in Table 20.9. The Nusselt number for liquid sodium is also shown as a function of the
Reynolds number in Figure 20.16. The parameters in this table are representative of the parameters that actual reactor
designers use. Moreover, as we will shortly see, these parameters must be adjusted slightly to account for differences in
the turbulent boundary layer between a square fuel assembly and a hexagonal one. These differences cause the Nusselt
numbers and heat transfer coefficients to be considerably different than those for circular pipes.
TABLE 20.9
The Nusselt Numbers for Some Common Reactor Coolants
The numbers in the last column are rounded off for highly turbulent flows.
FIGURE 20.16 In reactor fuel assemblies, the Nusselt numbers for laminar flows (Re < 2,300) do not change very much, and they
are always higher for metallic coolants than they are for water. At higher Reynolds numbers (Re > 2,300), the situation reverses
itself and the Nusselt numbers for liquid water become higher again. However, the convective heat transfer coefficients for liquid
metals are always higher because of their high thermal conductivity. The Nusselt number for liquid sodium at 500°C is shown above.
808 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
a good value for the Nusselt number is needed to find the correct value for h. In this section, we would like to illustrate
how the Nusselt number can be found for a reactor fuel assembly that does NOT have the same shape as a circular pipe,
and how this calculation affects the size of the convective heat transfer coefficient.
For single-phase flow, the size of the Nusselt number depends on the shape of a fuel assembly, but it is also different
for a square assembly than it is for a hexagonal one. In reactor work, the Nusselt number for a reactor fuel assembly can
be found by taking the Nusselt number for a circular pipe and multiplying it by a geometric correction factor C that
depends on the P/D ratio. This approach is surprisingly easy to implement, and it also has the advantage that it gives
more accurate values for the Nusselt number than more sophisticated approaches. Hence, the Nusselt number for a reac-
tor fuel assembly can be written as
Nu* = C ⋅ Nu (for a fuel assembly) (20.88)
where the asterisk indicates that we are referring to a fuel assembly rather than a circular tube. Normally, the geomet-
ric correction factor C is expressed as a function of the P/D ratio, where the pitch P is the center-to-center spacing
between the fuel rods, and D is the outer diameter of the rods. Representative values for the fuel rod pitch P, the fuel
rod diameter D, and the P/D ratio are shown in Table 20.10 for the fuel assemblies in different power reactors. The
actual values may differ from those shown here, but these values are representative of those that one actually finds
in practice. Notice that PWRs, BWRs, and LMFBRs all have P/D ratios between about 1.15 and 1.35, although the
fuel rods in LMFBRs tend to be much smaller and the fuel assemblies tend to be more compact. Also, the fuel rods in
BWRs are somewhat larger than they are in PWRs. The power density in commercial BWRs is also lower. For water
reactors, the values of C are given by
These particular expressions for the geometric correction factor were developed by Joel Weisman in 1959. Hence, they
have become known as the Weisman correction factors. For a PWR fuel assembly with a P/D ratio of 1.33, it is easy to
see that the Weisman correction factor is C = 1.385, and for a LMFBR fuel assembly with a P/D ratio of 1.15, it is easy
to see that C ≈ 1.04. Hence, these numbers show that the convective heat transfer coefficient in a reactor assembly is
considerably higher than it is in a circular pipe with the same hydraulic diameter.
Moreover, the relative difference between the Nusselt numbers gets larger as the P/D ratio gets larger. Hence,
when the flow becomes turbulent and fully developed, the Nusselt numbers become more sensitive to the geometry
of the lattice. This effect applies to both the interior channels and the edge channels. It is also possible to apply sepa-
rate correction factors to the corner channels, the edge channels, and the interior channels if additional accuracy is
required. These additional correction factors are discussed in Todreas and Kazimi (2008) and so we will not discuss
them here. Another correction factor that can be used in this case is the Markoczy correction factor, which was
developed in 1972. Now suppose that we attempt to plot the size of C for both a square lattice and a hexagonal one
TABLE 20.10
Representative P/D Ratios for Reactor Fuel Assemblies
as a function of the P/D ratio. The values of the geometric correction factors are shown in Table 20.11. Notice that
the correction factors get larger as the P/D ratios increase. The convective heat transfer coefficients for liquid water
are shown in Figure 20.17 and for liquid sodium in Figure 20.18. Again, the convective heat transfer coefficients are
very sensitive to the local Reynolds number and the spacing of the fuel rods.
TABLE 20.11
The Nusselt Number Geometric Correction Factors for Square and Hexagonal Reactor Fuel Assemblies with
P/D Ratios between 1.1 and 1.5
P/D Ratio Square Assembly Correction Factor C Hexagonal Assembly Correction Factor C
1.10 0.966 0.982
1.20 1.148 1.095
1.30 1.331 1.208
1.35 1.422 1.265
1.40 — 1.321
1.50 — 1.434
Note that the heat transfer coefficients are generally 10%–40% larger than those of an equivalent circular pipe
having the same Reynolds number, reactor coolant, and hydraulic diameter.
FIGURE 20.17 The convective heat transfer coefficients for liquid water range from about 200 W/m 2 °C for laminar flows to about
40,000 W/m2 °C for highly turbulent flows in PWR fuel assemblies.
810 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
FIGURE 20.18 The variation of the convective heat transfer coefficient in a reactor fuel assembly as a function of the
Reynolds number. In this case, the actual hydraulic diameters are used. At low Reynolds numbers, metallic coolants have
much larger heat transfer coefficients because their thermal conductivity is so high. During normal operating conditions,
the value of h is about three times higher in an LMFBR fuel assembly as it is in a PWR fuel assembly. This is necessary
because LMFBR fuel assemblies have much higher power densities than PWR fuel assemblies do, and liquid metals are
more effective at removing the heat.
FIGURE 20.19 The behavior of the turbulent Nusselt number in a fast reactor coolant channel as a function of the P/D ratio.
The results are shown for a hexagonal lattice with liquid sodium. Most commercial LMFBRs have a P/D ratio close to 1.15.
Therefore, the Nusselt number for fully developed flow in an LMFBR fuel assembly is about 25% higher than that for a
circular pipe. The Reynolds number in this case is taken to be 100,000. (Todreas and Kazimi.)
FIGURE 20.20 This figure shows the variation of the local Nusselt number as a function of the distance from the core inlet.
In general, the Nusselt number (and the convective heat transfer coefficient) is higher close to the entrance to the core, where the
velocity and temperature profiles are still developing. After some distance, the flow field becomes fully developed, and the Nusselt
number approaches an asymptotic value called the asymptotic Nusselt number Nu∞. Beyond this point, the Nusselt number does not
change as long as the axial heat flux remains constant.
812 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
Kazimi (2008) present specific correlations to find the ratio of the average heat transfer coefficient h to the fully devel-
oped one h∞. For a reactor subchannel with square edges, the McAdams correlation is sometimes used:
h h ∞ = Nu Nu ∞ = 1 + ( D e /L ) (20.90)
0.7
In other correlations where the heat transfer coefficient depends on the Reynolds number, the value of h /h∞ increases
as the subchannel Reynolds number increases.
FIGURE 20.21 A rendering of a U-tube steam generator (left), and the exposed tube bank from a U-tube steam generator (right)
used in a nuclear power plant. (Pictures provided by Westinghouse Nuclear.)
20.36 Typical Steam Generator Heat Transfer Rates 813
TABLE 20.12
Heat Transfer Data for a Modern Steam Generator in a Westinghouse PWR
about 75,000, the Nusselt number can be found from the Dittus–Boelter correlation, which we discussed earlier in the
chapter. The result is
Nu TURBULENT = 0.023Re 0.8 ⋅ Pr 0.4 = 0.023 × 7,945 × 0.997 ≈ 182 (from Dittus–Boelter)
and when the flow is laminar and the Reynolds number is 750, the corresponding single tube Nusselt number is
Hence, although the flow rate through a pressure tube is reduced by a factor of about 100 (from about 1.25 kg/s to about
0.0125 kg/s), the round tube Nusselt number only falls by a factor of 42 (182 divided by 4.36). The convective heat
transfer coefficients are then
h = ( k/D e ) ⋅ Nu ( W/m 2
°C ) or
h TURBULENT = (k/D)Nu TURBULENT = (0.53/0.0222) ⋅ 182 ≈ 4350 W/ m 2 °C
when the flow is completely turbulent, and when the flow is completely laminar,
These results are summarized together with other relevant data in Table 20.12.
drop across the length of an individual pressure tube (which can be between 15 and 20 m long) is about 40 °C, and the
average temperature differential is half this value or about 20°C. This means that the average temperature of the coolant
inside one of the pressure tubes is between 300°C and 310°C. Since the temperature of the tubes on the secondary side of
the steam generator is maintained at 270°C, the value of ΔT needed to find the heat transfer rate is ΔT = 305°C–270°C ≈
35°C. According to Newton’s Law of Convection, the heat flow rate through an individual steam generator tube is
where L is the length of the tube, and D is the tube inner diameter.
If a tube is 16.5 m long, the average heat transfer rate across a single tube during full power operation is
q = (4,350)·(1.15)·(35) ≈ 175 kW. Since there are 4,000 tubes in a typical steam generator, the total heat transfer rate
through the tube bank in the steam generator is 175 kW × 4,000 ≈ 700 MWT. Normally, there are four steam generators
in a 4-loop Westinghouse PWR. Thus, the total heat transfer rate from the primary to the secondary side of the plant
from all four steam generators is 4 × 700 MWT ≈ 2,800 MWT. Assuming the nuclear steam supply system in the plant
has an average thermal efficiency η of 35%, the total electrical power output of the plant is about Power (MWE) = Power
(MWT) × η = 2,800 × 0.35 = 980 MW. This compares favorably with the quoted value of 1,000 MWE which the reactor
vendor advertises for this plant. Consequently, although we used only the average temperatures for the coolant on the
primary and secondary sides, we were able to demonstrate how the heat transfer process works for a typical PWR steam
generator. In practice, many additional refinements are normally made to this process to calculate the exact value for the
heat transfer rate. However, it is gratifying to see that we were able to come very close to the actual heat transfer rates
with just a couple of simplifying assumptions. If more accurate estimates were required, the Log Mean Temperature
Difference (or LMTD) could be used to find the heat flux. The LMTD was discussed in Chapter 8.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi, India (2000).
Reactor Theory (Nuclear Parameters), DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
Questions for the Student 815
scientist ________, who first proposed its existence in 1820. The Peclet number has a value between _____ and
_____ in an LMFBR fuel assembly.
26. In practice, is the Peclet number most closely associated with water, metallic coolants, or gaseous coolants?
27. Name three metallic coolants that can be used in fast reactors.
28. Which metallic coolant is used in most fast reactors today?
29. Write an expression for the Nusselt number for a metallic coolant. How many components does the Nusselt number
have for a metallic coolant? What is the physical significance of each of these components?
30. What is a representative value of the Nusselt number in an LMFBR fuel assembly when the flow is laminar?
31. What is a representative value of the Nusselt number in an LMFBR fuel assembly when the flow is turbulent?
32. Write an expression for the Nusselt number for ordinary water. Why is this expression different for water than it is
for a metallic coolant? Why does this expression for the Nusselt number contain only a single term rather than two?
33. What is the mass transfer analog of the Nusselt number?
34. For what types of fluids and what types of flows can the Dittus–Boelter correlation be used?
35. Write down an explicit expression for the Dittus–Boelter correlation. In this expression, what power of the Reynolds
number is used?
36. When a reactor coolant is gaining energy, what exponent is used for the Prandtl number in the Dittus–Boelter
correlation, and when a reactor coolant is losing energy, what exponent is used for the Prandtl number in the Dittus–
Boelter correlation?
37. For what sort of coolants should the Dittus–Boelter correlation NOT be used? What is the physical reason for this
particular restriction?
38. How can the Nusselt number for a circular pipe be converted into the Nusselt number for a reactor fuel assembly?
39. Generally speaking, is the Nusselt number for a reactor fuel assembly greater than, less than, or the same as the
Nusselt number for a heated circular pipe?
40. For a square reactor lattice, at what P/D ratio does the Nusselt number become greater than that for a circular pipe
and at what P/D ratio does it become less?
41. Name two correlations that can be used to convert the Nusselt number for a heated circular pipe to the Nusselt
number for a reactor fuel assembly.
42. Which one of these correlations can be applied to a square reactor lattice and which one of these correlations can
be applied to a hexagonal reactor lattice?
43. Generally speaking, is the Nusselt number for fully developed flow higher or lower when the heat flux is constant
than when the rod temperature is constant? What is the reason for this discrepancy?
44. In general, is the Nusselt number in a reactor coolant channel higher or lower at the entrance to the channel than it
is downstream?
45. At what distance downstream does this difference tend to disappear? Is this discrepancy a function of the coolant
type—in other words, is it different for gaseous coolants than that for metallic coolants or liquid water?
to find the convective heat transfer rate between a fuel rod and the coolant. A similar problem arises when water is
flowing through a heated circular pipe. For a pipe of this type, calculate the bulk temperature T BULK needed to find the
value of q″. Assume the coolant flowing through the pipe is water with an average velocity of 1 m/s, and also assume the
pipe is 0.2 m in diameter. Assume the velocity profile is fully developed and the flow is laminar. When finding the value
of TBULK, also assume the temperature profile of the coolant is given by T(r) = 0.5TMAX[1 + ((r − R)/R))2] where TMAX =
200°C, R is the radius of the pipe, and r is the distance from the center of the pipe (in meters). Is the method we have just
outlined also applicable to nuclear fuel assemblies?
Exercise 20.2
Water, sodium, carbon dioxide, and helium are popular reactor coolants. Compare the Nusselt numbers for these cool-
ants under the following conditions: air at 1 MPa and 1,000°C; carbon dioxide and helium under the same conditions;
Exercises for the Student 817
sodium at 600°C and 150 kPa; water at 300°C and 15.5 MPa; water at 250°C and 6 MPa. Assume a Reynolds number
of 100,000.
Exercise 20.3
Using the same reactor coolants discussed in Exercise 20.2, calculate their convective heat transfer coefficients (in
W/m2 °C) assuming a coolant channel with an equivalent diameter of 0.005 m. Repeat the same exercise when the
equivalent diameter is changed to 0.01 m.
Exercise 20.4
Light water is the primary reactor coolant in the world today. In American PWRs, the average temperature of the water
flowing through the core is about 300°C. Determine the percent change in the convective heat transfer coefficient when
the bulk temperature is 200°C and 400°C, but the coolant is still a single-phase liquid. Assume the water is slightly
subcooled in each case.
Exercise 20.5
Determine the Nusselt number and the convective heat transfer coefficient for a water reactor where the coolant veloc-
ity is 6 m/s, the pressure is 2,250 PSI (15.5 MPa), the coolant temperature is ~300°C, the P/D is 1.33, and the equivalent
diameter of a coolant channel is 0.012 m.
Exercise 20.6
Determine the Nusselt number and the convective heat transfer coefficient for a sodium-cooled fast reactor where the
coolant velocity is 6 m/s, the pressure is 150 kPa, the coolant temperature is ~500°C, the P/D ratio is 1.15, and the equiva-
lent diameter of a coolant channel is 0.004 m.
Exercise 20.7
A liquid metal coolant has a Prandtl number of 0.01, a Reynolds number of 100,000, and a thermal conductivity of
50 W/m °C. It flows through a long circular pipe with a diameter of 1 cm. The wall temperature of the pipe is maintained
at 600°C, and the temperature of the coolant halfway between the wall surface and the centerline of the pipe is 580°C.
Find the temperature of the coolant at the center of the pipe, the bulk temperature of the coolant, and the heat flux at the
surface of the wall in W/m2.
Exercise 20.8
Find the Peclet number for liquid sodium in a reactor coolant channel when the velocity is 8 m/s, the hydraulic diameter
of 0.5 cm, and the average temperature of the sodium is 550°C. How is the Peclet number defined? Why does it not
appear in more correlations for water reactors?
Exercise 20.9
Derive an expression for heat transfer coefficient through a tube in a reactor heat exchanger where the convective heat
transfer coefficients are known on both sides of the tube and the thermal resistance of the tube wall is also known. What
is the value for the thermal resistance of the tube if the tube is circular in shape? Hint: refer to the previous discussion
of this subject in Chapter 8.
Exercise 20.10
Calculate the Nusselt number in a PWR steam generator tube (1) under normal operating conditions when the flow is
turbulent, and (2) after the core has been shut down, and the flow is laminar. Then, repeat the same exercise for a steam
generator tube in a sodium-cooled fast reactor. How do the values of the Nusselt numbers compare?
Exercise 20.11
A research reactor in India is evaluating a new core constructed of parallel vertical plates of 2.25 m high and 1.5 m wide.
The plates contain uranium dioxide fuel pellets, and the maximum temperature of the plate surfaces is limited to 975°C.
818 Fundamentals of Single-Phase Heat Transfer in Nuclear Power Plants
The lowest allowable temperature of the bismuth is 325°C. Using the Dittus–Boelter correlation, calculate the maximum
possible heat flux on each side of the plates when the Nusselt number is 10.
Exercise 20.12
The mass flow rate through the core of a commercial PWR is 10,000 kg/s. The convective heat transfer coefficient in
a typical fuel assembly under these conditions is 30,000 W/m2 °C. By how much does the mass flow rate have to be
increased to increase the convective heat transfer coefficient to 45,000 W/m2 °C? Assume the properties used to calculate
the heat transfer coefficient do not depend on the coolant temperature.
21
Correlations for Single-Phase
Nuclear Heat Transfer
21.1 Correlations for the Convective Heat Transfer Coefficient
Many correlations have been developed over the years to predict the convective heat transfer coefficients in reactor fuel assemblies—
both fast and thermal. The majority of these correlations were developed for circular pipe flow, but were later modified for reactor
work using the equivalent diameter De and the pitch-to-diameter (P/D) ratio of the fuel assemblies. Most correlations were intended
to be used for water and industrial gases, but in the 1960s and 1970s, additional correlations were developed for metallic coolants.
For nonmetallic coolants, the Nusselt number was correlated to the Reynolds number and the Prandtl number. For metallic coolants,
it was correlated to the Peclet number. In this chapter, we would like to discuss these correlations and show how they can be applied
to important problems in the field of nuclear heat transfer.
21.2 Values for Nusselt Number and the Convective Heat Transfer Coefficient
Once the shape of the nuclear heat flux is known, the Nusselt number becomes a function of the geometry of a coolant channel and
whether the flow is natural or forced. Nusselt numbers are smaller for natural convection than they are for forced c onvection. They
also depend on how the boundary layers develop (see Chapter 20). Nusselt numbers for water and gaseous coolants range from 50
to 2,000, whereas for metallic coolants, they range from about 5 to 20. Geometric correction factors (see Section 21.5) are then
used to convert between heat transfer correlations for circular pipe flow and those for noncircular ones. Normally, these correction
factors depend on the P/D ratio of the lattice. Representative values of the Nusselt numbers for reactor fuel assemblies are shown in
Table 21.1. Notice that water has the highest Nusselt number and liquid sodium has the lowest. However, metallic coolants generally
have higher heat transfer coefficients than other coolants because their thermal conductivity is so high. The convective heat transfer
coefficients are obtained by multiplying the Nusselt number by (k/De)
( )
h = k D e Nu (21.1)
where k is the thermal conductivity of the coolant. Representative values of the resulting heat transfer coefficients are shown in Table 21.2.
Notice that the heat transfer coefficients for forced convection are about ten times larger than they are for natural c onvection. This
indicates that there is a considerable economic incentive to force the coolant through the core as quickly as possible. Typical coolant
velocities are 5–6 m/s in the cores of water reactors and 8–12 m/s in the cores of liquid metal fast breeder reactors (LMFBRs). These
velocities are also a function of the power density of the core. When water boils, the heat transfer coefficient becomes larger as well.
TABLE 21.1
Nusselt Numbers for Some Common Coolants in a Reactor Coolant Channel
819
820 Correlations for Single-Phase Nuclear Heat Transfer
TABLE 21.2
Heat Transfer Coefficients for Some Common Coolants in a Reactor Coolant Channel
Here, the physical properties are evaluated at the bulk temperature of the fluid. However, the Dittus–Boelter correlation
does NOT take into account the effects of the wall temperature gradient on the thermophysical properties of the c oolant.
Therefore, it was originally intended to be used for isothermal or near isothermal flows. For some coolants such as
water, this equation can be simplified even further. For example, if we expand the heat transfer coefficient in terms of
its thermo physical properties, we get
Then, the quantity containing only the physical properties k(ρ/μ)0.8(μc/k)0.4 can be combined into one
t emperature-dependent term. For the case of water, the result (in British Thermal Units) is
where T is the temperature of the water in degrees Fahrenheit. As long as the temperature drop across the film does
not exceed 5°C (about 10°F), the value of T can be interpreted as the bulk fluid temperature. Otherwise, the actual
film temperature must be used instead. In most cases, the film temperature TFILM is the arithmetic mean of the wall
temperature and the bulk temperature
This approximation is required because the viscosity is temperature dependent. Now let us see what the Dittus–Boelter
c orrelation implies when we are dealing with a reactor coolant such as carbon dioxide (CO2). The Prandtl number for
most industrial gases has a value between 0.65 and 0.8. So when we substitute this value into the Dittus–Boelter equa-
tion, we raise the Prandtl number to the power of 0.4, and if we assume a value of Pr of about 0.70, the value of (Pr)0.4
becomes about 0.85. Then, the Dittus–Boelter correlation becomes
21.5 Single-Phase Heat Transfer Correlations for Reactor Fuel Assemblies 821
This equation can be used for carbon dioxide, hydrogen, and helium up to temperatures of about 1,200°C. In other
words, it works very well for gas reactors when the core inlet and outlet temperatures are high.
The Seider–Tate correlation is a very reliable correlation for water reactors. Remarkably, this simple “fix” to the fluid vis-
cosity allows the heat transfer coefficient to be calculated for non-isothermal flows or for flows where the wall temperature is
much higher than that of the surrounding coolant. In other words, it can be used in cases where the wall temperature gradient
is very steep. Here, the physical properties of the coolant must be calculated at the bulk temperature, except for the viscos-
ity close to the wall, μWALL, which must be evaluated at the surface of the wall. The dynamic viscosity of water at 280°C is
0.0002 N s/m2, at 300°C, it is 0.000184 N s/m2, and at 320°C, it is 0.00017 N s/m2. Hence, if the temperature of the coolant
at the wall surface is 320°C and the temperature of the coolant away from the surface is 300°C, the viscosity correction
term “(μWALL/μ)0.14” has a value of (0.00017/0.000184)0.14 = 0.989. Hence, the Seider–Tate correlation predicts a lower Nusselt
number than the Dittus–Boelter correlation in a typical pressurized water reactor (PWR) core. In both the Dittus–Boelter
and Seider–Tate correlations, the equivalent diameter De must be used when a reactor c oolant channel is noncircular in shape.
The heat transfer coefficients for noncircular coolant channels can be obtained from the correlations for circular ones as long
as they do not deviate appreciably from the geometry of a circular pipe. As we discussed earlier, we can use the equivalent
diameter De = 4A/P for coolant channels that are square, elliptical, rectangular, or even triangular in shape as long as the
aspect ratio is not too high. Here, A is the cross-sectional area of the flow channel, and P is its wetted perimeter. When the
aspect ratio becomes greater than about 1.2, an additional geometric correction factor may have to be used to obtain more
accurate results. We will have more to say about the form of this correction factor shortly.
FIGURE 21.1 Examples of center, side, and corner coolant channels in reactor fuel assemblies.
Under these conditions, a geometrical correction factor C (or GCF) is usually applied to the heat transfer correlations
for circular channels to obtain the heat transfer correlations for noncircular channels. Thus, for fully developed turbulent
flow, the Dittus–Boelter and the Seider–Tate correlations become
( )
0.14
Nu = C ⋅ 0.023(Re)0.8 (Pr)0.4 µWALL µ (Seider–Tate) (21.8b)
where the geometrical correction factor C is a function of the P/D ratio of the reactor lattice. The local heat transfer
coefficients are obtained by simply multiplying these expressions by (k/DE), where DE is the equivalent diameter that we
defined previously. There are a couple of correlations that have been developed over the years to find the value of C. For
reactor fuel assemblies, the most popular correlations are
1. The Weisman correlation
2. The Markoczy correlation
3. The Presser correlation
We would now like to briefly describe each of these correlations and demonstrate how they can be used. However, in
most cases, the geometrical correction factor C turns out to have a value between 0.95 and 1.05 (±5%). This range of
values applies as long as the predominant direction of flow is parallel to the surface of the fuel rods, turbulent, and fully
developed. If the flow is perpendicular to the surface of the rods (which is not a common occurrence in the cores of
water reactors), then an additional geometric correction factor may have to be used. However, even in this case, the total
geometrical correction to the heat transfer coefficient rarely exceeds 10% (0.90 ≤ C ≤ 1.10).
and for a triangular array which appears in a hexagonal lattice, the corresponding correction factor is
where R is the P/D ratio. For water reactors (with square fuel assemblies and typical P/D ratios), the Weisman c orrection
factor is about 1.3. The Weisman correction factor does not work well for liquid metals. For water reactors, it is valid for
P/D ratios R = P/D between 1.1 and 1.3. In some books, the Dittus–Boelter and Seider–Tate correlations are written as
( )
0.14
h = C′(k/D e )(Re)0.8 (Pr)0.4 µWALL µ (Seider–Tate) (21.10b)
For ordinary water flowing parallel to the fuel rods, the values of C′ are then*
and
Equation 21.10(c) applies to square reactor lattices where 1.1 ≤ P/D ≤ 1.3, and Equation 21.10(d) applies to triangular
reactor lattices where 1.1 ≤ P/D ≤ 1.5. For a typical square reactor fuel assembly with a pitch of P = 0.6 in. (1.52 cm) and
a fuel rod radius of 0.234 in. (0.6 cm), the value of C′ is then
In this case, the value of C′ in Equations 21.10a and 21.10b is almost identical to the value of C′ = 0.023 that is used in
the Dittus–Boelter and Seider–Tate correlations to find the convective heat transfer coefficients in long circular pipes!
For water reactors (that use square arrays), the value of C is about 1.1. Again, the Presser correction is not intended to
be used for metallic coolants. It is valid for P/D ratios between 1.1 and 1.9 for a square lattice, and between 1.1 and 2.2
for a hexagonal one.
* See Weisman, J., “Heat Transfer to Water Flowing Parallel to Tube Bundles”, Nuclear Science and Engineering, 6:78, 1979.
824 Correlations for Single-Phase Nuclear Heat Transfer
and 1.8, its average error is about 7%. It does NOT work well for liquid metals, but it works very well for water, organic
coolants, and most oils. The form of the Markoczy correction is
The Markoczy Geometric Correction Factor (for both Square and Hexagonal Lattice)
( )
C = 1 + 0.912Re −0.1Pr −0.4 1 − 2.004e − E (21.13)
and its value is typically greater than 1.1. Here, the exponent E is equal to the average equivalent diameter De of the
c oolant channels surrounding a fuel rod divided by the diameter of the rod:
E = D e /D ROD (21.14)
The Markoczy correction factor is designed to be applied primarily to the interior channels in a reactor fuel assembly,
and it works best when the fuel assembly is very large (i.e., when the size of the array is greater than 10 by 10). It was
initially applied to coolant channels in an infinite reactor lattice. It works well for both square and triangular arrays,
and can handle very low turbulent flow rates (Re ≅ 3,000) up to a Reynolds number of about 100,000. Hence, it is nearly
ideal for large reactor fuel assemblies where the central channels are identical. It is not designed to handle corner or edge
channels. The Weisman, the Markoczy, and the Presser correction factors are designed to work with the Dittus–Boelter
correlation, or a modified form of it (such as the Seider–Tate correlation). As such, they cannot be used with other cor-
relations without extensive modifications.
TABLE 21.3
The Correction Factors to the Heat Transfer Coefficient for a Square and Hexagonal Lattice
FIGURE 21.2 The ratio of the Nusselt number for parallel flow in reactor fuel assemblies to that in circular tubes.
P/D = 1.33, the Weisman correction factor (from Equation 21.9a) is C = 1.39. The convective heat transfer coefficient is
then h = C′(k/De)(Re)0.8(Pr)0.4 = 1.39 × 4.24 × 32,240 × 1.035 = 196,675 W/m2 °C. The flow is clearly turbulent in this
case because the Reynolds number is approximately 500,000. [Ans.]
Here, the value of z (z = 0) begins at the entrance and is measured in equivalent hydraulic diameters. In other
words, z is some multiple of De. This implies that if z = 10De, then h(0) = 1.2h∞. The effects of the changing axial
temperatures are taken into account by evaluating the physical properties of the coolant at specific points along the
length of the channel, that is, ρ(z) = ρ(T(z)) and μ(z) = μ(T(z)) where the heat transfer coefficient is to be found. If
these values cannot be determined from an energy balance, then the average of the inlet and the outlet temperatures
is sometimes used.
Generally speaking, entrance effects are not a major concern in the design of reactor fuel assemblies because the
heat transfer coefficients are always much higher than the asymptotic value of h∞ would imply. So if the entrance
effects are not well understood, one could simply use the asymptotic value for h∞ and the results would be conservative
826 Correlations for Single-Phase Nuclear Heat Transfer
because the axial fuel pin temperature profiles will be overestimated in the region where the flow is developing. In
most fuel assemblies, this effect completely negligible about 20% of the way into the core. However, entrance effects
can still be important if one is trying to prove that the fuel pin temperatures are actually lower than the ones that a
fully developed turbulent flow pattern might imply. The exercise below demonstrates how the value of h behaves from
near the bottom of a fuel assembly (where z ≅ De) to near the middle of the same assembly, where z ≅ 150De. Clearly,
entrance effects produce a beneficial result in this case. Even at a distance of only one equivalent diameter from the
start of the coolant channels (at z = De), the heat transfer coefficient h is approximately twice as large as it is further
into the bundle. Its variation with the length-to-diameter ratio in a circular coolant channel with a uniform heat flux
is shown in Figure 21.3.
Nu = f(Pe) (21.16)
FIGURE 21.3 The variation of the Nusselt number with the length-to-diameter ratio of a circular coolant channel with a uniform
heat flux. (See Kays and Crawford, Convective Heat and Mass Transfer, McGraw Hill (1980).)
21.11 Convective Heat Transfer Coefficients for Liquid Metals 827
TABLE 21.4
The Prandtl Numbers for Various Coolants in a Circular Coolant Channel at Typical Temperatures and Pressures
The Peclet number Pe is the product of the Reynolds number and the Prandtl number, and so we can write it as
The Peclet number is part of a larger set of dimensionless numbers that are used to measure the transport of thermo-
physical properties through a continuum. This number was originally named for French physicist Jean Claude Peclet,
whose picture is shown in Figure 21.4. The Peclet number can be thought of as the ratio of the rate of advection of a
physical quantity to the rate of diffusion of the same quantity driven by an appropriate gradient. Thus, in the context of
the reactor fluid flow, the thermal Peclet number is equivalent to the product of the Reynolds number and the Prandtl
number. In the context of mass transfer, the Peclet number is equivalent to the product of the Reynolds number and the
Schmidt number Sc, where the Schmidt number is defined by
In Equation 21.18 μ is the dynamic viscosity of the coolant (in N s/m2 or kg/ms), ρ is the density of the coolant (in kg/m3),
and D is its mass diffusivity (in m2/s). Thus, for mass transfer, the Peclet number becomes
FIGURE 21.4 A picture of French physicist Jean Claude Peclet, (left) for which the Peclet number is named, and German engineer
Ernst Heinrich Wilhelm Schmidt (right), for which the Schmidt number is named. (Pictures provided by Wikipedia.)
828 Correlations for Single-Phase Nuclear Heat Transfer
where L is a characteristic length over which the coolant flows, v is the local fluid velocity, D is the mass diffusion coefficient
(not to be confused with the equivalent diameter). Then for a related problem in reactor heat transfer, it is defined as
where α is the thermal diffusivity of the coolant, and α = k/ρc. Here, k represents the thermal conductivity, ρ represents
the density, and c represents the specific heat at constant pressure. In engineering applications, the Peclet number can
sometimes become quite large, and values of over 4,000 have been observed in LMFBR fuel assemblies with P/D ratios
of 1.3. At these Peclet numbers, the Nusselt number is about 20. The actual relationship between the Nusselt number and
the Peclet number is shown in Figure 21.5. Finally, identical flows will often have different Peclet numbers for heat and
mass transfer. This can lead to the phenomenon of double diffusive convection, which is discussed in other heat transfer
books. If we further expand Equation 21.17, we soon discover that
Pe = D e cρv/k (21.21)
and the viscosity μ is nowhere to be found. (It cancels itself from the numerator and the denominator.) This means that
the viscosity is NOT needed to determine the convective heat transfer coefficient for a liquid metal. For liquid metals,
the Nusselt number takes the form:
FIGURE 21.5 The Nusselt number as a function of the Peclet number in an actual LMFBR fuel assembly with P/D ratios of
1.15 and 1.30. Notice that the Nusselt number becomes larger as the P/D ratio increases. (Adopted from Kazimi and Carelli, Heat
Transfer Correlation for Analysis of CRBRP Assemblies, Westinghouse Report CRBRP-ARD-0034 (1976).)
21.13 The Lyon–Martinelli Correlation 829
and the corresponding expression for the heat transfer coefficient becomes
In this equation, A, B, and C are constants that depend on the type of liquid metal and the geometry of the flow chan-
nel. For liquid sodium, A has a value between 5 and 7, and C almost always has a value of 0.8. The fact that A is finite
means that the Nusselt number (and hence the heat transfer coefficient h) remains finite even when the fluid is essentially
stagnant (with v close to 0). Clearly, this is because metallic coolants have much higher thermal conductivities than
water and gaseous coolants do. Moreover, although LMFBRs have been in operation for many years, there are not a
lot of publicly available correlations to calculate the convective heat transfer coefficient. The first correlations for fully
developed convective flow through circular tubes and over plat plates were published in the early 1950s, and the first
correlations for liquid metals in rod bundles did not appear until the early 1970s. However, because there are so few of
them, we would like to discuss how they came to be developed in more detail. These correlations are the primary cor-
relations that are used today to design LMFBR cores and liquid metal heat exchangers.
The Lyon–Martinelli Correlation (for Liquid Metals with a Constant Heat Flux)
Nu = Nu CONDUCTION + Nu CONVECTION
(21.24)
= 7.0 + 0.025(Pe)0.8
For the case of a constant wall temperature, the Seban correlation (sometimes called the Seban–Shimazaki correlation)
may be used:
The Seban Correlation (for Liquid Metals with a Constant Wall Temperature)
Nu = Nu CONDUCTION + Nu CONVECTION
(21.25)
= 5.0 + 0.025(Pe)0.8
Both correlations were developed in the early 1950s, and they have an error range of about 10%. The Seban correlation
implies that the heat transfer coefficient approaches a constant value of
h = 5k/D e (21.26)
when the velocity of the metal approaches 0. Then, the value for h is simply the conductive component of the heat
t ransfer coefficient.
FIGURE 21.6 Normalized velocity profiles for fully developed flow through a circular pipe for water and liquid metal.
through a circular tube with a constant axial heat flux q″. When the flow is highly turbulent, the velocity profile is essen-
tially “flat” in the radial direction, and the value of v is about the same everywhere (see Figure 21.6). In other words, the
coolant behaves like a solid slug of metal as it moves through the channel. Since the metal slug has a very high thermal
conductivity, it is reasonable to assume that most of the heat transfer takes place by conduction alone. At some distance
from the entrance to the channel, the flow becomes fully developed and the temperature profile approaches its equilib-
rium shape. The temperature still rises as the coolant moves down the channel, but at every radius, it rises at the same
rate, and the edge-to-centerline temperature difference stays the same. For a circular channel, the heat flowing into the
liquid metal at a particular radial location r is
( ) ( πR ) = q′(2πR)(r/R) (21.27)
q(r) = q ′(2 πR) πr 2 2 2
where q′ is the heat input per unit length at the wall surface, and R is the channel radius. Notice that when r = R,
q(R) = q′(2πR) and when r = 0, q(0) = 0. This heat must travel through every point in the metal, so we know from
Fourier’s law (see Chapter 10) that
where the minus sign no longer appears because the heat is flowing in the negative r direction. Equation 21.28 may also
be written as
or
dT = (q ′/kR)rdr (21.30)
Suppose we then integrate this equation between r and R, where T(r) is the temperature at r and TWALL is the temperature
at the wall surface. We then obtain
( )
T(r) = TWALL − (q ′/2kR) R 2 − r 2 (21.31)
The bulk fluid temperature in the coolant channel is just the volumetric average temperature. Since the velocity is also
the same everywhere, we have
21.14 Liquid Metal Behavior in Reactor Coolant Channels 831
R
(
TBULK = 1/πR 2 )∫ 0
2πrT(r) dr = TWALL − q ′R/4k (21.32)
or
Thus, the Lyon–Martinelli correlation implies that the heat transfer coefficient approaches a value of
h = 8(k/D e ) (21.35)
as the velocity of the flow approaches 0. Then, the value for h is simply the conductive component of the heat transfer
coefficient, and the Nusselt number is
Thus, we have just shown that when the velocity profile is uniform, and when heat transfer takes place by conduction alone,
the Nusselt number for a liquid metal must be exactly 8! Clearly, this is an interesting result because it shows that when
the effects of convection are ignored entirely, the value of A must be 8 instead of 7. The Lyon–Martinelli correlation takes
a more holistic approach to this problem because it assumes that the heat transfer process consists of two distinct parts—
conduction (where NuCONDUCTION = 7) and convection (where NuCONVECTION = 0.025Pe0.8). The reason why the conduction
component is 7 rather than 8 is that the real turbulent velocity profile (which involves the 1/7th power law) is used instead
of a constant velocity profile, and for the Lyon–Martinelli correlation, this profile was obtained from a fit to experimental
data. Finally, for metallic coolants, the convective component does not approach the conductive component until Pe ≈
1,200. For liquid sodium, this does not occur until the Reynolds number reaches 300,000! In other words, up to a Reynolds
number of about 25,000 where NuCONVECTION = 0.025Pe0.8 ≈ 1.0, it is completely legitimate to use a value of Nu = 8 for a
heated circular pipe! In the Seban correlation, the heat transfer coefficient then approaches a constant value of
h = 5(k/D e ) (21.37)
as the velocity of the flow approaches 0 because the boundary condition on the wall surface assumes a much greater
importance for liquid metals than for nonmetallic coolants such as water. For a constant wall temperature, the value of
7 is then replaced by a value of 5. Thus, for any liquid metal, the conductive heat transfer coefficient is (7 − 5)/5 = 40%
higher when there is a uniform heat flux than when there is a uniform wall temperature. In fast reactors, the heat flux is
generally more uniform than the wall temperature in the axial direction.
where the Reynolds number was based on the maximum speed through the rod bundle, and the physical properties used
to find the Prandtl number were based on the average film temperature. The other properties were based on the average
bulk temperature. This correlation was eventually used in the design of cross-flow heat exchangers in LMFBRs. Then,
in the 1970s, several new correlations began to appear. The Kazimi and Carelli correlation was the first of these to
become widely used. It stated that
832 Correlations for Single-Phase Nuclear Heat Transfer
FIGURE 21.7 The Nusselt number for liquid metal flowing parallel to rod bundles with a uniform heat flux. Here, the Nusselt
number is plotted as a function of the Peclet number for P/D ratios of 1.2 and 1.5.
and it could be used when the Peclet number was less than 5,000, and the P/D ratio (R = P/D) was between 1.1 and 1.4.
However, it tended to underestimate the Nusselt number for high P/D ratios. The Kazimi and Carelli correlation was
used by Westinghouse to design the Clinch River Breeder Reactor near Oak Ridge, Tennessee, in the 1970s. Finally, two
additional correlations called the Schad correlation and the Borishanskii correlation later appeared in the literature.
The Schad correlation is good up to Peclet numbers of about 1,000, and the Borishanskii correlation is good up to Peclet
numbers of about 2,000. The specific forms of these correlations are
and
Notice that both correlations are more complicated than the earlier correlations they replaced. Both the Schad correla-
tion and the Borishanskii correlation are valid for P/D ratios (R = P/D) between 1.1 and 1.5. The Borishanskii correlation
is the most accurate of the three over the entire range of P/D ratios. However, all three of these correlations have been
used to design LMFBRs, and collectively, they correlate most of the experimental data that is used to do liquid metal
convective heat transfer today.
21.15 Comparing the Heat Transfer Coefficients for Water and Liquid Metals
Based upon what we have said so far, it is now possible to compare the single-phase heat transfer coefficients in PWR
and LMFBR fuel assemblies for some representative flow rates. Suppose that we are asked to do so for normal operating
conditions (e.g., those having representative temperatures and pressures) and for three flow rates: one for a “low” flow
rate of 1 m/s, the second for a “normal” flow rate of 5 m/s, and the third for a “high” flow rate of 10 m/s. First, for a PWR,
we would most likely use the Dittus–Boelter correlation or the Seider–Tate correlation with the Presser correction factor
C. For a LMFBR, we would use the Kazimi–Carelli correlation if the P/D ratio was not too high. So for a PWR, we will
assume a P/D ratio of 1.33, and for a LMFBR, we will assume a P/D ratio of 1.15. Under these conditions, the equivalent
diameter De is about 11.75 mm for a PWR and about 3.95 mm for a LMFBR. For a LMFBR at 1 m/s, the Peclet number
is about 64.8, and at 10 m/s, it is about 648. If we use the thermal property data shown in Table 21.5, the corresponding
values of the heat transfer coefficients are shown in Table 21.6.
Notice that the heat transfer coefficients in a LMFBR fuel assembly are always higher than those in a PWR fuel
assembly for the same coolant velocity. This is because the thermal conductivity of liquid sodium is so high. The heat
transfer coefficient in a LMFBR increases by about 30% when the flow rate increases by a factor of 10 (from 1 to 10 m/s),
whereas the heat transfer coefficient in a PWR increases by a factor of more than 6 under the same conditions. This
implies that the heat transfer rate in a LMFBR is driven more by conduction than it is by convection, whereas exactly
the opposite is true in a PWR. This is a classic example of how water and liquid metals behave in completely different
ways. The heat transfer coefficients for boiling water reactors are discussed in Chapter 24.
TABLE 21.5
The Physical Property Data Normally Used to Calculate the Convective Heat Transfer Coefficients in PWR and
LMFBR Cores
Physical Property Data Used to Calculate the Convective Heat Transfer Coefficient in PWR and LMFBR Cores
Physical Property Pressurized Water at 315°C Liquid Sodium at 540°C
ρ (kg/m3) 704 817.7
μ (kg/ms) 0.000087 0.000228
k (W/m °C) 0.5 62.6
cP (J/kg °C) 6,270 1,254
834 Correlations for Single-Phase Nuclear Heat Transfer
TABLE 21.6
The Convective Heat Transfer Coefficients in PWR and LMFBR Fuel Assemblies at Different Flow Rates Based on the
Dittus–Boelter Correlation, the Dittus–Boelter Correlation with the Presser Correction Factor, and the Kazimi and Carelli
Correlation
21.16 H
ow the Radial and Axial Power Profiles Can Affect the Value
of the Heat Transfer Coefficient for Turbulent Flow
Up to now, we have assumed that the radial and axial temperature profiles are uniform throughout the core. However,
this is almost never true in practice because the radial and axial power profiles can be completely different. As a
matter of fact, peak-to-average power ratios in the range of 2.5:1 are sometimes found in commercial reactor cores,
and these power ratios can also change due to fuel rod burn up and control rod motion. These power ratios are dis-
cussed in our companion book,* and they are summarized in Chapter 5 of this book. Now suppose that we would
like to calculate the heat transfer coefficients at three different axial elevations in the core—at the bottom of the core
(where z ≅ 0), at the middle of the core (where z ≅ H/2), and at the top of the core (where z ≅ H). Here, H is the core
height, which is usually about 4 m for a commercial power reactor. In all three cases, we shall assume that the mass
flow rate is exactly the same and that the coolant does not boil. Differences in fluid properties will then drive changes
in the heat transfer coefficients that are observed. Most of these changes will be caused by temperature and pressure
differences, although entrance effects (see Chapter 20) also tend to enhance the heat transfer coefficients near the
inlet to the core.
In addition, in order to capture this important behavior, we would like to calculate the heat transfer coefficients again
halfway between the middle for the core and the bottom, and halfway between the middle of the core and the top. This
will give us five different elevations where we can find the values for h. The values of the coolant properties can then
be found between these five points by the process of linear interpolation, which is illustrated in Figure 21.8. There are
actually two different scenarios that we would like to present to demonstrate this behavior. The first scenario assumes
that the heat flux in the axial direction is a constant. This can almost never happen in a real reactor, but it can provide
the “lower end of the envelope” when attempting to understand what the behavior of h might be. The second scenario
assumes that the heat flux in the axial direction is co-sinusoidal in shape. This second scenario is more common in
practice because there is less neutron leakage out of the center of the core than there is at the edges. We will assume that
this second scenario will provide the “upper end of the envelope” when attempting to understand how the convective
heat transfer coefficient behaves. Finally, we would like to assume that the flow becomes fully developed after it has
progressed 40–60 equivalent diameters into a fuel assembly. Again, this is a reasonable assumption to make if the flow
is highly turbulent and Re ≈ 500,000.
Both of these axial heat flux profiles (q″(z) = CONSTANT and q″(z) = CONSTANT cos (πz/H)) will give rise to
different axial temperature profiles for the coolant. Here, we will assume that H is the core height, so the bottom of
the core is located at z = −H/2, and the top of the core is located at z = H/2. We will only show what the temperature
profiles look like at this point in the book, but we will derive explicit expressions for them in Chapter 26. We can then
extrapolate what the heat transfer coefficients will look like from the material properties at each of these five equally
spaced axial elevations. Again, we will defer a discussion of how two-phase flow affects this picture until Chapter 24.
In the first case (where the axial heat flux is constant), the axial coolant temperature will increase in a linear manner
between the bottom and the top of the core. In the second case, the temperature rise of the fluid will be anything but
linear, and its general shape will resemble that shown in Figure 21.9. Notice in this case that it rises more slowly ini-
tially but then reaches a “broad plateau” near the top of the core. For a PWR, we will assume that the coolant enters
* See An Introduction to Nuclear Reactor Physics, by R.E. Masterson, CRC Press (2017).
2.16 RADIAL AND AXIAL POWER PROFILES 835
FIGURE 21.8 The actual, discrete, and interpolated temperature profiles in a hypothetical reactor core based on the use of a linear
interpolation scheme.
the core at 300°C and leaves the core at 330°C. Then, the bulk fluid temperatures at each of these five axial elevations
are shown in Table 21.7.
Now let us take a look at how the values of the heat transfer coefficients fit into this picture. If we use the
properties shown in the steam tables, the values of the heat transfer coefficients and relevant t hermophysical
properties (neglecting entrance effects) are shown in Table 21.8. Notice that for the same Reynolds number, the
heat transfer coefficients vary by about 4% between the bottom of the core and the top. The heat transfer coef-
ficients are generally highest near the bottom of the core and lowest near the top. Their values behave nearly
linearly between the interpolation points. We will leave it to the reader to show how a similar set of coefficients
behave in a LMFBR core (see Exercise 21.4). We will defer a discussion of what happens when the coolant boils
until Chapter 24.
836 Correlations for Single-Phase Nuclear Heat Transfer
FIGURE 21.9 Typical axial temperature profiles in a PWR core with a co-sinusoidal power shape.
TABLE 21.7
The Fluid Temperatures in a 4 m-high PWR Core with a Uniform and Co-Sinusoidal Axial Heat Flux
TABLE 21.8
The Axial Heat Transfer Coefficients in a PWR Core Based on a Coolant Channel Hydraulic Diameter of 0.0118 m
The physical properties of the coolant then become a function of axial elevation. These physical properties affect all of
the dimensionless numbers that are used in the correlations for the heat transfer coefficients. If the flow is turbulent, the
heat transfer coefficient (except for liquid metals) can be determined from the Dittus–Boelter correlation, repeated here
for the sake of clarity:
( )
0.8
h = k/De ⋅ 0.023 ρcD e µ (µc/k)0.4 (21.43)
The Prandtl number for liquid water varies by about 20% (from 0.90 to 1.10) between the top and the bottom of a water
reactor core. Also, for gases, the value of Pr does not vary by more than a couple of percent with temperature.
So at least in the core, the value of (Pr)0.4 does not have a major effect on the value of h as long as the coolant does
not boil. Also (unless we are near a grid spacer or a flow blockage) the equivalent diameter De is a constant, and the
mass flux G = ρv will be essentially the same for a given cross-sectional area of the coolant channel. Thus, the value of
h varies directly with k0.6 and inversely with μ0.8. For simple analyses, the recommended procedure is to use the physical
properties of the coolant evaluated at a temperature midway between the entrance and the exit of the channel:
However, today most computer programs (such as TRACE, RELAP, and COBRA) actually update these properties at
each axial elevation, and then, they are assumed to be constant within whatever node is located at that elevation. The
second major observation that we would like to make is that the temperature of the coolant lags the change in the heat
flux slightly. In Chapter 26, we will show that if the axial power shape is co-sinusoidal, then the power peaks at the
center of the core (z = H/2), but the fuel temperature peaks about three-fourths of the way between the top and the bot-
tom of the core and the coolant temperature peaks at the core exit (z = H). Entrance and exit effects do not affect this
location very much, and the “hot spot” does not move as long as the flow continues to be highly turbulent. However, the
location of the hot channel depends on where the radial heat flux is highest. This is generally at the location where the
enrichment is highest as well.
where the values for n and C are shown as a function of the Reynolds number in Table 21.9. For single tubes or tubes that
are arranged in such a way that the flow over them is not affected by other tubes, the Reynolds number is determined
from the outer diameter of the tube D and the approach velocity V∞:
Re = DV∞ ν (21.46)
When using these equations, the fluid properties must be evaluated at the film temperature on the surface of the tube.
Several correlations can be used to find the average Nusselt number under these conditions. Probably, the most compre-
hensive of these is the Churchill–Bernstein correlation, which has the following form:
838 Correlations for Single-Phase Nuclear Heat Transfer
TABLE 21.9
Constants for Calculating the Nusselt Number for Flow across a Circular Tube
Constants for Calculating the Nusselt Number for Flow across a Single Tube
Nu = (hD/k) = 0.3 + 0.62Re 0.5 Pr 0.33 1 + (0.4/Pr)0/67 ⋅ 1+(Re/282,000)0.625 0.8 (21.47)
0.25
The Churchill–Bernstein correlation works well for Reynolds numbers and Prandtl numbers greater than 0.2. In this
correlation, the properties of the coolant must be evaluated at the film temperature T FILM = ½(T∞ + TS) which is just
the average of the free stream temperature and surface temperature. Hence, the Churchill–Bernstein correlation is
appropriate to use for water and gaseous coolants but not metallic ones. Example Problem 21.4 illustrates an interesting
application of this correlation.
Normally, the local Nusselt number changes as one moves around the periphery of the pipe if it is subjected to cross-
flow. It is relatively high at the stagnation point (θ = 0°) when the flow first encounters the pipe, but then, it decreases
for a while until one passes the edge and θ ≅ 120°. Then, the boundary layer starts to separate, and the Nusselt number
increases again. The behavior of the heat transfer coefficient is illustrated in Figure 21.10. In general, the boundary
layer remains laminar for Reynolds numbers below about 200,000 and becomes turbulent for Reynolds numbers
greater than this. This behavior explains why so many different values are required in Table 21.9 to cover the entire
range of Reynolds numbers. Each set of coefficients represents a transition in the convective heat transfer coefficient as
the physical characteristics of the boundary layer change. The parameters in Table 21.9 then define the average value of
the Nusselt number around the pipe.
FIGURE 21.10 The variation of the local heat transfer coefficient along the circumference of a circular tube for cross-flow for a
coolant like air. (For more information, see Cengel 2003.)
FIGURE 21.11 An example of an in-line tube bank on the left and a staggered tube bank on the right.
this is called an in-line array. When the tubes are offset from each other vertically or horizontally, the array is called
a staggered array. The heat transfer rate Q depends on the heat transfer coefficient, the total surface area of the tubes,
the number of tubes, and the inlet and outlet temperatures. Just like parallel flow heat exchangers, the heat transfer rate
is given by
Q = h(NπDL)∆TLM (21.48)
where h is the “average” heat transfer coefficient of a tube, N is the total number of tubes in the bank, and ΔTLM is the
log-mean temperature difference (LMTD) between the start and the end of the tube bank given by
∆TLM = ( TWALL − TIN ) − ( TWALL − TOUT ) ln ( TWALL − TIN ) − ( TWALL − TOUT ) (21.49)
840 Correlations for Single-Phase Nuclear Heat Transfer
( )
TOUT = TWALL − ( TWALL − TIN ) exp −πDNh ( ρc p V∞ N T PT ) (21.50)
where NT is the number of tubes in the transverse direction, PT is the transverse pitch of the tubes, and V∞ is the aver-
age approach velocity of the coolant. To find the temperature of the coolant when it exits the array, we need to find the
convective heat transfer coefficient for an “average” tube. The value of h depends on the diameter of the tube, the veloc-
ity of the coolant at the tube’s surface, the geometric arrangement of the tubes, and the physical characteristics of the
coolant at various locations in the tube bank. In 1987, Zukauskas (see Cengel 2003) proposed a generic correlation for
the Nusselt number for an average tube:
where
( )
0.25
P* = Pr Prs (21.52)
is called the Prandtl correction factor. When using the above equations, the Reynolds number must be determined from
the maximum velocity VMAX of the fluid in the tube bank rather than the approach velocity V∞:
Re = DVMAX ν (21.53)
For an in-line array, the maximum velocity can be deduced from the approach velocity using
where P is the transverse pitch. If P happens to be two-dimensional, the maximum velocity is twice the approach velocity.
To use the Prandtl correction factor P*, Pr is evaluated at the arithmetic mean of the inlet and outlet t emperatures, and
Prs is Prandtl number evaluated at the surface temperature of a tube Ts. The values for m and n depend on both the
Reynolds number and the geometric arrangement of the tubes. The correlations for the Nusselt number for an “average”
tube are shown in Table 21.10. These particular correlations are good for flow across tube banks having at least 16 rows
and Prandtl numbers between 0.7 and 500. Usually, banks with several rows of tubes have lower average heat transfer
coefficients per tube than isolated tubes. To find the effect this has on the Nusselt number, the single tube Nusselt number
is usually multiplied by a geometric correction factor F whose values are shown in Table 21.11. The Nusselt number for
a tube in a bank consisting of NR individual rows is then
TABLE 21.10
Nusselt Number Correlations for Cross-Flow over Tube Banks with N > 16 and 0.7 < Pr < 500
TABLE 21.11
The Value of the Geometric Correction Factor F as a Function of the Configuration and the Number of Rows in the Tube Bank
Nu = F Nu (21.55)
For small numbers of rows, the “single tube” Nusselt number is reduced by 10%–30%. This is because the bank tends
to inhibit the flow of coolant around the tubes. Then, as the number of rows becomes larger, the flow becomes more
uniform and the average Nusselt number asymptotically approaches the value for a single tube. When there are more
than 10 rows in the bank, the numbers become virtually identical again. However, staggered arrays of tubes approach
this value more slowly than in-line ones. Example 21.5 illustrates how the average Nusselt number is determined under
these conditions. The heat transfer rate from the tube bank is then found by multiplying the heat transfer rate for a single
tube by the number of tubes. If steam happens to condense on these tubes, the condensation rate is given by
CONDENSATION = q ′ h lv (21.56)
m
where hlv is the latent heat of vaporization (about 2,256 kJ/kg at 100°C) and q ′ = q″/L = πDh(TSAT − T WALL). In a
condenser where h is about 1,500 W/m2 °C, and the water flowing through the tubes has a temperature of 30°C, the
condensation rate is about 0.005 kg/ms per tube. This is an example of what is called film condensation. Another form
of condensation called dropwise condensation is discussed in Chapter 34.
k = 0.64 W/m °C
Pr = 3.9
The Reynolds number is Re = DVMAX /ν = 0.25 × 4/0.0000006 = 1,666,666. Hence, the flow is clearly turbulent. The
Prandtl correction factor is P* = (Pr/Prs)0.25 = (3.9/1.95)0.25 = 1.19. From Table 21.10, the Nusselt number is
The heat transfer coefficient for an “average tube” is h = (k/D)Nu = (0.64/0.25) × 6,340 = 16,230 W/m2 °C. The sur-
face area of a tube is A = πDL = π/2 = 1.57 m2. The geometrical correction factor F can be set equal to 1.0 because the
tube bank is so large. The heat transfer rate per tube is Q = hA·ΔT, where the temperature difference is approximately
ΔT = T WALL − TMEDIAN = 45°C. In general, this temperature difference overpredicts the heat transfer rate slightly, and
a more accurate analysis would use the LMTD. However, let us assume for the moment that a value of ΔT = 45°C can
be used. Then, the heat transfer rate per tube is Q = 16,230 × 1.57 × 45 = 1,146,650 W = 1,146.6 kW = 1.146 MW. Since
there are 400 tubes, the total heat transfer rate is QTOTAL = N × Q = 400 × 1146.6 = 458,650 kW = 458.6 MW. Clearly,
this tube bank is designed to remove a lot of heat! [Ans.]
842 Correlations for Single-Phase Nuclear Heat Transfer
where NR is the number of rows, χ is a geometric correction factor depending on whether the bank is in-line or stag-
gered, and f is the Darcy friction factor, which can be found from charts similar to the Moody charts in Cengel (2003).
For a square bank or one consisting of equilateral triangles, the geometric correction factor is 1.0. When the lateral pitch
is different than the transverse pitch, χ can range from 0.3 to about 6 for an in-line array and from 0.9 to about 1.6 for
a staggered array. In asymmetrical arrays, it is also Reynolds number dependent. Hence, the pressure drop is directly
proportional to the number of rows. The power required to pump a fluid across the bank also increases with the number
of rows because it is proportional to the pressure drop:
Pumping power = υ′ ∆P = m
∆P/ηρ (21.58)
∆P = N R ⋅ χ ⋅ f ⋅ ρ VMAX 2 2
Because the P/D ratio is 2, the maximum velocity is twice the approach velocity. Hence, we should set V MAX = 5 m/s. For
a square in-line array, the geometric correction factor χ = 1. At 20°C, the density of water is 998.2 kg/m3. The pressure
drop is then ΔP = NR·χ·f·ρVMAX2/2 = 10 × 1 × 0.15 × 998.2 × 25 × 0.5 = 18,715 N/m2 ≅ 18.7 kPa or ~2.7 PSI. The pump-
ing power is given by Equation 21.57, where η = 0.70 because the pump is 70% efficient. The pumping power required
to keep the coolant flowing is then mΔP/ηρ = 100 × 18,715/(0.7 × 998.2) = 268 J/s = 268 W. [Ans.]
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Questions for the Student 843
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi, India (2000).
Reactor Theory (Nuclear Parameters), DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
22. In the context of the reactor fluid flow, the Peclet number is equivalent to the product of the Reynolds number and
the Prandtl number. In the context of mass transfer, the Peclet number is equivalent to the product of the Reynolds
number and the Schmidt number. How is the Schmidt number defined, and what abbreviation is used to represent it?
23. How does the Nusselt number for a metallic coolant differ from that for water?
24. For water and liquid sodium at a Reynolds number of 500,000, what is the ratio of their Nusselt numbers?
25. How many terms do the correlations for the Nusselt numbers contain for coolants such as water and carbon dioxide?
26. How many terms do the correlations for the Nusselt numbers contain for metallic coolants such as mercury and
liquid sodium?
27. What is the purpose of the Lyon–Martinelli correlation? Does it apply to water or liquid metals? What conditions
does it impose on the shape of the axial heat flux?
28. What is the purpose of the Seban correlation? Does it apply to water or liquid metals? What conditions does it
impose on the wall temperature profile?
29. For the flow of metallic coolants across a tube bank, such as those that can be found in reactor heat exchangers, what
correlations for the Nusselt number should be used?
30. For the flow of metallic coolants parallel to a tube bank, such as those that can be found in a fuel assembly, what
correlations for the Nusselt number should be used?
31. For the same mass flux and the same P/D ratio, which type of correlation predicts a larger heat transfer coefficient—
a correlation for flow parallel to a tube bank or a correlation for flow perpendicular to a tube bank?
32. What famous correlation for the Nusselt number was used by Westinghouse to help design the Clinch River Breeder
Reactor near Oak Ridge, Tennessee?
33. How accurate does this particular correlation happen to be?
34. Name two additional correlations that have been developed since that time for liquid metal heat transfer in reactor
fuel assemblies.
35. What range of P/D ratios and Peclet numbers are supported by the Kazimi–Carelli correlation?
36. What range of P/D ratios and Peclet numbers are supported by the Schad and the Borishanskii correlations?
37. At what coolant temperatures are the terms in these three correlations usually evaluated?
38. What is a typical mass flow rate through a fuel assembly in a LMFBR core?
39. What is a typical value for the mass flux through a fuel assembly in a LMFBR core?
40. What is the shape of one of these fuel assemblies?
41. How do the P/D ratios for a fuel assembly in a LMFBR core compare to the P/D ratios for a fuel assembly in a water
reactor core?
42. Which type of core has a higher enrichment or fissile content?
43. Which type of core has a higher volumetric power density?
Exercise 21.2
Calculate a representative value for the convective heat transfer coefficient in the same fuel assembly with a Reynolds
number of 800,000. Use the Seider–Tate correlation to perform this calculation, and assume the temperature drop across
the film at the surface of the fuel rods is 10°C. Assume a bulk fluid temperature of 310°C.
Exercise 21.3
Calculate a representative value for the convective heat transfer coefficient for a LMFBR fuel assembly with a Reynolds
number of 500,000. Use the Lyon–Martinelli correlation to perform this calculation, and assume a bulk fluid tempera-
ture of 600°C. How does the heat transfer coefficient compare to that of a typical PWR?
Exercise 21.4
Use the Seban correlation to perform the same calculation. Why is the heat transfer coefficient different in this case?
What is the fundamental difference between the Lyon–Martinelli correlation and the Seban correlation? How can this
difference be explained physically?
Exercises for the Student 845
Exercise 21.5
Calculate the Weisman correction factor for a square reactor lattice with a P/D ratio of 1.33. Apply the Weisman correc-
tion factor to the Dittus–Boelter correlation in Exercise 21.1 to find the heat transfer coefficient for the lattice. How does
the heat transfer coefficient change as the P/D ratio is increased or decreased?
Exercise 21.6
Calculate the Markoczy correction factor for a hexagonal reactor lattice with a P/D ratio of 1.15. Apply the Markoczy
correction factor to the Lyon–Martinelli correlation in Exercise 21.3 to find the heat transfer coefficient for the lattice.
How does the heat transfer coefficient change as the P/D ratio is increased or decreased?
Exercise 21.7
Calculate the equivalent diameter of an interior coolant channel in a PWR fuel assembly having a P/D ratio of 1.25
where the fuel rods are 1 cm in diameter. Approximately how far from the entrance to the channel does the velocity
profile become fully developed and does the convective heat transfer coefficient given by the Dittus–Boelter correlation
reach its asymptotic value? At this distance, approximately how much smaller is the value of h than it is 5 cm from the
entrance to the fuel assembly?
Exercise 21.8
For a hexagonal fuel assembly cooled by liquid sodium where the fuel rods are 8.5 mm in diameter and the pitch of the
rods is 9.8 mm, calculate the surface heat flux when the cladding temperature is 600°C and the bulk sodium temperature
is 590°C. Assume a Reynolds number for the coolant of 100,000.
Exercise 21.9
Liquid mercury was once used as a reactor coolant. Liquid mercury is pumped through a stainless steel pipe at the
rate of 1.6 kg/s. The pipe is 2 cm in diameter. The mercury enters the pipe at 15°C and leaves the pipe at 35°C. How
long is the pipe? Assume the wall temperature is 50°C and the heat flux is constant. What correlation would you use
to find the heat transfer coefficient? What is the value of the Reynolds number, the Prandtl number, and the Peclet
number?
Exercise 21.10
Water flows through a coolant pipe where the diameter of the pipe is doubled. If the fluid properties and the mass flow
rate remain the same, what is the percentage change in the heat transfer coefficient? Assume the flow through the pipe
is turbulent. Now consider an identical pipe where the diameter is held constant but the mass flow rate is doubled. What
is the percentage change in the heat transfer coefficient under these conditions?
Exercise 21.11
Liquid metal flows through a 5 cm stainless steel tube at the rate of 4.5 kg/s. It enters the tube at 415°C and leaves the
tube at 440°C. The wall temperature is 20°C higher than the bulk temperature, and a constant heat flux is maintained
along the tube’s surface. Calculate the amount of heat that flows into the coolant. Calculate the Reynolds number, the
Prandtl number, and the Nusselt number. Using this information, calculate the heat transfer coefficient and the total
length of the tube.
Exercise 21.12
A horizontal tube passes through a heat exchanger where water enters the tube at 20°C and leaves the tube at 40°C.
Saturated steam is then passed over the surface of the tube at 80°C where it is condensed. The heat transfer coef-
ficient on the outer surface of the tube is 10,750 W/m 2 °C. The tube is made of brass with a thermal conductivity of
k = 110 W/m °C. The tube has an ID of 1.34 cm and an OD of 1.59 cm. Calculate the Reynolds number, the Prandtl
number, and the Nusselt number for the water in the tube. What is the convective heat transfer coefficient on its
inner surface?
846 Correlations for Single-Phase Nuclear Heat Transfer
Exercise 21.13
Calculate the total thermal resistance of the brass tube in Exercise 21.12. Using the LMTD (see Chapter 8), calculate the
heat flow rate through the tube. Assume the tube is 1.58 m long. If a heat exchanger contained 1,000 of these tubes, and
they were arranged in such a way that the heat transfer rates were identical, how much steam would 1,000 of these tubes
condense in one second? Assume a latent heat of vaporization of 2,308 kJ/kg.
Exercise 21.14
A tube bank in a heat exchanger is arranged in the shape of a perfect square having 10 tubes per row and 10 rows per
side. The turbulent friction factor for flow across the bank is f = 0.16. If air flows through the bank with a maximum
velocity of 6 m/s, and the density of the air at this velocity is 1.2 kg/m3, estimate the total pressure drop across the tube
bank. Since the bank is an in-line bank, assume for the moment that NR = 10 and χ = 1. If the P/D ratio is P/D = 2, what
is the approach velocity of the air?
22
Natural Convection
in Nuclear Power Plants
22.1 Natural Convection in a Reactor Core
For a variety of reasons, the flow in a reactor core is designed to be as turbulent as possible. The turbulence allows the maximum
amount of heat to be removed from the fuel rods for a given mass flow rate through the core. This flow rate is sustained by pump-
ing the coolant past the fuel rods in a pressurized water reactor (PWR) at about 6 m/s, and as we discussed in a previous chapter, it
requires a significant percentage of the total power output of the plant (about 2% for a commercial PWR) to keep the pumps running.
A flow of this type is said to be forced because a pump is required to force the fluid to flow through the core. In other words,
Important Observation
During normal operation, a reactor is a forced convection machine
However, if the reactor is scrammed and we suddenly lose power to the pumps, then a significant amount of decay heat will continue
to be produced in the core. This decay heat will continue to be produced for some time, and under normal circumstances, it may
take as long as a month for the decay heat to subside to a manageable level. For this reason, the coolant pumps must be kept running
for at least a week after the core is to be shut down for refueling. The general time line to support a scheduled refueling is shown in
Figure 22.1. Normally it takes about one week after a reactor is shut down before the pressure vessel head can be pulled.
FIGURE 22.1 The heat generation rate in a typical light water reactor before and after shutdown.
847
848 Natural Convection in Nuclear Power Plants
where Po is the steady-state power level (in MW thermal). The exact numbers are presented in Chapter 5. The behavior
of the decay heat over time is shown in Figure 22.2. Again, shutdown of the reactor occurs at t = 0.
FIGURE 22.2 A graphic depiction of the decay heat produced by the nuclear fuel after a reactor is shut down. During the first few
years, most of the decay heat is generated by the fission products. Then after about 500 years, the actinides and transuranics generate
more decay heat. However, this long-term decay heat is not modeled particularly well by the Wigner–Way equation. This long-term
decay heat falls as the inverse three-quarter power of the time, that is, P ~ t−0.75, while the Wigner–Way equation predicts that the
decay heat for shorter periods of time falls as P ~ t−0.20.
precise equations to use are given in ANS Standard ANS-5.1-1994, which the reader is encouraged to read. For short
times after the reactor is shut down compared to the total operating time to (i.e., for t ≪ to), the Wigner–Way equation
reduces to
∫ 0.066P t −0.2 t
Q TOTAL (t) = o dt = 0.066 × 1.25Po t 0.8 = 0.0825Po t 0.8 − 1 (22.4)
0
0
where t is the time in seconds, Q is the decay heat in MJ, and Po is the steady-state power output in MW thermal. Thus
by substituting the elapsed time after shutdown into this equation, we find that it takes about 25 seconds before the decay
heat deposited in the core becomes equal to the energy generation rate Po during 1 second of full power operation. If we
plot the total energy deposition in the core as a function of time, the result is the integral heat buildup curve shown in
Figure 22.3. Note that after 1 calendar year, the decay heat deposited in the core is approximately 11,000 greater times
the instantaneous steady-state power output, and after only 1 day, the decay heat deposited in the core is approximately
1,000 times the instantaneous steady-state power level (in MJ), which indicates that 1,000/60 ≈ 17 min of full power
energy must be removed from the core one day after shutdown.
( )
Q TOTAL (t) = 0.0825Po t 0.8 − 1 (22.5)
Obviously, this decay heat must be removed in order to prevent the fuel rods from melting. More specifically, in the event
that the reactor coolant pumps fail, or there is no electricity available to power the pumps, reactors must rely on natural
convection rather than forced convection to remove this decay heat. The exact method for removing the decay heat is a
function of how the nuclear steam supply system and the reactor safety systems are designed.*
∫
* e t dt = (1/(t + 1))e(t +1).
850 Natural Convection in Nuclear Power Plants
FIGURE 22.3 The integrated decay heat buildup in a reactor core following a scheduled shutdown at t = 0.
FIGURE 22.4 The fuel assemblies in PWRs and LMFBRs have slightly different shapes, and the areas of the center channels, the
side channels, and the corner channels are also different.
when the coolant velocity falls to about 0.02 m/s, the flow becomes laminar. Table 22.1 compares the coolant velocities
and the Reynolds numbers where these flow transitions occur. Thus, the flow in a reactor core without coolant pumps
can be either laminar or turbulent. In most cases, it starts out being turbulent, and then becomes laminar as the core
cools down. During a normal shutdown process, the coolant pumps are allowed to run until the decay heat falls to about
one half of one percent of the initial power level Po. Then natural circulation alone can be used to remove the remaining
decay heat. In this case, the flow field becomes laminar approximately one week after a scheduled shutdown. By this
time, the head may be removed from the pressure vessel, and the fuel assemblies may be completely exposed.
TABLE 22.1
Typical Coolant Velocities and Reynolds Numbers in PWR and LMFBR Coolant Channels before and after a
Reactor Is Shut Down for Refueling
Flow Type Fluid Reactor Type Reynolds Number (Re) Coolant Velocity (m/s)
Natural Water PWR 10 3 0.01
Forced Water PWR 5 × 105 6
Natural Sodium LMFBR 103 0.10
Forced Sodium LMFBR 105 10
852 Natural Convection in Nuclear Power Plants
FIGURE 22.5 A reactor storage pool where spent nuclear fuel rods are kept. Pools of this type are also called “spent fuel pools.”
case, the situation becomes more complicated because the heat generation rates are so high. So the questions we need
to answer are
1. How fast does the fluid flow when this occurs?
2. Is the flow laminar or turbulent?
3. Does the hot fluid next to the wall dissipate all of the decay heat that is produced?
4. What does the flow pattern look like under these conditions?
Obviously, all of these factors must be considered when determining how the fuel rods behave. The hot fluid next to the
rod surface experiences an upward force due to a density difference between the hot fluid and the surrounding colder
fluid. The flow pattern over the surface of a typical fuel rod is shown in Figure 22.6. This picture approximates the flow
of fluid of the fluid over plate-type fuel rods, and it can also be used to emulate the flow of fluid over cylindrical fuel
rods. To illustrate the processes involved, we have exaggerated the horizontal scale by a factor of about five. At a certain
point, zTURBULENT, the flow becomes turbulent, and if more heat is added to it, its velocity will continue to increase. Once
the flow has become turbulent, turbulent eddies form in the flow, and turbulent mixing increases as one moves further
up the rod. Conceptually, these eddies are superimposed on the upward directed flow. People have had some success
in analyzing this situation when the flow is laminar. However, the analysis becomes much more difficult when the flow
becomes turbulent.
and the buoyancy force due to the displacement of the colder fluid by the hotter fluid is
or
F = ( ρc − ρ) gδ − (µdu/dy) (at y = 0) (22.9)
22.8 The Force Balance for Fluid Flowing over a Heated Surface 853
FIGURE 22.6 A heated surface showing how the Grashof number can be derived for vertical flow.
where δ is the boundary layer thickness. The average velocity vAVG of the fluid flowing past the wall can then be found by
inserting this force term into the momentum equation (see Chapter 15), and integrating it over the width of the boundary
layer itself (between y = 0 and y = δ). The result is
where β is the volumetric coefficient of expansion. (For water at 30°C, the value of β is about 0.0003/°C.). A complete
range of values for water and air are presented in Table 22.2. Note that because air can be considered to be an ideal gas,
β = 1/T, where the air temperature T is the absolute temperature. Sometimes β is also called the thermal expansion
coefficient. Moreover, when there is a temperature difference between the heated fluid and the surrounding colder fluid,
the definition for β becomes
ρ( T) = ρCOLD (1 − β ( T − T∞ )) (22.11)
TABLE 22.2
The Thermal Expansion Coefficients and Prandtl Numbers for Air and Water as a Function of Their Temperature
Here T is the temperature of the hot fluid, T∞ is the ambient temperature of the cold fluid some distance away from the
surface of the wall, and ρ COLD is the density of the cold fluid before heat is added to it. Hence, we now have a convenient
way to calculate the Reynolds number at the fuel rod surface. From Equations 22.10 and 22.11, it is simply
or
where L is the characteristic length or length scale of the heated surface. Notice that if the wall temperature is the same
as the bulk temperature (i.e., T WALL = T∞), then the overall fluid velocity vAVG becomes zero. Next, in order to find the
convective heat transfer coefficient under these conditions, we must define two additional dimensionless numbers called
the Grashof number and the Rayleigh number. These numbers are not as well known as the Reynolds number and the
Prandtl number, but they always appear when the convection is natural rather than forced. Now let us examine how these
numbers are actually defined.
22.9 T he Role of the Grashof Number and the Rayleigh Number in Natural Convection
The Grashof number plays the same role in natural convection as the Reynolds number does in forced convection.
Essentially the Grashof number is the ratio of the buoyancy force to the viscous force exerted on a drop of fluid near
a solid surface:
See Figure 22.7 for a physical interpretation of the Grashof number. The Grashof number and the Rayleigh number were
named for Lord Rayleigh and Franz Grashof, whose pictures are shown in Figure 22.8. They initially proposed these
numbers in the mid-1800s when the field of fluid mechanics was still in its infancy. In most textbooks, the Grashof num-
ber is given the symbol Gr. For flat vertical plates, it can be shown that the local value of the Grashof number is given by
where z is the distance from the bottom of the plate where the number is to be found. Just like the Reynolds number, the
Grashof number is dimensionless. However, during natural convection, its value increases with the cube of the distance
from the start of the surface where the convection occurs. The average value of the Grashof number is then found by
integrating the local Grashof number (Equation 22.14b) over the total length L of the surface: Thus, the integration in
Equation 22.15 must be performed from z = 0 to z = L.
In forced convection, this is equivalent to defining an average Reynolds number for the entire surface (see Chapter 17).
Another important dimensionless grouping that appears in the study of natural convection is the product of the Grashof
number Gr and the Prandlt number Pr. Since the Prandlt number is given by
Pr = µc/k (22.16)
22.9 ROLE OF GRASHOF NUMBER AND RAYLEIGH NUMBER 855
FIGURE 22.7 In practice, the Grashof number measures the relative magnitudes of the viscous and the buoyancy forces that act on
a drop of fluid at the surface of a nuclear fuel rod. The ratio of these forces determines the magnitude of the Grashof number Gr.
FIGURE 22.8 A picture of Lord Rayleigh on the top left and Franz Grashof below him. These men are responsible for today’s
modern definitions of the Rayleigh Number and the Grashof Number. These numbers play an important role in the study of natural
convection in reactor fuel assemblies when the coolant pumps are turned off. Specifically, the Rayleigh number can be used to deter-
mine the convective heat transfer coefficient. Before he was proclaimed Lord Rayleigh by the Queen of England, his original name
was John William Strutt. (Pictures provided by Wikipedia.)
856 Natural Convection in Nuclear Power Plants
and this latter product is called the Rayleigh number Ra. Note that this can also be written as
The Rayleigh number is used in reactor heat transfer because it is a convenient way to express the convective heat trans-
fer coefficient for single-phase flow when natural circulation is important but forced convection is not. In most prob-
lems, the Rayleigh number is a function of axial position. Thus, Equation 22.17 represents the local Rayleigh number,
and the value of Ra changes as the boundary layer develops. The average Rayleigh number can be found by integrating
Equation 22.17 over the length of the surface. The average Rayleigh number for a vertical plate of height L is then
L
Ra =
∫ 0
Ra z dz/L = (1/4) ⋅βg ( TWALL − T∞ ) L3 ν2 × Pr (22.19)
or
where ΔT WALL = T WALL − T∞. For natural convection, the heat transfer coefficient h can then be expressed as a function
of the Rayleigh number alone. That is to say,
h = C ⋅ Ra N (22.21)
where Ra = Gr·Pr, C is an empirically determined constant that varies slightly with the geometry of the heated
s urface, and N is an exponent having a value of about 0.25 for laminar flows. For a heated vertical plate where
T WALL ≠ T∞, the factor L/4 is then lumped into the value of C. This same methodology can be extended to other
reactor geometries but doing so requires different values for C and N. For now it is important to keep in mind that
most of the physics is tied up in the value of Ra. For a flat vertical plate, the Rayleigh number has a value of about
1 × 10 9 (or 1 billion) when the transition from laminar to turbulent flow occurs. Example 22.2 shows when the flow
becomes turbulent if the temperature difference between the wall and the colder fluid is known. Of course, this is
just an approximation to real reactor flow because the coolant is flowing over a flat plate rather than a cylindrical
fuel rod. For some geometries like heated horizontal and vertical plates, the transition points between the laminar
and turbulent flow regimes can be written as a function of the Rayleigh number alone. These transition points for
some simple geometric shapes are shown in Table 22.3. Notice that the Rayleigh number is proportional to the tem-
perature difference ΔT between the wall T WALL and the colder fluid T∞ some distance away from the wall. Table 22.4
then d iscusses when the Reynolds number should be used and when the Rayleigh number should be used. Hence,
whenever the reactor coolant pumps are turned off, the Rayleigh number Ra can be used to determine these transi-
tion points. When the coolant pumps are turned back on and the flow becomes forced again, the Reynolds number is
a more appropriate number to use.
TABLE 22.3
During Natural Convection, the Rayleigh Number Is Used to Determine the Heat Transfer Coefficient h
Type of Heated Plate Rayleigh Numbers for Laminar Flow Rayleigh Numbers for Turbulent Flow
Vertical Ra = 10 – 10
4 9 Ra = 109 – 1011
Horizontal Ra = 104 – 107 Ra = 107 – 1010
During forced convection, the Reynolds number serves the same purpose.
22.10 Rayleigh Numbers for Horizontal Surfaces 857
TABLE 22.4
During Forced Convection, the Reynolds Number Can Be Used to Determine when
the Flow Becomes Turbulent, and Many Heat Transfer Correlations Are Based on This
We can summarize the information presented in Tables 22.3 and 22.4 by making the following observations:
h FORCED = k/L ⋅ Re m Pr n
For natural convection, the convective heat transfer coefficient is determined by the product of the Grashof
number and the Prandtl number, which are normally raised to the same power:
h NATURAL = k/L ⋅ Gr n Pr n
The product of the Reynolds number and the Prandtl number is then defined as the Peclet number Pe, and
the product of the Grashof number and the Prandtl number is called the Rayleigh number Ra.
In reactor work, these observations allow us to find the convective heat transfer coefficient when the flow is either
natural or forced.
{
Nu = 0.6 + 0.387Ra 0.17 1.0 + (0.56/Pr)0.56 } (22.22)
0.30 2
This unusual expression was first proposed by Churchill and Chu in 1975,* and it can also be applied to a hot nuclear
fuel rod as long as the coolant surrounding the rod does not boil. The examples below illustrate how this correlation
can be used.
(
Ra = βg ⋅ ∆TWALL D3 4 ν2 Pr )
so Ra = [(0.003 × 9.8 × 60 × 0.000001)/(4 × 0.225 × 10 −12)] × 3 = 5.28 × 10 −6/0.9 × 10 −12 = 5.86 × 106. [Ans.]
{
Nu = 0.6 + 0.387Ra 0.17 1.0 + (0.56/Pr)0.56 }
0.30 2
* See Churchill, S.W. and Chu, H.H.S., “Correlating Equations for Laminar and Turbulent Free Convection from a Horizontal
Cylinder”, International Journal of Heat and Mass Transfer, 18:1049, 1975.
22.11 Natural Convection over a Hot Flat Surface 859
Spent plate-type fuel rods can be modeled using this approach as well. Only in the case of these rods, both surfaces must
be considered when calculating the rate at which the decay heat is removed. Now let us examine each of these orienta-
tions in more detail. First, consider the rate of heat loss from a flat vertical plate.
22.11.1 Finding the Heat Transfer Rate from a Hot Vertical Plate
Assume for the moment that the plate is in the form of a square 1 m on each side. Suppose one side of the plate is main-
tained at a temperature of 90°C and is to be air-cooled. The other side is insulated and does not radiate any heat. Now
let us examine the rate of heat transfer from the plate to the air which surrounds the plate. Assume the temperature of
the surrounding air is 30°C. Several correlations have been developed over the years to handle a situation such as this.
However, because the plate is flat and vertical, Equation 22.20 can be used. The characteristic length in this case is the
height of the plate, which in this case is L = 1 m. The film temperature is the average of the surface and ambient tem-
peratures or 60°C. For air at 60°C and 1 atm, its physical properties are
( )
Ra = gβ ⋅ ∆TWALL L3 ν2 Pr = 3.65 × 10 9
The flow appears to be slightly turbulent. The exact expression to use for the Nusselt number depends on the size of the
Rayleigh number. If the Rayleigh number is less than 109, the Nusselt number should be found from
and the Rayleigh number is greater than 109, the following correlation should be used:
These correlations are used in many applications, but the alternative correlation
{
Nu = 0.825 + 0.387Ra 0.17 1.0 + (0.49/Pr)0.56 }
0.30 2
( for Ra = 10 4
)
– 1013 (22.25)
is more accurate and can be used for any Rayleigh number between 104 and 1013. In other words, it can be used whether
the flow is laminar or turbulent. Now let us determine the rate of heat transfer from the plate. The Nusselt number is
( )
0.33
Nu = 0.10Ra 0.33 = 0.10 × 3.65 × 10 9 = 143
and the surface area of the plate is A = L2 = 1 m2. Since ΔT WALL = 60°C, the heat transfer rate Q from the plate to the
surrounding air is then
22.11.2 Finding the Heat Transfer Rate from a Hot Horizontal Plate with the Hot Side Facing Up
Now let us find the heat loss from the same surface when it is placed horizontally and when its hot side is facing up
rather than down. The characteristic length in this case is L c = A/P = L2/4L = L/4 = 0.25 m. The Rayleigh number is
( )
Ra = gβ ⋅ ∆TWALL L c 3 ν2 Pr = 5.7 × 10 7
860 Natural Convection in Nuclear Power Plants
The appropriate correlation to use for the Nusselt number in this case is
( )
0.25
Nu = 0.54Ra 0.25 = 0.54 × 5.7 × 10 7 = 46.9
and convective heat transfer coefficient and the surface area are
and A = L2 = 1 m2. Since ΔT WALL = 60°C, the heat transfer rate Q from the plate to the surrounding air is then
22.11.3 F
inding the Heat Transfer Rate from a Hot Horizontal
Plate with the Hot Side Facing Down
Finally, consider the rate of heat loss from the same surface when it is placed horizontally and when its hot side is facing
down. The Rayleigh number for this surface is
( )
Ra = gβ ⋅ ∆TWALL L c 3 ν2 Pr = 5.7 × 10 7
When the heated surface is facing down, an alternative correlation must be used for the Nusselt number. The form of
this correlation is
( )
0.25
Nu = 0.27Ra 0.25 = 0.27 × 5.7 × 10 7 = 24.95
and convective heat transfer coefficient and the surface area are
and A = L2 = 1 m2. Since ΔT WALL = 60°C, the heat transfer rate Q from the plate to the surrounding air is then
Notice that the heat transfer rate for natural convection is much lower when the hot side is facing down. This is because
the hot fluid becomes trapped beneath the surface of the plate and can only escape from beneath it by flowing around
the edges. As a result, the cooler fluid in the vicinity of the plate will have trouble reaching the plate, which reduces the
heat transfer rate. Hence, using the information at our disposal, we have found that
h = 5.25 W/m 2 °C (for a horizontal surface—hot side UP)
This is about what we would have expected based on physical arguments alone. Thus, the same surface with the hot side
up has the highest heat transfer coefficient and the identical surface with the hot side down has the lowest. When the
same surface is vertical, the heat transfer coefficient is approximately the average of these two values.
22.11.4 Accounting for the Effects of Radiative Heat Transfer during Natural Convection
Now let us return to our earlier discussion where we contended that it is not necessarily possible to ignore the effects of
radiative heat transfer when the surface of a heated plate is large. If the plate is hot enough, the plate will lose heat to its
surroundings by radiation as well as natural convection. Assuming the surface of the plate to be black (or nearly so), the
emissivity ε will be close to unity (ε ≈ 1.0) and the radiative heat transfer rate from the plate will be given by
( )
Q RADIATION = εAσ TWALL4 − TSUR 4 (22.28)
where TSUR is the absolute temperature of the plate’s surroundings (in °K), and σ = 5.67 × 10 −8 W/m2 − K4 is the Stefan–
Boltzmann constant derived from Quantum Theory (see Chapter 10). Thus, the radiative heat transfer rate from the
plate is
( )
Q RADIATION = εAσ TWALL 4 − TSUR 4 = 505 J/s = 505 W (from radiation alone)
Notice that the radiative heat transfer rate is larger than the convective heat transfer rate when a surface is large
and its temperature is high. Thus, the effects of radiative heat transfer cannot be ignored when a surface undergoing
natural convection is both large and hot. The techniques we have just discussed can also be applied to spent fuel stor-
age casks such as those that are located close to reactor sites to store spent nuclear fuel. Pictures of these casks are
shown in Chapter 11 of our companion book.* To determine the flow of heat from these casks, the side walls can be
approximated by flat vertical surfaces (see Section 22.11.1), and the top of the cask can be modeled as a flat horizontal
surface (see Section 22.11.2). The exercises at the end of the chapter then ask the reader to analyze the behavior of
these casks in more detail.
ρ( u ∂ u/ ∂z + v ∂ u/ ∂ y ) = µ ∂ 2 u ∂ y 2 − ∂P/ ∂z − ρg (22.29)
where the z-axis in this case is parallel to the surface, and the y-axis is perpendicular to it. Because there is a density
difference between the boundary layer and the fluid far away from it, we can replace the term −∂P/∂z in the momentum
equation with ρ∞g. The momentum equation then becomes
ρ ( u ∂ u/ ∂z + v ∂ u/ ∂y ) = µ ∂ 2 u ∂ y 2 + ( ρ∞ − ρ) g (22.31)
The final term (ρ∞ − ρ)g is the net upward force on a unit volume of the fluid inside the boundary layer. Hence, this force
term initiates and sustains the flow of coolant in a fuel assembly when the reactor coolant pumps are turned off. As long
* See Nuclear Engineering Fundamentals: A Practical Perspective by R.E. Masterson, CRC Press (2017).
862 Natural Convection in Nuclear Power Plants
as the fluid does not boil, we can also equate this density difference Δρ to the thermal expansion of the coolant which is
represented by the volumetric expansion coefficient β. Remembering that the definition of β is
u ∂ u/ ∂z + v ∂ u/ ∂y = µ ∂ 2 u ∂ y 2 + gβ ( T − T∞ ) (22.33)
In other words, Equation 22.33 governs the motion of fluid within the boundary later when there is a temperature
difference between the wall and the coolant. Now suppose we divide all of the independent and dependent variables
in Equation 22.33 by a characteristic length L, an arbitrary reference velocity v = Re(μ/ρL), and a suitable temperature
difference which we will call T WALL − T∞. The momentum equation then becomes
( )
U ∂U/ ∂ Z + V ∂U/ ∂Y = gβρ2 ( TWALL − T∞ ) L3 µ 2 δT Re 2 + (1/ Re) ∂ 2 U ∂Y (22.34)
2
where
is then the Grashof number which we discussed previously. Notice that the Grashof number is proportional to the
square of the fluid density ρ. It is also directly proportional to the temperature difference ΔT = (TWALL − T∞) between
the wall and the coolant. Thus, when the wall and the coolant are at exactly the same temperature ΔT = 0, Gr = 0, and
the momentum equation reduces to
2
U ∂U/ ∂ Z + V ∂U/ ∂Y = (1/Re) ∂ 2 U dY (22.37)
which is just the momentum equation for viscous flow. The term on the right-hand side can then be approximated by the
frictional pressure drop.
22.13 Finding the Convective Heat Transfer Coefficient during Natural Circulation
For simple reactor geometries, it is possible to estimate the size of convective heat transfer coefficient for a free vertical
surface. In general, the convective heat transfer coefficient becomes a function of axial position because it depends on
the difference between the surface temperature and the free stream temperature T∞. Because of this, most correlations
attempt to find the average Nusselt number over the surface. Almost all of these correlations have the same basic form:
where L is a characteristic dimension of the surface, and C is a constant of proportionality that is geometry and flow
regime dependent. The components of the Nusselt number are shown in Figure 22.9. For a vertical fuel rod, both Nu and
Ra are based on the fuel rod diameter D. In other words, both the Grashof number and the Prandtl number are raised
to the same power for natural circulation flows. The Rayleigh number Ra, which is again the product of the Grashof
number and the Prandtl number, is given by
Sometimes the Rayleigh number is also written in terms of the kinematic viscosity ν = μ/ρ as
FIGURE 22.9 During natural convection, heat transfer correlations for reactor heat transfer are usually expressed in terms of
the Rayleigh number raised to some power n, and then multiplied by another constant C, which is both geometry and flow regime
dependent. For most reactor geometries, the values of n and C must be determined experimentally.
The values of C and n which appear in the equations for Nu depend on the geometry of the surface and the flow regime.
Normally the value of n is 0.25 for laminar flow and 0.33 for turbulent flow. However, whether the flow is laminar or
turbulent must be determined by the value of the Rayleigh number Ra and not the value of the Reynolds number Re.
The value of C can vary from about 0.05 to 0.50 for a vertical surface. However, for reactor fuel rods, it is usually less
than 0.1 and typically closer to 0.05. There are a few specialized geometries where correlations have been developed
for the value of Nu and therefore the heat transfer coefficient h. These geometries and the correlations that can be used
with them are shown in Table 22.5. Unfortunately, none of these geometries map directly to nuclear fuel rods. Notice
that the characteristic lengths are geometry dependent and the fluid properties must be evaluated at the film temperature
which is given by TFILM = (TWALL + T∞)/2. A vertical cylinder can be treated as a vertical plate when the diameter of the
cylinder is sufficiently large that the curvature of the surface can be neglected. It can be shown that this condition can
be satisfied when
where L is the length of the cylinder. Unfortunately, most nuclear fuel rods do not always meet this condition. However,
the Nusselt number can still be found using the correlations presented in the references at the end of the chapter. For a
flat vertical plate, the plate height H is its characteristic length L. All of the correlations presented in Table 22.5 can be
used when the temperature of the plate is constant. Now let us see how these correlations can be applied to a plate-type
fuel rod with a constant nuclear heat flux. When the heat flux is constant, the rate of heat transfer Q = q″A is known, but
the surface temperature is not. The temperature of the fluid increases with height but so does the surface temperature
of the plate. However, many studies have shown that the Nusselt number is nearly the same for a constant surface tem-
perature and a constant heat flux. Therefore, the correlations shown in Table 22.5 can also be used for a plate-type fuel
rod with a constant axial heat flux provided that the plate midpoint temperature TH/2 is used in the evaluation of the film
temperature, the Nusselt number, and the Rayleigh number. The average Nusselt number in this case is then
and the average convective heat transfer coefficient for the plate is
h = (k/H) ⋅ Nu (22.43)
where H is the plate height. Naturally, horizontal cylinders and spheres have different expressions for the Nusselt num-
ber, and the tubes in horizontal steam generators can use these correlations when the flow is no longer forced. Again, the
characteristic length L of these surfaces is given by
L = A/PW (22.44)
TABLE 22.5
The Effect of Geometry on Nusselt Number When Natural Convection Occurs
FIGURE 22.10 An enclosed surface in which different types of natural circulation occur between hot and cold walls.
panes). An example of this flow pattern is shown in Figure 22.10. If the right-hand side of the enclosure is colder than
the left-hand side, a circulatory pattern may develop due to a temperature difference ΔT = THOT − TCOLD between the
surrounding walls. For laminar flow, the flow field will resemble the pattern shown on the left-hand side, and for turbu-
lent flows, it will resemble the pattern shown on the right. The coolant flows up the warm wall as it is heated and down
the cold wall as it is cooled. The shape of the flow field depends on the aspect ratio L/D of the enclosure as well as the
temperature difference ΔT. When the flow rate is very low, and the enclosure is very narrow, the convective heat transfer
process does not contribute much to the heat transfer coefficient because the distance the fluid has to travel is small
before it changes direction. Most heat transfer in this case occurs by conduction across the fluid gap. Using Fourier’s law
of conduction, the surface heat flux is given by
where D is the gap width. When the Nusselt number is based on the gap width alone, we find that
or
Nu = 1 (22.47)
when h = k/D. Thus, the Nusselt number is very close to unity for very low values of the Rayleigh number (say Ra < 100,000),
and this further implies that the value of C in Equation 22.21 is about 0.05 when the flow is laminar. At higher Rayleigh
numbers, the flow becomes turbulent and it resembles the pattern shown on the right-hand side of Figure 22.10. In this
case, turbulent eddies develop in the middle of the flow, and turbulent boundary layers develop on each side. A more
sophisticated analysis indicates that the turbulent Nusselt number in this case is given by
where the Rayleigh number is based on ΔT = THOT − TCOLD. Other experiments lead to similar results, except that the
value of C is about 0.06 when the Rayleigh number is very high (Ra ≈ 1010). Thus, the enclosed surfaces tend to exhibit
the following range of Nusselt numbers depending on the type of flow:
866 Natural Convection in Nuclear Power Plants
Again, the definition of what is laminar and what is turbulent depends on the value of the Rayleigh number Ra and NOT
the value of the Reynolds number Re. Other correlations with higher values of C (say C ~ 0.10) use Rayleigh numbers
where the temperature difference ΔT = T WALL − TBULK is based on the bulk temperature of the fluid instead of the
temperature difference between the walls.
22.15 Nusselt Numbers and Heat Transfer Coefficients for Natural Circulation
Not surprisingly, the Nusselt number can also be correlated to the Rayleigh number:
when the flow is natural, and L is a characteristic dimension of the surface. Normally the value of n is 0.25 for laminar
flow and 0.33 for turbulent flow, and the value of C is between 0.05 and 0.10. Now let us examine the value of Nu for
different values of Ra in a reactor fuel assembly. For this exercise, we will assume the coolant is water. The results
are shown in Table 22.6. In these fuel assemblies, laminar flows occur between Rayleigh numbers of Ra = 10 4 – 109
and turbulent flows occur between Rayleigh numbers of Ra = 109 – 1010. At low Rayleigh numbers, the Nusselt
number is between 1 and 10, and for high Rayleigh numbers, it is between 50 and 200. If the reader compares these
values to typical values of the Nusselt number for forced convection (see Chapters 20 and 21), it is easy to see that the
Nusselt numbers for natural convection are high enough to remove most of the decay heat from the core. The results
we have just quoted can be used for reactor coolants such as gases and water but they CANNOT be used for liquid
metals. During natural convection, metallic coolants have Nusselt numbers closer to 5, and their Prandtl numbers are
between 0.004 and 0.005. As a point of reference, the Prandtl numbers for water and most gaseous coolants are close
to 1.0 (see Chapter 21).
TABLE 22.6
Nusselt Numbers during Natural Circulation
in the pool for some time. The pool boiling accompanying this process is shown in Figure 22.11. Pool boiling occurs
when the convection is natural rather than forced, and when most of the surrounding coolant is stationary or nearly
so. We will have more to say about how the heat transfer coefficient behaves in Chapter 24. For a known burnup B
(which is usually measured in MWT/T), the amount of decay heat these fuel assemblies produce is proportional to
the time they have been in the core, and once they have been removed from the core, the heat generation rate can
be deduced from the Wigner–Way equation (see Section 22.2). We can then find the heat flux q″ at the surface of an
individual fuel rod by taking the decay heat that the fuel produces Q(t), and dividing it by the surface area of the rod:
A = πR 2 L (22.51)
Here L is the length of the fuel in the rod (about 5 m). The heat flux as a function of time is then given by
With this equation, one can calculate the wall temperature T WALL(t) that is needed to find the Rayleigh number Ra.
Normally, the wall temperature is also a function of axial position TWALL = TWALL (z, t). However, let us suppose for the
moment that we can use the average values for the heat flux q″(t) and the wall temperature TWALL(t) to find the Rayleigh
number. Then we simply need to find a relationship between the bulk temperature TBULK, the wall temperature T WALL,
and the heat flux q″(t). The easiest way to do this is to use Newton’s Law of Convection:
where h is the local heat transfer coefficient. Normally the values of q″(t) and TBULK can be thought of as independent
variables that are set by the boundary conditions. Solving for T WALL then gives
If we substitute this result into the previous equations for the Rayleigh number, we obtain
FIGURE 22.11 An example of pool boiling with the reactor coolant pumps turned off.
868 Natural Convection in Nuclear Power Plants
where we have assumed that TBULK = T∞. Normally the value of h is proportional to the Rayleigh number Ra raised to
some power less than unity. From our previous discussion, we also know that
h = C ⋅ Ra n (22.56)
where Ra = Gr × Pr, and C is an empirically determined constant that varies slightly with the geometry. Here the expo-
nent n has a value of about 0.25 for laminar flows. (For turbulent flows, it has a value closer to 0.33.) Thus, we can write
{ }
1/ (1− n )
Ra = ρ2 cgβq′′(t)z 3 µ kC (22.59)
If Ra < 109, the flow is laminar, and if Ra ≥ 109, the flow is turbulent. The heat transfer coefficient h can then be calculated
iteratively using this approach. The reader may have noticed that the equation for the Rayleigh number requires a value
for ρ, c, μ, and k, and that the value of n is problem dependent. In the approach we have just outlined, these values are
the values for the liquid only. However, if the coolant happens to boil, then we must use the appropriate values of ρ, c, μ,
and k for the liquid and vapor mixture. Thus, the density of the fluid ρ close to the surface of the rods must be the density
of the two-phase mixture when it boils. The same conclusion applies to the values for c, μ, and k.
22.17 Experimental Scaling Parameters for Pool Boiling for Reactor Fuel Assemblies
When the coolant surrounding a stationary surface becomes hot enough to boil (see Chapter 24), it becomes much
more difficult to predict how the flow patterns will behave when the flow rates are low. Normally, one tries to design
an experiment to simulate the behavior of a fuel assembly or a small number of fuel assemblies under these conditions.
However, the power density is usually so high in real reactors that it usually not possible to do this with more than a few
fuel rods. If we have to determine how a fuel assembly will behave when the flow is low, then there are two ways that we
can proceed. (1) First, we can run a computational fluid dynamics program like Fluent to simulate the behavior of a real
fuel assembly. (2) Second, we can perform an experiment to predict the behavior of the “real” fuel assembly. Whatever
parameters are measured in the experiment must then be “adjusted” using a scaling law to predict the behavior of a
real fuel assembly when the number of rods is increased. Normally, reactor vendors do a number of half-scale or 1/5th
scale tests to obtain this data, and then they apply a proprietary scaling algorithm to extrapolate the behavior of a real
fuel assembly from the experimental model. For single phase fluids, experiments like this are conducted all of the time.
However, for two phase flows, they are considerably more difficult.
When the coolant starts to boil, the process we have just described becomes a lot more messy. One usually tries
to compensate for these effects by using another coolant like Freon that has a behavior similar to water. Then if the
water boils, and we need to understand the flow patterns in the fuel assembly. We must create a scaled-down version
of the same experiment using Freon as the coolant and observe how the Freon behaves when the flow rates are low.
In a well-designed experiment, we can then predict the flow quality, the flow patterns, and the different flow regimes
that are present inside the real fuel assembly. Normally, two distinctly different types of flows have to be taken into
account. In boiling water reactors (or BWRs), the coolant in the lower part of the assembly undergoes subcooled boil-
ing, and the coolant in the upper part of the fuel assembly undergoes bulk boiling. (Both of these regimes are discussed
in Chapter 23.) If we use a simulated coolant like Freon-12 to model these conditions, then the size of the system, the
pressure of the system, and the temperature of the system must be “adjusted” so that the flow patterns behave in exactly
the same way. In a BWR with natural convection, van de Graff and van der Hagen* found that the linear dimensions of
a BWR fuel assembly where Freon was used instead of water had to be reduced by a factor of 0.46, the working pressure
had to be reduced to 22.6 bars (~313 PSI), and the power consumption of the scaled fuel assembly had to be reduced to
* van de Graaf, R. and van der Hagen, T.H.J.J., “Two-phase flow scaling laws for a simulated BWR assembly”, Nuclear Engineering
and Design, July, 1994.
22.20 The Passive Safety Systems in the Westinghouse AP-1000 869
2% to achieve exactly the same results as a real fuel assembly with a water–steam mixture. When the flow rates are low
and boiling occurs, this is generally how the behavior of an actual fuel assembly is simulated.
An Important Observation
Removing the decay heat from the core is the top priority of any reactor operator
When a reactor is shut down for refueling, and the refueling process goes as planned, the decay heat is removed by keep-
ing the coolant circulating through the Nuclear Steam Supply System for a week or two. This allows the coolant pumps
to send the decay heat to the condensers, where it is then rejected from the plant. By this time, the decay heat has fallen
to about ½ of 1% of the steady-state power level Po, and there is no longer enough residual decay heat in the core to melt
the fuel rods—even if one happens to turn off the coolant pumps entirely. However, the fuel pin temperatures can still
remain above 500°C, and although the reactor has been shut down safely, the fuel rods are normally “too hot to handle”
by a human being. If the coolant pumps are turned off earlier in the refueling cycle, the process of natural convection will
still be able to remove the decay heat from the fuel rods, but this heat must still be transferred to the coolant and “dumped”
to the condensers to prevent the core from melting. If there are no coolant pumps available to circulate the coolant, then
other safety systems must be utilized to ensure that this residual decay heat is completely removed. In the next section,
we would like to briefly discuss the evolution of these systems and explore the actual versions that are in use today. Like
most of the critical components of a nuclear power plant, they have become more refined and better designed over time.
generators. This eliminates the mechanical seals found on the previous generation of reactor recirculation pumps. The
pressurizer in the the AP-600 and the AP-1000 is about 30% larger than the pressurizer in the previous generation of
these plants. This redesign reduces the likelihood of opening a safety relief value (pressure relief valve) during an over-
power transient. The core power density is also lowered slightly to reduce the probability of core damage during some
overpower transients. Passive cooling is provided for the primary loops and for the containment building. In the primary
loops, these cooling systems automatically cool the core during a LOCA. There are two ways that these passive cooling
systems are implemented. First, during the initial stages of an LOCA, a high-pressure water makeup tank provides the
water needed to keep the core cool. This tank provides additional water to the core automatically without the need for
operator intervention. An automatic depressurization system is then used to reduce the pressure in the primary loop so
that coolant can flow from an external tank to the pressure vessel under the force of gravity alone. Together these two
systems are about a factor of ten more effective in reducing the probability of core damage than the previous systems.
This reduces the cost of insurance for the plant as well.
The second way that the AP series differs from earlier designs is that it uses a passive residual heat removal system
(or PRHRS) to remove heat from the core when the steam generators are not available. This system works by using a
large pool of water from the in-containment refueling water storage tank (IRWST) to quench the core and to serve as a
heat sink to absorb this additional decay heat. This large pool of water can also be used to help recondense the steam
generated during the early stages of the LOCA. The recondensation process limits the maximum pressure level that can
be reached in the containment building (see Chapter 34). Thus, it helps to suppress the pressure spikes using the same
approach that is employed with a BWR suppression pool. The third and final way that the AP series differs from earlier
designs is that it uses a PCCS to remove the decay heat from the containment building after an LOCA is initiated. This
system relies on natural convection or natural circulation to remove heat from the containment building without the
need for fans, pumps, and other electrically powered equipment. The idea behind the PCCS is illustrated in Figure 22.12.
The PCCS is the final way in which heat can be passively removed from the plant after a severe reactor accident. On
average, an AP-1000 has about 75% less piping in the primary loop and about 30% less piping in the secondary loop
than a conventional two-loop PWR. The number of components has also gone down as the components of the NSSS have
become better integrated. There are much smaller numbers of valves in the containment building as well.
FIGURE 22.13 The flow of coolant through the Westinghouse AP-1000 when the reactor is operating normally. Notice that the
valve to the IRWST is closed at this time.
872 Natural Convection in Nuclear Power Plants
FIGURE 22.14 The timeline for an AP-1000 reactor accident involving the loss of on-site power.
In general, the reactor coolant pumps require huge amounts of electric power to operate (normally between 5 and 10
MWe per pump), and under some conditions, this power is simply not available at the site. At the same time the core flow
begins to stagnate, the coolant pumps are also shut off to the spent fuel pool (because they also require power to operate).
This means that cold water is no longer pumped to the pool, and the spent fuel assemblies in the pool continue to dump
decay heat to the surrounding water. This causes the water temperature in the pool to rise—sometimes by several °C.
In earlier designs, this could have caused some very severe cooling problems. However, in the AP-1000, this additional
decay heat causes a separate set of passive safety systems to come online. After about 2 min, the steam generator water
level falls to the point that it activates what is called the passive core cooling system (or PCCS). The passive core cool-
ing system consists of a heat exchanger that is located inside a large water storage tank immediately above the reactor
pressure vessel. This water storage tank is sometimes referred to as IRWST.
Because of the difference in elevation, and the difference in density between the cold water in the tank and the warm
water in the core, the coolant automatically begins to circulate between the two systems. Heat is transferred to the stor-
age tank, and heat is removed from the core. This process continues under the influence of gravity alone for some time.
The entire process is illustrated in Figure 22.15. Eventually the decay heat begins to subside. After 2 h, the decay heat
decreases to about 1% of its full power value. After about 3 h, the cooling water in the spent fuel pool begins to boil as
well (although this boiling is restricted to just the hotter fuel assemblies in the pool). At this point, the decay heat from
the spent fuel pool is transferred from the water to the steam. This causes some of the water in the pool to evaporate.
Any evaporated water is replaced with water from a water supply tank that is located in the adjacent cask wash down
pit. This cold water is sent to the spent fuel pool under the action of gravity alone. Therefore, a separate coolant pump
is not required to replace the water in the pool. After about 5 h, the passive heat removal system has transferred enough
decay heat from the reactor core to the IRWST that the water inside of the tank begins to boil. This causes steam to be
produced in the containment building.
The pressure inside of the containment building begins to rise, and the water in the IRWST continues to boil. The
resulting steam begins to condense on the steel liner inside of the containment building. (The condensed steam flows
back into the IRWST.) After about 6 h, the control system decides to cool the containment building, and it opens up
another set of valves to begin the flow of cooling water from the PCCS. The water for this system comes from a large
water storage tank that is located near the roof of the containment building, and cold water from this tank is allowed to
run over the outside of the steel liner. This cools the top and the sides of the steel liner and helps to remove the waste
heat. The steam that is generated in the IRWST is transferred to the steel liner inside of the containment building
through a set of pipes at the top of the tank. The water flowing over the outside of the steel liner removes this decay
heat through the process of condensation. (The dynamics of this process are explored in Chapter 34.). The air flow-
ing between the steel liner and the outside of the containment building naturally cools the heated steel liner. In other
words, the air surrounding the containment building ultimately carries away the decay heat from the core. (This is the
same process that is used in a natural draft reactor cooling tower, which we discussed in Chapter 8.) After about 7 h,
enough of the heat is removed from the steel liner that some of the steam begins to recondense back into water inside
of the containment building. This water is redirected back into the IWRST for continued use in removing decay heat
from the core. This process of heat removal continues until the core has completely cooled. After about 36 hours, the
22.23 Natural Convection in Steam Generators and PWR Fuel Assemblies 873
FIGURE 22.15 The flow of coolant through the Westinghouse AP-1000 when the power to the pumps is lost. Notice that the valve
to the IRWST is opened, and that the decay heat is then sent to the IRWST. The core continues to be cooled passively without the
need for electric power. The control rods have been inserted into the core.
temperatures in the core and the containment building finally stabilize, and the reactor has been successfully shut down.
The decay heat removal process does not require an external power source.
At this point, the reactor’s decay heat is slightly more than one half of one percent of its full power output. The ancil-
lary diesel generators are then restarted within 3 days after the initial blackout occurs to provide power for post-accident
monitoring. Water makeup pumps are usually employed to transfer water from the ancillary water storage tank to the
passive containment cooling water storage tank to continue cooling the containment building. These pumps also transfer
additional water to the spent fuel pool. In most blackout scenarios, external power is assumed to be completely restored
to the plant within seven calendar days. This can occur by “rebooting” the power grid, or by restarting the ancillary
diesel generators. At this point, additional water is transferred to the ancillary water storage tank from other sources.
After 7 days, the reactor’s decay heat has fallen to about one-third of one percent of its full power output (see Chapter 5).
In the case of a 1,000 MWE PWR, this corresponds to about 3,300 MWT × 0.003 = 10 MW of thermal energy. (Note
that this is enough energy to heat a reasonably large cooling pond.) A more comprehensive discussion of the AP-1000
and its safety system systems can be found at the following URL:
http://ap1000.westinghousenuclear.com/station_blackout_home/passivecorecooling.html
which is maintained by the Westinghouse Nuclear corporation. The reader is encouraged to visit this site to view a short
movie of how the AP-1000’s passive safety systems work. The cooling of the fuel assemblies in the spent fuel pool is also
discussed. Eventually, all reactors will be equipped with a passive heat removal system of some type. A large number
of articles that describe how these systems work can be found on the Internet. The reader is encouraged to find these
articles with the help of the Google Search Engine.
FIGURE 22.16 An example of when the flow becomes laminar in a reactor fuel assembly and in a steam generator tube. The
velocities where this occurs are different because the hydraulic diameters are different. Also, there may be several steam generators
to help distribute the hydraulic load.
4,000 pressure tubes with an internal diameter of 7/8 of an inch (~0.0222 m). In a U-tube steam generator, the average
length of these tubes is about 16.5 m, which is about twice the height of the tube bank itself. At 1% of rated flow, the
Reynolds number in an average tube is
Flow in circular steam generator tubes becomes laminar below a Reynolds number of about 2,300, so the friction fac-
tors and heat transfer coefficients must also be laminar when the aggregate mass flow rate is 50 kg/s. In particular, the
friction factor at this flow rate is
h = (k/D) ⋅ Nu = (0.53/0.0222) ⋅ 4.36 = 23.9 × 4 ≈ 104 W /m 2 °C (for the steam generator) (22.62)
This heat transfer coefficient is based on a Nusselt number of 4.36 (which corresponds to the correct value to use for a
constant heat flux). Because the Reynolds number is 690 when the flow rate falls to 1%, the transition point between the
turbulent and laminar flows occurs when the mass flow rate is about 2,300/690 = 3.3% of its full power value, or 0.033
× 5,000 kg/s = 165 kg/s. The coolant velocity when this occurs is about 0.1 m/s in an average pressure tube, and when
the flow rate reaches 1% of its normal value, the velocity falls to about 0.03 m/s. Clearly, the flow through the tube bank
is natural rather than forced. Now let us turn our attention to the flow through the core. In PWR fuel assemblies, the
flow can also become laminar under certain conditions. In Chapter 17, we learned that the Reynolds number in a PWR
fuel assembly is about 550,000 when a modern PWR is operating at full power. This compares to a Reynolds number
of about 70,000 in a steam generator pressure tube under the same conditions. Thus, the Reynolds number in the core
must fall to slightly less than one half of one percent of its normal value before the flow becomes laminar. Although this
is less than it is in a steam generator pressure tube, it can still occur about a day after shutdown. The coolant velocity at
this time is between 0.01 and 0.02 m/s. The friction factor is again
Questions for the Student 875
The heat transfer coefficient is slightly higher in the core because the hydraulic diameter of the subchannels is less.
Now let us discuss the same processes in Chapter 23 when the coolant begins to boil. Obviously the correlations for the
Nusselt number and the heat transfer coefficient become more complicated than they are for single phase fluids, and an
additional level of refinement is normally required to be able to predict the correct values for the coolant velocities and
the nuclear heat flux.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi, India (2000).
Reactor Theory (Nuclear Parameters), DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
Fluid Property Calculator for Water, Air, and Many Industrial Gases
http://www.mhtl.uwaterloo.ca/old/onlinetools/airprop/airprop.html
1 0. At what coolant velocity in an LMFBR fuel assembly is the flow no longer considered to be turbulent?
11. How thick is the boundary layer over the surface of the nuclear fuel rods when the pumps are running normally?
12. How thick does the boundary layer become over the surface of the rods when the coolant pumps are turned off?
13. Under what conditions can natural convection occur in a spent fuel pool?
14. When the flow is forced, express the Nusselt number as a function of the Reynolds number and the Prandtl number.
15. Name two other components in a nuclear power plant where the flow can become either natural or forced.
16. What is the Nusselt number in a PWR fuel assembly when the flow is natural?
17. What is the Nusselt number in a PWR fuel assembly when the coolant pumps are at full power and the flow is
forced?
18. Fill in the following sentence with the appropriate word or phrase: The _______ number plays the same role in
natural convection as the Reynolds number does in forced convection.
19. How is the Grashof number defined, and how is it abbreviated when used in practice?
20. In the definition of the Grashof number, a constant β sometimes appears. What is the meaning of this constant, and
how can its value be found? What is its value for ordinary water?
21. For a flat vertical plate or a nuclear fuel rod, is the Grashof number position dependent? If so, what is its dependence
in the vertical direction?
22. What role does the Rayleigh number play when the convection in a reactor coolant channel becomes natural? How
is the Rayleigh number defined, and how is it abbreviated when it is used in practice?
23. Who was first responsible for proposing the Rayleigh number, and before he was proclaimed Load Rayleigh by the
Queen of England, what was his original name?
24. What fluid properties are required to evaluate the Rayleigh number? Is the fluid viscosity one of these properties?
25. What two other dimensionless numbers of reactor fluid mechanics appear in the definition of the Rayleigh number?
How are these dimensionless numbers defined, and what symbols are used to represent them?
26. Fill in the following sentence with the appropriate word or phrase: In classical heat transfer and fluid flow, the
Reynolds number is represented by the symbol ____, the Prandtl number is represented by the symbol ______,
the Grashof number is represented by the symbol _____, the Nusselt number is represented by the symbol ______,
the Peclet number is represented by the symbol ______, and the Rayleigh number is represented by the symbol
_____.
27. In practice, it turns out that each of these six dimensionless numbers (with the exception of the Nusselt number)
represent force ratios. Explain each of these force ratios, and explain how they are used.
28. What is a typical value of the Rayleigh number in a reactor fuel assembly when the flow is natural and laminar?
29. What is a typical value of the Rayleigh number in a reactor fuel assembly when the flow is natural and turbulent?
30. Both the Rayleigh number and the Grashof number are a function of the temperature of a surface and the temperature
of the fluid far away from the surface. When referring to these temperatures, what are they called? Is this r elationship
a linear one?
31. What is the relationship between the Nusselt number and the Rayleigh number?
32. How can the convective heat transfer coefficient in a nuclear power plant be deduced from the Rayleigh number, or
equivalently, from the Nusselt number when the coolant pumps have been shut off?
33. Are the Rayleigh number, the Prandtl number, and the Grashof number all dimensionless? If so, what force ratios
do they represent?
34. In a reactor fuel assembly, what characteristic dimensions are required to define these numbers? Are these
characteristic dimensions the same for each number?
35. For a flat, heated horizontal plate, at what Rayleigh number does the transition from laminar to turbulent flow
occur?
36. For a flat, heated vertical plate, at what Rayleigh number does the transition from laminar to turbulent flow occur?
37. Write the Nusselt number as a function of the Rayleigh number for a heated surface.
38. To what power is the Rayleigh number raised when the heated surface is horizontal?
39. To what power is the Rayleigh number raised when the heated surface is vertical?
40. For enclosed flows where both heated and cooled surfaces are present, express the Nusselt number as a function of
the Rayleigh number when the flow is laminar.
41. For enclosed flows where both heated and cooled surfaces are present, express the Nusselt number as a function of
the Rayleigh number when the flow is turbulent.
42. Fill in the following sentence with the appropriate word or phrase: In reactor fluid mechanics, the product of the
Reynolds number and the Prandtl number is called the ______ number, and the product of the Grashof number and
the Prandtl number is called the _______ number. What relevance do each of these particular combinations have
to nuclear heat transfer?
Exercises for the Student 877
Exercise 22.2
Immediately after a large commercial PWR is shut down for refueling, a single nuclear fuel rod can generate 4,300 W
of thermal energy from decay heat alone. If a rod such as this is put into a spent fuel pool, where the water temperature
is 20°C, and the surface temperature is about 90°C, what is the average heat flux at the surface of the rod? Assume the
active height of the fuel in the rod is 4 m.
Exercise 22.3
For the fuel rod in Exercise 22.2, the decay heat generated by the fuel falls relatively quickly with time. After 1 month,
what is the linear heat generation rate? Assume the rod has a diameter of 1 cm.
Exercise 22.4
You are put in charge of assessing the thermal behavior of a reactor core if the power to the coolant pumps is lost. Suppose
the core is to be shut down for refueling, and approximately 1 week after it is shut down, the power is actually lost. What
is the approximate velocity of the coolant after this occurs? Does the flow become laminar or turbulent? What correla-
tions should be used to find the Nusselt number and the convective heat transfer coefficient under these conditions?
Exercise 22.5
A heat transfer tube inside a reactor heat exchanger receives water from the primary loop at 320°C. The power to the
pumps in the secondary loop is suddenly lost. However, the water continues to flow due to natural convection, and the
water on the secondary side has a temperature of 250°C and a pressure of 5.5 MPa (~800 PSI). What is the approximate
value of the convective heat transfer coefficient on the secondary side of the tube?
Exercise 22.6
For gaseous coolants such as helium, air, and carbon dioxide show that the thermal expansion coefficient β that appears
in the definition of the Grashof number is given by β = 1/T, where T is the absolute temperature of the gas in °K. Why
does the absolute temperature appear in this equation, and what does it imply? Can a relationship such as this be used
for a liquid such as water?
Exercise 22.7
After the coolant pumps are turned off, fuel rods can still radiate a lot of heat. Consider a small axial segment of a typi-
cal rod 5 cm long which generates 200 W of heat. Moreover, assume this heat is removed by the coolant due to natural
convection alone. Suppose the average surface temperature of the rod is 340°C and the average temperature of the sur-
rounding coolant and structure is 300°C. The rod is 1 cm in diameter. If the rod acts as a black body, can it generate
more radiant energy than the heat convected away by the coolant? Exactly how much radiant energy does it produce?
Exercise 22.8
An unknown reactor coolant has a Grashof number of 1 × 109 and a Prandtl number of 0.5 when it flows over a 1 cm
diameter nuclear fuel rod. What is this coolant likely to be? What is the Rayleigh number for this coolant, and if its
thermal conductivity is k = 0.65 W/m °C, what is a representative value for its convective heat transfer coefficient?
Exercise 22.9
A research reactor in India is evaluating a new core design consisting of parallel vertical plates 2.25 m high and 1.5 m
wide. The plates contain uranium dioxide fuel pellets, and the maximum temperature of the plate surfaces is limited to
878 Natural Convection in Nuclear Power Plants
975°C. The lowest allowable temperature of the bismuth is 325°C. Using the correlation Nu = 0.13(Gr·Pr)0.33, calculate
the maximum possible heat dissipation from each side of the plates when the core must be cooled by natural convection
alone. Assume the Rayleigh number is evaluated at the mean temperature of the surface film.
Exercise 22.10
A steam pipe is used to provide heat to the auxiliary building of a nuclear power plant on a cold winter’s day. The pipe is
6 cm in diameter and is covered with a layer of insulation 2 cm thick. The insulation has a surface emissivity of 0.92. The
surface temperature of the insulation is 75°C, and the air surrounding the pipe has a temperature of 25°C. Considering
the heat loss from the pipe from both radiation and natural convection, calculate (1) the heat loss from a section of the
pipe 5 m long, and (2) the overall heat transfer coefficient and the heat transfer coefficient due to radiation alone.
Exercise 22.11
Nuclear power plants are frequently lit up like Christmas trees at night using long-lasting incandescent bulbs. Calculate
the heat transfer coefficient from a 60 W incandescent light bulb to the surrounding air, which has an ambient tempera-
ture of 25°C. Assume the light bulb is spherical in shape and has a diameter of 5 cm. Find the percentage of the energy
carried away by free convection. Calculate the Nusselt number using the correlation Nu = 0.60(Gr·Pr)0.25.
23
Fundamentals of Two-Phase
Flow in Nuclear Power Plants
23.1
Fundamentals of Two-Phase Flow
In reactor work, two-phase flow can occur in many different components of the NSSS including the core, the steam generators, the
steam turbines, and the condensers. This flow is usually comprised of different phases that belong to the same substance (i.e., water), and
when multi-phase flows occur, these flows are usually comprised of three or more components of which at least one belongs to a differ-
ent substance. The first phase is called the liquid phase and the second phase is called the vapor phase. Most of the time, these phases
have completely different densities and physical properties, and they can affect the components of the NSSS in different ways. In power
producing systems, the most common example of two phase flow is water and steam. A related example of a three phase flow is water,
steam, and air. In two phase flows involving water and steam, the vapor phase is produced from the liquid phase by adding energy to
the liquid phase. This causes a phase change to occur. Phase changes can occur from both pressure changes and temperature changes.
When it is the result of a temperature change, the temperature of the fluid becomes equal to the saturation temperature. When it is the
result of a pressure change, the pressure of a fluid falls below the saturation pressure. When either of these two conditions is met, some
of the liquid will be converted into a vapor and this conversion process continues until all of the liquid phase has been vaporized away.
The energy per unit mass (or the specific enthalpy) required to achieve this phase transition is called the enthalpy of vaporization.
The enthalpy of vaporization, which is represented here by symbol hlv, (or hfg), is fluid dependent, and for a reactor coolant like water
in the secondary loop of a PWR at 8 MPa, it is about 1441.4 kJ/kg. Moreover, the value of hlv is pressure dependent, and it normally
decreases as the pressure is raised. For example, for water at 100 kPa (~14.7 PSI) its value is 2257.5 kJ/kg, and at 15.5 MPa (~2250 PSI)
its value is approximately 965 kJ/kg. Hence smaller amounts of heat are required to produce a phase change as the pressure is raised.
In reactor work, two-phase flow occurs when the coolant starts to boil. Two phase flow occurs in the primary loop of BWRs and
in the secondary loop of PWRs. In these loops, phase changes are used to convert steam into electric power, and these phase changes
form the basis of the Rankine thermal cycle (see Chapter 9). The energy required to produce these phase changes is provided by the
fuel rods in the core. In nuclear power plants, there are many different types of fuel rods, and the thermal energy produced by these
rods is eventually used to convert at least some of the water in the plant into steam. This conversion process is regulated by the nuclear
steam supply system or NSSS (see Chapter 8). In some parts of the NSSS (and particularly in the core), there can be many different
types of two phase flow. Each type of two phase flow (see Figure 23.1) is characterized by a different flow pattern or flow regime. Five
or six of these flow regimes can be present in a boiling core at the same time. In some types of two phase flow, the water and the steam
move together in the same direction (and generally at the same speed). When the velocities of both phases are the same, these flows
are called homogeneous flows, and they are normally assumed to have the same temperature and pressure. However, when one phase
moves at a different speed (or occasionally in a different direction), the flow is called an inhomogeneous flow or slip flow. Flows of this
type are more common than homogeneous flows in nuclear power plants and they normally occur in BWR cores and in PWR steam
generators. Finally, the liquid and vapor phases can be at the same temperature, or they can be at different temperatures. When they
are at the same temperature, they are said to be in thermal equilibrium, and the resulting flow is called an equilibrium flow. When
the temperature of the vapor phase is higher than the temperature of the liquid phase, the flow is called a non-equilibrium flow, and
the phases in these flows are not in thermodynamic equilibrium. The flows that occur during Loss of Coolant Accidents (or LOCAs)
(see Chapter 30) are perhaps the best known examples of flows where the two phases are not in thermal equilibrium and where the
velocities of each phase can be different. Fortunately, LOCAs almost never occur in commercial nuclear power plants, although their
effects must be taken into account in preparation of the plant’s Final Safety Analysis Report or FiSAR.
The easiest way to understand two-phase flows is to understand the conditions under which they occur. When a phase change
occurs, the vapor phase will have a different density than the liquid phase, and this may cause the two phases to move in d ifferent
directions or at different speeds. In a gravitational field, the downward force exerted on the liquid phase is Fl = ρlg, and the downward
879
880 Fundamentals of Two-Phase Flow in Nuclear Power Plants
FIGURE 23.1 A picture illustrating various types of two-phase flow that can occur within an enclosed area.
force exerted on the vapor phase is Fv = ρvg. Thus, the density difference between the phases causes the vapor to rise
relative to the liquid. Using Newton’s first law, the net force exerted on the two-phase mixture is then
Fnet = ma = ( ρl − ρv ) g (23.1)
and the velocity may be found by integrating the previous equation. Moreover, an additional force term must be added
to Equation 23.1 to account for the friction between the phases. Normally, this term is proportional to the velocity
difference between the phases, and the friction that is created in this way is called interfacial shear. In general, there are
three factors that determine whether a phase change will occur. They are
☉☉ the temperature of the fluid
☉☉ the pressure of the fluid
☉☉ the heat of vaporization.
The heat of vaporization depends on the molecular forces that hold the fluid together. The primary difference between
the behavior of fluids in nuclear power plants and those in other thermal systems is that they are subjected to much
higher pressures in either the reactor core or the nuclear steam supply system (NSSS). Hence, two-phase flows in nuclear
power plants do not behave in the same way as flows in low-pressure applications associated with turbo machinery. In
pressurized water reactors (PWRs), only a small amount of nucleate boiling is allowed to occur in the core because of
the way the NSSS is designed. Hence, nucleate boiling usually occurs in only the hottest fuel assemblies near the top of
the core. However, in boiling water reactors (BWRs), a phase change is deliberately introduced into the core about one
quarter of the way between the inlet and the outlet. Thus if the inlet to the core is assumed to located at z = 0 and the
outlet of the core is assumed to be located at z = H, boiling will begin to occur when z = H/4. This boiling becomes vig-
orous when z = H/2, and by the time the outlet to the core is reached, most of the space in the coolant channels consists
of vaporized water or steam. The void fraction at this time can be about 80%. This phase change can be beneficial as
long as the cladding does not melt and the fuel rods do not fail. We will have more to say about this in Chapter 24 when
we discuss how two-phase flow can affect the value of the convective heat transfer coefficient. Normally the convective
heat transfer coefficients are higher for two-phase flow than they are for single-phase flow. Hence, in water reactors,
flow boiling is used to increase the heat transfer rate without necessarily increasing the mass flow rate.
23.2
Comparing Evaporation, Boiling, and Condensation
There are three basic ways that a reactor coolant can be converted from a liquid into a vapor or from a vapor into a liquid:
1. evaporation
2. condensation
3. boiling
23.2 Comparing Evaporation, Boiling, and Condensation 881
FIGURE 23.2 Whether a fluid decides to evaporate or boil depends on the vapor pressure. If the vapor pressure is less than the
saturation pressure (represented here by the atmospheric pressure), then a fluid will evaporate from an open container. However, if it
is greater than the saturation pressure, then additional energy must be added to the fluid to cause it to boil.
Each of these processes is completely different although they have many things in common. The conditions under which
a coolant can boil or evaporate are summarized in Figure 23.2. Now let us discuss each of these processes individually.
23.2.1
Evaporation
Evaporation is a process that occurs when a stationary pool of liquid (or one that is nearly so) converts itself from a liq-
uid to a vapor as the result of pressure differences between the pool and its environment. The molecules near the surface
of the pool are the first ones to evaporate. Evaporation occurs at the boundary of the pool when the vapor pressure is
less than the saturation pressure. For example, water in the Mississippi river at 20°C evaporates into the surrounding air
at 20°C if the relative humidity is 60%. This is because the saturation pressure of water at this temperature is 2.34 kPa,
whereas the vapor pressure of air at 20°C and 60% relative humidity is 1.40 kPa. In other words, evaporation occurs
when PVAPOR < PSAT:
Moreover, evaporation does not require bubble formation or growth. It only requires the vapor pressure to be less than
the saturation pressure at the liquid–vapor interface. The evaporation of water in a cup of coffee (see Figure 23.3) is
common example of this effect..
23.2.2 Condensation
Condensation is the inverse of evaporation. It occurs when a hot vapor transfers some of its energy to a colder fluid or
surface. Condensation occurs through two primary mechanisms called dropwise condensation and surface condensa-
tion (see Chapter 34). In dropwise condensation, the vapor condenses into larger droplets, and these droplets may be
attached to a colder surface, or they may not. When there is no surface to attach to, the condensation of water vapor into
larger droplets leads to rain. Filmwise condensation occurs when the cooling surface is completely covered by these
droplets, and the water vapor condenses to form what is called a surface film. Neither of these processes is very com-
mon in nuclear power plants except when there is an accident and steam is vented into the containment building to later
recondense. Figure 23.4 illustrates how these processes work. Condensation and evaporation are responsible for creating
882 Fundamentals of Two-Phase Flow in Nuclear Power Plants
FIGURE 23.3 Pictures showing the evaporation of water from a cup of coffee (left), and the boiling of a pot of hot water (right).
Boiling requires the existence of a heated surface, while evaporation does not.
FIGURE 23.4 In addition to their importance to nuclear science and engineering, condensation and evaporation are responsible for
the world’s water cycle. (Pictures provided by NASA—see earthobservatory.nasa.gov.)
the world’s water cycle. In other words, the evaporation of water is responsible for the formation of clouds, which in turn
leads to rain. In this cycle, high temperatures lead to evaporation and low temperatures lead to condensation.
23.2.3
Boiling
Boiling is entirely different than either evaporation or condensation. It occurs at the boundary between a liquid and
a solid surface (or “wall”) when the temperature of the surface exceeds the saturation temperature of the liquid. In
general, boiling is both temperature and pressure dependent. For example, at sea level where the atmospheric pressure
is 14.7 PSI or 101.325 kPa, water in contact with a hot surface boils at 101°C because the saturation temperature of the
water at atmospheric pressure is 100°C. Hence, boiling occurs when TSURFACE > TSAT:
Boiling also occurs in PWR fuel assemblies when surface temperature of the fuel rods exceeds the saturation tempera-
ture of the coolant, which is about 344.5 °C at 15.5 MPa or 2250 PSI. In power reactors, boiling is also accompanied
by the rapid motion of bubbles away from or at the surface. (In reactors, this heated surface is usually assumed to be a
fuel rod.) These bubbles move into the surrounding liquid and increase the thermal energy of the fluid as a whole. Boiling
in which this occurs is called pool boiling, and pool boiling can occur in any large pool or tank. It can also occur in a
pot of water on your stove (see Figure 23.3). Strictly speaking, pool boiling requires the surrounding pool of liquid to be
23.3 Types of Boiling in Nuclear Power Plants 883
stationary (or nearly so). If the same fluid is then set in motion with an external device such as a reactor coolant pump,
a fan, or a rotor, the boiling becomes what is called flow boiling. Flow boiling is the most common form of boiling in
nuclear power plants. In general, the convective heat transfer coefficients for flow boiling are much larger than they are
for pool boiling. In pool boiling, the convection is said to be natural, while in flow boiling, it is said to be forced.
23.3
Types of Boiling in Nuclear Power Plants
Flow boiling begins when bubbles start to form on a heated surface. Normally this boiling begins with what is called
nucleate boiling. Then as the surrounding fluid becomes hotter, nucleate boiling begins to become bulk boiling and
then film boiling. Bulk boiling and film boiling require additional energy beyond that which is required for nucleate
boiling to occur. Boiling in reactor fuel assemblies can be further classified as saturated boiling and subcooled boiling
or surface boiling. Each type of boiling has a specific meaning and occurs under a slightly different set of conditions.
For example, in BWRs, nucleate boiling begins when the specific enthalpy of the fluid reaches about 1,290 kJ/kg, and
bulk boiling begins when the fluid specific enthalpy reaches about 1,400 kJ/kg. Film boiling occurs at specific enthalpies
greater than 1,475 kJ/kg. (In PWRs these numbers are higher because the operating pressures are higher.) Now let us
examine the specific conditions under which each of these types of boiling can occur.
23.3.1
Pool Boiling and Bulk Boiling
Pool boiling (see Figure 23.5) occurs when a hot object in the middle of a large stationary pool causes the surrounding
liquid to boil. In pool boiling, vapor is created by an energy source internal to the pool. Bulk boiling is similar to pool
boiling except that it occurs when the fluid has acquired enough additional energy that almost all the fluid in the pool is
close to or at the saturation temperature. Then all the liquid in the pool participates in the boiling process, and the boil-
ing is not localized to a particular point in the pool. Sometimes this type of bulk boiling is also called volume boiling.
23.3.2
Nucleate and Film Boiling
Nucleate boiling is different than bulk boiling because it occurs much closer to a heated surface. In nucleate boiling,
vapor bubbles form around small cavities or imperfections in the heated surface called nucleation sites (see Section
23.9). These bubbles grow until they enter the flow stream, where they either join together to form larger bubbles or
voids or where they collapse back into the colder fluid. Nucleate boiling is important in reactor work because it defines
the amount of boiling that is permitted to occur before the convective heat transfer coefficient at the surface of the fuel
rods reaches its maximum permissible value. In PWRs, this is the only type of boiling that normally occurs. It almost
always occurs near the top of the core because this is where the coolant temperature is highest. Film boiling is different
than the nucleate boiling or bulk boiling because it occurs when a continuous film of liquid or vapor blankets the wall.
Film boiling usually occurs further downstream than nucleate boiling. It is normally accompanied by a sharp change
in the value of the convective heat transfer coefficient at the surface of a fuel rod. Under certain conditions, film boiling
and nucleate boiling may coexist for some time. This type of boiling is called film nucleate boiling. Sometimes it is also
called transition film boiling or partial film boiling. In reactor coolant channels, this type of boiling must be avoided
FIGURE 23.5 Pool boiling consists of several major regimes. In volume or bulk boiling, the bubble patterns away from a heated
surface resemble those shown on the left. Prior to the boiling pool achieving this state, several boiling regimes develop on the sur-
face being heated similar to those shown on the right. The figure on the right was provided by www.engr.iupui.edu.
884 Fundamentals of Two-Phase Flow in Nuclear Power Plants
because of the potential for the surface of the fuel rods to dry out. Film boiling and excessive nucleate boiling then lead
to what is called the boiling crisis (see Section 23.46).
23.3.3
Saturated and Subcooled Boiling
Boiling can be further subdivided into saturated boiling and subcooled boiling. Both of these types of boiling are based
on the bulk temperature of the fluid, TBULK, which represents the volume weighted fluid temperature at a specific spatial
location. For a circular coolant pipe having radius R, the bulk fluid temperature is given by
where v(r, z) is the velocity of the coolant at radius r and axial position z. Sometimes subcooled boiling is also called surface
boiling. In subcooled boiling, the bulk temperature is slightly lower than the saturation temperature. However, the local
temperature may be the same as the saturation temperature—particularly close to the boundary of the wall. In subcooled
boiling, vapor bubbles form where the temperature of the fluid is the highest, and they generally collapse back into the sur-
rounding fluid, where the bulk temperature is less than the saturation temperature. Subcooled boiling occurs in BWR and
PWR cores, and it is not necessarily undesirable as long as it can be controlled. Saturated boiling occurs after subcooled
boiling, and it begins when most of the fluid reaches the saturation temperature. In this case, vapor bubbles that are cre-
ated do not collapse back into the fluid. Instead, they enter the flow stream, and if there are enough of them, they combine
together to form larger bubbles or voids. Both saturated boiling and subcooled boiling may be either of the nucleate or film
type. However, if the convection process is not forced, the only type of saturated boiling that is allowed to occur is volume
boiling. Saturated boiling is almost always associated with a higher bulk fluid temperature than subcooled boiling. Within
each flow regime, the rate of bubble formation and growth depends on the mass flow rate and the surface heat flux.
23.4
Boiling Points in PWRs and BWRs
The specific conditions under which subcooled boiling, nucleate boiling, bulk boiling, and film boiling occur are pres-
sure dependent. In American PWRs, the primary loop operates at a pressure of about 2,250 PSI (15.5 MPa), and the
secondary loop operates at a pressure of about 1,000 PSI (7 MPa). The primary loop of a BWR operates at a pressure of
about 1,050 PSI (7.25 MPa). Table 23.1 shows the specific enthalpies at which subcooled boiling, nucleate boiling, bulk
TABLE 23.1
The Specific Enthalpies at Which Subcooled Boiling, Nucleate Boiling, Bulk Boiling, and Film Boiling Occur in PWRs and BWRs
Representative Values of the Coolant Enthalpy Where Different Types of Boiling Can Occur
7 MPa (Representative Pressure for a PWR Secondary Loop)
Saturation temperature = 285.83°C Subcooled boiling Nucleate boiling Bulk boiling Film boiling
Specific enthalpy h (kJ/kg) ~1,270 ~1,280 ~1,390 ~1,470
boiling, and film boiling occur under these conditions. In a BWR, the saturation temperature is about 288°C, and in a
PWR core, it is about 345°C.
23.5
Attributes of Two-Phase Flow
In fluid mechanics, two-phase flow is defined as a flow where both the liquid and vapor phases move together. The
phases may move at the same speed, or they may move at different speeds. Moreover, they may either have the same
temperature or they may have different temperatures. Two-phase flow may also be classified as single component flow
(where both the liquid and vapor phases come from the same substance), or two-component flow, where they do not.
In nuclear systems, the most common example of single component flow is water–steam flow, and the most common
example of two-component flow is air–water flow. Under certain conditions, air–water flow is used to emulate water
steam flow.
23.6
A ir Water Emulation of Steam–Water Flows
Sometimes air–water mixtures are used to simulate the behavior of steam–water mixtures. This is because it takes a
great deal of heat to accurately replicate two-phase flow at the temperatures and pressures in a reactor core, while using
air avoids this difficulty entirely because it is plentiful, easy to use, and cheap. However, air–water flows do not behave
in exactly the same way as steam–water flows because steam is denser than air. At high vertical flow rates, the difference
in their behavior is small, but at low vertical flow rates, it can be significant. Thus, the pressure drop correlations do not
behave in exactly the same way for up-flow as they do for downflow, and it is not always appropriate to infer the motion
of steam–water mixtures from the motion of air–water mixtures. Except for highly turbulent flow, where the Reynolds
number is high, the liquid and vapor phases do not move at the same speed, and generally speaking, the vapor moves
past the liquid. To describe this behavior, a slip ratio is used. In reactor work, the slip ratio S is defined as the ratio of
the velocity of the vapor phase to the velocity of the liquid phase.
Thus, the slip ratio is a dimensionless number. When both phases move at the same speed, then vl = vv and the slip ratio
becomes equal to unity. This condition is called a no-slip condition. We will have more to say about how the slip ratio is
used in Section 23.17. We would now like to turn our attention to how bubbles are formed before they enter the flow stream.
23.7
Bubble Formation and Growth
In reactors, bubble formation and growth signify the start of two-phase flow. Bubble formation requires the creation
of small pockets of fluid where the local pressure is greater than or equal to the saturation pressure PSAT. Bubble
growth occurs when the pressure difference ΔP = pV − p∞ driving the growth rate exceeds the value of the surface
tension 2σ/R at the liquid–vapor interface. That is, bubbles will begin to form when ΔP > 2σ/R.
23.8
Pool Boiling and Bubble Growth
In pool boiling, a hot object initiates the boiling process. When boiling first begins, at least a few isolated pockets of
fluid are created near the hot object where the fluid becomes superheated. The degree of superheat determines the rate
of bubble formation in the pool. Normally, there are four parameters that determine how many bubbles can form and
how quickly they can form. They are
☉☉ The degree of local superheat
☉☉ The amount of gas or vapor already present in the pool
☉☉ The wetting characteristics or the surface tension of the surface
☉☉ The presence of microscopic cavities, pores, or scratches on the surface itself.
886 Fundamentals of Two-Phase Flow in Nuclear Power Plants
Each of these parameters creates a nucleation site, and these sites are also called nucleation centers or nucleation
aids. In reactor cores, the number of nucleation sites can be very high because of the presence of ionizing radiation and
because of the production of larger numbers of hydrogen and oxygen ions in the coolant while it is continually bom-
barded by this intense radiation. These ions promote bubble formation and growth because they help to exert a repulsive
force on the electrically charged bubbles. As a result, the coolant in a reactor core will have a slightly higher heat trans-
fer coefficient for the same flow rate and surface heat flux than the identical coolant in a coal-fired power plant. In other
words, this additional ionization increases the size of the heat transfer coefficient.
23.9
Surface Effects and Bubble Growth
A rough surface will create more bubbles than a smooth surface will. This is because microscopic scratches, pores,
and cavities on the surface act as nucleation sites. In many materials, the number of nucleation sites is directly pro-
portional to the average surface roughness ε. However, these cavities generally have a distribution of sizes and shapes
about an average value or mean. A typical distribution of cavity sizes is shown in Figure 23.6. Here small cavities and
crevices act as nucleation aids because they tend to entrain the vapor, and they cannot always be completely filled with
liquid because of surface tension effects. Thus, once a bubble starts to form, it continues to grow because the tempera-
ture inside the crevice is greater than the temperature of the surrounding liquid. The heat flow is higher because the
total surface area that the crevice presents to the fluid is higher as well. The growth and detachment of bubbles from
these crevices is a relatively straightforward process that we would now like to discuss. As we mentioned earlier, the
rate of bubble growth depends strongly on the wetting characteristics of the surface. This is in turn a function of the
surface tension, which has the dimensions of force per unit length, or energy per unit area. These two representations
of the surface tension are equivalent, but the energy per unit area is also called the surface energy. Surface tension is
measured in Newton per meter (N/m) although it’s most common unit of measurement is dynes per centimeter (dyne/
cm). Bubble surface tension is given the symbol σ (rho), and it depends entirely on the physical characteristics of the
fluid in contact with the surface. In particular, it is defined as the ratio of the surface force F to the length L along which
the force acts:
FIGURE 23.6 A typical distribution of cavity sizes along a heated surface where bubbles can form.
23.10 A Bubble Force Balance 887
FIGURE 23.7 Examples of bubble growth and detachment from a heated surface and a representative distribution of bubble sizes.
Surface tension is due to intermolecular forces called Van der Waals forces, which were discussed in Chapter 7. These
forces cause the fluid particles to be attracted to each other more than they are to the surface. The stronger these forces
become, the stronger the surface tension will be. The surface tension falls as the t emperature rises because higher
temperatures tend to break the molecular bonds that hold the molecules together. In general, there are three separate
parameters that are required to predict how a bubble behaves. They are
☉☉ The pressure of the vapor inside of the bubble, which we will call PIN
☉☉ The pressure of the liquid outside of the bubble, which we will call POUT
☉☉ The surface tension at the surface of the bubble, which separates the liquid from the vapor. We will call this σlv.
These three parameters collectively determine how fast a bubble will grow, shrink, or detach from a surface. The pro-
cesses of bubble formation, growth, and detachment are shown in Figure 23.7. Notice that different values of the surface
tension result in different rates of bubble growth. The surface tension of water at 100°C is 58.85 dynes/cm, whereas
the surface tension of liquid sodium at 100°C is ~200 dynes/cm. Thus, sodium-cooled reactors normally have much
“sticker” coolants than water-cooled reactors do (see Chapter 4).
23.10
A Bubble Force Balance
When a bubble begins to form, the upward force on the bubble must be balanced by the downward force of the fluid
and the surface tension σ, which holds the bubble to the surface. When a bubble is attached to a heated surface and it is
hemispherical in shape, this bubble force balance can be represented by the equation
The force of the vapor pressing upward on the top hemisphere of the bubble is
The force of the liquid pressing downward on the top hemisphere of the bubble is
The force difference must be equal to the product of the surface tension σ and the bubble’s circumference 2πR:
or
For a hemispherical bubble that has only one surface, the surface tension required for these forces to balance (from
Equation 23.10 above) is
And when the bubble detaches from the surface, the downward force due to the surface tension is twice the surface
t ension times the circumference so that
Thus, the vapor pressure PIN inside the bubble is related to the pressure of the liquid POUT outside the bubble by Equation
23.13. Obviously, the vapor pressure inside of the bubble must be greater than the pressure of the liquid outside of the
bubble, or the bubble will collapse (see Figure 23.8).
FIGURE 23.8 Stable bubbles can only exist when the difference between the pressure inside and outside of the bubble is
balanced by the surface tension σ on the bubble’s surface. In general, the force balance for a hemispherical bubble is different
than it is for a spherical one.
23.11 Bubble Growth and Detachment 889
FIGURE 23.9 The amount of wetting of a heated surface is a function of the contact angle β. As the size of β is decreased, a
surface becomes better wetted.
Furthermore, these equations imply that the temperature of the vapor inside of the bubble must be greater than the
temperature of the liquid surrounding the bubble for bubble growth to occur. When a bubble is in thermal equilibrium
with the surrounding fluid, it can no longer grow or shrink, and the force balance implied by Equations 23.13 and
23.14 must continue to be maintained. If heat is added to the bubble, then PIN will increase, and σ must increase for a
given value of the radius R to keep the bubble stable. In some textbooks, this statement of the mechanical equilibrium
of a bubble is called Laplace’s law. In other words, Laplace’s law states that the tendency of a bubble to minimize its
surface tension is what causes it to become a sphere. Once the size of the bubble becomes larger than the cavity in
which it is born, a force balance can be performed to determine if the bubble stays in the cavity or detaches from it.
However, before discussing this force balance, we would first like to discuss the relationship between surface wetting
and surface tension.
23.10.1
Surface Wetting and Surface Tension
After a bubble forms, the heated surface to which it is attached can be characterized as either
1.
fully wetted
2.
partially wetted
3.
poorly unwetted.
Each of these conditions is shown in Figure 23.9. Once a surface becomes poorly wetted, a bubble will detach from the
surface. The angle of inclination between the edge of the bubble and the surface is called its contact angle. Sometimes
it is also called its wetting angle. For well-wetted surfaces (see Figure 23.9), the contact angle β is always less than 90°.
A partially wetted surface in one where the contact angle is 90°, and a poorly wetted surface is one where the contact
angle is greater than 90°. Examples these surfaces are shown in Figure 23.9. When the surface becomes well wetted,
the bubble has almost enough internal energy to detach from the surface. It eventually detaches when the contact angle
approaches 5°. At this point, the surface tension at the liquid–vapor interface is about the same as the surface tension in
the vapor–surface interface. The degree of wetting then depends on how far the contact angle deviates from this value.
23.11
Bubble Growth and Detachment
Before a bubble leaves a surface, the rate of bubble growth depends on the contact angle between the bubble and the sur-
face. It is easy to show that the maximum rate of bubble formation occurs when the surface is poorly wetted (e.g., when
β > 90°) because this is when the heat transfer rate between the surface and the bubble is the highest. However, poorly
wetted surfaces can sometimes create an undesirable side effect because if the bubbles do not detach from the surface,
they can totally immerse the surface in a vapor blanket or vapor barrier which causes the convective heat transfer coef-
ficient to fall dramatically. In a reactor core, this can also cause the fuel rods to dry out and fail. Normally, the fuel rods
are designed so that their surfaces are well wetted or partially wetted (i.e., β ≤ 90°). This implies that bubbles in the core
detach from these surfaces rather quickly, and this inhibits the formation of a surface film. Thus, a desirable surface for
nuclear fuel rods is one that is smooth and in which most (if not all) of the nucleation sites are small.
890 Fundamentals of Two-Phase Flow in Nuclear Power Plants
23.12
Bubble Formation and Superheat
As we just mentioned, the liquid film adjacent to a hot surface must be at least partially superheated in order for bubbles
to form. Otherwise, the vapor will not have enough energy to overcome the external pressure exerted on it by the cooler
liquid. The amount of superheat can be very high in nuclear power plants, which are classic examples of what are called
clean systems. In this context, a clean system is defined as one where
1. there are no dissolved gases in the liquid
2. the heated surfaces are smooth and devoid of nucleation sites.
In water reactors, water chemistry in the NSSS is closely monitored to keep the water “pure” as possible, and thus, it
is rare for any significant contaminates to reach the surface of the fuel rods under these conditions. An expression can
be derived for the amount of superheat required for a bubble to grow at the mouth of a conical cavity.* The amount of
superheat in this case is
where hlv is the enthalpy of vaporization. Equation 23.15 is also called the Young–Laplace equation for the mechani-
cal equilibrium of a stationary bubble. An example of a conical cavity to which Equation 23.15 applies is shown in
Figure 23.10. Here TBUBBLE is the vapor temperature at which the bubble forms, and TSAT is the saturation temperature
of the liquid surrounding it, P is the pressure of the surrounding liquid, υ is the specific volume of the vapor (which is
also used in the Clausius–Clapeyron equation, see Chapter 7), σ is the surface tension of the liquid (in N/m), hlv is the
enthalpy of vaporization (in J/kg), and R is a characteristic radius that depends on how completely the surface is wetted.
For contact angles up to 90° (β ≤ 90°), R is equal to the radius of the cavity, and for β > 90°, it is equal to the radius of the
FIGURE 23.10 The size of a cavity or crevice determines how quickly a bubble can form and detach from a hot surface.
In this case, bubble growth is illustrated for a conical cavity.
cavity divided by the sine of the contact angle (i.e., R = R/sin β). Thus, Equation 23.15 implies that as the contact angle
gets larger, the surface becomes less wetted (see Figure 23.9), and the amount of surface superheat required for bubble
formation to occur becomes smaller because the contact area increases. Hence, the maximum rate of bubble formation
occurs on poorly wetted surfaces (e.g., those when β > 90°). Example Problem 23.1 calculates the radius of a stable
bubble on a PWR fuel rod under these conditions.
Thus, a small amount of surface superheat must always be present before the bubble is able to leave the surface.
Otherwise, the bubble collapses into the surrounding liquid because the liquid has a higher ambient pressure. If the
bubble and the liquid are at the same temperature, then the liquid must be slightly superheated because the vapor inside
of the bubble already is. Normally, a temperature difference between one-tenth and one-hundredth of one degree Celsius
is required to keep the bubble from collapsing. Finally, the amount of the superheat ΔT = (TBUBBLE − TSAT) required to
sustain bubble growth is inversely proportional to the bubble diameter, DBUBBLE:
∆T ∝ 1 D BUBBLE (23.16)
Equation 23.16 also implies that the smaller the bubble is, the greater the degree of superheat ΔT must be in order to
sustain the nucleation process. Therefore, if one wants to initiate bubble formation near the saturation temperature,
one needs relatively large bubbles (on the order of several microns) in order for them to grow. Fortunately, the heated
surfaces in most nuclear power plants (including the surface of the nuclear fuel rods) happen to have this feature.
Finally, the surface temperature must be slightly above the saturation temperature for any significant amount of
nucleate boiling to occur. An additional discussion of the bubble nucleation process is provided in the following
references
☉☉ Boiling, Condensation and Gas-Liquid Flow by P.B. Whalley, Oxford Science Publications, 1987.
☉☉ Heat Transfer by J.P. Holman, McGraw-Hill Company, 1997.
Other discussions of this subject can be found in the references at the end of the chapter.
23.13
D efining the Equilibrium Void Fraction and Quality
After a bubble detaches fromthe surface, it enters what is called the flow stream. In the flow stream, it consumes space,
and some of the mass of the flow stream then consists of superheated vapor. The exact liquid to vapor ratio depends on
the temperature of the flow stream, and it also depends on what the vapor temperature inside of the bubbles is relative
to the temperature of the surrounding fluid. This vapor displaces some of the surrounding fluid and causes the average
density of the mixture to fall. In turn, this affects the power generation rate. In reactor work, two-phase flow can be
described by two important parameters called the void fraction α and the quality x. Their exact definitions depend on
whether the liquid and the vapor are moving at the same speed or at different speeds. In general, the vapor moves faster
than the liquid.
892 Fundamentals of Two-Phase Flow in Nuclear Power Plants
23.14
T he Void Fraction and the Equilibrium Quality for Homogeneous Mixtures
When a flow channel is vertical and the Reynolds number is high, it is usually possible to assume that the vapor is
moving at the same speed as the liquid. This can occur when
☉☉ there are only a couple of bubbles in the flow stream
☉☉ the flow is vertical and the convection is forced
☉☉ the coolant velocity is greater than the vapor velocity that would be possible when the vapor velocity is driven by
buoyancy forces alone.
Then the liquid and vapor phases move upward together, and they appear to be stationary with respect to each other in
the flow stream. When this occurs, we can define the static quality of this two-phase mixture as
where mv is the mass of the vapor, and ml is the mass of the liquid. Similarly, we can define the static void fraction as
where Vv is the volume occupied by the vapor, and Vl is the volume occupied by the liquid. The previous definitions
also allow us to write
The volume fraction of the vapor + the volume fraction of the liquid = α + (1 − α) = 1.0
The mass fraction of the vapor + the mass fraction of the liquid = x + (1 − x ) = 1.0
According to these definitions, the void fraction and the quality can never be greater than unity or less than zero:
Now consider the relationship between the void fraction and the quality in a coolant channel with cross-sectional area
A. Here Figure 23.11 can be used to facilitate our discussion. At any point in the coolant c hannel, the cross-sectional
area occupied by the liquid phase is Al, and the cross-sectional area occupied by the vapor phase is Av. The volume of
space occupied by both phases is then
V = ( A l + A v ) ∆z = A ⋅ ∆z (23.20)
Now let us see if we can use these relationships to express the static quality as a function of the static void fraction.
23.15 Relating the Quality and the Void Fraction for a Homogeneous Mixture 893
FIGURE 23.11 A reactor coolant channel where the liquid and vapor phases are moving at the same speed.
23.15
Relating the Quality and the Void Fraction for a Homogeneous Mixture
The static void fraction is
α = Vv ( Vv + Vl ) = A v ( A v + A l ) (23.23)
since Δz cancels from the numerator and the denominator. The static quality x (which is based on the mass ratios) is then
x = m v ( m v + m l ) = ρ v Vv ( ρ v Vv + ρ l Vl ) (23.24)
or
x = ρv A v ( ρv A v + ρl A l ) (23.25)
Equation 23.25 can be used to derive an explicit relationship between the static quality and the static void fraction by
dividing the numerator and the denominator by ρvAv. The result is
( )
x = 1 1 + ρl A l ρv A v (23.26)
( )
x(α) = 1 1 + ρl ρv × (1 − α)/α) (23.27a)
This relationship applies to non-flow systems and to flowing systems where the two phases move at the exactly same
speed. In other words, it is applicable to two-phase systems where the slip ratio is unity. We can also invert this equation
and solve for α as a function of x:
( )
α(x) = 1 1 + υ l υ v ⋅ (1 − x)/x (23.27b)
894 Fundamentals of Two-Phase Flow in Nuclear Power Plants
FIGURE 23.12 The effect of water pressure on the relationship between α and x for a nonslip system with S = 1.
Since the ratio of the density of the liquid to the density of the vapor is pressure dependent, it follows from Equation
23.27 that both x and α depend on the system pressure as well. Figure 23.12 shows the void fraction as a function of
the equilibrium quality at different system pressures. Note in particular how strongly the values of α and x depend on
the ratio of ρl to ρv, which is often called the density ratio. Close to atmospheric pressure (at approximately 14.7 PSI or
0.1 MPa), a small value of the quality (about 2%) corresponds to a very large value of the void fraction (about 97%). This
behavior is possible because the ratio of the liquid density to the vapor density is about 1,600 to 1 (ρl = 995.6 kg/m3 vs.
ρv = 0.60 kg/m3 at 14.7 PSI and 30°C). However, when the system pressure is increased to the pressure in the primary
loop of a BWR (which is about 1,050 PSI or 7.25 MPa), the ratio of α to x falls to about 22 (ρl = 802 kg/m3 vs. ρv =
35.9 kg/m3 at 1,000 PSI and 250°C), and when the pressure reaches 2,250 PSI or 15.5 MPa (which is the pressure in
the primary loop of a PWR), the value of α/x falls to about 6.9 (ρl = 679.25 kg/m3 vs. ρv = 98.45 kg/m3 at 2,250 PSI and
320°C). Thus, it is important to realize that the relationship between the quality and the void fraction depends heavily
on the ambient pressure as well as the reactor through which the coolant is flowing. Normally high pressures result in
lower values of the void fraction for the same value of the equilibrium quality. Now let us compare the behavior of water
to that of metallic coolants.
23.16
Comparing the Vapor Generation Rates for Water and Liquid Metals
In reactor coolant channels, the value of the void fraction depends on the liquid to vapor density ratio ρl /ρ v, which is
in turn a function of the ambient pressure. When the system pressure is raised, the value of ρ l /ρ v becomes smaller,
and the vapor does not expand as explosively as it does when the pressure is low. Then if the coolant is close to
atmospheric pressure, it can flash in an explosive manner if a phase transition occurs. Boiling results in a large vapor
to liquid ratio (about 1,600 to 1 for saturated water at atmospheric pressure. However, for liquid metal fast breeder
23.17 Defining the Slip Ratio 895
FIGURE 23.13 Pressure waves which are sometimes called shock waves can be created in an LMFBR fuel assembly when the coolant
boils. This is because of the extreme density difference between the liquid and vapor phases, which can approach 40,000 in some cases.
Shock wave propagation in LMFBRs is mitigated by the surrounding each fuel assembly with a metal can or “sheath.” This sheath
directs the shock waves downward and upward rather than laterally, and it also serves to protect the structural integrity of the core.
Normally the sodium in LMFBR cores is kept about 300°C below the saturation temperature to prevent shock waves from developing.
reactors (LMFBRs) cooled with liquid sodium, the same ratio is about 40,000 to 1 at atmospheric pressure and
~800°C). See
http://www.thermalfluidscentral.org/encyclopedia/index.php/Thermophysical_Properties:_Sodium
for a more detailed discussion of how liquid metals perform under these conditions. Thus, two-phase systems exhibit
much more explosive vapor generation rates when the operating pressure is low than when the operating pressure is
high. Because of this, boiling must always be avoided in reactors cooled with liquid metals because they are designed to
operate at very low pressures relative to PWRs and BWRs. Hence, if even a small amount of liquid sodium was allowed
to boil, a shock wave would develop that could damage the core as well its surrounding structural components. An
example of such a shock wave is shown in Figure 23.13. To limit the propagation of these waves, LMFBR fuel assem-
blies are surrounded by thin metal cans or sheaths. These “cans” force the shock waves to propagate vertically rather
than horizontally. BWRs also surround their fuel assemblies with cans or sheaths, but the density ratio in this case is
about 200 times lower than it is for sodium, so the shock waves are much less likely to damage the core. Normally, the
operating temperature in the primary loop of an LMFBR is kept about 200°C below the boiling point to keep the sodium
from flashing explosively. Conversely, in BWR fuel assemblies, controlled boiling is encouraged as long as there is a low
density difference between the phases. As we briefly mentioned in Chapter 5, the power level of the core depends on the
average coolant density. In water reactors, lower densities cause the power level to fall, and higher densities cause the
power density to rise. The reasons for this are discussed in our companion book.*
23.17
D efining the Slip Ratio
In two-phase flows, the liquid and vapor phases rarely (if ever) move at exactly the same speed or in exactly the same
direction. Usually the vapor slides past the liquid at a higher velocity even though it appears to be entrained within it.
When this occurs, the vapor is said to be “slipping past the liquid.” The ratio of the vapor velocity to the liquid velocity
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
896 Fundamentals of Two-Phase Flow in Nuclear Power Plants
is then used to define another important parameter of reactor fluid mechanics called the slip ratio S. Thus, the slip ratio
S is defined by
and just like the density ratio, it is a dimensionless number. In practice, the slip ratio is always greater than or equal to
unity, and it never less than this:
However, that this definition assumes the phases are moving in the same direction. If the phases are moving in opposite
directions, then the slip ratio could be either positive or negative (since the signs of the phase velocities could be differ-
ent). Then the relative velocities of the phases would have to be used instead. However, this is usually not the way the
slip ratio is implemented in practice, and a two-fluid model, which we will discuss in later chapters, would have to be
used to resolve this apparent inconsistency. At normal flow rates, slip ratios in PWR fuel assemblies are very close to
1.0, but in BWR fuel assemblies, they can sometimes exceed 4. The slip ratios near the top of most BWR cores are in
this range. Now let us show how the equilibrium void fraction and quality are related when the slip ratio is greater than
unity (S > 1.0), and both phases move in the same direction.
23.18
T he Void Fraction and the Quality with Interfacial Slip
When the liquid and vapor phases move at different speeds, the void fraction and the quality are no longer related by
Equation 23.27. Instead, the static quality must be replaced by another parameter called the flow quality in which the
effects of slip are taken into account. The flow quality is sometimes written as
v (m
x=m l ) = A v ρv v v ( A v ρv v v + A l ρl v l ) (23.31)
v +m
{ ( )( )( )}
x = 1 1 + ρl ρv ⋅ A l A v ⋅ v l v v (23.32)
Since (Al/Av) = (1 − α)/α, Equation 23.32 leads to the following relationship between the quality and the void fraction:
The Quality as a Function of the Void Fraction and the Slip Ratio
x(α) = 1 1 + ( ρl /ρv ) ⋅ (1/S)(1 − α)/α (23.33a)
This relationship can then be inverted to solve for the void fraction as a function of the flow quality:
The Void Fraction as a Function of the Quality and the Slip Ratio
( )
α(x) = 1 1 + ρv ρl ⋅ ( (1 − x)/x ⋅ S) (23.33b)
23.18 The Void Fraction and the Quality with Interfacial Slip 897
Normally, the void fraction α is found by performing an experiment, and the quality x is deduced from it for specific values
of vv and vl. The resulting expression is then called a “S–α–x” relationship. Notice that the relationship between α and x
depends on the amount of slip. In general, increasing the slip decreases the void fraction for the same value of the equilib-
rium quality. Figure 23.14a shows the void fraction in a BWR core as the slip ratio is changed. When temporal fluctuations in
the value of α can be ignored, the equations we have just derived can be used to establish the following relationship between
the slip ratio, the flow quality, the density ratio, and the void fraction for the flow of one-dimensional homogeneous mixtures:
The Slip Ratio as a Function of the Void Fraction and the Quality
( )
S = x (1 − x) ⋅ ρl ρv ⋅ (1 − α) α (23.34)
The slip ratio rarely exceeds 4 in a BWR core, so most fuel assemblies operate with values of S between 1 and 4. In
BWRs, S increases more rapidly near the center of the core, and then increases more slowly as one nears near the top.
Good experimental data covering some of these operating ranges is hard to find. In a fuel assembly, the geometry of the
coolant channel, the pitch to diameter ratio (P/D), and the axial heat flux also affect the relationship between α and x.
In general, no single correlation is able to cover all possible operating conditions, and increasing the complexity of the
correlation does not necessarily lead to better results. This has caused some engineers to suggest that a drift flux model
would be more appropriate to describe the relative motion of the liquid and vapor phases. We would now like to describe
the equations the drift flux model uses and explore their underlying assumptions. Unfortunately, a comprehensive dis-
cussion of the drift flux model is beyond the scope of this book; thus, the reader is referred to
http://ltcm.epfl.ch/files/content/sites/ltcm/files/shared/import/migration/COURSES/
TwoPhaseFlowsAndHeatTransfer/lectures/Chapter_17.pdf
for a more comprehensive discussion of its features.
{ ( ) }
α PWR = 1 1 + ρv ρ l ⋅ (1 − x)/x = 1 {1 + (1/6.9) ⋅ (0.99/0.01)} = 1/15.35 = 6.5%
and
{ ( ) }
α BWR = 1 1 + ρv ρ l ⋅ (1 − x)/x = 1 {1 + (1/23) ⋅ (0.85/0.15)} = 1/1.25 = 79.5%
{ ( ) }
α PWR = 1 1 + ρv ρ l ⋅ (1 − x)/x = 1 {1 + (1/6.9) ⋅ (0.99/0.01) ⋅ 2} = 1/29.7 = 3.3%
and
{ ( ) }
α BWR = 1 1 + ρv ρl ⋅ (1 − x)/x = 1 {1 + (1/23) ⋅ (0.85/0.15) ⋅ 2} = 1/1.51 = 66.2%
Thus, increasing the slip ratio S reduces the value of the void fraction for a given value of the equilibrium quality.
This reduction can be sometimes significant because of the effect that it can have on the reactivity of the core. This
relationship is discussed in more detail in our companion book.* [Ans.]
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
898 Fundamentals of Two-Phase Flow in Nuclear Power Plants
FIGURE 23.14 (a) Effect of the slip ratio S on α and x for pressurized water at 7 MPa. Flows in which the vapor phase slides
past the liquid phase at high speeds are said to have high slip, and flows at which the vapor phase slides past the liquid phase at
low speeds are said to have low slip. When both phases move at the same speed, and have no slip, and the slip ratio corresponding
to equal liquid and vapor velocities is 1.0. In general, increasing the slip ratio decreases the value of the void fraction when the
fluid quality is fixed. This effect is particularly pronounced in BWRs. (b) Predictions of the HEM model compared to those of
the Premoli and Chexal–Lellouche correlations as a function of the mass flux G at a pressure of ~1,050 PSI (7.25 MPa). In this case,
the predictions of the HEM model are mass flux independent.
23.19 UNDERSTANDING A DRIFT FLUX MODEL 899
23.19
Understanding a Drift Flux Model
Sometimes the slip ratio S cannot be used to establish a precise relationship between α and x—particularly when the mass
flow rate is low. Under these conditions, a drift flux model can be used to establish a better correlation between α and x.
The basic idea behind a drift flux model is to express the local vapor velocity vv as the algebraic sum of the average volu-
metric velocity of the two-phase mixture v, and a local drift velocity for the vapor phase which we will call vdrift. Hence,
This simple prescription has a number of practical advantages that will soon become apparent. It allows us to write the
volumetric velocity of the vapor vv as
v v = αv v = αv + α ( v v – v ) (23.36)
The second term on the right-hand side of this equation is called the drift flux.
Physically, the drift flux represents the rate at which the vapor flows through a unit area (perpendicular to the axis of the
channel) where the mixture is already travelling with an average velocity v. It is possible to derive a much more accurate
expression for the slip ratio under these conditions. While we will not go through all the details of how this is done, it is
possible to show that the slip ratio in a complex flow field consists of two separate terms:
☉☉ An initial term due to the local velocity differences between the liquid and vapor phases
☉☉ A second term due to a nonuniform distribution of voids in the flow channel
The first term represents the slip ratio we previously defined, except that it is expressed in terms of the effective drift
velocity ved of the vapor phase
v ed = α ( v v – v ) α (23.38)
Term 1
ρl v ed (1 – x)ρ⋅ v (23.39)
Term 2
C + (C – 1)·xρl (1 – x)ρv (23.40)
where C is an empirically determined constant. Thus, a more accurate expression for the slip ratio is the sum of these
two terms:
If either the void profile or the velocity profile is uniform across the entire channel, then it can be shown that C = 1, and
the slip ratio reduces to the conventional definitions we presented earlier. The values of ved and C are also a function of
900 Fundamentals of Two-Phase Flow in Nuclear Power Plants
TABLE 23.2
Values of n and the Bubble Rise Terminal Velocity VBUBBLE for Different Flow Regimes Encountered in Reactor Coolant Channels
the flow regime. C will have a value of 0 for very low void fractions, and a value of 1.0 for very high void fractions. For
intermediate states (like bubbly and slug flow), C has a value of about 1.2. The advantage to this approach is that we
can correlate the value of ved to a specific flow regime. An enormous amount of work has gone into understanding how
ved behaves for bubbly and slug flows, which frequently appear in BWR cores. This usually requires the effective drift
velocity to be expressed as
where VBUBBLE is the bubble rise terminal velocity in the liquid. For most problems, n has a value of 0–3. The most
commonly accepted values for VBUBBLE and n are shown in Table 23.2. Notice that they are flow regime dependent. In
slug flow and churn flow, n = 0, and so ved and VBUBBLE are the same. However, for bubbly flow, they are different. In the
1970s, drift flux correlations were extended to handle annular flow. However, in annular flow, there is little difference
between the volumetric flow rate of the vapor and the total volumetric flow rate, and so the effective drift velocity ved
is essentially 0. Then, the slip ratio reduces to S = C + (C − 1)·xρl/((1 − x)·ρv). Slip ratios in the annular flow regime are
usually much larger than they are for bubbly flow.
23.20
Comparing Conventional Correlations for the Void
Fraction to Drift Flux Correlations
If we actually go through the process of calculating the void fraction as a function of the flow quality, we get different
results depending upon what correlations we use. The standard equation for the void fraction as a function of the quality:
{ ( ) }
α(x) = 1 1 + ρv ρl ⋅ ( (1 − x)/x) ⋅ S) (23.43)
does not depend on the mass flux G = ρv in any way. However, the second term in the drift flux model does depend on the
mass flux, and this means that the relationship between void fraction and the quality is not as simple as Equation 23.43
implies. Consider for the moment a simple two-phase mixture in which the steam and the water are flowing vertically through
a smooth 20 mm circular pipe. At 1,050 PSI (7.25 MPa), and approximately 300°C, we can compare the void fraction versus
quality relationship for the standard slip model (sometimes called the homogeneous void fraction model) with the more
accurate drift flux model. For mass flow rates of 100, 1,000, and 3,000 kg/m2s, the results are shown in Table 23.3. Notice that
the two models do not compare particularly well. However, we can conclude a couple of things from our initial observations:
1. The standard slip ratio model is good for high pressure, high flow conditions.
2. The drift flux model is better for low pressure, low flow conditions.
TABLE 23.3
The Void Fraction α for Different Values of the Equilibrium Quality and Mass Flux at Approximately 1,050 PSI
The Void Fraction α for Different Values of the Quality and the Mass Flux
Model Mass Flux (kg/m2 s) x = 10% x = 30% x = 50% x = 70% x = 90%
Homogeneous Any mass flux—the model is 0.68 0.89 0.95 0.98 0.99
mass flux independent
Drift flux 3,000 0.56 0.74 0.79 0.81 0.83
Drift flux 1,000 0.54 0.73 0.78 0.81 0.82
Drift flux 100 0.40 0.63 0.71 0.75 0.76
23.21 Two-Phase Flow Regimes in a Nuclear Power Plant 901
In the 1990s, a more sophisticated correlation was developed by the Electric Power Research Institute (or EPRI) to over-
come these inconsistencies. The resulting model became known as the Chexal–Lellouche void fraction model. Today
this model is probably the most accurate way to predict the relationship between void fraction and the quality in a BWR
core. It takes into account the effects of the system pressure and the mass flow rate. In addition, it has the lowest average
error of any correlation in widespread use today.
However, it is hard to use, and it can also be difficult to apply. The reader is referred to the references at the end of
the chapter for more information concerning the specific form of this correlation (see for example Todreas and 2008).
Another void fraction versus quality correlation called the Permoli correlation provides similar results, but it has the
advantage that it is much easier to use. There is a great deal to like about the Chexal and Lellouche correlation. Its most
desirable characteristic is that it is applicable to all flow regimes. It is also applicable to a variety of different sub channel
geometries (vertical, horizontal, and inclined), and even to different flow directions (which represent the liquid and the
vapor flowing in the same direction, in different directions, and even situations where the liquid does not flow at all). It
can be used at many different system pressures and mass flow rates. Finally, it is relatively independent of the idiosyn-
cratic behavior of the fluid and the vapor within each of these flow regimes. It always predicts a void fraction that has a
physically realistic value (between 0 and 1.0). Its main drawback is that it requires an iterative approach to implement
it because some of the parameters depend on the void fraction as well. Thus, it is a set of implicit equations that relate
the void fraction to the quality. The results of the Permoli and Chexal–Lellouche correlations are compared to those of
a homogeneous equilibrium model (or HEM correlation) in Figure 23.14b. Note that both correlations predict lower void
fractions than a HEM model does.
23.21
Two-Phase Flow Regimes in a Nuclear Power Plant
The exact orientation of the liquid and the vapor phases in a flow channel depends on the pressure, the mass flux, and the
heat flux. The easiest way to understand these dependencies is to subdivide the flow into a number of different regions or
flow regimes. Strictly speaking, a flow regime represents a flow pattern where the liquid and the vapor phases behave in a
predictable way. When the coolant pumps are running, the flow fields in PWRs and BWRs can be further subdivided into
nine separate flow regimes with different hydrodynamic properties. These flow regimes are summarized in Table 23.4. The
best way to understand each regime is to examine the behavior of the coolant as it flows through the core. A reasonably
complete picture of how the liquid and the vapor phases behave is shown in Figure 23.15. The fuel rods are assumed to be
heated, and the flow is assumed to be forced. The flow is also assumed to be vertical. Thermal energy is continually added
to the coolant, and this is one of the factors that determines how the flow behaves. Normally, one flow regime transitions
into another when the heat flux is changed, or when the mass flux is raised or lowered. For example, subcooled boiling
can easily become saturated boiling, and saturated boiling can immediately become bulk boiling. The transition points
between these flow regimes are both mass flux and pressure dependent. They depend on the saturation temperature as well
as the coolant channel geometry. However, for a circular coolant channel, these regimes are relatively simple to under-
stand. A graphical depiction of how these flow regimes interact is called a flow regime map, and many reactors designers
use these maps to understand how the flow behaves as a function of the operating conditions. A flow regime map that a
reactor designer can use is shown in Figure 23.16. In almost all power reactors, the coolant enters the core from below as
TABLE 23.4
Representative Flow Regimes in the Cores of Thermal Water Reactors
FIGURE 23.15 The flow regimes in a heated channel with a uniform heat flux. Notice that nucleate boiling can begin almost
immediately in the channel if the degree of subcooling is low.
FIGURE 23.16 A typical flow regime map in which the regimes are plotted as a function of the value of the
liquid and vapor velocities.
23.21 Two-Phase Flow Regimes in a Nuclear Power Plant 903
a subcooled liquid and exits the core from above. In BWRs, it then leaves the core as a two-phase mixture. In most fuel
assemblies, the velocities of the liquid and vapor phases are different, and there is a considerable amount of slip. Thus, the
slip ratio and the heat transfer coefficients become flow regime dependent. We would now like to discuss these flow regimes
and illustrate how one flow regime can transition into another. We will begin with a discussion of subcooled liquid flow, and
we will then extend our discussion to each of the flow regimes that can be found as we move further up a heated channel.
23.21.1
Regime I: Subcooled Liquid Flow
When a coolant enters the inlet of a reactor fuel assembly, it is normally a subcooled liquid. The coolant is forced
through the core at a speed of several meters per second to optimize the heat transfer rate. In American PWRs, this is
accomplished by pumping the coolant through the core at between 5 and 6 m/s during normal operation. Since there are
no bubbles when the coolant is subcooled, it is a pure liquid, and its velocity profile is relatively uniform in the radial
direction. The Reynolds number in most fuel assemblies is about 500,000, and the flow is highly turbulent. A boundary
layer develops between the turbulent core and the surface of the rods, but at such high Reynolds numbers this layer is
very thin, and can be only a couple of molecules thick. The temperature of the coolant stays below the saturation temper-
ature until it progresses further into the core. The saturation temperature for a PWR core is about 345°C, and for a BWR
(which operates at about half the pressure), it is about 286°C. There are turbulent eddies in the flow field that enhance
the heat transfer rate. The value of the convective heat transfer coefficient in a typical fuel assembly is about 40,000 W/
m2 °C, and the convective heat transfer coefficient in fast breeder reactor fuel assemblies can approach 100,000 W/m2 °C.
23.21.2
Regime II: Bubbly Flow with Subcooled Boiling
Once the temperature of the cladding reaches the saturation temperature, small bubbles begin to form on the surface of
the rods. These bubbles detach from the surface and flow into the turbulent core where the bulk fluid temperature is at
least several degrees below the saturation temperature (TBULK < TSAT). Any bubbles that form in this way are immedi-
ately engulfed by the cooler liquid, and they collapse back into the flow stream. The creation of vapor bubbles at the wall
surface is called nucleate boiling and these bubbles do not merge together to form larger bubbles or voids. Hence, the
bulk fluid temperature stays below the saturation temperature, and the two-phase mixture tends to be highly turbulent.
In the subcooled boiling regime, the void fraction is less than 5%, and the fluid quality is typically below 0.5%. The
vapor does not take up much space relative to the liquid. Because of the enhanced turbulent mixing that nucleate boiling
provides, the convective heat transfer coefficient is 3–5 times larger than it is for the subcooled liquid regime (Regime I).
23.21.3
Regime III: Bubbly Flow with Saturated Boiling
Bubbly flow with saturated boiling is similar to bubbly flow with subcooled boiling except that the average temperature
of the fluid has now reached the saturation temperature. When this occurs, the bubbles that detach from the surface of
the cladding do not immediately collapse when they enter the flow stream. Instead, they either remain intact or combine
with other bubbles to form larger bubbles or voids. In BWRs, the equilibrium quality approaches 2%–3%, and the void
fraction approaches 20%–30%. The interior of the flow channel is now about 1/3 full of vapor, but there are still spaces
between the larger bubbles, and they do not coalesce into a single bubble that spans the entire width of the channel. In
BWRs, this happens slightly above the core midplane. Because bubbly flow enhances the turbulent mixing rate, the heat
transfer coefficient is about twice as high as it is in Regime II. Thus, the heat transfer coefficients in this regime are 6–10
times higher than they are for Regime I. There is still a liquid film on the surface of the fuel rods, and this film helps to
remove heat from the rods. The fuel rods do not dry out or melt because the film is stable.
23.21.4
Regime IV: Saturated or Bulk Boiling
Saturated or bulk boiling is similar to bubbly flow, except that the void fraction now rises to 40% to 60%. In BWRs, the
flow quality is relatively low (about 1/12th of this value), and the bubbles that form in the center of the channel can now
span the entire width. However, they do not do so on a permanent basis. The heat transfer coefficient increases slightly,
but the rate of increase (the first derivative of h vs. x) starts to decline. The boiling is stable, and does not oscillate back
and forth between flow regimes unless additional heat is added to or subtracted from the flow.
23.21.5
Regime V: Slug Flow
In the slug flow regime, the bubbles that have already formed coalesce into very large bubbles that not only span the
entire width of the channel but also remain long and continuous. In reactors, these vapor bubbles can be 4–6 times
longer than they are wide. In other words, they can have an aspect ratio of about 5 to 1. They look like slugs of vapor
904 Fundamentals of Two-Phase Flow in Nuclear Power Plants
separated vertically by small columns of liquid. There is still a thin film of liquid on the surface of the cladding, but this
film is much thinner than it is for bulk boiling (about 1–2 mm thick). The heat transfer coefficient is not much different
than it is for bulk boiling, but it no longer increases dramatically as the void fraction or the flow quality is raised. The
liquid film on the surface of the rods cannot be agitated any more because there is no additional turbulent mixing to
increase the heat transfer rate. In most BWRs, the fluid quality is now 8%–10%, and the void fraction is approximately
60%–70%. The voids now blanket the entire center of the coolant channels.
23.21.6
Regime VI: Churn Flow
In the churn flow regime, vertical columns of water separating the individual vapor slugs begin to break down. There is a
vapor core throughout the entire channel. However, the width of this vapor core is not exactly the same at all locations because
the water left over from the slugs still clings to the surface of the rods. Hence, this is one of the primary ways that churn flow
is distinguished from slug flow. In some books, slug flow and churn flow are simply combined together into an extended
slug flow regime, but in reality, there is enough of a difference in the void fraction and the quality to make this distinction—
particularly when these voids affect the design of the control system. The other major difference is that the liquid film on the
surface of the cladding can be moving in the opposite direction as the vapor core because of the effects of gravity and surface
tension. This difference is illustrated in Figure 23.17, which shows how slug flow and churn flow behave. In BWRs, the fluid
quality is 8%–10%, and the void fraction can rise above 70%. In practice, this is about as far as the boiling in a BWR fuel
assembly is allowed to progress. However, it can be extended a bit further when the churn flow transitions itself into annular
film flow or annular flow for short. Transitioning from churn flow to annular flow requires the presence of additional heat.
23.21.7
Regime VII: Annular Flow
The primary difference between the annular flow regime and the churn flow regime is that the liquid film on the surface
of the fuel rods is now moving in the same direction as the vapor. This causes the “churn” in the flow pattern to nearly
disappear because the interfacial friction between the two phases is now lower (due to a reduction in the relative veloc-
ity difference between the phases). Sometimes this interfacial friction is called interfacial drag. The liquid film on the
surface of the rods is now very stable, but it is even thinner than it was before (about 1 mm thick on average). More of the
liquid on the surface of the fuel rods is now entrained in the vapor core. In commercial BWRs, the fluid quality is now
at its maximum value (10%–15%), and the void fraction is almost 80%. This flow regime is the flow regime that exists
when the coolant exits a BWR core.
FIGURE 23.17 A picture showing the differences between slug, churn, and annular flow in a reactor coolant channel.
23.23 Useful Relationships for Two-Phase Mixtures 905
23.21.8
Regime VIII: Mist or Drop Flow
Adding more heat to the coolant causes the annular film to evaporate entirely from the surface of the rods, and this causes
the fuel rods to dry out. Thus, the liquid film on the surface of the fuel rods disappears. The remaining flow consists of
hot droplets of liquid that are entrained in the vapor core. All that remains is a fine “mist” of these droplets in the center
of the coolant channel. This mist is like the mist one encounters when driving on a foggy day. However, the liquid drops
are much hotter, and they move at much greater speeds. This regime is called the mist flow regime or the drop flow regime.
The flow quality is now greater than 20%, and the void fraction is over 85%. The few liquid drops that remain completely
disappear if any more heat is added to the flow channel. However, BWRs generally do not operate in this regime because
the liquid film on the fuel rods disappears, and the heat transfer coefficient (for the same mass flux) falls dramatically by
a factor of 10 or more. Hence, allowing a coolant channel to enter this regime could potentially cause the fuel rods to fail.
23.21.9
Regime IX: Pure Vapor Flow
In this final flow regime, the liquid completely disappears and all that remains is pure vapor flow. Hence, this regime is
called the pure vapor flow regime. All of the complicated physics has now disappeared, and the heat transfer coefficients
behave in exactly the same way as they do for single-phase flow. Only in this case, the single phase is just the vapor
phase, and the correlations that we developed for single-phase liquid flow can be used again provided that the physical
properties of the fluid (ρ, μ, k) are the same as the properties of the vapor. The amount of liquid left in the vapor is a
function of the degree of superheat. If the superheat exceeds a few degrees centigrade, then the quality increases to over
99%; however, because the Reynolds number is so high, the flow is not necessarily an equilibrium flow where one can
simply set the quality to 100%. Essentially the same process is used to superheat the steam in a steam generator before it
enters the steam turbines. For obvious reasons, the coolant is not allowed to become dry when it flows through the core.
Adding additional heat to the flow stream superheats the vapor and increases its pressure. The more energy the vapor
possesses, the more mechanical work it can do when it reaches the blades of the steam turbines. Thus, superheating the
vapor increases the thermodynamic efficiency of the plant (see Chapter 9).
23.22
F low Regime Maps
Sometimes it is helpful to visualize the interactions between these flow regimes using what is called a flow regime map.
Different reactor coolants have slightly different flow regimes, but the general idea is to show how these flow regimes
are affected by the operating state of the system. A flow regime map is a graphical depiction of the state of a two-phase
mixture as a function of its velocity and its equilibrium quality x or void fraction α. Maps can also be constructed in
which the flow regimes are plotted as a function of the mass flux G and the heat flux q″. A map of this type is shown
in Figure 23.18. There are many different types of flow regime maps. Flow regime maps are similar to state diagrams
except that they do not contain the same state variables that P–V, P–T, or T–S diagrams do. Flow regime maps can also
be used to map a specific flow regime (i.e., slug flow, bubbly flow, or annular flow) to specific pressure, velocity, and heat
flux combinations. A different heat transfer coefficient and pressure drop correlation can then be assigned to each flow
regime. This mapping technique is sometimes used to assign the appropriate heat transfer coefficient and pressure drop
correlations to each flow regime. For example, reactor safety analysis programs such as RELAP and TRACE sometimes
use this technique to assign the appropriate friction factors and heat transfer coefficients to the energy and momentum
equations. We will have more to say about this in Chapter 24.
23.23
Useful Relationships for Two-Phase Mixtures
There are a number of other important relationships that are sometimes encountered in the study of two-phase flow.
These relationships can be used to inter-relate the mass flux G the quality, and the void fraction at a specific axial loca-
tion. These relationships do not require one to “guess” the slip ratio S to find the velocities of the liquid and vapor phases.
To see how this can be done, consider the conservation of mass for a two-phase mixture. Suppose that the total flow rate
= ρvA, where ρ, v, and A are the density, the velocity, and the flow area of the two-phase mixture. This must be
is m
equal to the sum of the mass flow rates of the liquid and vapor phases:
=m
m l +m
v (23.44)
ρvA = ρv v v A v + ρl v l A l (23.45)
906 Fundamentals of Two-Phase Flow in Nuclear Power Plants
FIGURE 23.18 Another flow regime map obtained from data for low pressure air–water mixtures and high pressure water–
steam mixtures in round adiabatic tubes having internal diameters between 1 and 3 cm (Adopted from Todreas and Kazimi
(see Lamarsh 1972) with modifications.)
Here area of the coolant channel A is equal to the sum of the flow areas of the liquid and the vapor phases:
A = A l + A v (23.46)
ρl v l A l = (1 – x)ρvA (23.47)
ρv v v A v = xρvA (23.48)
Dividing through by A = Al + Av, and realizing that α = Av/(Av + Al), it immediately follows that
This leads us to the following relationships between the behavior of the mixture and the behavior of the individual phases:
In other words, the mass flow rate of the liquid m l is just the mass flow rate of the mixture multiplied by (1 − x)/(1 − α),
and the mass flow rate of the vapor m v is just the mass flow rate of the mixture multiplied by (x/α). These relationships
allow us to understand how the individual phases behave as long as we can measure the values of x and α. Normally, x
and α are position dependent and they depend on how much heat is added to the coolant. The location where a particular
flow regime develops depends on the total flow rate m and the shape of the axial heat flux q″(z).
23.24
Boiling Cores
In light water reactors, it is important to know where boiling begins in the core. The location at which boiling begins has
great practical importance because it determines not only how hot the fuel rods become, but it also determines what the
total pressure drop ΔP across the core will be. In American PWRs and BWRs, the coolant enters the core from below
and exits the core from above. In PWRs, the coolant is intentionally designed to remain a liquid except near the top of the
core, where a small amount of nucleate boiling may be allowed. In BWRs, the coolant starts to boil vigorously near the
core mid-plane, and it continues to boil until the top of the core is reached. The differences between these two designs
are shown in Figure 23.19. When boiling occurs, it is important to be able to distinguish between regions of the core
where boiling occurs and regions where it does not. To do this, we would like to define two new terms—the boiling and
non-boiling heights. For typical cores, these heights are shown in Figure 23.19. The non-boiling height Ho refers to the
FIGURE 23.19 An illustration of the boiling and non-boiling heights in a BWR coolant channel. In American BWRs, the coolant
stays a subcooled liquid in the lower 25% of the core although some nucleate boiling can occur long before the equilibrium quality
reaches zero. In American PWRs, the equilibrium quality always stays below zero, even though a small amount of nucleate boiling
can occur near the top of the core.
908 Fundamentals of Two-Phase Flow in Nuclear Power Plants
height within the core where the coolant remains a subcooled liquid, and the boiling height HB refers to the height above
this where at least some of the coolant boils. This boiling can consist of
1. Nucleate boiling
2. Bulk boiling
3. Film boiling
By convention, the sum of the boiling and non-boiling heights is equal to the total active core height H:
The Total Core Height and the Boiling and Non-Boiling Heights
H = H o + H B (23.52)
Normally, the non-boiling height can be thought of as the length of a fuel assembly where the flow remains a single-
phase liquid, LSP, and the boiling height can be thought of as the length of a fuel assembly LTP (usually above the non-
boiling height) where two-phase flow occurs. In other words, we can write LSP + LTP = LTOTAL = H. The non-boiling
height Ho ends when the coolant reaches its saturation temperature TSAT. This occurs when z = Ho. The coolant must also
reach its saturation enthalpy hSAT at this particular location. Thus, a relatively simple way to predict the location of Ho is
to find out how much heat must be added to the coolant to reach its saturation enthalpy hSAT. This approach can be used
for both PWRs and BWRs. If the enthalpy of the fluid at the core inlet is hIN, then bulk boiling occurs when the amount
of heat q added to the coolant is
( h SAT − h IN ) (23.53)
q=m
where m is the total mass flow rate. Normally, hSAT and hIN are known in advance, so Ho can be found by integrating q′(z)
from the bottom of the core (where z = 0) to the wherever q = m(h SAT − hIN). Thus
Ho
∫ 0
( h SAT – h IN ) (23.54)
q′(z) dz = m
is how the value of Ho can be found. For the case of uniform heat addition, q′(z) = q′ and performing this integration gives
( h SAT – h IN ) (23.55)
q′ ⋅ H o = m
where q′ is the linear heat generation rate (see Chapter 11). In practice, the ratio of Ho to H for an arbitrary coolant
channel can also be found by taking the ratio of (hSAT − hIN) to (hOUT − hIN):
where hOUT is the enthalpy at the outlet of the core. We can also write
FIGURE 23.20 The variation of the void fraction α and the equilibrium quality x in a uniformly heated reactor c oolant channel
in an American BWR as a function of axial elevation. (From El-Wakil, M.M., Nuclear Heat Transport, American Nuclear Society,
La Grange Park, IL, 1981.)
( ( ))
( h SAT – h IN ) q′ (23.61)
H/π ⋅ 1 – cos πH o H = m
and the value of HB is given by HB = H − Ho. The total amount of energy that needs to be added for the coolant to boil is
simply the area under the curve for q′(z) from z = 0 to z = Ho. For other modes of heat addition where the shape of q′(z)
is known, a similar procedure can be used to find Ho and HB. For example, if the heat flux increases linearly along the
length of the channel, then q′(z) = q′·z. If the shape of q′(z) cannot be represented by a simple function, then a computer
program can be used to find the value of Ho. Another approach is to simply find the elevation z where the fluid reaches
its saturation temperature TSAT. Large values of the mass flow rate result in large values of Ho, and small values result in
small ones. The behavior of the equilibrium quality is shown in Figure 23.20 for a uniformly heated channel (q′(z) = q′).
Notice that the quality increases linearly from the time that the boiling first occurs. The maximum value of the quality
xOUT is reached at the core exit. The void fraction follows a similar trend, but its shape is not linear because α (x) = 1/{1 +
(ρv/ρl)·((1 − x)/x)·S)}. The shape is shown for three different values of the slip ratio (S = 1), (S = 2), and (S = 4). Clearly,
the value of S is important in determining the size of the void fraction. Figure 23.21 shows the same parameters when
the heat flux is sinusoidal.
Notice that neither the quality nor the void fraction is particularly uniform. This is much more typical of what the
void fraction and the quality look like as a function of axial elevation in a boiling core. In American BWRs, the value
of the exit quality xOUT is 15%–17% at the core exit, and this normally implies that the flow field has transitioned from
bulk boiling into annular flow. The core pressure drop can then be found using the values we have just provided for Ho
and HB with the appropriate adjustments for ρ, α, and x. The coolant is acted upon by several different forces as it moves,
and these forces manifest themselves as pressure drops that can be measured. The largest of these are the pressure drop
due to friction and the acceleration pressure drop caused by the acceleration of the fluid as it receives heat and boils. The
pressure can also change if the fluid encounters an obstruction such as a rod spacer, a grid spacer, or an orifice plate. In
boiling cores, a combination of these factors determines the total pressure drop. Of course, this methodology is based
on using the bulk temperature of the coolant. If the coolant at the surface of the cladding has a higher temperature than
the coolant in the center of each channel, then the coolant can boil much earlier than Equations 23.56 and 23.61 predict.
In fact, it can boil as much as 20 cm below these values in an American BWR in spite of the fact that the equilibrium
quality is still less than zero.
910 Fundamentals of Two-Phase Flow in Nuclear Power Plants
FIGURE 23.21 The value of the void fraction α and the equilibrium quality x in a typical American BWR as a function of
axial elevation. Note: The values may differ somewhat from one fuel assembly to the next. (From El-Wakil, M.M., Nuclear Heat
Transport, American Nuclear Society, La Grange Park, IL, 1981.)
23.25
F luid Friction for Two-Phase Flow
In boiling cores, the frictional pressure drop consists of two components—the first component for the lower portion of
the core (where the coolant is a single-phase liquid), and the second component for the upper portion of the core (where
the coolant is at least partially a vapor). Normally, these components are calculated separately and then added together
to find the total pressure drop:
In this equation, ΔPSP refers to the single-phase pressure drop, and ΔPTP refers to the two-phase pressure drop. Calculating
the magnitude of these pressure drops requires a knowledge of the boiling and non-boiling heights—HB and Ho. In gen-
eral, the non-boiling height ends when the bulk temperature becomes greater than or equal to the saturation temperature:
where 0 corresponds to the bottom of the core. Above the non-boiling height, the coolant begins to boil, and this boiling
can take the form of either nucleate boiling or bulk boiling. The boiling height HB then extends from z = TSAT to the top
of the core, where z = H.
23.27 Pressure Drops for Homogeneous Mixtures 911
The two-phase length or the boiling length is subsequently defined as the length of the coolant channel where the coolant
temperature is greater than or equal to the saturation temperature:
Sometimes the boiling length begins where nucleate boiling begins. However, this is usually only a couple of centime-
ters below the location where bulk boiling begins. Along the boiling length, the two-phase pressure drop is much greater
than the single-phase pressure drop. This pressure drop is much larger even when the boiling and non-boiling heights
have approximately the same length.
23.26
Finding the Pressure Drop for Combinations of Single-
Phase Flow and Two-Phase Flow
Ideally, it would be nice to use different friction factors for each flow regime - both single phase and two phase. This
would then mean that we would need at least half a dozen friction factor correlations to span the entire operating range
of a BWR core. The frictional pressure drop could then be found from
where the subscripts SP, NB, BB, SF, and AF refer to single phase or subcooled flow, nucleate boiling, bulk boiling, slug
flow, and annular film flow. While reactor vendors have sometimes used this approach to determine the core pressure drop,
it has become more popular in modern times to use just one or two correlations to span the entire operating range. Normally,
this involves lumping the values of ΔPNB, ΔPBB, ΔPSF, and ΔPAF together into a single two-phase pressure drop called ΔPTP:
where again ΔPSP and ΔPTP refer to the single-phase and two-phase pressure drops. When BWRs were first built, people
tried to apply the pressure drop correlations for single-phase flow to two-phase flow. They tried to apply the same friction
factors as well. They did so by using an average value for the density and the viscosity in addition to some good engineering
judgement. For obvious reasons, this approach had limited success until someone developed what was called the two-phase
friction multiplier. Then engineers discovered that they could obtain reasonably good estimates for the total pressure drop
when the fluid properties were evaluated at the saturation temperature TSAT. It is instructive to repeat the initial arguments
that led to this approach because a great deal of insight can be gleaned into the overall behavior of the core.
23.27
P ressure Drops for Homogeneous Mixtures
The easiest way to understand two-phase friction is to imagine that the liquid and the vapor phases move together, and
that the friction factor is not a function of the flow regime. Obviously, this is not very realistic in practice because there
can be many different flow regimes in a boiling core. However, if we can adjust the densities, the velocities, and the
viscosities to account for the fact that at least some of the flow is a liquid, and some of the flow is a vapor, then we can
912 Fundamentals of Two-Phase Flow in Nuclear Power Plants
use the correlations we developed in Chapter 17 to estimate the pressure drop for each region. The individual pressure
drops can then be summed together to find the total pressure drop ΔPTOTAL.
23.27.1
O ption 1
In some applications, all we need to do is to find the quality of the mixture as a function of axial elevation in order to
determine how much of the coolant is a liquid, and how much of the coolant is a vapor. Once we know the densities and
the viscosities of the liquid and the vapor phases (e.g., ρl, ρv, μl, μv), we can then define an effective density and an effec-
tive viscosity based on the average properties of the mixture at each axial location. This specific approach is called a
homogeneous approach to two-phase friction because only the average properties of the mixture are used to determine
the friction factor. The effective density and the effective viscosity are then
where x m is the local quality, and α m is the void fraction of the mixture. When the flow is homogeneous, we can
also write
1 ρm = x m ρv + (1 − x m ) ρ1 (23.70)
The frictional pressure drop ΔPTP can then be found from our old friend, the Darcy equation:
Here LTP is the total length of the flow channel where the coolant boils. This boiling includes nucleate boiling, bulk boil-
ing, and film boiling. The parameters used to determine the two-phase friction factor are simply the weighted average
of the values for the liquid and vapor phases:
ρm = α m ρv + (1 − α m ) ρl and µ m = x m µ v + (1 − x m ) µ l (23.72)
However, to make this process work, the mass flow rate m and the coolant velocity vm must be those of the mixture. For
the velocity used to calculate the Reynolds number, the choice is relatively straightforward:
v = x m v v + (1 − x m ) v l (23.73)
Moreover, when there is no slip (S = 1), this is the same as the velocity of the mixture. However, when calculating the
Reynolds number,
we must use
G=m
/A = ρv = xρv v v + (1 − x)ρl v l (23.75)
23.27.2
O ption 2
the density, and the viscosity at the inlet
An alternative approach to find the friction factor is to use the mass flow rate m,
and the outlet of each channel. Then, we can use the average properties of the coolant between the inlet and the outlet
because the quality is already included in the average. In this case, we can write
The expressions for the viscosity can be handled in the same way. The average Reynolds number is then given by
Re = ρ v ⋅ D µ (23.77)
µ = ( µ IN + µ OUT ) 2 (23.78)
This approach works when the quality does not change very much between the top and the bottom of the core, or if it
changes linearly as a function of axial elevation. Unfortunately, most water reactors do not behave in this way. Thus, we
must either accept the errors that we have introduced into the Darcy equation, or we must find another way to compen-
sate for our lack of knowledge about the behavior of x(z). Fortunately, there are some simple prescriptions that can be
used to remove most of these uncertainties. First, we can use the fact that the coolant starts to boil only part of the way
between the bottom and the top of the core. The pressure drop can then be expressed as the sum of the pressure drops
for the boiling and non-boiling heights:
Alternatively, we can ignore the Darcy equation and adopt a more holistic approach. Actually, a combination of these
approaches usually gives the best results. Which approach works best depends on how the quality changes as a function
of elevation. The results also depend on the shape of the axial heat flux q″(z) and the mass flow rate m.
23.28
P ressure Drops with the Boiling and Non-Boiling Heights
When it comes to reactor design, one finds that two-phase friction is almost greater than single-phase friction for the
same geometry and the same mass flow rate. In commercial BWRs, two-phase friction can be up to five times greater
than single-phase friction in the core. To account for such a large difference, single-phase friction must be treated differ-
ently than two-phase friction. Thus, the Darcy equation (with a single-phase friction factor) can be applied to calculate
the single-phase pressure drop ΔPSP:
up to the point where boiling begins (z = LSP), and then one multiplies the single-phase pressure drop by a two-phase
friction multiplier to account for the additional two-phase friction beyond this point. The two-phase multiplier M is a
dimensionless number that always has a value greater than unity. In some books, the two-phase multiplier is given the
symbol R. Thus, the two-phase pressure drop above the non-boiling height can be written as a function of the single-
phase pressure drop:
where for a commercial BWR, M is a multiplier greater than unity and less than about ten.
The value of M is a function of the flow quality, and it is generally expressed as a function of the exit flow quality (the
value of the quality at z = H). The frictional pressure drop in a reactor coolant channel containing both boiling and non-
boiling sections is then
where ρ is the liquid density at the entrance to the channel. This expression may also be written as
where LSP is the length of the channel where the coolant is a single-phase liquid, and LTP is the length of the channel
where the coolant is a two-phase mixture. For modest amounts of subcooling, a reasonable value for f/De · ½ρv2 can be
found by evaluating ρ and μ at the saturation temperature. The value of the two-phase multiplier M can then be corre-
and P are
lated to the exit quality and the inlet mass flow rate. This approach works well as long as the values for x, m,
known. However, not all correlations are designed to account for the effects of m on M.
23.29
Correlations for the Two-Phase Multiplier
Many attempts have been made over the years to develop an accurate correlation for M. In the cores of commercial
BWRs, these attempts have met with varying degrees of success. The major problem is that M(x) is slightly different for
each flow regime (nucleate boiling, bulk boiling, slug flow, churn flow, annular flow, etc.) because the drag force changes
when the flow regime changes. Thus, the best way to determine a good value for M would be to use the average value
of the quality in the region where boiling occurs, but in many cases, this is neither practical nor possible because the
value of the quality x is only known at the exit of the core (i.e., at z = H). We would now like to discuss two correlations
that employ this approach.
23.29.1
T he Martinelli–Nelson Correlation
The first widely used correlation for M was developed by Martinelli and Nelson in 1948. This correlation, which
then became known as the Martinelli–Nelson correlation, was presented in the form of a chart showing the value
of M as a function of the ambient pressure and the exit quality. However, strictly speaking, the value of M is also a
function of the mass flow rate and the shape of the axial heat flux, but the first version of the Martinelli–Nelson cor-
relation did not account for these effects. The Martinelli–Nelson multiplier is shown as a function of exit quality in
Figure 23.22. The information they used to develop their original correlation was extracted from pressure drop data
for isothermal air–water mixtures at low pressures and temperatures. However, their original work did not account
for changes to M that were attributable to changes in the mass flow rate. Their data also did not account for the den-
sity differences between air and steam in two-phase mixtures, and these differences could be considerable (about a
23.29 Correlations for the Two-Phase Multiplier 915
FIGURE 23.22 The Martinelli–Nelson two-phase multiplier M depicted in graphical form. Notice that for an American BWR
with an exit quality of 17%, the two-phase multiplier is approximately 4.0, and for an American PWR with an exit quality of 0.5%,
it is almost exactly 1.0.
factor of 10) under certain conditions. Hence, the Martinelli–Nelson correlation usually underestimates the value of
M at low flow rates, and it does not do a particularly good job when the void fraction is high and the pressure is low.
23.29.2
T he Lottes–Flynn Correlation
About 10 years after Martinelli and Nelson published their work, a more sophisticated correlation was proposed
by Lottes and Flynn. In this correlation, it was hypothesized that the two-phase multiplier M could be written as a
function of only the exit void fraction αe. Based on their experimental observations, Lottes and Flynn hypothesized
that the two phase multiplier could be written as M = f(α e) if the exit void fraction α e was low. Interestingly, this
turned out to be a reasonable approach to use if the heat flux was spatially uniform, and if the flow was mostly
annular. A large amount of experimental data that had been collected over the years showed that M increased in
inverse proportion to the exit void fraction if the exit quality stayed below about 15%. In particular, Lottes and
Flynn found that the value of M was proportional to 1/(1 − α e) over this entire range if α e was defines as the exit
void fraction. Lottes and Flynn then deduced that most of the two-phase friction was due to the liquid flowing over
the coolant channel walls. Unfortunately, their model lacked a pressure term, and like the Martinelli–Nelson cor-
relation that preceded it, it did not account for how the two-phase multiplier was affected by changes to the mass
flow rate. As more advanced approaches were developed, the lack of a mass flow term eventually caused their work
to fall out of favor. However, the Lottes–Flynn correlation is historically significant for a number of reasons that
we would now like to discuss.
Initially, Lottes and Flynn hypothesized that M was proportional to some power of 1/(1 − α e). When the flow becomes
annular, 1/(1 − αe) is also the ratio of the total channel area to the liquid flow area at the exit. If the friction in the channel
is caused by the motion of the liquid past the wall, it can be shown (see El-Wakil 1981 and the other references at the end
of the chapter) that the two-phase pressure drop ΔPTP can be expressed in terms of the liquid velocity vl as
∆PTP = f D e ⋅ 1 2 ρl v1 2 ⋅ L TP ·M (23.85)
916 Fundamentals of Two-Phase Flow in Nuclear Power Plants
where LTP is the length of the channel where the boiling occurs, and
Thus, the two-phase multiplier becomes a quadratic function of (1/(1 − αe)), with all of the terms having an equal weight.
At first sight, this would seem rather strange because a more generally acceptable form of the correlation might be
M = A + Bm + Cm 2 (23.87)
Here the constants A, B, and C would have different values, and m = (1/(1 − α e)). However, Lottes and Flynn based their
work on a constant heat flux at the wall surface. Curiously enough, for this specific set of conditions, it can be shown
that each of the weighting coefficients is exactly the same. Thus, we can conclude that if the heat flux q″ is spatially
uniform (i.e., q″ is a constant that is independent of z), then a necessary and sufficient condition for the correlation to
be valid at low exit qualities is for A = B = C = 1/3. As we shall see shortly, more modern correlations do not have these
restrictions, and consequently, they can be applied to a wider range of operating conditions with more flow regimes.
23.29.3
O ther Popular Correlations
The biggest drawbacks to the Martinelli–Nelson correlation and the Lottes–Flynn correlation are that they are based
on the assumption that pressure drop can be correlated to either the exit flow quality x or the exit void fraction α e. They
do not necessarily care about what happens to the coolant between the time it starts to boil and the time that it exits the
core. Unfortunately, it is not possible to create a more accurate pressure drop correlation without including all of the
complicated physics that occur between the channel inlet and channel exit. In general, this turns out to be a daunting
task because it is necessary to account for all of the flow regimes that we discussed earlier. However, there are a variety
of ways this can be done without getting lost in the details, and so this is the next subject that we would like to discuss.
23.30
Using the Homogeneous Equilibrium or HEM Model
The key to developing a better correlation for the two-phase friction multiplier is to understand how the liquid and vapor
phases interact when their velocities change. The simplest model that can be devised to account for this interaction is one
where the liquid and vapor phases move at the same velocity, where the two velocities are uniform within a given flow
area, and where the phases are in thermodynamic equilibrium. If they are in thermodynamic equilibrium, the tempera-
ture of the liquid phase must be the same as the temperature of the vapor phase. When these assumptions apply, then we
can construct what is called a homogeneous equilibrium model for the two-phase pressure drop. Sometimes this model
is also called a HEM model. Then the Darcy equation becomes
where f TP is the two-phase friction factor, and ρm and vm are based on the average properties of the mixture. That is to
say, ρm = αρv + (1 − α)ρl is the effective density of the mixture, and vm is the effective velocity. If the phases are in thermal
equilibrium, it can further be shown (refer to Todreas and Kazimi 2008 at the end of the chapter) that
In other words, in this expression, the two-phase multiplier increases linearly with the local quality, and the constant of
proportionality between the two phases is (ρl − ρv)/ρv. For large density differences between the phases, this reduces to
( )
M(x) ≈ 1 + x ρl ρv (23.90)
In a BWR operating at approximately 1,050 PSI (7.25 MPa), the ratio of the liquid density to the vapor density is about
20 to 1. Hence, we can set ρl/ρv ≈ 20 and the two-phase multiplier becomes
M(x) ≈ [1 + 20x] (for a water − steam mixture in a typical American BWR) (23.91)
This is clearly an improvement over our previous results because the two-phase multiplier is now a function of just the
local quality x, rather than just the exit quality, as it should be. When there is no vapor, then x = 0 and M = 1. Moreover,
at the core exit where x ≈ 0.15, M ≈ 4. If we want to include the effects of the viscosity on the two-phase multiplier, as
McAdams originally suggested, then the expression above can be modified to read
where n = 0.20 or n = 0.25. In general, M increases with decreasing pressure or increasing flow quality, and the viscosity
becomes increasingly important as the flow quality becomes higher. Interestingly enough, this happens to be completely
consistent with our previous criticisms of the Martinelli–Nelson Correlation, where we stated that it did not do a particu-
larly good job of predicting the two-phase multiplier at low pressures and high void fractions. However, in most cases,
the HEM model takes care of these problems.
23.31
Separated Flow Models
It is possible to improve upon the results of the HEM model even further by recognizing that the velocities of the liq-
uid and vapor phases are usually not the same. In general, this requires that vl ≠ vv, and this is particularly true in the
annular flow regime where the slip ratio can be greater than 4. Then there must be a different momentum equation for
each phase, and the resulting model where the velocities are modeled independently is called a separated flow model.
It is a long and laborious task to derive an expression for the two-phase multiplier from the momentum equations under
these conditions. However, a number of more modern correlations have been developed using this basic premise. They
include the Lockhart–Martinelli correlation, the Thom correlation (which was developed at Cambridge from a large
steam–water database), the Baroczy correlation, and the Friedel correlation. The Friedel correlation is the most modern
of these correlations, and it has gained a fair amount of popularity over the years. Today the Friedel correlation is prob-
ably used more than any other correlation to predict the two-phase pressure drop in BWR cores.
23.32
T he Friedel Correlation for the Two-Phase Multiplier for BWRs
The Friedel correlation, which was developed in 1980, is probably the most accurate general-purpose correlation for
BWRs, and so we would like to devote a couple of paragraphs to explain how it works. It is written in terms of the total
l = ρlvlA), and it works for horizontal and vertical flows if the flow direction is
mass flow rate of the liquid phase (i.e., m
up. A similar correlation is available for vertical down flow (which can help to determine the flow patterns in the down-
comer of a steam generator), but it is not as accurate. In the Friedel correlation, the two-phase multiplier M is written as
where
( )
A = (1 − x)2 + x 2 fl ρv fv ρl (23.94a)
( ) (µ ) (1 − µ )
0.91 0.19 0.7
C = ρl ρv v µl v µl (23.94c)
f l is the friction factor for the liquid, and fv is the friction factor for the vapor. Normally liquid friction is between 10 and
20 times greater than vapor friction in a boiling core. Here Fr is a dimensionless number called the Froude number,
and We is another dimensionless number called the Weber number. The Froude number is used to measure the ratio of
the inertial force of a fluid to the gravitational force, and the Weber number is used to measure the ratio of the inertial
force to the surface tension. Their most common definitions are
although other definitions can be found in the literature. Here the symbol σ refers to the surface tension, which is expressed
in N/m or dynes/cm. This correlation can be a bit messy to implement in practice because of all of the restrictions that are
placed on it, but at least it captures the physics correctly. For liquid-only and vapor-only friction, Friedel recommends a
different set of correlations that are similar to the expressions for the single-phase friction factors that we presented ear-
lier. However, there are different ones for laminar flow than there are for turbulent flow. The Friedel correlation predicts
relatively large values for the two-phase multiplier if the flow rate is low or if the flow quality is high. Exercise 23.6 below
illustrates that the value of M can sometimes exceed 10 in this case. If the Martinelli–Nelson correlation was applied to
the same problem, the exit quality in BWR would have to be at least 60% at 1,050 PSI for the two-phase multipliers to be
the same. Thus, these correlations give similar results, but the Friedel correlation is more accurate over a broader range.
23.33
More on Two-Phase Flow
In reactor coolant channels, two-phase flow is always turbulent—even when the coolant pumps are turned off. However,
the greatest advantage to two-phase flow is that the convective heat transfer coefficients are least a factor of 10 greater
than they are for single-phase flow (for the same mass flow rate). For this reason, two-phase flow is of considerable inter-
est to the nuclear industry. It allows designers to remove heat from the core more easily than a single-phase fluid would
permit, and if properly used, it also allows the core to be smaller because the power density can be increased. Finally,
two-phase flow is helpful in controlling the power output of light water reactors because it tends to harden the neutron
energy spectrum (shift it to higher energies) when the number of voids increases. This makes a light water reactor easier
to control because the power goes down as the temperature and the void fraction increase. In reactor physics, this effect
is referred to as negative feedback, and it is used extensively in both PWRs and BWRs to regulate the power level in the
core.* So contrary to popular belief, a small amount of boiling in a reactor core is not necessarily a bad thing. However,
large amounts of boiling can be detrimental because they can cause the cladding to dry out and the fuel rods to fail. Hence,
surface dry-out must be avoided, or the fuel rods will fail. Sometimes surface dry-out is also called the boiling crisis (see
our discussion of this topic in Section 23.41). The parameters that determine when it is safe to boil the coolant and when it is
not are set by the design limits of the core. Normally, these design limits are set by the US Nuclear Regulatory Commission
(the NRC), and they are different for PWRs than they are for BWRs and LMFBRs. Furthermore, these design limits must
also account for statistical uncertainties in the calculated values of the pressure, the quality, and the mass flow rate (see
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
23.34 Conservation Equations for Two-Phase Flow 919
Chapter 28). When the fuel rod temperatures are calculated using these uncertainties, the temperatures tend to cluster about
an average value or mean. This is true even when deterministic equations are used to find the reactor temperature profiles.
23.34
Conservation Equations for Two-Phase Flow
In general, conserving mass, energy, and momentum for two-phase mixtures is more complicated than it is for single-
phase fluids. This is because the phases can move at different speeds, and they can also have different densities. In fact,
in some flow regimes (like vertical churn flow and downflow), they can even move in different directions and at different
speeds (see Figure 23.23). This unusual behavior has a number of consequences that we would now like to discuss. First
consider a homogeneous mixture where the phases are moving at the same speed v (and in the same direction), and the
density of the mixture is ρm. The mass flow rate of the mixture m is then
where the value of m is based on the properties of the mixture. However, if we look at the situation more closely, what
that guarantees that
we really need to find is an expression for m
when we consider the flow of the liquid and vapor phases independently. We can determine the mass flow rate for each
phase by assuming that they occupy different cross-sectional areas in a flow channel. If the flow channel has a total cross-
sectional area A, then the flow areas for the liquid and the vapor phases are Al and Av. It also follows from this observation
that A = Al + Av. The appropriate expressions to use for the mass flow rates of the liquid and the vapor are then
FIGURE 23.23 An illustration of the void patterns in a BWR fuel assembly (on the left) and in a PWR fuel assembly
of the same height (on the right).
920 Fundamentals of Two-Phase Flow in Nuclear Power Plants
We can then write the conservation of mass for the moving mixture as
ρm v m A = ρl v l A l + ρv v v A v (23.99)
However, we also know that A = AMIX = Al + Av. Thus, we can rewrite the conservation of mass as
ρm v m ( A l + A v ) = ρl v l A l + ρv v v A v (23.100)
or
G m = ρm v m = ρl v l A l ( A l + A v ) + ρv v v A v ( A l + A v ) (23.101)
where Gm = ρmvm is called the mass flux of the mixture. Now the reader may recall from our previous discussion that the
void fraction of the mixture is given by
α m = A v ( A l + A v ) (23.102)
This latter equation defines the conservation of mass for a two-phase mixture in a reactor coolant channel or pipe. Notice
that the mass flow rate depends on the value of the void fraction as well as the densities and the velocities of the individ-
ual phases. However, when the mixture is homogeneous, both phases move at the same speed, and we can conclude that
v MIX = v l = v v (23.104)
Under this condition, the phase velocities are identical, the slip ratio is equal to 1.0, and we can write
Thus, the only time we can legitimately volume-average the liquid and vapor densities to get the density of the mixture
is when the velocities of the phases are the same. This relatively simple expression also affects the way that we must
handle energy and momentum conservation for a two-phase mixture. In particular, the conservation of energy for the
mixture must be written as
and the conservation of momentum for the same mixture must be written as
If we write these expressions in any other way, then the energy and the momentum of the mixture will NOT be con-
served. One additional lesson to take away from this example is that the void fraction is the most natural way to conserve
mass, energy, and momentum because all three of these quantities are based on the volume of the liquid and the vapor
phases. If we were to use the flow quality x instead, then we would have to write
23.35 Finding Acceleration Pressure Drop for Two-Phase Mixtures 921
where γ = 1 + S(ρv/ρl)·(1 − x)/x = 1/αm. Clearly, using the void fraction has several practical advantages in this case, and it
is easier to apply in practice. Finally, the void coefficient of the reactivity (another important quantity that is encountered in
the study of reactor control theory) is also written as a function of the void fraction for obvious reasons. Its use is discussed
in our companion book,* which may also be consulted to learn about the use of power-flow feedback in two-phase systems.
23.35
Finding Acceleration Pressure Drop for Two-Phase Mixtures
In principle, the acceleration pressure drop for two-phase mixtures can be estimated using Bernoulli’s equation, which
was discussed in Chapter 16. Just like single-phase fluids, two-phase mixtures can accelerate if their density changes,
and they can also accelerate if there is a directional change or an area change. Normally these effects are modeled
independently, and then combined together. As we first mentioned in Chapter 16, there is a force F on the fluid equal
to its change in momentum. This force is equivalent to the acceleration pressure drop, ΔPACCEL multiplied by the cross-
sectional area of the flow channel:
Acceleration pressure drops are relatively small for single-phase flows, but they can become quite large for two-phase
flows. First consider a coolant channel with no changes in area, but where enough heat is added for the coolant to boil.
Then, the coolant pressures at the inlet and the outlet must obey Bernoulli’s equation:
This implies that the acceleration pressure drop due to a phase change alone must be given by
Here the density and the velocity at the inlet and the outlet are considered to be those of the mixture. If the liquid and
vapor phases do not move at the same speed, then vl ≠ vv, and we must use a more general approach to find the change
in pressure. This involves decomposing each of the pressure drops into the sum of their constituent components. Then
Bernoulli’s equation becomes
2 2 2 2
PIN + ρl IN v l IN + 1 2 ρv IN v v IN = POUT + 1 2 ρl OUT v l OUT + 1 2 ρv OUT v v OUT (23.112)
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
922 Fundamentals of Two-Phase Flow in Nuclear Power Plants
where the subscripts l and v refer to the liquid and vapor phases. However, if the fluid entering the coolant channel has
not yet boiled, then the term for the momentum of the vapor can be ignored at the inlet, and Bernoulli’s equation can
be written as
or
where x is the mixture quality at the exit. Sometimes this expression for the acceleration pressure drop ΔPACCEL is also
written as
∆PACCEL = (m/A)
2
⋅ M ACCEL = (ρv)2 ⋅ M ACCEL (23.118)
In this context, ρv is called the mass flux, which is the mass flow rate per unit area. Representing the mass flux by the
symbol G, the equation for the acceleration pressure drop becomes
= ρv/A. Thus, the value of MACCEL increases as α increases. Notice that we did not use any assumptions
where G = m/A
regarding the slip ratio S = vv/vl to arrive at this result. Moreover, the acceleration multiplier does not depend on how
heat is added to the coolant channel because it is based on only the exit quality x and the exit void fraction α. Finally,
because we used Bernoulli’s equation to derive this result, Equation 23.120 can be applied to any combination of flow
regimes.
23.36
Single-Phase Pressure Drops due to Sudden Area Reductions
If a blockage or other obstruction develops in a flow channel, the pressure in the channel will change because the fluid
must accelerate or decelerate around the blockage. The area change created by this blockage can then be modeled using
Bernoulli’s equation when the flow is incompressible. Consider a viscous two-phase mixture where we would like to find
the value of the area pressure drop ΔPAREA. Since ρl and ρv are constants (because of the incompressibility assumption),
Bernoulli’s equation can be written as
23.36 Single-Phase Pressure Drops due to Sudden Area Reductions 923
FIGURE 23.24 In single-phase flow, the pressure drop depends on how fast the flow area changes. The pressure drop consists
of (1) a kinetic energy term and (2) a viscous dissipation term. The kinetic energy term depends on the areas before and after the
contraction, and the viscous dissipation term depends on how fast the flow area contracts. Faster contractions result in more viscous
dissipation which leads to additional frictional losses and vortices at the contraction point.
where the final term represents the pressure drop due to the friction of the liquid. The dimensionless loss coefficient k
is called the Borda–Carnot loss coefficient. An expression for k can be derived by combining the continuity equation
with Bernoulli’s equation to express v2 as a function of v1. In this case (see Figure 23.24), the area changes suddenly
perpendicular to the direction of flow, and the continuity equation becomes
1=m
m 2 or ρ1v1A1 = ρ2 v 2 A 2 (23.122)
(
v 2 = v1 ρ1A1 ρ2 A 2 (23.123) )
Substituting Equation 23.123 back into Bernoulli’s equation gives
(
P2 − P1 = 1 2 ρ1 v1 2 − 1 2 ρ2 v 2 2 − 1 – A 2 A1( ) )⋅
2
1
2 ρ2 v 2 2 (23.124)
so
( (
k = 1 – A 2 A1 ) ) (23.125)
2
Suppose that the density of the coolant is the same everywhere in the flow stream. Then, we can rewrite Bernoulli’s
equation as
( (
P2 = P1 + 1 2 ρ v1 2 − v 2 2 − 1 – A 2 A1 ) )⋅
2
1
2 ρv 2 2 (23.126)
Using this same terminology, we also find that we can write Equation 23.124 as
P2 − P1 = 1 2 ρv1 2 − 1 2 ρv 2 2 − 1 2 ρ ( v1 2 – v 2 2 ) (23.127)
since A1v1 = A2v2. The first term on the right-hand side represents the pressure change due to a change in the kinetic
energy of the fluid, and the second term represents the pressure drop due to viscous dissipation. The size of the viscous
dissipation term ½ρ(v1 − v2)2 can be reduced by expanding the channel more gradually, which reduces the area change.
In practice, this keeps the flow stream in contact with the boundary layer, and this reduces the amount of viscous
924 Fundamentals of Two-Phase Flow in Nuclear Power Plants
dissipation that can occur. If the value of Ac at the choke point is not known exactly, then the expression for k is some-
times written as
(
k = β 1 – A 2 A1 (23.128) )
2
where β is a dimensionless number less than unity, and normally having a value of about 0.4. Repeating the same process
again, the equation for the pressure drop becomes
P2 − P1 = 1 2 ρ ( v1 2 − v 2 2 ) + 1 2 ρβ ( v1 2 – v 2 2 ) (23.129)
P2 − P1 = 1 2 ρ ( v1 2 − v 2 2 ) + 0.4 1 2 ρ ( v1 2 – v 2 2 ) (23.130)
In other words, the single-phase pressure drop consists of a term due to a change in the kinetic energy of the fluid, and
a loss term, which is equal to 40% of the kinetic energy term. If the channel contracts so that v2 > v1, both terms in
Equation 23.130 are negative, and the area reduction causes the pressure to drop. In Equation 23.130, the viscous dis-
sipation term is equal to 40% of the kinetic energy term. However, if the value of β were increased to 0.6, the frictional
loss would then be 60% of the kinetic energy loss. In addition, it is easy to see that Equation 23.129 can be written as
P2 − P1 = (1 + β) ⋅ k ⋅ 1 2 ρv1 2 (23.131)
where the value of β depends on how fast the flow area contracts. Thus increasing the value of β increases the size of
the pressure drop.
ρ( v1 – v 2 ) 1 2 ρv1 2 − 1 2 ρv 2 2 Ratio = ( v1 – v 2 ) v1 2 − v 2 2
2 2
Ratio = 1
2 or
Since A2 = 2A1, then v2 = 2v1. Inserting these numbers into the expression above yields
So the ratio of the pressure drop due to viscous dissipation (friction) to the pressure drop due to a change in the kinetic
energy is 1/3. This means that 25% of the pressure drop (one part in four) is due to viscous losses, and the other 75%
of the pressure drop (three parts in four) is due to a change in the kinetic energy. Thus, this is consistent with assigning
the loss factor a value of 0.25. [Ans.]
23.37
Two-Phase Pressure Drops due to Sudden Area Reductions
Now suppose we turn our attention to the same problem for two-phase flow. The vapor accelerates more quickly than
the liquid at the contraction point, and the value of β is approximately 0.2. This reduces the void fraction at the point
of maximum constriction. However, following the constriction, the shear forces between the liquid and the wall tend to
restore the flow stream to its original configuration. Then the void fraction assumes substantially the same value as it
had prior to the contraction. To predict the pressure drop under these conditions, the mixture density must be written as
ρm = α m ρv + (1 − α m ) ρl (23.132)
Since αmρv is small compared to (1 − αm)ρ, the mixture density can be written as (1 − αm)ρ. Then, the equation for the
two-phase pressure drop becomes
23.38 Single-Phase Pressure Increases due to Sudden Area Expansions 925
FIGURE 23.25 In two-phase flow, the pressure drop also depends on the value of the void fraction α at the contraction point. For
the same mass flow rate and area change, the two-phase contraction pressure drop is larger than the single-phase contraction drop.
Moreover, this pressure drop tends to increase as the void fraction becomes larger (see Equation 23.137.)
where v1L is the velocity of the liquid. When α = 0, Equation 23.133 reduces to its single-phase equivalent. Comparing
Equations 23.131 and 23.133 further, we see that for the same mass flow rate and area change, the two-phase contrac-
tion pressure drop is larger than the single-phase contraction drop. Moreover, this drop increases with the void fraction
(see Figure 23.25).
23.38
Single-Phase Pressure Increases due to Sudden Area Expansions
Coolants encounter area changes when they enter and leave reactor fuel assemblies, pass over grid spacers inside of the
assemblies, and flow through pipes and valves. Sometimes these area changes are gradual and sometimes they are not.
However, if a coolant channel expands, the pressure will rise in response to the area change. Moreover, this rise in the
local pressure will be greater for two-phase flow than it is for single-phase flow (for the same mass flow rate and area
ratio). First consider the pressure change for single-phase flow (refer to Figure 23.26). The thermophysical properties
FIGURE 23.26 In single-phase flow, area expansions lead to pressure increases. The pressure increase depends on the area ratio as
well as the rate of expansion of the flow areas. Gradual expansions lead to larger pressure increases than abrupt ones because there
is less viscous dissipation.
926 Fundamentals of Two-Phase Flow in Nuclear Power Plants
before the channel expands are given the subscript 1, and the thermophysical properties after the channel expands are
given the subscript 2. If there is no elevation change, the pressure change is given by
P2 − P1 = 1 2 ρ v1 2 − v 2 2 − k ⋅ 1 2 ρv1 2 (23.134)
where the first term is due to a kinetic energy change, and the second term is due to the presence of fluid viscosity.
However, we also know that k·½ρv12 = ½ρ(v1 − v2)2 since k = 1 − (A1/A2)2. Thus, the expression for the pressure increase
becomes
P2 − P1 = 1 2 ρ v1 2 − v 2 2 − 1 2 ρ( v1 – v 2 ) (23.135)
2
l = ρv1A1,
and since m
( ) ( )
P2 − P1 = (1/ρ) 1 A1A 2 – 1 A 2 2 m
1 2 (23.136)
In other words, whenever A2 > A1, the kinetic energy falls, and the fluid pressure (which can be correlated to the poten-
tial energy of the fluid) rises. This change is proportional to the term [(1/A1A2) − (1/A22)]. However, as Equation 23.135
shows, frictional losses, represented by the second term in Equation 23.135, also reduce the pressure to a certain extent.
The size of this term can be reduced even further when the expansion becomes more gradual. Then the boundary layer
does not separate from the surface of the wall, and this results in what is called a subsonic diffuser. In addition to being
used in nuclear power plants, diffusers are also used in heating, ventilating, and air-conditioning systems. If the fluid
velocity becomes so high that the flow also becomes critical or choked, then these equations do not predict the pressure
drop correctly, and the approach presented in Chapter 29 must be used.
23.39
Two-Phase Pressure Increases due to Sudden Area Expansions
In two-phase systems, the pressure also rises after a sudden flow area expansion. However, the rate of this increase
depends on the value of the local quality and the void fraction. The most commonly used expression for the two-phase
increase is
( ) ( )
∆PTP = P2 − P1 = (1/ρ) 1 A1A 2 – 1 A 2 2 m ( )( )
1 2 ⋅ (1 – x)2 (1 − α) + x 2 α ρl ρv (23.137)
which is derived in the references at the end of the chapter. This expression is reasonably accurate for high void fractions
(α > 0.4) and area ratios greater than 0.5. Thus, it can be used for many nuclear applications.
23.40
T he Vena Contracta
For both single-phase and two-phase flows, sudden area changes can lead to an acceleration or deceleration of the flow.
The sudden contraction and subsequent expansion of a flow channel is then a function of the amount of constriction
that occurs. The point of maximum constriction is called the Vena Contracta, and a typical Vena Contracta is shown
in Figure 23.25. The frictional losses due to the contraction of the flow field in front of the Vena Contracta (from point
0 to point 1) are relatively small, but those following the Vena Contracta (from point 0 to point 2) can be quite large—
especially when the boundary layer separates. This has some important consequences in the study of orifice flow, which
we will discuss in Chapter 29.
23.41
More Applications of Bernoulli’s Equation
Applying Bernoulli’s equation to other two-phase systems requires a few more assumptions. Suppose that we write
Bernoulli’s equation as
where the subscript v refers to the vapor, and the subscript l refers to the liquid. Using the fact that
and neglecting the effects of gravity, we can show that P1 and P2 are related by
(( ) (
P2 = P1 + ρv1 2 ⋅ M TP A1 A 2 – A1 A 2 ) ) (23.140)
2
or
(( ) (
∆PTP = ρv1 2 ⋅ M TP A1 A 2 – A1 A 2 ) ) (23.141)
2
where MTP = (k + other terms) is a dimensionless two-phase multiplier that depends on the area ratio A1/A2. For area
ratios of 0.5 and higher, and reasonably high void fractions, it can be shown that the approximate value for MTP is
{ ( )}
M TP = (1 − x)2 (1 − α) + x 2 α ⋅ ρl ρv (23.142)
This expression also predicts that the pressure will rise as the area expands, and that it will fall as the area contracts.
However, for the same mass flow rate and cross-sectional areas, the pressure change in a two-phase system is always
greater than it is in a single-phase system because the value of MTP is always greater than 1.0. In other words, the accel-
eration pressure drop is greater than it is for single-phase flow. We can show that this is true by noting that the value
of x is always smaller than the value of α, so MTP always increases with α and x. Finally, Equation 23.142 reduces to its
single-phase counterpart when x = 0 and α = 0. The example below illustrates how the two-phase multiplier changes in
a BWR fuel assembly. For modest area changes (up to about 50%), and high void fractions (>80%), the value of MTP is
typically 4.0–5.0. At lower void fractions, it becomes much closer to unity (See Figure 23.27).
FIGURE 23.27 For area expansions in two-phase systems, the pressure rise is greater than it is for single-phase flow for the same
mass flow rate and area ratio. In addition, it increases as the void fraction of the two-phase mixture increases.
928 Fundamentals of Two-Phase Flow in Nuclear Power Plants
{ ( )
α = 1 1 + ρv ρl ⋅ (1 − x)/x }
If x = 0.17 (17%), then α = 1/{1 + (1/22)(0.83/0.17)} = 1/1.222 = 81.8%. The two-phase area change multiplier is then MTP
= {(1 − x)2/(1 − α) + x2/α·(ρl/ρv)}. Since (ρl/ρv) = 22, x = 0.17, and α = 0.818, the value of MTP is MTP = {(1 − x)2/(1 − α) +
x2/α·(ρl/ρv)} = {(1.0 − 0.17)2/(1 − 0.818) + 0.172/0.818·(22)} = 3.785 + 0.0289/0.818·22 = 3.785 + 0.777 = 4.562. [Ans.]
23.42
Calculating Two-Phase Pressure Changes
Earlier, we demonstrated that pressure changes in two-phase systems could be attributed to the effects of
1. Fluid friction
2. Fluid expansion and contraction
3. Flow blockages and area changes.
Adding thermal energy to the flow creates an acceleration pressure drop that requires an acceleration multiplier
M ACCEL greater than 1.0. This is in addition to the two-phase friction multiplier M FRICTION that can be determined
from the Martinelli–Nelson, Lottes–Flynn, Baroczy, or Friedel correlations. Finally, a two-phase area multiplier
M AREA can be added to the previous multipliers if the channel area changes, but when there is no additional heat
addition or subtraction. These three two-phase multipliers can then be combined together to find the total pressure
change:
where SP refers to the single-phase pressure change. We can also write this as
Two-Phase Pressure Changes with Friction, Heat Addition, and an Area Change
∆P( TOTAL ) TP = ∆PFRICTION (SP) ⋅ M FRICTION + ∆PACCELERATION (SP) ⋅ M ACCELERATION
+ ∆PAREA (SP) ⋅ M AREA (23.143)
This innocent looking equation allows the two-phase pressure change to be found anywhere in a reactor core. The two-
phase multipliers provide a convenient way to convert the values of ΔPFRICTION (SP), ΔPACCELERATION (SP), and ΔPAREA (SP)
into their two-phase analogs. The values for these multipliers are a function of the local quality and void fraction,
Moreover, their values increase as x and α increase. When reactors were first being designed, the values of M were based
on the exit quality xe and void fraction α.
23.43
P ressure Losses in BWR Fuel Assemblies
In any BWR, the pressure can drop due to friction, gravity, and acceleration. In a BWR fuel assembly without grid spac-
ers, the magnitude of these losses can be calculated rather easily. The acceleration pressure drop is simply
M ACCELERATION = (1 – x e )
2
((1 − α)ρl ) + x e 2 (αρv )]OUT – (1 ρl )IN (23.146)
The frictional pressure drop is
where
and MFRICTION = A + 3.24·BC/Fr 0.045·We0.035 (see our earlier discussion of the Friedel correlation). Finally, the gravita-
tional pressure drop is
23.44 Pressure Losses in PWR Fuel Assemblies 929
FIGURE 23.28 This figure shows the overall pressure drop in the interior coolant channels of a BWR fuel assembly without grid
spacers. Here the components of the pressure drop including friction, acceleration, and gravity are depicted as a function of the mass
flux in kg/m2 s. Boiling is assumed to occur roughly ¼ of the way up the core. The frictional component of the pressure drop increases
rapidly with the mass flux while the other components barely change at all.
where Hl is the distance through the core where the coolant is a liquid, and HMIX is the distance through the core where
the coolant is a two-phase mixture. Hence, the total core height H is just the sum of Hl and HMIX; i.e., H = Hl + HMIX. For
typical BWR fuel assemblies with no grid spacers, an exit quality of 14.5% (x = 0.145), and a co-sinusoidal axial heat
flux, representative values of the acceleration pressure drop, the frictional pressure drop, and the gravitational pressure
drop are
respectively. These values are based on an average mass flux G of 1,600 kg/m2s. The total pressure two-phase drop
ΔPTOTAL (TP) is then
Hence, the frictional pressure drop is largest, followed by the gravitational pressure drop and finally by the acceleration
pressure drop. The total pressure drop is then shown in Figure 23.28 as a function of the mass flux in kg/m2s. Now let us
repeat the same process for an American PWR.
23.44
P ressure Losses in PWR Fuel Assemblies
In an American PWR (such as one sold by Westinghouse), the core-averaged mass flux is about 2.5 times what it is in the
BWR we discussed previously. Suppose that the mass flux is 3,800 kg/m2s. In a bare fuel assembly without rod spacers
where the effects of nucleate boiling can be ignored, the pressure drop consists of a single-phase friction term:
FIGURE 23.29 This figure shows the overall pressure drop in the interior coolant channels of a PWR fuel assembly without grid
spacers. Here the two major components of the pressure drop (friction and gravity) are depicted as a function of the mass flux in
kg/m 2 s. The acceleration pressure drop is essentially negligible without grid spacers. Notice that the frictional component of the
pressure drop increases rapidly with the mass flux while the gravitational component remains essentially constant.
where H is the total core height. The single-phase pressure drop is then
or
∆PTOTAL (SP) = 44.8 kPa + 26.8 kPa = 71.6 kPa (or 10.4 PSI) (23.154)
Hence, the frictional pressure drop is approximately twice the gravitational pressure drop. Moreover, because the cool-
ant does not boil, the friction factor f l is proportional to the mass flux to the −0.2 or −0.25 power (see Chapter 17), and
the frictional pressure drop is proportional to the mass flux G to the 1.75 or 1.8 power:
The value of ΔPGRAVITY (SP) depends on only the local density and the core height, and is essentially independent of the
value of the mass flux. The total pressure drop is then shown in Figure 23.29 as a function of the mass flux at the inlet
(in kg/m2s).
23.45
Negative Qualities in Reactor Fuel Assemblies
In reality, the assumption of thermodynamic equilibrium does NOT apply to most reactor components when the flow
rate is high and the coolant velocity is greater than a couple of meters per second. When this occurs, the true flow
quality is often different than the equilibrium flow quality determined by a simple energy balance. This is particularly
true in the core when the coolant enters the core as a subcooled liquid and then proceeds to boil for a while before the
equilibrium quality becomes zero. In these situations, the equilibrium quality cannot be found from the conventional
definition for the equilibrium quality
because vapor does not yet exist in the coolant channel from a thermodynamic perspective. For situations such as this
where the coolant is subcooled thermodynamically, but where bubbles can form and nucleate boiling can occur at the
surface of a fuel rod, a different approach is needed to calculate the equilibrium quality xeq. In this case, the equilibrium
quality before the zero quality point is reached can be found from the relationship
where hlv is the enthalpy of vaporization at the saturation pressure, hl is the enthalpy of the liquid at the saturation pres-
sure, and h(z) is the enthalpy of the coolant at axial location z. Normally, the quality before the zero quality point is
reached is a negative number, and it continues to remain so until the bulk temperature becomes equal to the saturation
temperature (TBULK = TSAT).
In Chapter 24, we will seek to quantify this behavior in a boiling core. However, the amount of subcooling at the inlet
to many BWRs is normally so great that the equilibrium quality may have a value as low as −0.06 (−6%) before it enters
the core. Moreover, its value may also be negative as much as a half a meter (50 cm) after the coolant boils because its
bulk temperature may still less than the saturation temperature. Hence, thermal non-equilibrium effects are generally
important in reactor cores, and in particular, they must be taken into account when examining the behavior of boiling
cores. We will have more to say about these effects in Chapter 24.
23.46
Boiling Heat Transfer and the Boiling Crisis
Having completed our introduction to two-phase flow, we would next like to turn our attention to the subject of two-
phase nuclear heat transfer. Two-phase heat transfer is more complicated than single-phase heat transfer because
the convective heat transfer coefficients are flow regime dependent. When the first commercial nuclear reactors were
built, these flow regimes presented a real challenge to the mechanical and nuclear engineers of the time. In particular,
the existence of these flow regimes was first confirmed in BWR fuel assemblies by the famous Borax experiments in
the 1950s. Since those days, two-phase flow has become better understood, but the existence of these flow regimes still
presents a significant engineering challenge—particularly during certain types of reactor accidents and hypothetical
LOCAs. In future chapters, we would like to discuss these challenges in more detail. Normally, the convective heat
transfer coefficients for two-phase flow are much larger than they are for single-phase flow. This means that two-phase
flow can actually remove more heat from the core than single-phase flow for the same mass flow rate. However, this heat
must be removed in such a way that the cladding does not dry out and the fuel does not melt. When the liquid film disap-
pears from the surface of the fuel rods, the convective heat transfer coefficient falls precipitously, and the fuel tempera-
ture begins to rise. In some cases, this can even cause the fuel to melt and redistribute itself through the core—resulting
in a possible recriticality event. In reactor design, this unfortunate chain of events is referred to as the boiling crisis. The
heat flux at or just before the dry-out point is called the critical heat flux. However, as we will see in Chapter 24, modern
reactors are designed in such a way that the fuel rods almost never dry out, and the cladding almost never fails. In PWRs,
the most generally accepted way to measure how close one is to fuel rod dry-out is the departure from nucleate boiling
ratio (or DNBR). In BWRs, the onset of the critical heat flux is predicted by another parameter called the critical power
ratio (or the CPR). Reactor designers must therefore demonstrate that the DNBR and CPR limits are never reached
during anticipated operation. This includes reactor transients and many commonly anticipated design basis accidents
or DBAs (see Chapter 28). The safety of the plant then depends on not exceeding these limits during both transient and
steady-state operation. Finally, overpower transients must be taken into account when performing this analysis.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Crowe, C. and Schwartzkopf, J., et al. Multiphase Flows with Droplets and Particles, CRC Press, Boca Raton, FL (2012).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
932 Fundamentals of Two-Phase Flow in Nuclear Power Plants
Web References
https://www.irc.wisc.edu/properties/.
http://nht.xjtu.edu.cn/paper/en/2011206.pdf.
27. For the same mass flow rate and the same coolant channel geometry, is the two-phase friction factor less than, equal
to, or greater than the single-phase friction factor?
28. In an HEM model, what hydrodynamic variables are usually required to calculate the two-phase friction factor?
29. What is the purpose of the Darcy equation? Is it ever used to describe a two-phase mixture?
30. For what specific flow geometries was the Darcy equation originally developed?
31. How can the Darcy equation applied to reactor fuel assemblies even though it was originally developed for single-
phase flow in circular pipes and tubes?
32. What are the boiling and non-boiling heights in a commercial BWR core that is 4 m high?
33. What is the highest value of the fluid quality that is usually allowed in a BWR fuel assembly during normal reactor
operation?
34. How many flow regimes can there be between a single-phase liquid and a single-phase vapor in a r eactor coolant
channel?
35. What is a typical value of the two-phase multiplier M that appears in the Martinelli–Nelson correlation?
36. What is the difference between a separated flow model and a homogeneous equilibrium model?
37. How many fluid conservation equations does an HEM model require?
38. What type of boiling immediately follows the nucleate boiling regime in a reactor coolant channel?
39. How many distinct components are there to the nucleate boiling curve?
40. What is an average value of the void fraction at the exit of a commercial BWR core?
41. What is the primary difference between slug flow and churn flow in a reactor coolant channel?
42. What is the condition required for a vapor bubble to develop on the surface of a nuclear fuel rod and subsequently
grow?
43. What is the difference between annular flow and mist flow?
44. What is the purpose of the Lottes–Flynn correlation?
45. How can one calculate the two-phase pressure drop when it can no longer be assumed that the two-phase mixture
in a reactor coolant channel is a homogeneous mixture?
46. How is the slip ratio defined in a homogeneous model of two-phase flow, and how is it defined in a drift flux model?
47. What advantage does a drift flux model have over a conventional model that uses a slip ratio?
48. What is the purpose of the RELAP 5 computer code, and where was it originally developed?
49. Under what set of conditions does the drift flux model of Fauske give its most accurate results?
50. Can you think of at least one condition where the two-phase flow in a reactor fuel assembly is not entirely turbulent?
51. How is the void coefficient of the reactivity defined for a thermal water reactor?
52. Is the value of the void coefficient of the reactivity positive or negative for a commercial BWR?
53. Why are the fuel assemblies in a BWR core surrounded by stainless steel “cans” or sheaths?
54. What is the mean neutron lifetime used for, and what is the difference between its value in a water reactor and an
LMFBR?
55. What causes the liquid and vapor phases in a two-phase mixture to depart from thermal equilibrium?
56. How can the two phases be brought back into thermal equilibrium when they are initially in separate states?
57. Under what conditions does the flow in a reactor coolant channel become choked?
58. What is the value of the equilibrium quality at the entrance to a typical BWR fuel assembly? Is it zero, positive, or
negative?
59. For a given value of the equilibrium quality, does the void fraction increase or decrease as the slip ratio is increased?
60. Name three of the most popular reactor safety analysis codes used in the United States today.
61. What is a reactor LOCA, and why is it so important to the nuclear industry as a whole?
62. In correlations for two-phase flow, what is the purpose of the two-phase multiplier M?
63. How many equations are used in a two-fluid model of reactor behavior (excluding the equations of state)?
64. Under what conditions does an LOCA generate a shock wave?
65. What is the largest single drawback to the Lottes–Nelson correlation for reactor design?
66. What factors eventually led to the Martinelli–Nelson and the Lottes–Flynn correlations being replaced by more
modern correlations for the two-phase friction multiplier?
67. Write an equation for the void fraction as a function of the equilibrium quality.
68. What is a typical value for the two-phase multiplier at the top of a BWR core?
69. What does the term “critical flow” refer to?
70. Is there any difference between the critical flow rate through a long pipe and a short pipe with the same cross-
sectional area and the same back pressure?
71. How can one increase the critical flow rate through an orifice in which the maximum velocity of the coolant is
limited?
934 Fundamentals of Two-Phase Flow in Nuclear Power Plants
7 2. Write a condition when bubbles will grow on the surface of a nuclear fuel rod.
73. For the same bubble radius R, is the bubble generation rate higher or lower on a well-wetted surface than it is or a
poorly wetted one?
74. What are some representative inlet and outlet temperatures from the steam generators in the primary loop of a PWR?
75. At what temperature does the steam leaving the secondary loop of these steam generators become c ompletely
saturated?
76. In a water–steam mixture, what is the definition of the equilibrium quality x?
77. Write an equation for the void fraction as a function of the equilibrium quality.
78. Some reactor steam generators are designed to produce superheated steam. What does the term “superheated steam”
refer to, and why is superheated steam preferable to saturated steam?
79. What are the inlet and outlet temperatures from the steam generators in the secondary loop of a PWR?
80. In a water–steam mixture, write an equation describing the relationship between the void fraction and the slip ratio.
81. Does the void fraction become larger or smaller for the same value of the quality when the slip ratio is increased?
82. Under what conditions can the equilibrium quality become less than zero in a reactor coolant channel, and what
does this value of the quality imply?
83. What is the value of the equilibrium quality at the inlet to a typical fuel assembly in an American PWR?
84. What are the three biggest contributors to the pressure drop in water reactor fuel assemblies?
85. In commercial PWRs, what is the largest single contributor to the single-phase pressure drop—gravity or friction?
86. In a contracting flow channel in a thermal water reactor, does the pressure of a frictionless fluid become larger or
smaller?
87. What does the term Vena Contracta refer to, and is it applicable to sonic or subsonic flows?
88. Fill in the following expression with the appropriate word or phrase: In a separated flow model, momentum
equations are used to describe the liquid and vapor phases.
89. In an American BWR, what is the largest single contributor to the single-phase pressure drop—gravity or friction?
90. Give two examples of stratified flow.
91. Fill in the following expression with the appropriate word or phrase: In a homogeneous equilibrium model,
momentum equation(s) and energy equation(s) are used to describe the behavior of the liquid and vapor
phases.
92. What two processes that occur in nuclear power plants are responsible for the existence of the world’s water cycle?
93. In an expanding flow channel in a thermal water reactor, does the pressure of a frictionless fluid become larger or
smaller?
94. What does the Friedel correlation attempt to do?
95. Write down the mathematical definitions for the Froude number and the Weber number.
96. What is the meaning of each of these numbers, and how are they normally used in two-phase flow?
97. How is the Reynolds number defined for a two-phase mixture? Do the phases have to be travelling at the same speed
for this definition to apply?
98. Write an expression for the density and the viscosity of a homogeneous two-phase mixture in terms of the densities
and viscosities of the individual phases
Exercise 23.2
In a typical BWR, the equilibrium quality at the top of the core is between 13% and 17%. If the pressure at this elevation
is 1,050 PSI (7.25 MPa), what is the void fraction when the slip ratio is 1.0 and then when the slip ratio is 4.0? Which of
these two scenarios more accurately represents the behavior of the core at this elevation?
Exercise 23.3
Liquid sodium in a commercial LMFBR boils at a temperature of about 880°C. If the sodium happens to boil, what is
the ratio of the specific volume of the vapor to the specific volume of the liquid at this temperature? What is the name of
the pressure wave that is generated under these conditions?
Exercises for the Student 935
Exercise 23.4
For water at atmospheric pressure, what is the ratio of the densities of the liquid and vapor phases when the water boils? If the
velocity of the vapor is then 3 times that of the liquid, what is the value of the void fraction if the equilibrium quality is 5%?
Exercise 23.5
In a condenser in a nuclear power plant, the void fraction is 99%. If the slip ratio is 1.0, what is the e quilibrium quality?
Exercise 23.6
Suppose that a BWR is to be designed where the pressure in the core is approximately the same as it is in the secondary
loop of a commercial PWR (800 PSI or 5.5 MPa). Under these conditions, what is the value of the two-phase multiplier
when the exit quality is 20%?
Exercise 23.7
In a commercial BWR, how far above the core inlet does subcooled boiling begin? What is the equilibrium quality and
void fraction at this location?
Exercise 23.8
In a commercial BWR, how far above the core inlet does bulk boiling begin? What is the equilibrium quality and void
fraction at this location?
Exercise 23.9
In a two-phase mixture, the liquid and vapor phases move in the same direction, but the vapor phase moves 50% faster
than the liquid phase. What is the slip ratio in this case?
Exercise 23.10
In a two-phase mixture, the liquid and vapor phases move in different directions, and the vapor phase has a relative
velocity twice that of the liquid phase. Can this situation actually occur in a nuclear power plant, and if so, where? What
is the value of the slip ratio in this case?
Exercise 23.11
For thermal efficiency reasons, most PWR steam generators are designed to support 10°C–30°C of superheat. What is the
enthalpy of the steam under these conditions? By what percent is its energy content greater than that of saturated steam?
Exercise 23.12
Using an HEM model, derive a simple expression for the two-phase multiplier in a BWR fuel assembly. If the equilib-
rium quality is 2%, what is the two-phase multiplier? At what axial elevation should this value of the two-phase multi-
plier be used to calculate the two-phase friction factor?
Exercise 23.13
The boiling length in the core of a BWR is 3 m. For a representative commercial BWR, what is the non-boiling length?
Exercise 23.14
Suppose the slip ratio at the mid-plane of a commercial BWR core unexpectedly doubles because of a surge in the reac-
tor power level. If the quality increases by 20%, what happens to the void fraction under these conditions?
Exercise 23.15
Calculate the frictional pressure drop in a 1 m section at the top of a BWR core using the Martinelli–Nelson correlation.
Assume the mass flux G at this location is 3,000 kg/m2 s and the average quality is 12%.
936 Fundamentals of Two-Phase Flow in Nuclear Power Plants
Exercise 23.16
Construct a table showing at least six two-phase friction and void fraction calculations. In this table, show the names of
the correlations, the fluids to which they can be applied, and their general range of applicability. Also include a com-
ment column where the strengths and weaknesses of each correlation are discussed. Which correlation would you use to
design a BWR today? Which correlation would you use to design a PWR?
Exercise 23.17
Calculate the equilibrium quality at the inlet and the exit of a commercial BWR core. Also calculate the equilibrium
quality at the inlet and the exit of a commercial PWR core. How do the equilibrium qualities compare? What does a
negative equilibrium quality imply? How important do you think non-equilibrium effects are in each of these cores?
Exercise 23.18
For a two-phase mixture where the relative velocities of the liquid and vapor phases are not known, but the void fraction
and quality are known at the contraction point (commonly referred to as the Vena Contracta), derive an expression for
the two-phase multiplier due to the area change alone. What does this expression say about how the pressure behaves
when the area expands, and how it behaves when the area contracts?
Exercise 23.19
In a BWR fuel assembly, what are the two largest components of the overall pressure drop, and what are their approxi-
mate values?
Exercise 23.20
In a PWR fuel assembly, what are the two largest components of the overall pressure drop, and what are their approxi-
mate values? Why are these values different than those in a BWR, and on average, are they higher or lower?
Exercise 23.21
A 5-m-long coolant channel in the core of a research reactor has an equivalent diameter of 1 cm. Water enters the chan-
nel from below at a rate of 25 kg/s and is 10°C subcooled. The velocity of the flow is 3 m/s. Ten MW of thermal energy
is added sinusoidally to the channel. The slip ratio is 1.8, and the average system pressure is 7 MPa. Neglecting the
extrapolation lengths discussed in Chapter 5, calculate the friction and acceleration pressure drops. Assume the fuel
rods are smooth.
Exercise 23.22
A 2-m-long coolant channel in a small BWR has a cross-sectional area of 10 cm2. It receives heat uniformly from the
sides of the channel at a rate of 1 × 106 W/m2. The average channel pressure is 8 MPa. Water enters the channel as a satu-
rated liquid at 1 m/s. The slip ratio is 2, and the turbulent friction factor is 0.03. Calculate the friction and acceleration
pressure drops in the channel.
Exercise 23.23
A 4-m-long BWR coolant channel supports a flow rate of 25 kg/s. Water enters the channel 15°C subcooled, and the sys-
tem pressure is 7.25 MPa. Ten MW of heat is added to the channel sinusoidally by the nuclear fuel rods. Neglecting the
extrapolation length at the top and the bottom of the core, how far from the channel inlet does bulk boiling begin? If the
slip ratio after the coolant boils is approximately 2.5, and the single-phase friction factor is 0.03, calculate the frictional
pressure drop in the boiling and non-boiling zones.
Exercise 23.24
For the coolant channel in Exercise 23.23, determine the quality and the void fraction just as the coolant is exiting the
channel. Assume a slip ratio of 4 at the channel exit. What is the velocity of the coolant at the channel inlet, and what
are the velocities of the liquid and vapor phases at the channel exit?
24
Two-Phase Nuclear Heat Transfer
24.1
Two-Phase Heat Transfer in Nuclear Power Plants
Two-phase heat transfer is an important aspect of power plant design. Sometimes two-phase heat transfer is also referred to as boiling
heat transfer. Boiling heat transfer occurs when a fluid (which contains both liquid and vapor phases) flows over a hot object, and in
the process the object transfers some of its heat to it (see Figure 24.1). Boiling heat transfer occurs in the nuclear steam supply system
(or NSSS), and it is also needed to understand the behavior of the core and the fuel assemblies inside of it. Boiling heat transfer is
more complicated than single-phase heat transfer, but it is also more interesting. Thermodynamically speaking, boiling heat transfer
begins when the enthalpy of the coolant becomes greater than or equal to its saturation enthalpy. When this occurs, some of the
mass of the coolant is converted into a vapor. The vapor is less dense than the liquid, and the total energy of the mixture is equal to
enthalpy of the liquid hl plus the enthalpy of the vapor, hv. If the quality of the mixture is x, then the specific enthalpy of the mixture is
The rate at which energy flows from a fuel rod to the coolant is proportional to the convective heat transfer coefficient h(TP). The
amount of thermal energy q added to or subtracted from the coolant is then
where TWALL is the temperature of the hot surface (commonly called the wall), TSAT is the saturation temperature of the coolant, and
A is the surface area through which the heat flows. In terms of the nuclear heat flux q″ = q/A, this can also be written as
FIGURE 24.1 An example of a pool of boiling water where boiling heat transfer occurs. When the flow is not forced, the boiling near the heated
surface is called pool boiling, and when the flow is forced, it is called flow boiling. Flow boiling is the predominant form of boiling heat transfer in
nuclear power plants. In this picture, the hot surface which causes the boiling to occur can be seen in red.
937
938 Two-Phase Nuclear Heat Transfer
TABLE 24.1
Values of the Convective Heat Transfer Coefficients in Reactor Fuel Assemblies during Natural Circulation
(with the Coolant Pumps Turned Off) and Forced Convection (with the Coolant Pumps Turned On)
Sometimes the two-phase heat transfer coefficient is called the boiling heat transfer coefficient. Boiling heat transfer
coefficients are larger for two-phase flow than they are for single-phase flow, and their respective values are com-
pared in Table 24.1. Normally the convective heat transfer coefficients for two phase flow are 5 to 10 times larger than
comparable heat transfer coefficients for single phase flow. Now let us take a look at these values in more detail.
24.2
Single-Phase Heat Transfer Coefficients for Forced
Convection in Reactor Fuel Assemblies
For single-phase fluids, the convective heat transfer coefficients in reactor fuel assemblies are 25,000–75,000 W/m2 °C for light
and heavy water, 55–555 W/m2 °C for gaseous coolants, and 40,000–120,000 W/m2 °C for liquid metals. In British thermal
units, these values become 4,500–13,500 BTU/h ft2 °F, 10–100 BTU/h ft2 °F, and 7,000–20,000 BTU/h ft2 °F, respectively.
They depend heavily on the Reynolds number and they can also vary by ±15% depending upon the subchannel geometry.
24.3
Single-Phase Heat Transfer Coefficients for Natural
Convection in Reactor Fuel Assemblies
Now let us turn our attention to how the convective heat transfer coefficients behave when the coolant pumps are turned
off. Normally, h is lower when the flow is natural than when the flow is forced. For single-phase fluids, the value of h lies
between 300 and 1,200 W/m2 °C when the core is cooled by light or heavy water, and between 5 and 25 W/m2 °C when the
core is cooled by hydrogen, helium, or carbon dioxide. In British thermal units, h ranges between 50 and 200 BTU/h ft2 °F for
water, and between 1 and 5 BTU/h ft2 °F for gaseous coolants. For metallic coolants, it lies between 1,000–3,000 W/m2 °C.
24.4
Two-Phase Heat Transfer Coefficients for Forced
Convection in Reactor Fuel Assemblies
Two-phase flow has much greater convective heat transfer coefficients than single-phase flow, and boiling heat transfer
coefficients are two to ten times higher than the single-phase heat transfer coefficients for the same mass flow rate. In
American made PWRs, it is not unusual for the boiling heat transfer coefficient to approach 270,000 W/m2 °C or 45,000
BTU/h ft2 °F in the hottest fuel assemblies. These numbers assume a core average heat flux of 600,000 W/m2 or 190,000
BTU/h ft2 with a radial power peaking factor of 2.5 (see Chapter 25). For fresh fuel assemblies, these numbers can be
even higher. Finally, in boiling cores, the two-phase heat transfer coefficients are 5–20 times higher than the single-
phase heat transfer coefficients for the same flow rate. Consequently, there is a strong practical incentive incentive to
allow the coolant to boil as long as the amount of boiling and the intensity of the boiling can be controlled.
Boiling heat transfer coefficients are sensitive to the temperature difference between the surface of the fuel rods
and the coolant. They are also sensitive to additional factors such as the surface roughness, the mass flow rate, and the
system pressure. Next and perhaps most importantly, two phase heat transfer coefficients are flow regime dependent.
Because of this dependency, they are different for nucleate boiling than they are for bulk boiling, and they are different
for bulk boiling that they are for film boiling. Finally, the convective heat transfer coefficients are a function of whether
the flow is natural or forced. Natural convection is associated with pool boiling, and forced convection is associated
with flow boiling. For pool boiling, h lies between 1,000 and 12,000 W/m2 °C in reactor fuel assemblies such as those
that are stored in spent fuel pools. In British thermal units, the comparable values for h are 170–2,000 BTU/h ft2 °F for
24.7 Bubble Behavior in Pool Boiling and Flow Boiling 939
light and heavy water. The convective heat transfer coefficient rises with the Reynolds number, and it is also a function
of the hydraulic diameter and the coolant velocity. The value of h can change between the bottom and the top of a verti-
cal surface, and therefore, it may have to be calculated at several different locations depending upon how the Reynolds
number is defined (see Chapter 22). Over horizontal tube banks, the heat transfer coefficients are different on the top of
a tube than they are on the bottom. We will have more to say about this in Section 24.20.
24.5
Characterizing Boiling Heat Transfer
As we just mentioned, two-phase heat transfer coefficients are a function of the conditions at an object’s surface. If the surface
is covered by a liquid film, the value of h is different than when it is covered with a vapor film. Each flow regime has a dif-
ferent liquid-to-vapor ratio, and this ratio affects the relative contributions of conduction and convection to the heat transfer
process. The velocities of each phase and their respective Reynolds numbers also affect the size of the heat transfer coeffi-
cient. The conditions under which a nuclear fuel rod can dry out are a function of the system pressure, the flow regime, and the
heat flux. Depending on the pressure and flow rate, the probability for dryout can be described by a set of parameters called
☉☉ The critical heat flux (CHF)
☉☉ The departure from nucleate boiling ratio (DNBR)
☉☉ The critical power ratio (CPR).
The critical power ratio is used to describe when dryout occurs in boiling water reactors (BWRs), and the DNBR is
used to describe when dryout occurs in PWRs. These parameters represent different ways of determining if a fuel rod
will fail due to inadequate cooling. There are several ways to keep a fuel rod cool, but they are all based on keeping the
convective heat transfer coefficient as large as possible so that the heat can be removed as quickly as possible from the
cladding. In general, the methods for determining when dryout occurs are related to the subject of bubble formation and
growth, which were introduced in Chapter 23.
24.6
Bubble Formation and Growth
In reactor work, it is widely believed that boiling heat transfer is the result of three separate and distinct processes called
1. Bubble formation, bubble growth, and departure
2. Liquid micro-layer evaporation from a hot surface
3. The resulting micro-convection of liquid moving with the bubbles and being replaced by cooler liquid reaching
the heated surface.
In general, bubble formation and growth begins when the temperature of the fluid exceeds the saturation temperature.
Normally, this occurs close to a heated surface, which in a nuclear power plant, happens to be a nuclear fuel rod. Boiling
requires the wall temperature T WALL at the surface to be greater than the saturation temperature TSAT. Of course, this
location can vary if the heat flux or the flow rate is changed, but in PWRs, it is usually greatest above the core midplane
(see Figure 24.2). Conversely, in BWRs, the coolant begins to boil about one-quarter of the way up the core, and it con-
tinues to boil until the top of the core is reached. This boiling can be beneficial as long as the fuel rods do not dry out
or fail. The axial elevations at which boiling occurs in American PWRs and BWRs are shown in Figure 24.2. These
locations can vary considerably from one fuel assembly to the next, and they are a function of the local heat flux and the
enrichment of the fuel.
24.7
Bubble Behavior in Pool Boiling and Flow Boiling
Boiling is usually classified into two broad categories called
1.
Pool boiling
2.
Flow boiling
Flow boiling is different than pool boiling because flow boiling requires the coolant to be actively flowing past a
heated surface (i.e., driven by a coolant pump), while pool boiling does not. Hence, in addition to just localized boiling,
flow boiling requires a flow to be mechanically superimposed on top of the local flow. Not surprisingly, each of these
processes leads to different heat transfer coefficients, and the convective heat transfer coefficients for flow boiling are
much larger than they are for pool boiling. Finally, before the convection becomes forced, the convective heat transfer
coefficients for nucleate pool boiling are greater than those for film pool boiling. To understand why this is true, we need
to examine the pool boiling curve next.
940 Two-Phase Nuclear Heat Transfer
FIGURE 24.2 A comparison of where flow boiling begins American PWRs and BWRs.
Normally, wall superheat is one of the most reliable indicators of when pool boiling begins. When the amount of super-
heat exceeds about 1°C, bubbles begin to form near the surface and their motion is driven primarily by buoyancy-related
forces. The bubbles begin to agitate the surrounding liquid, and this increases the convective heat transfer coefficient.
This form of pool boiling is called nucleate pool boiling. If the heat flux becomes even higher, a vapor film begins to
blanket the heated surface and the convective heat transfer coefficient begins to decline. This form of pool boiling is
called film pool boiling. Generally speaking, the convective heat transfer coefficients for pool b oiling are different
for horizontal surfaces than they are for vertical ones. Thus, flat plates have different expressions for the convective
heat transfer coefficient than vertical plates, and cylindrical surfaces have different expressions for the convective heat
transfer coefficient than flat surfaces do. In general, the value of h is affected by both the surface roughness and its
micro-geometry. Moreover, it depends on the wetability of the surface, which we discussed in Chapter 23. Finally, dif-
ferent correlations are required for nucleate pool boiling than film pool boiling. In general, these correlations attempt to
emulate the shape of the pool boiling curve, whose shape is shown in Figure 24.3.
24.9
T he Pool Boiling Curve
The first comprehensive studies of pool boiling were conducted in the eighteenth century when German physician and
theologian Johann Leidenfrost (see Section 24.17) observed the creation of a vapor film over a heated horizontal sur-
face. However, the first systematic studies of this effect were conducted by Shiro Nukiyama in 1934 when he studied
the effect of very large temperature differences between the bulk temperature of the fluid and the surface temperature
on the rate of bubble formation and growth. However, it was not until 1948 that this process was completely understood
24.10 Convective Heat Transfer During Pool Boiling 941
FIGURE 24.3 The pool boiling curve for water for a horizontal heated surface at atmospheric pressure. In this picture, the
Leidenfrost point is a function of the type of heated surface and its roughness.
when Farber and Scorah conducted a series of experiments to measure the heat flux in a boiling pool as a function of the
degree of wall superheat ΔTwall = Twall − Tsat. Their work then became the basis for what is called the pool boiling curve
today. An example of this curve for water at atmospheric pressure is shown in Figure 24.3. (Other fluids have similar but
not identical curves.) Farber and Scorah proposed that the pool boiling curve could be subdivided into at least four dif-
ferent regions, or regimes, based on the observable patterns of vapor production. Each regime then behaved as if it had
a different set heat transfer coefficients. In particular, they subdivided the pool boiling curve into four specific regions
or zones which they described as follows:
1. The natural or liquid convection region, where the coolant is a subcooled liquid
2. The nucleate boiling region, where bubbles form on the heated surface and are ejected into the coolant
3. The transition boiling region, where an unstable vapor film develops on the surface. This film may oscillate
between a state of nucleate boiling and a state of film boiling.
4. The stable film boiling region, where a stable vapor film forms on the surface and continues to stick to the surface
until it is swept away
These boiling regimes are plotted as a function of the surface heat flux and the wall superheat in Figure 24.4. Notice that
the surface heat flux in a boiling pool reaches a maximum value called the Critical Heat Flux (or CHF) at the end of the
nucleate boiling regime. For water at atmospheric pressure, the CHF is reached when the wall superheat approaches
30°C (or 55°F). Any additional increase in the wall temperature causes the surface heat flux to fall and the convective
heat transfer coefficient to decrease. Finally, when the amount of superheat approaches 120°C (~220°F), the liquid film
evaporates entirely from the surface of the wall and only a vapor film remains. In convective heat transfer, the point at
which this occurs is called the Leidenfrost point. We will have more to say about the Leidenfrost point in Section 24.17.
24.10
Convective Heat Transfer During Pool Boiling
Now let us examine how the convective heat transfer coefficient behaves as we move along the pool boiling curve. In general,
h can change by at least an order of magnitude between the time the boiling starts and the time a stable vapor film develops.
24.10.1
T he Liquid Convection Region
In region 1 (see Figure 23.4), the wall temperature is too low for bubble formation to occur. The coolant remains a
subcooled liquid, and convective heat transfer occurs due to density differences between the fluid at the surface of the
942 Two-Phase Nuclear Heat Transfer
FIGURE 24.4 The regions of boiling on the pool boiling and flow boiling curves.
wall and the fluid in the remainder of the pool. Small plumes of heated liquid occasionally rise from the surface. The
flow may be either laminar or turbulent, depending on the initial temperature of the coolant. However, if the number of
plumes is high, the flow is generally turbulent. Before boiling begins, the convective heat transfer coefficient is propor-
tional to the Rayleigh number (Ra) raised to a specific power N (see Chapter 22):
For laminar flows, the value of N is about 0.25, and for turbulent flows, it is about 0.33. To a certain extent, the value of
N is geometry dependent. For the same Rayleigh number, the heat transfer coefficient is usually larger for horizontal
surfaces than it is for vertical ones. Moreover, the value of C is geometry dependent—it can vary from about 0.05 to
0.50 for heated vertical surfaces. However, for nuclear fuel rods, it is usually less than 0.1 and typically closer to 0.05.
24.10.2
T he Nucleate Boiling Region
In region 2 of the pool boiling curve, nucleate boiling begins. The nucleate boiling region can be further subdivided into
two additional regions which Farber and Scorah called
1. The subcooled nucleate boiling region
2. The bulk nucleate boiling region.
These regions are distinguished from each other by the amount of wall superheat. At point 1, bubbles begin to form on
the heated surface. However, if the amount of superheat is low (say ΔTwall ≤ 5°C) and the bulk temperature of the fluid
is less than the saturation temperature, they simply collapse back into the colder fluid. This process is referred to as
subcooled nucleate boiling. The heat transfer coefficient in this region is greater than it is for free convection, and it is
generally a factor of 2 to 3 greater for water. At point B in Figure 24.4, the bulk nucleate boiling regime begins. Here,
the wall superheat becomes so high (ΔTwall = 5°C–30°C for water) that the bubbles no longer collapse back into the sur-
rounding liquid. When the amount of surface superheat exceeds about 10°C, additional nucleation sites become active,
and more bubbles begin to form. If the rate of bubble formation becomes high enough, the bubbles merge with each
other to form larger bubbles, voids, and slugs. The rate of bubble formation is high enough that the fluid near the surface
of the wall becomes highly agitated, and the heat transfer coefficient rises dramatically. The heat transfer coefficient
in this region is greater than it is for subcooled nucleate boiling, and it can become 10–100 times greater than it is for
single-phase flow. Many industrial processes take advantage of this increase in the heat transfer coefficient to remove
heat from the heated surface(s). Eventually, at about 30°C of surface superheat, the density of the bubbles becomes so
great that they begin to merge together to form columns and slugs of vapor, which decrease the contact area between the
heated surface and the coolant. This happens for both horizontal and vertical surfaces.
24.10 Convective Heat Transfer During Pool Boiling 943
q′′CRITICAL = Ch lv σgρv 2 ( ρl − ρv )
0.25
(24.5b)
where C is a constant whose value depends on the geometry of the heated surface. Approximately 10 years later, Zuber
arrived at a similar result using an entirely different approach. Values of C for different heated surfaces are shown in
Table 24.2. For flat plates, the value of C is 0.15, and for flat cylinders and spheres, it is about 0.12 and 0.11, respectively.
The value of C also depends on whether the heated surface is large or small. Equation 24.5b gives the CHF in W/m2 when
the other parameters are expressed in SI units. Notice that the CHF is independent of the fluid/surface combination, the
thermal conductivity, the viscosity, and the specific heat of the liquid. Moreover, since the CHF is directly proportional
to the enthalpy of vaporization hlv, large CHFs can be obtained using reactor coolants with a large enthalpy of vaporiza-
tion such as water. This is important in the design of boiling heat transfer equipment, where it is important to know the
maximum heat flux that can be withstood before burnout occurs. Finally, heat exchangers are classified as being large or
small depending on the value of what is called their characteristic dimension L′. This characteristic dimension is given by
L ′ = L g ( ρl − ρv ) σ
0.50
(24.5c)
where L is the physical dimension of the object. Hence, for cylinders, L corresponds to the radius or diameter of the cylin-
der, and for spheres, it corresponds to the radius or diameter of the sphere. Values of L′ for large and small heated objects are
shown in Table 24.2. Note in particular that large, flat surfaces have values for L′ greater than 25, whereas small cylinders
and thin wires have values for L′ as low as 0.15. Hence, a large range of values is theoretically possible in reactor work.
Example 24.2 calculates the maximum critical heat flux for an object submerged in a pool of boiling water. For large values
of the heat flux, the heating elements in the pool can burn out or fail. Of course, the failure of the surface can be prevented
by adding a coolant pump. It is then up to the designer of the pool to determine whether or not a pump is required
TABLE 24.2
Values of the Coefficient C Used to Determine the Critical Heat Flux in Pool Boiling Applications
24.10.4
T he Transition Boiling Region
Now let us turn our attention to what happens when the amount of wall superheat is increased even further. Beyond
point C, where the critical heat flux is reached, the rate of bubble formation exceeds the rate of bubble detachment, and
the bubbles merge together to form a continuous vapor film that covers small portions of the surface. This region is
called the transition boiling region, and it corresponds to region 4 of the pool boiling curve. In this region, the vapor
film is not particularly stable. Hence, it can temporarily detach from the surface before reattaching to it, and this leads
to a situation where colder fluid briefly comes in contact with the surface. The fluid flashes (or vaporizes) and then reat-
taches to the surface again. When this occurs, the surface temperature can oscillate violently because the heat transfer
rate is sometimes governed by the heat transfer coefficient for nucleate boiling and at other times by the heat transfer
coefficient for film boiling. This process can continue for some time until the surface superheat becomes large enough to
create a stable vapor film. Here, the heat transfer coefficient reaches its lowest possible value, and the heat flux reaches
its minimum value, q′′MIN . This point is called the Leidenfrost point, and it marks the end of the transition boiling regime.
Thus, the Leidenfrost point is equivalent to point D in Figure 24.4, and the degree of superheat at that point is called the
Leidenfrost superheat. We will have more to say about the Leidenfrost effect in Section 24.17.
24.10.5
T he Film Boiling Region
At wall temperatures above the Leidenfrost point, the liquid remaining in the pool is completely separated from the
wall by a stable vapor film. At this point, the boiling is no longer in transition, and one enters what is called the film
boiling regime. Here, the phase change occurs at the liquid–vapor interface, instead of directly at the heated surface.
The thermal energy from the heated surface reaches the liquid–vapor interface by convection in the vapor film as well as
by radiative heat transfer, and if the coolant pumps are turned off, the radiative heat transfer rate can exceed the con-
vective heat transfer rate. In this regime, the surface heat flux becomes a monotonically increasing function of ΔTwall.
Pool boiling continues until the surface reaches its maximum possible temperature. Here, the heated surface can either
melt or catastrophically fail. This point corresponds to point E on the pool boiling curve. However, the heat transfer
coefficient is higher than it is at the Leidenfrost point (point D).
24.11
Correlations for the Convective Heat Transfer Coefficient during Pool Boiling
Along the pool boiling curve, the heat transfer coefficients are flow regime dependent, and for pool boiling, they are
determined by the liquid-to-vapor ratio at the wall surface. The heat transfer coefficients increase until the critical heat
flux (CHF) point or departure from nucleate boiling (DNB) point (point C) is reached. Their values then fall through
the transition boiling regime until the Leidenfrost point (point D) is reached. Then, they increase again with increasing
superheat because the effects of radiative heat transfer become important. We would now like to discuss how the convec-
tive heat transfer coefficients are affected by the wall superheat ΔTwall. We would then like to discuss how the convective
heat transfer coefficients behave for the flow boiling instead of pool boiling.
24.12
Convective Heat Transfer Coefficients Prior to Nucleate Boiling
Prior to point A, the amount of wall superheat is too low for nucleate boiling to occur. This region of the pool boiling
curve is called the subcooled liquid region. The only form of convection that is possible in this region is natural convec-
tion. For flat heated surfaces immersed in a large pool, the heat transfer rate can be deduced from the Fishenden and
Saunders correlation:
24.13 CONVECTIVE HEAT TRANSFER COEFFICIENTS 945
where Ra is the Rayleigh number of the liquid (see Chapter 22), and all the properties used to find the Rayleigh number,
including the thermal conductivity k, are assumed to be those of the liquid:
(
Ra = gβ ( TWALL − TSAT ) L3 u l k l c pl ρ1 2 (24.7))
In terms of the wall superheat, the Rayleigh number is
( )
Ra = gβ ⋅ ∆Twall L3 u l k l c pl ρ1 2 (24.8)
The Fishenden and Saunders correlation is valid for Rayleigh numbers between 10,000,000 and 10,000,000,000.
In these expressions for the Rayleigh number, L is the “length” of the heated surface, and it is usually calculated by
assuming that L = A/P, where A is the surface area, and P is the perimeter. β is the thermal expansion coefficient of the
liquid, which in this case is assumed to be water. When using the Fishenden and Saunders correlation, the heated surface
is assumed to be oriented horizontally rather than vertically. For vertically oriented surfaces, the expression for the
heat transfer coefficient is different.
24.13
Convective Heat Transfer Coefficients during Nucleate Boiling
Following point A on the pool boiling curve, bubbles begin to form at the surface of the wall. In this regime, the con-
vective heat transfer coefficient becomes larger because the bubbles that form tend to increase the amount of thermal
mixing. Thermal energy is then transferred from the hot wall to the surrounding fluid to create these bubbles.
24.13.1
T he Rohsenow Correlation
One of the most popular correlations to predict the heat transfer coefficient under these conditions is the Rohsenow correla-
tion, which was proposed by Warren Rohsenow, an acquaintance of the author at the Massachusetts Institute of Technology
(MIT) in the 1950s. This correlation is probably the most widely used correlation for nucleate pool boiling today. However,
it does NOT work well for flow boiling, which we will discuss next. The exact form of Rohsenow’s correlation is
The Heat Transfer Coefficient for Nucleate Boiling (with the Pumps
Turned Off) Rohsenow Correlation (Pool Boiling Version)
h NB = u1h lv g ( ρl − ρv ) σ
0.5 3
c pl γ h lv Pr1 ∆TWALL (24.9a)
n 2
where
The heat flux q′′NB when nucleate boiling occurs can then be obtained from the following equation
or
Nu(L) = Ja 2 ⋅ Bo(L)0.5 ⋅ Pr −2 γ (24.11)
and Bo is the dimensionless Bond number (not to be confused with the movie character James Bond)
Here L is the characteristic length of the surface if it is a heated vertical plate. The Bond number (Bo) is also called the
Eötvös number (Eo), and it is used to measure the relative importance of gravity to surface tension when characterizing
the shape of the bubbles in the flow stream. Notice that the heat transfer coefficient for pool nucleate boiling is propor-
tional to the square of the wall superheat, which means that to a first approximation, the heat transfer coefficient h rises in
direct proportion to ΔTwall2. Here, γ is an empirical constant that depends on the liquid surface combination. For a boiling
pool of water near atmospheric pressure, the recommended value for γ is 0.013 for stainless steel. For other metals, it can
be higher or lower (see Cengel 2003). Hence, γ is meant to capture the effect of surface roughness (e.g., the number of
nucleation sites and crevices) on the rate of bubble formation. Rohsenow’s correlation can be used for subcooled nucleate
boiling as well as bulk nucleate boiling. Hence, it can be used to find the convective heat transfer coefficient anywhere
between points A and C on the pool boiling curve. The data points Rohsenow used to develop his original correlation are
shown in Figure 24.5. Not surprisingly, additional experimental data has become available over the years.
24.13.2
T he Zuber and Cooper Correlations
Occasionally, the location of the critical heat flux point (point C) can be difficult to find, and when this occurs, the CHF
in a boiling pool can be found using another correlation called the Zuber correlation:
24.13 CONVECTIVE HEAT TRANSFER COEFFICIENTS 947
FIGURE 24.5 The original data points used by Warren Rohsenow to develop his pool boiling correlation for aluminum, brass, and
stainless steel.
This correlation can be used for a boiling pool of water with a heat source imbedded in the pool. It gives the critical heat
flux as a function of the fluid pressure and the wall superheat. Notice that the critical heat flux increases as the pressure
increases, and this trend is also shown in Figure 24.6. However, after the critical heat flux point has been reached, the
flow transitions from nucleate boiling to film boiling and a different correlation for the heat transfer coefficient must
be used. Finally, for flat plates and horizontal cylinders, a simpler correlation called the Cooper correlation is some-
times used. The Cooper correlation is an interesting correlation because it can be used for fluids other than water. It is
applicable for pressures up to about 90% of the critical point. In other words, it can be used for almost all of the pressures
encountered in reactor work (0.1–16 MPa). The exact form of the Cooper correlation is
h = Ap1 α − log ( p1 )
−0.55
M −0.50 (q′′)0.67 (24.15)
where p1 is defined as the ratio of the local pressure p to the pressure at the critical point, that is, p1= p/pCR, ε is the surface
roughness (in microns where 1 micron = 10 −4 cm or 10 −6 m), α = 0.12 − 0.21 log (ε), M is the molar mass of the coolant
(which is 18 for water), A is a geometry factor equal to 55 for flat plates and 95 for horizontal cylinders, q″ is the heat
flux (in W/m2), and h is the convective heat transfer coefficient (in W/m2 °K) Hence, in the Cooper correlation, the con-
vective heat transfer coefficient is proportional to the two-thirds power of the heat flux q″. For water, the pressure at the
critical point is 217.7 atm (~22.1 MPa). Hence, at atmospheric pressure, the value of p1 is p1 = (1/217.7) = 0.0046, and at
1,000 PSI (~6.9 MPa), it is p1 = (6.9/22.1) = 0.312. Consequently, the critical pressure ratio p1 = p/pCR is a dimensionless
number that is greater than zero but less than 1.0.
FIGURE 24.6 The CHF versus the surface superheat for a platinum wire in a pool of boiling water at different pressures.
heat transfer coefficient from the surface of a horizontal tube in a reactor steam generator? Under what conditions would
it possibly apply? How accurate would you expect it to be under these conditions?
Solution As we stated earlier, the Cooper correlation can be used for hot stationary cylinders and horizontal plates
immersed in large pools of water and similar liquids. When there is no superimposed flow (i.e., the flow is not forced),
the convective heat transfer coefficient depends on the critical pressure ratio p1 = p/pCR, the surface roughness ε, and the
molar mass of the coolant. If ε is one micron (1 × 10 −6 m), the molar mass is 18, and the critical pressure ratio is 0.312,
the convective heat transfer coefficient is h = Ap1α[−log (p1)]−0.55M−0.50(q″)0.67 = 55 × (0.312)1.38 × 1.46 × 0.238 × (q″)0.67 =
55 × 0.20 × 1.46 × 0.238 × (q″)0.67 = 55 × 0.069 × (q″)0.67 = 3.82(q″)0.67. When the heat flux is 1 × 105 W/m2, the value of h
is 3.83 × 2,238.7 = 8,574 W/m2 °C (or 1,453 BTU/h ft2 °F). In principle, the Cooper correlation could be used to calcu-
late the convective heat transfer coefficient in steam generator tubes if they were horizontal. However, the flow must be
unforced, and this can only occur during natural circulation conditions. Finally, since the tubes in steam generators are
bound together into very tight bundles, a bundle correction factor (see Section 24.22) would have to be used to deter-
mine the actual boiling heat transfer coefficient. [Ans.]
24.14
T he Evaporation Rate from a Boiling Film
In nuclear power plants, tremendous amounts of steam are produced every second. This requires a very large amount
of thermal energy to be produced in the core. The transition of at least part of the water in the NSSS into steam can be
thought of as forced evaporation where a net transfer of mass between the two phases occurs. The mass transfer rate
then depends on the heat transfer rate. The constant of proportionality between the heat transfer rate Q = Aq″ and the
EVAP is called the latent heat of vaporization which is represented in the discussion that follows by
mass transfer rate m
the symbol hlv. Expressed quantitatively, the evaporation rate can be written as
where all of the parameters in Equation 24.16(a) must be expressed in the same unit system. For example, if the heat
transfer rate is required in W (or J/s), and the enthalpy of vaporization is known in J/kg, the evaporation rate will be
given in kg/s. Example 24.5 shows how the evaporation rate is affected by the local pressure. In particular, it shows that
the evaporation rate increases as the pressure is raised. At atmospheric pressure (or 14.7 PSI), the value of hlv for water
is 2,257 kJ/kg, whereas at 1,050 PSI (7.25 MPa), it is 1,500 kJ/kg. Thus, the evaporation rate is inversely proportional to
the value of hlv. In reactor work, Equation 24.16(a) can be used to find the evaporation rate from a spent fuel pool if the
24.15 Convective Heat Transfer Coefficient for Transition Boiling and Film Boiling 949
water in the pool happens to boil (see Section 24.19). It can also be used to determine the response of a reactor contain-
ment building to a loss of coolant accident because the condensation of the steam on the containment building walls
can be used to reduce the pressure level in the containment building following a LOCA. (see Chapter 34). Then when
condensation occurs instead of evaporation, Equation 24.16(a) can be written as
COND = Q/h lv
m (24.16b)
and this latter equation can be used to find the condensation rate instead of the evaporation rate.
24.15
Convective Heat Transfer Coefficient for Transition Boiling and Film Boiling
On the pool boiling curve, nucleate boiling stops at point C where transition boiling and then film boiling start. Here,
the amount of wall superheat for water is about 30°C if the fluid happens to be water at atmospheric pressure. Between
points C and D (see Figure 24.4) transition boiling and film boiling occur, the heat transfer coefficient can be expressed
as the sum of two terms:
☉☉ A convective heat transfer term
☉☉ A radiative heat transfer term
The convective term dominates the value of the heat transfer coefficient prior to the Leidenfrost point (between points
C and D), and the radiation term dominates the heat transfer coefficient after that (between points D and E). Hence, we
can write the total heat transfer coefficient as
The Convective Heat Transfer Coefficient in Transition Boiling and Film Boiling
h = h CONV + Wh RAD (24.16c)
where W is a weighting factor that expresses the relative importance of the two terms. Normally, W is a function of the
local quality x and the degree of wall superheat ΔTwall. For water, the Leidenfrost point is reached when the amount of
superheat reaches 150°C to 320 °C. The exact value depends on the material composition of the wall, the surface rough-
ness, and the degree of subcooling of the surrounding fluid. A range of Leidenfrost temperatures for many different
materials can be found at
https://engineering.purdue.edu/BTPFL/BTPFL%20Publications/81.pdf
950 Two-Phase Nuclear Heat Transfer
Thus, for the same heat flux, the Leidenfrost temperature is material dependent. In the transition and film boiling
regimes, the convective heat transfer coefficient can be found from what is called the Berenson correlation. The Berenson
correlation is normally written as
where ε is the surface emissivity having a value between 0 and 1, W = 0.75 and
γ = √ σ g ( ρl − ρv ) (24.19)
Here, the surface emissivity is a dimensionless number which represents the ratio of the radiation emitted by a hot object
at a given temperature to the radiation emitted by a blackbody at the same temperature. The heat transfer coefficient
increases beyond the Leidenfrost point because the radiative term becomes more important than the convective term.
Notice that the heat transfer rate in this region rises in direct proportion to the fourth power of the wall temperature
difference. Expressions of this type almost always appear in radiative heat transfer when a hot body emits thermal
radiation in the form of photons. This effect can also be described by the Stefan–Boltzmann law which we discussed
in Chapter 10. To apply this law, the absolute temperature of the heated surface that emits the radiation must be used.
Eventually, the vapor film becomes stable, and the pool boiling regime ends. Notice that only three correlations—the
Fishenden and Saunders correlation, the Rohsenow correlation, and the Berenson correlation were needed to cover the
entire range of conditions encompassed by the pool boiling curve. Now let us turn out attention to the behavior of
the heat transfer coefficient along the pool boiling curve.
24.16
Behavior of the Convective Heat Transfer Coefficients along the Pool Boiling Curve
In a boiling pool, the correlations for the convective heat transfer coefficients attempt to mirror the shape of the pool
boiling curve. Since h = q″/ΔT, the convective heat transfer coefficients become equal to the heat flux q″ divided by
the amount of wall superheat ΔT at each point along the curve. This implies that the heat transfer coefficients become
largest when the CHF point (point C) is reached, and they become smallest when the Leidenfrost point (point D) is
reached (see Figure 24.4). Thus, they are higher for nucleate boiling than they are for transition boiling, and they are
higher for transition boiling than they are for film boiling. However, eventually a point is reached (at very high levels of
film boiling) where they increase again because the radiative heat transfer term dominates the convective term. We can
summarize this behavior by drawing the following conclusions:
The Relative Sizes of the Heat Transfer Coefficients along the Flow Boiling Curve
h SP < h TB < h NB (for modest amounts of wall superheat)
(24.20)
h FB ≥ h TB (for very large amounts of wall superheat)
24.17 The Leidenfrost Effect 951
FIGURE 24.7 A picture of Johann Leidenfrost (left) who first explored the Leidenfrost effect.
Now let us examine the behavior of the fluid when the Leidenfrost point is reached.
24.17
T he Leidenfrost Effect
In all pool boiling applications, a point is eventually reached where the liquid film completely dries out and all that
remains on the heated surface in a vapor barrier. This stable vapor barrier only appears when the amount of superheat
is so high that no more fluid can touch the heated surface. Then, the heat flux reaches its minimum value at a point
(point D) which is called the Leidenfrost point. The Leidenfrost effect occurs on both vertical and horizontal surfaces,
and historically it was first studied by Johann Leidenfrost (see Figure 24.7) about 200 years before the first nuclear
power plants were built. However, the Leidenfrost effect does not apply to just nuclear power plants. Its effects can also
be seen in a frying pan on your stove. If you sprinkle a few drops of water on a hot s killet whose surface temperature
is well above the boiling point, the drops of water will skitter or “dance” across the surface, and in doing so, they will
take much longer to evaporate when the surface temperature is far above the saturation temperature. Then, the amount
of time tEVAPORATION it takes for a drop of water to evaporate from the surface depends on the temperature difference
ΔT WALL between the heated surface T WALL and the boiling point TBOIL (or TSAT):
The temperature at which the water droplet stays on the surface for the longest period of time before evaporating is called
the Leidenfrost temperature, and on the pool boiling curve, this point corresponds to the Leidenfrost point (point D) in
Figure 24.4. Here, the convective heat transfer coefficient reaches its minimum value, and if the temperature of the sur-
face becomes higher or lower, the heat transfer coefficient will actually increase. The vapor film blankets the hot surface
most effectively at this single temperature. In reactor work, the Leidenfrost point signifies the onset of stable film boiling
on the surface of the cladding. It represents the point on the pool boiling curve where the heat flux is at a minimum, and
where the hot surface is completely covered by a stable vapor film that prevents any additional liquid from touching the
cladding. Here, the heat transfer from the hot surface to the remaining liquid occurs through a combination of radiation
and conduction through the vapor barrier. These are the two processes that the Berenson correlation (Equation 24.17)
attempts to model. As the surface temperature increases beyond the Leidenfrost point, radiation through the vapor film
becomes more important and the heat flux q″ = q/A continues to rise. The heat flux where the Leidenfrost point occurs
can be found for simple objects like horizontal plates using a modified form of Zuber’s equation (see below). For liquids
other than water, and for water at very low pressures, a commonly used form of Zuber’s correlation that can be used to
determine where the heat flux reaches its minimum value is
where the properties in Equation 24.22 must be evaluated at the saturation temperature TSAT. Here, Zuber’s constant C
has a value of approximately 0.09 for most liquids at moderate pressures. Note that the form of Zuber’s correlation we
have just p resented is slightly different than Zuber’s original correlation, which can be commonly found in the litera-
ture. In g eneral, this form of the correlation is a better suited to reactor work at pressures below 6.9 MPa (~1,000 PSI).
However, at pressures above 1,000 PSI, the original form of Zuber’s equation becomes more accurate.
24.18
Effects of Surface Roughness on the Heat Transfer Coefficient
The heat transfer coefficients along most of the pool boiling curve are a function of the surface roughness ε, and they
are a function of the material from which the surface is made. Hence, they are different for smooth surfaces than they
are for rough ones. In almost all pool boiling applications, a term such as γ (see Berenson’s and Rohsenow’s correla-
tions) is added to the pool boiling correlations to express the fact that rough surfaces have higher heat transfer rates
than smooth surfaces. Of course, this is because the heat transfer rate in the nucleate boiling regime depends on the
number of nucleation sites on the surface. Therefore, increasing the surface roughness increases the number of nucle-
ation sites where bubbles and small pockets of superheated vapor can form. Around 1950, Berenson demonstrated that
the heat flux in the nucleate boiling regime could be increased by a factor of about 10 by intentionally roughening a
smooth surface.
For example, when a hot fuel rod or heating element is placed in a stationary pool, the heat transfer coefficient can
be increased by machining more cavities or pores into the surface (see Figure 24.8). However, these higher heat transfer
rates cannot be sustained indefinitely because the cavities and nucleation sites eventually become filled with minerals
that plate out of the coolant as it boils. Then, the heat transfer rate eventually drops to levels that are more indicative of
smooth surfaces. However, Berenson discovered that surface roughness has virtually no effect on the minimum CHF
in the film boiling regime. Because of these considerations and others, almost all heated surfaces in reactor cores are
designed to be smooth rather than rough. As a practical matter, fuel rods last longer and have lower failure rates when
their surfaces are smooth. Some of the reasons for these lower failure rates are discussed in our companion book (see
Masterson 2017).
FIGURE 24.8 The cavities on a rough surface can act as nucleation sites and increase the heat transfer rate in the nucleate boiling
regime. This is particularly true when a heating element or a hot wire is placed in a cooler s tationary pool. Unfortunately over time,
these cavities disappear as minerals are plated out of the water by the boiling process. Then, the heat transfer coefficients for pool
boiling are essentially the same for rough surfaces as they are for smooth surfaces.
24.20 Pool Boiling over Horizontal Tubes and Tube Banks 953
FIGURE 24.9 A spent fuel pool in which both natural convection and pool boiling can occur. (Picture provided by the United
States Department of Energy [US DOE].)
24.19
Pool Boiling in Spent Fuel Pools
Occasionally, the fuel rods in spent fuel pools (see Figure 24.9) can become hot enough for nucleate pool boiling and film
pool boiling to occur. However, spent fuel pools are normally designed with auxiliary pumps to pump cold water into the
pool and hot water out of it. As long as the pumps are running, the cold water added to the pool sets up natural circulation
loops and the coolant continues to stay a single-phase liquid. Then, the correlations in Chapter 22 can be used to find the
convective heat transfer coefficient. However, these pools require electric power to operate the pumps, and if the power
to the pumps is suddenly lost (which can sometimes occur during a reactor accident), the decay heat produced by the fuel
rods begins to build up in the pools. After a while, a point is reached where vapor bubbles begin to form on the surface
of the fuel rods, and this in turn causes some form of boiling to occur in the pool. Then, Rohsenow’s correlation and
Berenson’s correlation can be used to find the convective heat transfer coefficient for the nucleate boiling and transition
boiling regimes. However, the physical constants used in these correlations may have to be evaluated at different points
along the rods. If the rods have just been placed in the pool, then it is also possible for film boiling to occur.
24.20
Pool Boiling over Horizontal Tubes and Tube Banks
There are several places in the NSSS where one can find large numbers of heated horizontal and vertical tubes. These tubes
can be found in the plant’s heat exchangers where they are used to transfer heat from a high pressure fluid inside of the tubes
954 Two-Phase Nuclear Heat Transfer
to colder fluid which is designed to flow over the surface of the tubes. When the temperature of these tubes becomes high
enough, nucleate boiling can occur on the tube’s surface. Over flat plates, bubbles are formed at scattered nucleation sites,
and when they leave the surface of the plates, they take part of the superheated boundary layer with them. This causes an
inrush of colder fluid, and this inrush continues itself until the plates are cooled. The rate of convection can then be char-
acterized using the bubble diameter at departure (or the BDAD). However, the nucleation process is entirely different for
the fluid surrounding hot horizontal tubes. Nucleation occurs primarily on the lower surface of the tubes, and the bubbles
then slide parallel to the tube surface until they eventually detach and flow into the coolant. Thus, a bubble layer similar
to the one shown in Figure 24.10 is formed, and the void fraction increases from almost zero at the bottom of the tubes to
about 0.5 at the sides where the bubbles detach. When the bubbles detach from the surface, their upward velocity increases
as well. These processes cause a considerable variation in the value of the heat transfer coefficient around the surface of
the tubes. The Nusselt number for an individual heated tube surrounded by a fluid such as water can then be written as
where Re represents the Reynolds number for the vapor production rate in the bubbly layer surrounding the tube. Again,
C is a fluid dependent parameter, whose value can increase or decrease as the pressure of the fluid is changed. For water,
its value is approximately
C = 9.7 ( p1 ) (24.24)
0.5
The effect of pressure on the heat transfer coefficient can then be determined by an equation called the Mostinski equa-
tion that sometimes appears in pool boiling applications*. The most common form of Mostinski’s equation is
Mostinski’s Equation for the Effect of Pressure on the Heat Transfer Coefficient
F ( p1 ) = 1.8p1 0.17 + 4p1 1.2 + 10p1 10 (24.25)
This equation uses the Principle of Corresponding States to adjust the value of the heat transfer coefficient when the
ambient pressure changes. In Equation 24.25, p1 is the critical pressure ratio (see Section 24.13) and it represents
the ratio of the local pressure p to the pressure at the critical point, that is, p1 = p/pCR. The Mostinski equation can be
used for many different fluids including water. It can also be used for pool boiling applications (with no superimposed
flow) for tube diameters between 8 and 50 mm (0.3–2 in.) for both natural and “as machined” surfaces. It gives good
results for high and low pressure water and most industrial fluids. However, it is NOT appropriate to use for liquid
metals. Now let us discuss how the heat transfer coefficient is affected by these pressure changes.
* see Mostinski, I.L., “Application of the rule of corresponding states for calculation of heat transfer and critical heat flux”,
Teploenergetika, 4:66, 1963.
24.22 Pool Boiling over Heated Tubes in a Reactor Heat Exchanger 955
24.21
D ependence of the Heat Transfer Coefficient on the System
Pressure in the Nucleate Boiling Regime
As we just mentioned, the local pressure can effect the heat transfer coefficient when nucleate boiling occurs. At low
pressures, the rate of bubble formation over a hot surface is not affected very much by the pressure of the fluid imme-
diately above the surface. However, reactors operate at such high pressures that the rate of bubble formation is much
different than it is at low pressures, and this can have a surprisingly large effect on the value of the convective heat trans-
fer coefficient in a boiling pool. In 1963, Mostinski (see the discussion above) applied the Principle of Corresponding
States in an attempt to estimate the heat transfer coefficient for nucleate boiling in several reactor components. His
approach was quite innovative at the time because it did NOT require an explicit knowledge of the surface roughness or
the thermophysical properties of the fluid in the pool to determine the heat transfer coefficient. Instead, it just required
the system pressure and the local heat flux. His approach resulted in the following correlation for the convective heat
transfer coefficient in the nucleate boiling regime:
Sometimes this correlation is called the dimensional reduced pressure correlation because it contains a pressure
c orrection factor F(p1) that depends on the critical pressure ratio p1 = p/pCR. Notice that no physical properties of
the fluid are required to calculate the boiling heat transfer coefficient in this case! The heat transfer coefficient in the
Mostinski correlation then becomes a function of the critical pressure, the reduced pressure ratio, and the applied heat
flux q″ raised to the 0.7 power. The heat transfer coefficient is expressed in W/m2 °K, the applied heat flux q″ is expressed
in W/m2, the critical pressure of the fluid (pCR) is expressed in kN/m2 (i.e., in kPa), and the critical pressure ratio p1 =
p/pCR is dimensionless. The effects of pressure on the nucleate boiling rate are then given by Equation 24.25. Example
Problem 24.10 shows how much the heat transfer coefficient for water changes in a boiling pool when the pressure is
raised from atmospheric pressure (14.7 PSI) to the operating pressure of a commercial PWR (~2,250 PSI). Most other
fluids behave in a similar way.
24.22
Pool Boiling over Heated Tubes in a Reactor Heat Exchanger
Rod bundles in power plants can be positioned either horizontally or vertically to dissipate heat. Their orientation is deter-
mined by the design of the heat exchanger in which they are used. When a bundle of these tubes is positioned horizontally,
the convective heat transfer coefficients gradually increase from the bottom of the tube bank to the top. The individual
tubes within the bank also behave in this way. When the quality in the vicinity of the tubes is relatively low (say, less than
20%), the heat transfer rate becomes higher along the sides of the tubes because there is high velocity bubbly flow between
the tubes in the vertical direction and relatively motionless liquid in the horizontal direction (see Figure 24.11). Then, the
heat transfer coefficient for an average tube increases for low values of the wall superheat (typically less than 15°C) and
then decreases by a factor of at least two when the amount of superheat increases (see Figure 24.12). The heat transfer
coefficient for an average tube in the bundle can then be found by multiplying the heat transfer coefficient for a single tube
hST by a bundle correction factor FB and then adding a natural convection correction term to the result:
FIGURE 24.11 The flow of liquid and vapor around a bank of hot horizontal tubes during the pool boiling process.
FIGURE 24.12 A pool boiling curve for a horizontal tube bundle compared to that of a single horizontal tube for water.
The bundle correction factor FB in this case has a value of about 1.5, and hNC has a value of about 0.25 kW/m2 °C. This
approach works well for water and most industrial coolants For simpler geometric shapes, the heat transfer coefficients
can be deduced from the correlations shown in Table 24.3.
24.23
F low Boiling in Nuclear Power Plants
Flow boiling is more prevalent in nuclear power plants than pool boiling. In fact, it is dominant mode of convective
heat transfer in Boiling Water Reactors. In General Electric BWRs, the coolant velocity averages about 2 m/s at the
core inlet, 4 m/s at the core midplane, and 7 m/s at the core exit (see Appendix B). Most of this velocity change is due
to the conversion of the water into steam, and at the core exit, the flow quality can approach 12% to 17%. The exit void
fraction under these conditions is about 80%. (In LMFBRs, the coolant velocity can be even higher.) This enables a two
24.24 CONVECTIVE HEAT TRANSFER DURING FLOW BOILING 957
TABLE 24.3
Some Popular Pool Boiling Correlations and Their Intended Range of Applicability
Specialty Correlations
Kutateladze correlation Calculates the maximum ΔTWALL = ΔTCRIT No superimposed flow Water and organic
or CHF coolants
First Zuber correlation Calculates the maximum ΔTWALL = ΔTCRIT No superimposed flow Water
or CHF
Second Zuber correlation Calculates the minimum ΔTWALL = ΔTLEIDENFROST No superimposed flow Water
heat flux at the
Leidenfrost point
phase mixture to transverse the entire length of a fuel assembly in about 1 second. Under these conditions, flow boiling
occurs in a number of distinct stages that are conceptually similar to the behavior of the fluid in a boiling pool. When
the flow rate is high and the direction is vertical, these stages are shown in Figure 25.1 of Chapter 25. Usually, forced
convection encourages the process of bubble formation and growth, and because the flow is forced instead of natural,
it is not unusual for, the convective heat transfer coefficients to be much larger than they are for pool boiling. We will
have more to say about this in the next section.
24.24
Convective Heat Transfer during Flow Boiling
As we just mentioned, the heat transfer coefficients for flow boiling can be much greater than they are for pool boiling.
There are a number of reasons why this is true. First the additional convective forces caused by the superimposed flow
disrupt the rate of bubble formation at the fuel rod surface. This alters the shape of the pool boiling curve and the values
of the heat transfer coefficients prior to burnout. However, the flow boiling curve is still qualitatively very similar to the
pool boiling curve (see Figure 24.13). As the velocity of the imposed flow increases, the critical heat flux also increases
and the point at which it occurs shifts further to the right (where the wall superheat becomes greater). Thus, the amount
of wall superheat required to reach the CHF point increases as well. Dozens of correlations have been developed over the
years to simulate this behavior, but most of them are limited to just one or two flow regimes. When we then examine the
forced flow of water through a vertically heated coolant channel with a uniform heat flux, the boiling process resembles
the one shown in Figure 24.14. Exactly what happens next depends on the value of the heat flux q″. If the heat flux is high
enough, the boiling behaves in one way, and if the heat flux is lower, the boiling behaves in another. So in flow boiling,
the ratio of the heat flux q″ to the mass flux G (q″/G) determines the way in which the boiling occurs. Hence, are two
basic scenarios that can occur during flow boiling:
☉☉ Flow boiling with a low heat flux-to-mass flux ratio (q″/G)
☉☉ Flow boiling with a high heat flux-to-mass flux ratio (q″/G)
where, G is the imposed mass flux and G = m/A.
The first scenario leads to an undesirable condition called film dryout,
and the second scenario leads to an equally undesirable condition called departure from nucleate boiling or DNB.
Film dryout can lead to rod failures in BWR fuel assemblies and DNB can lead to rod failures in PWR fuel assemblies.
Moreover, both of these scenarios become more likely as the mass flux is lowered or the heat flux is raised. Each of these
scenarios is depicted in Figure 24.15. Normally film dryout occurs much more gradually than DNB does. Collectively,
958 Two-Phase Nuclear Heat Transfer
FIGURE 24.13 The shape of the boiling curve as the velocity of the coolant increases. Notice that increasing the velocity of the
coolant shifts the curve to the right, and when the imposed velocity is zero, the flow boiling curve reverts to the pool boiling curve.
these two scenarios are responsible for what is called the boiling crisis. We would now like to discuss each of these
scenarios separately and then show how they are related.
24.25
F low Boiling on a Heated Surface with a Low Heat
Flux-to-Mass Flux Ratio (Scenario 1)
When the heat flux is relatively low compared to the mass flux, a fuel rod will dry out in a gradual manner. Before boiling
begins, the bulk temperature is less than the saturation temperature, and the coolant can be considered to be a subcooled
24.26 FLOW BOILING ON A HEATED SURFACE 959
FIGURE 24.15 The values of the heat flux and the mass flux determine when dryout occurs in PWRs and BWRs. When both the
heat flux and the mass flux are low, the coolant channels assume the flow patterns shown on the left, and these flow patterns are
usually associated with American BWRs. When both the heat flux and the mass flux are high, the coolant channels assume the flow
patterns shown on the right, and these flow patterns are usually associated with American PWRs.
liquid with xeq ≤ 0. Then, as more heat is added to the coolant, some bubbles begin to appear in the flow stream. These
bubbles initially appear close to the surface of the wall, but they quickly collapse back into the surrounding liquid. As
more heat is added to the coolant, more steam is produced and the bulk temperature of the fluid reaches its saturation tem-
perature (TSAT). Then, large bubbles or voids begin to develop in the flow channel, and instead of collapsing back into the
flow stream, they merge together to form larger bubbles and voids. The temperature of the fluid stays at the saturation tem-
perature, but the quality starts to increase. Eventually, the flow transitions itself from bubbly flow (where the flow quality
is relatively low) to slug flow (where the flow quality has an intermediate value) and finally to annular flow (where the flow
quality is relatively high). If more heat is added to the coolant, the liquid film on the surface of the rods evaporates, and
only a high temperature mist remains. This type of flow is then called mist flow. The cladding temperature rises dramati-
cally as the heat transfer coefficient falls, and this leads to a condition called film dryout. Film dryout is one of the most
common causes of rod failures in BWRs. When the liquid film is disappears, the heat transfer coefficient falls sharply and
the cladding temperature doubles or triples. The wall temperature spikes and the cladding can fail or melt. This behavior
can be seen in Figure 24.16. Further heating of the mist flow causes the water droplets inside of the vapor core to evapo-
rate, and eventually, all we are left with is a superheated vapor. At this point, the flow quality reaches 100% (x = 1.0) and
cannot increase any further. However, the coolant temperature can still rise if more heat is added to the channel.
24.26
F low Boiling on a Heated Surface with a High Heat
Flux-to-Mass Flux Ratio (Scenario 2)
Now let us turn our attention to what happens when the heat flux-to-mass flux ratio (q″/G) is much higher. This typi-
cally occurs in the fuel assemblies of PWR cores. Then as heat is added to a coolant channel, some bubbles form on
the surface of the fuel rods, but they immediately collapse back into the coolant if the coolant is subcooled. Then, as
the fluid continues to flow, and more heat is added to it, the bubbles that are produced tend to become so numerous
that they blanket the wall rather than collapsing or merging into larger bubbles and voids. If the applied heat flux q″
becomes high enough and the mass flux G becomes low enough, the boiling will become so intense at the wall sur-
face that the wall can quickly become blanketed with a vapor film, which prevents any colder fluid from reaching the
960 Two-Phase Nuclear Heat Transfer
FIGURE 24.16 The ratio of the heat flux to the mass flux (q″/G) determines when dryout occurs in PWRs and BWRs. If this ratio
is low, the fuel rods fail at relatively high qualities after the liquid film disappears from the surface of the rods. If this ratio is high,
the fuel rods fail much earlier and at much lower values of the equilibrium quality. In the figure above, failure occurs due to DNB
at qualities of about 15%, while rod failures occur due to dryout at qualities approaching 80%. To illustrate these two scenarios, the
mass flux and the pressure are assumed to be the same.
surface. This causes the heat transfer coefficient to fall dramatically, and it also causes the cladding temperature to
abruptly rise. The fluid condition at the wall surface is then referred to as the DNB condition. When the DNB point is
reached, fuel rod dryout typically occurs at much lower equilibrium qualities than in the fuel rods in a BWR. The heat
transfer coefficient also falls more abruptly, and unless the heat flux is reduced immediately, the fuel rods will cata-
strophically fail. In fact, the heat transfer coefficient falls in a much more dramatic fashion than it does during film
dryout (Scenario 1). The process of DNB is compared to film dryout in Figures 24.15 and 24.16. Further up the cool-
ant channel, the liquid core disappears, and only a flow of water droplets remains. These droplets remain entrained in
the vapor core. The temperature of the fluid stays at the saturation temperature, but the equilibrium quality continues
to rise. Eventually, the water droplets vaporize, and all that is left is a superheated vapor. Not surprisingly, DNB is the
most likely cause of fuel rod dryout in PWRs.
Throughout the world, various names have been given to scenario 1 and scenario 2. However, in this text, we will
simply refer to them as (1) film dryout and (2) DNB. And finally - just to be clear, they are completely different phenom-
ena, and they affect the design margins in PWRs and BWRs in different ways. Film dryout is more common in BWRs,
and DNB is more common in PWRs. We will discuss how they affect the design margins in Chapters 27 and 28. Hence,
to summarize what we have just learned:
☉☉ The coolant normally enters the fuel assemblies in both PWRs and BWRs as a subcooled liquid, and the amount
of subcooling is usually greater at the entrance to a PWR than it is at the entrance to a BWR. (The specific values
can be found by comparing the inlet enthalpies to the saturation enthalpies).
☉☉ Between a pure liquid and a pure vapor, the coolant can change its behavior in a number of different ways, and
a single-phase heat transfer correlation cannot be used to describe the convective heat transfer coefficient across
all of these regimes.
☉☉ The value of the convective heat transfer coefficient is different for nucleate boiling than it is for bulk boiling, and
it is also different for bulk boiling than it is for annular film boiling.
☉☉ Once a phase change occurs, each flow regime is dominated by a completely different set of physical parameters,
and the heat transfer coefficients for each regime are considerably different. For the same mass flux, the heat
transfer coefficient always increases when the coolant boils, and for water–steam mixtures, it is at least ten times
greater than it is for subcooled water.
☉☉ A small amount of boiling in a reactor core is not necessarily a bad thing as long as its location and its intensity
can be controlled.
24.27 TEMPERATURE DEPENDENCE 961
☉☉ The CHF is normally used to determine when dryout occurs in BWRs, and the DNBR is normally used to deter-
mine when dryout occurs in PWRs. These ratios represent different ways to determine when a fuel rod will fail
due to inadequate cooling. We will explain the meaning of these terms more fully in Chapter 27.
24.27
T he Temperature Dependence of the Heat Transfer
Coefficients in the Flow Boiling Regime
The behavior of the heat transfer coefficient during flow boiling is entirely different for nucleate boiling that it is for bulk
boiling. When the coolant is still a single phase liquid it is relatively easy to demonstrate that the heat transfer rate q is pro-
portional to the wall superheat ΔTWALL = (TWALL − TBULK), where TWALL is the wall temperature, and TBULK is the bulk tem-
perature. Thus, Newton’s Law of Convection can be used to determine the value of the convective heat transfer coefficient,
and the constant of proportionality between (TWALL − TBULK) and q is a unique variable which can be correlated to the size of
the Reynolds number. However, after boiling begins, the dependence of the heat transfer rate on the wall temperature differ-
ence changes. At the onset of nucleate boiling, the heat transfer rate for both flow boiling and pool boiling is proportional to
the square of the wall superheat ΔTWALL2 = (TWALL − TSAT)2. Then, when the nucleate boiling becomes fully developed, the
heat transfer rate becomes proportional to the cube of the wall superheat ΔTWALL3 = (TWALL − TSAT)3. Thus the heat transfer
rate becomes proportional to different powers of the superheat ΔTWALL, and we can summarize this behavior by writing
Flow Regime Heat Flow Behavior
I. Single-phase liquid q = C ( TWALL − TBULK )
q = C ( TWALL − TSAT )
2
II. Two-phase fluid (initial nucleate boiling)
(24.28)
= C ( TWALL − TSAT )( TWALL − TSAT )
q = C ( TWALL − TSAT )
3
III. Two-phase fluid (fully developed nucleate boiling)
where C is an arbitrary constant that changes from one flow regime to the next. Thus, for Newton’s Law of Convection to
be valid, the heat transfer coefficients (neglecting for the moment differences in the physical properties of the coolant),
must behave for the same flow rates as
These observations vividly demonstrate that the convective heat transfer coefficients for two-phase flows are always
higher than the convective heat transfer coefficients for single-phase flows and that they increase at first linearly and
then quadratically with the amount of wall superheat ΔT WALL = (T WALL − TSAT). This means that increasing the wall
temperature beyond the saturation temperature will increase the heat transfer coefficient as long as nucleate boiling
persists because the value of T WALL will continue to increase while the value of TSAT will not. As we shall see in
Chapter 25, there are a number of commercially available correlations that are able to model these effects.
In addition to the fact that nucleate boiling can occur during flow boiling, it is also possible for film boiling to occur
at higher flow qualities. Film boiling will not become stable until the equilibrium quality approaches 10%–15%. The
heat transfer coefficient then continues to rise until dryout occurs, but the rate of increase is less than it is in the nucleate
boiling regime. After dryout, the value of h falls dramatically (often by a factor of 10 and sometimes by a factor as great
as 100) over very short distances (typically on the order of a few centimeters). Figure 24.16 compares the behavior of the
heat transfer coefficient under these conditions.
heat flux be increased in this case? Assume that the predominant form of convective heat transfer is nucleate boiling at
the bottom of the pool.
Solution According to Rohsenow’s correlation (see Equation 24.9), the heat transfer coefficient increases with the square
of the wall superheat and the surface heat flux increases with the cube of the wall superheat. If the wall superheat is
doubled from 10°C to 20°C, the surface heat flux must increase by a factor of 8 if the same heating element is used. [Ans.]
24.28
Rod Dryout for Different Heat Flux-to-Mass Flux Ratios
Finally, the equilibrium quality xDRY and the location zDRY at which dryout occurs can be theoretically different when
the heat flux to mass flux ratio is high than when the heat flux to mass flux ratio is low. This can then affect how the heat
transfer coefficient behaves as a function of the equilibrium quality. A typical curve governing this behavior is shown
in Figure 24.16. The pressure and the mass flux in both cases are the same. Note in particular that the heat transfer
coefficient peaks much earlier in the process when the heat flux is high and at a much smaller value of the equilibrium
quality. This can happen if a vapor film blankets the surface of a very hot wall and separates the wall from the onrush-
ing coolant. The heat transfer coefficient also reaches a higher value before it begins to decline precipitously. However,
when the heat flux is lower, there is not enough heat for the vapor barrier to develop. The heat transfer coefficient peaks
much further downstream, and this peak is generally followed by the dryout of the iquid film. Finally, the heat transfer
coefficient does not become as large before the maximum value of h is reached. The smaller heat transfer coefficient is
due to the fact that the amount of surface superheat ΔT = T WALL − TSAT is lower if the heat flux is lower. So in addition
to everything we have just said, the size of q″, which we can correlate to ΔT, determines the size of the heat transfer
coefficient under these conditions.
24.29
Behavior of the Enthalpy, the Equilibrium Quality,
and the Void Fraction in Boiling Cores
In reactor cores, the value of h depends on the temperature, the enthalpy, the quality, and the void fraction at each axial
elevation. Their values can change as a function of position, and they in turn depend on the shape of the axial heat flux
q″(z). If the heat flux is co-sinusoidal in shape, the energy deposition in the coolant is also co-sinusoidal. In General
Electric BWRs, this causes the coolant channels to segment into two distinct regions called a boiling length and a non-
boiling length. The non-boiling length contains a subcooled liquid, and the boiling length contains a two-phase mixture.
Inside the boiling length, the velocities of the liquid and vapor phases are usually different. Initially, the coolant enters
the core as a subcooled liquid, and for about one meter, its equilibrium quality is less than zero. The equilibrium quality
before the zero quality point is reached can be found from
where hlv is the enthalpy of vaporization at the saturation point, hl is the enthalpy of the liquid phase at the saturation
point, and h(z) is the enthalpy of the coolant as a function of axial position. In commercial BWRs, the coolant inlet tem-
perature is about 270°C, and the saturation temperature is about 286°C. At operating pressures of 1,050 PSI (~7.25 MPa)
and temperatures of 270°C, the specific enthalpy of water is 1,180,000 J/kg, while at the saturation temperature, it is
1,270,000 J/kg. The enthalpy of vaporization is approximately 1,540,000 J/kg. Hence, the equilibrium quality at the core
inlet is approximately
Then, as more heat is added from the fuel rods, the enthalpy starts to rise, and eventually, its bulk temperature becomes
equal to the saturation temperature (TBULK = TSAT). When this occurs, the coolant begins to boil vigorously although
some subcooled boiling may occur long before this point is reached. Figure 24.17 shows the temperature and enthalpy
profiles under these conditions. In American BWRs, the non-boiling length is about 1 meter, and the boiling length is
about 3 m. Thus, the active core height,* which is the sum of these two distances, is approximately 4 m (1 + 3). As more
heat is added to the coolant, its quality and its void fraction begin to rise. Figures 24.18 and 24.19 depict how these values
* In general, the active core height is the part of the core where heat is added to the coolant due to heat generation in the nuclear
fuel rods.
24.29 BEHAVIOR IN BOILING CORES 963
FIGURE 24.17 The temperature and the enthalpy in an American BWR as a function of axial elevation. Here, h is the average
enthalpy in the core, and the heat flux is assumed to be co-sinusoidal.
FIGURE 24.18 The equilibrium quality as a function of axial elevation in a modern BWR. Here, the axial heat flux is assumed to
be co-sinusoidal. Note that when the coolant first enters the core, it is highly subcooled, and its equilibrium quality is negative.
behave in a steady-state core. Each graph represents the average value of these parameters at each axial elevation. Thus,
the quality represents the equilibrium quality xeq in a thermodynamic sense. However, bubbles may form close to the
surface of the fuel rods long before the bulk temperature reaches the saturation temperature. In core thermal design,
this form of boiling is called subcooled nucleate boiling, and its occurrence in reactor coolant channels is discussed in
Chapter 25. The point zBOIL where bulk boiling begins then marks the line of demarcation between the boiling and non-
boiling lengths. In practice, this point can be found by integrating the equation
964 Two-Phase Nuclear Heat Transfer
FIGURE 24.19 The void fraction as a function of elevation in a modern BWR. Here, the axial heat flux is assumed to be
co-sinusoidal, and the slip ratio is assumed to be 1.0. An equilibrium thermodynamic model is assumed.
z
h(z) = h INLET + PH (GA)
∫ z INLET
q′′(z) dz (24.31)
from the inlet (where zINLET = −H/2) until a value of z is found where h(z) = hSAT. It can then be shown that the equilib-
rium quality as a function of axial elevation is
z
x(z) = x INLET + PH ( GA ⋅ h lv )
∫ z INLET
q′′(z) dz (24.32)
However, in Equation 24.32, xINLET is usually a negative number thermodynamically (xINLET ≅ −0.05) because the cool-
ant enters the core as subcooled liquid. Once the equilibrium quality is found, the void fraction can then be determined
from the following equation:
where M(z) = 1 + [(1 − x(z))/x(z)]·(ρv/ρl)·S(z), and the slip ratio S(z) is normally a function of axial position (see our
previous discussion of this subject in Chapter 23). Thus, the values for h(z), x(z), and α(z) can be determined either
numerically or analytically by solving Equations 24.31, 24.32, and 24.33. Their values can subsequently be used to select
the appropriate heat transfer coefficients for each flow regime. Finally, when the heat flux is co-sinusoidal, Equations
24.31 and 24.32 can be evaluated to give
[1 + sin(πz/H) ] (24.34)
h(z) = h INLET + (Q/2m)
and
(
x(z) = x INLET + Q 2m )
h lv [1 + sin(πz/H)] (24.35)
where Q = 2q′′o PHH)/π. Equations 24.34 and 24.35 provide the basis for the curves shown in Figures 24.17 and 24.18.
In these equations, the fuel rod fission gas plenums (see Chapter 5) are normally not included in the calculation of the
active core height.
24.30 DEPICTION OF THE FLOW BOILING CURVE 965
FIGURE 24.20 Depicting the flow boiling curve using the heat transfer coefficient and the equilibrium quality.
24.30
D epicting the Flow Boiling Curve Using the Heat Transfer
Coefficient and the Equilibrium Quality
For flow boiling in reactor cores, there is a rather interesting relationship between the size of the convective heat trans-
fer coefficient and the equilibrium quality xeq. This relationship is illustrated in Figure 24.20. When the coolant first
enters the core as a subcooled liquid, the equilibrium quality is zero (or less), and the convective heat transfer coefficient
is the heat transfer coefficient for the liquid. Then, when bubbles begin to form, and the equilibrium quality exceeds
zero, the heat transfer coefficient rises gradually as bubbles begin to detach from the surface. The liquid core remains
intact, but the rest of the flow field enters the nucleate boiling or bubbly flow regime (see Figure 24.20). This persists for
equilibrium qualities between 0 and about 40%. Eventually, as more heat is added to the coolant, the liquid core disap-
pears, and the coolant enters the saturated boiling and slug flow regimes. Here, bubbles can be found anywhere in the
channel, and in some cases, they join together to form larger bubbles or voids. In vertical flow applications, slug flow
occurs for equilibrium qualities between 40% and 65%. Then, when additional heat is added to the coolant, the bubbles
in the center of the channel begin to coalesce to form large slugs, and eventually, these slugs begin to merge until all
that remains is a vapor core. In this regime, which is called the annular flow regime, only a liquid film remains on the
surface of the walls. This annular film can normally be found in vertical channels with equilibrium qualities between
65% and 85%. Then, when the liquid film evaporates completely from the wall surface (at equilibrium qualities between
80% and 90%), the heat transfer coefficient falls dramatically because there is no liquid film remaining to remove heat.
This regime, which occurs at qualities close to 90%, is then called the mist flow regime. Finally, when all of the water
droplets in the vapor core evaporate, the vapor becomes superheated, and for water, it manifests itself as superheated
steam. Then, the equilibrium quality becomes 100% and all that remains is a superheated vapor. Practically speaking,
the transition points between these flow regimes depend of the mass flux G, the heat flux q″, and the system pressure P.
966 Two-Phase Nuclear Heat Transfer
Hence, the equilibrium qualities defining each flow regime may be different than those depicted in Figure 24.20; how-
ever, they are usually representative of the transition points that are actually encountered in reactor fuel assemblies.
24.31
F low Boiling Correlations and Heat Transfer Coefficients
Specific to Nuclear Power Plants
Before proceeding to Chapter 25, we would like to point out that there are about as many heat transfer correlations for
flow boiling as there are types of nuclear power plants in the world today. In general, flow boiling consists of at least four
distinct processes that we have decided to call
1. Nucleate boiling
2. Forced evaporation
3. Forced convection
4. At least some radiative heat transfer.
The relative contributions of these processes to the convective heat transfer coefficient are flow regime dependent.
Hence, they depend not only on the liquid-to-vapor ratios but also on how the liquid and the vapor phases are distributed
within a channel. When we start at the bottom of the flow boiling curve, nucleate boiling dominates the convective heat
transfer coefficient, and as we move further up the curve, evaporation and radiation become more important. Here, many
heat transfer correlations for flow boiling attempt to break the heat transfer process into a number of distinct regions
or regimes, and to mix the effects of nucleate boiling, forced convection, radiation, and forced evaporation together in
each of these regimes.
In Chapter 25, we will discuss how these processes are interrelated as we move along the flow boiling curve. When
we do so, we will find that the convective heat transfer coefficient increases for a while and then decreases until the
Leidenfrost point is reached. The heat transfer coefficient can change by at least an order of magnitude as we move up
and down this curve, and in some cases, it can change by as much as two orders of magnitude. In Chapter 25, we will
discuss this behavior for common reactor coolants such as light and heavy water. Because convection in nuclear power
plants is almost always forced, we will concentrate primarily on how the convective heat transfer coefficient behaves
for flow boiling (as opposed to pool boiling). Typically, the heat transfer coefficients for flow boiling are used for the
subchannels within a fuel assembly. Here, the hydraulic diameter is based on the subchannel cross-sectional area and
the wetted perimeter alone. However, in other applications, the hydraulic diameter can be approximated by replacing
the subchannels by an equivalent annulus that surrounds each fuel rod (see Figure 24.21). In this case, the fluid flow
resembles the flow of fluid through a heated vertical pipe. In some cases, the shape of the axial heat flux is constant, and
in others, it is not. In the next chapter, we would like to examine each of these scenarios in more detail.
FIGURE 24.21 A representative flow channel from a reactor fuel assembly and its equivalent annulus r epresentation. In this
representation, the coolant is assumed to flow through the center of the surrounding fuel rod which is constructed from portions of
its four nearest neighbors.
Questions for the Student 967
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, CRC Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Web References
https://www.steamtablesonline.com/steam97web.aspx.
https://www.irc.wisc.edu/properties/ (which contains a really nice steam table calculator).
http://www.peacesoftware.de/einigewerte/wasser_dampf_e.html (which contains a really nice calculator of subcooled water properties).
http://www4.ncsu.edu/~doster/NE402/Text/BoilingHeatTransfer/BoilingHeatTransfer.pdf.
21. How many regions are there on the pool boiling curve? What are the names of these regions and what is their physi-
cal significance?
22. If one knows the Nusselt number in a reactor fuel assembly just before nucleate boiling begins, what other param-
eters are required to calculate the convective heat transfer coefficient?
23. What components can the Rohsenow correlation be applied to in a typical nuclear power plant?
24. Suppose that we have reached the Leidenfrost point on the pool boiling curve, and radiative heat transfer begins to
dominate the heat transfer process. If the temperature difference between the wall and the fluid is ΔT, what power
of ΔT is radiative component of the heat transfer coefficient proportional to?
25. Which correlations, if any, are applicable to the hottest fuel assemblies in a spent fuel pool when the pumps which
cool the pool are out of operation for a significant period of time?
26. What happens if we spray water onto the spent fuel pool from above? Can any of the correlations that we have
discussed be used?
27. What is a typical value for the convective heat transfer coefficient in a PWR fuel assembly before flow boiling occurs.
28. What is a representative value of the Reynolds number in a PWR fuel assembly during normal operating conditions?
29. What is the typical velocity of the coolant flowing through a PWR fuel assembly when a reactor is o perating at full power?
30. What were the names of the researchers who first systematically measured the heat transfer rate in a boiling pool as
a function of the amount of wall superheat?
31. What are some typical values of the convective heat transfer coefficient in a pool of boiling water during natural
circulation conditions?
32. During normal operating conditions, what is a typical range of values for the convective heat transfer coefficient in
a BWR?
33. What is the definition of the for reactor work ? In what correlations for the convective heat transfer coefficient does
the critical pressure ratio sometimes appear ?
34. Before pool boiling begins, what correlation can be used to find the convective heat transfer coefficient when there
is a sizeable temperature difference between the top and the bottom of the pool?
35. In pools of boiling water close to atmospheric pressure, how much wall superheat (in °C) is normally required to
reach the CHF point?
36. In boiling nuclear heat transfer, what does the term “flow boiling” refer to? How is flow boiling different than pool
boiling ? For the same reactor coolant, which of these two types of boiling has a higher heat transfer coefficient ?
37. What correlation is usually used to find the value of the minimum heat flux at the Leidenfrost point? What is the
explicit form of this correlation?
38. In a boiling pool of water, derive an expression for the rate of evaporation from the pool if the latent heat of vapor-
ization is known.
39. What correlation(s) can be used to find the CHF in a stationary pool of boiling water? Why can these correlation(s)
no longer be used when the flow becomes forced?
40. After the Leidenfrost point is reached on the pool boiling curve, what form of heat transfer is primarily responsible
for increasing the heat transfer coefficient at the wall surface?
41. When pool boiling begins, the difference between the wall temperature and the saturation temperature is sometimes
given a specific name. What is this temperature difference called, and what is its physical significance?
42. Write an expression that can be used to find the boiling heat transfer coefficient around a hot horizontal tube in a
reactor heat exchanger. What type of boiling typically occurs around the tube and is the boiling on the sides of the
tube the same as it is on the top or the bottom?
43. What effect does surface roughness have on the boiling heat transfer coefficient for a hot object in a stationary pool?
44. In general, are the convective heat transfer coefficients for flow boiling smaller, larger, or the same as they are for
pool boiling?
45. Fill in the following expression with the appropriate word or phrase: the temperature at which the water droplets
stay on a heated surface for the longest period of time before evaporating is called the ______________.
46. In a reactor fuel assembly, how would you calculate the equivalent annulus for a square unit cell and a hexagonal
unit cell? In what types of reactors are these unit cells most likely to be found?
47. Prior to nucleate boiling, what correlation(s) can be used to find the convective heat transfer coefficient if the coolant
pumps are still running?
48. Dryout occurs differently in BWRs than it does in PWRs. What is the reason for this difference, and how much of
this difference is attributable to the heat flux, the mass flow rate, and the system pressure?
49. What are the definitions of the Jakob number and the Bond number, and what are these dimensionless numbers used
for in nuclear heat transfer?
Exercises for the Student 969
5 0. What is the Stefan–Boltzmann law used for, and to what power is the absolute temperature raised in this law?
51. Prior to nucleate boiling, what correlation(s) can be used to find the convective heat transfer coefficient if the coolant
pumps are turned off.
52. At what coolant velocities does flow boiling typically occur in a reactor fuel assembly?
53. At what coolant velocities does pool boiling typically occur in a reactor fuel assembly?
54. How is the decay heat normally removed from a spent fuel pool?
55. Fill in the following sentence with the appropriate word or phrase: _____ is the most likely cause of fuel rod dryout
in the cores of PWRs.
56. In addition to Rohsenow’s correlation, name at least two other correlations that can be used to find the convective
heat transfer coefficient when nucleate boiling occurs.
57. In large bundles of heated horizontal tubes, is the heat transfer coefficient for an average tube higher or lower than
it is for a single isolated tube having the same surface temperature and geometric orientation?
58. What correlation(s) can be used to find the convective heat transfer coefficient in a boiling pool when transition
boiling and film boiling occurs?
59. When nucleate boiling occurs in a stationary pool of water close to atmospheric pressure, how much wall superheat
is required to sustain this type of boiling on the pool boiling curve?
60. What is the equilibrium quality at the entrance and the exit of a modern BWR fuel assembly?
61. What is the equilibrium quality at the entrance and the exit of a modern PWR fuel assembly?
62. Thermodynamically, what does an equilibrium quality less than zero or greater than one represent?
63. Write an expression for the equilibrium quality as a function of axial elevation in a BWR fuel assembly with a co-
sinusoidal axial heat flux.
64. Write an expression for the equilibrium quality as a function of axial elevation in a PWR fuel assembly with a con-
stant heat flux.
65. The pressure in a tank of boiling water is slowly raised from atmospheric pressure to 15.5 MPa, which is the pres-
sure in the primary loop of a Westinghouse PWR. Does the nucleate boiling heat transfer coefficient increase,
decrease, or remain the same under these conditions?
66. In pool boiling applications involving flat plates and heated tubes, what does the term BDAD refer to?
Exercise 24.2
The coolant enters the core of a commercial BWR as a subcooled liquid with a temperature of 270°C. At a pressure of
1,050 PSI (7.25 MPa), how much energy must be added to the coolant for saturated boiling to begin?
Exercise 24.3
Calculate the equilibrium quality of the coolant at the core inlet in Exercise 24.2. If the equilibrium quality is negative,
what does a negative quality represent?
Exercise 24.4
In an experiment designed to study the effects of pool boiling in nuclear power plants, the heat flux and the temperature
and pressure of ordinary water are simultaneously increased so that saturated boiling occurs at all times. Burnout occurs
when the pressure reaches 2 MPa. Assuming for the moment that burnout is due solely to the effects of radiative heat
transfer, estimate the temperature of the heated surface when burnout occurs. Assume a radiative heat transfer coef-
ficient of 1,000 W/m2 °C.
Exercise 24.5
An electrically heated nuclear fuel rod 1 cm in diameter is placed in a large tank of water at 2,250 PSI (15.5 MPa).
Calculate the largest value of the surface heat flux the fuel rod can withstand without burning out.
970 Two-Phase Nuclear Heat Transfer
Exercise 24.6
An experiment is designed to study the effects of nucleate boiling on a flat, heated horizontal plate immersed in a pool
of saturated water at 300°C. When the surface heat flux is q″ = 1 × 105 W/m2, how much superheat is required for nucle-
ate boiling to begin?
Exercise 24.7
Calculate the convective heat transfer coefficient prior to the start of nucleate boiling on the surface of a heated rod using
the Fishenden and Saunders correlation. Assume the rod is 8 m long and 1 cm wide, and is positioned horizontally in a
tank of water at 250°C and 800 PSI (5.5 MPa).
Exercise 24.8
Using Rohsenow’s correlation, calculate the convective heat transfer coefficient on the surface of the same fuel rod
described in Exercise 24.7 after nucleate boiling begins. Assume the physical state of the coolant and the geometry of
the rod are the same. Is the heat transfer coefficient higher or lower than that in Exercise 24.7?
Exercise 24.9
The CHF on the surface of a nuclear fuel rod is q″ = 1 × 106 W/m2. If the rod is immersed in a tank of water at 300°C and
2,250 PSI (15.5 MPa), and the flow of coolant around the rod is natural rather than forced, use Rohsenow’s correlation
to estimate the temperature of the surface when burnout occurs.
Exercise 24.10
The bottom of a large water storage tank is heated electrically until nucleate boiling begins. Prior to the start of the heat-
ing process, the water in the tank is maintained at 30°C and 1 atm. When the temperature of the lower surface reaches
120°C, what is the value of the heat flux?
Exercise 24.11
Calculate the CHF at the bottom of the water storage tank in Exercise 24.10. At what surface temperature is the CHF
reached?
Exercise 24.12
A flat, hot plate is placed at the bottom of a reactor heat exchanger and positioned horizontally to optimize the heat
transfer rate. The water in the immediate vicinity of the plate has a temperature of 100°C and a pressure of 100 kPa (14.7
PSI). Using Zuber’s correlation, calculate the surface heat flux at the Leidenfrost point. Can the surface heat flux ever
get any lower? If the minimum heat flux is reached when the surface superheat is 120°C, what is the surface temperature
at the point where this occurs?
Exercise 24.13
Water is to be evaporated from a large water storage tank by heating the bottom of the tank. The water in the tank is
maintained at atmospheric pressure and 20°C. Suppose a heat flux of 1 × 106 W/m2 is applied to the bottom of the tank.
What is the evaporation rate from the tank?
Exercise 24.14
The Mostinski correlation is sometimes used to estimate how changes to the system pressure affect the value of the
convective heat transfer coefficient in the nucleate boiling regime. Suppose the convective heat t ransfer coefficient is
10,000 W/m2 °C when the pressure is 800 PSI (5.5 MPa). What is the value of the convective heat transfer coefficient
when the pressure is raised to 1,050 PSI (7.25 MPa)? Does the convective heat transfer coefficient become higher or
lower under these conditions? What is the physical explanation for this behavior?
25
Heat Transfer Correlations for Advanced
Two-Phase Nuclear Heat Transfer
25.1 Heat Transfer Mechanisms for Two-Phase Flow
Flow boiling is an important component of convective heat transfer in nuclear power plants. In fact, it is the primary mode of boiling
heat transfer in boiling water reactors (BWRs), and it also occurs in the steam generators and other components of pressurized water
reactors (PWRs). It is much more prevalent than pool boiling because the convection process is always assumed to be forced. To
describe the flow boiling process, most heat transfer correlations used for reactor work attempt to separate the physical processes that
occur at a heated surface into two separate components or regimes called
☉☉ A nucleate boiling (NB) regime where bubbly flow or slug flow determines the heat transfer coefficient
☉☉ An annular flow (AF) regime where forced evaporation (EV) and convection (CONV) determine the heat transfer coefficient
In the first regime, the heat transfer process is driven primarily by the creation and detachment of bubbles from the fuel rod surface.
In the second regime, the heat transfer process is driven by convection through the liquid film and through the forced evaporation of
the molecules at the film’s surface. For the same mass flow rate, the convective heat transfer coefficients are much larger than their
single-phase counterparts. Physically, the characteristics of each flow regime are illustrated in Figure 25.1. Before the bulk tempera-
ture reaches the saturation temperature (i.e., when TBULK < TSAT), the heat transfer process consists of a convective component and a
nucleate boiling component:
The fluid under these conditions is considered to be subcooled, but bubbles can be created and detached from the fuel rod surface.
Hence, the equilibrium quality can be negative in a thermodynamic sense (see Chapter 24), in spite of the fact that the vapor produc-
tion can be significant. Then as more heat is added to the coolant, its bulk temperature becomes equal to the saturation temperature
(TBULK = TSAT). After the saturation temperature is reached, the coolant is no longer considered to be subcooled thermodynamically,
and the heat transfer process consists of a nucleate boiling component and a forced evaporation component:
The exact weighting of these components depends on how the flow field is modeled. From Equation 25.2, it is possible to write
where A and B are weighting functions that determine the percentage of the heat transfer that is due to nucleate boiling and forced
evaporation. Similarly, we can adopt an identical approach when nucleate boiling begins and assume that the heat transfer process
consists of the following components:
where C and D are another set of weighting coefficients. The problem of determining the heat transfer coefficient is then reduced
to the problem of finding appropriate values for A, B, C, and D as a function of the flow quality x and the void fraction α. Clearly, it
must be true that
971
972 HEAT TRANSFER CORRELATIONS
FIGURE 25.1 The flow regimes and cladding temperatures encountered in two-phase nuclear heat transfer during safe and
unsafe operation.
and
It then becomes a curve fitting exercise to find the appropriate values for A(x), B(x), C(x), and D(x) that give the best fit
to available data. Many famous flow boiling correlations are based in part on this approach.
h TP = ACo Bh SP + CBo Dh SP
(25.6)
h TP = Convective boiling term + Nucleate boiling term
where A, B, C, and D are constants that can be determined by performing an experiment, Co is a number called the
convection number, and Bo is a number called the boiling number. Just as in the case of the Reynolds number Re and
the Prandtl number Pr, the convection number and the boiling number are dimensionless. Equation 25.7 then becomes
the foundational equation of two-phase nuclear heat transfer.
25.3 Heat Transfer Coefficient Behavior in the Subcooled Boiling Regime 973
Boiling Number
where G is the mass flux of the mixture (G = m/A). In practice, there are dozens of correlations that can be used to find
the convective heat transfer coefficient during forced convection and flow boiling. In general, these correlations assume
that h is a function of G, x, and p, and in some cases, they also include a dependence on ΔTwall or q″. However, most of
them either have very large margins of error or a limited range of applicability. In the discussion that follows, we would
like to concentrate on four particular correlations that do not exhibit these limitations. They include
☉☉ the Bernath correlation
☉☉ the Klimenko correlation
☉☉ the Chen correlation
☉☉ the Kandlikar correlation.
For any single-phase flow regime that immediately precedes these correlations, the Dittus–Boelter correlation (see
Chapter 20) can be used to find the convective heat transfer coefficient. The Dittus–Boelter correlation requires the flow
to be turbulent and fully developed. However, it is not appropriate to use for liquid metals, and so, its use is restricted
primarily to water, superheated steam, and industrial gases. As we mentioned in Chapter 21, the Dittus–Boelter correla-
tion has a very low overall error.
25.3
Heat Transfer Coefficient Behavior in the Subcooled Boiling Regime
The heat transfer coefficient when subcooled boiling occurs is a function of both the fluid velocity and the applied
heat flux. The combined effects of nucleate boiling and forced convection when the boiling is subcooled are shown in
Figure 25.2. The heat transfer coefficient without boiling (hSP) is plotted for water at atmospheric pressure as the straight
FIGURE 25.2 The variation of the convective heat transfer coefficient with the fluid velocity and the heat flux in the subcooled
boiling regime. Here the liquid velocity is in m/s. (Thermopedia.com.)
974 HEAT TRANSFER CORRELATIONS
line on the bottom of the chart. Then as the applied heat flux increases ( q1′′ > q′′2 > q′′3 ) and the flow velocity increases, the
heat transfer coefficient increases as well. At low values of the mixture velocity, the heat transfer coefficient increases
dramatically with the heat flux (up to a factor of about 2), but when the velocity becomes high enough, the applied heat
flux has little effect on the overall heat transfer coefficient. During subcooled boiling, the equilibrium quality is still
negative, although bubbles continue to form in the flow stream.
( )
h NB = 10,890 D e D′e + 48v D e 0.6 (25.10)
The temperature of the cladding (also called the wall temperature) is given by
In Equations 25.10, 25.11, and 25.12, T WALL is the surface temperature of the cladding when the bubbles begin to form,
TBULK is the bulk fluid temperature, p is the pressure in PSI, v is the coolant velocity (in ft/s), De is the equivalent diam-
eter of the coolant channel (in ft), and D′e = De + (PH/π) where PH is the heated perimeter of the same channel (in ft).
Thus, the Bernath correlation consists of three separate equations that require the bulk temperature of the coolant to
be known before the saturation temperature is reached:
The cladding temperature: TCLAD = 102.6 ln (p) − (97.2p)/(p + 15) − 0.45v + 32 (25.13b)
The Bernath correlation can be used for pressures between 23 PSI and 3,000 PSI (0.158–20.68 MPa), velocities between
4 and 54 ft/s (1.22–16.46 m/s), and hydraulic diameters between 0.14 in. and 0.66 in. (0.36–1.68 cm). The heat transfer
coefficient is calculated in BTU/h ft2 °F. It is an extremely useful correlation to find the convective heat transfer coef-
ficient in the subcooled boiling region. In general, it predicts a higher value for the convective heat transfer coefficient
than the Dittus–Boelter correlation does. It also requires the bulk fluid temperature TBULK to be known before the
equilibrium quality becomes positive. Thus, in the subcooled boiling region, the equilibrium flow quality is less than
zero (x < 0), and it is actually a negative number in a thermodynamic sense. This is because the coolant velocity is too
high for all the fluid molecules to be at exactly the same temperature across the entire channel. Example Problem 25.1
shows how the convective heat transfer coefficient predicted by the Bernath correlation compares to the convective heat
transfer coefficient predicted by the Dittus–Boelter correlation when subcooled boiling begins. The point at which vapor
bubbles first detach from the surface of the cladding and enter the flow stream is called the Onset of Nucleate Boiling
(ONB) Point. We will have more to say about the location of this point in Section 25.14.
☉☉ See Bernath, L., Transactions of the A.I.Ch.E (1955).
5.6 m/s (18.37 ft/s), the cross-sectional flow area is about 8.8 × 10 −5 m2 (0.00095 ft2), the cladding diameter is about
9.5 mm (0.031 ft), and the equivalent diameter is about 11.8 mm (0.0118 m or 0.039 ft). The operating pressure is
2,250 PSI or 15.5 MPA. Under these conditions, calculate the convective heat transfer coefficients using the Dittus–
Boelter correlation and the Bernath correlation in the subcooled boiling regime. Which correlation is more accurate, and
which correlation gives a higher convective heat transfer coefficient?
Solution To use the Bernath correlation (Equation 25.13a), we must know the heated perimeter of the coolant channel.
Since the cladding diameter is 9.5 mm, the heated perimeter of the central channel is PH = πD = 29.85 mm = 0.098 ft.
The value of the heat transfer coefficient is then h = 10,890(De/D′e) + 48v/D 0.6
e = 10,890·[0.039/(0.039 + (0.098/3.14))] +
48·18.37/(0.039)0.6 = 10,890·0.55 + 6,172.3 = 12,162 BTU/h ft2 °F = 68,950 W/m2 °C. The Dittus–Boelter correlation
for a rod bundle, which we discussed in Chapter 21, can be written as h = C·(k/De)·(Re)0.8·(Pr)0.4, where the geometrical
correction factor C for a typical PWR is C = 0.042·(P/D) − 0.024 for a square fuel assembly. Since (P/D) = 1.33, the
value of C is 0.032. The thermal conductivity at 2,250 PSI is 0.296 BTU/h ft °F. Therefore, the convective heat transfer
coefficient predicted by the Dittus–Boelter correlation is h = 0.032·(0.296/0.039)·36,239·0.997 = 8,775 BTU/h ft2 °F =
49,750 W/m2 °C. Thus, the Bernath correlation predicts a convective heat transfer coefficient that is approximately 40%
higher than the Dittus–Boelter correlation in the subcooled boiling region. The Bernath correlation is obviously a more
accurate correlation to use because it is specifically designed to calculate the convective heat transfer coefficient under
these conditions. [Ans.]
25.5
T he Klimenko Correlation for Flow Boiling (for 0 < x < ~0.80)
The Klimenko correlation was developed in 1988, and it can be used to find the convective heat transfer coefficient
until the liquid film disappears from the surface of the fuel rods. The Klimenko correlation attempts to distinguish
between the heat transfer rate for nucleate boiling and film boiling. For nucleate boiling, the Klimenko correlation uses
the following equation for the convective heat transfer coefficient
where Pe is the Peclet number of the mixture, and p is the operating pressure (in bars). The reader may recall that
1 bar = 0.1 MPa. In the film boiling region, where a large amount of forced evaporation can occur, the Klimenko cor-
relation uses the following equation for the convective heat transfer coefficient
( ) ( )
0.2
h FB = k l L ⋅ 0.087 ⋅ Re 0.6 ⋅ Prl 0.167 ρv ρl (25.15)
In both of these equations, p is the operating pressure, Prl is the Prandlt number of the liquid, Re is the mixture Reynolds
number, and the following definitions are used for Re, Pe, and L, the characteristic length:
Pe = q′′Lρlc pl h lv ρv k l (25.17)
L = √ σ g ( ρl − ρv ) (25.18)
Here G is the mass flux in the coolant channel. Klimenko’s correlation is valid up to the point that the wall dries out, but
it is not valid after that. Whether the first or the second equation should be used depends on a separate calculation for
the combined boiling coefficient, CB:
( ) ( )
0.33
CB = G ⋅ h lv q ′′ 1 + x(ρl − ρv ) ρv ρv ρl (25.19)
where G = m/A = ρv is the mass flux of the mixture. If CB < 16,000, then the first equation (Equation 25.14) should be
used, and the heat transfer is dominated by nucleate boiling. If CB ≥ 16,000, then the second equation (Equation 24.15)
976 HEAT TRANSFER CORRELATIONS
should be used, and the heat transfer is dominated by forced evaporation at the liquid film’s surface. Klimenko’s model
assumes that the two phases are in thermal equilibrium, while in fact, boiling begins well before TBULK = TSAT because
the temperature near the surface of the wall can be very high. Thus, his model is somewhat of an idealization, but it can
give respectable results if it is used properly.
25.6
T he Chen Correlation for Flow Boiling (for 0 < x < ~0.80)
The Chen correlation was originally developed by J.C. Chen. It is a very popular correlation, and it has enjoyed consid-
erable success since it was first introduced in the mid-1960s. It is applicable to any boiling condition prior to film dryout
or departure from nucleate boiling (DNB), and it is used in many reactor safety codes to predict when DNB occurs.
It has the lowest average error (about 12%) of all the common correlations for the two-phase flow of water in vertical
pipes and reactor coolant channels. However, it does not do a particularly good job for refrigerants such as methanol and
Freon. It is based on combining the predictions of two other famous correlations—the Dittus–Boelter correlation and
the Forster–Zuber correlation—and then applying a set of weighting functions to these correlations that have become
known as the forced convection enhancement factor F and the nucleate boiling suppression factor S. The exact form of
the Chen correlation is
where the values of F and S are defined by the individual flow regimes, and the subscripts DB and FZ refer to the
heat transfer coefficients predicted by the Dittus–Boelter and Forster–Zuber correlations. The enhancement factor can
have a value between 1 and 70, while the suppression factor can have a value between about 0.1 and 0.9. In general,
the enhancement factor has a value less than 5.0. The suppression factor is a function of the Reynolds number, and the
enhancement factor is a function of the convection number Co. However, the convection number is multiplied by a cor-
rection factor (μl/μv)0.1 to account for the viscosity difference between the liquid and the vapor. The value of F is then
expressed as a function 1/(Co·(μl/μv)0.1) which is sometimes called 1/χ. The values of F and S are shown in Figure 25.3a
and b. The Chen correlation can be quite challenging to implement because the convective part is a modified form of
the Dittus–Boelter correlation.
Dittus–Boelter Correlation
( )
0.8
h DB = k l D e ⋅ 0.023 G(1 − x)D e u l Prl0.4 (25.21)
Forster–Zuber Correlation
h FZ = 0.00122λ ⋅ ∆T 0.24 ⋅ ∆p0.75 (25.22)
(
λ = k 0.79 c P 0.45ρ0.49 ) (σ
l
0.5
)
µ l 0.29ρv 0.24 h lv 0.24 (25.23)
25.6 The Chen Correlation for Flow Boiling (for 0 < x < ~0.80) 977
FIGURE 25.3 (a) The convection enhancement factor F used in the Chen correlation for flow boiling nuclear heat transfer. (b) The
NB suppression factor S used in the Chen correlation for flow boiling nuclear heat transfer.
Here, ΔT is the amount of surface superheat (ΔT = T WALL − TSAT), Δp is the pressure difference between the wall
and the fluid core (Δp = p(T WALL) − p(TSAT)), and the suppression factor S is a function of the total Reynolds num-
ber Re. Some people have derived analytical expressions for F and S to avoid having to use the curves shown in
Figure 25.3. The values for the forced convection enhancement factor F can be found from
F =1 ( for C o )
< 0.1 (25.24a)
F = 2.35 ( 0.213 + Co )
0.736
( for C o )
≥ 0.1 (25.24b)
where Co is the convection number. The values for the nucleate boiling suppression factor S can be found from
978 HEAT TRANSFER CORRELATIONS
(
S = 1 1 + 2.53 × 10 −6 ⋅ Re1.17 (25.25) )
where the Reynolds number is calculated using
Re = Re l ⋅ F 5/ 4 (25.26a)
and
Re l = (1 − x) ⋅ G ⋅ D e u l (25.26b)
The value of the convection enhancement factor is then plotted as a function of the Greek symbol χ (chi) whose value
can be found from the convection number Co using
( ) (
= [ (1 − x)/ x ] ⋅ ρv ρl ) ⋅ (µ )
0.1 0.8 0.5 0.1
χ = Co µ l µ v l µv (25.28)
Hence, χ can be interpreted as the convection number corrected for the viscosity difference between the liquid and
the vapor. In the secondary loop of a commercial PWR, the value of χ in a steam generator tube is about 0.9. In steam
generators and reactor fuel assemblies with flow qualities of about 20%, the convective component of the heat transfer
coefficient, given by F times the value predicted by the Dittus–Boelter correlation, is about 14,000 W/m2 °C, where the
value of F is between 2.5 and 3.0. The nucleate boiling component, given by S times the heat transfer coefficient pre-
dicted by the Forster–Zuber correlation, is about 5,000 W/m2 °C, where the value of S ranges from 0.10 to 0.15. Hence,
the total heat transfer coefficient is about 20,000 W/m2 °C when the flow quality is high. Like the Klimenko correlation,
the Chen correlation can be used until the liquid film disappears from the surface of the rods.
25.7
T he Kandlikar Correlation for Flow Boiling (for 0 < x < ~0.90)
The Kandlikar correlation is not as complicated as the Chen correlation, but it is very useful because it is applicable to
fluids other than water. It was developed at the Rensselaer Polytechnic Institute (or RPI) in 1983. The basic form of the
correlation is
where A, B, C, and D are coefficients whose values are shown in Table 25.1. A separate set of coefficients must be used
for the convective region and the nucleate boiling region. Here again, Co is the convective number and Bo is the boiling
number. The convective coefficients must be used when Co < 0.65 and the nucleate boiling coefficients must be used
when Co ≥ 0.65. The value of h SP is determined from the Dittus–Boelter equation:
(
h SP = 0.023Re 0.8 Pr 0.4 k D e (25.29) )
where k is the thermal conductivity of the liquid, and De is the equivalent diameter of the channel. The properties used
for the Reynolds and the Prandtl numbers are both evaluated at the saturation temperature of the liquid. The Kandlikar
correlation has a number of advantages over other correlations because it can be extended to fluids other than water.
The basic way this is done is to incorporate another parameter into the correlation called the fluid-dependent parameter
F. The resulting correlation is then
TABLE 25.1
The Values of the Primary Coefficients Used in the Kandlikar Correlation
where F has a value of 1.0 for water and values between 1.1 and 4.7 for other fluids. Representative values for F are shown
in Table 25.2. This novel approach works well for vertical flows, and all one has to do is find the appropriate value for
F. The values for A, B, C, and D remain unchanged. One of the major advantages to the Kandlikar correlation is that it
can be used for two-phase flows in horizontal tubes where the effects of stratification become important. An example
of what stratified flow looks like is shown in Figure 25.4.
The effects of stratification are taken into account by introducing an additional parameter into the calculations called
the Froude number. The Froude number is a dimensionless number that represents the ratio of the inertial forces to the
gravitational forces acting on the fluid, and for most applications it is given by
Here v is the velocity of the flow, g is the force of gravity, and D is the ratio of the cross-sectional area to the width of
the flowing surface. For horizontal, stratified flows in tubes and ducts, various values for the Froude number can be used
to define the following conditions:
☉☉ Fr = 1, critical flow—a disturbance will just begin to develop on the heated surface
☉☉ Fr > 1, supercritical flow—the flow is very rapid and can have many disturbances in it
☉☉ Fr < 1, subcritical flow—the flow is slow and tranquil
The form of the Kandlikar correlation for stratified flow in horizontal tubes is then
{ }
h TP = h SP ⋅ ACo B ⋅ ( 25Fr ) + CBo D ⋅ F (25.32)
E
TABLE 25.2
The Values of the Fluid-Dependent Parameter Used in the Kandlikar Correlation
FIGURE 25.4 Various forms of stratified flow for water in horizontal tubes.
980 HEAT TRANSFER CORRELATIONS
where the Froude number is the liquid-only Froude number, and the value of E is about 0.3. Here again hSP is the
single-phase liquid-only heat transfer coefficient given by the Dittus–Boelter correlation. Also, for vertical flow, and for
horizontal flow with Fr > 0.04, the Froude number multiplier in the convective term is equal to 1.0. Then the Kandlikar
correlation reduces to the previous form that we presented for vertical flow. Like the Klimenko and Chen correlations,
it can also be used up to the point where dryout occurs.
25.8
Margins of Error in Common Flow Boiling Correlations
One of the great advantages to the Kandlikar correlation is that it is has a very low average error, or mean deviation,
compared to other correlations in the flow boiling regime. The percentage mean deviations are shown in Table 25.3.
Notice that it does a good job for water and an even better job for refrigerants and cryogenic fluids. The Chen correlation
does slightly better for water (with a mean error of about 12%), but it does not do as well for refrigerants and cryogenic
fluids. Hence, the Chen correlation gives slightly better results for two-phase heat transfer in reactor fuel assemblies
and steam generators, but it is much more complex. The Kandlikar correlation gives better results for a wider variety of
fluids, and it is much simpler to implement. Other correlations have an average error for water of between 17% and 30%,
depending on the quality of the fluid and the flow regime.
The Kandlikar correlation has been correlated to several thousand data points, and hence, the underlying physics
is modeled with greater precision than some of the more primitive correlations. Other correlations that are used from
time to time as an alternative to the Kandlikar correlation include the Shah correlation; the Bjorge, Hall, and Rohsenow
correlation; and the Gungor and Winterton correlation. The latter correlation was developed in the United Kingdom.
These correlations are summarized in Table 25.4. Each correlation is based on a slightly different set of data points, and
the average error is also different. The Kandlikar c orrelation was updated in 1990 to include a much larger data set than
Kandlikar originally used. It probably employs the largest underlying data set of any two-phase heat transfer correlation
in use today. Again, none of these correlations is perfect, and engineering judgment must be used to select the appropri-
ate correlation for a specific problem.
25.9
Heat Transfer in the Mist Flow Regime after Film
Dryout Occurs (for ~0.90 < x < 1.0)
When enough heat is added to the coolant, the liquid film eventually disappears from the surface of the cladding. This
is particularly true for flow boiling in BWRs, but it can also occur in PWRs after the DNB point is reached. The exact
conditions under which this occurs are shown in Figures 25.5 and 25.6. When the liquid leaves the wall, the heat transfer
coefficient falls dramatically and decisively. In PWRs, the flow regime in which this occurs is called inverted annular
vapor flow or inverted annular film boiling. In BWRs, the analogous condition is called the end of film boiling or just
film dryout. While it does not occur as abruptly in BWRs as it does in PWRs at the DNB point, the heat transfer coef-
ficient can easily fall by a factor of 10 over a very short distance. The convective heat transfer coefficient then ranges
from about 930 W/m2 °C (or 295 BTU/h ft2 °F) when the vapor film is stationary to about 2,480 W/m 2 °C (or 788 BTU/h
ft2 °F) when it is moving at very high speeds.* The wall temperature will then climb dramatically as the heat transfer
TABLE 25.3
Comparisons of the Average Deviations in Several Popular Flow Boiling Correlations versus Experimental Data
* See Whalley, P.B., Boiling, Condensation, and Gas-Liquid Flow, Oxford, Clarendon Press (1987).
25.9 HEAT TRANSFER IN THE MIST FLOW REGIME 981
TABLE 25.4
Some Common Flow Boiling Correlations for Two-Phase Flow in Vertical and Horizontal Tubes
FIGURE 25.5 In PWR coolant channels, dryout occurs when a vapor film covers the surface of the cladding. This vapor film cre-
ates a vapor barrier which causes the heat transfer coefficient to fall rapidly over a very short distance, and this vapor barrier also
causes the cladding temperature to rise. In PWRs, this occurs at lower equilibrium qualities than it does in BWRs, and it is called
Departure from Nucleate Boiling or DNB. The equilibrium quality at the DNB point is approximately 15% in most PWRs.
coefficient falls. As soon as the liquid film leaves the wall entirely, all that is left is a two-phase mist containing a number
of water droplets. These water droplets can have a variety of sizes and shapes, but they are mostly spherical and they are
all entrained by and move within the vapor flow. Some of them impinge on the surface of the wall, while others remain
in the vapor core.
At the point of initial dryout, the mixture quality is not 100%. In fact, the flow quality is usually in the range of 10%–15%
for PWRs and 30%–40% for BWRs (see Figure 25.5 and Figure 25.6). After dryout is reached, the Chen, Klimenko, and
Kandlikar correlations can no longer be used because there is no longer a liquid film covering the surface. In this flow
982 HEAT TRANSFER CORRELATIONS
FIGURE 25.6 In BWR coolant channels, dryout occurs when a vapor film covers the surface of the cladding. This vapor film cre-
ates a vapor barrier which causes the heat transfer coefficient to fall rapidly over a very short distance, and this vapor barrier also
causes the cladding temperature to rise. In BWRs, this effect occurs at higher equilibrium qualities than it does in PWRs, and it is
called film dryout. The equilibrium quality at the dryout point is approximately 40% in modern BWRs.
regime, which is generally called the mist flow regime, the Groeneveld correlation is normally used. The presence of water
droplets in the flow means that it is still a two-phase mixture, and the Dittus–Boelter correlation (which is applicable a pure
vapor) cannot be used. The Groeneveld correlation was first developed in the 1970s, although the underlying data set has
been updated several times since its introduction. It can be used for tubular geometries, annular geometries, and rod bundle
geometries where there is vertical steam-water flow. It is applicable to any equilibrium flow quality up to 100%. Thus,
the Groeneveld correlation can be used from −0.5 < x < 1.0, where a quality less than zero indicates that the boiling is
subcooled. The basic form of the Groeneveld correlation is
where the subscript MF refers to the mist flow or post film dryout and, the Reynolds number, the Prandtl number, and
the thermal conductivity are those of the vapor, that is,
Re = G ⋅ D e µ v (25.34)
Pr = u v c v k v (25.35)
(
Y = 1 − 0.1 ρl ρv − 1 ) (1 − x)0.4 (25.36)
0.4
The values for A, B, C, and D are shown in Table 25.5. The Groeneveld correlation can be used for horizontal and verti-
cal tubes and for an exceptionally wide range of heat fluxes, mass fluxes, and system pressures. Its root mean squared
or RMS error is 7% for annular geometries and 12% for horizontal or vertical tubes. It can also be used for the tubes
in reactor steam generators.
25.9 HEAT TRANSFER IN THE MIST FLOW REGIME 983
TABLE 25.5
The Parameters Used in the Groeneveld Correlation
There have been a significant number of improvements to the Groeneveld correlation since it was first introduced
to fit the data better at low pressures (P ≤7 MPa). The most significant of these improvements has been to replace the
parameter Y by
( )
−0.508
Y = q′′ 0.278 ⋅ k v k CR (25.37)
where the thermal conductivity kCR is the thermal conductivity at the critical point. This also eliminates the value of D
from Equation 25.33. The revised values for A, B, and C are then
A = 0.000116 B = 0.838
C = 1.810
The Groeneveld correlation has about the same statistical error as the Chen correlation immediately preceding the dry-
out point (about 12%), but after the dryout point, it is far more accurate. Notice that it is more similar to conventional
heat transfer correlations because it does not contain separate convective and nucleate boiling terms. In fact, if we were
to compare the Groeneveld correlation to the Dittus–Boelter equation, the comparison would reveal that
( )
Groeneveld: h = k D e ⋅ 0.000116 ⋅ λ ⋅ Re 0.838 Pr1.810 (25.38)
( )
Dittus – Boelter: h = k D e ⋅ 0.023 ⋅ Re 0.8 Pr 0.4 (25.39)
where all of the physics of the vapor film is lumped into the value of λ:
( )
B −0.508
λ = x + (1 − x) ρv ρl q′′ 0.278 ⋅ k v k CR (25.40)
Thus, it should come as no surprise that the two-phase mist that remains after the wall dries out has many similarities
to the heat transfer coefficient for a two-phase vapor. In fact, the dependence on the Reynolds number is almost exactly
the same. The primary difference is that the value of λ (for high values of the heat flux) has the effect of diminishing the
heat transfer rate slightly. Other than this, the physics is essentially the same. Normally, the Groeneveld correlation is
implemented in the form of a lookup table, and many lookup tables for it can be found in the literature. As we will see
in Chapter 27, it also provides a lookup table for the critical heat flux (CHF). Some of the most common lookup tables
for the convective heat transfer coefficient for the film boiling regime can be found at
☉☉ Groeneveld, D.C., et al., “A Lookup Table for Fully Developed Film Boiling Heat Transfer,” Nuclear Engineering
and Design, 225:83–97 (2003).
The lookup tables for the CHF can be found at
☉☉ Groeneveld, D.C., et al., “The 2006 CHF Lookup Table,” Nuclear Engineering and Design, 237:1909–1922
(2007).
More updates to the lookup tables have also been made since that time.
984 HEAT TRANSFER CORRELATIONS
Heat Transfer Coefficients for Saturated and Superheated Vapor (for x ≥ 1.0)
25.10
Once the equilibrium flow quality reaches 100%, the water droplets in the vapor core disappear, and any further addition
of heat to the flow will cause the steam to become superheated. When this occurs, the coolant can be considered to be a
pure vapor, and the Dittus–Boelter correlation can be used again to find the convective heat transfer coefficient but with
all of the properties being those of the vapor rather than those of the liquid.
( )
h VAPOR = k V D E ⋅ 0.023 ( Re V )
0.8
( PrV )0.4 (25.41b)
For flow boiling, Table 25.6 shows some of the most popular flow regimes and the heat transfer correlations that can
be applied to these regimes. The only scenario that these correlations do not cover is one where the heat flux is so high
initially that it causes a vapor film to blanket the surface of the wall before an annular flow can be established. As we
discussed in a previous section, this is known as the DNB condition.
TABLE 25.6
Some Popular Flow Boiling Correlations and Their Ranges of Applicability
25.11
Cladding Temperature Changes after the CHF Is Reached
As we first mentioned in Chapter 24, the critical heat flux or CHF occurs when the convective heat transfer coefficient
cannot become any larger and will fall drastically if a vapor film begins to blanket the surface of the wall. The critical
heat flux increases as the mass flux increases (see Figure 25.7), and it is also pressure dependent. Once the critical heat
flux is reached, higher values for the heat flux cause the cladding temperature to increase dramatically at the CHF point,
and this “jump” in the wall temperature is due to the fact that the value of ΔT WALL = T WALL − TSAT increases while the
value of the saturation temperature TSAT remains the same. In practice, the size of this “jump” in the surface temperature
(see Figure 25.8) can sometimes approach several hundred degrees C because there is no liquid left on the wall to remove
FIGURE 25.7 The behavior of the CHF as a function of the mass flux in a reactor coolant channel.
FIGURE 25.8 The jump in the cladding temperature that occurs when the heat flux on the surface of a nuclear fuel rod exceeds
the CHF limit. When this occurs, the fuel rods usually fail and this causes radioactive fission products to be released into the cool-
ant. This jump in the surface temperature is also the origin of the boiling crisis.
986 HEAT TRANSFER CORRELATIONS
the additional heat. As soon as this jump occurs (from point C to point E on the flow boiling curve in Figure 25.8), the
cladding melts and the fuel rod it surrounds catastrophically fails. Now let us examine how the heat transfer coefficient
behaves before, during, and after the CHF point is reached.
25.12
Visualizing How the Heat Transfer Coefficient Behaves during Flow Boiling
Now let us see if we can visualize the heat transfer coefficient as we move along the flow boiling curve. Figure 25.9
shows how the heat transfer coefficient behaves as a function of the equilibrium quality when the pressure, the mass flux,
and the heat flux are fixed and the applied heat flux by nuclear standards is relatively low. In this case, we can plot the
heat transfer coefficient for the subcooled flow regime, where the equilibrium quality is negative in a thermodynamic
sense, for the saturated flow regime, where the equilibrium quality is between zero and unity, and for the superheated
vapor flow regime, where the equilibrium quality is unity or more. Note that for water, the critical heat flux mechanism
by which the fuel rods dryout is different for PWRs than it is for BWRs. For high heat flux to mass flux ratios (q″/G), the
rods fail due to DNB, and for low heat flux to mass flux ratios, the rods fail due to film dryout. The reduction in the heat
transfer coefficient at the DBN point in PWRs (where the equilibrium quality is 10%–15%) is typically more abrupt than
it is in BWRs at the dryout point where the equilibrium quality is between 30% and 40%. Hence, if the imposed heat flux
is high enough, it is possible for the vapor generation rate in the DNB region to become high enough to create a vapor
film to separate the liquid completely from the fuel rod surface. This generally results in more abrupt and catastrophic
rod failures than are normally encountered at lower power to flow ratios.
25.13
Fuel Rod and Coolant Channel Temperature Profiles
When finding the temperature profiles for the cladding and the coolant, the axial heat flux q″(z) can be thought of as an
independent variable that is set by the neutronic design of the core. Hence, the temperature profiles for the cladding and
the coolant are dependent variables that can be deduced from the heat flux. This means that once the axial heat flux q″(z)
is specified, an energy balance can be performed to find the temperature difference ΔT(z) between the cladding and the
FIGURE 25.9 The flow boiling heat transfer coefficient with forced convection.
25.15 TYPICAL LOCATIONS 987
coolant at every axial elevation. We will discuss this energy balance in more detail in Chapter 26. In this section, we
would like to briefly discuss how flow boiling affects the shape of these temperature profiles. To begin our discussion,
assume that the axial heat flux is uniform and the flow is forced. We can then find the temperature profiles in each of the
flow regimes we have already discussed. Again, it is helpful to refer to Figure 25.1 to follow the progression of events.
25.14
Cladding-Coolant Temperature Differences before Nucleate Boiling Begins
Before nucleate boiling begins, the coolant is a single-phase liquid moving at considerable speed, and the convective
heat transfer rate is determined by Newton’s Law of Convection:
If the flow is fully developed and turbulent, we can use the Dittus–Boelter correlation to find the value of h. For repre-
sentative values of ρ, μ, k, v, and De; h is relatively constant, and so the difference between the wall temperature and the
coolant temperature can be expressed as
∆T = q′′ h (25.43)
where h depends on the Reynolds number. If q″ is constant, then ΔT is inversely proportional to h. This is shown
p receding the location of the ONB (or Onset of Nucleate Boiling) point in Figure 25.10. The Reynolds and the Prandtl
numbers in this case correspond to those of a subcooled liquid.
25.15
Typical Locations for the ONB Point, the Net Vapor
Generation Point, and the Bulk Boiling Point
In reactor fuel assemblies, nucleate boiling begins long before the equilibrium quality reaches zero. This is because the
cladding surface is much hotter than the surrounding fluid. Thus, while equilibrium thermodynamics implies that bulk
FIGURE 25.10 The vapor generation process in a boiling core showing the location of the ONB point, the NVG point, and the zero
quality point. Note: In this picture, the axial elevations are not drawn to scale.
988 HEAT TRANSFER CORRELATIONS
boiling begins just after the net vapor generation (or NVG) point (see Figure 25.10), in reality, bubbles begin to form
on the surface of the fuel rods at the ONB point. In American BWRs, the bulk boiling point (or saturated boiling point)
is located about a meter from the bottom of the core (see Figure 25.11). However, the ONB point is normally located
about 0.2 m from the core inlet, and the NVG point is normally located about 0.65 m from the core inlet. Hence, bubbles
can detach from the surface of the fuel rods as much as 0.8 m before bulk boiling begins! In practice, the NVG point is
normally considered to be the point where the flow transitions from subcooled boiling to bulk boiling. After this point,
conventional correlations can be used to find the boiling heat transfer coefficient. In some text books, the NVG point is
called the bubble detachment point (or the BDP). Thus, it represents the point where bubbles generated at the surface
of the fuel rods no longer collapse back into the surrounding coolant. In practice, the NVG point is located as much as
one-third of a meter before equilibrium thermodynamics says that bulk boiling begins. Thus, it is just another example
of how important thermal non-equilibrium effects can be in nuclear power plants.
25.16
Cladding-Coolant Temperature Differences during Subcooled NB
After nucleate boiling begins (see point B), bubbles start to form at the surface of the fuel rods. However, they collapse
back into the coolant as soon as they form because the bulk temperature is less than the saturation temperature. After
the ONB point is reached, the wall temperature profile begins to flatten out. Here, T WALL > TSAT but TFLUID < TSAT. In this
region, the heat transfer process becomes more efficient because the heat transfer coefficient is high. The temperature
difference ΔT between the wall and the coolant is generally very small (typically a few °C). The temperature difference
between points C and H is given by the Jens–Lottes correlation:
where ΔT WALL = (T WALL – T SAT) because the fluid temperature is equal to the saturation temperature at least at the
rod surface. Here, q″ is the heat flux in BTU/hr-ft 2, and p is the pressure in PSI. The Jens–Lottes correlation can be
used in both PWRs and BWRs, and it works particularly well for water reactors between pressures of 500 PSI and
2,250 PSI (~35–150 bar). An alternative version of the Jens–Lottes correlation that can be used with the metric unit
system is
FIGURE 25.11 The effect of the axial mass flux G on the equilibrium quality at the core exit in an American BWR. Here the axial
mass flux is measured in kg/m 2 s and a co-sinusoidal axial heat flux is assumed.
25.17 Finding the ONB Point 989
where the heat flux q″ is in MW/m2, the pressure p is in bars (where 1 bar = 100 kPa = 0.10 MPa), and ΔT WALL is in °C.
An alternative correlation that can be used over the same range of conditions is the Thom correlation:
where q″(z) is in BTU/h ft2, T is in °F, and p is again in PSI. To find the temperature change in °C, we simply find the
temperature change in °F and multiply by 5/9 or 0.5555.
In practice, the Thom correlation predicts a slightly lower value of ΔT WALL, but this normally a mounts to less than
1°C under most conditions. The Jens–Lottes correlation is designed to handle both nucleate boiling and bulk boiling.
However, it cannot be used for annular film flow. For typical PWRs operating at 155 bar (2,250 PSI or 15.5 MPa) and
a heat flux of 1 × 106 W/m2, the Jens–Lottes correlation predicts a temperature difference between the cladding and the
coolant of about 3°C. After the saturation temperature is reached (point H), the wall and the coolant temperatures stay
relatively constant until dryout occurs. Then the wall temperature shoots up precipitously, which can cause the cladding
to fail. Notice that the wall temperature flattens out when point C (the ONB point) is reached, but the fluid temperature
continues to rise until point H, where the bulk fluid temperature is now the same as the saturation temperature. If we
were to do a simple energy balance, we would find that the equilibrium quality of the flow is negative until point H is
reached. At point H, it becomes equal to 0, and after point H, it begins to steadily rise. This is a consequence of the fact
that the liquid and vapor phases are assumed to be in thermal equilibrium, when in reality, they are not.
25.17
Finding the ONB Point
In reactor fuel assemblies, nucleate boiling begins when the surface temperature of the cladding becomes infinitesimally
greater than the saturation temperature. Thus, the location of the ONB point zONB can be found by determining where
Normally, an energy balance is performed before the ONB point to establish an additional relationship between q″,
T WALL, and TBULK. This energy balance gives
or
where h is the single-phase heat transfer coefficient immediately preceding the ONB point. We can then substitute
Equation 25.49 into Equations 25.44, 25.45, or 25.46 to eliminate TWALL from the left-hand side. Using Equation 25.45
for the sake of illustration, we find that
Thus, when the value of q″(z) is known, we can solve Equation 22.50 to find the location zONB of the ONB point. In most
reactors, a representative expression for TBULK(z) is
is the mass flow rate (see Chapter 26). When q″(z) is co-sinusoidal in shape
where TINLET is the inlet temperature and m
the integral in Equation 25.51 can be evaluated easily, and the resulting expression for TBULK(z) is
This expression then specifies the bulk temperature of the coolant as a function of axial elevation with H being the core
height. To find the ONB point, the value of h is normally evaluated using the Dittus–Boelter correlation (see Chapter
21) where the properties of the coolant are assumed to be those of a subcooled liquid. Again, the ONB point is normally
reached about half a meter (0.5 m) before the NVG point in an American BWR. The equilibrium quality then reaches a
value of zero about 0.3 m after the NVG point.
25.18
Cladding-Coolant Temperature Differences after the NVG Point
Point G, which immediately precedes the zero quality point, is called the point of NVG. From the NVG point forward,
the flow is normally assumed to be a two-phase mixture. This means that the two-phase heat transfer coefficients we
discussed previously can be used. The NVG location can be determined from a number of empirical correlations. The
most frequently used ones are the Griffith correlation and the Browning correlation:
and
Here, the heat transfer coefficients and the densities are those of the subcooled liquid, and q″ is the heat flux flowing into
the channel. The pressure p is measured in Bar, the density ρ in kg/m3, and the mass flux G in kg/m2 s. Again ΔTSUB
refers to the amount of subcooling where the NVG point occurs:
The Griffith correlation was developed by Peter Griffith at MIT in the 1970s. Another correlation, called the Saha–
Zuber correlation, can also be used to calculate the NVG point. The Saha–Zuber correlation says that the bulk tempera-
ture of the coolant at the NVG point is given by
Pe = GD e c P k (25.58)
and k and cP are the thermal conductivity and the heat capacity of the liquid phase. The bulk temperature TBULK(z) is
found by performing an energy balance from the channel inlet up to the NVG point. A typical energy balance in this
case yields
∫
TBULK (z) = TINLET + 4 GD e c P ⋅ q′′(z) dz (25.59)
the integral in Equation 25.59 can be evaluated easily, and the resulting expression for TBULK(z) is
Then the location of the NVG point zNVG can be found by substituting Equation 25.61 into Equation 25.57 and solving for
the value of z that causes the combined equation to balance. When this is done, we find that the NVG point is about 0.50
m downstream of the ONB point and the NVG point occurs approximately 0.33 m upstream of the zero quality point
where the equilibrium flow quality is zero. Because it is a relatively modern correlation, the Saha–Zuber correlation is
used to find the NVG point in RELAP 5. However, a value for the critical enthalpy hCRITICAL is found in this case where
RELAP starts to treat the fluid as a two-phase mixture. The critical enthalpy is determined from the Peclet number
which can be written as
( )( )
Pe = D e ρv µ ⋅ cµ k (25.63)
A more detailed discussion of how the NVG point is found is presented in the RELAP 5 user’s manual (see the URL
we have provided at the end of the chapter). In general, the Saha–Zuber correlation predicts a much earlier location
for the NVG point than the Griffith and Browning correlations do. In fact, under some conditions, it predicts that the
NVG point is reached between 10 cm and 15 cm earlier in most fuel assemblies. Thus, it is a less conservative cor-
relation than either the Griffith or Browning correlations for reactor work. Table 25.7 summarizes the correlations we
have discussed so far.
TABLE 25.7
Some Popular Correlations Used to Describe Flow Boiling in Reactor Coolant Channels
25.19
Finding the Cladding Temperatures in the NB Regime
with the Jens–Lottes Correlation
In this section, we would like to illustrate how the cladding temperatures can be found after nucleate boiling begins, but
before the bulk temperature becomes equal to the saturation temperature. To do so, we would like to use the Jens–Lottes
Correlation, which we introduced earlier. Since the cladding temperature is equivalent to the fuel rod surface temperature
1
∆T(z) = TCLAD (z) − TCOOLANT (z) = 1.897e( − P / 900) q′′(z) 4 (25.65)
We can use this expression to find the cladding temperature after the coolant reaches its saturation temperature. In this
case, the cladding temperature is simply
1
TCLAD (z) = TSAT + 1.897e( − P / 900) q′′(z) 4 (25.66)
If the core average heat flux is q′′AVG, and the core power peaking factor is F, then when boiling occurs, the cladding
temperature at the point of maximum heat flux is
1
TCLAD (z) = TSAT + 1.897e( − P / 900) q′′(z) 4 (25.67)
where q′′MAX = F · q′′AVG. Typically, in a well-designed core F has a value of about 2.0. Example 25.4 illustrates how the
cladding temperature and the boiling heat transfer coefficient can be found under these conditions.
If we denote the maximum heat flux by q′′max , then the cladding temperature at this point is given by
0.25 − p / 900
TCLAD = TSAT + 1.897 q′′max e
At a system pressure of 2,250 PSI, the saturation temperature is 652.67°F (344.8°C). The maximum heat flux in the
coolant channel is then given by q′′max = F·q″ = (2.4) (189,800) = 455,500 BTU/h ft2 and the cladding temperature at that
point is TCLAD = 652.67 + 1.897(455,500)0.25e−2,250/900 = 652.67 + 4.05 = 656.7°F (347°C). Here, TCLAD is the temperature
of the cladding at its outer surface. [Ans.]
25.21 Estimating the Void Fraction as a Function of Axial Elevation 993
Notice that at the point of maximum heat flux, the heat transfer coefficient is much larger than it is for single-phase flow.
The relative magnitudes of the heat transfer coefficients for single- and two-phase flows are shown in Table 25.8, which
was initially presented to the reader in Chapter 24. Another lesson to be learned from this simple example is that a small
amount of nucleate boiling is not necessarily a bad thing as long as it can be controlled. Boiling increases the value of
the heat transfer coefficient, and this makes it easier to remove the heat. The amount of boiling that can be tolerated in
a reactor core is generally a function of its thermal design margins. These design margins are different for PWRs than
they are for BWRs, and the differences for these important reactor types are discussed in Chapters 27 and 33.
25.20
D ependence of the Equilibrium Quality on the Mass Flux
In BWRs, the equilibrium quality depends on the mass flux G. As the mass flux increases, the equilibrium quality gets
smaller, and as the mass flux decreases, it gets larger. In practice, the actual quality x(z) approaches the equilibrium qual-
ity xeq asymptotically beyond the NVG point. In some theories, this asymptotic behavior is described by the equation
where zeq is the location of equilibrium quality, zNVG is the location of NVG, and M = (xeq[zeq]/xeq[zNVG]) − 1. The depen-
dence of the equilibrium quality on the mass flux is shown in Figure 25.12. Thus, the outlet equality decreases more
than linearly with the mass flux, and at a mass flux of 6,500 kg/m2 s, it is only 0.02 (2%). Conversely, at a mass flux of
1,500 kg/m2 s, it is 0.14 (14%). In the literature, the stated exit quality for a General Electric advanced BWR is 14.5%.
25.21
Estimating the Void Fraction as a Function of Axial Elevation
In general, the void fraction affects the value of the heat transfer coefficient during flow boiling. In BWRs, the exit
quality is limited to about 18% for the hottest fuel assemblies in the core. Conversely, in PWRs, it is limited to about
1% at the top of the core. When the vapor phase slides past the liquid phase, the void fraction is related to the quality by
TABLE 25.8
The Convective Heat Transfer Coefficients in Reactor Fuel Assemblies with Single- and Two-Phase Flow
Typical Heat Transfer Coefficients for Reactor Fuel Assemblies (W/m2 °C)
Flow Type Water Liquid Metals Gases
Natural convection (SP) 300–1,200 1,000–3,000 5–30
Forced convection (SP) 25,000–75,000 40,000–120,000 50–500
Pool boiling (natural) 1,000–12,000 N/A N/A
Flow boiling (forced) 200,000–300,000 N/A N/A
In general, the heat transfer coefficients for two-phase flow are much larger than they are for single-phase flow.
994 HEAT TRANSFER CORRELATIONS
{ ( ) }
α = 1 1 + ρv ρl ⋅ (1 − x)/x ⋅ S (25.69)
(see Chapter 23) where S is the slip ratio, which represents the ratio of the velocity of the vapor phase to the velocity of
the liquid phase
S = v v v l (25.70)
If the liquid and vapor phases move together at the same speed, the slip ratio is unity (S = 1). For the purpose of this
discussion, consider the pattern of bubble formation and growth shown in Figure 25.10 again. Prior to point B, there are
no bubbles and the void fraction is 0. Between points B and C, bubbles develop at the surface of the wall, but they col-
lapse back into the liquid core because the bulk temperature is far below the saturation temperature. Therefore, although
bubbles are created, they only exist for a short period of time, and the void fraction is still 0. At point C (the ONB),
bubbles form close to the surface of the wall and they do not collapse back into the surrounding liquid. Here, the void
fraction becomes measurable, but from a thermodynamic perspective, the equilibrium quality and void fraction are still
negative because the bulk temperature has not yet reached the saturation temperature. Then, at point G (the point of
NVG and bubble detachment), the bulk temperature of the fluid is still 5°C–10°C below the saturation temperature, but
bubbles begin to form, and they have enough thermal energy that they do not collapse back into the surrounding liquid.
Finally, at point H, the flow transitions from subcooled boiling to saturated boiling, and the equilibrium quality x
has a value greater than zero (x > 0) for the first time. Hence, the bulk temperature of the fluid TBULK does not reach
the saturation temperature TSAT until we get to point H. This is consistent with our earlier comment that nucleate boil-
ing can begin with a modest amount of subcooling (5°C–10°C) in most reactor fuel assemblies. When we plot the void
fraction as a function of axial elevation, z, the real void fraction looks like what is shown in Figure 25.12, while the
void fraction that is calculated using a thermodynamic equilibrium model looks like the one shown with the blue line.
FIGURE 25.12 A picture illustrating the actual void fraction and the equilibrium void fraction in a boiling core. Note: The axial
elevations are not drawn to scale.
25.21 Estimating the Void Fraction as a Function of Axial Elevation 995
Obviously, there is a disconnect between the two models because there are real bubbles before we get to saturated boil-
ing (at the NVG point), but if we assume that the liquid and vapor phases are in thermal equilibrium (i.e., they have the
same temperatures), then there is not. For this reason, the ONB, and the real void fraction starting at the NVG point,
must be known with some precision or the reactivity calculation for the core, which depends on an accurate value for the
coolant density, will be in error. It is usually possible to estimate the surface temperature at which local boiling begins
from the Jens–Lottes correlation or the Thom correlation. Hence, although there is no single, fixed temperature where
the boiling actually begins, the value of the wall temperature TONB where boiling first occurs (the ONB point), can be
found from the simple formula
where h is the convective heat transfer coefficient. In commercial BWRs, the difference between the location of the
ONB point zONB, and the location of the bulk boiling point zBB, can be significant. For a typical hot channel, the bulk
fluid temperature can reach the saturation point somewhere between 40 and 50 in. (~100 cm) from the bottom of the
channel, while the ONB point will be reached approximately 12 in. (~25 cm) from the bottom of the core. For the hot
channel, the difference in this distance can be as great as 1 m! (Refer to the schematic in Figure 25.13.) Hence, it is usu-
ally NOT acceptable to use an equilibrium thermodynamic model (where Tl = Tv) to find where the ONB point occurs.
Doing so can lead to severe errors in the calculation of the void fraction, particularly in the lower quarter of the core. The
same effect can also occur in the hottest channel of a PWR if the assumption of thermal equilibrium is used. However,
in this case, the error is generally lower (about 0.5 m under standard conditions). The easiest way to avoid this problem
entirely is to model the flow of the liquid and the vapor phases independently, which gives rise to what is called a two-
fluid model. Then the liquid and vapor phases can have entirely different temperatures, and in general, Tl ≠ Tv. The heat
FIGURE 25.13 The locations of the ONB point, the NVG point, and the zero quality point in a typical 4 m core with a
standard power profile. Sometimes the NVG point is also called the bubble detachment point. Note: The axial elevations are not
drawn to scale.
996 HEAT TRANSFER CORRELATIONS
flow qlv between the phases is then equal to the temperature difference between the phases, multiplied by a constant of
proportionality C similar in function to the convective heat transfer coefficient h. The interfacial heat flow is then
q vl = CA c ( Tv − Tl ) (25.72)
where Ac is the surface contact area. For this approach to work, it should be apparent that Tv must always be greater than
Tl. Finally, in a two-fluid model, the velocities of the phases can be different (vl ≠ vv). The equations that are needed to
include these effects (unequal phase velocities and a lack of thermal equilibrium) are discussed in Chapter 29. However,
in practice, the solutions to these equations must be found numerically.
25.22
Closing Comments and Observations
Thus we have taken the opportunity in this chapter to discuss the critical heat flux and the amount of wall superheat
under which dryout occurs. In the next chapter (Chapter 26), we would like to show how the fuel pin, the cladding,
and the coolant temperatures can be found when the mass flow rate and the heat flux are known. In real reactors, these
calculations are normally performed using computer programs such as TRAC, TRACE, RELAP, COBRA, VIPRE, or
CATHARE. However, when it comes to designing a real core, a great deal of additional information can be learned
about how the core behaves by making some simplifying assumptions. Consequently, this is the path that we would
next like to pursue. This will lead us to a discussion of reactor design margins (e.g., thermal design limits) for BWRs
and PWRs. It will then lead us to a discussion of the DNB ratio (or the DNBR) for PWRs and the critical power ratio
(or the CPR) for BWRs. In practice, most design margins are statistical in nature in order to account for the fact that
statistical uncertainties exist in the values of the mass flow rate, the fuel rod dimensions, and the average enrichment of
the fuel. In the next few chapters, we will show how these statistical uncertainties can be included in the calculation of
the DNBR and the CPR. Generally speaking, they result in more conservative values for the DNBR and the CPR than
deterministic calculations provide.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Web References
http://wwwme.nchu.edu.tw/Enter/html/lab/lab516/Heat%20Transfer/chapter_4.pdf.
http://www4.ncsu.edu/~doster/NE402/Text/BoilingHeatTransfer/BoilingHeatTransfer.pdf.
http://tutorial.math.lamar.edu/Classes/DE/SeparationofVariables.aspx.
https://pdfs.semanticscholar.org/0824/790be46f9d59cd1c5440176037653a434532.pdf.
http://www.iaea.org/inis/collection/NCLCollectionStore/_Public/32/037/32037353.pdf.
Other References
RELAP-5/MOD3 Code Manuel: www.iaea.org/inis/collection/NCLCollectionStore/_Public/27/008/27008531.pdf.
35. Which one of the correlations in this chapter uses a fluid-dependent parameter F to show the difference various
fluids can have on the convective heat transfer coefficient after flow boiling occurs?
36. Which of the correlations we have presented in this chapter (if any) are applicable to the hottest fuel assemblies in
a spent fuel pool when the flow to the pool is lost for a significant period of time?
37. What is the boiling number, Bo, and how is it defined and used in boiling heat transfer work?
38. What is the convection number, Co, and how is it defined and used in boiling heat transfer work?
39. In a reactor fuel assembly, what would a Froude number of 1.0 imply?
40. What does a thermal design margin in a reactor fuel assembly represent? To what important nuclear parameters are
thermal design limits normally applied?
41. Name at least three popular computer programs that can be used to model boiling heat transfer in reactor fuel assem-
blies. If you were a reactor designer, which one of these programs would you select and why?
42. Where is the zero equilibrium quality point usually reached in a BWR core? Is the NVG point reached earlier or
later than this?
43. What is the Saha–Zuber correlation used for in nuclear heat transfer?
44. Fill in the following sentence with the appropriate word or phrase: The ONB point is normally reached about
________ m before the NVG point is reached in a modern BWR.
45. What is the physical significance of the NVG point in a BWR and why is it also called the BDP or bubble detach-
ment point?
46. Name at least two correlations that can be used to find the NVG point after boiling begins in a water reactor fuel
assembly.
47. What is the physical significance of the NVG point, and how is it different than the zero quality point or the ONB
point?
48. Fill in the following sentence with the appropriate word or phrase: The Jens–Lottes correlation is designed to handle
both NB and bulk boiling. However, it cannot be used for _________ in nuclear power plants
49. Once the bulk temperature reaches the saturation temperature in an American BWR, approximately how many
degrees above the saturation temperature does the cladding temperature usually remain?
50. How would you attempt to determine the location of the ONB point in a PWR?
51. The coolant pressure in a thermal water reactor is gradually increased from 7 to 14 MPa. According to the Thom
correlation, by approximately what percentage is the amount of wall superheat raised?
52. In the subcooled NB regime, what three important parameters can be inferred from the Bernath correlation?
53. What is the approximate velocity of the coolant in a BWR coolant channel? Does the coolant normally remain in
the channel long enough for thermal equilibrium to be reached?
54. What is the name of the heat transfer correlation that should be used to find the convective heat transfer coefficient
in a water reactor after film dryout occurs? What is the explicit form of this correlation and what is its stated range
of applicability?
55. What is the equilibrium quality at the entrance and the exit of a modern BWR fuel assembly?
56. What is the equilibrium quality at the entrance and the exit of a modern PWR fuel assembly?
57. Thermodynamically, what does an equilibrium quality less than zero or greater than one represent?
58. What range of equilibrium qualities can the Bernath correlation, the Chen correlation, and the Groeneveld correla-
tion be used for in reactor thermal-hydraulics work?
59. Which two popular flow boiling correlations have the lowest overall margin of error in reactor work?
60. What is the stated range of applicability of the Groeneveld correlation for reactor heat transfer? What is the lowest
value of the equilibrium quality and the highest value of the equilibrium quality for which it can be used?
61. At what values of the equilibrium quality can the Dittus–Boelter correlation be used to find the convective heat
transfer coefficient in reactor fuel assemblies?
62. What is the approximate value or range of the convective heat transfer coefficient in a PWR fuel assembly during
normal operating conditions?
63. What is the approximate value or range of the convective heat transfer coefficient in the boiling region of a BWR
fuel assembly during normal operating conditions?
64. Fill in the following sentence with the appropriate word of phrase: In general, boiling occurs in BWR fuel assem-
blies in the top _______% of the core.
65. Approximately how far upstream of the zero quality point in a typical BWR fuel assembly are the NVG point and
the ONB point? At which point in the assembly can it be assumed that bulk boiling begins?
66. Out of all the correlations presented in this chapter, which correlations would you attempt to use to find the convec-
tive heat transfer coefficient on the secondary side of a PWR steam generator? How accurate would you expect these
correlations to be for this particular application?
Exercises for the Student 999
Exercise 25.2
For the fuel rod in Exercise 25.1, calculate the convective heat transfer coefficient. What is the temperature of the clad-
ding at the surface of the rod? Can the boiling under these conditions be best described as NB or bulk boiling?
Exercise 25.3
At the entrance to a BWR fuel assembly where the bulk temperature of the coolant is 280°
C and the pressure is 1,050 PSI (7.25 MPa), calculate the equilibrium quality. Is the equilibrium quality zero, positive,
or negative under these conditions?
Exercise 25.4
Using the Bernath correlation and the equilibrium quality found in Exercise 25.3, calculate the temperature at the sur-
face of a nuclear fuel rod, the heat transfer coefficient, and the surface heat flux.
Exercise 25.5
Pressurized water flows over the surface of a nuclear fuel rod whose heated surface is 4 m in length. The mass flux in the
vicinity of the rod is G = 3,000 kg/m2 s. The Reynolds number is 500,000, the Prandtl number is 0.5, and the ambient
pressure is 1,000 PSI (6.9 MPa). Using the Klimenko correlation, calculate the convective heat transfer coefficient at the
rod’s surface. Under these conditions, can boiling be best described as NB or film boiling? Assume a heat flux of 5 × 106
W/m2 and an equilibrium quality of 7%.
Exercise 25.6
Using the Chen correlation, calculate the convection enhancement factor and the NB suppression factor for a two-phase
flow where the convection number is 0.2, the Reynolds number is 500,000, and the equilibrium quality is 5%.
Exercise 25.7
After film dryout, calculate the convective heat transfer coefficients over a nuclear fuel rod using the Dittus–Boelter and
Groeneveld correlations. Assume the fluid-dependent parameter λ = 1.0, the Reynolds number is 500,000, the Prandtl
number is 1.0. Also assume the equivalent diameter is that of a representative BWR subchannel and the appropriate
value of the thermal conductivity to use is 0.3 W/m °C.
Exercise 25.8
Boiling water at 7 MPa (~1,000 PSI) flows through the tube of a PWR steam generator. The surface temperature of
the tube is 290°C, and the saturation temperature of the surrounding water is 285°C. The mass flux G is 450 kg/m2 s.
Calculate the forced convection enhancement factor and the NB suppression factor when the flow quality is 20%.
Exercise 25.9
Using the information provided in Exercise 25.8, calculate the convective heat transfer coefficient and the NB heat
transfer coefficient. Which one is larger? What is the approximate value of the total heat transfer coefficient under these
conditions?
Exercise 25.10
Using the information provided in Exercise 25.9, calculate the surface heat flux based on the difference between the wall
temperature and the saturation temperature. How does this compare to the heat flux over the surface of a PWR fuel rod
in the primary loop?
1000 HEAT TRANSFER CORRELATIONS
Exercise 25.11
For a surface heat flux of 500,000 W/m2, calculate the amount of subcooling at the NVG point in a BWR fuel assembly
where the boiling heat transfer coefficient is 250,000 W/m2°C. If the saturation temperature is 286°C, what is the average
temperature of the coolant at the NVG point?
Exercise 25.12
Using the Jens–Lottes correlation, calculate the cladding temperature where the ONB occurs in a commercial PWR
where the saturation temperature is 345°C. Assume a boiling heat transfer coefficient of 250,000 W/m2 °C and a peak
surface heat flux of 1,000,000 W/m2. Where exactly in the core would you expect this to occur?
26
Core Temperature Fields
26.1 T hermal Behavior of the Fuel and the Cladding
The nuclear heat generation rate Q and the mass flow rate m determine how hot the fuel and the cladding can become. In this chapter,
we would like to discuss the temperature profiles in the fuel and the cladding. We would also like to discuss the coolant temperature
profiles in the radial and axial directions. About 95% of the reactors in the world today use uranium dioxide (UO2) for their primary
fuel.* Its melting point is between 2,760°C and 2,810°C (5,000°F–5,190°F) when it is fresh; however, its melting point declines at a
rate of about 22°C per 10,000 MWD/MTU when it is burned (see Chapter 11). Thus, at burn ups of about 50,000 MWD/MTU, when
it is removed from the core, its melting point is a pproximately 2,680°C (4,856°F). Its thermal conductivity follows a similar trend;
however, at beginning of life (or BOL) it is about 0.025 W/cm °C.
During normal operation, the fuel centerline temperature is limited to about 2,200°C because higher fuel pin temperatures tend to
release too many fission products into the fuel.* This behavior places certain restrictions on the thermal-hydraulic behavior of the core.
Thus, reactors are designed so that during normal operation, the fuel is never allowed to melt and the cladding is never allowed to fail.
These criteria are then responsible for letting limits on the departure from nucleate boiling (DNB) ratio and the critical power ratio,
which we will discuss in Chapter 27. The melting point of the cladding depends on the type of material that is used. The most common
types of cladding are alloys of zirconium and stainless steel. These alloys include Zircaloy-2 and Zircaloy-4, and Stainless Steel 304
and 316 (SS-304 and SS-316) (see Table 26.1 for a complete list). Zircaloy-2 is used in boiling water reactors (BWRs), and Zircaloy-4
is used in pressurized water reactors (PWRs). Stainless steel is more prevalent in fast and gas reactors. In general, the strength of the
cladding does not decline appreciably as the fuel is burned. However, in water reactors, the NRC limits its maximum permissible tem-
perature to about 2,200°F (1,204°C) to prevent the buildup of hydrogen gas in the core. When the cladding temperature exceeds this
limit, hydrogen gas is produced from what is called the zirconium water reaction. This is an extremely violent exothermic chemical
reaction that is difficult to control. We will have more to say about the dynamics of the zirconium water reaction in Chapter 34. Other
important thermophysical properties of the materials used nuclear fuel rods are shown in Table 26.2. Taken collectively, Tables 26.1
and 26.2 provide enough information to find the radial and axial temperature profiles in almost any nuclear fuel rod. In reactors where
the fuel is surrounded by metallic cladding, a small “gap” is usually included between the fuel and the cladding to allow for the thermal
expansion of the fuel. This gap is very small, and in most reactors, it is only 0.10 mm across. It is usually filled with an inert gas such
as helium to improve its initial conductivity. Then as the fuel is burned, the gap also serves as a repository for additional fission product
gases such as xenon and krypton. The thermal conductivity of helium is kHELIUM = 0.002 W/cm °C, and the thermal conductivity of the
xenon–krypton mixture is k MIX = 0.001 W/cm °C. Thus, the gap temperature drop changes as the fuel is burned. Finally, for cylindrical
fuel rods that do not contain a central hole or void, the temperature drop ΔTFUEL in the radial direction is given by
∆TFUEL = q′ 4 πk FUEL (26.1a)
Hence, the temperature drop across the fuel does not depend on the radius of the fuel. Instead, it depends on only the linear power
density q′ and the thermal conductivity of the fuel. This explains why the linear power density is such an important core parameter. In
most PWRs and BWRs, an average value for q′ is about 180 W/cm. In the hot channel, its maximum value is between 400 W/cm and
500 W/cm. These numbers depend on the burnup of the fuel, but they are very representative of most light water reactors.
26.2
Mechanistic Thermal Design
As we mentioned earlier, the volumetric heat generation rate q‴(x, y, z) = Q(x, y, z)/V FUEL determines how hot the fuel and the
cladding become. In general, q‴ can be thought of as an independent variable that reflects the isotopic composition of the core. In
addition, the value of q‴ depends on whether or not the core is surrounded by a reflector.† The higher the enrichment of the fuel,
* See Nuclear Engineering Fundamentals: A Practical Perspective by R.E. Masterson, CRC Press (2017).
† See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
1001
1002 Core Temperature Fields
TABLE 26.1
The Melting Points of the Fuel and the Cladding for Common Nuclear Materials
Melting Points of Common Materials Used for the Fuel and the Cladding
Material Temperature (°C) Temperature (°K) Temperature (°F)
Metallic uranium 1,132 1,405 2,069
Metallic plutonium 640 913 1,184
Fuel (UO2) a 2,800 3,120 5,156
Fuel (PuO2)a 2,390 2,663 4,334
Fuel (UC)a 2,365 2,638 4,289
Cladding (SS-304) 1,440 1,710 2,618
Cladding (SS-316) 1,490 1,760 2,708
Cladding (zirconium) 1,855 2,125 3,365
Cladding (Zircaloy-2) 1,845 2,115 3,347
Cladding (Zircaloy-4) 1,880 2,150 3,410
a Note: Appropriate for fresh fuel only.
TABLE 26.2
Important Thermal Properties of Common Nuclear Materials
the more likely it is that the volumetric heat generation rate will be high. Normally the value of q‴ is a function of
core geometry, and it is different for cores that are cylindrical in shape than those that are spherical or rectangular.
Finally, the value of q‴ depends on how the fuel is loaded into the core.* Once these parameters are known, we can
then derive explicit expressions for the temperatures of the fuel, the cladding, and the coolant. These expressions
also depend on the mass flow rate m. The fuel and cladding temperature profiles are different for PWRs than they
are for BWRs, and we would like to explain some of the reasons for these differences in the sections that follow.
Also, the design of the fuel can affect the shape of the temperature profiles, and annular fuel rods have different
temperature profiles than solid fuel rods (see Chapter 11). Perhaps the easiest way to find these temperatures in a
reactor core (see Figure 26.1) is to start with the shape of the heat flux in the axial direction and to use this shape
to determine the temperatures of the coolant, the cladding, and the fuel. In general, these temperatures are position
dependent, and they can vary from one fuel assembly to the next. In fact, they can even vary from one fuel rod to the
next. We can then use the coolant temperature profiles to derive the temperature profiles for the fuel and the clad-
ding. Although the cladding temperature, the fuel temperature, and the coolant temperature are a function of x, y, and
z, it is easiest to first derive an expression for the coolant temperature in the axial direction and then deduce the other
temperature profiles from this. To a first approximation, the shape of the axial power profile in a spatially uniform
and unreflected cylindrical core is given by
* See Nuclear Engineering Fundamentals: A Practical Perspective by R.E. Masterson, CRC Press (2017).
26.2 Mechanistic Thermal Design 1003
FIGURE 26.1 The temperature fields in a reactor core can be very complex. In this chapter, we would like to discuss these fields in
fuel rods, coolant channels, and reactor fuel assemblies.
MAX is the maximum volumetric heat generation rate (located at r = 0 and z = 0), z is the
where R is the core radius, q′′′
axial location, and Jo is called a Bessel function of the first kind. Bessel functions of the first kind were introduced to
the reader in Chapter 5, and so the reader is encouraged to consult that chapter for a more information regarding their
behavior. Their general shape as a function of their order is presented in Figure 26.2. For a uniform and unreflected
cylindrical core, Equation 26.1b applies to the entire core. Consequently, the power level reaches its maximum value
when r = 0 because at this point Jo(r) = 1. Here, the volumetric power generation rate q‴ becomes
FIGURE 26.2 There are several different types of Bessel functions, but the most important Bessel function in reactor design is a
Bessel function of the first kind. It is represented by the symbol Jo, and the shape of this function for a bare cylindrical core is shown
in red. Notice that it is a function of the argument xo, which in reactor work, is equal to xo = 2.405r/R. Hence, at the center of the
core (where r = 0), it has a value of 1.0, and at the edge of the core (where r = R), it has a value of 0. Bessel functions of the second
kind, which are represented by the symbol J1, are shown in black and appear in solutions to the neuron diffusion equation for radio-
active line sources. (See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).).
1004 Core Temperature Fields
FIGURE 26.3 When using the equation q ′′′ TOTAL (r, z) = q ′′′
MAX ·Jo(2.405r/R)·cos (πz/H) to describe the power profile for a cylindrical
core, the origin of the z and r coordinate systems is placed at the centerline of the core. Thus, the top of the core is located at z = H/2
and the bottom of the core is located at z = −H/2. The power profile in the axial direction is given by the term cos (πz/H), and
the power profile in the radial direction is given by the term Jo(2.405r/R). Most reactor physics books also use this convention to
describe the shape of the neutron flux.
where H is the core height, and z is equal to 0 at the core midplane. Hence, according to Equation 26.2, the bottom of
the core is located at z = −H/2, the center of the core is located at z = 0, and the top of the core is located at z = H/2 (see
Figure 26.3). Equation 26.2 can then be used to describe the volumetric heat generation rate for the central fuel assembly
or for the central fuel rod. For fuel rods and fuel assemblies that are displaced radially from the central rod, we must
apply a radial correction factor to Equation 26.2 to properly specify the three-dimensional power shape. The form of
this correction factor is F RADIAL(r) = Jo(2.405r/R) where r is the distance from the center of the core. In practice, this cor-
rection factor has a value greater than 0 and less than 1.0. However, if we ignore this correction factor for the moment,
then the axial heat generation rate is given by Equation 26.2 alone.
26.3
A Coolant Energy Balance
Consider for the moment the hot channel surrounding the central fuel rod where we would like to find the coolant tem-
perature. The axial power generation rate at this location is
q′′′(z) = q′′′
MAX ⋅ cos(πz/H) (26.3)
where all of the terms have their usual meaning. As the coolant moves along the surface of the fuel rod, heat is trans-
ferred between the fuel rod and the coolant at a rate given by Newton’s Law of Convection:
q = q′′′VFUEL = mc
p ∆TCOOLANT (26.4)
where m is the mass flow rate over the surface of the rod, cp is the specific heat of the coolant, and ΔTCOOLANT is the dif-
ference between the bulk temperature of the coolant TBULK and the surface of the cladding TCLAD (ΔTCOOLANT = TCLAD −
TBULK) at each axial elevation between z and z + Δz. In the discussion that follows, we will replace TCOOLANT by TCOOL
to simplify the notation. This expression for the heat transfer rate is valid along the entire surface of the rod as long as
the coolant does not boil. If the surface area of the fuel rod is A, then the amount of heat leaving the rod along the same
axial distance Δz is given by
26.4 Axial Coolant Temperature in the Hottest Coolant Channel 1005
Equating these two expressions to each other, an energy balance between the fuel rod and the coolant at each axial
elevation z requires that
or in differential form
dTCOOL dz = q′′′(z)A ( mc
p ) (26.7)
Equation 26.7 can then be used to determine the temperature of the coolant.
26.4
A xial Coolant Temperature in the Hottest Coolant Channel
When we substitute our previous expression for the maximum power generation rate
q′′′(z) = q′′′
MAX ⋅ cos(πz/H) (26.8)
into Equation 26.7, the equation for the temperature profile of the coolant becomes
dTCOOL dz = q′′′ (
MAX ⋅ cos (πz/H) A mc )
p (26.9)
If we then integrate Equation 26.9 from the point where the coolant enters the core (at z = −H/2) to an arbitrary point
along the z axis, we obtain
MAX ⋅ A ⋅ H ( πmc
TCOOL (z) = TIN + q′′′ p ) [1 + sin(πz/H) ] (26.10)
or alternatively,
where TIN is the coolant temperature at the inlet to the core, and VFUEL = AΔz is the volume of the fuel in the fuel rod.
Finally, since qMAX = q′′′
MAX V FUEL , we arrive at the following expression for the coolant temperature when the axial heat
flux is co-sinusoidal:
This equation gives the value of the coolant temperature at any axial location along the fuel rod in the hottest channel
in the core. In a subsequent section, we will learn that the temperature profile in another channel can simply be obtained
by multiplying the value for q′′′
MAX by a radial correction factor F RADIAL , which normally has a value less than 1.0 (i.e.,
F RADIAL ≤ 1.0). For a homogenous, cylindrical, and unreflected core, the radial correction factor is
where R is the radius of the core, Jo is a Bessel function of the first kind, and r is zero at the center. The values for Jo are
shown in Figure 26.2 in red as a function of the argument 2.405r/R. Notice that when r = 0, Jo has a value of 1.0, and in
this case, the axial temperature profile for the coolant in the hot channel reduces to the expression we derived previously.
However, for r > 0, the temperature profile has the same shape, but the overall increase in the coolant temperature will
be lower. According to Equation 26.12, the coolant temperature in the hot channel increases as we move up through the
channel and it reaches its maximum value at the top of the core. Here, the maximum coolant temperature is given by
and the total increase in the coolant temperature as it flows through the core is
FIGURE 26.4 The axial temperature profile of the coolant as a function of the distance from the inlet to the outlet. In this case, the
temperature refers to the bulk coolant temperature. The hot channel has the greatest temperature increase, and the cooler channels have
smaller temperature increases. However, for the same axial power shape q‴(z), all of the temperature profiles are essentially similar.
where again, VFUEL is the volume of the fuel (in m3), m is the mass flow rate (in kg/s), and cP is the specific heat of the
coolant. The reader may recall from our previous discussion that we must use the value of cP rather than the value of cV
because the pressure drop across the core is negligible compared to the average core pressure of approximately 2,250
PSI (15.5 MPa). The behavior of the coolant temperature is shown in Figure 26.4 for the hot channel as well as for a
number of adjacent subchannels with a co-sinusoidal axial heat flux. Note that the adjacent channels are exposed to a
slightly lower axial heat flux and therefore have lower values for FRADIAL. Consequently, the shape of the temperature
profile is the same in each case except that the temperature increase is lower. This implies that the hot channel is gener-
ally the most limiting channel when it comes to the thermal design of the core.
26.5
Corrections for Changes in the Radial Power Distribution
For channels other than the central hot channel, the value of q‴ is lower because radial power peaking factors are also lower.
For a homogenous, cylindrical, and unreflected core, we learned previously that the radial correction factor could be written as
where R was the radius of the core. However, in practice, a reactor core is almost always spatially heterogeneous, and the
value of this correction factor can vary from one fuel assembly to the next. In fact, it can even vary from one subchan-
nel to the next. If j is the index of the fuel assembly and there are J fuel assemblies in the core, and i is the index of the
coolant channel within the jth fuel assembly, and there are I coolant channels within each fuel assembly, then the radial
correction factor can be written as
where i is the index of the ith channel in the jth fuel assembly, and i = 1, 2, 3, … I, and j = 1, 2, 3, … J. The expression
for the temperature of the coolant in the ith channel of the jth fuel assembly then becomes
Obviously, this expression depends on the fact that power generation rate q‴ is co-sinusoidal in shape. If it departs from
this shape, then the coolant temperature profiles will be different as well. Example Problem 26.1 illustrates what this shape
becomes when the power generation rate is spatially uniform in the axial direction. In this case, the temperature of the coolant
26.6 Cladding Temperatures for the Hot Channel 1007
continues to rise as we move through the core, and it eventually reaches its maximum value at the core outlet. However, when
the axial power profile is completely uniform (i.e., when q‴ = q′′′ MAX everywhere), then the coolant temperature profile (for
single-phase flow) increases linearly as a function of axial elevation z. Neglecting the effects of the control rods, most PWRs
tend to have an axial coolant temperature profile that shares some of the characteristics of both (linear and co-sinusoidal).
The exact shape of the profile depends on the design of the core and where the individual fuel assemblies are placed.
26.6
Cladding Temperatures for the Hot Channel
Now that we have found the coolant temperature profile, the temperature at the surface of the cladding TCLAD can
be deduced as a function of axial position by applying Newton’s Law of Convection to the fuel rod and the coolant.
However, as we shall shortly see, this requires a specific knowledge of the convective heat transfer coefficient h between
the surface of the cladding and the coolant. Under most conditions, the flow is turbulent and fully developed. We can
then use the Dittus–Boelter correlation (see Chapter 21) to find the value of h for the hot channel. When the coolant
remains subcooled and does not boil, the heat transfer coefficient is
( ) ( )
h = k D e ⋅ Nu = k D e ⋅ 0.023(Re)0.8 (Pr)0.4 (W m 2
)
°C (26.19)
In general, the Dittus–Boelter correlation leads to lower than expected values for h in fuel assemblies with pitch-to-
diameter ratios close to 1.33 (the industry average for commercial PWRs). However, to obtain this conservatism, we must
calculate the fluid properties around the fuel rods at the bulk temperature of the coolant. Finally, to use the Dittus–Boelter
equation, we must assume that the coolant does not boil. If the coolant boils, the correlations for h become more compli-
cated, and they can even become flow-regime dependent (see Chapter 25). We will consider what this implies in future
sections. However, assume for the moment that we are able to calculate a value for h, and that the shape of the coolant
temperature profile in the axial direction is already known. Then Newton’s Law of Convection implies that
where ACLAD is the surface area of the cladding, TCLAD is the surface temperature of the cladding, and TCOOL is the bulk
temperature of the coolant. We also know from our previous discussion that
q(z) = q′′′
MAX VFUEL ⋅ cos(πz/H) (26.21)
where VFUEL is the volume of the fuel. Equating these two expressions gives
where the heat transfer coefficient for single-phase flow is assumed to be constant. This energy balance between the
coolant and the cladding is valid at every axial elevation along the fuel rod surface as long as the axial heat conduction
through the cladding can be ignored. Inserting the value of TCOOL from our previous discussion, and solving for the value
of TCLAD, we find that
where TIN is the coolant temperature at the core inlet. With a little rearrangement, we can also write this as
where RCONV is the thermal resistance to convective heat transfer at the fuel rod’s surface and q = q′′′
MAX VFUEL . Hence,
Equation 26.24 can be applied to either cylindrical fuel rods or plate-type fuel rods by simply using the appropriate
expression for RCONV (see Chapter 10).
26.7
Finding the Location of the Maximum Cladding Temperature
Equation 26.24 gives the value of the cladding temperature at any axial location in the hottest channel in the core.
According to this equation, the cladding temperature in the hot channel increases as we move up the channel and reaches
its maximum value somewhere between the midpoint of the core and the top of the core. The position where this occurs
can be found by simply taking the derivative of the cladding temperature with respect to z and setting it to zero to find
the location (zMAX) where it reaches its maximum value. Setting dTCLAD/dz = 0, and solving for zMAX, we obtain
The Location of the Maximum Cladding Temperature with a Co-sinusoidal Axial Heat Flux
z MAX = (H/π) cot −1 ( πmc
p RCONV ) (26.25)
where α = π m cp RCONV and TIN is the temperature of the coolant at the core inlet. In a reactor with a co-sinusoidal power
distribution, this equation demonstrates that the cladding temperature always reaches its maximum value before the top
of the core and almost always above the core midplane. Typically it reaches its maximum value about three quarters of
the way between the bottom and the top of the core. See Figure 26.5. This figure illustrates the cladding temperatures for
the hot channel and for a few adjacent subchannels that are exposed to a lower heat flux. Notice that the shape of the tem-
perature profile is the same in each case, except that the absolute temperature increase away from the hot channel is lower.
This implies that the hot channel is generally the most limiting channel when it comes to the cladding temperature as well.
26.8
T he Coolant and Cladding Temperatures when the Axial Heat Flux Is Uniform
If the axial power profile is uniform, then we can derive a similar equation for the axial temperature profile. The equa-
tion we must solve to find the cladding temperature in the hot channel is then
FIGURE 26.5 The cladding temperature in a typical PWR at the outer surface of the cladding. The temperature at the inner sur-
face of the cladding is often 60–70°C higher. Exact values are provided in Chapter 11. (Taken from the RELAP5 user’s manual.)
26.9 Temperature Drop from the Cladding to the Coolant 1009
However, as we stated previously, the coolant temperature in this case increases linearly with z. If the coolant enters the
core at z = −H/2 with a value of TCOOL = TIN, the cladding temperature as a function of axial elevation becomes
or more concisely
and M = 1/(πmc p). This means that when the axial power profile is the same everywhere, the cladding temperature
reaches its maximum value at the top of the core (where z = H/2):
TCLAD MAX = TIN + q ⋅ R CONV + M ⋅ H (26.32)
If we compare this expression to the cladding temperature when the axial power profile is co-sinusoidal, the location of
the peak temperature is different. This implies that we must always know the shape of the axial power profile if we want
to calculate the location of the peak cladding temperature correctly. If the peak cladding temperature is significantly
above the boiling point of the coolant, the point of minimum DNB will also occur very close to this location. However,
it will be offset slightly downstream for some of the reasons that we discussed earlier.
26.9
Temperature Drop from the Cladding to the Coolant
In the previous sections, we mentioned that the heat flux q″ from the surface of the cladding to the coolant is given by
Newton’s Law of Convection
which we derived in Chapter 10. Here h is the convective heat transfer coefficient at the cladding surface. The value of h
depends on the properties of the coolant as well as the axial mass flow rate m and the shape of the coolant channel. For
water, h can range from 25,000–75,000 W/m2 °C for single-phase forced convection to about 200,000–300,000 W/m2 °C
(if boiling occurs), to only 50–500 W/m2 °C for gaseous coolants such as helium in gas reactors. For a cylindrical fuel
rod where the radius of the cladding is RCLAD, Newton’s law can also be written as
For a linear heat generation rate of q′ = 500 W/cm (50,000 W/m), and a cladding radius of RCLAD = 0.45 cm (0.0045 m),
the value of ΔTCOOL is ΔTCOOL = 50,000/(0.28h). When h = 50,000 W/m2 °C, the temperature drop is ~3.6°C if no boiling
occurs. The surface heat flux is then q″ = q′/(2πRCLAD) = 500/(0.9π) = 177 W/cm2 = 17,700 W/m2. Temperature drops of
this magnitude are common in water-cooled reactors with single-phase flow. If flow boiling is allowed to occur, the heat
1010 Core Temperature Fields
transfer efficiency in the nucleate boiling regime is so high that the cladding surface temperature will rise no more than
3–6°C above the fluid saturation temperature in the top 75% of the core. Hence, the cladding temperature will stay well
below its melting point as long as a fuel rod is not completely engulfed by a vapor film.
26.10
Finding the Temperature Profiles for the Fuel
Finally, we would like to discuss how the fuel pin temperature profiles behave when the coolant does not boil. This is
almost always true in PWRs and it is also true in liquid metal fast breeder reactors (LMFBRs). We can use the results
we presented previously to find the profiles in this case. There are two basic ways in which this can be done:
1. First we can start with the cladding surface temperature and then work inward to the center of the pin. In this
approach, the cladding temperatures are given by Equations 26.24 and 26.30. Then, we must know the thermal
resistance for the fuel, the cladding, and the fuel to cladding gap. Alternatively, if we are only interested in the
fuel pin centerline temperature, then we can use the thermal resistance of the fuel rod to estimate it. The thermal
resistance for both cylindrical and plate-type fuel rods is
where RCONV is the convective resistance between the cladding and the coolant. Normally the convective resistance is
inversely proportional to the value of the convective heat transfer coefficient h:
where A is the surface area of the rod. Both approaches lead to the same results, but the second approach is a little easier
to illustrate. Normally, the fuel pin centerline temperature must be kept about 25% below the melting point of the fuel
during normal reactor operation. For uranium dioxide fuel rods where the melting point is ~2,800°C, the temperature
of the fuel at the centerline of the rod must be restricted to about 75% × 2,800°C ≈ 2,100°C in the hot channel during
steady-state operation. We would now like to demonstrate how these two approaches compare.
26.10.1
Approach #1: Start with the Coolant Temperature to Find the Fuel Temperature
Suppose that the coolant temperature around the hottest fuel rod is known as a function of axial elevation. Normally,
this requires us to average the coolant temperatures from the surrounding subchannels, such as the subchannels shown
26.10 Finding the Temperature Profiles for the Fuel 1011
in Figure 26.6a. Hence, in this case, both q‴(z) and TCOOL(z) are assumed to be known. The energy balance that must be
satisfied at each axial elevation is then
where ACLAD is the surface area of the cladding, VFUEL is the volume of the fuel inside of the rod, TFUEL is the maximum
temperature of the fuel (presumably at the fuel centerline), and RTOTAL is the thermal resistance for the entire rod. Since
q(z) = q‴(z)·VFUEL, we can also rewrite Equation 26.36 as
Since q(z) and TCOOL(z) are already known, we simply have to know the value of the thermal resistance to find the value
of TFUEL(z). For a cylindrical rod, we know from Chapter 10 that the total thermal resistance of the rod (including the
convective resistance at the surface) is
FIGURE 26.6 (a) The average temperature of the coolant surrounding a fuel rod can be found by averaging the temperatures from
the four surrounding subchannels in a PWR and the six surrounding subchannels in a LMFBR. (b) The axial temperature profile
for the fuel as a function of the distance from the inlet to the outlet. In this case, the temperature refers to the fuel pin centerline
temperature. The hot channel has the highest fuel pin temperatures, and the cooler channels have lower ones.
1012 Core Temperature Fields
where R FUEL = 1/(4πk F) is the thermal resistance of the fuel, RCLAD = tCLAD/(2πkC R FUEL ) is the thermal resistance of the
cladding, RGAP = 1/(2πRGAPhGAP) is the thermal resistance of the fuel to cladding gap, and RCONV = 1/(2πRChCONV) is
the convective resistance at the cladding surface. Here, R F is the radius of the fuel, RGAP is the radius of the gap, t is
the cladding thickness, and RC is the radius of the cladding. For plate-type fuel rods, the corresponding expressions
for the thermal resistance become
where W is the width of the fuel and A is its surface area. Hence, the expression for the maximum fuel temperature is
Thus, the maximum fuel temperature will always be higher than the coolant temperature adjacent to it by an amount
given by q(z)RTOTAL . Now let us examine what the fuel pin temperature profile looks like for the hot channel when we
vary the shape of q‴(z). If q‴(z) = q′′′
MAX cos (πz/H), then the expression for the maximum fuel pin temperature is
The Fuel Centerline Temperature when the Axial Heat Flux is Co-sinusoidal
and
The Fuel Centerline Temperature when the Axial Heat Flux is Constant
TFUEL (z) = TCOOL (IN) + q ( πmc
p ) ⋅ (z + H/2) + q MAX RROD (26.47)
For channels other than the hot channel, then we must then replace qMAX(z) by
q(i, j,z) = FRADIAL (i, j)q MAX (z) (26.48)
where F RADIAL(i, j) is the radial power peaking factor of the channel (typically less than 1.0), and the index (i, j) refers
to the ith channel in the jth fuel assembly in the core. If we plot the fuel pin centerline temperatures for the hot chan-
nel, as well as in some of the cooler channels around it, we obtain the graph shown in Figure 26.6(b). As Figure 26.6(b)
indicates, the maximum fuel pin temperature in a commercial PWR during normal operation is about 1,800°C. The
actual centerline temperature may be higher or lower at different points in the operating cycle. The temperature on the
outer surface of the fuel pin (see Chapter 11) is about 500°C (or 1,300°C lower). After reviewing these numbers, we find
that ΔTCOOL ≅ 5°C, ΔTCLAD ≅ 70°C, ΔTGAP ≅ 130°C, and ΔTFUEL ≅ 1,300°C. The total temperature drop between
the center of the pin and the coolant is ΔTROD = ΔTCOOL + ΔTCLAD + ΔTGAP + ΔTFUEL ≅ 1,500°C. Here, the average
coolant temperature is assumed to be 300°C.
26.10 Finding the Temperature Profiles for the Fuel 1013
The shape of the temperature profiles is shown in Figure 26.7 for a constant axial power profile and in Figure 26.8
for a more conventional co-sinusoidal one. Notice that increasing the mass flow rate m reduces the value of TMAX(z)
because the second terms on the right-hand side of Equations 26.46 and 26.47 are lower. If the power profile is constant,
FIGURE 26.7 Axial temperature profiles for a cylindrical fuel rod in a PWR core with a uniform heat flux.
FIGURE 26.8 Axial temperature profiles for a cylindrical fuel rod in a PWR core with a co-sinusoidal heat flux.
1014 Core Temperature Fields
the maximum centerline temperature is reached at the top of the core, but if the power profile has a more typical co-
sinusoidal shape, then the fuel pin temperature profile will peak somewhere between the midpoint of the core and
the top. In a typical core, notice that the cladding temperature always peaks after the fuel temperature does, and this
offset is on the order of 5%–10% of the total core height. However, if the power profile is completely uniform (q‴[z] =
constant), then the coolant temperature, the cladding temperature, and the fuel temperature all peak at the same location
(near the top of the core). If the coolant does not boil, every reactor with vertical core flow will follow this same basic
trend. There are two reasons why the maximum pin fuel temperature peaks above the core midplane under these condi-
tions. First, the temperature of the coolant continues to rise beyond the midplane as long as additional heat is added to
it. Irrespective of the shape of q(z), the fuel temperature profile must increase according to the equation
where R ROD is the total thermal resistance of the rod. Notice that if q(z) = 0 or q(z) = constant, the fuel temperature pro-
file has the same shape as the coolant temperature profile. Second, the heat flux q″(z) is determined solely by the value
of q‴(z). The combined effect of TCOOL(z) and q‴(z) determines what the value of TMAX(z) must be. If q‴(z) drops further
up the channel, but TCOOL is still increasing, then there must be a point somewhere between the middle of the core and
the top of the core where the fuel pin temperature reaches its maximum value. Typically, this point is about midway
between z = 0 (the middle of the core) and z = +H/2 (the top). The maximum cladding temperature, which determines
where the minimum departure from nucleate boiling ratio (MDNBR) point occurs, is then offset about 5%–10% of the
core height from this location.
26.10.2
Approach #2: Start with the Cladding Temperature to Find the Fuel Temperature
Sometimes the cladding temperature is known and we would like to infer the fuel pin temperature from it. This approach
is simpler than the previous approach because the convective heat transfer coefficient is not required at the cladding
surface. Suppose that the temperature of the cladding is known at the cladding surface as a function of axial elevation.
From a simple energy balance, we know that the fuel and cladding temperatures must be related by
where R ROD is the thermal resistance of the rod alone. In the case of plate-type fuel rod, the thermal resistance is
( )
RPLATE = (1/A) W 2k f + 1 h GAP + t k c (26.51)
where W is the width of the fuel, t is the thickness of the cladding, A is the surface area of the rod, and hGAP is the gap
convective heat transfer coefficient. Similarly, for a cylindrical fuel rod the thermal resistance is
( )
RCYLINDER = 1 4 πk f + 1 2πR GAP h GAP + 1 2πk c ⋅ t R F (26.52)
where R F is the radius of the fuel, t is the thickness of the cladding, and hGAP is the gap heat transfer c oefficient which
we also discussed in Chapter 11. The maximum fuel pin temperatures are then given by
TFUEL MAX (z) = TCLAD (z) + q MAX (z) ⋅ RPLATE ( for the plate) (26.53a)
TFUEL MAX (z) = TCLAD (z) + q MAX (z) ⋅ RCYLINDER ( for the cylinder ) (26.53b)
when the fuel rods are solid. If they have been in the core for a while and develop a central void, then another expres-
sion for RFUEL must be used (see Chapter 11). Hence when the axial power profile is constant we must set q(z) = q in
Equations 26.53a,b and when it is co-sinusoidal we must set q(z) = qMAX·cos (πz/H). Again, these equations are based on
the assumption that the origin is placed at the core mid-plane. This means that the bottom of the core is located at z =
−H/2, the core mid-plane is located at z = 0, and at the top of the core is located at z = H/2 (see Figure 26.2). Almost all
nuclear science and engineering textbooks follow the same convention as well.
Solution We know from our previous discussion that ΔTGAP = q″RGAP. In a typical PWR (see Problem 26.2)
q″ = 150 W/cm2. If ΔTGAP = 150°C, the value of RGAP must be RGAP = ΔTGAP/q″ = 1.0 cm2 °C/W. In practice, the value of
RGAP changes as the fuel is burned, but this is a very representative number. The convective resistance at the outer surface
of the cladding is RCONV = 1/(hA). Since h is typically 50,000 W/m2 °C (or 5 W/cm2 °C) and A = πD is 2–3 cm, the thermal
resistance across the gap is 10 to 20 times larger than the convective resistance at the outer cladding surface. [Ans.]
26.11
Finding the Locations of the Maximum Fuel and Cladding Temperatures
To find where the maximum fuel and cladding temperatures occur as a function of axial elevation, we simply take the
derivatives of TFUEL(z) and TCLAD(z) with respect to z and set them equal to zero. When we do so, we find that the maxi-
mum fuel and cladding temperatures are located at
where
( )
RROD = 1 ( 4 πk f ) + 1 ( 2πR GAP h GAP ) + 1 ( 2πk c ) ⋅ t R F + 1 ( 2πR C h ) ( for a cylindrical fuel rod ) (26.55b)
Similarly, we find that zCOOL MAX = H/2, which means that the coolant reaches its highest temperature at the top of the
core. These equations can be used to predict the approximate locations of the peak fuel, cladding, and coolant tem-
peratures in a water reactor with a vertical core flow pattern. The maximum values of TFUEL, TCLAD, and TCOOL can then
be found by substituting the values for zFUEL MAX, zCLAD MAX, and zCOOL MAX into the previous equations for the fuel, the
cladding, and the coolant temperatures. With a little work, it is easy to see that the maximum temperatures are
( ( )) α (26.56b)
TCLAD MAX = TCOOL (IN) + qRCONV 1 + √ 1 + α 2
where
α = πmc
p R CONV (26.57)
β = πmc
p R ROD (26.58)
The only major assumptions that have been made so far are
☉☉ The coolant does not boil
☉☉ The axial conductivity of the fuel and the coolant are small
In PWRs and gas reactors, these assumptions result in only small errors in the core temperature profiles. However, in
LMFBRs, where the coolant can have a high thermal conductivity, it is not possible to ignore the effects of axial heat
conduction completely. The expressions we have derived for the temperature profiles may then have an error on the
order of ±2% when the flow rate is low. However, when the flow rate is high, these errors are generally insignificant.
Finally, in boiling water reactors, the expressions we have just derived work well up to the point where the coolant starts
to boil. Beyond that point, we then have to modify the equations to account for the effects of two phase heat transfer.
In particular, we have to modify the equation that governs the bulk coolant temperature. We will demonstrate how this
can be done in Section 26.12.
1016 Core Temperature Fields
26.12
Finding the Fuel, the Cladding, and Coolant Temperatures in Real Cores
In real PWRs, there can be as many as 200 fuel assemblies, and between 200 and 300 fuel rods per fuel assembly. Hence
there can be between 40,000 and 60,000 fuel rods, and a similar number of coolant channels in a very large core. A typi-
cal reactor core in which this is true is shown in Figure 26.9. A representative fuel assembly is then shown in Figure 26.10.
In practice, the axial heat flux does not fall to zero exactly at the edge of the core. In fact, it maintains a large and finite
value over the entire active core height (see Figure 26.8). An energy balance over the length of the core gives
z OUT
h(z) = h IN + PH ( GA ) ⋅
∫
z IN
q′′(z) dz (26.59)
where the integration must be performed from zIN = −H/2 to zOUT = +H/2 and G is the axial mass flux. The actual shape
of q″(z) under these conditions can be approximated by replacing the standard expression for the heat flux
where now H′ = H + 2δ, and the additional distance δ on the top and the bottom of the core is called the extrapolation
distance.* When the coolant temperature is re-evaluated using Equation 26.61, the actual c oolant temperature profile
becomes
( ) ( ) ( )
p sin πH 2H ′ + sin πz H ′ (26.62)
TCOOL (z) = TIN + q′′o ⋅ 2R CLAD H ′ mc
Without an axial reflector, a light water reactor will have a value for δ of 10–15 cm. Hence if the active height of the core
is 4 m, the value of H/H′ is 400/430 ~ 0.93 (which translates to a 7% difference). Using our earlier work on conductive
FIGURE 26.9 A large commercial PWR contains approximately 200 fuel assemblies. The fuel assemblies are arranged in the
shape of a rough circular cylinder, and some of the fuel assemblies have different thermal power outputs than others. In general, the
power output is proportional to the enrichment of the fuel. Here the enrichments are represented by different fuel assembly colors.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
26.13 Coolant Temperatures in a Boiling Core 1017
FIGURE 26.10 A fuel assembly in a PWR core. Normally, there are between 250 and 350 fuel rods per assembly, and each of these
rods can generate a different amount of thermal energy.
heat transfer in cylindrical fuel rods, we can then work backwards to calculate the fuel and cladding temperature pro-
files. The resulting profiles are shown in Figure 26.8. In general, the fuel, the cladding, and the coolant temperatures
are higher because more thermal energy is deposited in the coolant. These expressions assume that the effect of the
control rods is not included in the shape of q″(z).* If we temporarily ignore the presence of the control rods, then the
shape of q″(z) will be similar from one coolant channel to another, and the coolant and cladding temperatures will follow
very similar trends. For PWRs, this implies that the location of the MDNBR will be somewhere near the top of the hot
channel where the cladding temperature peaks. Normally, it can be found just downstream of the point where the peak
cladding temperature occurs. We will have more to say about this in Chapter 27.
26.13
Coolant Temperatures in a Boiling Core
In PWRs, the coolant normally does not boil, but in BWRs, it does. When the coolant boils, it is helpful to subdivide the
core into two separate regions or zones called the boiling and non-boiling heights. The lower part of the core, where the
coolant does not boil, is referred to as the non-boiling height HNB, and the upper part of the core, where the coolant boils,
is referred to as the boiling height HB. Thus, the sum of the boiling and non-boiling heights is equal to the total core height:
H = H B + H NB (26.63)
See Figure 26.11. The exact transition point between HB and HNB is often a subject of debate because it depends on how
the elevation where boiling first occurs zBOIL is defined. Normally, zBOIL is defined to be the point where the bulk tem-
perature of the coolant becomes equal to the saturation temperature (i.e., TBULK = TSAT). However, as Figure 26.12 shows,
some local boiling can occur long before TBULK = TSAT. In fact, in some BWRs, it can occur as much as two-thirds of a
meter before the bulk temperature reaches TSAT.
When we look at this problem more closely, we find that the onset of nucleate boiling (ONB) (point B or the ONB
point) occurs long before TBULK = TSAT (point D). In BWRs, the difference between the ONB point zONB, and the bulk boil-
ing point zBOIL, can be considerable because the coolant never has a chance to reach an equilibrium state. For the hot chan-
nel, the bulk fluid temperature can reach its saturation point about 1 m from the bottom of the core, while the ONB point
is reached about one-third of a meter from the core inlet. Hence, the difference in the value of HNB (depending upon the
model used) can be as great as two-thirds of a meter! Exactly the same thing happens in the hot channel of a PWR when
the assumption of thermal equilibrium is used. However, in this case, the difference in the value of HNB is about 0.5 m
because the system pressure and the flow rate are higher. Thus an equilibrium thermodynamic model requires us to use
q(z) = h ⋅ A CLAD ( TCLAD (z) − TCOOL (z) ) (26.64)
when the boiling point is based on the bulk coolant temperature alone.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
1018 Core Temperature Fields
FIGURE 26.12 In many reactor cores, boiling can occur long before the equilibrium quality = 0 point is reached. This is because
the coolant velocities and the heat generation rates are so high that thermal equilibrium in the coolant channels is almost never
reached.
Hence, the coolant remains a subcooled liquid when TCOOL < TSAT, and it becomes a two-phase mixture when TCOOL =
TSAT. In other words, the effect of bulk boiling on the axial temperature profile can be reduced to the problem of finding
the appropriate values of the coolant temperature and the convective heat transfer coefficient in the boiling and non-
boiling heights, HB and HNB. If we assume that the energy balance mentioned previously applies, then for the non-boiling
height, the appropriate equation to use for TCOOL(z) is
26.14 Fuel and Cladding Temperatures in a Boiling Core 1019
where A FUEL is the surface area of the fuel, and TIN is the temperature of the coolant at the core inlet. Here, the integra-
tion of the axial power profile q‴(z) must be carried out between z = −H/2 (the bottom of the core) and z = HNB, the
non-boiling height. Thus, the values of h and TBULK must be found in such a way that this energy balance is maintained.
The appropriate value of q‴(z) can then be inserted into Equation 26.65 and integrated to obtain the axial temperature
profile. However, as soon as the coolant reaches its saturation t emperature, the boiling can no longer be considered to be
localized, and we must use the fact that TCOOL(z) = TSAT above z = HNB. This leads us to conclude that
Thus, to a first approximation, this is how the axial temperature profiles in a boiling core are found. Now let us see how
the fuel and the cladding temperatures behave under these conditions.
26.14
Fuel and Cladding Temperatures in a Boiling Core
In boiling cores, the previous equations can be used up to the point the coolant starts to boil. In this case, it is convenient
to subdivide the core into two regions (or axial zones) called the boiling and non-boiling regions (see Figure 26.12) and
to find the coolant temperature in the non-boiling region using Equation 26.65. However, when enough additional energy
is added to the coolant for the bulk temperature to equal the saturation temperature, then we must set TCOOL(z) = TSAT
from the point where bulk boiling begins to the top of the core. After this point is reached, the equations for the cladding
temperature become:
( )
TCLAD (z) = TSAT + qM z + H 2 + qRCONV ( when the axial heat flux is uniform) (26.68)
( ) (
TCLAD (z) = TSAT + qM 1 + sin π z H + ( qRCONV ) cos π z H )
( when the axial heat flux is co-sinusoidal ) (26.69)
and the equations for the fuel pin temperature profiles become
where M = 1/(πmc p). Together with our previous equations, these expressions can be used to find the coolant tempera-
tures, the cladding temperatures, and the fuel pin temperatures in both the boiling and non-boiling regions. Normally
the cladding temperature in BWRs is several °C above the saturation temperature when the coolant boils. However, the
value of the thermal resistance RCONV in Equations 26.65 and 26.66 is also flow regime dependent. In fact, a different
value of hCONV (from which RCONV is derived) must be used for the nucleate boiling, the bulk and film boiling, and the
mist flow regimes. In other words, the values of hNB, hBB, and hMF are then imbedded in the values of R NB, R BB, and R MF.
Correlations to find the correct values for h and R NB, R BB, and R MF were provided for the reader in Chapter 25. The loca-
tions where nucleate boiling, bulk and film boiling, and mist flow boiling occur are also shown in Figure 26.13. Hence
in a boiling core, at least three different correlations are required to find the appropriate values for RSP, R NB, R BB, and
R MF. These include the Dittus–Boelter correlation for hSP and RSP; the Chen, Klimenko, or Kandlikar correlations for
hNB, R NB, hBB, R BB, hFB, and RFB; and the Groeneveld correlation for hMF and R MF. Under some conditions, other combi-
nations of correlations can be used. The axial locations of each of these boiling regimes can then be inferred from the
local values of the quality, the void fraction, and the mass flow rate. Example Problem 26.5 illustrates a situation where
these correlations can be used.
1020 Core Temperature Fields
FIGURE 26.13 The flow regimes in a boiling core when a reactor is being operated normally. Notice that the coolant enters the
core as a subcooled liquid and nucleate boiling begins at the ONB point. This continues until the NVG or bubble detachment point
where bulk boiling is assumed to begin.
26.15
Comparing the Behavior of PWR and BWR Cores
BWRs and PWRs are classic examples of boiling and non-boiling cores. Notice that bulk boiling in a BWR begins about a
quarter of the way up the core (see Figure 26.11), and by the time the core midplane is reached, the boiling can become very
pronounced. In PWRs, the coolant stays a single-phase liquid, except in the hottest channels near the top of the core, where
a small amount of nucleate boiling is sometimes allowed to occur. Even there, the equilibrium quality rarely exceeds one-
half of 1%, while in BWRs at the same location, the equilibrium quality is in the 15%–18% range. These differences in the
behavior of the coolant cause the temperature of the coolant in a PWR to rise gradually until the top of the core is reached,
while in BWRs, the coolant temperature rarely changes above the core midplane. (Above the midplane, the remaining
water is just converted into more steam.) The heat transfer coefficients in BWRs gradually increase with elevation, while in
PWRs, they do not (except for minor changes to hSP). The higher heat transfer coefficients keep the fuel pin temperatures
about the same as they are in PWRs, although the surrounding coolant temperature does not change. Hence, both boiling
and non-boiling cores end up with approximately the same peak fuel pin temperatures (in the range of 1,700°C–1,800°C)
during steady-state operation. There is some variation in these numbers that we will seek to clarify in Section 26.16.
26.16 Accounting for Statistical Uncertainties in the Core Temperature Profiles 1021
26.16
Accounting for Statistical Uncertainties in the Core Temperature Profiles
In practice, many of the parameters that are used to calculate the core temperature profiles are not known with absolute
precision. This is because many of them (specifically the mass flow rate m, the heat transfer coefficient h, and the power
level P) tend to fluctuate about an average value or mean. For turbulent flow, this also applies to the velocity—see
Figure 26.14. Finally, some other parameters such as the specific heat cp and even the radius of the fuel rods are only
known with a certain level of precision, and their values can often change subtlety as the fuel is burned. Thus, many
of the parameters that are used to calculate the fuel and cladding temperature profiles can be represented by statistical
probability distributions (SPDs), and each of these distributions can have different medians, variances, and standard
deviations. Thus, the equations we have used so far (which are deterministic in nature) are based on the average value
a or the expected value of these physical and thermophysical properties. In some cases, the fluid velocity, the mass flow
rate, and the even equilibrium quality can be time dependent as well. This implies that the temperature profiles we must
calculate are subject to various statistical uncertainties, and thus, we must consider the effect of these uncertainties on
the behavior of the core as a whole. For example, in modern PWRs, the coolant in normally subcooled water. However,
the velocity V and the mass flow rate m have an expected or average value v or m , but they also have a range of other
possible values that cluster around these values (see Figure 26.15). In BWRs, the equilibrium quality x, the void fraction
α, and sometimes the flow velocity also fluctuate with axial elevation or when the power-to-flow ratio becomes too high.
FIGURE 26.14 In reactor cores where the flow is turbulent, the coolant velocity fluctuates about an average value or mean.
FIGURE 26.15 The distribution of probabilities as a function of the number of standard deviations σ in a normal probability
distribution.
1022 Core Temperature Fields
Thus, the heat transfer coefficients (which are a function of the void fraction and the q uality) can only be determined
with a certain degree of precision. This in turn introduces statistical uncertainties into the calculation of the fuel and
cladding temperatures. When finding the core temperature profiles, these uncertainties must be taken into account by
introducing engineering uncertainty factors into the calculation of the fuel, the cladding, and the coolant temperatures.
These statistical uncertainty factors then describe the probability that a particular variable (such as the maximum fuel
temperature or the maximum cladding temperature) will fall within a range of expected values a certain percentage
of the time. Reactor licensing then uses a detailed knowledge of these statistical uncertainty factors to determine the
conditions under which a plant can be safely operated.
26.17
SPDs and Engineering Uncertainty Factors
A probability distribution for the variables used to find the fuel pin temperatures can have virtually any shape at all, but
normally it follows a normal or bell-shaped probability distribution whose profile is shown in Figure 26.15. Also, some-
times the process of manufacturing a reactor component introduces additional statistical uncertainties into the values
of important reactor parameters such as the enrichment of the fuel, the radius of the fuel pins, and even the thickness of
the cladding. These statistical uncertainties effect the value of q‴(z), which, in turn, determines the temperature of the
fuel. To account for these statistical uncertainties, an engineering uncertainty factor or manufacturing uncertainty fac-
tor is normally applied to each variable that is used to calculate the fuel and cladding temperatures. This means that the
predicted temperature profiles must also follow SPDs (statistical probability distributions) that have an expected value
or mean. Thus, if the predicted value of T at a particular location is T , then the actual values of T will become less and
less likely as one departs further away from the mean (see Figure 26.15). The engineering uncertainty factor Fv for a
variable v that follows a normal or bell-shaped probability distribution is
Fv = ( v + 3σ (v) ) v (26.72)
or
Fv = 1 + 3 σ (v) v (26.73)
where v is the average value of the variable, and σ(v) is its standard deviation. For example, in the case of the tempera-
ture field T, the engineering uncertainty factor can be defined as
FT = 1 + 3σ (T) T (26.74)
The highest temperature TMAX that we will find within three standard deviations of the calculated value of the tempera-
ture T is then given by
TMAX = FT ⋅ T (26.75)
In other words, over 99% of the time, the measured value of the fuel, the cladding, and the coolant temperature will be
less than the maximum value defined by Equation 26.75. Normally, F T has a value between 1.05 and 1.10. Therefore,
when one is asked to specify the actual temperature that a specific component is designed to withstand, one must take
each of the individual values for T and multiply them by an appropriate engineering uncertainty factor FT to account for
all of the engineering uncertainties. In the case of the fuel, the cladding, and the coolant, these values become
In general, the values of F T FUEL, F T CLAD, and F T COOL are not identical because different assumptions and manufacturing
tolerances are used to determine their exact values. Exactly how these values are determined when hundreds of overlap-
ping probability distributions must be considered at the same time is a topic that is best left to the statisticians. However,
a value of about 1.12 for each factor is generally considered to be conservative when designing the core to withstand a
Category II accident without causing the fuel rods to fail. The actual temperatures that a reactor component can reach
depend on how far it can deviate from the average value T or mean. For example, in the probability distribution shown
in Figure 26.15 (which happens to be a normal probability distribution), there is about a 67% chance that the actual value
26.17 SPDs AND ENGINEERING UNCERTAINTY FACTORS 1023
TABLE 26.3
The Probability of Occurrence as a Function of the Number of Standard Deviations for a Normal Probability
Distribution
will fall within one standard deviation of the mean ( T ± σ), about 95% chance that the actual value will fall within two
standard deviations of the mean, ( T ± 2σ), and about 99.7% chance that the actual value will fall within three standard
deviations ( T ± 3σ) of the mean. The degree to which a normally distributed engineering variable v can be found out-
side of different points on this distribution is shown in Table 26.3. For a normal probability distribution, this deviation is
often described by what is called the 67-95-99.7 rule. The 67th percentile, the 95th percentile, and the 99.7th percentile
then correspond to points on the distribution that are one standard deviation, two standard deviations, and three standard
deviations from the mean of the distribution. The two points that are one standard deviation from the mean are some-
times called the inflection points of the distribution. When distributions of this type are applied to nuclear fuel rods,
one finds that the maximum fuel pin temperature in a PWR under normal operating conditions is about 1,800°C with
standard deviation of about 120°C (~7%), and the maximum fuel pin temperature in a BWR is about 1,700°C with a
standard deviation of about 50°C (~3%). These values are based on the analysis of many commercial power reactors at
different points in their operating cycles. Example Problem 26.6 illustrates the application of this concept to a nuclear
fuel rod where some statistical uncertainties exist in the density of the fuel (e.g., the density of UO2). This in turn causes
uncertainties in the value of q″(z) which can be significant when the standard deviation becomes large.
26.18
Reactor Nodal and Subchannel Analysis
A similar approach can be used to analyze the thermal-hydraulic behavior of individual fuel rods and coolant chan-
nels. In reactor fuel assemblies, these coolant channels are called subchannels and their characteristics were dis-
cussed in Chapter 18. In PWRs and BWRs, these subchannels are arranged in square arrays, and in LMFBRs, they
are arranged in triangular or hexagonal ones. The subchannels for each reactor type are shown in Figure 26.16.
Notice that there can also be corner, side, or central channels, and these channels depends in part on whether the
fuel assembly is surrounded by a sheath or a can. Large BWR cores have about 800 fuel assemblies, and each fuel
assembly contains between 64 and 81 rods. Large PWR cores have about 200 fuel assemblies, and each assembly
contains between 289 and 361 rods. Thus, for the same electrical output, the number of fuel rods in both reactor
types is about the same. In American PWRs, the fuel assemblies do NOT have sheaths or cans, and the cores of these
reactors are referred to as open cores. In “open” fuel assemblies, some of the fuel rods are replaced by control rod
guide tubes and instrumentation channels.
When the control rods are withdrawn from the core, the power level increases, and the guide tubes become filled
with water. This water helps to moderate the fission neutrons and produce more thermal neutrons. In the center
of each fuel assembly, four fuel rods normally deposit their thermal energy into a single coolant channel, and in
LMFBRs, the number of fuel rods that deposit this thermal energy into a single subchannel is usually six. Hence,
the expressions for q‴ we have been using so far must account for the fact that there may be contributions from more
than one fuel rod to the thermal energy of the coolant. The energy which is added to each channel is then found by
partitioning the fuel rods in the manner shown in Figure 26.17. In some text books, this is referred to as segmenting
the fuel rods. Suppose that the fuel rods are numbered with the prefix R and the subchannels are numbered with the
prefix S. In square fuel assemblies, the heat that flows into a single subchannel is then equal to the sum of the heat
flows from the surrounding fuel rods. For example, in the case of subchannel 6 which is surrounded by fuel rods 7, 8,
12, and 13 (see Figure 26.17), we have:
q S6 = 1 4 q R7 + 1 4 q8R + 1 4 q12
R
+ 1 4 q13
R
(26.79)
and the individual heat flows from these fuel rods are
R
q12 = h ( TROD12 − TCOOL6 ) (26.82)
R
q13 = h ( TROD13 − TCOOL6 ) (26.83)
FIGURE 26.16 The center, edge, and corner subchannels in light water reactor fuel assemblies (on the left) and in LMFBR fuel
assemblies (on the right).
26.19 Using Super Cells for Reactor Thermal-Hydraulic Analysis 1025
FIGURE 26.17 The flow of heat from four fuel rods into a single subchannel.
Then for example, the coolant channel temperature profile for subchannel 6 can be written as
TCOOL6 = TIN + q 6S ( πm
6 c p ) (z + H/2) (when the axial heat flux is uniform) (26.84)
where q S6 is given by Equation 26.79. For regulatory purposes, most reactor designers focus their attention on the tem-
perature profiles in the hottest fuel assemblies in the core. Normally, these fuel assemblies contain the hottest fuel rods,
and the assemblies that contain these rods normally have the highest average enrichment.
26.19
Using Super Cells for Reactor Thermal-Hydraulic Analysis
Unfortunately, it is not always practical or even feasible to find the temperature profiles in 50,000 to 60,000 fuel rods
and a comparable number of subchannels. To reduce the amount of work involved, smaller channels are sometimes
lumped together to form larger ones. This process is called subchannel homogenization . It begins by identifying the
unit cells from which the fuel assemblies are made. Different unit cells can be defined for PWRs, BWRs, and LMFBRs.
Examples of these cells are shown in Figure 26.18. These unit cells also affect the neutronic behavior of the core because
they determine the rate at which thermal energy is produced. This process is also described in our companion book*.
Normally there are two ways in which reactor unit cells can be defined:
1. First, they can be defined to consist of a single subchannel and a fraction of each of the fuel rods that surround
the channel. In the center of a PWR assembly, a subchannel of this type is then surrounded by four fuel rods, and
this is illustrated in Figure 26.17. If the fuel assembly is also surrounded by a can (which it can be in a BWR), the
peripheral channels will then receive heat from two fuel rods along the edges and one fuel rod at the corners.
2. Secondly, a unit cell can be defined to consist of single fuel rod surrounded by the subchannels that are adjacent
to it. Thus, in the center of a PWR or BWR fuel assembly, the fuel rod is surrounded by four subchannels, and the
corner fuel rods (if the fuel assembly happens to be surrounded by a can) are surrounded by two, three, or four
subchannels (depending on how the subchannels are defined). A PWR fuel assembly that uses this approach is
shown in Figure 26.18.
The biggest advantage to defining unit cells in this way is that the hottest rod can usually be identified within an indi-
vidual fuel assembly based on the enrichment of the fuel alone. The hottest rod can then be surrounded by a representa-
tive coolant channel having the same cross-sectional area and temperature as the subchannels from which it is formed.
The behavior of the hot rod can then be analyzed using the energy balances we presented earlier. Sometimes reactor
designers attempt to aggregate a number of unit cells together to form a larger cell known as a super cell. In principle,
it is then possible to treat an entire fuel assembly as if it were a single super cell. See Figures 26.19, 26.20, and 26.21 to
see how this can be done. In reactor design, a combination of unit cells and super cells is then used to determine how
1026 Core Temperature Fields
FIGURE 26.18 On the left, a unit cell can be defined to as a subchannel and four fuel rods or a fuel rod and four subchannels. On
the right, it can be defined as a subchannel and three fuel rods or a fuel rod and six subchannels. Both representations are used in
reactor work today.
FIGURE 26.19 A PWR fuel assembly divided into 225 unit cells. Here a fuel cell contains a fuel rod and a control cell
contains a control rod*.
the temperature profiles behave. Normally, sophisticated computer programs are used to automate this process. This
approach can also be applied to hexagonal fuel assemblies except that the number of subchannels surrounding the hot-
test rod is usually six and each of these subchannels is normally triangular in shape.
26.20
Conservatism in Subchannel Models
Once the subchannel temperatures are known, a temperature profile can be constructed showing all the fuel pin tempera-
tures in the core. For single-phase flow, this profile is generally conservative because the subchannels are assumed to be
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
26.20 Conservatism in Subchannel Models 1027
FIGURE 26.20 Sometimes reactor unit cells can be merged together to create larger thermal-hydraulic nodes called “super cells.”
FIGURE 26.21 These “super cells” can then be merged together to model the behavior of an entire fuel assembly. In reactor
thermal-hydraulics, this is sometimes called a “nodal approach.”
isolated and do not interact with one another. This assumption is called the isolated channel assumption. However, reac-
tor coolant channels with different temperatures and flow rates can exchange both mass and energy between them in the
lateral or transverse direction. The lateral mass flow that occurs between these channels is then called cross flow. In cores
with predominately vertical flows, this cross flow usually flows from hot subchannels to cooler ones (see Figure 26.22)
because the coolant accelerates as it expands. Fluid mechanically, this behavior can be described by the material deriva-
tive which we introduced in Chapter 15, and the pressure between the channels can change because the coolant will either
accelerate or decelerate. Any mass advected between the channels in the lateral direction causes the temperature in the
hotter channels to fall, and the temperature in the cooler channels to rise. This lateral mass transfer results in a lateral
advection of energy between the channels that reduces the temperature of the hotter channels by several °C. (The exact
1028 Core Temperature Fields
FIGURE 26.22 Cross flow between adjacent subchannels can occur if the temperature in one reactor coolant channel is differ-
ent than the temperature another. In general, the coolant flows laterally from the hot subchannels to the colder ones. Cross flow is
caused by lateral pressure differences induced by different power generation rates in a reactor’s fuel rods.
amount of energy transferred depends on the lateral mass flow rate, and the result of this lateral energy flow is shown
in Figure 26.23. Hence the energy transfer rate becomes proportional to the lateral mass flow rate and the energy that
is advected can be expressed as the product of the crossflow rate w (in kg/s) and the specific enthalpy h* (in J/kg) of the
coolant that is advected from the donor channel. However, this is not the only way that thermal energy can be exchanged
FIGURE 26.23 Turbulent mixing and cross flow reduce the local temperature peaks in reactor fuel assemblies because of the
lateral transfer of energy and momentum between adjacent subchannels. Collectively, these processes can reduce the local coolant
temperature peaks by several °C.
26.20 Conservatism in Subchannel Models 1029
between adjacent channels. In the addition to the advection of mass between the channels (which carries some of the ther-
mal energy with it), some energy and momentum can also be exchanged between the channels by what is called turbulent
mixing . In turbulent mixing, turbulent eddies randomly move between the channels in the lateral direction, and as they
move, they can carry both energy and momentum with them. Hence, even when there is NO net transfer of mass between
the channels, these turbulent eddies can carry enough energy and momentum between adjacent channels to flatten the
transverse temperature profiles. Moreover, the turbulent mixing rate increases as the flow becomes more turbulent. Most
of the time, the amount of turbulent mixing can be correlated to the Reynolds number. The resulting transfer of energy
and momentum between the channels in the lateral direction then consists of two separate and distinct parts:
(1) A turbulent energy and momentum transfer term (Term 1) which requires the existence of cross flow
(2) A turbulent energy and momentum transfer term (Term 2) which does not require the existence of cross flow
because no mass is exchanged between the channels by the eddies
In the first case, the rate of lateral enthalpy transfer is proportional to the cross flow rate w multiplied by the specific
enthalpy h* carried from the donor channel to the receptor channel:
Thus, when the lateral cross flow rate is small, the enthalpy transfer rate is small, and when the lateral cross flow rate is
large, the enthalpy transfer rate is large. The size of this cross flow is usually small compared to the axial mass flow rate
(normally it is only 1% to 2% of it), and in the absence of major changes to the geometry of the channels such as those
which are caused by fuel rod deformation during severe accidents or debris accumulation , it only affects the channels in
close proximity to another channel. For example, increasing the flow of energy into channel 6 in Figure 26.17 affects the
temperature profiles in channels 2, 5, 7, and 10, but it does not necessarily affect the temperature profiles in channels 12,
13, 14, 15, or 16. Hence only the adjacent channels are affected by this temperature change. Now let us turn our attention
to how the coolant temperature profiles are affected by each of these processes. When there is no crossflow and w = 0,
the only way that energy can be transferred between two adjacent channels is by turbulent mixing. Then rate of turbulent
energy exchange is proportional to axial mass flow rate m multiplied by a turbulent mixing coefficient CMIX. Within a
fuel assembly, the turbulent mixing coefficient between two neighboring channels is defined by
where m is the average axial mass flow rate in the two channels that exchange the turbulent eddies, and S is the surface
area per unit length through which the turbulent eddies flow. Normally, the cross flow rate between these subchannels is
measured in kg/ms or lb/ft s and S is measured in m2/m = m or ft2/ft = ft. In rod bundle geometries, the value of S then
becomes S = P − D, where P is the fuel rod pitch and D is the fuel rod diameter (see Figure 26.22). Thus, when there is
turbulent mixing in addition to lateral cross flow, the total enthalpy exchange rate can be written as
where h′ is the turbulent enthalpy that is exchanged, and h* is the enthalpy of the fluid in the donor channel. Normally,
the values for h′ and h* are similar. The turbulent mixing coefficient depends on the Reynolds number and the axial
mass flux G and it increases as either of these parameters increase. Secondly, it depends on the pitch-to-diameter (P/D)
ratio of the lattice. Then if the Reynolds number happens to fall, CMIX falls as well, and in the limit where the flow
becomes laminar, CMIX -> 0 because there are no longer any turbulent eddies to transfer turbulent energy and momentum
between the channels. Similarly, as the fuel rods get closer together and the pitch to diameter ratio (P/D) approaches 1,
the spacing between the rods become so small that the turbulent eddies have trouble moving laterally. Then CMIX also
approaches zero because there is no way for the turbulent energy carried by the eddies to be exchanged.
Thus, lateral pressure differences, which can lead to both cross flow and turbulent mixing when the Reynolds num-
ber is high, can transfer thermal energy from the hotter channels to the colder ones. Including both of these effects in
1030 Core Temperature Fields
FIGURE 26.24 An illustration of the subchannel model used for turbulent mixing and cross flow.
the calculation of the subchannel temperatures reduces the conservatism in the estimates of the coolant temperatures
and the coolant enthalpies by several percent. Finally, popular computer programs like COBRA and VIPRE define a
turbulent momentum factor (or FTM) to determine how efficiently turbulent flow transfers momentum between the
channels. In reactor work, the amount of this lateral momentum transfer is not very sensitive to the value of FTM. Values
of FTM close to 0.8 seem to provide the best overall match to experimental data, and the value of the FTM is normally
constrained between 0 and 1. The turbulent momentum factor used in VIPRE is closely related to the thermal diffusiv-
ity that we first defined in Chapter 20. For subcooled liquids, the turbulent diffusivity of momentum εM increases the
effective viscosity according to the equation
µ ′′ = µ + ρε M (26.89)
where ρ is the average density of the coolant. Thus the effective viscosity μ″ can be decomposed into the dynamic viscos-
ity μ, which is a tabulated property, and the turbulent viscosity ρεM, which is due to the turbulent motion of the fluid.
Again recalling that the kinematic viscosity ν can be written as ν = μ/ρ, we can also express the effective viscosity as
Effective Viscosity
µ ′′ = ρ( ν + ε M ) (26.90)
Consequently, the turbulent momentum factor (FTM) provides a direct link between the energy transfer rate and the
momentum transfer rate between two subchannels when the flow becomes highly turbulent. A typical subchannel model
that uses this methodology is shown in Figure 26.24. Most reactor fuel assemblies are analyzed using this approach.
Now let us discuss the exact form of the equations that reactor designers use.
Solution From Equation 26.87, it can be seen that H ′TOTAL = wh* + CMIXm h′. In this problem h* = h′ = 1,500 kJ/kg,
CMIX = 0.01 cm s/kg, m = 0.35 kg/s, and w = 0.01m . The total enthalpy transfer rate per centimeter is then H ′TOTAL =
(0.01 + CMIX)m h = 0.02m
h = 10.5 kJ/s cm = 10.5 kW/cm. Over a core that is 400 cm high, this means that 4,200 kJ of
thermal energy is transferred from a hot coolant channel to a colder one each second. This lateral energy transfer tends
to flatten the transverse temperature profiles, and it provides a much more accurate picture of the coolant temperatures
than an isolated channel model does. [Ans.]
26.21
Applying the Homogenous Equilibrium Model
Equations to Reactor Subchannel Design
Because reactor cores are so complicated both geometrically and spatially, famous reactor thermal-hydraulic analysis
programs like RELAP, TRACE, COBRA, and VIPER do not attempt to solve the fluid conservation equations in the
same way that Computational Fluid Dynamics Programs like FLUENT do. Instead, they attempt to model each fuel
assembly as a collection of subchannels that are interconnected in the radial and axial directions. Within each subchan-
nel or computational cell, the properties of the fluid are assumed to be spatially uniform. For example, if a bulk fluid
property B is required for the conservation equations within one of these cells, the value of B is defined by
B = (1/V) ⋅
∫ V
BdV (26.91)
where V is the volume of the cell. This then allows a reactor designer to lump together groups of subchannels in the
manner shown in Figure 26.25 to model either the whole core or just portions of it. The conservation equations required
to find the axial mass flow rates, the lateral cross flow rates, and the coolant pressure are then written as a set of Eulerian
equations similar to the conservation equations we derived from first principles in Chapter 15. In particular, the continu-
ity equation for the ith subchannel is written as
∂(ρA)i ∂ t + ∂(ρuA)i ∂z + ∑w
k
i, j = 0 (26.92)
where i is the index of subchannel i, Ai is the flow area of subchannel i, ρ and u are the fluid density and the fluid velocity,
j is the index of subchannel j which borders subchannel i, and k is the index of the lateral interface between subchannels
i and j though which the coolant flows. The cross flow wi,j = ρvSk is then defined as the lateral mass flow rate per unit
length across gap k between subchannels i and j, v is the velocity of the coolant in the lateral direction that supports this
cross flow, and Sk is the width of gap k. The axial momentum equation for subchannel i is
FIGURE 26.25 A picture illustrating how multiple subchannels are then lumped together to model reactor fuel assemblies and
then the entire reactor core.
1032 Core Temperature Fields
where Pi is the pressure in subchannel i, g is the acceleration of gravity, f i is the wall friction factor for subchannel i,
Di is the hydraulic diameter for subchannel i, and u* is the axial velocity of the donor channel across interface k between
subchannels i and j. In other words, if the lateral flow is from subchannel i into subchannel j, then u* = ui, and conversely,
if the flow is from subchannel j into subchannel i, then u* = uj. Finally, wTURBULENT i,j is the turbulent mixing mass flow
rate per unit length across gap k in the lateral direction. This turbulent cross flow is related to the turbulent eddy dif-
fusivity ε (see Chapter 17) by
w TURBULENT i, j = ε i ρi Sk l k (26.94)
where lk is the distance between the centroids of subchannels i and j along interface k through which this turbulent
momentum can flow (see Figure 26.24). In many thermal-hydraulic analysis codes, wTURBULENT i,j is determined using
where GAVG = ½((ρu)i + (ρuj)) is the average axial mass flux through subchannels i and j, and CMIX is the turbulent mixing
coefficient which we discussed for reactor fuel assemblies earlier in the chapter. The energy equation for subchannel i
is then
∂(ρhA)i ∂ t + ∂( (ρhA)i u i ) ∂z + ∑w k
i, j h* + ∑w k
TURBULENT i, j (hi − h j )
+ ∑ n
PHEATED i,n H n ( Ti − Tn ) = 0 (26.96)
where h i is the specific enthalpy for subchannel i, h* is the specific enthalpy which moves across interface k from the
donor channel (which can be i or j), PHEATED i,n is the heated perimeter for rod n in subchannel i, n is the number of
rods which add heat to subchannel i, Ti is the bulk temperature of the coolant in subchannel i, H n is the convective
heat transfer coefficient at the surface of rod n, and Tn is the temperature of the surface of the nth rod facing subchan-
nel i. The temperature at the surface of each fuel rod facing subchannel i is obtained by solving the transient heat
conduction equation
where T is the temperature of the rod, q‴ is the volumetric heat generation rate within the rod, ρ′ is the density of the
materials in the rod, cp is the specific heat capacity of the materials, and k is their thermal conductivity. Normally, the
value of q‴ is assumed to be an independent variable which is determined from a reactor physics calculation. Finally,
the lateral momentum equation used to find the cross flow between subchannels i and j is
where ⟨u⟩= ½(ui + uj) is the average axial velocity in the two subchannels, ρ = ½(ρi + uj) is the average density of the
coolant in the two subchannels, KGAP is the lateral loss coefficient that accounts for the friction and form pressure loss
caused by the area change in the lateral direction, and lk is the distance between the centroids of subchannels i and j
in the lateral direction. These four conservation equations are then the equations reactor designers solve to find the
flow rates and temperature fields in reactor fuel assemblies. Turbulent mixing terms are added to the momentum and
energy equations to account for turbulent energy and momentum transfer. The exact form of these equations depends on
Bibliography 1033
how the turbulent mixing coefficients are defined. A procedure for evaluating these terms was presented earlier in the
chapter. Normally, turbulent mixing reduces the temperature gradients between the coolant channels and it also makes
the radial enthalpy profiles more uniform. This flattening of the temperature profiles reduces the conservatism inher-
ent in an isolated channel approach and it also makes it possible to predict the fuel rod and coolant temperatures more
accurately. The exact forms of the energy and momentum equations that are used when the effects of turbulent mixing
become important are provided at the following URL:
https://lwrs.inl.gov/RiskInformed%20Safety%20Margin%20Characterization/INL-EXT-14-33102_9-25-14.pdf
which is maintained by the Idaho National Engineering Laboratory (or INEL). A similar approach is also employed by
the RELAP-5 computer program.
26.22
Constructing Three-Dimensional Core Temperature Profiles
Finally, we would like to illustrate how we can use the equations we have just derived to find the fuel rod temperature
profiles in three-dimensional cores. Since we already know what the radial and axial temperature profiles look like, this
is a relatively easy thing to do. All we need to do is to assemble the pieces together, and to then present the results. The
fuel rod temperature profiles are normally subdivided into radial and axial components:
Thus, we must simply assemble these components together to find T FUEL(r, z). In most cases, the radial heat flux (and
therefore the radial temperature profile) has very little dependence on the axial heat flux (and therefore the axial tem-
perature profile). When this is true, the radial and axial temperature profiles can be considered as independent variables
and the three-dimensional temperature profiles can be found by simply multiplying them together; that is,
In a differential equations class, this approach is known as the separation of variables method. Likewise, a similar
approach can also be used for the cladding:
Thus, it is relatively straightforward to multiply the radial and axial components together to find the three-dimensional
temperatures as long as the radial and axial profiles are loosely coupled. Normally, the temperature gradients in a fuel
rod are much greater in the radial direction r than they are in the axial d irection z. If we were to compare these gradi-
ents, we would find that the average ratio of these gradients dT/dr and dT/dz is about 1,000 to 1 in a thermal water reactor
and between 50 and 100 to 1 in a reactor cooled with liquid sodium. For these reasons and others, a reactor designer is
usually justified in neglecting heat conduction in the axial direction when the flow rate is high because dT/dr ≫ dT/dz.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Crowe, C. and Schwartzkopf, J., et al. Multiphase Flows with Droplets and Particles, CRC Press, Boca Raton, FL (2012).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
1034 Core Temperature Fields
Web References
https://lwrs.inl.gov/RiskInformed%20Safety%20Margin%20Characterization/INL-EXT-14-33102_9-25-14.pdf.
http://www-pub.iaea.org/MTCD/publications/PDF/IAEA-THPH_web.pdf.
Kao, H.S., Morgan, T.D. and Parker, W.B., “Prediction of Flow Oscillation in Reactor Core Channel”, Transactions of the American
Nuclear Society, 16:212–213, 1973.
Kitchen, T.J., “Generic Safety Evaluation for 17×17 Standard Robust Fuel Assembly (17x17 STD RFA),” SECL-98-056, Revision 0,
September 30, 1998.
Kjaerheim, G. “In-Pile Measurements of Center Fuel Temperatures and Thermal Conductivity Determination of Oxide Fuels,” Paper
IFA-175 Presented at the European Atomic Energy Society Symposium on Performance Experience of Water-Cooled Power
Reactor Fuel, Stockholm, Sweden, October 1969.
Kjaerheim, G. and Rolstad, E. “In-Pile Determination of UO2, Thermal Conductivity, Density Effects, and Gap Conductance,” HPR-
80, December 1967.
Kruger, O.L., Heat Transfer Properties of Uranium and Plutonium Dioxide, Paper 11-N-68F, presented at the Fall Meeting of Nuclear
Division of the American Ceramic Society, Pittsburgh, September 1968.
Lahey, R.T. and Moody, F.J. “The Thermal Hydraulics of a Boiling Water Reactor”, American Nuclear Society, Hinsdale, IL, 1977.
Leech, W.J., et al. “Revised PAD Code Thermal Safety Model,” WCAP-8720, Addendum 2, October 1982.
Letter from A.C. Thadani (NRC) to W.J. Johnson (Westinghouse), January 31, 1989, Subject: Acceptance for Referencing of Licensing
Topical Report, WCAP-9226-P/9227-NP, “Reactor Core Response to Excessive Secondary Steam Releases.”
Letter from C. Berlinger (NRC) to E.P. Rahe, Jr. (Westinghouse), Subject: “Request for Reduction in Fuel Assembly Burnup Limit for
Calculations of Maximum Rod Bow Penalty,” June 18, 1986.
Lucks, C.F. and Deem, H.W. “Thermal Conductivity and Electrical Conductivity of UO2,” in Progress Reports Relating to Civilian
Applications, BMI-1448 (Revised) for June 1960,BMI-1489 (Revised) for December 1960, and BMI-1518 (Revised) for May 1961.
Lyons, M.F., et al. “UO2 Powder and Pellet Thermal Conductivity During Irradiation,”GEAP-5100-1, March 1966.
McFarlane, A.F., “Power Peaking Factors,” WCAP-7912-P-A (Proprietary) and WCAP-7912-A (Non-Proprietary), January 1975.
Motley, F.E., Cadek, F.F. “DNB Test Results for R-Grid Thimble Cold Wall Cells,” WCAP-7695-L Addendum 1, October 1972.
Motley, F.E., Wenzel, A.H. and Cadek, F.F. “The Effect of 17 × 17 Fuel Assembly Geometry on Interchannel Thermal Mixing,”
WCAP-8298-P-A (Proprietary) and WCAP-8290A (Non-Proprietary), January 1975.
Nelson, R.G., et al. “Fission Gas Release from UO2 Fuel Rods with Gross Central Melting,” GEAP-4572, July 1964.
Nishijima, T., Kawada, T. and Ishihata, A. “Thermal Conductivity of Sintered UO2 and 4Al2O3 at High Temperatures”, Journal of the
American Ceramic Society, 48:31–44, 1965.
Nodvick, R.J., “Saxton Core II Fuel Performance Evaluation,” WCAP-3385-56, Part II, “Evaluation of Mass Spectrometric and
Radiochemical Analyses of Irradiated Saxton Plutonium Fuel,” July 1970.
Ohtsubo, A. and Uruwashi, S. “Stagnant Fluid due to Local Flow Blockage”, Journal of Nuclear Science Technology, 9(7):433–434,
1972.
“Partial Response to Request Number 1 for Additional Information on WCAP-8691,Revision 1,” Letter from E.P. Rahe, Jr.
(Westinghouse) to J.R. Miller (NRC), NS-EPR-2515, October 9, 1981; “Remaining Response to Request Number 1 for Additional
Information on WCAP-8691, Revision 1,” Letter from E.P. Rahe, Jr. (Westinghouse) to R.J. Miller (NRC), NS-EPR-2572,
March 16, 1982.
Poncelet, C.G., “Burnup Physics of Heterogeneous Reactor Lattices,” WCAP-6069, June 1965.
Rowe, D.S. and Angle, C.W. “Crossflow Mixing Between Parallel Flow Channels During Boiling, Part II Measurements of Flow and
Enthalpy in Two Parallel Channels,” BNWL-371, Part 2, December 1967.
Rowe, D.S. and Angle, C.W. “Crossflow Mixing Between Parallel Flow Channels During Boiling, Part III Effect of Spacers on Mixing
Between Two Channels,” BNWL-371, Part 3, January 1969.
Saha, P., Ishii, M. and Zuber, N. “An Experimental Investigation of the Thermally Induced Flow Oscillations in Two-Phase Systems”,
Journal of Heat Transfer, 98:616–622, November 1976.
Skaritka, J. (Ed.) “Fuel Rod Bow Evaluation,” WCAP-8691, Revision 1 (Proprietary) and WCAP-8692, Revision 1 (Non-Proprietary),
July 1979.
Slagle, W.H. (Ed.) et al. “Westinghouse Improved Performance Analysis and Design Model (PAD 4.0),” WCAP-15063-P-A, Revision
1 (Proprietary) and WCAP-15064-NP-A, Revision 1 (Non-Proprietary), July 2000.
Smith, L.D., et al. “Modified WRB-2 Correlation, WRB-2M, for Predicting Critical Heat Flux in 17×17 Rod Bundles with Modified
LPD Mixing Vane Grids,” WCAP-15025-P-A (Proprietary) and WCAP-15026-NP (Non-Proprietary), April 1999.
South Texas Project Electric Generating Station FSAR, Chapter 4, Houston Lighting and Power Company, Docket No. 50-498.
Stewart, C.W., et al. “VIPRE-01: A Thermal-Hydraulic Code for Reactor Core,” Volume 1–3 (Revision 3, August 1989), Volume 4
(April 1987), NP-2511-CCM-A, Electric Power Research Institute.
Stora, J.P. “In-Reactor Measurements of the Integrated Thermal Conductivity of UO2-Effect of Porosity,” Transactions of the
American Nuclear Society, 13:137–138, 1970.
Stora, J.P., et al. “Thermal Conductivity of Sintered Uranium Oxide under In-Pile Categorys,” EURAEC-1095, 1964.
Stora, J.P., et al. “Thermal Conductivity of Sintered Uranium Oxide Under In-Pile Categorys,” EURAEC-1095, August 1964.
Sung, Y.X., et al. “VIPRE-01 Modeling and Qualification for Pressurized Water Reactor Non-LOCA Thermal-Hydraulic Safety
Analysis,” WCAP-14565-P-A andWCAP-15306-NP-A, October 1999.
Thom, J.R.S., et al. “Boiling in Subcooled Water During Flowup Heated Tubes or Annuli”, Proceedings of the Institution of Mechanical
Engineers Part C, 180:226–246, 1955–1966.
Tong, L.S. “Prediction of Departure from Nucleate Boiling for an Axially Non-uniform Heat Flux Distribution,” Journal of Nuclear
Energy, 21:241–248, 1967.
Tong, L.S. “Critical Heat Fluxes in Rod Bundles, Two Phase Flow and Heat Transfer in Rod Bundles,” Annual Winter Meeting ASME,
November 1968, p. 3146.
Tong, L.S. “Boiling Crisis and Critical Heat Flux,” AEC Critical Review Series, TID-25887, 1972.
Virgil C. Summer Nuclear Station FSAR, Chapter 4, South Carolina Electric & Gas Company, Docket No. 50-395.
1036 Core Temperature Fields
Vogt, J., Grandel, L. and Runfors, U. “Determination of the Thermal Conductivity of Unirradiated Uranium Dioxide,” AB Atomenergi
Report RMB-527, 1964, Quoted by IAEA Technical Report Series No. 59, “Thermal Conductivity of Uranium Dioxide.”
Weiner, R.A., et al. “Improved Fuel Performance Models for Westinghouse Fuel Rod Design and Safety Evaluations,” WCAP-10851-
P-A (Proprietary) and WCAP-11873-A (Non-Proprietary), August 1988.
Weisman, J., “Heat Transfer to Water Flowing Parallel to Tube Bundles”, Nuclear Science Engineering, 6:78–79, 1959.
22. Approximately how many times greater is the radial temperature gradient than the axial temperature gradient in a
commercial PWR fuel rod?
23. Derive approximate expressions for the maximum fuel temperature and the maximum cladding temperature in an
American PWR assuming that the axial nuclear heat flux is co-sinusoidal in shape.
24. In PWR fuel rods, what is the temperature drop between the inner surface of the cladding and the outer surface of
the cladding?
25. For the same fuel rods, what is the average temperature drop across the fuel to cladding gap? Is this temperature
drop larger or smaller than the temperature drop across the cladding?
26. What is a realistic value for the temperature drop from the center of a nuclear fuel pin to its outer surface? Is the fuel
temperature drop larger or smaller than the temperature drop across the rest of the rod?
27. Derive separate analytical expressions for the maximum fuel, cladding, and coolant temperatures in a PWR assum-
ing that the nuclear heat flux is co-sinusoidal in shape. Do the maximum temperatures for the fuel, the cladding, and
the coolant occur at the same locations or different ones?
28. What type of cladding is used in commercial PWRs today? What type of cladding is used in commercial BWRs
today? Are the melting points of these two types of cladding different? What is the primary difference in their
molecular composition, and how thick does the cladding around a PWR and BWR fuel rod normally have to be?
29. What is the melting point of uranium dioxide in a nuclear fuel rod? Does the melting point increase, decrease, or
stay the same as the fuel is burned?
30. In conductive heat transfer, what is the thermal resistance and how is it defined?
31. For the PWR fuel rods used in this chapter, calculate the thermal resistance for the fuel, the cladding, and the fuel to
cladding gap. What is a representative value for the convective resistance at the outer surface of the cladding when
the coolant is flowing normally?
32. During normal operation, how large is the temperature drop between the cladding and the coolant for a representa-
tive PWR fuel rod? How large is the temperature drop for a similar BWR fuel rod?
33. How can the boiling and non-boiling lengths be determined in BWR cores? On average, how much larger or smaller
is the boiling length than the non-boiling one?
34. Derive an expression for the enthalpy rise is a fuel assembly as a function of axial elevation. At what value of the local
enthalpy does the equilibrium quality become positive? What is the name of the axial location where this occurs?
35. In principle, is it possible for localized boiling to occur in a PWR or BWR core even when the equilibrium quality
is negative? If so, what is the reason for this apparent behavior?
36. Approximately how far above the saturation temperature does the outer surface of the cladding get in a commercial
BWR? Ay which axial elevation does this initially occur?
Exercise 26.2
A pressurized water reactor operates at a pressure of 2,250 PSI or 15.5 MPa. The mass flux at the entrance to the core is
4,000 kg/m2 s and the incoming water is 50°C subcooled. The heat flux at the surface of a typical fuel rod is 1 × 106 W/m2
and the energy generating portion of each rod is 4 m long. If the fuel rods are 1 cm in diameter and the average area of
the flow channels surrounding them is 0.000088 m2, what is the enthalpy increase along the length of a rod? What is the
exit temperature? Does any subcooled boiling occur?
Exercise 26.3
A PWR fuel rod with an external diameter of 1 cm dumps heat into a reactor coolant channel having a flow rate of
0.35 kg/s. If the heated portion of the rod is 4 m long and the axial heat flux is uniform, what is the temperature dif-
ference between the top and the bottom of the rod if the applied heat flux is 1,200 kW/m 2?
1038 Core Temperature Fields
Exercise 26.4
A cylindrical fuel rod in a PWR has a linear heat generation rate q′ of 50,000 W/m (500 W/cm). The diameter of the rod
is 1 cm. What is the temperature drop between the outer surface of the cladding and the coolant? Assume a convective
heat transfer coefficient of h = 50,000 W/m2 °C.
Exercise 26.5
Suppose that the lateral mass flow rate (i.e., cross flow) between two adjacent subchannels in a PWR fuel assembly is
1% of the axial mass flow rate and that the turbulent mixing coefficient is 0.005 cm s/kg. Estimate the total enthalpy
transfer rate between these channels when the axial mass flow rate is 0.50 kg/s and the specific enthalpy of the coolant
in the donor channel is 1,800 kJ/kg. Assume that h′ = 1,800 kJ/kg.
Exercise 26.6
A cylindrical fuel rod in a light water reactor has a linear heat generation rate of 500 W/cm. If the fuel is UO2, what is
total temperature drop ΔTFUEL across the fuel in the radial direction? How many times larger is this than the temperature
drop across the cladding?
Exercise 26.7
In nuclear fuel rods, the fuel to cladding gap can be very small but the temperature drop across it can be very large. If the
average temperature drop across the gap is 200°C, calculate the thermal resistance RGAP. How many times larger is this
than the convective thermal resistance? At approximately what burn up does the temperature drop become this large?
Does the temperature drop become larger or smaller as the burn up increases?
Exercise 26.8
A large commercial PWR is fueled with fresh UO2 having an average density of 10.4 g/cm3. Suppose the standard
deviation in its measured density is 0.25 g/cm3. What is the engineering uncertainty factor Fρ governing the amount of
uranium in the core? The core of a large 1,000 megawatts electric PWR contains about 100 metric tons of UO2. What is
the uncertainty in the total amount of uranium in the core in kg?
Exercise 26.9
The first nuclear-powered cargo ship, the USS Savannah, was powered by a pressurized water reactor where the core
operated at 1,750 PSI. Compared to the nuclear-powered aircraft carriers of today, the power density was relatively low.
The coolant entered the bottom of the core at 258°C and left the core at 270°C. The flow rate through the reactor p ressure
vessel was 1,200 kg/s. What was the average enthalpy change of the coolant between the bottom and the top of the core?
What was the total thermal power output of the core?
Exercise 26.10
Pressurized water flows through a reactor core at a rate of 20,000 kg/s. When passing through the core, its enthalpy
increases from 1,300 to 1,500 kJ/kg. What is the total thermal output of the core if the operating pressure is 2,250 PSI
(15.5 MPa)? What is the equilibrium quality at the core outlet? If the plant containing the core was 34% efficient, how
much electric power would it generate? If it was a Westinghouse AP 1000, how many fuel assemblies would it have?
27
Nuclear Hot Channel Factors, the
Critical Heat Flux, and the DNBR
27.1
T he Boiling Crisis in Water Cooled Reactors
Earlier we discussed specific combinations of the pressure, the temperature, the flow rate, and the heat flux that could result in the
coolant vaporizing from the surface of a fuel rod. In general, fuel rods fail when small bubbles formed during the process of nucleate
boiling coalesce to form a vapor film that covers the entire surface of the rod. Once this occurs, the convective heat transfer coefficient
falls dramatically over a very short distance and the temperature of the cladding suddenly rises by several hundred degrees C. The
cladding then melts and the fuel rods fail. In pressurized water reactors (PWRs), this condition is called Departure from Nucleate
Boiling or DNB, and in boiling water reactors (BWRs), it is called film dryout. Although the exact processes which initiate dryout in
PWRs and BWRs are different, they have many similarities that we would now like to expound upon.
27.2
D ryout of the Fuel Rods in PWRs
Dryout in PWRs occurs when localized nucleate boiling becomes so intense that a vapor film blankets the entire surface of the rod
(see Figure 27.1). This film causes the convective heat transfer coefficient to fall dramatically (typically by a factor of 10–100 over
a very short distance) because the coolant can no longer reach the cladding surface. The fuel temperature rises, the cladding melts,
and fission gases are released into the coolant. At typical temperatures and pressures, this occurs at equilibrium qualities as low
as 12% in PWRs (xeq = 0.12) because the bubbles very quickly isolate the surface of the cladding from the liquid surrounding it.
27.3
D ryout of the Fuel Rods in BWRs
A similar phenomenon occurs in BWRs, but the dryout process is different because the flow rates and the pressures are lower (see
Table 27.1), and this results in the liquid film remaining on the surface for a longer period of time. Once bulk boiling begins, the
coolant starts to develop a vapor core, but most of the remaining liquid still clings to the surface of the cladding. Thus, the surface
of the cladding remains wet, while a vapor mist occupies the center of the channels, which are called the vapor cores. This behavior
is illustrated on the left hand side of Figure 27.1 and the flow immediately before dryout is called annular film flow. Dryout under
these conditions occurs at equilibrium qualities between 30% and 40%, and thus dryout occurs much more gradually than it does
in PWRs. (see Table 27.2). In other words, dryout occurs in BWRs after the flow becomes annular, while in PWRs, it occurs before
the flow becomes annular. This behavior can be attributed to differences in the system pressure, the heat flux, and the mass flux.
Normally, reactors are designed so that dryout can never occur during start-up, shutdown, or full-power operation.
27.4
Coolant and Cladding Temperatures in Typical PWR Cores
When a reactor is operating normally, a thick liquid film removes heat from the surface of the cladding. In PWRs, the turbulent
eddies in this film agitate the convective layer, and this results in a very high convective heat transfer coefficient; however, the coolant
below the core midplane usually stays below the saturation temperature. Then as one moves beyond the core midplane, the cladding
temperature rises 3°C–6°C above the saturation temperature. The bulk temperature stays below the saturation temperature, but near
the top of the core small bubbles start to form in the hotter channels. However, the coolant (in a thermodynamic sense) continues to
remain subcooled. Hence, a small amount of nucleate boiling may occur, but it does not reduce the convective heat transfer coefficient
1039
1040 NUCLEAR HOT CHANNEL FACTORS, CHF, AND DNBR
FIGURE 27.1 The boiling crisis in a PWR compared to the boiling crisis in a BWR.
TABLE 27.1
Flow Rate and Power Density Comparisons between a BWR-5 and PWR for Approximately the Same Core
Thermal Output
Representative Core Flow Rates and Power Densities for Select PWRs and BWRs
Parameter BWR-5 PWR
Core thermal power output (MW) 3,325 3,410
Total core flow rate (kg/s) 13,670 17,475
Core average power density (KW/L) 52.3 104.5
Average core pressure (PSI/MPa) 1,050/7.25 2,250/15.5
TABLE 27.2
Representative Equilibrium Qualities Where Dryout Occurs in PWRs and BWRs
FIGURE 27.2 The temperature profiles in a typical PWR subchannel before nucleate boiling occurs. In this case, a uniform axial
heat flux is assumed.
at the cladding surface—in fact, it temporarily increases it.* For a uniform axial heat flux, the temperature profiles for
the fuel and the cladding are shown in Figure 27.2. When the heat flux becomes co-sinusoidal, the temperature profiles
for the fuel and the cladding are shown in Figure 27.3(a).
{ }
α PWR DRYOUT = 1 1 + (ρv ρl ) ⋅ (1 − x)/x = 1 {1 + (1/6.9) ⋅ (0.90/0.10)} = 1/2.30 = 43.5% and
Hence, the minimum void fraction at which dryout occurs in BWRs is essentially twice what it is in PWRs. [Ans.]
27.5
T he Critical Heat Flux and the Boiling Crisis
Now let us take a look at representative values for the critical heat flux (CHF) in modern PWRs. After completing this
discussion, we will then turn our attention to BWRs. The critical heat flux in BWRs is different than it is in PWRs
because the mass flux G, the system pressure p, and the quality x are different. In general, the critical heat flux is a
function of the equivalent diameter De as well. Hence, we can write the functional dependence of the CHF as
Now let us examine how each of these parameters affects the value of the CHF.
FIGURE 27.3 (a) The temperature profiles in a typical PWR subchannel before nucleate boiling occurs. In this case, a
co-sinusoidal heat flux is assumed. Notice that the cladding temperature peaks about three-fourths of the way up the core.
(b) The effect of the pressure on the critical hear flux.
27.5.1
Representative Values for the CHF in Modern PWRs
If one keeps the mass flow the same, and increases the heat flux even further, eventually a point is reached where an
unstable transition occurs from nucleate boiling to a vapor film. The vapor film completely blankets the surface of the
cladding, and no more liquid can reach the cladding to remove the heat. Then the cladding temperature rises dramati-
cally over a very short distance (see Figure 27.4), and the value of h falls by a factor of 10 to 100. The value of the heat
flux at which this transition occurs is called the critical heat flux, and when the critical heat flux is exceeded, the fuel
temperature rises, the cladding melts, and fission gases are released into the coolant. In most PWRs, the critical heat
flux occurs when the heat flow rate reaches between 2 × 105 W/m2 and 4 × 105 W/m2. (The exact value depends on the
flow rate and the equilibrium quality at the core exit.) However, to a first approximation, the relationship between the
CHF and the exit quality is a linear one.
27.5 The Critical Heat Flux and the Boiling Crisis 1043
FIGURE 27.4 The temperature profiles in the hot subchannel of a PWR when the liquid film surrounding the cladding dries out.
This is caused by a DNG that is sometimes referred to as the boiling crisis.
27.5.2
D ependence of the CHF on the Exit Flow Quality
If the exit quality is negative from a thermodynamic perspective and x = −0.20, the coolant leaving the core is highly
subcooled, and the CHF is about 4 × 105 W/m2. If the exit quality rises to 10%, the CHF then becomes 2 × 105 W/m2. Finally,
for an exit quality near zero, which is representative of most American-made PWRs, the CHF is about 3 × 105 W/m2.
These values are based on an average mass flux of G = 3,000 kg/m2 s. The exact dependence of the CHF on the exit
quality is shown in Figure 27.5. The equivalent diameter of the flow channel De also affects the value of the CHF.
27.5.3
D ependence of the CHF on the Mass Flow Rate
In general, higher mass flow rates (and hence higher mass fluxes) result in higher values for the CHF and lower mass
fluxes result in lower ones. To a first approximation, this dependence is also linear. Hence, if one were to reduce the mass
flux through a fuel assembly by 50% (say from 3,000 kg/m2 s to 1,500 kg/m2 s) while keeping the power level the same,
the CHF would fall from 3,000 kW/m2 (300 W/cm2) to approximately 1,500 kW/m2 (or 150 W/cm2). We will have more
to say about this dependency later.
27.5.4
D ependence of the CHF on Grid Spacers
The value of the CHF also depends on whether grid spacers or wire wraps are used. In general, the CHF increases
for a short distance downstream of a grid spacer or wire wrap. This effect is illustrated in Figure 27.6, and it was first
observed by Groeneveld and Cheng in 1998. More grids lead to higher values of the CHF and no grids lead to lower
ones. Phenomenologically, this effect is due to the increased turbulence of the flow field immediately downstream of
the grid spacers or wire wraps, which suppresses the onset of the CHF due to the effects of improved turbulent mixing.
Various attempts have been made over the years to take advantage of this effect through smarter grid spacer design. The
“Robust Fuel Assemblies” in the Westinghouse AP 1000 are an example of such a design. These fuel assemblies and the
DNB (Departure from Nucleate Boiling) correlations they use are described in Appendices I and J. Normally, these grid
spacers are used to reduce the conservatism in the existing CHF correlations the U.S. NRC uses to grant an operating
license for a plant. To reduce this conservatism, Westinghouse developed a new CHF correlation for the AP-1000 called
the WRB correlation (where the term WRB is an acronym for a Westinghouse Rod Bundle). This correlation is intended
to eventually replace the Westinghouse W-3 correlation (see Chapter 25) as the primary correlation Westinghouse uses
to calculate the critical heat flux in its reactor fuel assemblies. It appears to be far better at predicting the shape of the
critical heat flux in the upper third of the core where the third and fourth mixing vanes in the Robust Fuel Assemblies
1044 NUCLEAR HOT CHANNEL FACTORS, CHF, AND DNBR
FIGURE 27.5 The dependence of the CHF on the exit flow quality in circular flow channels with different diameters. The system
pressure in this case is 2.3 MPa (333 PSI).
FIGURE 27.6 Behavior of the CHF a short distance downstream from a grid spacer or a wire wrap. In general, the CHF increases
beyond the spacer because of the additional turbulence induced in the flow field.
are located. Hence, it more accurately accounts for the effects of turbulent mixing on the shape of the CHF in the upper
third of the core where the DNBR (see Figure 27.7) is usually the lowest.
27.5.5
D ependence of the CHF on the System Pressure
Finally, as one might expect, the critical heat flux is affected by the system pressure. As low steam qualities, increasing
the system pressure lowers the CHF. However, at high steam qualities, increasing the system pressure increases the
CHF. This behavior is due to the interplay between the heat capacity, the rate of vaporization, and surface tension at the
27.6 Variation of the CHF and the DNBR with Position 1045
FIGURE 27.7 An illustration of the actual heat flux, the computed CHF, and the DNB or CHF ratio for a typical PWR. Notice that
the value of the DNBR reaches its lowest value about three-fourths of the way up the core, and this is where the potential for rod
dryout is greatest.
fuel rod surface. Exactly how the CHF behaves depends on which of these factors dominates the other two. Normally,
this effect is rather small at reactor operating pressures—tending to vary by no more than ±10% as the system pressure
is changed. The exact dependence is shown in Figure 27.3b. Naturally, this effect must be quantified before a nuclear
power plant can be granted an operating license.
27.6
Variation of the CHF and the DNBR with Position
The critical heat flux in reactor work is a calculated quantity. It is normally found as a function of position in the h ottest
channel using the correlations presented in Section 27.8. It depends on the local mass flux, the pressure, the exit quality,
and the axial power profile. For a co-sinusoidal shaped power profile in a typical PWR, the calculated CHF is shown in
Figure 27.7. In general, the CHF falls as one moves up the core. However, the actual heat flux (see Figure 27.6) peaks near
the core midplane, and its value can be thought of as an independent variable determined by the neutronic characteristics
of the core. When the heat flow is primarily radial, the applied heat flux or actual heat flux at a fuel rod’s surface is
where q‴(z) is the volumetric heat production rate in the fuel, AFUEL is the cross-sectional area of the fuel, and CCLAD is
the circumference of the cladding. This equation applies to both cylindrical and plate-type fuel rods. In addition to the
thermal limits that are placed on the fuel, the cladding, and the coolant (see Chapter 11), an additional design limit is
placed on the heat flux at the cladding surface. Above certain values of this heat flux, the cladding dries out, and if the
heat generation rate in the fuel does not change, the value of h will fall low enough for the cladding to fail. The heat flux
′′
where this occurs is called the critical heat flux, and in practice, it is represented by q CRITICAL ′′ (z).
(z) or q CR
Thus, the ratio R of the critical heat flux (CHF) to the actual heat flux (AHF) (or q ′′CRITICAL (z) /q ′′ACTUAL (z) = q ′′CR (z)/
q ′′(z) ) determines the location where dryout occurs. Thus, reactors must be designed so that the actual heat flux q′′ACTUAL
′′
(z) is always less than the critical (or burnout) heat flux qCRITICAL (z) everywhere in the core
′′
Normally, the ratio of q CRITICAL (z) to q ′′ACTUAL (z) is called the critical heat flux ratio (the CHFR) or the departure from
nucleate boiling ratio (the DNBR).
1046 NUCLEAR HOT CHANNEL FACTORS, CHF, AND DNBR
FIGURE 27.8 In practice, statistical uncertainties in the enrichment of the fuel and the thickness of the cladding can produce a
small envelop of uncertainty around the computed values of the CHF and the DNBR even through calculating the CHF itself is a
deterministic exercise. In this case, engineering judgment must be applied to determine the exact values of the CHF and the DNBR
to be used in the FSAR.
Notice that this ratio depends on the axial position (z) in the core. For a typical PWR fuel rod, its value is shown in
Figure 27.7. Normally it reaches its minimum value about three-fourths of the way up the core where the potential for
dryout is highest. Thus, reactor designers must always ensure that the actual heat flux (AHF) never exceeds the critical
heat flux (CHF) at this location. As we will see later, the current design limit for American BWRs is a minimum DNBR
(MDNBR) of 1.9, and the current design limit for American PWRs is a MDNBR of 1.3. By setting these limits on the
DNBR, there is a reasonable probability that the fuel rods will never fail. The value of the DNBR is also affected by
statistical uncertainties in the values of key core parameters such as the enrichment of the fuel, the thickness of the
cladding, and the local mass flux G. These statistical uncertainties create a small envelope of uncertainty around the
computed value of the CHF. This envelope of uncertainty is shown Figure 27.8. Consequently, engineering uncertainty
factors (see Section 27.26) are applied to understand the size of these uncertainties. The U.S. Nuclear Regulatory
Commission (or NRC) then uses these statistical uncertainties to determine the most probable value of the MDNBR in
the hot channel. The resulting value(s) are then used in the final safety analysis report or FSAR.
27.7
Some Boiling Crisis Terminology
The breakdown of the liquid film on the cladding surface has many different names. Over the years, it has been referred to as
1. DNB
2. the burnout condition
3. catastrophic dryout
4. the boiling crisis.
Not surprisingly, each of these four terms refer to essentially the same thing. Empirical correlations have then been
developed to determine the location of the DNB point. Almost all of these correlations attempt to determine the magni-
tude of the CHF at which DNB occurs and to then use this value to infer the DNB point at the location where the actual
heat flux and the critical heat flux overlap. However, the point at which the critical heat flux is reached depends on the
shape of the coolant channel, the surface conditions, the physical characteristics of the coolant, and the local mass flux
27.9 The Tong Correlation 1047
G = m/A. The CHF point also depends on the shape of the applied heat flux, which is discussed in Section 27.13. Thus,
its location depends on many different factors. We would now like to discuss some of the most popular correlations that
are used to determine the dryout location and the CHF point.
27.8
CHF Correlations
Correlations for the critical heat flux in reactor fuel assemblies have existed since the 1950s. The first correlations to
calculate the CHF for PWRs were developed in the 1950s, and there have been many improvements to these correlations
since that time. In PWRs, the bulk temperature of the coolant (TBULK) is intended to stay below the saturation temperature
(TSAT), which is about 344.5°C at 2,250 PSI (15.5 MPa). In modern PWRs, the equilibrium quality at the core exit, where
the coolant temperature is closest to the saturation temperature, is about −0.05. However, vapor bubbles can still form
on the surface of the fuel rods, and during overpower transients, DNB can occur in the hottest channels when the flow
quality becomes high enough for a vapor film to blanket the surface of the cladding. This typically occurs when the local
quality (as opposed to the equilibrium or thermodynamic quality) is between 10% and 15%. The first correlations for the
critical heat flux in PWRs were developed for circular flow channels, but they were later extended to other subchannel
geometries including those in rod bundles. The first popular correlation for the critical heat flux in a circular flow channel
was developed by L.S. Tong in the mid-1960s. Not surprisingly, this correlation became known as the Tong correlation.
For rod bundles geometries, the most famous of these correlations is the W-3 correlation, which was also developed at
Westinghouse by L.S. Tong after he completed his original work with circular pipes. In all these correlations, the critical
′′
heat flux q CRITICAL is expressed as a function of the mass flux G = m/A, the equilibrium quality x (which can be either
the local quality or the exit quality), the geometry of the flow channel, and the system pressure p. Hence, for PWRs,
critical heat flux correlations are expressed by Equation 27.1 where De is the equivalent diameter of the flow channel.
Under most conditions, the critical heat flux increases almost linearly with the mass flux G = m/A, but it also decreases
as the exit quality increases (when the mass flow rate is held constant). In the next section, we would like to present five
correlations that can be used to determine the value of the CHF:
☉☉ The Tong correlation
☉☉ The Bernath correlation
☉☉ The Bowring correlation
☉☉ The Groeneveld correlation
☉☉ The Westinghouse W-3 correlation
The first three of these correlations (the Bernath correlation, the Tong correlation, and the Bowring correlation) are appli-
cable to circular flow channels, while the W-3 and Groeneveld correlations are applicable to other flow channel geometries,
such as those encountered in reactor fuel assemblies. The primary virtue of the Bernath correlation is that it can be used to
illustrate the underlying principles in a much simpler way than some of the other correlations. Its primary drawback is that
it relies on the English system of units to do so. The Bowring correlation is an interesting correlation because it probably
has a wider range of applicability (e.g., pressures and flow rates) than any of the other correlations for circular flow chan-
nels. It can also be used to find the location of dryout in a BWR, while some of the other correlations cannot. Hence, the
Bowring correlation is capable of handling both film dryout in BWRs and localized boiling in PWRs. Not surprisingly, it
is more complex than the Bernath and the Tong correlations, which have a much narrower range of applicability.
27.9
T he Tong Correlation
The Tong correlation was originally developed by L.S. Tong in 1966 to predict where rod dryout would occur. It is one
of the most famous correlations for calculating the critical heat flux in heated circular flow channels. It is intended to be
used primarily to find the value of the critical heat flux during departure from nucleate boiling in PWRs. Sometimes it
is called the Tong 68 correlation. The specific form of this correlation is
where
( )
x = − c pl h lv ⋅ ( TSAT − TBULK ) (<0 for most PWRs) (27.7)
1048 NUCLEAR HOT CHANNEL FACTORS, CHF, AND DNBR
In this correlation, notice that the critical heat flux increases as the 0.4 power of mass flux G. In the Tong correlation,
the value of x that is used to find the value of C is the equilibrium quality at the exit of the coolant channel, and it is
calculated based on the assumption of thermodynamic equilibrium. Consequently, it can be used to find the value of the
critical heat flux during subcooled boiling in PWRs where the average temperature of the coolant (TBULK) barely reaches
or exceeds the saturation temperature TSAT. During normal operation, x typically has a value between −0.10 and −0.05.
′′
Thus, the value of q CRITICAL at a particular axial elevation z depends on the value of x. Moreover, the value of x that is
needed to find the value of C never quite reaches the point where x = 0 and C = 1.76. This means that the value of C must
always be less than 1.76 in a PWR (for relatively low negative values of the exit quality). Here again, G is the mass flux
G = m/A, De is the equivalent diameter of the flow channel (which is the same as the physical diameter D if the chan-
nel is circular), hlv is the latent heat of vaporization (which can be found from the steam tables), and ul is the dynamic
viscosity of the liquid at the saturation temperature TSAT. For all of these parameters, SI units should be used. The Tong
68 correlation is the least conservative of all the major CHF correlations, and it is also one of the simplest ones. More
information about the Tong correlation can be found in the following reference
☉☉ Tong, L.S. and Weisman, J., Thermal Analysis of Pressurized Water Reactors, 3rd edition, American Nuclear
Society, Lagrange Park, Illinois, 1996.
27.10
T he Bernath Correlation
The Bernath correlation is another popular correlation that can be used to determine the value of the critical heat flux
during subcooled boiling, but it does not match experimental data particularly well for low-pressure bulk boiling. It was
originally presented in the A.I. Chem. Eng. Transactions in 1955, and it enjoyed a fair amount of popularity at the time.
Since then, it has been superseded by more modern correlations with larger data sets. A good discussion of the history
of the Bernath correlation can be found at
☉☉ http://www.trtr.org/Ann_ Mtg/2007%20meeting/07%20Presentations/Feldman/TRTR_2007_Earl_Feldman.
ppt#281,12,Bernath Correlation (1960).
The critical heat flux in the subcooled boiling region can be found from the following equation
where
( )
TWALL = 57 ln p − 54 ⋅ p (p + 15) − v/4 (27.9)
′′
is the wall (or cladding) temperature where the critical heat flux q CRITICAL occurs, TBULK is the bulk temperature of the
fluid (which is generally less than the saturation temperature) at the point of the CHF, and hCRITICAL is the heat transfer
coefficient where the critical heat flux occurs.
Here
C = 48 D e ⋅ 0.6 ( for D
e )
≤ 0.1 ft (27.11a)
T WALL and TBULK are measured in °F, p is measured in PSI, v is the velocity of the coolant (in ft/s), De is the equivalent
diameter of the coolant channel (in feet), and DH is the heated perimeter of the coolant channel in feet, divided by π
(3.71828). The Bernath correlation can be used for fluid velocities between 4 and 54 ft/s, pressures between 23 and
3,000 PSI, and equivalent diameters between 0.145 and 0.600 in. Because the v alues of the critical heat flux, the criti-
cal wall temperature, and the critical heat transfer coefficient can be found separately, this correlation can be used to
illustrate the relationships between them more easily than other correlations do. The following example illustrates the
process involved.
27.11 The Groeneveld Correlation 1049
Compared to other correlations, the Bernath correlation is not particularly conservative because it uses the c enterpoint
of a large number of data points to arrive at an analytical expression for the critical heat flux. Other correlations either
use much tighter envelopes of experimental data or they avoid the problem entirely by defining a limit line based on the
lowest measured values of the CHF. Then, there is no reason to find the “center of gravity” for the resulting data set.
Most modern correlations also have a lower RMS (or Root Mean Squared) error than the Bernath correlation, and these
correlations have average errors between 7% and 10%.
27.11
T he Groeneveld Correlation
Several years after the Bernath correlation was introduced, another correlation called the Groeneveld correlation
appeared in the literature. The Groeneveld correlation was subsequently implemented in RELAP5, and over time, it has
become the “politically correct” correlation in the NRC’s primary reactor licensing code—the TRACE computer pro-
gram. This correlation is based on a set of lookup tables where the critical heat flux q ′′CRITICAL is presented as a function of
☉☉ the pressure p (in kPa)
☉☉ the mass flux G (in kg/m2 s)
☉☉ the local flow quality x.
Therefore, a complicated set of analytical expressions is not used to compute the critical heat flux with the Groeneveld
correlation. The Groeneveld correlation was initially based on water flowing through an 8 mm diameter vertical tube
that was heated from the periphery, but it was extended in 2006 to include other channel diameters and subchannel
geometries. Initially, it was applicable to the following range of conditions:
0.1 ≤ p ≤ 21 MPa
0 < G ≤ 8,000 kg m 2s
3 < D ≤ 25 mm
However, it has been extended to a much wider range of conditions since that time. A negative value of the quality is then
used to represent regions where subcooled nucleate boiling occurs. For representative PWR fuel assemblies with a mass
flux of 3,800 kg/m2 s, a system pressure of 2,250 PSI (15.5 MPa), and an exit q uality of −0.05, the Groeneveld correla-
tion predicts a CHF of 2.94 × 10 5 W/m2. An extensive discussion of the Groeneveld correlation can be found in Todreas
and Kazimi (2008; refer to the references at the end of the chapter) and in the following reference on the internet:
☉☉ http://www.trtr.org/Ann_ Mtg/2007%20meeting/07%20Presentations/Feldman/TRTR_2007_Earl_Feldman.
ppt#282,14,1986, 1995, and 2006 Groeneveld CHF Look-Up Tables.
1050 NUCLEAR HOT CHANNEL FACTORS, CHF, AND DNBR
TABLE 27.3
The Meanings of the Eight Multiplicative Correction Factors That Are Used with the
Groeneveld Lookup Tables to Calculate the CHF in Reactor Fuel Assemblies
Different regions of the lookup tables are shaded in different colors to specify the degree of uncertainty in the estimated
values for the CHF. In addition, multiplicative correction factors are provided with the tables to account for specific dif-
ferences in subchannel geometries that were not included in the original data set. The actual corrections to the original
data set are then contained in eight correction factors that attempt to account for changes in the geometry and the heat
flux. In the literature, these correction factors are represented by the symbols K1 to K8, and their meanings are shown in
Table 27.3. The actual critical heat flux q ′′CR (z) is then found by multiplying the value provided in the Groeneveld lookup
table ( q CR
′′ (z) )LUT by the correction factors which apply. For example,
q ′′CR (z) = K1K 3K 5K 6 ⋅ ( q ′′CR (z) )LUT (27.12)
Hence, the critical heat flux for virtually any rod bundle geometry and combination of radial and axial heat fluxes can
be found in this way. Updates to the CHF lookup tables in 2006 improved the accuracy of the Groeneveld correlation
significantly in the subcooled boiling region and in the limiting quality region. Today, the Groeneveld correlation is
looked upon as one of the best critical heat flux correlations available. The Groeneveld correlation is also one of the
best correlations to use for TRIGA reactors, which are discussed in our companion book.*
27.12
T he Bowring Correlation
The Bowring correlation is the last of the critical heat flux correlations for circular flow channels that we will discuss. It
uses four constants C1, C2, C3, and C4 that are a function of the system pressure p. It has a very low RMS error (about
7%), and it can be used the find the critical heat flux for both DNB and annular film dryout. In other words, it can be used
for both PWRs and BWRs. Its database also supports the widest data set of any correlation when it comes to the values of
p and G. It is valid for system pressures between 0.2 and 19.0 MPa, and mass fluxes between 135 and 18,600 kg/m2 s. It
can be used for equivalent diameters between 3 and 25 mm, and it can be used for equilibrium qualities between −50%
and +90% (−0.50 and +0.90). The reference pressure pR in the Bowring correlation is given by
pR = 0.145p (27.13)
where p is the system pressure in MPa. The exact form of the Bowring correlation is
where
(
A = 0.579 ⋅ C1 ⋅ h lv DG 1.0.0143 ⋅ C2 ⋅ D 0.5G (27.15a))
B = 0.25DG (27.15b)
(
C = 0.077C3 ⋅ DG 1 + 0.347 ⋅ C4 G 1,356 ( ) ) (27.15c)
n
* See, for example, an Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
27.13 The Westinghouse W-3 Correlation 1051
n = 2.0 − 0.5pR, and the values of C1, C2, C3, and C4 are given by two separate sets of equations, depending upon the
value of pR. For pR ≥ 1 MPa (145 PSI)
C3 = pR 0.219 (27.16c)
C4 = C3 ⋅ pR 1.649 (27.16d)
( )
C2 = C1 pR 1.316e 2.444 (1− pR ) + 0.309 1.309 (27.17b)
C4 = C3 ⋅ pR 1.649 (27.17d)
The critical heat flux is calculated in W/m2. The Bowring correlation is good for the following range of conditions:
0.2 ≤ p ≤ 19 MPa
′′
Notice that once the system pressure p is set, the value of q CRITICAL becomes a function of the mass flow rate (i.e., the
mass flux G) and the local quality x. For a heated 10 mm vertical tube with a mass flux of 2,000 kg/m2 s, a pressure of
6.9 MPa, and an inlet temperature of 205°C, the Bowring correlation predicts a critical heat flux of about 1.5 × 105 W/m2
(or 150 W/cm2). Neither the Bowring correlation nor the Bernath and Tong correlations make any distinction between
the value of the critical heat flux when the axial power profile is uniform and when it is not. However, a correlation
like the W-3 correlation, which we will discuss next, actually attempts to take the shape of the axial power profile into
account. As we shall see, even a small change in the shape of the heat flux can have a significant impact on the value of
the critical heat flux.
27.13
T he Westinghouse W-3 Correlation
The W-3 correlation is the most famous and most widely used correlation for the critical heat flux in PWRs. This c orrelation
was developed by L.S. Tong at Westinghouse Nuclear Energy Systems several years after the first Tong 68 correlation
appeared. It can be used to predict the critical heat flux in circular, rectangular, and realistic rod bundle geometries when
a DNB condition is initiated by subcooled nucleate boiling. It is highly optimized to solve this specific problem and it is
not recommended for critical heat flux calculations in BWRs, where the flow quality at the dryout point is much higher
and where annular flow can occur. The W-3 correlation was originally developed to be used with a uniform axial uniform
heat flux (i.e., q″(z) = constant), but it also has a shape correction factor FSHAPE that can be used to account for the effects
of a nonuniform axial power distribution on the value of q ′′CRITICAL. The shape correction factor was also proposed by L.S.
Tong, although other people like Silvestri developed the same general approach independently at about the same time. See
☉☉ Silvestri, M., On the burnout equation and the location of burnout points, Energia Nucleare, 13(9); 469–479,
1966.
1052 NUCLEAR HOT CHANNEL FACTORS, CHF, AND DNBR
Finally, the effects of grid spacers on the value of the critical heat flux are taken into account by introducing other f actors
such as the grid correction factor FGRID into the basic model. In its simplest form, the W-3 correlation can be expressed
as the product or quotient of four specific factors:
where the shape correction factor FSHAPE can have a value less than 1.0, equal to 1.0, or greater than 1.0, and PFACTOR,
GFACTOR, and HFACTOR are specific factors that attempt to account for the effects of pressure, mass flux, and enthalpy on
the value of the critical heat flux. The shape correction factor usually has a value between 0.95 and 1.25, and the values
of PFACTOR, GFACTOR, and HFACTOR are given by
Notice that the GFACTOR is approximately a linear function of the mass flux. Hence, doubling the mass flux nearly dou-
′′
bles the value of the critical heat flux. In the W-3 correlation, the value of q CRITICAL is given in kW/m2, the mass flux G is
in kg/m2 s, the pressure is in MPa, the enthalpy h is in kJ/kg, and the equivalent diameter De is in meters. For a coolant
channel with an axially uniform heat flux, the shape correction factor, FSHAPE = 1.0. The W-3 correlation was originally
intended to cover the following range of conditions:
Notice that the equilibrium quality, when calculated from a simple energy balance, can have a value between −0.15
and +0.27. It only achieves a value of 0.0 when the bulk temperature TBULK becomes equal to the saturation tempera-
ture TSAT (i.e., TBULK = TSAT). It reflects the fact that DNB can occur in PWRs even when the boiling is subcooled. For a
representative PWR fuel assembly with a mass flux of 3,800 kg/m2 s, a uniform heat flux, a system pressure of 2,250 PSI
(15.5 MPa), and an equilibrium quality of −0.05, the W-3 correlation predicts a critical heat flux of 2.90 × 105 W/m2.
Hence, its predictions are nearly identical to those of the Groeneveld correlation, which we discussed earlier.
27.14
T he W-3 Shape Correction Factor
The value of the shape correction factor (FSHAPE) in the W-3 correlation depends on whether the axial heat flux
☉☉ increases
☉☉ decreases
☉☉ stays the same.
as it approaches the CHF point zCRITICAL. (See Figure 27.9 to see how it behaves in more detail.) The reason for this
behavior is that the rate of energy addition is different for the same average value of q ′′MAX if the shape of the axial heat
flux is position dependent. Thus, adding less heat to the channel earlier in the process (lower in the core) reduces the
value of the shape factor and increases the value of the CHF. Conversely, if more heat is added to the coolant earlier (case
II in Figure 27.9), the opposite situation occurs. Hence, the shape correction factor accounts for the memory effect of the
27.14 The W-3 Shape Correction Factor 1053
FIGURE 27.9 The shape correction factor in the Westinghouse W-3 correlation.
TABLE 27.4
The Shape Factor Matrix for the W-3 Correlation
axial power profile on the critical heat flux. The relationship between the shape correction factor and the axial heat flux
is shown in Table 27.4. For case I (with a symmetrical cosine for the shape of q"(z)), the value of FSHAPE is about 1.17 at
z = 1 meter. For most cores (where the value of the heat flux is higher in the center), the values of the shape correction
factor can range from about 1.12 to 1.24. Thus, it is very important to take the axial variation of the heat flux into account
′′
when attempting to calculate an accurate value for the critical heat flux q CRITICAL . Today computer programs like VIPRE
calculate this value automatically, and properly coded algorithms are used to keep errors in the calculation of the critical
heat flux to a minimum. The actual value of the shape correction factor is given by the equation
( ))
z
FSHAPE (z) = C
∫ 0
(
q ′′ ( z ′ ) e − C(z − z′ )dz ′ q ′′ ( z ) 1 − e − Cz (27.21)
C = 185.60 ⋅ (1 − x(z) )
4.31
G −0.478 (27.22)
In the equation for C, G is the mass flux in kg/m2 s and x(z) is the equilibrium quality at the axial location of interest.
In general, the value of C decreases as the quality increases, and this amplifies the size of the memory effect. Finally,
when the equilibrium quality becomes very high (say x > 0.2), DNB no longer becomes the dominant form of CHF in a
coolant channel. Then when the critical heat flux occurs, it is more of a global phenomenon with very little dependence
on the value or the shape of the local heat flux. Example 27.3 illustrates how the critical heat flux can be found when the
axial heat flux is given by q″(z) = q ′′MAX cos( πpz/H), and H is the core height. Thus, the shape correction factor can be
thought of as a convenient way for the critical heat flux to remember the axial power profile that was used to create it.
27.15
Understanding Film Dryout in BWRs
In an earlier section, we mentioned that dryout in BWRs does not occur in exactly the same way it does in PWRs.
This behavior is attributable to differences in the mass flux, the flow quality, the heat flux, and the system pressure.
When a BWR is operating normally, the coolant begins to boil about 0.5 m (1.5 ft) from the inlet to the core. However,
this nucleate or subcooled boiling soon transitions into bulk boiling (see Figure 27.10), and about two-thirds of the
way up the core, this bulk boiling transitions into film boiling. For a representative coolant channel and even for the
hottest coolant channel, the cladding and coolant temperatures follow the general trend shown in Figure 27.10. These
FIGURE 27.11 Cladding and coolant temperatures during unsafe operation accompanying dryout.
temperatures are similar because the coolant temperature always stays at the saturation temperature. However, in the
hottest channel, the cladding temperature is higher. This is true for both a uniform heat flux and a nonuniform heat
flux. In Figure 27.12, this behavior is shown for a uniform axial heat flux. Now let us illustrate what happens to the flow
field if the flow rate remains constant, but the reactor power level is suddenly increased. This can occur if a control
rod is suddenly withdrawn from the core. Under these conditions, the heat flux q″ increases in direct proportion to the
reactor power level. The liquid film on the surface of the cladding vaporizes in the upper half of the core, and annular
flow transitions into what is called mist flow. At this location, the cladding temperature rises sharply and the convective
heat transfer coefficient falls sharply. The fuel rod dries out and the cladding fails. For the hottest channel, this scenario
is illustrated in Figure 27.11. Beyond the dryout point, the equilibrium quality may become equal to 1.0, and the steam
may even become superheated. Now let us turn our attention to how the critical heat flux and the dryout location can
be found under these conditions.
27.16
CHF Correlations for BWRs
In the early days of the nuclear industry, critical heat flux calculations for BWRs were based on the same model that was
used for PWRs. This approach used the MDNBR to define the critical heat flux, and a value of 1.9 was widely accepted
as the industry standard for the MDNBR. The most successful critical heat flux correlation for BWRs was the GE cor-
relation, which was developed by Janssen and Levy in the early 1960s. This correlation was based on the equilibrium
flow quality x and the mass flux G = m/A. The specific form of this correlation was
where
( )
x1 = 0.197 − 0.108 G 10 6 (27.24a)
( )
x2 = 0.254 − 0.026 G 10 6 (27.24b)
the mass flux G was measured in English units (lb/h ft2) and the critical heat flux qCRITICAL was given in BTU/h ft2.
Notice again that the critical heat flux is a linear function of the mass flux, and the constant of proportionality changes
with the equilibrium quality. The Janssen–Levy correlation was originally developed for pressures between 600 PSI
and 1,450 PSI, mass fluxes between 0.4 × 106 lb/h ft2 and 6.0 × 106 lb/h ft2, values of the quality up to 45% (x = 0.45),
coolant channels between 2.4 and 9.0 ft in length, and having equivalent diameters De between 0.25 and 1.25 in. In
some circles, it was simply called the GE correlation. Like most of the correlations before it, it was based largely on
data that was available for circular pipes. Eventually, more modern correlations were developed that were not subject to
this limitation. However, in the early days of the nuclear business, this was the primary correlation that was used. The
Janssen–Levy correlation used a different approach than some of the correlations that preceded it because it attempted
to define a limit line that was based on the lowest measured values for the critical heat flux. Thus, it was a very conservative
correlation, and this was one of the reasons for its initial success.
Equation 27.23b also applies and q ′′CRITICAL = 0.45 × 106 BTU/h ft2. Noting that 1,200 kW/m2 = 0.38 × 106 BTU/h ft2, the
′′
value of the DNBR at the center of the assembly is DNBR = q CRITICAL /q″ = 0.92 × 106/0.38 × 106 = 2.42. [Ans.]
However, after its initial success, it became apparent that the dryout of the liquid film in BWRs needed to be modeled
differently than the process of nucleate boiling in PWRs. There were a couple of reasons for this change in thinking.
First, the average power density in BWRs was much lower than it was in PWRs (55 kW/L vs. 110 kW/L). Next, because
of the way the core is designed, the heat does not enter the coolant as quickly as it does in a PWR. This means that
boiling is not localized at just a single axial elevation, and it is a core-wide effect where knowledge of the rate of heat
addition before the rods begin to dry out can be very important.
Finally, because the heat flux to mass flux ratio (q″/G) is not as high, nucleate boiling is much less pronounced in a
BWR than it is in a PWR. The fuel rods can only dry out when the liquid film completely evaporates from the surface of
the cladding, and this usually does not occur until the local quality reaches 20%–30%. By comparison, DNB can occur
in PWRs when the local quality is about 0.12 (12%). However, the computed quality would still be a negative number
(about −0.05) if an equilibrium model were used. For all of these reasons, it became more convenient to define a value
of the flow quality where the liquid film would completely disappear from the surface of the rods. This flow quality
eventually became known as the critical flow quality (CFQ), xCRITICAL, and today it is the primary way that the dryout is
determined in BWRs. It is the real flow quality, and it is much higher than the equilibrium quality at a comparable axial
elevation would suggest. Figure 27.12 illustrates how these two quantities compare a typical BWR fuel assembly with a
sinusoidal power shape. Hence, about half way up a 4 m core, the critical quality is nearly twice the equilibrium quality
Needless to say, this is a very important distinction to make. In BWRs, the critical quality has been determined to be a
function of the system pressure, p, the mass flux, G = m/a, the fuel assembly geometry, and the axial elevation, z. Hence,
with no loss of generality, we may write
( )
x CRITICAL = f p, G, z, D e , L (27.26)
27.16 CHF CORRELATIONS FOR BWRs 1057
FIGURE 27.12 The CFQ and the equilibrium flow quality in a typical BWR fuel assembly.
where De is the equivalent diameter of the flow channel and L represents the fuel assembly length. In this section, we
would like to present three correlations that can be used to find the location at which dryout occurs. However, these
correlations are based on an additional concept called the critical power level (or CPL), and the CPL can be thought of as
the reactor power level where the flow quality becomes equal to the CFQ. The CPL turns out to be a much more accurate
way to predict when the cladding will dry out in BWRs, and it is also less conservative than the concept of the minimum
CHFR (the MCHFR) under comparable conditions. One of the three correlations that we will present to find the critical
quality is the CISE-4 correlation. This correlation was originally developed at the CISE laboratories in Milan, Italy, by
G.P. Gaspari et al. in the mid-1970s. See
☉☉ Gaspari, G.P., et al., A rod centered subchannel analysis with turbulent (enthalpy) mixing for critical heat flux
prediction in rod clusters cooled by boiling water Proceedings of the 5th International Heat Transfer Conference,
Tokyo, Japan, 197; CONF-740925, 1975.
It is applicable to circular flow channels containing two-phase mixtures. The other two correlations we will present
include the GE GEXL correlation and the Hench–Gillis or the EPRI-II correlation. These correlations can be
applied to actual flow channels, while the CISE correlation is more appropriate for heated circular ones. All three of
these correlations assume that the key parameter that determines when the liquid film evaporates is the critical flow
quality xCR, which is defined in the following way
In general, the critical flow quality is a function of axial position z, and each correlation that we will present will create
a slightly different curve for xCR(z) for specific values of the mass flux G and the system pressure p. Thus, the power
level at which the critical flow quality is reached is called the critical power level, and it represents the limit line below
which a BWR must always operate. To illustrate this concept, suppose that we draw a curve of the critical quality as a
1058 NUCLEAR HOT CHANNEL FACTORS, CHF, AND DNBR
FIGURE 27.13 The CFQ is defined as the steam quality where dryout occurs. It depends on the mass flux G, the pressure p, and
′′ is known.
the axial elevation z. The CPR can then be deduced from the critical quality once the CHF q CR
function of axial elevation z. This is illustrated in Figure 27.13. As the reactor power level increases, the flow quality
progressively increases until it reaches the critical quality level. Here, film dryout will occur, and the fuel rods will fail.
This establishes the upper power level or the critical power level (CPL) at which a standard BWR can operate. A BWR
cannot operate at a higher power level without explicit authorization from the NRC.
27.17
T he Critical Power Ratio and Its Uses
Once the CPL (in MW) has been found, one can determine how far below this power level the reactor should be
allowed to operate. The ratio of the CPL to the normal power level is known as the critical power ratio (or the CPR),
and it is defined by
Physically, the critical steam quality xCRITICAL where dryout occurs depends on the values for G, p, and z. Hence, for a
BWR coolant channel
Using an appropriate correlation, we can then plot xCRITICAL as a function of z. The value of the reactor power P where
x = xCRITICAL is then defined as PCRITICAL. The CPR is therefore
For a BWR to be licensed in the United States, the CPR must always be greater than 1.2; that is, a BWR cannot be
licensed unless
In other words, a BWR must always be operated at least 20% below the critical power limit. This is equivalent to the
MDNBR for a PWR. However, at the present time, the minimum acceptable value of the DNBR is about 1.3. Several
publically available correlations (like the CISE-4 correlation) can be used to find the value of xCRITICAL. We would next
like to discuss how some of these correlations can be used.
27.18
T he CISE-4 Correlation for the CFQ
The CISE-4 correlation is a modern correlation that can be used to calculate xCRITICAL in a BWR. It is based on the
assumption that the flow quality will continue to increase in a coolant channel as a function of axial elevation z until the
fuel rods dry out. Then the value of xCRITICAL will be reached. The generic form of the CISE-4 correlation is
where
( )
A = 1 − p pcr (G/1,000)0.33 (27.33a)
( )
0.4
B = 0.199 pcr p − 1 ⋅ G ⋅ D e 1.4 (27.33b)
and z is the axial elevation (in meters). In the CISE-4 correlation, pcr is the thermodynamic critical pressure of water
(22.1 MPa) and the units for the mass flux G and the equivalent diameter De are kg/m2 s and m, respectively. The CISE-4
correlation is based on several earlier correlations (called the CISE-2 and CISE-3 correlations) that were first developed
at CISE laboratories in Milan, Italy, in the early 1970s. As we mentioned earlier, the CISE-4 correlation is also based
on the premise of being able to calculate the critical flow quality. It is designed only for BWRs, and so it is optimized
for typical BWR pressures, mass fluxes, and hydraulic diameters. Its stated range of applicability is between 7 MPa and
9 MPA (1,000 PSI to 1,300 PSI). In the CISE-4 correlation, the critical pressure pCRITICAL is the thermodynamic critical
pressure of water (~22.1 MPa), and G and De must be in SI units (kg/m2 s and m, respectively). PH is the heated perim-
eter, and PW is the wetted perimeter. The critical power level PCRITICAL is reached when the curve for the flow quality
x intersects with the curve for the critical flow quality xCRITICAL. In practice, a BWR must operate at least 20% below
this power level for film dryout to be avoided. We would now like to discuss some other correlations that can be used
to estimate the values for xCRITICAL and zCRITICAL. Again, the terminology is more complex because flow boiling can be
more complicated in BWRs than it is in PWRs.
27.19
T he GEXL Correlation
The GEXL correlation is a well-accepted correlation developed in the 1970s by GE to overcome the objections to
the original Janssen–Levy correlation. The GEXL correlation is more accurate (and hence less conservative) than the
Janssen–Levy correlation, and it is said to have a standard deviation (based on a very large amount of experimental data)
of about 3.5%. The word GEXL is an acronym for the General Electric Quality versus Boiling Length Correlation. It
is based on large experimental data sets that were obtained at GE’s ATLAS heat transfer facility. This correlation is
capable of predicting the critical quality where dryout occurs in full-scale 7 × 7 and 8 × 8 BWR fuel assemblies. In its
original form, the GEXL correlation was a function of G, p, and a number of important geometric parameters including
the hydraulic diameter and the boiling length:
Here, xCRITICAL is the critical flow quality (a number having a value from 0.0 to 1.0), LBOIL is the boiling length (the
distance from the point of zero equilibrium quality to the channel exit), LH is the total heated length, DH is the heated
diameter (four times the flow area divided by the heated rod perimeter), and PP is a parameter that is used to account for
the amount of power peaking in a fuel assembly relative to its most limiting rod. Unlike the Janssen–Levy correlation,
which attempts to define a lower limit line for the critical heat flux, the GEXL correlation is based on a least squares fit
to all of the available experimental data. Hence, it is much less conservative (and presumably more accurate) than the
Janssen–Levy correlation. Once the critical flow quality is found, the critical power level can be found by finding when
the power level causes the flow quality to intersect the CFQ curve. This allows us to calculate the CPR
which we discussed earlier. Unfortunately, the exact form of the GEXL correlation has continued to remain proprietary,
and hence, GE has not released its specific contents to the rest of the industry. For this reason, the specific form of the
correlation cannot be presented here. However, it is generally believed to be one of the best correlations available today,
and it is the primary design tool that GE uses to determine the values of xCRITICAL and the critical power ratio.
27.20
T he Hench–Gillis Correlation
Historically, the Hench–Gillis correlation was developed for BWR fuel assemblies by EPRI about the same time as the
GEXL correlation. It was initially developed because the need existed for a publically available correlation that could do
the same things that the GEXL correlation did. However, it was intended to be based on readily accessible experimental
data rather than the proprietary data that the GEXL correlation used. The Hench–Gillis correlation was developed for the
Electric Power Research Institute in Palo Alto, California, by J.E. Hench and J.C. Gillis, and over time, it also became known
as the EPRI-II correlation. Like the GEXL correlation, its goal was to find the critical flow quality, and its basic form is simi-
lar to that of the CISE-II correlation, which preceded it by several years. This correlation was based on publically available
data that was published for BWR fuel assemblies at the time. The exact form of the Hench–Gillis correlation is
where
R(z) = Boiling heat transfer area(z)/bundle flow area which may be a function of axial position, and PCF is a pressure
correction factor that can be found from the equation
where F = (p − 800)/1,000. Here, the pressure p is in PSI, the mass flux G is in Mlb/h ft2, D is the rod diameter, n is the
number of heat producing rods, and LBOIL is the boiling length of the coolant channel. The Hench–Gillis correlation is
used to find the value of the critical flow quality where dryout occurs. A and B are functions of the mass flux G, and the
mass flux must be specified in millions of lbs/h ft2. Note: One million lbs/h ft2 is equivalent to 1,356.5 kg/m2 s. Here, R(z)
is the ratio of the boiling heat transfer area to the bundle flow area:
The values for LBOIL(z), D, and A FLOW can be expressed in any units system (mm, cm, m, in., or even ft) as long as it is
used for all three variables because the value of R is dimensionless. The variable W is a weighting factor that depends on
the relative powers of the rods surrounding a particular rod. It accounts for local power peaks, and it has different values
for the side, the corner, and the central rods. For central rods, W has a value of about 1.03. However, values between
0.98 and 1.05 are commonly used. The references at the end of the chapter provide additional information about how the
values for W can be found. In general, a fuel rod surrounded by higher power rods will have a higher value of W, and a
fuel rod surrounded by lower power rods will have a lower one. A typical distribution of radial power peaking factors for
a 9 × 9 BWR fuel assembly is shown in Figure 27.14. The values of these factors go into the calculation of W. The cross
flow between the adjacent coolant channels is generally neglected when finding the value of G. Because the weighting
factors in the Hench–Gillis correlation vary as one moves around a fuel assembly (from rod to rod), different combina-
tions of fuel rods will have different values for xCRITICAL. However, the critical flow quality in an overpower transient will
usually occur about halfway between the midpoint of the core (z = H/2) and the top (z = H). Hence, dryout is reached
about 80% of the way up the core. For typical BWR fuel assemblies with a co-sinusoidal power shape, dryout occurs
between 3 and 3.5 m from the bottom of the core. If the power level is increased even further, the location of dryout will
move closer to the core centerline.
Of course, in real BWRs, there are many fuel assemblies with different radial and axial power shapes (see
Figure 27.14), and the location of dryout will be different from one fuel assembly to the next. Here, the first rod to fail
will usually be located in the hottest fuel assembly, and this rod will dryout at a lower axial elevation than it does in
27.20 The Hench–Gillis Correlation 1061
FIGURE 27.14 Radial power peaking factors used to calculate the weighting factor W in the Hench–Gillis correlation.
TABLE 27.5
Some Popular Correlations Used to Calculate the CHF in PWRs and BWRs
other fuel assemblies. For this reason, the critical quality calculation is usually performed for the hottest fuel rod in
the hottest fuel assembly. This then defines the CPR, and the maximum power level (CPL) at which the reactor can be
operated. The critical heat flux correlations we have discussed so far are summarized in Table 27.5. In correlations like
the Hence–Gillis correlation, the value of the boiling length LBOIL (see Equation 27.39) determines the shape of the
1062 NUCLEAR HOT CHANNEL FACTORS, CHF, AND DNBR
critical quality curve. The boiling length starts at the point where the equilibrium quality xEQ becomes equal to zero.
Hence, its value is given by
and its value is zero at the location where the equilibrium quality is zero. Its value then increases monotonically as a
function of axial position (see Figure 27.15), and in a 4 m high core, its value is about 3 m at the top of the core. When the
equilibrium quality and the critical quality are then plotted at a function of axial position, we obtain the curve shown in
Figure 27.16. Thus, both the critical quality and the equilibrium quality increase after the zero quality point is reached,
and in general, the critical flow quality increases much faster than the equilibrium quality. At the top of the core, the
critical flow quality is about 35%. This means that if the exit quality exceeds this value, the fuel rods will fail. Once
FIGURE 27.16 The equilibrium quality curve and the critical quality curve for a commercial BWR. Normally, when the
equilibrium quality exceeds about 35% at the core exit, the fuel rods will dry out and fail.
27.21 Thermal Design Limits and Regulatory Requirements 1063
FIGURE 27.17 As the power level increases, the applied heat flux approaches the CHF. At the point where these two curves
intersect, the cladding dries out and the fuel rods fail. Normally, this point is located near the top of the core.
the shape of the critical quality curve is known, it is then possible to increase the power level to see the effect it has
on the value of the equilibrium quality xEQ(z). As the power level is raised, the equilibrium quality curve comes closer
to the critical quality curve (see Figure 27.17). At the power level where the two intersect, the fuel rods dry out and the
cladding fails. As we mentioned earlier, the CPR for an American BWR is always greater than 1.2. In other words, the
power level in the hottest channel must be approximately 20% less than the power level at which the critical quality is
reached. Thus for BWRs, the CPR defines the operating limit as a simple function of the reactor power. However, for
PWRs, this operating limit is defined differently because the DNBR represents the ratio of the critical heat flux to local
heat flux at each axial location
Hence, the value of the MDNBR for a PWR may be different than the value of the CPR for a BWR. It is important to
keep this distinction in mind when attempting to determine the operating power. The difference between this power
level (PCRITICAL) and the operating power level (P = 100%) then defines the CPR
27.21
T hermal Design Limits and Regulatory Requirements
Over the years, regulatory agencies have employed several approaches to ensure that the heat flux does not exceed
the CHF and the flow quality does not exceed the CFQ. These safety limits or safety margins for PWRs are normally
expressed in terms of the DNBR, and for BWRs, they are normally expressed in terms of the DNBR or the CPR.
Sometimes the Departure from Nucleate Boiling Ratio is called the DNB ratio or the DNBR. Once the critical heat flux
has been calculated using the correlations presented earlier, the DNBR can be defined as the ratio of the critical heat
flux to the maximum heat flux in the core:
Since the heat flux is position dependent, the DNBR for a PWR can be also written as
′′ = q CRITICAL
where q DNB ′′ is the critical heat flux at a particular axial location z = zCRITICAL where DNB occurs. If the heat
flux q″(z) at a particular axial elevation is less than the critical heat flux q ′′CRITICAL (z) at that location, for example,
then DNB will not occur and the cladding will not fail. Equation 27.45 is sometimes called the DNB limit, and all PWRs
are expected to respect this limit during normal operation.
27.22
PWR DNBR Limits
For commercial PWRs in the United States (see Section 27.5), the CHF on the outer surface of the cladding occurs at
about 300 W/cm2. The DNBR is then required to remain above 1.3 at all axial elevations to keep the core in a safe and
stable state. In other words, the core must be operated about 30% below the DNBR limit. Consequently, the design limit
for commercial PWRs in the United States is
Not surprisingly, this design limit applies to both steady-state and time-dependent operation. In other words, it applies
to both a 15% overpower condition and a 15% overpower transient. Example 27.5 illustrates how this design limit deter-
mines the steady-state power rating. Figure 27.18 shows the critical heat flux and the actual heat flux q″(z) in a uniformly
heated core with a standard axial power shape q″(z) = q″ cos (πz/H). For a PWR to be operated in the United States, the
value of DNBR must always exceed 1.3 at any axial location. Notice that the minimum value of the DNBR is usually
reached about three-fourths of the way up the core. However, if the power shape happens to be axially uniform instead
(i.e., q″(z) = constant), and there was no radial variation in its value, then the point of MDNBR would be reached much
FIGURE 27.18 The DNBR and the CHF as a function of axial elevation in a uniformly loaded core.
27.23 BWR DNBR Limits 1065
FIGURE 27.19 How the DNBR changes as the power is increased and decreased. The MDNBR moves closer to the core midplane
when the heat flux is high and further away from the core midplane when the heat flux is low. As long as the DNBR stays above 1.3,
the reactor is considered to be safe.
closer to the top of the core. (However, the neutron leakage at the top of the core usually precludes this from happening.)
Moreover, although the shape of the curve shown in Figure 27.18 is essentially the same from one PWR to the next, it
tends to move lower as the core power level is raised. Thus, it intersects with the axial heat flux curve sooner at higher
power levels than at lower ones. This has the effect of moving the MDNBR point closer to the core midplane at very high
values of the nuclear heat flux and further away from the midplane for very low levels of the nuclear heat flux. Again,
this behavior is acceptable as long as the value of the MDNBR does not fall below 1.3. The power level for the hottest
fuel assembly is then determined in this way. Figure 27.19 shows how the MDNBR changes as the heat flux is raised and
lowered. As long as the DNBR stays above 1.3, the reactor is assumed to be safe.
27.23
BWR DNBR Limits
A similar approach is used to define the DNBR limits in BWRs, but because the flow regimes that lead to film dryout
are different than they are in PWRs (see Section 27.3), the core must be operated considerably below the DNBR limit
to ensure that dryout does not occur. This then results in an alternative DNBR limit for commercial BWRs operated in
the United States:
TABLE 27.6
Some Common PWR and BWR DNBR and CPR Limits in the United States
Because of the conservatism of this limit, the DNBR has been replaced in many countries by a more accurate measure-
ment of film dryout called the CFQ. The CFQ then defines the CPR and the maximum power level (CPL) at which a
BWR can be safely operated. In the United States, the CPL is generally set to 20% above the rated power level because
the CPR is 1.2.
In other countries, the CPR can have higher or lower values. Thus the CPR also attempts to limit the CHF at which rod
dryout will occur. Table 27.6 lists the current CPR and DNBR limits imposed by the U.S. NRC. Most countries in the
western world apply similar limits to ensure the safety of their PWRs and BWRs.
27.24
Comparison of Critical Quality and CHF Correlations
Each of the correlations we have discussed so far creates a slightly different curve for the CHF and the CFQ. These
curves can then be used to create curves for the behavior of the CPR in BWRs and the MDNBR in PWRs as a func-
tion of the thermal energy deposited in a typical coolant channel. For reactor fuel assemblies, these curves are shown in
Figures 27.20 and 27.21. These curves then create an envelope of statistical uncertainty regarding the values of the CPR
and the MDNBR that should be used for a particular design. Normally, the midpoint of these curves is used to establish
the operating power level. Thus, if the CPR is set at 1.2, the power the hot channel in a BWR can absorb is set at about
FIGURE 27.20 The values of the CPR for a number of popular correlations for the hottest channel in a BWR core. Because these
correlations predict slightly different values for the CPR, the midpoint of these values is commonly used to set the critical condition.
27.25 Other Thermal Limitations on Core Design 1067
FIGURE 27.21 The values of the MDNBR predicted by a number of popular correlations for the hottest channel in a PWR core.
Because these correlations predict slightly different values for the MDNBR, the midpoint of the values shown is commonly used to
set the appropriate power level.
130 kW. Similarly, if the MDNBR is set at 1.3, the power the hot channel in a PWR can absorb is set at about 175 kW.
Higher or lower values are possible depending upon which correlation is used to establish the critical condition. Whatever
line is selected to represent the CPR or MDNBR limit then defines the highest power that is allowed for any coolant chan-
nel in the core. Normally, this limit applies to the hottest rod or to the hottest coolant channel surrounding the hottest rod.
Since this line of demarcation defines the upper allowable operating power for a PWR or BWR, it is sometimes called the
limit line for the core. This terminology is also used to represent the maximum acceptable value of any parameter that
affects the overall safety of the plant. One final point we would like to make is that all of the CHF correlations we have
discussed so far predict essentially the same dependence between the CHF, the real heat flux, and the core power level
when the pressure, the mass flux, and the inlet enthalpy are the same. Then the behavior of both q″ and q ′′CRITICAL are driven
primarily by the core power level P. This dependence is illustrated for the interior hot channel of an American PWR in
Figure 27.22. In this case, the actual heat flux and the critical heat flux curves intersect at a value of about 300 W/cm2
(3,000 kW/m2). Here, the cladding dries out and the fuel rods fail. Thus, American PWRs are usually operated about 30%
below this limit, and this corresponds to a heat flux in the hot channel of about 200 W/cm2 (2,000 kW/m2) at 100% of rated
power. During a 15% overpower transient, the critical heat flux increases to approximately 230–240 W/cm2.
27.25
O ther Thermal Limitations on Core Design
Rational core design is predicated on the assumption that the fuel does not melt and that the cladding does not fail. For
American PWRs, these design restrictions usually limit the linear power density to about 650 W/cm, and the surface
heat flux to about 300 W/cm2. However, not all of the fuel rods in the core operate at exactly the same temperature, and
the surface heat flux is not exactly the same for each rod. In fact, the radial power generation rate (and hence the surface
heat flux) can vary by a factor of three or more in a well-designed core. Moreover, most fuel rods operate well below
these critical thermal limits, and only a few rods in the core actually approach the design limits that are encountered
in practice. Consequently, most initial approaches to core design attempt to investigate how closely the hottest fuel
assemblies in the core come to exceeding these thermal design limits. Then, if one can show that the hottest channels
and the hottest fuel assemblies fall below these design limits, then it is relatively easy to demonstrate that the remaining
channels in the core will also do the same.
This approach to core design assumes that the hottest coolant channels in the core are those where the surface heat
flux and the enthalpy rise are the greatest. Thus, if all the fuel rods are identical, the hottest rods and the hottest channels
1068 NUCLEAR HOT CHANNEL FACTORS, CHF, AND DNBR
′′
FIGURE 27.22 The dependence of the CHF q CRITICAL and the actual heat flux q″ on the core power level in a typical PWR. At about
130% of rated power, the two curves overlap, the cladding dries out, and the fuel rods fail. To avoid this, the MDNBR must be kept
above 1.0.
will also be the most limiting ones. Following this approach, the hot spot in the core can be defined as the point of
maximum heat flux or linear power density, and the hot channel can be defined as the coolant channel along which
the enthalpy rise is greatest. The remaining channels will then have lower fuel pin temperatures and lower enthalpy
increases than the hot channels do. These inherent variations in the core temperature profile and enthalpy rise can then
be described using what are called nuclear hot channel factors. Consequently, these hot channel factors can be used to
gain additional insight into the thermal-hydraulic performance of the core.
27.26
Nuclear Hot Channel Factors
Sometimes additional parameters are required to account for different radial and axial heat generation rates in PWR and
BWR cores. The most important of these parameters is an interesting parameter called the nuclear hot channel factor
F NUCLEAR. The hot channel factor is defined as the ratio of the maximum heat flux in the hottest channel in the core q ′′MAX
(measured in W/cm2) to the average heat flux q ′′AVERAGE in a representative channel in the core (also measured in W/cm2):
Hence, F NUCLEAR is a dimensionless number greater than 1.0, and for zone-loaded PWR cores, it typically has a value of
about 2.5. Normally, the total hot channel factor is found by multiplying together the hot channel factors for the radial
and axial directions, which are used to take into account additional variations in the radial and axial power profiles.
Thus, the radial and axial hot channel factors are defined by
27.26 Nuclear Hot Channel Factors 1069
and
The product of these two factors then defines the total nuclear hot channel factor:
For unreflected cylindrical cores with a single uniform enrichment, the values of FRADIAL and FAXIAL are a pproximately
2.32 and 1.57, respectively. This results in a value for the total nuclear hot channel factor of F NUCLEAR = FRADIAL·FAXIAL =
2.32 × 1.57 = 3.64. Thus, in these cores, the hottest fuel rod and coolant channel operate with temperatures and enthalpy
increases over 3.5 times greater than the average rod or coolant channel. Radially and axially zoning the core, and sur-
rounding it with a reflector (see the extensive discussion of neutron reflectors in our companion book*) then reduces
this number to about 2.5. Hot channel factors for the fuel assemblies in a typical BWR core are shown in Figure 27.23.
FIGURE 27.23 Representative values for the nuclear hot channel factors for a one-fourth symmetry section of a commercial
reactor core. Here the hottest fuel assemblies are shown in red and the coolest ones are shown in yellow. The power distribution is
controlled by varying the enrichment of the fuel. (See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press,
Boca Raton, FL (2017).)
27.27
Engineering Uncertainty Factors and Their Effect
on the Nuclear Hot Channel Factor
Normally, the values of the nuclear hot channel factor differ from the values we calculated earlier due to s tatistical
uncertainties in the values of the core components over which the designer of the core has little or no control. For
example, the amount of U-235 in the fuel rods can vary slightly from what is called for in the design specifica-
tions. Hence, the actual enrichment of the fuel rods may be different than the intended enrichment (usually ±1%–2%)
because of the inherent uncertainties in the manufacturing process—even though the average values of the actual
enrichment and the intended enrichment may be the same. Similarly, the cladding is subject to the same engineering
uncertainties as well. There may be variations in the cladding thickness from one fuel rod to the next. Finally, after the
fuel rods are placed in the core, they can bow or deform slightly at different radial and axial locations. This bowing
can cause the coolant flow in the vicinity of one fuel rod to be different than that in the vicinity of another. Thus, if the
flow is reduced at the point where q″ is the highest, then the value of the nuclear hot channel factor will be higher than
its intended value. These manufacturing and operational factors create statistical or engineering uncertainties in the
value of the nuclear hot channel factor F NUCLEAR that may cause it to become higher or lower than its calculated value.
To address the effect of these manufacturing and operational uncertainties on the nuclear heat flux, an engineering
uncertainty factor, F ENGINEERING, is sometimes added to Equation 27.49. A value of F ENGINEERING = 1.0 means that there
are no statistical uncertainties in the calculation, and a value greater than 1.0 indicates that these uncertainties exist.
For example, if the value of the nuclear hot channel factor is 2.5, and the engineering/operational uncertainty factor is
1.05, then the expected value of the nuclear hot channel factor will range from a minimum value of 0.95 × 2.5 = 2.38
to a maximum value of 1.05 × 2.5 = 2.63. Normally, the value of F ENGINEERING in a modern PWR is about 1.04, although
values of 1.12 have sometimes been reported. The exact value of F ENGINEERING can be deduced by analyzing each of the
steps in the fuel rod manufacturing process. The more uncertainties that are present (or the lower the manufacturing
tolerances are), the greater the value of F ENGINEERING will be. (In a perfect world where there are no uncertainties, the
value of F ENGINEERING would be 1.0.) When these engineering uncertainties are taken into account, overall hot channel
factor becomes
where the values of F RADIAL and FAXIAL are determined by the core power profile, and the value of F ENGINEERING is a func-
tion of the manufacturing process itself. The reader should refer to Chapter 26 for a more extensive discussion of how
these engineering uncertainty factors are measured and defined. Typically, engineering uncertainty factors are assigned
to each component of a fuel rod including the enrichment of the fuel, the thickness of the cladding, and the average
width of the fuel to cladding gap. These individual uncertainty f actors are then referred to as engineering uncertainty
sub factors. If each of the components of a fuel rod is produced by a separate process, then the engineering sub factors
will be statistically independent, and the total engineering uncertainty factor FENGINEERING, will be the product of the
individual engineering sub factors:
For example, if the values of F ENRICHMENT, FCLADDING, and FGAP are measured to be 1.01, 1.01, and 1.02, respectively, the
total engineering uncertainty factor will be
Hence, it should come as no surprise that the U.S. NRC requires each reactor vendor to provide the values of the engi-
neering sub factors that they intend to use to conduct their hot channel analysis. This information is then fed into the
values for the nuclear hot channel factor and the CHF calculation. Today, computer programs can calculate FENGINEERING
from engineering data for each lot of fuel. These calculations are more complex than the ones we have just described, but
the approach they employ is essentially the same. We would next like to turn out attention to how statistical uncertain-
ties in the values of these key components enter into the licensing process for a commercial nuclear power plant. This
will be one of the most important topics of our discussion in Chapter 33. Before beginning this discussion, we would
Questions for the Student 1071
first like to turn our attention to how LOCAs (Loss of Coolant Accidents) and other DBAs (Design Basis Accidents)
occur. This will require us to first understand compressible flow and choke flow, which are described in great detail in
the next few chapters.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Web References
http://ocw.mit.edu/courses/nuclear-engineering/22-06-engineering-of-nuclear-systems-fall-2010/lectures-andreadings/MIT22_06F10_
lec06b.pdf.
http://digbib.ubka.uni-karlsruhe.de/volltexte/fzk/6825/6825.pdf.
http://www.trtr.org/Ann_Mtg/2007%20meeting/07%20Presentations/Feldman/TRTR_2007_Earl_Feldman.ppt#281,12,BernathCorrela
tion(1960).
h t t p : / / w w w. t r t r. o rg / A n n _ M t g / 2 0 0 7 % 2 0 m e e t i n g / 0 7 % 2 0 P re s e n t a t i o n s / Fe l d m a n / T R T R _ 2 0 0 7 _ E a r l _ Fe l d m a n .
ppt#282,14,1986,1995,and2006 Groeneveld CHF Look-Up Tables.
Additional References
Gaspari, G.P., et al., A rod centered subchannel analysis with turbulent (enthalpy) mixing for critical heat flux prediction in
rod clusters cooled by boiling water, Proceedings of the 5th International Heat Transfer Conference, Tokyo, Japan, 197;
CONF-740925, 1975.
Silvestri, M., “On the burnout equation and the location of burnout points”, Energia Nucleare, 13(9):469–479, 1966.
Tong, L.S. and Weisman, J., Thermal Analysis of Pressurized Water Reactors, Third edition, American Nuclear Society, Lagrange
Park, IL, 1996.
13. In general, how does the value of the CHF vary as a function of axial position in a PWR or BWR core?
14. The breakdown of the coolant on the surface of a nuclear fuel rod has many different names. Name at least three
common terms that can be used to describe this effect.
15. What types of flow channels were used to develop the first CHF correlations?
16. What does the term “limit line” refer to in a commercial PWR or BWR?
17. A boiling water reactor in the United States is designed and licensed to be operated at a power level of up to
3,600 MWT. At what power level will the fuel rods begin to dry out and fail? What is the equilibrium quality at the
point where dryout occurs?
18. Name at least four popular correlations that can be used to calculate the CHF in a commercial PWR.
19. The CHF in a Westinghouse PWR has a value of 2.5 × 106 W/m2 just above the core midplane. If the actual heat flux
at this location is 1.5 × 106 W/m2, what is the value of the DNBR at this point?
20. What was the first poplar correlation to be used to calculate the value of the CHF in commercial PWRs?
21. What is the definition of the critical quality, and at the burnout point, is its value larger than, the same as, or less
than the equilibrium quality?
22. Fill in the following sentence with the appropriate word or phrase: The correlation is probably the most
accurate correlation for finding the CHF in water reactors today. It is the default correlation used by the U.S. NCR
in its reactor licensing code .
23. The Groeneveld correlation uses a set of lookup tables to calculate the value of the CHF. What are the three primary
parameters needed to find the CHF when the Groeneveld correlation is used?
24. What is the term DNBR an abbreviation for, and what is the minimum acceptable value of the DNBR for PWRs and
BWRs to be currently licensed in the United States?
25. In general, is the CFQ a more representative measure of the true flow quality than the equilibrium flow quality?
26. In the Groeneveld correlation, multiplicative correction factors are used to account for specific differences in
subchannel geometries that were not included in the original data set upon which the Groeneveld correlation
was based. How many of these correction factors does the Groeneveld correlation use, and can you name at least
three of them?
27. Fill in the following sentence with the appropriate word or phrase: In water reactors, the boiling length starts at the
point where the equilibrium quality xeq becomes equal to . In a 4 m high core, where does this point
generally occur?
28. What is the only major CHF correlation to be used for both PWRs and BWRs?
29. What is the definition of the nuclear hot channel factor?
30. Suppose that the nuclear hot channel factor for the hottest fuel rod in a reactor fuel assembly is 1.3 and the engineer-
ing uncertainty factor for the same fuel rod is 1.04. What is the overall hot channel factor for the fuel rod?
31. What person was responsible for developing the W-3 correlation, and what major reactor vendor is it owned by
today?
32. In the Hench–Gillis correlation, the CFQ xCRITICAL is expressed as a function of the boiling length. How is the
boiling length defined in this correlation?
33. Name at least three correlations that can be used to calculate the CFQ.
34. How is the CPR deduced from the CFQ?
35. What is the range of applicability of the CISE-4 correlation, and where was this correlation originally developed?
36. At what axial elevation does dryout normally occur in a PWR with a co-sinusoidal heat flux? What is the value of
the MDNBR at this location?
37. At what value of the equilibrium flow quality does dryout occur in a PWR fuel assembly? Is this quality higher,
lower, or the same as the equilibrium quality at which dryout occurs in a comparable BWR fuel assembly?
38. In all water reactors, the CHF is a function of the axial mass flow rate. If the axial mass flow rate increases, does
the CHF increase or decrease?
39. Is the following statement true or false? DNB can occur in PWRs even when the boiling is subcooled.
40. In most CHF correlations for water reactors, the CHF is a function of the exit flow quality. If the exit flow quality
increases, does the CHF increase or decrease?
41. If you were to design a new CHF correlation today, what variables would you use to calculate the CHF? Could a
concept such as the CPR also be applied to PWRs?
42. Based on the information presented in this chapter, do you believe that the CHF, the DNBR, the CFQ, and the CPR
are the most effective tools to predict when and where dryout occurs?
43. If you were put in charge of building a nuclear power plant today, what type of nuclear power plant would you build
and why?
Exercises for the Student 1073
Exercise 27.2
A coolant channel in the core of a commercial BWR has a cross-sectional area of 0.00015 m2. The operating pressure is
7 MPa (~1,000 PSI). Water enters the channel from below as a saturated liquid and leaves the channel from above with
an exit quality of 14%. The slip ratio is 3 and the acceleration pressure drop is 0.1 PSI. How much heat is generated by
the fuel rods surrounding the channel?
Exercise 27.3
A boiling water reactor operates at 7 MPa (~1,000 PSI). The coolant channels are 4 m high and contain 1 cm diameter
fuel rods. Water enters the core from below as a saturated liquid at 5 m/s. The flow area per fuel element is 0.00015 m2.
Thermal energy is generated sinusoidally, and the maximum heat flux is 1 × 106 W/m2. Calculate the CHF at the entrance,
the center, and the exit of the channel. Sketch the variation of the CHF and the actual heat flux as a function of axial
position. Sketch the ratio of the CHF to the actual heat flux along the channel.
Exercise 27.4
A boiling water reactor operates at 7 MPa (~1,000 PSI). The coolant channels are 4 m high and contain 1 cm diameter
fuel rods. Water enters the core from below as a subcooled liquid with an equilibrium quality of −0.03. Between the
bottom of the core and the top of the core, heat is added to the coolant at a rate of 265 kJ/kg. The mass flow rate in a
typical coolant channel is 0.18 kg/s. At approximately what elevation above the inlet does the coolant first start to boil
(as defined by the point where the equilibrium quality becomes greater than zero)?
Exercise 27.5
Using the data provided in Exercise 27.4, calculate the specific enthalpy at the inlet and the exit of the core. What is the
average enthalpy increase between the bottom and the top of the core? If the peak to average power ratio in the radial
direction is 2 to 1, what is the enthalpy increase in the hottest channel?
Exercise 27.6
Using the data provided in Exercise 27.5, calculate the void fraction and the equilibrium quality at the inlet and the
exit of the core assuming a slip ratio of 1.0. If the slip ratio at the top of the core is 4.0, what does the value for the void
fraction become?
Exercise 27.7
Using the Groeneveld lookup tables, calculate the CHF in a commercial PWR assuming a pressure of 15.5 MPa, a mass
flux of 4,000 kg/m2 s, and an equilibrium quality of—0.05. Sketch a picture of the equilibrium quality as a function of
core height. Does the equilibrium quality ever become positive before leaving the core?
Exercise 27.8
Using the Hench–Gillis correlation, calculate the CPR 1 m above the core midplane in a commercial BWR. Assume
the system pressure is 7.25 MPa, the inlet coolant temperature is 278°C, the core average mass flux is 1,570 kg/m2 s, the
active core height is 3.6 m, and the maximum heat flux is 1,350 kW/m2. If the normal power output is 1,000 MWE, at
what power output does dryout occur?
1074 NUCLEAR HOT CHANNEL FACTORS, CHF, AND DNBR
Exercise 27.9
In a Westinghouse PWR, the MDNBR of 2.2 is reached in the hottest channel 1 m from the top of the core. If the axial
heat flux at that location is 1 × 106 W/m2, what is the CHF? Suppose the mass flux is doubled from 3,500–7,000 kg/m2 s.
According to the W-3 correlation, what happens to the CHF?
Exercise 27.10
The U.S. NRC requires the DNBR for a commercial PWR in the United States to stay above 1.3 everywhere in the core
during a 15% overpower transient. If the CHF predicted by the W-3 and Groeneveld correlations is 2.5 × 106 W/m2, what
is the largest value of the heat flux that can be reached in the core during steady-state operation?
Exercise 27.11
Assume the equilibrium quality at the inlet, the midpoint, and the exit of the hottest fuel assembly in a BWR core
is—5%, 5%, and 15%, respectively. Using the Janssen–Levy correlation, predict the CHF at each of these locations.
Assume the mass flux is 1 × 106 lb/h ft2 and the core power profile is co-sinusoidal. Assume the maximum applied heat
flux is 1,500 kW/m2 at the core midplane. What is the DNBR at this location?
Exercise 27.12
In uniformly heated circular tubes (instead of reactor coolant channels) the Tong correlation can be used to predict the
CHF. The Tong correlation states that
G is the mass flux, and D is the inner diameter of the heated tube (all in SI units). For a tube with a heated diameter of
1 cm at 7 MPa and with a mass flux of 1,000 kg/m2 s, calculate the CHF when the exit quality is 10%.
28
Particle Transport and Entrainment
during Reactor Accidents
28.1
Particle Transport during Reactor Accidents
Particle transport during reactor loss-of-coolant accidents (or LOCAs) occurs as the result of two distinct processes called (1) advection
and (2) diffusion. Advection occurs very rapidly while diffusion occurs more slowly and is generally more limited in scope. In this
chapter, we would like to discuss how these processes effect the dispersion of radioactive particles produced by a hypothetical reactor
accident. We will then discuss how these particles are dispersed if they happen to escape the containment building entirely. This will
lead to a discussion of particle entrainment and Stokes flow as well as the motion of radioactive plumes and clouds. We will then
conclude our discussion with an explanation of the wedge model of atmospheric dispersion.
28.2
Understanding Advection, Diffusion, and Convection
Advection is the dominant form of mass and energy transfer in nuclear power plants, and this is particularly true during reactor acci-
dents and LOCAs. Advection occurs whenever the coolant is in motion. Diffusion does not require a fluid such as air or water to be in
motion. Instead, it requires a concentration gradient to exist between two points in the flow field. All fluids can facilitate the transfer
of mass and energy, but in nuclear engineering, the most important fluids are air and water (or alternatively, water and steam). The
air performs this function inside the containment building, and the water/steam performs this function in the nuclear steam supply
system (NSSS). In general, a substance can be advected if the fluid containing the substance moves—that is, it is capable of bulk
motion. In single-phase flow, this motion can be described by a velocity field. The velocity field has both magnitude and direction.
Hence, velocity fields can the thought of as vector fields. Two examples of these fields are shown in Figure 28.1. The substance or
property that is advected by the velocity field is usually considered to be embedded in the field itself—and this pertains to matter
particles as well as fluid particles. Hence, the velocity field is “the construct” that allows material particles to move through the NSSS,
and in the event of a reactor accident, into the containment building. Sometimes it is helpful to think of advection as a process that
requires the existence of both vector and scalar fields. The velocity field is the vector field in which the advected substance or prop-
erty moves. The magnitude of the property or substance within the field is then described by a scalar field. The combination of these
two fields (scalar plus vector) then defines the transport capabilities of the fluid. Advection is also frequently confused with the more
encompassing process of convection. However, convection is normally used to refer to both advection and diffusion:
Energy transport during diffusion occurs at a molecular level, while energy transport during advection occurs because of the macro-
scopic or bulk motion of the material. Several examples of advective processes are shown in Figure 28.2.
28.3
A Mathematical Description of Advection
Mathematically, advection can be described by what is called the advection equation. The advection equation is simply an expression
of the fact that a scalar quantity ψ flowing with a fluid must be conserved as the fluid moves and is advected. The most general form
of this equation is
1075
1076 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
FIGURE 28.1 Two examples of vector fields where both advection and diffusion can occur. (Pictures provided by
Matlab—see www.mathworks.com.)
FIGURE 28.2 An example of the process of advection involving the mixing of warm air and cold air (left) and an example of
advection and diffusion in an airstream that is moving past a snow flake. (Pictures provided by MIT.)
where ψ is the scalar quantity being advected, and the velocity vector V at each point in the field is given by
V = iv x + jv y + kv z (28.2)
or
V = iu + jv + kw (28.3)
where ∇ is the divergence operator and u = vx, v = v y, and w = vz. Frequently, it is assumed that the flow is incom-
pressible, that is, the velocity field satisfies ∇·V = 0 and V is said to be solenoidal. If this is so, Equation 35.1 can be
rewritten as
∂ψ / ∂ t + V ⋅ ∇ψ = 0 (28.4)
28.4 A Molecular View of Diffusion 1077
V ⋅ ∇ψ = 0 (28.5)
which shows that ψ is constant along a streamline (see Chapter 14). Hence, ∂ψ/∂t = 0 and ψ does not vary with time.
Hence, if we would like to determine the amount of enthalpy H = ρh that is advected between two points in a reactor
coolant channel when the flow is steady, the equation we must solve is
V ⋅ ∇H = 0 (28.6)
or
where u, v, and w are the components of the velocity vector V along the x, y, and z directions. When heat is added to the
fluid, Equations 28.6 and 28.7 can be rewritten as
V ⋅ ∇H = Q (28.8)
and
The complete set of continuity, momentum, and energy equations required in this case is presented in Chapter 15. In
reactor work, one sometimes encounters a situation where solid particles become entrained in the fluid and move with
it. However, the fluid does not have to move for these particles to diffuse through it. Then, the motion of the particles can
be described by what is called a diffusion equation. If the fluid is also moving, then the motion of the solid particles will
be described by a combination of advection and diffusion. We then need to solve what is called the particle transport
equation. The particle transport equation combines the effects of both advection and diffusion, and it is much more
complex than the advection or diffusion equations alone.
28.4
A Molecular View of Diffusion
Perhaps the easiest way to visualize the process of diffusion is to think of it as the random walk of a very large number
of very small particles. These particles can be either atoms or molecules, and they can even be small clumps of dirt or
debris as long as the particles in question do not have an electric charge. (Note: If they happen to possess an electrical
charge, then we must treat them as a plasma, and the equations used to describe them are considerably different because
of electromagnetic effects.) In this process, the diffusing particles are propelled in different directions by their own ther-
mal energy and through random collisions with other molecules in the medium in which they are located. The motion of
these particles is completely random, and if they are suspended in a fluid, then their motion is known as Brownian motion.
Brownian motion was first discovered by Englishman Robert Brown (see Figure 28.3) in 1827 when he observed the
motion of grains of pollen under a microscope in what appeared to be a drop of stagnant water. He could not explain
why the grains of pollen moved when the water was standing still. However, later Einstein and others provided an
elegant mathematical explanation of why this occurred. This molecular view of diffusion is very popular among physi-
cists and mathematicians because classical mechanics can be used to describe the motion of the particles and the rate
at which they diffuse. Figure 28.4 illustrates this process using a concentration gradient. Suppose that, we have a glass
of stagnant water and decide to add some particles to it. If some of the particles are dissolved in the water, the particles
will initially reside in only one part of the glass. Then over a period of time, the particles will randomly move around,
and they will diffuse in different directions. Eventually, over a longer period of time called the diffusion time, they will
eventually become d istributed randomly and uniformly throughout the glass of water. The rate at which this diffusion
occurs depends on the temperature of the water in the glass, and it can be expressed mathematically by what is called
the diffusion coefficient.
1078 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
FIGURE 28.3 Pictures of Robert Brown (left) and Albert Einstein (right) who were responsible for discovering and later
developing the theory of Brownian motion. (Pictures provided by Wikipedia.)
FIGURE 28.4 A figure illustrating the diffusion of particles through a glass of water. The particles are propelled through the
water by random molecular collisions that can be described by a diffusion equation. The rate of diffusion is then proportional to the
concentration gradient dC/dx.
Later Einstein extended Robert Brown’s work by providing an elegant mathematical description of how Brownian
motion works. In Latin, the word “diffuse” means to spread out. However, at a molecular level, diffusion is a statistical
process that requires the concept of the random walk to thoroughly appreciate the underlying physics. In one dimension,
this turns out to be relatively easy to do, and the overall process can be illustrated by playing a simple children’s game
(see the discussion that follows). However, in two or three dimensions, tracking the paths of the individual particles can
become much harder to do because the collisions occur in two or three dimensions. Depending on the properties of the
particles, a number of other factors can also affect how far they scatter or move. For example, a monatomic gas has only
three degrees of freedom (one for each coordinate direction), and so, collisions are almost entirely elastic in this case.
28.6 Derivation of the Particle Transport Equation 1079
FIGURE 28.5 The mean free path λ of a diffusing particle is found by adding the distance the particle travels between successive
collisions and dividing it by the total number of collisions.
However, as soon as the structure of the particle deviates from that of a simple “billiards ball,” the p hysical and
mathematical description of the collisions becomes a lot more complex. For example, diatomic molecules have three
translational degrees of freedom (x, y, and z) and two rotational components to their motion (θ and ψ) which com-
plicates the outcome of the collision. Polyatomic molecules have three translational and three rotational degrees of
freedom, in addition to other modes where their bonds can vibrate back and forth like a spring being compressed
and released as well as side-to-side motions that can occur when the bonds bend. This vibrational motion ordinarily
does not increase the number of degrees of freedom until the temperature of a gas reaches about 400°C. However,
when it does, these effects can also effect the mean free path between successive collisions. Nevertheless, as a
whole, the particle population will still follow a Maxwell–Boltzmann probability distribution. The properties of
this distribution are discussed in our companion book (see Masterson 2017). The molecules in any substance will
normally travel a different distance d between successive collisions, and the average of this distance over many
thousands or millions of collisions is called the mean free path. Normally, a particle will travel a different distance
between each collision, and the mean free path λ represents the average distance, which is independent of the details
of the individual collisions. An example of how this occurs is shown in Figure 28.5. The mean free path λ may then
be found from
λ= ∑d i
i N (28.10)
where N is the total number of collisions and di is the distance the particle travels for collision i. In a liquid or a gas, this
process is repeated millions of times a second because the molecules are moving very quickly and because the number
of collisions N is very large.
28.5
Writing the Particle Transport Equation with Advection and Diffusion
The next thing we would like to do is to combine the effects of advection and diffusion together into a single equation
that is called the particle transport equation. The particle transport equation has many practical applications, and some
of them are specific to the field of nuclear power. Outside of the nuclear business, this equation can be used to predict the
behavior of a bottle of ink that is dumped into a river. It can also be used to tell how fast the smog in Los Angeles will
be blowing on a p articular day. In a reactor core, it can be used to predict the rate at which the enthalpy flows between
two coolant channels. In a containment building, it can be used to predict where radioactive particles released from the
NSSS during a LOCA will go. Hence, computer programs such as GOTHIC (an acronym for the “Generation of Thermal
Hydraulic Information in Reactor Containments”) use some form of this equation to predict where the radioactive par-
ticles discharged from a hypothetical reactor accident will go.
28.6
D erivation of the Particle Transport Equation
The general particle transport equation used by nuclear engineers can be derived from an alternative form of the trans-
port equation that is sometimes called the Smoluchowski equation, in honor of Polish Physicist Martin Smoluchowski (see
1080 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
FIGURE 28.6 Sometimes the particle transport equation is also called the Smoluchowski equation in honor of Polish Physicist
Martin Smoluchowski, who first formulated it around 1900. Among other things, Martin Smoluchowski helped to explain why
the sky is blue (due to atmospheric diffraction), and he gave one of the first credible explanations of the phenomena of Brownian
Motion, which was done independently of Robert Brown and Albert Einstein.
Figure 28.6) who first formulated it around 1900. Among other things, it can be used to model any fluid where a scalar
quantity ψ such as the mass, the energy, or the number of particles is transported through a fluid by a combination of diffu-
sion and advection. In the discussion that follows, we would like to derive the particle transport equation and demonstrate
how it can be used. Most nuclear engineering text books tend to shy away from this subject because diffusion is not con-
sidered to be as important as advection in the core as a whole. However, when the flow through the core becomes highly
turbulent, and the Reynolds number approaches 50,000, turbulent mixing can become very important, and we will obtain
better results when we consider both processes simultaneously rather than just the process of advection alone. The particle
transport equation can be derived from the continuity equation if we are willing to make a couple of minor modifications
to the way we treat the inflows and outflows. In our initial discussion of this equation in Chapter 15, we learned that the
continuity equation applies to the fluid as well as the particles inside of it. Hence, in its most general form, we can write
∂C i ∂ t + ∇ ⋅ J i = Si (28.11)
where i is the particle species that we are trying to count. Here Ji is the total particle flow (i.e., the current) of that spe-
cies, and Si is the rate that particles of concentration Ci are added or deleted from the control volume. Sometimes, Si is
also called the particle source term.
Generally speaking, the source term must be present when we are injecting particles into a fluid. Sometimes, the
particles can also be created or destroyed when they react chemically with other particles in the fluid itself. Hence, the
most general expression for S is
S = So + R (28.12)
where R > 0 means that a chemical reaction is creating some additional particles, and R < 0 means that a chemical reac-
tion is destroying them. For a heat transfer problem, the value of R can be greater than 0 if the fluid has friction and
generates thermal energy as it moves. In most problems, there can be two sources for the particle current J. The first is
the diffusive flux, which is the movement of particles due to diffusion alone. The second is the overall fluid flow, which
is caused by the advective flux. The total convected particle current JCONVECTION (for a stationary system) is then given
by sum of these two terms:
Hence, the particle transport equation for a single species may be written as
It then remains for us to determine appropriate expressions for JDIFFUSION and JADVECTION. The particle current (or “flux”)
due to diffusion alone is given by Fick’s first law:
J DIFFUSION = − D ⋅ ∇C (28.15)
J ADVECTION = VC (28.16)
where V is the velocity vector. The total particle current due to convection (which consists of both diffusion and advec-
tion) is then
Hence, the transport equation is a second-order differential equation that combines the features of both parabolic and a
hyperbolic equations, and V is the average velocity of the particles that are moving through the fluid. In a pressurized
water reactor (PWR), C might be the concentration of soluble boron in the coolant. In a boiling water reactor (BWR),
C might be the concentration of the bubbles in the core. In other words, the particle transport equation is sufficiently
general that any scalar quantity that can be advected or diffused through the fluid can be thought of as a “particle.” The
enthalpy is another example of a scalar variable that can be treated in this way (see Section 28.4).
28.7
Simplifications to the Particle Transport Equation
The particle transport equation is a very difficult equation to solve. Except for a few special cases, it can only be solved
numerically. However, we can make it a lot more tractable if we can make a couple of simplifying assumptions. Suppose
that there are no particle sources or sinks, that the diffusion coefficient is constant, and that the flow is incompressible.
The first assumption means that we can set S = 0, the second assumption means that D is constant, and the third assump-
tion implies that
∇ ⋅ ( VC) = V∇ ⋅ C (28.21)
since ∇·V = 0. (i.e., the fluidics field has a zero divergence). Then the particle transport equation becomes
This is the standard form of the particle transport equation that most containment analysis codes use, where C is again
the particle concentration. Normally, the particle concentration is expressed in mol/cm3 or mol/m3. The flow of particles
from one region of space to another is then expressed in terms of the molar current J which is also called the molar
flux. The molar flux J is the chemical equivalent of the mass flux G which appears in the equations of fluid mechanics.
While the mass flux G = ρv has the units of kg/m2 s, the molar flux has the units of mol/m2 s (or mol/cm2 s), and physi-
cally, it represents the amount of a substance that will flow across a unit area per unit time (e.g., the soluble boron flow
through a pipe in the reactor coolant system). When there is no advection, the dissolved particles travel from regions
1082 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
of high concentration to regions of low concentration, and the flow rate is proportional to the negative concentration
gradient −∇C. Alternatively, when advection occurs, the advective component of the molar flux is
where C is given in mol/m3. Note that the molar flux in both the presence and absence of advection is considered to be
a vector because when there is advection, it moves with the flow, and when there is no advection, its direction is defined
by the concentration gradient. The rate of diffusion is then given by Fick’s first law of particle diffusion:
where D is the diffusion coefficient or the diffusivity having the units of cm2/s or m2/s. Equation 28.23 is known as Fick’s
first law of diffusion. The value of D is proportional to the square of the average velocity of the diffusing particles, and it
also depends on the temperature of the fluid, the viscosity of the fluid, and the particle size. In dilute aqueous solutions,
the diffusion coefficients for most ions have a value of about 1 × 10 −9 m2/s. For biological molecules, the diffusion coef-
ficients are 10 to 100 times smaller. When a fluid is at rest, the advective term V∇·C in the general transport equation
can be ignored, and the transport equation reduces to
This equation is called Fick’s second law of diffusion. Fick’s second law is a special case of the convection–diffusion
equation in which there is no advective flux and no net volumetric source. It is a second-order partial differential equa-
tion whose solutions are well known. In one dimension, it reduces to
∂C/ ∂ t = D ⋅ ∂ 2 C ∂x 2 (28.25)
The driving force for one-dimensional diffusion in this equation is the quantity ∂C/∂x, which for ideal mixtures is the
same as concentration gradient. In chemical systems other than ideal solutions or mixtures, the driving force for the
diffusion of each species is the gradient of the chemical potential 𝜕ϕi/𝜕x for the species. Then, Fick’s first law can
be written as
( )
For an individual species: J i = − DCi RT ∂φi ∂x (28.26)
where the subscript i refers to the ith species in the mixture, C is the concentration of that species (in mol/m3), R is the
universal gas constant (having the units of J/K mol), T is the absolute temperature (in °K), and ϕi is the chemical poten-
tial (in J/mol). If Fick’s law is expressed in terms of the mass fraction (mi, given, e.g., in kg/kg), rather than the chemical
potential, then the correct version to use is
where ρ is now the density of the mixture (in kg/m3). Notice that the density appears outside the gradient operator in
this case.
28.8
Steady-State Solutions
When a reactor is operating normally, engineers are usually interested in only steady-state solutions to the particle
transport equation. Then the equation reduces to
D∇ 2 C − V∇ ⋅ C = S (28.28)
Fortunately, this is a tractable equation to solve if V and D are constants. In one dimension, it simply becomes
d 2 C dx 2 + AdC/dx = S′ (28.29)
where A = −u/D, S′ = S/D, and u is the fluid velocity. When there is no diffusion, Equation 28.28 reduces to
V∇ ⋅ C = −S (28.30)
udC/dx = −S (28.31)
which then requires a specific knowledge of u to find the concentration gradient dC/dx. The molar current, from
Equation 28.23, is then
28.9
A nalogies to the Navier–Stokes Equations
Particle transport equations like the Smoluchowski equation have a great deal in common with the Navier–Stokes equa-
tions which we derived in Chapter 15. In fact, the flow of momentum in the Navier–Stokes equations is mathematically
similar to the flow of particles in the transport equation because it can be both advected and diffused. It is easiest to see
this analogy if we write the momentum equation for an incompressible Newtonian fluid. In this case, the momentum
equation becomes
where M is the momentum of the fluid (per unit volume) at each point (equal to the density ρ multiplied by the
velocity V), μ is viscosity, P is fluid pressure, and F is any body force such as gravity which is exerted on the fluid. In this
equation, the term on the left-hand side describes the change in momentum at a given point; the first term on the right-
hand side describes the viscosity, which is really the diffusion of momentum; the second term on the right describes the
advective flow of momentum; and the last two terms on the right describes the external and internal forces that can act
as sources or sinks of momentum. Later, we will find that the first term on the right-hand side can be formulated in terms
of a conventional fiction factor f. The momentum equation then becomes
where
V = iu + jv + kw and M = ρV (28.35)
where F = ρg. Here, u (du/dx) is the material derivative, which we introduced to the reader in Chapter 14.
1084 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
28.10
Particle Transport with Turbulent Mixing
In the previous section, we derived an expression for the time-dependent particle concentration by counting the particles
in an arbitrarily shaped control volume V. We found that the scalar transport equation, which represented the transfer of
mass due to molecular diffusion and advective particle transport, could be written as
( )
∂C/ ∂ t = u ∂C/ ∂x + v ∂C/ ∂ y + w ∂C/ ∂z + D ∂ 2 C ∂x 2 + ∂ 2 C ∂ y 2 + ∂ 2 C ∂z 2 (28.37)
where u, v, and w were the components of the fluid velocity V in the x, y, and z directions. This equation is known as
the advective–diffusion equation or the Smoluchowski equation, and it has many similarities to the heat conduction
equation that we derived in Chapter 11. The diffusion coefficient D depends on the physical properties of the fluid or
the substance that is dissolved in it (the solute). For low concentrations of the solute (i.e., a dilute fluid), the value of D
is constant, and its values can be found in many reference books (see, e.g., the Handbook of Chemistry and Physics).
Normally, the effects of molecular diffusion can be neglected in a reactor core because the diffusion coefficient is so
small (about 1 × 10 −4 m 2/s). However, they CANNOT be neglected in a reactor containment building during or after a
LOCA. To get a better appreciation for why this is true, suppose that we would like to calculate the distance L that a
substance diffuses in time t if its diffusion coefficient is D. If the fluid is not moving, a simple dimensional analysis
would lead us to conclude that
L ∼ √( Dt ) (28.38)
t ~ L2 D (28.39)
So if we know the values of D and L, we can determine how much time it takes for the molecules to u niformly diffuse
across a distance L. Ordinarily, this time can be on the order of hours or days. Example 28.2 shows exactly how long it
takes the molar flux to do the same.
Obviously, molecular diffusion is a very slow process, and when it comes to the core, it does not occur rapidly enough to
be of major importance. However, because a reactor core can generate a great deal of heat, the flow rate through the core
must be highly turbulent to remove the heat as efficiently as possible. Therefore, as soon as the flow becomes turbulent,
a related process called turbulent diffusion or turbulent mixing comes into play. This process takes place much more
rapidly than molecular diffusion, and it can sometimes affect the rate of mass and energy transfer in a dramatic way. In
essence, turbulent diffusion can be thought of as the transfer of mass, energy, or momentum due to random temporal
fluctuations in the velocity field. In this case, the turbulent eddies are the “particles” that we would like to observe. To
see how this affects the transport process, suppose that we would like to apply the advective–diffusion equation to a tur-
bulent flow. Turbulent eddies are produced because of the shear forces that particles of fluid exert on each other. These
eddies can carry mass, energy, or momentum along with them (see Figure 28.7), and in general, they transport energy
and momentum from regions of high temperature to regions of low temperature. In other words, they help to flatten the
core temperature profile by transferring thermal energy from hotter channels to cooler ones.
28.11
T he Particle Transport Equation with Turbulence
As the degree of turbulence increases, the number of turbulent eddies increases as well. We can derive an expression
for the size of this effect by taking the advection–diffusion equation and decomposing the velocity and the particle
concentrations into the sum of their average values and their temporally fluctuating parts. We can then time-average the
28.11 The Particle Transport Equation with Turbulence 1085
FIGURE 28.7 Turbulent eddies of many different sizes are exchanged during the turbulent mixing process. (Pictures provided by
Yale University.)
results. So for the solute, the particle concentration at any point in time is the time-averaged concentration c , plus the
instantaneous fluctuation c′, or deviation from the mean:
C = c + c ′ (28.40)
Similarly, the velocity V along each coordinate direction must have an average value V as well as a temporally fluctu-
ating component V′. We can express this by writing
u = u + u ′ (28.41)
v = v + v ′ (28.42)
w = w + w ′ (28.43)
Now let us consider a one-dimensional problem, where the mathematics is much simpler. We can substitute the expres-
sion for u into the particle transport equation, as well as the expression for the time-dependent concentration C, and
average the results over time. When we do so, we find that
∂ c ∂ t + u ∂ c ∂ x = D ∂ 2 c ∂ x 2 − ∂ u ′c ′ ∂ x (28.44)
Just as before, the time-averaged transport equation is identical to the instantaneous transport equation, except that a
new term ∂ u ′c ′ ∂x has been added to the right hand side. This term represents the transport of substance c due to tur-
bulent fluctuations in the velocity field. We can then write the right-hand side of the transport equation a little differently
if we group the terms together. For the x direction, the result is
∂ / ∂x D ∂ c ∂x − u ′c ′ (28.45)
Thus, both terms are responsible for the transfer of mass. The first term is the mass transfer rate due to molecular diffu-
sion alone, and the second term is the mass transfer rate due to the correlation between u′ and c′. Because D is usually
a very small number, u ′c ′ D ∂ c ∂x, and molecular transport term is usually neglected compared to the turbulent
mixing term when the flow is turbulent and u ′c ′ is large. However, if the flow becomes laminar, then u′ = 0, and
the molecular transport term dominates the turbulent mixing term. At this point, it is sometimes appropriate to drop
the brackets from c and u , v , and w to simplify the notation. When we do this, the particle transport equation
becomes
( )
∂c/ ∂ t + u ∂c/ ∂x + v ∂c/ ∂ y + w ∂c/ ∂z = − ∂ u ′c ′ ∂x + ∂ v ′c ′ ∂ y + ∂ w ′c ′ ∂z (28.46)
1086 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
Unfortunately, there is no way to solve this equation analytically because we do not know the values of u ′c ′ , v ′c ′ , and
w ′c ′ in advance. However, if we can measure these values in an experiment or replace them with something else that
we can measure, then we will be able to predict how the value of c behaves. In a reactor coolant channel, the standard
“trick” for doing this is to replace u ′c ′ , v ′c ′ , and w ′c ′ by the following expressions
u ′c ′ = −ε x ∂c/ ∂x (28.47)
v ′c ′ = −ε y ∂c/ ∂ y (28.48)
w ′c ′ = −ε z ∂c/ ∂z (28.49)
where the constants of proportionality εx, εy, and εz are called the eddy diffusivity coefficients. In other words, each of
these expressions assumes that the turbulent mixing rate is proportional to the average concentration gradient. While
this model can be effective in some cases, it requires us to measure these values in the geometry we want to simulate.
This means that we have to rely on experiments to determine the values of εx, εy, and εz. Moreover, the values of εx, εy,
and εz can be strongly flow dependent, and they can even vary within the flow field itself. However, if we are willing to
make this assumption and the turbulent mass transfer is much greater than the molecular mass transfer (ε ≫ D), then we
can write
(( ) ( )
∂c/ ∂ t + u ∂c/ ∂x + v ∂c/ ∂ y + w ∂c/ ∂z = ∂ ε x ∂c ∂x ∂x + ∂ ε y ∂c ∂ y ∂ y + ∂ ε z ∂c ∂z ∂z ( ) ) (28.50)
Then when the eddy diffusivity does not change spatially, we can rewrite the particle transport equation as
( ) ( ) (
∂c/ ∂ t + u ∂c/ ∂x + v ∂c/ ∂ y + w ∂c/ ∂z = ε x ∂ 2 c ∂x 2 + ε y ∂ 2 c ∂ y 2 + ε z ∂ 2 c ∂z 2 ) (28.51)
This equation is the starting point for most calculations of the particle concentration in a turbulent flow field. We can
also add a source term S to the equation if new particles are injected into the flow field. Then the particle transport
equation becomes
( ) ( ) ( )
∂c/ ∂ t + u ∂c/ ∂x + v ∂c/ ∂ y + w ∂c/ ∂z = ε x ∂ 2 c ∂x 2 + ε y ∂ 2 c ∂ y 2 + ε z ∂ 2 c ∂z 2 + S (28.52)
This equation is usually solved numerically by computer programs such as GOTHIC and FLUENT. However, if the
terms on the right-hand side are replaced by empirical coefficients (i.e., mixing coefficients) that are essentially constants
for a particular fuel assembly geometry (hexagonal or square), then this becomes the form of the transport equation
that is widely used in a lumped parameter approach. Under these conditions, the values of εx(∂ 2c/∂x2), εy(∂ 2c/∂y2), and
εz(∂ 2c/∂z2) are assumed to be spatially averaged over each computational cell. In practice, this is normally how the rate
of enthalpy transport is determined between a hot sub-channel and a cooler one. To perform such a calculation, one
usually measures the value of the turbulent mixing coefficient and then recasts the right-hand side of the particle trans-
port equation (Equation 28.52) to use the experimentally determined values. The references at the end of the chapter
describe how this can be done. Normally, turbulent mixing reduces the temperature of the coolant in the hottest channels
and increases the temperature of the coolant in the coldest channels. In other words, turbulent mixing allows the fluid
temperature profile to become much more uniform in the radial or transverse direction. In PWRs, this type of turbulent
mixing can reduce the temperature differences between adjacent coolant channels by as much as 5% to 10%. While
this does not seem like a particularly large number, it means that the onset of nucleate boiling can be delayed by the
same amount for the same power output from the core. Turbulent mixing can also be used to reduce the temperature
differences between adjacent fuel assemblies, although the overall difference is not as great as it is between adjacent
sub-channels. This occurs because the distance the turbulent eddies travel falls off dramatically as the pitch P between
the coolant channels increases. Normally, the effects of turbulent mixing do not propagate between adjacent coolant
channels when they are more than two rows apart. Figure 28.8 shows the effect that turbulent mixing has on the overall
temperature profiles. In BWRs, the situation becomes more complicated because the fuel assemblies are “sheathed” or
“canned” and the coolant within them is allowed to boil. Hence, the degree of turbulent mixing is greater than what
would be if no boiling were to occur.
In this case, the eddy diffusivity for the two-phase mixture εTP can be found by multiplying the eddy diffusivity for
the single-phase mixture εSP by a two-phase correction factor or multiplier M that is a function of the Reynolds number
and the local void fraction or quality:
ε TP = ε SP ⋅ M(x,Re) (28.53)
28.13 Stokes Law and Stokes Flow 1087
FIGURE 28.8 Turbulent mixing reduces the temperature differences between adjacent coolant channels and creates a more
uniform radial temperature profile.
or
ε TP = ε SP ⋅ M(α,Re) (28.54)
Normally, the value of M is determined by performing an experiment over a targeted range of conditions. The values for
M can range from 1.5 to about 5. Thus, the amount of turbulent mixing is always greater for two-phase mixtures than
or mass flux G = m/A.
it is for single-phase flow for the same mass flow rate m Unfortunately, in BWRs, the amount of
turbulent mixing is also a function of axial elevation because of the different flow regimes as one moves up through the
core. Hence, the two-phase correction factor M can also have a strong dependence on z.
28.12
Particle Transport with Drag
In addition to diffusion and advection, which apply to relatively small particles, larger radioactive particles can also be
transported by a moving fluid. The fluid velocity in this case is determined by the fluid pressure gradient ∇P. The fluid
momentum equation which determines the amount of particle transport in the presence of frictional drag is
ρu ∂ u/ ∂x = µ ∂ 2 u ∂x 2 − ρg − ∂P/ ∂x (28.55)
where the terms on the left-hand side are the advective terms, and the first and second terms on the right-hand side rep-
resent the effects of friction and gravity, respectively. When the velocity is V = iu + jv + kw, a drag force will be exerted
on the particles in the fluid. This drag force is proportional to the square of the fluid velocity and may be represented by
the simple expression
where CD is a dimensionless quantity known as the drag coefficient, and A is the cross-sectional area of the particle.
Normally, the drag force predicted by Equation 28.56 is expressed in Newton.
28.13
Stokes Law and Stokes Flow
For low Reynolds numbers and predominantly horizontal flows, Equation 28.55 reduces to
∂P/ ∂x = µ ∂ 2 u ∂x 2 (28.57)
1088 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
∇P ≈ µ∇ 2 V (28.58)
Sometimes the flow of fluid under these conditions is called Stokes flow or creeping flow. When the Reynolds number
is low (Re ≪ 1), the drag force on a three-dimensional object moving at speed V with a characteristic dimension L can
be written as
If the particle moving in the fluid is spherical in shape, then the constant in the drag equation (Equation 28.59) is equal
to 3π when we interpret L as the diameter D of the sphere. Then the drag equation for the particle becomes
This equation is called Stokes law, and it forms the basis for radioactive particle motion in a viscous fluid when the
Reynolds number is low. Notice that when the velocity of the particle becomes the same as the velocity of the fluid, then
V = 0 and there is no longer any net force on the particle. It simply moves along with the fluid at its local speed. When
we use the more conventional equation for the drag force
FDRAG = CD 1 2 ρV 2 A (28.61)
CD = 24/Re (28.62)
where Re is the Reynolds number of the particle, which for the case of a sphere is
Re = ρVD/µ (28.63)
Equation 28.60 is called Stokes Law, and it shows that at very low Reynolds numbers, the drag force acting on a s pherical
particle is proportional to its diameter, its velocity, and the viscosity of the surrounding fluid. Stokes law is often used to
predict the motion of dust particles in air as well as solid particles suspended in a pool of water.
28.14
Particle Motion Induced by Buoyancy Forces and Gravity
In addition to drag forces that are exerted on a particle by a fluid, a buoyancy force may also be exerted on the same
particle. This force is proportional to the amount of fluid it displaces. This buoyancy force is described by Archimedes
Principle, which states that the buoyant force acting on a particle is given by
where m is the mass of the particle, ρ′ is the density of the particle, (m/ρ′) is the volume of the particle, and (m/ρ′)·ρ is
the mass of the fluid displaced by the particle. The total force acting on the particle is then
or
( )
mdV /dt = − CD 1 2 ρV 2 A − m ρ′ ρg (28.66)
28.16 Aerosol Behavior 1089
dV dt = ( ρ′ − ρ) g ρ′ − CDρV 2 A m (28.67)
where we have reversed the signs on the right-hand side to indicate that the drag force and gravitational force will tend to
drag the particle downward if we define the positive velocity to be pointing upward. This last equation (Equation 28.67)
can then be used to find the motion of a particle entrained in a fast moving fluid. Among other things, it can be used to
predict the terminal velocity of the particle as well as its gravitational settling time.
28.15
E quations of Motion for Particles Suspended in a Viscous Fluid
Equation 28.67 is the equation of motion for a solid particle moving through a fluid such as water or air. It states that the
acceleration on the particle is given by
when the particle is moving at velocity V with respect to the fluid as a whole. If the particle also happens to be spherical
in shape, then the drag coefficient is given by
CD = 24/Re (28.69)
where Re is the Reynolds number of the particle. If no external forces are applied to the particle, then the particle will
fall through the surrounding fluid under the force of gravity alone. The fluid will exert a drag force on the particle, and
the particle will continue to decelerate until it reaches a terminal velocity which is the maximum velocity it can attain
under these conditions. Sometimes this velocity is also called the settling velocity. During a severe reactor accident or
LOCA, a great deal of particulate matter can be released into the containment building, and after some period of time,
this particulate matter will also obey Equation 28.68. The settling velocity of the particles in the air can be found by
setting dV/dt = 0 and solving Equation 28.68 for the value of V. The result in this case is
(
VSETTLING = √ 2mg (ρ′ − ρ) ( Aρρ′CD ) ) (28.70)
If the particles happen to be spherical in shape, m = ρπD3/6, and A = πD2/4. The equation for the settling velocity is then
This equation also applies to larger particles that are suspended in the reactor coolant. They also settle to the bottom of
the containment building floor, but the values of ρ, D, and CD which determine the settling velocity are different than
they are for lighter particles.
28.16
Aerosol Behavior
In addition to reactor accidents, the equations we have just presented can also be used to predict the behavior of aerosols.
An aerosol is defined as a suspension of fine solid particles or liquid droplets in a gas. Common examples of aerosols
include the clouds in the sky, the smog in the city of Los Angeles (see Figure 28.9), and the smoke in a room when some-
one is smoking a cigarette. Aerosols are also emitted from coal-fired power plants as they generate electricity. In nuclear
1090 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
FIGURE 28.9 The smog and the dust in the air of Los Angeles are two examples of common aerosols that are easy to see.
science and engineering, aerosols are particularly important because they provide a way to measure the effects of radio-
active contamination and dispersion. There are many other examples of aerosols, but at the end of the day, they are small
solid or liquid particles entrained within the molecules of a gas. How far the particles travel is determined by the size of
the particles, the mass of the particles, and their overall shape.
The most important parameter in determining how many particles can reach a particular location is the number of par-
ticles that are suspended in a cubic centimeter or cubic meter of the atmosphere. In aerosol calculations, this is referred
to as the particle concentration of the aerosol. The size of the particles and their distribution of sizes can also affect
how the aerosol behaves. The diameter of the particles (or their radius) determines the ratio of the viscous force to the
initial force that each of these particles possesses. If the particles have the same size, the aerosol is known as a mono-
dispersal aerosol. If the particles have a range of different sizes, the aerosol is known as a poly-dispersal aerosol. While
liquid droplets are nearly always spherical in shape, solid particles can have a variety of different shapes. To model the
behavior of irregularly shaped solid particles, the concept of the equivalent diameter is used. The equivalent diameter
DE is defined as the diameter of a spherical particle that has the same cross-sectional area as the area of an irregular
particle. Sometimes the equivalent diameter is also known as the aerodynamic diameter. For a uniform aerosol, a single
number—the particle diameter D—describes the size of all of the particles. However, for a poly-dispersal aerosol, the
size of the aerosol must be described by a particle size probability distribution. This probability distribution is repre-
sented by P(D), where D is the particle diameter. Normally, the probability distribution is constructed so that the prob-
ability of a particle having some diameter between the minimum and maximum diameter is 100%. In other words, the
probability distribution is normalized so that
∫ P(D) dD = 1.0 (28.72)
where the lower limit of integration is DMIN and the upper limit of integration is DMAX. A normal probability distribution
is usually not suitable for modeling the behavior of most aerosols because they always contain a number of relatively
large particles. Because of this, most aerosols have skewed probability distributions with a long tail of larger sized par-
ticles. In radiological dispersion, most aerosols can be described by a log-normal size probability distribution, where the
probability of a particle having a particular diameter between D and D + dD is
P = (1/σ)√ 2πe − ( D − D )
2
2σ2
(28.73)
where
☉☉ σ = the standard deviation of the size distribution
☉☉ D = the average or mean diameter
In containment analysis, different classes of particles can be entrained within the atmospheric fluidics field. Each
class of particles can then have a different probability size distribution (P iD), where the subscript i refers to the
particle class.
28.17 Forces on the Particles in an Aerosol 1091
28.17
Forces on the Particles in an Aerosol
How far an individual particle travels after it enters an atmospheric field depends on the forces that act on the par-
ticle. In general, there will be a drag force that the molecules of the atmosphere will exert on the particle. Finally,
the force of gravity will tend to pull the particle out of the atmospheric field if its velocity is low enough. This will
tend to drag the particle down to the containment building floor. A simple way to understand the interplay between
these forces is to calculate the magnitude of these forces if the particle is spherical in shape. The drag force on the
particle is given by
FDRAG = CD A ρV 2 2 (28.74)
where A is the projected area of the particle normal to the direction of the flow, V is the velocity of the fluid, ρ is the
density of the atmospheric fluid, and CD is an empirically determined coefficient known as the drag coefficient. For
spherical particles with a diameter D, the drag force is then
Stokes Law
FDRAG = 6 πµRV (28.76)
where R is the radius of the sphere. In a containment building, some of the aerosol particles ejected from the core are
so small (about one-tenth of a micron) that a molecular correction factor has to be added to the drag coefficient. The
Knudsen number Kn can be used to determine when this correction factor becomes important. The Knudsen number
is defined by
where λ is the mean free path of the particles in the fluid and D is the particle’s diameter. The mean free path is the aver-
age distance between molecular collisions in this case. The mean free path of an air molecule in a reactor containment
building is about 6.5 × 10 −6 cm (0.065 microns). If the particle diameter D is much greater than λ, the Knudsen number
will be very small, and the gas appears to be a continuous fluid. However, if the particle diameter D is much less than
λ, the Knudsen number becomes very large, and the particle acts just like another gas molecule. If the particle size is
between these two values, the gas has several characteristics of each. Sometimes this intermediate region is called the
transition region. This means that particles having a diameter greater than about 4 × 10 −5 cm will have normal values
for the drag coefficient, while particles having a diameter less than about 2 × 10 −6 cm must have corrected values for the
drag coefficient. In between these two extremes, a combination of drag coefficients must be used.
1092 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
28.18
Finding the Terminal Settling Velocity
Sometimes we would like to know how fast a particle will settle to the containment building floor. To do this, we have
to determine the particle’s trajectory using Newton’s laws of motion. However, for the special case where the air has
become still and the particle is at rest at t = 0, the vertical velocity of the particle is given by
( )
V(t) = τg 1 − e − t / τ (28.79)
where τ = C′DρPD2/18μ, and ρP is the density of the particle. Notice that for t ≫ τ, the particle attains an asymptotic
v elocity called the terminal settling velocity, which is given by
Thus, τ can be thought of as the characteristic time constant for the particle as it approaches its terminal velocity. If the
particle enters a moving airstream, then it will also approach the velocity of the stream with the same characteristic time
constant τ. Not surprisingly, this same equation also applies to the motion of skydivers (see Figure 28.10). However,
for skydivers, the terminal velocity is about 195 km/h (122 mph) in a belly-to-earth position (i.e., face down) and about
320 km/h (200 mph) if the arms are pulled in. When the same equations are applied to the hot air in the containment
building following a reactor accident, we find that it takes between 3 seconds and 10,000 seconds (166 min or 2.77 hours)
for the water droplets to fall 1 cm. The finer particles also settle out more slowly than the larger particles do. However,
this does not account for the fact that the air in the containment building may be moving, and when it does, an additional
calculation must be performed. If the particles ejected by the LOCA have a distribution of sizes, the particles used to
represent a particular particle class will follow the same trajectory as the “average” particle within that class, but some
particles will fall more slowly and some particles will fall more rapidly than others. Under these conditions, a probability
distribution P(D) is normally required to determine where all of the particles will fall. If the particles are spherical in
FIGURE 28.10 The figure on the left shows the forces that are exerted on a Stokes sphere to calculate the terminal settling veloc-
ity. The skydiving team on the right has a terminal velocity that can be found by applying the identical force balance. However,
because the skydivers are heavier, their corresponding terminal velocities are much higher (about 200 km/h). (Skydiver picture
provided by Wikipedia.)
28.19 Particle Dispersion Outside of the Containment Building 1093
FIGURE 28.11 The paths of radioactive particles released from a LOCA can sometimes be modeled by assuming the particles
are entrained in a jet of high pressure water and steam similar to that emanating from a high pressure hose. The water within the
jet spreads out quickly and loses its parabolic definition. The heaviest particles fall out of the jet first and the higher particles travel
much further. The distribution of these trajectories can be modeled using a probability distribution function.
shape and the Reynolds numbers are low enough (Re < 0.1), then Stokes law can be used to replace P(m P) with P(D). A
probability distribution can then be generated for τ(D):
and the equations of motion can be solved with this distribution function to find the particle trajectories. The results of
such a calculation are shown in Figure 28.11. Notice that heavier particles fall to the floor first, and they follow more
defined trajectories (with less statistical scatter) than the lighter particles do. Now let us turn our attention to how the
particles behave after they escape from the containment building.
28.19
Particle Dispersion Outside of the Containment Building
If some particulate matter escapes the containment building, the particles may become entrained in the surrounding air,
and they may flow with the air as they fall to the ground. However, if they are released at a higher temperature than the
surrounding air, they may cause the air in which they are entrained to rise because it expands. The particulate matter will
then create what is called a radioactive plume. The distance that a plume rises above the ground is determined by the tem-
perature difference between the executants and the surrounding environment. The plume will continue to rise a distance
ΔH until it mixes with the surrounding atmosphere and achieves thermal equilibrium with it. Then the particles inside
the plume settle out of the air with a settling velocity VSETTLING, which is given by Equation 28.80. Larger and heavier
particles will settle out of the plume first (because VSETTLING is higher), and then smaller and lighter particles settle out
next because VSETTLING is lower. The time they take to reach the ground is then given by
FIGURE 28.12 Plume types commonly emitted from power plants into the Earth’s atmosphere.
Hence, heavier particles tend to settle close to the containment building, and lighter particles tend to fall fur-
ther away. The shape of the plume also depends on the temperature of the atmosphere at the time of the accident.
Normally, the air temperature falls with increasing altitude, but this does not always have to be the case. Sometimes
temperature inversions occur where the cold air stays closer to the ground and the hot air stays further above it.
Finally, in some weather patterns, the atmospheric temperature profiles may change in more complex ways where
different inversion layers can have different temperatures. Each of these temperature profiles will have an important
effect on the shape of the plume after it leaves the containment building. Some of the most common types of plumes
are shown in Figure 28.12.
28.20
Radioactive Plumes and Their Shapes
The particles in a radioactive plume will also separate from one another as the result of local atmospheric turbulence.
This turbulence gives rise to what is called turbulent diffusion and turbulent mixing. We discussed both of these pro-
cesses in Section 28.10. When the air around the containment building is stationary, the particle concentration C obeys
the usual diffusion equation
( )
∂C/ ∂ t = D ∂2 C ∂x 2 + ∂2 C ∂ y 2 + ∂2 C ∂z 2 (28.83)
where D is the diffusion coefficient. However, if the wind is blowing with an average velocity v in the x direction, the
diffusion equation must be modified to account for the fact that the air in which the diffusion is taking place is in motion.
The diffusion equation in this case becomes
( )
∂C/ ∂ t + u ⋅ ∂C/ ∂x = D ∂2 C ∂x 2 + ∂2 C ∂ y 2 + ∂2 C ∂z 2 (28.84)
28.21 Modeling the Motion of Plumes Close to the Ground 1095
Hence, the concentration gradient ∂C/∂x is advected with an average velocity u in the x direction. Now let us attempt
to solve this equation to see what it implies. Normally, the rate of particle transport due to advection in the x direction
is many times greater than the particle transport due to diffusion. This implies that u·∂C/∂x ≫ D·∂ 2C/∂x2, and because
of this, we can neglect the term D·∂ 2C/∂x2 in the x direction. The steady-state diffusion equation we would like to
solve is then
(
u ⋅ ∂C/ ∂x = D ∂2 C ∂ y 2 + ∂2 C ∂z 2 ) (28.85)
Now suppose that the particles are emitted from the containment building at a constant rate of S particles per second,
where S can be thought of as a particle source term. The solution to the diffusion equation which satisfies all of the
required boundary conditions is
In other words, the particles in the plume which move in the x direction spread out in a Gaussian probability distribution
in the y and z directions. The standard deviation of this distribution is
1
σ = (2xD/u) 2 (28.87)
which means that the equation for the concentrations in the y and z directions may also be written as
( )
( (
C(x, y,z) = S 2 πuσ 2 e )) − y 2 + z 2
2 σ 2
(28.88)
This equation assumes the diffusion coefficients are the same in the y and z directions. The σ′s are then referred to as
the horizontal and vertical dispersion coefficients. Normally, these coefficients are determined from Equation 28.87
once the value of the atmospheric diffusion coefficient D is known. In some cases, the diffusion coefficient in the y or
horizontal direction may be different than the diffusion coefficient in the z or vertical direction. Then the equation for
the spatial concentration becomes
( ) ( ) (28.89)
( )
– y 2 2 σ y 2 + z 2 2 σ z 2
C(x, y,z) = S 2 πuσ y σ z e
28.21
Modeling the Motion of Plumes Close to the Ground
To derive Equation 28.89, we assumed that the particles released from the containment building were emitted into an
infinitely large atmosphere far away from the ground. In reality, they are generally released at some altitude h above the
ground, which then becomes the origin of a ground-based coordinate system. If we define z to be the vertical coordinate,
then the revised solution to the diffusion equation becomes
− y2 ( 2σ y2 ) + (z + h )2 ( 2σz 2 ) − y 2 ( 2 σ y 2 ) + ( z − h )2 ( 2 σ z 2 )
(
C(x, y,z) = S 2 πuσ y σ z e ) + e
(28.90)
This equation implies that the concentration of particles at ground level, where z = 0, must be given by
( ) ( ) (28.91)
( )
− y 2 2 σ y 2 + h 2 2 σ z 2
C(x, y,z) = S 2 πuσ y σ z e
Furthermore, since the centerline of the plume is located at y = 0, the largest concentration of the particles will occur
directly along the centerline, and at this location, the concentration is
( ) (28.92)
(
C(x, y,z) = S 2 πuσ y σ z e ) − h2 2σ z 2
1096 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
28.22
Time-Dependent Particle Dispersion
If the rate of particle emission S becomes time dependent so that S = S(t), the concentration of dispersed particles can be
found by integrating Equation 28.92 over the period of time where additional particles are being added to the plume. If
∫
N = S(t) dt (28.93)
is the total number of particles released, then the total concentration of particles that reaches the ground is
( (
C = N πuσ 2 exp − h 2 2σ 2 )) (28.94)
28.23
T he Wedge Model of Atmospheric Dispersion
Atmospheric dispersion models based on particle diffusion theory work well for distances up to about 10 km (10,000 m)
from the location where the radioactive particles first enter the atmosphere. However, for larger distances, diffusion
theory begins to break down and the diffusion coefficient D can no longer be tweaked enough to account for unexpected
fluctuations in the air flow. Moreover, changes in atmospheric conditions will begin to disperse the particles in unpre-
dictable ways. When this occurs, another competing model of atmospheric dispersion called the wedge model can be
used to determine where the particles will go. The wedge model assumes that the particle concentration C can be treated
as being uniform within a wedge of angle θ and height h as shown in Figure 28.13. If the particles in the wedge do not
decay while they are airborne and they stay in the wedge rather than falling to the ground, then the number of particles
between points r and r + dr within the wedge is
C(r)dV = C(r)rhθdr (28.95)
where C(r) is the number of particles at point r and dV = rhθdr. If the particles are emitted from the reactor at a rate of S
particles per second, then the number of particles emitted into the wedge during a time interval dt is N = Sdt. Equating
the total number of particles injected into the wedge to C(r)dV, we find that
C(r)rhθdr = Sdt (28.96)
Solving for C(r) gives
C(r) = Sdt/rhθdr (28.97)
However, dr/dt is also equal to the average wind speed u. This allows us to write
The Wedge Model for the Particle Concentration as a Function of Distance and Velocity
C(r) = (1/r) ⋅ S/uhθ (28.98)
In other words, Equation 28.98 states that the particle concentration falls as (1/r) and that it is also inversely proportional
to the dispersion angle θ and the wind speed u. If the particles also decay and the radioactive decay constant for the
particles is λ, then the correct form of Equation 28.98 to use is
An equation similar to this was used to predict the radioactive fallout from the reactor accident at Fukushima Japan in
2011. Normally, Equation 28.99 only applies to one radioisotope that is released from the plant. If N radioisotopes are
released from the plant and the release rate for radioisotope N is SN, then the total time-dependent concentration for all
N radioisotopes is
In a severe reactor accident, the number of radioisotopes that may have to be tracked by this equation can range between
20 and 30. A picture showing the radioactive contamination released by the Chernobyl reactor accident is shown in
Figure 28.14. In this case, the cloud covered large regions of the former Soviet Union, Eastern and Western Europe, and
Scandinavia. The radioactive materials carried by the cloud took several weeks to completely disperse, and before they
did so, they affected millions of people.
FIGURE 28.14 A map of the radiation released from the Chernobyl nuclear accident in 1986. The distribution of radiation released
was modeled using some of the techniques discussed earlier in the chapter.
1098 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
28.24
Release Rates for Various Types of Reactor Accidents
Not all reactor accidents release radiation to the environment in exactly the same way, and in most cases, any radioactive
materials that are released are assumed to remain within the friendly confines of the containment building. When the
WASH-1400 report was first released in 1975 (see Chapters 30 and 33), it was decided to classify reactor accidents into
nine categories of increasing severity with category 1 accidents being the least severe and category 9 accidents being the
most (see Table 33.5 in Chapter 33 to see the actual accident descriptions). Category 1 accidents are considered trivial
and involve releases of radioactivity that are not substantially different from those accompanying normal plant operation.
Accident categories 2 through 7 entail the release of increasingly significant amounts of radiation into the environment.
Category 8 accidents involve the Design Basis Accidents (DBAs) that are analyzed in the preliminary safety analysis
report (PSAR) and final safety analysis report (FSAR). Category 9 accidents are assumed to be so rare or extreme that
they can be ignored when analyzing their safety and environmental impact. Such accidents involve either an incredibly
unlikely sequence of events or a single improbable event such as a meteor strike or an atomic attack. An example of a
category 9 accident is the simultaneous failure of all on-site and off-site power at the exact moment that a LOCA occurs.
Another example of a category 9 accident is the sudden and catastrophic rupture of the entire reactor pressure vessel.
Hence, accidents of this type are not specifically defined, and they do not appear in Table 33.5, per se. Generally speak-
ing, only category 8 and 9 accidents are assumed to release significant amounts of radiation into the environment.
Table 28.1 shows the duration of the radiation release from each category of hypothetical accident, the elevation of
the radiation release, and the fraction of the core inventory that is assumed to be released. Normally, a 1000 MWE PWR
or BWR core will have a total activity inventory of approximately 8 × 109 Curies half an hour after it is shutdown. This
number ultimately declines as the radioactive fission products begin to decay. Notice that only category 8 and 9 accidents
release a significant percentage of the total core inventory outside of the pressure vessel. Now let us examine the doses
that are typically received following BWR and PWR LOCAs, which are normally considered to be category 8 DBAs.
28.25
Radiological Doses from Typical LOCAs
The off-site dose rates received from a LOCA are different for PWRs than they are for BWRs. In Sections 28.26 and
28.27, we would like to briefly describe the specific differences between them. They are based on an additional federal
guideline called Regulatory Guide 1.4, which was created by the U.S. Nuclear Regulatory Commission to estimate the
dose rates that might be received by the general public. Regulatory Guide 1.4 is also referenced in 10 Compendium of
Federal Regulations 50, and the exact name for it is:
A free copy of it can be obtained from the NRC at the following URL:
www.pbadupws.nrc.gov/docs/ML0037/ML003739614.pdf
This important guideline specifies the regulatory assumptions that are used to estimate the amount of off-site
radioactivity that is released during a design basis LOCA. Regulatory Guide 1.4 is intended primarily for PWRs.
There is a separate regulatory document that also pertains to BWRs.
28.26
Dose Rates for PWR LOCAs
The amount of fission product gas produced in a PWR fuel rod can be estimated using a diffusion equation that accounts
for the flow of gas from the surface of the UO2 as a function of the fuel pin temperature. Normally, the flow of gas from
the fuel increases dramatically above 2,150°C, and it also increases almost linearly with burnup. The fission product
inventory at the end of a fuel cycle is shown in Table 28.2. Regulatory Guide 1.4 assumes that 100% of the Noble gases
in the fuel are immediately released into the containment building, and that 25% of the radioactive iodine inventory is
released immediately into the containment building. In general, these are very conservative assumptions. Regulatory
Guide 1.4 assumes that the core has been operating at full power since the beginning of the fuel cycle where the LOCA
occurs. It also assumes that ALL of the radioactivity magically leaks out of the containment building at the end of
the LOCA. In other words, no credit is given for the fact that the containment building may actually contain the fis-
sion products released by the LOCA! Iodine has several radioactive by-products, but the most important ones are
TABLE 28.1
Estimated Release Rates for Hypothetical Reactor Accidents That Involve the Core
7 5.0 1.5 0 0.8 0.006 0.2 0.2 0.3 0.02 0.03 0.003
6 2.0 3.0 0 0.6 0.002 0.09 0.04 0.03 0.005 0.003 0.0004
5 2.0 4.0 0 0.3 0.002 0.03 0.009 0.005 0.001 0.0006 0.00007
4 12.0 10.0 0 0.3 0.002 0.0008 0.0008 0.001 0.00009 0.00007 0.00001
3 10.0 10.0 0 0.006 0.00002 0.00002 0.00001 0.00002 0.000001 0.000001 ~0.0
2 0.5 0.5 0 0.002 0.000005 0.00010 0.00050 0.000001 ~0.0 0 0
1 0.5 0.5 0 ~0.0 ~0.0 ~0.0 ~0.0 ~0.0 ~0.0 0 0
BWR Accidents
5 2.0 2.0 25 1.0 0.007 0.4 0.4 0.7 0.05 0.5 0.005
4 30.0 3.0 0 1.0 0.007 0.9 0.5 0.3 0.10 0.03 0.004
3 30.0 3.0 25 1.0 0.007 0.1 0.1 0.3 0.01 0.02 0.003
2 5.0 2.0 25 0.6 0.0007 0.0008 0.0008 0.004 0.0006 0.0006 0.0001
1 3.5 5.0 150 0.0005 ~0.0 ~0.0 ~0.0 ~0.0 ~0.0 0 0
Source: WASH-1400—see “Reactor Safety Study: An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants”, WASH-1400, U.S. Nuclear Regulatory Commission,
October 1975.
Notice that the amount of radioactivity released is a function of the severity level of the accident. Here, category 9 accidents are the most severe and category 1 accidents are the least
a Includes Mo, Rh, Tc, and Co.
b Includes Nd, Y, Ce, Pr, La, Nb, Am, Cm, Pu, Np, and Zr.
1099
1100 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
TABLE 28.2
The Inventory of Volatile Fission Products in the Gap and the Plenum of a 1,000 MWE PWR at a Burnup of About 50,000
MWD/Ton
Iodine-133 and Iodine-135. The NRC takes a s omewhat more conservative approach with respect to the how this iodine
is distributed. The current reg. says that
Twenty-five percent of the equilibrium radioactive iodine inventory developed from maximum full power operation of the core
should be assumed to be immediately available for leakage from the primary reactor containment. Ninety-one percent of this
25 percent is to be assumed to be in the form of elemental iodine, 5 percent of this 25 percent in the form of particulate iodine,
and 4 percent of this 25 percent in the form of organic iodides.
During Three Mile Island, which was the most serious reactor accident to ever occur in the United States, about 50%
of the Noble gases, 50% of the iodine, and 50% of the cesium were released from the fuel. However, almost all of the
cesium and the radioactive iodine were trapped in the water in the core and in the containment building, and most of
the Noble gases were trapped in the air inside of the containment building as well. Only a small fraction of the Noble
gases (primarily krypton and xenon) were released to the atmosphere. The fraction of the iodine that was released to
the outside atmosphere was only about two parts in ten million. No cesium ever escaped from the plant. These doses
were well below the levels that would be expected to produce any health problems for the general public. The accident at
Fukushima was an entirely different type of event, and it involved a number of GE BWRs rather than an American PWR.
28.27
Dose Rates for BWR LOCAs
The radiological release rates for a BWR LOCA are specified in another regulatory document called Regulatory
Guide 1.3. Like Regulatory Guide 1.4, this document was also authored by the NRC. The official name for Regulatory
Guide 1.3 is:
This document is specific to BWRs, and a free copy of it can be obtained from the NRC at the following website:
www.pbadupws.nrc.gov/docs/ML0037/ML003739601.pdf
Like the dose rates for PWR LOCAs, the dose rates for BWR LOCAs are calculated on the basis of two different but
similar sets of operating assumptions. The first assumption is that 100% of the Noble gases are released immediately into
the containment building following the LOCA. The second assumption is that 25% of the radioactive iodine is released
28.28 REDUCING PARTICLE DISPERSION 1101
immediately into the containment building as a result of the LOCA. No credit is given for any of the iodine or cesium that
is released from the fuel rods and trapped in the suppression pool. With an additional air filtration system, the off-site
doses are essentially the same as those calculated earlier for PWRs. The LOCA is assumed to occur at full core power,
and all of the fuel rods are assumed to fail. However, as we pointed out earlier, it is generally not realistic to assume that
more than 10% of the rods will fail if the Emergency Core Cooling System (or ECCS) works properly. Hence, the radio-
logical dose rates that are incurred by the general public are really determined by how well the ECCS works and whether
or not the containment building does the job that it is supposed to do. If both the ECCS and the containment building
work the way they are supposed to, the amount of radioactivity that is released to the general public is quite minimal. In
fact, it may not be much different than the background levels that are normally measured in your home.
28.28
Reducing Particle Dispersion with the Containment
Building Overheat Spray System
All modern containment buildings are also equipped with what is called an overhead spray system. Technically, the
overhead spray system is related to the ECCS but it is not part of it. The overhead spray system (see Figure 28.15) con-
tains a number of nozzles which are located near the top of the containment building. In the event of a severe LOCA,
these nozzles spray water onto the piping system from above. These nozzles are connected to the refueling water storage
tank, and when that empties, they draw water released by the LOCA from the containment sump. The overhead spray
system is designed to perform two primary functions:
☉☉ First, it is used to condense the steam generated by the LOCA in order to limit the pressure buildup inside of the
building as the LOCA unfolds. The nozzles are designed to spray enough cold water on to the steam to keep the pres-
sure inside the containment building below the design pressure for the building, which is usually 60 PSI (~ 400 kPa).
This helps to preserve the structural integrity of the building and to ensure that no radioactive materials are released.
☉☉ Second, the overheat spray system is used to reduce the inventory of iodine and other nongaseous fission products
that might otherwise become suspended within the atmosphere of the building. This reduces the possibility that
any of these airborne particulates would be able to leave the building in the event a leak develops in the contain-
ment building itself.
FIGURE 28.15 Overhead sprayers are used in both PWR and BWR containment buildings to cool the containment building and
scrub any particular matter from the air that is released during a large break LOCA. When the inventory of cold water emitted from
sprayers is large enough, they can also help to limit the pressure buildup inside of the containment building as the LOCA unfolds.
1102 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
Some containment buildings are also equipped with additional filters called scrubbers to trap this radioactivity by
recirculating the contaminated air through the filters. These filters reduce the amount of particulate matter within the
atmosphere of the building as the LOCA subsides. Thus, the amount of airborne radioactivity within the building may
be considerably less than the amount that is assumed in the FSAR. However, no credit is generally taken for this when
calculating the dispersion rates outside the containment building.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Crowe, C. and Schwartzkopf, J., et al. Multiphase Flows with Droplets and Particles, C&R Press, Boca Raton, FL (2012).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
17. When Stokes law is applied to a skydiver jumping out of an airplane, what does the terminal settling velocity
become?
18. If the skydiver moves his hands closer to his body, what happens to his velocity as he falls?
19. Can the same concept be applied to the particles released into the containment building by a LOCA?
20. What is the definition of an aerosol? Is the air on a smoggy day in Los Angeles an example of such as aerosol?
21. If some radioactive particles happen to escape the containment building and get entrained in the surrounding air,
can the resulting air–particle mixture be treated as an aerosol?
22. How can this air–particle mixture be modeled if the particles suspended in the mixture have a distribution of sizes
that are described by a statistical probability distribution?
23. Write down a plausible equation for this distribution of particle sizes.
24. Suppose that a plume of particles escapes the containment building and then becomes entrained in the surrounding air.
What determines the distance this plume rises above the ground, and what determines the direction this plume moves?
25. Name a famous mathematical model that can be used to describe the motion of this plume after it enters the
surrounding air. Can this model be applied to more than one type of particle or particle size?
26. Write down the equations that govern the motion of the plume through the surrounding air.
27. How is the radiological exposure to the general public calculated from the plume as it moves over the countryside?
28. What are the most important radioisotopes that have to be considered in such an analysis?
Exercise 28.2
A radioactive particle of mass M and radius R is eventually released from a containment building and enters the sur-
rounding air. The particle is released at a height of 100 m, and the air is initially stationary. Write an equation describing
the rate at which the particle will settle to the ground. Approximately how long will it take to reach the ground if the
radius of the particle is 1 micron, the density of the particle is 1 kg/m3, and the temperature of the air is 30°C?
Exercise 28.3
A plume of radioactive steam is released from a nuclear power plant following a loss of coolant accident. If the radioac-
tive particles in the plume have very long half-lives and they do not decay appreciably for a couple of days, how much
smaller will the particle concentration be at a distance of 10 km from the power plant than at a distance of 1 km?
Exercise 28.4
Calculate the drag coefficient for a spherical particle of radius 1 mm at sea level. Assume the temperature of the
surrounding air is 20°C and the air flow around the particle is laminar. Calculate the drag force on the particle using
Stokes law. Assume the velocity of the particle is 0.01 m/s.
Exercise 28.5
Calculate the terminal settling velocity for a spherical particle having a diameter 0.1 mm in calm air at 20°C. Assume
the particle weighs 0.1 g and has a drag coefficient of 0.5.
Exercise 28.6
Using the wedge model of atmospheric dispersion, calculate the particle concentration 1 km away from a reactor
containment building if one million particles are released per second into a wedge of air having a height of 10 m and an
angular displacement of 30°. Assume the lateral velocity of the particles in the center of the wedge is 1 km/h.
1104 PARTICLE TRANSPORT AND ENTRAINMENT DURING REACTOR ACCIDENTS
Exercise 28.7
Using the results obtained from Exercise 28.6, calculate the number of radioactive particles remaining in the wedge at a
distance of 10 km from the release point if the decay constant for the particles in the wedge is 0.5/h.
Exercise 28.8
A plume of radioactive material is released from a reactor containment building following an accident. The plume is
travelling at a distance of 1 km above the ground. The horizontal and vertical dispersion coefficients have a value of 20 m
and the wind velocity is 10 km/h. Calculate the concentration of particles at ground level if the containment building
is emitting 10 million particles per second into the plume. Assume the material in the particles has a very long half-life
and does not decay away.
29
Equilibrium and Non-Equilibrium
Flows, Compressible Flows,
and Choke Flows
29.1 Critical Flow and Choke Flow
Before discussing reactor accidents, we would first like to turn our attention to non-equilibrium flow, compressible flow, and choke
flow. A working knowledge of these flow types is required to understand how reactors behave when the coolant suddenly passes from
a region of high pressure PHIGH (such as the core) to a region of low pressure PLOW (such as the containment building) during a loss-
of-coolant accident (or LOCA). In this chapter, we would like to describe how the mass flow rate behaves when each of these factors
comes into play. During LOCAs, choke flow (which is sometimes called critical flow) is also related to the subject of orifice flow.
Now let us provide a brief introduction to the differences between conventional pipe flow, choke flow, and orifice flow. At relatively
low pressures and flow rates, the mass flow rate m through a circular pipe is directly proportional to the square root of the pressure
gradient dP/dz between the high-pressure and low-pressure sides (i.e., between the inlet and the outlet):
where G = m /A, ΔP = PHIGH − PLOW, and L is the length of the channel through which the coolant flows. Equations 29.1, 29.2, and 29.3
are the equations of classical pipe flow, and they show that the mass flow rate increases as the flow area increases, and the mass flow
rate decreases as more friction is added to the flow (i.e., when the value of f becomes higher). However, an interesting thing happens
when the pressure difference ΔP between the inlet and the outlet becomes very large, as it often does in reactor work. Then, the outlet
pressure does not fall below a critical pressure called the exit pressure and even though the mass flow rate is proportional to the
square root of the pressure difference between the high-pressure and low-pressure sides, the flow velocity is limited to a maximum
velocity called the critical velocity, determined by the exit pressure, which cannot be exceeded no matter what the pressure imme-
diately beyond the exit to the pipe becomes. When this occurs, the flow is said to be restricted, throttled, or choked. Choked flow is
then defined as
1105
1106 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
FIGURE 29.1 A picture of a pipe break in the reactor coolant system or nuclear steam supply system. When the break develops,
the flow of fluid through the break can be described by treating the channel through which the fluid is flowing as an orifice. The flow
in this case can be either sonic or subsonic, and at very high flow speeds, a shock wave can develop.
Choked flow can occur in liquids, gases, and two-phase mixtures. Thus, when a crack develops in a high-pressure coolant
pipe (see Figure 29.1), the flow will become choked for at least some period of time. When this concept was first applied
to reactor accidents, nuclear engineers found that critical flow could affect the way that a LOCA evolved. In particular,
the behavior of the fluid on the upstream side of the leak (where the pressure was high) no longer seemed to depend on the
pressure of the fluid on the downstream side of the leak (where the pressure was low). In other words, as the fluid velocity
approached the speed of sound on the upstream side, the pressure gradients on the upstream and downstream sides became
physically disconnected, and the mass flow rate could no longer be described using equilibrium fluid mechanics.In other
words, information could no longer be transferred between the downstream side and the upstream side.
= ρvA o (29.4)
m
because the streamlines converge at the vena contracta. Thus, the effective flow area of the orifice Ao′ becomes less than
its physical flow area Ao, and for sharp-edged or “ideal” orifices, this ratio Cd = AVC/Ao is about 0.61. For orifices in the
form of short tubes with rounded edges, it is much closer to unity, and we can set Cd = 0.99 to 1.00. Then the value of Cd
is sometimes called the coefficient of discharge. To find the flow rate through an orifice under these conditions, we can
simply replace Equation 29.4 with
= Cd ⋅ ρvA o (29.5)
m
where v is the average velocity of the coolant at the contraction point (i.e., at the Vena contracta). Thus, the coefficient
of discharge allows the average velocity and the physical area of the orifice to be used to calculate the correct rate
of discharge. It can then be shown that Equation 29.5 applies to both single-phase and two-phase flows if we write the
mass flow rates as
29.2 Orifice Flow 1107
FIGURE 29.2 (a) The discharge of fluid through an orifice. The mass flow rate in this case is given by m = Cd·ρvA, where Cd is the
coefficient of discharge. Normally, the value of Cd lies between 0.60 and 1.0. It depends on the length and the shape of the orifice.
(b) The behavior of a shock wave and a pressure pulse through a reactor coolant pipe. Normally, shockwaves are very thin, and the
pressure and the density within the wave are much higher than those of the surrounding fluid.
(Continued)
1108 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
FIGURE 29.2 (CONTINUED) (c) Examples of diffusers, nozzles, and orifices that may be found in nuclear power plants. In these
devices, the flow speed is highest at what is known as the throat. If the Mach number Ma becomes equal to unity at the throat, the
flow becomes restricted or choked and the properties at the throat become the critical properties.
where v(TP) is the velocity of the two-phase mixture, and Cd (TP) is the coefficient of discharge for the mixture. In general,
the value of Cd (TP) is a function of the void fraction at the discharge point. Normally, the vapor accelerates more rapidly
than the liquid at the Vena contracta, so if the temperature is held constant, there is a considerable decrease in the void
fraction at the point of maximum constriction.
( )
where Δp = p2 − p1 is the pressure drop across the orifice, A ′o = Cd A o √ 1 – A o A c is the effective flow area of the
2
orifice (normally about 60% of its actual area), Ac is the cross-sectional area in front of the orifice, and Cd is the coefficient
of discharge (dimensionless) that we discussed earlier. For orifices with sharp edges, Cd has a value of about 0.62, and this
value can be used if an exact value is not known. For orifices with larger L/D ratios that resemble short tubes, the coeffi-
cient of discharge is approximately 1.0.* Thus, for single-phase flow, the pressure drop across an orifice can be written as
* See Murdock, J.W., “Two-Phase Flow Measurements with Orifices”, Transactions of the ASME, Journal of Basic Engineering,
82:419–433, 1962.
29.4 Two-Phase Subsonic Orifice Flow 1109
( )
2
∆pSP = m
A ′o (2ρ) (29.9)
Now, let us see what happens to this equation when the flow contains more than one phase.
☉☉ See https://mysite.du.edu/~jcalvert/tech/fluids/orifice.htm for more information on the subject of orifices.
m (
l = A ′ol √ 2ρl ⋅ ∆ ( pTP )l ) (for the liquid phase) (29.10)
v = A ′ov
m √ ( 2ρ v ⋅ ∆ ( pTP ))
v
(for the vapor phase) (29.11)
The effective cross-sectional areas through the orifice for each phase are
( )
A ′ol = A ol Cd √ 1 − A o A c (29.12)
2
(
A ′ov = Aov Cd √ 1 − A o A c (29.13) )
2
Thus, we can write
( )
A ′o = ( A ol + A ov ) Cd √ 1 − A o A c (29.14)
2
and if Cd is approximately the same for the liquid and the vapor, then Equations 29.10–29.13 can be combined to give
(
√ ( ∆pSP )l ( ∆pTP )l ) + √ (( ∆pSP )v ( ∆pTP )v ) = 1 (29.15)
Then, assuming that Δp TP L and Δp TP V are the same as the pressure drop for the two-phase mixture Δp TP, we see that
In other words, the single-phase pressure drops can be measured individually, and they can then be combined together to
find the pressure drop for the two-phase mixture Δp TP. At least when it comes to water, experiments have shown that the
liquid phase should be given a greater weight in Equation 29.16 than the vapor phase, where representative values of these
weights become 1.26 and 1.0, respectively.* A more accurate expression for the two-phase pressure drop then becomes
* See Murdock, J.W., “Two-Phase Flow Measurements with Orifices”, Transactions of the ASM E, Journal of Basic Engineering,
82:419–433, 1962.
1110 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
Again, Equation 29.17 applies to any flow that is subsonic but is cannot be used for flows that are critical or choked. If
the flow becomes choked, the methodology presented in Section 29.6 should be used instead. Thus, irrespective of the
number of phases the flow possesses, the mass flux G = ρv (in kg/m2-s) up to the point where the flow becomes critical
is proportional to
where the constant of proportionality C is fluid and geometry dependent, pHIGH is the pressure on the upstream side of
the orifice, and pLOW is the pressure on the downstream side of the orifice. Hence, in conventional orifice flow, the flow
rate never reaches a Mach number of unity (Ma = 1.0) where the flow stream becomes restricted or choked. Suppose
that the flow area of the liquid phase through the orifice is Al and the flow area of the vapor phase through the orifice is
Av. Then, the subsonic liquid and vapor mass flow rates are
Subsonic Mass Flow Rates for the Liquid and Vapor Phases through an Orifice
m (
l = Cd ⋅ A ′l √ ( 2ρl ) ( pHIGH – pLOW )l ) (for the liquid) (29.19)
v = Cd
m ⋅ A ′ √ (( 2ρ ) ( p
v v HIGH – pLOW ))
v
(for the vapor) (29.20)
where Ac is the cross-sectional area of the flow channel immediately preceding the orifice, and
( (
A ′l = A l √ 1 – A o A c ) ) (29.21)
2
√ (1 – ( A ) ) (29.22)
2
A ′v = A v o Ac
Hence, two-phase orifice flow is similar to single-phase orifice flow except that each of the phases must be treated indepen-
dently when calculating the pressure drop and the mass flow rate. Finally, it can be shown with some additional work that
TP = m
m v = Cd ·A ′l ⋅ √ ( 2ρl ) ( pHIGH – pLOW ) + Cd ⋅ A ′v ⋅ √ ( 2ρv ) ( pHIGH – pLOW ) (29.23)
l +m
or
when both phases have the same discharge coefficient Cd. It is important to bear in mind that neither of these equations
places any restrictions on the speed that the fluid can pass through the orifice, as long as it is less than the speed of sound
(i.e., v < c). Now, let us take a look at how the speed of sound affects the mass flow rate.
Δp = (pHIGH − pLOW) through the orifice is 6.9 kPa (1 PSI) and the coefficient of discharge is 1.0. The area of the orifice is
0.1 m2, and the area of the channel immediately preceding the orifice is 0.2 m2. Assume the flow is subsonic.
Solution If the flow is subsonic, the mass flow rate of each phase is simply m (
l = A ′ol √ 2ρl ⋅ ∆ ( pTP ) )
and m ( )
v = A ′ov √ 2ρv ⋅ ∆ ( pTP ) . A ′ol and A ′ov are the same because the volumes are the same. Hence,
( (
A ′ol = A ′ov = A ′o 2 = 0.5Cd A o √ 1 – A o A c ) ) where
2
Ao = 0.1 m2, Ac = 0.2 m2, and Cd = 1.0. We then find that
( )
A ′ol = A ′ov = 0.05 1 – 0.52 = 0.066, ρl = 958 kg/m3 and ρv = 0.6 kg/m3. The mass flow rates are then m
l = 0.066√(2 × 958 ×
6,900) = 236.4 kg/s and m v = 0.066√(2 × 0.6 × 6,900) = 5.9 kg/s. The total mass flow rate is m
=ml+m v = 242.3 kg/s.
[Ans.]
c 2 = ∂P/ ∂ρ (29.26)
where γ = CP /CV is the specific heat ratio of the fluid which is essentially the ratio of its enthalpy to its internal energy.
Furthermore, if the coolant can be treated as an ideal gas, we know from our earlier discussion in Chapter 7 that the
equation of state is relatively simple and the pressure is related to the temperature by P = ρRT. Substituting this expres-
sion into Equation 29.28 and differentiating it leads to
c 2 = γRT (29.29)
or
The Speed of Sound for a Fluid that can be treated as an Ideal Gas
c = √ γRT (29.30)
Equation 29.30 provides a relatively simple way to find the speed at which a sonic wave will propagate through the
fluid. As we shall shortly see, it is also applicable to compressible flows as long as the fluid behaves like an ideal gas.
The temperature in Equation 29.30 is then the absolute temperature of the fluid in degrees Kelvin. Closely related to
this propagation speed is another important number of fluid mechanics called the Mach number Ma, named in honor of
Austrian physicist Ernst Mach (1838–1916). The Mach number then becomes the ratio of the actual velocity of the fluid
(or an objected imbedded in it in still air) to the speed of sound in the same fluid at the same state:
Thus, the Mach number depends on the speed of sound, and the speed of sound depends on the state of the fluid.
For any fluid that can be treated as an ideal gas, the propagation speed is a function of the temperature only (see
Equation 29.29). The flow of fluid through a reactor can then be described in terms of the Mach number of the flow.
The flow is called subsonic when Ma < 1, transonic when Ma ≅ 1, supersonic when Ma > 1, and hypersonic when
Ma ≫ 1. Fluid flow during reactor accidents can be described in this way. For water at 1 atm and 30°C, the speed
of sound is c = 1,512 m/s; for water at 8 MPa and 270°C, it is c = 1,082 m/s; and for water at 15.5 MPa and 300°C, it
is c = 977 m/s. Conversely, for air at 30°C, it is 349 m/s. Now, let us discuss the relationship between the speed of
sound and various types of orifice flow. In general, flows through devices like nozzles, diffusers, and orifices behaves
differently at sonic speeds than they do at subsonic ones. In each device, the flow area changes as a function of posi-
tion, and the pressure falls at the location where the flow area is least because this is the location where the velocity
is highest. In nozzles, the flow area is smallest at the exit of the nozzle, and the flow area at this point is called the
throat (see Figure 29.2c). In diffusers, which are designed to perform the opposite function of nozzles, the flow area
is greatest at the exit and least at the entrance. Hence, the throat of a diffuser can usually be found at the entrance.
Finally, in a nozzle that suddenly converges and then diverges, the throat can be found somewhere near the center of
the device. Nozzles of this type are called converging–diverging nozzles, and these nozzles are sometimes used to
accelerate gases to supersonic speeds (see Chapter 34). However, they should not be confused with Venturi nozzles,
which are used strictly for fluids that are incompressible. Orifice flow is similar to converging–diverging nozzle flow
except that the edges are sharper and the L/D ratio at the throat is lower. The fluid velocity is greatest at the point of
maximum contraction if the density is constant. Hence, the fluid velocity reaches the sonic velocity at the location of
the throat first as the flow rate is increased. The size of the Mach number in the throat then determines whether the
flow is sonic or subsonic. The flow through an orifice becomes restricted or choked when the discharge speed at the
throat of the orifice becomes equal to the sonic speed. Hence, when the Mach number of the flow becomes equal to
unity (Ma = 1.0), the flow reaches its maximum possible speed and this speed then becomes known as the critical
speed. Increasing the pressure difference between the inlet and the outlet then has no further effect on the speed
at which the flow exits the opening. Only in this case, the length of the channel (or pipe) through which the fluid is
flowing determines the point at which the critical flow speed is reached. Whether the flow becomes critical or choked
depends on how the actual velocity compares to the critical velocity. Example Problem 29.5 illustrates how this can
be determined for a gas reactor.
FIGURE 29.3 As fluid flows through an orifice or a pipe, the pressure drops as the velocity increases. Eventually, a point is
reached where the exit pressure reaches the critical pressure and the exit velocity reaches the critical velocity. At this point, the pres-
sure gradient becomes fully developed, and the exit pressure cannot fall any further. When this occurs, the fluid velocity at the exit
becomes equal to the sonic speed c.
side) and the orifice area determine the discharge rate. Thus, the critical flow rate becomes the maximum rate at which
fluid can exit the orifice. The velocity of the fluid at this point is called the critical velocity, and it is usually represented
by vCRITICAL. Hence, the maximum flow rate occurs when Ma = 1 and v = vCRITICAL = c (c being the sonic speed of the
fluid). Thus, the flow under these conditions becomes restricted or choked. Choked flow is the most common type
of flow during the early stages of reactor LOCAs; moreover, it can occur in both single-phase fluids and two-phase
m ixtures. Before discussing how this flow occurs for water–steam mixtures, we would first like to discuss its behavior
for gaseous coolants such as carbon dioxide and helium.
29.7
Applying t he Ideal Gas Law to Orifices
As we mentioned earlier, the ideal gas law can be thought of as an equation of state that relates the density of a gas to its
temperature and its pressure. The ideal gas law was first proposed by Émile Clapeyron when he combined Boyle’s law
and Charles’s law together in 1834. The ideal gas law can also be derived from the kinetic theory of gases, but we will
not show how this can be done here. One of the most common forms of the ideal gas law is
where p is the pressure of the gas, V is the volume of the gas, n is the number of moles of the gas present, T is the abso-
lute temperature of the gas, and R is the ideal or universal gas constant. Hence, T is measured in degrees Kelvin (°K).
The universal gas constant R has a value of 8.314 J/mol °K, and from statistical mechanics, it can also be shown that it
is equal to the product of Boltzmann’s constant and Avogadro’s number. The utility of the ideal gas law is that it allows
us to relate the temperature and the pressure of a gas to its density ρ. As a practical matter, the ideal gas law can also
be written as
where M is the molar mass (in grams per mole). In SI units, p is measured in pascals, V is measured in cubic meters, and
n is measured in moles. For state diagrams, the two specific forms of the relationship between the pressure, the density,
and the temperature are
( )
– (1/γ )
Based on pressure: p1 = p0 ρ1 ρ0 (29.34)
( )
(γ − 1)γ
Based on temperature: T1 = T0 ρ1 ρ0 (29.35)
where state 0 and state 1 are two arbitrary states, and γ = cp/cv. Here, the value of γ is determined solely by the physi-
cal properties of the gas. Because γ is the ratio of the specific heat at constant pressure to the specific heat at constant
volume, it is also a dimensionless number.
= ρvA (29.36)
m
where A is the area of the pipe. From Bernoulli’s equation, we also know that
pHIGH + ( 1
2 ρv 2 ) HIGH
= pLOW + ( 1
2 ρv 2 ) LOW
(29.37)
Now, if the reservoir that feeds the fluid into the pipe is very large, the velocity of the fluid in the reservoir will be very
small and we can set vHIGH = 0. Bernoulli’s equation then becomes
pHIGH − pLOW = ( 1
2 ρv 2 ) LOW
(29.38)
∆p = ( 1
2 ρv 2 ) LOW
(29.39)
Taking the derivative of Equation 29.39 with respect to the fluid velocity and multiplying through by the cross-sectional
area A gives
d(pA)/dv = m
(29.40)
Hence, we can combine Equation 29.40 with the continuity equation, which gives
If we also differentiate the continuity equation with respect to p (with dA/dp = 0), we find that
(1/m)dm/dp = (1/ρ)dρ/dp + (1/v)dv/dp (29.42)
29.8 Single-Phase Critical Flow 1115
or
dv/dp = −(v/ρ)dρ/dp = − m
ρ2 A dρ/dp (29.44) ( )
Combining this result with Equations 29.36 and 29.40, we come to the conclusion that
( m )
2
MAX A = ρ2 dp dρ (29.45)
G MAX 2 = ρ2 dp dρ (29.46)
This equation has enormous practical importance because it allows us to deduce the highest possible mass flux GMAX
through any pipe or orifice where the flow becomes critical or choked. Moreover, we can solve Equation 29.47 for the
critical flow velocity vCRITICAL or vMAX at the point where choking occurs. In this case, vMAX is simply
or equivalently,
Equation 29.49 can be used to find the critical velocity for either a liquid or a gas. Now, let us apply this result to predict
the critical flow rate for an ideal gas, for example, one that follows the ideal gas law. We know that for a steady-state,
inviscid, isentropic, and adiabatic flow, the specific enthalpies of the high- and low-pressure regions are related by
h HIGH − h LOW = ( 1
2 ρv 2 ) LOW
(29.50)
when the reservoir velocity in the tank that feeds the orifice or pipe is low. We also know that dh = cpdT. The exit velocity
on the low-pressure side is then reached when
or
{ (
v LOW = √ 2c p THIGH 1 − TLOW THIGH (29.53) )}
However, we also know that
1116 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
( )
(γ − 1)/γ
TLOW THIGH = pLOW pHIGH (29.54)
and
( )
1/γ
ρLOW ρHIGH = pLOW pHIGH (29.55)
( ) { (
v LOW = ρHIGH ρLOW √ 2c p THIGH pLOW pHIGH
)
2/γ
(
− pLOW pHIGH )
(γ − 1)/γ
}
(29.56)
The critical velocity on the low-pressure side is then reached when dvLOW/dpLOW = 0. From Equation 29.56, we see that
this only occurs when
where R is the molar gas constant R = Ru/M. Furthermore, since p = RρT, and the speed of sound c in an ideal gas is
given by
c = √ γdp/dρ (29.58)
it is easy to see that c = vCRITICAL = √γRTLOW. Thus, for an ideal gas, the velocity at which choking occurs is the criti-
cal velocity vCRITICAL corresponding to a Mach number of 1.0. For an ideal gas (or a substance which behaves as one),
steady-state choked flow occurs when the ratio of the absolute exit pressure (pLOW) to the absolute upstream pressure
(pHIGH) becomes less than or equal to [2/(γ + 1)]γ/(γ − 1), which is called the critical pressure ratio (CPR) RCRITICAL. Here, γ
is the specific heat ratio of the gas (discussed earlier) which is also called the isentropic expansion factor. So, the critical
pressure ratio RCRITICAL where the flow becomes choked is
where the size of γ is determined solely by the molecular composition of the gas. For most gases, values for γ range from
about 1.10 (for butane) to about 1.67 (for monatomic gases), and so the value of [2/(γ + 1)]γ/(γ − 1) ranges from about 0.48–0.59.
This means that choked flow for most gases occurs when RCRITICAL falls between 0.48 and 0.59 (or equivalently, when the
ratio of PHIGH to PLOW falls between 1.7 and 2.1). Reactor coolants such as helium and carbon dioxide exhibit behavior as
well. The minimum critical pressure ratios for choking to occur are shown in Table 29.1. These values were obtained from
1. Perry, R.H. and Green, D.W. (1984). Perry’s Chemical Engineers’ Handbook, Table 2-166, (6th Edition). McGraw-
Hill Company. ISBN 0-07-049479-7.
2. Phillips Petroleum Company (1962). Reference Data for Hydrocarbons and Petro-Sulfur Compounds (Second
Printing). Phillips Petroleum Company.
TABLE 29.1
Choked Flow Parameters for Common Industrial Gases
where Cd is the aforementioned discharge coefficient (dimensionless), Ao is the discharge hole cross-sectional area (in
m2), ρ is the density of the gas (in kg/m³) upstream of the orifice, m
MAX is the maximum mass flow rate in kg/s, and γ =
cp/cv (also dimensionless).
v MAX = c = √ ( γRTc )
where Tc is the absolute temperature of the air surrounding the pipe. Now, for carbon dioxide, γ = 1.289, R = 188.9 J/kg °K,
and T = 300 K. So, the maximum velocity the carbon dioxide can escape from the pipe is vMAX = √(γRTc) = 271.4 m/s.
The maximum velocity is this case is also the speed of sound for the CO2. According to Table 29.1, the flow becomes
choked when the pressure outside the pipe is roughly 55% of the pressure inside the pipe. [Ans.]
An excellent online calculator that can be used to find the discharge rate from a large tank of high-pressure gas when the
flow becomes critical or choked can be found at
https://www.lmnoeng.com/Gas/choke.php
The reader is encouraged to visit this site for more examples of how the critical flow rate can be found for a variety of
gases and piping system geometries. It is applicable to any adiabatic gas flow. Hence, it can be used to find the correct
flow rate as long as no heat is added to or taken away from the fluid.
29.9 Effect of the Back Pressure and the Exit Pressure on the Critical Flow Rate
The pressure pLOW in Equation 29.53 is the absolute pressure at the exit to the orifice or pipe. Sometimes this pressure is
called the exit pressure; that is, pLOW = pEXIT. Thus, the critical flow rate m
MAX or m
CRITICAL is determined by the pressure
at the inlet of the pipe as well as the pressure at the exit. When the downstream pressure pLOW falls far enough below
the inlet or upstream pressure pHIGH, the flow becomes choked. However, the pressure beyond the exit can fall even
lower, and the ambient pressure at this location is called the back pressure pBACK. Normally, the back pressure pBACK is
lower than the exit pressure, that is, pBACK < pEXIT. If the back pressure is less than the exit pressure, there is no further
increase in the mass flow rate m MAX or decrease in pEXIT. This behavior is shown in Figure 29.4. As the back pressure is
reduced even further, the exit pressure remains the same but the fluid expands beyond the discharge point if it happens
to be compressible. Then, the flow takes on a characteristic parabolic shape. This effect occurs for both single-phase
and two-phase flows. The pressure at the entrance to the pipe is then called the upstream pressure pHIGH or the stagna-
tion pressure po (refer again to Figure 29.4). Because the exit pressure can never exceed the stagnation pressure, the exit
pressure ratio can only vary between 0 and 1.0. The exit pressure pEXIT can be reduced until it reaches a critical pressure
pCRITICAL where reducing it any further has absolutely no effect on the mass flow rate. At this point, the flow is said to be
critical or choked. Figure 29.4 shows how the flow behaves under these conditions.
FIGURE 29.4 The effect of the exit pressure on the mass flow rate when the flow becomes choked in short and long pipes. The
ambient pressure beyond the exit is called the back pressure pBACK, and it stays below the exit pressure (called pEXIT or pLOW)
even when the flow becomes choked. Notice how the discharge rate increases as a pipe becomes shorter. The curve on the right
then shows how the flow behaves up to the exit pressure where choking occurs. The ratio of the exit pressure to the inlet pressure
(pLOW/pHIGH) is then called the CPR RCRITICAL.
This equation establishes a simple relationship between the critical velocity and the speed of sound for single-phase
liquids and gases. However, finding the speed of sound is not as simple for multiphase mixtures, and most of the time,
each of the phases can have a different sonic speed or speed of propagation.
☉☉ Second, the velocities of the phases must be known. If the velocities of the liquid and the vapor are the same, then
the slip ratio will be 1.0, and if they move at different speeds, then the slip ratio will be greater than 1.0. Finally,
if the temperatures of the two phases are exactly the same, then the phases will be in thermal equilibrium. Under
these conditions, it is possible to describe the behavior of the two-phase mixture using what is called a homoge-
neous equilibrium model or HEM.
Now, let us apply these two conditions to predict the flow rate of the mixture before choking occurs. For the moment,
assume that the liquid and vapor phases are moving at different speeds, and the slip ratio is greater unity (S > 1.0). In the
absence of interfacial drag, the momentum equations that describe the behavior of the mixture are
If we multiply these equations by the flow areas for the liquid and vapor phases Al and Av, we obtain
(
∆ ( pv A v ) = ∆ ρv v v 2 A v ) (for the vapor) (29.64)
Now, the pressure drops for the two phases must be equal between the beginning and the end of the orifice through
which the two-phase mixture passes. This means that Δp TP = Δpl = Δpv, and this allows us to write the total two-phase
pressure drop as
(
∆pTP ( A v + A l ) = ∆ ρv v v 2A v + ρl v l 2A l ) (for the mixture) (29.66)
or
where A = Al + Av. Now, the flow rate for the mixture is the sum of the flow rates for the liquid and vapor phases:
( ) ( )
ρv = ρv v v A v A + ρl v l A l A (29.69)
Using the previous expression for the void fraction, we then see that
where G is the mass flux of the mixture. This leads us to the following relationships between the void fraction and the
equilibrium quality:
Relationships between the Void Fraction and the Quality for an Equilibrium Flow
(1 – x)ρv = (1 – α)ρl v l (29.71)
xρv = αρv v v (29.72)
1120 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
Finally, noting that we have set α = A v /A and (1 − α) = Al/A, the equation for pressure drop of the two-phase mixture
(in differential form) becomes
( ) ( )
dpTP dz = −(G)2 ⋅ d/dz (1 – x)2 (1 – α) ρl + x 2 α ρv (for the mixture) (29.73)
or
dpTP dz = −(G)2 ⋅ dC/dz (29.74)
where C = [((1 − x)2/(1 − α))ρl + (x2/α)ρv)]. Here, the mass flux G is given by ρv, α and x are the equilibrium void fraction
and quality, and we have assumed that we can write Δp/Δz = dp/dz. Thus, for critical two-phase flow, the maximum flow
rate for a compressible mixture is
( )
G MAX 2 = −(1/dC/dz) ⋅ dpTP dz (29.75)
Alternatively, we can eliminate either α or x from our previous equations using one of the following relationships:
{ ( ) }
x = 1 1 + ρl ρv ⋅ ((1 − α)/α) ⋅ (1/S) (29.76)
After doing so, the pressure drop becomes a function of just one variable (α or x) and the slip ratio S. Finally, it can be
shown that when the equilibrium quality is 0, the term in the brackets
( ) ( )
C = (1 – x)2 (1 − α) ρl + x 2 α ρv (29.78)
reduces to the specific volume υ (the inverse of the density) and the mass flux becomes
which is exactly what we would expect for a single-phase fluid. Notice that Equation 29.79 can also be written as
Hence, the equations we have just derived can be used to predict the maximum mass flow rate through an orifice or pipe
when the flow becomes choked. Now, let us turn our attention to how the maximum mass flux is affected by the length
of the orifice or the coolant channel.
29.12 Critical Flow Rates for Long and Short Coolant Channels
As Figure 29.4 illustrates, critical two-phase flow behaves differently in long coolant channels than it does in short
ones. Critical flow starts when the pressure gradient achieves its maximum possible value at the channel exit. In long
channels, the residency time is long enough for the phases to come into thermal equilibrium. When this happens, the
temperatures of the two phases become identical. It then becomes possible to correlate the critical pressure ratio to the
L\D ratio of the orifice. For water–steam mixtures, this relationship is presented in Figure 29.5. The maximum flow rate
29.14 Discharge Rates for Equilibrium and Non-equilibrium Flows 1121
FIGURE 29.5 The critical pressure ratio for orifices and short, medium, and long pipes. Here, the CPR is expressed as a function
of the L/D of the orifice or pipe. The coolant channels in this case are assumed to have sharp edge rather than smooth ones.
is then obtained when the slip ratio is equal to √(υv /υl), where υv is the specific volume of the vapor phase and υl is the
specific volume of the liquid phase. The critical pressure ratio (or CPR) is approximately 0.55 for long pipes, but it can
be 0.33 for short pipes and less than 0.10 for orifices. Thus, the length of the channel through which the flow passes has
a great deal to do with when it becomes choked.
FIGURE 29.6 How two-phase critical flow behaves in orifices and short pipes. Under these conditions, thermal equilibrium
is never fully achieved. In the orifice on the left (part a) flashing occurs outside of the orifice and the flow immediately becomes
choked. In the short pipe in the center (part b) thermal equilibrium is only partially achieved and the critical pressure ratio becomes
larger. In the pipe on the right (part c), core breakup only occurs near the exit to the pipe and the critical pressure ratio approaches
that of a very long pipe. For even larger L/D ratios, the critical pressure ratio (or CPR) approaches an asymptotic value and does not
change any further.
inside of it. In this case, the flow immediately becomes choked, and the mass flow rate can be estimated for single-phase
fluids from the conventional orifice equation for incompressible flow:
Here, we must assume that Δp = po − pBACK because the flow behaves as if there is no critical pressure (pCRITICAL = 0)
(see Figure 29.6a). For short pipes, which are described by region 1 of the curve in Figure 29.5, and where 1 < L/D < 3,
the liquid accelerates through the pipe, and its core becomes a metastable jet. Liquid then evaporates from the surface
before it reaches the exit (see Figure 29.6b). Here, the mass flow rate for single-phase coolants such as water and liquid
sodium is given by
where Δp = po − pCRITICAL and pCRITICAL is obtained from Figure 29.5. Next, in region 2, where 3 < L/D < 12, the liquid
core breaks up, and the critical flow rate is less than that predicted by Equation 29.76 (see Figure 26c). The critical flow
rates in this case are shown in Figure 29.7 as a function of the upstream or stagnation pressure po. Finally, when thermal
equilibrium is attained in region 3 and pitch-to-diameter (P/D) ratio > 12, the critical flow rate can be expressed as
where Δp = po − pCRITICAL and pCRITICAL = 0.55po. In other words, when thermal equilibrium is finally reached, we can
set Δp = 0.45po. Again it must be emphasized that the equations we have just presented apply to orifices, pipes, and pipe
cracks that have very sharp edges. If these objects have well-rounded edges (particularly at the entrance), the liquid core
in region 1 remains in closer proximity to the outer walls, and these entrances result in much higher CPRs and somewhat
greater flow rates than the data in Figure 29.5 might suggest. The effect of rounded entrances becomes progressively
less important for L/D ratios between 3 and 12, and for long channels, where L/D > 12, it can be completely ignored
because thermal equilibrium has been attained. Finally, gas or vapor bubbles in the coolant can affect the critical flow
rate since they can alter the stability of the vapor core as it progresses through the flow channel. Examples 29.7 and 29.8
are designed to illustrate these points.
29.14 Discharge Rates for Equilibrium and Non-equilibrium Flows 1123
FIGURE 29.7 The critical flow rate as a function of the upstream pressure for region 2 of the CPR curve. The upstream pressure Po
is sometimes called the stagnation pressure, and the pressure at the exit is sometimes called the exit pressure. The ambient pressure
beyond the exit is then called the back pressure pBACK. In nuclear applications, the back pressure is generally lower than the exit
pressure, that is, pBACK < pEXIT.
Now, Ao = πR2 = 0.314 m2, po = 16 MPa, and ρ ≅ 710 kg/m3. Hence, the discharge rate is m = 0.61 × 0.314 × √(2 × 710 ×
0.45 × 16,000,000) = 19,367 kg/s. Note that this is nearly equal to the core mass flow rate. [Ans.]
(
= 0.61 A o √ (2ρ∆p) = 0.61 A o √ 2ρ 0.72 po
m )
Now, Ao = πD2/4 = 0.000078 m2, po = 15.5 MPa, and ρ ≅ 710 kg/m3. Hence, the discharge rate is m = 0.61 × 0.000078 ×
√ (2 × 710 x 0.72 × 16,000,000) = 6.08 kg/s. Because the discharge rate is low, the water lost from the pipe can be easily
made up from the water stored in the accumulator tanks. [Ans.]
1124 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
Finally, upon further reflection, it can be seen that the pressure difference Δp = po − pCRITICAL that drives the discharge
rate depends on the physical length L of the orifice or pipe. When the flow area is fixed, pCRITICAL falls as the pipe
becomes shorter. Thus, the value of Δp increases, while the value of po remains the same. In other words, shorter chan-
nels of the same diameter have greater discharge rates than longer channels of the same diameter do, and the discharge
rate increases with decreasing length because more of the coolant remains a liquid. Thus, the effective pressure drop
Δp = po − pCRITICAL increases as the L/D ratio becomes smaller, and this causes the discharge rate m , which is propor-
tional to the square root of Δp (see Section 29.13) to increase. Example 29.8 shows how the discharge rate from the
primary loop of a PWR is affected by a pipe attached to the pressure relief valve (PRV). Thus, short pipes can always
remove more fluid from the core than longer pipes can when the transverse dimension of the pipe is the same.
The first assumption is equivalent to a condition of no slip and the second assumption is equivalent to a condition of
thermal equilibrium. Now, consider a very long pipe where the residency time is sufficiently long for the two phases to
have the same temperature. When this occurs, the phase relationships for single-phase critical flow can be translated
directly to the phase relationships for two-phase critical flow. If the flow is assumed to be isentropic (i.e., frictionless
and adiabatic), then the resulting pressure drop will be due to the acceleration of the fluid as it moves from pHIGH to pLOW.
This inexorably leads us to conclude that
where G is the mass flux of the mixture. Now, for an isentropic fluid, the sound speed c is given by c2 = dp/dρ, and from
the chain rule we see that
Equation 29.85 then enables us to find the critical sound speed for the mixture if we can find the value of (dp/dυ). To find
the critical velocity vCRITICAL = c, we multiply (dp/dυ) by the derivative of the specific volume υ = υl + xυlv, and take the
square root of the result. After doing so, we find that
( )
c 2 = υ 2 dυ v dρ + (1 − x) dυ l dp + υ lv dx/dp (29.86)
Thus, the sound speed and the critical velocity can be expressed as a function of the equilibrium quality:
and in the limit where x → 0, we recover our previous expression for the speed of sound of the liquid phase again.
Alternatively, consider how the speed of sound behaves for an air–water mixture as the gas content of the mixture
changes. The results are shown in Figure 29.8. At atmospheric pressure, the sound speed falls dramatically (from
about 1,500 to about 100 m/s) when only a small amount of air is present. The critical flow rate GCRITICAL for the two-
component mixture then becomes
when there is no slip. Equation 29.82 is one of the key predictions of the homogeneous equilibrium model. We must
keep in mind that the homogeneous equilibrium model is not a physically complete model, but it is not a totally unre-
alistic one either. It can produce reasonable results as long as the transit time is long enough for the phases to achieve
thermal equilibrium. The derivatives must also be evaluated at the surface of the orifice or the channel exit since
this is where the “choking” actually occurs. In real systems, the critical flow rate may be higher than Equation 29.88
implies because the flow may possess some internal inhomogeneities and at least one phase may have a sound speed
that is higher than the speed of sound of the mixture. However, this is partially compensated for by the fact that the
resulting pressure waves may propagate at much lower amplitudes than they do for a pure liquid or a pure vapor. Not
surprisingly, the homogeneous equilibrium model produces reasonable results when the system pressure is high and
when the equilibrium quality is also high. However, it does not do as well when the system pressure is low because the
density ratios are larger. It also does not do as well at low flow qualities when non-equilibrium effects become important.
Finally, as the stagnation pressure and the quality increase, the homogeneous equilibrium model tends to give better
results, which is a conclusion we can also draw from our previous phenomenological discussion.
☉☉ Fauske, H.K. “Two Phase Critical Flow” paper presented at the MIT Two-Phase Gas-Liquid Flow Special
Summer Program, 1964.
In practice, the solutions to Fauske’s equations are normally presented in tabular form. Sometimes they are presented
in the form of a chart or a graph. Notice that the critical flow rate decreases near the exit to a long pipe as the fluid
quality increases. This is because a two-phase mixture has more friction near the exit than a single-phase mixture does.
It also occurs because there is less fluid (in liquid form) to flow through the pipe at this point. The other conclusion that
we can immediately draw from Fauske’s analysis (see Figure 29.8) is that increasing the system pressure (i.e., the value
of pHIGH) always increases the critical flow rate because the pressure gradient at the exit to the channel becomes higher.
Results from many experiments have confirmed this observation. Because the slip equilibrium model assumes that the
phases are in thermal equilibrium, it gives much better results for L/D ratios greater than 12. Consequently, it is ideally
suited for long flow channels and orifices (where the phases have enough time to come into thermal equilibrium), while
its results are not as good for shorter ones (L/D < 12) (see Figure 29.5).
The channel or orifice lengths required to attain a state of thermal equilibrium have been studied extensively for
water–steam mixtures over the years. Normally, a length of 0.1 meters (about 4 in.) appears to be the magic number
at which thermal equilibrium is attained. For shorter channels, the discharge rate increases strongly with decreasing
length because more of the coolant remains a liquid. Thus, the effective pressure drop Δp = po − pCRITICAL increases
1126 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
FIGURE 29.8 The speed of sound for an air–water mixture as a function of the equilibrium quality. (Taken from http://wins.engr.
wisc.edu/teaching/mpfBook/node19.html but then revised.)
☉☉ Richter, H.J., “Separated Two-phase Flow Model: Application to Critical Two Phase Flow,” International Journal
of Multiphase Flow, 9(5):511–530, 1983.
In the literature, it is also called a separated flow model. It first became popular in the 1980s, and it is still used in LOCA
analysis codes like RELAP and TRACE today. It allows the liquid and vapor phases to move at different speeds, and it
uses two continuity equations, two momentum equations, and a mixture energy equation to describe the processes asso-
ciated with choked flow. It is a difficult model to implement, and so a thorough discussion of its features is beyond the
scope of this book. Finally, when thermal equilibrium is attained, it can be shown that the critical mass flux, GCRITICAL,
in a long coolant channel or orifice having L/D > 12 or L > 10 cm is given by
{ } (29.90)
2 −1
ρ′ = x ρv + (1 − x)S ρl ⋅ x + (1 − x) S2
Here, ρ′ is the effective density of the fluid flowing through the orifice which depends on the slip ratio (S = vv/vl), and in
many books, ho is referred to as the stagnation enthalpy. Sometimes pHIGH and THIGH are known as the reservoir states.
However, the slip ratio can change as one moves through an orifice, and the values of the phase densities and x can
also change because the pressure can fall. Therefore, ρl, ρv, and x are usually taken to be the densities and the quality
at the critical back pressure (pLOW). For a homogeneous equilibrium model, the slip ratio S would be 1.0. However, for
separated flow, finding the correct value of S is more difficult to do because there can be several flow regimes involved.
29.17 The Equilibrium Rate Model 1127
For LOCAs, two prominent models have emerged for calculating the slip ratio when the flow becomes choked. The
first one was proposed by Fauske, and the second one was proposed by Moody. The values of the slip ratios for these
models are
S = (ρ ρ )
0.33
l v (Moody’s model) (29.92)
In both cases, the densities are evaluated at the critical back pressure PLOW, that is, downstream of the orifice itself. More
information regarding these models can be found at
☉☉ Fauske, H.K. “The Discharge of Saturated Water Through Tubes”, Chemical Engineering Symposium Series,
61:210, 1965.
☉☉ Moody, F.J. “Maximum Two-Phase Vessel Blowdown from Pipes”, The Journal of Heat Transfer, 88:285, 1966.
In general, neither model predicts the critical flow rate correctly under all conditions. The homogeneous equilibrium
model works well for pipe lengths and orifices greater than about 30 cm and for pressures greater than about 2.0 MPa
(300 PSI). When L/D > 40, the HEM is better than either Fauske’s model or Moody’s model. Fauske’s model gives
slightly better results than Moody’s model for lower pressures and lower L/D ratios. The Richter model does better than
any other model when non-equilibrium effects become important and when a channel is very short (say, L/D < 6). In
other words, there is no perfect model to predict the critical flow rate during a LOCA under all conditions.
The ERM Equation for the Critical Mass Flux for Saturated Liquids
( ) ( )
G ERM = h lv υ lv ⋅ √ 1 Tc l (29.93)
where G is the mass flux (measured in kg/m2 s), hlv is enthalpy of vaporization (measured in J/kg), υlv is the change in
specific volume of the flashing fluid (measured in m3/kg), T is the absolute temperature of the liquid, and cl is the specific
heat of the flashing liquid (measured in kJ/kg °K). The ERM is compared to other models as well as to experimental
data in Figure 29.9.
To use this model, all of the properties must be evaluated at the stagnation pressure po. Equation 29.93 works well
for liquids close to the saturation point (i.e., saturated ones). However, if the pipe break happens to occur at a location
where the coolant is no longer close to the saturation temperature, a correction factor may have to be applied to Equation
29.93 to obtain acceptable results. The effect of subcooling on the discharge rate is then obtained by adding a correction
factor F of the form
where To is the temperature at the stagnation point. Then, the critical mass flux for subcooled fluids that immediately
flash can be deduced from
* See Fauske, H.K. “Flashing flows – some practical guidelines for emergency releases”, Plant Operations Progress, 4:132, 1985.
1128 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
FIGURE 29.9 A comparison of the discharge rates predicted by the ERM, the HEM, and the orifice equation for saturated water at
different pressures and L/D ratios.
The ERM Equation for the Critical Mass Flux for Subcooled Liquids
G CRITICAL = √ F + G ERM 2 (29.95)
Notice that when there is no subcooling, Equation 29.95 reduces to Equation 29.93 because pSAT(To) = po. Because of the
correction factor, Equation 29.95 also shows that the critical mass flux for a subcooled liquid is always greater than the
critical mass flux for a saturated one. Thus, the critical mass flux can vary between different points in the reactor piping
system depending on the degree of subcooling! Moreover, as we will see shortly, this conclusion applies to both equi-
librium and non-equilibrium flows. Now, let us turn our attention to non-equilibrium flows. For non-equilibrium flows,
Fauske proposed that the critical mass flux given by the ERM equation (Equation 29.93) could be properly predicted for
non-equilibrium flows by incorporating a non-equilibrium parameter N into Equation 29.93:
The non-equilibrium parameter in this case is a function of Δp = po − pb (in pascals), the discharge coefficient for the hole
CD = 0.61 if it has sharp edges, and the length of the hole in meters, ranging from 0 to 0.1m. The exact expression for N is
Hence, for a non-equilibrium flow, the discharge rate is lower because N > 1. Interestingly enough, Equation 29.96 can
also be written in terms of the L/D ratio. When this is done, we are led to the following results:
For orifice flow (L/D ≈ 0)
In this case, the residence time is zero and the flashing time is effectively zero. Equation 29.96 then reduces to the
orifice equation for an incompressible fluid:
where the value of the critical pressure is obtained from Figure 29.5.
29.18 A
pplications of the Stagnation Enthalpy, the Stagnation
Temperature, and the Stagnation Pressure
Many fluid properties required to calculate the behavior of choked flows (such as the stagnation enthalpy, the stagna-
tion temperature, and the stagnation pressure) are based on conditions that exist when a fluid is brought to rest from a
more energetic state. Normally, the stagnation enthalpy that is cited under these conditions (e.g., in Figure 29.10) is the
enthalpy after the fluid is brought to rest adiabatically, that is, without heat being added or lost. To bring a flowing fluid
into a stagnation state, any kinetic or potential energy it possesses must be used to increase its specific enthalpy, which
can be thought of as its internal energy plus its flow energy:
FIGURE 29.10 The critical mass flux predicted by a slip equilibrium model for the blowdown of a water–steam mixture through a
long unheated pipe. Here, the maximum mass flux G is plotted as a function of the fluid pressure and stagnation enthalpy at the pipe
entrance. The results shown here were measured by Fred Moody for the blowdown of a water–steam mixture in the early 1990s.
More information regarding his work can be found in The Thermal-Hydraulics of a Boiling Water Reactor, by Lahey and Moody,
Second Edition, Published by the American Nuclear Society in 1993.
1130 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
It should therefore come as no surprise that this deceleration process causes the temperature and pressure of the stag-
nating fluid to rise. The properties of the fluid in this state are then called its stagnation properties. Normally, these
stagnation properties are represented by the subscript 0. Consequently, stagnation enthalpy is written as
h o = h + v 2 2 (29.101)
where h and v are the enthalpy and the velocity of the fluid when it is in motion. When the same fluid is considered to be
an ideal gas with constant specific heats, its enthalpy can be replaced by cpT and Equation 29.101 becomes
c P To = c P T + v 2 2 (29.102)
or
To = T + v 2 2c P (29.103)
Here, To is called the stagnation temperature, and it represents the temperature that an ideal gas attains when it is
brought to rest adiabatically. The term “v2/2cP” corresponds to the temperature rise during this process, and it is called
its dynamic temperature. Thus, the stagnation temperature is the sum of the initial temperature and the dynamic
temperature:
or To = T + dynamic temperature. To understand why this distinction is so important, suppose that we compare the initial
temperature to the dynamic temperature. The dynamic temperature of air when it is moving at 200 m/s is approximately
20°C. Hence, when air at 30°C and 200 m/s is brought to rest adiabatically, its temperature rises to a stagnation value of
30°C + 20°C = 50°C. Thus, for low-speed flows, the stagnation and static temperatures are approximately the same, but
for high-speed flows, the temperature measured by a stationary probe placed in the fluid may be significantly higher than
its static temperature implies. These temperature and enthalpy differences must be taken into account in the analysis
of high-speed flows, and they are particularly important in the analysis of critical flows with high Mach numbers. The
pressure a fluid attains when it is brought to rest adiabatically is then called its stagnation pressure Po. For many gases
with constant specific heats, the stagnation pressure is related to the static pressure by
( )
γ /(γ − 1)
Po = P To T (29.104)
( )
1/( γ − 1)
ρo = ρ To T (29.105)
Using these relationships allows us to automatically account for any changes in the kinetic energy of the flow stream.
Example 29.10 illustrates the advantages to this approach.
Another example of how these relationships can be applied to a nozzle that first converges and then diverges is illus-
trated in Example 29.11. Nozzles of this type are sometimes used in rocket engines to accelerate the flow to supersonic
speeds. Here, these relationships are used to determine the Mach number at the nozzle inlet, at the choke point, and
at the nozzle outlet. For this particular problem, the combination of the temperature, the density, and the area change
29.19 Critical Flow for Gaseous Coolants 1131
actually accelerate the fluid to Mach numbers greater than 1.0. This is a classic example of what we referred to earlier as
a converging–diverging nozzle. In a nozzle with this particular design, the pressure drops quickly as we progress from
the inlet to the outlet. The flow first achieves a Mach number of unity (Ma = 1) at the throat of the nozzle where the area
is least, and it then becomes progressively more supersonic until it reaches the nozzle exit. While nozzles of this type
are rarely used in nuclear power plants, they are commonly used in jet fighter engines and rocket engines. Adiabatic flow
channels that do not converge and then diverge quickly cannot have Mach numbers greater than 1.0 because there is no
way to accelerate the fluid beyond its critical velocity.
P = ρRT (29.106)
a pressure reduction by a factor of 4 can cause the density to fall by a factor of 4 as well (see Equation 29.106). In pipes
where the flow area is constant, the velocity must increase by the same amount for the mass flow rate m to remain the
same. In frictionless flows, which are usually associated with high-speed flows through short devices with large cross-
sectional areas (such as large nozzles and short pipes), this increase can cause a subsonic gas flow to become a sonic one
because the Mach number changes so rapidly. An example of this behavior is shown in Figure 29.11, which presents the
results obtained from Example 29.11. These results are presented in Table 29.2.
A shock wave then emanates from the nozzle or pipe at the point where the flow first becomes sonic and the tem-
perature and density at this point can be determined from the equation c = √γRT where T is measured in °K, and the gas
constant R, which is usually expressed in kJ/kg °K, must be multiplied by 1,000 m2/s2 per kJ/kg to calculate the sonic
speed in m/s. For gaseous coolants, the temperatures, pressures, and densities are tabulated as a function of the Mach
number at the point of maximum contraction. For an ideal gas with no heat addition, these values are presented in
Table 29.3 as the ratio of their values at the sonic point to the stagnation point (before the fluid enters the nozzle or the
pipe). These values assume the flow is compressible and isentropic. Virtually, all compressible fluids behave in this way.
At high speeds, it becomes more convenient to express the variation in fluid properties as a function of the Mach number
directly. Properties at the sonic state are usually easy to determine, and thus, the critical state corresponding to a Mach
number of unity (Ma = 1) serves as a convenient reference point to determine the properties at other values of the Mach
number (Ma ≠ 1). For isentropic flows that are adiabatic and have negligible friction, the temperatures, the densities, and
the velocities at the sonic state can be deduced from the properties at the stagnation state using
ρ* = [ 2/(1 + γ ) ]
1/( γ − 1)
ρo (29.109)
1132 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
FIGURE 29.11 The variation of the temperature, the density, the velocity, and the Mach number of the fluid passing through the
nozzle described in Example 29.11. In this case, the fluid is assumed to have an equation of state similar to the ideal gas law.
TABLE 29.2
The Fluid Properties between the Inlet and Outlet along the Converging–Diverging Duct Described in Example 29.11
Properties between the Inlet and Outlet along the Channel Presented in Example 29.11
Pressure (MPa) Temperature (°C) Velocity (m/s) Speed of Sound (m/s) Mach Number Density (kg/m3) Flow Area (cm2)
1.4 200 0a 339.4 0 15.7 ∞
1.2 184 164.5 333.6 0.493 13.9 13.1
1.0 166 240.7 326.9 0.736 12.1 10.3
0.80 144 306.5 318.8 0.962 10.1 9.64
0.767 140 317.2 317.2 1.000 9.82 9.63
0.60 118 371.5 308.7 1.203 8.12 10.0
0.40 84 442.0 295.0 1.498 5.93 11.5
0.20 33 531.0 273.0 1.946 3.45 16.3
a Assumes a stagnation condition at the inlet.
29.20 Rayleigh Flows for Gaseous Coolants 1133
TABLE 29.3
Temperature, Pressure, and Density Ratios for a Compressible Gas as a
Function of the Mach Number
TABLE 29.4
The Critical Pressure, Temperature, and Density Ratios for Reactor Coolants When They Are Treated as Ideal Gases
The Critical Pressure, Temperature, and Density Ratios for Reactor Coolants When They Are Treated as Ideal Gases
Ratio Superheated Steam (γ = 1.3) Air (γ = 1.4) Helium (γ = 1.667) Carbon Dioxide (γ = 1.289)
P*/Po 0.545 0.528 0.487
T*/To 0.570 0.833 0.750
ρ*/ρo 0.627 0.634 0.649
where the asterisk * refers to the sonic state, and the subscript 0 refers to the stagnation state. The values for these ratios
are shown in Table 29.4 as a function of the specific heat ratio γ = CP/Cv. Notice that for superheated steam (with γ =
1.3), the critical pressure ratio (or CPR) is 0.545. In other words, when the flow becomes choked and the Mach number
becomes equal to unity, we can set P* = 0.545Po.
Pressure, Temperature, and Density Ratios for Critical Rayleigh Flows with Heat Addition
( )
2
T* = 1 + γ ⋅ Ma 2 (1 + γ ⋅ Ma) T (29.110)
P* = (1 + γ ⋅ Ma ) (1 + γ ) P (29.111)
2
TABLE 29.5
The Critical Pressure, Temperature, and Velocity Ratios for Rayleigh Gas Flow with γ = 1.4
Similarly, the velocity V at any point in the flow field is related to the sonic velocity V* by
( )
V = (1 + γ )Ma 2 1 + γ Ma 2 V* (29.113)
where Ma is the local Mach number. The values of these ratios are shown in Table 29.5 as a function of the Mach number
for a specific heat ratio of 1.4. Hence, when a flow becomes a Rayleigh flow, the properties away from the sonic condition
can be determined from the properties at the sonic condition if the Mach number at other locations is known.
( )
T = (1 + γ ) 2 + (γ − 1)Ma 2 T* (29.114)
( )
1/2
P = (1/Ma) ⋅ (1 + γ ) 2 + ( γ − 1)Ma 2 P* (29.115)
( )
1/2
V = Ma ⋅ (1 + γ ) 2 + ( γ − 1)Ma 2 V* (29.116)
and
ρ = ( V*/V ) ρ* (29.117)
29.22 The Sonic Length 1135
TABLE 29.6
The Critical Temperature, Pressure, Velocity, and fL*/D Ratios for Fanno Gas Flow with γ = 1.4
TABLE 29.7
The Critical Temperature, Pressure, Velocity, and fL*/D Ratios for Fanno Steam Flow with γ = 1.3
Equations 29.114 through 29.117 are called the Fanno functions for compressible fluid flow, and their values can be
found for air and steam in Tables 29.6 and 29.7. Hence, compressible flows with friction are often called Fanno flows
because wall friction and not just acceleration determines when and where the flow becomes sonic. For coolants that
behave like ideal gases, it is also possible to determine the length of a flow channel where the flow becomes critical if
the conditions at the inlet of the channel are known. The length of the channel where the flow first becomes choked is
then called the sonic length or critical length, and if this particular length is known, increasing the length of the channel
further will not increase the discharge rate. In fact, when the effects of wall friction are taken into account, it usually
decreases it. We will have more to say about this effect in the next section.
f for a given value of γ. This is usually done by correlating the value of fL*· De as a function of the Mach number Ma.
The appropriate values to use for air and steam are shown in Tables 29.6 and 29.7. If we extend the length of the coolant
channel beyond the critical lengths shown in Tables 29.6 and 26.7, an interesting thing happens. We simply move the
critical state further downstream and decrease the mass flow rate! This causes the inlet state to change because the inlet
velocity decreases. Increasing the length any further simply reduces the inlet velocity and the discharge rate. Thus, wall
friction plays an important role in determining the critical channel length L* in many nuclear applications. For a given
Mach number, L* is large for pipes with smooth surfaces and small for pipes with rough ones. In practice, the Reynolds
number at the inlet is used to determine the friction factors in Tables 29.6 and 29.7. The value of f is usually found from
the Moody charts or from reliable correlations such as the Haaland or Colebrook correlations, which we discussed in
Chapter 18. Roughness correction factors are then applied to these correlations to find the friction factor if the relative
surface roughness is known. For pipes, the relative roughness is usually defined by the ratio ε/D. Example 29.12 shows
how the critical length, or sonic length, can be found for a reactor coolant such as carbon dioxide. The relationship of the
critical length to the friction factor is also explored. Obviously, the same methodology can be applied to any pipe or flow
channel when the cross-sectional area is fixed. This approach can be applied to gas flows in gas-cooled reactor LOCAs
as well. Additional tables covering different specific heat ratios such as 1.2 and 1.67 are available at
http://www.potto.org/tableGasDynamics.pdf
Thus, the critical length depends on the specific heat ratio of the coolant.
The first condition is equivalent to assuming that the liquid and vapor phases have no slip, and the second condition is
equivalent to assuming that the liquid and the vapor phases are in thermal equilibrium. We would now like to extend
our analysis to problems when one (or both) of these conditions no longer applies.
TABLE 29.8
The Governing Equations of a HEM
TABLE 29.9
The Governing Equations of a Separated Flow Model
Momentum phase 1 ( )
∂ / ∂ t ⋅ ( α l ρl u1 A ) = − ∂ / ∂x ⋅ α l ρl u1 2A − ( α l ρl gA ) sin θ − τ w A w − τ i A i L i − α l ⋅ ∂ / ∂x ⋅ (pA)
Momentum phase 2 ( )
∂ / ∂ t ⋅ ( α 2ρ2 u 2 A ) = − ∂ / ∂x ⋅ α 2ρ2 u 2 2A − ( α 2ρ2 gA ) sin θ − τ w A w − τ i A i L i − α 2 ⋅ ∂ / ∂x ⋅ (pA)
See http://wins.engr.wisc.edu/teaching/mpfBook/node14.html.
Note: This table considers only two phases at the same time, and the frictional dissipation terms (viscous and turbulent) have been
neglected similar to the approach used by most applications.
29.28 Equations Used by Separated Flow Models 1139
numerically, but numerical solutions are usually best when energy is added to or subtracted from the flow. Homogeneous
equilibrium models are easy to use, and when applied to reactor subchannels, the physical properties of the fluid are
averaged over the cross-sectional area of the subchannels. This spatial averaging scheme results in a set of partial dif-
ferential equations that have a temporal domain, t, and a space domain, z. Now, let us take a look at how these equations
are implemented by reactor designers.
DRAG lv = τ lv Ai Li (29.118)
where τlv = u·|vl − vv| and τlv is the interfacial shear stress between the liquid and the vapor. Hence, most separated flow
models assume the momentum exchange between the phases is proportional to their relative velocity difference Δv = |vl –
vv|. The relaxation of the equal velocity assumption is the most important feature of separated flow models. Relaxing this
TABLE 29.10
The Governing Equations of a Two-Fluid Model
Momentum phase 2 ( )
∂ / ∂ t ⋅ ( α 2ρ2 u 2 A ) = − ∂ / ∂x ⋅ α 2ρ2 u 2 2A − ( α 2ρ2 gA ) sin θ − τ w A w − τ i A i L i − α 2 ∂ / ∂x ⋅ (pA) + Γ M2
assumption becomes particularly important when the liquid and vapor phases begin to move in different directions. For
example, the liquid phase may move downward and the vapor phase may move upward in the annulus of a reactor steam
generator. This is also important when the coolant pumps are turned off, and the flow enters what is called a natural circu-
lation condition. Moreover, buoyancy-related effects can induce a drift velocity between the lighter phase and the heavier
one. One indicator of the amount of drift that occurs under these conditions is the Atwood number, At, which is defined by
The Atwood number (At) is used to study hydrodynamic instabilities in stratified flows. As the Atwood number
approaches zero, the HEM becomes more accurate because the drift velocity is reduced as the buoyancy of the lighter
phase is diminished. The relative motion between the phases then creates an additional pressure drop as well. This
pressure drop is the source of the two-phase friction multiplier M that we first introduced in Chapter 23. Like the
Reynolds number and the Prandtl number, the Atwood number is also a dimensionless number. However, because of
the way it is defined, it can never be less than zero or greater than one. Normally, stratified flows are studied using four-
equation models.
The Fourier number also appears in the study of transient heat conduction, which was first discussed in Chapter 12.
In Equation 29.120, α = k/(cpρ) is called the thermal diffusivity (which is measured in m2/s), and L is the characteristic
length of the system. Then, thermal non-equilibrium effects become progressively more important at high Fourier
numbers, and one must account for these effects by writing a separate energy equation for reach phase. Energy is then
transferred between the phases in proportion to their relative temperature difference.
E lv = q lv′′ A i L i (29.121)
which allows heat to flow between the vapor and the liquid until thermal equilibrium is reached. The exact form of the
interphase heat flux q′′lv in Equation 29.121 is
q′′lv = h lv ( Tv – Tl ) (29.122)
where hlv is the heat transfer coefficient between the liquid and vapor phases. Notice that the flow of heat between the
phases only ceases when Tv = Tl, and a state of thermal equilibrium has been reached. Hence, heat generally flows from
Bibliography 1141
the vapor energy equation to the liquid energy equation. In computer programs like RELAP 5, thermal equilibrium and
non-equilibrium are simulated by multiplying the equation for q′′lv by a relaxation factor called f. When equilibrium is
on, f = 0, and q′′lv = 0, irrespective of the values of Tv and Tl. Otherwise, q′′lv will have a finite value, and energy will flow
from the vapor phase to the liquid phase. Sometimes there is a time lag associated with this process, and this time lag is
usually modeled by adjusting the value of hlv.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Web References
http://www.potto.org/tableGasDynamics.pdf.
https://www.ansys.com/products/fluids/ansys-fluent.
https://mysite.du.edu/~jcalvert/tech/fluids/orifice.htm.
http://wins.engr.wisc.edu/teaching/mpfBook/node14.html.
www.iaea.org/inis/collection/NCLCollectionStore/_Public/27/008/27008531.pdf.
Other References
Fauske, H.K. “Two Phase Critical Flow” paper presented at the MIT Two-Phase Gas-Liquid Flow Special Summer Program, 1964.
Fauske, H.K. “The discharge of saturated water through tubes”, Chemical Engineering Symposium Series, 61:210, 1965.
Fauske, H.K. “Flashing flows – some practical guidelines for emergency releases”, Plant Operations Progress, 4:132, 1985.
Moody, F.J. “Maximum two-phase vessel blowdown from pipes”, The Journal of Heat Transfer, 88:285, 1966.
Murdock, J.W. “Two-phase flow measurements with orifices”, Transactions of the ASME, The Journal of Basic Engineering,
82:419–433, 1962.
Perry, R.H. and Green, D.W. “Perry’s Chemical Engineers’ Handbook”, Table 2–166, Sixth edition, McGraw-Hill Company, New
York (1984). ISBN 0-07-049479-7.
Phillips Petroleum Company. “Reference Data for Hydrocarbons and Petro-Sulfur Compounds”, Second printing, Phillips Petroleum
Company, Bartlesville, OK (1962).
Richter, H.J., “Separated two-phase flow model: Application to critical two phase flow, International Journal of Multiphase Flow,
9(5):511–530, 1983.
1142 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
38. What is the difference between a HEM, a separated flow model, and a two-fluid model? How many equations does
each of these models usually have?
39. Which of these models is usually the most comprehensive in terms of its scope and its overall range of applicability?
40. How is the interfacial drag coefficient in a two-fluid model usually defined?
41. How many sound speeds can a two-phase mixture actually have?
42. Does a flow have to become fully developed to become an equilibrium flow?
43. What is the equation of state called for an ideal gas?
44. What is the definition of the isentropic expansion factor?
45. Write an expression for the CPR for a coolant such as helium in a gas reactor.
46. Write an expression for the speed of sound in a two-phase mixture.
47. Write an expression for the CPR for single-phase flow through a coolant channel.
48. Under the same physical conditions, is the discharge rate through a “short” orifice the same, larger, or smaller than
the discharge rate through a “long” orifice.
49. Approximately how long does a pipe have to be before a slip equilibrium model can be used?
50. Name two common slip equilibrium models that can be used to find the critical flow rate in a reactor coolant channel.
51. Write an expression for the pressure drop for a homogeneous two-phase mixture. Does the pressure drop increase,
decrease, or stay the same as the equilibrium quality is increased?
52. How is the interfacial heat transfer coefficient in a two-fluid model usually defined?
53. Do cracks with rounded edges have higher or lower CPRs than cracks with sharp edges?
54. In critical flow models, what does the term “stagnation enthalpy” refer to?
55. Name a famous separated flow model that is used in RELAP and TRACE to account for the effects of thermal
non-equilibrium on the critical flow rate.
56. What effect does the residency time of a fluid in a reactor coolant channel have on whether or not a flow becomes
an equilibrium flow?
57. Write an expression for the effective density of a two-phase mixture in a slip equilibrium model.
58. In very long reactor coolant pipes, the exit pressure required for choked flow to occur becomes a certain p ercentage
of the upstream pressure. Exactly, what percentage of the upstream or stagnation pressure does this happen to be?
(
dA/A = dP ρV 2 1 – (V/c)2 )
where c is the speed of sound of the fluid flowing through the pipe. When the flow is subsonic, how does the pressure
change when the area increases and how does the pressure change when the area decreases?
Exercise 29.2
Using some simple thermodynamic relationships and the continuity and energy equations, show that a shock wave
propagates through a compressible fluid with a speed that is given by c2 = (∂P/∂ρ)s where c is the speed of the wave.
Assume the flow is isentropic. What is a popular name for the value of c? What is the value of the Mach number Ma
when the velocity of an object is equal to c?
Exercise 29.3
A coolant pipe 0.3 m in diameter carries pressurized water from the steam generators to the reactor pressure vessel. The
pressure of the water in the pipe is 2,000 PSI (13.8 MPa), and the temperature of the water is 290°C. The pipe suddenly
breaks cleanly 1 m from the pressure vessel. What is the initial rate of discharge through the pipe?
Exercise 29.4
For gas reactors, the relationship c2 = (∂P/∂ρ)s can also be written as c2 = γ(∂P/∂ρ)T where γ is the specific heat ratio of the
gas (γ = Cp/Cv). For helium and carbon dioxide, what is the value of γ? For an ideal gas, show that the value of c can also
1144 EQUILIBRIUM AND NON-EQUILIBRIUM FLOWS, COMPRESSIBLE FLOWS, AND CHOKE FLOWS
be written as c = √γRT where R is the molar gas constant. If the temperature of the coolant is 1,000°C, and the coolant
is carbon dioxide, what is the value of c?
Exercise 29.5
A coolant pipe 0.3 m in diameter carries pressurized water from the reactor pressure vessel to the steam generators. The
pressure of the water in the pipe is 2,250 PSI (15.5 MPa), and the temperature of the water is 320°C. The pipe suddenly
breaks 10 m from the reactor pressure vessel. The break is clean and perpendicular to the pipe axis. The back pressure
is atmospheric. What is the rate of discharge from the pipe at the instant the break occurs?
Exercise 29.6
A small hole develops in the core shroud of a boiling water reactor after the coolant pumps have been turned off. The
hole has a diameter of 0.001 m. A two-phase mixture at 5,515 kPa (800 PSI) with an equilibrium quality of 10% spills
into the downcomer. The pressure in the downcomer is maintained at 5,508 kPa (799 PSI). The coefficient of discharge
through the hole is assumed to be 0.6. What is the rate of spillage into the downcomer?
Exercise 29.7
A flow restrictor in the form of a concentric disk is placed in a pipe to change the pressure drop. Suppose the pipe has
an inner diameter of 0.02 m and the disk is a concentric disk with an inner diameter of 0.01 m. The flow rate through
the pipe before the restrictor is installed is 1 kg/s. What is the coolant velocity at the center the restrictor? Assume the
coolant is water and the pipe is maintained at atmospheric pressure. What is the pressure drop across the restrictor if it
is only 0.001 m long?
Exercise 29.8
A flow channel with a cross-sectional area of 0.001 m2 suddenly develops a small obstruction that cuts the flow area in
half. The obstruction is very thin and behaves like an orifice. Upstream of the obstruction, saturated water flows through
the pipe at 800 PSI (5.5 MPa) at a velocity of 1 m/s. Calculate the pressure drop caused by the obstruction if the coef-
ficient of discharge is Cd = 0.6.
Exercise 29.9
During a small pipe break LOCA, the flow becomes choked. If the system pressure at the choke point is 15 MPa, and
the flow discharges into a reactor containment building where the ambient pressure is 50 kPa, calculate the slip ratios
between the liquid and vapor phases just downstream from the choke point. Calculate these slip ratios using Fauske’s
and Moody’s models. In which model is the vapor velocity higher?
Exercise 29.10
An ERM can be used to predict the critical mass flux when hot water flashes into steam through a short pipe, such as one
attached to a safety relief valve. Suppose a pipe of approximately half a meter long and 0.2 m in diameter is attached to
a safety relief valve in the secondary loop of a commercial PWR. The water in the vicinity of the safety relief valve is
saturated at the time the valve opens, and the valve discharges some of this water through the pipe where it immediately
flashes to steam. Using Fauske’s approach, determine the critical flow rate through the pipe.
Exercise 29.11
Superheated steam having a Reynolds number of one million is discharged through a 10 cm-diameter pipe at a tempera-
ture of 280°C. What is the Mach number at the entrance to the pipe, and how long does the pipe have to be before the
flow becomes choked? If the pipe is made longer, what happens to the mass flow rate?
Exercise 29.12
A coolant with a specific heat ratio of 1.4 escapes from a gas reactor through 5 cm-diameter pipe whose inner surface is
very smooth. If the temperature of the coolant is 600°C, what is the maximum possible velocity the coolant can attain in
the pipe? If the Mach number at the entrance to the pipe is 0.2, what is the turbulent friction factor for the coolant? How
long does the pipe have to be before the flow becomes restricted or choked?
30
Reactor Accidents, DBAs, and LOCAs
30.1 D esign Limits for Reactor LOCAs
The most serious design basis accident (or DBA) that can occur in a nuclear power plant is a loss-of-coolant accident or LOCA. The
probability of most LOCAs occurring is relatively small (see Chapter 33), and since the 1980s, not a single LOCA has occurred in
a commercial nuclear power plant that has led to significant damage to the core. The accident at Three Mile Island (TMI) in 1979
involved a stuck pressure relief valve on the pressurizer (which controls the pressure in the primary loop), but if the sticky valve had
been diagnosed properly, even this LOCA would not have occurred. In the United States, numerous design limits are imposed by
regulatory authorities to ensure that hypothetical LOCAs are properly anticipated, categorized, and contained. These limits include
1. Thermal design limits
2. Structural design limits
3. Oxidation or chemical design limits.
Some of the most important of these design limits are summarized in Table 30.1. Pressurized water reactors (PWRs) and boiling
water reactors (BWRs) also have hydrodynamic stability limits that we will elaborate upon in Chapter 31. Regulatory authorities also
impose rather strict thermal and structural limits on the response of the cladding and the fuel. These limits can take the form of strain
limits, fatigue limits, and oxidation limits. In light water reactors, the cladding temperature is also required to stay below 2,200°F
(or 1,204°C) to prevent the zirconium in the cladding from oxidizing, which can lead to what is called the zirconium–water reaction
(see Chapter 34). Sometimes this reaction is also called the metal–water reaction.
TABLE 30.1
Typical Fuel Pin Thermal and Structural Design Limits during Normal Operation and Reactor Accidents
1145
1146 REACTOR ACCIDENTS, DBAs, AND LOCAs
In general, Equation 30.1 describes a highly exothermic reaction that produces more energy than it consumes. Above
1,200°C (~2,200°F), the energy produced by combining the zirconium and the water together is about 580 kJ/mol of
zirconium metal or about 6.3 MJ/kg. In 2011, the zirconium–water reaction was ultimately responsible for the contain-
ment building failures that occurred in reactors 1 and 3 at the Fukushima site in Fukushima, Japan. The hydrogen
ignited, and this is what caused the explosions in the containment buildings to occur.* For these reasons and others, the
cladding temperature is limited to 2,200°F (or 1,204°C) during commonly hypothesized LOCAs. This helps control
the oxidation rate and it also serves to limit the amount of hydrogen gas that is produced. Moreover, it helps to maintain
the structural integrity of the cladding so that the fuel rods can reach burnups of at least 50,000 MWD/T without failing.
* See Nuclear Engineering Fundamentals: A Practical Perspective by Robert E. Masterson, CRC Press (2017).
30.7 Visualizing a Reactor LOCA 1147
In other words, a LOCA is essentially a leak in the reactor’s primary or secondary loops that results in a leak rate beyond
the normal makeup capacity of the plant. There are several ways that LOCAs can occur, and some types of LOCAs are
more likely to occur than others. In PWRs, the most limiting LOCA is currently believed to be a double-ended guillo-
tine pipe rupture of one of the primary coolant loops in the cold leg, and in BWRs, the most limiting LOCA is a rupture
of one of the recirculation loops. On average, large pipe break LOCAs are believed to occur, once every 100,000 to
1,000,000 years of anticipated operation. Small pipe break LOCAs are assumed to occur more frequently.* The frequency
of a particular LOCA occurring is estimated using a statistical model that is based on the assumption that certain reactor
components will fail or that the operators will make inappropriate or unwise operating decisions at some point during a
plant’s operating life. The probability of occurrence is then calculated from a statistical analysis of hundreds or even thou-
sands of different fault trees and event trees. Some examples of these event trees are provided in Chapter 33. The statistical
probability distributions that are used to arrive at these probabilities of occurrence can then be symmetrical or asym-
metrical in shape.† The probabilities of LOCAs and other hypothetical DBAs occurring in commercial nuclear power
plants in the United States were initially calculated in a famous government report called the WASH-1400 report, which
was published in 1975. The findings of this report are discussed in Chapter 33. Using an approach called probabilistic risk
assessment or PRA, the WASH-14000 report determined that PWR LOCAs have different probabilities of occurring than
BWR LOCAs. In addition, the radiological consequences of these LOCAs can be considerably different as well.
* See http://www.iaea.org/inis/collection/NCLCollectionStore/_Public/28/021/28021737.pdf.
† See An Introduction to Nuclear Reactor Physics by Robert E. Masterson, CRC Press (2017).
1148 REACTOR ACCIDENTS, DBAs, AND LOCAs
FIGURE 30.1 A diagram showing a hot water–steam mixture being ejected from a crack in a coolant pipe during a typical LOCA.
Here, the reactor is assumed to be a PWR. (Picture drawn by the author.)
process. Immediately after the core is scrammed, the heat generation rate drops to about 7% of its steady state value, and
after about an hour, it falls another 6% to about 1% of its original value (see Chapter 5). A coolant leak can also occur in
one of the pipes in the secondary loop that connect the steam generators to the steam turbines. However, the steam in the
feed lines is not radioactive and the reactor can continue to operate normally if the pipe break is isolated and contained.
Normally, the most limiting LOCA in the primary loop occurs immediately before the coolant enters the core. This LOCA
is sometimes called a cold leg rupture or cold leg leak. In a BWR, this is equivalent to the loss of a r ecirculation loop.
TABLE 30.2
LOCA Category Definitions
LOCA Category Flow Rate Threshold, gpm (lpm) LOCA Classification
1 >100 (380) SB
2 >1,500 (5,700) MB
3 >5,000 (19,000) LB
4 >25,000 (95,000) LB a
5 >100,000 (380,000) LB b
6 >500,000 (1,900,000) LB c
Note: Here, gpm refers to gallons per minute, and lpm refers to liters per minute.
TABLE 30.3
Break Size to Leak Rate Correlations
break develops in the secondary loop of a PWR, the initial pressures will be lower. The blowdown stage ends when there
is no longer a significant pressure difference between the inside and the outside of the cracked pipe.
TABLE 30.4
The Criteria in Appendix K of 10 CFR 50 for Limiting the Consequences of a Water Reactor LOCA
TABLE 30.5
The Time Lines and Sequences of Events for a Best Estimate Large Pipe Break LOCA
TABLE 30.6
The Projected Peak Cladding Temperatures during Each Phase of a Large Pipe Break LOCA
Core Component Blowdown Peak First Reflood Peak Second Reflood Peak
Average peak cladding temperature <1,508°F (820°C) <1,681°F (916°C) <1,384°F (751°C)
Top cladding temperature (hottest 5%) <1,760°F (960°C) <1,976°F (1,080°C) <1,964°F (1,073°C)
Maximum cladding oxidation <11%
Total cladding oxidation <0.90%
The cladding melting point is about 1,845°C for Zircaloy-2 (which is used in BWRs) and 1,880°C for Zircaloy-4 (which is used in
PWRs).
30.13 PRIMARY OR LB LOCAs 1153
FIGURE 30.2 A sketch of the PCCS in a modern PWR. This specific design is used in the Westinghouse AP-1000.
FIGURE 30.3 The pressure in a PWR pressure vessel during a large break LOCA. The depressurization of the pressure ves-
sel only takes about 25 seconds, at which point the LOCA is assumed to be officially over. For small break LOCAs, the same
depressurization process can take several minutes to several hours. Thus small break LOCAs behave in an entirely different
way than large break LOCAs do.
1154 REACTOR ACCIDENTS, DBAs, AND LOCAs
TABLE 30.7
The Time Line for a Standard Large Pipe Break LOCA
TABLE 30.8
The Time Line for a Standard Small Pipe Break LOCA
break LOCA, it only takes about half a minute for the core to completely empty. In a very small pipe break LOCA, the same
process could take several hours. The sequence of events following the beginning of the pipe break is shown in Table 30.7. Of
course, these numbers depend on the design of the individual plant. After about 5 minutes, the LOCA is assumed to be over.
FIGURE 30.4 The active and passive subsystems in the ECCS of a commercial PWR.
requires no electrical power to operate. The passive subsystem consists of several large water storage tanks, or accu-
mulator tanks, with at least one tank being devoted to the primary loop. These tanks contain borated water which is
maintained at a pressure above 4.5 MPa (~650 PSI). When the pressure in the reactor pressure vessel falls below 4.5 MPa,
the reversal of pressure across a check valve causes the tanks to open and discharge their coolant into the cold leg of the
primary loop. The cold legs then send the water directly into the core. This accumulator injection is shown in Figure 30.4.
The active safety subsystem requires electrical power to operate, and it is in turn subdivided into a low-pressure injection
system and a high-pressure injection system. The low-pressure injection system takes water from a storage tank (which is
normally located inside of the containment building), heats up the water using a heat exchanger, and injects the water into
one of the hot legs in the primary loop. From there, it flows directly into the core. Following the normal shutdown of the
reactor for maintenance or refueling, the low-pressure injection system is also used to remove decay heat from the core.
Hence, it serves as both an emergency function and as necessary maintenance function. The high-pressure injection sys-
tem performs a similar function, but it is activated much earlier when the coolant pressure in the reactor pressure vessel is
still high. The high-pressure injection system uses another set of pumps (called charging pumps) which are used to inject
soluble boron into the coolant while the reactor is still critical. The boron supplied by this system is used to maintain the
correct neutron production rate in the core. When a signal is received to begin the high-pressure injection process, these
pumps circulate additional coolant from a water storage tank through the boron shim tank and into the cold leg of the
primary loop. The function of the active and passive injection systems is shown in Figure 30.4.
by a pump which is driven by high-pressure steam extracted from the main steam line. This water is initially provided
by a large water storage tank located inside of the containment building, and then, it is provided by the suppression pool.
The additional water enters the pressure vessel from the feedwater system or from a ring of nozzles located directly
above the core. The nozzles spray the makeup water directly onto the core. The exact path that the coolant follows var-
ies from one type of BWR to the next. If the high-pressure injection system fails to maintain the proper water level, the
reactor pressure vessel will begin to depressurize. The rate of depressurization depends on the size of the pipe break.
The reactor pressure vessel is depressurized by opening some pressure relief valves on the main steam line. Normally,
these valves operate automatically. The steam from the main steam line enters the dry well (see Chapter 3) and is then
diverted into the suppression pool (see Figure 30.5) where it is condensed. If the pipe break is large enough, the depres-
surization occurs without the pressure relief valves opening. After the reactor pressure vessel has been depressurized,
the low-pressure injection system and the low-pressure core spray system are used to cool the core. The core spray sys-
tem uses pumps to carry the water to the core from the suppression pool. The low-pressure injection system uses another
set of pumps to pump water from the suppression pool directly into the reactor recirculation loop. The water flowing out
of the broken pipe is collected in the suppression pool and is recycled back into the core. During normal operation, the
low-pressure injection system also removes decay heat from the core. Normally, this decay heat does not exceed 7% of
the full power output of the plant. Again, the goal of the ECCS is to keep the fuel rods cool and to prevent a large number
of the fuel rods from failing. In general, it is not realistic to assume that more than 10% of the fuel rods will fail if the
ECCS works properly. We will have more to say about these numbers in a subsequent section. The containment building
then attempts to keep the radiation released by the rods inside of the plant.
rushes out of the break and then flashes to steam. Almost immediately after this, the pressure in the reactor vessel drops
to about 650 PSI (~4.5 MPa), and the injection of cold water from the accumulators begins. A mixture of steam and water
flows out of the break until pressure in the reactor pressure vessel and in the containment building becomes about the same.
Normally, the pressure at this point is about 60 PSI. This then corresponds to the end of the blowdown phase.
FIGURE 30.6 The peak cladding temperature as a function of time during a large pipe break LOCA. Note: During these LOCAs,
the fuel temperatures are approximately 1,000°C higher than the cladding temperatures, and there is a slight time lag between the
cladding temperatures and the fuel temperatures.
1158 REACTOR ACCIDENTS, DBAs, AND LOCAs
When the core is reflooded and refilled again, the cladding and fuel pin temperatures spike for a second time and they
normally spike at higher temperatures than are observed during the blowdown phase. These higher temperatures are
due to the fact that some of the fuel rods may no longer be covered with a liquid film, which means that the heat transfer
coefficients at the rod surface are lower than they are in the blowdown phase. In most credible scenarios, this causes the
average cladding temperature to approach 915°C (1,680°F) and the highest cladding temperature to approach 1,200°C
(~2,200°F). While the cladding still does not melt under these conditions, it is possible for some of the hottest rods in
the core to fail due to thermally induced stresses. Like the cladding temperature, the peak fuel temperature also occurs
about 3 minutes after the LOCA begins. Then, the fuel temperature begins to slowly subside again. Most licensing
calculations are usually performed using very conservative assumptions that normally overestimate the fuel and clad-
ding temperatures during an LB LOCA. In the FSAR, these assumptions are described in excruciating detail, and a
much less conservative analysis is also provided. Best estimate calculations for the cladding temperatures are usually
100°C–200°C below the temperatures that the licensing calculations use. An example of the conservatism built into
these calculations is shown in Figure 30.6. Hence, regulatory authorities and reactor vendors have gradually reduced
some of the restrictions on the fuel and cladding temperature limits during LB LOCAs. However, at least in the United
States, the current fuel and cladding temperature limits (2,200°C and 1,200°C, respectively) are still used to license
commercial nuclear power plants.
TABLE 30.9
Computer Programs Used to Perform Nuclear Fuel Rod Behavioral Studies in the World Today
Common Computer Programs Used to Perform Nuclear Fuel Rod Behavioral Studies
Name of Computer Program or Code Country of Origin or Country Where Program Is Primarily Used
Frapcon and Bison United States
Elestres (CANDU reactors only) Canada
Enigma United Kingdom
Copernic China
Rtop Russia
Transuranus Italy
Source: Todreas and Kazimi (2008).
FIGURE 30.7 A process used to predict fuel rod failure rates during a reactor LOCA.
codes are then “chained together” to take the output from one computer program and load it into the input of another. A
typical flowchart for a deterministic calculation of the fuel failure rates is presented in Figure 30.7. The failure thresh-
old for each specific rod is calculated as a function of the rod power and the burnup. Most of the time, higher powered
fuel rods with higher burnups have a much higher probability of failing than lower powered fuel rods with much lower
burnups do. In contrast to a truly mechanistic model, the empirical approach that the NRC uses to predict rod failures
1160 REACTOR ACCIDENTS, DBAs, AND LOCAs
in NUREG-0630 ignores the effects of cladding creep. Consequently, it provides a relatively simple and straightforward
relationship between the mechanical load on the undeformed cladding and the rupture temperature of the cladding. The
specific form of the correlation that is presented in NUREG-0630 is
where
☉☉ TRUPTURE is the rupture temperature of the rod (in °C).
☉☉ σ is the hoop stress (in kpsi).
☉☉ H is the ramp ratio (°C/s to 28°C/s).
and the hoop stress σ is computed from the following formula:
σ = (D/2L) ⋅ ∆p
where
☉☉ D is the deformed cladding mid-wall diameter (in inches).
☉☉ L is the undeformed cladding thickness (in inches).
☉☉ Δp is the differential pressure (in kpsi) across the cladding wall at the time of rupture.
Needless to say, this correlation is extremely conservative. It uses a limit-line approach to associate the failure of a
p articular fuel rod with the lowest possible value of the cladding temperature that is consistent with the rupture of the
rod. Nevertheless, it is currently the “politically correct” way to estimate the percentage of the fuel rods that will fail
during a reactor LOCA in the United States. At some point in time, the approach that the NRC uses in the United States
will probably be updated to resemble the approach used in Germany and most of Europe. In any event, it is safe to
assume that no more than 10% of the fuel rods will fail during most credible LOCAs.
FIGURE 30.8 Probability of a fuel rod failing as a function of the linear power q′ at different burnups.
fail higher up in the core, but the temperature profiles between a PWR and a BWR can be completely different. This is
due to the fact that some boiling in a BWR can occur along roughly three-quarters of the length of the fuel rods (during
steady-state conditions). During a LOCA, the progression of the boiling front in a BWR and a PWR is also completely
different. During steady-state conditions (prior to the LOCA), subcooled boiling is allowed in the hot channel of a PWR,
but generally only in the top 10% of the core. Thus, the LOCA in a BWR starts with a significant amount of bulk boiling,
whereas the LOCA in a PWR it does not.
Zr + 2H 2 O → ZrO 2 + 2H 2
1162 REACTOR ACCIDENTS, DBAs, AND LOCAs
This reaction can occur at many different temperatures. However, the speed of the reaction is greatly accelerated if tempera-
ture of the cladding rises above 500°C. This reaction is highly exothermic (meaning that it will release additional heat after
it begins), and it also produces a great deal of hydrogen gas. Although it occurs mostly at very high temperatures (such as
those that can be found during a LOCA), it closely resembles the kind of reaction that can occur when an alkaline metal (such
as sodium or potassium) is put in a bucket of water. This reaction was responsible for a small hydrogen gas explosion that
occurred inside the reactor containment building at TMI in 1979. This same reaction also occurred in several of the BWRs
at Fukushima, Japan, in 2011 after the cooling pumps to reactors 1, 2, and 3 were shut off. After the hydrogen gas was vented
into the Mark I containment buildings, the hydrogen gas detonated and severely damaged several of the service buildings
and at least one of the containment buildings. The hydrogen igniters that were present in the buildings did not function in
the way they were intended because the electrical power from the onsite diesel generators and from the primary power grid
was lost. In many PWR containment buildings, a catalyst-based recombination system is used to convert the hydrogen gas
and the oxygen into liquid water at room temperature before an explosion can occur. However, this type of recombination
system was not installed in the containment buildings at Fukushima, Japan, which used the old Mark I design. Under most
conditions, the dependence of the oxidation rate R on the temperature and the pressure can be expressed as
The oxidation rate R is generally very small (1 × 10 −20 g/m2/s) at 0°C, 6 × 10 −8 g/m2/s at 300°C, 5.4 × 10 −3 g/m2/s at
700°C, and 300 × 10 −3 g/m2/s at 1,000°C. Normally, there is no clear threshold where the oxidation rate becomes apoca-
lyptic, but during a LOCA, it can become quite significant above 600°C. At normal cladding temperatures of 300°C, it
is generally not considered to be a big deal. However, to keep this reaction from generating serious amounts of hydrogen
gas, the U.S. NRC limits the maximum permissible cladding temperature during a LOCA to 2,200°F (1,204°C). There
are actually two specific reactions that can occur between the hydrogen atoms in the surrounding water molecules and
the zirconium metal. One reaction is with very hot water, and this leads to a corrosive oxide film that builds up on the
surface of the cladding. However, the film is usually very thin as long as the temperature of the cladding is kept below
300°C. The second and more important reaction occurs when the zirconium metal interacts with steam. This can lead
to the production of large amounts of hydrogen gas, and it can also lead to the possible structural failure of the cladding.
This is clearly an exothermic reaction, and additional energy will be released as the zirconium corrodes. However, at
300°C, the oxidation rate is only 6 × 10 −8 g/m2/s. Although there is no exact consensus on whether zirconium and the dif-
ferent alloys of zircaloy that can be found in BWR and PWR fuel rods have exactly the same oxidation rate, Zircaloy-2
(used in BWRs) and Zircoloy-4 (used in BWRs and PWRs) appear to behave in a similar way.
This is also an exothermic reaction, but more energy is released (about 10% more per mol than at the temperature of
the water during normal operation), and the oxidation rate per second is about one million times higher. Hence, at the
maximum cladding temperature of ~1,204°C, the amount of hydrogen gas that is produced can be quite significant.
dR/dt = B ( R − R o ) e − G/T
where R is the oxidation rate, B and G are empirically determined constants, Ro is a reference oxidation rate, and T is the
surface temperature of the cladding. The specific form of the Baker–Just correlation that is used in RELAP5 is
where
☉☉ R is the oxidation rate in g/(cm2 s).
☉☉ T is the temperature of the outer surface of the cladding (in °K).
More information regarding the Baker–Just correlation can be found in the following reference, which outlines the
approach Baker and Just used to arrive at their original correlation:
☉☉ L. Baker, Jr. and L.C. Just, “Studies of Metal-water Reactions at High Temperatures,” ANL-6548, May 1962.
The amount of radioactivity that is released during a LOCA is directly proportional to the number of the fuel rods that
fail, and how long the fuel rods have been in the core. More highly burned fuel rods tend to release more fission products
than lightly burned fuel rods. As a result of this, two basic methods are used to predict the release of radioactive mate-
rials during a LOCA. The first model assumes that all of the fuel rods will fail and that all of the radioactivity within
the rods is released into the coolant. The second model assumes that only a fraction of the fuel rods will fail (usually
between 1% and 10%) and that the rest of the radioactivity will remain in the core. In the next section, we would like to
discuss the environmental implications of each of these models. These models also distinguish between the noble gases,
which can move around freely and which are generally all released, and heavier radioactive materials, like the iodine
inventory, various types of particulate matter, and the transuranics, which may be only partially released under certain
conditions.
TABLE 30.10
Important Temperature Thresholds in an American PWR
Important Temperature Thresholds for Normal Operation and LOCAs in an American PWR
Event Temperature (°C)
Water in a PWR begins to undergo subcooled nucleate boiling 330
Water in a PWR reaches its saturation temperature 345
Maximum cladding temperature is reached during normal operation ~330 (outer surface); ~400 (inner surface)
The zirconium–water reaction becomes important 1,200
Steel fuel rods start to melt 1,450
The zirconium cladding loses its structural integrity 1,600
The zirconium–steam reaction can become apocalyptic 1,600
Various alloys of zirconium (zircaloy) begin to melt 1,900
Most of the fission gases are released from the fuel 2,300
The core structure melts and loses its structural integrity 2,500
The UO2 in the fuel rods melts 2,700
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi, India (2000).
Reactor Theory (Nuclear Parameters), DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
Web References
h t t p s : / / d s p a c e. m i t . e d u / b i t s t re a m / h a n d l e / 1 7 2 1 . 1 / 4 5 5 3 3 / 2 2 - 3 9 Fa l l - 2 0 0 5 / N R / rd o n l y re s / N u c l e a r- E n g i n e e r i n g /
22-39Fall-2005/9ABDFB04-CDB9-402B-B5BB-4E9263DF1C7E/0/lec9_nt.pdf.
http://www.iaea.org/inis/collection/NCLCollectionStore/_Public/28/021/28021737.pdf.
https://www.nrc.gov/docs/ML0715/ML071580895.pdf.
Other References
Baker, Jr., L. and Just, L.C., “Studies of Metal-water Reactions at High Temperatures,” ANL-6548, May 1962.
Cathcart, J.V., et al., “Zirconium Metal-Water Oxidation Kinetics IV. Reaction Rate Studies,” ORNL/NUREG-17, August 1977.
“Improvement of Computer Codes used for Fuel Behavior Simulation,” IAEA-Tecdoc-1697, International Atomic Energy Agency,
Vienna, Austria, 2013.
Powers, D.A. and Meyer, R.O., “Cladding swelling and rupture models for LOCA analysis,” NUREG-0630, April 1980.
“Reactor Safety Study: An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants,” WASH-1400, U.S. Nuclear
Regulatory Commission, October 1975.
32. Why did the WASH-1400 report only provide five categories of accident severity for BWRs, while it provided nine
categories of accident severity for PWRs?
33. Can PWR LOCAs become more energetic than BWR LOCAs? Do these differences have anything to do with
differences in the plant pressure levels and power densities?
34. Following the release of WASH-1400, the AEC at the time (and then the NRC following it) began formulating a
number of design criteria to specify the “ground rules” under which a LOCA would be considered to be s uccessfully
contained. Can you name five of these “ground rules” that later became design requirements in Appendix K of
10 CFR 50?
35. From the perspective of a reactor designer, an LB LOCA can be subdivided into five distinct phases. What are the
names of these phases and in what order do they occur?
36. In what phase does coolant exit the primary or secondary loops at a high rate of speed?
37. What is this rapid depressurization of the NSSS called?
38. When high-pressure water immediately turns itself into steam as it leaves the NSSS, what is this process called?
39. How long does this process continue to occur? Is the flow sonic or choked during most of this time?
40. In what phase is the core refilled and returned to a stable state?
41. For a PWR, what are the approximate time frames associated with each of these five phases?
42. For a BWR, what are the approximate time frames associated with each of these five phases?
43. How many seconds after the blowdown begins in a PWR does the blowdown end? Assume a large pipe break LOCA.
44. When during this rapid depressurization of the NSSS does the low-pressure injection system begin to inject cold
coolant back into the core?
45. What is the role of the ECCS during this core stabilization process?
46. How many subsystems does the ECCS contain?
47. How many of these subsystems are active, and how many of these subsystems are passive?
48. Approximately when is the peak cladding temperature reached during an LB LOCA?
49. What is the maximum allowable cladding temperature under these conditions?
50. During a large pipe break LOCA, what percentage of the fuel rods are assumed to fail by regulatory authorities in
the United States?
51. During a large pipe break LOCA, what percentage of the fuel rods are assumed to fail by regulatory authorities in
Germany, France, England, and Japan?
52. How do these failure probabilities affect the amount of radioactivity that is released into the containment building?
53. What is the purpose of a reactor suppression pool?
54. Are these pools installed today in most PWRs or BWRs?
55. What is the economic justification for building a suppression pool?
56. By what percentage can a well-designed suppression pool reduce the peak air–steam pressures in a commercial
nuclear power plant following a large pipe break LOCA?
57. What is the average temperature of the water in a suppression pool before a LOCA begins?
58. What is the maximum pressure that is reached in a PWR containment building after a large pipe break LOCA?
59. What is the maximum pressure that is reached in a BWR containment building after a large pipe break LOCA?
60. After an LB LOCA begins, approximately when are these peak pressures reached?
61. What is the purpose of the Baker–Just equation?
62. Why is it used in RELAP5, TRACE, and other famous accident analysis programs?
Exercise 30.2
A coolant pipe of 0.3 m in diameter carries pressurized water from the steam generators to the reactor pressure v essel.
The pressure of the water in the pipe is 2,000 PSI (13.8 MPa), and the temperature of the water is 290°C. The pipe
suddenly breaks cleanly 1 m from the pressure vessel. What is the initial rate of discharge through the pipe?
1168 REACTOR ACCIDENTS, DBAs, AND LOCAs
Exercise 30.3
A coolant pipe 0.3 m in diameter carries pressurized water from the reactor pressure vessel to the steam generators. The
pressure of the water in the pipe is 2,250 PSI (15.5 MPa), and the temperature of the water is 320°C. The pipe suddenly
breaks 10 m from the reactor pressure vessel. The break is clean and perpendicular to the pipe axis. The back pressure
is atmospheric. What is the rate of discharge from the pipe at the instant the break occurs?
Exercise 30.4
Water flows from the pressure vessel in a PWR to the steam generators through a 0.5 m-diameter pipe. The pressure in
the pipe is 15 MPa. The pipe develops a SB just before the pipe enters the steam generator. At the break, the quality is
found to be 5%. What is the void fraction just as the coolant is exiting the pipe?
Exercise 30.5
During a small pipe break LOCA, water flows from a 0.01 m-diameter hole in a coolant pipe that is 0.02 m thick. The
water pressure in the pipe is 5.5 MPa, and the pipe discharges into the containment building where the pressure is
50 kPa. What is the discharge rate of the water from the pipe? Does the flow of water from the pipe ever become choked?
Exercise 30.6
During an LB LOCA, the ejection rate of water from the core of a commercial PWR is approximately 500,000 L/min.
After the blowdown stage is complete, approximately how much water is left in the core? Assume the core contains
300 m3 of high-pressure water that weighs 225,000 kg.
Exercise 30.7
In most hypothetical LOCAs, the flow remains choked above 100 PSI (0.7 MPa) on the upstream side, and so the flow
rate is generally not proportional to the square root of the pressure gradient across the break location. Using this fact,
calculate the discharge rate from a 1 m2 pipe break assuming (1) the flow is choked and (2) the flow is not choked. When
making this estimate, assume the stagnation pressure is 15 MPa and the back pressure is 100 kPa. Approximately how
much faster does the core drain when the flow is not restricted or choked?
Exercise 30.8
During PWR LOCAs, saturated water flashes into steam if a pipe break develops on the primary side. Using an equilib-
rium rate model (see Chapter 29), calculate the discharge rate from a pipe having a hole with a cross-sectional area of
0.3 m2. Assume the water exiting the hole has an initial temperature of 320°C and an initial pressure of 15 MPa. Does
the discharge rate become higher or lower when the water is subcooled?
31
Flow Oscillations, Density Waves,
and Hydrodynamic Instabilities
31.1
Types of Hydrodynamic Instabilities
Under certain conditions, both boiling water reactors (BWRs) and pressurized water reactors (PWRs) can exhibit local and global
oscillations in their flow fields that are known as hydrodynamic instabilities. These instabilities occur in both the primary and sec-
ondary loops. While these instabilities are not necessarily detrimental to the plant as a whole, f ailure to address them can sometimes
lead to unexpected or undesirable reactor behavior. In this chapter, we would like to explain how these instabilities develop and
how they can be controlled. In practice, BWRs are much more susceptible to these instabilities than PWRs are. Normally, a large
amount of boiling is required to initiate a hydrodynamic instability. In PWRs, these instabilities occur mostly in the secondary loop.
In BWRs, they occur in the primary loop. Normally, hydrodynamic instabilities are considered to be Category II Design Basis
Accidents (DBAs) (see Chapter 33), and thus, reactor designers must demonstrate that their occurrence does not result in appreciable
damage to the plant. In the 1990s, the NRC began to promulgate guidelines intended to limit the size and scope of these hydrody-
namic instabilities. More information regarding these guidelines can be found at
https://www.nrc.gov/reading-rm/doc-collections/nuregs/staff/sr1793/.../chapter4.pdf
31.1.1
Static Instabilities Including Ledinegg Instabilities
Ledinegg instabilities are static instabilities that develop in the flow field when the mass flow rate m in a boiling channel transitions
itself from one steady-state value to another (and generally lower) steady-state value. In general, these instabilities can be produced
whenever the slope of the RCS pressure drop-flow rate curve (which is an internal characteristic of a flow channel) becomes smaller
than the loop supply (pump head) pressure drop-flow rate curve (which is an external characteristic of the same channel). Therefore,
they do NOT occur when the partial derivative of the pressure drop with respect to the flow rate of the RCS is greater than or equal
to the derivative of the head with respect to the flow rate of the reactor coolant pump head-capacity curve. We will have more to say
about this as we proceed further into the chapter.
31.1.2
D ynamic Instabilities Including Density Wave Oscillations
As opposed to Leginegg instabilities, density wave oscillations (sometimes called concentration wave oscillations) are considered
to be dynamic instabilities that begin when an inlet fluctuation in a coolant channel produces an enthalpy perturbation in the same
channel. If the coolant boils, this perturbation perturbs the length and the pressure drop of the single-phase regime and causes qual-
ity or void perturbations in the two-phase regimes that travel along the channel with the flow. These quality and length perturbations
create perturbations in the two-phase pressure drop (see Chapter 23). However, because the total pressure drop across the core is
determined by the design of the plant, these two-phase pressure drop perturbation feeds back to the single-phase regime. Depending
on the configuration of the core, these perturbations can be either attenuating or self-sustaining. Similar oscillations occur in PWRs at
very high power levels—but these usually begin at about 150% of the rated power level. However, they occur much more frequently in
BWRs because of the large number of voids. These density waves can create very large loads on the reactor piping system which are
1169
1170 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
sometimes interpreted as unwanted mechanical vibrations. Consequently, density waves are similar in some respects
to a water hammer, which sometimes develops in reactor piping systems when a valve is suddenly opened or closed.
However, their frequency of vibration is different, and if they are not attenuated, they can continue to persist for very
long periods of time. Both Ledinegg instabilities and density wave oscillations can also result in what are called coupled
power-flow oscillations (see Section 31.3). These oscillations can cause the power level to oscillate wildly because of
coupled neutronic-thermal hydraulic feedback. This is particularly true in BWRs because of the void coefficient of the
reactivity. The void coefficient of the reactivity is discussed in more detail in our companion book.*
31.2
Hydrodynamic Instabilities in BWR Cores
In BWRs, hydrodynamic instabilities develop between the core inlet and the core outlet when the coolant is allowed to
boil. Each flow regime has a different frictional pressure drop, and the flow regimes at higher void fractions are not par-
ticularly stable (see Chapter 23). If the pressure drop (i.e., the pressure head) is the same for all fuel assemblies between
the top and the bottom of the core, then there are many different flow regime configurations that can theoretically
support the same values of PIN and POUT. These hydrodynamic instabilities manifest themselves as large fluctuations in
the value of the axial mass flux G = m/A. They can result in large swings in the void fraction, and they can also induce
vibrations in reactor fuel assemblies under certain conditions. They are called Leginegg instabilities, and they occur
primarily at low flow rates and high void fractions. Ledinegg instabilities involve a sudden change in the core mass flow
rate as the core transitions from one flow regime to another. This instability occurs when the slope of the reactor coolant
system pressure drop curve versus mass flow rate curve:
becomes algebraically smaller than the loop supply (pump head) pressure drop–flow rate curve:
Hence when
a Ledinegg instability can develop. However, for reasons that we will subsequently discuss, these instabilities are gener-
ally not a problem in PWRs.
31.3
Coupled Power–Flow Oscillations
In BWRs, the power and flow rate are much more tightly coupled than they are in PWRs. If one increases the mass flux
G = m/A, the power will increase, and if one decreases the mass flux, the power will decrease. Physically, this occurs
because the fission rate increases when there are fewer voids, while it decreases when there are more voids. Figure 31.1
shows how this occurs. This coupling can be very intricate at times and can have many unexpected consequences. The
core-wide reactivity change Δρ induced by changes in the flow rate or the power level depends on both the void coef-
ficient of the reactivity dρ/dα and the temperature coefficient of the reactivity dρ/dT:
Here, the symbol ρ refers to the reactivity and not the density of the coolant. Under some conditions that we will sub-
sequently discuss, it is possible for the mass flow rate through the core to vary at the same time the power level of the
core oscillates in phase. (However, there may be a slight time delay between the flow peak and the power peak.) For
the same pressure drop, it may also be possible for the power and the flow to oscillate out of phase. Finally, under still
other conditions, some parts of the core will oscillate in phase while other parts of the core will oscillate out of phase.
In other words, they will oscillate independently but with random phase differences between them. Coupled power–flow
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
31.3 Coupled Power–Flow Oscillations 1171
FIGURE 31.1 The physical coupling between the power and the coolant in a BWR core. When the number of voids goes up, the
power goes down, and when the number of voids goes down, the power goes up. More information about this effect can be found in
our companion book*.
oscillations induced by phase changes, reactivity changes, and density waves were first observed in the 1980s when two
BWRs—the Caorso BWR in Caorso, Italy, and the LaSalle 2 BWR in the United States—began to exhibit this behavior.
In 1984, the Caorso BWR began oscillating wildly and did not want to stop. In this reactor, the power level in upper half
of the core would increase, while the power level in the lower half of the core would decrease. The two halves would then
reverse this oscillation, and the power increased below the core midplane while decreasing immediately above it (see
Figure 31.2). In other words, the reactor unexpectedly entered a state where the two halves of the core would oscillate
out of phase. However, the total power output remained the same. Because of this, the control system did NOT scram
the reactor and the local power levels continued to oscillate wildly for some time. (Eventually the operators were able
to figure out what was happening, and they scrammed the core manually.) Then in 1988, a second BWR of similar design
(the LaSalle BWR located in LaSalle, Illinois), exhibited a different type of oscillation, known as an in-phase oscillation.
In this case, the power level of the entire core oscillated in phase. After a while, these oscillations became so violent that the
control systems scrammed the core (in response to a predetermined change in the total power output). The control systems
FIGURE 31.2 An out-of-phase instability in which the power level in the top of the core oscillates 180° out of phase with the power
level in the bottom of the core. (Adopted from https://technicalreports.ornl.gov/1992/3445605784933.pdf.)
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
1172 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
for these reactors were subsequently redesigned so that they would scram the core automatically based on local power oscil-
lations as well as global ones. Later authorities discovered that these oscillations had occurred in other reactors (but not
every reactor) during Start-up tests (see Table 31.1). Hence, in a relatively short period of time, power oscillations in BWRs
became a topic of considerable interest in the scientific and regulatory community. We would now like to examine some of
these oscillations in more detail and explain how they can affect the thermal-hydraulic behavior of the core. We would also
like to explain why certain types of BWRs are more susceptible to oscillations than others. In principle, a BWR consists of
several hundred fuel assemblies (see Figure 31.3) that are isolated from each other hydraulically because they are surrounded
by “cans” or sheaths. However, they are not necessarily isolated from one another neutronically because the neutron flux
in one fuel assembly can affect the flux in another. Normally, this form of neutronic feedback is restricted to just the four
neighboring assemblies (which are also called the nearest neighbors).
Conversely, in fast reactors, the neutron mean free paths are longer and other assemblies can be affected as well. As far
as a reactor is concerned, these fuel assemblies are considered to be loosely coupled neutronically. So, the power genera-
tion rate in one fuel assembly (say fuel assembly A in Figure 31.4) can be affected by the power generation rate in another
fuel assembly (say fuel assembly B) and vice versa. Thus, if the c oolant boils in fuel assembly A, it will have some effect
on the neutron production rate in fuel assembly B. If the flow regimes in fuel assembly A become unstable when the void
fraction is high, this can induce a pulse of neutrons to enter fuel assembly B at a slightly later time. This can create slightly
different flow patterns (or density wave) in fuel assembly B than there is in fuel assembly A. These waves can also vary
in intensity as they move from the bottom to the top of the core. If one examines this problem more closely, one finds
that there are certain core loading patterns that can prevent these oscillations from occurring, and there are other loading
patterns that can encourage them to develop. In other words, the radial and axial power profiles in BWRs strongly affect
the hydrodynamic stability, and there are certain combinations of these profiles that can bring the core into configurations
that are more prone to inducing neutronic and thermal-hydraulic instabilities than others. Also, a specific transient may
be better at inducing a particular instability than another transient of comparable magnitude and duration.*
31.4
Types of BWR Instabilities
BWRs can have several different types of hydrodynamic instabilities, and each of these instabilities can have a different
vibrational mode. In this section, we would like to briefly discuss some of the more common instabilities that have been
TABLE 31.1
A Table Summarizing the Reported Power Oscillations in Commercial BWRs since 1980
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
31.4 Types of BWR Instabilities 1173
FIGURE 31.3 A modern BWR core containing about 800 fuel assemblies.
FIGURE 31.4 Four modern BWR fuel assemblies surrounding a cruciform-shaped BWR control rod.
observed. The names we will give to them are not universal; however, they are representative of the physical processes
that enable them to occur. The instabilities in BWR cores fall into three broad categories:
☉☉ Control system instabilities
☉☉ Channel thermal hydraulic instabilities
☉☉ Coupled neutronic and thermal hydraulic instabilities
Each of these instabilities affects the core in an entirely different way; however, they also have several things in common.
The size and frequency of the disturbances can vary from one category to the next. In general, oscillations induced by the
1174 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
control system are not very large (between 1% and 5% of the average power level or flow rate), and they can be hard to
detect because they happen so slowly. Their time constants (as defined by the time between one peak or trough and the next)
are relatively long, and in some cases, their cycles can be on the order of minutes. They can put an unexpected strain on
some of the reactor’s components (such as the fuel rods, the pumps, and the structural materials) in much the same way as a
low-frequency seismic wave would. Hence, if they can be avoided, these critical systems will experience less wear and tear,
and they will have to be serviced less often. The second major type of instability that a BWR is susceptible to is a thermal-
hydraulic channel instability. Sometimes it is called a channel instability for short. Channel instabilities can refer to indi-
vidual fuel assemblies or to a group of fuel assemblies. They can have more variations than other instabilities do. Channel
instabilities have a period of several seconds. However, they can be quite harmful because if they go undetected for a long
period of time, there is a good chance that some of the fuel rods will fail. Channel instabilities are generally more localized
than other instabilities, and this is one of the reasons why they are so hard to detect. The third major type of instability that
a BWR is susceptible to is called a coupled thermal-hydraulic and neutronic instability. This instability affects the entire
core (or at least a significant portion of it). It is the largest of all of the common instabilities. Since 1980, there have been
approximately 20 events that have been attributed to this instability worldwide. Some of these events have been relatively
serious, and others have been reported during Start-up tests. In addition, there are probably a few additional examples
that have not been reported. The Caorso and LaSalle 2 BWRs both reported this type of instability. The occurrences of
this instability are summarized in Table 31.1. The general characteristics of these three types of instabilities are described
in Table 31.2. However, these are only broad descriptions of these instability types, and these instabilities do not always
manifest themselves in the manner described here. We would next like to discuss each of these instabilities in more detail.
31.5
Control System Instabilities
Control system instabilities are usually the most benign instabilities in commercial BWRs. They can be produced
when the control system (which is similar to the operating system of a personal computer) does not adjust the feedwater
flow, the feedwater temperature, the control rod positions, or the pressure level properly. These actions can create an
unwanted low-frequency power or flow oscillation that is typically 3%–5% of the average value for these parameters.
Instabilities of this type develop when one of the plant’s controllers sends a signal to the control system that regulates
a key operating parameter for the plant. For example, instead of the core flow being increased, it should be decreased.
Normally, these controllers are designed to respond to more common types of events than less common ones. Hence, if
a controller is programmed to send a certain type of signal at a flow rate of 100%, it may send an entirely different signal
(for optimization reasons) when the flow rate is 20% or 30%. The first signal may cause the reactor to perform flawlessly,
while the second signal may cause it to become unstable (because a boiling core can be prone to nonlinear feedback).
Hydrodynamic instabilities can also produce low-frequency oscillations (see Figure 31.5).
These oscillations can be found in the feedwater flow rate, the feedwater temperature, or the pressure in the pres-
sure vessel. A typical oscillation appears when a valve follows an incorrect set of instructions that are issued by the
TABLE 31.2
A Summary of the Hydrodynamic Instabilities That Can Occur in BWRs and PWRs
Reactor(s)
Type of Instability Affected Scope Character Description of Instability
Channel instabilities BWR Local Static or dynamic Can occur in one or more fuel assemblies
Ledinegg instability BWR Local Static Involves a flow regime transition
Flow transition instability BWR Local Static Usually occur at low power-to-flow ratios
Geysering/chugging instability BWR Local Static Can occur if the power is high and the flow is low
Density wave oscillations BWR Local or global Dynamic Sometimes referred to as “shock waves”
Pressure oscillations BWR Local or global Dynamic Caused by rapid void fraction fluctuations
Flow oscillations BWR Local or global Dynamic Caused by phase transitions in the core
Control system instabilities BWR Local or global Dynamic Usually caused by poor control system design
Coupled neutronic-thermal BWR Local or global Dynamic Also called reactivity-induced oscillations
hydraulic instabilities
Primary loop instabilities PWR Local or global Dynamic Can only occur at very high power levels
Secondary loop instabilities PWR Local or global Dynamic Can only occur at very low flow rates (usually
involving the steam generators)
31.7 Thermal-Hydraulic Instabilities 1175
FIGURE 31.5 An oscillation in the inlet flow rate caused by a control system instability in a BWR5. Normally, instabilities of this
kind are relatively benign and do not pose a significant safety concern.
control system. These instabilities tend to be relatively benign, and the reactor is usually so forgiving that they tend to go
unnoticed. In other words, they show up in the “noise” level of the plant. However, they can still result in low frequency
power or flow oscillations that are between 1% and 5% of the total core power or flow rate. These oscillations are nor-
mally inaudible. However, they can lead to unwanted vibrations in the core that are actually audible to the human ear.
In addition, these vibrations can cause premature wear to some reactor components. Hence, these instabilities should be
detected and corrected, if for no other reason than to avoid excessive wear of the control actuators that may cause a more
serious problem later on. An example of such an instability is shown in Figure 31.5. This instability was first observed
in BWR 5s. The total core flow rate can be seen to oscillate with a period of about 50 seconds. However, because the
oscillations in this case are relatively minor (±0.5% of the average core flow), they can be very hard to detect. Before
these oscillations were known to exist, many reactor operators simply attributed them to “noise.”
31.6
Types of BWR Controllers
There are two main controllers that affect how the core of a BWR behaves:
31.7
T hermal-Hydraulic Instabilities
Channel thermal-hydraulic instabilities are instabilities that occur in BWR fuel assemblies when each fuel assembly is
considered hydraulically to be a single coolant channel. These instabilities can be of either the static or dynamic type.
Each instability has three separate subtypes which we would also like to discuss. Examples of static thermal-hydraulic
instabilities include
☉☉ Ledinegg instabilities
☉☉ Flow regime “relaxation” or transition instabilities
☉☉ Chugging or geysering instabilities.
1176 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
These instabilities can occur as soon as a stable operating state is reached (e.g., at full power operation). It does not take
a change in the state of the reactor to trigger one of these instabilities. It simply takes a r andom fluctuation in the core
flow rate, the pressure, or the power which corresponds to a particular operating condition. Now let us turn our attention
to dynamic instabilities. Examples of dynamic thermal-hydraulic instabilities include
☉☉ Density wave oscillations
☉☉ Pressure wave oscillations
☉☉ Instabilities that can be induced by flow regime changes.
Dynamic instabilities are different than static instabilities because they only occur when the operating state of the
reactor is changed (e.g., by changing the flow rate, the pressure, or the control rod locations). They are generally
triggered by a reactor transient. The most common type of dynamic instability is a density wave instability, which is
also called a channel flow instability. All channel thermo-hydraulic instabilities tend to be local in nature. By this,
we mean that they are generally restricted to a single fuel assembly or a group of fuel assemblies. (A large BWR core
can have up to 800 of these assemblies.) However, more than one fuel assembly can participate in some of these insta-
bilities. Hence, there can be both single-channel instabilities and parallel-channel instabilities. In parallel-channel
instabilities, the flow rate in one fuel assembly decreases while the flow rate in an adjacent fuel assembly increases.
The void fractions also behave in a similar way. This mode of operation is an example of what is called an out of phase
flow oscillation. Examples of these oscillations in the flow rate are shown in Figure 31.6 for two, three, and four fuel
assemblies.
If the pressure drop across the fuel assemblies is the same (due to the core pressure head), the void fraction will
vary inversely with the flow rate. Under the right conditions, three and four channel oscillations are also possible.
During two-channel oscillations, two fuel assemblies (or groups of fuel assemblies) oscillate 180° out of phase with
respect to each other. In three-channel oscillations, three fuel assemblies (or groups of fuel assemblies) oscillate
120° out of phase, and in four-channel oscillations, four fuel assemblies (or groups of fuel assemblies) oscillate 90°
out of phase. Oscillations in the void fraction can usually be seen on the control room’s local power range monitors
(or LPRMs), but they can be very hard to distinguish on the average power range monitors (or APRMs). The phase
differences tend to mask the underlying processes. Local flow oscillations also have the potential to cause transition
boiling and prolonged periods of rod dryout, which can lead to the failure of some of the fuel rods. Fortunately, if they
are detected early enough, they will not cause the rods to dryout. In one reactor in the 1960s (the Garigliano BWR in
Italy), flow meters were installed at the outlet of some of the fuel assemblies during Start-up. When one of the flow
meters got stuck in the wrong position, it caused a significant increase in the two-phase pressure drop between the top
and the bottom of the assembly. This caused the flow rate to become unstable, and the flow oscillated almost indefi-
nitely in the way that a single-channel instability would. In the adjacent channels, where the flow meters behaved
normally, the same instabilities did not develop.
FIGURE 31.6 Three examples of parallel channel instabilities in which the mass flow rate oscillates 90°, 120°, and 180° pit of
phase with the neighboring fuel assemblies. In some cases, the oscillations can be up to 10% of the total flow rate.
31.8 Coupled Neutronic and Thermal Hydraulic Instabilities 1177
31.8
Coupled Neutronic and Thermal Hydraulic Instabilities
Coupled neutronic-thermal hydraulic instabilities are the most significant instabilities in commercial BWRs. They are
also the largest common instabilities. Sometimes reactor physicists refer to them reactivity instabilities. Approximately
20 examples of these instabilities have been reported in commercial power reactors since the 1980s, although some of
these instabilities occurred during Start-up tests. The remainder of them occurred during normal operation, and they
caused the reactors to scram. All of these instabilities occurred in BWRs. The known incidents are shown in Table 31.1.
Density waves (see Section 31.5) can also cause the reactivity to change as the void fraction is changed. The amount of
reactivity feedback is given by
∆ρ = ∂ρ/ ∂α ⋅ ∆α (31.5)
where ρ is the reactivity again and Δα is the average change in the void fraction. If the density wave takes a while to
develop and propagate, and the void coefficient of the reactivity ∂ρ/∂α is large enough, a reactor will become unstable.
The power level then oscillates with a frequency close to the inverse of the density wave time constant. This type of
instability is called a coupled neutronic-thermal hydraulic instability because it involves two separate feedback effects:
one due to the neutron production rate in the fuel (which changes the power level) and the second due to core pressure
changes (which changes the flow rate). The combination of these two effects (with phase delays) then causes the reactor
power and flow levels to oscillate wildly. When these instabilities develop, the time constant for the power is normally
different than the time constant for the flow
For example, when additional heat is generated in the fuel, it takes a while before this heat can reach the coolant. Thus,
even when the fuel temperature begins to rise, the void coefficient does not. Normally, it takes between 6 and 10 seconds
before the additional heat in the fuel reaches the coolant. Therefore, the delay times for density wave oscillations and the
void coefficient of the reactivity are not exactly the same. Differences in these delay times can lead to phase delays, and
these phase delays can cause the void feedback to reinforce the density wave oscillations (which can create even more pos-
itive feedback). Decreasing the time constant for the fuel (by making the fuel rods thinner) generally has a destabilizing
effect under these conditions. There are four basic modes of oscillation that have been observed in commercial BWRs:
☉☉ Mode I: A core-wide mode where the power level and the flow rate oscillate in phase for all of the channels
☉☉ Mode II: A core-wide mode where the power level of half of the core oscillates out of phase with respect to the
other half. Here, the inlet flows of both halves are also out of phase with respect to one another.
☉☉ Mode III: A core-wide mode where two opposite quadrants of the core oscillate in phase, and the other two quad-
rants oscillate out of phase (similar to the oscillation of a membrane on the top of a drum).
☉☉ Mode IV: A core-wide mode where the center of the core oscillates out of phase with respect to the rest of the core.
These modes of oscillation are illustrated in Figure 31.7. Each mode is due to the fact that the neutron flux oscillates with
a different set of harmonics. The spatial modes of the neutron flux are then coupled to the thermal hydraulic behavior
of the core through the void coefficient of the reactivity. Because these oscillations can sometimes become very large,
they can be easily detected by the LPRM,* but they may not show up at all on the APRM† because in some oscillations,
the total power level may barely change.
Experiments have shown that these core-wide oscillations are not always symmetrical. The difference between the
size of the local oscillations and the global oscillations can be quite large. In fact, in some cases, it can be as great as 10
to 1. This implies that by the time the set point for a global scram is reached (at a power level of 120%), the local oscil-
lations may be as large as 600%–1,200%. Because of this, BWRs are now designed to scram automatically if one of the
LPRMs detects a power oscillation in excess of 20%. The exact set point depends on the country in which the reactor
is operated. The most famous set of these oscillations in the United States occurred at the LaSalle BWR in 1988. Here,
the whole core oscillated in phase, but the core flow rates and the power levels oscillated out of phase (see Figure 31.8).
In the Ringhals BWR in Sweden, an entirely different oscillation occurred. Here, both halves of the core oscillated out
of phase in a pattern similar to that shown in Figure 31.2. These oscillations caused the LaSalle reactor to automatically
scram when the power level reached 118% of its nominal value. However, the power level in the Ringhals BWR did not
change as dramatically because phase effects cancelled each other out. The oscillations could only be stopped when the
* See https://www.nrc.gov/docs/ML1125/ML11258A333.pdf.
† See Radiation Detection and Measurement by G. Knoll, John Wiley & Sons (2010).
1178 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
FIGURE 31.7 Modes of oscillation for a coupled neutronic-thermal-hydraulic instability. In the areas in red, the power increases,
and in the areas in green, the power decreases. The pattern then reverses itself after a while.
FIGURE 31.8 Some measurements taken during an in-phase oscillation at the LaSalle BWR in 1988. (Taken from http://web.ornl.
gov/info/reports/1992/3445605784933.pdf.)
31.9 DENSITY WAVES IN BWRs AND THEIR ORIGIN 1179
operators decided to manually scram the core. In most BWRs, the outputs from approximately 20 LPRMs are summed
together to provide input to one of the seven or eight APRMs. Each APRM then provides input to both the control and
radiation protection systems.
31.9
D ensity Waves in BWRs and Their Origin
Density waves are dynamic instabilities, while Ledinegg instabilities are static ones. Density waves can introduce time
delays (i.e., phase delays) into the power production process because the power generated in the fuel rods takes a while
to reach the flow.
In liquids and gases, the speed of propagation of density waves is limited by the speed of sound c, which is a function
of the density, the temperature, and the pressure. In general, the maximum speed of a density wave is the same as the
speed of sound for the wave:
c = √( M/ρ) (31.7)
where M is the bulk modulus of elasticity for the material, measured in Pa, and ρ is the average density. For an ideal gas,
the bulk modulus is related to the pressure p by
M = γ ⋅ p (31.8)
where γ is called the adiabatic index. It is also called the isentropic expansion factor. For an ideal gas, it is the ratio
of specific heat of the gas at a constant pressure cP to the specific heat of the gas at constant volume cV, so γ = cP/cV,
and it arises because a classical sound wave induces an adiabatic compression, in which the heat of the compres-
sion does not have enough time to escape the pressure pulse and, thus, contributes to the pressure induced by the
compression. We can then write the equation for the speed of sound in the gas as
1 1
c = γ 2 (p/ρ) 2 (31.9)
Thus, the speed of sound is directly proportional to the square root of the pressure and inversely proportional to the
square root of the density. We can also use the ideal gas law to replace p with nRT/V, and ρ with nM/V. Then the expres-
sion for the maximum speed of the wave becomes
where
☉☉ CIDEAL is the speed of sound for an ideal gas.
☉☉ R is the molar gas constant, which has a value of approximately 8.3145 J mol−1 K−1.
☉☉ k is Boltzmann’s constant.
☉☉ γ (gamma) is the adiabatic index (sometimes assumed to equal 1.40 for diatomic molecules from kinetic theory,
assuming from quantum theory a temperature range at which thermal energy is fully partitioned into rotation
[rotations are fully excited] but none into vibrational modes. Gamma is actually experimentally measured over
a range from 1.3991 to 1.403 at 0°C, for air. Gamma is assumed from kinetic theory to be exactly 5/3 = 1.6667
for mono-atomic molecules such as noble gases).
☉☉ T is the absolute temperature of the gas in degrees Kelvin (°K).
☉☉ M is the molar mass in kilograms per mole. The mean molar mass for dry air is about 0.0289645 kg/mol.
☉☉ m is the mass of a single molecule in kg.
1180 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
In reactor fuel assemblies, density waves behave differently. They develop in BWRs because large packets of voids are
sometimes produced in the core. These void packets take several seconds to propagate through the core, and their rate of
propagation depends on the pump speed. Normally, the propagation time is between 1 and 2 seconds. The effective time
for the voids to propagate through the core is referred to as the density wave propagation time. During normal operation,
these voids (and thus the wave) move upward from the bottom of the core to the top. The density change created by these
voids can be relatively large when the power-to-flow ratio is high. The mixture density ρ at a specific axial location z is then
where α(z) is the local void fraction. The density distribution versus time then resembles the graph shown in Figure 31.9.
As the voids move upward, the density changes as they move. However, the total pressure drop across the core is fixed.
The local pressure drop is higher when there are more voids and it is lower where there are fewer ones. So if the pressure
drop Δp across the core is fixed (between the inlet and the outlet), then the flow rate must be less in channels that have
a high void fraction than it is in channels that have a low void fraction. This implies that at any given time, if the radial
power profile changes from one fuel assembly to the next, the flow rates must also change to produce the same core-wide
pressure drop. In BWRs, the propagation time of the pressure wave can cause the pressure drop in some fuel assemblies
to get out of phase with the inlet flow rate. This chain of events is illustrated in Figure 31.10.
In two-phase mixtures, the axial pressure drop can get 180° out of phase with the inlet mass flow rate m IN if the density
wave moves at just the right speed. So if the timing is right, a higher void fraction causes the flow rate to fall even more in the
hot fuel assemblies to maintain the same Δp across the core. This causes additional instabilities to develop because the fric-
tional feedback is now positive rather than negative. So instead of the density waves damping themselves out (which is gener-
ally what friction tends to do), they actually start to increase in magnitude if the propagation time between the bottom of the
core and the top of the core is just right. In this case, flow oscillations will develop. (The channel where the oscillation occurs
has an effective friction factor that is close to zero at the frequency of oscillation.) A density wave oscillation can be triggered
FIGURE 31.9 A density wave propagating through a BWR fuel assembly. Normally, it takes 1–2 s for the wave to traverse the
entire assembly.
31.10 Stability Maps and Operating Maps 1181
FIGURE 31.10 An illustration of a density wave propagating through a BWR fuel assembly. As the density wave moves upward,
the inlet flow rate oscillates out of phase with the pressure drop between the top and the bottom of the core.
by a small perturbation in the inlet flow rate, and if the propagation time of the density waves through the core is just right,
there is no compensating friction to turn it off. (It will simply sit there and oscillate forever until the reactor is scrammed.)
Hence, even a simple perturbation in the inlet flow rate or power level will become self-sustaining over time. In general, water
reactor fuel assemblies with high void fractions (and high power peaking factors) tend to be more prone to density wave
instabilities than fuel assemblies with low void fractions (and low power peaking factors). Reducing the flow rate during a
transient can also cause these instabilities to develop. The best way to prevent these instabilities from developing in the first
place is to keep the flow rate high enough and the void fraction low enough to keep a large, slow moving density wave from
developing. In the early days of the nuclear power industry, it was generally believed that the frequency of these instabilities
was roughly equal to the residence time of the density waves in the core. So if the density waves took 1 second to move from
the bottom to the top of the core (which would imply a coolant velocity of about 5 m/s), then the frequency of the instability
would be 1 cycle per second. In recent years, it has been shown that the frequency of these instabilities can be as high as four
or five times the residence time of the density waves in the core. This is because a second harmonic can develop at twice the
standard residence time, and a fourth harmonic can occur at four times the expected residence time. Thus, this is no different
than exciting the higher harmonics on a guitar string. How unstable the flow becomes depends on the amount of pressure
feedback to the inlet flows, and certain transients tend to be better at sustaining these instabilities than others. When one does
a more detailed analysis of these instabilities, starting with the fluid mechanics equations, one can develop an operating map
that shows where these instabilities can develop and where they do not. The operator of a BWR is then required to operate the
plant in only the stable regimes of this operating map. In the next section, we would like to discuss what one of these maps
looks like and show how they can be used. Operating maps are also discussed in our companion book.*
31.10
Stability Maps and Operating Maps
Two-phase flows can create many different types of hydrodynamic instabilities. When a BWR is operating normally,
these instabilities are normally relatively benign, but when they do occur, they can sometimes affect the safe and effi-
cient operation of the plant. In an extreme case, they can cause some of the fuel rods to fail or they can cause the reactor
to scram. Their effect on a reactor is generally not as severe as that of a loss-of-coolant accident (LOCA), but they are
nothing to be complacent about. A reactor designer then has the job of predicting how, when, and where these instabili-
ties can occur so that they can be designed around or otherwise compensated for. Normally, most BWR instabilities
occur when a reactor is being started up. One of the most important tools in predicting how, when, and where these
instabilities develop is a reactor stability map. In some textbooks, it is also called a reactor operating map or ROM. An
example of such a map for a BWR is shown in Figure 31.11. Operating maps are also discussed in Chapter 3. In modern
* See Nuclear Engineering Fundamentals: A Practical Perspective by R.E. Masterson, CRC Press (2017).
1182 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
FIGURE 31.11 An operating map showing regions of the map where reactor operation is allowed and other regions where it is not.
BWRs, there is a significant difference between the density of the coolant at the core inlet and outlet. This density dif-
ference causes the flow at the outlet to behave entirely different than the flow at the inlet. The coolant at the inlet is
subcooled and has an equilibrium quality of about −3%. The coolant at the outlet is a saturated water–steam mixture,
and the quality is between 10% and ~16%. Example Problem 31.2 illustrates how these values are determined.
x eq = ( h − h l ) h lv
Plugging in the appropriate numbers, we see that the equilibrium quality at the core exit is xeq = (h − hl)/hlv = (1,452 − 1,279)/
1,491 = 11.6%. This is rather low for a modern BWR, but it indicates a very conservative design. [Ans.]
Stability maps were first introduced into the nuclear power industry by Zuber and Ishii in 1970, and their work was
considered to be revolutionary at the time. Two references that describe their initial work are:
☉☉ M. Ishii and N. Zuber. Thermally induced flow instabilities in two phase mixtures. Proceedings of the fourth
international heat transfer conference, B5.11, 1970.
☉☉ M. Ishii. Thermally Induced Flow Instabilities in Two-Phase Mixtures in Thermal Equilibrium. PhD thesis,
School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, June 1971.
The equations they presented revealed the existence of two nondimensional scaling parameters that could be used to
determine when a two-phase system would become stable and when it would not. These parameters have subsequently
become known as the phase change number N PC and the subcooling number NSC. Together they define a set of operating
parameters where the flow is stable and where it is not. Normally, they are plotted on the x and the y axes of what has
31.11 The Phase Change Number 1183
become known as a two-phase stability map. There are certain regions of this map where a BWR can be operated safely,
and there are other regions where it cannot. In the next section, we would like to discuss the phase change number and
the subcooling number. We would then like to illustrate how they can be used.
31.11
T he Phase Change Number
The phase change number and the subcooling number are based on the enthalpy and the density of the coolant at a par-
ticular axial location. Hence, they depend on the core height. It is usually easiest to discuss the phase change number for
a system that is in thermodynamic equilibrium. In other words, the phase change number under these conditions implies
that the location of the boiling boundary (the position of the net vapor generation or NVG point) occurs when the bulk
fluid temperature TBULK becomes equal to the saturation temperature TSAT and not before this. If we need to find the
location of the boiling boundary when the boiling is subcooled, we can always use the Griffith correlation (which was
discussed in Chapter 25) or a similar correlation to find the point of NVG. There are three relatively good derivations
of the phase change number that can be found in the literature. These include the original derivation of Ishii and Zuber
(see the reference mentioned previously) and other derivations dealing with non-equilibrium effects by Guido and Saha.
☉☉ P. Saha and N. Zuber. An analytical study of the thermally induced two-phase flow instabilities i ncluding the
effect of thermal non-equilibrium. International Journal of Heat and Mass Transfer, 21(4):415–426, April 1978.
ISSN 0017-9310.
☉☉ G. Guido, J. Converti, and A. Clausse. Density-wave oscillations in parallel channels-an analytical approach.
Nuclear Engineering and Design, 125(2):121–136, February 1991. ISSN 0029-5493.
The derivation by Guido et al. is the most recent of the three. For an equilibrium state, the phase change number NPC
is given by
or
N PC = F ⋅ (q/m)
(31.12b)
where F = ((ρl − ρv)/ρv)·(1/hlv) is sometimes called the phase change factor. Here, hl and ρl are the enthalpy and the den-
sity of the liquid phase, hv and ρv are the enthalpy and the density of the vapor phase, and hlv is the enthalpy of vaporiza-
tion. Thus, the phase change number is proportional to the amount of heat q = q″PHLH that is added to the channel (in
J/s), and it is inversely proportional to the mass flow rate m = GA (in kg/s). Here, A is the area of the flow channel (in m2),
LH is its heated length (in m), and PH is the heated perimeter of the channel (in m) through which heat flows. Normally,
the phase change number increases with the length of the channel because of the additional heat that is added. The phase
change number is dimensionless, and it can be thought of as a dimensionless scaling parameter that is proportional to
the ratio of the reactor power to the average flow rate. High values of the phase change number mean that the power
level is high and the flow rate is low. Low values mean that the power level is low or that the mass flow rate has been
increased. The phase change number for a commercial BWR is calculated in the following exercise.
Fluid Properties
Solution For a commercial BWR, the average heat flux is about q″ = 1,000 kW/m2, m = 0.175 kg/s (for an interior
subchannel), PH = 35.2 mm, LH = 4 m, and q = q″PHLH = 1,000 × 0.14 = 140 kW. Hence, q/m = 800 and F = 0.0123. The
phase change number is then NPC = F·(q/m) = 9.85. [Ans.]
31.12
T he Subcooling Number
The subcooling number is a different number than the phase change number. It is a measure of the degree of inlet sub-
cooling, and it can be thought of as the dimensionless residence time of a fluid particle in the single-phase flow regime.
The subcooling number, NSC, is defined by
where hIN is the mixture enthalpy at the inlet to the boiling channel (in J/kg). Notice that the subcooling number does
not depend on the geometry of the flow channel. Typical values for the subcooling number range from about 0 to 10,
and typical values of the phase change number can range from about 0 to 15. A value of 0 for each number implies that
a single-phase liquid exists (at least in a thermodynamic sense because x = 0 and q = 0). Traditionally, a ratio of the
phase change number to the subcooling number between 1.0 and 2.0 means that the flow will be stable and a ratio
greater than about 2.0 means that it will not. The easiest way to understand the implications of these numbers is to draw
a stability map where the phase change number is plotted on one axis and the subcooling number is plotted on the other.
Certain areas where the values intersect will be stable and other areas where the ratios are greater than two will not. An
example of a typical stability map is shown in Figure 31.12. In this particular map, we have overlaid three theoretically
determined stability thresholds on top of the underlying experimental data—one based on the equilibrium stability
FIGURE 31.12 A diagram showing different conditions where Ledinegg instabilities can develop. PWRs operate to the left of the
solid blue line and BWRs operate to the right of it. Operation to the right of the solid red line is not stable and is normally avoided
in practice.
31.13 Ishii’s Stability Criterion for Hydrodynamic Instabilities 1185
theory of Ishii and Zuber (1970), another based on a non-equilibrium model developed by Saha and Zuber (1978), and a
third model based on some simple stability criteria proposed by Ishii (1971). If a reactor operates to the left-hand side of
the solid blue line (where a PWR normally does), the thermodynamic exit quality xexit at the exit to the core is less than 0,
and we are only dealing with a single-phase liquid (or a two-phase mixture that is slightly subcooled). The other regime
to the right of the dotted blue line indicates that there is a superheated vapor at the exit to the core because xexit = 1.0.
Between these two extremes, there is a stable boiling regime and an unstable one.
Fluid Properties
Using the phase change number calculated in Problem 31.3, determine if the flow is stable or unstable.
Solution From Equation 31.13, the subcooling number is NSC = ((hl − hIN)/hlv)·((ρl − ρv)/ρv. The subcooling number is then
NSC = ((1,279 − 1,228)/1,491)·((735.3 − 38)/38) = 0.034 × 18.35 = 0.62. The ratio of the phase change number to the subcool-
ing number is NPH/NSC = 9.85/0.62 = 15.9. The flow is definitely not stable because this ratio is greater than 2.0. [Ans.]
31.13
Ishii’s Stability Criterion for Hydrodynamic Instabilities
In the stable boiling regime between the zero exit quality line and the various stability thresholds, the flow is considered
to be stable because flow boiling damps out small perturbations in the flow field and returns the core to a stable state.
Normally, this is because the frictional forces in the fluid are large enough to prevent an instability from growing. If one
moves immediately to the right of any of the stability lines (for example the solid red line), a perturbation in the flow field
will produce diverging oscillations that will continue to increase over time. The frictional forces under these conditions
are not sufficient to damp out the effects of gravity and acceleration on the two-phase oscillations. Hence, the flow field
continues to oscillate, and the mass flow rate, the density, or the pressure will also oscillate as well. This effect does not
include the effects of neutronic feedback that we discussed previously (see Section 31.4), and these feedback effects are
usually more important on a global scale. The simplified stability model proposed by Ishii (in 1971) was based on the
assumption of thermodynamic equilibrium between the phases. The stability criterion for this model was
where
( ) ( )
S = 2 K IN + fL 2D e + K OUT 1 + 0.5 fL 2D e + K OUT (31.14)
and this is shown by the dashed black line in the center of the stability map. Here, K IN and KOUT are the inlet and the
outlet restriction pressure drop coefficients, f is the two-phase friction factor of the mixture (assumed to be constant
along the channel’s length), De is the equivalent hydraulic diameter, and L is the length of the boiling regime (in meters).
For a variety of reasons, Saha proposed that the stability criteria of Ishii was valid for NSC > π.
1186 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
Thus one can always achieve a stable operating state by driving the ratio of the phase change number to the subcool-
ing number to about 1.0 (NPC/NSC ≈ 1.0). This can be done by increasing the inlet flow rate (which reduces the value of
the phase change number) or by reducing the amount of heat q that is added to the coolant channel (which increases the
subcooling number). In any event, both approaches reduce the size of the void fraction and a system with a smaller void
fraction is generally a more stable one. Conversely, if we increase the power and reduce the flow, we will drive the ratio
of the phase change number to the subcooling number to a much larger value, and the flow will become unstable again.
Similarly, an increase in the degree of subcooling stabilizes the system, and a decrease in the degree of subcooling
destabilizes it. Finally, increasing the pressure for the same power level tends to stabilize the system because fewer voids
are formed. If we suddenly decrease the system pressure, than many more voids will form, and unstable oscillations in
the flow field will begin to develop. Reactor operators and reactor vendors use stability maps to ensure that BWRs stay
in a stable operating range—particularly during Start-up. PWRs are generally operated to the left of the single-phase
stability line (which is shown in solid blue) and at least during normal operation (which is defined by normal power and
flow levels), instabilities in the flow field in PWRs are not a significant problem. Under most conditions, they will not
develop in the primary loop until the power level exceeds about 150% of rated power. However, in the secondary loop,
instabilities may develop in the steam generators at very low flow rates (typically about half of normal flow), and when
this occurs, the steam generators may behave abnormally.
In general, Homogenous Equilibrium Models or HEMs are more conservative at low subcooling numbers, and
Homogenous Non-equilibrium Models (or HNEMs) are more conservative at high subcooling numbers. The specific
model one chooses to use is dictated by how important these thermal non-equilibrium effects become. Occasionally a
BWR coolant channel can enter into another unstable hydrodynamic state that has become known as geysering. Geysering
occurs at very low flow rates, and it is conceptually quite similar to what happens when Old Faithful erupts periodically
in Yellowstone National Park (see the illustration in Figure 31.13). However, geysering only occurs at very low pressures
and extremely low flow rates. In a reactor core, subcooled chimneys can sometimes develop in fuel assemblies where
some of the larger voids collapse as they rise due to buoyancy forces. This interaction of the colder fluid with these hotter
bubbles creates instabilities in the flow field as the bubbles begin to condense. Here an entire BWR fuel assembly will
simply “cough up” a hot water–steam mixture when the flow rate is abnormally low. In BWRs, loop-type instabilities
can sometimes follow a geysering instability and here the flow in the downcomer can oscillate in phase or out of phase
with the core flow. Finally, density wave oscillations can occur at much higher power-to-flow ratios, and they generally
have a higher characteristic frequency than geysering does. Density waves usually have a frequency of about 1 second.
31.14
Hydrodynamic Instabilities in PWRs
As we mentioned before, hydrodynamic instabilities can also occur in PWRs. These instabilities can be divided into
two basic categories:
1. Hydrodynamic instabilities in the primary loop
2. Hydrodynamic instabilities in the secondary loop
FIGURE 31.13 Geysering in BWR fuel assemblies is similar to the eruption of Old Faithful in Yellowstone National Park. The
geysering frequency depends on the degree of inlet subcooling.
31.15 More on Ledinegg Instabilities 1187
We would now like to discuss these instabilities and show how they can affect the operation of a PWR. Normally, hydro-
dynamic instabilities in PWRs are much less common than they are in BWRs, and they are also less severe. However,
they can still occur if certain conditions are met.
31.14.1
I nstabilities in the Primary Loop
Hydrodynamic instabilities in PWRs do not occur very often, and normally, the power level must be increased to about
150% of its rated value before they appear. This behavior is caused by the fact that PWRs normally operate in the
subcooled boiling regime where the ratio of the phase change number to the subcooling number is 1.0 or less. For the
primary loop, Ishii’s simplified stability test (see Section 31.9) can be used to determine whether the core will develop a
density wave oscillation. Moreover, this test is used to determine the stability of AP1000 cores because Ishii’s method
is more conservative in the open-channel architecture of PWRs. The fuel assemblies in the core are not “sheathed” like
they are in BWRs, so there is very little resistance to flow in the lateral direction. Thus, significant enthalpy exchange
can occur between channels of high power density and channels of low power density. Thus, the power level in PWRs
must reach 150% of its nominal value before these oscillations can develop. The higher operating pressure (15.5 MPa
vs. 7.25 MPa) also helps to dampen these waves. Flow instabilities have been observed in closed-channel geometries in
BWR cores, but they have almost never been observed in open-channel geometries representative of PWR cores above
2,000 PSI (15.5 MPa). For example, Kao, Morgan, and Parker (see the reference below) conducted a number of stability
experiments in closed-channel geometries that simulated typical reactor core flow rates.
☉☉ Kao, H.S., Morgan, T.D., and Parker, W.B., “Prediction of Flow Oscillation in Reactor Core Channel,” Transactions
of the American Nuclear Society 16, pp. 212–213, 1973.
Their experiments were conducted between pressures of 800 PSI and 2,200 PSI. They showed that for the higher
power and flow rates that are usually encountered in PWRs, no flow oscillations could be induced above a pressure of
approximately 1,200 PSI (8.3 MPa). Thus, many regulatory authorities believe that BWRs operate near the upper limit
of the system pressures where these oscillations can develop. Their results also suggested that if a BWR core looked
more like a PWR core, it would be much more difficult (without flow and power feedback) for significant hydrodynamic
instabilities to develop.
31.14.2
I nstabilities in the Secondary Loop
Static hydrodynamic instabilities were first observed in the steam generators of coal-fired power plants in the 1930s.
At the time, they were believed to be caused by fluctuations in the inlet mass flow rate. However, it later turned out that
these instabilities were, in fact, Ledinegg instabilities, and they occurred whenever the following condition was met:
It should therefore come as no surprise that similar instabilities can develop in the secondary loop of PWRs where the
ambient temperatures and pressures are lower. In particular, these instabilities can develop in the steam generators when
the flow rate becomes low. In fact, Ledinegg instabilities can occur during Start-up in PWRs at about 50% of normal
power or at 100% of full power and at about 50% of normal flow. These instabilities can cause the tubes in the steam
generators to vibrate wildly, and so Start-up procedures have been devised by reactor vendors to prevent them from
occurring. The most common procedure to eliminate these instabilities is to increase the feedwater flow rate through
the steam generators until they disappear.
31.15
More on Ledinegg Instabilities
Ledinegg instabilities, in their original form, were first studied theoretically and experimentally by German scientist
Maximilian Ledinegg in 1938. His original explanation of how these instabilities developed showed that they could occur
during both natural circulation and forced convection. More information regarding his original work can be found at
☉☉ Ledinegg, M. (1938) “Instability of flow during natural and forced circulation,” Die Wärme, 61, 8.
Ledinegg instabilities are not just a problem in nuclear power plants, but they are also a problem in the boilers of
coal-fired power plants at low flow rates. However, in nuclear reactors, they are a more common concern for BWRs.
Specifically, BWR 4’s and BWR 6’s have had to change their operating procedures to ensure that they do not operate
1188 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
in a range of conditions where these instabilities can develop. Fortunately, enough work has been done in this area that
the range of void fractions, flow rates, and pressures where these instabilities develop and propagate is relatively well
understood. Thus, they can be largely prevented if a reactor is o perated properly. This requires adhering to the ROM,
which is described in our companion book* (see Figure 31.11). We would now like to present a more detailed description
of how these instabilities are formed. To begin our discussion, we would like to draw a chart of how the mass flow rate
m can change as a function of the channel pressure drop. In single-phase flow, this is a relatively simple relationship, and
for the most part, the flow rate increases as the pressure drop (inlet to outlet) increases. However, in two-phase flow, this
does not always have to be the case because a number of different flow regimes are theoretically possible for the same
pressure drop. Figure 31.14 shows how the pressure drops and the flow rates are related in this case. This is not an exact
representation of what can happen when the pressure drop is changed, but it is also a reasonably representative one.
There are three distinct regimes to consider:
☉☉ Regime I which occurs between points A and B
☉☉ Regime II which occurs between points B and C
☉☉ Regime III which occurs between points C and D
Each regime has its own unique behavior that we would now like to discuss.
31.15.1
Regime I: Single-Phase Liquid Flow
In flow regime I, which occurs between points A and B, the fluid is a single-phase liquid. The liquid can be either satu-
rated or subcooled. There are no voids in the coolant, and the single-phase friction factor is
f = C Re n (31.16)
( )
∆PSP = fSP ⋅ 1 2 ρv 2 ⋅ L D E (31.17)
where n is a constant having a value between 0.10 and 0.31. Again, ρ is the density of the liquid.
FIGURE 31.14 The flow regimes in a BWR core during safe and unsafe operation. During safe operation, the fuel rods never dry
out, but during unsafe operation, dryout can sometimes occur near the top of the core.
31.16 Ledinegg Stability Criteria 1189
31.15.2
Regime II: Two-Phase Flow
In regime 2, which occurs between points B and C, the fluid is a two-phase mixture. Depending on the v elocity of the
flow and the applied heat flux, the flow field can be a combination of nucleate boiling, bubbly flow, bulk boiling, slug
flow, churn flow, and annular flow. There are always some voids in the coolant, and the void fraction can vary from 0 to
about 99%. When the coolant boils, the two-phase phase friction factor is given by
( )
fTP = M TP ⋅ C Re n (31.18)
where MTP is the two-phase friction multiplier we discussed in Chapter 23, which normally has a value between 1.0 and
4.0. Here, the Reynolds number must be evaluated using the properties of the mixture. In the single-phase region, the
single-phase friction factor is given by
( )
fSP = C Re n (31.19)
Here, the Reynolds number is evaluated using the properties of the liquid. The value of MTP can be correlated to the
mixture quality or void fraction and it normally increases as either of them increase. Thus, the total pressure drop is
given by the sum of the pressure drops across the boiling and non-boiling heights:
or
∆P = f D e ⋅ 1 2 ρv 2 ⋅ L SP + f D e ⋅ 1 2 ρv 2 ⋅ M ⋅ L TP (31.21)
where ρ is the liquid density at the entrance to the channel. This expression may also be rewritten as
∆P = f D e ⋅ 1 2 ρv 2 ⋅ ( L SP + M ⋅ L TP ) (31.22)
where LSP is the length of the channel where the coolant is a single-phase liquid and LTP is the length of the c hannel
where the coolant is a two-phase mixture. In general, these lengths will be different, and the two-phase pressure drop
will generally be greater than the single-phase pressure drop. If the degree of subcooling is low, then a reasonable value
for f/De·½ρv2 can be found by evaluating the values of ρ and ν at the saturation temperature.
31.15.3
Regime III: Single-Phase Vapor Flow
In regime 3, which occurs between points C and D, the coolant is now a vapor. The coolant has left the surface of the
fuel rods, and only small water droplets remain in the flow. The single-phase friction factor for this high energy vapor
is given by
fSP = C Re n (31.23)
( )
∆P = ∆PSP = fSP ⋅ 1 2 ρv 2 ⋅ L D e (31.24)
where ρ is now the density of the vapor. After the equilibrium quality becomes equal to 1.0, the vapor becomes super-
heated and the water droplets disappear.
31.16
L edinegg Stability Criteria
Close to point B, or slightly before it, the coolant starts to boil. When point B is reached, the NVG rate becomes suf-
ficiently large to reverse the trend of ΔP falling when the mass flow rate m falls. Physically, this occurs because the
increase in the frictional pressure drop is large enough to overcome the reduction in the pressure drop caused by the
lower mass flow rate. Moreover, when one includes the effects of gravity in this analysis, the decrease in the gravitational
head may be greater than the increase in the frictional pressure drop caused by the two-phase mixture. This causes the
1190 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
total pressure drop to decrease. The total pressure drop along the channel is then the sum of the pressure drops in these
three regimes.
If the total pressure drop ΔP is held constant because of the boundary conditions imposed on the inlet and the outlet,
then the rate of change of the mass flow rate (dm/dt) is given by
where z is the height of the channel, L is its length, A is the flow area, and ΔPf is the total frictional pressure drop. Small
perturbations in the mass flow rate around this value then require that
Now suppose that the change in the mass flow rate is given by Δm = m oe±λt, where m
o is the “normal” mass flow rate
and λ is the “time constant” governing the rate of its growth or shrinkage. Equation 31.27 then leads us to conclude that
λ = ∂( ∆P − ∆Pf ) ∂m
⋅ (Az/L) (31.28)
For the flow in the channel to be stable, small perturbations in the flow rate should shrink or stay constant with time.
This requires the value of λ to be 0 or negative. The resulting condition to avoid a Ledinegg instability is then
Mapping this condition back to Figure 31.14, we can see that a stable operating condition only exists when we stay away
from regime II (B to C). When this analysis is repeated in more detail, it can be shown that any points on the upward
sloping side of the pressure drop/mass flow rate curve will cause the flow to remain stable. Thus, points 1 and 3 are stable
operating conditions, but point 2 is not. Figure 31.15 shows a typical Ledinegg stability map for a BWR-4. This map was
originally created by Richard Lahey. There are many other features of two-phase instabilities. For this reason, a great
FIGURE 31.15 Regions of an operating map where the flow is stable and other regions where it is not. Here the phase change
number is plotted on the x axis and the subcooling number is plotted on the y axis.
31.18 Thermal-Hydraulic Instabilities and Coupled Power-Flow Oscillations 1191
deal of time and effort has been spent trying to explain and control them. A more detailed discussion of their attributes
and behavior can be found in the references at the end of the chapter.
31.17
F low Oscillations during Start-up and Shutdown
If enough heat is generated in the core, and the mass flow rate is low enough, it is possible for Ledinegg instabilities to
develop. These instabilities typically do not occur when a plant is operated normally. However, they can occur during
Start-up, and in some cases, after shutdown. These instabilities occur when the power-to-flow ratio is high and when
the reactor pressure vessel is not fully pressurized. Then the pressure head may cause different combinations of flow
regimes to develop. Small perturbations in these flow regimes can then cause the void fraction to change for the same
value of the pressure drop. This may cause rapid oscillations to develop in the mass flow rate in the absence of sufficient
pumping power. During Start-up, this is primarily a problem for BWRs. Raising the system pressure to its normal value
(~1,000 PSI) before raising the power level is the best way to circumvent these instabilities. However, the power cannot
be raised too rapidly because this may also develop a temporary instability if the pumps are not operating at full power.
In the United States, these instabilities are typically avoided by requiring reactor operators to follow what is called a
ROM. An example of such a map for the Peach Bottom 3 BWR is shown in Figure 31.16. Accordingly, the U.S. NRC has
required GE to adjust the allowable operating range in BWR 4’s and BWR 6’s to avoid operating in Regime II (which
corresponds to the yellow area in the map) during reactor Start-up. A similar problem can also occur after shutdown if
the flow rate in the core is decreased too rapidly. The plant control system can also be designed to compensate for these
instabilities if they develop.
31.18
T hermal-Hydraulic Instabilities and Coupled Power-Flow Oscillations
In addition to any mechanical or structural problems that these instabilities can cause, the primary reason that these
instabilities need to be controlled is that they can affect the power level as well. Any oscillations in the local coolant
density (which are sometimes referred to as density waves) can cause the neutron energy spectrum in a water reactor to
FIGURE 31.16 A picture of the power-flow operating map for Reactor 3 at the Peach Bottom nuclear power plant in Harrisburg,
Pennsylvania. This is an example of what the operating map looks like for a commercial BWR. Power control in BWRs is accom-
plished by changes in the recirculation flow rate rather than by moving the control rods. Operation in the yellow part of the map is
not allowed because of thermal-hydraulic and neutronic instabilities. These instabilities can lead to density waves and Ledinegg
instabilities. (Picture provided by GE.)
1192 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
harden or soften unexpectedly. If the density falls at a particular point in the core (because the coolant boils and there are
too many voids), then the power level at that point will decrease. Conversely, if the coolant density rises at that point, then
the power level will also rise at that point (due to the increased moderation), and this can cause the power level to sud-
denly increase (and then decrease). In American PWRs, this is not necessarily a problem because changes in the coolant
density are carefully controlled, and the coolant is not allowed to boil except near the top of the core during normal reac-
tor operation. However, in BWRs, changes in the coolant density can be much larger (because the coolant is designed to
boil), and this can sometimes create a problem where the flow and the power oscillate if the power-to-flow ratio becomes
too high. Under unusual conditions, this can cause a reactor to inadvertently shutdown or “scram.” In the early days of
the nuclear business (in the 1970s and the 1980s), this actually caused several reactors in the United States and Europe to
scram. In fact, in 1984, the Caorso BWR in Caorso, Italy, actually oscillated in an out-of-phase mode, where half of the
core had the power level increase, while the other half of the core had the power level decrease. This continued to occur
for some time before the plant was finally able to be shut down. The best way to prevent this problem from occurring is
to operate the plant in accordance with the operating map. In the next section, we would like to conclude our discussion
of this effect by showing how the plant power level P, which is related to the reactivity ρ, is affected by these density
wave oscillations. This will require us to draw upon our knowledge of the relationship between the power level and the
coolant density that we developed in the previous chapters. In this context, the reactivity ρ is not the same as the density
although the two of them are closely related. Hence, when the symbol ρ appears in the discussion that follows, it refers to
the reactivity (a measure of the neutron production or destruction rate) and not to the density of the coolant.
31.19
Effects of Temperature on the Reactivity
Reactors are equipped with many different types of control systems (such as LPRMs and APRMs) to prevent rapid
increases in the power level or the temperature from causing the power level to continue to increase even further. The
extent to which a change in the temperature of a reactor’s materials will produce a corresponding change in the power
level is called the temperature coefficient of the reactivity, αT. This coefficient is defined by
so that if αT and ΔT are known, then the change in the reactivity Δρ is given by
∆ρ = α Τ ⋅ ∆Τ (31.31)
As such, αT has the units of per °C or per °K, where a centigrade degree (C) is the same size as a Kelvin degree (K), and
C = K − 273 is the conversion factor between them. It can also be shown that the power level of the core P is related to
the value of the reactivity ρ by
where Po is the steady-state power level, and t is the time in seconds after the reactivity is changed to ρ. The term T = 1/ω
is called the reactor period, and it is a measure of how long it will take for the neutron flux and, therefore, the power
level to increase or decrease by a factor of e or 1/e. We also know that the reactor period T is given by T = Λ(1 − p)/ρ and
ω = 1/T = ρ/Λ(1 − p) where Λ is an important constant for thermal water reactors called the mean neutron lifetime. The
mean neutron lifetime is the time it takes for a new fission neutron (after it is created) to be absorbed or to leak out of
the system. For light water reactors, it generally has a value of about 10 −4 seconds. Thus, a change in the power level of
the core ΔP is related to a change in the reactivity Δρ by
assuming that the reactor is initially operating close to steady-state conditions (where ρ = 0). Under steady-state condi-
tions, the value of the reactivity ρ is very close to 0, and this also implies that
31.20 Void Formation and the Reactor Power Level 1193
which is exactly what one would expect. The temperature coefficient of the reactivity αT can be further subdivided into
three separate coefficients—one for the thermal expansion of the moderator (before it boils), αM,—the second for the
thermal expansion of the fuel, αF, and the third for the increased leakage of neutrons from the core due to the reduction
in the density of the moderator and the fuel, αL. This then allows us to write
α Τ = α Μ + α F + α L (31.35)
or
∆ρ = ( α Μ + α F + α L ) ⋅ ∆T (31.36)
Ordinarily, the neutron leakage coefficient αL is smaller than the other two, and so, we can neglect it in the analysis that
follows. The change in the core power level can then be predicted from the equation
where
☉☉ αM is the moderator temperature coefficient
☉☉ αF is the fuel temperature coefficient
if the reactivity change is not particularly large. Hence, to a first approximation, the change in the power level ΔP due to
a change in the temperature of the coolant is proportional to eα Μ ∆T. For small values of the temperature change and the
elapsed time, the value of the term (αM + αF)/Λ·ΔT·t can be much less than 1.0 because (αM + αM)/Λ for a water reactor is
on the order of about 0.1. The actual values for αM, αM, and Λ for different reactor types are discussed in our companion
book.* Under these conditions, the exponential term in the equation for the power can be expanded in a Taylor Series.
For small values of x,
Since x in this case is given by x = (αM + αF)/Λ·ΔT·t, the change in the power level is then
∆ P ≈ Po ( α Μ + α F ) Λ ⋅ ∆T ⋅ t (31.38)
where ΔT is in °C and t is in seconds. Due to the fact that αM and αF are normally negative, the power level in a water
reactor initially decreases as the temperature increases. Sometimes this is called nuclear temperature feedback.
31.20
Void Formation and the Reactor Power Level
We can look at this problem in a slightly different way for BWRs by considering what happens when the coolant
boils. In this case, the temperature is always at the saturation temperature TSAT so the value of ΔT will not change
as the power level is increased. However, the reactivity still changes because voids (or bubbles) are produced in the
coolant, and these voids then reduce the average density as a function of axial elevation. We can represent the effect
of this density change on the reactivity by defining a new coefficient called the void coefficient of the reactivity α V.
The void coefficient is defined in the following way
and so Δρ = αV·Δα, where α is the void fraction. If we can neglect temperature related effects for the moment while the
coolant is boiling, then the total change in the power level is given by
− α V Λ ⋅∆αt
∆P = Po e (31.40)
1194 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
because the value of αV in water reactors is always negative. Notice that the power level changes in direct proportion
to the change in the void fraction Δα. This equation has enormous practical importance because it means that to a first
approximation, the rate of void formation in the coolant controls the reactor power level ΔP. Fortunately, the value of
αV is always negative in water reactors because a decrease in the density will harden the neutron energy spectrum. So at
least in a BWR, this can help control how quickly the power level increases (or decreases). This is important because it
means that the neutronic instabilities will tend to get damped out as more bubbles are formed. In electrical engineering
terminology, this is an example of how negative power feedback can be used to keep a reactor under control. In fact,
reactor control systems in both PWRs and BWRs are designed to take advantage of this fact when assisting the operator
in controlling the core. However, thermal feedback and void formation are also the underlying source of the neutronic
feedback that causes the Ledinegg instabilities to propagate if the power-to-flow ratio becomes too high. Finally, the
void coefficient can also be used to provide BWRs with some form of automatic load following. For small values of Δα
and small values of the elapsed time, the exponent in Equation 31.40 can also be expanded in a Taylor Series. The power
change due to the formation of the additional voids is then
∆P = − Po α V Λ ⋅ ∆αt (31.41)
In practice, Δα varies from 5–10% (Δα = 0.05–0.10) unless there is a severe overpower transient. So again creating more
voids will cause the power level to fall. Reducing the number of voids (which corresponds to a negative value of Δα) will
cause the power level to rise. These effects are illustrated in Figures 31.17 and 31.18.
31.21
Regulatory Requirements Governing Hydrodynamic Instabilities
For the reasons we have just discussed, the Nuclear Regulatory Commission has adopted the policy that the temperature
coefficient of the reactivity and the void coefficient of the reactivity must be negative across the entire operating range
of a reactor before it can be licensed in the United States. Typically, this is done by increasing the power level in the
FIGURE 31.17 In water reactors, the temperature coefficient of the reactivity, which is normally associated with the fuel, is nega-
tive. As the fuel temperature goes up, the power level goes down, and as the fuel temperature goes down, the power level goes up.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
31.21 Regulatory Requirements Governing Hydrodynamic Instabilities 1195
FIGURE 31.18 In BWRs, the void coefficient of the reactivity is negative. As more voids are created, the power level goes down,
and as fewer voids are created, the power level goes up.
core incrementally, and at each new power level, measuring the increase in the temperature of the coolant and the total
change in the reactivity that is observed. If the reactor passes this test, then the NRC will award the reactor owner with
an operating license. In the United States, the Nuclear Regulatory Commission will not license a reactor unless the
value for αF is negative over the entire anticipated range of operation. The specific regulatory requirements governing
this behavior are spelled out in 10 Compendium of Federal Regulations Part 50 under the General Design Criterion
11 (GDC 11) of Appendix A. This regulation states that “the prompt inherent feedback characteristics of the fuel must
compensate for a rapid increase in the reactivity in the power operating range.” In other words, the prompt temperature
coefficient of the reactivity must be shown to be negative over the entire operating range of the reactor before an operat-
ing license can be granter. The total change in the system-wide reactivity due to a change in the operating temperature
is usually measured separately for the fuel and the coolant, and then the two of them are combined together to generate
an overall temperature reactivity coefficient (see Figure 31.19). The procedure for calculating this coefficient is shown
for a low-power thermal water research reactor in Table 31.3. Once the data has been collected, it can then be plotted in
the manner shown in Figure 31.19.
FIGURE 31.19 The temperature coefficient of the reactivity and the change in the reactivity versus fuel temperature for a typical
thermal water reactor.
1196 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
TABLE 31.3
Measurements of Temperature Coefficient of the Reactivity for a Hypothetical Test Reactor
Average Core
Reactor Power (kW) Temperature (°C) Core ΔT (°C) Core Δρ (cents) αT = Δρ/ΔT (cents/°C)
0.1 25.8 0.0 ~0.0 0
5.0 31.9 3.1 −6.8 −2.19
15.0 35.1 6.2 −9.7 −1.56
25.0 44.3 9.2 −10.7 −1.15
50.0 58.9 14.6 −15.8 −1.08
75.0 82.2 15.9 −23.3 −1.15
100.0 98.4 31.2 −23.1 −1.18
150.0 121.7 23.3 −31.8 −1.24
200.0 147.5 25.8 −33.4 −1.29
250.0 161.3 29.8 −23.3 −1.40
300.0 179.9 23.6 −31.6 −1.53
Once a system-wide temperature coefficient has been found, and has been demonstrated to be n egative everywhere,
a final test must be conducted to demonstrate that αF—the temperature coefficient of the fuel—is also negative in the
intended operating range. Since the temperature coefficient of the fuel reacts immediately to changes in the reactor
power, this represents the first way in which the system can respond to an unanticipated increase in the power level.
This also helps to clarify some of the time frames over which each of these coefficients comes into play. Typical results
for the thermal research reactor we have been discussing are shown in Figure 31.18. The process described here also
applies to commercial power reactors as well. It is important to realize that the temperature coefficients also change with
temperature and that none of the coefficients we have discussed so far are constant over the entire temperature range in
question. So while we have written these coefficients as
α Τ = α Μ + α F + α L + α V (31.42)
The effect of density wave oscillations on the power level of a BWR is accounted for in essentially the same way.
31.22
Void Coefficients for Fast Reactors
In fast reactors the void coefficient of the reactivity behaves differently than it does in thermal reactors. This is primar-
ily due to the fact that there are almost no thermal neutrons in the core. In water reactors, increasing the power causes
the coolant to boil, and this in turn enables the power level to fall. Conversely, reducing the power level will cause the
density of the coolant to rise, and this will also cause the power level to rise. However, exactly the opposite situation
occurs in fast reactors because boiling the coolant causes the power level to increase. So if voids are produced in a
sodium-cooled fast reactor, the neutron energy spectrum actually hardens (i.e., shifts to higher energies), and this causes
more fission reactions to occur. To understand the consequences of this behavior, we can use the previous expressions we
derived to estimate the change in the power level, but the void coefficient of the reactivity αV is now a positive number
rather than a negative one. So when liquid sodium is used as the coolant, a change in the power ΔP is related to a change
in the void fraction Δα by
+ α V Λ′ ⋅∆αt
∆P = Po e (for fast reactors) (31.44)
Hence, a positive change in the void fraction Δα = (α′ − α) causes the power level to rise. Here, α′ is the new void fraction
and α is the previous one. The other challenge involved in designing and operating a fast reactor is that the value of the
mean neutron lifetime Λ is usually much smaller than it is in a thermal water reactor because the neutrons do not live
Bibliography 1197
as long before they are absorbed or leak out of the core. In fact, the value of Λ in most fast reactors is about 10−7 seconds,
which means that any change in the void fraction will cause a change in the power level approximately 1,000 times
larger for the same value of Δα in a water reactor. This explains why the coolant in sodium-cooled reactors is NOT
allowed to boil and why the fuel assemblies are surrounded by protective sheaths or cans. In other words, a Ledinegg
instability will NOT develop in the primary loop of a sodium-cooled fast reactor. Instead, the sodium vapor expands
explosively and creates shock waves which can propagate through the cores of these reactors vertically (see Chapter
23). Because the void coefficient is now positive, the voids continue to grow, and some large, dynamic density waves
can develop. This is an important problem that is generally addressed by keeping the coolant 150°C to 200°C below the
saturation temperature (see Chapter 4). Near atmospheric pressure, the boiling point of liquid sodium is about 882°C.
An excellent discussion of shock waves is presented in Cengel and Boles (2007).
∆ P Po = − α V Λ ⋅ ∆αt
For the BWR, this expression evaluates to ΔP/Po = −|0.0001/0.0001|·0.0001t = −0.0001 when t = 1 s. For the LMFBR,
it evaluates to ΔP/Po = +|0.000001/0.0000001|·0.0001t = +0.001 when t = 1 s. Hence, the power change in the LMFBR
core is ten times greater and positive, while the power change in the BWR core is ten times smaller and negative. The
BWR is clearly a more stable reactor under these conditions because of negative reactivity feedback. [Ans.]
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Knoll, G. Radiation Detection and Measurement, John Wiley and Sons, New York (2010).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Web References
https://www.nrc.gov/docs/ML1125/ML11258A333.pdf.
http://caltechbook.library.caltech.edu/51/01/chap15.pdf.
http://web.ornl.gov/info/reports/1992/3445605784933.pdf.
https://www.nrc.gov/reading-rm/doc-collections/nuregs/staff/sr1793/.../chapter4.pdf.
1198 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
Other References
Guido, G., Converti, J., and Clausse, A. “Density-wave oscillations in parallel channels-an analytical approach”, Nuclear Engineering
and Design, 125(2):121–136, February 1991. ISSN 0029-5493.
Ishii, M. and Zuber, N. “Thermally induced flow instabilities in two phase mixtures”, Proceedings of the fourth international heat
transfer conference, B5.11, 1970.
Ishii, M. “Thermally induced flow instabilities in two-phase mixtures in thermal equilibrium”, PhD thesis, School of Mechanical
Engineering, Georgia Institute of Technology, Atlanta, June 1971.
Kao, H.S., Morgan, T.D., and Parker, W.B. “Prediction of flow oscillation in reactor core channel”, Transactions of the American
Nuclear Society, 16:212–213, 1973.
Ledinegg, M. “Instability of flow during natural and forced circulation”, Die Wärme, 61, 1938.
Saha, P. and Zuber, N. “An analytical study of the thermally induced two-phase flow instabilities including the effect of thermal non-
equilibrium”, International Journal of Heat and Mass Transfer, 21(4):415–426, April 1978. ISSN 0017-9310.
32. How much larger can an undetected local power or flow oscillation be than an average power or flow oscillation in
a modern BWR core?
33. How are BWRs designed to detect these power or flow oscillations and prevent them from occurring?
34. What is the sign of the void coefficient of the reactivity for fast reactors? For fast reactors, does the void coefficient
create a form of positive or negative reactivity feedback?
35. If the reactivity of a commercial power reactor increases, does the core power level increase, decrease, or remain
the same?
36. For a modern BWR, how much larger is the pressure drop for two-phase flow than it is for single-phase flow under
the same operating conditions?
37. What is a typical value of the two-phase multiplier M used in the Martinelli–Nelson correlation?
38. Why is boiling not allowed to occur in LMFBRs?
39. What does a homogeneous model for the two-phase friction factor imply?
40. How can the Darcy equation be used to find the two-phase friction factor in this case?
41. What are representative values of the boiling and non-boiling heights in a relatively modern BWR?
42. What is the highest value of the fluid quality that is usually allowed in a BWR during normal operating conditions?
43. How many flow regimes can there be between a single-phase liquid and a single-phase vapor in a r eactor coolant
channel?
44. What is the difference between a separated flow model and a HEM model?
45. How many conservation equations does a HEM model use?
46. What type of boiling immediately follows the nucleate boiling regime in a reactor coolant channel?
47. How many parts are there to the nucleate boiling regime?
48. What is the condition required to keep a Ledinegg instability from developing?
49. What is an average value of the void fraction at the exit to the core in a BWR?
50. What is the primary difference between annular flow and slug flow?
51. What is the condition required for a vapor bubble to develop on the surface of a fuel rod?
52. How is the speed of sound computed for a two-phase mixture?
53. How many different values of the speed of sound can there be in a two-phase mixture?
54. What is the Atwood ratio and what is it used for?
55. What is the difference between annular flow and mist flow?
56. What does the Lottes–Flynn correlation attempt to correlate?
57. How can you calculate the two-phase pressure drop when you can no longer assume that the two-phase mixture in
a flow channel is completely homogeneous?
58. How is the slip ratio defined in the standard model of two-phase flow, and how is it defined using the drift flux model?
59. What advantage does the drift flux model have over the standard slip ratio model?
60. What is the RELAP 5 computer code and where was it developed?
61. Under what conditions does the drift flux model of Fauske give its best results?
62. Can you think of at least one condition where the two-phase flow in a reactor fuel assembly is not entirely turbulent?
63. What is a typical use for the CATHARE code?
64. What are the fundamental differences between the CATHARE code and RELAP?
65. What is the void coefficient of the reactivity in a water reactor?
66. Is the value of the void coefficient positive or negative in a BWR?
67. What is the mean neutron lifetime used for, and what is the difference between its value in a water reactor and a
LMFBR?
68. What causes the liquid and the vapor in a two-phase mixture to depart from thermal equilibrium?
69. How can the two phases be brought back into thermal equilibrium when they are initially in separate states?
70. Under what conditions does the flow in a two-phase system become choked?
71. What does the term CFM refer to?
72. For a given value of the fluid quality, does the void fraction increase or decrease as the slip ratio is increased?
73. Name three of the mainstream safety analysis codes used in the United States
74. What is a LOCA and why is it so important?
75. In two-phase flow, what is the two-phase multiplier M and how does it work?
76. How many equations are there in a two fluid model (excluding the equations of state)?
77. Under what conditions will a LOCA generate a shock wave?
78. What is the biggest single drawback to the Lottes–Nelson correlation?
79. What factors eventually caused the Martinelli–Nelson and the Lottes–Flynn correlations to be replaced by the more
modern correlations for the two-phase multiplier that exist today?
1200 Flow Oscillations, Density Waves, and Hydrodynamic Instabilities
8 0. What is the relationship between the void fraction and the quality in a flowing system?
81. What is a typical value for the two-phase multiplier in a BWR?
82. What does the term critical flow refer to?
83. Is there any difference in the critical flow rate through a long pipe and a short pipe with the same cross-sectional
area and the same back pressure?
84. How can we increase the critical flow rate through an orifice in which the maximum velocity of the coolant is
limited?
85. For the same bubble radius R, is the bubble generation rate higher or lower on a well-wetted surface than it is or a
poorly wetted one?
Exercise 31.2
In BWRs, there can be significant density differences between the top and the bottom of the core. These density differ-
ences can cause hydrodynamic instabilities to occur. Calculate the density change between the top and the bottom of
the core for the BWR described in Appendix B. Calculate the equilibrium quality at the core inlet, the core outlet, and
the core midplane.
Exercise 31.3
Coolant enters a BWR fuel assembly from below with a velocity of 2 m/s. The temperature of the coolant is 278°C and
the operating pressure is 7.25 MPa. What is the saturation temperature of the coolant, and at approximately what axial
elevation does bulk boiling in the fuel assembly first occur?
Exercise 31.4
For the fuel assembly described in Exercise 31.3, calculate the mixture velocity of the coolant at the core midplane and
at the core exit. If the slip ratio at the core midplane is 1.5 and the slip ratio at the core exit is 3.5, calculate the velocity
of the steam at the core midplane and the core exit. How many times greater are these velocities than the liquid velocity
at the core inlet?
Exercise 31.5
Using the Griffith correlation (see Chapter 25), calculate the distance upstream of the equilibrium quality point where
NVG first occurs. Assume the heat flux at this location is 1 × 106 W/m2 and the convective heat transfer coefficient is
250 kW/m2 °C. Why is the equilibrium quality not necessarily an appropriate measure of the true flow quality at this point?
Exercise 31.6
A commercial BWR is operated in such a way that the phase change number is 5. What must the value of the subcooling
number be for the core to be hydrodynamically stable? If the saturation temperature is 288°C, what value of the inlet
enthalpy is required for this to occur? What value of the inlet coolant temperature is required to create a hydrodynami-
cally stable system?
Exercise 31.7
Using the information provided in Example Exercise 31.3, calculate the phase change number when the average heat flux
is increased to 1,200 kW/m2 and the average mass flux is increased by 30%.
Exercise 31.8
Calculate the subcooling number for a commercial BWR where the core inlet temperature is 270°C and the core outlet
temperature is 288°C. Assume the core is operated at 7.25 MPa (1,050 PSI). How does the s ubcooling number change
as the inlet enthalpy is decreased?
32
Containment Buildings
and Their Function
32.1 Containment Buildings and Their Function
The final topic we would like to discuss in this book is the reactor containment building. Normally, not enough attention is paid to these
important structures even though they are the final barrier that prevents radioactive material from being released into the environment in
the event of a reactor accident. In this chapter, we will discuss the design and function of reactor containment buildings in general, and
then we will discuss some reactor-specific containment buildings. Later in the chapter, we will discuss how the containment building
responds to a reactor accident in more detail. Historically, each type of reactor has had its own specific containment building, although the
design of these containment buildings has tended to converge over the years. In this chapter, we will concentrate primarily on the design
of pressurized water reactor (PWR) and boiling water reactor (BWR) containment buildings. However, other reactors, such as gas-cooled
reactors, use similar designs for their containment building as well. The amount of energy that can be released from the core in the event
of a reactor accident or loss-of-coolant accident (LOCA) generally determines the size or volume of the containment building. Equations
that can be used to predict how a containment building responds to a LOCA are discussed in Chapter 34. Normally containment
buildings are designed to withstand internal pressures of about 60 PSI (~ 415 kPa). In most accidents, however, the peak pressures
rarely exceed 40 PSI. Hence, containment buildings also have safety margins just like the safety margins that regulatory authorities
use to license commercial reactor cores. In addition, more modern containment buildings also use passive cooling systems to remove
the heat that is generated by a LOCA. These systems do not require human decision making to function properly.
Simply put, the purpose of a containment building is to prevent the release of radioactive material to the environment in the event
that a reactor accident occurs. There are many different types of containment buildings, and the role of these containment buildings
has continued to evolve over time. However, the designs of most of these buildings have tended to converge into one or two standard
architectures or designs as the industry has become more sophisticated and mature. In this chapter, we would like to describe the
role of the containment building in more detail and to discuss some of its design principles in more detail. The relationship of the
containment building to the reactor pressure vessel is also discussed.
1201
1202 Containment Buildings and their Function
much higher temperature and pressure (~2,250 PSI and ~330°C) than the surrounding temperature and pressure (~10 PSI
and ~30°C). A relatively severe LOCA can also occur if one of the secondary loops connected to the SGs fails. However,
the temperature and pressure of the resulting water–steam mixture is somewhat lower than it is on the primary side
(~1,000 PSI and ~285°C), and because of this, the total energy released is less as well.
The size and severity of the shock wave that develops from this pipe break depends on the size of the pipe break and
exactly where the pipe break occurs. If it occurs in the hot leg of the reactor pressure vessel subcompartment (where the
pressurizer is also located), it is generally modeled as a double-ended guillotine pipe break involving only the discharge of
water from the primary loop. This normally produces the most severe temperatures and pressures, although relatively HPs
can also be found in one of the subcompartments containing the reactor coolant pumps (RCPs) and the SGs if this is where
the pipe breaks happen to occur. The buildup of the pressure in one of these subcompartments as a function of time is shown
in Figure 32.2. Notice that the subcompartment pressure reaches its maximum value about 1 s after the pipe break occurs.
Containment 32
Differential pressure (psi)
SG
Pressurizer 24
RCP RCP
SG
Core
16
SG
RCP 8
0
Cross-sectional view of the subcompartments 0 1 2 3 4 5 6
(a) Time (seconds after the accident)
(b)
FIGURE 32.2 (a) The time-dependent differential pressure buildup in the vicinity of a large break LOCA near the entrance to the
SG compartment in a typical 1,000 MWE PWR. A PWR of this size normally contains three or four SGs, and each of these is housed
together with the RCPs in a separate containment subcompartment. A cross-sectional view of some of the subcompartments is also
shown. Note that most modern PWRs have between 6 and 12 separate subcompartments depending upon the design of the containment
building itself. The differential pressure shown is the pressure difference between the SG subcompartment and the rest of the
containment building. (From North Anna UFSAR.) (b) Shapes of some common PWR containment buildings in widespread use today.
32.4 Containment Building Missile Defense Shield 1203
Differential pressure levels in the subcompartment will usually reach a peak value between 25 and 30 PSI. However,
the local values of the pressure may be somewhat higher immediately in the vicinity of the pipe break itself. In order to
keep these pressure levels from exceeding the design limit of the building, most reactor subcompartments within the con-
tainment building are designed with what is called a blowdown panel. When the pressure level in the subcompartment
exceeds the maximum design pressure (which is usually determined by the difference between the maximum pressure
and the ambient pressure), the blowdown panel allows the excess pressure to escape to the next largest subcompartment
in the reactor containment building. This effectively limits the rate and speed of propagation of the LOCA. One other
interesting effect that occurs during a PWR LOCA is that the hot two-phase mixture that leaves the pipe break does so
at such a high rate of speed that the flow becomes choked. When this occurs, the rate of propagation of the shock wave
cannot exceed the local Mach number, Ma = v/c, for the water–steam mixture, where v is the velocity of the flow and c
is the speed of sound of the mixture. In other words, the velocity of the escaping fluid is limited to a Mach number Ma
less than or equal to 1.0 (v ≤ Ma of 1.0). In fluid mechanics, this condition is sometimes referred to as choked flow. When
the flow becomes choked, the mass flow rate dm/dt from the pipe break is no longer a function of the square root of the
pressure difference between the inside and the outside of the pipe. Then the methodology presented in Chapter 29 must
be used to determine the rate at which the hot water-steam mixture exits the pipe.
(F = ma) (32.1)
dp
F= (32.2)
dt
where dp/dt is the rate of change of momentum of the object. If the momentum of the object is p = mv and the mass m
of the object does not change, we can write Equation 32.2 as
d(mv) mdv
F= = = ma (32.3)
dt dt
Thus, the force the object exerts on the building is proportional to how fast the velocity changes, dv/dt = Δv/Δt. In the
field of collisions, the impulse is equal to this momentum change:
Thus, the impulse depends on how fast the relative velocity between the two objects changes. Moreover, Equation 32.4
shows that the impact force is directly proportional to the mass m, directly proportional to the velocity change Δv, and
inversely proportional to the collision time Δt. Hence, doubling the mass of a projectile (from 1 to 2 m) will double the
force and halving the time (from Δt to Δt/2) will also double the force. Doing both at the same time will cause the
impact force to be four times as great. The pressure P that a flying object exerts on the containment building can be
found by dividing the impulse by the cross-sectional area A of the containment building, which is affected by the fol-
lowing collision:
So in effect, this is the expression that mechanical engineers use to determine the peak pressure that a flying object
exerts on a containment building when it collides with it. Now, let us use this expression to show you how much pressure
a flying telephone pole can exert on the containment building in a direct collision.
Solution In this collision, we know that m = 4,400 kg, Δv = vinitial − vfinal = 45 m/s, A = π(0.15)2 = 0.07 m2, and Δt = 0.1 s.
The pressure P that the telephone pole exerts on the containment building can be found from Equation 32.5. Plugging
in the necessary numbers, we find that
[Ans.]
FIGURE 32.3 The BWR containment buildings at Fukushima, Japan, prior to the earthquake in 2011. The containment buildings
are the four small square buildings in the middle of the picture. (From Wikipedia.)
The secondary building is maintained slightly below the atmospheric pressure (10.7 PSI). This lower pressure keeps
unwanted gases from escaping from the building during normal operation and during refueling. In BWRs, the contain-
ment buildings have evolved considerably over the years. Today, there are three different generations of BWR containment
buildings in widespread use, and they are referred to as the Mark I, Mark II, and Mark III containments. Each one of these
designs represents an evolution in the way that containment buildings are designed.
Spent fuel
pool
Reactor service
floor
Concrete reactor
building
Reactor pressure
vessel
Primary containment
drywell
Suppression pond
wetwell
FIGURE 32.4 A picture of a Mark I containment building used in early BWRs. (Courtesy of Ge.)
32.10 Mark III Containment 1207
FIGURE 32.5 A picture of a Mark II containment building used in more modern BWRs. (Courtesy of Ge.)
*Reactor building*
1. Shield building
2. Freestanding steel containment
3. Upper pool
4. Refueling platform
5. Reactor water cleanup
Mark III containment 6. Reactor vessel
7. Steam line
8. Feedwater line
9. Recirculation loop
2 10. Suppression pool
11. Weir wall
12. Horizontal vent
15
13. Daywell
4 1 14. Shild wall
15. Polar crane
5 *Auxiliary building*
16. Steam line tunnel
17. NHR system
6 18. Electrical equipment room
7 14
16 24 23
8 13 21
29
22 19
18
25
9 11
12
17 10
*Fuel building*
19. Spent fuel shipping cask
20. Fuel storage pool
21. Fuel transfer pool
22. Cask loading pool
23. Cask handling crane
24. Fuel transfer bridge
25. Fuel cask srid on railroad car
FIGURE 32.6 A picture of a Mark III containment building in a BWR with all of its subcomponents. (Courtesy of Ge.)
Feeder
pipes
Condenser
Feed
Control rods water
pumps
Torus
Suppression pool
FIGURE 32.7 A cross-sectional view of the NSSS and the suppression pool in a BWR. (Picture provided by Ge.)
Moisture separator
Secondary and rehector
loop
Generator
Turbine
Pressurizer
Circulating pump
Steam
generator
Condensate pump
Coolant
pump
Tertiary loop
Reactor
Primary loop
Containment
wall
FIGURE 32.8 The NSSS for a typical four-loop PWR. Only one loop is shown in the picture. Notice that the heat content of
the coolant divides itself rather neatly into three distinct regions—one for the core, one for the SGs, and one for the condensers.
(Courtesy of Westinghouse Nuclear energy Systems.)
3
Wturbine
Qout
Wpump
2
1
Qin
FIGURE 32.9 A pictorial representation of the four steps in the Rankine thermal cycle. (Picture provided by Wikipedia.)
from the time the steam leaves the reactor core or the SGs until the time it reaches the power turbines, generates electric-
ity, and returns to be reheated again. The steam turbines are coupled to an electrical generator by a long rotating shaft.
Spinning the shaft at high speed causes the electrical generators attached to the steam turbines to produce electricity. The
pressurized steam hits the blades of the turbines at a very high speed, and this causes them to turn. An example of what
these blades look like is shown in Figure 32.10. The drier the steam (and the higher energy content it has), the faster the
blades will spin. The steam is always delivered to the blades of the turbine as dry as possible, and after it hits the blades,
it starts to lose energy and subsequently begins to recondense. The pressure drops, and the hot steam can no longer be
used to drive the turbine blades. At this point, it is sent to the condenser where it is then returned to the SGs or to the core.
To increase the thermal efficiency of most plants, the steam is generally sent through turbines which consist of two
separate stages—an HP stage and a low pressure or LP stage. First, the steam is sent through the HP stage, and then,
after some of its energy is lost, it passes through a moisture separator before being sent on to one or two LP stages. The
role of the moisture separator is to remove water droplets from the steam before it enters the LP stages. The remaining
steam and water mixture must then be returned to the core or the SGs to be reheated again. Before doing so, it passes
32.13 Electrical Generators 1211
FIGURE 32.10 A large steam turbine used in a nuclear power plant. (Courtesy of Siemens.)
through an intermediate device called a condenser, which condenses the remaining steam and discharges the waste heat
to a cooling pond, a river, a lake, or a cooling tower, which allows the thermal cycle to be completed. The efficiency
of the thermal cycle, η, is then the amount of electric power produced per second, P, divided by the amount of heat, Q,
produced per second by the reactor core. Hence, the efficiency of any power plant thermal cycle can be written as
As we discussed previously, both P and Q are usually expressed in units of watts. P is usually the number of megawatts of
electrical energy produced, and Q is typically the thermal output of the power plant in megawatt thermal (MWT). Hence,
P is usually given in terms of MWE, and Q is usually given in terms of MWT. However, as we previously discussed, the
ratios are essentially fixed for a given reactor type. The efficiency of the power plant can be then be increased by using
some “tricks” to recycle some of the remaining waste heat. Two standard techniques used to do this are called reheat and
regeneration. Exactly how this is done is described in many reactor thermal hydraulics books (see, for example, Todreas
and Kazimi (2008)), and it is also discussed in great detail in Chapter 9 of this book. The specific implementation of
the reheating process in a nuclear power plant is shown in Figure 32.11. An example of how the regeneration process is
implemented in a nuclear power plant is shown in Figure 32.12. Reheat and regeneration can also be used in coal-fired
power plants. The thermal efficiencies of most common reactor types are compared in Table 32.1. Note that it is very
difficult to achieve an overall thermal efficiency in excess of about 42%.
FIGURE 32.11 A simplified diagram of the reheat process in light water reactor. The reheating process occurs in the reheating
section between the LP and HP stages.
FIGURE 32.12 A simplified diagram of the regeneration process in a light water reactor. The regeneration occurs in the red box
involving the feedwater heaters.
TABLE 32.1
Thermal Efficiencies for Various Types of Nuclear Power Plants
Coolant in the Primary Coolant Inlet Coolant Outlet Temperature Difference Overall Thermal
Reactor Loop Temperature (°C)a Temperature (°C)a (°C) Efficiency η (%)
CANDU Heavy water 270 310 40 28–30
BWR Light water 275 285 10 32–34
PWR Light water 290 325 35 33–34
LMFBR Liquid sodium 400 550 150 40–42
AGR Carbon dioxide 290 640 350 42–44
HTGR Helium 320 740 420 46–48
Source: Todreas, N. and Kazimi, M.S., Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2012).
a Primary loop.
32.14 Condensers 1213
(a) (b)
FIGURE 32.13 A large electrical generator used in a dam (a) and (b) a similar generator used in a nuclear power plant.
Note the immense scale and size that these electrical generators can have. (Courtesy of Wikipedia.)
of about 20,000 V AC (Figure 32.11). The alternating current is then sent on to an electrical transformer, where the
voltage is increased to either 230,000 or 345,000 V. The power is sent to a switchyard, where it is further filtered,
refined, and sent to a customer site. Normally, each steam turbine (or power turbine) has a separate electrical gen-
erator attached to it, but this does not always have to be the case. The exact configuration depends on the design of
the power plant and how much electricity it is designed to produce. For example, most modern nuclear power plants
have an electrical output between 500 and 1,500 MWe and are operated as base-loaded units. Each primary loop
then has a separate set of power turbines attached to it. Most modern designs have two electrical generators attached
to the power turbines. Therefore, a two-loop plant can have four electrical generators, a three-loop plant can have
six electrical generators, and a four-loop plant can have eight electrical generators, and so on. This achieves some
modularity in the overall design of the plant, and it also allows one to scale up the capacity of the plant by simply
adding more identical components.
32.14 Condensers
From the steam turbines, the spent steam is then passed on to a device called a condenser. Condensers are passive
devices whose sole purpose is to condense the steam into water before the water can be returned to be reheated again.
A condenser can be thought of as a large cooling tank with no moving parts. A picture of a typical condenser is shown
in Figure 32.14. In this case, the condenser is cooled by water pumped into it from a nearby river, a lake, and, in some
cases, an ocean. Inside of the condenser, the steam passes over a large number of very cold tubes. These tubes contain
the water that is being pumped into it from an external source. As the steam passes over the tubes, it is condensed back
into ordinary water, and it is then sent back to the SGs or the core. The steam enters the condenser with a high energy
content and leaves the condenser with a low energy content. (Sometimes, this energy content is called the enthalpy of
the fluid.) The mass flow rates from the primary and secondary sides are carefully adjusted so that the amount of heat
that enters the condenser is exactly balanced by the amount of heat that leaves the condenser. The mass flow rates are
generally found by doing a simple energy balance. The enthalpy used to perform this energy balance has the units of J/
kg. (The exact value of the enthalpy for a water–steam mixture can be found in a compendium of thermodynamic prop-
erties known as the stream tables.) The enthalpy in these calculations is usually given the symbol h. Various versions of
the steam tables are available on the Internet and in elementary thermodynamics books.
It is also possible to bypass the condenser directly and dump the spent steam back into the atmosphere or into another
“heat sink” outside of the power plant. However, when this is done, the cooling water cannot be recycled or reused again.
Most of the time, the cool water from the condensers is simply pumped back into the plant, where the whole process is
repeated again. An NSSS that does this is called a closed-loop design. After the condensed water leaves the condenser, it
is pumped through a water purifier called a demineralizer before it is sent back to the steam-producing side of the NSSS.
This removes any impurities that may have been added to the water by the previous stages. Sometimes, the condensed
water that enters the water purifier is called the condensate.
1214 Containment Buildings and their Function
Flanges
Cooling water
from river or lake
Hotwell
Return to river or
lake Flanged Tubesheet Tubesheet
(or cooling tower) cover
plate
Water inlet Condensate
Condensed water to pumps
FIGURE 32.14 Two views of the condenser in a nuclear power plant. The picture on the left represents a simplified view of the
condenser and the picture on the right represents the actual condenser. (Courtesy of Wikipedia.)
FIGURE 32.15 A feedwater heater from a Russian VVER reactor. (Courtesy of Wikipedia.)
Solution The flow rate in the secondary loop can be determined by a simple energy balance. For the heat to be continu-
ously removed from the reactor core, we require that ΔEprimary = ΔEprimary or mprimary × Δhprimary = msecondary × Δhprimary,
where m is the mass flow rate and Δh is the enthalpy. Solving for msecondary gives
( )
m secondary = m primary × ∆h primary ∆h secondary = 15,000 × (0.2/2.6) = 1,155 kg / s
Thus, the flow rate in the secondary loop is 13 times smaller than the flow rate in the primary loop. [Ans.]
Fill
Drift
eliminators
Casting
Louvers
Collecting dash
FIGURE 32.16 A cross-sectional view of an induced draft cooling tower that is used in a nuclear power plant.
(From DOE-Handbook-1018/2-93.)
1216 Containment Buildings and their Function
FIGURE 32.17 A picture of some induced draft cooling towers used by nuclear power plants.
This is a picture of a single natural draft cooling tower that is used at a European nuclear power plant. A natural draft
cooling tower differs from an induced draft cooling tower because it has no moving parts. This particular tower is about
400 ft (120 m) high. The differential pressure difference between the cold outside air and the hot humid air on the inside
of the tower is used as the driving force to cool the steam. No circulating fans are required in a design of this type. (The
PCCS that is used in the AP-1000 is based on a similar principle.)
Whether a natural or mechanical draft cooling tower is used depends on climate that a nuclear power plant is sup-
posed to be operated in. In colder climates, natural draft cooling towers tend to be the more popular than induced
draft cooling towers are. In this particular plant, the water used to feed the cooling tower is supplied by the river on
the left-hand side of the plant. For comparison, the picture on the right-hand side of Figure 32.17 shows the Watts Bar
Nuclear Plant in Knoxville, Tennessee. This particular power plant has two cooling towers, and they are again fed with
water from the Tennessee River (near the top of the picture). This particular power plant is operated by the Tennessee
Valley Authority, and it is used for electric power generation and tritium production. The cooling towers at the Watts
Bar Power Plant are 506 ft (150 m) high and have a diameter at the base of 405 ft (about 120 m). These two pictures can
give you a good idea of how large a reactor cooling tower can be. As a point of comparison, the Empire State Building,
one of the highest buildings in the North America, is only about 1,473 ft (449 m) high. However, it is also much narrower
at the base.
FIGURE 32.18 A picture of a spent fuel pool in which spent fuel assemblies are stored. (Courtesy of the American Nuclear
Society—see www.ans.org.)
where the electrons emitted from the spent fuel rods emit photons in the visible part of the electromagnetic energy
spectrum (in the form of light blue light) as they move. This is part of the natural process of beta decay where the fis-
sion products transform themselves into more stable and less radioactive elements. The tops of the fuel assemblies are
about 6 m (~20 ft) beneath the surface of the pool. The pumps attached to the pool then remove heat from the pool and
replace it with colder water from below. Exactly how this is done depends on the size and the shape of the pool. The
colder water flowing over the surface of the fuel rods cools them through a process known as natural convection. It
does not require a pump, a fan, or another mechanical device to occur. Hence, it is a passive cooling process as opposed
to an active one.
4
1. Fuel-handling area 7. Reactor vessel
3 2. Concrete shield building 8. Integrated head package
3. Steel containment
4. Passive containment 9. Pressurizer
cooling water tank 10. Main control room
2
5. Steam generators (2) 11. Feedwater pumps
6. Reactor coolant pumps (4) 12. Turbine generator
1
5
12
9
6
8
10
11
FIGURE 32.19 A visual depiction of the AP-1000 containment building and auxiliary building built by Westinghouse. A typical
PWR containment building can be treated as a radiological barrier. Note that later versions of this building are considerably more
complex than the simple model shown here. Sometimes the spent fuel pool is located inside of the containment building and
sometimes it is not. (Courtesy of Westinghouse Nuclear.)
usually about 4 m long. The primary difference between the 412-NSSS and the 414-NSSS is that the fuel assemblies in
the 414 are 2 ft longer (14 vs. 12 ft). The 414 can also use an 18 × 18 fuel assembly.
The tertiary loop in each of these designs then contains a heat rejection system, which rejects the waste heat into a
nearby lake, a river, or an ocean, and a condenser, which returns the condensed steam (in the form of feedwater) back
to the SGs again. A simplified schematic of this design was presented in Chapter 1. All PWRs employ the same basic
architecture. The design parameters for each of the four Westinghouse NSSS systems in widespread use in the world
today are shown in Table 32.2. These designs are probably used in at least half of the world’s commercial nuclear power
plants. As we shall soon see, the use of SGs allows a PWR to prevent any radioactive materials released by the fuel rods
from getting into the secondary loop and the steam turbines. This also allows the plant architect to confine the radioac-
tive materials to a single building during normal power operation, and this eliminates the maintenance problems that
would occur if radioactive steam were passed directly into the steam turbines.
TABLE 32.2
Design Parameters for the Most Common NSSSs Offered by Westinghouse Today
Design Parameters for the Most Common Versions of the Westinghouse NSSS
Model 212 312 412 414
Number of loops 2 3 4 4
NSSS power output (MWT) 1,882 2,785 3,425 3,819
Electrical output (MWE) 600 900 1,150 1,280
Steam pressure (secondary side) 920 960 1,000 1,100
SG model F F F H
Pressure vessel diameter (inner) 3.35 3.99 4.39 4.39
RCP 93A1 93A1 93A1 93A1
Average pump horsepower 7,000 7,000 7,000 7,000
Hot leg ID (cm) 73.7 73.7 73.7 73.7
Cold leg ID (cm) 69.9 69.9 69.9 69.9
Number of fuel assemblies 121 157 193 193
Fuel assembly length 3.65 3.65 3.65 4.37
Fuel assembly type 16 × 16 17 × 17 17 × 17 17 × 17
Source: Westinghouse Nuclear.
* See An Introduction to Nuclear Reactor Physics by R.E. Masterson, CRC Press (2017).
1220 Containment Buildings and their Function
during the normal operation of the plant. These gases escape from the fuel primarily through the process of molecular
diffusion, which occurs at a very slow rate. If they happen to escape from the surface of the UO2 fuel pellets, then they
will tend to accumulate in the fuel-cladding gap between the fuel and the cladding. They also accumulate in a small
empty cavity called a plenum that is located at the end of each fuel rod. The plenum is a generic feature of most modern
fuel rod designs, and in fact, several patents have been filed over the years to keep the design of some plenums protected
and proprietary.
The release of fission gases from the fuel depends on two primary factors—(1) the temperature of the fuel and (2) its
burnup. Up to a burnup of about 25,000 MW days per ton, the UO2 tends to be very good at holding the gases inside of it,
and only 1% of them escape naturally (due to the process of molecular diffusion) into the fuel-cladding gap. This is true
as long as the temperature of the fuel pellet is less than about 3,000°F (1,700°C). At temperatures above this, the UO2
begins to recrystallize, and a large percentage of the gases (between 50% and 90%) can be released before the fuel melts.
After the melting point is reached, 100% of the fission product gases are released. It must be kept in mind that not all of
the gases in the fuel rod may be released at the same time. The axial and radial temperature distributions in the rod are
never completely uniform. Therefore, the gases within the fuel pellets are likely to be released first about three-fourths
of the way between the bottom of the rod and the top (where the rod is hottest), and even then, only the inner layers of
the fuel may melt while the outer layers may not (see Chapter 11).
To keep them dimensionally stable as a reactor is being operated, the change in the average temperature of the pres-
sure vessel is usually limited to about 25°C (~45°F) per hour. The design standards for a modern LWR pressure vessel
are spelled out in detail in Section III of the ASME Boiler and Pressure Vessel Code. When the pressure in the pressure
vessel exceeds the design limit (about 10% higher than the nominal operating pressure), the pressure vessel (or the cooling
system), depending upon the reactor type in question, is designed to “vent” the gases produced in the pressure vessel (usu-
ally from the buildup of hydrogen gas as the cladding begins to melt). This venting process is designed to keep the liquid
in the pressure vessel from leaking out and exposing the core further. In both Three Mile Island (TMI) and Fukushima,
the pressure vessel “venting” system operated normally and vented the excess pressure caused by the buildup of excess
hydrogen in the core into the reactor containment building. In other words, these were “designed events” in which the
containment building was supposed to contain the radiation until it could be removed by automatic scrubbers and other
containment injection systems. However, in the case of Fukushima, hydrogen gas actually caught on fire.
Reactor building
1. Reactor
7 8 2. Storage pool
3. Containment dome
6 4. Shielding plugs
5. Hatch
5 6. Bridge crane
4 7. Refueling platform
3 8. Fuel storage pool
2
1
30
29
21. Containment vessel
28 22. Wet well
27 23. Heating and ventilating units
26 24. Steam tunnel
25 25. Feedwater
26. Main steam
27. Recirculating system pump
28. Dry well
24 29. Sacrificial shield
23 30. CRD module
22 21
FIGURE 32.20 An example of the Oyster Creek BWR containment building. A typical BWR containment building that can be
treated as a radiological barrier. Note that this building is similar to the one that was used at the Fukushima nuclear power station in
Japan. (From The U.S. NRC.)
1222 Containment Buildings and their Function
Therefore, the containment building must be large enough to accommodate this increased pressure until the steam is
eventually able to condense back into liquid form. In newer designs, the pressure in the containment building is designed
to be slightly lower than the normal atmospheric pressure outside of the building. This guarantees that any minor amounts
of radiation that leak into the building will not leak out of the building because of the pressure difference. The “air” in the
containment building is also circulated through a number of filters to absorb airborne particulates and to remove excess
radioactivity. Finally, most newer designs have what is known as a “containment suppression system” consisting of a
large number of cold water injectors at the top of the containment building to condense steam released during a LOCA.
32.26 T he Final Line of Defense: The Site Location and Exclusion Zone
Finally, all modern reactors are located at sites that provide some degree of isolation from the general public and
population centers. This constitutes the sixth and final line of defense in an in-depth defense strategy. In other words,
they are not located with 30 or 40 miles of a major metropolitan area. If radiation is released from the plant, then the
majority of this radiation will be carried away by the wind, and it will diffuse according to a combination of the inverse
square law and molecular diffusion that was discussed in Chapter 28. In other words, the further one moves away from
the plant, the less the concentration of the radiation will be. Most nuclear power plants are surrounded by an exclusion
zone that is about 2 miles (or 4 km) wide. The general public is not allowed to enter this zone, except when a reactor tour
in progress. The normal exclusion zone is surrounded (by statute) by a high metal fence. Other security systems, such
as dogs and security guards, may also be used. In addition to the normal exclusion zone, the utility owners and opera-
tors, as well as local civil defense officials, have the option of extending the exclusion zone out further in case a reactor
accident becomes serious. In the case of the accident at Fukushima, this zone was extended to a radius of about 15 km
(or 10 miles) from the site of the plant.
In general, this is an exothermal reaction that produces much more energy that it consumes. At cladding temperatures
above 1,200°C (~2,200°F), the reaction becomes so violent that it becomes hard to control. At these temperatures, the
FIGURE 32.21 A hydrogen gas monitoring system in a modern reactor containment building. (Courtesy of Google Images.)
32.27 Preventing the Buildup of Hydrogen Gas 1223
energy produced by combining the zirconium metal and the oxygen together is about 580 kJ/mol of zirconium metal
or about 6.3 MJ/kg. In 2011, the zirconium–water reaction was ultimately the cause of the containment building
failures that were observed in reactors 1 and 3 at the Fukushima site. The hydrogen ignited, and this is what caused
the explosions in the containment buildings to occur. The reactor pressure vessel itself did not explode in this case.
Refer to Figure 32.22 for some pictures of what the reactors looked like after the roofs were blown off the containment
buildings.
In the case of the TMI, some of the gas was simply vented to the atmosphere (unfortunately right into a helicop-
ter containing a number of TV reporters that were equipped with a TV camera and a Geiger counter). In the case of
Fukushima, the containment building (for an older BWR) was equipped with hydrogen igniters to burn off the hydrogen
as it was produced, but because the electric power to the igniters was cut off after the plant was shut down and the emer-
gency generators were taken out by the ensuing earthquake and tsunami, there was no power left to turn on the igniters
and burn off the gas in a controlled manner. The hydrogen gas thus continued to build up until it blew the top off Mark
I containment buildings. In retrospect, this entire scenario could have been avoided by simply locating an emergency
generator on the cliffs immediately above the site, where the waves from the tsunami would not have been able to take it
out. It could also have been avoided by driving down to the Japanese equivalent of the local Ace, Home depot, or Lowes
hardware store and buying a portable Honda generator (which were also manufactured in a plant less than 100 km away)
and using it to power the hydrogen igniters.
Unfortunately, the power plant supervisor and the local authorities were not trained to think “out of the box” and
improvise in this way. This lack of creativity in addition to the decision to shut down the plant entirely immediately after
the earthquake and tsunami ultimately resulted in the disaster that was seen on television. Many uranium mining and
resource stocks subsequently tumbled after the disaster itself, and the value of TEPCO stock (the utility that owned the
plant) fell by almost 50% in a single day. However, both of them have subsequently recovered most of their value, and
some of them have even gone higher. In the future, the U.S. NRC will probably require all nuclear power plants in the
United States to have hydrogen igniters with multiple power sources, so that if the main power and diesel generators fail,
the plant will still be able to burn off the excess hydrogen gas produced by the zirconium–water reaction. Hence com-
mercial reactor containment buildings must be designed today so that it is very difficult (or nearly impossible) to release
any radiation from the plant - even after a serious accident occurs. This radiation simply stays inside of the containment
building until it decays away. The remaining substances which are produced by this decay process eventually become
inert and are generally NOT radioactive. The amount of time this takes to occur depends on severity of the accident as
well as the type of fuel that is used. After a small accident (corresponding to a Category 4 and 5 accident on the INES
scale - see Chapter 33), the reactor can generally be brought back into service in a year or two after the accident occurs.
The INES scale and the accident categories it uses are discussed in Chapter 33. Only two or three reactor accidents since
the beginning of the commercial nuclear power industry—including Chernobyl and Fukushima—have achieved the
highest level of severity (level 7) on the INES scale. In these accidents, significant amounts of radiation were released
into the environment. The accident at TMI only received a rating of 5 on the INES scale. The types of radiation that
FIGURE 32.22 A picture showing the roofs that were blown off the containment buildings for units 1, 3, and 4 at the Fukushima
site. The damage to unit 4 (far left) was apparently caused by some hydrogen gas that leaked into the unit from unit 3 (further to the
right). (From Wikipedia.)
1224 Containment Buildings and their Function
were released from the accident at Chernobyl are discussed in our companion book*. Only about 10% of the radioactive
materials from the core of the Chernobyl plant were actually released.
Core Core
Hot water
FIGURE 32.23 An idealized PWR containment building before and after a reactor LOCA.
* See Nuclear Engineering Fundamentals: A Practical Perspective, by Robert E. Masterson, CRC Press, Boca Raton, FL (2017).
32.28 PRESSURE, TEMPERATURE, AND ENERGY RELATIONSHIPS 1225
where
☉☉ P is the pressure of the gas
☉☉ V is the volume of the gas
☉☉ n is the number of moles of gas
☉☉ R is the universal gas constant, which is given by the product of Boltzmann’s constant and Avogadro’s number
☉☉ T is the temperature of the gas (in °K)
In Equation 32.8, P is measured in Pa, V is measured in m 3, n is measured in moles, and T is measured in degrees
Kelvin. In SI units, the universal gas constant R has a value 8.314 J/°K mol. Of course, if we want to estimate the
containment building pressure using the ideal gas law, the estimate will have some errors associated with it, but
when the air and the steam–water mixture come into thermal e quilibrium with each other (which means that they
have eventually the same temperature and pressure), then we can deduce the final temperature of the air–steam
mixture from a simple energy balance. The pressure change ΔP is then proportional to the temperature change ΔT.
When we go through this analysis in more detail (see GOTHIC 2012) and also invoke the law of partial pressures,
we find that the final internal pressure in an average PWR containment building is about 0.445 MPa (65 PSI), while
the initial pressure is 0.101 MPa (slightly below atmospheric pressure). The exact temperatures and pressures before
and after a LOCA are shown in Tables 32.3 and 32.4. Table 32.3 shows the temperatures and pressures for a large
pipe break LOCA in the primary loop, and Table 32.4 shows the temperatures and pressures for a large pipe break
LOCA in the secondary loop.
Notice that the maximum containment building pressure is between 0.445 MPa and 0.523 MPa (65 PSI and 76 PSI)
depending upon the accident scenario that is assumed. In most cases, the final pressure is higher when a pipe breaks in
the primary loop because there is no other place for the thermal energy to go. For this reason, PWR containment build-
ings are normally designed to sustain continuous pressures between 50 and 75 PSI and peak pressures of up to 100 PSI.
BWR containment buildings have tended to follow a similar but not identical trend. If one attempts to perform a more
realistic analysis of a reactor blowdown with GOTHIC, one finds that the pressures in the containment building tend to
peak at about 35 PSI and that the pressure levels tend to peak about 2 s after a large-scale LOCA begins (see Figure 32.2).
Eventually, the NRC will probably agree to use a less conservative approach after the dynamics of this process is more
completely understood. However, the buildup of hydrogen gas (H2) in the containment building still has to be addressed,
and any relaxation of these restrictions will probably require the reactor vendor to show that the hydrogen igniters in the
reactor containment building will still function adequately even if all of the off-site power to the plant is lost for a con-
siderable period of time. In general, there are two ways that the peak pressure in a reactor containment building can be
reduced during a large break LOCA. First, the reactor vendor can make the containment building larger. This will cause
the peak pressures to be lower because there is more space for the hot air–steam mixture to occupy. Second, overhead
TABLE 32.3
Temperatures and Pressures in a PWR Containment Building Immediately before and after a PWR Hot Leg LOCA
TABLE 32.4
Temperatures and Pressures in a PWR Containment Building Immediately before and after a PWR Secondary Loop LOCA
sprayers and a suppression pool (see Section 32.11) can be used to absorb some of the energy released by the LOCA. This
reduces the final containment building pressure because the temperatures are reduced as well. The following example
illustrates how the first scenario can be effective. The second scenario is discussed in one of the homework problems.
Bibliography
Books and Textbooks
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
GOTHIC Thermal Hydraulic Analysis Package Technical Manual and User Manual, Numerical Applications, Inc., Richland,
Washington, DC (2012).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Web References
http://en.wikipedia.org/wiki/AP1000.
http://en.wikipedia.org/wiki/Containment_building.
http://en.wikipedia.org/wiki/Three_Mile_Island_accident.
http://en.wikipedia.org/wiki/European_Pressurized_Reactor.
http://en.wikipedia.org/wiki/Fukushima_Daiichi_nuclear_disaster.
http://en.wikipedia.org/wiki/Boiling_water_reactor_safety_systems.
http://www-pub.iaea.org/MTCD/publications/PDF/Pub1189_web.pdf.
http://www.nuclearstructures.com/Types_of_Nuclear_Containment_Structures_48.html.
Additional References
☉☉ Important physical constants used by nuclear scientists and engineers can be found in Appendix A.
☉☉ Unit systems and their conversion factors can be found in Appendix B.
☉☉ Physical properties of important nuclear materials can be found in Appendix D.
☉☉ Atomic masses, half-lives, and natural abundances can be found in Appendix E.
5. What is considered by the U.S. NRC to be the most severe pipe break that can occur during a PWR LOCA?
6. What is the highest level of severity of a reactor accident on the INES scale?
7. What is the size of a modern PWR containment building?
8. Approximately what fraction of the height of the Empire State building is a modern PWR reactor cooling tower?
9. In the nuclear business, what does the term NSSS traditionally refer to?
10. What types of reactors were destroyed at Fukushima, Japan, by the earthquake and ensuing tsunami?
11. What is the peak pressure level in a typical compartment of a BWR containment building during a LOCA?
12. What passive device is used in a BWR to limit the buildup of pressure waves that are produced during a LOCA?
13. What types of steels are used in the construction of light water reactor pressure vessels?
14. What type of radioactive decay process is responsible for most of the decay heat that is produced in a spent fuel
pool?
15. What happens if the power is shut off to the pumps feeding cold water to a spent fuel pool?
16. What is the average temperature of the water in a BWR suppression pool?
17. How many types of BWR containment buildings are in widespread use today, and what are their names?
18. What are the two most common types of cooling towers used in nuclear power plants today?
19. What does the term LOCA normally refer to?
20. Write down the equation for the zirconium–water reaction that can occur during a reactor LOCA.
21. Approximately how much energy is released by each mole of the zirconium cladding during a hypothetical LOCA?
22. At what temperature does the zirconium–water reaction begin to become autocatalytic?
23. What is the radius of the exclusion zone around most nuclear power plants?
24. Where can the spent fuel pool be found in a nuclear power plant that uses a Mark I containment building?
25. What are the two primary factors that determine the amount of radioactive fission gas that can be released from a
reactor fuel rod?
26. In a colder climate like Norway or Finland, what type of reactor cooling tower would you be most likely to find?
27. If a nuclear power plant is 33% efficient and generates 1,000 MW of electric power, how much thermal power must
be removed by its cooling towers?
28. What type of process is used to remove the heat from a natural draft cooling tower?
29. What is the oldest BWR containment building called, and what is the shape of the suppression pool in this building?
30. What is the thermal efficiency of the high-temperature gas reactor?
31. What type of BWR containment building most closely resembles a modern PWR containment building?
32. What is the purpose of a blowdown panel in a reactor containment building?
33. Where can the spent fuel pool be found in a nuclear power plant that uses a Mark III containment building?
34. During a reactor accident, what does the term “choked flow” refer to?
35. What is the purpose of the containment building missile defense shield?
36. With respect to the NSSS, what does the term “enthalpy” refer to?
37. What two thermodynamic processes starting with the letter “R” are used to optimize the thermal efficiency of a
nuclear power plant?
38. What is the first line of defense that limits the release of radioactive materials from a nuclear power plant?
39. In what famous book can the thermodynamic properties of a water–steam mixture normally be found?
40. How does a suppression pool affect the size of a BWR containment building?
41. Which nuclear utility had its stock damaged as a result of the reactor accident at Fukushima, Japan?
42. At what speed do the rotors of an electrical generator turn in a country where the alternating current is 60 Hz AC?
43. What containment structure in a BWR can be found just outside of the dry well?
44. After the spent steam leaves the steam turbines in a nuclear power plant, what is the next device in the NSSS that it
encounters?
45. How many types of turbines are there in the NSSS of a typical nuclear power plant?
46. In what device is the cooling water returning from the condensers reheated before it is sent back into the core of a
nuclear power plant again?
47. Where is the steam from the HP turbine reheated before it is sent to the LP turbine?
48. What is the process of reheating the steam called before it is sent onto the LP turbines?
49. Why is it necessary to reheat the steam before it is sent to a LP turbine?
50. How much additional thermodynamic efficiency can be added to a PWR by using the processes of reheat and
regeneration?
51. Except for Chernobyl, what have been the two major causes of reactor accidents since the 1950s?
52. How many physical barriers are built into a nuclear power plant to prevent the radiation that the plant produces from
reaching the general public?
1228 Containment Buildings and their Function
Exercise 32.2
Water in a reactor suppression pool is normally maintained as close to the freezing point as possible (usually at a tem-
perature of about 4°C). Suppose that the suppression pool contains 1,000 m3 of this very cold water and that each cubic
meter weighs 1,000 kg. If every 1°C rise in the temperature of the water allows it to absorb 4,186 J of energy per kg,
how much energy is the suppression pool theoretically capable of absorbing if 100% of the water is allowed to reach the
boiling point (100°C)?
Exercise 32.3
In one of the damaged cores at the Fukushima nuclear power plant, it was estimated that approximately 25,000 kg of zir-
conium metal reacted with the cooling water to form hydrogen gas. This gas was then vented to the containment build-
ing (which subsequently exploded). If the temperature of the cladding exceeded 1,200°C when this reaction occurred,
approximately how many joules of thermal energy were generated by this chemical reaction?
Exercise 32.4
Suppose that the thickness of a reactor containment building (if it is in the shape of a right circular cylinder) is approxi-
mately proportional to the square root of the internal pressure. Suppose that a reactor containment building is designed
to withstand a maximum design pressure of 40 PSI. Later on, before the building is built, it is learned that the design
pressure must be increased to 60 PSI. How much thicker must the walls of the containment building be in this case if
the initial wall thickness was 1 m?
Exercise 32.5
A blunt metal object weighing 1,000 kg hits the side of a reactor containment building in a head-on collision. If the
object is travelling at 100 km/h (27.75 m/s) and the collision is over in two-tenths of a second, how much force does the
object exert on the walls of the containment building?
Exercise 32.6
A reactor containment building has an internal volume of 75,000 m3. A decision is made to reduce the internal volume
of the building to 50,000 m3 based on the results of several GOTHIC computer calculations (see Section 32.27). If the
containment building was originally designed to withstand a maximum internal pressure of 50 PSI, how much conser-
vatism did the GOTHIC calculations apparently remove from the original design estimates? Assume that the design
pressure remains unchanged at 50 PSI.
33
Thermal Design Limits,
Operating Limits, and Safety Limits
33.1 T hermal Design Limits
Reactor design begins with the premise that the nuclear fuel rods must never be allowed to fail—no matter how hot the core becomes,
no matter how poorly the plant is operated, and no matter what the burnup of the fuel may be. Unfortunately, in real life, unexpected
events can sometimes occur that make it difficult to predict exactly how, when, or where a fuel rod will fail. To compensate for these
events, the core is designed to operate at a specific power level, a specific flow rate, and a specific temperature and pressure. This is
sometimes called its base state or its expected state. Then a safety margin is added to the base state to account for the fact that one or
more unexpected events may occur. This safety margin, plus the expected state, then creates what is known as its range of anticipated
operating states:
Sometimes the design margin, which represents the difference between the operating states and the base state, is called the plant’s
safety margin. The safety margin is determined by the probability that certain types of events will occur and that some events may be
more severe (or may be more likely to occur) than others. Finally, it is possible that more than one type of event can occur at the same
time. Thus, the safety margin can be thought of as the physical difference between the established safety limit and the values of the
temperature, the pressure, and the flow rate at which the reactor is expected to operate normally. The size of this margin is usually
stipulated by regulatory authorities after a detailed analysis of the various types of uncertainties involved.
1229
1230 Thermal Design Limits, Operating Limits, and Safety Limits
FIGURE 33.1 An operating map for a typical American BWR. Operation is allowed outside of the shaded red area,
where hydrodynamic instabilities can develop.
When this design limit is exceeded, at least some of the fuel rods will fail. If the event that leads to these rod failures
becomes even more severe, then it may be classified as a reactor accident. Today most reactor accidents are classified using
what is called the INES scale of accident severity. This scale, which is logarithmic in nature, is defined and elaborated upon
in Figures 33.2 and 33.3. This scale, which is an acronym for the International Nuclear and Radiological Event scale, was
first proposed by the International Atomic Energy Agency (the IAEA) in 1990 to allow the prompt communication of infor-
mation regarding the severity and radiological consequences of commercial reactor accidents. Since then, it has become
widely accepted as a reliable way to convey the radiological consequences of a reactor accident to the general public. The
relationship between a reactor’s design limits, operating limits, and safety limits is shown in Table 33.1.
Normally, design uncertainties are classified into five or six categories depending on their severity and fre-
quency of occurrence. These categories then define the difference between the expected state of the system and an
FIGURE 33.3 An explanation of the events that define the INES scale of accident severity.
TABLE 33.1
The ANS Limit Hierarchy That Defines the Operational Limits for a Commercial Nuclear Power Plant
unanticipated state. For severe accidents, additional factors may have to be added to the range of anticipated states to
arrive at what is called the design safety limit. Normally, the design safety limit is more conservative than the operational
design limit. Then the severity of an accident is ultimately determined by the containment design limit (see Table 33.1),
which is discussed in Chapters 32 and 34.
For boiling water reactors (BWRs), similar events are described in a related document called
☉☉ ANSI/ANS 52.1:
Nuclear Safety Criteria for the Design of Stationary Boiling Water Reactor Plants (1988)
According to the ANS, the types of events a reactor must be designed to withstand fall into four basic categories that
are called Category I, II, III, and IV events. These categories of events are described in Table 33.2. Sometimes these
categories are also called conditions. Category V events refer to accidents that a reactor is NOT ordinarily designed
to withstand. Each category or condition is then associated with a specific reactor state or set of operating conditions.
These categories and the specific sets of events associated with them are shown in Table 33.2. Their frequency of occur-
rence varies with their severity as well. Notice that Category IV events, which are defined as operational events that may
result in a significant accident, are only assumed to occur on very rare occasions. However, Category II events can occur
relatively frequently (up to once a year), and Category I events can even occur on a daily, weekly, or monthly basis. In all,
there are a total of 13 safety design standards that nuclear power plants in the United States are required to meet. These
standards were created by the ANS, and they are also listed as ANSI Standards (see www.ansi.org). The ANS Standards
that apply to nuclear power plant safety are shown in Table 33.3. In some cases, these standards are similar to those the
American National Standards Institute (ANSI) applies to other industrial processes as well.
From time to time, these standards may be subject to major changes and revisions. The publication date (e.g., 2005
for ANS 5.1) is the way that the actual revisions are tracked. The last major revisions to ANS Standard 51.1 (for PWR
safety) were made in 1988 after the accident at Three Mile Island (TMI). A new version of ANS Standard 52.2 (for BWR
safety) is expected to be released because of the lessons learned at Fukushima, Japan in 2011. This updated standard
will probably require reactors to be equipped with additional off-site (or on-site) power sources, hydrogen igniters that
do not rely on the power grid for electricity, and video cameras with battery backups. There may also be some additional
requirements pertaining to the cooling of spent fuel pools when the coolant pumps used for these pools lose power.
ANSI/ANS Standard 51.1 also describes the operating states of a plant and the safety objectives a reactor designer is
TABLE 33.2
The ANS Classification of Reactor Accidents
TABLE 33.3
A List of the Current ANS Safety Standards That Apply to Nuclear Power Plants in the United States
expected to meet for each state. These states are described in Table 33.4, where their expected rates of occurrence are
also shown.
This definition has evolved over time, and today, it is the standard terminology used to describe a DBA. To evaluate
the expected response of a plant to these accidents, each accident is analyzed within a framework of highly conser-
vative assumptions that are intended to exaggerate the consequences of the expected accidents. These assumptions
are then used to provide a justification for licensing the plant. In the United States, the NRC classifies these acci-
dents into nine distinct categories that are specified in 10 CFR 50—the Compendium of Federal Regulations. These
accident classes are defined in an order of increasing severity from 1 to 9 with category 1 accidents being the least
severe and category 9 being the most. In general, these accident classifications are different than the classifications
used by the ANS and the IAEA.
Class 1 DBAs are normally considered to be trivial in nature because they result in levels of radioactivity that are
not substantially higher than those which occur due to natural causes. Accident categories 2 through 7 are increasingly
more severe in nature and usually involve the release of more radioactive materials. Finally, Class 8 DBAs include
the most severe credible accidents that are hypothesized to occur over the life of the plant. Class 8 DBAs also include
LOCAs. These accidents must be analyzed in both the PSAR and FSAR. A partial list of these accidents is shown in
Table 33.5. The most severe DBAs that have a significant chance of occurring over the life of the plant are then called
maximum credible accidents. Other accidents can also occur under extraordinary circumstances, but these accidents
occur so infrequently that it is possible to assume that they might occur once every 100,000 to 1,000,000 operating
TABLE 33.4
The Operating States of a Nuclear Power Plant as Defined by the ANS and the ASME
TABLE 33.5
Categories of Reactor Accidents Defined by the U.S. NRC
Accident Classifications
Used by the NRC Accident Descriptions within Each Classification
Category 9 An extreme event so improbable that it is not classified or regulated
Category 8 LOCAs
Category 8 A break in the instrumentation line controlling the primary systems
Category 8 A rod ejection accident (applicable to PWRs)
Category 8 A rod drop accident (applicable to BWRs)
Category 8 Steam line breaks outside the containment building (applicable to PWRs)
Category 8 Steam line breaks inside the containment building (applicable to BWRs)
Category 7 Spent fuel-handling accidents
Category 6 Refueling accidents
Category 5 Release of fission products into the primary or secondary loops caused by cladding
defects and steam generator tube leaks (specific to PWRs)
Category 4 Release of fission products into the primary loop caused by cladding defects and
unexpected fuel rod failures (specific to BWRs)
Category 3 Release of radioactive wastes from on-site storage tanks
Category 2 Small releases of radiation outside of the containment building
Category 1 Trivial operational incidents and other events
Here, Category 1 accidents are trivial in nature and Category 8 accidents are DBAs that must be analyzed in the PSAR
and FSAR. Category 9 accidents are so improbable that they are not specifically regulated or defined by the NRC.
years. Because of their rarity, they are sometimes called maximum improbable accidents. In fact, these accidents occur
so infrequently that it is generally not necessary to include an analysis of them in the PSAR or FSAR. Accidents of
this type are called Class 9 accidents. Examples of Class 9 accidents include a commercial airliner striking a nuclear
power plant during a magnitude 8 earthquake—such as the one that occurred at Fukushima, Japan. Another Class 9
accident might involve the simultaneous loss of on-site and off-site power during a large pipe break LOCA. Needless
to say, Class 9 accidents are extremely rare and disruptive.
This rule ensures that there is a reasonable certainty that the parameters governing the behavior of each major compo-
nent will fall within the specified design limits (or design envelope) for Category I and II events at least 95% of the time.
Category III and IV events, which are much more severe and also occur less frequently, are subject to a different set of
design rules. The components or operational states for which design limits must be specified include
☉☉ The fuel
☉☉ The cladding
☉☉ The core flow
☉☉ The critical heat flux (CHF)
☉☉ The departure from nucleate boiling (DNB) ratio
☉☉ The overall hot channel factors (both nuclear and statistical).
33.6 Design Limits for the Fuel 1235
Calculations of these factors are usually conducted at rated power and then at an elevated power level (corresponding to
a 115% or 120% overpower condition) and for various types of predefined transients and operational events—including
startup, shutdown, loss of feed water flow, turbine trips, and DBAs. Each design limit is then described by a separate
ANS or ANSI standard.
Burnups including:
☉☉ 0 MWD/Ton (fresh fuel), 15.000 MWD/Ton, 30,000 MWD/Ton, 45,000 MWD/Ton, and 60,000 MWD/Ton
FIGURE 33.4 A normal probability distribution used to define the 95-95 rule.
1236 Thermal Design Limits, Operating Limits, and Safety Limits
with a confidence level of 95%. Modern PWRs are designed so that the fuel will not melt during Category I or II events
for burnups up to 75,000 MWD/Ton. However, in the United States, the fuel in light water reactors is normally replaced
at burnups less than 60,000 MWD/Ton.
In most PWRs, the currently accepted value for the radial power peaking factor F RADIAL is about 1.59, and the currently
accepted value of the axial power peaking factor FAXIAL (which corresponds to a chopped cosine) is about 1.61. Hence,
the total power peaking factor that is used to find the minimum DNB ratio (MDNBR) is
Today, these numbers are believed to be conservative during steady-state operation and during most expected transients.
The MDNBR is then calculated for the design power shape assuming a peak-to-average power ratio of 2.56. To establish
the DNBR operating limits when manufacturing uncertainties, flow uncertainties, instrumentation uncertainties, fuel
rod bowing (described in a subsequent section), and other anomalies are taken into account, design calculations are per-
formed to DNBR limits higher than the normal DNBR design limits. The difference between the normal design limit
DNBRs and the safety limit DNBRs is then the DNBR safety margin (see Figure 33.5).
FIGURE 33.5 Thermal design limits for typical PWR cores. (Note: MDNBR values shown here were taken from the South Texas
PWR FSAR.)
reasonable certainty that the plant can be brought back on line after the failed fuel rods have been replaced. This damage
can be greater than the damage for Category I and II events.
Normally, about 94% of the total core flow in a PWR is assumed to pass directly over the fuel rods. The remaining 6% is
assumed to pass around the control rod guide tubes and through the shroud between the fuel assemblies and the reactor
pressure vessel.
FIGURE 33.6 A blockage at the inlet of a PWR fuel assembly. Notice how the flow travels around the blockage and recovers about
30 in. (~0.75 m) downstream.
33.11 Effects of Flow Blockages on DNBR Estimates 1239
(see Figure 33.6), the coolant from the adjacent fuel assemblies flows back into the blocked fuel assembly as soon as
the blockage is passed. In a typical PWR, the flow recovers about 30 in. (~0.75 m) downstream from the point where the
blockage occurs. More information on how blockages behave under these conditions can be found at
☉☉ Stewart, C.W., et al., “VIPRE-01: A Thermal-Hydraulic Code for Reactor Cores” Volume 1–3 (Revision 3, August
1989), Volume 4 (April 1987), NP-2511-CCM-A, Electric Power Research Institute.
Computer codes such as COBRA and VIPRE are normally used to predict the location where flow reversal occurs. Most
of the time, partial blockages do not affect the DNBR more than 5%–10%. Other blockages that must be considered
during Category II accidents include those which occur within a fuel assembly rather than at the entrance of a fuel
assembly. These blockages (such as the one shown in Figure 33.7) typically have less effect on the value of the MDNBR
than an external blockage does. In open-lattice cores (such as American PWRs), flow blockages can cause perturbations
to the flow field immediately in front of and behind the blockage. During normal operating conditions, the coolant flow
usually recovers between 4 and 6 in. (10–15 cm) downstream of the point where the blockage occurs. For a reasonable-
sized blockage (on the order of 40%–50% of the area of the flow channel), experiments have shown that the stagnant
region behind the blockage disappears approximately 1.65 L/DE inches downstream from the point where the blockage
is located. Here, L is the length of the coolant channel and DE is its equivalent diameter. Consequently, flow blockages
within reactor fuel assemblies do not have much effect on the enthalpy rise Δh in the blocked channel. Blockages also
increase the amount of turbulent mixing between the channels, and like grid spacers, they serve to enhance the amount
of lateral cross flow in their immediate vicinity. More information about how blockages behave under these conditions
can be found at
☉☉ Ohtsubo, A., and Uruwashi, S., “Stagnant Fluid due to Local Flow Blockage,” Journal of Nuclear Science
Technology, No. 7, pp. 433–434, 1972.
FIGURE 33.7 The behavior of the coolant inside a fuel assembly that contains a partial blockage. In this case, the flow recovers
about 6 in. (~15 cm) downstream from where the blockage is located. Also, the “backfilling” of the blocked channel with coolant
from the adjacent channels may cause the flow to temporarily stagnate or even reverse its direction directly behind the blockage.
This behavior is typical of how the flow field behaves when the Reynolds number is high.
1240 Thermal Design Limits, Operating Limits, and Safety Limits
☉☉ Basmer, P., Kirsh, D., and Schultheiss, G.F., “Investigation of the Flow Pattern in the Recirculation Zone Downstream
of Local Coolant Blockages in Pin Bundles,” Atomwirtschaft 17, No. 8, pp. 416–417, 1972 (in German).
Thus, most flow blockages do not have an appreciable effect on the MDNBR.
Each accident is then simulated using sophisticated computer programs or computer codes (see Section 33.17) to under-
stand its radiological and environmental consequences. Normally, only the most severe accidents are assumed to release
radiation into the environment.
* See An Introduction to Nuclear Reactor Physics by Robert E. Masterson, CRC Press (2017).
33.13 Additional Safety Guidelines 1241
far the individual components in the core can degrade before radiation is released from them. For example, in the United
States, federal policy mandates that the following design limits must not be exceeded during any postulated DBA:
Again, these limits are controlled by the reactor’s safety systems, and so exactly how they are controlled is at the
discretion of the designer and the operator. However, most safety systems have a number of these features in common.
They can be classified for PWRs and BWRs in a similar way.
FIGURE 33.8 An event tree taken from the WASH-1400 report for calculating the probability of different reactor accidents. The
probability of occurrence is expressed in terms of a probability of occurrence per operating year. The aggregate probability is found
by multiplying the individual probabilities of each event together along each branch of the accident event tree. This particular tree
is one of the sub-trees that was used to calculate the probability of a large pipe break LOCA releasing radiation from the plant.
Sometimes a tree of this type is also called a reduced event tree.
killed the dinosaurs*, or an atomic attack. An example of a category 9 accident is the simultaneous failure of all on-site
and off-site power at the exact moment that a LOCA occurs. Another example of a category 9 accident is the sudden and
catastrophic rupture of the entire reactor pressure vessel. Hence, accidents of this type are not specifically defined, and
they do not appear in Table 33.5, per se.
* See Nuclear Engineering Fundamentals: A Practical Perspective by Robert E. Masterson, CRC Press (2017).
1244 Thermal Design Limits, Operating Limits, and Safety Limits
TABLE 33.6
Calculated Probabilities of Reactor Accidents up to Those That Can Lead to Core Damage in the United States
models within these computer programs are very sophisticated, and up to recently, they required a supercomputer to
run. In this section, we would like to briefly discuss some of the mainstream licensing codes that are used by regula-
tory authorities today to predict the behavior of nuclear power plants during LOCAs and other Category III and IV
events. The reactor vendors, who design and build the reactors, have their own proprietary computer programs, and
sometimes the programs the NRC uses give slightly different results than the ones the reactor vendors use. These dis-
crepancies are generally driven by differences in the underlying physical assumptions that are used (such as a 4 equation
vs. a 5 equation vs. a 6 equation model), or because the reactor vendors use proprietary data that the NRC does not use
for its analysis. The phenomenological and fluid-mechanical reasons for these differences are discussed in Chapter 15.
However, the overall trends that these different computer codes predict are normally in good agreement. In general,
there are eight categories of accident analysis codes the NRC uses:
☉☉ PRA codes
☉☉ Fuel behavior codes
☉☉ Reactor kinetic codes
☉☉ Thermal-hydraulics codes
☉☉ Severe accident codes
☉☉ DBA codes
☉☉ Health effects/dose calculation codes
☉☉ Radionuclide transport codes (for license termination/decommissioning)
Reactor thermal-hydraulic analysis codes are considered to be a subset of this larger code package. Further information
regarding how to acquire a copy of these computer codes can be obtained from the NRC at www.nrc.gov. Since the 1950’s,
literally hundreds of computer programs have been written to cover all eight of these important phenomenological catego-
ries, but only about a dozen of them have achieved “rock star” status. Some of these programs are designed to analyze the
33.21 Reactor Thermal-Hydraulics Codes 1245
behavior of the core, while others are designed to simulate the behavior of the plant as a whole. Some examples of pro-
grams specifically designed to analyze the core are COBRA and VIPRE, while other programs such as RELAP, RETRAN,
and TRAC were designed to model the behavior of the entire plant. Specialized computer programs such as FRAPCON
were written to analyze the behavior of nuclear fuel rods, while other programs such as CONTAIN and GOTHIC were
written to model the behavior of the containment building. Finally, programs such as MELCOR and MAAP were devel-
oped to understand the consequences of severe reactor accidents where the structural integrity of the core could be
lost. Eventually the NRC decided to combine the capabilities of many of these programs together into a single modern
computer program called TRACE (The TRAC-RELAP Advanced Computational Engine). Today, TRACE is the primary
tool used by the U.S. NRC to license commercial nuclear power plants in the United States. It is very sophisticated and
has extensive visual display tools. More information regarding its capabilities can be found at the following URL:
https://www.nrc.gov/docs/ML0710/ML071000097.pdf
which is also maintained by the NRC. This program uses many of the correlations we have introduced in this book to
predict the thermal and hydraulic behavior of the core. It is also capable of modeling the behavior of the liquid and vapor
phases independently. In other words, TRACE can be applied to both equilibrium and non-equilibrium flows. Its primary
fluid mechanical model is a two fluid, separated flow model that uses separate continuity, energy, and momentum equations
for the liquid and vapor phases (e.g., water and steam). Other programs the NRC uses to perform its regulatory functions are
explained in Sections 33.18 through 33.25. The descriptions for some of these programs were taken directly from the NRC’s
main website www.nrc.gov. The reader is encouraged to visit this site for more information regarding these programs.
☉☉ FRAPCON-3 is a computer code used for steady state and mild transient analysis of the behavior of a single fuel
rod under near-normal reactor operating conditions.
☉☉ FRAPTRAN is a computer code used for transient and design basis accident analysis of the behavior of a single
fuel rod under off-normal reactor operation conditions.
☉☉ PARCS: The Purdue Advanced Reactor Core Simulator (PARCS) is a computer code that solves the time-
dependent two-group neutron diffusion equation in 3D Cartesian geometry using nodal methods to obtain the
transient neutron flux distribution. The code may be used in the analysis of reactivity-initiated accidents in light-
water reactors where spatial effects may be important. It may be run in the stand-alone mode or coupled to other
NRC thermal-hydraulic codes such as RELAP5.
of their agreements, the NRC provides these member countries the latest versions of its thermal-hydraulic systems
analysis computer codes to help evaluate the safety of planned or operating plants in each member’s country. To help
ensure these analysis tools are of the highest quality and can be used with confidence, the international partners perform
and document assessments of the codes for a wide range of applications, including identification of code improvements
and error corrections. The thermal-hydraulic codes developed by the NRC include the following:
☉☉ TRACE: The TRAC/RELAP Advanced Computational Engine A modernized thermal-hydraulic code designed
to consolidate and extend the capabilities of NRC’s three legacy safety codes—TRAC-P, TRAC-B, and RELAP.
It is able to analyze large/small break LOCAs and system transients in both PWRs and BWRs. The capability
exists to model thermal-hydraulic phenomena in both 1D and 3D spaces. This is the NRC’s flagship thermal-
hydraulics analysis tool.
☉☉ SNAP: The Symbolic Nuclear Analysis Package is a graphical user interface with preprocessor and post-processor
capabilities, which assists users in developing TRACE and RELAP5 input decks and running the codes.
☉☉ RELAP5: The Reactor Excursion and Leak Analysis Program is a tool for analyzing small-break LOCAs and
system transients in PWRs or BWRs. It has the capability to model thermal-hydraulic phenomena in 1D volumes.
While this code still enjoys widespread use in the nuclear community, active maintenance will be phased out in
the next few years as usage of TRACE grows.
☉☉ Legacy tools that are no longer actively supported include the following thermal-hydraulic codes:
1. TRAC-P: Large-break LOCA and system transient analysis tool for PWRs. Capability to model thermal-
hydraulic phenomena in 1D or 3D components.
2. TRAC-B: Large- and small-break LOCA and system transient analysis tool for BWRs. Capability to model
thermal hydraulic phenomena in 1D or 3D components.
3. CONTAIN: Containment transient analysis tool for PWRs or BWRs. Capability to model thermal-hydraulic
phenomena (within a lumped-parameter framework) for existing containment designs.
☉☉ VARSKIN: The NRC has sponsored the development of the VARSKIN code in the 1980s, to assist licensees in
demonstrating compliance with Paragraph (c) of Title 10, Section 20.1201, of the Code of Federal Regulations
(10 CFR 20.1201), “Occupational Dose Limits for Adults.” Specifically, 10 CFR 20.1201(c) requires licensees to
have an approved radiation protection program that includes established protocols for calculating and document-
ing the dose attributable to radioactive contamination of the skin. Since that time, the code has been significantly
enhanced to simplify data entry and increase efficiency. VARSKIN 3 is available from the Radiation Safety
Information Computational Center (RSICC). For additional information, see NUREG/CR-6918, “VARSKIN 3:
A Computer Code for Assessing Skin Dose from Skin Contamination.”
☉☉ Since the release of VARSKIN 3 in 2004, the NRC staff has compared its dose calculations for various energies
and at various skin depths, with doses calculated by the Monte Carlo N-Particle Transport Code System (MCNP)
developed by Los Alamos National Laboratory (LANL). That comparison indicated that VARSKIN 3 overesti-
mates the dose with increasing photon energy. For that reason, the NRC is sponsoring a further enhancement to
replace the existing photon dose algorithm, develop a quality assurance program for the beta dose model, and
correct technical issues reported by users.
☉☉ DandD: A code for screening analyses for license termination and decommissioning. The DandD software auto-
mates the definition and development of the scenarios, exposure pathways, models, mathematical formulations,
assumptions, and justifications of parameter selections documented in Volumes 1 and 3 of NUREG/CR-5512.
Probabilistic RESRAD 6.0 and RESRAD-BUILD 3.0 Codes: The existing deterministic RESRAD 6.0 and
RESRAD-BUILD 3.0 codes for site-specific modeling applications were adapted by Argonne National Laboratory
for NRC regulatory applications for probabilistic dose analysis to demonstrate compliance with the NRC’s license
termination rule (10 CFR Part 20, Subpart E) according to the guidance developed for the Standard Review Plan for
Decommissioning. (The deterministic RESRAD and RESRAD-BUILD codes are part of the family of codes developed
by the U.S. Department of Energy. The RESRAD code applies to the cleanup of sites and the RESRAD-BUILD code
applies to the cleanup of buildings and structures.)
33.26 Additional Information Regarding Reactor Accident Analysis and Safety Codes
Modern reactor safety analysis programs such as the ones listed in Sections 33.18 through 33.25 are often used in con-
junction with other NRC reports and regulations. Most of these regulations can be obtained from the data sources shown
below. The ASME also provides access to some of these reports, and the ANS provides access to others.
1. All NRC regulations listed herein are available electronically through the Public Electronic Reading Room on the
NRC’s public Web site, at http://www.nrc.gov/reading-rm/doc-collections/cfr/part050. Copies are also available for
inspection or copying for a fee from the NRC’s Public Document Room at 11555 Rockville Pike, Rockville, Maryland;
the PDR’s mailing address is USNRC PDR, Washington, DC, 20555; telephone (301) 415-4737 or (800) 397-4209; fax
(301) 415-3548; e-mail [email protected].
2. All NUREG-series reports listed herein were published by the U.S. Nuclear Regulatory Commission. Copies
are available for inspection or copying for a fee from the NRC’s Public Document Room at 11555 Rockville
Pike, Rockville, Maryland; the PDR’s mailing address is USNRC PDR, Washington, DC, 20555; telephone
(301) 415-4737 or (800) 397-4209; fax (301) 415-3548; e-mail [email protected].
3. All regulatory guides listed herein were published by the U.S. Nuclear Regulatory Commission. Where an
Agencywide Documents Access and Management System (ADAMS) accession number is identified, the speci-
fied regulatory guide is available electronically through the NRC’s ADAMS at http://www.nrc.gov/reading-rm/
1248 Thermal Design Limits, Operating Limits, and Safety Limits
adams.html. All other regulatory guides are available electronically through the Electronic Reading Room on the
NRC’s public Web site, at http://www.nrc.gov/reading-rm/doc-collections/reg-guides/.
4. Commission Papers (SECY) listed herein are available electronically through the Public Electronic Reading
Room on the NRC’s public Web, site at http://www.nrc.gov/reading-rm/doc-collections/commission/secys/.
Copies are also available for inspection or copying for a fee from the NRC’s Public Document Room at 11555
Rockville Pike, Rockville, Maryland; the PDR’s mailing address is USNRC PDR, Washington, DC, 20555; tele-
phone (301) 415-4737 or (800) 397-4209; fax (301) 415-3548; e-mail [email protected].
5. Copies may be purchased from the American Society of Mechanical Engineers, Three Park Avenue, New York,
New York, 10016-5990; phone (212) 591-8500; fax (212) 591-8501; www.asme.org.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi, India (2000).
Reactor Theory (Nuclear Parameters), DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
Web Reference
http://www.iaea.org/inis/collection/NCLCollectionStore/_Public/28/030/28030311.pdf.
15. According to the ANS, how many different categories of events must a commercial PWR or BWR in the United
States be able to withstand?
16. According to the ANS, how many distinct safety design standards is a nuclear power plant in the United States
required to meet?
17. In which ANS documents are these design standards defined?
18. What is the NRC’s definition of a reactor accident?
19. What is the NRC’s definition of a design basis accident or DBA? Is this definition different than that of an ordinary
accident?
20. How many categories of reactor accidents does the NRC use to classify reactor accidents in the United States?
21. Of these accident categories, how many categories describe DBAs and how many categories describe accidents that
are not considered to be DBAs?
22. Which category of reactor accident is considered to be the most severe and which category of reactor accident is
considered to be the least severe?
23. Into which of these categories does a LOCA fall?
24. Fill in the following sentence with the appropriate word or phrase: Category ______ accidents are considered
trivial in nature and involve releases of radioactivity that are not substantially different from those accompany-
ing normal plant operation. Accident categories _____ through ____ entail the release of increasingly significant
amounts of radiation into the environment. Category _____ accidents involve the DBAs that are analyzed in the
PSAR and FSAR. Category ____ accidents are assumed to be so rare and extreme that they can be ignored when
analyzing their safety and environmental consequences.
25. What information is required to be provided in a PSAR or a FSAR? Who is responsible for providing this
information?
26. What is an environmental impact statement? Approximately when is it expected to be filed?
27. What is the 95-95 rule, and how does it pertain to nuclear power plants?
28. Fill in the following sentence with the appropriate word or phrase: For category I and II events, reactor designers
must show that there is at least a ______% chance (with a ______% confidence level) that the fuel rods having the
highest total power output (when the power level is measured in kW/L) will not exceed the melting point of UO2.
29. For PWRs and BWRs operating in the United States, what is the currently established temperature limit for the
fuel?
30. What types of transients are to be considered when defining this temperature limit?
31. For PWRs and BWRs operating in the United States, what is the current temperature limit for the cladding? Why
are the temperature limits for the cladding defined in this way?
32. For PWRs and BWRs operating in the United States, what is the current DNBR limit? How is this limit defined? Is
the DNBR limit different for PWRs than it is for BWRs?
33. As an alternative to the DNBR limit, can the critical power ratio (CPR) be used to license commercial BWRs
instead? How is this limit defined?
34. For BWRs, what advantage does using the CPR have over using the DNBR?
35. Name six distinct categories of DBAs. Is a loss-of-flow accident (LOF) considered to be more severe or less severe
than a fuel-handling accident?
36. What are the relative probabilities of these accidents occurring? How are these probabilities calculated?
37. What is the meaning of PRA?
38. What was the WASH-1400 report intended to do? Approximately when was this report published?
39. According to this report, what is the probability of a large pipe break LOCA occurring in the United States? In what
units of time is this probability quoted?
40. Many of the findings of the WASH-1400 report were based on the use of event trees. What is the difference between
a full event tree and a partial or reduced event tree? Does using a reduced event tree have any effect on the accident
probabilities that are found?
41. In PRA, how is the probability of a specific sequence of events defined? If every event is independent of every other
event, is the probability of a specific sequence of events occurring additive or multiplicative? How long can a typical
event tree become?
42. Why did the WASH-1400 report only provide five categories of accident severity for BWRs, while it provided nine
categories of accident severity for PWRs?
43. Can PWR accidents become more energetic than BWR accidents? Do these differences have anything to do with
differences in the plant pressure levels and power densities in the core?
4 4. What computer programs are used by NRC to estimate the probability of a reactor accident? What other com-
puter programs are used by the NRC to understand the consequences of LOCAs and related DBAs?
1250 Thermal Design Limits, Operating Limits, and Safety Limits
Exercise 33.2
PWRs built by Westinghouse normally operate with a MDNBR of 2.2. According to the W-3 correlation, how much can
the steady-state mass flow rate be reduced if the core power output remains the same before the fuel rods dry out? How
much can the steady-state mass flow rate be reduced before the American DNBR design limit is reached?
Exercise 33.3
Suppose the radial and axial power peaking factors in a commercial PWR are 1.5 and 1.8, respectively. What power
peaking factor must be applied to the hot channel to calculate the minimum acceptable DNBR?
Exercise 33.4
Suppose 3% of the core flow through a commercial PWR flows over the control rods, 2% flows through the instrumenta-
tion channels, and 3% flows through the core shroud. The core contains 200 fuel assemblies and the total mass flow rate
is 20,000 kg/s. What is the flow rate of the coolant over the fuel rods alone in a single fuel assembly?
Exercise 33.5
A modern PWR has 200 fuel assemblies and generates 3,000 MW of thermal energy. If the peak-to-average power ratio
is 2.2, how much power does the hottest fuel assembly generate? Using the mass flow rates calculated in Exercise 33.4,
estimate the MDNBR.
Exercise 33.6
Compare the average enthalpy increase of the coolant through an American PWR to the average enthalpy increase of the
coolant through an American BWR. If the linear power density in the PWR is 110 kW/L and the linear power density in
the BWR is 55 kW/L, what does the ratio of the average mass flux though the cores have to be so that the MDNBR for
the PWR will be 1.3 and the MDNBR for the BWR will be 1.9?
Exercise 33.7
In the United States, the NRC limits the maximum cladding temperature in water reactors during a LOCA to 1200°C to
keep the zirconium metal in the cladding from reacting violently with the surrounding water and steam to initiate what
is called the zirconium–water reaction (see Chapter 30). Write down an equation that accurately describes this chemical
reaction. When the zirconium oxidizes, what additional combustible gas is produced?
Exercise 33.8
A reactor accident occurs where the zirconium–water reaction (see Exercise 33.7) becomes apocalyptic and produces
6.3 MJ of thermal energy by oxidizing a single kilogram of zirconium metal in the core. Assuming there are 20,000 kg
of zirconium metal in a single fuel assembly and that there are 200 fuel assemblies in the core, estimate the thermal
energy that is released if just 50% of the zirconium in these fuel assemblies is oxidized. Also estimate the volume of the
hydrogen gas that is produced. Do you think this gas had anything to do with the explosions that occurred during the
reactor accidents at Fukushima, Japan?
Exercise 33.9
Suppose a PRA is performed on a PWR involving four independent events having a probability of occurrence per year of
0.01, 0.002, 0.001, and 0.004, respectively. All four events must occur in a specific sequence for an accident condition to
develop. The probability of them occurring in this specific sequence is 0.001. What is the probability that this accident
condition will actually develop?
34
Response of a Containment
Building to a Reactor LOCA
34.1
Containment Buildings and Their Function
As we mentioned in Chapter 32, containment buildings are the final line of defense against a Category IV loss-of-coolant accident
(or LOCA). Category IV LOCAs are extremely rare, but when they do occur, they can be quite severe, and in most cases, the core will
be uncovered for some period of time (see Figure 34.1). A classic example of a Category IV accident is a pipe break in the hot leg or
the cold leg of the primary loop. Then, the primary loop loses some or all of its coolant, and as the pressure level falls, the hot cool-
ant immediately flashes and turns to steam. The rate of discharge of the steam from this break depends on the size of the break and
exactly where it is located (see Chapter 30). LOCAs can occur over a period of several minutes to several hours, and at the end of the
process, the pressure in the containment building stabilizes between 40 PSI and approximately 60 PSI (or 275 to 415 kPa). When the
coolant surrounding the fuel rods is lost, the cladding melts, and the rods will fail. This releases radioactivity into the containment
FIGURE 34.1 An artist’s rendering of the TMI core after it melted down in 1979. (Picture provided by Wikipedia.)
1251
1252 Response of a Containment Building to a Reactor LOCA
building in the form of both gaseous fission products (like iodine) and other particulate matter. If the containment build-
ing is strong enough to withstand the initial pressurization process, the accident will be contained, and no radioactive
materials will be released to the surrounding environment. However, the resulting decay heat will generally cause the
core to lose its structural integrity. For example, at the end of the LOCA at Three Mile Island (TMI) in 1979, the core
melted down to resemble the picture shown in Figure 34.1. Because of this core damage, the plant had to be shut down
and permanently decommissioned.
Regulatory authorities then mandated that all commercial nuclear power plants in the United States should be
designed to withstand certain unintended operational events that have become known as “design basis accidents” or
DBAs. DBAs are classified according to their severity, and a complete classification of these DBAs can be found in
Chapter 33. These accidents are normally described in the plant’s final safety analysis report or FSAR. In most coun-
tries, a copy of the FSAR is maintained for public viewing in a library close to the plant site. Most reactor containment
buildings operate at negative pressures slightly below atmospheric pressure to keep small amounts of radioactive par-
ticulates from being released from the buildings during Category I and II DBAs. In the United States, the thickness of
the containment building walls is mandated by the federal government, and its thickness is specified in a regulatory
document called 10 CFR 50.55a (sometimes called 10 CFR 50, Section 55a). The procedures that must be used to esti-
mate the temperatures and pressures within the containment building after a LOCA begins are also specified in another
document called ANSI/ANS 56.4: Pressure and Temperature Transient Analysis for Light Reactor Containments (1988),
which was originally developed by the American Nuclear Society (the ANS) and then adopted by the American National
Standards Institute (ANSI). Here, the DBAs the containment building must be designed to withstand are also speci-
fied. Both the containment building missile defense shield and the inner steel liner must be strong enough to withstand
the impact of a fully loaded passenger plane (equivalent to a Boeing 777) without compromising the building’s struc-
tural integrity. In theory, this allows a commercial nuclear power plant to be located close to a commercial airport.
However, since 9/11, this practice has been frowned upon, and now, virtually all nuclear power plants are located in
more remote locations.
34.3
LOCA Heat Sources and Their Time Lines
There are several different heat sources (and sinks) that can affect the evolution of a reactor LOCA. Each of these heat
sources has a different intensity and time frame, and in general, there are different time frames before, d uring, and after
a LOCA when the effects of these heat sources and sinks become important. We would now like to describe the primary
heat sources and sinks in more detail.
34.4
Heat Sources Preceding a Reactor LOCA
Prior to the beginning of a LOCA, the primary source of heat is the thermal energy QTHERMAL or QNUCLEAR stored in the
primary and secondary loops of the NSSS. This thermal energy is generated by the process of nuclear fission in the core.
This source of heat generation stops when a LOCA begins, and then, no more thermal energy is produced by the fission
process directly. This occurs because either the control rods are dropped into the core, and the reactor is scrammed, or
the coolant leaves the core, which shuts down the fission process by hardening the neutron energy spectrum. In other
words, after a LOCA begins, no additional thermal energy is produced by the fi ssion process directly. However, addi-
tional heat is still produced by the fuel rods when radioactive materials within them continue to decay. (see Section 34.5).
This radioactive decay heat can be significant, and it can continue to be generated for long periods of time.
34.6 Heat Sources after a Reactor LOCA 1253
34.5
Heat Sources during a Reactor LOCA
In addition to the thermal energy stored in the primary and secondary loops, additional thermal energy can be produced
by what is called the zirconium–water reaction (see Chapter 33). This chemical reaction is an exothermic reaction that
can generate significant amounts of heat above 1,200°C. In fact, it can quickly become apocalyptic above 1,600°C if the
core is not quenched. Since this is chemical in nature, it is usually represented by the symbol QCHEMICAL. Normally, the
zirconium–water reaction is modeled in two ways—first with the core quenched and then without the core quenched.
The second scenario produces more thermal energy and creates higher containment pressures than the first scenario
(the quenched core), but even in scenario #1, significant amounts of thermal energy are released. The hydrogen gas H2
produced by this reaction can even allow a hydrogen explosion to occur. In fact, several hydrogen explosions related
to this reaction have been observed in nuclear power plants over the years. The sources of thermal energy before, dur-
ing, and after a LOCA are summarized in Table 34.1. A more extensive description of the sizes and severities of these
LOCAs can be found in Chapter 30, where the characteristics and consequences of small, medium, and large break
LOCAs are discussed.
34.6
Heat Sources after a Reactor LOCA
At the conclusion of a LOCA, additional thermal energy may be added to the containment building by the radioactive
decay heat QDECAY produced in the core. This heat is produced by the decay of radioactive materials in the fuel rods.
Normally, this decay heat is very important during the first week or two following a LOCA. When a reactor is initially
shut down, the amount of decay heat generated by the fuel rods is equal to about 7% of the heat generated during
normal full power operation. In other words, if the LOCA starts at t = 0, this decay heat is equal to about 7% of the
steady-state power level P at that time. This decay heat then falls to about 0.1% of the steady-state power level after
about one week. However, as the accident at Fukushima (Japan) showed, the removal of this decay heat is one of the
most important priorities of a reactor operator after the core has been scrammed. The sum of these three heat sources
(nuclear heat, chemical heat, and decay heat) then determines the total amount of thermal energy Q(t) that the contain-
ment building must be able to absorb over time.
Every credible containment analysis code has the capability to model these sources of thermal energy in great detail.
The final pressure and temperature levels in the containment building are then affected by the values of QNUCLEAR(t),
QCHEMICAL(t), and QDECAY(t).
TABLE 34.1
Sources of Heat Addition and Absorption during Typical Reactor LOCAs
34.7
T he Effects of Heat Sources and Sinks on the Evolution of a LOCA
Earlier, we mentioned that significant amounts of decay heat are produced by the fuel rods after the core is scrammed.
This decay heat builds up incrementally when the core is shut down, and if it is not removed by the NSSS or the ECCS,
it can cause the cladding to melt and the fuel rods to fail. The total thermal energy deposited in the containment building
by this decay heat can be determined from the following equation:
Q DECAY =
∫Q DECAY (t) dt (34.2)
where the limits of integration are assumed to be from t = 0 (when the reactor scrams) to t ≈ 30 days (when the decay heat
becomes almost negligible). A typical graph of the decay heat as a function of time is shown in Figure 34.2. The value
of QDECAY can then be found in one of three ways:
☉☉ It can be found numerically.
☉☉ It can be found from a lookup table provided by the American Nuclear Society (the ANS) or the American Society
of Mechanical Engineers (the ASME).
☉☉ It can be found by integrating the Equation 34.2 to obtain an analytical expression for QDECAY.
34.8
Time Dependence of the Decay Heat
Perhaps the most popular expression for QDECAY(t) is
which is also called the Wigner–Way equation. The Wigner–Way equation is commonly used to determine the time-
dependent decay heat after a reactor has been scrammed. In this equation
FIGURE 34.2 Radioactive decay heat released by nuclear fuel rods after reactor shutdown as a percentage of the reactor power
level prior to shutdown.
34.9 The Decay Heat Deposited in the Containment Building over Time 1255
FIGURE 34.3 The decay heat produced from the fuel rods in a nuclear power plant after the plant is shut down with an average
burnup of 50,000 MWD/tom. The numbers shown here represent the ratio of the total decay heat produced to the rated power output
of the plant in 1 s.
Consequently, if a reactor has been operating for 1 year prior to shutdown, the power level at t = 1 s after shutdown is given by
When t = 2.628 × 106 s (about a month after shut down), the value of Q3 falls to about 0.001P. Hence, the decay heat
falls very rapidly for the first week or two after a reactor has been shutdown, and then it becomes progressively less
important to the safety of the plant. The total decay heat is shown as a function of time in Figure 34.3. This graph shows
the total decay heat that is deposited in the containment building (or the NSSS) due to the radioactivity in the core.
This decay heat is due to the decay of both the fission products and the Actinides. See our companion book* for a more
comprehensive description of how this decay heat is produced.
34.9
T he Decay Heat Deposited in the Containment Building over Time
We can integrate the Wigner–Way equation to find the amount of decay heat QDECAY that is deposited in the core (and ulti-
mately in the containment building) from the time the core is shut down (at t = 0) until an arbitrary later time t. The result is
* See Nuclear Engineering Fundamentals: A Practical Perspective by R.E. Masterson, CRC Press (2017).
1256 Response of a Containment Building to a Reactor LOCA
Accordingly, the amount decay heat deposited in the containment building after about 1 day (86,400 s) is
Q DECAY (1 day) = 0.066P 86,400 0.8 − (86,400 + 31,536,000)0.8 + 31,536,000 0.8 (34.7a)
or
Q DECAY (1 day) = 0.066P[8,896.3 − 999,990.5 + 997,804] = 0.0777P × 6,709.8 = 521.35P (34.7b)
for a reactor that has been operating for 1 year (to = 31,536,000 s) prior to shutdown. For a 3,000 MW thermal PWR with
P = 3,000 MWT, this corresponds to a total energy deposition of
Needless to say, this is a very large amount of thermal energy for the containment building to absorb. The energy
released by the alpha decay of the heavier elements in the fuel rods increases this value slightly.
34.10
Estimates of the Thermal Energy Generated by the Zirconium–Steam Reaction
The final source of thermal energy produced by a LOCA is the thermal energy released by the zirconium–water reac-
tion. Sometimes this reaction is also called the zirconium–steam reaction. The total oxidation rate R of the fuel rods as
a function of the local temperature and the pressure is
where
☉☉ R is the oxidation rate in g/(cm2 s).
☉☉ P is the pressure in atmospheres.
☉☉ k is Boltzmann’s constant = 8.617 × 10 −5 (eV/°K).
☉☉ T is the absolute temperature in degrees Kelvin (°K).
Thus, the oxidation rate R is generally very small (1 × 10 −20 g/m2/s) at 0°C, 6 × 10 −8 g/m2/s at 300°C, 5.4 × 10 −3 g/m2/s
at 700°C, and 300 × 10 −3 g/m2/s at 1,000°C. Normally, there is no clear threshold where the oxidation rate becomes
apocalyptic, but during a LOCA, it can become significant above temperatures of 1,000°C. Today, 1,204°C (2,200°F)
is the maximum permissible design limit for the cladding temperature during a water reactor LOCA. If the cladding
temperature exceeds this limit, very large amounts of additional thermal energy (and hydrogen gas) will be produced.
We can estimate how much thermal energy QCHEMICAL is produced by noting that one mole of oxidized zirconium will
produce 583,000 J of heat. Hence, if the oxidation rate is 1 g/m2 s, the total energy released is 6.46 kJ/m2 s, since one
mole of zirconium = 91.224 g. It is worth mentioning that the Nuclear Regulatory Commission currently assumes that
100% of the fuel rods in a commercial reactor core will fail and release their radioactivity into the containment building
during a large break LOCA. However, almost all realistic studies of this problem have shown that the fuel rod failure
rate is much closer to 10%, and so most countries (including those in Europe) have adopted the position that only 10%
of the fuel rods will fail by the time the core is refilled, and that only 1% of the zirconium in these rods will be oxidized
by the zirconium water and zirconium steam reactions. This then tends to limit the thermal energy production by these
exothermic chemical reactions. The problem below asks the reader to estimate the amount of heat the zirconium-water
reaction produces during a modern PWR LOCA. The amount of heat produced here is actually 50 times higher than
what most reactor vendors assume when designing their containment buildings.
Solution From Appendix B, each fuel assembly produces ~20 MW of thermal power. Therefore, there are approximately
3,500/20 = 175 assemblies in the core. Each fuel assembly contains about 20,000 kg of zirconium. Therefore, there is
approximately 175 × 20,000 = 3.5 × 106 kg of zirconium in the entire core. From our earlier discussion, the oxidation of
1 kg of zirconium metal produces 6.3 MJ of thermal energy. If half the zirconium is the core is oxidized, 1,750,000 kg of
zirconium will be oxidized by the water and the steam. This amount of oxidation will produce 1,750,000 × 6.3 MJ/kg =
1.1 × 107 MJ of thermal energy which must be absorbed by the containment building. [Ans.]
34.11
BWR Suppression Pools
In addition to these three heat sources, a reactor containment building can also contain a number of heat sinks. In
BWRs, the primary heat sink is the suppression pool. Suppression pools are designed to absorb some of the thermal
energy produced by a LOCA. As we mentioned in Chapter 32, these pools are used to suppress the pressure levels in the
containment building. Normally, the water in a suppression pool is maintained at a temperature of about 27°C (80°F). If
the primary loop ruptures, a steam–air mixture is funneled through vents between the dry well and the wet well, and it
eventually enters the suppression pool through a set of pipes a couple of feet (about 1 m) below the surface of the water.
The cold water condenses the hot steam, and this helps to cool the air bubbles as they rise to the surface. Thus, the water
in a suppression pool acts as a “heat sink” to soak up the excess thermal energy produced by the LOCA. It heats up to
nearly 200°F (~93°C) during the course of an accident, but it is sized so that the water in the pool never boils. The cooler
water reduces the pressure levels in the containment building because it absorbs most of the excess thermal energy.
Hence, for a given volume V, the pressure and the temperature in the containment building are reduced. The contain-
ment building would have to be either far stronger or far larger if it were not for the presence of the pool. In the United
States, suppression pool temperature limits are specified in a regulatory document called NUREG-0783. A picture of a
typical BWR suppression pool is shown in Figure 34.4.
In recent years, PWRs have started to use water-filled pools as well. As Figure 34.4 shows, the pipes feeding the pool
are always a couple of feet (about 1 m) below the surface of the water. BWRs are also equipped with safety relief valves
(SRVs) to control the pressure level in the primary loop. If these SRVs are opened, air is initially injected into the pool
from the discharge lines, and after the air clears, essentially pure steam is injected into the pool during the remainder
of the LOCA. Experiments have shown that the steam jet that enters the pool is relatively stable when the local pool
temperature is low. Hence, the condensation process, which is proportional to the temperature difference between the
water and the jet, proceeds in a relatively orderly manner until the water temperature begins to rise. However, as the
blowdown continues, adding more steam to the pool eventually causes the local pool temperature to increase to what is
known as the threshold temperature. During some LOCAs, the water temperature in the vicinity of the steam jets may
become high enough (around 200°F or 93°C) that the condensation rates at the steam–water interface fall below those
that are needed to readily condense the discharged steam. When this occurs, the condensation process develops insta-
bilities, and steam bubbles may form at the exit to the discharge lines. These bubbles grow and collapse with predict-
able frequency, and in the process, they can create density waves that impact against the boundaries of the pool. These
density waves can cause unwanted hydrodynamic forces to rattle the pool’s structure.
To limit the impact these waves have on the pool, the current practice is to restrict the suppression pool operating
temperature so that the threshold temperature is never reached. NUREG-0661 and NUREG-0487 describe how these
specific temperature limits have evolved over time. The current temperature limit for the pool as a whole is about 175°F
(80°C) for BWRs in the United States, although local temperatures near the discharge lines can sometimes reach 200°F
(93°C). Depending on the design of the pool, the difference between the local temperature and the bulk temperature can
be 2°C–7°C (3°F–12°F). Normally, density waves start to develop where the hot steam enters the pool. These waves usu-
ally develop 2–3 min after a LOCA begins. These waves can vary in both intensity and duration, and how long they persist
in the pool depends on the dynamics of a particular LOCA. The hydrodynamic loads exerted on the structure of the pool
can sometimes be considerable. In many scenarios, the bulk temperature in the pool reaches its maximum of about 30 min
(~2,000 s) after a LOCA begins. The temperature profile in this case resembles that shown in Figure 34.5. Normally, BWR
suppression pools have a volume of approximately 250,000 ft3 (7,080 m3), and about half of this volume contains water,
whereas the other half contains air. In the Mark I containment, the suppression pool has an average diameter of about
112 ft (34 m), and the diameter of the torus is about 30 ft (9 m). The average water depth is between 13 and 15 ft (~4 m). The
air space may be slightly pressurized to reduce the temperature of the water below the threshold temperature. This tends
to increase the ambient air temperature in the torus as well. The total volume of the water–steam mixture in the reactor
vessel is about 21,000 ft3 (600 m3). Hence, the amount of water in the suppression pool is approximately six times greater
than the mass of the coolant in the pressure vessel. After the LOCA begins, the average enthalpy of the water–steam mix-
ture falls in direct proportion to the volume of the water in the pool. Since the air pressure is proportional to the enthalpy
of the mixture, it falls in direct proportion to the temperature as well. Normally, the suppression pool acts as an energy
sink that appears in the energy equation for the entire containment building. If we write down this energy equation and
assume that the temperature of the cladding does not rise above 1,200°C, then the energy equation becomes
FIGURE 34.5 The bulk water temperature in a representative suppression pool during a large-break BWR LOCA.
34.12 Other Energy Sources during Severe Accidents 1259
before the effects of the decay heat are introduced into the equation. In Equation 34.10, mWS is the mass of the water–
steam mixture and m AIR is the mass of the air. Notice that the value of QPOOL is negative because the pool absorbs
thermal energy from the coolant and acts as an energy sink. The energy of the pool is simply m POOL·hl POOL. Thus, the
containment building volume and energy equations must be simultaneously satisfied to arrive at the temperature of the
air. In their most elemental form, these equations are
m WS ( h l + xh lv ) + c V m AIR R ′TAIR
′ = − m POOL h l POOL (34.11)
VC = m WS ⋅ ( υ l + xυ lv ) (34.12)
where we have neglected the volume of the air at the top of the suppression pool, υl is the specific volume of the liquid,
and m POOL is the mass of the water in the pool. For reasonable values of mWS, m AIR, and m POOL, the suppression pool tends
′ and also the pressure inside of the containment building. Without a suppression
to reduce the final air temperature TAIR
pool, the peak pressure in a BWR containment building could get as high as 200 PSI (~1.4 MPa). With a suppression
pool, the peak pressure falls to about one-third of this value (approximately 60 PSI or 0.40 MPa). If a suppression pool
was not present, the containment building would have to be larger or stronger to contain the energy of the LOCA. Hence,
it is generally more economical to build containment buildings with a suppression pool than pay for the additional con-
crete and steel that would be required to reinforce the buildings without it. In PWRs, overhead sprayers and auxiliary
water sources are used for essentially the same purpose (see Chapter 32). However, the energy sinks they create on the
right-hand side of the energy equation have a slightly different value. Once the zirconium has been consumed, the free
carbon that remains interacts with the water to produce more hydrogen gas H2. The governing reaction is
34.12
O ther Energy Sources during Severe Accidents
During very severe accidents (which occur very infrequently), the core can melt down into a mixture of molten materials
called corium. This corium consists of a combination of melted steel, zirconium, uranium dioxide, and other materi-
als and gases. In liquid metal fast breeder reactors (LMFBRs), liquid sodium is also added to this mix. If an accident
is severe enough, the temperature of this molten mass will rise above 2,300°C, and the resulting mixture will melt
through the bottom of the pressure vessel and empty onto the containment building floor. If this occurs, the corium can
react chemically with the concrete and a number of other by-products can be produced. The results of this reaction are
depicted in Figure 34.6, and it is commonly called the corium–concrete reaction. The ensuing set of chemical reactions
adds another 4% to 5% to the total amount of energy that is produced during an extremely large LOCA.
Ordinarily, at least some of the zirconium that is not oxidized by the zirconium–steam reaction will be oxidized when it
reacts with the CO2 gas that is produced when the corium comes into contact with the concrete floor. Thus, the corium
decomposes the concrete to create large quantities of CO2 and H2O. The specific amounts of water and carbon dioxide
that are released depend on the molecular composition of the concrete. The composition of a few common types of
concrete is shown in Table 34.2. Notice that the amount of carbon dioxide in the concrete can vary from about 4% by
weight when the silicon content is high to about 33% by weight when the silicon content is low. The additional CO2 and
H2O that are produced then oxidize at least some of the zirconium metal that remains in the molten corium. The set of
chemical reactions that allow this to occur are summarized in Table 34.3. However, this reaction happens to be slightly
1260 Response of a Containment Building to a Reactor LOCA
FIGURE 34.6 The corium pool at the bottom of a PWR pressure vessel following a severe reactor accident in which all of the
cooling to the core is lost. This accident was simulated using the Modular Accident Analysis Program (or MAAP) which is owned,
licensed, and administered by the Electric Power Research Institute. The MAAP program is available to nuclear utilities and other
interested institutions on request. (The simulation above was provided by Fauske and Associates—see http://www.fauske.com.)
TABLE 34.2
The Composition of Modern Concrete Mixes
Material In Siliceous Form (wt%) From Common Sand (wt%) From Limestone (wt%)
SiO2 70 28 6
CaO 13 26 45
MgO 1 10 4
Al2O3 3 3 2
H2O 6 6 5
CO2 4 20 33
Other materials 3 7 5
Totals 100 100 100
TABLE 34.3
Energy Sources and Energy Production during the First Day of a Severe LOCA
endothermic, and it requires about 155 kJ/mol °C at 2,300°C to produce the hydrogen gas. If the hydrogen gas explodes,
then additional energy will be generated as well. The relative amounts of energy released by all of these energy sources
are also shown in Table 34.3. Notice that decay heat is by far the most important source of additional energy during a
conventional LOCA. However, it is not the only one.
34.13
Performing an Energy Balance on the Reactor Containment Building
In a typical containment building, the energy of the fluids before a LOCA is
1. The energy of the water in the pipes, EWATER
2. The energy of the steam in the pipes, ESTEAM
3. The energy of the air in the containment building, EAIR
The total thermal energy E of these three components is
where EMIX is the energy of the water–steam mixture. This can also be expressed in terms of the specific enthalpy as
Here, we are neglecting the energy of the water vapor in the air, EWV. After a LOCA, the energy of the fluids in the
containment building is
1. The energy of the water left in the containment building, E ′WATER
2. The energy of the air–steam mixture in the containment building, E ′MIX
′
The total thermal energy E TOTAL of these components is
where the energy of the mixture in this case is the energy of the steam plus the energy of the air:
E ′MIX = E STEAM
′ + E ′AIR = m STEAM
′ ′
h STEAM + m ′AIR h ′AIR (34.18)
Hence, when the fluids finally settle and come into thermal equilibrium with each another, the total energy of the fluids
before the LOCA must equal to the total energy of the fluids after the LOCA. So when equilibrium is reached, we can
write
where the primed terms refer to the energy after the LOCA, and the non-primed terms refer to the energy before the
LOCA. Here, we are assuming that none of the heat transported by the fluids leaks out of the containment building or
participates in additional chemical reactions (like the hydrogen–water reaction). Thus, we are assuming that there are no
additional heat sources QIN or heat sinks QOUT when we write the conservation equations in this way. We will add these
heat sources to the energy equation in another section to illustrate the effects they have.
1262 Response of a Containment Building to a Reactor LOCA
34.14
T he Containment Building Energy Equation with Different Heat Sources and Sinks
Normally, the energy equation can be used to find the temperature of the air–steam mixture in the containment building
after a LOCA has subsided. The energy of the water and the steam when the LOCA begins EWATER + ESTEAM are simply
equal to the thermal power output of the core. The energy of the air in the containment building at that time is
where
☉☉ m AIR is the mass of the air (in kg).
☉☉ R is the ideal gas constant for the air (expressed in J/kg °K).
☉☉ T is the absolute temperature of the air (in °K).
☉☉ cV is the heat capacity of the air at constant volume (always dimensionless).
In practice, additional energy can be added to the containment building from the decay heat QDECAY produced in the fuel
rods and from the zirconium–steam reaction if the cladding temperature happens to get high enough. We will refer to
this latter form of heat as chemical heat, and as previously stated, we will represent it by the symbol QCHEMICAL. Thus,
the energy equation to be solved to predict the final temperature of the air–steam mixture inside of the containment
building is
E ′WATER + E STEAM
′ + c V m AIR RTAIR
′ = E WATER + E STEAM + c V m AIR RTAIR + Q DECAY + Q CHEMICAL (34.22)
E ′WATER + E STEAM
′ + c V m AIR RTAIR
′ = E WATER + E STEAM + c V m AIR RTAIR + Q DECAY + Q CHEMICAL − Q POOL (34.23)
where QDECAY, QCHEMICAL, and QPOOL all have different time constants associated with them. Finally, the value of EWATER +
ESTEAM is equal to the reactor power level P. The final form of the energy equation to be solved is then
E ′WATER + E STEAM
′ + c V m AIR RTAIR
′ = P + c V m AIR RTAIR + Q DECAY + Q CHEMICAL − Q POOL (34.24)
Normally, the terms on the right-hand side are evaluated or estimated in advance. Suppose that we call the sum of these
terms QTOTAL. Then, we can write
Q TOTAL (t) = P + c V m AIR RTAIR + Q DECAY (t) + Q CHEMICAL (t) − Q POOL (t) (34.25)
and the energy equation we must solve to find the containment temperature and pressure at the end of the LOCA is
where
Q POOL =
∫Q POOL (t) dt t = 0 to t ≈ ∞ (34.27)
Q DECAY =
∫Q DECAY (t) dt t = 0 to t ≈ ∞ (34.28)
Q CHEMICAL =
∫Q CHEMICAL (t) dt t = 0 to t ≈ ∞ (34.29)
and the limits of integration are from t = 0 to t → ∞. If the reactor does not have a suppression pool, then we can simply
set QPOOL = 0.
34.17 Dalton’s Law of Partial Pressures for Containment Building Analysis 1263
34.15
A Mass Balance in the Containment Building
Normally, the mass of the materials in the containment building (unless it ruptures and releases its contents to the sur-
rounding environment) is also conserved. Hence, the mass of the air, the steam, and the water before the LOCA must
equal the mass of the air, the steam, and the water after the LOCA. Obviously, these materials will redistribute them-
selves differently at different times during the LOCA, and they will have different energies, temperatures, and pressures
than their original energies, temperatures, and pressures. However, the total mass of these materials will be exactly the
same. Hence, with very little error, it is possible to write
where m′ refers to the mass of the individual components (in kg) after the LOCA, and m refers to the mass of the
individual components (in kg) before the LOCA. These equations for the conservation of mass, plus the previous
equations for the conservation of energy, can be used to find the temperature and the pressure inside of the contain-
ment building at the end of a LOCA if we know the temperatures and the pressures inside of the building before the
LOCA begins.
34.16
D etermining the Pressures and Temperatures in a
Containment Building after a LOCA
At the end of the LOCA, most of the water inside of the reactor pressure vessel and the piping system has turned to steam
(which can be either saturated or superheated), and this steam will be mixed with the air already in the containment
building. Thus, the final state of the system is an air–steam mixture that determines the temperature and the pressure. If
the temperature TMIX of this mixture is known, then we can calculate the pressure of this mixture from what is known as
Dalton’s law of partial pressures. We would now like to illustrate how Dalton’s law works, and we can present an actual
example to show you how the final pressure and the temperature can be found.
34.17
Dalton’s Law of Partial Pressures for Containment Building Analysis
Around 1800, an Englishman named John Dalton (see Figure 34.7) discovered that the pressure of a mixture of several
gases could be written as the sum of the pressures of each of the gases individually. Hence, the partial pressure is the
FIGURE 34.7 A picture of John Dalton and an application of his law of partial pressures.
(Picture of Dalton provided by Wikipedia.)
1264 Response of a Containment Building to a Reactor LOCA
pressure that a single gas would exert if it existed independently at the same temperature and volume. In some parts of
the world, Dalton’s law of partial pressures is also known as Dalton’s law of additive pressures. However, both expres-
sions refer to exactly the same thing. Dalton’s law of partial pressures can be written as
where T is the temperature of the mixture (in °K), pMIX is the pressure of the mixture (in Pa, atmospheres, bars, or PSI),
and pN is the pressure of the nth component. In Equation 34.31, all of the gases are assumed to coexist in the same vol-
ume of space V. Since we would like to find the pressure in a containment building when the air and steam come into
equilibrium, the volume of the gases in this case is simply VC, the volume of the containment building. By this, we mean
the free space in the building and not the total volume of the building itself (which may include the containment building
structures, the NSSS, the stainless steel, and the concrete).
34.18
A magat’s Law of Additive Volumes
There is another law of gas behavior that we also need to perform this analysis. This law is called Amagat’s law of addi-
tive volumes, and Amagat’s law states that “The volume of a gaseous mixture is equal to the sum of the volumes that
each gas would occupy individually if it existed alone at the temperature and the pressure of the mixture.” That is to say,
Amagat’s law of additive volumes can be written as
In Dalton’s law, pi is called the component pressure of component i, and in Amagat’s law, Vi is called the component
volume of component i. The ratios pi/pMIX and Vi/VMIX, respectively, are then called the pressure fraction and the volume
fraction of each component i in the gaseous mixture.
34.19
T he Ideal Gas Law for Containment Building Analysis
The final equation we need to find the pressure of the air–steam mixture is the ideal gas law, which we discussed at
great length in Chapter 6. Essentially, the ideal gas law allows us to establish a relationship between the volume of space
V that N moles of a gas occupy, its temperature T (in °K), and its pressure p. The most common form of the ideal gas
law is
where R′ is a universal gas constant, and N is the number of moles. In many simple experiments, it can be shown that
the value of the universal gas constant is R′ = 8.314 J/(mol °K). It can also be shown at standard temperature (273°K) and
pressure (1 atm or 101,325 N/m2 [STP]) that one mole of a gas occupies a volume of exactly 22.4 L. The ideal gas law
predicts the behavior of most gases with an accuracy of ±3%, and under the temperatures and pressures that one finds
during a LOCA, it typically has an even lower margin of error. Sometimes it is convenient to write the ideal gas law in
terms of the mass of a particular gas (in kg) rather than in terms of the number of moles. In this case, the appropriate
form of the ideal gas law is
where m is the mass of the gas (in kg), and R is the ideal gas constant, in J/kg °K. For air, the value of R is 286 J/kg °K,
and for water vapor (steam), the value of R is 462 J/kg °K. Other values for the ideal gas constant are computed below.
34.20 Adding Partial Pressures during a LOCA 1265
TABLE 34.4
The Chemical Composition of Ordinary Air
These numbers are typically used to perform a containment analysis. The air in the containment building is a mixture
of many gases, and its molecular composition is summarized in Table 34.4. The molecular weights of N2, O2, Ar (for
argon), and CO2 are
M ( N 2 ) = 14 + 14 = 28
M ( O 2 ) = 16 + 16 = 32
M(Ar) = 40
M ( CO 2 ) = 12 + 2(16) = 44
and from our earlier discussion of the ideal gas law, we know that the values for R are
R ( N 2 ) = 8.314/28 = 0.297
R ( O 2 ) = 8.314/32 = 0.259
R(Ar) = 8.314/40 = 0.208
R ( CO 2 ) = 8.314/44 = 0.189
since R = R′/M.
34.20
Adding Partial Pressures during a LOCA
Sometimes we would like to find the peak pressure pPEAK in the containment building, given the volume and the initial
air temperature. If the steam in the building is superheated, then we have a situation where the ideal gas law applies, and
1266 Response of a Containment Building to a Reactor LOCA
Dalton’s law of partial pressures also applies. In this case, the total pressure in the containment building is the sum of
the air pressure and the pressure of the superheated steam, which is given by the following equation:
Obviously, the pressure will change as the temperature inside of the building changes, but since the volume of the con-
tainment building VC is fixed, all we need is the temperature of the air–steam mixture from a simple energy balance to
find the pressure pMIX of the air–steam mixture. Applying the ideal gas law to the equation above gives
However, at the end of the LOCA (after the temperatures have had a chance to finally equalize), the air temperature
TAIR and the steam temperature TSTEAM will be the same, and they will also be equal to the temperature of the air–steam
mixture TMIX. In this case, we can write the equation for the pressure inside of the containment building as
Then, all we need to find the pressure is the value of TMIX, since the values of VC, m AIR, R AIR, and mSTEAM and RSTEAM
are already known. Next, we would like to illustrate this principle for the blowdown of the cold leg of a large PWR.
The conditions that exist in the containment building prior to a primary or secondary-side LOCA and at the end of each
LOCA are shown in Table 34.5. Example 34.4 illustrates how the final containment building pressure level can be found
using these equations.
TABLE 34.5
Conditions in a Typical PWR Containment Building Prior to and after a Primary-Side LOCA and a Secondary-Side LOCA
Secondary-Side LOCA
Coolant in secondary loop (prior to LOCA) 89 6.89 285 Mostly a saturated liquid
Containment building air (prior to LOCA) 50,970 0.100 27 Φ = 80%
Containment building mixture (following 51,060 0.446 (64.7 PSI) 205 x = 17%
secondary-side LOCA)
34.20 Adding Partial Pressures during a LOCA 1267
The opposite problem, where one knows the pressure that one would like to design to, but one does not know the volume
of the containment building, can be solved by inverting the equation for the mixture pressure PMIX, and solving for the
containment volume VC assuming that the temperature of the air–steam mixture TMIX is known. In this case, the relevant
equation is
( )
VC ( TMIX ) = TMIX p MIX ⋅ ( m AIR R AIR + m STEAM R STEAM ) (34.38)
and the only unknown variable on the RHS is the value of TMIX if we specify pMIX. Clearly, the more steam that is
injected into the containment building, the larger the containment building must be for a given value of p. To a first
approximation, this relationship is linear, and so as the power level of the reactor is increased, the value of mSTEAM
must also increase, and the size of the containment building (or the peak pressure) must increase as well. Obviously,
everything we have said so far is based on the assumption that the ideal gas law and Dalton’s law of partial pressures
apply. It is also based on the assumption that the steam in the air–steam mixture is superheated. In general, this is a
reasonable assumption because most of the water will flash to superheated steam during the initial phases of the LOCA
if the pressure in the containment building is close to atmospheric pressure when the LOCA occurs. However, if all of
the water does not flash to superheated steam, then the remaining steam will have too much water in it to be ignored,
and the temperature of the resulting air–steam mixture will not exceed the saturation temperature TSAT at the combined
pressure. In this case, we must replace the temperature of the air–steam mixture TMIX by the saturation temperature of
the steam TSAT, and we must replace the mass of the steam mSTEAM by the fraction of the total mass of the steam–water
mixture that has been converted into steam. (The remaining volume of the hot water can usually be ignored when it
comes to these types of pressure calculations.) The correct equation to use in this case is
if we want to solve for the volume VC. Here, mWS is the total mass of the water–steam mixture, and x is the final
e quilibrium quality. Hence, mSTEAM = x·mWS, and mWATER = (1 − x)mWS that if the saturation temperature and the satura-
tion pressure are known, then this equation can also be solved for the equilibrium quality x. The final mass of the water
and the steam in the containment building are then mSTEAM = x·mWS, and mWATER = (1 − x)mWS, where mWS is the initial
inventory of the water–steam mixture. This mixture can consist of the water–steam inventory mWSP in the primary loop
or the water steam inventory mWSS in the secondary loop. Obviously, the total water steam inventory in the entire plant
is then
where the subscript P refers to the primary loop, and the subscript S refers to the secondary loop. The total thermal
energy in both loops can be found from a similar energy balance:
or
where the subscripts WP and SP refer to the water and the steam in the primary loops, and the subscripts WS and SS
refer to the water and the steam in the secondary loops.
1268 Response of a Containment Building to a Reactor LOCA
34.21
I ncluding the Effects of Water on the Air Pressure in the Containment Building
If we would like to include the effects of the water that is left in the containment building on the final system pressure,
then we must estimate the volume of the water and subtract it from the volume VC that we have assumed for the contain-
ment building. The volume of the water can be found from
where the density of the water ρWATER must be evaluated at the saturation temperature TSAT and the final pressure PMIX.
The effective volume of the containment building, accounting for the presence of the water on the containment building
floor, is then
where VC is the initial volume. Thus, the equation for the final system pressure becomes
( )
p MIX ( TSAT ) = TSAT VC′ ⋅ ( m AIR R AIR + x ⋅ m WS R STEAM ) (34.46)
or
( )
p MIX ( TSAT ) = TSAT ( VC − VWATER ) ⋅ ( m AIR R AIR + x ⋅ m WS R STEAM ) (34.47)
Finally, using the expression for the volume of the water VWATER (Equation 34.44), the final pressure in the containment
building becomes
( ( ))
p MIX ( TSAT ) = TSAT VC − (1 − x) m WS ρWATER ⋅ ( m AIR R AIR + x ⋅ m WS R STEAM ) (34.48)
Thus, the presence of the water reduces the internal volume slightly, and this causes the final pressure to be somewhat
higher than it would be if the water could be removed from the building at the end of the LOCA. However, if any addi-
tional water were added to the building from overhead sprayers, or we include the effects of the additional water in the
suppression pool (in a BWR), the volume of this extra cold water will more than offset the additional space it occupies
because it causes the final quality of the water–steam mixture after the LOCA to be lower, and the final temperature
to be lower as well. Hence, if one performs a more detailed analysis, the presence of the cold water in the suppression
pool or from the overhead sprayers actually causes the size of the containment building to be reduced. This can be
determined by performing a simple energy balance where the energy of the final mixture (with the cold water added) is
compared to the energy of the mixture without it. The final pressure is lower because the final temperature is also lower.
Example 34.5 shows how large this effect can be.
34.22
Corrections for the Humidity of the Containment Building Air
Most of the time, the air in the containment building before the LOCA begins is not dry enough to neglect the effects
of humidity entirely. In a typical containment building, there can easily be 1,000 kg of water in the air due to the pres-
ence of normal humidity alone. In a BWR, there is also a large amount of water in the suppression pool. This additional
water tends to absorb some of the energy released by the LOCA, and even a small amount of water at a much lower
temperature can reduce the peak pressures that are observed. Under steady-state conditions, the water vapor in the air
contributes to the total pressure as well as the air molecules themselves. Dalton’s law of partial pressures can then be
used to find the total atmospheric pressure from the sum of these two terms:
Here, p1 is the pressure due to the presence of the air molecules (component 1), and p2 is the pressure due to the pres-
ence of the water molecules or the humidity in the air (component 2). The amount of water vapor in the air is usually
determined by the relative humidity or from a chart such as the one shown in Figure 34.8. The relative humidity of an
air–water mixture is defined as the ratio of the partial pressure of the water vapor (H2O) in the mixture to the saturated
vapor pressure of the water at a specific temperature. The relative humidity Φ is normally expressed as a percentage
between 0% and 100%, and it is calculated from the following equation:
where pWV is the partial pressure of the water vapor in the air, and pSAT is the saturation pressure of the water vapor.
If the relative humidity in the containment building is known, then we can find the partial pressure of the water vapor
p2(T) from
FIGURE 34.8 The amount of water vapor in ordinary air as a function of the relative humidity Φ and the temperature T.
1270 Response of a Containment Building to a Reactor LOCA
In a typical containment building at standard temperature (300°K) and standard pressure (100,000 Pa), the vapor pres-
sure is about 3,000 Pa, and the pressure of the air (without the water molecules in it) is about 97,000 Pa. Hence, the total
atmospheric pressure that we measure is
p MIX (T) = p1 (T) + p2 (T) = p AIR (T) + p WV (T) = 97,000 Pa + 3,000 Pa = 100, 000 Pa (14.7 PSI)
This approach can be helpful because it allows us to find the mass of the free air and the free water in the containment
building before a LOCA begins. The mass of the free air is simply
or
m WV = p WV ⋅ VC R WV TWV (34.55)
Earlier, we showed that the equation for the internal energy of an ideal gas is important because it allows us to calculate
the energy of the gas (and hence its temperature) from an energy balance when the air and the steam are mixed together
with boiling water at the end of a LOCA. The pressure in the containment building is then given by Dalton’s law of
partial pressures:
where
If the air and the steam both have the same temperature at the end of the LOCA, then the gas temperature T derived from
an energy balance can be used to find the containment pressure, and
or
( )
pCONTAINMENT (T) = T VC ⋅ [ m AIR R AIR + m STEAM R STEAM ] (34.60)
In the next section, we would like to discuss how this energy balance can be written, and how the value of T can be
found.
34.23 Finding the Final Quality from a Static Energy Balance 1271
34.23
Finding the Final Quality from a Static Energy Balance
In Section 34.12, we introduced the containment energy equation which we were able to write at the time as
E ′WATER + E STEAM
′ + c V m AIR RTAIR
′ = Q TOTAL (34.61)
where E ′WATER is the energy of the water at the end of the LOCA, and E STEAM
′ is the energy of the steam. Normally, all that
is known at the end of the LOCA is the total mass of the water–steam mixture mWS. Hence, for computational purposes,
it is usually more convenient to write this equation as
m WS ( h l + xh lv ) + c V m AIR RTAIR
′ = Q TOTAL (34.62)
where
☉☉ mWS is the mass of the water–steam mixture (in kg).
☉☉ hl is the enthalpy of the liquid in the saturated mixture.
☉☉ hlv is the enthalpy of vaporization.
☉☉ x is the quality of mixture.
We can determine the value of the final quality x from this equation if the final temperature inside the containment
building is known. Normally, one guesses at the final temperature of the air–steam mixture, and in doing so, we assume
that TAIR = TSAT. If we know the final temperature, then we can find the final pressure (from Dalton’s law and the ideal
gas law), and from this, we can determine the values of hl and hlv from the steam tables, which are presented for the
reader at the end of the book. Then, the equation above can be used to find the quality x of the water–steam mixture at
the end of the LOCA. This process is not particularly gratifying because in reality, Equation 34.62 always contains two
unknowns—T and x. We can find much more accurate values for the final quality and the final temperature if we have
another equation relating x and T. Then, we will have two equations for x and T with two unknowns. Normally, the
second equation comes from the fact that both the air and the steam occupy the same volume of free space VC within the
containment building and that they both contribute to the total pressure. If we know the specific volumes of the liquid
and the vapor (υl and υv), then we can write the specific volume of the two-phase mixture υMIX as
or
υ MIX = υ l + xυ lv (34.64)
where υlv is the specific volume of vaporization. Strictly speaking, υl and υlv are functions of the final temperature and
pressure, so we can think of these as coefficients that are not exactly constant under all conditions but become constants
when we specify T (and hence p). The final equation that we need to relate x to T is then
VC = m WS ⋅ υ MIX = m WS ⋅ ( υ l + xυ lv ) (34.65)
where υl = f(T) and υlv = f(T). Hence, the two equations that we need to simultaneously solve to find the final values of
x and T are
Normally, the approach that is used is to assume that we know the final value of T in the first equation, and to solve it
for x, given the values of p from Dalton’s law. With these values of T and p, we can then estimate the values of υl and
υlv, and solve the second equation for x. We must then iterate between the values of x and T until both of these equations
are simultaneously satisfied. If we obtain a value of x greater than 1.0 (which means that the steam is superheated and
that there is no liquid water left in the containment building), then the search technique fails, and we must use another
approach to find the final values of T, x, and p. Hence, the energy equation, the equation for the specific volume of the
mixture, and Dalton’s law of partial pressures are applied to estimate the initial equilibrium state. An iterative process
is then used to determine the final equilibrium state. Normally, one begins with T and p and then proceeds to find x.
However, computer programs are used to automate this process today, and in more sophisticated containment analysis
codes, the computation is considerably more complex because of the time dependence of the heat sources and sinks. We
would now like to comment on how this time dependence affects the containment building pressure calculation.
34.24
Containment Building Pressures before, during, and after LOCAs
Once a LOCA begins, the pressure inside a containment building depends on the size of the building, the capacity of
its pressure suppression systems, and the size of the LOCA. Normally, large-break LOCAs result in higher pressures
than small-break LOCAs. In PWR containments, one can perform an iterative energy balance on the air, the water, and
the steam to see how the pressure behaves as a function of time. Once the LOCA begins, the high-pressure water in the
primary loop immediately turns to steam. As we mentioned earlier, this process is called flashing. During the blowdown
phase, the pressure in the primary loop drops from 2,250 PSI to 1,000 PSI in a very short period of time. For large-break
LOCAs, this pressure reduction lasts less than one-tenth of a second, and it is dominated by sonic decompression waves
that propagate through the primary loop. These waves create very large transient pressure pulses across the fuel rods,
the core barrel, and other structural components. These pressure waves resemble shock waves, and they have many
similarities to water hammers. As high-pressure water enters the containment building, it immediately flashes and turns
to steam. This causes the pressure and the temperature in the containment building to rise. Thus, as the pressure in the
primary loop falls, the pressure in the containment building rises. The containment pressure PTOTAL is then the sum of
the steam pressure PSTEAM and the air pressure PAIR:
The pressure buildup continues until the passive low-pressure injection system begins to reflood the reactor pressure
vessel. Normally, this system comes on line about 10 s after the start of the LOCA (see Table 34.6). The water this sys-
tem uses is stored in what are called accumulator tanks. Then, the emergency core cooling system (ECCS), which is
an active system, provides additional cooling to the core about 5 seconds later. The timelines for the activation of these
systems are shown in Table 34.6. Sometimes these systems are also called engineered safety features (ESFs). Then, as
Table 34.6 indicates, the residual heat removal system (RHRS) and the containment overhead spray system are brought
into play. These systems are activated about 20 seconds after the LOCA begins. These systems reduce the temperature
in the containment building and eventually help to depressurize it. At the start of the LOCA, the conditions in Table 34.7
are assumed. Figure 34.9 shows the pressure buildup in the containment building as a function of time. The LOCA is
assumed to start at t = 0. The pressure rises rapidly to about six times its initial value within 15 seconds. Then, the ECCS
begins to inject cold water into the core which tends to decelerate the rate of pressurization. After about 20 seconds,
the containment building pressure reaches its design limit, which in many countries is 40–50 PSI (275–345 kPa). Then,
the overheat sprayers and the recirculation fans come on line to reduce the containment building temperature. They
continue to operate until the pressure drops to its previous value of about 6 PSI 160 min (or 10,000 s) later. The effect of
these engineered safety systems (ESSs) on the containment building pressure levels is shown in Figure 34.10.
Once the LOCA has subsided, heat in the building is removed by rejecting it to a large pool of water at the bottom
of the containment building called a sump. The sump is filled by dropwise condensation from the overheat sprayers and
the walls of the containment building as the temperature inside the building falls. The hot water in the sump is then
recirculated through the heat exchangers before it is pumped to the overhead sprayers again. Usually, large LOCAs
34.25 Condensation in Reactor Containment Buildings 1273
TABLE 34.6
Engineered Safety Systems That Come into Play Following a PWR LOCA
TABLE 34.7
Initial Conditions in the Containment Building and the
Primary Loop before a PWR LOCA
cause the containment pressure to increase more rapidly than small- and medium-sized LOCAs do. The pressurization
and depressurization times for small, medium, and large PWR LOCAs are shown in Table 34.8. The pressurization and
depressurization times for BWRs are slightly different. Here, a small LOCA is defined as one where 2% of the inventory
in the primary loop is lost, a medium LOCA is defined as one where 15% of the inventory in the primary loop is lost, and
a large LOCA is defined as one where 25% or more of the inventory in the primary loop is lost. If 100% of the inventory
of the primary loop is lost, the peak pressure in the containment building approaches 60 PSI (415 kPa)—which is some-
times regarded as the design limit for PWR containments in the United States. Other factors such as the size of the water
droplets released from the overhead sprayers (see Figure 34.11) can affect the rate of depressurization after a large LOCA.
Normally, droplets with diameters between 2 and 4 mm reduce the pressure in the containment building the fastest.
34.25
Condensation in Reactor Containment Buildings
As we mentioned earlier, modern PWRs use passive containment cooling systems (PCCS) in addition to the systems we
have already discussed. The Westinghouse AP-600, and its bigger brother the Westinghouse AP-1000 are two examples
of reactors that employ a PCCS. This system relies on natural convection or natural circulation to remove heat from the
containment building without the need for fans, pumps, and other electrically powered equipment. Hence, it functions
independently of overheat spray system and the containment building air recirculation fans. The idea behind the PCCS
was discussed in Chapter 22 (see Figure 22.13).
1274 Response of a Containment Building to a Reactor LOCA
FIGURE 34.9 The pressure buildup in a PWR containment building following small, medium, and large-break LOCAs.
If 100% of the inventory of the primary loop was lost, the peak pressure would be closer to 60 PSI. Here, the ESSs have not
yet been activated.
FIGURE 34.10 The effect of ESSs on the containment building pressure following small-, medium-, and large-break LOCAs.
Notice that it can take up to 10,000 s (about 3 h) before the pressure in the containment building returns to its initial value.
This assumes that heat is rejected from the building by circulating hot water through heat exchangers before returning it to the
overhead sprayers.
The water for the PCCS comes from a large water storage tank that is located near the roof of the containment building,
and cold water from this tank is allowed to run over the outside of the steel liner inside of the building. This cools the
top and the sides of the steel liner and helps to remove the waste heat. The steam that is generated in the in-containment
refueling water storage tank is transferred to the steel liner inside of the containment building through a set of pipes at
the top of the tank. The water flowing over the outside of the steel liner removes this decay heat through the process of
34.26 Fundamentals of Condensation Heat Transfer 1275
TABLE 34.8
Typical Pressurization and Depressurization Times for a PWR Containment Following Small-, Medium-, and Large-Break LOCAs
FIGURE 34.11 Condensation occurs in two forms called film condensation and dropwise condensation. Both forms of condensa-
tion are shown above. In either case, a small pool of water may accumulate at the base of the surface if the surface is inclined or
vertical. In film condensation, the thickness of the film δ is determined by the condensation rate. Both processes occur when a high-
temperature vapor such as steam is exposed to a surface with a temperature below TSAT. The rate of heat and mass transfer is then
proportional to the temperature difference between the vapor and the surface.
condensation. The air flowing between the steel liner and the outside of the containment building naturally cools the
heated steel liner. In other words, the air surrounding the containment building ultimately carries away the decay heat
from the core. (This is the same process that is used in a natural draft reactor cooling tower, which we first discussed
in Chapter 8.). After about 7 hours, enough of the heat is removed from the steel liner that some of the steam begins to
recondense back into water inside of the containment building. This water is redirected back into the IWRST for con-
tinued use in removing decay heat from the core. This process of heat removal continues until the core has completely
cooled (see Figure 34.12).
34.26
Fundamentals of Condensation Heat Transfer
In classical heat transfer, one learns that condensation is an extremely effective way to transfer and remove heat.
Condensation occurs when the temperature of a vapor is reduced below its saturation temperature TSAT. Most of the
time, this is done by bringing the vapor into contact with a solid surface whose wall temperature T WALL is below the
saturation temperature TSAT of the vapor. In reactor work, two distinct forms of condensation occur. The first form is
called film condensation, and the second form is called dropwise condensation. In film condensation, the condensate
wets the surface and forms a liquid film that slides down the surface under the influence of gravity alone. This film acts
as a thermal boundary layer that transfers heat from the condensate to the cooler surface. The thickness of the boundary
1276 Response of a Containment Building to a Reactor LOCA
FIGURE 34.12 The normalized heat transfer coefficient for laminar and turbulent flow of the condensate film on a vertical plate.
The normalization factor (g/ν2)0.33 commonly appears in the calculation of the heat transfer coefficient for a condensing film.
layer increases in the direction of flow as more vapor condenses on the film. In dropwise condensation, the vapor that is
condensing forms small droplets instead of a continuous film. The surface eventually becomes covered by these droplets
in various sizes and shapes. Most of these droplets are spherical in shape, and their sizes can be described by a statistical
probability distribution. Eventually, these droplets become so large that they slide off of the surface if it is vertical, and
this may cause a small pool of water to form at the base of the surface. These processes are compared in Figure 34.11.
Now let us examine the flow of heat from the vapor to the walls of the containment building when this occurs.
34.27
Film Condensation on Vertical Surfaces such as Containment Building Walls
In reactor containments, the walls on which the vapor condenses can be modeled as flat vertical plates. For film conden-
sation, the heat transfer rate is roughly proportional to the temperature gradient across the film—adjusted for the film’s
downward velocity. The velocity of the condensate is zero when it touches the surface of the wall because of the no-slip
boundary condition. The temperature of the condensate is TSAT at the outer surface of the film, and it decreases gradu-
ally until it becomes the same as the temperature of the wall TWALL. Just as in forced convection, the heat transfer rate
depends on whether the flow of the condensate is laminar or turbulent. Again the criterion that is used to define these
regimes is the Reynolds number Re. For the film on the surface of the wall, the Reynolds number is given by
where
δ is the thickness of the liquid film (m).
P is the wetted perimeter of the condensate (m).
DH = 4A/P = 4δ is the hydraulic diameter of the condensate flow (m).
A = Pδ is the wetted perimeter x film thickness, which is the cross-sectional area (m 2) of the condensate flow at the
lowest point in the flow.
ρ is the density of the liquid condensate (in kg/m3).
μ is the viscosity of the liquid condensate (in kg/ms).
V is the average velocity of the condensate (in m/s) at the lowest point in the flow.
m = ρVA is the mass flow rate of the liquid condensate (in kg/s) at the lowest point in the flow.
34.27 FILM CONDENSATION ON VERTICAL SURFACES 1277
Now, consider how fast the vapor condenses on the containment building walls. When the vapor condenses, the latent
heat of vaporization hlv is the heat released as a unit mass of the vapor condenses, and it normally represents the heat
transfer rate per unit mass of condensate deposited on the walls as condensation occurs. Thus, the mass transfer rate m
of the condensate at the surface of the film is proportional to the heat transfer rate Q. This means that the heat transfer
rate can be expressed as
Q = hA ( TSAT − TWALL ) = mh
lv (34.69)
of the
where A is the surface area of the wall. Rearranging this equation slightly, we see that the condensation rate m
vapor is given by
can then be substituted back into Equation 34.69 to obtain yet another expression for the Reynolds
The value for m
number:
This expression can then be used to determine whether the flow is laminar or turbulent. The temperature of the conden-
sate varies from TSAT at the liquid–vapor interface to TWALL at the wall surface. Therefore, the properties of the liquid film
must be evaluated at TFILM = (TSAT + TWALL)/2 which is approximately the average temperature of the liquid. However,
the value of hlv must still be evaluated at TSAT because it is not affected by the liquid subcooling. Various expressions
for h have been developed over the years for different condensates. For vertical plates, which closely resemble contain-
ment building walls, the value of h can be seen in Figure 34.12. Notice that this picture shows the normalized value of
h, where the normalization factor is (vl2/g)1/3/k l. A justification for using this factor is presented in many heat transfer
books. For steam, the Prandtl number required to determine the value of h is close to 1.0. The heat transfer coefficient in
Equation 34.69 again depends on the Reynolds number of the condensate in the film. For very smooth films where the
streamlines are well defined and the flow is laminar, the heat transfer coefficient is given by
( )
0.33
h vertical = 1.47k ⋅ Re −0.33 g ν2 (0 < Re < 30) (34.72)
where ν = μ/ρ is the kinematic viscosity of the liquid condensate, and k is its thermal conductivity. When the Reynolds
number becomes greater than about 30, waves form at the liquid–vapor interface although the flow inside the film
remains primarily laminar. The flow in this case is said to be wavy laminar. The waves at the liquid–vapor interface tend
to increase the heat transfer coefficient, and the increase in the heat transfer due to the wave effect is, on average, about
20%, although it can exceed 50% in some cases. The average heat transfer coefficient under these conditions is given by
( )( )
0.33
h vertical = k ⋅ Re 1.08Re1.22 − 5.2 g ν2 (30 < Re < 1,800) (34.73)
Finally, at a Reynolds number of about 1,800, the condensate flow becomes turbulent and the heat transfer coefficient
is given by
( )( )
0.33
h vertical = k ⋅ Re 8,750 + 58Pr −0.5 Re 0.75 − 253 g ν2 (Re > 1,800) (34.74)
This expression for heat transfer coefficient again assumes a flat vertical plate.
Solution The properties of water at 100°C are ρl = 965.3 kg/m 3 and ρ v = 0.60 kg/m 3. The properties of the liquid
film must be evaluated at T FILM = (TSAT + T WALL)/2. The heat of vaporization is approximately 2,315 kJ/kg, ν =
μ/ρ = 0.33 × 10 −6 m 2/s, and k = 0.675 W/m °C. Since the Reynolds number is assumed to be 1,000, the heat trans-
fer coefficient is given by hvertical = k·Re/(1.08Re1.22 − 5.2)(g/ν 2)0.33 = (0.675 × 1,000)/[1.08 × (1,000)1.22 − 5.2)](9.81/
(0.33 × 10 −6)2)0.33 = (675/4,930) × 4,025 = 550 W/m 2 °C. The heat transfer rate is Q = hA(TSAT − T WALL) = 550 × 10 ×
(100 − 80) = 110,000 W.
Part 2: The rate of condensation of the steam is m CONDENSATION = Q/hlv=110 kJ/s/2,315 kJ/kg = 0.0475 kg/s. That is, the
steam will condense on the surface of the plate at a rate of 47.5 g/s. [Ans.]
34.28
Film Condensation on Inclined Surfaces
Inclined surfaces can be treated in almost the same way as vertical ones, but for these surfaces, the heat transfer coef-
ficient must be adjusted for the angle of inclination θ. If the condensation occurs on the upper surface of a plate which
is inclined by an angle θ with respect to the vertical direction, the value of g in the equations for the heat transfer
coefficient must be replaced by g cos θ. This approximation gives acceptable results especially when θ ≤ 60°. The con-
densation heat transfer coefficient for the inclined surface is then
when the flow of the condensate is laminar. A similar correlation can be developed for turbulent film flow.
34.29
Comparing Dropwise Condensation to Film Condensation
Dropwise condensation is a much more prevalent form of heat transfer in a reactor containment building after a
LOCA. In fact, after a LOCA subsides, it is the dominant form of heat transfer on the containment building walls. This
form of condensation is characterized by countless droplets of varying diameters instead of a continuous liquid film.
The heat transfer rates for dropwise condensation are higher than those for film condensation. As a matter of fact,
after a LOCA subsides, they can be 10–20 times higher. However, in unoptimized scenarios, they may be only three
to five times higher:
In dropwise condensation, the small droplets that form at the nucleation sites on the walls grow as a result of
c ontinued condensation, and when they reach a certain size called the critical size, they coalesce and slid down
the side of the walls, exposing them to more vapor. There is no liquid film is this case to reduce the heat transfer
coefficient. Consequently, heat transfer coefficients for dropwise condensation are a factor of 10 higher than those
that can be achieved for film condensation. Dropwise condensation of steam on a vertical surface is illustrated in
Figure 34.13.
34.30 HEAT TRANSFER COEFFICIENTS FOR DROPWISE CONDENSATION IN REACTOR 1279
34.30
Heat Transfer Coefficients for Dropwise Condensation
in Reactor Containment Buildings
Dropwise condensation has been studied extensively for many surface–fluid combinations. Of these, studies involv-
ing the condensation of steam on steel and copper surfaces have attracted the most attention because of their potential
applications in nuclear and coal-fired power plants. In the early 1980s, Peter Griffith at MIT proposed the following
correlation for the dropwise condensation of steam on a copper surface:
h DROPWISE = 51,104 + 2,044TSAT (for 22°C < TSAT < 100°C) (34.77a)
In this correlation, TSAT is in °C, and the heat transfer coefficient hDROPWISE is in W/m2 °C. Copper is a popular metal in
power plant heat exchangers because of its excellent thermal conductivity. Similar correlations have been applied to the
stainless steel liners in reactor containment buildings. However, the thermal conductivity of steel is about 15 times lower
than that of pure copper and so the heat transfer coefficients for dropwise condensation are a factor of 10–20 lower as
well. The reader is referred to the references at the end of the chapter for more information regarding these correlations.
Therefore, dropwise condensation is the preferred form of passive heat removal from a reactor containment building
following a LOCA. Subsequently, LeFevre and Rose* developed an alternative correlation for pure, quiescent steam. In
their correlation, the heat transfer coefficient was expressed as a function of the vapor to surface temperature difference
ΔT = TSAT − T WALL (in °C) and the Celsius temperature of the vapor. The explicit form of their correlation was
h DROPWISE = ( TSAT )
0.8
(5 + 0.3∆T ) ( in kW m 2
)
°C (34.77c)
The first term indicates that the heat transfer coefficient increases as the pressure of the steam increases. The second
term represents the temperature difference between the vapor and the surface. Now let us apply this equation to a con-
tainment building wall.
* See www.Thermopedia.com.
1280 Response of a Containment Building to a Reactor LOCA
Solution Using the correlation of LeFevre and Rose, the heat transfer coefficient for the wall at the pressure peak is
hDROPWISE = (TSAT)0.8(5 + 0.3·ΔT) = (200)0.8(5 + 0.3 × (200 − 80)) = 69.3(5 + 36) = 2,841 kW/m2 °C. Following the pressure
peak, the saturation temperature subsides to 100°C and the heat transfer coefficient becomes hDROPWISE = (TSAT)0.8(5 +
0.3·ΔT) = (100)0.8(5 + 0.3 × (100 − 80)) = 39.8(5 + 6) = 438 kW/m2 °C. [Ans.]
34.31
Calculating the Release of Fission Products within the
Containment Building
During a really severe accident, such as those addressed in the WASH-1400 report (see Chapter 30), some corium (see
Section 34.10) can be expelled from the reactor pressure vessel into the containment building at very high speeds. This
molten core debris can pressurize the containment building through a process called direct containment heating (DCH).
For years, the process of DCH has been the subject of extensive analytical analysis and experimental validation by the
NRC and other regulatory agencies. These efforts eventually led to the development of a number of severe accident analy-
sis programs that track the progression of the corium mixture through the containment building and also the behavior of
subcompartments (SCs) within the building if a less severe LOCA were to occur. Normally, this requires a computer pro-
gram to model the behavior of air–steam mixture, the condensation and evaporation of materials on the structures inside
of a containment building, and the behavior of the aerosol particles to get the predictions to agree with experimental data.
This means that fission product decay and transport processes also have to be modeled with a fair degree of accuracy. Most
computer programs that are used for this purpose divide the containment building into a number of SCs, and each of these
SCs contains a major reactor subsystem (such as the core, the steam generators, and the primary or secondary piping).
If a high energy pipe rupture occurs, an asymmetric pressurization of the containment building can occur, where
the pressure in some sections of the building will initially be higher than the pressure in others. This asymmetric pres-
surization can lead to large local pressure differences between the SCs or even within an individual SC. This can result
in pressure waves or unexpected loads that can damage structures inside of each SC. To simulate a serious accident,
the containment building is usually partitioned into a network of interconnected control volumes or computational
cells. Each control volume has a number of external surfaces or geometrical interfaces associated with it that allow it to
communicate with its nearest neighboring cells. The layout of these cells for a typical containment building is shown in
Figure 34.14. The cells can represent an actual SC or group of interconnected SCs, although sometimes an individual SC
can be partitioned into a number of individual nodes to model phenomena such as natural convection and stratification
within the compartment. The mesh cells communicate with each other through the flow of mass and energy between
the cells or by the conduction of heat between the cells through heat transfer through the surrounding structures. The
arrows in Figure 34.14 indicate some directions that this mass and energy transfer can occur. Usually, the environment
outside of the containment building can be modeled by a single computational cell that occupies a very large volume
of space (often larger than the volume of the containment building itself). This approach makes it much easier to apply
the boundary conditions from outside of the containment building to the internal cells, and it also allows the release of
fission products to the environment to be tracked in terms of the inventories of the individual cells. Usually, a relatively
small number of computational cells are used to model a severe accident, but each of these cells can have a relatively
large number of degrees of freedom associated with them. Most of the time, one tries to model the physics inside of the
containment building with two thermal-hydraulic fields: a global fluidics field for the atmosphere and another global
fluidics field for the coolant within each cell. The atmospheric field consists of
1. The atmosphere bulk fluid, which includes gases, coolant vapor, and any homogeneously dispersed liquid coolant
2. Aerosols represented in terms of a number of size classes, each with a separate material composition
3. Fission product distributions assigned to the gas and each aerosol component material
4. Finely dispersed core debris, represented in terms of an arbitrary number of debris droplet fields. Such dispersed
core debris can result, for example, from high-pressure melt ejection from the reactor pressure vessel.
The coolant field consists of
1. The pool bulk fluid
2. Aerosols deposited into the pool
3. Fission product distributions assigned to the pool.
These fields can interact with each other in very complex ways, and they can be modeled within a cell or between
groups of interconnected cells. Some of the physical processes that are modeled are shown in Figure 34.15. Note
in particular that a large amount of physical feedback can occur between each of the fields. Some of the processes
in the feedback loop are intercell flow, hydrogen combustion, heat and mass transfer processes (e.g., convection,
34.32 Containment Analysis and Safety Codes 1281
FIGURE 34.14 A reactor containment building subdivided into interconnected computational cells to find where the radioactivity
will go. In this picture, the boundaries of the cells are represented by dotted lines, and the flow of mass or energy across the cells is
represented by red arrows.
condensation, condensate film flow, thermal radiation, conduction, and concrete out-gassing), aerosol behavior (e.g.,
agglomeration, deposition, and condensation), fission product behavior (e.g., decay, heating, and transport), ESFs (e.g.,
containment sprays, fan coolers, and ice condensers), processes associated with but not limited to BWRS (e.g., vent
clearing, gas-pool equilibration, and aerosol scrubbing), DCH caused by high-pressure ejection of finely divided core
debris from the reactor vessel, and core–concrete interactions. These modeling categories can be further subdivided
into numerous separate phenomenological models, a few of which are noted parenthetically above. Taken collectively,
these models provide containment analysis codes with the capability to analyze a wide variety of light water reactors
and DBAs. Through appropriate user input, large-dry, subatmospheric, ice condenser PWR containment buildings
can also be modeled. BWR containment buildings and advanced light water reactor containment buildings can be
modeled using similar techniques. Normally, time step control during a simulation of this type can be very important
because there is so much number crunching involved, and because some of the reactions happen very quickly (on the
order of seconds), while others can happen very slowly (on the order of minutes). A typical containment analysis code
can still not be run on a high-speed personal computer because even a single time step can take several hours (or days)
of processor time. Hence, most serious accident analysis codes are designed to be run on only large IBM mainframe
and supercomputers. In the next section, we will examine some of these containment analysis codes and discuss their
important features and functions.
34.32
Containment Analysis and Safety Codes
Because reactor accidents are so complex, the NRC has historically used sophisticated computer programs to determine
the response of a reactor to certain hypothetical DBAs. The physical models within these computer programs can be
very complex, and the computer programs within the nuclear industry that use them are known as computer codes. In
this section, we would like to briefly discuss some of the mainstream licensing codes that are used to predict the behav-
ior of nuclear reactors during LOCAs and other unusual Category III and IV events (A comprehensive summary of the
codes the NRC uses is presented in Chapter 33). The reactor vendors, who design and build the reactors, have their own
1282 Response of a Containment Building to a Reactor LOCA
FIGURE 34.15 An illustration of the physical processes modeled by the CONTAIN Code to predict the response of a reactor
containment building to a LOC.
proprietary computer programs, and sometimes the programs the NRC uses give slightly different results than the ones
the reactor vendors use. These discrepancies are usually driven by differences in the underlying physical assumptions
(such as a four-equation vs. a five-equation vs. a six-equation model), or because the reactor vendors use proprietary
data that the NRC does not rely upon in its standard licensing models. However, the overall trends predicted by these
computer programs are generally in good agreement. In the next few sections, we would like to discuss some of the
mainstream containment analysis codes that are in widespread use today. We have mentioned some of them before, but
now, we would like to discuss them from a more holistic point of view, To do this, we also need to describe the numerical
methods and the physical models they use. A large amount of information about these computer codes can be found at
the following website, which the NRC maintains for the benefit of the general public:
http://www.nrc.gov/about-nrc/regulatory/research/safetycodes.html
We will defer a detailed discussion of how these programs work until a future edition.
34.33
Popular Containment Analysis Codes
There are three major containment analysis codes in widespread use today and a number of less popular ones. The pri-
mary containment analysis codes used by the nuclear industry include the Integral Containment Analysis (CONTAIN)
code, the Methods for Estimation of Leakages and Consequences of Releases (MELCOR) code, and another code
known as GOTHIC, the Generation Of Thermal Hydraulic Information for Containment buildings code. All three
of these codes have been around since the 1980s, and some of their development was accelerated after the accident at
TMI. We would now like to describe some of their most important features and give you an idea of how they work.
The GOTHIC code is more suitable for modeling LOCAs that do not involve the meltdown of the core and the creation
of large amounts of corium. However, the CONTAIN and MELCOR codes are specifically designed to model reactor
accidents where this can occur.
34.36 GOTHIC 1283
34.34
T he CONTAIN Code
CONTAIN was developed to model severe reactor accidents, and it uses very detailed mechanistic models to do so.
Unfortunately, its development was terminated by the NRC in the mid-1990s, and it has not contained any major updates
since that time. In general, it can only be run on supercomputers. More detailed information about the models that
CONTAIN uses can be found in the following article, which the NRC still provides:
☉☉ Murata, K.K., Williams, D.C., Tills, J., Griffith, R.O., Gido, R.G., Tadios, E.L., Davis, F.J., Martinez, G.M.,
and Washington, K.E., Code manual for CONTAIN 2.0, A Computer Code for Nuclear Reactor Containment
Analysis, NUREG/CR-6533, SAND97-1735, 1997.
The CONTAIN code can predict the physical, chemical, and radiological conditions inside the containment and
connected buildings of a nuclear reactor in the event of an accident. CONTAIN was developed at Sandia National
Laboratories under the sponsorship of the US Nuclear Regulatory Commission (the U.S. NRC) for analyzing contain-
ment phenomena under severe accident and DBA conditions. It is designed to predict the thermal-hydraulic response
inside containment buildings and the release of radionuclides to the environment in the event of a containment failure.
The modeling capabilities of CONTAIN are sufficiently flexible that it can be applied to the analysis of nonreactor
problems, such as the migration of radioisotopes in waste repositories, and the thermal-hydraulic response of nonnuclear
facilities under accident conditions. The CONTAIN code usage in reactor safety applications is its primary targeted
application. Such analyses are an integral part of the U.S. NRC’s Severe Accident Research Program, where reactor
safety issues are addressed through an appropriate mix of experimental and analytical research.
34.35
M ELCOR
MELCOR has similar capabilities to CONTAIN, but they are less detailed in some areas. However, the NRC rec-
ommends using MELCOR instead of CONTAIN because active development of its underlying models continues at
this time. MELCOR was also developed at Sandia National Laboratories, and development on it continues today in
Albuquerque, New Mexico. MELCOR applies specifically to light water reactors (both PWRs and BWRs). It can be
used for LMFBRs and gas reactors with some basic differences in the modeling assumptions. It is a second-generation
plant risk assessment tool, and it is the successor to the Source Term Code Package. A broad spectrum of severe accident
phenomena in both boiling and PWRs is handled in MELCOR within a unified framework. These include the thermal-
hydraulic response of the reactor coolant system, the reactor cavity, and the containment and confinement buildings;
core heatup, degradation, and relocation; core–concrete attack; hydrogen production, transport, and combustion; and
fission product release and transport behavior. MELCOR applications include estimation of severe accident source
terms, and their sensitivities and uncertainties under many different conditions. MELCOR is also used to analyze DBAs
for advanced plant applications (including ESBWRs, EPRs, and APWRs).
34.36
G OTHIC
GOTHIC is a modern computer program that was developed by a group of thermal-hydraulic engineers and methods
developers who came out of Battelle Pacific Northwest Laboratories and worked on the COBRA series of computer
codes at the Hanford site in Richland, Washington in the early 1980s. Most of the author’s initial experience with
nuclear reactor thermal-hydraulics came from working with this team. Today, GOTHIC is developed and marketed
by Numerical Applications Inc. (NAI), a company that is also headquartered in the Federal Building in Richland,
Washington. The URL for their website is
www.numerical.com
GOTHIC was developed by NAI with funding from the Electric Power Research Institute, and it has been reviewed and
approved by the U.S. NRC for containment analysis applications for several U.S. utilities. The latest version of GOTHIC is
called GOTHIC 8. It was developed specifically to analyze reactor containment buildings; however, it can be used to model
the thermal-hydraulic performance of auxiliary buildings outside of the containment building as well. GOTHIC solves a
set of time-dependent fluid mechanics equations involving water, steam, and non-condensable gases in containment build-
ings and containment sub-compartments. The nodal connection methods are very similar to those that were used in the
COBRA and VIPER codes, and the physical processes are modeled using a lumped-parameter approach. GOTHIC has
been benchmarked against several multidimensional containment experiments, including the Battelle Model Containment
System. For large and complex problems, Gauss–Seidel, conjugate gradient and multidimensional iterative techniques are
used to solve the fluid conservation equations. Time steps within the program are automatically calculated and adjusted
1284 Response of a Containment Building to a Reactor LOCA
so that the solution always remains stable. However, the largest and the smallest values of the time step must generally
be specified by the user. Westinghouse has recently decided to use GOTHIC for their future containment analysis work.
34.37
Regulatory Documents Pertaining to Reactor Containment Buildings
In the United States, numerous regulatory documents are used to specify how containment buildings are expected to be
constructed and behave. The most relevant of these documents is
pbadupws.nrc.gov/docs/ML0932/ML093200703.pdf
As far as regulatory documents go, it is a relatively new document and its last major revision occurred in 2010. It
deals primarily with the structural loads that the containment building is expected to withstand. In addition, the
ANS has established an alternative set of technical standards that can be used to calculate the time-dependent
temperature and pressure profiles during a LOCA. This set of procedures can be found in ANS 56.4: Pressure and
Temperature Transient Analysis for Light Reactor Containments (1988). A list of the specific DBAs a containment
building is expected to handle are also specified in this document. Since the accidents at Fukushima, Japan, com-
bustible gas control systems (hydrogen igniters and suppression systems) have been installed in most commercial
nuclear power reactors. For light water reactors, the U.S. government now lists the requirements that these systems
are expected to meet in
10 CFR 50, section 44, “Combustible Gas Control for Nuclear Power Reactors”
This regulatory guide provides a list of acceptable methods for meeting the requirements of 10 CFR 50.44 for small-,
medium-, and large-break LOCAs.
Bibliography
Books and Textbooks
Cengel, Y. Heat Transfer: A Practical Approach, Second edition, McGraw Hill, New York (2003).
Cengel, Y. and Boles, M. Thermodynamics: An Engineering Approach, Sixth edition, McGraw Hill, New York (2007).
Cengel, Y. and Cimbala, J. Fluid Mechanics: Fundamentals and Applications, McGraw Hill, New York (2006).
Crowe, C. and Schwartzkopf, J., et al. Multiphase Flows with Droplets and Particles, C&R Press, Boca Raton, FL (2012).
Duderstadt, J. and Hamilton, L. Nuclear Reactor Analysis, John Wiley and Sons, New York (1976).
El-Wakil, M.M. Nuclear Heat Transport, American Nuclear Society, La Grange Park, IL (1981).
Glasstone, S. and Eedlund, M.C. The Elements of Nuclear Reactor Theory, D. Van Nostrand Company, Inc., New York (1952).
Glasstone, S. and Sesonske, A. Nuclear Reactor Engineering, Third edition, Van Nostrand Reinhold Company, New York (1981).
Holmann, J. Heat Transfer, McGraw Hill, New York (2003).
Incropera, F. and DeWitt, D. Introduction to Heat Transfer, Fourth edition, John Wiley and Sons, New York (2002).
Khartchenko, V. Advanced Energy Systems, Second edition, C&R Press, Boca Raton, FL (2014).
Lamarsh, J.R. Introduction to Nuclear Reactor Theory, Second printing, Addison-Wesley Publishing Company, Inc., Reading,
MA (1972).
Lamarsh, J.R. and Baratta, A.J. Introduction to Nuclear Engineering, Prentice Hall, Upper Saddle River, NJ (2001).
Masterson, R.E. Introduction to Nuclear Reactor Physics, CRC Press, Boca Raton, FL (2017).
Masterson, R.E. Nuclear Engineering Fundamentals: A Practical Perspective, CRC Press, Boca Raton, FL (2017).
Rajput, R. Heat and Mass Transfer, S. Chand & Company, New Delhi, India (2000).
Reactor Theory (Nuclear Parameters), DOE-HDBK-1019/2-93.
Shultis, K. and Faw, R. Fundamentals of Nuclear Science and Engineering, Second edition, C&R Press, Boca Raton, FL (2008).
Thomas, L. Heat Transfer, Prentice Hall, Englewood Cliffs, NJ (1992).
Todreas, N. and Kazimi, M.S. Nuclear Systems, Vol. 1, CRC Press, Boca Raton, FL (2008).
Volk, M. Pump Characteristics and Applications, Third edition, C&R Press, Boca Raton, FL (2014).
Questions for the Student 1285
Web References
http://www4.ncsu.edu/~doster/NE405/Manuals/PWR_Manual.pdf.
http://www.iaea.org/inis/collection/NCLCollectionStore/_Public/28/030/28030311.pdf.
Other References
Ayyaswamy, P.S., Chung, J.N., and Niyogi, K.K., “Reactor containment heat removal by passive heat sink f ollowing a LOCA”, Nuclear
Technology, 33:243–247, 1977.
Berman, M. “A critique of three methodologies for estimating the probability of containment failure due to steam explosions”, Nuclear
Science and Engineering, 96:173–191, 1987.
Braun, W.K., et al. “The reactor containment of standard design German pressurized water reactors”, Nuclear Technology, 72:268–290,
1980.
Burnell, J.C. “Flow of boiling water through nozzles, orifices and pipes”, Engineering, 164:572, 1947.
Gafuku, A., et al. “Diagnostic-techniques of a small break loss-of-coolant accident at a pressurized water reactor plant”, Nuclear
Technology, 81:313–332, 1988.
International Atomic Energy Agency “Design of reactor containment systems in nuclear power plants”, IAEA-SG-D 12 (1983).
Linn, J.D.M., Maskell, S.J., and Patrick, M.A., “A note on heat and mass transfer to a spray droplet”, Nuclear Technology, 81:122–125,
1988.
Moody, F. “Maximum flow rate of single component, two phase mixture”, Journal of Heat Transfer, 87:134, 1965.
Murata, K.K., Williams, D.C., Tills, J., Griffith, R.O., Gido, R.G., Tadios, E.L., Davis, F.J., Martinez, G.M. and Washington, K.E., Code
manual for CONTAIN 2.0, A Computer Code for Nuclear Reactor Containment Analysis, NUREG/CR-6533, SAND97-1735,
1997.
Peterson, C.E., Chexal, V.K., and Clements, T.B. “Analysis of a hot-leg small break loss-of-coolant accident in a three-loop Westinghouse
pressurized water reactor plant”, Nuclear Technology, 70:104–110, 1985.
Tanaka, M. “Heat transfer of a spray droplet in nuclear reactor containment”, Nuclear Technology, 47:268–273, 1980.
Whitley, R.H., Chan, C.K., and Okrent, D., “On the analysis of containment heat transfer following a LOCA”, Annals Nuclear Energy,
3:515–525, 1976.
Westinghouse Electric Corp., “Pressurized water reactor plant”, U.S.A. Feb. (1979).
22. In addition to nuclear energy sources, name two additional sources of thermal energy produced by chemical
reactions during a LOCA.
23. What does the term “corium” refer to, and when can this substance be produced?
24. If the resulting corium interacts with the concrete floor of the containment building, what normally happens to the
concrete? Is this an example of an endothermic or exothermic reaction?
25. How does the thermal energy released from this reaction compare to the thermal energy released from the
zirconium–water reaction?
26. Taken as a whole, what fraction of the thermal energy produced by a LOCA is nuclear in origin, and what fraction
of this energy is chemical in origin?
27. What law of thermodynamics can be used to estimate the temperatures and pressures of the air–steam mixture in
the containment at the conclusion of a large-break LOCA?
28. What corresponding gas law can be used to estimate the temperatures and pressures of the air–steam mixture in the
containment at the conclusion of a large-break LOCA?
29. How is Dalton’s law of partial pressures related to Amagat’s law of additive volumes?
30. What is the definition of the relative humidity, and how can it be used with each of these laws?
31. Name at least one containment analysis code and two severe accident analysis codes.
32. Which one of these codes has been approved by the NRC for the licensing of commercial nuclear power plants in
the United States?
33. How many fluid and temperature fields do these codes use? How is the transport of particulate matter through
the containment building modeled by these codes? Are these codes based on the same methodologies that were
presented in Chapter 33?
34. In the United States, numerous regulatory documents have been developed to specify how containment buildings
are expected to be constructed and behave. What is the number of the regulatory guide the U.S. NRC uses for this
purpose? What is its exact name?
Exercise 34.2
Humid air in a containment building has a temperature of 40°C. What is the vapor pressure of the water in the air? If the
relative humidity is 50%, what is the partial pressure of the water vapor in the containment building?
Exercise 34.3
Dry air in the same containment building has a pressure of 0.2 MPa at 40°C. What is the pressure of the combined
m ixture when the relative humidity is 50%?
Exercise 34.4
A BWR suppression pool contains 5,000 m3 of water at 20°C and 1 atm. During the blowdown stage of a LOCA, the
average water temperature rises to 70°C. How much thermal energy is deposited by the LOCA in the pool?
Exercise 34.5
Saturated steam at 1 atm condenses on a flat vertical plate that is maintained at a temperature of 90°C in a reactor heat
exchanger. The plate has a surface area of 1 m2. Determine the rate of heat transfer by condensation to the plate and then
the rate at which the condensate drips off the lower edge of the plate. Assume the Reynolds number of the condensate
flowing over the plate is 2,000.
Exercises for the Student 1287
Exercise 34.6
Including the pressure vessel, the steam generators, and other reactor components, a containment building has an
internal volume of 70,000 m3. Following a LOCA, the final temperature of the air–steam mixture is 450°K (177°C).
The mass of the water vapor in the air is 70,000 kg, and the mass of the air itself is 70,000 kg. What is the pressure
of this mixture?
Exercise 34.7
Calculate the thermal energy produced by the zirconium–water reaction in a Westinghouse AP-600 following a Class
IV LOCA. Assume the core normally produces 2,000 MW of thermal energy and that 30% of the fuel assemblies in the
core are completely oxidized.
Exercise 34.8
A containment building roof has a surface area of 2,000 m2. The roof is maintained at a temperature of 30°C using a
passive cooling system. Following a LOCA, a mixture of hot air and steam begins to condense on the roof and then falls
to the floor of the containment building where it is collected in a storage area called a sump. Calculate the rate that the
mixture condenses on the roof. Assume the roof is exposed to a steam/air flow of 200 kg/s, and approximately half of
this flow is steam.
Exercise 34.9
Calculate the specific enthalpy of an air–steam mixture consisting of equal parts air and steam at 200°C and 0.2 MPa.
Suppose the mixture is then mixed with an equal amount of water from a water storage tank at 20°C. What is the final
temperature of the mixture?
Exercise 34.10
Assume the temperature of the steam in a reactor containment building is 180°C immediately following a LOCA,
but after the overheat sprayers come online; it subsides to 90°C in a couple of minutes. What is the heat transfer
coefficient on the containment building walls before and after the pressure peak? Assume the temperature of the walls
is maintained at 60°C.
Appendix A: Unit Systems
and Conversion Factors
A.1 Common Unit Systems
Today, there are two common systems of units in the world—the SI unit system and the English unit system. The SI unit system is a
more modern unit system based on the metric unit system. Conversions within the SI unit system are normally measured in increments
of 10. The SI unit system is now used throughout most of the civilized world. The one exception to this rule is the United States, where
a combination of English and SI units is currently used. Eventually, the United States is expected to convert entirely to the SI unit sys-
tem. However, this may not be completed for another 20 or 30 years. The reactor vendors in the United States use a combination of SI
and English units. In fact, it is not uncommon for a reactor vendor to use metric or SI units, CGS units (which are commonly used in
the field of nuclear physics), and English units for other important parameters such as temperature, pressure, and length measurement.
Sometimes this causes confusion outside of the United States, but because so much of the current U.S. nuclear infrastructure is
based on the English unit system, it is expected to be a long time before this transition can be complete. For this reason, conver-
sion tables between the SI or metric unit system, the CGS unit system, and the English unit system are presented in this appendix.
However, as the reader may have already gathered, this book is based primarily on the SI unit system. The units of length, mass, time,
and temperature, in addition to the units of force, pressure, energy, power, and heat, are compared to each other in Table A.1. In this
table, a total of nine different types of units are compared.
TABLE A.1
A Comparison of the Primary Units in Each Unit System
Quantity SI Unit SI Symbol CGS Unit CGS Symbol English Unit English Symbol
Length meter m centimeter cm inch, foot in, ft
Mass kilogram kg gram g or gm pound lb
Time second s second sec second s
Force newton N dyne dyne pound lb
Temperature kelvin °K celsius °C fahrenheit °F
Pressure pascal Pa dyne/cm2 dyne/cm2 pound/in2 PSI
Energy joule J erg erg foot pound ft lb
Power watt W erg/sec erg/sec foot pound/sec ft lb/s
Heat joule J calorie cal British thermal unit BTU
1289
1290 APPENDIX A
TABLE A.2
The Prefixes Associated with the SI Unit System
TABLE A.3
The Primary Units in the SI Unit System
TABLE A.4
Derived Units in the SI Unit System and Conversion Factors between the SI Unit System and Other Unit Systems
TABLE A.5
The World’s Primary Temperature Scales and Their Values
Power Output
3,000–4,500 MW(T)
1,000–1,500 MWe
Average power density: ~110 kW/L
Average thermal efficiency: 33%–36%
* For commercial nuclear plants. Note: The design parameters for research reactors can be very different.
1293
1294 APPENDIX B
TABLE B.1
The Coolant Temperatures and Enthalpies in a Typical PWR Core at 2,250 PSI or 15.5 MPa. Note: At these pressures, the
inlet temperatures tend to be between 290 °C and 300 °C, and the core outlet temperatures tend to be between 325 °C and
335 °C. These numbers can vary slightly depending upon the actual design.)
Tin = 300°C Tavg = 315°C Tout = 330°C TSAT = 344.7°C
hin = 1,338 kJ/kg havg = 1,423 kJ/kg hout = 1,517 kJ/kg
hl(@TSAT) = 1,630 kJ/kg hv(@TSAT) = 2,596 kJ/kg hlv(@TSAT) = 966 kJ/kg
APPENDIX B 1295
TABLE B.2
Coolant Properties in the Core of a Typical PWR Evaluated at 15.5 MPa
Tin = 300°C Tavg = 315°C Tout = 330°C TSAT = 344.7°C
ρ(@Tin) = 726.5 kg/m 3 ρ(@Tavg) = 692.9 kg/m 3 ρ(@Tout) = 651.5 kg/m 3 ρ(@TSAT) = 594.8 kg/m3
h(@Tin) = 1,337.6 kJ/kg h(@Tavg) = 1,422.7 kJ/kg h(@Tout) = 1,517.1 kJ/kg h(@TSAT) = 1,629 kJ/kg
cp(@Tin) = 5.46 kJ/kg °C cp(@Tavg) = 5.93 kJ/kg °C cp(@Tout) = 6.76 kJ/kg °C cp(@TSAT) = 8.93 kJ/kg °C
k(@Tin) = 0.56 W/m °C k(@Tavg) = 0.53 W/m °C k(@Tout) = 0.50 W/m °C k(@TSAT) = 0.46 W/m °C
υ(@Tin) = 0.00137 m /kg
3 υ(@Tavg) = 0.00144 m /kg
3 υ(@Tout) = 0.00153 m /kg
3 υ(@TSAT) = 0.00168 m3/kg
μ(@Tin) = 8.85 × 10 Pa s
−5 μ(@Tavg) = 8.25 × 10 Pa s
−5 μ(@Tout) = 7.61 × 10 Pa s
−5 μ(@TSAT) = 6.84 × 10−5 Pa s
ν(@Tin) = 0.12 × 10−6 m2/s ν(@Tavg) = 0.12 × 10−6 m2/s ν(@Tout) = 0.12 × 10−6 m2/s ν(@TSAT) = 0.11 × 10−6 m2/s
1296 APPENDIX B
Power Output
3,000–4,500 MW(T)
1,000–1,500 MWe
Average power density: ~52.5 kW/L
Average thermal efficiency: 32%–35% (the General Electric average is 33%)
TABLE B.3
Coolant properties at the core inlet, the core outlet, and the core mid-plane
Tin = 278°C Tout = 288°C TSAT = 288°C
hin = 1,228 kJ/kg hout = 1,452 kJ/kg* hl(@TSAT) = 1,279 kJ/kg hv(@TSAT) = 2,770 kJ/kg hlv(@TSAT) = 1,491 kJ/kg
* At the outlet, the equilibrium quality in this example is xeq = 12%
TABLE B.4
Equilibrium qualities at the core inlet, the core mid-plane, and the core outlet. Note: The equilibrium
quality at the core mid-plane is slightly less than the core average
Tin = 278°C Tout = 288°C TSAT = 288°C
xeq(@z = zin) = −0.03 xeq(@z = zmid) = 0.04 xeq(@z = zout) = 0.12
xeq = (h − hl)/hlv
1298 APPENDIX B
Fuel Assembly Flow Rates, Hot Channel Factors, and the CPR
Mass flow rate per assembly: 15.5 kg/s (typical core average)
Typical fuel assembly flow area: 0.0097 m2
Typical subchannel flow rates
Interior channels: A = 0.175 kg/s
Edge channels: A = 0.135 kg/s
Corner channels: A = 0.092 kg/s
Typical nuclear hot channel factors: F TOTAL = 2.4 − 2.8 (F TOTAL = FRADIAL × FAXIAL)
Critical power ratio (CPR) (for design transients): CPR > 1.2
CPR (for normal operation): CPR ≅ 1.38
TABLE B.5
Coolant Properties Evaluated at P = 7.25 MPa
Steam Generators
BWRs have no steam generators
https://www.nrc.gov/docs/ML0715/ML071580895.pdf
http://www.peacesoftware.de/einigewerte/wasser_dampf_e.html
https://www.tlv.com/global/TI/calculator/steam-table-pressure.html
TABLE C.1
Properties of Water and Steam
Light water
Saturation Density Density Specific Enthalpy Enthalpy Enthalpy of Thermal
Pressure Temperature (Liquid) (Vapor) Heat (J/ (Liquid) (Vapor) Vaporization Conductivity
(kPa) (°C) (kg/m3) (kg/m3) kg °K) (kJ/kg) (kJ/kg) (kJ/kg) (W/m °K)
100 100 958 0.59 4,217 417.436 2,674.95 2,257.51 0.025
7,000 285.8 740 36.63 5,400 1,267.44 2,772.57 1,505.13 0.572
7,250 288.2 735 36.92 5,450 1,267.44 2,772.57 1,505.13 0.568
15,500 344.8 102 101.93 9,200 1,629.85 2,596.22 966.366 0.121
For light water and liquid sodium, the specific heat, the thermal conductivity, and the dynamic viscosity are those of the liquid phase. Here, 100 kPa
is approximately the same as atmospheric pressure (1 bar or 14.7 PSI), 7,000 kPa is the typical pressure on the secondary side of the nuclear steam
supply system, 7,250 kPa is the typical pressure in a commercial BWR core, and 15,500 kPa is the typical pressure in a commercial PWR core.
TABLE C.2
Prandtl Numbers and Volumetric Expansion Coefficients for the Coolants listed in Table C.1
Light Water
Volumetric Specific
Saturation Expansion Specific Volume
Pressure Temperature Prandtl Number Prandtl Number Coefficient Volume (Liquid) (Vapor) Surface
(kPa) (°C) (Liquid) (Vapor) (1/°K) (m3/kg) (m3/kg) Tension (N/m)
100 100 1.75 1.02 0.0029 0.00104315 1.69402 0.0589
7,000 285.8 0.862 1.61 0.00275 0.00135186 0.0273796 0.0176
7,250 288.2 0.867 1.63 0.00283 0.00135186 0.0273796 0.0171
15,500 344.8 1.36 2.70 0.0185 0.00168249 0.00981114 0.0047
For light water, the volumetric expansion coefficients are those of the liquid phase. For liquid sodium, the expansion coefficient β is essentially zero.
1301
1302 APPENDIX C
TABLE C.3
Properties of Liquid Sodium
TABLE C.4
Properties for Coolants used in Gas Reactors—Helium
TABLE C.5
Properties for Coolants Used in Gas Reactors—Carbon dioxide
http://webbook.nist.gov
https://www.irc.wisc.edu/properties/
http://thermodynamik.hszg.de/fpc/index.php
https://www.steamtablesonline.com/steam97web.aspx
https://www.tlv.com/global/TI/calculator/steam-table-pressure.html
Other sources can be found in classical heat transfer and fluid flow books.
TABLE D.1
Some Commonly Used Dimensionless Numbers and Their Physical Significance
1305
1306 APPENDIX D
TABLE D.1 (Continued )
Some Commonly Used Dimensionless Numbers and Their Physical Significance
The applications of these numbers to nuclear power plants are discussed in Chapters 11–25.
Appendix E: Properties of
Nuclear Fuel Assemblies
E.1 Types of Fuel Assemblies
Nuclear fuel assemblies can have many different sizes and shapes, but about 90% of the fuel assemblies in the world today are either
square or hexagonal in shape. Some reactors such as CANada Deuterium Uranium reactors (or CANDU reactors) use circular fuel
assemblies, but they are considerably less common than either rectangular or hexagonal ones. In square fuel assemblies, the unit
cells can be either square or rectangular in shape. In hexagonal fuel assemblies, the unit calls can be either triangular or hexagonal
in shape. A hexagonal unit cell can then be constructed from six triangular unit cells. Pictures of these fuel assemblies can be found
in Chapters 1–5, 11, 17–18 and in many other sections of this book.
E.3 Formulas for the Flow Area, the Wetted Perimeter, and the Equivalent Diameter
For square fuel assemblies, the area of an interior coolant channel is given by A = P2 − πD2/4, the wetted perimeter is given by
PW = πD, and the equivalent diameter is given by De = 4A/PW = (4P2 − πD2)/πD. For hexagonal fuel assemblies, the area of an interior
coolant channel is given by A = 0.433P2 − πD2/8, the wetted perimeter is given by PW = πD/2, and the equivalent diameter is given by
De = 4A/PW = (0.866P2 − πD2/4)/πD. These results are summarized in Table E.1. Values of these parameters for modern PWRs and
BWR fuel assemblies are also presented in Appendix B. These values can be used to determine the mass flux, the flow rate, and the
Reynolds number of the subchannels.
TABLE E.1
Lattice Parameters for Square and Hexagonal Fuel Assemblies
1307
1308 APPENDIX E
in shape. These fuel pellets are normally constructed from UO2 or PuO2. The melting points for UO2 and PuO2 are very high (about
2,850°C and 2,390°C, respectively), and these extreme temperatures are almost never reached during normal reactor operation. The
density of UO2 is about 10.97 g/cm3, and the density of PuO2 is about 11.45 g/cm3. Other physical characteristics of these important
nuclear materials are discussed in Chapter 11. Since Y2K, about 95% of the reactors in the world use either uranium dioxide or plu-
tonium dioxide for their fuel.
Appendix F: Properties
of Subcooled Water
F.1 Subcooled Water
Subcooled water is defined as water that has not yet reached its saturation temperature. Subcooled water is used in the primary loops
of Westinghouse and AREVA pressurized water reactors (PWRs). It is the primary form of the coolant in most of the PWRs in the
world today. The properties of subcooled water are well known and well understood. They are listed here for temperatures below the
saturation temperature TSAT, which for the primary loop of most PWRs in the Western world is 345°C at 15.5 MPa or 2,250 PSI. Other
properties can be found in classical fluid flow and heat transfer books. In general, the properties of subcooled water are temperature
and pressure dependent. Where possible, the Prandtl numbers and volumetric expansion coefficients are also shown.
F.1.1 Properties of Subcooled Water for Westinghouse and AREVA PWRs at 15.5 MPa
Properties of subcooled water for the primary loops of Westinghouse and AREVA PWRs are presented in Table F.1 below. The
subcooled water properties shown in this table are the same as those of light water, where there are 6,420 ordinary water molecules
for each molecule of heavy water. A detailed discussion of the difference between light water and heavy water is presented in our
companion book*. Heavy water is about 10% heavier than ordinary water and has a boiling point that is 1.4% higher. At 15.5 MPa or
2,250 PSI, the saturation temperature of light water is 344.5 °C or 652 °F.
TABLE F.1
Properties of Subcooled Water at 15.5 MPa (2,250 PSI)
Light Water
Specific
Dynamic Kinematic Internal Specific Specific Specific Heat Specific Heat Thermal
Temperature Density Viscosity Viscosity Energy Enthalpy Entropy Capacity, Cp Capacity, Cv Conductivity
(°C) (kg/m3) (10−5kg/m s) (10−6 m2 s) (kJ/kg) (kJ/kg) (kJ/kg °K) (kJ/kg °K) (kJ/kg °K) (W/m °K)
280 764.3 9.64 0.126 1,212.3 1,232.6 3.04 5.07 3.07 0.596
285 755.4 9.44 0.125 1,237.7 1,258.2 3.08 5.15 3.06 0.588
290 746.2 9.24 0.124 1,263.4 1,284.2 3.13 5.24 3.05 0.589
295 736.6 9.05 0.123 1,289.6 1,310.6 3.18 5.34 3.04 0.572
300 726.5 8.85 0.122 1,316.3 1,337.6 3.23 5.46 3.03 0.562
305 715.9 8.65 0.121 1,343.6 1,365.2 3.27 5.59 3.03 0.553
310 704.8 8.45 0.120 1,371.6 1,393.6 3.32 5.74 3.03 0.543
315 692.9 8.26 0.119 1,400.4 1,422.7 3.37 5.92 3.03 0.532
320 680.2 8.05 0.118 1,430.1 1,452.9 3.42 6.14 3.03 0.521
325 666.5 7.84 0.117 1,461.0 1,484.2 3.47 6.41 3.03 0.510
330 651.5 7.61 0.116 1,493.3 1,517.1 3.53 6.76 3.03 0.498
335 634.9 7.38 0.116 1,527.6 1,552.0 3.59 7.22 3.04 0.485
340 616.0 7.11 0.115 1,546.5 1,589.7 3.65 7.89 3.06 0.472
344.5a 595.8 6.85 0.115 1,601.2 1,627.2 3.71 8.88 3.11 0.459
a At 15.5 MPa or 2,250 PSI, the saturation temperature for light water is 344.5°C. Here, 15,500 kPa or15.5 MPa is the typical pressure in a
c ommercial PWR core.
* See “An Introduction to Nuclear Reactor Physics” by Robert E. Masterson, CRC Press (2018).
1309
1310 APPENDIX F
TABLE F.2
Properties of Subcooled Water at 7.25 MPa (1,050 PSI)
Light Water
Specific
Dynamic Kinematic Internal Specific Specific Specific Heat Specific Heat Thermal
Temperature Density Viscosity Viscosity Energy Enthalpy Entropy Capacity, Cp Capacity, Cv Conductivity
(°C) (kg/m3) (kg/m s) (10−6 m2 s) (kJ/kg) (kJ/kg) (kJ/kg °K) (kJ/kg °K) (kJ/kg °K) (W/m °K)
250 802.6 0.00011 0.133 1,076.6 1,085.6 2.78 4.82 3.15 0.625
255 794.9 0.00010 0.132 1,100.8 1,109.9 2.83 4.88 3.14 0.619
260 786.9 0.00010 0.130 1,125.2 1,134.4 2.87 4.94 3.12 0.613
265 778.6 0.00010 0.129 1,150.0 1,159.3 2.92 5.01 3.11 0.606
270 770.0 0.00010 0.127 1,175.1 1,184.5 2.97 5.08 3.10 0.599
275 761.0 0.00010 0.126 1,200.6 1,210.1 3.02 5.17 3.09 0.591
280 751.7 0.00009 0.125 1,226.5 1,236.2 3.06 5.26 3.08 0.583
285 741.9 0.00009 0.124 1,253.0 1,262.8 3.11 5.37 3.07 0.574
288a 735.7 0.00009 0.123 1,269.2 1,279.0 3.14 5.45 3.06 0.568
a At 7.25 MPa or 1,050 PSI, the saturation temperature for light water is 288 °C. Here, 7,250 kPa or 7.25 MPa is the typical pressure in a c ommercial
BWR core.
TABLE F.3
Properties of Subcooled Water at 6 MPa (870 PSI)
Light Water
Specific
Dynamic Kinematic Internal Specific Specific Specific Heat Specific Heat Thermal
Temperature Density Viscosity Viscosity Energy Enthalpy Entropy Capacity, Cp Capacity, Cv Conductivity
(°C) (kg/m3) (kg/m s) (10−6 m2 s) (kJ/kg) (kJ/kg) (kJ/kg °K) (kJ/kg °K) (kJ/kg °K) (W/m °K)
240 816.1 0.00011 0.137 1,030.4 1,037.8 2.70 4.74 3.18 0.635
245 808.8 0.00010 0.135 1,054.2 1,061.6 2.74 4.78 3.16 0.630
250 801.2 0.00010 0.133 1,078.2 1,085.7 2.79 4.84 3.15 0.624
255 793.4 0.00010 0.131 1,102.4 1,110.0 2.83 4.89 3.13 0.618
260 785.3 0.00010 0.130 1,127.0 1,134.6 2.88 4.96 3.12 0.611
265 776.9 0.00010 0.129 1,151.8 1,159.6 2.92 5.03 3.11 0.604
270 768.2 0.00010 0.127 1,177.1 1,184.9 2.97 5.11 3.10 0.597
275.5 758.1 0.00009 0.125 1,205.3 1,213.3 3.03 5.20 3.09 0.588
a At 6 MPa or 870 PSI, the saturation temperature for light water is 275.5 °C. Here, 6,000 kPa or 6 MPa is the typical pressure in the secondary loop
of a commercial PWR (where the steam generators are located). The actual pressures in the secondary loop can vary from about 5.7 to 7.6 MPa—
depending on the actual design.
APPENDIX F 1311
TABLE F.4
Properties of Subcooled Water at 101.3 kPa (14.7 PSI)
Light Water
Specific
Dynamic Kinematic Internal Specific Specific Specific Heat Specific Heat Thermal
Temperaturea Density Viscosity Viscosity Energy Enthalpy Entropy Capacity, Cp Capacity, Cv Conductivity
(°C) (kg/m3) (kg/m s) (10−6 m2 s) (kJ/kg) (kJ/kg) (kJ/kg °K) (kJ/kg °K) (kJ/kg °K) (W/m °K)
10 999.7 0.00130 1.306 42.02 42.12 0.151 4.20 4.19 0.580
20 998.2 0.00100 1.003 83.91 84.01 0.296 4.18 4.16 0.598
30 995.7 0.00080 0.800 125.73 125.83 0.437 4.18 4.12 0.615
40 992.2 0.00065 0.658 167.52 167.62 0.572 4.18 4.07 0.630
50 988.0 0.00054 0.553 209.31 209.41 0.708 4.18 4.04 0.643
60 983.2 0.00046 0.474 251.12 251.22 0.831 4.18 3.97 0.654
70 977.8 0.00040 0.413 292.97 292.07 0.955 4.18 3.92 0.663
80 971.8 0.00035 0.365 334.88 334.99 1.075 4.19 3.87 0.670
90 965.3 0.00031 0.325 376.88 376.99 1.192 4.20 3.82 0.675
99.9 958.4 0.00028 0.294 418.57 418.67 1.306 4.21 3.77 0.679
a At 101.3 kPa or 14.7 PSI, the saturation temperature for light water is 100 °C. The saturation temperature varies slightly depending upon whether
one is above or below sea level.
http://webbook.nist.gov
https://www.irc.wisc.edu/properties/
http://thermodynamik.hszg.de/fpc/index.php
https://www.steamtablesonline.com/steam97web.aspx
http://www.peacesoftware.de/einigewerte/wasser_dampf_e.html
https://www.tlv.com/global/TI/calculator/steam-table-pressure.html
Other sources of this data can be found in classical heat transfer and fluid flow books.
G.1.1 Properties of Saturated Water and Steam for PWRs and BWRs at Low Pressures
The properties of saturated water and steam are presented in Table G.1 for pressures between 0.001 MPa (14.5 PSI) and 1 MPa
(145 PSI). These pressures are typically encountered in low pressure nuclear applications.
TABLE G.1
Properties of Saturated Water and Steam below 1 MPa (1 kPa–1 MPa)
1313
1314 APPENDIX G
TABLE G.1(Continued)
Properties of Saturated Water and Steam below 1 MPa (1 kPa–1 MPa)
G.1.2 Properties of Saturated Water and Steam for PWRs and BWRs at High Pressures
The properties of saturated water and steam are presented in Table G.2 for pressures between 1 MPa (145 PSI) and 20 MPa (2,900
PSI). These pressures are typically encountered in high pressure applications, and pressures between 7 MPa (1,050 PSI) and 15.5 MPa
(2,250 PSI) are typically be found in BWR and PWR cores.
TABLE G.2
Properties of Saturated Water and Steam above 1 MPa (1–22 MPa)
These relationships are based on the assumption of phase equilibrium. Sometimes the subscript “mix” is dispensed with entirely in
which case the equations for the properties of the mixture become
G.2.1 Densities and Specific Volumes of Saturated Water and Steam for PWRs and BWRs
These relationships can be used to define the properties of any mixture where the phases are in thermal equilibrium and are moving
at the same speed. That is, they are valid for any mixture where the slip ratio is 1.0. Many useful properties for saturated water and
steam are presented in Table G.3 for pressures between 0.001 MPa and 22 MPa. These pressures include virtually all the pressures
that are encountered during routine reactor operation.
TABLE G.3
Densities and Specific Volumes
ρ = 1/υ (G.9)
−1
ρ = 1 x ρv + (1 – x) ρl = x ρv + (1 – x) ρl (G.11)
Hence, the properties of any saturated two-phase mixture can be determined in this way. Similar relationships can be used to
d etermine the dynamic viscosity, the kinematic viscosity, and the thermal conductivity of the mixture. Notice that the density of the
mixture can also be determined directly from the void fraction which leads us to the alternative definition for ρ:
Equating Equations G.11 and G.12 then leads to the following relationship between α and x:
( )
(x/α) 1 ρv = (1 – x) (1 − α) 1 ρl ( ) (G.13)
or
(ρl )
ρv = (1 – x) x α (1 − α) (G.14)
which is frequently encountered in the study of two-phase mixtures. Hence, Equation G.14 establishes a unique relationship between
x and α for a nonslip system. The effect of slip on Equation G.14 is discussed in Chapter 23.
http://webbook.nist.gov
https://www.irc.wisc.edu/properties/
http://thermodynamik.hszg.de/fpc/index.php
https://www.steamtablesonline.com/steam97web.aspx
https://www.tlv.com/global/TI/calculator/steam-table-pressure.html
Others can be found in classical heat transfer and fluid flow books.
TABLE H.1
Properties of Superheated Steam at 4.0 MPa (250.4°C)
1317
1318 APPENDIX H
TABLE H.2
Properties of Superheated Steam at 4.5 MPa (257.4°C)
TABLE H.3
Properties of Superheated Steam at 5.0 MPa (263.9°C)
TABLE H.4
Properties of Superheated Steam at 6.0 MPa (275.6°C)
TABLE H.5
Properties of Superheated Steam at 7.0 MPa (285.8°C)
TABLE H.6
Properties of Superheated Steam at 8.0 MPa (295.0°C)
http://webbook.nist.gov
https://www.irc.wisc.edu/properties/
http://thermodynamik.hszg.de/fpc/index.php
https://www.steamtablesonline.com/steam97web.aspx
https://www.tlv.com/global/TI/calculator/steam-table-pressure.html
Other sources of this data can be found in classical heat transfer and fluid flow books.
The properties in Equation I.1 are evaluated at bulk fluid conditions (see Chapter 20). Some vendors attempt to account for the
temperature dependence of the viscosity at the wall surface by adding a viscosity correction factor, which then leads to the Sieder–
Tate correlation:
Here, the viscosity correction factor C = (μWALL/μ)0.14 is normally evaluated using the wall temperature T WALL or the average of the wall
temperature and the bulk temperature (T WALL + TBULK)/2. This correction factor changes the value of h given by the Dittus–Boelter
correlation by about 1% when the flow is forced. Both correlations have been shown to be conservative for rod bundle geometries with
P/D ratios in the range of those used by modern PWRs (1.30 < P/D < 1.35).
where:
ΔTSAT = TCLAD(z)—TSAT is the amount of wall superheat (°F)
q″(z) = wall heat flux (Btu/h ft2)
P = pressure (PSI)
TCLAD(z) = outer clad wall temperature (°F)
TSAT = saturation temperature of coolant at pressure P (°F)
1321
1322 APPENDIX I
The WRB correlation was first developed using turbulent mixing vane data obtained from the fuel assemblies used in the
AP 1000, which were subsequently renamed Robust Fuel Assemblies or “RFAs” (to differentiate them from the fuel assemblies
Westinghouse used previously). The WRB correlation is applicable to the heated rod spans above the first mixing vane grid,
and it predicts a less conservative DNB ratio (DNBR) limit than the W-3 Correlation discussed in Chapter 27. Its stated range of
applicability range is
More recently, Westinghouse released a revised version of this correlation called the WRB-2 correlation, which is also used to find the
DNB limit in their RFA assemblies. Using this correlation, Westinghouse has shown that relatively large non-uniformities in the flow
distributions at the core inlet (in the range of 5%–10%) have very little effect on the value of the DNBR in the hottest fuel assembly.
Other factors that affect the calculation of the DNBR are discussed below.
1325
1326 APPENDIX J
TABLE J.1
Core Design Data
Thermal and Hydraulic Design Parameters AP1000 AP600 Cores with XL Fuel
Core Power Output
Core power output (MWT) 3,400 1,930 3,800
Core thermal power output (1 × 106 BTU/h) 11,600 6,600 13,000
Percent of heat deposited directly in the fuel 97.4 97.4 97.4
Coolant Temperatures
Nominal inlet temperature (°F) 535.0 532.8 561.2
Average temperature rise in pressure vessel (°F) 77.2 69.6 63.6
Average temperature rise in core (°F) 81.4 75.8 68.7
Average temperature in core (°F) 578.1 572.6 597.8
Average temperature in pressure vessel (°F) 573.6 567.6 593.0
TABLE J.2
Comparison of Different Fuel Assembly Designs
Thermal and Hydraulic Design Parameters AP1000 AP600 Cores with XL Fuel
Fuel Assemblies
Hot channel factor for the nuclear heat flux 2.60 2.60 2.70
Peak fuel center line temperatures (°F) (to prevent fuel melt) 4,700 4,700 4,700
Fuel assembly configuration 17 × 17 XL Robust 17 × 17 17 × 17 XL Robust
Fuel Fuel/with no IFM
Number of fuel assemblies 157 145 193
(Continued)
APPENDIX J 1327
Thermal and Hydraulic Design Parameters AP1000 AP600 Cores with XL Fuel
Fuel Assemblies
Uranium dioxide rods per assembly 264 264 264
Rod pitch (in.) 0.496 0.496 0.496
Overall dimensions (in.) 8.426 × 8.426 8.426 × 8.426 8.426 × 8.426
Fuel weight, as uranium dioxide (lb) 211,588 167,360 261,000
Clad weight (lb) 43,105 35,555 63,200
Fuel Pellets
Material UO2 sintered UO2 sintered UO2 sintered
Density (% of theoretical) 95.5 95 95
Diameter (in.) 0.3225 0.3225 0.3225
Length (in.) 0.387 0.387 0.387
TABLE J.3
Comparison of Other Design Parameters
Core Structure
Core barrel, ID/OD (in.) 133.75/137.75 133.75/137.75 148.0/152.5
Thermal shield Neutron Panel None Neutron Panel
Baffle thickness (in.) Core Shroud Radial reflector 0.875
(Continued)
1328 APPENDIX J
Structural Characteristics
Core diameter, equivalent (in.) 119.7 115.0 132.7
Core height, cold, active fuel (in.) 168.0 144.0 168.0
References
1. Letter from N.J. Liparulo (Westinghouse) to J.E. Lyons (NRC), Transmittal of Response to NRC Request for Information on Wolf Creek Fuel
Design Modifications, NSD-NRC-97-5189, June 30, 1997.
2. Letter from N.J. Liparulo (Westinghouse) to R.C. Jones (NRC), Transmittal of Presentation Material for NRC/Westinghouse Fuel Design
Change Meeting on April 15, 1996, NSD-NRC-96-4964, April 22, 1996.
3. Letter from Westinghouse to NRC, Fuel Criteria Evaluation Process Notification for the 17 × 17 Robust Fuel Assembly with IFM Grid
Design, NSD-NRC-98-5796, October 13, 1998.
4. Letter from H.A. Sepp (Westinghouse) to T.E. Collins (NRC), Notification of FCEP Application for WRB-1 and WRB-2 Applicability to the
17 × 17 Modified LPD Grid Design for Robust Fuel Assembly Application, NSD-NRC-98-5618, March 25, 1998.
5. Letter from H.A. Sepp (Westinghouse) to T.E. Collins (NRC), Fuel Criteria Evaluation Process Notification for the Revised Guide Thimble
Dashpot Design for the 17 × 17 XL Robust Fuel Assembly Design, NSD-NRC-98-5722, June 23, 1998.
6. Davidson, S.L. and Kramer, W.R., (Ed.), Reference Core Report Vantage 5 Fuel Assembly, WCAP-10444-P-A (Proprietary), September 1985
and WCAP-10445-A (Non-Proprietary), December 1983.
7. Davidson, S.L. (Ed.), VANTAGE 5H Fuel Assembly, Addendum 2-A, WCAP-10444-P-A (Proprietary) and WCAP-10445-NP-A
(Non-Proprietary), February 1989.
8. Davidson, S.L. and Nuhfer, D.L., (Ed.), VANTAGE+ Fuel Assembly Reference Core Report, WCAP-12610-P-A (Proprietary) and WCAP-
14342-A (Non-Proprietary), April 1995.
9. Davidson, S.L. (Ed.), Fuel Criteria Evaluation Process, WCAP-12488-A (Proprietary) and WCAP-14204-A (Non-Proprietary), October 1994.
10. NTD-NRC-94-4275 Westinghouse’s Interpretation of Staff’s Position on Extended Burnup, August 29, 1994.
Index
Abnormal reactor states, 1231 Angular momentum, 542 Average temperature differences, 300
Absolute pressures, 199, 200 Angular velocity vector, 539, 541 Average values of various parameters, 1021
Absolute temperature, 366, 402, 853, 861, 950, Annular/film boiling regime, 85 Avogadro’s number, 1114
1111, 1114 Annular flow regimes, 904, 965, 971 AWK scripts, 1160
Absolute viscosity, 486, 487, 559, 647, 748 Annular fuel pins Axial blanket, 100
ABWR, See Advanced boiling water reactors average temperature, 439–441 Axial coolant temperature, 1005–1006
(ABWR) thermal conductivity, 390 Axial flow pumps, 761
Acceleration pressure drop, 616, 921 Annular fuel rods, 389, 427, 448–452, 456 Axial heat flux, 1008–1009
Accumulators, 1151, 1155, 1157 Annular tube heat exchangers, 293 Axial heat generation rate, 452–454
Actinides, 167, 168, 171 ANS, See American Nuclear Society (ANS) Axial mass flow rate, 1029
Active cooling system, 869 ANS-5.1–1994, 173 Axial power peaks, 153–154
Active core height, 793, 962 ANSI/ANS 56.4: Pressure and Temperature Axial power profiles, 156–157
Active fuel height, 443 Transient Analysis for Light Axial power shapes, 153–154
Active injection systems, 1151 Reactor Containments (1988), 1254 Axial temperature profile, 452
Active length, 1309 ANSI/ANS Standards 51.1, 1234 Axial zoning schemes, 154–155
Active safety systems, 869 ANSI N18.2a-75 requirements document, Azimuthal angle, 405
Active subsystems, 1154, 1155 1233
Actual heat flux, 1045 ANSI standards, 1234 Backfilling process, 1241
Actual vs. idealized Rankine cycles, 340–341 ANS 56.4: Pressure and Temperature Back pressure, 1117, 1118, 1123
Adiabatic flow channels, 1131 Transient Analysis for Light Backup diesel generators, 850
Adiabatic processes, 341 Reactor Containments (1988), Baffles used in steam generators, 294, 307
Advanced boiling water reactors (ABWR), 1286 Baker–Just equation, 1163, 1164
78–79 ANS safety standards, 1234 Baroczy correlation, 917
Advanced gas reactors (AGRs), 96–97, 407, Anticipated operating states range, 1231 Barometers, 523
411, 739 Antoine chart, 501 Base loaded units, 284
Advection process, 353, 564, 1075–1077 Antoine equation, vapor pressure, 264, 500 Base-loading model, 1
diffusion vs. convection, 533–534 AP 1002 design parameters, 1327–1330 Base state, 1231
of momentum, 577–578 Applied heat flux, 1045 Beattie–Bridgeman equation of state,
Advective acceleration, 536 APRM, See Average power range monitors 225, 226
Advective derivative (APRM) Bell Coleman cycle, 341
conservation equations of fluid mechanics, Aqueous coolants, 735 Benedict–Webb–Rubin equation of state, 225,
555 Area change, 522, 614, 628 226
observations, 536–537 AREVA EPR, 60 BEP, See Best efficiency point (BEP)
space-dependent term, 535 AREVA PWR, 41, 48, 276, 1152 Berensen correlation, 950, 951, 953, 957
Advective–diffusion equation, 1084 Armatures, 766 Bernath correlation, 974–975, 1048–1049
Aerodynamic diameter, 1090 ASME, 1235 Bernoulli effect, 751
Aerosol, 1089–1090 Asymmetric pressurization, 1282 Bernoulli head, incompressible fluid, 757
forces on particles, 1091–1092 Atmospheric dispersion, wedge model of, Bernoulli’s equation, 587, 591, 921, 1114
Agencywide Documents Access and 1096–1097 flow blockage, 616–617
Management System (ADAMS), Atmospheric field, 1282 laminar flows, 637
1249 Atmospheric pressure, 1271 incompressible flows, 609, 637
AGRs, See Advanced gas reactors (AGRs) Atmospheric turbulence, 1094 streamline flows, 637–638
Air refrigeration cycle, 341 Attack angle, 763 loss coefficients, 698
Air–steam mixture, 1151, 1152, 1264, 1266, Attenuation coefficients, 130–131, 369–370 with loss coefficients and friction factors,
1268 Atwood number, 1140 636–637
Air–water flow, 885 Automobile radiators, 304 to nuclear science and engineering,
Air–water mixtures, 245–247, 885, 1271 Auxiliary building, 284 612–613
Alpha decay, 128–129 Auxiliary pumps, 953 to pipes and other simple structures,
Alpha particles, 128 Available head, 759 613–616
Amagat’s law of additive volumes, 1266 Average approach velocity, 840 practical applications of, 637
Ambient pressure, 282, 630 Average axial velocity, 596 and principle, 612
Ambient temperature, 356, 472, 478 Average drag coefficient, 601, 725, 729 start-up times for RCP, 752–755
American boiling water reactors, types of, Average fuel pin temperatures, 439–441, 443 from estimating, 633–636
70–72 Average power range monitors (APRM), 1176, to steady-state flows, 637
American Nuclear Society (ANS), 1233–1235, 1179 time-dependent, 632
1240 Average roughness, 691 of two-phase flow, 926–928
AMUs, See Atomic mass units (AMUs) Average subchannel, 792–794 Bessel function, 158, 159, 467–470, 1003,
AndreaTek, horizontal steam generator Average temperature, 381, 472, 777, 778, 1130, 1005
examples, 278 1161 Best efficiency point (BEP), 758
1329
1330 Index
Beta decay, 128, 167 Hench–Gillis correlation, 1060–1063 fuel pin temperature profile, 445, 453
Beta particles, 128 and nuclear hot channel factor, 1070–1071 fuel rod design, 408, 456
Beta rays, 167 nuclear hot channel factors, 1068–1069 fuel temperatures vs. coolant
Bettis Atomic Research Laboratory, 104 PWR temperatures, 454
Bhatti and Shaw correlation, 493 vs. BWR, 1040 global water and steam profile, 72
Binary vapor cycles, 344 coolant and cladding temperatures in, hydrodynamic instabilities, 1170, 1174
Biological feedback loops, 365 1039–1042 instabilities, 1172–1174
Biot number DNBR limits, 1064–1065 LOCAs, 1155–1156, 1160–1161
description of, 469 dryout of fuel rods in, 1039 operating conditions, 1295
dimension for geometric shapes, 477 representative values for CHF in, operating maps, 88–91
geometrical shapes, 470 1042–1043 operating restrictions, 88
length scales, 474 and regulatory requirements, 1063–1064 power control, 74, 75
physical interpretation of, 478 terminology, 1046–1047 pressure behavior in, 528
physical significance of, 478–479 thermal design limits, 1063–1064 pressure point, 516
time-dependent temperature profiles, 473, thermal limitations on core design, pressure vessel, 15–16
477 1067–1068 vs. PWR, 1020, 1040
BISO coating for circular fuel pellets, 106 Tong correlation, 1047–1048 Rankine cycle for, 315, 316
Bjorge–Hall–Rohsenow correlation, 980 in water cooled reactors, 1039 in reheating process, 324
Blackbody curve, 366 Westinghouse W-3 correlation, reported power oscillations, 1172
Blackbody radiation, 366 1051–1052 representative operating conditions,
Black box representation of a reactor core, 332 W-3 shape correction factor, 1052–1054 1299–1301
Blade angles, 761, 767 Boiling heat transfer, 931 single-phase flow, 609
Blade design, steam, 769–770 and boiling crisis, 931 stability maps and operating maps,
Bladed wheels, steam turbines, 282 coefficient, 938 1181–1183
Blasius correlations, 666, 670–671, 748 Boiling heights, 908, 913, 1017 suppression pools, 1259–1261
Blockages, 1240–1242 Boiling length, 911, 962, 1060 temperature drops in, 412–413, 435–436
Blowdown panel, 1205 Boiling number, 972, 973 temperature profiles, 76–78, 87
Blowdown stage, 1150, 1151, 1157 Boiling point, 211, 884–885 thermal-hydraulic instabilities, 1175–1176
Boiling, 882–883 thermodynamic properties two-phase flow pressure losses in,
characteristics of, 939 Antoine equation, 264 928–929
nucleate and film, 883–884 Clausius–Clapeyron equation, void fraction and quality, 81, 83–85
pool boiling and bulk, 883 234–235, 264 density, 81–83
saturated and subcooled, 884 of reactor coolants, 264 volumetric power density, 411
temperature and pressure dependent, 882 Boiling water reactors (BWRs), 20, 69–70, Boltzmann’s constant, 259, 260, 1114
Boiling cores, 907–910 194, 199, 208, 224, 238, 247, 264, Bond number, 946
coolant temperatures internal to, 1017–1019 269, 270, 272, 275, 289, 363, 372, Borda–Carnot coefficient, 618
fuel and cladding temperatures in, 392, 393, 399, 488, 493, 501, 1145, Boric acid, 46, 64–66
1019–1020 1147, 1149, 1232, 1234, 1243 Borishanskii correlation, 832
Boiling crisis, 884, 918, 958, 985 boiling point, 884–885 Boundary conditions
Bernath correlation, 1048–1049 characteristics, 80–81 for annular fuel rod, 449
Bowring correlation, 1050–1051 cladding temperatures in, 439 for cladding, 422
BWR containment buildings, 1207–1208, 1223 flow, 599
DNBR limits, 1065–1066 controllers, 1175 fluid, 596–597
dryout of fuel rods in, 1039 control rods vs. pressurized water reactors free surface, 598
vs. PWR, 1040 control rods, 22–24 for fuel, central void, 420, 426
CHF and DNBR, 1045–1046 control system instabilities, 1174–1175 heat conduction equation, 413–415
CHF correlations, 1047 coolant pump, 751(See also Jet pumps) inlet/outlet, 598
for BWRs, 1055–1058 coupled neutronic-thermal hydraulic interface, 597–598
critical quality vs., 1066–1067 instabilities, 1177–1179 no-slip, 597
CISE-4 correlation for CFQ, 1059 data sources for reactor coolant properties, viscous effects and, 588
critical heat flux and, 1041 1301 for nuclear fuel rods, 416
critical power ratio and its uses, density waves, 1179–1181 other types of, 598
1058–1059 design parameters, 78, 1298–1299 pressure, 580, 598–599
dependence of CHF DNBR limits, 1065–1066 specification of, 461
on exit flow quality, 1043 dryout of fuel rods in, 1039 velocity, 580, 598
on grid spacers, 1043–1044 efficiency of, 329 Boundary layers, 589, 657, 779
on mass flow rate, 1043 enthalpies for, 329 approximations, 589–590
on system pressure, 1044–1045 film dryout in, 1054–1055 effects, 801
eight correction factors, 1050 flow regimes, 85–88 equations, 590
engineering uncertainty factors, Friedel correlation for two-phase flow
1070–1071 multiplier, 917–918 conservation equations, 600–601
film dryout, 1039 fuel assemblies, 16, 22, 72–74, 81, transition points in, 602
in BWRs, 1054–1055 139–140, 142, 1173 region, 590
GEXL correlation, 1059–1060 fuel design parameters, 80 solutions, 601–602
Groeneveld correlation, 1049–1050 fuel pellets, 136 thicknesses, 666–667
Index 1331
Bowring correlation, 1050–1051 fuel assembly, 143–144 materials in reactor designs, 433
Boyle’s gas law, 224, 1113 Cans, 526, 528 plate-type fuel rod without, 415
Brayton cycle, 1, 6, 25 Caorso BWR, 1171 rupture equation, 1160
Brayton thermal cycle, 19, 96, 106, 107, 611, Carnot cycle, 314 temperature drop, 413, 435–436, 442
737, 742, 1211 Carnot heat engines, 185, 215–216 temperature profile, 418, 422
description of, 314, 341 thermal efficiency of, 269, 320, 340 thermal conductivities, 436
gas reactors, 270, 272, 332 T–S diagram, 314, 318 thermal properties of, 410
ideal Brayton cycle, steps in, 341–344 Carnot thermal cycle, 215–216, 253 Cladding-coolant temperature
thermal efficiency of, 348 Cartesian coordinates, 408 after NVG Point, 990–991
Break size, LOCAs, 1149 Cartesian coordinate system, 563, 567 before NB, 987
Breeder reactors, 149 continuity equation in, 568 during subcooled NB, 988–989
Brownian motion, 1077 Category 6 LOCA, 1149 Cladding oxidation, 1150, 1163
Browning correlation, 990 CATHARE code, 1137 Cladding strain limit, 1146
Brute force, 557 Cathcart correlation, 1164 Cladding temperature, 1146, 1154, 1157–1158,
Bubble detachment point (BDP), 988, 995 Cauchy’s equations, 584 1161, 1162, 1164
Bubble diameter, 954 for compressible fluid, 584–585 Class 9 accidents, 1236
Bubble force balance, 887–889 to handle incompressible Newtonian Clausius–Clapeyron equation, 209–210,
and surface tension, 889 fluids, 585–586 234–235, 264
surface wetting, 889 Cavitation, 771–772 Clausius inequality, 187–189
Bubble formation, 511 bubbles, 771 Clean tube surface, 305
and growth, 885, 939 number, 772 Clementine reactor, 96
and superheat, 890–891 preventing cavitation in RCPs, 771–772 Closed cores, 526
Bubble growth Centerline temperature, 477–478, 1011, 1012 Closed feedwater heaters, 326
and detachment, 889–890 Centipoise, 647 Closed loop cycle, 316, 335
and formation, 885, 939 Central void, 413, 425–428, 431 Closed loop design, 288
and pool boiling, 885–886 Centrifugal coolant pumps, 761–764 Closed thermal cycle, 313
and surface effects, 886–887 Centrifugal pumps, 750 Coal-fired power plant, 19
Bubble profiles, 72, 73 Centrifugal separation process, 13 efficiency of, 329
Bubble rise terminal velocity, 900 Ceramic materials, 360 enthalpies for, 328
Bubbly flow, 971 CF, See Correction factors (CF) regeneration process for, 327
regime, 965 CFD, See Computational fluid dynamics Coal-fired power plants, 7, 30, 221, 222, 269,
with saturated boiling, 903 (CFD) 272, 274, 284
with subcooled boiling, 903 CFR 50, 1235 COBRA, 1030, 1031
Bucket-shaped blades, 770 10 CFR 50.55a, 1254 Code of Federal Regulations (CFR), 1249
Bulk boiling, 363, 528, 868 10 CFR 50, section 44, 1286 Coefficient of discharge, 1106–1108
flow regime, 903 Chain rule, 535, 788 Coefficient of proportionality, 1137
point, 987–988 Channel thermal-hydraulic instabilities, 1175, Coefficient of skin friction, 602
and pool boiling, 883 1176 Cold leg, 1151
Bulk coolant temperature, 1006, 1017 Characteristic length, 473, 474, 652, 854, Colebrook correlations, 1136
Bulk motion, 353 946 Colebrook equation, 692
Bulk nucleate boiling region, 942 Charles’s law, 1113 Columnar grain region, 425, 428–429
Bulk strain, 545 Chemical composition, of ordinary air, 1267 Combined delivery rate, 759
Bulk temperature, 381, 392, 689, 777, 778, Chen correlation, 976–978, 1019 Combined gas vapor cycles, 344
866, 884, 909 Cherenkov radiation, 168–169, 1218–1219 Commercial power reactors
and Newton’s law, 777–779 Chernobyl accident, 74 coolant pressures, 200
single-phase heat transfer correlations, Chernobyl reactor accident, 1097 thermal efficiency, 195
820 Chexal–Lellouche correlation, 901 Commercial power reactors, number of, 4
Bundle-averaged friction factors, 684–685 Chexal–Lellouche void fraction model, 901 Compendium of Federal Regulations (CFR),
Bundle correction factor, 948, 955 CHF, See Critical heat flux (CHF) 1150, 1235
Buoyancy force, 1088–1089, 1091 CHF ratio (CHFR), 1045 Composite flow channels, 667
Buoyancy forces, 850 Choke flow, 1105–1106, 1113, 1116 Composite pipes, 667
Buoyancy-induced forces, 515 Churchill–Bernstein correlation, 837 Composite thermal resistances, 377–378
Buoyancy-related effects, 851, 1140 Churn flow regime, 904 Compressibility chart, 503
Burnable poisons, 152–153, 163 Circular pipes, 672 Compressibility coefficient, 488, 501–503
Burnup function, 428–431, 434, 437, 447, vs. reactor coolant channel, 682 Compressibility factors (gases), 225, 503
1160, 1237 Circular tubes, 829 Compressible flows, 488, 1111
BWR 6 fuel assembly, 75 Circulation loop, 506 Compressible fluids, 583, 586, 1131
BWRs, See Boiling water reactors (BWRs) Circulation measures, 539 Cauchy’s equations for, 584–585
CISE-4 correlation, 1057 single-phase flow, 610, 611
Calandria, 51–53 for CFQ, 1059 treatment of fluid friction in, 583–584
Calza-Bini equation, 446–447 Cladding, 381, 383, 386, 387, 389, 392, 1222 Compression ratio, ideal Brayton cycle, 343
CANada Deuterium Uranium (CANDU) absolute temperature of, 409 Compressive force, 516
reactor, 1, 8, 16, 45, 50–54, 58–59, boundary conditions, 418, 422 Compressors, pumps and turbines, 764–765
75, 149, 275–276, 411, 412, 507, design limits, 1238 Computational fluid dynamics (CFD)
739, 1163, 1243 heat conduction equation, 421 programs, 546, 557, 1031, 1136
1332 Index
Computational fluid mechanics (CFM) model, fluid viscosity and friction, 558–561 fission products, 1282–1283
557 lateral flow and cross-flow, 557 hydrogen gas, 1224–1226
Computer codes, 301, 1158–1159, 1245–1246 material derivative, 555–556 ideal gas law, 1266–1267
Concentric tube heat exchangers, 304 modeling effects of turbulence, 595 Mark I, 1208
Condensate, 288, 336 momentum equations, 573–575 Mark II, 1209
Condensation, 872, 881–882 adding turbulence to, 592 Mark III, 1209–1210
Condensation heat transfer, fundamentals of, differential, 575–577 mass balance in, 1265
1277–1278 time-dependent, 578–579 missile defense shield, 1205
Condensation process, 1260 Navier–Stokes equations, 561 pressures, 1276, 1277
Condensation rate, 1279 approximate solutions to, 595–596 pressure, temperature and energy
Condensers, 274, 283, 287–288, 295, 528, 869, in cylindrical coordinates, 581–583 relationships, 1226–1228
1215–1216 origin of, 596 pressurized water reactors, 1203, 1204
defintion of, 313 singular solutions to, 561–563 protection from flying objects, 1206–1207
NSSS, 317 for turbulent flows, 594–595 as radiation barrier, 1223–1224
reduce temperature and pressure, 319, Newtonian vs. non-Newtonian fluids, 583 shapes and sizes, 1207
331–332 number of single-phase flow, 556 subcompartments and function,
waste heat rejection process, 320, 324, for one-dimensional flows, 564–566 1203–1205
335–336 pressure field from velocity field, 580 Containment energy equation, 1273
Condition of slip, 526 Reynolds decomposition, 594 Containment heat removal system, 1151
Conduction, 351–354, 393, 399 simplifications to CFD models for nuclear Containment liner, 1152
and convection, 781 reactors, 557 Containment pressure, 1268, 1275
Conduction of heat, 478 simplifications to fluid, 556–557 Containment suppression system, 1224
Conductive cladding, 417 thermal-hydraulic correlations and their Continuity equation, 1114, 1119
Conductive heat transfer, 352, 354–355, 393, uses, 566–567 in Cartesian coordinate system, 568
395, 455–456, 466, 791 treatment of fluid friction in compressible in different coordinate systems, 569
Conductive resistance, 372 fluid, 583–584 for incompressible flows, 569
Conductor, 400 for two-phase flow, 919–921 for incompressible fluid, 568
Conservation, 402 velocity fluctuations in turbulent flows, with material derivative, 568
Conservation equations 592–594 Continuous vapor film, 944
advection of momentum, 577–578 Conservatism, 494–495 Contraction factor, 619, 699
boundary conditions Constant, 613 Control rod drive mechanisms (CRDMs), 40
flow, 599 Constant axial heat flux, 794–796, 829 Control rods, 21–22, 163–165
fluid, 596–597 Constant of integration, 550 Control surface, 744
free surface, 598 Constant pressure processes, ideal Brayton Control system instabilities, 1174–1175
inlet/outlet, 598 cycle, 341, 342 Control volume, 145, 146, 534, 557, 560
interface, 597–598 Constant rod temperature, 795 conservation of energy, 402
no-slip, 597 Constant speed, 767 energy balance, 402
other types of, 598 Contact angle, 889 in reactor fluid mechanics, 495–497
pressure, 580, 598–599 Contact pressure, 434, 447 Convection, 351–354, 393, 399, See also
velocity, 580, 598 Contact zone, 378 Newton’s law of convection
viscous effects and, 588 CONTAIN Code, 1247, 1248, 1284, 1285 and conduction, 781
boundary layer approximations, 589–590 Containment analysis and safety codes, diffusion vs. advection, 533–534
boundary layer flow, 600–601 1283–1284 number, 972, 973, 977
transition points in, 602 Containment building, 1146, 1147, 1151, 1152, Convection of heat, 478
boundary layer solutions, 601–602 1155–1157, 1162, 1165 Convection process, 1075
Cauchy’s equations asymmetric pressurization, 1282 Convective behavior of dye, 534
for compressible fluid, 584–585 overhead spray system, 1101–1102 Convective component, 971
to handle incompressible Newtonian particle dispersion outside of, 1093–1094 Convective heat transfer, 351, 352, 361–363,
fluids, 585–586 pressure equation, 1269 393, 395
comparing various forms of, 592 volume equation, 1269 CHF region, 943–944
continuity equation, 567–569 Containment buildings, 1253–1254 coefficients, 300, 362, 363, 381, 937
working with, 569–570 behavior during loss-of-coolant accident, affecting factors, 783–785
in different coordinate systems, 563–564 1254 correlations for, 819
energy equation, 570–572 boiling water reactors, 1207–1208, 1223 equivalent diameter, 819
working with, 572–573 condensation, 1275–1277 fine-tuning, 800–801
Euler’s equation conditions, 1275 geometric correction factors, 819
into Bernoulli’s equation, 591–592 corrections for humidity of, 1271–1273 for liquid metals, 826–829
and its origins, 588–589 Dalton’s law of partial pressures, 1265–1266 natural convection, 862–864
fluid flow types, 557–558 decay heat deposition, 1257–1258 values for Nusselt number, 819–820
fluid friction in nuclear power plant, determining pressures and temperatures correlations for, 944
586–588 in, 1265 evaporation rate, 948–949
fluid mechanical modeling effects of water on air pressure, 1270 film boiling region, 944
with CFD program, 557 energy equation, 1264 horizontal tubes over, 953–954
of reactor core, 603–604 film condensation, 1278–1280 liquid convection region, 941–942
Index 1333
nucleate boiling, 942–943 maximum cladding temperature, 1008 boiling crisis and, 1041
resistance, 473 and fuel, 1015 cladding temperature changes after,
in spent fuel pools, 953 mechanistic thermal design, 1001–1004 985–986
system pressure in nucleate boiling PWR vs. BWR cores, 1020 correlations, 1047
regime, 955 radial power distribution, 1006–1007 for BWRs, 1055–1058
transition boiling reactor nodal and subchannel analysis, vs. critical quality, 1066–1067
and film boiling, 949–950 1024–1025 dependence of
region, 944 reactor subchannel design, 1030–1033 on exit flow quality, 1043
tube banks, 953–954 SPDs and engineering uncertainty factors, on grid spacers, 1043–1044
tubes in reactor heat exchanger, 955–956 1022–1023 on mass flow rate, 1043
Convective nucleate boiling, 954 statistical uncertainties in, 1021–1022 on system pressure, 1044–1045
Convective resistance, 1010 super cells for reactor thermal-hydraulic and DNBR with position, 1045–1046
Conventional orifice equation, 1122 analysis, 1025–1026 region, 943–944
Conventional orifice flow, 1110 temperature drop from cladding to Critical isentropic flows, 1131
Converging–diverging nozzles, 1112, 1131, 1132 coolant, 1009–1010 Critical length, 1135, 1136
Coolable core geometry, 1150 three-dimensional cores, 1033 Critical mass flux, ERM equation
Coolant Corium, 1261 saturated liquids, 1127
behavior of, 226–227 Corium pool, 1262 subcooled liquids, 1128
channels, 1120–1121, 1134 Correction factors (CF), 493, 672–673, 808 water–steam mixture, 1129, 1141
energy balance, 1004–1005 for isothermal flows, 689–690 Critical orifice flow, 1112–1113
field, 1282 for non-isothermal flows, 689–690 Critical point, 983
leg, 275, 759 viscosity, 821 in Clausius–Clapeyron equation,
lines, 275 Correlations 234–235
pumps, 337 for convective heat transfer during pool in property diagrams, 230
temperature profiles, 55, 262–263 boiling, 944 P–V diagram, 226
thermodynamic properties of, 221–222 for two-phase flow, 914 T–V diagram, 232
velocity profiles, 659–660 Cosmic rays, 129, 369 Critical power level (CPL), 1058
Cooling spent nuclear fuel, 370–372 Counterflow Critical power ratio (CPR), 939, 996, 1001,
Cooling towers, 289–291, 1217–1218 heat exchanger, 281, 294–297 1146
Core average void fraction, 742 SG, 279, 281 and its uses, 1058–1059
Core bypass flow, 40 vs. parallel-flow heat exchangers, 302 Critical pressure ratio (CPR), 947, 954, 1116,
Core design correlations, 1323 Coupled neutronic-thermal hydraulic 1118, 1120–1121
Core flow instabilities, 1177–1179 Critical properties, 1108
with orifice plate, 707–708 Coupled power–flow oscillations, 1170–1172, Critical quality vs. CHF correlations,
patterns, 713–716 1191–1192 1066–1067
and pressure management, 708–709 CRDMs, See Control rod drive mechanisms Critical Reynolds number, 653–655, 722, 812
Core flow effects, 1239–1240 (CRDMs) Critical speed, 1112
Core flow rates, 1237 Creep, 282 Critical two-phase flow, See Two-phase
Core isolation, 1151 Critical back pressure, 1126 critical flow
Core power density, 145–146, 411 Critical cavitation number, 772 Critical velocity, 1105, 1113, 1115, 1117, 1118
Core pressure drop, 909 Critical distance, 726, 789 Cross-coupling effect, 564
Core spray system, 1156 Critical enthalpy, 991 Cross-flow, 1027, 1031
Core temperature Critical flow correction factor, 304
axial coolant temperature, 1005–1006 computer modeling of, 1136 coupling coefficient, 712
axial heat flux, 1008–1009 description of, 1105, 1106 definition of, 495
boiling core with friction, 1134–1135 fluid statics and dynamics, 525, 526
coolant temperatures in, 1017–1019 for gaseous coolants, 1131–1133 heat exchangers, 303, 304, 831
fuel and cladding temperatures in, LOCA, 1149 intra-assembly, 712–713
1019–1020 with multiple sound speeds, 1118–1120 and its origins, 711
cladding and coolant single-phase, 1114–1117 subchannel, 711–712
in real cores, 1016–1017 two-phase, 1117–1120 Cruciform-shaped control rods, 22, 74, 163
temperatures, 1008–1009 Critical flow quality (CFQ), 1056 Cruciform-shaped cross section, 455
cladding temperatures for hot channel, CISE-4 correlation for, 1059 Crud buildup, 304–307, 372
1007–1008 Critical flow rate Curve fitting exercise, 972
conservatism in subchannel models, back pressure and exit pressure, 1117 Cylindrical coordinates, 408, 449
1026–1031 critical orifice flow, 1112–1113 Cylindrical fuel rod
coolant energy balance, 1004–1005 for long and short coolant channels, steady-state heat conduction equation,
fuel and the cladding, 1001 1120–1121 426–428
fuel temperature single-phase critical flow, 1115 temperature profile of cladding in,
cladding, 1014–1015 sound speed, 1118 421–422
coolant, 1010–1014 thermal equilibrium on, 1122, 1125 temperature profile of fuel in, 420–421
homogenous equilibrium model equations, thermal non-equilibrium on, 1121 thermal resistance of, 422–424
1030–1033 Critical heat flux (CHF), 939, 983, 985, 1042, Cylindrical objects, heat conduction, 379–380
hottest coolant channel, 1005–1006 1238 Cylindrical power profiles, 158
1334 Index
Dalton’s law of partial pressures, 246, for fuel, 1237–1238 Dynamic flow instabilities, 1169–1170
1265–1266, 1272, 1274 Design margin, 1231 Dynamic pressure, 630
Darcy equation, 622, 640, 664, 748, 912 Design rules, 1236 Dynamic temperature, 1130
to composite flow channels, 667 Design safety limit, 1233 Dynamic thermal-hydraulic instabilities, 1176
to composite pipes, 667 Deterministic model, 1158 Dynamic viscosity, 486–487, 494, 508–509,
formula, 662 Deuterium, 51 559, 560, 1030
for frictional pressure drop, 662 Deuterium atoms, 736 laminar and turbulent flows with friction,
Darcy–Weisbach friction factor, 662, 664, 667 Developed flow, 492–493, 1118 647
Design basis accident (DBA) categories Diesel generators, 871 single-phase flow, 611
categories Differential form, 564, 571
equipment failure, 1243 Differential momentum equations, 575–577 ECCS, See Emergency core cooling system
federal policy, design limits, Diffusers, 628–630 (ECCS)
1243–1244 Diffusion ECR, See Equivalent clad reacted value
fuel-handling accidents, 1243 advection vs. convection, 533–534 (ECR)
loss of coolant accidents, 1243 coefficient, 407 Effective density, 912, 916
loss-of-flow transients, 1242 equation, 399, 407 Effective diameter, 623, 653
reactivity initiated event, 1242 process, 353, 393, 399, 1075, 1076, Effective drift velocity, 899, 900
reactor accidents, 1243 1077–1079 Effective flow area, 1106, 1108
codes, 1248 of radioactive particles, 551 Effective gap heat transfer coefficient, 387
DCH, See Direct containment heating (DCH) Digital pressure sensors, 519 Effective viscosity, 912, 913, 1030
Decay constant, 475 Dimensional similarity, 1307–1308 Elastic scattering, 125
Decay heat, 127, 464, 871 Dimensionless numbers, 652, 885, 1114, 1140, Electrical analog, 450
ANS standards governing, 173–174 1307 Electrical conductivity, 358
from beta decay, 167, 168 Direct containment heating (DCH), 1282 Electrical distribution system, 30
as function of burnup, 169 Discharge coefficient, 1117 Electrical generators, 10, 27–29, 528, 739,
over time, 848–850 Discharge rates, 1121–1124, 1128 766, 1213–1215
production, 848 Distillation process, 736 function of, 284–287
removal, 169–173 Dittus–Boelter correlation, 793, 800, 813, matched pairs, 765
sources of, 167 820–821, 976, 1007, 1019 NSSS component, 272, 274
temporal dependence, 172 Divergence operator, 568 in steam turbines, 283
time dependence, 1256–1257 DNBR, See Departure from nucleate boiling Electrical power production measurements,
Decay heat deposition, 1257–1258 ratio (DNBR) 29–30
Deep ocean thermal cycle, 313 DNBR estimation Electrical resistance, 372, 384
Degree of superheat, 885 core flow effects on, 1239–1240 Electric meters, 29
Demineralizer, 288–289 flow blockages on, 1240–1242 Electric Power Research Institute (ePRI),
Density changes, 431–432 rod bowing effects, 1239–1240 1205, 1285
Density differences, 524, 852, 861 DNBR operating limits, 1238 Electromagnetic waves, 365–366
Density of liquid water, 498 Donor channel, 1029 Elliptic boundary conditions, 590
Density ratio, 894 Dose calculation codes, 1249 Elliptic equations, 600
Density wave oscillations, 1177, 1180–1181, Dose rates Emergency core cooling system (ECCS), 869,
1186 for BWR LOCAs, 1100–1101 1151, 1154–1155, 1158, 1274
Density wave propagation, 1180 for PWR LOCAs, 1098 Emergency shutdown, 464
Density waves, 501, 1179–1181 Double diffusive convection, 828 Empirical correlations, 566
Departure from nucleate boiling (DNB), 1039, Double-ended guillotine pipe breaks, 1243 Empirical failure model, 1158
1323–1324 Double-ended LOCAs, 1243 Empirical friction factors, 556
condition, 960, 984 Downcomer, SG, 277 Empirical models, 1137
limit, 1064 Downflow, 716 Enclosed flows, 485, 489, 649, 652, 653, 655
Departure from nucleate boiling ratio Draft cooling tower, 1218 End of film boiling, 980
(DNBR), 939, 996, 1001, 1045, Drag coefficient, 723, 724, 1091 Energy balance, 287, 295, 297–299, 332, 402,
1063, 1146, 1236 effect of surface roughness, 721–723 462, 472, 628, 1263
effect of power shape, 1325 local, 725 multiple feedwater heaters, 330, 331
effect of rod bowing, 1325 Drag force, 723, 1087, 1091 NSSS, 318–320
effect of turbulent mixing, 1325 on nuclear fuel rod, 727–728 nuclear power plants, 228
limit, 1324 on plate-type fuel rod, 729 primary loop, PWR, 334
PWR, 1064–1065 vs. pressure drag, 726–727 thermal efficiency, 313
Depleted fuel, 430 Drift flux, 899 using enthalpy, 263–264
Depressurization rate, 1152–1154, 1156 model, 899–900 Energy cascade, 595
Design assumptions, 1238–1239 void fraction to correlations, 900–901 Energy conservation, 376
Design basis accidents (DBAs), 1098, 1145, Drift velocity, 899, 1140 Energy conservation equation, 613
1154, 1235, 1254, 1286 Drop flow regime, 905 Energy decrement per collision, 133
Design criteria, LOCA, 1150 Dropwise condensation, 841, 881, 1274, 1277, Energy, definitions of, 195–197
Design limits 1278, 1280–1281 Energy deposition, 367–369
for cladding, 1238 Dry Falls National Monument, 365 Energy equation, 570–572, 596, 1140–1141
for core, 1236–1237 Dry steam, 240 material derivative, 571
description of, 1231–1233 Dry well, 1156 working with, 572–573
Index 1335
Energy flow rate, 314 ERM, See Equilibrium rate model (ERM) Film nucleate boiling, 883
Energy of photon, 353 ERM correction factor, 1127 Film pool boiling, 939, 940
Energy storage capacity, 254 ERM equation, 1127–1129 Film temperature, 837, 863
Energy–velocity equation, 132 ESFs, See Engineered safety features (ESFs) Filmwise condensation, 881
Engineered safety features (ESFs), 1274 ESSs, See Engineered safety systems (ESSs) Final safety analysis report (FSAR), 428,
Engineered safety systems (ESSs), 1274–1276 Eulerian fluid mechanics, 497 1046, 1098, 1155, 1158, 1235, 1236,
Engineering uncertainty factors, 1046, acceleration, 533 1244, 1254
1070–1071 descriptions of, 531, 533, 538–539 Finger-shaped control rods, 22–23, 24
and statistical probability distributions material derivative in, 537–538 Finite difference equations, 481
(SPDs), 1022–1023 views of, 532–533, 555, 561 First law of thermodynamics, 179–181
Engineering uncertainty sub factors, 1070 Eulerian reference frames, 534–536 heat energy, 184–185
English unit system, 1291 Euler’s equation, 589 thermal efficiency, 185
Enhance methods, 785 into Bernoulli’s equation, 591–592 Fishenden–Saunders correlation, 944, 945,
Enthalpy, 962–965 and its origins, 588–589 950
definition of, 202–203, 227–228, 313 material derivative, 589 Fissile isotopes, 10
for ideal gas, 343 single-phase flow, 612 Fission gas plenums, 448, 964, 1309
and internal energy, 227 European power plant, 290–291 Fission gas release, 430–431
in reactor heat exchanger, 263–264 Evaporation, 209, 234, 291–293, 881, 1152 Fission neutrons, kinetic energy of, 131–133
specific, 570 rate, 948–949 Fission process, 122, 124, 126–127, 1147
in thermodynamic property table, EV component, 971 Fission product inventory, 1098
226–227 Event trees, 1244–1245 Fission products, 167, 365, 368, 371, 1164
transport equation, 538 Exit flow quality, 914, 1043 Five-equation models, 1137
of two-phase mixture, 228–229 Exit pressure, 1105, 1112, 1113, 1117, 1118, Flashing flows, 1127
of vaporization, 236–237, 962 1123 Flat plates, 954
Enthalpy H, 202 Exothermic reaction, 1146, 1162 Flat vertical plates, 854
Enthalpy of vaporization, 203, 207 Expansion factor, 619, 699 Flow area, 1112
Entrance effects, 657–659 Expected state, 1231 Flow area formulas, 1309
in single-phase heat transfer correlations, Exponential heat sources, 453–455 Flow blockage, 495, 616–617, 1240–1242
825–826 Extensive properties, fluid, 497 effects of, 716–717
Entrance length, 657–659, 797–798 External events, 1243 fluid statics and dynamics, 539
for laminar flows, 797–798 External neutron source, 407 Flow boiling, 883, 937, 938, 939, 956, 971
Entropy Extrapolation distance, 1016 convective heat transfer for, 957–958
definition of, 252, 253 correlations, 966
of two-phase mixture, 239 Fanning friction factor, 667 curve, 965–966
Entry length, 493 Fanno functions, 1135 and forced convection, 956–957
Envelope of statistical uncertainty, 1066 Fanno gas flows, 1134, 1135 regimes, 85
Envelope of uncertainty, 1046 Fanno steam flow, 1135 Flow boundary conditions, 599
Environmental impact statement, 1235 Faraday disk, 285 Flow channels, 495–497
Eötvös number, 946 Fast breeder reactor, 95 Flow constrictors, 618
EPRI-II correlation, 1057, 1060 Fast reactor coolants, 739–740 Flow energy, 183, 745
Equation of motion, for particles, 1089 Fast reactors Flow oscillations, 1191
Equation of state, 497, 499, 611 advantages, 95 Flow pattern, 879
Equations of state, See also State equation; breeding ratios, 101 Flow quality, 83–85, 896
Thermodynamic properties clementine reactor, 116–117 Flow regime, 879
description of, 222–223, 225–226 coolants, 95–96 dependent, 528, 901, 938, 966, 1007, 1019
for ideal gas law, 223–224 thermal cycles, 345–347 two-phase flow
Equiaxed grain region, 425, 428–429 void coefficients, 1196–1197 annular flow, 904
Equilibrium flow, 879 Fauske’s model, 1125, 1127 bubbly flow with saturated boiling, 903
definition of, 1118 Feedwater heaters, 289, 328, 330, 331, bubbly flow with subcooled boiling,
discharge rates for, 1121–1124 1216–1217 903
void fraction and quality, 1119 Feedwater ring, 276 churn flow, 904
Equilibrium models, 1136–1137 Fertile materials, 11 maps, 905, 906
Equilibrium quality, 962–965, 1119, 1124 Fick’s first law, of particle diffusion, 1082 mist or drop flow, 905
and heat transfer coefficient, 965–966 Fick’s law, 407 pure vapor flow, 905
location of, 993 Fick’s second law, of particle diffusion, 1082 saturated/bulk boiling, 903
on mass flux, 993 Field variables, 532 slug flow, 903–904
Equilibrium rate model (ERM), 1127–1129 Film boiling, 883, 944, 961 subcooled liquid flow, 903
Equilibrium void fraction, 891 and nucleate, 883–884, 965 Flow separation, 727
Equivalent annulus, 966 region, 975 Flow stream, 891
Equivalent clad reacted value (ECR), 1163 Film condensation, 841, 1277 Flow tube, 550
Equivalent diameter, 622, 623, 673, 1090 dropwise condensation vs., 1280–1281 Flow work, 183, 250, 255, 259, 744–745
formulas, 1309 on inclined surfaces, 1280 Fluid boundary conditions, 596–597
Equivalent electrical network, 385, 386 on vertical surfaces, 1278–1280 Fluid continuity equation, 567
Equivalent inertial length, 753 Film dryout, 957, 959, 980, 982, 1039 Fluid deformation, 545–546
Equivalent thermal resistance, 304 in BWRs, 1054–1055 Fluid-dependent parameter, 978
1336 Index
Fluid drag coefficient, 804 volume changes natural convection and, 782–783, 857
Fluid dynamics, 515, 531 natural convection driven by, 506–507 nuclear heat transfer, 793
behavior of fluids in motion, 531 with temperature and pressure, Nusselt number, 793
fluid kinematics and, 531–532 504–505 in reactor fuel assemblies, 938–939
fluid statics (See Fluid statics) water hammers, 503–504 Forced evaporation, 948
Fluid energy balance, 298 Fluid properties, 307–309 Forced flows, 489–490, 506
Fluid energy equation, 571 Fluid statics, 515 Form coefficient, 697–698
Fluid flow, 1112 advection, 533–534 Form drag, 727
types, 557–558 advective derivative, observations, Form factor, 618, 697–698
Fluid friction, 340 536–537 Forster–Zuber correlations, 976
correlations, 566 behavior of fluids in motion, 531 Fossil-fired power plants, 19, 270
in nuclear power plant, 586–588 cross-flow in rod bundles, 526 Fossil-fueled power plants, 284
single-phase flow, 611–612 deformation and volumetric strain, Fouled heat exchanger tube, 305
for two-phase flow, 910–911 545–546 Fouling factors, 305–307
and viscosity, 558–561 diffusion and convection, 533–534 Four-equation models, 1137, 1140
Fluidic work, 198 and dynamics, 531–532 Fourier number, 469, 474, 479, 1140
Fluidity, 508 Eulerian fluid acceleration of, 533 Fourier’s equation
Fluid kinematics, 531–532 Eulerian reference frames, 534–536 boundary conditions and, 461
Fluid mechanical modeling fluid kinematics, 531–532 for conductive heat transfer, 466
with CFD program, 557 fluid mechanics definiton of, 461
of reactor core, 603–604 Lagrangian and Eulerian views of, heat conduction equation, 408
Fluid mechanics 532–533, 538–539 higher order solutions, 471–472
bubble formation, 511 material derivative in Eulerian, for reactor geometries, 467–469
compressibility coefficient, 501–503 537–538 solutions, 476
compressibility factors, 503 horizontal planes pressure in, 516–517 Fourier’s law
compressible and incompressible flows, Pascal’s law of pressure, 518–519 application, 399–401
488 path lines, 546–547 assumptions in, 355–356
control volumes in, 495–496 observing streak lines and, 550–551 of conduction, 354–355, 379, 399
developed and undeveloped flows, and streamlines, 546 energy balance, 401, 417
492–493 pressure behavior heat flow, 422
dynamic viscosities, 508–509 in BWRs, 528 heat flux, 379, 403, 418
extensive properties of, 497 in PWRs, 526–527 integral form of, 356, 379
fluidity uses, 508 pressure drops in thermal resistance, 372
forced and unforced flows, 489–490 in horizontal pipe with monometer, in vector notation, 355
ideal gas law, 499 522–524 Fourier’s law of heat conduction, 798
intensive properties of, 497 in open and closed cores, 526 Fourth law of thermodynamics, 180, 190–191
internal flows with and without friction, pressure levels FRAPCON, 433, 448
485–487 in nuclear steam supply system, FRAPCON-3, 1247
kinematic viscosities, 508–509 528–530 FRAPTRAN, 1247
Lagrangian and Eulerian, 532–533, in reactor cores, 524 Free delivery rate, 759
538–539 in reactor pressure vessel, 519–520 Free stream velocity, 789
laminar and turbulent flows in tank/pipe, 519–520 Free surface boundary conditions, 598
description of, 488–489 pressure point, 516 Fresh fuel, 429
entry length for, 493 pressures for layered fluids, 521–522 Friction, 723
lumped parameter approach, 496–497 reactor accidents and particle entrainment, Frictional drag, 726
mach number, 488 551 coefficients, 727
material derivative in Eulerian, 537–538 reactor fuel assemblies force, 723, 726
partial pressures and vapor pressure, fluid pressures in, 520–521 Frictional forces, 485
499–501 heat and energy transfer in, 538 on nuclear fuel rods, 723–726
reactor coolant flows classification, vs. pressure equalization, 524–526 and other support structures, 723–726