J. R. Simonson (Auth.) - Engineering Heat Transfer-Palgrave Macmillan UK (1988)

Engineering Heat Transfer
Other Macmillan titles of related interest

An Introduction to Engineering Fluid Mechanics
J. A. Fox
Principles of Engineering Thennodynamics, second edition
E. M. Goodger
Polymer Materials: An Introduction for Technologists and Scientists
Christopher Hall
Reinforced Concrete Design by Computer
R. Hulse and W. H. Mosley
Prestressed Concrete Design by Computer
R. Hulse and W. H. Mosley
Civil Engineering Materials, third edition
edited by N. Jackson
Reinforced Concrete Design, third edition
W. H. Mosley and J. H. Bungey
Microcomputer Applications in Structural Engineering
W. H. Mosley and W. J. Spencer
Strength of Materials, third edition
G. H. Ryder
Surveying for Engineers
J. Uren and W. F. Price
Engineering Heat Transfer
J. R. Simonson
Senior Lecturer
The City University, London
SECOND EDITION
M
MACMILLAN
EDUCATION
© J. R. Simonson 1975,1988
All rights reserved. No reproduction, copy or transmission

of this publication may be made without written permission.
No paragraph of this publication may be reproduced, copied
or transmitted save with written permission or in accordance
with the provisions of the Copyright Act 1956 (as amended),
or under the terms of any licence permitting limited copying
issued by the Copyright Licensing Agency, 33-4 Alfred Place,
London WC1E 7DP.
Any person who does any unauthorised act in relation to
this publication may be liable to criminal prosecution and
civil claims for damages.
First edition 1975

Reprinted 1978,1981,1983,1984
Second edition 1988
Published by
MACMILLAN EDUCATION LTD
Houndmills, Basingstoke, Hampshire RG21 2XS
and London
Companies and representatives
throughout the world
British Library Cataloguing in Publication Data

Simonson, J. R.
Engineering heat transfer.-2nd ed.
1. Heat-Transmission
I. Title
621.402'2 TJ265
ISBN 978-0-333-45999-7 ISBN 978-1-349-19351-6 (eBook)
DOI 10.1007/978-1-349-19351-6
Contents
Preface to First Edition viii
Preface to Second Edition x
Nomenclature xi
1 Introduction 1
2 The Equations of Heat Conduction 8

2.1 The Nature of Heat Conduction 8
2.2 The Differential Equation of Conduction in a Cartesian
Coordinate System 10
2.3 The Differential Equation of Conduction in a Cylindrical
Coordinate System 13
3 One-dimensional Steady State Conduction 16

3.1 Conduction in Plane Slabs 16
3.2 Effect of a Variable Conductivity in a Plane Slab 24
3.3 Radial Conduction in Cylindrical Layers 25
3.4 Critical Thickness of Insulation 28
3.5 Radial Conduction in Spherical Layers 29
3.6 Conduction with Heat Sources 31
4 Two-dimensional Steady State Conduction 39

4.1 A Numerical Solution of Two-dimensional Conduction 40
4.2 Iterative Computing Procedures for Two-dimensional
Steady State Conduction 44
4.3 The Electrical Analogy of Conduction 52
5 Transient Conduction 58
5.1 The Uniform Temperature, or Lumped Capacity, System 58
5.2 The Solution of Transient Conduction Problems in
One Dimension 61
5.3 Two-dimensional Transient Conduction 67
5.4 Periodic Temperature Changes at a Surface 68
vi CONTENTS
6 Forced Convection: Boundary Layer Principles 78
6.1 Introduction 78
6.2 Equations of the Laminar Boundary Layer on a
Flat Plate 80
6.3 Laminar Forced Convection on a Flat Plate 87
6.4 Laminar Forced Convection in a Tube 92
7 Forced Convection: Reynolds Analogy and Dimensional

Analysis 101
7.1 Reynolds Analogy 101
7.2 Dimensional Analysis of Forced Convection 111
7.3 Empirical Relationships for Forced Convection 115
8 Natural Convection 124

8.1 The Body Force 125
8.2 Dimensional Analysis of Natural Convection 125
8.3 Formulae for the Prediction of Natural Convection 126
9 Separated Flow Convection 136

9.1 Relationship between heat Transfer and Pressure
Loss in a Complex Flow System 137
9.2 Convection from a Single Cylinder in Cross Flow 139
9.3 Convection in Flow across Tube Bundles 139
10 Convection with Phase Change 144

10.1 Description of Condensing Flow 144
10.2 A Theoretical Model of Condensing Flow 145
10.3 Boiling Heat Transfer 149
11 Extended Surfaces 157

11.1 The Straight Fin and Spine 157
11.2 Limit of Usefulness of the Straight Fin 164
11.3 Fin and Finned Surface Effectiveness 164
11.4 Overall Coefficients of Finned Surfaces 165
11.5 Numerical Relationships for Fins 170
CONTENTS vii
12 Heat Exchangers 176
12.1 Types of Heat Exchanger, and Definitions 176
12.2 Determination of Heat Exchanger Performance 181
12.3 Heat Exchanger Transfer Units 185
12.4 Heat Exchange in Cross Flow 191
12.5 Plate Heat Exchangers 200
12.6 Batch Heat Exchangers 202
13 The Laws of Black- and Grey-body Radiation 208

13.1 Absorption and Reflection of Radiant Energy 209
13.2 Emission, Radiosity and Irradiation 210
13.3 Black and Non-black Bodies 210
13.4 Kirchhoff's Law 213
13.5 Intensity of Radiation 215
13.6 Radiation Exchange Between Black Surfaces 217
13.7 Grey-body Radiation Exchanges 224
13.8 Non-luminous Gas Radiation 228
13.9 Solar Radiation 231
Appendix 1 Heat Transfer Literature 236
Appendix 2 Units and Conversion Factors 237
Appendix 3 Tables of Property Values 239
Appendix 4 Gas Emissivities 257
Index 261
Preface to First Edition
The aim of this book, which is a revised edition of a book previously

published by McGraw-Hill, is to introduce the reader to the subject
of heat transfer. It will take him sufficiently along the road to enable
him to start reading profitably the many more extensive texts on the
subject, and the latest research papers to be found in scientific
periodicals. This book is therefore intended for students of engineer-
ing in universities and technical colleges, and it will also be of assis-
tance to the practising engineer who needs a concise reference to the
fundamental principles of the subject. The engineering student will
find most, if not al~ aspects of the subject taught in undergraduate
courses and, thus equipped, he will be in a position to undertake
further studies at postgraduate level.
The aim throughout has been to introduce the principles of heat
transfer in simple and logical steps. The need for an easily assimilated
introduction to a subject becomes more urgent when the subject
itself continues to grow at an ever-increasing rate. It is hoped that the
material selected and presented will be of value at all levels of reader-
ship. Indebtedness is acknowledged to all those, past and present,
who have contributed to the science of heat transfer with their origi-
nal work, and as far as possible detailed references are given at the
end of each chapter. Also grateful thanks are extended to various
persons and organizations for permission to use certain diagrams,
tables, and photographs; credit for these is given at appropriate
points throughout the text.
It is also hoped that in this edition the changes made will further
enhance the value of the book. Greater attention has been given to
numerical methods in conduction, and some basic procedures in
digital computing are included. The chapter on radiation has been
extended to include an introduction to non-luminous gas radiation
and a short section on solar radiation. Numerous small changes have
viii
PREFACE TO FIRST EDITION ix
been made throughout in the light of reviews and criticisms received.

New worked examples are included to extend the range of applicabi-
lity, and some of the original problems set have been replaced by more
recent ones. SI units are now used exclusively, and conversion factors
for British units are included in appendix 2.
Many of the problems included are university examination
questions; the source is stated in each case. Where necessary the units
in the numerical examples have been converted to SI. Indebtedness
is acknowledged to the owners of the copyright of these questions for
permission to use them, and for permission to convert the units. The
universities concerned are in no way committed to the approval of
numerical answers quoted.
Much of the material in this book has been taught for a number of
years at undergraduate level to students at The City University.
Grateful thanks are due to Professor J. C. Levy, Head of the Depart-
ment of Mechanical Engineering, and to Mr B. M. Hayward, Head
of the Thermodynamics Section. Discussions with colleagues at City
and elsewhere have also contributed in numerous ways, and for this
help sincere thanks are expressed.
Finally, thanks are due to Malcom Stewart, of The Macmillan
Press, who has been responsible for the production of both editions,
and also to my wife, who has typed the manuscript revisions.
Department of Mechanical Engineering, JOHN R. SIMONSON

The City University
Preface to Second Edition
The essential aims of this new edition remain unchanged. While the
subject matter of heat transfer at undergraduate level has not greatly
altered, the student now has a powerful computational tool available to
him, which in its use enables him to plot, explore and appreciate the
mechanisms of heat transfer and their contributions in solving engineer-
ing problems. The use of the personal computer is increasing in all
subject areas of undergraduate courses; and with growing emphasis on
Design as an essential concept in the presentation of undergraduate
studies, the introduction of computing methods into this edition forms
the essential new material.
Relatively elementary computing procedures may be introduced in
the subject matter of steady state and transient conduction, extended
surfaces and heat exchangers, and the bulk of the new material lies in
these areas. Since a considerable amount of valuable and relatively
simple computing practice is possible in the field of cross flow heat
exchange and in rotary regenerators, new sections have been added in
these areas, and in order to make room for all the new computing
material the chapter on mass transfer has been removed. More advanced
computing techniques arise in convection studies and this subject 'is well
covered in the literature. It is hoped that this new edition will help the
student become familiar with the possibilities of computer literacy in
the more elementary aspects of the subject of heat transfer.
The language of the computer listings is BASIC, which is the most
popular language in use in the programming of personal and micro-
computers. Some minor editing may be required to enable the given
listings to run on particular machines. No claim is made for elegance in
the programming presented; it is intended merely to present relatively
simple examples with which the majority of students may gain in com-
puting experience.
Some of the older problems have been removed and new ones intro-
duced. At the same time, some earlier misprints and one or two miscon-
ceptions have been rectified. The author is grateful for comments and
suggestions received since the first edition appeared, and he is grateful,
too, for the support received which has made this second edition
possible.
JOHN R. SIMONSON
X
Nomenclature
a distance increment
A area
b, 1, t, w linear dimension
c capacity ratio of heat exchanger
C,K constants of integration
Cd average friction factor
Cf skin friction coefficient
Cp specific heat at constant pressure
Cp volumetric specific heat at constant pressure
d diameter
E effectiveness of heat exchanger
f friction factor
F geometric configuration factor
~ geometric emissivity factor
/o drag factor
g gravitational acceleration
G irradiation, mass velocity
Gz Graetz number, Re Pr(djx)
hR convection coefficient
H product hA
hrg latent enthalpy of evaporation
hr radiation coefficient
i current density
I current
I intensity of radiation
J radiosity
k thermal conductivity
L,D,T, W linear dimension
L,M,T,O dimensions of length, mass, time, temperature
m mass flow, or mass in transient conduction
m mass flow, where a non-flow m also occurs
n coordinate direction
n frequency of temperature variation
NTU number of transfer units
p,P,flp pressure, difference of pressure
p perimeter
PN plate number
ql heat transfer per unit area and time
q heat generation per unit volume and time
Q heat transfer per unit time, or a physical variable in dimension-
less analysis
xi
XII NOMENCLATURE
r radius, radial direction

r residual value
R resistance
Rm universal gas constant
s scaling factors in electrical analogy
Si electrical shape factor
Sq thermal shape factor
t temperature
T absolute temperature
t, At, T time, time increment, time constant
U,UA,UL overall heat transfer coefficients
u velocity of temperature wave
v velocity
v specific volume
v electrical potential, volume
x,y,z coordinate direction, linear dimension
X length of temperature wave
a thermal diffusivity
a absorptivity
~ coefficient of cubical expansion
li boundary layer thickness
lib thickness of laminar sub-boundary layer
lit thermal boundary layer thickness
lit equivalent conducting film thickness
£ emissivity
£ eddy diffusivity
tq eddy thermal diffusivity
'T/r fin effectiveness
'Tire equivalent effectiveness of finned surface
8,8m temperature difference, logarithmic temperature difference
8 angle in cylindrical coordinate system
X wave-length
IJ. dynamic viscosity
v kinematic viscosity
p density
p electrical resistivity
p reflectivity
a Stefan-Boltzmann constant, surface tension
'T shear stress
'T transmissivity
'Tt turbulent shear stress
tP angle in spherical coordinate system
Dimensionless groups
F Fourier number, Ata.ja 2
Gr Grashofnumber,~g8p2 z3tp.2
J Colburn /-factor, St. Pr 2 13
Nu 1 Nusselt number, hl/k
NOMENCLATURE Xlll
Pr Prandtl number, CpJJ/k

Ra Rayleigh number, Gr.Pr
Re 1 Reynolds number, pvl / JJ
St Stanton number, hfpvcp
Suffices
a at axis of tube
b black body
b limit of laminar sub-boundary layer
c cold fluid
c convection
d, /,X length terms used in dimensionless groups
e equivalent
f fluid
h hot fluid, heated length
i, 0 inlet, outlet (in heat exchangers)
l liquid
m mean value
M metal, in heat-exchanger wall
n direction of component
0 datum length
p constant pressure
r radial direction, or radial position
R radiation
s surroundings, of free stream
sat. saturated temperature
t temperature, turbulent
v constant volume, vapour
w wall
x,y,z direction of component
(J angular component
A monochromatic
Superscript
average value
1
Introduction
One of the primary concerns of the engineer is the design and

construction of machines many times more powerful than himself
or any of his domestic animals. The development of this skill over
the centuries has been fundamental to the growth of civilization.
Man's early efforts to harness the power of wind and water owed
very little to engineering science, and indeed the early steam engine
was a practical reality before the science of thermodynamics was
firmly established. In contrast, there is now a vast fund of engineer-
ing knowledge behind the present day prime movers.
Much engineering activity is directed to the controlled release of
power from fossil and nuclear fuels, and with making that power
available where it is needed. The laws of heat transfer are of the
utmost importance in these activities. The generation of power from
the energy changes of chemical and nuclear reactions involves the
transfer of vast quantities of thermal energy. Further, chemical
processes of combustion yield temperatures at which most construc-
tional materials would melt; adequate protection by heat transfer
processes is therefore vital. The distribution of energy as electricity
is accompanied, at all stages, by certain wastages manifested as
rising temperature of the equipment. Heat transfer considerations
enable these temperatures to be controlled within safe limits.
The laws of heat transfer find application in many other fields of
engineering. Chemical and process engineering, and manufacturing
and metallurgical industries are examples. In addition, the civil and
constructional engineer and environment control engineer need
considerable knowledge of the subject. Large city buildings must
be economically heated and insulated, and air conditioning is
increasingly necessary.
To the mechanical engineer heat transfer is a subject closely
allied to applied thermodynamics. The first and second laws of
1
2 ENGINEERING HEAT TRANSFER
thermodynamics state the relations between the physical entities of
heat and work, and the limit to the amount of work that may be
obtained from any source of heat. Even this limit cannot be reached
in practical engineering processes because of their inherent irreversi-
bility. These irreversibilities may be accounted for in calculations
but, even so, thermodynamics alone leaves a lot of questions un-
answered. There is no time scale and, consequently, thermodynamics
will not permit the calculation of physical sizes necessary to achieve
a given objective. In a steam power plant it is necessary to transfer
the thermal energy of the hot combustion gases of the burnt fuel to
the water in the boiler tubes. The actual rate of transfer to produce
a required flow rate of steam may be known, but without the laws
of heat transfer and knowledge of the properties of the engineering
materials to be used, it is not possible to calculate the size and
surface area of the tubes required. From an economic point of
view, the boiler must be made as small as possible, hence the heat
transfer rate must be as high as possible. Elsewhere in the plant,
heat transfer considerations are necessary in insulating the steam
delivery lines and in condensing the low pressure turbine exhaust.
Heat transfer processes, then, are described by equations which
relate the energy to be transferred in unit time to the physical area
involved. Other factors entering the equations are the temperatures,
or the temperature gradient, and some coefficient which depends on
various physical properties of the system and on the particular
mechanism of heat transfer involved. Three basic mechanisms of
heat transfer are recognized. They may occur separately, or simul-
taneously. Separate equations may be written to describe each
mechanism, and when two or more mechanisms occur simul-
taneously it is sometimes possible to add the separate effects; but
sometimes it is necessary to consider the equations of the participat-
ing mechanisms together. The subject matter thus conveniently
sub-divides itself into the separate basic mechanisms of heat transfer,
and the combinations of them.
Heat is transferred by conduction, convection, and radiation.
Before describing these processes, it is desirable to clarify what is
meant by 'heat'. In the study of thermodynamics, heat is defined as
an energy transfer between communicating systems, arising solely
from a temperature difference. Thus a heat transfer is strictly a
phenomenon occurring only at boundaries of systems, and a heat
transfer elsewhere in a system is more correctly a redistribution of
INTRODUCTION 3
internal energy within the system. As it is convenient to keep to the

conventional language of heat transfer, this should be kept in mind,
and the word heat will not in most cases be in accord with the
thermodynamic usage.
Conduction is the mode of heat transfer in a solid material and
occurs by virtue of a temperature difference between different parts
of the material. Conduction also occurs in liquids and gases but is
generally associated also with convection, and possibly with radia-
tion as well in the case of gases. Conduction within a solid is a
transfer of internal energy; this energy is, in fact, energy of motion
of the constituent molecules, atoms, and particles of which the
material consists. The kinetic energy of the motion is proportional
to the absolute temperature; molecular collisions lead to energy
transfer to regions of lower kinetic energy. Under steady conditions
a molecule will pass on the same amount of energy that it receives.
Under non-steady conditions the flow of energy is governed by the
changing energy levels.
The theory of conduction heat transfer was established by Joseph
Fourier whose work was published in Paris in 1822, 1 but pioneer
work was done by Biotin 1804 2 and 1816. 3 Conduction is described
by an equation known as the Fourier rate equation
dt
Qx =- kA-
dx
(1.1)
The rate of heat flow (in only the x-direction, see Fig. 1.1) is propor-
tional to the product of the area of flow and the temperature gradient,
the constant of proportionality being the thermal conductivity k
which is a property of the material. The negative sign results from
the convention of defining a positive heat flow in the direction of a
negative temperature gradient. The property k may be a function of
temperature and direction of heat flow. Materials with directional
dependence of thermal conductivity are said to be anisotropic.
The units involved depend on the system chosen. In the SI system,
the unit of heat or internal energy is the joule, hence rate of heat
transfer is measured in Jjs or W. However, the kilojoule, (kJ), and
kilowatt, (kW~ are accepted multiples of the SI unit, and to be consis-
tent with general usage in thermodynamics, the kJ and kW are the
preferred units in this book. With the area in m 2 and the temperature
gradient in K/m, the units of k are kW /(m K). This follows the British
Standards recommendation 4 for the presentation of complex units.
uniform temperatures over each face
I 41
Q = -kA dt
" dx
Fig. 1.1. Follrier's hlw for ou-tlinuuio•fll stellll:y st11te co1Uiuctio11 ;, 11
pl~~~~e sld of 11111terill/.
In the British system of units, used in the majority of publications in

English up to the mid 1960s, the rate of heat transfer is measured in
British thermal units/hour, or Btu/h. and with the area in ft 2 and the
temperature gradient in °F/ft, k is measured in Btu/(ft h °F).
Conduction in fluids generally forms a very small part of the total
heat transfer, convection being the predominating mechanism.
Convection is the name given to the gross motion of the fluid itself,
so that fresh fluid is continually available for heating or cooling.
Apart from the bulk movement of the fluid, there is generally a
smaller motion of eddies which further assists in distributing heat
energy. Convection heat transfer is sub-divided into two different
kinds, natural and forced. Heat transfer by natural convection
occurs between a solid and a fluid undisturbed by other effects
when there is a temperature difference between the two, as in a
kettle of water. It is not often that a fluid can be regarded as entirely
at rest, so frequently there is a small amount of forced convection as
well. But true forced convection requires a major applied motion of
the fluid in relation to the source or sink of heat, so that natural
convection effects are negligible. An important aspect of natural
convection is that the fluid motion which does occur is due entirely
to natural buoyancy forces arising from a changing density of the
fluid in the vicinity of the surface. Within the realms of both natural
INTRODUCTION 5
and forced convection there are two sub-divisions of laminar and

turbulent flow convection. In forced convection separation of flow
can occur from the surfaces of immersed bodies, for example, in the
flow across the outside of a pipe. A further type of forced convection
involves a phase change of the fluid, as in boiling and condensing.
It is thus evident that many factors enter into heat convection,
including the shape and magnitude of the solid-fluid boundary,
characteristics of the fluid flow, such as the magnitude of turbulent
eddies, and the conductivity of the fluid itself.
Because of these complexities many convection problems are not
amenable to mathematical solution, and recourse is made to tech-
niques of dimensional analysis and experiment. Thus many empirical
dimensionless relationships are now available in the literature to
enable the engineer to design his heat transfer apparatus, whether
it be an industrial heat exchanger or domestic convector.
Newton (1701) 5 proposed a general equation to describe convec-
tion heat transfer.
(1.2)
Figure 1.2 indicates that heat transfer is occurring from a surface
of area A at temperature t 1 to a fluid at a lower temperature t 2 .
fluid, t 2
Q = hA(t 1 - t 2)
Fig. 1.2. Dillgrtllllllllltic representation of convection from a flat plate,
i/butrating the use of Newton's equation.
h is the convection coefficient and has the units kW/(m 2 K) in the

SI system, or Btu/(ft 2 h °F) in the British system. It takes care
of the many factors entering a particular example of convection,
and the value of h will vary for differing flow regimes, fluid oroperties,
and temperature differences. The main problem in the analysis of
Table 1.1
Convection system Range of h,* kW/(m 2 K)
Natural convection 0·004-0-05

Forced convection (air) 0·01-0·55
Forced convection (liquids) 0·1-5·5
Boiling heat transfer (water) 1·0-11(}0
Condensation (steam, filmwise) 0·55-25·0
Forced convection (liquid metals) 3·0-11(}0
• For numerical conversion factors, see the Appendix
convection is to predict values of h for design purposes. The value

of h in different regimes and for different fluids is generally within
the ranges indicated in Table 1.1.
The third mode of heat transfer known as radiation is rather
different in nature from the first two. Conduction and convection
occur within solid or fluid material and often are present simul-
taneously. In contrast, radiation is an energy transfer which is
transmitted most freely in a vacuum. It occurs between all material
phases. All matter at temperatures above absolute zero emits electro-
magnetic waves of various wave-lengths. Visible light together with
infra-red and ultra-violet radiation forms but a small part of the total
electromagnetic spectrum. The mechanism by which radiation is
propagated is not of any direct concern to the mechanical engineer,
who is mostly interested in overall effects rather than in molecular
detail. It is sufficient to say that radiation is energy emitted by vibrat-
ing electrons in the molecules of material at the surface of a body, and
the amount emitted depends on the absolute temperature of the
body.
The third equation to be introduced at this stage dates from 1884
when the work of Boltzmann 6 consolidated the earlier work of
Stefan (1879). 7 Known as the Stefan-Boltzmann equation, it is
Q = aAT 4 (1.3)
where T is the absolute temperature, A is the surface area of a per-

fectly radiating body and a is the Stefan-Boltzmann constant and
has the value of 56·7 x 10- 12 kW /(m2 K 4 ), or 0·171 x 10- 8 Btu/
(ft 2 h 0 R4 ). Stefan established this relationship experimentally,
subsequently Boltzmann proved it theoretically. A perfectly
radiating or black body emits at any given temperature the maxi-
INTRODUCTION 7
mum possible energy at all wave-lengths. The energy emitted will

be less for real materials. This equation defines an energy emission,
rather than an energy exchange. The area A will also be absorbing
radiation from elsewhere, which must be taken into account in an
energy exchange relationship. The emitting and absorbing charac-
teristics of surfaces, and the 'view' that surfaces have of each other,
are factors which enter the consideration of radiation exchanges.
REFERENCES
1. Fourier, J. B. Theorie analytique de Ia chaleur, Paris, 1822. Translated by
A. Freeman, Dover Publications, New York, 1955.
2. Biot, J. B. Bibliotheque Britannique, Vol. 27, 310 (1804).
3. Biot, J. B. Traite de physique, Vol. 4, 669 (1816).
4. British Standards Institution, The Use of SJ Units, PD5686: 1972.
5. Newton, I. Phil. Trans., Roy. Soc., London, Vol. 22,824 (1701).
6. Boltzmann, L. Wiedemanns Annalen, Vol. 22, 291 (1884).
7. Stefan, J. Sitzungsber. Akad. Wiss. Wien. Math.-naturw. Kl., Vol. 79, 391
(1879).
2
The equations of heat conduction
2.1 The Nature of Heat Conduction

The Fourier equation of heat conduction (1.1) has already been
introduced. This equation is for one-dimensional heat ftow, and
may be written in a more general form:
ar
Q = -kA- (2.1)
" on
where Q, is the rate of heat conduction in then-direction, and ot/on
is the temperature gradient in that direction. The partial derivative
is used since there may exist temperature gradients in other direc-
tions. One-dimensional conduction does not often occur in practice
since a body would have to be either perfectly insulated at its edges or
so large that conduction would be one-dimensional at the centre.
Equation (2.1) expresses an instantaneous rate of heat transfer. It
may be re-written
q = Q, = _kat (2.2)
II A on
where q, is the heat ftux in heat units per unit time and per unit
area in the n-direction. This is a vector quantity since it has magni-
tude and direction. The greatest heat flux at an isothermal surface
will always occur along the normal to that surface.
Heat conduction within a solid may be visualized as a heat flux
which varies with direction and position throughout the material.
This follows from the fact that temperature within the solid is a
function of position coordinates of the system (e.g., x, y, z). In addi-
tion, temperature may be a function of time, (t), so in general
t = f(x, y, z, t).
The problem of determining the magnitude of heat conduction
8
THE EQUATIONS OF HEAT CONDUCTION 9
resolves itself to finding first the isotherms within the system and
the way in which their positions vary with time. In steady state
conduction the isotherms remain stationary with time, and one may
visualize a large number of isothermal surfaces throughout the
system, differing incrementally in temperature. The heat flux
normal to any one surface will vary with position depending on the
distance between surfaces. It is then necessary to sum the heat flow
through the boundary surfaces if internal heat sources are present, or,
if not, through any isothermal surface. In unsteady conduction the
problem is complicated by the fact that isothermal surfaces are no
longer fixed, and the rate at which heat is being stored must be taken
into account.
Before taking the first step, which is to develop the equation for
temperature as a function of position and time, it is opportune to
introduce some facts about different conducting materials.
Solid materials may be divided into two groups, metallic and
non-metallic, for which there is a marked contrast in the values of
conductivity. The Appendix lists properties for some of the more
useful materials. The high values of conductivity of metals are
attributable to the well ordered crystalline structure of the material.
The close arrangement of molecules permits a rapid transfer of
energy and, in addition, free electrons play a considerable part.
Metals such as copper which are good electrical conductors also
conduct heat well. There is also a marked similarity between
conduction heat transfer and the flow of electricity, and the electrical
analogy is often used in the solution of conduction problems.
In contrast, non-metals do not have a well ordered crystalline
structure and, in addition, are often porous in nature. Thus energy
transfer between molecules is seriously impeded, and the values of
conductivity are much lower. The small pores within the material,
being full of air, further restrict the flow of heat since gases are poor
conductors. This is because molecules of a gas are relatively widely
spaced and the transfer of energy depends on collisions between
these molecules.
The thermal conductivities of most substances V< cry with tempera-
ture, and for accuracy such variation should be allowed for in
conduction problems. However, this is a complication which may
be ignored in an introductory study of the subject because the
variation with temperature is not great. Over a reasonable tem-
perature range the relationship between conductivity k, and tern-
perature t may be assumed linear:
k, = k 0 (1 + ext) (2.3)
where k0 is the conductivity at temperature t 0 , and ex is a constant.
In most practical applications it is sufficient to assume a mean
uniform value for conductivity.
A complication more serious than temperature variation of con-
ductivity occurs in certain engineering materials, viz., that conduc-
tivity may vary with the direction of heat flow. This arises commonly
in laminated materials used in electrical engineering. Thus the
conductivity parallel to the laminates is different to the value
perpendicular to the laminates. Most types of wood also exhibit
this property, the conductivity parallel to the grain being different
to that across the grain. Conducting materials exhibiting this
property are said to be anisotropic. In the absence of this property
the material is said to be isotropic. For anisotropic materials the
analysis of conduction is more difficult and is not included in this
introductory text The basic ideas are given by Eckert and Drake, 1
and the general treatment may be found in the work of Carslaw and
Jaeger. 2
Differential equations of the temperature field will now be
developed in two coordinate systems, Cartesian and cylindrical.
2.2 The Differential Equation of Conduction in a Cartesian

Coordinate System
The material of the system is assumed to be isotropic and the conduc-
tivity is assumed invariable with temperature. Consider the infini-
tesimal element of the material represented by the volume dx dy dz
in Fig. 2.1. The heat flowing into and out of the element is resolved
in the x-, y- and z-directions. Thus from the Fourier equation the
rate of heat flowing into the element in the x-direction is
ar
dQ,. = - kdydz ox
since the area of flow normal to the x-direction is dy dz and the

temperature gradient is ot/ox. The rate of heat flowing out of the
element in the x-direction is
ar dx)
a (t +ox
dQ(.x+d.xl = - kdydz ox
THE EQUATIONS OF HEAT CONDUCTION II
dQ(z+dz)
dQ,
Fig. 2.1. Conduction in an element of11Uiterial ill Carte1ilut coordillate1.
ot
- kdydz ox- kdxdydz 0 x 2
o2 t
Therefore the net rate of heat flow into the element in the x-direction
is
(2.4)
In a similar manner, the net rates of heat flow into the element in the
y- and z-directions are given by
ot 2
dQY - dQ(y+dy) = k dX dy dz iJyl (2.5)
iJlt
dQZ - dQ(z+dz) = k dx dy dz OZ2 (2.6)
The total rate of heat flow into the element is the sum of the right-
hand sides of equations (2.4), (2.5), and (2.6), which is
o2 t o2 t o2
k dx dy dz (iJx2 + oy2 + oz2
t)
In addition to heat flowing into and out ofthe element, the possi-
bilities of heat being generated within the element (e.g., due to the
flow of electricity) and of heat being stored within the element (as
in the case of unsteady conduction) have to be considered.
If q' is the rate at which heat is generated per unit volume, the
rate of heat generation within the element is
q'dxdydz
The rate at which heat is being stored within the element is
governed by the rate of temperature change otjot. If pis the density
of the material and eP the specific heat at constant pressure, the
rate of heat storage will be
dx dy dz peP
at
at
This assumes the element may expand or contract freely at constant
pressure. The rate at which heat is being stored within the element
is equal to the sum of the net rate of heat flow into the element and
the rate of heat generation within it, hence:
ar (o2 t o2 t o2 t)
pep ot = k ox 2 + oy 2 + oz 2 + q'
where tX = k/ peP and is known as the thermal diffusivity of the

material. It is a ratio of the heat conduction to heat storage qualities
of the material.
Equation (2.7) is the general differential equation of conduction
in a Cartesian coordinate system and may be simplified to suit
any particular application. Thus the equation for unsteady conduc-
tion in one dimension without heat generation is
(2.8)
since q', o2 tjoy 2 and o2 t/oz 2 are equal to 0.

For any problem of steady conduction, ot/Ot = 0, since there is
then no variation of temperature with time. The equations for
two- and one-dimensional steady conduction with heat generation
are
(2.9)
and
(2.10)
it being permissible to use the total derivative in the one-dimensional

case. In the absence of heat generation the equations reduce to
_ (o2t o2t) and 0= oc(~)

0 - DC OX2 + oy2 dx
2
and, consequently,
(2.11)
and
d2t
dx2 = 0 (2.12)
Problems involving equations (2.8) to (2.12) will be considered in

later chapters.
2.3 The Differential Equation of Conduction in a Cylindrical

Coordinate System
Often, conduction problems involve heat flow in solid or hollow
round bars and, consequently, the cylindrical coordinate system,
Fig. 2.2, is used. The general approach is exactly the same as before
except that heat flows in radial, circumferential, and axial directions
have now to be considered. The element to be considered has volume
rdO dr dz. Heat flowing into the element in the radial direction is
or
dQ, = - k dz rdO or
and out of the element in the radial direction,
dQ(r+dr) = - kdz(r + dr)dO :r(t + ;; dr)
Hence
ot o2 t
dQ, - dQ 1,+dr) = k dz dr dO or + k dz rdO or 2 dr (2.13)
neglecting a term of higher order.

r dQ(r+dr)
Fig. 2.2. CoNhu:tion itt tu1 element ofm~~teritll in cylindricfll coordiutes.
In a similar manner, the net heat flowing into the element in the
circumferential direction is found to be
iJ2t
dQ 11 - dQ111 +dlll = k dr dz , 2062 rd(J (2.14)
and, in the axial direction,

iJ2t
dQ: - dQ(:+dz) = k drrd6 oz 2 dz (2.15)
The rate of heat generation within the element is

q' rd6drdz
and the rate at which heat is being stored within the element is
at
rd6 dr dz pep ot
Then an energy balance for the element leads to the general dif-
ferential equation for heat flow in three dimensions in a cylindrical
coordinate system. i.e.,
at (o 2t 1 at 1 o2t o2t)
Ot = IX or 2 + Orr + r2 iJ(J 2 + oz 2 + pep
q'
(2.16)
This equation may similarly be simplified to suit any particular

problem. Steady one-dimensional heat flow in the radial direction
only will be considered in later chapters. With heat generation
within the material the equation is
2
( d -t +1-dt)
a q'
- +peP
- -- 0 (2.17)
dr2 r dr
and without heat generation,
d 2 t +! dt = 0 (2.18)
dr 2 r dr
PROBLEM
Show that the general equation of heat conduction in spherical coordinates
is given by
~= Jo 2 t + ~ ot + __1_ ~(sin ~) + 1 o2 t] + .!(_

Ot Lar 2 r ar r 2 sin 4> aq, 4> aq, r 2 sin 2 q, o82 pep
and transform the equation in rectangular coordinates (2.7) into spherical
coordinates by making the substitutions
x = r sin 4> cos 8
y = r sin 4> sin 8
z = rcos 4>
REFERENCES
l. Eckert, E. R. G., and Drake, R. M. Introduction to the Transfer of Heat
and Mass, 2nd ed., McGraw-Hill, New York (1959).
2. Carslaw, H. S., ·and Jaeger, J.C. Conduction of Heat in Solids, Oxford
University Press (1947).
3
One-dimensional steady state
conduction
The simplest example of steady state conduction in one dimension

is the transfer of heat through a single plane slab. Many simple
problems, such as conduction through the wall of a building,
approximate to this.
3.1 Conduction in Plane Slabs

To calculate the conduction rate in a single slab of isotropic invariable
thermal conductivity materiaL Fourier's law applied to an infinite-
simal layer within the slab, Fig. 3.1, may be integrated directly. Thus
dt
Q.., = - kA dx
qx
• X
tl
R=
tl
-, dt
~ j_
- ~
t
tz
q,. dx
XI
X
Xz
thermal condu ctivity, k

lC
Fig. J.l. One-dimensioiUII steady sttlte cond11ction in tl pltute slab.
16
ONE-DIMENSIONAL STEADY STATE CONDUCTION 17
and hence
- kA(t 2 - t 1)
Q =---- (3.1)
X (x2 - x.)
where (x 2 - x.) is the thickness of the slab and A is the area of the
slab. Using consistent units, the heat transfer calculated will be in
heat units per unit time.
The same result will be obtained if the appropriate differential
equation is integrated. Integration twice of equation (2.12),
gives
(3.2)
where C 1 and C 2 are constants of integration to be determined from
the boundary conditions, i.e., the temperatures at x 1 and x 2 •
Equation (3.2) indicates that the temperature variation through the
slab is linear. The temperature gradient from equation (3.2) used in
Fourier's law gives equation (3.1). Equation (3.1) may be re-written
as
Qx k(t1 - t2)
-A = qx = x - x ' (3.3)
2 1
in which form it may be compared with Ohm's law describing the

flow of electricity, i.e.,
. (VI - V2) Potential difference
Current density (i) = -(- - - ,
p x 2 - x.) Resistance of unit area
where p is the resistivity of the material, in units of ohms x length.
The heat flux q" is analogous to current density i; the temperature
drop (t 1 - t 2 ) is analogous to potential difference (V1 - V2 ); and
the resistance per unit area to heat transfer, (x 2 - x d/k, is analogous
to electrical resistance per unit area, p(x 2 - x 1 ). The usefulness of
this similarity will be made more apparent later.
Conduction through a system of plane slabs of different material
has often to be considered. A partition wall comprising two layers
of plaster board separated by a thickness of glass-fibre insulation,
or a furnace wall consisting of a layer of fire brick and a layer of
insulating brick, are typical examples. Further, such a system may
separate two fluids of different temperatures, when the actual wall
temperatures are not known. The processes of heat transfer between
the wall surfaces and the adjacent fluid are by convection and
radiation. Figure 3.2 shows such a system. The Newton equation
for convection may be written in the sign convention of equation (3.1).
Thus
(3.4)
In this equation, q. is the heat flux due to convection at the solid/fluid
interface, and tw is the wall temperature and tr the fluid temperature.
The region in the fluid where the temperature changes from tr to
tw is known as the boundary layer. h. is the convection coefficient
and is assumed known. Its determination forms the subject matter
of later chapters, where the suffix c is dropped.
ft UJ'dboun dary ayer
\
ftu;~
--, t
tfl I
Ifluid 2
·~ q. I qc + q, = q"
l x~'>
' I
x2 \:
tf2
kl k2 X31
Fig. 3.2. A nudtipk pltute slllb separating two jblids, one-dimemioiUII stetuly
state colfliMctio11.
As a convenience, the radiation exchange between the wall and
fluid or some other surface beyond the fluid may be expressed by an
analogous equation
(3.5)
Since radiation exchanges are a function of the fourth power of the
absolute temperatures involved, the radiation coefficient ~ is
heavily temperature-dependent.
The total heat flow or conduction flux q" from the wall by convec-
tion and radiation is found by adding (3.4) and (3.5):
q" = q. + qr = - hc(tf - tw) - hR(tr - tw)
(3.6)
A multiple slab of two layers of conductivities k 1 and k2 which
separates two fluids f1 and f2 at temperatures tfl and tr 2 is now
considered. For exchange between fluid f 1 and wall surface at t 1 :
qx = -(he+ hR)fl(tl - tfl) (3.7)
For conduction through the two layers of material:
- kl(t2 - tt} - k2(t3- t2)
q" = x2 = (3.8)
- x1 x3 - x2
For exchange between the wall surface at t3 and the fluid f2 :
q" = - (h. + hR)r 2 (tr 2 - t 3) (3.9)
Re-arranging and adding :
q" [
1
+ hR)fl
+
x2 - x1
+
x3 - x2
+ -c:----:---
1 J
(h. kl k2 (h 0 + hR)f2
= - (tf2- ted
and hence
(3.10)
where
1 1 x2 - x1 x3 - x2 1
- = + + + (3.11)
U (he + hR)fl k1 k2 (he + hR)C2
1/U is the overall thermal resistance per unit area between fluids
and U is the overall heat transfer coefficient. The resistances to heat
flow due to convection and radiation act in parallel and the
resistances due to the conducting layers act in series. The heat flow is
calculated from (3.10), once the overall coefficient U is found from
(3.11), and interface temperatures follow from (3.7), (3.8), and (3.9).
Sometimes in composite structures slabs of differing thermal
conductivity are present as shown in Fig. 3.3. This situation may be
treated one-dimensionally provided it is assumed that the y-z faces
ofthe intermediate slabs have uniform temperatures. The total resis-
tance may be deduced by adding the intermediate resistances in
parallel before adding the others in series.
-,...-
A2 k2
-Q
AI A3
kl
k3 ks
A4 k4
I I I I
XI x2 XJ x4
II 12 13 14
II R2 14
Rl
R4
-Q
Fig. 3.3. Olle-dimeiiSiolllll at~tuly attlte colflblctioll ;, tl a~riea-ptll'tllkl
ayat~m of altlba; tz tUUl t3 tu~ •lliform temp~rat•rea ;, the y--z pl1111e.
EXAMPLE 3.1
The reduction of heat loss from buildings is of very great practical
and economic importance. The Chartered Institution of Building
Services Engineers in the CIBS Guide Book A, give values of overall
heat transfer coefficient (known as U values and expressed as
W/(m 2 K) and not kW/(m 2 K)) for various types ofwal~ window, and
roof. Some typical values are given in Appendix 3. Use the U values
given below to calculate the heat transfer rate through a house
structure in cases (i) and (ii). In both cases the wall area is 110m2 ,
wood-frame window total area 14m2 , upstairs ceiling area 36m 2 ,
environment temperature difference 21 oc.
Case (i) 335 mm solid brick wall, U = 1·5 W/(m 2 K); pitched
roof with felt, foil-backed board ceiling, U = 1·5 W/(m 2 K);
single-glazed windows, U = 4·3 W/(m 2 K).
Case (ii) 335 mm solid wall plus 30 mm foam board lining,
k = 0·026 W /(m K); pitched roof as before plus 50 mm glass-fibre
insulation U = 0·5; double-glazed windows, U = 2·5 W/(m 2 K).
Solution. For parallel heat flow through walls, windows and roof, in
case (i)
Q = 21(110 X 1·5 + 36 X 1·5 + 14 X 4·3) = 5860 W
= 5·86 kW
In case (ii), the thermal resistance of the insulated wall is the original
resistance plus the insulation resistance which equals 1/1·5 +
0·03/0·026 = 1·82.
New U value = 1/1·82 = 0·55
Q = 21(110 X 0·55 + 36 X 0·5 + 14 X 2·5) = 2380 W
= 2·38 kW
A saving of 3·48 kW is achieved. Actual heating requirements will be
greater than the figures calculated on account of air changes, and
some losses through the ground floor.
EXAMPLE 3.2
Heat transfer through a double-glazed window is an example of the

application of resistances in parallel and in series. A section of the
window together with the corresponding diagram of circuit resistances
is shown in Fig. 3.4. Convection and radiation coefficients act in
parallel on both inside and outside surfaces of the frame and glazing,
and the overall resistances of the frame and glazing are in parallel. The
glazing itself consists of three resistances in series. For a small width,
heat transfer across the air gap is by pure conduction. As the width
increases, some convection commences so that an optimum width of air
gap occurs. The convection effect is accounted for by an empirical term
in the air gap resistance, R 5 • The BASIC names of all variables may be
deduced from the listing, with the exception of TKF, TKA and TKG
which are the thermal conductivities of the frame, air gap and glass,
respectively. The BASIC program for this problem is listed below; vari-
ables in the data are given in answer to questions appearing. A sample
run is also shown. The variation of heat transfer as a function of air gap
is given in the table:
Air gap, mm Heat transfer, W
3 157·32
7 107·92
11 99·56
15 103·96
20 114·85
25 126·87
heA - -- ----heR
hRA ___ _-
----hRR
-heR
-hRR
Troom
conv. conv.
rad. rad.
Fig. 3.4. Diagrammatic reprelentation of a double-glazed window and the

equivalent thermal re1i1tance circuit.
BASIC Program Listing
10 PRitU,"PLEASF. TYPE CONVECT! Oil COEFF!C!EIITo AIR SIDE [II W/112 K:"
20 IUPUT HCA
30 PR [lllo"PLEAH TYPE RAD!AT!OII COrFF !CI EUT, AIR SIDE Ill W/H2 K:"
40 Ill PUT liP A
50 rRif~T,"PLEASE TYPE CONVECTION COHF!C!ENTo ROOM SIDE [II W/H2 K:"
60 IUPIJT IICR
70 PRIIITo"PLEASE TYPE RADIAT!O~ COF.FFICIEIITo ~001-1 SIDE Ill W/"2 K :••
81) Ill PUT IIRR
90 PAftiT,"PLEASE TYPE WINDOW Fr.AIH AREA, [~ 112: ..
1r.0 triPUT WF.\
110 PRINT,"rLEASE TYPE ~[liD OW FRAIIF. TIIICY.NESS, [II H:u
120 !IIPUT IHT
130 rR!IIIo"PLEASE TYPE Wir~DOU GLASS AP.EA, IU 112: ..
140 [II PUT \IGA
150 PRIUT,"PLEA$[ TYPE WI~DOW GLASS TH!CKUESS, I~ MM:"
160 !UPHT W~T
165 UGT•WGT/10no.n
170 PRI~To"PLEASE TYP~ AIR GAP TH!CK"E5$o IH MM:"
180 IHPUT AGT
1r5 AGT•AGT/1000.0
19(\ TKF•0.166
1?5 TKA•O.O?~
200 TKG•0.76?
205 PRINTo"PLEASE TYPE ROOM TEMPERATURF IN OCG C:"
210 IUPUT TROQn
215 rni~T,"PL~A5E TYPE E~TER~AL AIR T~MPEQATURE IN DEG C:"
220 IUPUT TAIR
23(' r1=1.0/(IIH*(HCA+HRA))
240 P2=WFT/cwrA•Tr.r>
250 R3•1.0/(WFA•(HCR+HRR))
260 ~4=1.0/CWGA•CHCA+~RA))
270 R5•2.0•UGT/CWGA•TKGl+1.0/(T~A•WGA/AGT+(AGT-0.007l•350.0l
280 ~6•1.0/CWGA•(HCR+HRR))
290 RF•P1+R2+R3
300 r.W•R4+P5+R6
31C HOV•RF•RH/(RU+Rf)
320 QT•(TROOM-TAIR)/ROV
330 Of•(TROOM-TAIRl/RF
340 QW•(TROOM-TAIRl/RW
350 T1•TROO"-(R2+R3l•(TROOM-TAIRl/r.r
360 T2•TROOM-R3•CTROOM-TA!R)/~F
370 T3•TROO"-(R5oR6)•(TROOM-TA!P)/RW
380 T4•TROO"-R6•(TROOH-TAINI/R~
390 PRINT," RESULTS FOR AlP ~AP T~!CKNESS:"oAGT
400 PA!NTo"T,TAL HEAT THAIISFEP RATC 1:1 W:"oQT
410 rRUIT,''IIEAT TRI\tiSFER RAT£, FRAr"!F., IU ·.~:",QF
420 rAtfiT,''II~AT TR4fiSFER RATE, WifiDOW, I~ W:",OW
43~ rRJ"T,"OUT5tP( GLASS TEHPERATIJR~, I~ DEG C:",J3
440 rRIHT,"I:ISIDE GLASS TEMP(nATUR(, Ill DEG C:'',T4
450 PRINTo"QUTS!DE FRA~E TEMPERATURCo !U DEG C:"oT1
460 PRINT,"l~S!D[ fRAME TE~PENATURCo I~ PEG C:"oT2
470 PRI~T," DO YOU WISH TO PO A RfPEAT CALCULATIO~? 1-YESo2-NO:"
480 INPUT IX:
490 IFC!X~<2>GO TO 10
500 STOP
Sample Result from Listing
PLEASE TYPE CONVECTION COEFF!CIEIIT, AIR SlOE IN II/HZ K:

•12.0
PLEASE TYPE q-OIAT!OII COEFF I C I EIIT, AIR SlOE IN W/M2 K:
•5.0
PLEASE TYPE CON'IECT ION COEFFICIEIIT, ROOn SlOE HI W/H2 K:
24 .o
PLEASE TYPE RAOIATIO'I COEFFICIENT, ROOI·l SIDE IN \1012 K:
=9.0
PLEASE TYPE YIN-OW F RAilE AREA, trl 1·12:
=0.4
PLEASE TYPE YIll DOW FP.M1E THICKIIESS, IN 11:
=0,12
PLEASE TYPE WHID011 GLASS ARF.A, ItI M2:
=1.6
PLEASE TYPE WINDOII GLASS TIIICKNESS, IN rm:
=4.0
PLEASE TYPE AIR GAP THICKNESS, Ill Mn:
=25,0
PLEASE TYPE ROOn TEMPERATURE IN OEG C:
=25. 0
PLEASE TYPE EXTERNAL AlP. TEMPERATURE IN DEG C:
=0.0
;ESULTS FOR AIR GAP TIIICKIIESS:0.(}25
TOTAL HEAT TRAII5FER RATE Ill 11:126,9
HEAT TRANSFER RATE, FRAME, Ill 11:11,6
HEAT TRAIISFER RATE, IIIIIDOW, Ill 11:115.?
OUTSIDE GLASS TEMPERATUQE, Ill OEG C: 4,2
INSIDE GLASS TEMPERATURE, IN DEG C:19.5
OUTSIDE FRAME TEMPERATURE, Ill OEG C: 1,7
!~SlOE FRA~E TEMPERATURE, Ill OEG C:22.8
00 YOU WISH TO 00 A REPEAT CALCULATIOII? 1-YES,Z-110:
=2
3.2 Effect of a Variable Conductivity in a Plane Slab

In considering the variation of k with temperature in the case of
one-dimensional flow in a plane slab, equation (2.3) for the relation-
ship between k and temperature will be used.
For conduction in a single plane slab,
dt
qx = - ko(l + cct)-
dx
and
(3.12)
It will be found that equation (3.12) can also be obtained by taking

an average ofthe conductivities at temperatures t 2 and t 1 and substi-
tuting into equation (3.3). Equation (3.12) may be used to find the
interface temperature between two plane slabs, e.g., for two materials
where k 11 = k 10(1 + tXt) and k 2 , = k20(1 + {Jt). The heat flux
through both slabs is the same, hence
- k 10 [2 + 1X(t2 + t 1)](t2 - t 1) _ - k20 [2 + fJ(t 3 + t 2)](t 3 - t 2)
2(x 2 - xt) - 2(x 3 - x2)
This equation may be solved to find t 2 , and then qx may be calcu-

lated.
EXAMPLE 3.3
The heat flux through a plane slab 0·1 m thick is 146kW/m2 for
surface temperatures of 120° and 30°C. Find the value and sign of
IX in the thermal conductivity function given that k0 = 0·16 kW/(mK)
Solution. Using equation (3.12)

146 = -0·16 [2 + tX(30 + 120)](30- 120)/(2 X 0·1)
2 + 150 IX = 146 X 0·2/(0·16 X 90)
= 2·03
:. IX= +0·03/150 = +2·0 x w- 4 K- 1
3.3 Radial Conduction in Cylindrical Layers

Conduction througp thick walled pipes is a common heat transfer
problem, and may be treated one-dimensionally if surface tempera-
tures are uniform. The heat flow is then in the radial direction only.
Fig. 3.5 illustrates the situation for a single layer. Fourier's law may be
applied to a cylindrical layer at radius r:
dt
Q, = - kA dr
Here A is the surface area at the radius r, and obviously A will vary
between the inner and outer radii. It is therefore convenient to
consider a unit length of cylinder, when the radial heat transfer is
dt
Q, = - k(2nr) dr (3.13)
2nr is the area per unit length. Since the same quantity Q, is flowing
through a steadily increasing cylindrical area, the temperature
gradient must decrease with increasing radius. Integrating:
Q = - 2nk (t 2 - td (3.14)
r In r2/r I
By analogy with Ohm's law, the thermal resistance per unit length
of cylinder in this case is [In (r 2/r 1)]/2nk.
-Q,
• c=:J •
ft In r21 'I t2
R=--
2nk
Fig. 3.5. Steady state radial conduction in a cylindrical layer.
Equation (3.14) may also be derived from the general equation

for the cylindrical coordinate system (2.16) which simplifies to
d2t + ! dt = 0 ((2.18))
dr2 r dr
for the case of steady radial conduction in the absence of internal

heat generation. This equation may be integrated to give
(3.15)
where C 3 and C 4 are constants of integration to be found from boun-
dary conditions. Thus, if t = t 1 at r = r 1 and t = t 2 at r = r2 , it
is found that
(t 2 -td 1 r
t= n-+t1 (3.16)
In r2 /r 1 r1
To obtain equation (3.14), the temperature gradient is found by
differentiating (3.16) and substituting back in (3.13).
A thick walled steam pipe with lagging is a familiar example of
multiple cylindrical layers, and the treatment is similar to the
multiple plane layer. Fig. 3.6 shows two cylindrical layers separat-
1 In r2!r1 In r3!r2 1
2nrlhcl 2nkl 2nk2 2n r3 hc 2
Fig. 3.6. Steady state radial conduction in concentric cylinders separating
two fluids.
ing two fluids f1 and f2 . It is assumed that heat transfer at the
surfaces is in each case predominantly by convection. Considering
unit length, at the inside surface:
Q, = - 2nr 1 hct(t 1 - tfl)
The same quantity is conducted through the two layers, hence
Q, = - 2nk (t2 - _!.!2 = - 2nk2 (t3 - t2)
1 In r /r In r 3 /r 2
2 1
It is also convected from the outside surface, so
Re-arranging and adding these equations gives:
or
Q, = - U(tr2 - tn) (3.17)
where
1 1 In r 2/r 1 In r 3 /r 2
U = 2nr 1 hct + 2nk 1 + 2nk + 2nr hc
2 3 2
U is the overall heat transfer coefficient per unit length between the
two fluids and 1/U is the thermal resistance of unit length and consists
of the sum of individual thermal resistances to conducted and
convected heat flow.
3.4 Critical Thickness of Insulation

Closer inspection of the thermal resistance per unit length of a
cylindrical system, given in (3.17), shows that the external convec-
tion resistance decreases with increasing radius, so that the sum of
the conduction and convection resistances of a layer of insulation will
at some radius pass through a minimum value. For a layer of insu-
lation having internal and external radii of r; and ro, a thermal
conductivity k, and an external convection coefficient h, the thermal
resistance per unit length will be (l/2nr0 h) +((In r0 /r;)/2nk) and this
will have a minimum value obtained by putting
or
k = CritlCa1 rad"IUS
ro = h 0 0
(3.18)
This is the value of outer radius for which heat transfer through the
system will be a maximum. It follows that if for a given k and h the
external radius is less than k/h, then increasing the thickness of
insulation up to a radius of k/h will increase, rather than decrease
the heat loss from the system. The situation is likely to arise if k has
a relatively high value, and h a relatively low value.
3.5 Radial Conduction in Spherical Layers

Another simple instance of one-dimensional conduction is that
which can occur in a spherical layer. Conduction will be only in
the radial direction if the temperatures of the two spherical surfaces
are uniform. The radial conduction is given by:
(3.19)
and the overall heat transfer coefficient for a double spherical layer
separating two fluids f 1 and f2 is
1 1 r2 - r1 1 r3 - r2 1 )
1/( 2
4nr1hc1
+ 4nk1
-- 0
r1r2
+ 4n
-k-
2
0
r2r3
+ 4nr32h c2
The value of the critical r 0 for a sphere is 2k/h.
EXAMPLE 3.4
In this program the critical radius of an insulated pipe may be investiga-

ted. The internal radius of the insulation Rl, and the internal and
external temperatures T1 and TA may be regarded as fixed in any one
problem, and the effects of the insulation thickness TH, insulation
thermal conductivity TK, and the outer convection coefficient H, may
be investigated.
The BASIC program list and sample results are given below. In the
example Rl = 0.1 m, TK = 0.85 W/m K, and H = 5.0 W/m 2 K. This
gives a critical outer radius of 0.17 m and, for insulation thicknesses
between 25 and 200 mm, the heat transfer variation is given in the
following table:
Insulation thickness, mm Outer radius, m Heat transfer, W/m
25 0·125 0·607
40 0·140 0·620
55 0·155 0·626
70 0·170 0·628
90 0·190 0·625
200 0·300 0·577
10 R1=0.1
20 T1=200.0
30 TA=20.~
40 PRINT,"PLEASE TYPE INSULATIOH THICKUESSo MM:"
50 l:tPUT T•f
60 R2=TH/1~00.0+R1
70 PRI~To"PLEASE TYPE IHSULATIOH THERMAL CONDUCTIVITY, W/M DEG C:"
80 llfPU T TK
90 PRIHTo"PLEASE TYPE OUTER RADIUS COHVECTIOU COEFFICIENT, W/M2 DEG C:"
100 llfPUT H
11~ RCRIT•TK/H
120 RESIS=1.0/(6.2832•R2•Hl+ALOG<R2/ R1l/(6.2832•TK)
13U QfLOW•(T1-TAl/RESIS
1~0 PRII~T,''RESULTS FOR Atl OUTER RADIUS OF, H:'',R2
150 PRI~T,"CRITICAL RADIUS fOR THE COEFFICIEIHS USED:"oRCRIT
160 PRINT,~tiEAT FLO~, U/11 PIPE LEHGTII:",OFLOU
170 PRIUT,"OD YOU UISH TO DO A REPEAT QUH! 1-YES; 2-HO"
180 IriPUT IX::
190 lf(IX~<2l&O TO 40
200 STOP
PLEASE TYPE I~SULATIOH THICKNESS, MM:

=25.0
PLEASE TYPE INSULATION THERMAL CONDUCTIVITY, W/M DEG C:
=0.85
PLEASE TYPE OUTER RADIUS CONVECTION COEFFICIENT, W/MZ DEG C:
=5.0
RESULTS FOR AN OUT~R RADIUS Of, M: 0.12500000E 00
CRITICAL RADIUS FOR TilE COEFFICIEIITS USED: 0.17000000E 00
HEAT fLOWo W/M PIPE LENGTH: 0.60722832E OJ
DO YOU WISH TO DO A REPEAT RUII! 1-YES; 2-HO
•1
PLEASE TYPE INSULATION THICKNESSo MH:
=70.0
PLEASE TYPE INSULATION THERMAL CO~DUCTIVITYo W/M DEG C:

•0. 8S
PLEASE TYPE OUTER RADIUS CONVECTION COEFFICIE~T, W/M2 DE~ C:
=S.O
RESULTS FOR AN OUTER RADIUS OFo H: 0.17000000E 00
CRITICAL RADIUS FOR T~E COEFFICIENTS USED: 0.17000000E 00
HEAT FLOWo W/M PIPE LENGTH: 0.62806211E 03
.,
DO YOU WISH TO 00 A REPEAT RUN? 1-YES; 2-NO
PLEASE TYPE l~SULAT!Otl THICKNESS, ~IH:

•200.0
PLEASE TYPE INSULATION THERMAL CONDUCTIVITY, 11/11 DEG c:
a0.8S
PLEASE TYPE OUTER RADIUS CONVECTION COEFFIC!Etll, 111M2 DEG c:
•S.O
RESULTS FOR AH OUTER RADIUS OF, M: 0.3000QOOQE 00
CRITICAL RADIUS FOR T~E COEFFICIENTS USED: 0.17000000E 00
HEAT FLOW, W/M PIPE LENGTH: O.S7727842E 03
DO YOU WISH TO DO A REPEAT RUN? I-YES; 2-NO
•2
3.6 Conduction with Heat Sources

The flow of electricity in a material gives rise to ohmic heating
and, generally, the resulting heat flow is at least two-dimensional.
However, if the flow of current in a flat wide bar, or the heating of
a flat plate by eddy currents is being considered, then the heat flow
is essentially one-dimensional if edge effects are neglected. (See
Fig. 3.7. where 1 >b).
The general equation for the rectangular coordinate system, when
applied to this problem, reduces to
d2 t )
(
O=IX- +q'- ((2.10))
dx2 peP
Assuming q' uniform in space, equation (2.10) is integrated to give
(3.20)
where C 5 and C 6 are constants of integration to be determined

from boundary conditions.
If the boundary condition is convection to a known fluid tempera-
ture, so that the solid boundary temperatures, t, are unknown, two
further equations are obtained by equating conduction to convection at
the boundary, for example
h(t
a x1
-t)=-(-k(dt))
a dx x,
(3.21)
32 ENGINEERI NG HEAT TRANSFER
and
hb (t -t
x2 a
)=- k(dt)
dx ~
(3.22)
where ha and ta apply at face x 1 and hb and tb apply at face x 2 • Four

equations are now available to give C5 , C6 and tx 1 and tx 2 •
Fig. 3. 7. One-dimensional conduction in a plane layer with internal heat

generation.
The heat transfer at any plane, x, is obtained by differentiating

equation (3.20) and applying Fourier's law. Thus
q = -
X
k(~)
dx X
(3.23)
If dt/dx = 0, the temperature is a maximum and the heat flux is

zero. Thus if one face of the slab is insulated, it will also be the hottest.
The maximum temperature is found by putting the value of x at
which dt/dx = 0 into equation (3.20).
One-dimensional conduction in the radial direction will occur in
a rod or hollow cylindrical bar if surface temperatures are uniform.
The maximum temperature will occur at the centre of a rod, and at
an intermediate radius in a hollow bar if both surfaces are cooled.
The general equation in cylindrical coordinates (2.16) reduces to
d2t
ex (- + -1 -dt) + -q' = 0 ((2.17))
dr 2 r dr peP
for this situation. The solution of this is

q'r2
t = - 4k + C 7 In r + C 8 (3.24)
which may be obtained by making the substitution dt/dr = p.

Values of C 7 and C 8 , the constants of integration, may be found by
substituting the known boundary conditions, see Fig. 3.8. The value
r
Fig. 3.8. Radial conduction in a cylindrical layer with internal heat generation.
of rat which dt/dr = 0 gives the position of the maximum tempera-
ture, and this substituted in (3.24) gives the value of the maximum
temperature.
The following example illustrates the way in which ohmic heating
problems may be solved.
EXAMPLE 3.5
An internally cooled copper conductor of 4 em outer diameter and
1·5 em inner diameter carries a current density of 5000 amp/cm 2 •
The temperature of the inner surface is maintained at 70°C, and it
may be assumed that no heat transfer takes place through insulation
surrounding the copper. Determine the equation for temperature
distribution through the copper, hence find the maximum tempera-
ture of the copper, the radius at which it occurs, and the heat
transfer rate internally. Check that this is equal to the total energy
generation in the conductor. For copper, take k = 0·38 kW/(mK)
and the resistivity p = 2 x 10- 11 ohm metre.
Solution. If i is the current density,
q' = pi 2 = 2 X 10- 11 X (5000 X 104 ) 2 kWjm 3
= 5 x 104 kW/m 3
The boundary conditions are that at r = 0·75 em, t = 70°C and
that at r = 2 em, dt/dr = 0. This is because the heat transfer is zero
at r = 2 em. It follows that the maximum temperature also occurs
at r = 2 em. The constants of integration in equation (3.24) may
now be found.
dt q'r C7
-= --+-=0 at
dr 2k r
5 X 104 X 0·()2 + C7 = O
2 X 0·38 0·02
- 1318 + C 7 j0{)2 = 0
c, = 26·3
C 8 is given by
70 = - 45 xX ()-10
4
(0·75) 2 + 26·3ln (Q-0075) + C 8
38 x 100
-1·85- 128·7 + C8
C 8 = 200·6
The equation for temperature is therefore:
t = - 32,900r2 + 26· 3 In r + 200·6
with r in metres.
The maximum temperature occurs at the outer radius. Substituting
r = 0·02 m in the above equation gives
t max= - 13·17 - I 02·8 + 200·6

= 84·6°C
To calculate the heat transfer rate internally, it is first necessary to
find the temperature gradient at r = 0·0075 m. Thus
- 5 X J0 4 X 0·0075
------~--
26·3
+- -
2 X 0·38 0·0075
- 494 + 3510
+ 3016
The heat transfer internally is in the direction of negative radius.
hence
Q(-r) = - (-kA:~)
+ 0·38 X (2n X 0·0075) X 3016
53·9 kW/m length
This result may be checked since all the heat generated m the
conductor must be dissipated internally.
Q(-rl = (volume/m length) x q'
= n(0·02 2 - 0·0075 2 ) X 5 X \0 4
= 53·9 kW/m length.
PROBLEMS
I. The walls of a refrigerator for a shop consist of slag wool 0·1522 m thick
sandwiched between sheet iron. 0·0794 em thick, on one side and asbestos
board, 0·953 em thick, on the other. The total surface effective for heat transfer
is 37·2 m 2 The atmospheric temperature is 18·3°C and the temperature in
the cold room is - 3·9oC.
The thermal conductivity of iron, slag wooL and asbestos board may be
taken as 69·1, 0·346, 1·21 x 10- 3 respectively and the surface heat transfer
coefficient as 1·705 x 10- 3 ; in kW, m, K, units.
Compute the heat leakage into the refrigerator. (Ans. 0·51 kW.) (King's
College, London}.
2. A spherical container 1·22 m internal diameter is made of sheet metal of
negligible thermal resistance and covered by cork insulation 0·457 m thick.
The interior contains a liquefied gas at - 62·2°C for which a surface heat
transfer coefficient of 1·06 k W/(m 2 K) may be considered to apply. The
atmospheric temperature is 18·3oC. Moisture vapour permeates the cork and
freezes at a suitable position to form an ice barrier. The mean surface co-
efficient for the outside may be regarded as 0·021 kW/(m 2 K). Calculate the
thickness of the ice assuming that the conduction characteristics of the cork
remain constant throughout.
Assume the thermal conductivity for cork is 43·2 X 10- 6 kW/(m K). (Ans.
Q-305 m.) (Queen Mary College, London).
3. A 30 mm diameter pipe at 100oc is losing heat by natural convection to
the atmosphere at 20°C at the rate of 0·1 k W per m length. It is required to
cut down this loss to 0·05 k Wjm. Two insulating materials A and Bare avail-
able. There is sufficient of A to use it at the rate of 3·14 x w- 3 m 3 jm length,
and ofB to use it at the rate of 4·0 x 10- 3 m 3 frn. The thermal conductivities
of A and B are 0·005 and Q-001 kW/(m K) respectively. Js it possible to
achieve the required degree of insulation? Assume the convection coefficient
applicableto the bare pipe is also applicable to the outer surface of insulation.
(Ans. B inside 0·0437 kWjm; A inside 0·0742 kWjm.) (The City University)
4. Calculate the surface temperature and the maximum temperature of a

10 mm diameter steel conductor carrying 5000 amps and forced convection
cooled to the atmosphere at 15°C with a convection coefficient of 5·55 kW/
(m 2 K). For the conductor, take the electrical resistivity as 8 x w- 8 ohm m,
and the thermal conductivity as 0·12 kW/(m K). (Ans. 161·3°C and 178·2°C.)
5. (i) Define the term thermal resistance and show that, when heat flows
through a number of individual resistances in series, the overall resistance is
equal to the sum of the individual resistances.
(ii) A double-glazed window consists of two sheets of glass separated by a
gap. The gap is filled with a gas, but is sufficiently thin to prevent convection
between the two sheets of glass. The area of the window in elevation is A,
the thickness of each sheet of glass is x and the thickness of the gap is y. The
thermal conductivities of the glass and of the gas in the gap are kx and ky
respectively. The surface heat-transfer coefficients inside and outside the
building are h 1 and h 2 respectively; the corresponding air temperatures are
t 1 and t 2 . Neglecting radiation, obtain an expression for the heat transfer
rate q, in terms of A, x, y, kx, ky, h 1 , h 2 , t 1 , and t 2 .
(iii) Find the percentage reduction in heat loss when a single-glazed window
is replaced by a double-glazed window. Assume that the values of A, x, k.,
h 1 , h 2 , t 1 , and t 2 are the same for both windows, the symbols having the same
meaning as in section (ii). Numerical data:
x = 0·318cm; y = 0·635cm;
kx = 865 X 10- 6 kW/(m K) k, = 26 X 10- 6 kW/(m K)
h 1 = 8·52 x 10 3 kW/(m 2 K) h 2 = 14·2 x 10- 3 kW/(m 2 K)
(Ans. 56·5 per cent) (Imperial College, London).
6. The inner surface of a 0·23 m furnace wall is at 800°C. The outer surface
convects to the atmosphere at 21 oc, with a coefficient of 0·012 kW/(m 2 K).
The conductivity of the furnace wall is 870 x 10- 6 k W/(m K). To cut down
heat loss, an additional wall 0·23 m thick of insulating brick is added on the
outside. having a conductivity of 260 x 10- 6 kW/(m K). For the same outer
surface coefficient, calculate the percentage reduction of heat loss, the brick
interface temperature, and the brick outer surface temperature. (Ans. 7H %.
633°C, 73°c.)
7. A wide copper strip 10 mm thick carries an electric current of density
50 amp/mm 2 . The heat generated is dissipated from the two wide faces
of the strip by convection. On the left-hand face the convection co-
efficient is 5 kW/m 2 K and on the right it is 10 kW/m 2 K. The surround-
ings are at 25°C.
You may assume for one-dimensional conduction with heat genera-
tion, that
where the symbols have their usual meanings.

Calculate:
(i) The constants of integration in the temperature distribution equa-
tion for the copper strip,
(ii) The temperatures of the two faces of the copper strip.
(iii) The heat fluxes (kW /m 2 ) at the two faces, and show the sum of the
fluxes equals the heat generation in the volume of copper per unit
area of wide face.
For copper, p = 2 x 10- 11 ohm metre, k = 0·38 kW/(m K).
(Ans. (i) 456·32K/m; (ii) 59·67°C and 57·66°C. (iii) 173·4 and 326·6
kW/m 2 .)
8. The walls of a house consist of two skins of brick each 115 mm thick,
separated by an air gap of 50 mm. The heat transfer coefficients on the
outside and inside walls of the house, respectively, are:
outside inside
by convection 7 W/m 2 K; by convection 3 W/m 2 K
and
by radiation 5 W/m 2 K; by radiation 4 W/m 2 K
Across the air gap, the overall heat transfer coefficient is 4 Wfm 2 K,
taking into account conduction, radiation and convection effects.
To reduce heat loss the air gap is filled with insulation, so that heat
transfer across this region is now by conduction only.
(a) For a wall area of 70m 2 calculate, for the uninsulated wall, the
annual heat loss in kWh, assuming an annual mean ambient temper-
ature of l2°C and a mean internal temperature of 22° C maintained
24 hours a day.
(b) Calculate the percentage reduction in annual heat loss across the
walls by the insertion of insulation, of thermal conductivity 0·03 2
W/mK.
The thermal conductivity of brick is 0·81 W/m K.
(Ans. (a) 8066·3 kWh. (b) 63·32 per cent.)
9. A hot water storage cylinder consists of a copper cylinder with hemi-

spherical ends. The cylindrical section is 0·4 m diameter by 0·6 m long,
and the hemispherical ends have a radius of 0·2 m, so that the overall
length is 1·0 m. The copper surface has a temperature of 60°C, and
there is a surface heat transfer coefficient of 5·4 W/m 2 K to the ambient
and surroundings at 18°C.
To reduce heat loss, 100 mm of insulation is added to the whole
surface, for which the thermal conductivity is 0·04 W/m K. The surface
coefficient of 5-4 W/m~ K continues to act now on the outside surface
of the insulation.
(i) Calculate the original heat loss rate for the whole tank and the per-
centage reduction in heat loss by adding the insulation.
(ii) Calculate the temperature on the outside surface of the insulation,
both on the cylindrical surface and the hemispherical surface. Why
are the two values not the same?
You may assume that the thermal resistance of a spherical layer of radii
r 1 and r 2 and of thermal conductivity k is given by (r2 -r 1 )/( 47Tkr 1 r 2 ).
(Ans. (i) 285 W; 90·6 per cent reduction. (ii) Sides 20-41°C; ends
l9·98°C.)
4
Two-dimensional steady state
conduction
It is important to realize that in many cases a conduction problem

is over simplified by the use of one-dimensional treatment, which
means the neglect of edge and corner effects which must be present
in any finite object. The error involved in this neglect will depend
on the dimensions of the system. Consider. for example. the wall of
a building some 6 m long and 200 mm thick. In the absence of doors
and windows. conduction through such a wall will be one-dimen-
sional over the greater part of the 6 m length and the error involved
in neglecting the corner effects will not be great. In contrast. conduc-
tion through a chimney, say, 300 mm square internally and 1 m square
externally, is essentially two-dimensional. Again a simplifying
assumption is being made, since near the base and top of the chimney
conduction will be three-dimensional. Thus those problems will be
considered in this chapter which may be assumed to be two-
dimensional without introducing significant error. This will cover
the majority of heat conduction problems which are sufficiently simple
to include in an introductory text.
Two-dimensional problems in rectangular coordinates only are
to be considered. The two equations, with and without heat genera-
tion, are:
o( iJ2t iJ2t)
:-+- +-=0
q'
((2.9))
ox 2 oy 2 pep
iJ2t iJ2t
((2.11))
ox2 + oy2 = 0
Solutions to these equations are, of course, possible, but the more

readily obtained ones depend on the choice of somewhat unrepre-
sentative configurations or boundary conditions. As an alternative,
39
therefore, numerical procedures will be described. These have an
advantage of being applicable to any two-dimensional shape. The
first method, involving simple arithmetic, is suited to problems
having specified boundary temperatures. The second method, which
can consider more general and complicated boundary conditions,
involves the use of a digital computer.
The main difference between the analytical solution and
numerical methods is that the former will give an equation from
which the temperature may be obtained anywhere in the solid,
whereas the latter will give values of temperature at chosen specific
points only. The accuracy will depend on how close together are
the chosen points; however, many points will entail much more
work than a few.
4.1 A Numerical Solution of Two-dimensional Conduction

A numerical method involving a process known as relaxation 1 will
be introduced by consideration of a typical example of two-
dimensional conduction, the right-angled comer. The method is
suited to simple problems involving only a few specific points. For
conduction fields involving many points. some elementary computing
procedures are considered in Section 4.2.
A right-angled comer, forming part of a square hollow section,
such as a chimney, is shown in Fig. 4.1. For boundary temperatures
uniform on the inside and outside surfaces, a one-eighth unique part
exists as shown, involving only 11 mesh points in this example.
It is necessary to replace the differential equation (2.11) by finite
difference approximations relating temperatures around a mesh
point, and this is possible if temperatures vary continuously in the
x- and y- directions. expressible as t = f(x) and t = f(y~ Using
MacLaurin's series, the temperatures at points 1 and 3 may be
expressed in terms of t0 at point 0, the differential coefficients of
t = f(x) at x = 0 at point 0, and the mesh size a. Thus:
(4.2)
TWO-DIMENSIONAL STEADY STATE CONDUCTION 41
y
t specified at boundary
m'
I t specified at
boundary
'"""-~ ,. 1/s th unique part

-+-+--+-~----~lines of
symmetry
Fig. 4.1. Construction for a numerical solution oftwo-dimensional conduction,

in a qlllll'ter of a hoUow sqiUlre section.
neglecting higher powers. When added together these give
or
(4.3)
Writing similar equations for t 2 at y = +a and t 4 at y = -a, it is

possible to obtain in like manner
iPt) _ t2 + t 4 - 2t0
( iJy2 14.4)
o- az
Both (4.3) and (4.4) contain discretization errors involving terms
containing fourth and higher powers of a. Equation (2.11) may thus
be replaced by
ot) + (ooy2t)
(ox2
2
0
2
0
= tl + tz + tJ + t4 -
a2
4to = 0 (4.5)
or simply
(4.6)
With heat generation equation (2.9) would be replaced by

t1 + t2 + t3 + t 4 + a 2 q'/k - 4t0 = 0 (4.7)
Equations of the form of (4.6) or (4.7) may be written for every

mesh point in the field and the problem is then to obtain a solution of
the set of simultaneous equations. In hand calculations, the relaxation
process is used. Guesses are made of temperatures at the points,
bearing in mind the boundary temperatures and the proximity to the
boundaries. On substitution of the guessed temperatures into (4.6)
or (4.n a solution other than 0 is likely to be obtained. This is known
as the residual. When all mesh points have their correct temperatures,
the residuals will be zero. To eliminate a residual of + r or - r, it will
be seen that t0 must be increased by + r/4 or - r/4 respectively.
Carrying out this operation will alter the residuals at surrounding
field points by the same amount, i.e., +r/4 or -r/4. Care must be
exercised around lines of symmetry. Thus tm = tm' in Fig. 4.1, and
eliminating a residual of ±r at m will alter the residual at n by
± r/2 and not by ± r/4. Eliminating a residual at n will have the normal
effect on the residual at m and m'.
To carry out the relaxation process, the initial residuals due to
guessed temperatures are calculated, then the residuals are elimi-
nated one by one starting with the largest Subsequent operations
may re-introduce residuals previously eliminated, so that some points
must be treated more than once. The process generally continues
until residuals are ±2 or smaller, indicating temperatures are within
to of their correct value. The magnitude of error arising depends
on the overall temperature range involved.
To illustrate a step in the procedure, let t 4 = 120° (boundary value),
t0 = 80°, t 2 = 40°, t 1 = 100° and t 3 = 70° as original guesses in
Fig. 4.1. Then equation (4.6) gives: 100 + 40 + 70 + 120 - 320 =
+ 10. The residual of + 10 could be reduced to -2 by adding 3°
onto t0 to give 83°. Residuals at points 1, 2 and 3 are then increased
by +3. H heat generation is present the constant term a 2 q'fk is
included in the equation. It does not enter into the relaxation proce-
dure, but it is seen to have the effect of raising the general level of
temperature.
Once the temperatures are known, heat conduction in the section
is obtained from a summation of conduction along the mesh lines.
Heat transfer through unit length of the whole section would be the
average of heat conduction in and out at the hotter and cooler
boundaries respectively. Now imagine point 0 to be at the centre of
a square of size a at a temperature t 0 • Similarly there are squares at
t 1' t 2 and t 3 , etc., and at the boundary there is a half-square at t 4 • Heat
conduction between 4 and 0 per unit length of section is given by
Q(4-0l = - k(a x 1)
(to - tJ = k(t 4 - t 0 )kW/m (4.8)
a
Total heat conduction at the boundary is a summation of sinrilar

terms.
When there is heat generation in the section, heat transfer out of the
section at point 4 is given by:
(4.9)
Similarly heat transfer into the section at point 4 is:
Q(4-0) - q'a 2 /2 = k(t 4 - t 0 ) - q'a 2 /2 (4.10)
These relationships arise from an energy balance on the boundary half

square, i.e., heat generated= net heat flow out.
EXAMPLE4.1
Establish the temperatures a-k in the duct shown, by relaxation, and
calculate the conduction heat transfer through the duct. k for the duct
is 0·1 kW/(mK).
Solution. Take initial guesses for the temperatures as: a = 50°,

b = 48°,C = 41°,d = 25°,e = 41°,f = 46°,g = 47°,h = 48°,j = 50°,
k = 50°. Residuals are calculated first, e.g., for point a: 80 + 20 +
48 + 48 - (4 x 50) = -4. The relaxation table with initial resi-
duals shown is as follows:
Mesh point: a b c d e f g h j k
Initial residual: -4 -1 +9 +22 +7 +4 +6 +5 -2 0
Operation:
+6atd=31° +15 -2 +13
+4 ate= 45° +3 -1 +2
+3ate=44° +5 +1 +7
+2 at f = 48° +3 -1 +8
+2 at g =49° +1 0 +7
+2 at h = 50° +2 -1 0
+1 at d = 32° 0 +1 +4
+1 ate= 45° +2 0 +2
-1 at a =49° 0 +2
+1 atf= 49° +1 -2 +3
+1 at g = 50° 0 +2 0 +2 +1 -1 -1 0 0 0
Final temperatures are: a = 49°, b = 48°, c = 45°, d = 32°, e = 45°,

f = 49°, g = 50°, h = 50°,j = 50°, k = 50°.
For the whole duct, heat conduction in = 4[fk(80-49) +
k(80-48) + k(S0-45) + k(80-45) + k(80-49) + k(80-50) + k(80-50)
+ k(80-50) + fk(80-50)] = 101·4kW/m.
Heat conduction out = 4[fk(49-20) + k(48-20) + k(45--20) +
2k(32-20) + k(45--20) + k(49-20) + k(50-20) + k(S0-20) + k(50-20)
+ fk(S0-20)] = 100·2 kW/m. The average figure is 100·8 kWfm.
d e f g h k
-i-
c
goo .. 1
L- . . ;.;o:~ l __ ----
Fig. Extlmple 4.1. Unique ptlrt ofrecltlllflllW tiMet.
4.2 Iterative Computing Procedures for Two-dimensional Steady

State Conduction
The relaxation procedure is suitable for fields of few mesh points, and
is most easily performed by hand. In an iterative process, it is the new
value of a temperature calculated at a mesh point, in comparison with
the old value, that is of interest, rather than the residual obtained at the
point when the temperature equation is solved. For small fields, indivi-
dual programs may be written for each problem, as in Example 4.2.
EXAMPLE4.2
The figure shows a hollow square section, with the centre hollow rotated
through 45°. A square mesh is superimposed so that temperatures in the
section may be determined. Then by considering heat flow along mesh
lines terminating at the boundaries, the heat flow into and out of the
section is found, per unit length. A simple program in BASIC is given to
carry out this task. The principles involved may be deduced by studying
the listing. All points are solved to form one complete iteration, and
iterations continue until all points have changed in temperature by less
than 0·1 °C since the previous iteration. In this example this is achieved
in 9 iterations. The average heat flow for a temperature drop of 80°C is
1510·54W/m, so that the actual surface values vary by 6·67 per cent.
This may be attributed to the coarse mesh with the inside surface set at
45° to the mesh lines.
2 3 4 5
""" "6 7 8 9 10
""" """ 11 12 13 14
"""
Fig. Example 4.2.

10 OIAENSIO~ T(16l,TT<16),0T(16l
20 TK•2.32
30 FOR I%•1 TO 16:READ T<IXl:NEXT 1%
40 oArA zo.o,zo.o.zo.o,zo.o,zo.o,so.o,so.o,so.o.so.o,so.o,so.o,so.o.
so &so.o,1oo.o,so.o,1oo.o
60 Nx~o
70 N%~117.+1
80 TT(6l•Z.O•!T(1l+T(7ll /4.0
90 Dl(1l~ABS(TT(6l-T(6ll
100 T!6l=TT(6l
110 TT(10l=!TC5l+TC14l+2 .0•T(9ll/4.0
120 DT!2l•A0S(TT(10)-T(10l l
130 T<10l=TTC101
140 TT(111•2.0•(f(7l+T!12 ll/4.0
150 DT(3l=AOS(TT(11l-T(11 ll
160 T( 11l=TT< 111
170 TT<15l=2.0•(T(12l+T(1 6ll".O
180 DT"l=AOS(TT(15l-T<15) )
190 T<15l•TT(15l
200 FOR 1%=7,9
210 TT(I%l=(T(I%-1l+T(I%+1l+T(I~-5l+T(I%+4ll/4.0
220 DT(I%-Zl=ABS(TT(J%1-T( !%)1
230 T(IXI=TT<!%1
235 NEXT 17.
240 FOR 17.=12,13
2 50 T T (IX l = ( T ( I %-1 ) +T ( I X+ 1 l + T ( 17.-4 l + T (IX+ 3 l l /4. 0
260 DT(J%-4l=ADS(TT(!%l-T (IXll
270 TC!Xl=TT<IXI
275 IIEXT 1:
280 LX=O
290 FQR U•1 ,9
300 IF(DT(l%l.LT.0.1lGO TO 320
310 LX=L%+1
320 NEXT IX
330 IF(L7..GT.OIGO TO 70
340 QIN=0.5•(T(14l-T(10l+T (14l-T(13ll•TK
350 QJN•QJII+(T(16l-T(15l+ T(161-T(13ll•TK
360 OIN=8.0•QIN
370 QOUT=O.O
380 FOR IX=1,4
390 QOUT=QOUT+(T(I7.+5l-T( IXII•TK
400 IIEXT IX
410 QOUT=OOUT+0.5•(T(10l- T(5ll•TK
420 QOUT=QOUT•8.0
430 PRINT," POINT TUIPERATURE"
435 FOR 1%=1,16
440 PRIIIT,II,T<I%1
445 NEXT U
450 PRINT," NUMUER OF RELAXATIONS",q
460 PRINT," HEAT TRANSFER IN",Q!N," WATTS/M LENGTH"
465 PRINT," HEAT TRANSFER OUT'',QOUT,'' YATTS/H LENGTIJ''
470 STOP
POl'~T TE11PERATURE
n.2n0r.ooone 02
0.20000DOOE 02
0.2oonnorJoE 02
0.znonoonoe oz
0.200fl0f!'l0E 02
6 0.26?1449~E 02
O.BSS2788E U2
r. 0.4090~473£ 02
~ n.,30~52H4E nz
,,
,n n. 5403'-!'\·J~e G?
0.477:.5155E 02
1~ 0.r>17~5517E 02
1~ C. 771,~7700E G2
,.,
14 0.10000UOOE 03
O.:HJSL.U663E 02
H fl.10000000E 03
tlUMUER OF ~ELAXATIOrJS
ll<AT TRAUSHW !II 0.14097312E 04 IIATTS/~1 LEt/GTII
IIEAT TRANSFER OUT 0.16113510E C4 WATTS/~ LENGTH
The simplest possible case has been considered so far, i.e. that of fixed
boundary temperatures. If these are uniform then the boundary is
said to be isothermal. What happens beyond the boundary to create
the isothermal condition is outside the scope of the problem, and
in this sense the exercise is rather unrealistic. A boundary which is
convecting or radiating, or perhaps is insulated, is the more practical
situation. Mesh points occurring on such boundaries will have
temperature relationships other than (4.6) or (4.7) for points in the
field Also, internal boundaries between different conducting
materials may exist With these and other complexities such as the
boundary shape, the work soon becomes too complicated to be
treated by relaxation methods. However, whatever these complexi-
ties may be, the problem always reduces to solving a set of simul-
taneous equations, and two computer-based methods are available.
The first is that of a direct solution using the Gaussian elimination
method 3 , and the second is an iterative solution. The essentials of the
second method will be described since one basic program with minor
changes may be applied to a wide range of problems.
4.2.1 Mesh Point Temperature Relationships for Boundary Points

The first step in preparing a computer program is to consider all
the different mesh point temperature relationships that will be
involved. Later these can be translated into a form suitable for the
program. The relationships to be used are based on the electrical
resistance analogue, a technique used extensively for conduction
problems prior to the availability of computers. It should be pointed
out that in some instances this method produces the same result as
the true finite difference relationship (e.g. equations (4.6) and ( 4.11)
below), but when a difference does occur the order of accuracy is
lower. For further discussion the reader is referred to Bayley, Owen,
and Tumer 2 •
To introduce the method involved, consider the field point shown
in Fig. 4.2. The square mesh is of size a, and the material has thermal
conductivity k. The resistances between the centres of squares 1 to 4
and square 0 are therefore equa~ and are af(k x a x 1) for unit
thickness of the field The conduction heat transfer (or 'current)
across this resistance, for temperature t 1 at point 1, and t0 at point 0
is therefore [ -(k x a x 1)/a](t0 - t 1) = k(t 1 - t 0 ). The summation
of heat transfers from all mesh points to point 0 must be zero in
steady state and hence
4
electrical equivalent
Fig. 4.2. Field node i11 two-dime11siolllll stetuly state co1Uhtctio11.
k(t 1 - t0) + k(t 2 - t 0 ) + k(t 3 - t0 ) + k(t4 - t0) = 0

t 1 + t 2 + t3 + t4 - 4t 0 =0 ((4.6))
Thus equation (4.6) has been confirmed by this method.
Some representative examples of boundary mesh points will now
be given.
Convecting Boundary. Fig. 4.3 shows the physical situation at a

convecting boundary. Double resistances (or half conductances)
fluid
2
Fig. 4.3. Bo111Ulary node with co11f!ectio11.
exist between points 1 and 3 and point 0, and a convection resistance

Rc exists beyond the solid boundary, of magnitude 1/h(a x 1). The
energy balance is
k(t -t)
1
2 0 + k(t 3 2- t 0) + k(t 2 - t o) + ha(tr - t o
) -- 0
tl + tJ
:. - 2- + t 2 + (ha/k)tc- (2 + ha/k)t0 = 0 (4.11)
fluid
.....---.----.. 0
~~ tr
I 1 I -
L----l--
I· a .I T h
l_2 _r 2
-\ a/2 f-
Fig. 4.4. External boundary corner witlr con11ection.
Boundary Corners with Convection. An external corner is shown in

Fig. 4.4. and the energy balance is given by
k(t 1 - t 0 ) k(t 2 - t 0 ) ha( ) ha( ) 0
2 + 2 +2 tr - to +2 tr - to =
:. t 1 + t 2 + (2ha/k) tr - (2 + 2ha/k) tr = 0 (4.12)

For the internal corner shown in Fig. 4.5, it is left to the reader to
show that
[(t 1 + t 2 )/2] + t 3 + t 4 + (ha/k)tr- (3 + ha/k)t0 = 0 (4.13)
r----~-
1a
r-
l
I 2
fluid
1 a/2 I
I
I 4 I 1
L ____ L_
I
I
3
~ 3 I
L ____ j
Fig. 4.5. Internal boundary cor~~er wit, con11ection.

4.2.2 Preparing the Field and the Equations for the Program
Examples have been given of mesh point temperature relationships.
The reader should now be in a position to write further relationships
for other situations which may arise.
In preparing the program, the first step is to define the field within
an array of I rows and J columns, and to write the temperature equa-
tions in terms of (I, J) subscripting. It should be noted that because of
differences in (I, J) subscripting. more than one equation may be
required for a given physical situation, as for example, in four separate
boundary corners. Figure 4.6 shows the field for a sample program,
for part of a square duct, having 6 rows and 10 columns. Nine separate
equations exist in the field, and of these equations 3, 4, 7 and 8 are used
only once at the points shown, equation 2 is used for 8 points, equation
6 for 3 points, equations 5 and 9 for 4 points, and equation 1 for 22
points.
J=1, 10 TF1,HCONI
2
3
'"' I'
!'-,~
"'""~ -~1- l
f- 5
- - I""~ ' '

I = 1'6 9 i
-1 }-
6
_A_
mesh stze
"""-8~ ,~,
TF2, I
7
lines of symmetry
Fig. 4.6. Field for an elementary BASIC program-part of a hollow duct.
TFI, TF2, fluid temperatures; HCON I, HCON 2, convection coefficients;
TK, thermal conductivity.
Equation ( 4.6) for the field is written in BASIC as

X=(T(I+ 1, J)+ T(I-1,J)+ T(I, J + 1)+ T(I, J -1))/4-Q (4.14)
where X is the temperature calculated in an iteration from previously
calculated values. This is line 210 in the program, and together with
the other nine equations, appears in the listing below.
BASIC Symbols and Iterative Sequence

50 CT1=1iCON1·•A·~tTF1/TK
:JI C::T2=HCON2*A*TF2/TK
52 01a1.0+HCON1•AITK
53 02=2.0+HCON1•A/TK
54 D3=3.0+HCON2•AITK
55 04=2. OHICON2*A/TI(
60 ITEH7.=0
70 L7.=0
80 FOH ~Y.=I TO 10
90 FOR 17.=1 TO 6
100 1<7.=MY. ( I7., ~7.)
110 IF K7.=1, GOTO 210
120 IF KY.=2, GOTO 220
1:10 IF K7.=3, GOTO 230
140 IF KY.=4, GOTO 240
150 IF K7.=5• GOTO 250
160 IF KY.=6, GOTO 260
170 IF KY.=7, GOTO 270
100 IF KY.=B, GOTO 280
1'10 IF KY.=9, GOTO 290
200 IF K7.=IO, GOTO 300
210 X=ITII7.+1,~Y.l+TIIY.-1,JY.l+TIJY.,~Y.+1l+T(J7.,~7.-1ll/4.0
215 GOTO 310
220 X=IO. 5•1T( I7., ~/.t-1 )+T( I/., J7.-1 l. l+TI I7.+1, J7.)+CT1 l/02
225 GOTO 310
230 X=ITII7..~7.+1l+CT1l/01
235 GOTO 310
240 X= IT! 17., ~:r.-1 l+TI IX+1, ~/.l+CTl )/02
245 GOTO 310
250 X= ( Tl I /.+1, ,17. l +T( IY.-1, ~7. l +2. O*TI JY., ~Y.-1) l /4. 0
255 GOTO 310
260 X=IO. 5*1TIIY..~Y.+Il+T<Jt..~Y.-1ll+T(l7.-1,~/.l+CT2l/04
265 GOTO 310
270 X=ITIIY..~7.-Il+TIIY.-l.~Y.l+CT2l/04
275 GOTO 310
200 X= IT( I/., ,17.+1 >+2. O*TI IY.-1, ~/.l+C1'2l/03
285 GOTO 310
290 X=IT( I/.-1, ,JY.l+TIJY., ~/.·t-1) l/2. 0
295 GOTO 310
300 X=TI~Y., 17.)
310 DT=ADSITIIY..~Y.l-Xl
320 IFIOT>O.OOSlGOTO 340
330 L7.=L7.+1
340 TIIY..~Y.l=TIIY..~7.l+1.9*1X-TIIY.,J/.ll
350 NIOXT 17.
360 NEXT ~Y.
370 ITER7.=ITERY.+1
300 lFIITER7.>200lGOTO 500
390 IF<L7.<bOlGOTO 70
400 CONTINUE
500 PRINT,"200 ITERATION(; REACHED"
4.2.3 The Iterative Technique

Starting with a set of given temperatures in the field, the iterative
method consists of solving every equation in the field to obtain a new
set of temperatures, each temperature being compared with the corres-
ponding value in the previous iteration. If the difference in temperature
is less than 0·005°, then convergence is assumed at that point and itera-
tions are continued until convergence has been obtained at all points.
This method is known as the Gauss-Siedel iterative technique. The part
of the program that carries out the selection of the correct equation at
each point, and tests for convergence in the iteration, is reproduced on
page 51. Initially, equation numbers are read into a storage array as
integer values, M(I, J). In lines 340-380 an accelerated convergence
technique 3 is used. This is known as the extrapolated Liebmann
method. The test for convergence at all points takes place in line 390,
lines 400 and 500 being the continuation and end of the program,
respectively.
Preparation of the complete program should now present no undue
difficulties to readers conversant with BASIC. For further background
material the reader is referred to F enner4 •
The advantage of the method is that provided the field is correctly
specified, and an appropriate set of equations is written, then any
two-dimensional problem can be solved.
4.3 The Electrical Analogy of Conduction
The mathematical similarity between Fourier's law and Ohm's law

has already been referred to. Thus, one-dimensional composite
systems may be represented by a number of resistances in series, and
it has been seen in Section 4.2 that two-dimensional fields may be
represented by grids of resistances, as in Fig. 4.2 to Fig. 4.5. The
technique may be extended to transient work by adding a capacitance
at each node, the value being proportional to the thermal capacity
of the node. Comparison of the two fundamental laws results in
scaling factors being defined. Thus, let Q = 8/R,, and I = V /R, so
I V R
sl = -; s2 = -; s3 =-
Q () R,
It is seen that values for only two ofthese can be chosen independently
since S 1 = S 2 /S 3 • For transient work, two further scaling factors
must be introduced. Thus S 4 = T e /T, the ratio of electrical to thermal
time constant, and S5 = C/C,, the ratio of electrical to thermal capa-
citance. Since the time constant is the product of resistance and
capacitance it also follows that s4 = s3 X Ss.
Complex two-dimensional shapes in steady state may be simulated

by a continuous analogue using electrically conducting paper. Thus
Fig. 4.7a shows a simple example with isothermal (constant voltage)
boundary conditions. These are made using high-conductivity silver
paint with an additional copper wire buried in the paint. Using the
probe shown it is possible to plot the constant voltage lines between
the boundaries.
(a)
vd
(b)+~
~w~l I
Fig. 4. 7. Electrical analogy of two-dimensional conduction.
An element of conducting paper, length I, width w, and thickness

t, is shown in Fig. 4.7b. There is a potential difference of V between
the ends of the length I, and a current 1 is flowing.
If R is the resistance of the element, then I = VIR,
and R = pl/wt, where p is the resistivity of the material; (units of
ohms x length). Then.
1 =-
wt
pi
v = s.-pt v
I
(4.15)
where Si is the 'shape factor'. The shape factor for this rectangular
element is wjl. Any other geometrical shape which passed the same
current for the same voltage drop would have the same shape
factor.
A similar equation may now be written for a geometrically similar
element, length L, width W and thickness T, along which heat is
conducting. (}is the temperature difference and k the conductivity,
hence from Fourier's law,
Q = k WT (} = k S TO (4.16)
L q
where Sq = WjL and is the shape factor for the element. As the
elements are geometrically similar, the shape factors are equal.
Dividing (4.16) by (4.15):
(4.17)
Q will be in heat units per unit length and time, with T as unit
thickness. Thus from measurements of I and V and from a knowledge
of(} and k for the conducting problem, the heat ftow may be calcu-
lated. Equation (4.17) is valid for any geometrical shape, provided
the prototype and· model are geometrically similar, when the shape
factors are equal. pjt is a property of the conducting paper and is
supplied by the manufacturers. It has a value of about 2000 ohms
per square. (R = plfwt = p/t for a square, regardless of its size.)
Conducting paper is rolled during manufacture, and this can result
in an anisotropic effect. To account for this let the element in Fig.
4.7b have an equal resistance in both principal directions. Then
p 1 1jwt = p 2 wjlt, where p1 is now the resistivity in the direction of
current flow shown. It follows that ljw = J(p 2 / p 1). The model in
Fig. 4.7a would then be made rectangular to the extent indicated by
measurements of p 1 and p2 •
Structures of composite material having differing thermal
conductivities may be simulated by punching regular holes in the
paper, to produce an area having an increased resistance. Convecting
boundaries may be included either by cutting a band of paper outside
the boundary into strips normal to the boundary, or by adding carbon
resistors Rc at the edge 4 , so that Rc/R = (1/h)/(L/k WT) where his the
convection coefficient.
Three-dimensional shapes of rectangular form may be modelled
using a number of sheets of paper to represent layers in the third
coordinate direction, with additional resistances joining the centres
of corresponding elements. Complicated three-dimensional shapes
may be simulated in electrolytic tanks 5 .
PROBLEMS
1. The diagram shows a plan view of the vertical insulation round the walls
of a liquefied natural-gas storage tank. The inside and outside surface temp-
eratures of the insulation are - 161 oand + 1oc. Calculate the heat transfer
rate into the tank per metre height. Treat points a and b as fixed boundary
temperatures of- 53° and -107°C and assume one-dimensional conduction
through the tank sides. Take k for the insulation as 50 x 10- 6 kW/(m K).
The mesh size is 0·2 m. (Ans. 0·1764 kW/m.) (The City University).
-161°C
I. 2·4m
.I
2. The diagram represents a square section with 81 mesh points.
Boundary temperatures and convection coefficients are subscripted to
allow for varying input information. Write a BASIC program to calculate
temperatures at the 81 points. Note that there are only three separate
physical situations: points 2, 3, 4 and 5 are similar, also points 6, 7, 8
and 9 are similar. Point 1 applies throughout the interior field. To check
the program, the same convection coefficient on all faces will produce a
symmetrical temperature field. Subsequent runs may be made with
alternative boundary conditions.
Tl(l)= 10.0
HI(!)
--
T!(J) = 10.0 T2(J) = 10.0
HI(J) + H2(J)
--I
4
T2(1) = 100.0
H2(1)
3. Write a BASIC program (deriving all the necessary equations) to

calculate the heat transfer through the duct shown in the figure. The inside
and outside fluid temperatures are 200° and 30°C, the thermal conductivity
is Q-005 kW/(m K), and inside and outside convection coefficients are 0·1
and 0·05 kW/(m 2 K). The mesh size is 3 em. (Ans. 2·796 kW/m.)
1 2 3
15
4 9 10 12
to-1 W/(m:tK
11 16 /1
2 0°C /1
/ 14 8
/
/
13
5 6 7
4. An H-section copper conductor (see diagram) carries an overload current

of 54,000 amps. In steady state conditions, the surface temperature is 60°C.
Using a 0·5 em grid, determine the temperatures within the copper. Calculate
the total heat transfer at the surface, k Wfcm length. The electrical resistivity
of copper is 2 x 10- 8 ohm m, and the thermal conductivity is 0·381 kW/
(m K). (Ans. 0·73 kW/cm.)
5. The figure shows a unique 1/8th part of a hollow square duct having
inside and outside surface temperatures of 150°C and 15°C. Using the
initial values of temperature given, relax until residuals have values not
greater than± 3. Calculate the heat transfer rate through the duct perm
length b¥ taking the average of the conduction rates at the 15°C, 70° C
and 150 C isothermals. The thermal conductivity of the material of the
duct is 4 Wjm K. (Ans. 4240 W/m.)
15°
T
" "'L4° 35° I 46o I 510 I 520
!"-.
"" 60°
I'.
\.85° 195°
I
J!9
"'."'
I
1
II£
""
150°
I
'
REFERENCES
1. Southwell, R. V. Relaxation Methods in Theoretical Physics, Oxford

University Press (1946).
2. Bayley, F. J., Owen, J. M. and Turner, A. B. Heat Transfer, Nelson (1972).
3. McCracken, D. D. and Dorn, W. S. Numerical Methods and Fortran
Programming, Wiley ( 1966).
4. Fenner, R. T. Computing for Engineers, Macmillan (1974).
4. Simonson, J. R. An Electrical Analogy of Extended Surfaces, Bull.
Mech. Engng. Educ., vol. 8, 215-25 (1969).
5. Karplus. W. J., and Soroka, W. W. Analogue Methods, McGraw-Hill
Book Company, New York (1959).
5
Transient conduction
In any thermal system, transient heat transfer generally occurs before

and after steady state operating conditions. The time duration of the
transient condition can be of importance in design, and further,
excessive thermal stress may arise. A simple approach is to assume
the system is lumped, i.e. the temperature is uniform in space and is a
function of time only. In a more detailed analysis, temperature will
also be a function of position.
5.1 The Uniform Temperature, or Lumped Capacity, System

The ratio of internal thermal resistance to external convection
resistance of a system is known as the Biot number, and when the
Biot number is smalL say <0·1, the system will effectively follow a
single heating or cooling curve, as in Fig. 5.1.
Considering the cooling curve in Fig. 5.1, for a system of mass m,
and specific heat cP, having a surface area A, a convection coefficient
h, and excess temperature() cooling by d(J in timedt, it follows that
- mcP d(J = hA(J dt
or
d(J hA dt
-=--dt=--
(J mcP T
where T = mcJhA = time constant = product of thermal resistance
and capacitance.
Cooling from 61 to 62 will take timet given by
()2 t
In-=--
(Jt T
or
(5.1)
58
TRANSIENT CONDUCTION 59
(}
---+-----===-----==---
1
I
/
l
j et heating curve
time
t
8
cooling curve
dt
1-
Fig. 5.1. Heating t111d cooling curtJesfor lumped systems.
The heating curve in Fig. 5.1 is the result of an internal heat source,
and the maximum temperature rise (}max is achieved when the rate of
cooling is equal to the rate of heat release. For a volume V and heat
release rate q'
q'V = hA (}max (5.2)
At some temperature(} the energy balance with a density p is given by
pVcP d(} = (q'V - hA(}) dt
:. - t 1 (q'V
= [ - -ln - hA(}) ]6t (5.3)
pVcP hA 0
where 01 is the temperature rise at time t.
• .!._ = _ l [q'V - hAOfl

.. T n q'V j
e-lrr = 1 - hA()t
q'V
Introducing ()max from (5.2) this then gives
01 = ()max (1 - e- 1rr) (5.4)
When()= 0, (5.3) becomes pVcPd() = q'Vdt
. (d(J) q' V hA ()max () _,.
· · dt ll=o = pVcP = pVcP = T
Hence the time constant will appear on the heating curve as shown in
Fig. 5.1. This provides an experimental method for measuring T and
deducing, for example, the convection coefficient, h.
EXAMPLE 5.1
An electric motor is designed to operate on a repetitive load duty
between temperature limits of 30° and 55°C, assuming an ambient of
15°C Heat dissipation on load is 0·38 kW, and the off-load period is
294 seconds. The motor has an effective mass of 3·5 kg, specific heat
of 0·45 kJ/(kg K~ and a surface convection coefficient of 0·15 kW/
(m 2 K). Determine: (i) the cooling area to be provided; (ii) The maxi-
mum temperature that would be achieved if the motor ran indefini-
tely; (iii) The duration of the allowable load period. (The City
University)
Solution. The motor cools from 55° to 30°C in 294 seconds, hence the
time constant may be found: (3(}-15)/(55-15) = e- 1rr wheret = 294
therefore t/T = 0·982 and T = 300 sees. But
T = mcJhA = (3·5 x 0·45)/(0·15 x A)
:. A = (3·5 x 0·45)/(0·15 x 300) = 0-()35 m 2
At the maximum temperature rise, all energy dissipation is convected
away, hence ()max hA = 0·38
.". ()max = 0·38/(0·15 X 0·035) = 72·5
The maximum temperature is therefore 72·5 + 15·0 = 87·5°C.
To establish the load period, use equation (5.4) to determine the
times to reach temperature rises of 40° and 15° from ambient Thus:
40 = 72·5 (1 - e-•·' 300) :. t 1 = 241 sees
15 = 72·5(1- e-•z/ 300) t = 69·6secs
:. 2
The load period is (t 1 - t 2 ) = 171·4 sees.
5.2 The Solution of Transient Conduction Problems in

One Dimension
The discussion is limited to one-dimensional transient conduction in
rectangular coordinates for which the equation is
:: = « ( ;:~) ((2.8))
For the general problem, numerical procedures will be described.
It is necessary to replace equation (2.8) by a finite difference
relationship. Figure 5.2 shows a plane slab uniformly divided into
sub-slabs of thickness a, with a temperature contour at some time t0 •
Recalling the argument of section 4.1 it will be seen that the tempera-
tures t 3 • 0 , t 4 • 0 , and t 5 • 0 are related
t 3, 0 + t 5, 0 = 2t4 •0 + (;:~) 4,0

a2
and hence
(fPt)
ox2 4,0
(5.5)
With a forward time step, the fmite difference relationship for (otfo t)
(at)
IS
= t4,1 - t4,o
Ot 4,0 L1t
where _t 4 • 1 is the temperature at point 4 at time tl' which is At after t 0 •
Equatton (2.8) can now be replaced by
t4,1 - t4,o = oc(t3,o + ts,o- 2t4,o)
(5.6)
L1t a2
This may be re-arranged as

t4,1 = F(t3,o + ts,o) + t4,o{l - 2F) (5,7)
where F is the Fourier number, Ata./a 2 = Atkfpc.a2• This compares
energy conducted in time At, proportional to At kfa, to energy stored,
proportional to pc•a, and henre gives a measure of temperature
response.
I
I
I
I
I
II I lt4o'1
I I I
1-Q-1 sub-slab
I I
I-= sutface half sub-slab

total slab width
Fig. 5.2. Tre•tmellt of" pllute sW for trtuuiellt colllbu:tiotl;, OM tlimertsio-.

(First sllbscript de110tes positiD11, secollllsllbscript de110tes time.)
5.2.1 A Numerical Method

Equation 5.7 is suitable for performing a numerical solution. Values of
~t and a are chosen to give a suitable value of F. The coefficient of t 4 , 0
must remain positive for the solution to be stable, hence F ~ IfF= t.
t then t 4 , 1 is simply the average of t 3 , 0 and t 5 , 0 • For smaller values of
F increased accuracy will be obtained. In order to solve a problem it is,
of course, necessary to know the boundary temperatures after each
time interval ~t.
EXAMPLE 5.2
The surfaces of a brick wall, 300 mm thick initially 20°C throughout,
rise in temperature at a constant rate of 10°C every 2500 seconds.
Dividing the wall into six equal slabs find the temperature distribu-
tion in the wall after 1<f seconds. Use F = !, a = Q-05 x 10- s m 2 fsec.
Solution. It is seen that F = at rxja 2 = 2500 x 0·05/(10 5 x 0·05 2 ) =

0·5. From equation (5.7), with F = f, the following table may be
written:
Node Surface 1 2 3 4 5 Surface
Initial temperatures 20 20 20 20 20 20 20
At: 2500 sees 30 20 20 20 20 20 30

5000sees 40 25 20 20 20 25 40
7500sees 50 30 22·5 20 22·5 30 50
10000 sees 60 36·25 25 22·5 25 36·25 60
Convection at a solid boundary can be allowed for in a numerical

solution by setting up an energy equation for the boundary slab.
The change in stored energy over the chosen time interval is equal
to the conduction from the adjoining slab plus the convection from
the boundary fluid. A relation between the fluid temperature and
temperatures in the 'iolid is thus obtained.
This procedure is based on the electrical analogy of transient
conduction in which a capacitance is added at each node. Referring
to the surface half-sub-slab in Fig. 5.2
(a x 1)
- - - x p x ( .p (t 1 1 - t 1 .0
)-kxl(t 2 ,0 - t 1 ,0 )at
- ---'--='"'--~~
2 . a
+h x l(t~, 0 - t 1 , 0 )LH
. t
. - 1,1
= (2k
pcPa2
at) t 2,0
+ (2h
pcPa
at) t t,o
+t
1,0
( 1 _ 2k
pcPa2
At _2hpcPaAt)
:. t 1 , 1 = 2Ft 2 , 0 + 2FBtr,o + t 1 , 0 (1- 2F(l +B)) (5.8)
where B = hajk, which is the Biot number of a sub-slab, and F is the
Fourier number as before. A new stability criterion applies to the
t.
boundary equation, i.e., F(l + B) ~ for the coefficient of t1. 0 to
remain positive. This means that F < t and the numerical procedure
using equations (5.7) and (5.8) is more complicated throughout the
whole field.
5.2.2 BASIC Program for Transients in One Dimension with

Convection at the Boundary
A one-dimensional problem with convecting boundary conditions can
readily be solved for as many time steps as are required by a BASIC
program. Equation (5.8) is used at the two boundaries and equation
(5.7) is used at all points in between. Temperatures are not normally
stored beyond one time step, hence they may be printed at chosen
intervals throughout the transient. A program is given, by way of
example, to solve the following problem.
EXAMPLE 5.3
It is required to investigate temperature rise and heat penetration in a

fire door. The door is 50 mm thick, and is considered as ten slabs 5 mm
thick (hence there are two boundary temperatures from equation (5.8),
and nine interior temperatures from equation (5.7).) Initially the temp-
erature of the door and environment on both sides may be taken as
25°C. On one side of the door a fire causes the effective environment
temperature to rise linearly to 500° C in 200 seconds, and then to
remain constant at that temperature, with a combined convection and
radiation coefficient to the door of 120 W/m 2 K. On the other side, the
environment temperature remains constant at 25°C, and a convection
coefficient of 15 W/m 2 K applies. The problem is shown in the figure.
120W/m 2 K 15 W/m 2 K
1--
~..--~...--~...-L--L---L--L---L--
25°-500°C
in 200 sec f . - - - ,...-f--,...-f--,...-
f.4-----50 mm - - - - - . J
Fig. Example 5.3.
Determine the temperature distribution throughout the door every

200 sec for a total time of 1800 sec, the heat flow rates to and from the
door and the energy stored in the door at these times.
For the door,k= 0·19 W/mK,p = 577 kg/m 3 , and CP = 816 Jfkg K.
Solution. The program listing and results are given below. Heat flow
and heat stored results are all per m 2 of door, and they are calculated in
lines 500, 510 and 540. It is seen that initially heat transfer rates to the
door are high, and the rate falls as the outer surface of the door rises in
temperature. The temperature of the inner face of the door rises slowly,
however by 1800 sec the temperature is significant and the heat flow
rate is approaching 1 kW /m 2 ; but the initial slow response and corres-
ponding escape time for occupants are clearly apparent.
10 DIME"SIO~ TI11J,TNEWI11)
20 A=0.005
30 TK=0,19
40 DEN=577,0
50 CP=816.0
60 111=120,0
70 112=15.0
80 T1=25.0
90 T2=25.0
100 TTOP=500,0
110 TRT=200,0
120 TLIM=1800.0
130 TPRINT=200.0
140 DTPRI"T=200,0
150 D1•111•A/TK
160 B2=H2•A/TK
170 f•TK/IDEn•CP•A•AJ
180 TIIH1=1,0/I?..O•F•I1.0+01 ))
190 TIME2=1.0/12,0•F•I1,0+B2ll
200 IFITIHE1>TIME2JGO TO 230
210 TIME=TinE1
220 GO TO 240
230 TIME=TIHE2
Z40 PRINT,"TIHE STEP IS:'',TIME
250 PRI~T,"~O YOU WISH TO ROUND OFF AT A SHALLER VALUE OF TIME STEP?"
260 PRINT,"1 - YES; 2 - NO''
270 [~PUT IX%
280 IF!IX%>1JGO TO 300
290 !~PUT TI"E
300 F=F•TitlE
310 F2•2.0•F
320 FB12=2.0•F•01
330 F022=2.0•F•02
340 FT1=1.0-2.0•F•I1,0+B1l
350 FT2=1.0-2.0•F•(1,0+92J
360 FT=1.0-2.0•F
370 DTR=I500,0-25.0l•TIME/TRT
380 TR UII=O, 0
390 T1=T1+DTR
395 IFIT1>TTOPJTHE~ T1=TTOP
400 TNEW!1J=f2•TI2J+F012•T1+111)•FT1
410 FOR 1=2,10
420 THEW!I:J=F•!TIIZ-1J+T!IX+1Jl+TIIZJ•Fr
430 ~IEXT IZ
440 TNEWI11l=F201(10l+FB220J2+T!11)•FT2
450 TRU"=TRUN+TIME
460 FOR 1~=1,11
470 TII%)•TNEWIIZJ
480 NEXT U
490 IF!TRUN<TPRINTJGO TO 390
500 II EAT Ill= ( T 1- T ( 1 ) ) • H 1
510 HEATOUT•!TI11J-T2J•H2
520 HSTORE=O.O
530 FOR 1%=1,11
510 HSTORE=HSTORE+(T(I%l-25.0l•A•DEN •CP
550 IIEXT 17.
560 PAifiT,"IIEAT TO DOOR ='',IIEATIH," WATTS"
561 PRI~T,"IIEAT FRO~ DOOR ='',ti£ATOUT,'' WATTS"
562 PRINT,''IIEAT STORED :",HSTORE,'' JOULES''
565 PRINT,"TEMPERARURES -"
570 PRINT T<l:l FOR 1%=1,11
580 TPRINT=TPRINT+DTPRINT
590 IF!TPRINT>TLI11lGO TO 620
600 GO TO 3?0
620 STOP
Results from Listing
TIME STEP IS: 0.74498731£ 01

DO YOU WISH TO ROU~D OFF AT A SMALLER VALUE OF TIME STEP?
1 - YES; 2 - NO
=1
•5.0
TIME 200.0 SECONDS
HEAT TO DOOR = 8694.08 WATTS
HEAl FROM DOOR • 0.16 WATTS
HEAT STORED • 1791252.77 JOULES
TH1PERATURES -
427.5 227.9 118.6 64.1 10.0 30.2 26.6 25.5 25.1 25.0 25.o
TIME = 100.0 SECO~DS
HEAT TO ODOR = 1539.30 WATTS
HEAT FROM DOOR • 17.69 WATTS
HEAT STORED = 3005670.75 JOULES
TEMPERATURES -
462.2 344.6 212.1 162.2 105.1 68.9 17.3 35.6 29.8 27.1 26.2
TIME • 600.0 SECO~DS
HEAT FROM DOOR • 110.10 WATTS
HEAT STORED = 1R08177.53 JOULES
TEI~PERATURES -
170.6 378.5 2?3.7 220.2160.1 111.6 81.6 59.2 45.0 36.h 32.3
TIME 800.0 SECO~DS
HEAT TO DOOR = 2?94.03 WATTS
HEAT FROM DOOR = 266.80 WATTS
HEAT STORED = 1438R35.19 JOULES
TEI1PERATURES -
175.0 396.8 323.0 256.3 198.8 151.3 113.? 85.6 65.1 51.6 12.8
TinE = 1000.0 SECOnDS
HEAT FROM DOOR = 13?.52 WATTS
TEI4PERATURES -
178.0 408.7 342.6 281.5 227.1 180.2 111.2 110.0 35.8 67.6 51.3
TinE = 1200.0 SECONDS
HEAT TO DOOR = 23?4.35 WATTS
HEAT FROM DOOR = 600.89 WATTS
TEMPERATURES -
180.0 417.3 356.? 300.3 248.8 203.2 163.? 131.0 104.1 82.4 65.1
TIME = 1400.0 SECOnDS
HEAT fROn DOOR = 741.11 WATTS
HEAT STORED = 5700~36.75 JOULES
TEfiPERATURES -
481.6 123.9 367.? 314.9 265.9 221.7 182.6 118.7 119.8 95.3 71.1
TIME = 1600.0 SECO!IDS
HEAT FROM DOOR = 85?.95 WATTS
HEAT STORED • 5?76912.13 JOULES
TEI-1PERATURES -
482.9 429.0 376.6 32~.5 279,7 236.7 197.9 163.3 132.9 106.1 82.3
TIME • 1800,0 SECONDS
HEAT TO DOOR • 1933.17 WATTS
HEAT fROM DOOR • 953.35 WATTS
HEAT STORED • 6202445.25 JOULES
THlPERATURES -
483.9 433.2 383.6 335.9 290.9 248,9 210.4 175.4 143.7 115.0 88.9
It may be seen from Fig. 5.2 that the averaging of temperatures can be
carried out by drawing. This is the basis of the Binder-Schmidt
method 1 • 2 • Graphical constructions are also possible for convecting
boundaries. These rely on the fact that the conduction rate at the
boundary must equate to the convection rate. Various cases are con-
sidered by Hsu 3 •
5.3. Two-dimensional Transient Conduction

In two-dimensional transient conduction in rectangular coordinates
the differential equation is
Ot
at
= IX (0ox
2
t
2
+ iJ 2 t)
oy 2
(5.9)
and referring back to the nomenclature of Fig. 4.1, the finite difference
relationship for a field point with a forward time step can be seen to
be
to, 1 - to o = IX (tl,O + tz,o + t3,o + t4,o - 4to,o) ( 5 .lO)
Llt a2
This is re-arranged to give
to,t = F(tl,o + tz,o + t3,o + t4,o) + to,o(l - 4F) (5.11)
with the stability requirement that F ~ ±-
In transient work, alternate finite difference relationships having
backward time steps may be used. The equivalent of equation (5.10)
would be:
to,l- to,o = IX(tl,l + tz,t + t3,1 + t41- 4to,t) ( 5 . 12 )
.D.t a2
Then t 0 ,0 is the only known temperature, and equations for all points
must be solved simultaneously to obtain the temperatures after the
next time step. However, there is no stability restriction in this case.
The reader is referred to Bayley4 for a full discussion of these methods.

The two-dimensional steady state computing methods discussed
in Section 4.2.2 may be modified to deal with transient problems
using equations with forward time steps. by substituting transient
equations for the steady state ones and by replacing the iterative
technique by a scheme for solving the equations throughout the field
for as many time steps as are required. Boundary equations may be
derived following a similar procedure to that in Section 5.2.1. To
illustrate, it may be verified that the transient equation for point 2
in Fig. 4.6 is, in BASIC
T(l, J)=F•(TP(I, J + 1)+ TP(I, J -1)+2·0•TP(I+ 1, J))

+2·0•F•BhTF1 + TP(I, J)•S(2)
where T(I, J) denotes the new temperature, TP(I, J) the existing
temperature, and
TK = thermal conductivity of material
DT = time interval
D = density of material
C = specific heat of material
A =mesh size
F = TK•DT/(D•C•A..2)
S(2)=(1·0--4·0•F-2·0•F•B1)
B1 = HCONhA/TK
Since different stability criteria exist the coefficient S(l) is sub-
scripted to enable a DT value to be determined to satisfy all equations.
5.4 Periodic Temperature Changes at a Surface

A periodically changing surface temperature can also be dealt with
by numerical or graphical methods, but the work involved is
probably not justified in view of the fact that an analytical solution
is not too lengthy for this particular boundary condition.
The problem to be considered is one in which a plane slab of
material, referred to as a 'semi-infinite solid', is regarded as being
infinitely thick, the periodic surface temperature existing at the face
of the slab where x = 0. The surface temperature varies in a sinu-
soidal manner and, because of the assumption of infinite thickness,
the temperature history within the material is controlled only by
the surface variation. Further, conduction takes place in only one

dimension, so that edge effects are neglected or the specimen is
regarded as being sufficiently large in they-direction for conduction
to be one-dimensional over the area of material of interest. An
additional assumption is that the cyclic variation of temperature at
the surface has been going on for a time sufficiently long for tempera-
tures elsewhere in the slab to be repeated identically in each cycle.
The general result obtained, as will be seen, is that the interior
temperature cycle lags behind the surface variation, depending on
the depth and, in addition, has a diminished amplitude compared with
the maximum surface values. This type of analysis finds application
wherever a cyclic variation of temperature occurs, as in annual or
daily temperature variation of buildings or the ground exposed to
solar radiation, and, in the other extreme, in the cylinders of recipro-
cating engines. The chief restriction on the validity of the analysis is
whether the object in question may be regarded as infinitely thick.
The depth in the material at which the temperature amplitude has
become, say, only 1 per cent of the surface value is the criterion by
which this is judged.
The surface of the slab has a mean temperature t. It varies in a
sinusoidal manner between an upper temperature limit oft + (Om)o
and a lower limit of t - (Om)o. Thus, if 9 is the temperature dif-
ference between the actual temperature at any instant and the
mean then, at the surface where x = 0, 9 varies between ± (Om)o
where Om denotes the maximum difference. Further, at some depth
x in the slab, 9 varies between ±(Om>x· The frequency of the tem-
perature variation is n cycles per unit time, so 1/n is the period of
the variation. The boundary conditions of the problem are set by
the sinusoidal temperature variation at the surface, given by
9 = (Om)o sin(2nnt) (5 .13)
which is the value of 9 at x = 0, and t = t. At x = 0 and t = 0,

() = 0. In equation (5.13) 2rrn is the angular velocity of the sine wave
in rads/unit time.
Since 9 is the temperature variation about a mean value t, 9 may
be regarded as the temperature variable since tis constant. For this
case the one-dimensional unsteady equation, (2.8), becomes
ao = oc(a2o)2 (5.14)
at ax
Since (} varies sinusoidally at the surface, it can also be expected to

do so within the solid, but between reducing limits and further, the
phase shift will depend on the time to penetrate to depth x, hence
the form of solution chosen is
(} = Ce- px sin(2n:nt - qx) (5.15)
where C, p, and q are constants to be determined. The constants p
and q may be found by substituting equation (5.15) in equation (5.14).
The partial differential coefficients found from (5.15) are
ae
- = 2nnce-P" cos(2nnt- qx)
at
ae =
ox
- pCe-px sin(2nnt- qx)- qce-px cos(2nnt -- qx)
oe = p
~
2
2 Ce-pxsin(2nnt- qx) + pqCe-pxcos(2n:nt- qx)
t•X
+ pqCe-P" cos(2n:nt- qx)- q 2 Ce-P" sin(2nnt- qx)
Hence, equation (5.14) becomes, noting that ce-P" may be cancelled
from all terms:
2nn cos(2nnt - qx) = 1X[p2 sin(2nnt - qx)
+ 2pq cos(2nnt - qx) - q 2 sin(2nnt - qx)]
(2nn - 2pq1X) cos(2nnt - qx) = 1X(p 2 - q 2 ) sin(2nnt - qx)

Since there is no cosine term on the right, it follows that
(2nn - 2pq1X) = 0
pq = nn/IX
Further, as there is no sine term on the left, it follows that

p2- q2 = 0
or p = q
Thus, from these results,

p = q = ± (nn/1X)0 ·S
The negative solution means an exponential increase of() with x,
hence taking the positive result only, equation ( 5 .15) becomes:
(} = C exp[- x(nn/1X) 0 . 5 ] sin[2nnt - x(nn/1X) 0 . 5 ] (5.16)
This result may now be compared with the boundary condition at
x = 0 and t = t. Thus (5.16) gives
(} = C sin 2nnt
and the boundary condition gives
(} = (lJm)o sin 2nnt
Thus comparing these two equations shows that C = (lJm)o. The
final solution is therefore
(5.17)
This equation shows that the maximum variation of 8 decreases
exponentially with x, the distance into the solid, according to the
equation
(5.18)
The general form of the result given by equation (5 .17) is shown in
Figs. 5.3 and 5.4. In Fig. 5.3 the temperature variation with distance
distance x
x=O
representation of a semi-infinite solid
00
J
X=O -+--------..-00
00
Fig. 5.3. Established temperature variation vs. distance into solid, at t =J/2n.
at a chosen time is shown, and in Fig. 5.4 temperature variations

with time at the surface and depth x are shown. It will be seen that
a temperature wave propagates into the solid, and also that the
cyclic variation of temperature at some depth x lags behind the
surface variation. The phase difference in temperature variation
t------period 1/n----~
x = 0 time lag at depth x,
~ Jc:lJ
Fig. 5.4. Temperature variation with time at x = 0, and at depth x.
at depth x given by x(nnfrx.)0 "5• Hence the time lag of a certain

temperature excess (diminished in value at depth x) will be given
by
2nnAt = x(nn/rx.)0 "5
At=~-
( 1 )0·5 (5.19)
2 nnrx.
At for a complete temperature wave of length X is 1/n hence, for
a complete wave, the wave-length is given by
2n = X(nn/rx.)0 "5
X = 2(nrx./n)0 "5 (5.20)
Also, the velocity of propagation of the temperature wave into the
solid is
U = -X = 2(nnrx.) 0 .5 (5.21)
1/n
The ratio of maximum temperature variations about the mean may
be compared using equation (5.18). Thus:
(0 )
= exp[- x(nn/rx.) 0 "5 ]
(0m)o
m "
Jf it is required to determine the distance x at which (Om)" has
decreased to a certain percentage of (Om)0 , this result may be re-
arranged to give
(5.22)
Finally, the heat transfer rate at the wall surface, at x = 0, may

be determined from
and from equation (5 .17),
oO)
(OX = - (Om)o{1tn/oc)0 "5 (sin 21rnt + cos 2nnt)
x=O
Using the identity,

sin(27tnt + 7t/4) = sin 21tnt cos 7t/4 + sin 7t/4 cos 21tnt
= (1/j2)(sin 27tnt + cos 27tnt)
it follows that
( ~0 )
uX x=O
= - (Om)o{27tn/oc)0 "5 sin(27tnt + 7t/4)
and hence,
q = k(Om)o{27tn/oc) 0 "5 sin(27tnt + 1t/4) (5.23)
From equation (5.23) it will be seen that the surface heat transfer
rate varies sinusoidally and with the same frequency as the surface
temperature, but leading by a period of 1/Sn. The total heat transfer
at the wall is given by
Jq dt = Jk(Om) {27tn/oc)
0 0 "5 sin(27tnt + 7t/4) dt
= -k(Om)o{1/27tnoc)0 "5 cos(27tnt + 7t/4) heat/unit area (5.24)
Thus the energy stored, as represented by an integral of heat transfer
rate at the surface, also varies with the same frequency, but it will
be found that it lags behind the surface temperature variation by a
period of 1/Sn. Further, it will be apparent that the surface heat
transfer is both to and from the solid and that the energy stored is
in sequence both positive and negative relative to the mean
temperature.
EXAMPLE5.4
In a cyclic heating process the inside of a furnace wall is subjected to
a sinusoidal temperature variation. The temperature rises from a
minimum of 100°C to a maximum of 750°C in 3 hours. With k =
692 x 10- 6 kW/(m K) and ex = 0·0516 x 10- 5 m 2 /sec, determine:
(i) the velocity of the temperature wave penetrating the wall;
(ii) the time lag of the wave function at a depth of0·2 m compared with
the surface; (iii) the maximum and minimum temperatures at a depth
of 0·2 m; (iv) the temperature at a depth of 0·2 m when the surface
temperature is a maximum; and the surface temperature when the
temperature at a depth of0·2 m is a maximum (The City University).
Solution. (i) The period (1/n) is 6 hours. From (5.21)

U = 2(7t X ()-0516 X 3600/6 X 1()5 )0·5
= 0·0622mfh
(ii) The time lag is given by (5.19),
()-2 ( 6 X 105 ) 0 "5

At = T 7t X Q-0516 X 3600
= 3·22hours
(Check: Q-2 mat 0·0622 mfh takes 3·22 hours.)

(iii) The maximum and minimum temperatures at Q-2 mare obtained
from (5.18). (8m)o is 325
(8m)x = 325 X exp( -0·2(7t X 105/6 X 0·0516 X 3600)0 "5)
= 325 X ()-03477 = 11·3
The mean temperature is 425°C, hence the maximum is 436· 3oc, and
the minimum is 413·7°C.
(iv) Using equation (5.17); 8 = 325 x 0·03477 sin (21Tnt - 1·07?T).
Surface temperature is a maximum at t = 1f hours = in (mean to
maximum~ hence 8 = 11·3 sin (Q-5 7t - 1·07n) = 11·3 sin (- 0·577t)
= -11·3sin0·437t = -11·02°. Hence temperature at 0·2m is
425 - 11·02 = 414°C. The temperature wave at Q-2 m must advance
in phase by 1·077t to reach its maximum value, when the surface wave
will be at 1·57n. Hence () = 325 sin (1·57n) = -325 sin (0·43n) =

-317°
:. the surface temperature is 425- 317 = 108°C
PROBLEMS
1. Steel strip of thickness 1·27 em emerges from a rolling mill at a temperature
of 538°C and with a velocity of 2·44 mjsec. The strip is cooled in such a way
that its surface temperature falls linearly with distance from the mill at a rate
of ll0°Cjm.
Derive a finite difference method for dealing with this case oftransient heat
conduction assuming that heat flows only in the direction normal to the strip
faces.
Subdividing the strip into six increments of thickness, determine the tem-
perature distribution in the strip and the heat flux from the surface at a
position 2·74 m from the mill. (For steel take thermal conductivity 43·3
x 10- 3 kW/(m K), thermal diffusivity 0·98 x 10- 5 m 2 js.) (Ans. 3·41 x 103
kWj(m 2 )(University of Manchester).
2. A steel pipe, 2·54 em wall thickness, is initially at a uniform temperature

of l6°C when a liquid metal at 572°C is pumped through it for a time of lO sec
and with a surface coefficient of 2·84kW/(m 2 K). It may be assumed that
the pipe diameter is large enough for the wall to be considered plane, that
no heat loss occurs from the outside of the pipe and from the inside after the
flow of liquid metal has ceased.
Derive a numerical method to deal with this case using finite increments
of thickness and making the simplification that the heat capacity of the
surface half-increment is negligible. Using four increments determine the wall
temperature distribution after 18 sec. (For steel take thermal conductivity
0·041 k W/(m K), density 7530 kg/m 3 and specific heat 0·536 kJ/(kg K.) (Ans.
231 oc, 120°C.) (University of Manchester).
3. At a certain instant in transient one-dimensional conduction through a

3 em thick slab of chrome steel, 40 em square, the temperature distribution
along the 3 em thickness is given by t = (60 + l·2x 2 + 0·3x 3 )"C where xis
the distance from one 40 em square face. Calculate the rate of energy storage
in the slab, and the rate of change of temperature at each square face, at the
particular instant. Take p = 7833 kgfm 3 , k = 0·0398 kW/(m K), cP =
0·46kJ/(kgK). (Ans. +9·74kW, +0·266K/s and +0·865K(s.) (The City
University).
4. Given the differential equation (otfor:) = rx.(o 2 tfox 2 ), for unsteady conduc-
tion in a 'one-dimensional' wall, show that the temperature t•.p+ 1 at some
section n and time instant (p + 1) can be calculated approximately from
t•.p+l = F[t•+l,p + t.- 1.p + (~- 2)t.. p]

The temperatures in the right-hand bracket are values at equidistant sections
(n - 1), n, (n+ 1), preceding (p + 1) by a finite time interval M; F = rxl'J,:r:jtu 2
is the Fourier number.
Plane 1 is a distance ll.x/2 to the right of a wall surface. Prove that, if the
convection coefficient and temperature of the fluid to the left of the surface is
h and t, respectively,
t l,p+l = F [ t 2,p + -2B

2+B'·P
1 2+
- t + ( --
F - -3B)
- t l,p
2+B
J
where B = Mxlk. (University of Bristol).
5. An insulating screen is intended to withstand the penetration of high

temperature for as long as possible. Select either material A, B, or Cas being
best for this purpose.
Material k kW/(m K) p kg/m 3 cP kJ/(kg K)
600 x w- 6
-~---~----
A 1500 0·84
B 600 X 10- 6 1200 1·60
c 280 X 10- 6 750 1·10
The screen is 5·3 em thick and is divided into 5 increments. It is initially l5°C
throughout; the temperature of one face rises linearly by 20oc per minute.
Regard the other face as insulated. Determine, for the chosen material, and
by a numerical technique, the time before the insulated face temperature
starts to rise, and the temperature of this face after 18 minutes. (Ans. B,
15 mins, 24·4°C) (The City University).
6. In order to carry out an approximate analysis of a butt-welding process, it

is assumed that there is a uniform rate of heat generation at the contact face
between the two bars, that heat conduction occurs only in a direction normal
to the contact face and that the physical properties of the bars are constant.
Derive a numerical method to deal with this case by sub-dividing the bars into
finite increments of length.
Apply the method to obtain the approximate temperature distribution in
two similar steel bars, initially at 16°C, after 10 sec. The heat generation rate
is 1·005 kJj(cm 2 s) and this acts for a period of 5 sec. Use 0·635 em increments
of length and, for steel, take thermal conductivity= 45 x 10- 3 kW/(m K),
density= 7690kg/m3 and specific heat = 0·545 kJ/(kg K). (Ans. 1249°C
maximum at joint, after 10 sec: 534°C.) (University of Manchester).
7. A slab of material is considered as six layers of equal thickness, and
is initially at 20° C throughout. The two surfaces of the slab rise in
temperature by 10°C per time interval when a transient analysis is
undertaken with F = ~. Calculate the centre-line temperature of the
slab when the surface temperature is 60°C. Repeat the calculation for
F = -! to give a more accurate solution, and find the percentage error in
the fresh answer. (Ans. 22·5°C, 23·32°C, 3·52 per cent.)
8. A large steel plate 7·62 em thick initially uniformly at 816°C is quenched

in oil at 38•c. If the oil temperature remains constant and there is negligible
surface resistance, estimate the time required to reduce to 427"C:
(a) the average temperature ofthe slab,
(b) the centre-line temperature.
Thermal diffusivity of steel = 1·032 x 10- s m 2 /s. (Ans. 26· 7sec, 52·0 sec.)
(University of Leeds).
9. A current of 3 amps is passed along a 1 mm diameter wire of resistance

3·5 ohms/m. The wire reaches a steady temperature of 60°C in an atmosphere
of 20°C. Calculate the initial rate of temperature rise of the wire, and the
temperature after a time lapse equal to the thermal time constant of the wire.
The mass is 25 gjm and the specific heat ()-460 kJ/(kg K). (Ans. 2·74 K/s,
4S·2°C.) (The City University).
10. An internal combustion engine runs at 2500 r.p.m. The thermal diffusivity
of the carbon steel ofthe cylinder walls is H6 x 10- s m 2 /s. The temperature
of the cylinder wall varies sinusoidally between 5000 and l00°C. Assuming
that the cylinder wall behaves like a semi-infinite solid, determine the depth
into the wall in em at which the temperature amplitude has decreased to I per
cent of the surface value, and plot the heat transfer rate at the wall surface
over a complete cycle. k = 40 x 10- 3 kW/(mk). (Ans. 0·194cm, limits
±2·68 x 104 kW/mz.)
11. Write a transient program in BASIC for the section in Question

3, Chapter 4. Assume the duct is initially at 30°C throughout, and then gas at
200oc enters the duct. Find the temperature distribution and the total heat
transfer into and out of the duct at 30 and at 360 seconds after the hot gas
enters. (Ans. 30 sees: 279 kJ in, 78 x 10- 6 kJ out; 360 sees: 2304 kJ in.
18·95 kJ out.)
REFERENCES
1. Binder, L. Dissertation, Mtinchen (1910).
2. Schmidt, E. Festschrift zum siebzigsten Geburtstag August Foppl, Springer,
Berlin (1924).
3. Hsu, S. T. Engineering Heat Transfer, D. Van Nostrand Company, Inc.,
New.York, 103, (1963).
4. Bayley, F. J., Owen, J. M. and Turner, A. B., Heat Transfer, Nelson (1972).
6
Forced convection: boundary
layer principles
6.1 Introduction
In Chapter 1 it has been shown that to evaluate convection heat
transfer, the magnitude of the coefficient h in Newton's equation
has to be found. The study of convection centres round the behaviour
ofthe fluid flowing past a surface, and the subject matter divides itself
under various headings concerned with the type of flow situation
or the 'method of analysis. This chapter shows how the convection
free stream, v5
laminar transition • : • turbulent
:---------'-====---;--= I
-----x
Fig. 6.1. Bo11tulary layer growth 011 a .flat plate.
coefficient may be determined by an approximate analytical method

for simple cases of laminar flow. References to other methods will
also be given.
Some familiarity with the flow of viscous fluids is assumed, and
the growth of laminar boundary layers is illustrated in Figs. 6.1 and
78
FORCED CONVECTION: BOUNDARY LAYER PRINCIPLES 79
6.2 where the turbulent boundary layer is also shown. Thus Fig. 6.1
shows the growth of laminar and turbulent boundary layers on a
flat plate with a transition region occurring at
pv.x
--~
5 X lOs (6.1)
Jl.
where Jl. is the coefficient of molecular viscosity. The boundary layer
exists as a result of the action of viscous shear within the fluid, the
shear stress being proportional to the velocity gradient
dvx
T =Jl. dy (6.2)
The group in equation (6.1), pv.x/Jl., is the dimensionless Reynolds

number, and is the ratio of momentum forces oc pv~, to shear forces oc
Jl.Vjx. Fig. 6.2 shows the growth of a laminar boundary layer in a tube
-Et-~~tl~~:>f)-8-
f--- starting length Ideveloped
fully fully
developed
laminar turbulent
flow flow
Fig. 6.2. BoiUUimy layer growth in a tube.
with fully developed turbulent flow shown for comparison. The

starting length is the length of tube required for the boundary layer
to become fully developed. The velocity profiles follow closely the
following equations:
for laminar flow: ~v. = ~(2-

r
~)
r
(6.3)
for turbulent flow:~= (~)t (6.4)

v. r
where v is the velocity at distance y from the tu :>e wall, v. is the
velocity at the axis.
Thermal boundary layers also exist. These are flow regions where
the fluid temperature changes from the free stream value to the value
at the surface. Examples in flow over a flat plate are shown in Fig. 6.3.
+,
/.}/"
I ' .........
I ·--..... ·--......
-"-~ ...........
~t:
(b)
heanransfer from
fluid to waU
(a)
heat transfer from
wall to fluid
Fig. 6.3. Exlllllpk!l oftemperlltlll'e bollllllary layer!l.
6.2 Equations of the Laminar Boundary Layer on a Flat Plate
In an exact analysis of laminar flow convection over a flat plate,

for example, see Bayley, Owen and Turner 1, the differential equations
of momentum and energy of the flow are used to obtain the tempera-
ture gradient in the fluid at the walL and hence the convection
coefficient. In this Section the laminar flow differential equations
will be derived together with integral equations for an approximate
analysis to be introduced in the next Section.
6.2.1 The Differential Equations of Continuity, Momentum and

Energy
The control volume within the boundary layer in Fig. 6.4 is to be con-
sidered. For continuity, assuming steady state conditions with unit depth
and fluid desnity p, the mass flow rates in and out in the x-direction are
pvxdy
and
respectively and hence the net flow into the element in the x-direction
is
- povxd
- X dy
OX
Similarly the net flow into the volume in the y-direction is
OV
- p.::...:.l.dydx
oy
The total net flow in must be zero, hence
- p(OVx
ox + ~)dxdy
oy =0
Since p, dx and dy are not zero, it follows that
~~+~=0 (6.5)
ox oy
The equation of momentum arises from the application of New-
ton's second law of motion to the element, assuming the fluid is
p dy
Fig. 6.4. Elemellt of bo11111lary lllyer for co11tiludty flllll mo~Uittllm IHUfulee.
-kd ~ at a2t )
-kdy ( -+-dx
!Yax ax ax2
-kdx.1..!..
ely
Fig. 6.5. Eleme11t ofbo•tulluy layer for eM'1JY baltutce.
Newtonian (i.e., viscosity constant), an absence of pressure gradients

in the y-direction, and viscous shear in the y-direction is negligible.
The rates of momentum flow in the x-direction are pv;dy and
p[v, + (ov,/ox)dx] 2 dy for the fluid flow across the left- and right-
hand vertical faces. The flow across the horizontal faces will also
contribute to the momentum balance in the x-direction. In the
x-direction, for the bottom face the momentum flow entering is
pv,v,dx, and for the top face the momentum flow leaving is
p(v, + ~dy )(v, + ~;dx )dx

The viscous shear force on the bottom face is - p.(ov)oy) dx and on
the top face is
p.dx[av, + ~(av,)dy]
ay ay ay
so that the net viscous shear in the x-direction is p.dx(o 2 v)oy 2 )dy.
The pressure force on the left face is pdy, and on the right - [p +
(opjox)dx] dy giving a net pressure force in the direction of motion
of -(opjox)dxdy. Equating the sum ofthe net forces to the momen-
tum flow out of the control volume in the x-direction gives, after
neglecting second-order differentials and using the continuity
equation: ( avx OVX) a2 vx op 6 6)
p vx ax + vy oy = ll iJy2 - ox ( .
The energy equation may now be deduced assuming constant

properties and an absence of shear work as in a low velocity flow.
Fig. 6.5 shows the energy terms involved, and it will be seen that
there are four convective terms in addition to the conduction terms
used in deriving equation (2.7). The energy balance is simply that
rate of net conduction in + rate of net convection in = 0, hence
- [pcP(vY:; + ~t + ~; :; dy)]dxdy = 0
Using the continuity equation and neglecting the second-order terms:
at at ( iJ 2t iJ 2t)
Vx iJx + Vy iJy = IX OX2 + iJy2 (6.7)
The conduction in the x-direction is usually neglected in comparison

with other terms and hence iJ 2tjiJx 2 may be dropped from equation
(6.7). If in equation (6.7) the pressure gradient is assumed small and
is neglected a similarity is then apparent between the equations of
momentum and energy:
avx OVX - (iJ 2vx)

vx ox + vY iJy - v iJy2
at at (o 2t)
vx iJx + vY iJy = IX iJy2
v is the kinematic viscosity or momentum diffusivity, p.jp, and
vjiX = (p./p)/(k/peP) = p.c~k, which is called the Prandtl number, Pr.
If v = IX, then Pr = 1, and the pair of equations will lead to identical
non-dimensionalised solutions of vx and t as functions of y. The
Prandtl number is the ratio of fluid properties controlling the
velocity and temperature distributions, and it can vary between
around 4 x 10- 3 for a liquid metal to the order of 4 x Hf for a
viscous oil.
6.2.2 The Integral Momentum and Energy Equations of the Laminar
Boundary Layer
To consider the motion in the boundary layer, an elemental control
volume is chosen that extends from the wall to just beyond the limit of
the boundary layer in they-direction, is dx thick in the x-direction, and
has unit depth in the z-direction. This is shown in Fig. 6.6. An equation
is sought which relates the net momentum outflow in the x-direction to
the net force acting in the x-direction.
Fig. 6.6. Elementlll control JJOhune inl'~'llillllr bmllllillry t.yer.
The momentum flow across the face AB will be
fo pv~ dy
6
Similarly, the momentum flow across the face CD will be

Fluid also enters the control volume across the face BD at the rate
J"
d opvxdydx
dx
This is the difference between the fluid leaving across face CD and
entering across face AB. The fluid entering across face BD has a
velocity vs in the x-direction, hence the flow of momentum into
the control volume in the x-direction is
Vsdx
d J" PVx dy dx
0
Hence the net outflow of momentum in the x-direction is
J"
d o pv; dy dx - vsdx
dx d o PVx dy dx J"
Pressure forces will act on faces AB and CD, and a shear force
will act on face AC. There will be no shear force on face BD since
this is at the limit of the boundary layer and dvx/dy = 0. The net
force acting on the control volume in the x-direction will be
Px(> - ( px + -dpx )
dx dx (> - T
w
dx = - dpx dx - r dx
~
dx w
(6.8)
The pressure gradient may be neglected as small compared with

the shear force at the wall, and the equality of the net momentum
outflow to the net force gives
dx
d J" PVx(Vs -
0 vx) dy = Tw (6.9)
This is the integral equation of motion in the laminar boundary

layer, and was first derived by von Karman. 2
The integral energy equation may be derived in much the same
way. In this case, a control volume extending beyond the limits of
both temperature and velocity boundary layers may be considered
initially, Fig. 6.7. The principle of conservation of energy applied
to this control volume will involve the enthalpy and kinetic energy
of fluid entering and leaving, and heat transfer by conduction at the
wall. Kinetic energy may be neglected as being small in comparison
velocity boundary
layer, b
Fig. 6.7. Control wbune in tempertltUTe tuUl flelocity boiUfllary layers.
with other quantities. The enthalpy flow rate across face AB is
and across face CD
f y. CpPVxt dy + dxd fy. CpPVxt dy dx

0 0
Fluid will also enter the control volume across face BD at the rate
d fy.
dx o PVxdydx
Again this is the difference between the flow rate out at face CD
and in at face AB. The enthalpy flow will be
Finally, heat transfer by conduction across the wall at AC will

amount to
-k dx( 01 )
oy y=o
For conservation of energy:
d
cpts -d
X
J"" PV.x dy dx - -dd JY· pcPtv.x dy dx -
0 X 0
k dx (ot)
,--
uy y=O
= 0 (6.10)
Beyond the limit of the temperature boundary layer, the temperature
is constant at t., and hence the integration need only be taken up
to y :r <5 1• Equation (6.10) therefore gives
-d J,)' (t. - t)v.x dy - a (ot)

- = 0 (6.11)
dx 0 oy )'=0
This, then, is the integral energy equation of the laminar boundary
layc:r.
6.3 Laminar Forced Convection on a Flat Plate
The integral equations (6.9) and (6.11) will now be applied to the
problem of laminar forced convection on a flat plate. The method
is due to Eckert. 3 The analysis assumes the viscosity is uniform with
temperature. The first step is to use the integral equation of motion
to derive an equation for boundary layer thickness. The velocity
contour may, for example, be assumed a polynomial
v.x = a + by + cy 2 + dy 3
where a, b, c, and d are constants. The constants may be found by
applying known boundary conditions. Thus vx = 0 at y = 0, and
hence a = 0. Also v.x = v. at y = <5, and (ovJoy),) = 0 at y = <5.
Further, since both v"' and vy are zero at y = 0, it follows from (6.5)
that rPvJoy 2 = 0, at y = 0. These results lead to
b = ~2 v. d v.
<5' c = 0, = - 2<53
and hence
~: = ~(~) - ~(~r (6.12)
Applying the integral equation of motion,
d
-d-
J,) pv.x(v. - v.x) dy = <w
X 0
The wall shear stress is found by considering the velocity gradient

at y = 0; this is found to be 3vJ2a. The above equation leads to
d 2 39(J 3 v.
dx pv. 280 = 2f'-g
pv2 d(J = ~. 280 JlVs dx
• 2 39 (J
140 v
(Jd(J = -·-dx
13 v.
On integration
(J 2 l40vx
-=--+C
2 13v.
C = 0, since (J = 0 at x = 0
(J 2 = 280vx
l3v.
or
4·64
(6.13)
X RetX
This result, due to Pohlhausen,4 is required later on in considering
the integral energy equation.
The temperature distribution in the boundary layer is assumed to
follow a similar law to the velocity distribution. Thus:
(} = (t - tw) = dy + ey2 + fy3
where, again, d, e, and f are constants. The boundary conditions
are that at y = b1 (the thickness of the temperature boundary
layer), (} = 0, and also (il(}joy)t}, = 0. Also, from equation (6.7) it
follows that (o 2 0joy 2 )y=o = 0 because vx and vY are both zero at
y = 0. From these conditions it follows that
e
o.
= 0, != - 2(J3
I
and hence
(6.14)
Turning to the integral energy equation, the substitutions
e= (t - tw) and e. = (t. - twl are made to give
j_fb, (e. -
dx o
e)vx dy - a (:e)y y:O
= 0 (6.15)
From the temperature equation (6.14) it follows that
ae)
(oy 3e.
a y;O = a2b 1
This result is substituted in equation (6.15) together with the ex-
e
pressions for and vx to give:
ddxf:. [e.- ~(i)e. +~(ite.J ·[~(~)v.- ~(~r v.J cty =a~::

A useful substitution is that). = b1/b.
3e.
= a2).J
This then leads to
d [ (3). 3). 3 )] 3e.

dx e.v.bt 20 - 280 = a2).15
It is convenient here to neglect the term 3). 3 /280 as being small in

comparison with 32/20. This is justified since ). has the value of 1
if Pr = 1, and will not be far removed from 1 at other values of Pr
fairly close to 1. Hence
--=a--
d 3.A.J.
dx 20
3
2.A.Jv.
d 2 lOa
dx (). b) = v•.A.J
22 e+
dx
22 dJ
dx
= 1oa
v.A.J
2).2J2d). + ).3~ = lOa

dx dx v.
Equation (6.13) for~ may now be substituted.
~ = 4 ·~. and hence ~= 4·64(xv)t

x Re" v.
2A. 2 21·6xv.dA. + A. 3 .21·6v = lOoc
v. dx 2 v. v.
(6.16)
This equation may be solved by making the substitution A. 3 = p,

and p = xn, and the solution obtained is:
( ~~1 ) 3 = ()-93
Pr
+M
xt
noting that oc/v = Pr, and M is a constant of integration. The thick-
ness of the thermal boundary layer will be 0 at the beginning of the
heated section, at x = xh, say, and hence
M = _ 0·934
Pr
and finally :
This result may be simplified further by assuming that the plate is

heated along its entire length, or xh = 0, in Fig. 6.8,
hence ~~ = (<>-93)t
Pr
}
i = p~t
(6.17)
or, approximately,
r~u X
Fig. 6.B. Ltuniluu forced cOIIW!Ctioll 011 a flat p/tlte. Helltillg COIIIIIIeiiCel .lia
from tl¥ letuling ··dge.
Using the equations for !50 !5, and the value of(o8joy)y=o it is possible
to determine the heat transfer at the wall, since
qw = k(~uy8 )y=O = k 38•, from (6.14)

2!51
The heat transfer rate at the wall is expressed non-dimensionally.
qw/8, is the heat transfer coefficient h, and the group hxjk is the
dimensionless Nusselt number, Nu. It is interpreted as the ratio
of two lengths, the characteristic linear dimension of the system,
and an equivalent conducting film of thickness ~5;. Figure 6.9 shows
~--t, ~
lt,.-.i
Fig. 6.9. To illustrate the significance of the Nusselt number.
how ~5; is defined. The heat transfer at the wall is qw = h8, and may
be expw;sed as qw = (k/!5;)8,. It follows that h = k/15; and hence
hx x
Nu=k=-;5;
The linear dimension of the system is generally large in comparison

with ~5;.
A Nusselt number may therefore be obtained:
qwx 3x 3x Re! Prt
Nu =-=-=---7---
x O,k 2!51 2(0·93)t4·64x
using (6.17) to eliminate b0 and (6.13) to eliminate !5.
Nux = 0·332Re! Prt (6.18)
This gives the local Nusselt number at some distance x from the
leading edge of the plate. The average value of the convection
coefficient h, over the distance 0 to x is given by:
1i 1
=-
fx h dx
X 0
where
h = 0·332k ( ~~r Pr-i, from equation (6.18)
fi k 0· 332
= - - - --
(v·x) t Prt
X f V
= 0·664k ( ~~) t p,t

and
Nu X = 0·664 RetX Prt (6.19)
This equation expresses in non-dimensional form the heat transfer
by convection at the surface of a flat plate.
ExAMPLE 6.1
Air flows at 5 m/s along a flat plate maintained at 77°C. The bulk
air temperature is 27°C. Determine at 0·1 m ,0·5 m and 1·0 m from
the leading edge, the velocity and temperature boundary layer
thicknesses, and the local and average convection coefficients. Use
mean properties of air from Table A6.
Solution. At 325 K, p = 1·087 kgjm\ k = 28·1 x 10- 6 kW/(m K),

Jl. = 1·965 x 10- 5 Pas, and Pr = 0·703.
The Reynolds numbers at x = 0·1, 0·5 and 1·0 m with v. = 5·0 m/s;
together with the boundary layer thicknesses using fJjx = 4·64/Re!,
equation (6.13), and [)J[) = 1/Prt, equation (6.17), and the local
and average coefficients using equations (6.18) and (6.19), are cal-
culated and tabulated below:
{Jmm
X= 0·1 2·76 X 104 2·79 3·14 13-8 X 10- 3 27·6 X lQ- 3

0·5 1·38 X lOS 6·26 7·04 6·15 X 10- 3 12·3 X 10- 3
1·0 2·76 X lOS 8·92 10·02 4·37 x w- 3 8·74 x w- 3
6.4 Laminar Forced Convection in a Tube

Laminar forced convection in a tube will be considered for the case
of fully developed flow and constant heat flux at the wall. For fully
developed flow it may be assumed that the velocity profile has a
parabolic shape. It is first necessary to derive the energy equation
for flow in a tube. To do this, a small cylindrical element of flow
may be considered, as in Fig. 6.1 0. The element is of length dx,
radius r on the inside, and radius r + dr on the outside. Energy will
flow into and out of the element in the radial direction by conduction,
I I
· dx1
;,
Fig. 6.10. Element of laminar flow in a tube.
and in the axial direction by convection. Conduction into the

element is
. ot
Q, = -k2nrdx-
or
Over the change of radius dr, this conduction rate will change by
oQ, dr = - k2n dx ~ (rot ) dr (6.20)

or or or
This change in conduction rate is accounted for by the difference
between the convection rates into and out of the element in the
axial direction. The axial velocity through the element is constant
but the temperature changes in the axial direction. The rate of
convection into the element is
2nr dr pvcPt
and out of the element, it is
2nr dr pvcP( t + ;~ dx)

Hence, the difference is
ot
21rr dr pvcPox dx
The sum of this and the right-hand side of (6.20) is zero, hence on
cancelling terms,
~~(/')=peP~ (6.2.1)
vr or or k ox
This is the energy equation for laminar flow in a tube. With a con-
stant wall heat flux qw, and constant fluid properties, the temperature
of the fluid (at any radius) must increase linearly in the direction of
flow. so that
ot
ox = constant
Other conditions applicable are that at r = 0, (tube axis), otfor = 0

=
and at r rw, t =tw. Also at r =rw,
the heat flux is related to the
temperature gradient,
Since otfox is assumed constant, equation (6.21) reduces to a

total differential equation. The velocity vis a function of the velocity
at the axis of the tube, v., and the radius r. The assumed parabolic
velocity distribution, equation (6.3), expressed in terms of r measured
from the axis is
~=
v.
1- (!_)2
rw
where r w is the wall radius.
This result must be substituted into equation (6.21) before inte-
grating. Hence, after re-arrangement,
~ (r01 ) = ~a ~v.
or or ox
[1 - (!_)
rw
]r
2
This is integrated to give
rot = ~ ~v. ( r2 -
or a OX 2 4r;
_c_) + c 1
and after a second integration, gives
ot ('- 2 - -r 4- )
t = -1 -va + C 1 In r + C 2 (6.22)
ex ox 4 16r!
C 1 and C 2 are constants of integration to be found from boundary

conditions. Since otjor = 0 at r = 0, it follows that c I = 0. The
other boundary condition is that at r = rw, t = tw. Hence,
c2 = t
w
-!CX ~v
OX a
3';
16
Hence equation (6.22) becomes., after some re-arrangement,
t = ~v r [!(!._)
!cxoxaw4rw
2
2
- _!_(!._)
16rw
4
- 2_]
16
+ t
w
(6.23)
This equation may be expressed as a temperature difference,

(} = t - tw. Further, if (}a is the temperature difference between the
axis, where r = 0, and the wall, (}a may be found from equation
(6.23) by putting r = 0. Hence,
(} = !CX ~v r 2 (- 2_) (6.24)

a OX a w 16
The temperature profile may be expressed non-dimensionally by

dividing equation (6.23) by equation (6.24).
(6.25)
The equation for heat transfer at the wall may be obtained by con-
sidering the temperature gradient at r = rw. Thus, from equation
(6.25)
and
In terms of the Nusselt number, Nud
(6.26)
This value of Nusselt number is based on the difference in tem-

perature between the tube axis and the wall. However, from a
practical point of view it is more convenient to consider the difference
in temperature between the bulk value and the wall. The bulk
temperature is the mean temperature of the fluid, and the tempera-
ture difference required is given by
(}
r- 2nr dr pvc/}
=-o_ _ _ __
m J~" 2nr drpvcp

Introducing equations for v and (}, this becomes
This, on integration, gives (}m = #(}a.

The heat transfer at the wall is,
q = _ k(d(}) = 4k(}a = 4k . 72 (}
w dr rw 3r w 3r w 44 m
This is now equivalent to
Hence
h=~.n
3rw 44
and
hd 4k 72 2rw
Nu 11 =-=-·-·-
k 3rw 44 k
8 72
= 3.44
= 4·36 (6.27)
These results are independent of Reynolds number because, for
fully developed flow, the boundary layer thickness is equal to the
tube radius.
ExAMPLE 6.2
Water at a mean temperature of 40°C flows at a mean velocity of
o-1 m/s in a 3 mm bore tube having a constant wall flux of 1-Q k W jm 2 •
Determine the temperature of the water as a function of radius,
using equation (6.23). Use fluid properties from Table AS.
Solution. It is necessary to obtain iJtjiJx, also v. the velocity of water

at the centre of the tube.
Consider an energy balance on 1 m of tube:
qw1td x 1 = mass flow x cP x temperature rise
at =
.·. ox qw1tdj(mass flow x cP)
1·0 X 1t X 3 4 X 106
= 4·178 X 1000 X -1t-X-32,.--x_0_·_1_x_9_9_4-·6
where cP = 4·178 kJ /(kg K~ and p = 994·6 kg/m 3

:. iJtjiJx = 3·22K/m
The laminar flow velocity distribution is given by
(v/v,) = 1 - (r/rw) 2
where v. = velocity at r = 0, hence for continuity:
Vm1tr! = t"' 21tr[v.- V8 (rfrw) 2 ] dr
21tr 2 v
= 1trw2 va - ~
4 = 1trw2 va ;2
:. V8 = 2vm
With IX= 15·1 x 10- 8, equation (6.23) becomes:
15·1 10 4 rw
2
t = 108 X 3·22 X 0·1 X 2 X 1.526 [!(!_) - _!_(!_)
16 'w
4
- ~]
16
+t w
2
= 9·6[!(!_) - _!_(!_)
4 r.., 16 rw
4
- ~]
16
+ tw
At r = rw> t = tw; at r = 0, t = tw - 1·8°C.
PROBLEMS
1. Derive the heat flow equation of the boundary layer
d f'
dx o (0. - O)U dy = rx (d(J)
dy o
and apply this equation to 'slug' flow of a liquid metal along a plate of uniform
temperature to find the thickness of the temperature boundary layer. (J would
be the liquid metal temperature relative to the plate temperature. Assume
that the temperature ptofile in the boundary layer can be described by an
equation of the form
where a, b, and care constants to be determined from the boundary conditions.

Hence prove that the local Nusselt number N" is given by
N" =j(x ~ 2)J(RJ>) = 0·534J(RJ>)
It can be shown that if the velocity profile can be approximated by an

equation of the form
the velocity boundary layer thickness is then given by
0
~=
j[ 2x
(4- x)R"
2
J
Show that for a liquid metal of P = 0·01 the temperature boundary layer
thickness is approximately equal toM. (University of Bristol).
2. Prove that, in hydrodynamically fully-developed laminar flow through a

tube, the temperature field is determined by the following partial differential
equation
1 a (at) a1(at)
Ur ar rar = ax
where r is the distance from the axis of the tube, and U is the velocity at r.
Hence derive an equation for the fully developed temperature profile,
when the heat flux qw is constant along the wall of the tube. You may assume
that the velocity profile is given by
Show that the temperature profile can be put into dimensionless form as
~ = .! = 1 - ~(!..)2 + !(!..)4
to - tw Bo 3 R 3 R
where t, 10 , and tw are the local, axial, and wall temperatures respectively,
and R is the radius of the tube. Also show that the Nusselt number
qwd 8
80 k 3
Explain, by writing down the initial equations, how you would derive the
Nusselt number qwdf(Jmk, where (Jm is the bulk temperature of the fluid relative
to the wall. (University of Bristol).
3. Show that if a flat plate has a heated section commencing at xh from the
leading edge, the local Nusselt number at distance x from the leading edge,
(x > xh), is given by:
Nu" = 0·332 Re! Prt(I - (xh/x)*)-t
Determine the velocity and th~rmal boundary layer thicknesses and the local
heat transfer rate at 1 m from the leading edge of a plate heated 0·5 m from
the leading edge, for air at 27aC flowing over the plate at 0·5 mjs, if the
temperature of the heated section is 127°C. (Ans. ~ = 0·0298 m, ~. =
0·0243 m, 0·184 kWjm 2 .)
4. The velocity in the boundary layer of a stream of air flowing over a flat
plate can be represented by
~ = ~(~)- ~(~r
where U is the main stream velocity, u the velocity at a distance y from the
flat plate within the boundary layer of thickness b. The variation of boundary
layer thickness along the plate may be taken as
b/x = 4·64(Re,rt
If the plate is heated to maintain its surface at constant temperature show
that the average Nusselt number over a distance x from the leading edge of
the hot plate is
Nu = 0·66(Pr)t(Re,.,>t
(University of Leeds).
5. If in laminar flow heat transfer on a flat plate the velocity distribution

is given by V, = V.(yfb1 and assuming in this case that there is no shear at
the limit of the boundary layer, show that the boundary layer thickness is
given by
bfx = 3-46/Re!
where {J is the boundary layer thickness at x from the leading edge. Also
show that the average Nusselt number at x is given by
Nu, = 0·73 Re! Prt
with heating commencing at x = 0.
REFERENCES
1. Bayley, F. J., Owen, J. M. and Turner, A. B. Heat Transfer, Nelson (1972).
2. Klinruin, T. von, Z. angew. Math. u. Mech., Vol. 1, 233 (1921).
3. Eckert, E. R. G. and Drake, R. M. Analysis of Heat and Mass Transfer,
McGraw-Hill, New York (1972).
4. Pohlhausen, K. Z. angew. Math. u. Mech., Vol. 1, 252 (1921).
7
Forced convection: Reynolds
analogy and dimensional analysis
Consideration of convection has so far been limited to laminar flow

For turbulent flow, it is possible to introduce additional terms into
the momentum and energy equations to account for the presence of
turbulence, and to obtain numerical solutions to the finite difference
forms of the equations. 1 • 2 However, these methods have only
become possible with the use of the more recent and more powerful
generations of digital computer, and at an introductory level the
more classical approaches will be followed.
7.1 Reynolds Analogy

The approach to forced convection known as Reynolds analogy is
based on similarities between the equations for heat transfer and
shear stress, or momentum transfer. The original ideas were due
to Reynolds 3 • 4 and the analogy has been subsequently modified and
extended by others.
The equation for shear stress in laminar flow, (6.4), may be written
as
dv
't = pv-
dy
(7.1)
where v is the kinematic viscosity, Illp. A similar equation may be

written for shear stress in turbulent flow. A term e, eddy diffusivity,
is introduced, which enables the shear stress due to random turbulent
motion to be written
dv
-r, = pe dy (7.2)
101
When turbulent flow exists, the viscous shear stress is also present
which may be added to <1 • The total shear stress in turbulent flow
is thus
dv
• = p(v + e) dy (7.3)
e is not a property of the fluid as v is. It depends on several factors

such as the Reynolds number of the flow and the turbulence level.
Its value is generally many times greater than v.
7.1.1 Shear Stress at the Solid Surface
In developing Reynolds analogy the heat transfer at the surface of a flat
plate or of a tube is ultimately compared with the shear stress acting at
that surface. This shear stress is obtained by substituting (dv/dy)y=O into
the equation forT. Thus, for laminar flow on a flat plate, x from the
leading edge, with the Reynolds number Rex based on the free stream
velocity and x,
Cf = 0·647 (7.4)
Rei
where Cf is the skin friction coefficient defined as <w/!pv:. v. is the
free stream velocity. An average value Cd for the length x is found to
be 2Cjfor laminar flow, where Cfis the local value at x. The derivative
of the turbulent velocity profile substituted into (7.2) leads to an
infinite shear stress at the wall. This is overcome by assuming the
existence of a laminar sub-layer, as in Fig. 6.1. For turbulent flow
on a flat plate, Cf and Cd are given by
Cf = 0·0583(Rex)-t (7.5)
and
Cd = 0·455 (7.6)
(log Rexf·SS
Equation (7.6) is an empirical relationship, 5 which takes into account
the laminar and turbulent portions of the boundary layer.
The ratio of the velocity at the limit of the laminar sublayer to
the free stream velocity is also of importance, as will be seen later;
this is a function of the Reynolds number at x from the leading edge:
vb 2·12
v, (Rex)O·l
(7.7)
FORCED CONVECTION: REYNOLDS ANALOGY 103
Corresponding relationships for flow in tubes are usually ex-

pressed in terms of a friction factor f, which is four times larger
than Cf in terms of the surface shear stress. Thus f = 4rwftpv~,
where vm is the mean velocity of flow
In laminar flow, !=~ (7.8)

Red
and in turbulent flow, f = 0·308 (7.9)

(Red)*
and vb 2·44
(7.10)
Vm = (Red)t
The derivations of these relationships may be found in the more

advanced texts on heat transfer, or fluid mechanics, e.g., refs. 6, 14.
The friction factors quoted above are for smooth surfaces.
Values are increased if the surface is rough. For any tube surface,
the average wall shear stress rw acting over a length L can be found
by considering the forces acting. Thus, if tlp is the pressure loss and
d the tube diameter, the pressure force ilpnd 2 /4 is equal to the wall
shear force rwndL, assuming the tube is horizontal.
7.1.2 Heat Transfer across the Boundary Layer
Equations for heat transfer across the boundary layer are written in
analogous form to (7.1) and (7.3). Thus in laminar flow, heat transfer
across the flow can only be by conduction, so Fourier's law may be
written as dt
q = -pc a~ (7.11)
p dy
In turbulent conditions energy will also be carried across the flow

by random turbulent motion, and the heat flux may be written
dt
q = - pcp(a + eq) dy (7.12)
where eq is the thermal eddy diffusivity, a term analogous to e.

The basis of Reynolds analogy is to compare equations (7.1) and
(7.11) for laminar flow, and equations (7.3) and (7.12) for turbulent
flow.
In equations (7.3) and (7.12) it has been seen that the ratio vjrx is
the Prandtl number; similarly ejea is known as the turbulent Prandtl
number, though this is not a property of the fluid as is vjr:x.
Some initial assumptions must now be made. The first is that
e = eq. This means that if an eddy of fluid, at a certain temperature
and possessing a certain velocity, is transferred to a region at a
different state, then it assumes its new temperature and velocity in
equal times. This assumption is found by experiment to be approxi-
mately true. (eq/e varies between 1 and 1·6. For a review of this
subject, see ref. 6.) A second assumption is that q and -r have the
same ratio at all values of y. This will be true when velocity and
temperature profiles are identical. Identical profiles occur in
laminar flow when the Prandtl number of the fluid is 1. In turbulent
flow, with e = eq, the groups responsible for velocity and tem-
perature distributions, (v + e) and (r:x + eq), are also equal when
Pr = 1. Further, even when the Prandtl number is not 1, (v + e)
and (r:x + eq) will be nearly equal, since e and eq are very much
greater than v and r:x.
The simple Reynolds analogy is valid when Pr = 1, and the
Prandtl-Taylor modification 7 • 8 which takes into account a varying
Pr is valid for a fairly restricted range, say 0·5 < Pr < 2·0.
7.1.3 The Simple Reynolds Analogy
With the assumptions noted above it is now possible to proceed to a
consideration of the simple analogy. Flow is assumed to be either all
laminar or all turbulent, and Pr = 1. By comparing equations (7 .1) and
(7 .11) for laminar flow, it follows that
q k dt
(7.13)
T ,u dv
This gives the ratio of qj-r at some arbitrary plane in the flow.
Noting that qj-r has the same value anywhere in the y-direction,
it is possible to express qw/Tw at the wall in terms of free stream and
wall temperatures and velocities.
Thus
(7.14)
v.
Details of the nomenclature are shown in Fig. 7.1. vw at the wall is

zero.
Fig. 7.1. Velocity and temperature distributions for the simple Reynolds
tuUllogy.
For turbulent flow, equations (7.3) and (7.12) give

fl. = pcp(r:x. + sq) dt
-r p(v + s)dv
Thus, between the free stream and wall :

qw (t, - tw)
-=C--- (7.15)
'rw p V,
Equations (7.14) and (7.15) for laminar and turbulent flow are
clearly identical if Pr = 1, i.e., if cP = k/ Jl., or Jl.Cp/k = 1. Re-arranging
equation (7.15) gives
where e. = (t. - tw), and where h is the surface heat transfer co-
efficient.
Substituting the skin friction coefficient Cf gives
or
h Cf
(7.16)
pv,cP = 2
This is one form of the result obtained from the simple Reynolds
analogy; it gives the convection coefficient h in terms of the skin
friction coefficient Cf. h/pv,cP is the Stanton number St. It is the
N usselt number divided by the product of the Reynolds and Prandtl
numbers. Further re-arrangement is possible; for example, con-
sidering laminar flow at dist~nce x from the leading edge of a flat
plate, both sides of(7.16) are multiplied by xfk to give
hx Cf pv.xcP
T=T-k-
But cpJ.L/k = 1, or cP/k = 1/J.L, hence
hx Cf pv.x
---
k 2 J.l
or
Cf
Nu X =-Re
2 X (7.17)
Cfmay be replaced by 0·647(Rex)-t from equation (7.4) to give

Nux= 0·323(Rex)t (7.18)
for laminar flow on a flat plate. This result may be compared with
equation (6.18) obtained by consideration of the integral boundary
layer equations. If Pr = 1 in this equation then the result is
Nux = 0·332(Rex)t
Reynolds analogy may also be applied to flow in tubes, and for this
o.
purpose and v. in the above analysis may be replaced by the mean
values ()m and vm, since the velocity and temperature distributions
are identical. The linear dimension is now the diameter of the tube, d.
The relationship will be
hd Cf pvmd
---
k 2 J.l
or
(7.19)
For turbulent flow in tubes, f = 0·308(Re4 )-t from (7.9) and

Cf = !f from the definition off Substituting for Cf in (7.19)
gives
(7.20)
7.1.4 The Prandtl-Taylor Modification of Reynolds Analogy

The simple Reynolds analogy agreed quite well with experiment in
laminar flow and also with results where Pr = 1 in both laminar
and turbulent flow. The modification proposed by Prandtl and
Taylor goes a long way to meeting the discrepancies generally
found in turbulent flow when there is no restriction on Pr. A laminar
sublayer is considered in addition to the turbulent boundary layer.
This makes an important difference to the analysis even though the
sublayer is quite thin. The fact that it is thin is also important in
that it makes it possible to assume a linear temperature and velocity
distribution with negligible error.
For turbulent heat and momentum exchange between the free
stream and the laminar sublayer, as in Fig. 7.2, applying equation
(7.15) gives:
(7.21)
v.
Fig. 7.2. Velocity fllfll temperature distributions for the Prtllldti-Tay/or

modification of Reynolds flll(l/ogy.
In the laminar sublayer, the equations are

and
and hence
(7.22)
Because the velocity and temperature distributions are straight

lines in the laminar region qw = qb, and •w = •b· Hence the right-
hand sides of (7.21) and (7.22) are equal.
cp(ts - tb) k (tb - tw)
and
If (t,- tw) is written as 0,, then the above may be re-arranged to

give
(), vb
1 + -(Pr- 1)
v,
and eliminating (tb - tw)/vb between this result and equation (7.22)
gives
Pr
Jl.Vs 1 + vb (Pr - 1)
v,
(7.23)
This equation is Reynolds analogy as modified by Prandtl and

Taylor. It may be noted straight away that if Pr = 1 in this equation,
then the relationship reduces to equation (7.15), i.e., Reynolds
original equation. Further, if vb = 0, i.e., there is no laminar sub-
layer so that flow is entirely turbulent, the equation again reduces
to the original relationship. A further simplification is that if flow
is all laminar, which means that vb = v,, equation (7.23) becomes
qw cp()s k(),
'tw v,Pr V5 Jl.
Equation (7.23) may now be treated in a similar manner to (7.15)

by re-arranging and introducing the coefficient Cf. Thus:
qw CJ
-0 = pv.cP-
1
(7.24)
2 vb
1 + -(Pr- 1)
s
v.
For turbulent flow on flat plates, both sides are multiplied by x/k
and 11 is introduced to the right-hand side to give
qwx pv.x Cp/1 Cf 1
O.k = 7
k 2 . 1 + vb (Pr _ 1)
v.
Cf RexPr
Nux = 2 ----'-'---
1 + vb (Pr - 1)
v.
Also, for turbulent flow on flat plates, equations (7.5) and (7.7) are
introduced to give t
0·0292RexPr
NUx = 1 + 2·12Re;io(Pr - 1) (7·25 )
This is the lpcal Nusselt number. To obtain an average Nusselt

number over some total length of plate, Cd from equation (7.6)
may be substituted for Cf in this analysis.
An alternative to this result was suggested by Colburn, 9 in which
the denominator in equation (7.25) was replaced by Prt. Re-
arranged, this gives
StxPrt = 0·0292Re; 0 · 2 (7.26)
and if Cf is substituted, this gives
Cf
St Prt =- = J (the Colburn ]-factor) (7.27)
X 2 '
This result reduces to equation (7.16) when Pr = 1.

For turbulent flow in round tubes, equation (7.23) may be suit-
ably modified. (}. becomes (}m, the temperature difference between
the mean fluid temperature and the wall, and v. similarly becomes
vm. Introducing k, Jl, and the linear dimension d, gives
qwd = PVmd CpJl. Cf . 1
(}mk J1 k 2 1 Vb ( )
+-Pr-1
vm
Nu 11 = Cf. Re11Pr
2 vb
1 + -(Pr- 1)
Vm
Finally, equations (7.9) and (7.10) are introduced to eliminate Cf

and vJvm, and remembering that f = 4Cf, the result obtained is
Nud = 0·0386RejPr (7.28)
1 + 2·44Re11 *(Pr - 1)
This is an average Nusselt number, because an average friction
factor was used.
The relationships (7.25) and (7.28) agree remarkably well with
experiment over a small range of Prandtl number.
EXAMPLE 7.1
Compare the heat transfer coefficients for water flowing at an
average fluid temperature of 100°C, and at a velocity of 0·232 m/s
in a 2·54 em bore pipe, using the simple Reynolds analogy, equation
(7.20), and the Prandtl-Taylor modification, equation (7.28). At
100°C, Pr = 1·74, k = 0·68 X w-J kWj(m K), and v = 0·0294 X
10-s m 2 js.
Solution. The Reynolds number is:

vd 0·232 X 0·0254 X 10 5 = 20 ()()()
v 0·0294 '
In the simple analogy, Nud = 0·038Re~· 75 , and Re~· 75 = 1643
and r: = 62·5 X 0·68 X J0- 3

Nud = 62·5, n
0·0254
= 1·675 kW/(m 2 K)
In the Prandtl--Taylor modification,
- 0·0386Re~· 7 5 Pr
Nu 11 = --------:-----
1 + 2·44(Re11)-t(Pr - 1)
Rej = 3·45
Nu = 0·0386 x 1643 x 1·74 = 72 .4
d 1 + (2·44/3·45) X 0·74
72·4 X 0·68 X lQ- 3
1i = 0 .0254 = 1·937 kW;(m 2 K)
The first answer is thus 13·5 per cent lower than the second, which
may be assumed more correct. This solution is for flow in smooth
pipes.
7.2 Dimensional Analysis of Forced Convection

Convection heat transfer is an example of the type of problem
which is difficult to approach analytically, but which may be solved
more readily by dimensional analysis and experiment.
The process of dimensional analysis enables an equation to be
written down which relates important physical quantities, such as
flow velocity and fluid properties, in dimensionless groups. The
precise functional relationship between these dimensionless groups
is determined by experiment.
Suppose that in a given process there are n physical variables
which are relevant. These variables, which may be denoted by
Q 1 , Q 2 , •.. , Q". are composed of k independent dimensional
quantities such as mass, length, and time. Buckingham's pi theorem 10
states that if a dimensionally homogeneous equation relating the
variables may be written, then it may be replaced by a relationship
of (n - k) dimensionless groups.
Thus, if
then
</J 2( 1T: 1 • 1T: 2 • · · · ' 7t(n- k)) = I
Each n term will be composed of the Q variables, in the general
form
1T: = Q~Q~Q3 ... Q~,
and will be dimensionless. The set of n terms will include all inde-
pendent dimensionless groupings of the variables. No n term can
be formed by combining other n terms. A set of equations for
a, b, c, ... , x is obtained by equating the sum of the exponents of
each independent dimension to zero. This will yield k equations
for n unknowns. One method of solution is to choose values for
(n - k) of the exponents in each term. The selected exponents must
be independent, which can be shown to be true if the determinant
formed from the coefficients of the others does not vanish.
An alternative procedure is to select k of the Q variables and to
combine them in turn with each of the other (n - k) Q variables.
The selection k of the Q quantities must together involve all the
independent dimensions, but they must not form a dimensionless
group by themselves. Further, each of the (n - k) Q variables in
each n term is given the exponent 1. This facilitates the algebra,
as will be seen, and is allowable since it only amounts to reducing
the n term by some unknown root. Thus, if there were six Q variables
and four independent dimensions, the two 1t terms would be:
7tt = Q~·Q~•Q)•Qi•Qs
7t2 = Q~2Q~2Q32Qi2Q6
In each 1t term there are therefore four simultaneous equations for

the four unknown exponents.
This procedure will now be applied to forced convection. For a
detailed mathematical proof of the pi theorem, the reader is referred
to Langhaar.U
The physical variables are selected by consideration of the govern-
ing differential equations, e.g. (6.6) and (6.7) for laminar flow. The
dependent variable is the convection coefficient h, and for an incom-
pressible fluid in the absence of viscous dissipation, the independent
variables are a velocity v, a linear dimension I, and the fluid pro-
perties of thermal conductivity k, viscosity p., specific heat cP, and
density p. The presence of turbulence does not add any further
variables. The velocity and linear dimension are normally those
which define the Reynolds number for the flow, e.g., free stream
velocity and distance from leading edge for flow along a flat plate,
and mean velocity and diameter for flow in a tube.
The independent dimensional quantities to be used are mass M,
length L, time T, temperature (}, and heat H. Heat, of course, is not
independent as it has the same dimensions as kinetic energy,
ML2 jT 2, but for present purposes it can be regarded as independent
provided there is no transference of energy from one form to
another. Heating effects due to fluid friction are consequently
neglected, and the results are invalid for high speed flow. Inspection
of the dimensions of the physical variables shows that when the
dimensions of H and (} occur, (in h, k, and cp), they do so in the
same combination of H/0. Thus H/0 can be regarded as an inde-

pendent dimensional quantity.
In forced convection there are therefore seven physical variables
involving four dimensional quantities. Consequently, three n terms
will be obtained. Four variables which together involve all four
dimensions, and which do not themselves form a dimensionless
group, are v, I, k, and J.L. Then h, cP, and p will each appear in a
separate independent n term. The n terms are
1t 1 = va'l"'it'J.Ld'h
1[2 = va2J"2kt2J.ld2cp
nJ = va'l"'k•'J.Ld'P
The n 1 term may be written
(L)a'
T
( L)"'(_!!__).'( M)d'__!!_
LTO LT UTO
which is dimensionless. The following equations for a 1, b 1, c 1,

and d 1 are obtained:
L a 1 + b 1 - c 1 - d1 - 2 = 0
T -a 1 - c1 - d1 - 1= 0
Hj() c1 +1= 0
M d1 =0
It is found that a 1 = 0, b 1 = 1, c 1 = -1, and d 1 = 0. The n 1 term
is thus hljk. In a similar manner, it is found that the n 2 term is
fLCp/k and the n 3 term pvl/J.L. These groups are recognized as the
Nusselt, Prandtl, and Reynolds numbers, and the result may be
expressed:
c/J 2 (Nu, Pr, Re) = 1
or, more usually,
Nu = c/J(Re, Pr) (7.29)
since the Nusselt number contains the dependent variable h.
Equation (7.29) agrees in form with Reynolds analogy, in that the
Nusselt number is a function of the Reynolds and Prandtl numbers.
Actual functional relationships have been determined for various
fluids, geometries, and flow regimes; these may be used to predict
h in similar circumstances, provided the Reynolds and Prandtl
numbers fall within the same ranges. There is, of course, no restric-
tion to the system of units which may be used, provided they are
consistent.
Scale model testing is a valuable practical application of the use
of these dimensionless relationships. By means of experiments on
a model, the performance of a projected design may be estimated.
The requirements are that the model must be geometrically similar
to the full scale design; also that Reynolds and Prandtl numbers
must be reproduced exactly. Then the flow patterns and fluid and
thermal boundary layers will be correctly modelled and, conse-
quently, the Nusselt number determined on the model will be the
correct value for the real thing.
Some of the more useful results will now be summarized. It
should be pointed out first, however, that the dimensional analysis
just considered was based on the assumption of constant fluid
properties and also that a single linear dimension was sufficient to
describe the system. Both of these assumptions are invalid in
certain circumstances. Viscosity is often the most temperature
dependent fluid property, and a varying viscosity will have a
considerable effect on the fluid boundary layer. If this is allowed for
in the dimensional analysis, an additional term, such as a viscosity
ratio to some power, will appear. In a result of the form of equation
(7.29), fluid property values at some mean temperature· are used.
Consequently, when these equations are used to predict heat transfer
coefficients, property values at the appropriate mean temperature
must be inserted. For pipe4low, an average or mean fluid temperature
is used. If the flow across a certain section of pipe were to be
thoroughly mixed, then an average fluid temperature would be
obtained. It will depend on the velocity profile as well as the tem-
perature profile. To evaluate an average heat transfer coefficient
over a length of pipe, then property values at a mean of the average
temperatures at the two ends must be inserted. When flow over a
flat plate is being considered, a mean film temperature may be used.
This is the average of the free stream fluid temperatures and the
wall temperature. In addition, an average of two mean film tem-
peratures may be used when considering an average convection
coefficient over a length of plate.
When an additional linear dimension is required, as in the case
of thermal boundary layer development in pipe flow, a length ratio

to some power will appear in the analysis.
7.3 Empirical Relationships for Forced Convection

Some of the more important relationships are now listed.
Laminar flow in tubes. An average Nusselt number between entry and

distance x from entry is given by
_= r
Nud 1·86 LRed Pr :;lJ (/J.)0-14
(~t
IJ.w (ref. 12) (7.30)
All physical properties are evaluated at the arithmetic mean bulk

temperature between entry and x, with the exception of llw which is at
the wall temperature, and the equation is valid for heating and cooling
in the range of Re Pr (djx) > 10. The group Re Pr (d/x) is the Graetz
number, Gz, and is significant in describing laminar flow heat transfer.
The Nusselt number, as both an average and local value, depends heavily
on the ratio (d/x) as well as on the Reynolds and Prandtl numbers. A
more recent equation is:
Nud = 1·4(Gz)t(IJ.:r (ref. 13) (7 .31)
where n = 0·05 for heating and 1/3 for cooling.

An equation for the local value of the Nusselt number is:
Nud = 3·66 + 0 "066 (Gz) 2 (ref. 14) (7 .32)

1 + 0·4 (Gz)3
As the entry length x in the Graetz number increases, the value of Nud
approaches the constant value of 3·66. This compares with the value of
4·36 (equation 6.27), deduced from an assumed parabolic velocity
distribution in fully developed flow.
In recent years, fresh analyses of laminar flow have been undertaken
by means of numerical solutions of the momentum, energy and conti-
nuity equations. The work of Collins 15 has shown a close agreement
with the results of Test 13 , equation 7.31, and a considerable divergence
from Sieder and Tate, equation 7 .30.
EXAMPLE 7.2
A heat exchanger thermal wheel consists of a porous disc of available
disc area A, having a porosity P. The mass flow to pass through the disc
is m, and the flow area A x P consists of N holes of diameter d. The
flow of gas is laminar through the holes and it is required to find the
group hAh/(mCp) for the wheel,for a range of hole sizes, where his the
laminar heat transfer coefficient, and Ah is the area for heat transfer.
(This group is the NTU, Number of Transfer Units; see also Chapter 12
and Example 12.3.)
Solution. For the wheel, P A = N 1T d 2 /4, hence N = 4 P A/(1T d 2 )
andm =pPA V
V =m/(pPA)
m
andRe= d/(P A ll)
The heat transfer area Ah = 1T d L N, where L is the hole length, equal

to the wheel thickness. Hence
Ah = 4PA L/d
In the following program listing, the heat transfer coefficient is calcula·
1
ted from Nud = 1·86(Re Pr d/LP, and WD = the wheel thickness L,
W = the mass flow rate, CP = the fluid specific heat, DEN= the fluid
density p, PR = the fluid Prandtl number, TK =the fluid thermal con-
ductivity, VIS = the fluid viscosity, and D is the hole diameter, d.
Basic Program Listing and Results
10C Calculation of NTU for a Thermal Wheel

20 wo=0.1
50 A=O .• 25
60 P=0.3
70 W=3.0
80 CP=1 .02
90 DEN=O. 75
100 pR=0.63
110 TK=O.D37
120 VIS=O.OQ0025
130 0=0.001
140 RE=W•D/CP•A•VIS)
150 H=1.86•TK•CCRE•PR•D/wo>••0.333)/D
160 AREA=4.0•P•A•WD/D
170 TU=H*AREA/CW•CP)
180 GZ=RE•P't•D/WD
190 MOL=P•A•4.0/(3.1417•D•D)
200 LPRINT " Matrix Hole Diameter=";
210 LPRINT USING "#I.#I##";D;
220 LPRINT " m, ~umber of Holes =";
230 LPRINT USING "#I##II":MOL
240 LPRI~T" Reynol1s Number ".,
250 LPRINT USING "#####I.#";RE;
260 LPRI~T " Graetz Nu~ber =";
270 LPRINT USING "II##U.U";Gz
280 LPRINT" Heat Transfer Coefficient :";
290 LPRINT USING "I###I.#";H;LPRINT " W/m2 K"
300 LPRINT " Heat Transfer Area =";
310 LP~INT USING "#####.#";AREA;
323 LORINT " mz, NTU =";
330 LPRINT USING "#II.#";TU
340 LPRINT
350 D=D+n.rnn5
360 IF(D<0.3035)GO TO 14n
370 STOP
Matrix Hole Di~meter = 0.00100 w, Number of Holes = 95439

Retnolds Number = 1600.0 Graetz Number 10.9
lleat Transfer Coefficient= 152.4 ·~lf.12
Heat Transfer Ared = 30.0 m2, N T U =
f'latrix Hole l>iarneter = 0.00150 ,,,,!lumber of lloles 42439

ReynolJs Number = 2400.0 Graetz Number 24. 5
Heat Transfer Coefficient = 133.1 W/;J2
lleat Transfer Are" = 20.0 m2, r1 T U =
Matri~ Hole Diameter = 0.00200 m, Number of Holes 23372

ReynolJs Number = 3200.0 Graetz Number 43.5
Heat Transfer Coefficient = 120.? :.1/.12 K,
Heat Transfer Area = 15.0 m2, N T U = 5?2. 6
Matrix Hole l>iameter = 0.00250 m, Number of Holes = 15278

Reynolds Number = 4000.0 Graetz Number 68.0
Heat Transfer Coefficient = 112.2 ~/M2 K,
Heat Transfer Area= 12.0 m2, N T U = 440.0
Matrix Hole l>iameter = 0.00300 m, Number of lloles = 10609

Reynolds Number =4800.0 Graetz Number 97.9
Heat Transfer Coefficient = 105.6 W/M2 K,
Heat Transfer Are~ = 10.0 m2, N T U 345.0=
Turbulent flow in tubes. For fluids with a Prandtl number near unity,
and only moderate temperature differences between the fluid and the
wall, (5°C for liquids, 55°C for gases), McAdams 16 recommends:
(7 .33)
where n = 0·4 for heating, and 0·3 for cooling, and Re11 > 10,000.
This is for fully developed flow, i.e., (xjd) > 60, and all fluid pro-
perties are at the arithmetic mean bulk temperature.
For both larger temperature differences and a wider range of
Prandtl number:
(7.34)
In this equation (}7 < Pr < 16,700, and all other details are as
before, with J.lw taken at the wall temperature.
In many situations where tube lengths are relatively short, fully
developed flow is not achieved, and the following relationship may be
used:
Nud = 0·036 Re/" 8 p,t (d/L)o-oss (7 .35)
for 10 <L/d < 400

This relationship is originally attributed to Nusselt.
The group 0·036 (d/L) 0 "055 varies between 0·0317 and 0·0259 for
the range of L/d given; these figures may be compared with 0·023
and 0·027 as the numerical constants in equations (7 .33) and (7 .34).
Turbulent flow along flat plates. For this type of flow, Chapman
recommends:
Nux = 0·036Prt(Re~·S - 18,700) (ref. 17) (7 .36)
This is based on a consideration of laminar flow (for which
Nux = 0·664(Rex)t(Pr)t) and turbulent flow after transition at
Rex = 400,000, for 10 > Pr > 0·6. Fluid properties are evaluated
at the mean film temperature.
Heat transfer to liquid metals. Liquid metals are characterised by
their very low Prandtl numbers. Experimental correlations are
for uniform wall heat flux and constant wall temperature in turbulent
flow in smooth tubes. Thus:
uniform heat flux, Nud = 0·625(RedPr)0 "4 (ref. 18) (7 .37)
constant wall temperature,
Nud = 5·0 + 0·025(Re11 Pr)0 "8 (ref. 19) (7 .38)
All properties are evaluated at the bulk temperature of the fluid,
with (x/d) > 60, and 102 < (Re4 Pr) < 104 •
The temperature profile becomes very peaked compared with the

velocity profile, when the Prandtl number is very small, as shown
in Fig. 7.3.
velocity
profile
Fig. 7.3. Normalized temperature and velocity profiles for flow in a tube at
very low values of Pr.
EXAMPLE 7.3
Freon at a mean bulk temperature of -1oac flows at 0·2 m/s in a
20 mm bore pipe. The freon is heated by a constant wall heat flux
from the pipe, and the surface temperature is 15°C above the mean
fluid temperature. Calculate the length of pipe for a heat transfer
rate of 1·5 kW. Use fluid properties from table AS.
Solution. At -10°C, v = 0·0221 X w-

5 ,k = 72·7 X 10- 6 kw/(mK),
Pr = 4·0, J.t = 31·6 X 10- 5 Pas. At +5°C, J.t = 28·8 X 10- 5 Pas.
A comparison of results using equations (7 .33) and (7 .34) may be
obtained. Re = 20 x 0·2 x 105 /1000 x 0·0221 = 18,100. Therefore
Re 0 "8 = 2547. Pr = 4·0, hence Pr 0 "4 = 1·74 and p,t = 1·588.
(J.t/J.tw} 0 " 14 = (31·6/28·8} 0 "14 = 1·013
From equation (7 .33),Nud = 0·023 x 2547 x 1·74 = 102·0
From equation (7 .34),Nud = 0·027 x 2547 x 1·588 x 1·013 = 110·7
Using the second result, which is 8·5 per cent larger than the first,
h = 110·7 X 72·7 X w- 6 X 103 ;20 = 0·402 kW;(m 2 K)
The pipe length required is calculated from Q = ndLh(tw - t1)
where tw and t 1 are the wall and fluid temperatures, hence
L = 1·5j(1t X 20 X 10- 3 X 0·402 X 15) = 3·96 m
PROBLEMS
I. The expression, Stanton number = ! x friction factor, may be derived
from the simple Reynolds analogy. Briefly explain this analogy, discussing
any assumptions made and stating limitations to the application of the above
expression.
Air at a mean pressure of 6·9 bar and a mean temperature of 65·5oc flows
through a pipe of 0·051 m internal diameter at a mean velocity of 6·1 mjs.
The inner surface ofthe pipe is maintained at a constant temperature and the
pressure drop along a 9·14 m length of pipe is 0·545 bar. Determine: (a) the
Stanton number, and (b) the mean surface heat transfer coefficient. (Ans.
0·288, 12·56kW/(m2 K.) (University of London).
2. Deduce the Taylor-Prandtl equation
~ ~[I + a(~r -
= 1)]
which gives the heat transfer per unit area and time, H, in terms of the drag
force per unit area, F, and in which Pr denotes the Prandtl number CJJ./k;
the other symbols having their usual meaning. (a= vJv•. )
Use the Taylor-Prandtl equation to show mathematically the following
deductions, and explain them in simple terms:
(a) For gases the Taylor-Prandtl equation approximates closely to the
Reynolds equation. (Reynolds equation is !!_ = cO but for liquids the

F v
divergence is considerable.)
(b) For turbulent flow the Taylor-Prandtl equation reduces to the Reynolds
' stream 1·me ftow It
. b ut 10r
equatiOn kO.
H = -
. red uces to -F
JI.V
(c) lfthe value of the Prandtl number is unity, then the form of the Taylor-
Prandtl equation for streamline and turbulent conditions is identical.
(d) With liquids of very low thermal conductivity, the whole of the tem-
perature drop occurs in the boundary layer. (King's College, London).
3. Discuss the effects of boundary layers on heat transfer by convection, and

show that, if Reynolds analogy between friction and heat transfer applies,
h f
CpiJU 2
It was found during a test in which water flowed with a velocity of2·44 mjs
through a tube 2·54 em inside diameter and 6·08 m long, that the head lost
due to friction was I· 22 m of water. Estimate the surface heat transfer coeffi-
cient, based on the above analogy. For water p = 998 kgjm 3 , cP = 4·187
kJ/(kg K). (Ans. 21·4 kW/(m 2 K.) (Queen Mary College, London).
4. Air at a temperature of ll5·6°C enters a smooth pipe 7·62 em diameter.

the wall of which can be maintained at a constant temperature of l5·6oC.
The rate of flow of air is 0·0226 m 3 jsec. Estimate the length of pipe necessary
if the air is to be cooled to 65·5°C, using the following assumptions: Prandtl
1;1umber for air = 0·74; f = 0·007; velocity at boundary of sublayer is half
the mean velocity in the pipe. (Ans. 12·55 m.) (University College, London).
5. A transformer dissipates 25 kW to cooling oil entering at 40° and leaving

at 60°C. The oil is subsequently divided equally into 16 tubes in a heat
exchanger. Calculate the convection coefficient of the oil in the heat exchanger
tube, given: Internal tube diameter, 10 mm; oil properties: p = 870 kgjm 3 ,
cP = 2·05 kJjkg K), J.t = Q-073 Pas, Pr = 1050, k = 140 x 10- 6 kW/(m K);
for laminar flow: Nu 4 = 0·125 (Re4 Pr)t

for turbulent flow: Nu 4 = 0·023 (Re4 ) 0 . 8 (Pr)t
(Ans. Flow is laminar, 0·0722 kW/(m 2 K).) (The City University).
6. It is proposed to test the cooling system of an oil-immersed transformer

by means of a model. The transformer dissipates 100 kW, the model is
} 0 linear size, with 4 ~ 0 surface area. Assuming the basic mechanism of heat
transfer is forced convection in a cylindrical duct, (0·5 em diameter on the
model), determine the energy dissipation rate and the velocity in the model.
Mean temperature differences are the same in the transformer and model.
Ethylene glycol is used in the model. Use Nu4 = 0·023 Re~·s Pr 0 · 4 ; Re =
2200; for oil: k = 131·5 x 10- 6 kW/(m K), Pr = 80; for ethylene glycol:
k = 256 x 10- 6 kW/(m K), Pr = 80, v = 0·868 x 10- 5 m 2 js. (Ans. 9·75 kW,
3·82 mjs.)
7. (a) Describe the following dimensionless quantities used in the study of

heat transfer: Nu, Re, Pr, Gr, St, giving their physical interpretations in a
form of simple ratios.
(b) Describe, using suitable formulae, what is known as Reynolds analogy.
Show that under certain conditions,

St = 2rjpv 2
(See also chapter 8.) (University of Oxford).
8. Air at mean conditions of 510°C, 1·013 bar, and 6·09 mjs flows through
a thin 2·54 em diameter copper tube in surroundings at 272°C.
(a) At what rate, per metre length, will the tube lose heat?
(b) What would be the reduction of heat loss if 2·54 em of lagging with
k = 173 x 10- 6 kW/(m K)were applied to thetube?Take N 4 = 0·023(R4) 0 . 8
P 0 . 33 with all the properties taken at the bulk air temperature. Assume the
surface heat transfer coefficient from the outside of the unlagged and lagged
tube to be 17-Q and 11-3 x 10- 3 kW/(m 2 K) respectively. (Ans. 0·174 kW/m,
32 per cent.) (University of Bristol).
9. A 100 MW alternator is hydrogen cooled. The alternator efficiency is

98·5 per cent and hydrogen enters at 27° and leaves at 88°C. It then flows
in a duct at a Reynolds number of 100,000. Calculate the mass flow rate of
coolant and the duct area. For hydrogen: cP = 14·24 kJ/(kg K1 J.l = 0·087 x
10- 4 Pas. (Ans. 1·73 kgjs, 5·0 m 2 .)
10. Explain and derive the simple Reynolds analogy between heat transfer
and fluid friction. Outline the Prandtl-Taylor modification to the simple
theory.
2·49 kg/s of air is to be heated from 15 to 75oc using a shell and tube heat
exchanger. The tubes which are 3·17 em in diameter have condensing steam
on the outside and the tube wall temperature may be taken as 100°C. Specify
the number of tubes in parallel and their length if the maximum allowable
pressure drop is 12·7 em of water.
Assume that/= 0·079Re-t and that the air has the following properties:
density 1-123 kgjm 3, kinematic viscosity 1·725 x 10- 5 m 2 js. (To solve this
problem, see also chapter 13.) (Ans. 94 tubes, 3·75 m long.) (University of
Leeds).
REFERENCES
1. Patankar, S. V. and Spalding, D. B. Heat and Mass Transfer in Boundary

Layers, 2nd ed., International Textbook Company, Scranton, Pa.
(1970).
2. Bayley, F. J., Owen, J. M. and Turner, A. B. Heat Transfer, Nelson
(1972).
3. Reynolds, 0. Proc. Manchester Lit. Phil. Soc., Vol. 14, 7 (1874).
4. Reynolds, 0. Trans. Roy. Soc. Lond., Vol. 174A, 935 (1883).
5. Schlichting, H. Boundary Layer Theory, McGraw-Hill Book Company,
Inc., New York (1955).
6. Knudsen, J. G. and Katz, D. L. Fluid Dynamics and Heat Transfer,
McGraw-Hill Book Company, Inc., New York (1958).
7. Prandtl, L. Z. Physik., Vol. 11, 1072 (1910).
8. Taylor, G. I. British Adv. Comm. Aero., Reports and Mem. ,Vol. 274,423
(1916).
9. Colburn, A. P. Trans. AIChE, Vol. 29,174 (1933).
10. Buckingham, E. Phys. Rev., Vol. 4, 345 (1914).
11. Langhaar, H. L. Dimensional Analysis and Theory ofModels, John Wiley,
New York (1951).
12. Sieder, E. N. and Tate, G. E. Ind. Eng. Chem., Vol. 28, 1429 (1936).
13. Test, F. L. 'Laminar flow heat transfer and fluid flow for liquids
with temperature dependent viscosity', J. Heat Trans., Vol. 90,
385-393 (1968).
14. Hausen, H. VDIZ, No.4, 91 (1943).
15. Collins, M. W. 'Finite difference analysis for developing laminar
flow in circular tubes applied to forced and combined convection',
Int. J. Num. Meth. Eng., Vol. 15,381-404 (1980).
16. McAdams, W. H. Trans. AIChE, Vol. 36, 1 (1940).
17. Chapman, A. J. Heat Transfer, 3rd ed., The Macmillan Company,
New York (1974).
18. Lubarsky, B. and Kaufman, S. J. NACA Tech. Note 3336 (1955).
19. Seban, R. A. and Shimazaki, T. T. Trans. ASME, Vol. 73, 803
(1951).
8
Natural convection
Forced convection heat transfer has now been considered in some

detail. The energy exchange between a body and an essentially
stagnant fluid surrounding it is another important example of
convection. Fluid motion is due entirely to buoyancy forces arising
from density variations in the fluid. There is often slight motion
present from other causes; any effects of these random disturbances
must be assumed negligible in an analysis of the process. Natural
convection is generally to be found when any object is dissipating
energy to its surroundings. This may be intentional, in the essential
cooling of some machine or electrical device, or in the heating of
a house or room by a convective heating system. It may also be
unintentional, in the loss of energy from a steam pipe, or in the
dissipation of warmth to the cold air outside the window of a room.
Fluid motion generated by natural convection may be laminar
or turbulent. The boundary layer produced now has zero fluid
velocity at both the solid surface and at the outer limit, and the profile
is of the form shown in Fig. 8.1. In laminar flow natural convection
bulk fluid temp.

_L _ _ _ _ _ _ _
I
direction of
induced motion
Fig. 8.1. Natural convection boundary layer on a vertical flat plate.
124
NATURAL CONVECTION 125
from a vertical plate, it is possible to obtain a solution ofthe bound-
ary layer equations of motion and energy, if a body force term is
included. This approach is limited in general application and, conse-
quently, the method of dimensional analysis will be used.
8.1 The Body Force

Before undertaking a dimensional analysis of natural convection it
is necessary to consider the nature of the body force. If Ps is the
density of cold undisturbed fluid, p is the density of warmer fluid,
and 0 is the temperature difference between the two fluid regions,
then the buoyancy force on unit volume is
(p. - p)g
and Ps is related to p by
Ps = p(l +PO)
where P is the coefficient of cubical expansion of the fluid. Thus
the buoyancy force is
[p(l + po) - P]g = pgpO (8.1)
The independent variables on which the natural convection

coefficient h depends may now be listed. A buoyancy force term would
appear in the differential equation of momentum, hence p, g, and 0
appear in addition to the fluid properties p, p., cP and k, and the
linear dimension characteristic of the system, l. This is the dimension
which would be used in the Reynolds number for a forced flow in the
same direction as the natural convective flow. p and g are usually
combined as a single variable pg since variation of g is unlikely.
8.2 Dimensional Analysis of Natural Convection

The procedure outlined in chapter 7 will now be followed to obtain
the dimensionless groups relevant to natural convection. There are
eight physical variables and five dimensional quantities, so that
three n terms are expected. H and 0 may not be combined to form
a single dimensional quantity, since temperature difference is now
an important physical variable.
Five physical variables selected to be common to all n terms are
p, p., k, 0, and l. These fulfil the necessary conditions. h, cP, and pg
will each appear in a separate 1t term. The 1t terms are :
1t 1 = p"' Jl.b,kel(Jdl/e'h
1t2 = p"'Jl.bzke•(Jd•Le•cp
1t 3 = p"3Jl.b3k&3(Jd3/e3pg
After wntmg the necessary equations to obtain the exponents

a to e in each 1t term, it is found that
The 1t 3 term is the Grashof number and the dimensionless relation-

ship may be expressed as
c/J (Nu, Pr, Gr) = 0
or.
Nu = c/J(Gr, Pr) (8.2)
The Grashof number is the ratio of buoyancy force to shear
force, where the buoyancy force in natural convection replaces the
momentum force in forced convection. pgpO is the buoyancy force
per unit volume, therefore pgpfJ x I would be for unit area. The
ratio of buoyancy to shear force per unit area is pgpfJI/(JJ.vfl). But
velocity is a dependent variable proportional to (Jl./pi), ·hence the
ratio of buoyancy to shear force becomes pgp 2 fJI 3 fJJ. 2 •
Many experiments have been performed to establish the func-
tional relationships for different geometric configurations convect-
ing to various fluids. Generally, it is found that equation (8.2) is of
the form
Nu = a(GrPrt (8.2a)
where a and b are constants. The product GrPr is the Rayleigh
number Ra. However, results are generally quoted in terms of(GrPr)
since it is often necessary to vary Gr at some fixed Pr. Laminar and
turbulent flow regimes have been observed in natural convection,
and transition generally occurs in the range 107 < GrPr < 109
depending on the geometry.
8.3 Fonnulae for the Prediction of Natural Convection

Some of the more important results obtained will now be presented.
These may be used for design calculations provided the system under
consideration is geometrically similar and that the value of (GrPr)

falls within the limits specified. Generally, there are no restrictions
on the use of any specific fluid. Example 8.1 shows how the formulae
are used. Figure 8.2 shows the principal geometries with external
flow in the direction of the arrows. For more extensive reviews of
available information, see for example, refs I and 2. In addition,
ref. 2 may be consulted for details of natural convection in enclosed
spaces and natural convection effects in forced flow when the
Reynolds number is very small, a situation known as combined or
mixed convection.
JI 1 ,lI
Horizontal cylinders
J
Vertical surfaces
Horizontal flat surfaces

Fig. 8.2. Priltcipal geom~tries;, IUltura/ confH!ction systems slwwiltg direction
of COIIfH!CPfH! flow.
8.3.1 Horizontal Cylinders

Detailed measurements indicate that the convection coefficient varies
with angular position round a horizontal cylinder, but for design pur·
poses values given by the following equations 3 are constant over the
whole surface area, for cylinders of diameter d.
Nu 4 = 0·525(Gr 4Pr) 0 ' 25 (8.3)
when 104 < Gr4 Pr < 109 (laminar flow) and
(8.4)
when 109 < GrdPr < 10 12 (turbulent flow). Below GrdPr = 104 ,
it is not possible to express results by a simple relationship, and
ultimately the Nusselt number decreases to a value of 0·4. At these
low values of GrdPr the boundary layer thickness becomes appre-
ciable in comparison with the cylinder diameter, and in the case
of very fine wires heat transfer occurs in the limit by conduction
through a stagnant film. Fluid properties are evaluated at the average
of the surface and bulk fluid temperatures, which is the mean film
temperature. If the surface temperature is unknown, a trial and
error solution is necessary to find h from a known heat transfer rate.
8.3.2 Vertical Surfaces

Both vertical flat surfaces and vertical cylinders may be considered
using the same correlations of experimental data. The characteristic
linear dimension is the length, or height, of the surface, l. This follows
from the fact that the boundary layer results from vertical motion of
fluid and the length of boundary layer is important rather than its
width. Again average values of Nu 1 are given, even though in the case of
Gr1Pr > 109 the boundary layer is initially laminar and then turbulent.
With physical constants at the mean film temperature the numerical
constants as recommended by McAdams3 are
Nu 1 = 0·59(Gr1 Pr) 0 " 2 ~ (8.5)
when 104 < Gr1Pr < 109 (laminar flow) and
Nu 1 = 0·129(Gr,Pr) 0 ·3 3 (8.6)
when 109 < Gr1Pr < 10 12 (turbulent flow).
For vertical flat plates, alternative relationships are:

0·8 (Gr1 Pr}0"25
Nu
l
= (I + (1 + 1/Pr0.5)2)0.25 (8.7)
(ref. 4), for Laminar flow,

and
0-0246 (Gr1Pr)0 "4
Nu 1 = - - - - - - - - - (8.8)
(Pr-1"/(1 + 0·444Pri-)}f
(ref. 5), for turbulent flow
For moderately inclined plates, rp to the vertical, Gr may be multiplied

by cos rp, to give the correct Grashof number based on the vertical com-
ponent of the buoyancy force.
8.3.3 Horizontal Flat Surfaces

Fluid flow is most restricted in the case of horizontal surfaces, and the
size of the surface has some bearing on the experimental data. The heat
transfer coefficient is likely to be more variable over a smaller flat surface
than a large one, when flow effects at the edges become less significant.
Further, there will be a difference depending on whether the horizontal
surface is above or below the fluid. Similar, though reversed, processes
take place for hot surfaces facing upwards (i.e., cold fluid above a hot
surface), and cold surfaces facing downwards (i.e., hot fluid below
a cold surface). In either case, the fluid is relatively free to move due
to buoyancy effects and be replaced by fresh fluid entering at the
edges. The following relationships are generally recommended for
square or rectangular horizontal surfaces up to a mean length of
side (I) of 2 ft:
Nu 1 = 0·54(Gr1Pr)0 "25 (8.9)
when 105 < Gr1Pr < 108 (laminar flow).and
Nu 1 = 0·14(Gr1Pr)0 "33 (8.10)
when Gr1Pr > 108 (turbulent flow). Thus turbulent flow is possible
in this geometrical arrangement.
The converse arrangement is the hot surface above a cold fluid,
or hot surface facing downwards, and a hot fluid above a cold
surface, or cold surface facing upwards. In either case, it is obvious
that convective motion is severely restricted since the surface itself
prevents vertical movement. Laminar motion only has been
observed, and the recommendation is
Nu 1 = 0·25(Gr1Pr) 0 "25 (8.11)
when Gr1Pr > 10 5 . Fluid properties are again taken at the mean
film temperature.
8.3.4 Additional Geometries
For two vertical parallel plates at the same temperature, convecting to
air:
Nu 1 = 0·04 (Grz Pr)(djl) 3 (ref. 6) (8.12)

where l is the vertical height of the plates and d is the space between
them. This geometry occurs, for example, in the space between the two
panels of a double hot water radiator. When the parallel plates have
different temperatures, as in two panes of glass in a double-glazed win-
dow, for example:
for laminar flow: Nud = 0·18 (Grd Pr)t (lfdr +(Prr t (8.13)
for 2 x 104 < Gr < 2 x 105
-- J. J.
= 0·065(Grd Pr) 3 (//d)-
I
for turbulent flow: Nud 9 (Prr T (8.14)
for 2 x 105 < Gr < 10 7 (ref. 7)
The vertical height is again l, the width d, and (J in Gr is the tempera-
ture difference between the plates. The relationship holds for 1/d > 3.
For a hollow vertical cylinder with an air column
Nu 1 = 0·01 (Gr 1Pr) (d/{) 3 (ref. 6) (8.15)
where lis the height of the cylinder and dis the diameter. In the case of
small rectangular solid objects there are generally insufficient vertical
and horizontal surfaces to be treated separately, and the natural cooling
of such objects in closed or open environments is a fairly common
occurrence. The interaction of separate boundary layer growth means
that an overall characteristic length is less than either of the actual
principal dimensions, so that
where the characteristic length is x, and /h and lv are the principal hori-
zontal and vertical dimensions, respectively. Then:
-
Nux = 0.6 ( Grx Pr)0·25 (ref. 8) (8.16)
for 1if < GrPr < 109
8.3.5 Approximate Formulae for use with Air
A great deal of natural convection work involves air as the fluid medium
and the fluid properties of air do not vary greatly over limited tempera-
ture ranges. Thus it is possible to derive simple formulae from equations
(8.3) to (8.6) and (8.9) to (8.11) in which the physical properties in the
Nusselt, Grashof, and Prandtl numbers are grouped together and
assumed constant. From equation (8.2a)
h = constan{kl-b(pg:
2CprJ!'f/3b-t
= constant x 0"1 3 b- 1 (8.17)
It will have been noted that b = 0·25 in laminar flow and 0·33 in
turbulent flow, so that the index of I is -0·25 in laminar flow and 0
in turbulent flow. The simplified expressions become
h = C (l(})0·25 in laminar flow (8.18)
and
h = C(J0 ' 33 in turbulent flow (8.19)
where the value of C, the constant, depends on the configuration

and flow, and I is the characteristic dimension.
The resulting expressions for horizontal cylinders. vertical and
horizontal surfaces. based on the relationships given by McAdams, 1
are:
Horizontal cylinders { h~ ~00131 d

(oy·ls laminar flow
d =diameter
(oy·ls
h = 0·00124(}0 "33 turbulent flow
Vertical surfaces { h ~ 000141 I laminar flow
I= height
h = 0·00131(}0 "33 turbulent flow
Horizontal surfaces
(oy·ls
I = length of side
Hot, facing upwards } { h ~ 000131 I laminar flow
Cold, facing downwards h = ()-001528o-3 3 turbulent flow

Hot, facing downwards
Cold, facing upwards } h = o-ooo5s
(oy·ls
I laminar flow
The numerical constants in these equations give h in kW/(m 2 K)

with(} in oc and linear dimensions in m. To determine whether flow is
laminar or turbulent it is necessary to find the approximate value of
(GrPr) and to see to which flow regime the value corresponds, as given
in earlier Sections. For this purpose take f3gp 2 c/JJ.k = 6·4 x 107 •
This is multiplied by (linear dimension) 3 x (} m K to obtain the
dimensionless (GrPr).
ExAMPLE 8.1
An oil filled electric heating panel has the form of a thin vertical
rectangle, 2m long by 0·8 m high. It convects freely from both
surfaces. The surface temperature is 85°C and the surrounding air
temperature 20°C. Calculate the rate of heat transfer by natural
convection, and compare the result with that obtained from the
simplified formula for air.
Fluid properties at the average of surface and bulk air tempera-
tures, 53°C, are f3 = 1/326, Pr = 0·702, p = 1·087kg/m 3 , Jl. =
1·965 x 10- 5 Pas and k = 28·1 x 10- 6 kW/(m K).
Solution. The characteristic linear dimension is the panel height,
Q-8m. The product (Gr1Pr) must first be found.(}= 85- 20 = 65.
Gr1Pr = [(1/326) x 9·81 x 65 x 1·087 2 x 0·8 3 /(1·965 x 10- 5 ) 2 ]
X ()-702 = 2·15 X 109
Hence the flow is turbulent, and Nu 1 = 0·129(Gr1Pr)0 ' 33 may be used.
(Gr1Pr) 0 ' 33 = 1·29 x 103 Nu 1 = 0·166 x 103
and
0·166 X 103 X 28·1
1i = 0.8 x 106 = 5·85 x 10- 3 kW/(m 2 K)
Using the simplified relationship, 1i = 0·00131 X (} 0 ' 33 = 0·00131 X

65°'33 = 5·26 x 10- 3 kW/(m 2 K)
This represents 10·1 per cent error on the value from equation (8.6).
The convection from both sides of the panel is
Q = 2 X 2 X 0·8 X 65 X 5·85 X 10- 3 =1·215kW
using the more accurate value of 1i
PROBLEMS
1. Describe briefly how experimental data on heat transfer by convection
obtained from small scale experiments may be applied to full-scale industrial
plant, and specify the conditions which must be satisfied for this to be possible.
Define the Nusselt, Prandtl, and Grashof numbers and show that they are
dimensionless. Calculate the rate of heat transfer by natural convection from
the outside surface of a horizontal pipe of 15·2 em outside diameter and
6·1 m long. The surface temperature of the pipe is 82°C and that of the
surrounding air 15·6°C.
The following relations are applicable to heat transfer by natural convection
to air from a horizontal cylinder; for laminar flow, when 104 < (GrPr) < 109
Nu = 0·56(GrPr)*
and for turbulent flow, when 109 < (GrPr) < 10 12
Nu = 0·12(GrPr)i
The properties of air given below, corresponding to the 'mean film tempera-
ture', i.e., 49°C, may be used.
Kinematic viscosity v = 1·8 X w-s m 2 fs
Thermal conductivity k = 0·0284 X 10- 3 kW/(mK)
Coefficient of cubical expansion fJ = 3 ~2 K- 1
Prandtl number Pr = 0·701. (Ans. 1·275 kW.) (Queen Mary College,

London).
2. The transfer of heat by natural convection from vertical planes may be

calculated by using the following formula which is valid for all P, for R less
than 109 and for N greater than 5.
N 4 /R = 2P/(5 + }Qpt + lOP)

where N = Q" Lfk8,
R = (!:J.p)gL 3 pcJJJ.k,
p = Jl.Cp/k,
and where
L = height of plane, k =thermal conductivity
(} = temperature difference, g = gravitational acceleration
cP = isobaric specific heat-capacity
Jl. = viscosity, p = density,
(!:J.p) = difference between density of fluid near plane and density of fluid
far away,
Q" = surface density of rate of heat transfer.
Some busbars are in the form of strips which run horizontally and are ten
times as high as they are thick. They are made of copper for which the
resistivity is 2 X 10- 8 ohm m.
They are to be designed for operation at 87·8oc in an atmosphere which is
at 32·2oC and at 1·013 bar. Calculate the height of busbar for use with a
current of 10,000 A. Assume that both radiation and that part of the convec-
tion which is from the top and bottom edges of the bars are negligible.
(Ans. 0·35 m.) (Queen Mary College, London).
3. By dimensional analysis show that for natural convection of a perfect gas
~ = t{( 1}). (T. ;o To)· (JlZp)} = f(GrPr)

where v is the kinematic viscosity, T. is the surface temperature and T0 is
the temperature in the bulk of the fluid. Give a brief statement of the assump-
tions made.
A metal plate, 0·609 min height, forms the vertical wall of an oven and is at
a temperature of 171 oc. Within the oven is air at a temperature of 93-4°C
and atmospheric pressure. Assuming that natural convection conditions hold
near the plate, and that for this case Nu = 0·548(GrPr) 0 ' 25 , find the mean
heat transfer coefficient, and the heat taken up by the air per second, per
metre width. For air at 132·2°C, k = 32·2 X 10- 6 kW/(m K), Jl = 0·232 X
10- 4 Pas. (Ans. 4·11 X 10- 3 kW/(m 2 K), 0·195 kW/m.)(Queen Mary College,
London).
4. A factory is heated by a bank of eight 100 mm diameter steam pipes

placed under grilles in the floor. Steam at 139°C passes through the
pipes and the mean air temperature in the factory is l5°C. Assuming
each pipe convects freely calculate the length of the bank of pipes
necessary to give 1OkW of heating. (Ans. 4.19 m.)
5. A tubular heater mounted horizontally is 25 mm diameter and dissipates

0·075 kW/m length. The surroundinf air temperature is 30°C. Assuming
h = Q-00127 (8/diameter) 0 ' 25 kW/(m K) with the diameter in metres,
calculate the surfa~ temperature and the value of the natural convection
coefficient. (Ans. 126°C, ()-01 kW/(m 2 K).)
6. The surface of an electric immersion heater is equivalent to that of a
given length of 10 mm diameter tube. The surface temperature must
not exceed 150°C when the water temperature is 65°C and the output
is 3 kW. Calculate the length of equivalent tube. (Ans. 0-44 m.)
7. The vertical wall of a building is 3 m high and it receives solar radia-

tion at the rate of 300 W/m 2 . Calculate the surface temperature of the
wall if two-thirds of this is reradiated back to the environment, one-fifth
is conducted through the wall and the remainder is dissipated by natural
convection to the atmosphere at l5°C. (Ans. 43°C.)
REFERENCES
l. Hsu, S. T. Engineering Heat Transfer, D. Van Nostrand Company, Inc.,

Princeton (1963).
2. Holman, J. P. Heat Transfer, 3rd ed., McGraw-Hill Book Company,

New York (1972).
3. McAdams, W. H. Heat Transmission, 3rd ed., McGraw-Hill Book Com-
pany, Inc., New York (1954).
4. Ostrach, S. NACA TN 2635 ( 1952).
5. Eckert, E. and Jackson, T. NACA TN 2207 (1950).
6. Ede, A. J. 'Advances in Free Convection', in Advances in Heat
Transfer, Vol. 4, Academic Press, New York (1967).
7. Jacob, M. Heat Transfer, Vol. 1, Wiley, New York (1957).
8. King, W. J.Mech. Eng., Vol. 4, 347 (1932).
9
Separated flow convection
Separation is an important characteristic of the type of flow

encountered in many modern heat transfer devices. Design require-
ments of compactness have resulted in the rapid growth of the use
of complex geometrical heat transfer surfaces, which have developed
from the single tube and tube bank placed across the line of flow.
A single tube or cylinder placed in a cross-flow is completely sub-
merged in the fluid and it therefore forms an obstacle around which
the fluid must flow. A boundary layer exists on the cylindrical surface
with free stream velocity at its extreme and zero velocity at the wall.
However, the free stream velocity increases around the front of the
cylinder and at low approach velocities flow within the boundary
layer also accelerates. Behind the cylinder free stream and boundary
layer flow decelerates again in a more or less reverse pattern. At
higher approach velocities the increased velocity around the front
of the cylinder which is accompanied by a drop in static pressure
is not followed by a similar increase in velocity in the boundary layer,
due to the increased viscous stress at the higher velocity gradients.
Thus, in the boundary layer the fluid has lost velocity before it
starts to decelerate behind the cylinder and it is then opposed by a
'surplus' of static pressure which forces the boundary layer away
from the surface. Separation, or break-away, results in the formation
of turbulent eddies which are carried downstream behind the
cylinder. Separation occurs nearer the front of the cylinder as the
approach velocity increases, and occurs much more readily in flow
over blunt ended obstacles.
Local heat transfer coefficients have been measured around the
circumference of cylinders in cross-flow. 1 They have minimum
values at the point of separation and increase forwards towards the
point of stagnation, but they increase more towards the rear of the
cylinder. This may be attributed to the scrubbing action of the eddies
136
SEPARATED FLOW CONVECTION 137
formed in that vicinity. Average values of heat transfer coefficient

have also been extensively determined as these are required for
design purposes. Owing to the degree of turbulence produced in a
tube bank, convection coefficients are high and average values for
tubes several rows back are found to be higher than for those at
entry due to the action of eddies shed from the leading rows. 2
The pattern of events in the tube bank has led to the evolution
of the compact heat transfer surface which is in general a complex
of finned cross-flow passages. The use of fins as a means of increasing
heat transfer coefficients is discussed in chapter 12. Flow through
such a system is largely composed of turbulent eddies, and even at
low approach velocities a high degree of turbulence is to be found.
For this reason the usual transition between laminar and turbulent
flow at Reynolds numbers around 2500 does not exist, and turbulent
flow has been found to persist to Reynolds numbers as low as 800. 3
In any arrangement of this type in which high heat transfer rates
may be obtained in a small space, the advantages have to be balanced
against the effect of increased pressure loss on overall performance.
Pressure loss is due to the total drag of the shapes involved and due
to shear over the fins. It may be measured across the whole system
and related to a friction coefficient by an expression similar to the
equation for flow in pipes:
(9.1)
Such a form is useful since it has been found generally that / 0 can
be related to the Reynolds number of flow. In the determination of
/ 0 from tip, the values of L, d, and Vm have to be defined in relation
to the particular geometry. The symbol / 0 is used to indicate that
it represents essentially a drag loss rather than a loss due to viscous
shear.
9.1 Relationship between Heat Transfer and Pressure Loss in a

Complex Flow Sytem
In the experimental determination of the performan~ of complex
heat transfer surfaces, Schenck4 found that an 'experimental
analogy' exists between heat transfer and friction, even though the
net friction effects involved are essentially due to drag forces. Thus
Fig. 9.1 shows the Colburn J-factor plotted against f D> as defmed in
equation (9.n for a wide range of different surfaces including plain
fins on tubes, plain and dimpled tubes, tube and spiral fins, flattened
tubes with plain grooved and wavy fins, pin fins and interrupted
plate fins. This particular plot is valid for Reynolds numbers in
excess of 5000.
The use of this information is illustrated in the following example.
0·02
0·01 /
...
~0·005
.
::,... ""•
~
...:, ... 1" .2'!:
.
,_.,. :
0·002 q1GGER-FULL_
eX ·:11-PART
" -NONE
0·001 I I II II
0·005 0·01 0·02 0·05 0·1 0·2
fo
Re > 5,000
Fig. 9.1. Relations/rip between the /-factor and fo for flow in a compkx
system. By courtesy of H. Schenck, Jr., and The American Society of NaMI
Engineers, Inc.
ExAMPLE 9.1
A compact forced convection oil cooler has a front area of 0·15 m2

and a surface area of 40m2 /m 2 frontal area. Air at 28°C enters the
cooler at 30 m/s. The average temperature of the heat transfer sur-
face is 126°C. fD is found to be 0·1. Estimate the heat transfer
performance.
Solution. From Fig. 9.1 the J-factor is o-01. The J-factor is given
by equation (7.27): J = StPrt, for mean conditions where St =
'Ft./pv.cP. At a mean temperature of 71°C, p for air is 0·998 kgfm 3, cP =
1·009 kJ/(kg K), Pr = 0·697. Hence the heat transfer coefficient is
given by
'ft. X (0·697)i = 0·0 1
0·998 X 30 X 1·009
:. 'Ft.= 0·386kW/(m2 K)
The heat transfer rate is liAO, where A = (surface area/unit area) x

(frontal area),
.". hA = 0·386 X 40 X 0·15 X (126- 28)
= 227kW
9.2 Convection from a Single Cylinder in Cross Flow
Much experimental work has been done to determine the heat
transfer coefficient from a single cylinder in cross flow. Investiga-
tions have included both fine heated wires and large pipes. A recent
examination of available data is that of Douglas and ChurchilP
and the equation which represents their results is
(9.2)
This equation is only valid for Red > 500. Nusselt and Reynolds
numbers are based on the cylinder diameter d, velocity is the free
stream, or undisturbed fluid velocity, and fluid properties are
evaluated at the average film temperature. Hsu 6 has proposed that
for Red < 500 the following equation may be used:
Nud = 0·43 + 0·48(Red)t (9.3)
Both of these equations are valid only for the simpler gases with
similar Prandtl numbers, since the small Prandtl number effects
are accommodated in the numerical constants. Both equations are
valid in heating as well as cooling of the cylinder.
9.3 Convection in Flow across Tube Bundles

Many examples of heat transfer across tube bundles occur in
industry, e.g. in cross-flow heat exchangers, and on the shell side
of shell and tube heat exchangers, (see Chapter 13). It is therefore
necessary to be able to predict convection coefficients in such
situations.
Snyder 2 found that the local Nusselt number on tubes in cross
flow achieved a constant value after the third row of tubes, and a
useful correlation is that of Colburn,? for the average Nusselt
number for all tubes, for ten or more rows of tubes in a staggered
arrangement:
Nud = o-33( dGJlma.. y·6 (Pr)t (9.4)
LL \ J I ~ ,{ ........ ~ /'-
' '
I I \ _1_ _1
,
J I I
L \ I ! l '- '-./ '/
disc-and-doughnut baffles segmental baffles
Fig. 9.2. Types of sheU baffle; see also Fig. 13.2 which shows doughnut and
900 segmental baffles.
dis the tube diameter, Gmax = mass velocity = p x v where vis the
velocity through the smallest free-flow area between tubes, fluid
properties are evaluated at mean of wall and bulk fluid temperatures,
and 10 <Red < 40,000. For a more exact dependence on Reynolds
number, the numerical constant 0.33 may be replaced by 0·193 for
4000 < (dGmaxfJ.J.) < 40,000 and 0·0266 for 40,000 < (dGmaxfJ.J.) <
400,000. The corresponding indices of the Reynolds number are 0·618
and 0·805 (ref. 8). A much more detailed analysis for staggered and
in-line tube arrangements of different spacings was carried out by
Grimison. 9
On the shell side of shell and tube heat exchangers, two relation-
ships proposed by Donohue 10 may be used. The baffle arrangements
quoted are illustrated in Fig. 9.2.
For disc-and-doughnut baffles:
Nu4 =
dG
0·033d~' 6 ( ---:
)0·6 (Pr)t (:w)o·t4 (9.5)
For segmental baffles, the (0·033 d~' 6 ) in (9·5) is replaced by 0·25. Note
that G. = .J(GwG), where Gw = mass velocity through the baffle
window area, and G. = mass velocity based on flow area at the
diameter of the shell. Fluid properties are evaluated at the fluid
bulk temperature, with the exception of f.l.w which is at the tube wall
temperature. It is important to note that in using equation (9.5) all
terms are dimensionless groups except for (0·033 d~' 6 ). Here d. is an
equivalent diameter = 4(SrSv - nd 2/4)/(nd) where ST =tube trans-
verse spacing, Sv = tube vertical spacing, d = tube diameter, and
d. is in mm.
The above equations give only very basic correlations of cross-
flow convection, for further information the reader is referred to
Kays and London. 11
EXAMPLE 9.2
In a shell and tube heat exchanger, the tubes are 25·4 mm diameter
and are spaced at 5(}8 mm centres both horizontally and vertically.
Water flows at 24·6 kg/sin the shell, and the baffie window area is
0·0125 m 2 and the net shell area is 0·05 m 2 • The water bulk tempera-
ture is 60°C and the tube wall temperature is 90°C. Calculate the
shell side heat transfer coefficient.
Solution. Property values of water are taken from Table AS. Thus,
= p x v, and at 60°C Jl. = 47·0 x 10- 5 , at 90°C Jl. = 31·9 x
Jl.
10- 5 Pas. Pr = 3·02, k = 651 x 10- 6 kW/(m K).
Equation 9.5 will be used. First calculate d., the equivalent
diameter.
d. = 4(STSv - 1td2 /4)/nd
= 4(50·8 2 - 1t x 25·4 2/4)/n x 25·4 = 104 mm
0·033 d~· 6 = 0·033 X (104) 0 "6 = 0·533

Gw = mass velocity through baffie window = p x velocity. But,
p x velocity x area = 24·6 kg/s.
Gw = 24·6/0·0125 = 1970
Gc = mass velocity through the shell = 24·6/0·05 = 492

G. = .j(GWGC) = ../(1970 X 492) = 984
R = 984 X 2·54 X 105 = 5·31 X 104 (Re)0·6 = 684

e 100 X 47·0 '
(Pr)t = (3·02)t = 1·445

(-
Jl. )0·14 = (-47 )0·14 = 1·056
Jl.w 31·9
Nu 4 = 0·533 X 684 X 1·445 X 1·056 = 556·0
1i = 556 x 651 x 10- 6 = 14.3 kW/(m 2 K)

0·0254
PROBLEMS
1. A gas is blown across two geometrically similar tube banks. In case (a)
there are 10 tubes 15 mm diameter by 200 mm long, the gas velocity is 50 mjs,
the gas temperature is 18°C, the tube surface temperature is 80°C and the
heat transfer rate is 1·26 kW. In case (b) the ten tubes are 30 mm diameter
by 400 mm long, the velocity is 30 mjs, and gas and surface temperatures
are 15° and 70°C, and the heat transfer rate is 2·78 kW. With the following
gas properties, determine A and B in the relationship Nud = A(Red) 8 for
the tube banks.
(a) k = 30 X w- 6 kW/(mK~ p = 1·0 kg/m 3 and J.l = 2·05 X 10- 5 Pas;
and(b)k = 26 x 10- 6 kW/(mK~p = 1·18kgjm3 andJ.1 = 1·85 x 10- 5 Pas.
(Ans. A = 0·0219, B = 0·81.) (The City University).
2. Air at 1· 5 bar and 1oooc passes through a compact heat exchanger at

107 mjs. The pressure drop is 0·2 bar. Given that the values of Land dare
0·5 m and 10 mm respectively, calculate the drag loss factor f 0 , the J-factor,
and the heat transfer in the exchanger, assuming a flow area of 0·2 m2 and
a surface area of 15m2 per m2 flow area. Take cP = 1·012 kJ/(kg K),
Pr = 0·692. (Ans. f = 0·05, J = 0·0072, 630 k W, at lw = 250°C.)
3. Hydrogen passes through a staggered bank of 200 tubes, 1·8 m long, and
25·4 mm diameter. The mass velocity is 1·5 kgj(m 2 s). Calculate the rate of
heat transfer for a mean gas temperature of 373 K and a tube surface to gas
temperature difference of 50 K. Calculate also the heat transfer rate if air
at twice the mass velocity is substituted for hydrogen. (Ans. 499 kW, 70·2 kW.)
4. Carbon dioxide flows in the shell side of a shell and tube heat exchanger.
There are 36 tubes 15 mm diameter by 2m long. The shell area for flow is
0·025 m2 and the baffie window area is 0·0125 m 2 • The vertical and hori-
zontal spacing of the tubes is 22·5 mm between centres. The mass flow of
carbon dioxide is 0·6 kgfs at a mean temperature of 400 K. The mean tube
surface temperature is 300 K. Calculate the convective heat transfer coef-
ficient on the shell side of the tubes and the heat transfer rate. (Ans. 0·168
kW/(m 2 K), 56·8 kW.)
REFERENCES
1. Schmidt, E. and Wenner, K. Forschung, Gebiete lngenieurw., Vol. 12,65
(1933).
2. Snyder, N. W. Chem. Eng. Progr., Symposium Series, Vol. 49, No.5, 11
(1953).
3. Schenck, H. Jnr. Heat Transfer Engineering, Longmans, Green and Co.
Ltd. (1960).
4. Schenck, H. Jnr. J. Amer. Soc. Naval Eng., Vol. 69, 767 (1957).
5. Douglas, M. J. M. and Churchill, S. W. Chem. Eng. Propr., Symposium
Series, Vol. 52, No. 18, 23 (1956).
Princeton (1963).
7. Colburn, A. P. Trans. AIChE, Vol. 29, 174 (1933).
8. Holman, J. P. Heat Transfer, McGraw-Hill Book Company, Inc.,
New York (1981).
9. Grimison, E. D. Trans. ASME, Vol. 59, 583 (1937).
10. Donohue, D. A. Ind. Eng. Chem., Vol. 41, 2499 (1949).
11. Kays, W. M. and London, A. L. Compact Heat Exchangers, McGraw-
Hill Book Company, Inc., New York (1964).
10
Convection with phase change
Convection processes with phase change are of great importance,

particularly those involving boiling and condensing in the fluid
phase. Such processes occur in steam power plant and in chemical
engineering plant. Convection in the liquid to solid phase change
is also of importance, as for example in metallurgical processes,
but this cannot be considered here.
10.1 Description of Condensing Flow

Two types of condensation are recognized, in which the condensing
vapour forms either a continuous film of liquid on the solid surface,
or a large number of droplets. Film condensation is the more
common; drop formation occurs generally in an initial transient
stage of condensing flow, or if for any reason the surface is un-
wettable. A condensing vapour generally forms droplets around
nuclei of minute solid particles, and these droplets merge into a
continuous film as they grow in number and size. The film then
flows under the action of gravity so that the process may continue.
As condensation depends on conduction of heat away through
the solid surface, the growth of a liquid film will impede the con-
densation rate. Condensation is also impeded if a non-condensable
gas is mixed with the vapour, since the concentration of gas tends
to be greater at the surface as the vapour changes its phase, and
this acts as a thermally insulating layer. It is thus desirable to
prevent the film growing in thickness, and for this reason horizontal
tubes are most commonly used as the condensing surface. Cold
water flows inside the tube whilst the vapour condenses outside.
The tubes are staggered vertically to prevent too great a build-up
of film on the lower tubes as liquid drips off the upper ones. In
comparison with the horizontal tube a vertical tube or flat surface
144
CONVECTION WITH PHASE CHANGE 145
will allow the liquid film to grow in thickness considerably, and

the average heat transfer rate per unit area is somewhat smaller
than for the horizontal tube.
10.2 A Theoretical Model of Condensing Flow

N usselt proposed an analysis of condensation in 1916. 1 This was
applied first to a vertical surface and the same mechanism was then
extended to the horizontal tube. The results agree well with experi-
ment. The analysis of the vertical surface will be given here to
illustrate the method, and the reader may refer to the literature for
the more lengthy analysis of the horizontal tube. 2 •3
Certain simplifying assumptions are made in the analysis. The
film of liquid formed flows down the vertical surface under the
action of gravity and flow is assumed everywhere laminar. Only
viscous shear and gravitational forces are assumed to act on the
fluid, thus inertial and normal viscous forces are neglected. Further,
there is no viscous shear between the liquid and vapour phases, so
there is no velocity gradient at the phase interface. (The temperature
of the surface is assumed constant at tw and the vapour is saturated
at temperature t 881 .}. The mass flow rate down the surface increases
with distance from the top ; this increase is associated with the
amount of fluid condensing at any chosen point. The model to be
considered is shown in Fig. 10.1. The velocity profile is of the form
shown, with v" = 0 at the surface, and (ovxfoy).,= 6 = 0 at the liquid-
vapour interface.
Assuming that the vertical surface has unit width, it is necessary
to consider an element of fluid dx dy and unit depth, at a distance
x from the top of the plate. The body force on this element is
pg dx dy. The shear stress at y is
OVx
r =p.-
., oy
The shear stress at y + dy is
These shear stresses act over an area 1 x dx. Balancing the forces
1------y
'
b --j (dv,) = ..()
dy y=l
Fig.lO.J. COIIIMMflliDII 011 • fln1it:tll :rllrjtiCe.
gives
pgdxdy = ('r,- ty+dy)dx = -JJ. 0Y;dxdy

ov2
d2vx pg
dy2 = ---;;
and on integration,
pgy2
v, = - 2il +ely+ C2
The boundary conditions are that v, = 0 at y = 0 and dvJdy = 0
at y = a, the thickness of the film. Hence
pga
c2 = 0, and - - + cl = 0
Jl.
The equation for v, is thus
v
X
=- pg(y2
Jl 2
- ya) (10.1)
The mass flow at x can then be obtained by integrating over the
J: J:- p~g(~ -
film thickness b. Thus
2
m= pvxdy = yb) dy
p2gb3 p2gb3 p2gb3

---+--=--
6J1 2J1 3J1
But b is a function of x, and
dm p 2 gb 2 db
dx = -Jl-· dx (10.2)
Next, the heat transfer, dQ, resulting from the condensation of an

element of matter, dm, may be considered. This quantity of energy
is conducted across the film to the wall, so by Fourier's law,
dQ = k dx(tsat. - tw) = k dxOw
(10.3)
b b
where dx is the area of the element of surface of unit depth. dQ may
also be expressed as dmhrg• assuming the vapour is saturated and
there is no undercooling of liquid. From these relationships, dm
may be expressed as
or
dm kOw
(10.4)
dx hrgb
Equations (10.2) and (10.4) may be combined to give
p 2 gb 2 db = k(Jw
J1 dx hrgb
This result may be integrated between the top of the surface down
to x to give
or
(10.5)
This is the relationship between film thickness and distance x from

the top of the surface. From equation (10.3) a convection coefficient
may be obtained as dQ k
h =-=-
X dxOw (j
and hence
Nu = hxx = ~ = (hrsP2gx3)*
x k (; 4f.lk0w
Thus the local Nusselt number may be written as
Nu = 0·706(hrsP2gx3)t (10.6)
x f.lkOw
An average Nusselt number is then obtained by integrating hx from
0 to x and dividing the result by the area x x unit depth, to give
4
Nu = -Nu = 0·943 rsP 2 gx 3)* (h (10.7)
X 3 Xf.lkOW
The analysis on the horizontal tube of diameter d yields a similar
expression for the average Nusselt number, thus
Nud = o·n5(hr~~d 3 r oo.8)

EXAMPLE I 0.1
Steam at 0·25 bar absolute condenses on 30 mm diameter horizontal
tubes which have a surface temperature of 40°C. Calculate the
average heat transfer coefficient.
Solution. The saturation temperature is 65°C at which hr1 =
2345·7 kJjkg. The mean film temperature (at which liquid fluid
properties are taken) is 53°C. Hence p = 986 kg/m 3, J1. = 526 x
w- 6 Pas, and k = 646 X 10- 6 kW/(m K). ow= (t.., - tw) = 25°C.
Equation (1 0.8) gives

- _ .
Nud- 0725
(2345·7 X 986 2 X 9·81 X 0·03 3
526 x 646 x 10 12 x 25
)*
= 0·725 X (712 X 108)*
= 0·725 X 517 = 375·0
li _ . k _ 375·0 X 646 X 10- 6

- 375 0 X d- 30 X 10 3
= 8·08 kW/(m 2 K)
Equation (10.7) for a vertical surface may be applied to a vertical

tube provided the diameter is not small, when the liquid film
becomes two-dimensional, and hence it is possible to compare the
relative merits of horizontal and vertical tubes. Thus
hd = 0·770(~)*
hJC d
If (x/d) is 75, say, it follows that h4 = 2·26 h". Thus over twice the
fluid is condensed with the tubes arranged horizontally, h" being
the coefficient for the vertical tube. Condensation inside a tube is a
process of some interest since it occurs in refrigeration and heat pump
condensers, but it is a process of considerable complexity and is
conveniently described by empirical relationships. For low vapour
]*
velocities Chato4 gives:
~d p!
3 (p 1 - Pv) g (hrg + 0·68 Cpl(tsat - tw ))
Nud = 0·555 (10.9)
k1J.L1 (tsat- tw)
for short tubes at Re (vapour)< 35,000.

Suffices I and v refer to liquid and vapour states respectively. The
subject of condensing flow inside tubes is described further by Akers,
Deans and Crosser. 5
For more advanced topics on condensation the reader is referred
to the literature. It is not possible to consider in this introductory
text the effects of turbulence in the liquid film, 3 velocity of the con-
densing vapour 6 or, superheat?
10.3 Boiling Heat Transfer

Heat transfer to boiling liquids is a subject at present under intensive
study. It is of paramount importance in the power generation
industry. Several fairly well defined regimes of heat transfer are
now recognized, and values of heat transfer coefficient associated
with each have been measured.
Thus when there is a free liquid surface above the heated surface,
the regime is known as pool boiling, and sub-cooled boiling occurs
when the bulk liquid temperature is below the saturation value. As
the temperature rises to saturation, saturated boiling occurs, increas-

ing in intensity as the surface temperature rises to give bulk boiling.
The term nucleate boiling is associated with these regimes as bubbles
leave nucleation sites, leading to film boiling as bubbles completely
cover the surface.
A simple experiment involving an electrically heated wire im-
mersed in water illustrates the simpler boiling mechanisms. 6 The
variation of heat flux with the difference in temperature between the
wire and liquid has been observed by numerous investigators and
the general form of the result is shown in Fig. 10.2. As the wire warms
up initially heat transfer is by natural convection. As the wire
temperature reaches a few degrees in excess of the saturation
temperature streams of tiny bubbles will be observed to leave the
surface of the wire. These bubbles are produced at nucleation sites,
since a minor roughness of the surface is necessary for the bubble
to form. Higher temperatures are found to be necessary for nuclea-
tion to begin if the surface is made especially smooth. Part 1-2 of
the curve in Fig. 10.2 is natural convection, and this becomes
steeper in region 2-3 as boiling proceeds. This initial boiling is
known as nucleate boiling. The heat transfer rate is significantly
8, (wire- fluid)
Fig. 1 0.2. The boiling curve, after Farber and Scorah. 1
improved by the stirring action of the bubbles. Bubble formation

becomes increasingly energetic as point 3 is approached. At this
point the bubbles tend to merge together to form a continuous
vapour enclosure round the wire. When this happens nucleate
boiling gives way to film boiling and there is a reduction in heat flux
due to the thermally insulating effect of the vapour. This situation
leads to a rapid increase in wire temperature and possible melting,
unless the current input is quickly reduced. Once film boiling is
safely established, the heat flux will again increase with temperature
until the wire melts, the mechanism here being convection and
radiation through the vapour.
Many useful calculations on boiling may be made from the
Rohsenow correlation 8 which is in terms of the difference in tem-
perature between the surface and the fluid saturation value and the
heat flux per unit area, for a number of surface/liquid combinations
(10.10)
where
cP1 = specific heat of saturated liquid
hr1 = enthalpy of vapourisation
Pr1 = Prandtl number of saturated liquid
Jl1 = viscosity of saturated liquid
p1 = density of saturated liquid
Pv = density of saturated vapour
a = surface tension of liquid vapour interface
9 = heated surface saturation temperature difference
Q/A = heat flux per unit area
g = gravitational acceleration
C.r =experimental constant
The value of C.r is 0·013 for water-copper and water-platinum, and

(H)06 for water-brass. The equation is dimensionless, so any system
of units may be used without correction.
The use of this correlation may be extended to flow in tubes,
when Rohsenow and Griffith9 recommend that a convective heat
flux may be calculated from (7 .34) and added to that from (10.10) to
obtain a total heat flux for the boiling flow.
Boiling processes may be further sub-divided when considering

the flow of fluid vertically in a tube. The process may be associated
with the type of flow. 10 Various flow regimes are shown in Fig. 10.3
These are: sub-cooled liquid flow, 'frothy' or 'bubbly' flow at low
dryness fraction, 'churn' or 'slug' flow in which slugs of vapour
appear, annular or climbing film flow, fog or dispersed liquid flow,
and finally dry wall flow at the saturated steam condition. Associated
boiling processes are tabulated in Fig. 10.3. Sub-cooled nucleate
and film boiling are examples of local boiling. There is no overall
production of vapour; this is condensed in the main bulk of the
fluid after being produced at the wall of the tube. Very high con-
vection coefficients result because of the activity at the wall, and
this heat transfer mechanism is finding application in other situ-
Order ofmagi
nitude of co-
Quality Mechanisms efficients I
kW/(m2 Kl
Convection to
. . uperheated I· 7
I Gas
- vapour
I X= 10 Convection 17
Dry wall
only
1
I
! Fog or
I dispersed
I liquid
I
Annular Comectivc I 15 to
or Boiling
U1
260
climhing 1 depending on
I film velocity and
!
heat flux.
I Churn
or
0 Bulk boiling (but reducing
to 0· 3-6
0
I
slug
I· roth)
C>
..
oo.•o 1 with film
Saturated boiling)
I ~
or I •.•
. ·. nucleate
huhbl) boiling
l
I Suh~coolcd II 7=0 Suh-conled
nuclcatt: hodin'
licjllld I ConH~ction to
6
"-illl'r
Fig. 1 0.3. Flow and boiling regimes in a vertical heated tube. From data of
Firman, Gardener, and Clapp. 10 By courtesy of the Institution of
Mechanical Engineers
ations where a high convection coefficient is valuable. Saturated

nucleate boiling occurs wnen the bulk fluid temperature has
reached the saturation value, and is therefore associated with
flow at low dryness fraction. This mechanism persists into the slug
flow regime when it is termed bulk boiling. When, with the in-
creasing velocities, annular flow is established, convective heat
transfer between the annulus of liquid and the core of vapour takes
place and the nucleate process tends to be suppressed. This is
known as convective boiling. Initially, the vapour core is thought
to be fairly dry, but with accelerated flow the liquid annulus is
entrained as a dispersed spray or fog in the core. Once the liquid
phase has left the tube wall, as in the dry wall region, the heat
transfer coefficient drops rapidly. The mechanism is by convection
and by conduction to individual droplets impinging on the wall.
Finally, when the steam becomes superheated, heat transfer is by
convection only. Film boiling is avoided in the foregoing as far as
possible. It occurs with excessive heat fluxes and results in drastic
reductions in the boiling coefficient and very high metal tem-
peratures. The order of magnitude of the heat transfer coefficients
associated with the type of flow and mechanism of heat transfer
are also shown in Fig. 10.3. It will be observed that the coefficients
vary over a considerable range.
It will be appreciated from what has been said so far that boiling
heat transfer is a complex subject and to take the subject any further
is beyond the scope of this text. Working formulae and procedures
exist in the literature for the determination of boiling coefficients
for design purposes, and the reader may refer to Bagleyll for a recent
statement from the boiler industry, and to J akob 12 and to Hsu 3 for
more comprehensive treatments of the subject.
ExAMPLE 10.2
Using the Rohsenow equation, calculate the heat transfer coeffi-
cient for boiling when water boils at atmospheric pressure in
a copper pan with the copper surface at l20°C, and compare with
the convection coefficient for water flowing in a 40 mm diameter
tube at 1 m/s under the same conditions, using equation (7 .34).
Use Csr = 0·013, cP 1 = 4·216 kJ/(kg K), h~ = 2256·7 kJfkg, Pr1 =
1·74, J1.1 = 279 x 10- 6 Pas, p1 = 957 kgfm , Pv = 0·598 kg/m 3 , a=
0·0587 N/m. At a mean temperature of ll0°C, p1 = 950 kg/m3 ,
J1. 1 = 252 x 106 Pas, Pr = 1·56; k = 684 x 10- 6 kW/(mK) and at
120oC J1.1 = 230 X 10- 6 Pas.
Solution. The Rohsenow equation will give Q/ A from which h may
J(
be found. Thus:
4·216 X 20 = ()-Ol3 [ Q/A X 106 X 0·0587 \] 0 · 3 3
2256·7 X (1·74) 1"7 279 X 2256·7 9·81(957-0·598)1
Q/A = 358·0 kW/m 2
and h = (Q/A);fJ = 358·0/20 = 17·9 kW/(m 2 K). From equation

(7 .34),
lid = 0.027 (950 X 1 X 40 X 106 ) 0 8

" ( 1.56)t (252) 0 · 14
k X 252 X 103 X X 230
= 442
1i = 442 X 684 X 103 = 7·57kW/( 2K)
106 x 40 m
PROBLEMS
To solve Question I see also chapter 12.
1. An air heater consists of horizontal tubes 30 mm diameter and 23 mm
bore arranged in vertical banks of twenty. Air passes inside the tubes and
is heated from 32°C to 143°C by saturated steam at 180°C which passes
over the tubes. The mean air velocity is 23 m/s and the air flow 3·82 kg/s.
Calculate the number and length of tubes required. The heat transfer coef-
ficient for saturated steam to tube surface (h,.) can be found from
h = 0·725(k~p;ghr,)t kW/(m 2 K)
•• NdJl/it
where the suffix c denotes condensate properties evaluated at the saturation
temperature, g is the gravitational acceleration in m/s 2 , N is the number of
horizontal tubes in a vertical bank, d is the outside diameter in m, t is the
temperature difference between the saturated vapour and the tube surface
and may be assumed to be 11 a c. The other symbols have their usual
meaning. (Ans. 400 tubes, 2·34 m.) (Unirersity of Glasgow).
2. Water flows in a 0·8 em bore copper tube at a Reynolds number of 10,000.

The saturation temperature is 105°C and the wall temperature 130°C.
Calculate the boiling heat flux using the Rohsenow equation and hence the
total heat flux. Use the following property values: u = 0·0525 N/m, hr1 =
2244 kJ/kg, p1 = 954 kgjm 3, P. = 0·71 kg/m\ cP 1 = 4·23 kJ/(kg K), Pr1 =
1·64, J-11 = 265 X 10- 6 Pas, Jlw = 230 X 10- 6 , k = 687 X 10- 6 kW/(mK).
(Ans. 956 kW/m 2 , 979 kWjm 2 .)
3. Describe the 'Farber-Scorah Boiling Curve' together with the mechanism

of heat transfer relating to each section of the curve. Discuss the following
topics in relation to the heat transfer to a fluid in which nucleate boiling occurs:
(a) Temperature distribution in the fluid:
(b) The nature of the heating surface:
(c) The operating pressure.
(U nit:ersity of Leeds).
4. Steam is being condensed on flat vertical surfaces. If the drag on the steam
side of the condensate film can be neglected, derive an expression for the local
and mean heat transfer coefficient on the surface.
Discuss the assumptions which you make in the derivation.
If the surfaces are parallel and steam enters the space between two surfaces
at the top, show how you would correct the derivation for the drag of the
flowing steam on the condensate film. (University of Leeds).
5. Outline the Nusselt theory of film condensation, indicating the steps which
lead to the following formula for the average surface heat transfer coefficient
hm during the condensation of a saturated vapour on a plane vertical surface:
N = hmL = 0·943(p1gL3hcg)*
Urn K J1Kf1T
Lis the height ofthe surface, g the acceleration due to gravity, hc 1 the enthalpy
of evaporation, 11 T the difference between the temperatures of the vapour
and the surface and p, Jl, and K are respectively the density, absolute viscosity,
and thermal conductivity of the condensate at the saturation temperature.
Saturated steam at 149°C is to be condensed in a cylinder of diameter
1·217 m and length 0·305 m, having its axis vertical. The curved wall is main-
tained at 10°C by external coolant and no condensation takes place on
the two horizontal surfaces. The steam is fed in through a pipe in the top
surface of the cylinder.
Determine the initial average surface heat transfer coefficient, and estimate
the time taken to fill the container with water which may be assumed to
remain at 149°C. (Ans. 4·85 kW/(m 2 K), 0·976 h.) (University of Cambridge).
REFERENCES
1. Nusselt, W. Z. d. Ver. deutsch. lng., Vol. 60, 541 (1916).
2. Nusselt, W. Z. d. Ver. deutsch. lng., Vol. 60, 569 (1916).
Princeton (1963).
4. Chato, J. C. J. Am. Soc. Refrig. Air Cond. Eng., Feb., 52 (1962).
5. Akers, W. W., Deans, H. A. and Crosser, 0. K. Chem. Eng. Progr.,
Symposium Series, Vol. 55, No. 29, 171 (1959).
6. Carpenter, F. G. and Colburn, A.P. 'General Discussion on Heat
Transfer', I. Mech. E. London (19 51).
7. Farber, E. A. and Scorah, R. L. Trans. ASME., Vol. 70,369 (1948).
8. Rohsenow, W. M., Trans. ASME, Vol. 74,969 (1952).
9. Rohsenow, W. M. and Griffith, P. AIChE-ASME Heat Transfer
Symposium, Louisville, Ky (1955).
10. Firman, E. C., Gardner, G. C., and Clapp, R. M. /. Mech. E.
Symposium on Boiling Heat Transfer, Manchester, Review Paper 1
(1965).
11. Bagley, R. I. Mech. E, Symposium on Boiling Heat Transfer,
Manchester, Paper 13 (1965).
12. Jakob, M. Heat Transfer, Vol. 2, John Wiley, New York (1957).
11
Extended surfaces
Convection from a solid surface to a surrounding fluid is limited by
the area of that surface. It would seem reasonable, therefore, that if
the surface area could be extended, then a gain in total heat transfer
would be achieved. This is done by adding fins to the surface. Heat
transfer is then by conduction along the fin, and by convection from
the surface of the fin. It is likely that the convection coefficient of
the basic surface will be altered by the addition of fins, due to the
new flow pattern involved and the fact that the temperature of the
fin surface will not be uniform. Though the average surface tempera-
ture is reduced by the addition of fins, the total heat transfer is
increased. In the treatment that follows it is assumed that the
convection coefficient is known. The Nusselt numbers of finned
surfaces may be determined experimentally.
There are various types of fin, the most common being the
straight fin, the spine, and annular fin. The straight fin is rectangular
in shape and generally of uniform cross-section, and the spine is
simply a short thin rod protruding from the surface. Annular fins
are often found ifthe primary or basic surface is cylindrical. Examples
are to be found in heat exchangers and air-cooled petrol engines.
Extended surface nuclear fuel cans are shown in Fig. 11.1. These
are both straight and spiral in form.
Only the straight fin and spine will be considered here in detail.
Fins of non-uniform cross-se_ction and annular fins are more
complex mathematically, and the reader is referred elsewhere for
details. 1 •2 •3
11.1 The Straight Fin and Spine

These are shown in Fig. 11.2. The straight fin has length L, and
height I (from root to tip). These definitions are used whatever the
actual orientation ofthe fin may be. In developing the theory of heat
157
transfer in a fin it is assumed that the thickness, or diameter of the
spine, is small compared with the length. Conduction along the fin
may then be assumed to be one-dimensional. The conduction and
convection heat transfers involved are shown in Fig. 11.3. Two
important dimensions of fins are their area of cross-section A, and
their perimeter P. In the straight fin it is convenient to assume that
a is small compared with L. Thus:
Straight fins A= La, P = 2L

Spines A = !1rd 2 , p = 1!d
Consider an element of a fin or spine as shown in the figure.

Conduction into the element at x is Qx. This must be equal to
the sum of the conduction out of the element at x + dx and the
Fig. 11.1 Magnox fuel element, as used in the U.K. gas cooled reactor pro-
gramme. Note the external extended surface configuration. This illustration is
reproduced by courtesy of Nuclear Fuels, pic.
EXTENDED SURFACES 159
convection from the surface of the edge of the element. Thus

dt
Q = - kA-
" dx
dt d 2t
Q(x+dx) = - kA-- kA-2 dx
dx dx
Qh = hP dx(t - t.)
and
Qx = Q(x+dx) + Qh
d2t
- kA dx 2 dx + hP dx(t - t.) = 0
d 2t hP
dx2 - kA (t - t.) = 0
Fig. 11.2. The straight fin and the spine.

Surroundings
t,
Qo - - Qx+dx
.-----------+-~--------~
-- - X
x=O x + dx
Fig. 11.3. Heat transfer from an extended surface. This diagram implies heat
transfer from the fin to the surroundings. For the reverse case an inverted tem-
perature profile would be obtained.
Since t. is assumed a constant surroundings temperature, (t - t.)

may be replaced by (}, and d 2 t/dx 2 becomes d 2 9/dx 2 •
d 2 (} hP (} _ O
dx 2 - kA -
This differential equation in (}has a solution of the form:
(11.1)
where
m = (:~r (11.2)
and C 1 and C 2 are constants of integration to be determined from

boundary conditions.
The first boundary condition is that (} = 9 0 at x = 0. Therefore,
from equation (1 1.1):
(1 1.3)
The second boundary condition depends on the heat transfer

from the tip of the fin. If the fin may be assumed long and thin this
is very small and may be assumed to be zero with very little error.
(11.4)
Solution of equations (11.3) and (11.4) yields the values of C1 and

c2. i.e.,
u 0 e -ml
l1 l1 ml
cl = eml + e ml' and C 2 = emluo+ee-ml
Substitution of these values back into equation (11.1) gives
- [em(l-x) + e-m(l-x)J
0- Oo ml ml
e +e
0 cosh m(l - x)
-= (11.5)
00 coshml
Even though it was assumed that (dO/dx)(x=n = 0, the temperature
at the end of the fin is still above t 1 , and is given by
0 _ Oo
(11.6)
1 - coshml
This is obtained by putting x = lin equation ( 11.5).
The total heat transfer from the fin is obtained by considering
the conduction into the fin at the root. Thus:
Qo = - kA(dO)
dx x=O
= mkAOo[sinh m(l- x)l

coshml Jx=o
= mkA80 tanh ml (11.7)

This result applies equally to the straight fin and spine, the appro-
priate value of m has merely to be substituted.
If the fin is comparatively short the assumption of no heat transfer

from the tip of the fin is not valid. Under these conditions the heat
transfer at the tip is given by
- kA(d(}) = + hA01
dx x=l
- k(mC 1 eml - mC 2 e- ml) = + h01 ( 11.8)
The constants C 1 and C 2 may now be obtained by solving equations
(11.3) and (11.8). Substituting for C2 in (11.8) and eliminating 01 by
using (11.1):
-k[mC 1 eml- m(0 0 - C 1 )e-m1] = +h[C 1 em 1 + (0 0 - C 1)e-m1]
This then gives:
Oo[e-ml- (h/km)e-m 1]
CI - -~-'---;------o--=----,-
- (eml + e-ml) + (h/km)(eml _ e-ml)
and
C - Oo[eml + (h/km)eml]
2 - (eml + e ml) + (h/km)(eml _ e ml)
and on substituting back into equation (11.1) gives
(} em(l-x) + e-m(l-x) + (h/km)[em(l-x)- e-m(l-x)]
Oo = (eml + e-ml) + (h/km)(eml- e-ml)
which may be expressed as

(} cosh m(l - x) + (h/km) sinh m(l - x)
(11.9)
00 cosh ml + (h/km) sinh ml
The temperature difference at the end of the fin is given by
(} - Oo
(11.10)
~-
cosh ml + (h/km) sinh ml
The heat transfer from the fin is obtained as before by considering
(dOjdx)x= 0 • Thus
Q0 = - kA(d(})
dx x=O
= -kAOo[-msinhm(l- x)- (h/k)coshm(l- xn

cosh ml + (h/km) sinh ml Jx=o
= mkAB [siuh ml + (h/km) cosh'"']

0 cosbml + (h/km)siuhml
tanh ml + h/km
= mkABo [ 1 + (h/km) tanh ml
J (11.11)
EXAMPLE 11.1
A transistor beat sink is a 100 mm length of aluminium section as

shown consisting of a 70 mm x 100 mm plate with 12 integral fins
2S mm hiah by 1 mm thick. H the plate is at 4S K above the surround-
ings find the percentage of beat transfer from the sink that occurs
from the fins. k = 0·1S kW/(m K~ h = 0.03 kW/(m 2 K). Neglect
beat transfer from the plate and fin edges. (The City University).
Solution. The plate surface temperature excess is 45 K, so heat

transfer by convection from the plate is (40 + (4 x 6)) x 100 x 2 x
0.03 X 4S X 10- 6 - 0.0173 kW.
For the fins, m = (2 x ()-03 x 1000/()-15 x 1)0 "5 = 200
For fins 2S mm in height, ml = 2()-0 x 25-Q/1000 = 0·5
Tanh ml = ()-462 Hence heat transfer from 12 fins 100 mm long
= 20-() X ()-15 X 1 X 4S X ()-462 X 100 X 12/106 = 0·075 kW
The total heat transfer is ()-()923 kW, 81·3 per cent ofthis being from
the fins.
~'--
r-
40mm
-ll-
lmm
Fig. Example 11.1.

11.2 Limit of Usefulness of the Straight Fin

It is important to recognize the fact that fins may not necessarily
improve heat transfer from a surface, and the conditions under which
fins will not be useful must be investigated before any design work is
contemplated. There are in fact three possibilities which arise from the
particular value of the dimensionless grouping (h/km) which occurs in
equation (11.11). By inspection:
when h/km > 1 then Q0 < h A 8 0

when h/km = 1 then Q0 = h A 8 0
when h/km < 1 then Q0 > h A 8 0
Clearly, the requirement for increasing heat transfer from the surface is
that h/km < 1, so that Q0 is greater than hA 8 0 which is the heat
transfer from the surface having no fins. The possibility of using fins for
insulation arises when h/km > I, and Q0 < h A 8 0 • It will be found that
to achieve a value of h/km > 1, the resulting fins are so thick that the
surface is virtually completely covered in insulation.
11.3 Fin and Finned Surface Effectiveness
A fin effectiveness may be defined by relating the actual fin per-

formance to that of an 'ideal fin' which has a uniform temperature
all along its surface equal to the temperature at the root. Such a
fin would result if constructed of a material having infinite thermal
conductivity. The heat transfer from an ideal fin would be defined by
Q~ = Plh60 (11.12)
neglecting heat transfer from the end.
Taking the heat transfer from the actual fin to be given by (I 1.7)
Q0 = mkA60 tanh ml
then the fin effectiveness, 'lr· would be given by
Q0 mkA60 tanh ml
'lr = Q~ = Plh6 0
This reduces to
Q0 tanhml
(11.13)
Q~ = ml
If the fin which has a significant end heat transfer is compared with
the ideal fin as defined by ( 11.12) then
tanh ml + h/km
'1r = (11.14)
ml + (hl/k) tanh ml
The fin effectiveness is a useful idea in relation to the next topic
to be considered, the overall heat transfer coefficients of surfaces
which have fins. In Chapter 3, overall coefficient were derived for
plane and cylindrical surfaces. Similar coefficients can be written
for surfaces, both plane and cylindrical, on which fins have been
added.
In the derivation of ( 11.13) it is seen that
('1rPl)h0 0 = mkA00 tanh ml
so '1r may be interpreted as the fraction of fin area which may be
regarded as being at 00 all over for purposes of calculating heat
transfer. A function 'Ire is now introduced which is the fraction of
area of a finned surface at 00 • If A 8 and All are the total fm surface
area and fin root area per unit area of primary or basic surface,
respectively, then the total area of surface at 00 is 1 - All + 'frA 5 .
As the total area is 1 - All + A 5 , the ratio of actual to ideal heat
transfer from a finned surface is
(1 - All + 'frA 8)h00 1 - All + 'frAs

{11.15)
'Ire = (1 - All + As)h0 0 = 1 - All + As
An alternative method of assessing a finned surface is to compare
its performance with that of the surface without fins, thus a surface
'coefficient of performance' would be given by
11.4 Overall Coefficients of Finned Surfaces
Fins are often added to only one surface to reduce the thermal
resistance on that side. However, Fig. 11.4 shows a plane surface
fluid a fluid b
t. tb
Heij!ht of
~I ,.. Height of
...,
a-fins b-fins
(b)
Fig. 11.4. (a) Plane finned surfaces &eparating two fluids. (b) Temperature pro-
files for (a). Heat transfer is to the a-fins from the fluid at t0 and from the b-fim
to the fluid at tb-
with fins both sides. Heat transfer to the a-fins per unit plane area is
given by:
Q= - h.(l - AR + '1cAJ. (t 1 - t.)

Similarly, from the b-fins to fluid b
Q = -hb(l - AR + '7rAs)b (tb- t2)
The heat transfer by conduction across the slab is
These three quantities are equal and combine to give

Q = - U(tb - t.)
where U is the overall heat transfer coefficient, given by
(11.17)
In this analysis, the group (1 - AR + 'TeAs). means that AR, '1c and As
all refer to the a-fins, and similarly for the b-fins.
A similar result may be obtained for a tube finned internally and
externally, as shown in Fig. 11.5. Unit length of tube may be con-
sidered and the following three equations for heat transfer may be
written:
Convection inside: Q = - h.2nr 1 (1 - AR + '1cAJ.(t 1 - t.)

2nk
Conduction : Q = - - - ( t 2 - t1)
ln r2 /r 1
Convection outside: Q= - hb2nr 2 (1 - AR + '7rAs)b(tb- t 2 )
Fig. 11.5. Cylindrical finned surfaces separating two fluids.

These equations then lead to the result
Q- -UL(tb- tJ,
where
U _ 1/{ 1 +lnr 2 /r 1
L - 27tr 1(1 - Aa + '1rAJ.h. 27tk
+ 27tr2(1 - A:+ '1rAJbhJ

(11.18)
This has units ofkW/(m K) or the equivalent. It is to be noted that

(1 - Aa + ,,As). is the fully effective surface area per unit area of
the a-primary surface. This must then be multiplied by the area of
primary surface per unit length, 27tr 1 • (SuffiX L denotes U for unit
length of tube).
Alternative expressions for U in terms of unit area of tube surface
may also be obtained. If UL in equation (11.18) is divided by 2rrr2 ,
the area of external primary surface per unit length. the result is
UA - 1/{r1(1 - Aa
r2
+ 'lrAJ.h.
+ r21n rJrt
k
+ (1 - Aa ! '7rAJbhb}
(11.19)
This has units of kW/(m 2 K) where the area refers to the external tube
surface (primary) area. The corresponding result for the unfmned
surface is
(11.20)
This is again in terms of external tube surface area.

These results are used in heat exchanger theory, in Chapter 12.
EXAMPLE 11.2
A stainless steel heat exchanger tube is 25 mm outside diameter with

a 2·5 mm wall thickness. Convection coefficients inside and outside
are 6-Q and 1-Q kW/(m 2 K) and the thermal conductivity is 0·04
kW/(m K). A similar tube has 20 axial fins 14 mm high by 2 mm thick.
Find the overall coefficient in both cases, and in each case state which
thermal resistance is controlling.
Solution. In the first case, equation (11.18) simplified for no fins
becomes equation (3.17) with only one conduction term.
. U 1/{ 1 In 1·25 1 }
• . L = 27t X ()-01 X 6-() + 21t X ()-04 + 27t X ()-0125 X 1·0
= 1/(2·65 + 0·887 + 12·75)
= 1/16·3 = Q-0614 kW/(m K)
The convection resistance on the outside is clearly the largest and is

therefore controlling, meaning that to reduce the overall resistance
greatest benefit will be obtained by reducing this part of it.
In the second case, fins are added to the outside surface. The
fm efficiency, 'I =(tanh ml)/ml. Considering 1m length,
m = ..j(hP/ka) = ..j(1·0 x 2·0/Q-04 x 0-()()2)
= 158 and ml = 2·22

'I = (tanh 2·22)/2·22 = Q-977/2·22 = 0·44
For the finned surface, 27tr0 = 27t x Q-0125 = 0·0785 m2 /m
27tr0 Aa. = root area/m length = 20 x 0·002 x 1 = ()-04 m 2/m
27tr0 ,A8 =effective fin area/m length = 20 x 2 x 0·014 x 0·44
= Q-246m 2/m
1/27tr0 (1 - Aa. + '1As)h 0 = 1/(0·0785 - ()-04 + 0·246) x 1·0
= 3·51
Equation (11.18) now gives
U L = 1/(2·65 + ()-887 + 3·51)
= Q-142 kW/(m K)
Although the inside and outside resistances are now similar, the
outside one is just still controlling.
11.5 Numerical Relationships for Fins

The range of fin problems that may be analysed is greatly increased
by the introduction of simple numerical relationships. Thus it is
possible to include a variable convection coefficient, or even a
tr
1--r--.---r----,----.,t,......ho--,---,.-,1_
I • I • I • I • b
~~~~~~~~T
Fig. 11.6. Straight sided fin with h dependent on position.
transient analysis of a straight sided fm. Further, flat rectangular

fins of the type encountered in air-conditioning equipment, can
readily be analysed. The necessary relationships are deduced by the
methods outlined in Chapters 4 and 5. Some examples are now given.
In Fig. 11.6 is shown a straight sided fin having a convection
coefficient dependent on position. Separate relationships would be
required for the root node, the tip node, and a further one for nodes
in between. For central nodes:
(b x l)k (t 2 - to) + (b x l)k (t 1 - to) + h0 (2x x 1) (tr - t 0 ) = 0
X X
2h 0
tl + t2 + ( bk tc -
x) (2 + bk
2h x)
to = 0
0
In BASIC the statement would be:

X= (T(I- 1) + T(l + 1) + BTF(I)•TF)/(2·0 + BTF(I)) (I 1.21)
Used in an iterative procedure X is the new value of T(I~ and
BTF(I) = 2·0•H(I)•X/(B• TK)
where H(l) is the value of the convection coefficient at node I,
X= x. B = b, TK = k, and TF = tr. Corresponding BASIC statements
for root and tip nodes are:
(11.22)
X=(T(I-1)+0·5•BTF(N)•TF)/(1·0+0·5•BTF(N)) (11.23)
Hence it is seen that the root node is at the surface temperature T(l)
and BTF(N) refers to the end node at I = N.
Fig. 11.7 shows the layout of a flat rectangular fin having a circular
or elliptical root, which is approximated to the rectangular grid.
Such a fin is usually symmetrical, so only one quarter need be con-
sidered. The general BASIC program in Chapter 4 is suitable for a
outline of root
,...----,
: I
1 12 I
,----:-- -~-l--.
I I I :
I 13 t1 1
I I to I Y
L ___ _j _ _ _ L ___ .J
1
I I f
___I I_
I
X
element thickness b, : 14
convecting to tr on both L ___ ..J
faces
Fig. 11. 7. F1at rectangular fin with circular root.
steady state analysis of temperatures in such a fin. The equation for a

field node is obtained from:
kyb ( - to
t1 -- +tJ--- to) + k b (t2---to+t4-- -to)
X
X X y y
+ 2hxy(tc - t0 ) =0 (11.24)
Other equations are needed for side and corner boundary nodes,
obtained from similar energy balances.
Other aspects of fin performance may be readily studied using

programs in BASIC. For example, a fin root which is at the highest
temperature in the fin also carried the greatest conduction heat load.
Consequently the possibility arises of improving fin performance by
using a material having temperature-dependent thermal conductivity, so
that the highest thermal conductivity occurs where it is needed, i.e., at
the root.
The iterative sequence for two-dimensional steady state conduction
introduced in Chapter 4 may be modified to allow for temperature-
dependent thermal conductivity. In each iteration, and at each field
point, the thermal conductivity for conduction with each adjacent node
must be calculated from the temperatures then existing. As convergence
occurs, both correct temperatures and correct thermal conductivity
values will be obtained.
A mesh grid for a rectangular plate fin mounted on a circular tube is
shown in Fig. 11.7. For this fin the mesh size is A and the fin thickness
is B, and the surface convection coefficient is H. By way of example,
the thermal conductivity is given by k = 50·0 + 0·1T, so that at 0°C,
k =50, and at 100°C, k = 60 W/m K. The conductivity between adja-
cent nodes is taken at the average temperature of those nodes. The
iterative sequence is given below.

500 CK=H*A*A/B
502 X=50.0
503 Y=O. 1
510 ITER7.=0
520 L7.=0
530 FOR J%=2 TO 15
540 FOR I7.=2 TO 16
550 Cl=Xl+!Yl*<T< I:r., ,17.-1 l+T! 17., ..J7.)) l/2. 0
560 C2=Xl+!Yl*!T!IZ-1,JZ)+T!IZ,JZ>>>I2. 0
570 C3=X1+!Y1*!T! 17., J7.+1 l+T! I7., ..J7.) > )/2. 0
580 C4=Xl+!Y1•<T<J7.+1,J7.>+T!J7.,J7.)l)/2. 0
590 C!l=Cl*!T!J:Y.,J:Y.-1))
600 C22=C2*<T(Ii.-1,JZ>>
610 C33=C3*!T!J7.,J7.+1>>
620 C44=C4*!T!J7.+1,J7.))
630 K:r.~M:r.!I7.,J7.)
640 IF K7.=1, GOTO 650
641 IF K7.=2, GOTO 670
642 IF K7.=3, GOTO 690
643 IF K7.=4, GOTO 710
644 IF K7.=5, GOTO 730
645 IF K7.=6, GOTO 750
646 IF K7.=7, GOTO 770
647 IF K7.=8, GOTO 790
648 IF K7.=9, GOTO 8!0
649 IF K7.=JO, GOTO 830
650 X=TIN
660 GOTO 840
670 X=!C11+2.0*C44+C33+2.0*CK•TC0N)/!C!+C3+2. O*!C4+CK))
bBO GOTO 840

b90 X=<CII+C44+CK*TCONl/(CI+C4+CK>
700 GOTO 840
710 x~<CII+(C44+C22l/2.0+CK•TCONl/(CI+(C2+C4l/2.0•CKl
720 GOTO 840
730 X•<Cil+C22+CK•TCONli<Cl+C2+CKl
740 GOTO 840
750 X•<<Cil+C33l/2.0+C22+CK~TCONl/((CI+C3l/2.0+C2+CKl
7b0 GOTO 840
770 X=(C22+C33+CK*TCONl/(C2+C3+CKl
780 GOTO 840
790 X=<C~2+2.0*C33+C44+2.0*CK•TCQN)/(C2+C4+2. O•<C3+CK))
BOO GOTO 840
810 X=<C11+C22+C33+C44+2.0•CK•TCONl/<Cl+C2+C3+C4+<2. O•CKll
820 GOTO 840
830 X-.£>.0
840 DT•ABS<T<IY.,JY.l-Xl
850 IF<DT)0. 005lGOTO 870
8b0 LY.=L%+1
870 T< 17.. JY.l=l"< IY., J%1+1. 9* ( X-T< 17.. J7.l l
880 NEXT 17.
890 NEXT JY.
900 ITER7.=ITER7.+1
910 IF<L7.(210lGOTO 520
920 IF<<210-L7.))QlGOTO 520
PROBLEMS
1. The diagram shows the cross-section of a nuclear reactor fuel element
consisting of a uranium fuel rod 28 mm diameter contained in a magnox
can which has longitudinal finning on its external surface. The fuel rod/can
interface temperature is 430°C and the heat release rate is 65·6 kW per m
length. Calculate the maximum temperature within the fuel rod, the tern-
fuel rod
urrounding
graphite
now area of --'\~::::::::::::

coolant gas
longitudinal fins
magnox can
perature drop across the 2·5 mm wall of magnox, and the total surface
area of longitudinal tinning per unit length, given that the fin root is 40 per
cent of the external can surface area, fin efficiency is 90 per cent (based on
fins + splitters which also act as fins), mean coolant gas temperature 280°C,
surface convection coefficient 0·8 kW/(m 2 K), k for magnox 0·15 kW/(m K~
k for uranium fuel 0·0325kW/(mK). (Ans.: 591°C, 11·5°C, 0·589m 2/m.)
(The City University).
2. Write a program in BASIC (or FORTRAN if preferred) which will

calculate the heat transfer rate from a 100 mm square fin, of thickness
0·5 mm, having a root approximating to a 25 mm diameter circle. Use a
temperature-dependent thermal conductivity for the fin material, and
investigate the percentage increase in fin heat transfer for a material
which shows (a) SO per cent and (b) 100 per cent increase in thermal
conductivity over the temperature range between the root and surround-
ings temperature. In addition, investigate the effect of surface convec-
tion coefficient, for values typically in the range of (i) natural convection
and (ii) forced convection.
3. A bar simulating a gas turbine blade, cooled at the root, is 10·2 em long
and has a cross-sectional area (A) 1·93 cm 2 , and a perimeter (p) of 7·6 em.
Gas at 815°C streams across it, and one end is cooled to 483oC. The mean
heat transfer coefficient for the gas flow conditions can be assumed constant
over the surface at ()-284 kW/(m 2 K), and the thermal conductivity of the
material of the bar (k) is 26 x 10- 3 kW/(m K~ Show that
9" cosh mL(1 - x/L)
9, = coshmL
where
9" = t1 - t", 9, = t1 - t, m = / :~
L is the length of the bar and x the distance from the cooled end, t 1 being the
gas temperature and t_, and t, the appropriate blade temperatures. Hence
find the heat passing to the cooled end ofthe bar in kW. Neglect the heat lost
from the uncooled end of the bar. (Ans.: 0·11 kW.) (Queen Mary College,
London).
4. Heat flows from a body A along a wire of diameter d and length I, the other
end of the wire being connected to a body B. The conductivity of the wire is
k and the surface coefficient of heat transfer ia 2 kd.
The temperature of the body A is maintained at 8A above the temperature
of the environment, and the temperature of the body B is 88 above the tem-
perature of the environment. Derive an expression for the temperature of
the wire at x from A, and deduce the particular values of 88 for which
(a) heat flow into B is one-half of the heat flow from A,
(b) heat flow into B is zero. (University of Oxford).
5. The cooling system of an electronic package has to dissipate 0·153 kW

from the surface of an aluminium plate 100 mm x 150 mm It is proposed
to use 8 fins each 150 mm long and 1 mm thick. The temperature difference
between the plate and surroundings is 50 K, the thermal conductivity of
plate and fins is ()-15 kW/(m K), the convection coefficient is ()-04 kW/(m 2 K).
Calculate the height of fm required and the effectiveness of the whole cooling
surface. (Ans.: 3()-3 mm, 88·4 %.) (The City University).
6. A vertical pipe is 3 m long by 50 mm diameter. The surface is at
70°C and it convects to the surroundings at l5°C. Twelve rectangular
fins are integral with the surface of the pipe; they are 3 m long by
40 mm in height (i.e., they extend 40 mm radially from the pipe surface
and run the entire length of the pipe), and they are 2 mm thick. Show
that natural convection over the pipe and fin surface is turbulent and
=
that the heat transfer coefficient is to be determined from Nu 0·129
(GrPr) 0 "333 (where(}= 55°C).
Determine: (a) the heat transfer coefficient, (b) the fin efficiency and
(c) the rate of convection from the total pipe and fin surface. (Ans.
(a) 5·526 W/m K, (b) 0·926, (c) 931·9 W.)
7. A shell and tube heat exchanger is to use copper tubes of outer

diameter 15 mm having a wall thickness of I mm. The outer surface
will have integral copper fins 1 mm thick and 10 mm in height, running
axially alonp the tube. The fluid convection coefficient inside the tube
is 480 W/m K, and outside the tube it is 100 W/m 2 K.
Determine to the nearest whole number, the number of fins required to
give an equal thermal resistance on both surfaces of the tube, and calcu-
late the overall heat transfer coefficient per unit length of the tube.
Take k for copper as 368 W/m K. (Ans. 7 fins, (6·364), 8·384 W/m K.)
REFERENCES
1. Jakob, M. Heat Transfer, Vol. 1, John Wiley and Sons, Inc., New York
(1949).
2. Eckert, E. R. G., and Drake, R. M. Analysis of Heat and Mass Transfer,
3. Chapman, A. J. Heat Transfer, 3rd ed., The Macmillan Company, New
York ( 1974).
12
Heat exchangers
Much of the basic conduction and convection theory finds its
greatest application in the heat exchanger. Whenever it is necessary
to transfer energy from one fluid to another in large quantities,
some form of heat exchanger is used. The most common form of
heat exchanger is that in which two fluid streams pass through in
steady flow, and heat transfer takes place through a separating wall.
Mechanisms involved are therefore convection to or from the solid
surface and conduction through the wall. The wall may be corrugated
or finned to increase turbulence and the heat transfer area.
The thermal capacity of a heat exchanger is usually kept small,
and is of significance only in transient conditions. However, a
regenerative type of heat exchanger does have a large thermal
capacity matrix through which the hot and cold fluids _pass alternately.
By this means energy is transferred indirectly between the fluids.
This chapter is concerned only with non-regenerative heat ex-
changers in which the fluids are separated. Other types of heat
exchanger exist in which the fluids mix. These include cooling
towers and jet condensers, for example. The basic principles will
be considered in relation to the simplest types only.
12.1 Types of Heat Exchanger, and Definitions

The two basic types of heat exchanger are the in-line or uni-
directional flow exchanger and the cross-flow exchanger. Flow is
along the same axis in the in-line exchanger, but the two fluids may
flow in the same or opposite directions giving rise to the names
parallel and counter flow. The in-line exchanger may consist simply
of two concentric tubes, one fluid flowing in the inner tube and the
other in the annulus. Alternatively, there may be a number of tubes
within a large tube or shell and to increase heat transfer the shell
fluid is made to flow partly across the tubes by means of batHes.
176
HEAT EXCHANGERS 177
Counter and parallel flow also occur in plate heat exchanges in
which the fluids flow between closely spaced plates sealed at the
edges. Fig. 12.1 shows some simple in-line arrangements and Fig.
12 .2 shows a part section of a shell and tube heat exchanger with
baffies in the shell of the segmented and 'doughnut' type.
The cross-flow exchanger is, as its name implies, one in which
the two fluid streams flow at right angles. Gas-to-gas heat exchangers
(a) Counter flow concentric (b) Parallel flow

tube exchanger concentric tube
exchanger
~~~;1 I .1 !J -~~~~j
(c) Tube and shell exchanger in parallel flow.
Tube fluid may be reversed to give
oount" How. I~
IIU
(I f--
tl ~
(d) Two-pass tube and shell exchanger.
Flow is partly counter, partly parallel.
Fig. 12.1. Basic in-line heat exchangers
are often of this type. Their analysis is complicated because fluid

temperatures vary in both the direction of flow and at right angles
to that direction.
The temperature variations of the fluids in parallel and counter
flow are shown in Fig. 12.3. Temperatures are plotted against length
or area of heat exchanger surface. The inlet end, where length or
area is zero is regarded as being the end where the hotter of the two
fluids enters. The fluids are regarded as being hot or cold, for con.-
venience, and rh is a temperature of the hot fluid, t 0 a temperature
Fig. 12.2. A liquid/liquid shell and tube heat exchanger. This type of unit is used
for cooling transformer oil, with water as the cooling medium. Pressure drops: oil
flow in the shell, 5-12 psi, water flow in the tubes, 1-5 psi. The heat transfer area
is in the range 110-1090 ft 2 , and the heat transfer rate is in the range 70-1950
kW. Photograph by courtesy ofAssociated Electrical Industries Limiteq,
HEAT EXCHANGERS 179
of the cold fluid. Suffixes 1 and 2 are used for inlet and outlet of
individual streams, and ei is the temperature difference between
fluids at the inlet end and eo the difference at the outlet end of the
exchanger. An important term in heat exchanger theory is the
capacity ratio C. It is a ratio of the products of mass flow rate and
(a) Parallel flow 0;
'I-0 0
(b) Converging at inlet end (c) Diverging at inlet end
Counter flow
thl th I th2 _l
r ·rI
I Oo
0; I
0; t c2
.l j_
tel t, I
(d) Boiling (e) Condensing
Fig. 12.3. Temperature distributions.
specific heat of each stream. It is always the ratio of the smaller

product to the larger, since they are not necessarily equal. Thus,
if mhcph is the 'capacity' of the hot stream and mccpc is that of the
cold stream, then
(12.1)
(12.2)
In counter flow, the temperature distributions are rather different

depending on the capacity ratio. Thus, in Fig. 12.3b, the temperatures
are converging at the inlet end when mhcph > mccpc and equation
(12.1) applies. In Fig. 12.3c the temperatures are diverging at the
inlet end when mccpc > mhcph and equation (12.2) applies.
In parallel flow it is obvious that tc 2 will approach th 2 for an infi-
nitely long heat exchanger, but can never exceed th 2 . In counter flow
it is quite normal for tc 2 to exceed th 2 and, consequently, the counter
flow exchanger is the more 'effective'. Effectiveness is the ratio of
energy actually transferred to the maximum theoretically possible.
Again, the definition depends on the relative thermal capacities of
the streams. The maximum theoretical transfer will take place in
counter flow in an exchanger of infinite length and, in such a case,
tc 2 -+ th 1 when mhcph > mccpc• and th 2 -+ tc 1 when mhcph < mccpc·
Thus the maximum transfers in the two cases are:
mccpc(th 1 - tc 1 ) when mhcph > mccpc

mhcph(th 1 - tcJ when mhcph < mccpc
The actual transfers in the two cases are mccpc(tc 2 - tc 1 ) and

mhcph(th 1 - th 2 ), and hence E, the effectiveness, becomes
(12.3)
and
E = thl - thz when ( 12.4)
thl -tel
These definitions may be used in either counter or parallel flow,
but the value of E will be lower in parallel flow.
Temperature distributions with a change of phase are also
shown in Fig. 12.3. These will occur in boiling, Fig. 12.3d, and
condensing, Fig. 12.3e. Only the phase change takes place in the
exchanger, so the temperature of the boiling or condensing fluid
does not change. The temperature distributions are the same for
both parallel and counter flow. The capacity ratio C becomes 0
HEAT EXCHANGERS 181
for both boiling and condensing since the larger thermal capacity
is in each case infinite. This follows, since by definition, cP = dh/dt
= oo when dt = 0. Equations (12.3) and (12.4) may be used in
condensing and boiling, respectively.
The other limit of capacity ratio is C = 1 and occurs when the
thermal capacities of the two streams are equal. This is not illustrated,
but it results in the temperature distributions being parallel straight
lines in the case of counter flow, () being a constant over the whole
heat exchange area.
12.2 Determination of Heat Exchanger Performance

The primary purpose of a heat exchanger is to achieve the required
transfer rate using the smallest possible transfer area and fluid
pressure drop. A large exchanger can mean unnecessary capital
outlay and high pressure drop means a reduced efficiency of the
plant considered overall. Generally, a smaller exchanger can be
produced by tinning surfaces to increase the overall heat transfer
coefficient. However, this leads to a higher fluid pressure drop,
and the best design is often a compromise between conflicting
requirements. In fact, a number of different designs for a given duty
may be acceptable.
The heat transfer requirement, Q, can be expressed in three ways:
(12.5)
(12.6)
( 12.7)
()m is a mean temperature difference between the fluids, and U A
and U L are mean coefficients, in kW/(m 2 K) and kW/(mK) or equiv-
alent units, applicable over the entire area A or length L of the ex-
changer. It is general practice to work in terms of the external
surface area of the tubes in heat exchanger design, and the overall
coefficient UA in terms of this area is given by equations (11.19)
for finned surfaces and ( 11.20) for plain surfaces.
12.2.1 Counter and Parallel Flow
If the mass flow rates and inlet and outlet temperatures are known, the
heat transfer Q will be known, but further details of the exchanger
cannot be specified until Om is known. Bm can be derived as follows:
Consider an incremental area of heat exchanger surface as shown

for either counter or parallel flow in Fig. 12.4. The heat transfer
over the area dA can be expressed in three ways as before, thus
(12.8)
(12.9)
(12.10)
Parallel flow
l A or L
0 dA or dL
Counter flow
Fig. 12.4. For the determination of logarithmic mean temperature difference
The temperature difference at the point in question is
and the increment in temperature difference is
dO = d(th - tc)
= dth- dtc (12.11)
HEAT EXCHANGERS 183
If d8 is expressed in terms of dQ from (12.9) and (12.10),
(12.12)
and
(12.13)
since dth is negative in both cases, and dtc is positive for parallel
flow and negative for counter flow; It follows that dO/dQ has
constant but different values for parallel and counter flow, and
therefore
(12.14)
with + for parallel flow and - for counter flow. Further, dQ from
equation (12.8) may be substituted in (12.12) and (12.13) to give
dO= - ( - 1- + -1- ) UdA

-
0 n1hCph - n1cCpc
This is integrated from 0 to A to give
(12.15)
The term in parentheses is now eliminated between (12.15) and (12.14)

to give
(12.16)
This result is clearly identical in form to equation (12.5) and it is

seen that
(12.17)
This is the required logarithmic mean temperature difference. It is

the same for counter and parallel flow, though 00 and 0; in terms
of values of th and tc are different as can be seen from Fig. 12.3.
EXAMPLE 12.1
0·2 kg/s of an alcohol is to be cooled from 75 to 35°C in a counter
flow heat exchanger. Cooling water enters the exchanger at l2°C
and at the rate of 0·16 kg/s. The convection coefficient between the
alcohol and the tube wall is 0·34 kW/(m 2K), and between the
tube wall and the water, 0·225 kW/(m 2K). The tubes may be assumed
thin. cP for the alcohol is 2· 52 kJ/(kg K) and for water is 4·187
kJ/(kgK).
Calculate the capacity ratio, the effectiveness, and the area of
the heat exchanger surface.
Solution. For the hot stream, alcohol,
mhcph = 0·2 x 2·52 = 0·504 kJ/(s K)
For the cold stream, water,
mccpc = 0·16 X 4·187 = 0·671 kJ/(s K)
From equation (12.2), C = mhcphfmccpc = 0·504/0·671 = 0·75. An

energy balance gives
0·2 X 2·52 X (75- 35) = 0·16 X 4·187 X (tc 2 - 12)

20·15 = 0·671tc2 - 8·05
tc 2 = 41·8°C
From equation (12.4),
E = tbt - th2 = 75 - 35 = 0·635

tbl - tel 75 - 12
The heat exchange area may be found from equation (12.5). To find
()0 = 35 - 12 = 23, ()j = 75- 41·8 = 33·2

() = 23 - 33·2 = -10·2 = 28 K
m In (23/33·2) -In 1·44
HEAT EXCHANGERS 185
Since the tubes are thin, r 1 = r 2 in (11.20), so UA is given by
1 1 1 1 1
-=--+--= - + -
uA halcobol hwater 0· 34 0· 225
uA = 0·1355 kW/(t;n 2 K)
Equation (13.5) gives 20·15 = UAAOm = 0·1355 x A x 28
A = 5·31 m 2
12.2.2 Cross Flow

Analysis of the cross-flow heat exchanger is more complicated owing to
temperature variation across the flow. This variation will depend on
whether the fluid is mixed or unmixed. A mixed fluid is free to move
across the flow direction; an unmixed fluid is constrained in parallel
flow passages. Thus, if an exchanger consisted of a bank of tubes
placed across a duct, the fluid in the duct would be mixed while the
fluid in the tubes would be unmixed.
Results of analyses of this type of exchanger are available as correc-
tions factors. 1•2 Equation (12.5) would become
Q = UAAFOm
where F is a factor to be obtained from the appropriate graph, and
Om is the mean temperature difference ( 12.17), calculated for counter
flow with the same inlet and outlet temperatures as for cross flow.
Figure 12.5 shows F for a cross-flow exchanger with one fluid
mixed and one fluid unmixed. In applying the factor F it does not
matter whether the hotter fluid is mixed or unmixed.
12.3 Heat Exchanger Transfer Units

One would now expect to be able to go ahead and design a heat
exchanger, using equations (12.5) to (12.7) and information from
earlier chapters to evaluate U A for the particular configuration
in mind. However, U A cannot be determined until something is
known of the tube sizes and velocities of flow, and the method of
procedure from theory so far developed can be extremely involved
and iterative. For example, supposing the tube sizes, length and
UA were decided upon, in order to check the design performance
the value of Q and outlet temperatures of the fluids must be regarded
as unknowns and equations (12.5) to (12.7) cannot be solved directly
'"- 1·0
~ 0·9
0 "' 1'\
.l! 0·8
.B=o-7
1\
~
\81 "!i i\~ ~
~ 0·6 II
~
0
u 0·5 6 . \ ll
0 0·1 U· ij·j ·4 ur:> U·6 0·7 0·8 0·9 1·0
t,2- t,l
t.l - ttl
Z = tsl - ts2 = (mc.,)tube
t,2 - ttl (mcp)shell
Fig. 12.5. Logarithmic temperature difference correction factor for cross flow,
one fluid mixed, one fluid unmixed. From R. A. Bowman, A. E. Mueller, and
W. M. Nagle, Trans. ASME, Vol. 62, p. 283 (1940). By permission of the American
Society of Mechanical Engineers.
for Q, tc 2 , and th 2 , because of the logarithmic form of Om. The

approach using transfer units is very useful from this point of view.
The method was developed by Kays and London. 3
The effectiveness E, and capacity ratio C of a heat exchanger
have already been defined. These quantities will now be used in
conjunction with a new term, Number of Transfer Units, NTU,
to determine heat exchanger performance. As with E and C, the
definition of NTU depends on the relative magnitudes ofthe thermal
capacities of the fluid stream. Thus,
NTU = UAA' W hen mhcph > mcCpc (12.18)

mcCpc
and,
(12.19)
Thus the denominator is always the smaller thermal capacity. The

performance of heat exchangers will now be examined using the
HEAT EXCHANGERS 187
definitions of C, E, and NTU in equations (12.1) to (12.4) and (12.18)
and (12.19).
12.3.1 Counter Flow Exchanger
Let mhcph be assumed the smaller quantity, then the definitions of
NTU, C, and E are
NTU = U AA, C = mbcph E = thl - th2
mhcph m.cpc, thl - tel
Equations (12.9) and (12.10) for counter flow (where temperature

increments are negative) give
(12.20)
Now, dO = d(th - t.) = dth - dt.,

and mhcph(dth - dt.) = dt.(m.cP• - mbcph)
using equation (12.20). Again, using (12.20), dtc may be eliminated
to give
= -dQ(l- C)
Using equation (12.8) to eliminate dQ gives

UAdAO
dth - dt. = - (1 - C)
mhcph
dO
0
Integrating:
= -NTU(l- C)
The left-hand side of this equation may be manipulated as follows:
thl - tel thl - tel - (thl - tb2)

thl - tel thl - tel - (tel - tel)
= 1-£ =e-NTU(l-C)
1- CE
from the right-hand side, above. This final result is now rearranged
to give
1 _ e-NTU(l-C)
E = 1 - C e NTU(l c> (12.21)
If mecpe had been assumed the smaller quantity, the same equation
would have been obtained, where E, NTU, and C would have
then been defined by the alternative expressions.
A relationship exists, then, between E, NTU, and C given by
equation (12.21). Using this result it is possible to determine outlet
temperatures tel and th 1 , and Q, the overall heat transfer for a given
design, without using a trial and error solution.
EXAMPLE 12.2
Determine the effectiveness and fluid outlet temperature of an oil
cooler handling 0·5 kg/s of oil at an inlet temperature of 130°C. The
mean specific heat is 2·22 k1/(kg K). ()-3 kg/s of water entering at 15°C
passes in counter flow at a rate of 0· 3 kg/s. The heat transfer surface
area is 2·4 m1 and the overall heat transfer coefficient is known to be
1·53 kW/(m1 K)
Solution. The thermal capacities are: oil, 0·5 x 2·22 = 1-11 kJ/(s K),
water, 0·3 x 4·182 = 1·255 kJ/(s K)
c = 1-11/1·255 = Q-885
HEAT EXCHANGERS 189
and,
1·53 X 2·4
NTU = = 3·31
1·11
Then,
1_ e-3·31(1-o 885)
1 - e- 0 . 38 0·316
1 - 0·885 e 0 ' 38 = 0·395 = 0· 8
1300 - th2
130° - 15°
(oil outlet)
By enthalpy balance
1·11 X (130- 38)
(tc2 - tel) = = 81·5 K
1· 255
(water outlet)
When U A is not known, this must be determined from either

equation (11.19) or (11.20), with the individual convection co-
efficients determined from the equation appropriate to the fluid,
flow geometry and type of flow, as given in earlier chapters. It is
convenient to use standard tube sizes to give a suitable value of
Re and number oftubes for the specified mass flow. Several attempts
may be necessary to achieve a suitable U A combined with a fluid
pressure loss which is acceptable.
12.3.2 Parallel Flow Exchanger

A similar analysis in parallel flow will yield the result
1 _ e-NTU(I +C)
E=------ (12.22)
!+C
Again this result is independent of which fluid stream has the
smaller thermal capacity, provided the appropriate definitions of
E, NTU, and Care used.
12.3.3 Limiting Values of C

It has already been noted that C = 0 in both condensing and boiling.
When this is so both equation (12.21) and (12.22) reduce to
E = 1- e-NTu (12.23)
Thus, the effectiveness is the same for both counter and parallel
flow.
The other limiting value is C = 1 for equal thermal capacities
and, in this case, for parallel flow equation (12.22) gives
1 _ e-2NTU
E=--2-- (12.24)
In the case of counter flow for C = 1 it is necessary to do a fresh

analysis from first principles since equation (12.21) becomes
indeterminate. For this case it is possible to write
and also
Also
E may be written as
E = NTU when C = 1 (12.25)

NTU + t'
12.3.4 Cross-Flow Exchanger
Convenient graphical plots of effectiveness as a function of NTU and

capacity ratio are available for cross flow. Figure 12.6 is for one fluid
HEAT EXCHANGERS 191
mixed and one fluid unmixed. When the capacity ratio of mixed to
unmixed fluid is greater than 1, the NTU is then based on (mcp) of the
unmixed fluid.
100
unmae _1
l 1.-- 1--
/ ".... ~ - --
- - - ~~ -
0·25
80
~ ;.::: ~ --
- - --
/ 0·75
~60 ~ ~ ,.,r- 1-33
I
.:: 40 ~
-
~
~ f..- I'Cmc.J,I,..d
(mc,),...laod
=1
u L
~
l.l.l 20 I
I
2 3 4 5
Number of transfer units, NTU = U AA/(mcp)mln
=f=:;~-
Fig. 12.6. Effectiveness vs. NTU for a cross-flow exchanger, one fluid mixed,
one fluid unmixed. From Compact Heat Exchangers, by W. M. Kays and A. L.
London, McGraw-Hill Book Company, Inc., New York (1958). Used by
permission of McGraw-Hill Book Company.
12.4 Heat Exchange in Cross Flow

Cross-flow heat exchangers may be analysed most readily by numerical
techniques, and Fig. 12.7 shows a typical model for consideration. For
the nodal element, one fluid enters a control volume at t 1 behind the
dividing wall at tM, and the second fluid, flowing upwards, enters the
control volume in front at t 2 • For transient analysis, a second subscript
is added to denote time, e.g., t 1, 0 , t 2 ,0 , tM,o. For the fluid flowing left
to right, the following equations may be written:
(12.26)
Transient: c)
(m• p 1 (t 3,0 - t 1,0 ) -- H 1 [tM,o - t1,0 + t3,0J (12.27)
2
or: • [ lto +t3o]

flT (mCp) 1 (t 3 ,o - lt,o) = Ht l:lT tM,o - ' 2 '
- (mCv)t(t3,1 - t3,o)
(12.28)
Mass within
--
f-....
I
I
I
tl J_...-
---- lc·"-.
(mcph
.....
t4
tI
~
1
t3
M
(2
Fig. 12. 7. ThertrUJI model for cross-flow heat exchange.

HEAT EXCHANGERS 193
The second transient equation is more accurate and complete. The

additional term, which is the increase in stored energy in the control
volume, is very small for gas to gas heat exchange, and is eliminated in
equation (12.27). Equation (12.28) when re-arranged may be expressed
as: increase of stored energy = heat flow across wall of control volume
+enthalpy flow in~ enthalpy flow out. In practice, equation (12.28) is
not often used. In equations (12.27) and (12.28), H 1 = h 1 A 1 =con-
vection coefficient x area of wall, for fluid 1; and similarly for fluid 2,
H 2 = h 2 A 2 • In equation (12.26) where the wall temperature is not of
importance, the overall thermal resistance between fluids is given by:
Hence the overall heat transfer coefficient is
Introducing the groups:
Equation (12.27) becomes, for the unknown t 3 , 0 :
t _ [ 2NTUt
3' 0 - 2 + NTU1
Jt M,o
+ [22 +~NTU
NTU
1]
ft,o
(12.29)
1
A similar equation for t 4 , 0 would be:
J 2 J
t4,o =
[22+NTU
2
NTUz tM,o +
[22 ~NTU
+ NTUz t2,o
(12.30)
In order to calculate t 3 and t 4 after the next time step, an equation for
a new tM must be obtained. Thus:
(12.31)
and hence
(12.32)
where (mCp)M is the product of mass and specific heat of the metal
element. For steady state, an enthalpy balance holds:
(12.33)
From this equation, and equation (12.26), the following relations for
t 3 and t 4 are obtained:
t =[ 2+NTU+CxNTU
3
2NTU ]t 2
+ [1 - 2NTU
2+NTU+CxNTU
]t 1
(12.34)
[ 2xCxNTU
t 4 = 2+NTU+CxNTU t 1 +
J c-
f, 2xCxNTU
2+NTU+CxNTU tz
J
(12.35)
where C = (m Cp) 1 /(m Cp)z and NTU = H/(m Cp) 1
The above equations may be used in the analysis of transient and
steady state behaviour of cross-flow heat exchangers. As shown in Fig.
12.7 outlet temperatures from one node become the inlet temperatures
of the next node, so that the field of nodes may be solved in simple
computer routines. Examples are given in Ref. 4. The equations to be
used in any analysis are (12.29), (12.30), (12.32), (12.34), and (12.35).
In each case the last term of the equation must remain positive for the
solution to be stable. This sets a limit on the values of the variables
occurring.
12.4.1 Rotary Regenerators

Rotary regenerators or thermal wheels are being used increasingly in
energy conservation measures, and the steady state equations (12.26)
and (12.33) above may be applied to their analysis. Figure 12.8 shows
HEAT EXCHANGERS 195
an element of gas flow in heat exchange with an element of matrix flow.
Results corresponding to equations (12.34) and (12.35) are obtained:
t2 =[ 2NTU1
2 + NTU1 + C1 x NTU1
JrA+[l- 2NTU1
2 +NTU1 + C1 x NTU1
Jr 1
(12.36)
tB = [
2xC 1 xNTU1
2 + NTU1 + C1 x NTU1
~
t1 + 1 -
2xC1 xNTU1J t
2 +NTU 1 + C1 xNTU1 A
J
(12.37)
!z
./
/ /
/.
,::· /
/
j', /
I ·, -y· / /
/
/
/
i i /
/
/
I /
(mcp)l 1.::" /
gas flow '- , '- . I / / ...- / t~
'- !..- (mcp)M
matrix flow
Fig. 12.8. Cro~~·flow heat exchange in a regenerator.
For the element shown, the gas stream entering at t 1 is in heat exchange
with a moving solid matrix stream entering at t A• and having a thermal
capacity rate of (m Cp)M. In the two equations above:
Later in the cycle, the matrix is in heat exchange with a second gas
stream, and similar equations may be written, including the terms:
Usually
NTU2 = NTU 1 and C2 = C 1
The elements of the form considered in Fig. 12.8 are pieced together
as shown diagrammatically in Fig. 12.9. Imagine Fig. 12.9 as a fixed
grid or series of control volumes with the fluids and matrix flowing
through in the direction of the arrows. The actual thermal wheel is
shown schematically in Fig. 12.1 0. The matrix passes through sealing
zones between the two gas streams, and it is assumed that there is no
change in the matrix temperature across this zone. However, as the
matrix passes out of the second gas stream, the matrix temperature at
all nodes must equal those values at inlet to the first gas stream, for
steady state conditions. This means that for a correct solution an itera-
tive procedure is necessary, since the matrix inlet temperature tA, tK,
tL and tM must be assumed or guessed at the start of the analysis.
Values obtained at outlet in the first iteration are inserted back in the
inlet and the process repeated until a close convergence is obtained.
Heat transfer between the gas streams is obtained by summing up the
enthalpy loss or gain of each fluid stream, and average gas stream outlet
temperatures may be found to give the regenerator effectiveness. For
steady state conditions the cooling of the matrix by the cold stream
must always equal the heating by the hot stream, and hence the capacity
ratio of the two gas streams is usually (though not necessarily) unity. In
this case it follows that C 1 = C2 and NTU 1 = NTU2 •
197
!
HEAT EXCHANGERS
,..,, ), t (mcph
I 14 I 15 I 16 I 17 I 18 I 18 118 I 18
110 Ill I 12 I 13
I6 17 I8 t9
12 t3 14 t5
IB tc lo IF IG IH lj lA
Fig. 12. 9. Nodal system for the Thermal Wheel.
tc
Fig. 12.1 0. Diagrammatic representation for the Thermal Wheel.

EXAMPLE 12.3
A Thermal Wheel may be analysed by means of the following program.

The input data are the initial stream temperatures, capacity ratios, and
number of Transfer Units for the whole wheel. For this, H is the
product of heat transfer coefficient and heat transfer area for that part
of the wheel exposed to one or other gas stream (assumed equal). If
desired, in a modified program H could be calculated for different
matrix hole sizes and hence surface areas for a specified wheel size.
Similarly C1 and C2 could be varied, depending on different wheel
sizes, rotating at different speeds. Results are given for Effectiveness as
a function of Transfer Units for C 1 = C2 for values of 0·1, 0·5 and 1·0.
The form of the curves are seen to be similar to Fig. 12.6 for a cross-
flow heat exchanger.
Basic Program Listing
20C THERMAL WHEEL ANALYSIS

30 DIH A(10o10),T(10o10loAAC1Uo10) 1 TT(10 1 10)
40 r1·~ou.o
SO T2•1UO.O
60 C1•0.1
70 CZ•O. 1
80 TN•S.O
90 TN1•TN/10.0
100 TNZ•TNI10.0
110 f1•Z.O•TN1•C1f(TH1•C1+2.0+TU1)
120 F2•1.0-f1
130 F3•2.0•TN2•C2f(TNZ•CZ+2.0+TN2)
140 F4•1.0-F3
150 f5•Z.O•TH1f(TN1+Z.O+TN1•C1l
160 f6•1.0-FS
170 F7•2.0•TH2f(TH2+2.0+TH2*C2)
180 f8s1.0-F7
190 IF(F1>1.0)GO TO 740
200 lf(F3>1.0)GO TO 740
210 IFCF5>1.0)GO TO 740
220 IFCF7>1.0lGO TO 740
225 LPRIUT " T1 ~ ";LPRINT USIUG "~UH.RN";rt;
210 LPMlHT " TZ • ";LPHlNT USING ''nNU.#N":rz;
215 L~RJHT " C1 • ";LPHINT USlHG "H.D#N";ct;
240 LPHliH " C2 • ":LPIIII.T USIIIG "ll.lllllr':cz;
250 LPRIUT " NTU a ":LPRJHT USING ''R.##~'';Tu
260 FOR JXa1 TO 10
270 AA(J~o10l•T1:HEXT J~
280 ITER:•O
290 A(1,1J•F1•T1+F2•AA(10o10)
300 T(1,1J•FS•AAC1Uo1Ul+f6•T1
310 FOR J%•2 TO 10
320 A(J%o1)•FI•T(J%-Io1)+f2*AACIO-J%+1oi0J
330 T<JXo1l•FS•AA(10-J%o10)+F6*T(J~-1,1)
llS IIEXT J<:
340 FO~ 1::•2 TO 10
350 A(1,t::J•FI•T1+f2•A<Iol%-1)
360 T(1ol%)•f5•A<Iol%-1)+F6•TI
365 NEXT IX
370 FOR J%•2 TO 10
380 FOR 1%•2 TO 10
390 A<JXolX>•F1•T<JJ-1,1%l+FZ•A(J%ol:-tl
HEAT EXCHANGERS 199
400 TIJX,J%l~f5•AIJ4,JX-1J+F6•TIJX-1,1Xl
404 UEX T IX
405 IHXT J%
410 AA(1,1l~F3•TZ+F4•AI10,10l
420 TT(1,1l~F7•AI10,10l+F8•T2
430 FOR n ~z TO 10
440 AAIJ%,1l~F3•TTIJ%-1,1)+f4•AI10-J4+1,10l
4SO TTIJ~,1l~F7•AI1u-J~+1,10l+f3•TTIJ~-1,1l
4S5 llEXT J4
460 fOR 14•2 TO 10
470 AA(1,J:l•F3•T2+F4•AA(1,1~-1)
480 TT(1,J~l=F7•AA(1,J%-1)+f~•TZ
485 NEXT I~
490 FOR J%~7. TO 10
500 FOR IX•Z TO 10
510 AAIJ%,Jil~FJ•TTIJ%-1,JXl+F4•AAIJX,J~-1)
520 TTIJ%,1~)•f7•AAIJ%,1X-1l+FB•TTIJ%-1,1%l
524 IIEXT n
525 :lEXT J:':
530 X•O
540 FOR J:•1,TO 10
5SO X•X+AA(JX,10l;NEXT J%
560 Z•AOS(X-Y)
570 lf(l<0.001lGO TO 610
580 Y•X
590 ITERX•ITER%+1
600 GO TO 290
610 TOUTZ•O.O
620 FOR 1%•1 TO 10
630 TUUTZ•TOUTl+TT(10,1%)/10.0
635 NEXT IX
640 Eff•(TOUT2-T2)/(T1-T2l
650 LPRINT " Gas Strea• Outlet TeMperatures:"
660 LPRINT USING •nNNN.MH";T(1Q,J%) FOR IX•1 TO 10
670 LP~INT " Air Strea• Outlet Te•perdtorrs:"
680 LPRINT U~ING •nnnn.gn•;Tr<10,JXl FOR 1%•1 TO 10
690 LPRINT '' Effectiveness ~ ";LPRINT USIUG"I.#II";EFF;
700 LPRI~T '' for '':LP~IUT U~lNG"nU.N#I":TII:LPRlUT'' Trdnsfer Units"
710 LPRINT ''lluYber of Iterations to Conver~e • ";
720 LPRIUT USIJ4ti ''U~M":tTEit
730 GO TO 7?0
740 LPRINT " Check Data - C•lculMtions Unstable:"
750 LPI«INT " f1 a ";LPNIIIT IJSlriG "##.d#";Fl;
760 LPRIUT •• FJ • ":LPHINT US!UG "nR.MH":F3:
770 LPR!UT " F5 • ":LPRINT USliJG "##,#N";rs;
780 LP:liur •• F7 =- ":LPRtNT ustuG "nn.M"":F7
790 STOP
'i-;lil
1.0 C=
0.1
;:;s 0.5
6-t····~-l·~
0.8 1.0
.L.,~~~~
E
0.6
:~·.·.·~~~ f1T
0.4 !), '
'
L
t
1-
j
)..
+-
+-
,='·
,_,_
tr +·
0.2
i --1=
0 5 10 15 20
NTU
Fig. Example 12.3. Thermal Wheel characteristics.
12.5 Plate Heat Exchangers
The plate type of heat exchanger is basically of the in-line type, but the
construction is very different from the conventional shell and tube
concept. A plate heat exchanger consists of a frame in which a
number of heat-transfer plates are supported and clamped between
a header and a follower. Each plate has four ports and the edges of
the plates and ports are sealed by gaskets so that hot and cold
fluids flow in alternate passages formed between the plates. This
means the fluids flow in very thin streams having a high heat-transfer
Fig. 12.11. A typical flow diagram of a plate heat exchanger showing a two·
pass arrangement (diagram by courtesy of the A.P. V. Company Ltd).
area, and corrugations on the plates promote turbulence and very

high heat-transfer rates. Since the plates are usually arranged for
general counter-current flow, very close approach temperatures
are obtained. Figure 12.11 shows a typical flow diagram. Because
of these advantages, the plate heat exchanger is being used exten-
sively in an increasing number of industrial applications.
The performance of a plate heat exchanger may be expressed in
terms of equations (12.5) to (12 .7), but since the overall coefficient is
obtained from empirically determined charts, the characteristics are
expressed in terms of chosen parameters only. Thus, using equations
(12.5) and (12.7)
For a plate heat exchanger A is the product of n, the number of

plates, and a, the individual plate area, so
mhc ph(thl - th2) = U Ana ()m
HEAT EXCHANGERS 201
Fig. 12.12. A Paraflow type Rl45 plate heat exchanger, capable of accepting
up to 955 m 3 per hour at 1 (). 7 bar, and up to 130°C; plate size is 2122 X 849 mm
(photograph courtesy of the A.P. V. Company Ltd).
mhcph (thl - th2)

n = PN . (}
m
where PN is the plate number, U..1 a. For mhcph being the minimum
capacity rate, or for equal rates as defined previously, it is seen from
equations (12.19), (12.5) , and (12.7) that (th 1 - th 2 )/8m =NTU, the
number of transfer units, and hence
me
n=~ x NTU (12.38)
PN
The performance of a particular plate design can be expressed
graphically in terms of the plate number, the NTU value, and the
pressure drop plotted against the plate rate, or the mass flow rate
across a plate, see Fig. 12.13. Separate curves would exist for different
capacity ratios, and from such information for various plate designs,
the required unit for a particular duty can be selected. Certain
correction factors have to be introduced, on account of concurrency
and other effects which depend upon the particular plate arrange-
ment, and on account of uneven distribution along the plate pack
due to pressure losses along the ports. For exactness liquid proper-
ties have also to be considered, and separate relationships would
apply to laminar and transitional flow.
pressure
loss
plate rate
Fig. 12.13. Plate number, pressure loss and NTU characteristics of a plate
heat exchanger for turbulent flow.
12.6 Batch Heat Exchangers

In certain industries where a fluid to be heated is of limited amount, or
alternatively is of a very viscous nature, as for example in food process-
ing, the batch heat exchanger is used. The entire mass of fluid to be
heated is contained in a stirred vessel and is in thermal contact with a
constant temperature source of heat, such as a coiled pipe through
which steam is passing. The mass of fluid is essentially of uniform
HEAT EXCHANGERS 203
temperature at any given time, and the theory of Section 5.1 applies.
Figure 12.14 shows the temperature time curve for the heat exchange
process, and from equation (5.1), the temperature-time curve is given
by:
8 2 /8 1 = exp (- t/T) (12.39)
where t is the heating time between temperature differences of 8 1 and
8 2 and T is the time constant mCp/ hA, where m is the mass of heated
and stirred fluid, CP is the specific heat, his the convection coefficient
between the pipe and fluid, and A is the area of the heating surface
(e.g., A= rrdL for a pipe) in contact with the heated fluid.
Heating source temperature
Batch temperature
Time ~
Fig. 12.14. Temperature-time curves for batch heat exchanger.
PROBLEMS
1. A tubular heater of the counter flow type is used to heat 1·26 kg/s of
fuel oil of specific heat 3·14 kJ/(kg K) from 10° to 26·7°C. Heat is supplied by
means of 1·51 kgfs of water which enters the heater at 82°C.
(a) Derive an equation relating the temperatures of oil and water at any
section of the heater.
(b) Determine the necessary surface if the rate of heat transfer is 1-135
kW/(m 2 K). (Ans.: 1·013 m 2) (University College, London).
2. In a test on a steam condenser the rate of flow of cooling water was varied
whilst the condensation temperature was maintained constant. The following
results were obtained:
Overall heat transfer coefficient K, kW/(m 2 K) 2·7 2·98 3·39 3·59
Water velocity V, m/s 0·986 1·27 1·83 2-16
Assuming the surface coefficient on the water side to be proportional to
yo·&, determine from an appropriate graph, the mean value of the steam side
surface coefficient. The thickness of the metal wall is 0·122 em and thermal
conductivity of tube material 0·111 kW/(m K). (Ans.: 6·04 kW/(m 2 K.)
(University of Manchester).
3. A counter flow ·heat exchanger consists of a bundle of 20 mm diameter

tubes contained in a shell. Oil flowing in the tubes is cooled by water flowing
in the shell. The flow area within the tubes is 4·4 x 10- 3 m 2 • The flow of oil
is 2·5 kg/s; it enters at 65°C and leaves at 48°C. Water enters the shell at
2·0kg/s and at 15°C. Calculate the.area of tube surface and the effectiveness
of the exchanger. For the oil in the tubes take Nu~ = 0·023 (Re~) 0 " 8 (Pr) 0 " 33 ,
c = 2·15 kJ/(kg K), ll = 2·2 x 10-s Pas, p = 880kg/m 3 , k = 190 x 10- 6
kW/(m K); for water 1i = 1·2 kW/(m 2 K), cP = 4·19 kJ/(kg K). (Ans.: 2·61 m 2 ,
34 %.) (The City University).
4. (i) Define the term 'mean temperature difference' as applied to a heat

exchanger and show that, for a counter flow heat exchanger, it is given by
where ~tm is the mean temperature difference, ~t 1 is the temperature difference

between the two fluids at one end of the heat exchanger, and ~t 2 is the tem-
perature difference at the other end. State any necessary assumptions.
(ii) A tubular, counter flow oil cooler is to use a supply of cold water as
the cooling fluid. Using the following data, calculate the mean temperature
difference and the required surface area of the tubes.
Data: Oil Water
Entry temperature, oc 121 15·6
Exit temperature, oc 82·3
Mass flow rate, lrg/s 0·189 0·378
Specific heat, kJ/(kg K) 2·094 4·187
Mean overall coefficient of heat transfer, referred to outside surface of
tubes, Q-454 kW/(m 2 K~ (Ans: 80·0 K, Q-422 m 2 .) (Imperial College, London).
5. Two counter flow heat exchanger schemes are shown in the diagrams.
In each scheme it is required to cool a fluid from 1400 to 90°C using a counter
flow rate of water of 1·2 kgfs entering at 30° and leaving at 80°C. In scheme
(b) each unit takes half the flow of the fluid. The overall heat-transfer coeffi-
cient is 0·9 k Wf(m 2 K) in both cases. Calculate the total area of heat exchange
surface in each case, assuming a capacity ratio of 1. (Ans.: (a) 4·65 m 2 ,
(b) 4·83 m 2 .) (The City University).
HEAT EXCHANGERS 205
water 30"C
water 30°C
-----<---tll!llll!lllll!llllll!llllll water 80"C
fluid
90oC
(b)
6. An industrial fluid is cooled by oil in a parallel flow heat exchanger,

from 280° to 160°C while the oil enters at 64° and leaves at 124°C. Find the
minimum temperature to which the oil could be cooled in parallel flow
and also in counter flow for the same entry temperatures. Find the ratio
of heat exchange area in parallel flow to that in counter flow, for an outlet
fluid temperature of 160°C (Ans.: 136°C, 64°C, 1·23 to 1.)
7. An oil cooler consists of a straight tube, of inside diameter 1·27 em, wall
thickness 0·127 em enclosed within a pipe and concentric with it. The external
surface of the pipe is well lagged.
Oil flow through the tube at the rate of 0·063 kg/s and cooling water flows
in the annulus between the tube and the pipe at the rate of 0·0756 kg/s and
in the direction opposite to that of the oil. The oil enters the tube at 177°C
and is cooled to 65·5oC. The cooling water enters at 10°C.
Estimate the length of tube required. given that the heat transfer co-
efficient from oil to tube surface is 1·7 kW/(m 2 K~ and that from the surface
to water is 3·97 kW/(m 2 K). Neglect the temperature drop across the tube
wall The specific heat of the oil is 1·675 kJ/(kg K~ (Ans.: 2·67 m.) (University
of London).
8. A tank contains 272 kg of oil which is stirred so that its temperature is

uniform. The oil is heated by an immersed coil of pipe 2·54 em diameter in
which steam condenses at 149°C. The oil, of specific heat 1·675 kJ/(kg K) is
to be heated from 32·2° to 121oc in 1 hour. Calculate the length of pipe in the
coil if the surface coefficient is 0·653 k W/(m 2 K~ (Ans.: 3·47 m.)
9. Explain briefly what is meant by the term 'surface or film coefficient' in

heat transfer considerations.
A counter-flow heat exchanger having an overall heat transfer coefficient
of0·114 kW/(m 2 K) is used to heat to 329°C the air entering the combustion
chamber of a gas turbine cycle. The pressure ratio of the cycle is 5: 1 and the
heating fluid is the exhaust from the turbine which expands the gas from
65o•c with an isentropic efficiency of 82 per cent. If the air conditions initially
are 1·013 bar and 21 •c and the isentropic efficiency of the compressor
is 80 per cent, calculate the area of heat exchanger for a total fluid mass flow
of 22·7 kgjs.
Assume a logarithmic mean temperature difference and constant specific
heat of 1·0 for the air and 1·09 kJ/(kg K) for the products. y = 1·4 for air
and products. (Ans.: 424m 2 .) (University of Manchester).
10. Define the terms 'effectiveness' and 'number of transfer units' as applied
to heat exchangers stating any assumptions involved. Obtain a relationship
between effectiveness and number of transfer units for a counter-current heat
exchanger and plot this relationship when the ratio of the stream heat
capacities is 0·5.
20·15 kgjs of an oil fraction at a temperature of 121 oc is to be cooled in a
simple counter-current heat exchanger using 5·04 kg/s of water initially at
1o•c. The exchanger contains 200 tubes each 4·87 m long and 1·97 em outside
diameter; the resulting heat transfer coefficient referred to the outside tube
area is 0·34 kW/(m 2 K). If the specific heat of the oil is 2·094 kJ/(kg K)
calculate the exit temperature of the oil. (Ans.: 90·8°C.) (University of Leeds).
11. Write a program in BASIC to consider heat exchange in steady state

for a single pass unmixed flow cross-flow heat exchanger.
Determine the Effectiveness for Capaciiy Ratios of value 0·25, 0·5,
0·75 and 1·0 for NTU values up to 5, and compare with analytical
results given by Kays and London. 3
12. Write a program in BASIC to consider transient response of a single

pass unmixed flow cross-flow heat exchanger. Consider two versions of
the program, based on equations (12.27) and (12.28). Obtain steady
state effectiveness results from the previous question for a Capacity
Ratio of 1·0, for NTU values of 1 and 5. Then determine and plot the
percentage of steady state effectiveness achieved at given real times up
to 1200 seconds, for fluid to wall capacity ratios of 5·0 and 2·5 for
NTU values of I and 5. Study the effect of using the two different
equations, (12.27) and (12.28). The result of this analysis is given in
Ref. 4.
REFERENCES
I. Smith, D. M. Engineering, Vol. 138, 479, 606 (1934).
2. Bowman, R. A., Mueller, A. C., and Nagle, W. M. Trans. ASME, Vol. 62,
283 (1940).
3. Kays. W. M., and London, A. L. Compact Heat Exchangers, McGraw-Hill
Book Company, Inc .. New York (1964).
HEAT EXCHANGERS 207
4. Simonson, J. R. 'Transient and steady state analysis of cross flow
heat exchangers by programs in Fortran', Trans. Inst. Chern. Engrs.,
Vol. 55,53-58 (1977).
13
The laws of black- and grey-body
radiation
The processes of heat transfer considered so far have been intimately

related to the nature of the material medium, the presence of solid-
fluid interfaces, and the presence of fluid motion. Energy transfer
has been observed to take place only in the direction of a negative
temperature gradient, and at a rate which depends directly on the
magnitude of that gradient.
It is now necessary to consider the third mode of heat transfer
which is characteristically different from conduction and convec-
tion. Radiation occurs most freely in a vacuum, it is freely trans-
mitted in air (though partially absorbed by other gases) and, in
general, is partially reflected and partially absorbed by solids.
Transmission of radiation, which can occur in solids as well as
fluids, is an interesting phenomenon because it can occur through
a cold non-absorbing medium between two other hotter bodies.
Thus the surface of the earth receives energy direct by radiation from
the sun, even though the atmosphere at high altitude is extremely
cold. Similarly, the glass of a green house is colder than the contents
and radiant energy does not stop there, it is transmitted to the
warmer absorbing surfaces inside. Radiation is also significantly
different from conduction and convection in that the temperature
level is a controlling factor. In furnaces and combustion chambers,
radiation is the predominating mechanism of heat transfer.
As already mentioned in Chapter 1, radiant energy is but part of
the entire spectrum of electromagnetic radiation. All radiation
travels at the speed of light and, consequently, longer wave-lengths
correspond to lower frequencies, and shorter wave-lengths to higher
frequencies. The entire spectrum of electromagnetic radiation extends
from about 10- 4 angstrom units (10- 14 metres), the wavelength
208
THE LAWS OF BLACK- AND GREY-BODY RADIATION 209
region of cosmic rays, up to about 20,000 metres, in the region

of Hertzian or electric waves. The wave-length region generally
associated with thermal radiation is 10 3 -106 angstrom units, which
includes some ultra-violet, all the visible, and some infra-red radia-
tion. Figure 13.2 shows part of the spectrum of electromagnetic
radiation.
Since radiation energy exchange depends on the rates at which
energy is emitted by one body and absorbed by another, it is neces-
sary to establish definitions relating to these characteristics of
surfaces. Further, not all of the energy emitted by one body may
necessarily fall on the surface of another due to their geometric
arrangement, and this too must be investigated. This then forms
the general approach by which engineers may consider radiant
energy exchange.
13.1 Absorption and Reflection of Radiant Energy

Three possibilities may follow the incidence of radiation on the
surface of a body. Some may be transmitted through the body leaving
it unaltered. Some may be absorbed on the surface, resulting in an
increase in temperature of the body at the surface. The remainder
will have been reflected. This can take place in two ways, either as
specular reflection where the angle of reflection is equal to the angle
of incidence, or as diffuse reflection where the reflected energy leaves
in all directions from the surface. Thus polished surfaces tend to be
specular and rough surfaces diffuse.
The percentage of incident energy absorbed by a surface is defined
as IX, the absorptivity; the percentage reflected is p, the reflectivity,
and the percentage transmitted is r, the transmissivity. Thus it must
follow that
1X+p+r=1 (13.1)
Energy absorbed on the surface is, in fact, absorbed in a finite
thickness of material, and if the body is very thin less absorption
and more transmission may take place. It will be assumed that
'thick' bodies only will be considered, for which r = 0. Hence
IX+ p = 1 (13.2)
In engineering applications of radiation, there will generally be a
gas separating solid bodies, and often this gas is air which may be
assumed to have no absorptivity or reflectivity, so r == 1. Combus-
tion gases containing carbon dioxide and water vapour behave
very differently, however, and an elementary treatment of non-
luminous gas radiation appears later in this chapter.
13.2 Emission, Radiosity, and Irradiation

To be consistent with previous nomenclature, Q is the energy
emitted by a surface in heat units per unit time. This energy emission
results from the surface temperature and the nature of the surface.
However, Q may not be the total energy leaving that surface, there
may also be some reflected incident energy. Thus J is defined as the
Radiosity, which is the total radiant energy leaving the surface, in
unit time. Similarly, G is defined as the Irradiation which is total
incident energy on a surface, some of which may be emission and
some reflection from elsewhere.
If G is the incident energy, pG ~ill be reflected. Thus
J = Q + pG (13.3)
13.3 Black and Non-black Bodies

All materials have values of ex and p between 0 and 1. However, it is
useful and important to imagine a material for which ex = 1 and
p = 0. A body composed of this material is known as a black body;
it absorbs all incident energy upon it and reflects none. For real
materials the highest values of ex are around 0·97. Artificial surfaces
may be arranged in practice which are virtually black. Consider
Fig. 13 .1. The hollow enclosure has an inside surface of high absorp-
tivity. Incident energy passes through the small opening and is
Fig. 13.1. Artificial black-body surface.
absorbed on the inside surface. However, some is reflected, but

most of this is absorbed on a second incidence. Again, a small
fraction is reflected. After a number of such reflections the amount
unabsorbed is exceedingly small and very little of the original
incident energy is reflected back out of the opening. The area of
the opening may thus be regarded as black.
The work of Stefan and Boltzmann led to the law named after
them which gives the emission of radiant energy from a black
body. Thus
(13.4)
is the Stefan-Boltzmann law for plack-body radiation. T is the
absolute temperature and (j is the Stefan-Boltzmann constant and
has the value 56·7 x 10- 12 kW/(m 2 K 4 ). A derivation of this law
is given by Jakob. 1
Black-body radiation consists of emission over the entire range
of wave-length. Most of the energy is concentrated in the wave-
length range already mentioned. The point to note is that the
energy is not distributed uniformly over this range. Thus qb;. may
be defined as the monochromatic emittance, the energy emitted
per unit area at the wave-length A., for a black body. It must follow
that
qb = s= qb .. d.A. = (jT
4 (13.5)
The variation of qb;. with wavelength was established by Planck 2 in
his quantum theory of electromagnetic radiation, thus
q -
c 1A. -5
b.l. - exp (C 2 jA.T)- I
where ). = wavelength, J.lm, T = absolute temperature, C 1 = 3·743
x 105 kWJ.1 4 fm 2 , C 2 = 1·439 x 104 J.1K. The form of the varia-
tion of qbx is shown in Fig. 13.2, and it is seen that there is a peak
value of qb.l. which occurs at a wave length which is related to the
absolute temperature by Wien's displacement law:
).max T = 2897·6!-1 K
Real materials that are not black will have monochromatic
emittances that are different from qbA.• and hence it is useful to
define a monochromatic emissivity e;. by the equation
q;, = e;.qb'-
or
(13.6)
1·0 2·0 4-Q 6·0 8·0

()-3- Qo7JI visible range oflight
~ wavelength A
u/v light --I I-- infrared light (to 10-3m ) -
Fig. 13 .2. Variation of black-body emissive power Qb;>.. with wave-length and
temperature
The black and non-black emittances which give e;. are measured at
the same temperature. In general, e;. is a function of wave-length,
temperature and direction. Real surfaces often exhibit directional
variation in emissive power, thus non-electrically conducting
materials emit more in the normal direction whereas for conducting
materials often the reverse is true. For practical calculations, quoted
emissivities are total hemispherical values. Most real materials
exhibit some variation in e;. with wave-length. These are known as
selective emitters. However, there is a second type of ideal surface,
known as a grey surface, where the emissivity is constant with
wave-length. Some real materials approximate closely to this ideal,
but the concept reduces calculations to the extent that it is worth-
while to accept the error introduced in exchange for the simplifica-
Fig. 13.3. Comparison of the emission of black, grey, and selective emitting
surfaces; ex= q-,.,/qbx.
tion afforded. Both grey-body and selective emission are shown in

Fig. 13.3. It must follow that for a grey body
(13.7)
The value of e used for a grey body is generally a function of the

temperature of the surface, but again a simplifying assumption
enables a suitable constant value to be used, irrespective of tempera-
ture, provided the range is not too large. Values of e for real materials,
and the temperatures at which they are valid, are given in Table A.7
(see page 254).
It is now apparent that materials exist for which a < 1 and also
for which the emission is not equal to the black-body emission. By
means of Kirchhoff's law the relationship between a and e may be
established.
13.4 Kirchhoff's Law 3

Consider a small black body of area A 1 completely enclosed by a
larger body with an internal black surface area A 2 , as in Fig. 13.4.
Both surfaces are at the same temperature. The small body will emit
at the rate A 1 aT 4 and must also absorb energy at the same rate
otherwise the temperature of the body will change. The concave
surface A 2 will emit A 2 aT 4 , but only A 1 aT 4 of this is incident upon,
Fig. 13.4. To demonstrate Kirchhoff's law.
and absorbed by, A 1 . If F is the fraction of energy leaving A 2

which is absorbed by A 1 , then
F- A1uT4- AI (13.8)
- A 2 uT 4 - A 2
The remainder of the energy emitted by A 2 will be re-absorbed by
A 2 as it will miss A 1 •
Now consider what happens when the black body of area A 1 is re-
placed by a grey body of identical dimensions, with an absorptivity
of ac and an emissivity of e, the temperature throughout remaining
at T. Since there is again thermal equilibrium the energy actually
absorbed on A 1 must equal the energy emitted by A 1 • The energy
emitted by A 2 is A 2uT4 and this is also the radiosity of A 2 since
nothing is reflected by A 2 • Of this, only FA 2 u T 4 will fall ·on A 1
and only acF A 2 uT4 will be absorbed. A 1 will itself emit eA 1 uT 4
and this must equal the energy absorbed.
But
from (13.8)
Therefore
e=oc {13.9)
Thus, Kirchhoffs law, as stated by equation (13.9), says that the

absorptivity is equal to the emissivity at any given temperature.
It follows that for a black body for which oc = 1, that e = 1 and,
consequently, e < 1 for a grey body. Since it is possible to use a
suitable value of e for grey bodies over a temperature range, the
value of a over that range is the same. This does not hold for real
materials that are true selective emitters when the temperature
difference is very large, because the bulk of the energy absorbed
by either body is in a very different wave-length region than the
energy emitted by that body.
13 .S Intensity of Radiation
The radiation from a unit area of black body is qb = uT4 • For
diffuse radiation from a small flat area of black surface dA, the
entire emittance Qb must pass through a hemispherical surface
surrounding the emitting area. It is necessary to consider the
distribution of radiant energy per unit area over the spherical
surface, before calculations can be made of radiation exchanges.
The intensity of black-body radiation, /, is the radiation emitted
per unit time and unit solid angle subtended at the source, and per
unit area of emitting surface normal to the mean direction in space,
and may be expressed as
(13.10)
This is shown in Fig. 13.5. dA 2 /r 2 is the solid angle subtended by

dA 2 • The radiant energy per unit area at the hemispherical surface is
the radiant flux dQb/dA 2 • The surface of dA 1 has been specified as
diffuse, thus Lambert's Iaw4 states that I is constant in the hemi-
spherical space above dA 1 . From the above definition of 1 it thus
Fig. 13.5. To evaluate intensity of radiation.

follows that dQ.JdA 2 will have a maximum value at any given r

when cjJ = 0, i.e., when dA 2 is on the normal to dA 1 . Further,
dQ.,/dA 2 is zero when cjJ = 90° and, in addition, dQb/dA 2 will vary
inversely as r 2 • In general,
where the suffix n implies on the normal to dA 1 .

For Lambert's law to be true, I for a black surface must depend
on the absolute temperature only. From equation (13.1 0),
(13.11)
and from Fig. 13.6 it is seen that dA 2 = rdcp (r sin cfJ d8) =
r 2 sin cjJ dc/J dO. Hence
Fig. 13.6. Detail from Fig. 13.5.
dQb = I dA 1 sin cjJ cos cjJ dc/J dO
The total radiation passing through the hemispherical surface is
then
f"'="'2s6=2"
Qb = I dA 1 tJ>=O B=o sin cos dcjJ dO cjJ c/>
= 2nl dA f"'= sin cos de/>

t/>=1</2
1 0 cjJ cjJ
= nl dA 1
I = qb = uT4 (13.12)
n n
13.6 Radiation Exchange between Black Surfaces
It is now possible to consider the radiation exchange between two
arbitrarily disposed black surfaces of area A 1 and A 2 , and at
temperatures T 1 and T2 • Small elements of each surface dA 1 and
dA 2 are considered as shown in Fig. 13.7. They are distance r
normal/
Fig. 13. 7. Arbitrarily disposed black surfaces exchanging radiation.
apart, and the line joining their centres makes angles </l 1 and </l 2
to their normals. Each element of area subtends a solid angle at the
centre of the other; these are dw 1 subtended at dA 1 by dA 2 , and
dw 2 subtended at dA 2 by dA 1 . The solid angles are given by:
d w 2 = dA 1 cos </l1
an d
r2
From equation ~13.11) the radiant energy emitted by dA 1 that
impinges on dA 2 is given by:
(13.13)
Since both surfaces are black this energy is absorbed by dA 2 • A
similar quantity of energy is also radiated by dA 2 and absorbed by
dA 1 expressed as
(13.14)
The net exchange is

dQb(l-2)- dQb(2-1) = dQb(12)
and
Equation ( 13 .12) is now used to give the final result
dQ _ adA 1 dA 2 cos<P1 cos<jl2(T 4 T4)

b(12J- nr 2 1- 2 (13.15)
The total radiation exchange between the two surfaces A 1 and A 2

amounts to a summation of the net energy exchange between dA 1
and all elements of area A 2 , and the net exchange between all other
elements of A 1 and all elements of A 2 • From equation (13.13), the
fi
total energy radiated by A 1 that falls on A 2 is given by
Q b(l-2) -_ I cos<jl 1cos<jl2 dA 1dA 2

I
r2
A, A2
But the total energy radiated by A 1 is

QbllJ = A 1 aTi
f
Hence the fraction of energy radiated by A 1 that falls on A 2 is
Qb( 1- 21 =_1_f cos<jl 1cos</l2dA 1dA2

Qb(l) A I A, A> n~
= F1-2 (13 .16)

F 1 _ 2 is known as the geometric configuration factor of A 1 with
respect to A 2. Thus the energy radiated by A 1 that falls on A 2 may
be expressed as
(13.17)
Similarly, from equation (13.14) the total energy radiated by A 2
ff
that falls on A 1 is given by
4 COS </J 1 COS </J 2 dA 1 dA 2

Qb(2-IJ=aT2 nr2
A, A2
l
and the total energy radiated by A 2 is A 2 aTi, so that
Qb(2-o=__!_J cos¢ 1 cos¢ 2 dA 1 dA 2

Qb(2) A 2 A, Az v 2
= Fz-1 (13.18)
and
(13.19)
From equations (13.16) and (13.18) it is seen that F 1 _ 2 andF2 _ 1 are
simply related:
(13.20)
The net radiation exchange from equations (13 .17) and ( 13 .19) can
be expressed in terms of either configuration factor, thus
Qb(12) = F1-2A1a(Ti- Ti)
(13.21)
It is necessary to know or to be able to calculate configuration
factors before black-body radiation exchanges can be determined.
Only a few results will be considered here, and the reader is referred
elsewhere for further information on this subject. 1 •5 · 6
13.6.1 Examples of the Black-Body Geometric Configuration Factor
(i) Cases where F 1 _ 2 = l. The simplest case is when surface A 1 is
entirely convex and is completely enclosed by A 2 . Then F 1 _ 2 must
be 1, since all the energy radiated by A 1 must fall on A 2 . It follows
also that F2 _ 1 is Ad A 2 . In this case. the net black-body radiation
exchange is
(13.22)
Another simple example is when surfaces A 1 and A 2 are parallel
and large, and radiation occurs across the gap between them, so
that in this case A 1 = A 2 and all radiation emitted by one falls on
the other if edge effects are neglected. Hence,
Fl-2 = F2-t = 1
Concentric surfaces may be included if the gap between them is
small so that little error is introduced by the small difference
between the area of A 1 and A 2 . The net radiation exchange is again
given by equation (13.22).
(ii) Small arbitrarily disposed areas. In some circumstances it is pos-

sible to use equation (13.15) as it stands, if the areas dA 1 and dA 2
are small. Thus the energy received by a small disc placed in front
of a small window in a furnace could be approximately calculated
this way.
(iii) Thermocouple in a circular duct. A simple practical example of
the geometric configuration factor is found in consideration of a
thermocouple in a circular duct. It may be assumed that the thermo-
couple joint is represented by a small sphere and, further, that it is
situated at the centre of a duct of length 2L and radius R. It is
illustrated in Fig. 13.8. The line joining elements of area always
y;f
·----~---
dAl . 2
i
f---1- r-dl
L--L I L ___j
Fig. 13.8. The thermocouple configuration factor.
strikes the thermocouple joint normally, so cos ¢ 1 is always 1. The

element of area of the duct wall is 2nR dl. Since A 1 is a very small
sphere of radius rc, dA 1 is the disc area nr~, and is constant.
f
Applying equation (13 .16) gives
Qb<J- 2 > = dA 1 cos </1 2 ~nR dl
Qb(ll A1 A, nr
But cos ¢ 2 = R/r and r = (R 2 + 12 )t
=
1[ U
4 (R2 + F)t
J+L
-L = (R2
L
+ U)t (13.23)
ExAMPLE 13.1
A thermocouple situated at the centre of a circular duct 10 em

diameter by 0·25 m long has a spherical bead 2 mm diameter. It
reads 185°C with gas at 200°C flowing along the duct; the wall of
the duct is at 140°C. Determine a convection coefficient for heat
transfer between the gas and the bead, assuming radiating surfaces
are black.
Solution. Convection to the thermocouple from the gas is equal to

the radiation exchange between the thermocouple and the wall.
. r . 0·125 f
T he confi gurat10n ,actor IS 0 052 2 t = 0·93. I h is the
(· +0·125)
convection coefficient, and A the area of the bead, then
Qb = 0·93 x A x 56·7 x 10- 4 [(~~r- (~~rJ = hA(}

where(}= 200- 185 = 15
52·7 x w- 4 (441 - 292) = 15 h
h = 0·0523 kW/(m 2 K)
1·0 2·0 3·0 4·0 5·0 6·0

R2 = Wj D
Fig. 13.9. Configuration factors for parallel opposed rectangles. (From A. J.
Chapman, Heat Transfer, The MacMillan Company, New York (1974). By
permission of the publisher.)
(iv) Parallel and perpendicular rectangles. Radiation exchanges

between finite parallel rectangles and perpendicular rectangles
with a common edge occur in furnaces, etc., and details of the
application of equation (13.16) to these cases may be found in ref. 6.
Calculated values ofthe configuration factor are available in graphi-
cal form, shown in Fig. 13.9 for parallel rectangles and Fig. 13.10
for perpendicular rectangles.
0·5 R, LID_ 0·02

>--- 0·05
1..--
0·4 f.-- 0·1
7 L---
·2
. 0·3
I
0·4
0·6
~ 0·2 /' /' 1·0
/'
I·~
~
0· )/' :;;;-- ...... 4·0
6·0 \0·
0 20·0
0·1 0·2 0·3 0·4 0·6 0·81·0 2·0 3·0 4·0 6·0" 8·0 10
R 2 = WjD
Fig. 13.10. Configuration factor for perpendicular rectangles with a common
edge. (From A. J. Chapman, Heat Transfer, The MacMillan Company, New
York (1974). By permission of the publisher.)
13.6.2 Configuration Factor Algebra

Since all radiation emitted by a black surface must be accounted for it
is possible to deduce algebraic relationships between configuration
factors. Suppose that three surfaces make up a complete enclosure as in
Fig. 13.11. Any one of those three surfaces may exchange black body
radiation with the other two, and all the emission from surface 1 must
be incident on either surface 2 or surface 3. Thus, assuming surfaces 2
and 3 have the same temperature, T2 :
+ A1F1-3 a (Ti- TD
(13.24)
For any number of surfaces, the configuration factor from one to the
remainder is the sum of the component factors. This result is useful
since unknown factors may be deduced from known standard cases.
I
I
I
I
I
I I
/ ) - - - - - - - - - - - _L __ _
/'
/
/
/
//
............
Fig. 13.11. Three surfaces making an enclosure.
EXAMPLE 13.2
It is required to find F 1 - 2 for the case in Fig. 13.12, i.e., rectangles at

right angles, but not sharing a common edge. The sides of the rectangles
are of unit or half unit length as shown.
From Fig. 13.10, F 1 - 2 + 3 = 0·32, for W/D = 2·0 and L/D = 0·5. Also
F 1 _3 = 0·29 for W/D = 1·0 and L/D = 0·5. Hence, from equation
(13.24):
F 1 _2 = 0·32- 0·29 = 0·03

Fig. Example 13.2.
13.7 Grey-Body Radiation Exchanges

When radiating surfaces are grey, the emissivities of those surfaces
must be taken into account as well as their geometric configuration.
To enable the equation for a net energy exchange to be written in
a similar manner to that for black-body radiation, Hottel 8 intro-
duced a new factor §. Thus a net exchange is expressed as
Q(t2) = A 1 .~-2u(Tj- Ti> (13.25)
The derivation of§ will be considered by means of an electrical
analogy of radiation. 7 In the case of a net black-body radiation
exchange (13.21) is compared with Ohm's law, so that
Qb(t2l = F 1 - 2 A 1<1(Tt- T1) is equivalent to I = llV/R
Hence
Qb(12)=1;
THE LAWS OF BLACK-AND GREY-BODY RADIATION 225
The corresponding electric circuit is shown in Fig. 13.12.

R=-l-
AtFt-z
Surface 1 ~Surface 2
V1 = uT1 V2 = aT~
Fig. 13.12. An equivalent electric circuit for a net black-body radiation
exchange Qb(ll) =A 1F 1-2 a(T1 - T~).
An important initial assumption is that each radiating surface

has a constant value of p and e over the whole surface. From the
definitions of radiosity and irradiation in section 13.2 it follows that
the net rate at which energy leaves a grey surface is the difference
J- G, and from equation (13.3)
J = eQb + pG
1 _ G = 1 _ J - eQb
p
and since p + e = 1 for opaque surfaces, this reduces to
e
J- G = -(Qb- J)
p
If two surfaces only are involved, and these form an enclosure, this
is also the net energy exchange between them, Q( 1 2 J, and the equa-
tion may be compared with Ohm's Jaw so that QbiA, which is uT4 ,
and J IA are potentials and PI Ae is the resistance. The correspond-
ing circuit element for either surface is shown in Fig. 13.13.
R = _!!__
Ae
o----c=J--o
V= QJA V= JfA
Fig. 13.13.
Further, for surfaces of area A 1 and A 2 (at temperatures T1 and

T2 ) which have configuration factors of f 1 _ 2 and F2 _ 1 , the net
energy exchange is also the difference between the total radiation
leaving A 1 which reaches A 2 , and the total radiation leaving A 2
which reaches A 1 • Thus
Q1121 = (~:)A1F1-2- (~:)A2F2-1

But, from the reciprocal relationship, A 1F 1 _ 2 = A 2 F2 - 1,
Jl J2)
Q<l21 = ( A! - A2 A!Fl-2
This may also be represented by a circuit element, with potentials

J 1/A 1 and 1 2 /Az and resistance 1/A 1F 1- 2, as shown in Fig. 13.14.
R=-l-
A•F•-2
o---c=:::J--0
V=J 1/A 1 V=J 2/A 2
Fig. 13.14.
To simulate completely an energy exchange between the surfaces

A 1 and A 2 , three circuit elements may be joined in series as shown
in Fig. 13.15, the whole circuit now being compared to equation
(13.25). aT1 and aT~ are the end potentials, (equivalent to Qb(l )lA 1
and Qb<dA 2), and the total resistance is
~+ 1 +~
A1e1 A1F 1_ 2 A 2e2
1
R=--
A•Ft-2
1
= A2F2-t
Fig. 13.15. Complete circuit for radiation exchange between two grey surfaces
forming an enclosure.
From equation (13.25) the resistance is also given by 1/A 1 ~1 - 2 ,

hence
Multiplying both sides by A 1 and substituting 1 - e for p gives
1 (1-e 1 ) 1 A 1 (1-e 2 )
~-2 = -e-1- + F1-2 + A2 _e_2_
§,_, ~ (e_lc- 1) + _1_1

F
+ "-'(_lc- 1)
e A2
(13.26)
1 1_ 2 2
This result may be used for any two surfaces of area A 1 and A 2 ,
provided they form an enclosure, for which the configuration factor
is F 1 -2- Equation (13.26) simplifies for the special case of infinite
parallel or concentric grey planes for which F 1 _ 2 = 1 and A 1 = A 2 •
Then,
~-2 = - - - - - (13.27)
1 1
-+- -1
~>1 ~>2
This result can also be readily achieved without reference to the

equivalent electric circuit.
A further simple result which is useful is that if A 1 is completely
enclosed by A 2 , so that F 1 _ 2 = 1, and A 2 is large compared with
A 1 sothat
then equation (13 .26) reduces to

(13.28)
EXAMPLE 13.3
A small oven measures 0·4 m by 0·5 m by 0·3 m high. The floor of
the oven receives radiation from all the walls and roof which are at
300°C and have an emissivity of 0·8. The floor is maintained at
150°C and has an emissivity of0·6. Calculate the radiation exchange.
Solution. A 1 is the total area of walls and roof, which is 0·74 m 2 •
A 2 is the area of the floor, 0·2 m 2 • Since A 1 encloses A 2 , F2 _ 1 =1

and F 1 _ 2 = A 2 /A 1 = 0·27.
The grey-body factor ~ _ 2 is ( 1 )

--1
0·8
= 0·156
The radiation exchange is
4 4
0·156 X 0·74 X 56·7 X 10- 4 [( 573) 100 ] = 0·495 kW
100 - (423)
13.8 Non-luminous Gas Radiation

The more simple gas molecules such as the monatomic gases, and
diatomic gases of symmetric structure such as oxygen and nitrogen,
are effectively transparent to thermal radiation. However other gases
and vapours are good absorbers and emitters, and gas radiation
plays an important part in many practical heat transfer calculations.
The absorptive properties of a gas layer or volume depend on
the wave-lenth of the incident radiation, and on a function of the
number of gas molecules in the path length, expressed as a function of
path length and partial pressure. Thus
di;. = -I;.a;.pdx
expresses the reduction of intensity of the monochromatic incident
beam in passing through a distance dx at partial pressure p, where a;.
is the absorption coefficient If I;. has the value I ;.o at x = 0, this will
integrate to give
(13.29)
In terms of the transmissivity -r;.• I;. = I ;.o-r;., and hence the trans-
missivity is given by
(13.30)
Since -r;. + IX;. = 1 for a gas, it follows that
(13.31)
This is also equal to the emissivity s;. if Kirchhoff's law is a9sumed to
be valid.
In order to calculate a radiation exchange between a gas volume

and an enclosing surface, account must be taken of radiation entering
the gas volume from all directions. This would lead to the determina-
tion of emissivity (e 1) and absorptivity (oc1) for a particular size and
shape of gas volume. Such calculations are necessarily complex.
However Hottel9 has shown that gas volume shapes of practical
interest can be compared to equivalent hemispheres, where the
radiation from the surface to the centre of the base has a constant
path length, the radius, which is known as the mean beam length
for gas volume. Table 13.1 shows examples of mean beam length
for simple shapes.
Table 13.1. Beam lengths of some simple gas volume shapes

Characteristic Mean beam
Shape dimension, D length
Sphere Diameter 0·66D
Infinite Cylinder Diameter D
Cube Length of side 0·66 D
Space outside infinite bank of tubes,
centres on equilateral triangles, Clearance HD
diameter = clearance
The mean beam length of any shape may be calculated approxi-

mately as L = 3·4 x (volume of gas)/(area of enclosing surface).
Hottel has also made available extensive empirical data on the
emissivities of radiating gases which are presented as a function of
the product of partial pressure and beam length. Data for carbon
dioxide and water vapour are given in Appendix 4, together with
the procedure for calculating e1 and oc1 for gas mixtures.
13.8.1 Calculation of Radiation Exchange between Non-luminous

Gases and Containing Surfaces
(a) Black Surfaces. If the gas volume is enclosed by a black surface of
area A, the rate of radiation from the gas to the surface is
Q(g-s) = e 1 Acr~ {13.32)
where ~ is the absolute temperature of the gas. The rate of radiation
from the surface absorbed by the gas is
Q(s-g) = oc1 Acr~
Hence the net exchange between the gas and surface is
Q<P> = oA(e1 r: - cx1r.> (13.33)

(b) Grey Surfaces. For the case of grey surfaces, the electrical network
analogy of Section 13.7 may be used. Figure 13.16 shows a gas
contained by two grey surfaces at ~ and I; together with the analogy
circuit. It is assumed all radiation leaving surface 1 is transmitted
through the gas to reach surface 2.
Fig. 13.16. Electrical analogy of gas radiation involving two grey surfaces.
The radiation leaving surface 1 transmitted to surface 2 is 11 F 1 _ 2 't'l'

and that leaving surface 2 transmitted to surface 1 is J2 F2 _ 1 't'1 .
Hence the net exchange between surfaces by transmission is
J1 F1 _ 2't'1 - J2F2 _ 1't'8 . Using the reciprocal relationship A 1 F 1 _ 2 =
A2F2-1•
Q<•••2> = AlFl-2(1- cx,)(JtfAl- J2/A2)

This gives the top resistance in the circuit between the J 1/A 1 and
J 2/A 2 nodes. The gas emits e,ur;, r;
of this F1 _ 1e1 reaches surface 1.
The radiation leaving surface 1 absorbed by the gas is J 1 F1 _ 1a.1,

hence the exchange between surface 1 and the gas is
Q(s,g) = A1~ -l;a(J1/A1 - O'r.-)
assuming a.1 = e1, and the reciprocal relation A 1 F1 _ 11 = A1 F1 _ 1 . A
similar relation exists between the gas and surface 2. Thus the resis-
tances between the J 1/A 1 and O'r:
nodes, and the J 2/A 2 and O'r:
nodes are obtained. The O'r:node is a source or sink node unless
there is no chemical reaction when it becomes a floating potential
between the JtfA 1 and J 2 /A 2 values. For a gas contained by a
single grey surface the circuit consists of the two resistances between
the O'Ji, J 1/A 1 and O'r: nodes only.
13.9 Solar Radiation
Interest in solar radiation as an energy. source has increased sub-

stantially in recent years as the finite limits of fossil fuel reserves
have become more apparent. Even in Great Britain useful amounts
of sunshine are available either for direct conversion to electricity in
photo-voltaic cells or for absorption as low grade heat in flat plate
collectors.
The energy reaching the outer atmosphere of the earth may be
calculated approximately by assuming the earth to be seen as a
disc of radius 6436 Ian at a radial distance of 15()-6 x 106 km from
the sun. Hence the percentage of the sun's radiation reaching the
earthis(n X 6436 2 )/(411: X 10 12 X 150·62 ) X 100 = 4·56 X 10- 8 per
cent. The sun may be assumed a black-body emitter having a surface
temperature of 6000 K and surface area of 6·131 x 10 18 m2 ,
therefore the emission is 56·7 x 10- 12 x 6·131 x 10 18 x 6()()(f
= 4·506 x 1023 kW. Consequently the amount reaching the outer
atmosphere based on the earth's disc area is 21·4 x 10 13 kW, or
1·646 kW/m 2 of earth's surface.
The measured quantity of energy received on unit area perpen-
dicular to the sun's rays outside the earth's atmosphere at the mean
distance between the sun and the earth is known as the solar constant.
This is 2·00 calf(cm 2 min) or 1·388 kW/m 2 , rather less than the result
of the approximate calculation.
Amounts reaching the surface of the earth which may be put to
use are very much less, and indeed, at Kew the annual mean radia-
tion received averaged over 24 hours is of the order of 100 Wfm 2 •
Corresponding figures for the United States and Australia are about
180 and 200 W/m 2 •
From these figures it is possible to carry out feasibility studies of,
say, flat plate collectors for domestic hot water and lighting, bearing
in mind a collection efficiency of 1Q-15 per cent for conversion to
electricity using semi-conductor devices, and 5Q-70 per cent for
conversion to low-grade heat, by absorption on black surfaces
under glass.
EXAMPLE 13.4
A flat plate solar collector has a selective surface with an absorptivity
of 0·92 and an emissivity of 0·15. The coefficient for heat loss by
convection is 0·003 kW/(m 2 K). If the area is 20m2 calculate the
rate of energy collection, and the collection efficiency at a time when
the solar insolation is 800 W /m 2, for a collector temperature of
60°C and an ambient temperature of 18°C.
Solution. The rate of collection is 0·92 x 0·8 k W, less the losses,

per m 2 • The radiation loss is uer per m 2 = 56·7 X w - l l X 0·15
x 333 4 = 0·105 kW/m 2 • The convection loss is 0·003 (60 - 18)
= 0·126 kW/m 2 .
Hence the net rate of collection is 0·92 x 0·8 - 0·105 - 0·126
= 0·505 kW/m 2 • Hence the rate of collection for 20m 2 = 10·1 kW.
The collection efficiency = (0·505/0·8) x 100 = 63·1 per cent.
PROBLEMS
1. An electric oven is internally a 0·3 m cube. The back wall is maintained at

300°C and other surfaces may be assumed to be at a uniform l00°C. The
emissivity of all surfaces is (}8. Calculate the grey-body configuration
factor and the net radiation transfer from the heated wall. (Ans.: 0·77;
0·348 kW.)
2. A spherical thermocouple is situated at the middle of the length and on

the axis of a pipe of length L and diameter D to measure the temperature of
the gas flowing through the pipe. Assuming that the couple is so small in
comparison with the duct that the surface of the couple is always perpendicular
to the direction of radiation, deduce from first principles the expression
L
Fh-c = .J<D2 + Ll)
THE LAWS OF BLACK-AND GREY-BODY RADIATION 233
which gives the 'area factor' of the system. Both the couple and the pipe may
be considered to be black bodies.
In an installation similar to that described above, the thermocouple is
3 mm in diameter and the pipe is 0·92 m long and 0·31 min diameter. The gas
temperature is 149°C and the internal surface of the pipe is 65·6°C. Heat is
being transferred from the gas to the couple at the rate of0·79 kW/m 2 • What
would be the reading of the thermocouple if both the couple and the pipe
can be considered as black? The expression for the area factor given above
may be used. (Ans. 136°C.) (King's College, London).
3. A small disc of area At is concentric and parallel with a large disc of

area A 2 and diameter D, and is spaced at a distance R from it. Show
that for the discs:
A furnace chamber is a cylinder 2m long by 1 m diameter. At one end

there is a window 20 mm diameter on the axis of the cylinder. A
pyranometer with an element temperature of 50°C placed at the
window receives a radiation transfer of 1·97 W from the opposite
circular end. Calculate the temperature of that end. (Ans. 900°C.)
4. Show that for a grey body of area At and emissivity Et enclosed by

a surface of area A 2 and emissivity e2 the grey body configuration
factor is given by:
~-2 = ---------
1/Et +(At /A2)(1/€2-1)
which may be approximated to.F1 - 2 = e 1 ifA 2 islargeincomparison

withA 1 .
A billet of surface area 1·8 m 2 is heated in a rectangular furnace
3 m by 3·5 m by 2·5 m, the whole surface of which is radiating. For
e 1 = 0·8 and e2 = 0·7, calculate the percentage error in using the
approximate result for ~1 - 2 in calculating the radiation exchange. (Ans.
1·05 per cent.)
5. Two large parallel plates, 1 and 2, having emissivities on their inner faces
of 0·5 and 0·8 are maintained at 300° and 100oc respectively. A third plate
having unknown emissivities on its faces A and B is placed between the
other two plates. When face A is pointing towards plate 1, the third plate
reaches an equilibrium of 278°C. When the third plate is turned round so
that face B is pointing towards plate 1, its equilibrium temperature drops to
140°C. Determine the emissivities of the two faces A and B. (Ans. Face A,
e = 0·916, Face B, e = 0·102.) (The City University).
6. A plane surface of emissivity € 1 and at temperature T 1 radiates to a
parallel shield at temperature T 2 • The shield has an emissivity e 2 facing
the plane surface, and an emissivity e3 facing the surroundings at T3 ,
which may be assumed large. Show that the radiation from the plane
surface to the surroundings, via the shield, per unit area, is given by:
For the situation described T 1 = 500°C, e 1 = 0·4, e 2 = 0·2, e 3 = 0·05,

and T 3 = 30°C. Calculate the radiation from the plane surface to the
surroundings per unit area, and calculate the temperature of the radia-
tion shield. (Ans. 742 W/m 2 , 449°C.)
7. Calculate the rate of energy absorption on a flat plate heat collector

positioned normally to the sun's radiation, given the information in Section
14·9, plus the following: collector area 15 m2 , absorptivity to solar radiation
0·96, emissivity of plate 0·2, transmissivity of upper atmosphere 0·626,
collector surface temperature 68°C, atmospheric temperature 20°C,
natural convection coefficient from collector 0·003 kW/(m 2 K). (Ans. 10·35
kW.)
8. A billet reheating furnace has a brick interior 10m x 3m x 1 m. The

brick surface has an emissivity of 0·85 and is maintained at 900°C. Billets
pass slowly and continuously through the furnace on a moving floor and
may be assumed to receive heat transfer to an exposed surface area of 30m 2 .
Combustion gases, assumed transparent to radiation, at 900°C pass through
the furnace. The convection coefficient between gases and billets is 0·05
kW/(m 2 K). The emissivity of the billets is 0·7. Calculate the total furnace
heat output for a mean billet temperature of 450°C, and the percentages of
this output which are due to (a) radiation, and (b) convection. (Ans. 2485 kW,
(a) 72·8 %, (b) 27·2 %.) (The City University).
9. A rectangular furnace chamber has a floor of 5 m by 3·5 m. The

height is 4 m and the upper 1·5 m of all four vertical surfaces are
covered in radiant heaters having a temperature of 600°C. These
heaters radiate to the floor at 150°C. By considering off-set perpendicu-
lar rectangles, calculate the configuration factor between the whole area
and the floor. Assuming the surfaces are black, calculate the radiation
rate to the floor. (Ans. 0·126, 99·67 kW.)
10. An annular combustion chamber is contained between an inner

cylinder of 1·83 and an outer cylinder of 2·44 m diameter. Combustion gases
within the annular space have a mean temperature of 870°C and they con-
tain 8 per cent by volume of each of carbon dioxide and water vapour. The
pressure within the combustion chamber is 2 atmospheres. The outer
cylinder is maintained at 424°C and the inner cylinder at 488°C. Calculate
the net radiation from the gas perm length of chamber, and the heat extracted
from each wall. The emissivity of the inner wall is 0·9, and of the outer wall
0·6. (Ans. 311 kW/m. inner 155·9kW/m. outer 154·9 kW/m.) (The City
University).
11. A room having a floor of 5 m by 4 m and a height of 3 m is heated

by underfloor heating elements, so that the floor temperature is 35°C.
The wall and ceiling temperature is l8°C. The floor emissivity is 0·8
and for the walls and ceiling the emissivity is 0·7. The room is heated
by radiation and convection. For natural convection it may be assumed
that Nu = 0·14(Gr.Pr) 0 "33 with the appropriate length being the floor
mean side of 4·5 m. Calculate the total energy input to the floor heaters
assuming this is dissipated entirely by radiation and convection to the
room. (Ans. 2179·68W, giving 1517·92W radiation and 661·76W
convection.)
REFERENCES
I. Jakob, M. Heat Transfer, Vol. I, John Wiley and Sons, Inc., New York
(1949).
2. Planck, M. The Theory of Heat Radiation (Translation) Dover (1959).
3. Kirchhoff, G. Ostwalds Klassiker d. exakten Wissens .. 100, Leipzig (1898).
4. Lambert. J. H. Photometria (1860).
5. Sparrow, E. M. and Cess, R. D. Radiation Heat Transfer. Brooks/Cole
Publishing Company (1970).
6. Chapman, A. J. Heat Transfer, 3rd ed. The Macmillan Company, New
York (1974).
7. Oppenheim. A. K. Amer. Soc. Mech. Engs. Paper 54-A75 (1954).
8. Hottel, H. C. Notes on Radiant Heat Transmission, Chern. Eng. Dept..
M.I.T. (1951 ).
9. McAdams, W. H. Heat Transmission (Chapter 4, Hottel, H. C.), 3rd ed.,
Appendix 1: Heat Transfer
Literature
The following is a list of journals, proceedings, and bibliography
which may be consulted in order to keep abreast of the most recently
published work in heat transfer.
The International Journal of Heat and Mass Transfer, Pergamon
Press, monthly
The Journal of Mechanical Engineering Science, The Institution of
Mechnical Engineers, bi-monthly.
Journal of Heat Transfer, Transactions of the American Society of
Mechanical Engineers, Series C, quarterly.
Proceedings of the International Heat Transfer Conferences, e.g.,
4th 1970 (Paris~ 5th 1974(Tokyo), Elsevier Publishing Company,
Amsterdam.
Progress in Heat and Mass Transfer, Monograph Series of the
International Journal of Heat and Mass Transfer, Pergamon
Press.
Advances in Heat Transfer, Academic Press, New York.
Proceedings of the Heat Transfer and Fluid Mechanics Institute,
Stanford University Press, California.
Heat Bibliography, HMSO London, annual.
Reports of the National Engineering Laboratory, East Kilbride,
(available on request).
The Engineering Index, Engineering Index, Inc., New York.
Applied Science and Technology Index, The H. W. Wilson Company,
New York.
The British Technology Index, The Library Association, London.
ISMEC Bulletin, Information Service in Mechanical Engineering.
The Institution of Mechanical Engineers.
Science Abstracts A, Physics Abstracts, The Institution of Electrical
Engineers.
Science Abstracts B, Electrical and Electronic Abstracts, The
Institution of Electrical Engineers and The Institute of Electrical
and Electronic Engineers, Inc.
236
Appendix 2: Units and
Conversion Factors
SI units are used exclusively in this book. However, much of the

existing heat transfer literature is in British units, and SI-British
conversion factors are therefore included. The kJ and kW are
accepted alternatives to the J and W in the use of SI units. They are
the units of energy and power generally used in the teaching of
engineering thermodynamics and are the preferred units used in this
book. For a complete discussion see The UseofSI Units, published
by the British Standards Institution, PD 5686: 1972.
The Basic SI units are:
Mass 1 kg = 2·2046lb
Length 1m = 3·2808 ft
Time 1 s = 2·778 x 1o- 4 h
Temperature 1 K = 1·8 aRankine
Derived SI units are:
Force 1 N = 0·2248lbf (1 newton = 1 kg m/s 2 )
Pressure 1 Pa = 14·5 x 10- 5 lbf/in 2 (1 pascal= 1 N/m 2 )
1 bar = 105 Pa = 14·5lbf/in2
Density 1 kgfm 3 = 0·06243 lb/ft 3
Specific
volume 1 m 3 fkg = 16·0179 ft 3 flb
Energy 1 J = 1 Nm; 1 kJ = 103 Nm = 737·6ft lbf
Power 1 W = 1 Nmjs; 1 kW = 737·6 ft lbf/s = 1·341 h.p.
237
N
w
00
Conversion Factors for Heat Transfer Units

Physical Conversion Reciprocal tTl
quantity SI British units factor* conversion z
factor* C':l
Q kW Btufh 2·931 X 10- 4 3·412 X 103

z
tTl
tTl
q kW/m 2 Btu/(ft 2 h) 3-155 X 10- 3 3-170 X 102 :::0
h, u 2
kW/(m K) 2
Btu/(ft h °F) 5·678 X 10- 3 1·761 X 102 z
-
k kW/(mK) Btu/(ft h °F) 1·731 X 10- 3 5·777 X 102 C':l
cP kJ/(kg K) Btu/(lb °F) 4·1868 Q-2388 ::II
tTl
J1. Pas lb/(ft h) 4·134 x w- 4 2·419 X 103 >
(Note: 1 Pas = 10 dyn sjcm 2 = 10 poise) >-l
v, IX, e, D 2·581 X 10- S 3·874 X 104 >-l
m 2 fs ft 2 /h :::0
(Note: 1 m 2fs = 104 cm 2 fs = 104 stokes, unit of dynamic viscosity) >
't',P,p Pa lbf/ft 2 9·931 X 105 1·007 X 10- 6 z
rJ)
't',P,p 6·897 X 103 1·450 X 10- 4 >Tj
Pa lbf/in 2 tTl
:::0
* Multiply the numerical value in British units by the conversion factor to obtain the equivalent in SI; multiply the numerical value in SI
by the reciprocal conversion factor to obtain the equivalent in British units.
Appendix 3: Tables of Property Values
Table A.l. Thermal Properties of Solids: Metals
Properties at 20°C k x 10 3,kW/(m K)
p CP X103 /{ X103 IX
100 200 300 400 600

(~3) (k~K) (!:) (~2) (OC)
Aluminium, pure 2707 896 204 8-42 X 10- 5 206 215 229 249
Duralumin, 94-96 AI, 3-5 Cu 2787 883 164 6·68 182 194
Lead 11.370 130 34·6 2·34 33-4 31·5 29·8
N Iron, pure 7897 452 72-7 2·03 67·5 62·3 55·4 48·5 39·8
w
'CJ Iron, wrought, C < 0·5% 7849 460 58·9 1·63 57·1 51·9 48·5 45·0 36·4
Iron, cast, C ::::: 4% 7272 419 51·9 1·70
Carbon steel, C ::::: 0·5% 7833 465 53-7 1-47 51·9 48·5 45·0 41·5 34·6
Carbon steel. C = 1·5% 7753 486 36-4 0·97 36·3 36·3 34·6 32-9 31·2
Nickel steel, 10% 7945 460 26·0 0·72
Nickel steel, 30% 8073 460 12·1 0·33
Nickel steel, 50% 8266 460 13-8 0·36
Nickel steel, 70% 8506 460 26·0 0·67
Nickel steel, 90% 8762 460 46·7 1·16
Chrome steel, I % 7865 460 60·6 1·67 55-4 51·9 46·7 41·5 36·4
Chrome steel, 5 % 7833 460 39·8 1·11 38·1 36·4 36·4 32·9 29·4
Chrome steel, 10% 7785 460 31·2 0·87 31·2 31·2 29·4 29·4 31·2
Cr-Ni steel, 18% Cr, 8% Ni 7817 460 16·3 0-44 17·3 17·3 19·0 19·0 22·5
Ni-Cr steel, 20% Ni, 15% Cr 7865 460 14·0 0·39 15·1 15·1 16·3 17·3 19·0
Manganese steel, 2% 7865 460 38·1 1·05 36·4 36-4 36·4 34·6 32·9
tv
-1>-
0
Table A. I. Continued
Properties at 2o•c kx 10 3 kW/(mK)
p cP x 103 k X 103 IX
tT'l
100 200 300 400 600
zQ
(~~) (~2) (•q z
(k~K) (!:) tT'l
tT'l
~
Tungsten steel, 2% 7961 444 62-3 1·76 x w-~ 58·9 53·7 48·5 45·0 36·4
Silicon steel, 2 % 7673 460 31·2 0·89
z
Q
Copper, pure 8954 383 386 ll·2 379 374 369 364 353 ::z::
Bronze, 75 Cu, 25 Sn 8660 343 26·0 0·86 tT'l
Brass, 70 Cu, 30 Zn 8522 385 111 3-41 128 144 147 147 >
....,
German silver, 62 Cu 15 Ni, 22 Zn 8618 394 24·9 0·73 31·2 39·8 45·0 48·5 ....,
~
Constantan, 60 Cu, 40 Ni 8922 410 22·7 0·61 22·2 26·0 >
Magnesium, pure 1746 1013 171 9·71 168 163 158 z
(/)
Molybdenum 10.220 251 123 4·79 ll8 ll4 111 109 106 "l"l
Nickel, 99·9% pure 8906 446 90·0 2·27 83·1 72-7 64·0 58·9 tT'l
~
Silver, 99·9% pure 10.520 234 407 16·6 415 374 362 360
Tungsten 19,350 134 163 6·27 151 142 133 126 113
Zinc, pure 7144 384 112 4·ll 109 106 100 93-5
Tin, pure 7304 227 64·0 3-88 58·9 57·1
Adapted from Table A-1, E. R. G. Eckert and R. M. Drake, Jr., Heat and Mass Transfer, McGraw-Hill Book Company, New York (1959).
APPENDIX 3 241
Table A.2. Thermal Properties of Solids: non-Metals
cP x 103 p t k X loJ ac
(k~K) (~~) (OC) (!:) (~2)

Bakelite 1590 1273 20 0·232 0·0114 x 10-s
Bricks:
Common 837 1602 20 0·692 0·0516
Face 2050 20 1-32
Chrome 837 3011 200 2·32 0·0929
550 2-48 0·0981
900 1·99 0·0800
Diatomaceous earth 204 0·242
(fired) 872 0·312
Fire clay
(burnt 1450°C) 963 2323 500 1·28 0·0568
800 1·37 Q-0619
1100 1·402 0·0619
Magnesite 1130 204 3·81
648 2·77
1204 1·90
Concrete 879 1906- 20 (}814- Q-0490-
2307 1·40 0·0697
Glass, plate 837 2707 20 0·762 0-()336
Plaster, gypsum 837 1442 21 (}485 0-<>413
Stone:
Granite 816 2643 1·73- 0·0800-
3·98 0·183
Limestone 908 2483 99 1·26 Q-0568
299 1·33 Q-0594
Marble 808 2499- 20 2·77 Q-0394
2707
Sandstone 712 2163- 20 1·63- 0·106-
2307 2·08 0·127
Wood, cross grain :
Cypress 464 30 0·097
Fir 2721 417 24 0·109 0·0095
Oak 2387 ~81 30 0·166 Q-0126
Yellow pine 2805 641 24 0·147 0·0083
Wood, radial:
Oak 2387 609- 20 0·173- 0·0111-
481 (}207 (}0121
Fir 2721 417 20 0·138 0·0124
Table A.2. Continued
CP X 1()3 p t k X 103 ~
(k:JK) (~) c·q (~:) (~2)
Asbestos 816 577 0 0·151

816 577 100 0·192
Cotton 1298 80·1 20 ()-0589 0·194
Cork, board 160 30 ()-0433

Cork, expanded scrap 1884 44·8- 20 0-()363 ()-0155--
119 0·0439
Earth, coarse gravelly 1842 2050 20 0·519 D-0139
Felt, wool 330 30 ()-0519
Fibre, insulating board 237 21 ()-0485
Glass wool 670 200 20 ()-0398 ()-0284
Ice 1926 913 0 2·22 0·124
Silk 1382 57·7 20 ()-0363 ()-0439
Adapted from A. J. Chapman, Heat Transfer, The Macmillan Company, New York
(1960); L. S. Marks, Mechanical Engineers' Handbook, 5th ed., McGraw-Hill Book
Company, Inc., New York (1951); W. H. McAdams, Heat Transmission. 3rd ed.,
McGraw-Hill Book Company, Inc., New York (1954); and E. R. G. Eckert and R. M.
Drake, Jr., Heat and Mass Transfer, McGraw-Hill Book Company, Inc., New York
(1959).
APPENDIX 3 243
Table A3. Thermal Conductivity of Some Building Materials
k(Wj(mK))
Asbestos cement sheet 1520 0·29--0·43

Asbestos felt 144 0·078
Asbestos insulating board 720-900 0·11-0·21
Asphalt, roofing 1920 0·58
Brick, common, dry 1760 0·81
Brick, wet 2034 1-67
Chipboard 350-1360 0·07-0·21
Concrete, gravel1:2:4 2240-2480 1·4
vermiculite aggregate 400-880 0·11-0·26
cellular 320-1600 0·08-0·65
Cork, granulated, raw 115 0·046
slab, raw 160 0·05
Fibreboard 280-420 0·05-0·08
Glass, window 2500 1·05
Glassfibre, mat 50 0·033
Hardboard 560 0·08
Plasterboard, gypsum 1120 0·16
Polystyrene, expanded board 15 0·037
Polyurethane foam 30 0·026
Polyvinyl chloride, rigid foam 25--80 0·035-0·041
Roofing felt 960-1120 0·19--0·20
Tiles, clay 1900 0·85
Tiles, concrete 2100 1-10
Tiles, PVC asbestos 2000 0·85
Urea formaldehyde foam 8-30 0·032-0·038
Vermiculite granules 100 0·065
Wilton carpet 0·058
Table A3. Continued

U values for Building structures, based on the dift'eren~ between
inside and outside environment temperatures, and for sheltered,
normal and severe external exposure, in W/(m 2 K).
Sheltered Normal Severe
260 mm cavity wall, lOS mm inner

and outer leaves, plus 16 mm
lightplaster on inner face H H H
220 mm solid wall, with 16 mm light
plaster 1·8 1·9 2·0
335 mm solid wan, with 16 m light
plaster 1·4 1·5 1·6
Pitched roof, tiles on battens with
roofing felt, roof space, foil
backed plasterboard ceiling 1·4 H 1-6
As above, plus SO mm glass fibre loft
insulation ()-49 o-s ()-51
Window, single glazing, 30% area
due to wood frame 3·8 4·3 s-o
As above, double glazing 2-3 2·5 2·7
From the CIBS Guide Book A, The Chartered Institution of Building Services
Engineers, London. The above U values and thermal conductivities are a brief
extract only (used by permission of the Institution).
Table A.4. Physical Properties of some Common Low Melting Point Metals
Melting Boiling p Jl cP x 103 k X 103 Pr

point point Temp.
(oC) (oC) (OC)
(~~) Pas (~~)
(k:K)
Bismuth 272 1480 315 10,010 1·62 x 10- 3 144 16·4 0·014
760 9467 0·79 164 15·6 0·0084
Lead 328 1738 371 10.540 2·40 159 16·1 0·024
704 10,140 1·37 155 14·9 0·016
Lithium 179 1318 204 506 0·59 4187 38·1 0·065 >
"t::
"t::
983 442 0·42 4187 til
Mercury -39 357 10 13.570 1·59 138 8·14 0·027 z
0
315 12,850 0·87 134 14·0 0·0084
Potassium 64 760 149 807 0·37 796 45-() 0·0066 ><
w
704 674 0·13 754 33·1 0·0031
Sodium 97 884 204 902 0·43 1340 80·3 0·0072
704 779 0·18 1256 59·7 0·0038
Sodium-Potassium, 22% Na 19 826 93·5 849 0·49 946 24·4 0·019
760 690 0·16 883
Sodium-Potassium, 56% Na -II 795 93·5 887 0·58 1130 25·6 0·026
760 740 0·16 1042 28·9 0·058
Lead-Bismuth, 44·5% Pb 125 1670 288 10,350 1·76 147 1(}7 0·024
649 9835 1·15
N
Adapted from Table' 1&-1, J. G. Knudsen and D. L. Katz, Fluid Dynamics and Heat Transfer, McGraw-Hill Book Company, Inc., New York t;
(1958).
N
-1>-
0\
Table A.S. Thermal Properties of Saturated Liquids
t p cP x t()l v (% Pr {J
I kxW
(OC) (kgfm 3 1 kJ/(kgK) (m 2/s) I kW/(mK) (m 2/s) (1/K) m
z
C'l
Water (H 2 0) z
tT1
0 1002 4218 0·179 X 10-S ()-552 13-1 X 10- 8 13-6 m
:;e
20 1001 4182 ()-101 ()-597 14·3 7-()2 I ()-18 X 10- 3
40 994·6 4178 ()-0658 ()-628 15·1 4·34
zC'l
60 985-4 4184 00477 ()-651 15·5 3-()2 =
4196 m
80 974·1 00364 0·668 16·4 2·22 >
96()-6 00294 ()-680
....,
100 4216 16·8 1·74 ....,
120 945·3 4250 00247 ()-685 1H 1·446 :;e
140 928·3 4283 00214 0·684 17·2 1·241 >
160 909·7 4342 00189 ()-680 17-3 I -o99 z
V>
()-675 1-()04 'Tj
180 889·0 4417 00173 17·2
200 866·7 4505 00160 ()-665 17·1 ()-937
m
:;e
220 842-4 4610 00149 ()-653 16·8 ()-891
240 815·7 4756 00143 ()-635 16·4 ()-871
260 785-9 4949 00137 ()-611 15·6 ()-874
280 752·5 5208 00135 ()-580 14·8 o-910
300 714·3 5728 00135 ()-540 13-2 1-()19
Methyl Chloride (CH 3 Cl)
-50 1053 1476 0·032o x to- s 0·215 13·9 x 10- 8 2·31
-40 1033 1483 0·0318 0·209 13-7 2·32
-30 1017 1492 0·0314 0·202 13-4 2·35
-20 999·4 1504 Q-0309 0·196 no 2·38
-10 981·4 1519 0·0306 0·187 12·6 2·43
0 962-4 1538 0·0302 0·178 12·1 2·49
10 942·4 1560 D-0297 0·171 11·7 2·55
20 923·3 1586 Q-0292 0·163 Il-l 2·63
30 903-1 1616 0·0287 Q-154 10·6 2·72
40 883·1 1650 0·0281 D-144 9·96 2·83 >
'"t:l
50 861·2 1689 Q-0274 0·133 9·21 2·97 '"t:l
ttl
Freon (CCI 2 F 2 )
z
t:i
-50 1547 875·0 o-o310 x 10- s Q-0675 5·01 x 10- 8 6·2 2·63 x 10- 3 ><
w
-40 1519 884·7 D-0279 0-0692 5·13 5-4
-30 1490 895·6 0·0253 0·0692 5·26 4·8
-20 1461 907·3 0·0235 0.0710 5·39 4·4
-10 1430 920·3 D-0221 0.0727 5·50 4·0
0 1397 934·5 Q-0214 0.0727 5·57 3·8
10 1364 949·6 Q-0203 0·0727 5·60 3·6
20 1330 965·9 Q-0198 0·0727 5·60 3·5
30 1295 983·5 0·0194 0·0710 5·60 3·5
40 1257 1002 Q-0191 0·0692 5·55 3·5
50 1216 1022 0·0189 0·0675 5·44 3·5
w
~
-.J
t0
00
"""
t p c, X J()l v k X 103 IX Pr p
("C) (kl/m 3 ) kJ/(kg K) (m 1/s) kW/(mK) (m 2 /s) (1/K)
tT1
z
C1
Glycerin (C 3 H 5 (0Hh) z
0 1276 2261 8·31 x w- 3 0·282 9·83 x to- s 84·7 X 103 tT1
tT1
10 1270 2320 3-()() Q-284 9·65 31·0 ::e
20 1264 2387 H7 Q-286 9·47 12·5 I 0·504 x w- 3 z
30 1258 2445 ()-50 Q-286 9·29 5·38 C1
40 1252 2512 Q-22 0·286 9·13 2-45 ::I:
tT1
50 1245 2583 Q-15 0·287 8·93 1·63 >
....,
....,
Ethylene glycol (C 2 H 4 (0Hh) ::e
0 1130 2294 5-75 x w- 5 Q-242 9·34 X JO-S 615 >
z
~
20 1117 2382 1·92 Q-249 9·39 204 1 o-648 x w- 3 'Tl
40 1101 2474 Q-869 Q-256 9·39 93 tT1
60 1088 2562 Q-475 Q-260 9·31 51
::e
80 1078 2650 Q-298 Q-261 9·21 32-4
100 1059 2742 Q-203 Q-263 9-()8 22-4
Engine oil (unused)
0 899 1796 4·28 X JO-l ()-147 9·11 x lO-a 47,100
20 888 1880 ()-9() ()-145 8·72 10,400 I o-102 x w- 3
40 876 1964 ()-24 ()-144 8·33 2870
60 864 2047 00839 ()-140 8-()() lOSO
80 852 2131 00375 ()-138 7·69 490
100 840 2219 00203 ()-137 7-38 276
120 829 2307 00123 ()-135 HO 175
140 817 2395 00080 0·133 6·86 116
160 806 2483 00056 ()-132 6·63 84 >
:g
Mercwy(Hg) tr1
zt1
0 13.630 140-3 00124 x 10-s 8·21 430 X lO-a 00288
20 13.580 139-4 00114 8-69 461 00249 )·82 X 10- 4 xw
so 13,510 138-6 00104 9·40 S02 00207
100 13,390 137-3 000928 1o-5 571 00162
ISO 13,260 136·5 0-()()853 11·5 635 ()-0134
200 13,150 136·1 000802 12·3 691 ()-0116
2SO 13,030 135-7 000764 13-l 740 ()-0103
316 12.8SO 134.() 000673 14.() 815 0-()()83
Adapted from Table A-3, E. R. G. Eckert and R. M. Drake, Jr., Heat and Mass Transfer, McGraw-Hill Book Company, Inc., New York (1959).
N
~
\0
N
Table A.6. Thermal Properties of Gases at Atmospheric Pressure Vl
0
T p C0 X 1()3 v k X J()l Q( Pr
Jl
(oK) (kg/m 3 ) kJ/(kg K) (m 2/s) kW/(mK) (m 2/s) Pas
Air tr1
250 1·413 1005 ()-949 X 10-S ()-0223 1·32 X 10-S 1·60 X w-s z
0·722
.300 H77 1006 1·57 ()-0262 2·22 1·85 ()-708 z0
350 ()-998 1009 2-()8 ()-0300 tr1
2·98 2·08 0·697 tr1
400 Q-883 1014 2·59 0·0337 3-76 2·29 0·689 :;tl
450 ()-783 1021 2·89 ()-0371 4·22 2·48 ()-683 -z
500 ()-705 1030 3-79 ()-0404 5·57 2·67 0
0·680
550 ()-642 1039 4·43 ()-0436 6·53 ::I:
2·85 0·680 tr1
600 ()-588 1055 5·13 0·0466 7·51 3·02 0·680 >
650 ()-543 1063 o-l
5·85 ()-0495 8·58 3·18 0·682
()-503 o-l
700 1075 6-63 ()-0523 9·67 3-33 0·684 :;tl
750 ()-471 1086 7-39 ()-0551 10·8 3-48 0·686
800 o-441 1098 8·23 ()-0578 12-() 3·63 ()-689
zen>
850 o-415 1ll0 9-()7 ()-0603 'TI
IH 3·77 ()-692 tr1
900 Q-392 ll21 9·93 ()-0628 14·3 3·90 ()-696 :;tl
950 ()-372 ll32 1()-8 ()-0653 15·5 4·02 0·699
1000 ()-352 1142 11·8 ()-0675 16·8 4·15 ()-702
1100 Q-320 ll61 13·7 ()-0723 19·5 4·40 0·706
1200 ()-295 ll79 15·7 ()-0763 22-() 4·63 0·714
1.300 ()-271 1197 17·9 ()-0803 24·8 4·85 0·722
Hydrogen
250 Q-0981 14,060 8·06 X lO-S 0·156 11·3 x w- s 7·92 x w- 6 G-713
300 0·0819 14,320 10·9 0·182 15·5 8·96 Q-706
350 0·0702 14.440 14·2 Q-206 20·3 9·95 0·697
400 O-o614 14.490 17-7 0·229 25·7 1Q-9 0·690
450 0·0546 14,500 21·6 0·251 31·6 11·8 0·682
500 0·0492 14.510 25·7 0·272 38·2 12·6 0·675
550 0·0447 14.330 30·2 0·293 45·2 13·5 0·668
600 0·0408 14,540 35·0 0·315 53·1 14·3 0·664
650 0·0349 14,570 45·5 0·351 69·0 15·9 0·659
700 0·0306 14,680 56·9 0·384 85·6 17-4 0·664 >
"'C
750 0·0272 14,820 69·0 0·412 102 18·8 0·676 "'C
800 0·0245 14.970 82·2 0·440 120 20·2 0·686 z0tr.l
850 0-()223 15.170 96·5 0·464 137 21·5 0·703
Oxygen ><
-
w
200 1·956 913·1 Q-795 X lO-S 0-()182 1·0? X JO-S 14·9 x w- 6 0·745
250 1·562 915·6 1·144 Q-0226 1·58 17·9 0·725
300 1·301 920·3 1·586 Q-0267 2·24 2Q-6 0·709
350 1-113 929·0 2·080 Q-0307 2·97 23·2 0·702
400 Q-976 942-Q 2·618 Q-0346 3-77 25·5 0·695
450 0·868 956·7 3·199 Q-0383 4·61 27·8 Q-694
500 Q-780 972-2 3·834 Q-0417 5·50 29·9 0·697
550 0·710 988·1 4·505 Q-0452 6"·44 32·0 0·700
600 Q-650 1004 5·214 Q-0483 7-40 33·9 0·704
N
01
IV
...,.
Table A.6. Continued IV
T p c, X 1()3 v k X 103 IX IJ Pr
("K) (kl/m 3 ) k.J/(kg K) (m 2 /s) kW/(m K) (m 1 /s) Pas
--
Nitrogen ttl
200 1·711 1043 o-757 x w- 5 00182 1-o2 x w- 5 12·9 x w- 6 (}747 0
z
300 1-142 1041 1·563 00262 2·21 17·8 (}713 zttl
-
400 0·854 1046 2·574 00333 3-74 22·0 0·691 ttl
500 0·682 1056 3-766 00398 5·53 25·7 0·684 ~
600 0·569 1076 5·119 (}0458 7-49 29·1 (}686 -z
700 0·493 1097 6·512 00512 9·47 32·1 0·691 0
800 (}428 1123 8·145 00561 11·7 34·8 (}700 :::
ttl
900 (}380 1146 9·106 Q-0607 13-9 37·5 (}711 >
o-,l
1000 0·341 1168 11·72 00648 16·3 4(}0 (}724 o-,l
1100 (}311 1186 13·60 (}0685 18·6 42-3 (}736 ~
1200 (}285 1204 15·61 00719 20·9 44·5 (}748 >
ztil
Carbon dioxide "!1
250 2·166 803-9 Q-581 x w- 5 00129 0·740 x w- 5 12·6 x w-" (}793 ttl
~
300 1·797 87(}9 (}832 00166 1-()6 15·0 0·770
350 1·536 9002 1-119 00205 1·48 17·2 (}755
400 1·342 942-() 1·439 00246 1-95 19·3 (}738
450 H92 979·7 1·790 00290 2-48 21·3 (}721
soo 1-()73 1013 2-167 (}0335 3-()8 23-3 (}702
sso (}974 1047 2·574 00382 3-75 25-1 (}685
600 (}894 1076 3002 00431 4·48 26·8 (}668
Carbon monoxide
250 ()-841 1043 1-128 X 10- 5 (){)214 1·51 X 10- 5 15·4 X 10- 6 Q-750
300 1-139 1042 1·567 (){)253 2-13 17·8 ()-737
350 ()-974 1043 2-()62 (){)288 2·84 2()-1 ()-728
400 ()-854 1048 2·599 ()-0323 3·61 22·2 ()-722
450 ()-762 1055 3-188 ()-()436 4·44 24·2 ()-718
500 ()-682 1063 3·819 (){)386 5-33 26·1 ()-718
550 0·620 1076 4·496 ()-0416 6·24 27·9 ()-721
600 ()-568 1088 5·206 ()-0445 7-19 29·6 ()-724
Water vapour
>
380 ()-586 2060 ()-216 X 10- 4 (){)246 2·04 X 10- 5 12·7 X 10- 6 1-()60 :g
()-554 2014 ()-242 (){)261 2·24 13-4 1·040 ttl
400 zt)
450 ()-490 1980 ()-311 (){)299 3·07 15-3 1·010
500 ()-441 1985 ()-386 (){)339 3·87 17·0 ()-996 ><
550 0·400 1997 ()-470 (){)379 4·75 18·8 ()-991 w
600 0·365 2026 ()-566 ()-()422 5·73 2()-7 ()-986
650 ()-338 2056 ()-664 (H)464 6·66 22·5 ()-995
700 0·314 2085 0·772 (){)505 7-12 24·3 HXlO
750 ()-293 2119 ()-888 (){)549 8·83 26·0 1·005
800 ()-274 2152 1-()20 (){)592 1(){) 27·9 1-()10
850 Q-258 2186 H52 ()-0637 11·3 29·7 1-()19
Adapted from Table A-4, E. R. G. Eckert and R. M. Drake, Jr., Heat and Mass Transfer, McGraw-Hill Book Company, Inc., New York (1959).
(Note: At pressures other than atmospheric, the density can be determined from the ideal gas equation, p == p/RT. Hence at any given tem-
perature p = p 0 (pjp 0 ) where Po is atmospheric pressure and Po is given in the table. k, p, and c• may be assumed independent of pressure. tv
v and 11 are inversely proportional to the density; hence at a given temperature are inversely proportional to the pressure.) ~
Table A.7. Normal Total Emissivity of Various Surfaces
Ref. t Emissivity
(OC)
Aluminium:
Highly polished plate, 98·3% pure 11 237-576 Q-039--0057
Rough polish 1 100 Q-18
Commercial sheet 1 100 {)-()9
Heavily oxidized 2 93-505 Q-20-0-31
At-surfaced roofing 5 38 Q-216
Brass:
Highly polished, 73-2 Cu, 26·7 Zn 11 247-357 0028-0031
Polished 1 100 006
Rollcd plate, natural surface 10 22 006
Chromium. polished 1 100 0075
Copper:
Carefully polished electrolytic copper 6 80 Q-018
Polished 1 100 Q-052
Molten 3 1076-1278 Q-16-o-13
Iron and steel :
Steel, polished 1 100 0066
Iron, polished 12 427-1028 Q-14-0-38
Cast iron, polished 9 200 Q-21
Cast iron, newly turned lO 22 ()-44
Wrought iron, highly polished 16 38-249 Q-28
Iron plate, completely rusted lO 19 ()-69
Sheet steel, shiny oxide layer 10 24 Q-82
Steel plate, rough 5 38-372 Q-94-0-97
Cast iron, molten 15 1300-1400 Q-29
Steel, molten 7 1522-1650 Q-43-o-40
Stainless steel, polished 1 100 Q-074
Lead, grey oxidized 10 24 Q-28
Magnesium oxide 8 278-827 ()-55-Q-20
Nichrome wire, bright 14 49-1000 ()-65-Q-79
Nickel-silver, polished 1 100 0·135
Platinum filament 4 27-1230 0036-o-192
Silver, polished, pure 11 227--fJ27 ()-02....()-032
Tin, bright tinned iron lO 23 ()-043, ()-064
Tungsten filament 18 3320 Q-39
Zinc, galvanized sheet iron, fairly
bright lO 28 Q-23
APPENDIX 3 255
Ref. t Emissivity
c·q
Asbestos board 10 23 0·96
Brick:
Red, rough 10 21 (}93
Building 14 1000 0·45
Fireclay 14 1000 0·75
Magnesite, refractory 14 1000 0·38
Candle soot 17 97-272 (}952
Lampblack, other blacks 14 5{}-1000 0·96
Graphite, pressed, filed surface 8 249-516 (}98
Concrete tiles 14 1000 (}63
Enamel, white fused, on iron 10 19 0·90
Glass, smooth 10 22 0·94
Oak, planed 10 21 0·90
Flat black lacquer 5 38-94 0·96--0·98
Oil paints. 16 different, all colours 13 100 0·92-0·96
Aluminium paints, various 13 100 0·27-0·67
Radiator paint, bronze 1 100 0·51
Paper, thin, pasted on blackened plate 10 19 (}92, 0·94
Plaster, rough lime 16 1{}-87 (}91
Roofing paper 10 21 (}.91
Water (calculated from spectral data) {}-100 0·95-0·963
(Note: When temperatures and emissivities appear in pairs separated by dashes,

they correspond; and linear interpolation is permissible.)
By courtesy of H. C. Hottel, from Heat Transmission, 3rd ed., by W. H. McAdams,
REFERENCES
I. Barnes, B. T., Forsythe, W. E., and Adams, E. Q. J. Opt. Soc. Amer.,
Vol. 37, 804 (1947).
2. Binkley, E. R., private communication (1933).
3. Burgess, G. K. Nat/. Bur. Stand., Bull. 6, Sci. paper 121, Ill (1909).
4. Davisson, C., and Weeks, J. R. Jr. J. OpL Soc. Amer., Vol. 8, 581 (1924).
5. Heilman, R. H. Trans. ASME, FSP 51,287 (1929).
6. Hoffman, K. Z. Physik, Vol. 14, 310 (1923).
7. Knowles, D., and Sarjant, R. J. J. Iron and Steel Inst. (London), Vol. 155,
577 (1947).
8. Pirani, M. J. Sci. Instrum., Vol. 16, 12 (1939).
9. Randolf, C. F., and Overhaltzer, M. J. Phys. Rev., Vol. 2, 144 (1913).
10. Schmidt, E. Gesundh-Ing., Beiheft 20, Reihe 1, 1-23 (1927).
11. Schmidt, H., and Furthman, E. Mitt. Kaiser-Kilhelm-Inst. Eisenforsch.

Dusseldorf, Abhandle., Vol. 109, 225 (1928).
12. Snell, F. D. Ind. Eng. Chem., Vol. 29, 89 (1937).
13. Standard Oil Development Company, personal communication (1928).
14. Thring, M. W. The Science.of Flames and Furnaces, Chapman and Hall,
London (1952).
15. Thwing, C. B. Phys. Rev., Vol. 26, 190 (1908).
16. Wamsler, F. Z. Ver. deut.lng., Vol. 55, 599 (1911); Mitt. Forsch., Vol. 98,
1 (1911).
17. Wenzl, M., and Morawe, F. Stahl u. Eisen, Vol. 47, 867 (1927).
18. Zwikker, C. Arch. neerland. sci., Vol. 9, 207 (1925).
Appendix 4
Gas Emissivities
The curves in Figs. Al and A2 give respectively emissivities of carbon
dioxide and water vapour. In each case there are separate curves for
constant values of the product of partial pressure and mean beam
length. As the total pressure is increased, the lines of the C0 2
spectrum broaden, and a correction factor from Fig. A3 is applied
for pressures other than 1 atmosphere. In the case of water vapour,
the emissivity depends on the actual partial pressure and the total
pressure as well as on the product of partial pressure and beam length.
()o02
()oOl
()o008
()o006
2500K
Fig. Al. Emissivity of carbon dioxide; adapted from W. H. McAdams
Heat Transfer, McGraw·Hill Book Company, 3rd ed., New York (1954);
by permission of the publishers.
257
ENGINEERING HEAT TRANSFER
()-8
()-6
()-5
()-4
o-3
()o()l
<roo!
~~K~~~K~~~~--~~~--~~-
Fig. Al. Emiuillity ofwtlterNpO.,.; tulllpt«lfrom W. H. McAtlturu, Hetlt

Tr11111mislio11, 3nl ed. McGNw-Hi/1 Book Co111JH111y, New York (1954);
by permislio11 of tM pdlUMrr.
Hence Fig. A2 is for actual partial pressures extrapolated to zero,

and the emissivity is multiplied by a correction factor from Fig. A4.
When carbon dioxide and water vapour are both present the sum of
emissivities is reduced by a value & obtained from Fig. AS, to
allow for mutual absorption. Thus e1 = Ba 1o + Bco1 - lie. To
estimate absorptivities to radiation from enclosing surfaces, which
depend on the gas temperature as well as the surface temperature,
Hottel recommends an emissivity figure (e) is first determined at
the surface temperature and at (pL)(T./T.~ Then
!Xcoz = e(T"./1'.)0·65
IXnzo = e(T"./T.)0·45
APPENDIX 4 259
2·0 3·0 5·0

total pressure, atm
Fig. A3. Adllptetl from W. H. McAtlams, Hut Trtu~smissioll, 3rtl etl.,
McGraw-Hill Book Co~~~pt~11y, New York (1954); by permissi011 of tu
p•blh,ers.
(total pressure + PH 2o) /2, atm

Fig. A4. Adllptetl from W. H. McAtlams, Hetlt Tr1U1smissio11, 3rtl etl.,
McGraw-Hill Book Comptu~y, New York (1954); by permissioll of tu
pllblisurs.
810K >HOOK
0
PH20 PH.p PHil
Pco2 + PH 20 Pco2+ PHp Pco2 + 1 JH20
Fig. AS. Adllptetlfrom W. H. McAtlams, Heat Tra11smissioll, McGraw-Hill
Book ColllfHIIIy, New York (1954); by permusio11 of the p•blisurs. For
Iiiia ofcouttutt P002 L + P820 L, ilf m btu, 1-1·5 m btu, :Z-HJ m btu, 3-0·6
m btu, 4-0·5 m btu, 6-0•:Z m hr, 7-0·1 m btu.
Then the correction factors are applied as in the case of emissivity
determination, and finally the mutual absorption correction is
similarly made.
ExAMPLE
A 1·5 m cubic chamber contains a gas mixture at a total pressure of

2·0 bar and a temperature of 1000 K. The gas contains 5 per cent
by volume of carbon dioxide and 10 per cent water vapour. Determine
the emissivity of the gas mixture.
Solution. The beam length is (2/3) x 1· 5 m = 1·0 m.

pL(C0 2 ) = 0·1 m bar, e = Q-112
pL(H 2 0) = 0·2 m bar, e = 0·18.
The correction factor for C0 2 at 1·97 atm = H5 from Fig. A3, and
for H 2 0 at (0·197 + 1·97)/2 = 1·083 atm, is 1·5, from Fig. A4
ec02 = 0·112 x H5 = Q-129
f:H 2 0 = (}18 X 1·5 = 0·270
The correction for mutual absorption is at PH 2o/(p002 + PH 2o) = 0·66,
and pL(C0 2 ) + pL(H 2 0) = 0·3 m bar. From the set of curves at
1100 K, lle = Q-035, at 810 K, = 0·016. Hence & may be taken as
0·023.
eII = 0·129 + 0·270- Q-023 = 0·376
Index
absorptivity laminar 78
definition of 209 separation of 136
of black body 210 sub-layer 78, 107
of grey body 213 thermal 80
Akers, W. W. 149 thickness of 90
algebra, configuration factor, in thickness of 8 7-8
radiation 222-4 turbulent 78-9
analogy, Reynolds 101-11, see velocity distribution in 79
also Reynolds analogy velocity distribution in 79-87
analogy in complex flow 137 boundary mesh points 47-9
analogy of conduction 52-5 British Nuclear Fuels, plc 158
analogy of radiation 224-8, Buckingham's pi theorem 111
230-1 building materials, thermal con-
anisotropic materials 10 ductivities of 243
Bagley,R. 153 capacity ratio in heat exchangers

BASIC listings 23, 30, 45, 51, 65, definition of 179
116, 172, 198 limiting values of 180
batch heat exchangers 202-3 Carslaw, H. S. 10
Bayley, F. J. 47, 68, 80 Chapman, A. J. 118
beam length in gas radiation 229 Chato,J.C.J. 149
bibliography, heat transfer 236 Churchill, S. W. 139
Binder, L. 67 Clapp, R. M. 152
Biot, J. B. 3 Colburn, A. P. 109, 139
black body 6, 210 ColburnJ-factor 109, 137
artificial 21 0 Collins,M.W. 115
emission 211 condensation
radiation 210-24 general discussion of 144-5
boiling inside a tube 149
coefficients 15 1-3 on a horizontal tube 148
general discussion of 149-54 on a vertical surface 145-8
mechanisms of 149-50 conducting film, equivalent 91
vertical tube, in a 152-3 conduction
Boltzmann, L. 6, 211 definition of 3
boundary condition in transient differential equation of
conduction 6 2-3, 68-9 in cylindrical coordinates
boundary layer 13-15
equations of 80-7 in rectangular coordinates
growth in a tube entrance 79 10-13
integral equations of 84-7 in fins 157-60
261
262 INDEX
conduction cont'd diffusivity

in multiple plane slabs 170-3 eddy, definition of 101
one-dimensional thermal, definition of 12
in cylindrical layers 25-9 thermal eddy, definition of
in parallel systems 20 103
in plane slabs 16-20 dimensional analysis
in spherical layers 29 of forced convection 111-15
steady state 16-3 5 of natural convection 125-6
transient 61 -7 dimensionless groups 111
with heat sources 31-5 Donohue, D. A. 140
two-dimensional double glazed window
steady state 39-52 analysis 21-4
with heat sources 42-3 program list 23
conductivity Douglas, M. J. M. 139
of metals 9 drag loss coefficient 13 7
of non-metals 9 Drake, R. M. Jnr 10
conductivity, thermal
definition of 3 Eckert, E. R. G. 10,87
temperature dependent 10 eddy diffusivity 101
in a plane slab 24-5 effectiveness of heat exchangers
configuration factor 180
algebra 222-4 electrolytic tanks 55
in radiation 218-24, see also emission 21 0
radiation configuration of black body 211
factor of grey body 213
convection emissivities of various surfaces
at boundary 254-55
in transient conduction 63-7 emissivity, monochromatic 211
in two-dimensional con- of black body 214
duction 48-52 ofgreybody 214
coefficient 5, 18, see also emittance, monochromatic 211
Nusselt number emitters, selective 212
discussion of treatment 78 empirical results
forced see forced convection of forced convection 115,
in cross flow 139-42 118, 139-40
in separated flow 136-42 of natural convection 12 7-30
in tube bundles 139-42 energy equation for laminar flow
natural see natural convection in a tube 92-5
with phase change 144-54 energy equation of laminar
conversion factors 23 7-8 boundary layer 83
counter flow in heat exchangers integral form 85-7
176 energy stored in transient con-
critical radius in insulation 28-30 duction 62
program list 30 entry length, laminar flow 115
cross flow heat exchange 191-4 extended surfaces 15 7-73, see
Crosser, 0. K. 149 also fins
Farber, E. A. 150
Deans, H. A. 149 Fenner, R. T. 52
INDEX 263
film, equivalent conducting 91 Grashof number, definition of
fin analysis, program list 172 126
finite difference relationships grey body 212
in steady state conduction 42, emission 213
47-50 Griffith, P. 151
in transient conduction 61 , Grimison, E. D. 140
63,67
finned surface heat, definition of 2
equivalent effectiveness of heat exchangers
164-5 basic types of 176-7
overall coefficient of 165-8 batch 202-3
fins cross flow 177, 191-4
conduction in 158-62,170-2 determination of performance
effectiveness of 164-5 of 181-203
limit of usefulness of 164 in counter and parallel flow
numerical relationships in 181-91
170-3 in cross flow 191-4
temperature distribution in effectiveness of
160-3 at limiting value of capacity
fire-resistant door analysis 64-7 ratio 190
program list 65 in counter flow 187-8
Firman, E. C. 152 in cross flow 190-1
forced convection in parallel flow 189
definition of 4 general discussion of 176
dimensional analysis of 111-15 in-line 176-7
empirical results of 115, 118, thermal wheel 116, 194-9
139-40 transfer units 18 5-91
in laminar flow 78-98 heat flux 8
flat plates 87-92 heat sink, transistor 163
in tubes 92-8 heat transfer across boundary layer
in tubulent flow 117-19 in laminar flow 103
forces, buoyancy 4, 124-5 in turbulent flow 103
Fourier number, definition of 62 heat transfer coefficient 19, 28-9,
Fourier's law 3, 8 165-8, see also convection
friction coefficient coefficient and Nusselt number
for flat plates 102 in complex flow system 137-9
for tubes 103 in fins 158-64
in liquid metals 118-19
Gardner, G. C. 152 in uniform temperature system
gas emissivities 257-60 58-61
gas radiation, non-luminous 228- heat transfer in building structures
31 20-1
gases, thermal properties of horizontal surfaces, natural con-
250-3 vection 12 7-8
Gaussian elimination method 4 7 Hottel, H. C. 224, 229
Gauss-Siedel iterative method 52 Hsu,S.T. 67,127,139,153
Graetz number 115
graphical solution of transient insulation, critical thickness of
conduction 67 28-30
264 INDEX
integral energy equation of laminar metals, thermal properties of

boundary layer 85-7 239-40
integral equation of motion of mixed fluid in heat exchangers
laminar boundary layer 187
84-7 models, testing of 114
intensity of radiation 215-17 modes of heat transfer, discussion
irradiation 210 of 3-7
in grey body exchanges 225 momentum diffusivity, definition
isothermal surfaces in conduction of 83
8,9 monochromatic emissivity 211
isotropic materials 10
iterative technique 51-2 natural convection 4, 124-32
program list 51 approximate results, in air
130-2
Jaeger, J. C. 10 buoyancy force 125
Jakob, M. 153,211 definition of 4
/-factor 109, 137-8 dimensional analysis of 125-6
joule, definition of 237 empirical results of 126-32
in laminar flow 127-32
Karmam, T. von 85 in tubulent flow 127-32
Kays, W. M. 141, 186 newton, definition of 237
Kirchhoff's law 213-15 Newton's equation of convection
5, 18,78
Lam bert's law 215 Newton's second law 81
laminar boundary layer 78 number of transfer units, defini-
equations of 80-7 tion of 186
laminar convection numerical relationships in fins
in tubes 92-7 170-3
on flat plates 87-92 in steady state conduction 41-
laminar sub-layer 78 3,47-9
velocity at limit of in transient conduction 61-8
in tubes 103 numerical solution of cross-flow
on a flat plate 102 heat exchange 191-4
Langhaar, H. L. 112 of transient conduction 62-8
Liebmann method 52 of two-dimensional steady state
liquid metals conduction 40-52
heattransferin 118-19 Nusselt, W. 145
thermal properties of 245 Nusselt number
liquids, saturated, thermal proper- definition of 91
ties of 246-9 for laminar flow on flat plates
London, A. L. 141,186 91
lumped capacity systems 58-61, average value of 92
202-3 in pipes 96, 97
of condensation 148
MacLaurin's series 40 of finned surfaces 157
McAdams,W.H. 117,127,131
metals, liquid Ohm's law 17, 52, 224
heat transfer in 118-19 one-dimensional steady state
thermal properties of 245 conduction 16-35
INDEX 265
program list 23-4 surfaces 219

one-dimensional transient con- grey surfaces 227
duction 61-7 for parallel and perpendicular
program list 65 rectangles 221-2
overall heat transfer coefficient for thermocouple in a duct 220
19,28 radiation exchange
finned surfaces 165-8 between black bodies 217-24
heat exchangers 181 between grey bodies 224-8
Owen, J. M. 47,80 radiation in black enclosures 219
radiation in gases 228-31
parallel flow in heat exchangers radiosity 21 0
179, 182, 189 in grey body exchanges 225
parallel plates, natural convection radius, critical 28-9
129 Rayleigh number, definition of
pi theorem 111 126
Planck, M. 211 rectangular solids, natural con-
plate heat exchangers 200-2 vection 130
Pohlhausen, K. 88 reflectivity, definition of 209
Prandtl number, definition of 83 relaxation method 40-4
pressure loss program list 45
in a complex flow system resistivity 17
137-9 Reynolds, 0. 101
in pipe flow 103 Reynolds analogy 101-6
properties, thermal assumptions in 104
of building materials 243 in laminar flow 102
of gases 250-3 in turbulent flow 108
of liquid metals 245 in tubes 109-10
of metals 239-40 on flat plates 108-9
of non-metals 241-2 in laminar flow 104
of radiating surfaces 254-5 on a flat plate 105-6
of saturated liquids 246-9 in turbulent flow lOS
in tubes 106
radiation 208-32 Prandtl-Taylor modification
definition of 6 of 107-11
electrical analogy of 2 24-8, Reynolds number, definition of
230-1 79
general discussion of 208-9 Rohsenow, W. M. 151
intensityof 215-17 rotary generator 194-9
real surface 212
solar 231-2 Schenck, H. Jm 137
radiation coefficient 19 Schmidt, E. 67
radiation configuration factor Scorah, R. L. 1 SO
218-24 selective emitters 212-13
for arbitrarily disposed black shape factor
surfaces 218-19 electrical 54
for black bodies 217-24 thermal 54
for grey bodies 224-8 shear stress at wall 102
for grey enclosures 227 shear stress equation 79
for infinite parallel black in laminar flow 10 1
266 INDEX
shear stress equation cont'd program list 198-9

in turbulent flow 101 time constant 58
SI units 3, 23 7 transients, in cross flow heat
Sieder, E. N. 115 exchange 192-4
Snyder, N. W. 139 transistor heat sink, analysis 163
solar constant 231 transmissivity, definition of 209
solar energy, flat plate collectors turbulent boundary layer 78, see
for 232 also boundary layer
solar radiation 231-2 Turner, A. B. 47,80
solid, semi-infinite 68 two-dimensional steady state
spines, conduction in 158-63 conduction 39-55
Stanton number, definition of program list 45-6, 51
105 two-dimensional transient con-
Stefan-Boltzmann constant 211 duction 67-8
Stefan-Boltzmann law 211
system, uniform temperature, units, discussion of 3, 23 7
heat transfer in 58-61 unmixed fluids in heat exchangers
185
Tate, G. E. 115 U-values for building structures
temperature, periodic changes of,
in transient conduction variables
68-75 in forced convection 112
temperature distribution in natural convection 125
in fins 161-2 velocity of temperature wave 72
in laminar pipe flow 95 velocity profile
in thermal boundary layer 88 in condensing flow 145
temperature residuals 42 in laminar flow on flat plates
temperature wave 87
velocity of propagation of 72 in pipes 79, 94
wave-length of 72 in turbulent flow on flat plates
Test,F.L. 115 78, 107
thermal boundary layer 80 in pipes 79
on a flat plate 88 vertical cylinder, hollow, natural
thickness of 90 convection in 130
thermal diffusivity, definition of vertical surfaces, natural con-
12 vection 128-9
thermal eddy diffusivity, defini- viscosity
tion of 104 kinematic, definition of 83
thermal properties molecular, definition of 79
of building materials 243 temperature dependent 114
of gases at atmospheric pressure
250-3 wall shear stress 102
of liquid metals 245 watt, definition of 237
of saturated liquids 246-9 wave-length of temperature wave
of solids 239-43 72
thermal wheel 116, 194-9 wheel, thermal 116, 194-9

J. R. Simonson (Auth.) - Engineering Heat Transfer-Palgrave Macmillan UK (1988)

Uploaded by

Copyright:

Available Formats

J. R. Simonson (Auth.) - Engineering Heat Transfer-Palgrave Macmillan UK (1988)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

J. R. Simonson (Auth.) - Engineering Heat Transfer-Palgrave Macmillan UK (1988)

Uploaded by

Copyright:

Available Formats

Engineering Heat Transfer

Other Macmillan titles of related interest

All rights reserved. No reproduction, copy or transmission

First edition 1975

British Library Cataloguing in Publication Data

Preface to First Edition viii

Preface to Second Edition x

2 The Equations of Heat Conduction 8

3 One-dimensional Steady State Conduction 16

4 Two-dimensional Steady State Conduction 39

7 Forced Convection: Reynolds Analogy and Dimensional

8 Natural Convection 124

9 Separated Flow Convection 136

10 Convection with Phase Change 144

11 Extended Surfaces 157

13 The Laws of Black- and Grey-body Radiation 208

Appendix 1 Heat Transfer Literature 236

Appendix 2 Units and Conversion Factors 237

Appendix 3 Tables of Property Values 239

Appendix 4 Gas Emissivities 257

The aim of this book, which is a revised edition of a book previously

been made throughout in the light of reviews and criticisms received.

Department of Mechanical Engineering, JOHN R. SIMONSON

r radius, radial direction

Pr Prandtl number, CpJJ/k

One of the primary concerns of the engineer is the design and

internal energy within the system. As it is convenient to keep to the

In the British system of units, used in the majority of publications in

and forced convection there are two sub-divisions of laminar and

h is the convection coefficient and has the units kW/(m 2 K) in the

Natural convection 0·004-0-05

• For numerical conversion factors, see the Appendix

convection is to predict values of h for design purposes. The value

where T is the absolute temperature, A is the surface area of a per-

mum possible energy at all wave-lengths. The energy emitted will

2.1 The Nature of Heat Conduction

2.2 The Differential Equation of Conduction in a Cartesian

since the area of flow normal to the x-direction is dy dz and the

Fig. 2.1. Conduction in an element of11Uiterial ill Carte1ilut coordillate1.

where tX = k/ peP and is known as the thermal diffusivity of the

since q', o2 tjoy 2 and o2 t/oz 2 are equal to 0.

it being permissible to use the total derivative in the one-dimensional

_ (o2t o2t) and 0= oc(~)

Problems involving equations (2.8) to (2.12) will be considered in

2.3 The Differential Equation of Conduction in a Cylindrical

and out of the element in the radial direction,

dQ(r+dr) = - kdz(r + dr)dO :r(t + ;; dr)

neglecting a term of higher order.

Fig. 2.2. CoNhu:tion itt tu1 element ofm~~teritll in cylindricfll coordiutes.

and, in the axial direction,

The rate of heat generation within the element is

This equation may similarly be simplified to suit any particular

and without heat generation,

~= Jo 2 t + ~ ot + __1_ ~(sin ~) + 1 o2 t] + .!(_

The simplest example of steady state conduction in one dimension

3.1 Conduction in Plane Slabs

thermal condu ctivity, k

Fig. J.l. One-dimensioiUII steady sttlte cond11ction in tl pltute slab.

in which form it may be compared with Ohm's law describing the

Re-arranging and adding :