Advanced Aircraft Flight Performances

ADVANCED AIRCRAFT FLIGHT PERFORMANCE
This book deals with aircraft flight performance. It focuses on commer-

cial aircraft but also considers examples of high-performance military
aircraft. The framework is a multi-disciplinary engineering analysis,
fully supported by flight simulation, with software validation at sev-
eral levels. The book covers topics such as geometrical configurations,
configuration aerodynamics and determination of aerodynamic deriva-
tives, weight engineering, propulsion systems (gas turbine engines and
propellers), aircraft trim, flight envelopes, mission analysis, trajectory
optimisation, aircraft noise, noise trajectories and analysis of environ-
mental performance. A unique feature of this book is the discussion
and analysis of the environmental performance of the aircraft, focusing
on topics such as aircraft noise and carbon dioxide emissions.
Dr. Antonio Filippone’s expertise is in the fields of computational and

experimental aerodynamics, flight mechanics, energy conversion sys-
tems, propulsion systems, rotating machines (helicopter rotors, pro-
pellers, wind turbines), systems engineering, and design and optimisa-
tion. He has published more than eighty technical papers, ten book
chapters, and two books, including Flight Performance of Fixed and
Rotary Wing Aircraft (2006).
Cambridge Aerospace Series
Editors
Wei Shyy
and
Vigor Yang
1. J. M. Rolfe and K. J. Staples (eds.): Flight Simulation

2. P. Berlin: The Geostationary Applications Satellite
3. M. J. T. Smith: Aircraft Noise
4. N. X. Vinh: Flight Mechanics of High-Performance Aircraft
5. W. A. Mair and D. L. Birdsall: Aircraft Performance
6. M. J. Abzug and E. E. Larrabee: Airplane Stability and Control
7. M. J. Sidi: Spacecraft Dynamics and Control
8. J. D. Anderson: A History of Aerodynamics
9. A. M. Cruise, J. A. Bowles, C. V. Goodall, and T. J. Patrick: Principles of Space
Instrument Design
10. G. A. Khoury (ed.): Airship Technology, Second Edition
11. J. P. Fielding: Introduction to Aircraft Design
12. J. G. Leishman: Principles of Helicopter Aerodynamics, Second Edition
13. J. Katz and A. Plotkin: Low-Speed Aerodynamics, Second Edition
14. M. J. Abzug and E. E. Larrabee: Airplane Stability and Control: A History of
the Technologies that Made Aviation Possible, Second Edition
15. D. H. Hodges and G. A. Pierce: Introduction to Structural Dynamics and
Aeroelasticity, Second Edition
16. W. Fehse: Automatic Rendezvous and Docking of Spacecraft
17. R. D. Flack: Fundamentals of Jet Propulsion with Applications
18. E. A. Baskharone: Principles of Turbomachinery in Air-Breathing Engines
19. D. D. Knight: Numerical Methods for High-Speed Flows
20. C. A. Wagner, T. Hüttl, and P. Sagaut (eds.): Large-Eddy Simulation for
Acoustics
21. D. D. Joseph, T. Funada, and J. Wang: Potential Flows of Viscous and
Viscoelastic Fluids
22. W. Shyy, Y. Lian, H. Liu, J. Tang, and D. Viieru: Aerodynamics of Low
Reynolds Number Flyers
23. J. H. Saleh: Analyses for Durability and System Design Lifetime
24. B. K. Donaldson: Analysis of Aircraft Structures, Second Edition
25. C. Segal: The Scramjet Engine: Processes and Characteristics
26. J. F. Doyle: Guided Explorations of the Mechanics of Solids and Structures
27. A. K. Kundu: Aircraft Design
28. M. I. Friswell, J. E. T. Penny, S. D. Garvey, and A. W. Lees: Dynamics of
Rotating Machines
29. B. A. Conway (ed.): Spacecraft Trajectory Optimization
30. R. J. Adrian and J. Westerweel: Particle Image Velocimetry
31. G. A. Flandro, H. M. McMahon, and R. L. Roach: Basic Aerodynamics
32. H. Babinsky and J. K. Harvey: Shock Wave–Boundary-Layer Interactions
33. C. K. W. Tam: Computational Aeroacoustics: A Wave Number Approach
34. A. Filippone: Advanced Aircraft Flight Performance
Ignoranti quem portum petat nullus suus ventus est.
No wind is favourable to a sailor who does not know at which port
to land.
[Lucius A. Seneca (4 BC–AD 65), Moral Letters to Lucilius (letter 71)]
Advanced Aircraft Flight Performance
Antonio Filippone
The University of Manchester
cambridge university press
Cambridge, New York, Melbourne, Madrid, Cape Town,
Singapore, São Paulo, Delhi, Mexico City
Cambridge University Press
32 Avenue of the Americas, New York, NY 10013-2473, USA
www.cambridge.org
Information on this title: www.cambridge.org/9781107024007
© Antonio Filippone 2012
This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2012
Printed in the United States of America
A catalog record for this publication is available from the British Library.
Library of Congress Cataloging in Publication Data

Filippone, Antonio, 1965–
Advanced aircraft flight performance / Antonio Filippone.
p. cm. – (Cambridge aerospace series ; 34)
Includes bibliographical references and index.
ISBN 978-1-107-02400-7 (hardback)
1. Airplanes – Performance. 2. Airplanes – Design and construction. I. Title.
TL671.4.F449 2012
629.132 3–dc23 2012015394
ISBN 978-1-107-02400-7 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of

URLs for external or third-party Internet Web sites referred to in this publication
and does not guarantee that any content on such Web sites is, or will remain, accurate
or appropriate.
Contents
Tables page xvii

Preface xxi
Nomenclature xxiii
Technology Warning xxvii
1 Prolegomena . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Performance Parameters 2
1.2 Flight Optimisation 4
1.3 Certificate of Airworthiness 4
1.4 The Need for Upgrading 6
1.5 Military Aircraft Requirements 7
1.6 Review of Comprehensive Performance Programs 9
1.7 The Scope of This Book 10
1.8 Comprehensive Programs in This Book 13
Bibliography 14
2 Aircraft Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.1 Model for Transport Aircraft 16
2.2 Wire-Frame Definitions 20
2.2.1 Stochastic Method for Reference Areas 21
2.3 Wing Sections 23
2.4 Wetted Areas 24
2.4.1 Lifting Surfaces 24
2.4.2 Fuselage 25
2.4.3 Nacelles and Pylons 28
2.4.4 Winglets 29
2.4.5 Flaps, Slats and Other Control Surfaces 30
2.4.6 Model Verification: Cross-Sectional Area 30
2.5 Aircraft Volumes 31
2.5.1 Case Study: Do Aircraft Sink or Float on Water? 32
2.5.2 Wing Fuel Tanks 33
2.6 Mean Aerodynamic Chord 34
vii
viii Contents
2.7 Geometry Model Verification 35

2.7.1 Case Study: Wetted Areas of Transport Aircraft 36
2.8 Reference Systems 37
2.8.1 Angular Relationships 40
2.8.2 Definition of the Aircraft State 41
Summary 41
Bibliography 42
Nomenclature for Chapter 2 42
3 Weight and Balance Performance . . . . . . . . . . . . . . . . . . . . . . . . . 45

3.1 A Question of Size 45
3.2 Design and Operational Weights 47
3.3 Weight Management 51
3.4 Determination of Operational Limits 52
3.5 Centre of Gravity Envelopes 53
3.5.1 CG Travel during Refuelling 54
3.5.2 CG Travel in Flight 55
3.5.3 Design Limits on CG Position 57
3.5.4 Determination of the Zero-Fuel CG Limit 59
3.5.5 Influence of CG Position on Performance 59
3.6 Operational Moments 60
3.7 Use of Wing Tanks 61
3.8 Mass and Structural Properties 62
3.8.1 Mass Distribution 64
3.8.2 Centre of Gravity 68
3.8.3 Moments of Inertia 68
3.8.4 Case Study: Moments of Inertia 73
Summary 75
Bibliography 75
4 Aerodynamic Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.1 Aircraft Lift 78
4.1.1 Calculation of Wing Lift 79
4.1.2 Wing Lift during a Ground Run 79
4.1.3 Lift Augmentation 81
4.1.4 Maximum Lift Coefficient 84
4.2 Aircraft Drag 85
4.2.1 Lift-Induced Drag 85
4.2.2 Profile Drag 87
4.2.3 Wave Drag 93
4.2.4 Interference Drag 94
4.2.5 Drag of the Control Surfaces 95
4.2.6 Landing-Gear Drag 96
4.2.7 Environmental Effects 100
4.2.8 Other Drag Components 102
Contents ix
4.2.9 Case Study: Aerodynamics of the F4 Wind-Tunnel Model 102

4.2.10 Case Study: Drag Analysis of Transport Aircraft 103
4.2.11 Case Study: Drag Analysis of the ATR72-500 104
4.2.12 Case Study: Drag Analysis of the Airbus A380-861 104
4.3 Transonic Airfoil Model 105
4.4 Aircraft Drag at Transonic and Supersonic Speeds 108
4.4.1 Drag of Bodies of Revolution 110
4.5 Buffet Boundaries 113
4.6 Aerodynamic Derivatives 114
4.7 Float-Plane’s Hull Resistance in Water 115
4.8 Vortex Wakes 116
Summary 118
Bibliography 118
5 Engine Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126

5.1 Gas Turbine Engines 126
5.2 Thrust and Power Ratings 128
5.2.1 Engine Derating 129
5.2.2 Transient Response 130
5.3 Turbofan Engine Model 130
5.3.1 Aero-Thermodynamic Model 132
5.3.2 Determination of Design Point 133
5.3.3 Case Study: General Electric CF6-80C2 134
5.3.4 Rubber Engines 137
5.3.5 Effects of Contamination 138
5.3.6 Performance Deterioration 139
5.3.7 Data Handling 140
5.4 Turboprop Engines 141
5.4.1 Case Study: Turboprop PW127M 143
5.5 Turbojet with After-Burning 143
5.6 Generalised Engine Performance 145
5.7 Auxiliary Power Unit 147
5.7.1 Case Study: Honeywell RE-220 APU 149
Summary 149
Bibliography 150
6 Propeller Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152

6.1 Propeller Definitions 152
6.1.1 Propeller Limitations 156
6.2 Propulsion Models 156
6.2.1 Axial Momentum Theory 157
6.2.2 The Blade Element Method 160
6.2.3 Propeller in Non-Axial Flight 163
6.2.4 Case Study: Hamilton-Sundstrand F568 Propeller 165
x Contents
6.3 Flight Mechanics Integration 168

6.3.1 Propeller’s Rotational Speed 171
6.4 Propeller Installation Effects 173
6.4.1 Gearbox Effects 175
Summary 175
Bibliography 176
7 Airplane Trim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179

7.1 Longitudinal Trim at Cruise Conditions 179
7.1.1 Trim Drag 183
7.1.2 Solution of the Static Longitudinal Trim 183
7.1.3 Stick-Free Longitudinal Trim 184
7.2 Airplane Control under Thrust Asymmetry 186
7.2.1 Dihedral Effect 186
Summary 192
Bibliography 192
8 Flight Envelopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

8.1 The Atmosphere 195
8.1.1 International Standard Atmosphere 195
8.1.2 Other Atmosphere Models 198
8.2 Operating Speeds 203
8.3 Design Speeds 206
8.4 Optimum Level Flight Speeds 208
8.5 Ceiling Performance 210
8.5.1 Pressure Effects on Human Body 210
8.5.2 Cabin Pressurisation 211
8.6 Flight Envelopes 211
8.6.1 Calculation of Flight Envelopes 213
8.6.2 Case Study: Flight Envelopes of the A320 and
G550 215
8.7 Supersonic Flight 216
8.7.1 Supersonic Dash 216
8.7.2 Supersonic Acceleration 217
8.7.3 Supersonic Flight Envelopes 218
Summary 220
Bibliography 220
9 Take-Off and Field Performance . . . . . . . . . . . . . . . . . . . . . . . . . 224

9.1 Take-Off of Transport-Type Airplane 224
9.2 Take-Off Equations: Jet Airplane 228
9.2.1 Ground Run 229
9.2.2 Rolling Coefficients 231
Contents xi
9.3 Solution of the Take-Off Equations 232

9.3.1 Case Study: Normal Take-Off of an Airbus
A300-600 Model 234
9.3.2 Effect of the CG Position on Take-Off 236
9.3.3 Effect of Shock Absorbers 236
9.4 Take-Off with One Engine Inoperative 238
9.4.1 Decelerate-Stop 239
9.4.2 Accelerate-Stop 240
9.5 Take-Off of Propeller Aircraft 242
9.6 Minimum Control Speed 245
9.7 Aircraft Braking Concepts 248
9.8 Performance on Contaminated Runways 250
9.8.1 Contamination Drag 251
9.8.2 Impingement Drag 253
9.9 Closed-Form Solutions for Take-Off 254
9.9.1 Jet Aircraft 255
9.9.2 Propeller Aircraft 259
9.10 Ground Operations 260
9.10.1 Ground Manoeuvring 261
9.10.2 Bird Strike 262
Summary 264
Bibliography 264
10 Climb Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269

10.1 Introduction 269
10.2 Closed-Form Solutions 270
10.2.1 Steady Climb of Jet Airplane 270
10.2.2 Steady Climb of Propeller Airplane 271
10.2.3 Climb at Maximum Angle of Climb 272
10.3 Climb to Altitude of a Commercial Airplane 273
10.3.1 Climb Profiles 273
10.3.2 OEI Take-Off and Go-Around 277
10.3.3 Governing Equations 277
10.3.4 Boundary-Value Problem 278
10.3.5 Numerical Issues 281
10.3.6 Initial Climb with One Engine Inoperative 282
10.4 Climb of Commercial Propeller Aircraft 282
10.5 Energy Methods 285
10.5.1 Total-Energy Model 286
10.5.2 Specific Excess Power Charts 288
10.5.3 Differential Excess Power Charts 290
10.6 Minimum Problems with the Energy Method 291
10.6.1 Minimum Time to Climb and Steepest Climb 291
10.6.2 Minimum Fuel to Climb 292
10.6.3 Polar Chart for the Climb Rate 292
xii Contents
10.6.4 Case Study: Climb to Specified Mach Number 293

10.6.5 Minimum Flight Paths 295
Summary 296
Bibliography 296
11 Descent and Landing Performance . . . . . . . . . . . . . . . . . . . . . . . . 300

11.1 En-Route Descent 300
11.2 Final Approach 303
11.3 Continuous Descent Approach 307
11.4 Steep Descent 308
11.5 Unpowered Descent 311
11.5.1 Minimum Sinking Speed 311
11.5.2 Minimum Glide Angle 312
11.5.3 General Gliding Flight 313
11.5.4 Maximum Glide Range with the Energy
Method 314
11.6 Holding Procedures 315
11.7 Landing Performance 316
11.7.1 Airborne Phase 317
11.7.2 Landing Run 318
11.7.3 Crab Landing 320
11.8 Go-Around Performance 323
Summary 324
Bibliography 325
12 Cruise Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328

12.2 Point Performance 329
12.2.1 Specific Air Range at Subsonic Speed 330
12.2.2 Figure of Merit 331
12.2.3 Weight-Altitude Relationship 332
12.3 Numerical Solution of the Specific Air Range 332
12.3.1 Case Study: Gulfstream G550 335
12.3.2 Case Study: ATR72-500 338
12.3.3 Effects of Atmospheric Winds on SAR 338
12.4 The Range Equation 339
12.4.1 Endurance 341
12.5 Subsonic Cruise of Jet Aircraft 341
12.5.1 Cruise at Constant Altitude and Mach Number 342
12.5.2 Cruise at Constant Altitude and Lift Coefficient 343
12.5.3 Cruise at Constant Mach and Lift Coefficient 343
12.5.4 Comparison among Cruise Programs 344
12.5.5 Fuel Burn for Given Range 345
Contents xiii
12.6 Cruise Range of Propeller Aircraft 346

12.7 Cruise Altitude Selection 347
12.8 Cruise Performance Deterioration 349
12.9 Cost Index and Economic Mach Number 350
12.10 Centre of Gravity Position 352
12.11 Supersonic Cruise 353
12.11.1 Cruise at Constant Altitude and Mach Number 354
12.11.2 Cruise at Constant Mach Number and Lift Coefficient 355
Summary 355
Bibliography 356
13 Manoeuvre Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360

13.2 Powered Turns 361
13.2.1 Banked Turn at Constant Thrust 362
13.2.2 Turn Power and High-Speed Manoeuvre 363
13.2.3 Turn Rates and Corner Speed 365
13.2.4 Minimum-Fuel Turn 367
13.3 Unpowered Turns 369
13.4 Manoeuvre Envelope: V-n Diagram 370
13.4.1 Sustainable g-Loads 374
13.5 Roll Performance 374
13.5.1 Mach Number Effects 378
13.6 Pull-Up Manoeuvre 379
13.7 Flight in a Downburst 380
13.7.1 Aircraft Manoeuvre in a Downburst 383
13.7.2 Case Study: Flight in a Downburst 386
Summary 387
Bibliography 387
14 Thermo-Structural Performance . . . . . . . . . . . . . . . . . . . . . . . . . 392

14.1 Cold-Weather Operations 392
14.1.1 Aircraft Icing 394
14.2 Aviation Fuels 397
14.3 Fuel Temperature in Flight 400
14.4 Fuel-Temperature Model 402
14.4.1 Fuel-Vapour Model 404
14.4.2 Heat-Transfer Model 404
14.4.3 Numerical Solution 405
14.4.4 Numerical Solution and Verification 407
14.5 Tyre-Heating Model 409
14.5.1 Numerical Simulations 416
14.6 Jet Blast 418
xiv Contents
Summary 419
Bibliography 419
15 Mission Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423

15.1 Mission Profiles 423
15.1.1 Operational Parameters 425
15.2 Range-Payload Chart 426
15.2.1 Case Study: Range Sensitivity Analysis 429
15.2.2 Case Study: Payload-Range of the ATR72-500 430
15.2.3 Calculation of the Payload-Range Chart 430
15.3 Mission Analysis 432
15.3.1 Mission Range for Given Fuel and Payload 434
15.4 Mission Fuel for Given Range and Payload 435
15.4.1 Mission-Fuel Prediction 435
15.4.2 Mission-Fuel Iterations 436
15.5 Reserve Fuel 438
15.5.1 Redispatch Procedure 441
15.6 Take-Off Weight Limited by MLW 442
15.7 Mission Problems 443
15.7.1 Cruise with Intermediate Stop 443
15.7.2 Fuel Tankering 444
15.7.3 Equal-Time Point and Point-of-No-Return 446
15.8 Direct Operating Costs 448
15.9 Case Study: Aircraft and Route Selection 453
15.10 Case Study: Fuel Planning for Specified Range, B777-300 455
15.11 Case Study: Payload-Range Analysis of Float-Plane 460
15.11.1 Estimation of Floats Drag from Payload-Range Chart 460
15.12 Risk Analysis in Aircraft Performance 463
Summary 465
Bibliography 466
16 Aircraft Noise: Noise Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . 470

16.2 Definition of Sound and Noise 471
16.2.1 Integral Metrics: Effective Perceived Noise 472
16.2.2 Integral Metrics: Sound Exposure Level 475
16.3 Aircraft Noise Model 475
16.3.1 Polar-Emission Angle 477
16.4 Propulsive Noise 478
16.4.1 Noise-Propulsion System Interface 478
16.4.2 Fan and Compressor Noise 479
16.4.3 Combustor Noise 483
16.4.4 Turbine Noise 484
16.4.5 Single-Jet Noise 489
Contents xv
16.4.6 Co-Axial Jet Noise 491

16.4.7 Far-Field Noise from a Subsonic Circular Jet 493
16.4.8 Stone Jet Noise Model 494
16.4.9 Jet-Noise Shielding 501
16.5 APU Noise 508
16.6 Airframe Noise 509
16.6.1 Wing Noise 510
16.6.2 Landing-Gear Noise 512
16.7 Propeller Noise 516
16.7.1 Propeller’s Harmonic Noise 517
16.7.2 Propeller’s Broadband Noise 521
Summary 523
Bibliography 524
17 Aircraft Noise: Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533

17.1 Airframe Noise Shielding 533
17.2 Atmospheric Absorption of Noise 535
17.3 Ground Reflection 538
17.3.1 Ground Properties 541
17.3.2 Turbulence Effects 542
17.4 Wind and Temperature Gradient Effects 543
17.4.1 Numerical Solution 545
Summary 548
Bibliography 549
18 Aircraft Noise: Flight Trajectories . . . . . . . . . . . . . . . . . . . . . . . . 553

18.1 Aircraft Noise Certification 553
18.2 Noise-Abatement Procedures 560
18.2.1 Cumulative Noise Index 561
18.2.2 Noise-Program Flowchart 562
18.3 Flight-Mechanics Integration 564
18.3.1 Noise Data Handling 565
18.4 Noise Sensitivity Analysis 566
18.5 Case Study: Noise Trajectories of Jet Aircraft 568
18.6 Case Study: Noise Trajectories of Propeller Aircraft 570
18.7 Further Parametric Analysis of Noise Performance 572
18.8 Verification of the Aircraft-Noise Model 574
18.9 Noise Footprint 578
18.9.1 Noise Maps Refinement 580
18.10 Noise from Multiple Aircraft Movements 581
18.10.1 Noise Reduction and Its Limitations 584
Summary 584
Bibliography 585
xvi Contents
19 Environmental Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589

19.1 Aircraft Contrails 589
19.1.1 Cirrus Clouds 591
19.1.2 Cruise Altitude Flexibility 593
19.1.3 The Contrail Factor 595
19.1.4 Effects of Propulsive Efficiency 596
19.1.5 Heat Released in High Atmosphere 599
19.2 Radiative Forcing of Exhaust Emissions 599
19.3 Landing and Take-Off Emissions 600
19.4 Case Study: Carbon-Dioxide Emissions 604
19.5 The Perfect Flight 606
19.6 Emissions Trading 608
19.7 Other Aspects of Emissions 609
Summary 610
Bibliography 611
20 Epilogue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 614
Appendix A: Gulfstream G-550 . . . . . . . . . . . . . . . . . . . . . . . . . . 617
Appendix B: Certified Aircraft Noise Data . . . . . . . . . . . . . . . . . . . 622
Appendix C: Options for the FLIGHT Program . . . . . . . . . . . . . . . 624
Index 627
Tables
2.1 Cross-sectional areas of selected supercritical wing sections page 23

2.2 Volume breakdown of selected aircraft; all volumes in [m3 ] 33
2.3 Calculations of MAC for the Airbus A320-200 aircraft; graphs on
the same scale 35
2.4 Analysis of the geometry of the F4 aircraft model 37
2.5 Wetted-area breakdown for the selected aircraft (calculated). All
areas are in [m2 ]; ()* data are approximate 38
3.1 Payload data for very large aircraft; X is the range at maximum
payload 47
3.2 Standard passenger weights (rounded to full kg) 52
3.3 Fuel tanks of some Airbus airplanes. ACT = Additional Central
Tanks; Jet-A1 density at 15 ℃ = 0.804 kg/l 63
3.4 Weight breakdown of Airbus airplanes; mass in [kg] 73
3.5 Airplane mass properties at take-off-empty (no fuel) configuration
(calculated) 73
3.6 Coefficients of Equation 3.44 74
4.1 Profile drag sensitivity for the Airbus A380-861 resulting from
Awet = 2%. All drag coefficients are given as drag counts 105
4.2 Aircraft separation following ICAO rules 118
5.1 Power ratings for PW127 turboprop engine variants, sea level;
maximum temperatures as indicated 129
5.2 Turbofan-engine parameters used for flight and aircraft-noise
calculations 133
∗
5.3 Selected engine data for the CF6-80C2A3; data with an asterisk are
estimated 135
5.4 Typical APU fuel flow [kg/s], depending on load type and
atmospheric conditions 148
5.5 Estimated APU power and emission database 148
6.1 Design limitations of the Dowty propeller R391; w is the wind
direction 156
6.2 Some notable propellers and their applications 157
7.1 Stability derivatives for calculation of airplane response to
asymmetric thrust; model Boeing B747-100 190
xvii
xviii Tables
8.1 Sea-level data of the International Standard Atmosphere 196

8.2 Recognised international symbols for design air speeds and Mach
numbers 208
9.1 International symbols for take-off of a transport airplane 225
9.2 Delay in response time after activation for selected systems 240
9.3 Average rolling coefficient for some runway conditions 255
9.4 Estimated fuel burn during a taxi-out 261
10.1 Approximate limit speeds for selected commercial aircraft 275
10.2 Key events in the OEI take-off and go-around procedure 278
10.3 Climb report for the Airbus A320-200 with CFM56-5C4P turbofan
engines and 331-9 APU; standard day, no wind 280
10.4 Climb report for the case shown in Figure 10.6 285
10.5 Climb time and fuel for the flight paths shown in Figure 10.15 295
11.1 Flap and slat settings for the Airbus A320-200 305
11.2 Descent report for the A320-200, conventional descent 308
11.3 Descent report for the A320-200, continuous descent approach 308
11.4 Definition of landing speeds 318
11.5 Limit crosswind speeds coupled with runway conditions 321
12.1 Summary of subsonic cruise conditions, jet aircraft 345
12.2 SAR penalty due to non-optimal cruise altitude for some Airbus
airplanes 349
14.1 Characteristics of aviation fuels, at 15 ◦ C; data are averages 398
14.2 Characteristics of turbine fuels Jet-A and Jet-A1 398
15.1 Fuel use for mixed long- and short-range service of the Boeing
B777-300 (calculated) 444
15.2 Summary of parameters for DOC model 452
15.3 Calculated payload fuel efficiency for long-haul commercial flight 453
15.4 Operational data for mission analysis in case study 456
15.5 Summary of flight-planning analysis 457
15.6 Taxi-out report of fuel/weight-planning analysis 458
15.7 Take-off report of fuel/weight-planning analysis 458
15.8 Cruise report of fuel/weight-planning analysis 459
15.9 Basic performance data of model float-plane 460
15.10 Estimated floats’ dimensions 460
16.1 Summary of integral noise metrics 472
16.2 Polar directivity levels 485
16.3 Empirical constants for turbine acoustic power 486
16.4 Spectrum function for broadband noise 487
17.1 Numerical coefficients for Equation 17.7 537
17.2 Typical values for flow resistivity and inverse effective depth 542
18.1 Microphone positions for aircraft-noise measurements at London
Heathrow 557
18.2 Noise sensitivity matrix for a Boeing 777-300 for ± 2 dB on take-off
and landing trajectories (simulated data) 567
18.3 ATR72-500 noise trajectories; All noise levels are in dB 572
Tables xix
18.4 Calculated noise metrics (in dB) over a conventional and steep
landing trajectory at a FAR/ICAO landing point and point 1,000 m
upstream 573
19.1 ICAO flight modes, times and thrust rating as % of maximum thrust 602
19.2 LTO emissions summary for Airbus A320-200 with CFM56 engines 603
19.3 Analysis of a perfect flight with an Airbus A320-200 model 608
A.1 Weights and capacities of the G-550 618
A.2 Basic dimensions of the G-550 618
A.3 Operational limits of the G550 619
A.4 Selected data of the Rolls-Royce BR710 C4-11 gas-turbine engine 619
A.5 Landing gear of the G550 620
B.1 Certified noise levels for commercial aircraft 622
B.2 Certified noise levels for commercial aircraft (part 2) 623
Preface
This book is a derivative of an earlier textbook on flight performance. This new

work reflects my increased wisdom on the subject and represents an almost complete
departure from closed-form solutions that are traditionally taught in under-graduate
and post-graduate programs. Over the past several years, I have benefited from the
experience of teaching a flight performance course to senior engineers from industry,
government departments and academia. In the process, I learned a few new things
that now find a place somewhere in the book.
There is an increase in numerical methods in all fields of engineering; neverthe-
less, flight performance has remarkably resisted change. Some closed-form solutions
have been retained for those engineers who need a quick answer. The modern air-
plane is a complex engineering machine governed by systems, software and avionics.
Primitive methods are still widely used, which are then applied to aircraft design
and produce results of dubious accuracy that cannot be assessed. Worryingly, these
methods are used in most “conceptual design” and “multi-disciplinary optimisation”
methods. Now assume, more realistically, that you have been hired to provide flight
prediction tools to an airline operator or a manufacturer of engines or airframes, a
national or international aviation authority, an air traffic control organisation. Why
should they trust your performance software? What is the risk of under-predicting
the mission fuel for an intercontinental flight?
As we worried about conceptual design, the world has moved on. There is
increased emphasis on airplane evolution and upgrading, which is now reflected in
my thinking. At the same time, the environmental performance of the aircraft has
become very prominent. Therefore, part of this book is devoted to a wide spectrum
of environmental aspects of flight. My initial concerns have slowly shifted from noise
to engine emissions. Noise disappears as the aircraft moves away from the receiver,
although not many would like to agree. Exhaust gases remain with us for the next
few generations. In particular, aircraft condensation trails are there to remind us that
aviation is having a measurable impact on our skies. The lack of flexibility in aircraft
levels, stepped cruise and descent, and the use of holding patterns in congested air
space are all problems that need a solution in the coming years.
The book contains considerable advanced material across several disciplines,
including aircraft noise, environmental performance, airframe-propulsion integra-
tion, thermo-structural performance and flight mechanics. I am conscious of the
xxi
xxii Preface
audacity of the task I have undertaken, but I am confident that this work meets the
expectations of the aviation industry and the academic world.
I have developed some fully comprehensive flight codes. One code in particular,
FLIGHT, to simulate aircraft performance and mission analysis of transport aircraft,
contains most of the cross-disciplinary aspects of performance discussed in this book.
In its present form it consists of about 160 KLOCS (thousand lines of code). Other
codes discussed in the book include the propeller code, that is fully integrated with
FLIGHT, as well as a supersonic flight performance code (SFLIGHT). Several block
flowcharts have been included to help with the understanding of computer programs,
numerical models, system analysis and flight performance. The following material is
made available to readers:
r Computer code FLIGHT (demo version)
r Computer code Prop/FLIGHT (demo version)
r Computer code SFLIGHT (demo version)
r All charts and figures in any suitable graphical format
Separate technical documents will be issued to the readers wishing to work with
these computer models.
Dr. Z. Mohammed-Kassim, my long-time associate, has actively contributed to
the work on aircraft noise and to considerable code debugging. My doctoral student
Nicholas Bojdo took great care in reading some chapters. I am indebted to my editor,
Peter Gordon, who has been enthusiastic about my work from the beginning of the
project to the end. The editorial and production work was efficiently managed by
Peggy Rote at Aptara, Inc.
Finally, I thank my wife, Susan, for having the patience to tolerate my late nights
at the desk, especially when I reached the tunnel phase of my work, that is, when
I thought the book was finished but in fact there was no end in sight. A sabbatical
leave from the University has allowed me to step up my efforts. I am grateful to the
University, and the School, for the opportunity they have given me.
Nomenclature
Organisations
Below is a list of organisations that publish regularly documents (technical reports,
papers, journals, regulations) as well as more general information of aviation.
AAIB Air Accidents Investigation Branch, United Kingdom (www.aaib.gov.uk)

AIAA American Institute of Aeronautics & Astronautics (www.aiaa.org)
ANSI American National Standards Institute (www.ansi.org)
ASTM American Society for Testing and Materials (www.astm.org)
BTS Bureau of Transportation Statistics, USA (www.bts.gov)
CAA Civil Aviation Authority (www.caa.co.uk)
EASA European Aviation Safety Agency (www.easa.eu.int)
ESDU Engineering Data Unit (www.esdu.com)
FAA Federal Aviation Administration (www.faa.gov)
FSF Flight Safety Foundation (www.flightsafety.org)
IATA International Air Transport Association (www.iata.org)
ICAO International Civil Aviation Organisation (www.icao.int)
IPCC Inter-governmental Panel for Climate Change (www.ipcc.ch)
Jane’s Jane’s Information Systems (www.janes.com)
MIL Military Standards (www.mil-standards.com)
NASA National Administration for Space and Aeronautics (www.nasa.gov)
NATO Advisory Group, Aerospace Research & Development (www.rta.nato.int)
NATS National Air Traffic System, United Kingdom (www.nats.co.uk)
NTSB National Transportation Safety Board, United States (www.ntsb.gov)
RAeS The Royal Aeronautical Society (www.aerosociety.org)
SAE Society of Automotive Engineers (www.sae.org)
SAWE Society of Allied Weight Engineers (www.sawe.org)
Acronyms Used in This Book

ACT Additional Centre Tank
AEO All Engines Operating
AF Activity Factor
xxiii
xxiv Nomenclature
APU Auxiliary Power Unit

ASDA Accelerate-Stop Distance Available
ASI Air Speed Indicator
ASK Available Seat per Kilometre
ATC Air Traffic Control
AUW All-Up Weight
BFL Balanced Field Length
BPR By-pass Ratio
BRGW Brake-Release Gross Weight
CAS Calibrated Air Speed
CASK Cost per Available Seat per Kilometre
CDA Continuous Descent Approach
CG Centre of Gravity
CTOL Conventional Take-off and Landing
DOC Direct Operating Costs
DOCG Dry Operating Centre of Gravity
DOF Degree of Freedom
DOW Dry Operating Weight
EAS Equivalent Air Speed
EBF Externally Blown Flap
ECS Environmental Conditioning System
EGT Exhaust Gas Temperature
EPNdB Effective Perceived Noise, in dB
EPNL Effective Perceived Noise Level
ETOPS Extended Twin-Engine OPerationS
FADEC Full Authority Digital Engine Control
FCA Final Cruise Altitude
FCOM Flight Crew Operating Manual
FDR Flight Data Recorder
FL Fuselage Line; Flight Level
FLS Flight Level Separation
FMS Flight Management System
GPS Global Positioning System
GPU Ground Power Unit
GRW Gross Ramp Weight
GTOW Gross Take-off Weight
IAS Indicated Air Speed
ICA Initial Cruise Altitude
ICW Initial Cruise Weight
IDA Initial Descent Altitude
IGE In Ground Effect
ILS Instrument Landing System
ISA International Standard Atmosphere
KCAS Calibrated Air Speed in knots
KEAS Equivalent Air Speed in knots
KIAS Indicated Air Speed in knots
Nomenclature xxv
KTAS True Air Speed in knots

LRM Long-Range Mach number
MAC Mean Aerodynamic Chord
MBGW Maximum Brake-Release Weight
MCP Maximum Continuous Power
MEW Manufacturer’s Empty Weight
MIL Military Standards (USA)
MLW Maximum Landing Weight
MRM Maximum-Range Mach number
MRW Maximum Ramp Weight
MSP Maximum Structural Payload
MTOP Maximum Take-off Power
MTOW Maximum Take-off Weight
MZFW Maximum Zero-Fuel Weight
NADP Noise Abatement Departure Procedure
OASPL Overall Sound Pressure Level
OAT Outside Air Temperature
ODE Ordinary Differential Equation
OEI One Engine Inoperative
OEW Operating Empty Weight
OGE Out of Ground Effect
OPR Overall Pressure Ratio
PAX Passengers
PNL Perceived Noise Level
PNLT Perceived Noise Level, Tone Corrected
PWL One-third octave band Power Level
SAR Specific Air Range
SAT Static Air Temperature
SEL Sound Exposure Level
SEP Specific Excess Power
SFC Specific Fuel Consumption
SHP Shaft Horse Power
SI International Units System
S/L Sea Level
SPL Sound Pressure Level
STOL Short Take-off and Landing
TAS True Air Speed
TAT Total Air Temperature
TMA Terminal Manoeuvre Area
TOCG Take-off Centre of Gravity
TOD Top Of Descent
TODA Take-off Distance Available
TODR Take-Off Distance Required
TOGA Take-off and Go-Around
TORA Take-off Distance Required
TORR Take-Off Run Required
xxvi Nomenclature
TOW Take-off Weight

TSFC Thrust-Specific Fuel Consumption
ULD Unit Load Device
VMC Minimum Control Speed
VMCA Minimum Control Speed in Air
VMGC Minimum Control Speed on the Ground
VMO Maximum Operating Speed
VNE Velocity Not to Exceed
WAT Weight-Altitude-Temperature
WBM Weight and Balance Manual
ZFCG Zero-Fuel Centre of Gravity
ZFW Zero-Fuel Weight
The U.S. Department of Defense and NATO publish a dictionary of acronyms

and aviation jargon. A detailed list of symbols follows each chapter.
Technology Warning
This book makes reference to real flight vehicles in realistic flight conditions. The
data used to model these vehicles have been extracted, elaborated, interpolated or
otherwise inferred from documents available in the public domain. These documents
are either published by the manufacturer or the operators, or both. They are sup-
plemented with official data published by several aviation authorities at the national
and international level. Many of these documents are freely available to the public
in electronic format from the manufacturers, through their websites, or the websites
of their customers, or by third parties. No commercial, sensitive or restricted data
have been disclosed anywhere. All sources have been cited when appropriate. There
is no implication that the data refer to any particular aircraft owned or operated by
any organisation. The flight performance shown is often validated, but sometimes it
is not. Whenever figures or tables report the term “simulated” or “validated”, they
refer to simulations carried out with the comprehensive performance code FLIGHT
and its related software technology (available from the author).
Readers should be made aware that the statements made in this book are the
author’s own. Readers should use judgement before making technical, commercial,
military, marketing or business decisions. The author cannot take responsibility for
any action resulting in damage, accident or loss, as a consequence of statements
made in this book. None of the graphs, figures and tables shown in this book can be
used to make a final judgement on any airplane, any manufacturer, any flight, any
service or any design. Use of the graphs for flight planning is prohibited. If you are
in doubt, please consult the author, or use the performance codes from the aircraft
manufacturers.
xxvii
1 Prolegomena
Commercial aviation has grown to become the backbone of the modern transport
system. Demand for commercial air travel has grown exponentially since the 1960s.
The expansion of the aviation services is set to increase further. Even in the worst
moments of recession, air transport has only suffered a blip in its expansion. Our
global economy has become so dependent on air transport that any inconvenience
caused by weather and external factors can cause mayhem. To understand the size
of commercial aviation, reflect upon the fact that by the year 2000, there had been
35.14 million commercial departures worldwide, for a total of 18.14 million flight
hours 1 . In some countries, air transport accounts for one quarter of all transported
goods by value. It is estimated that there are about 50,000 airports and airfields
around the world 2 and 18,000 commercial airplanes flying every day. A number
of large airports are being constructed in some rapidly expanding regions. Mod-
ern airports are the size of a city: London Heathrow covers about 12 km2 ; this
is an area large enough to park about 1,800 Airbus A380s nose-to-tail. The sup-
port required by the aircraft is extraordinary and involves engineering, logistics,
integrated transport systems, security systems, energy and people. However, at the
heart of everything is the aircraft itself, interpreted as a flight system: this is the
subject of our book. There will be only a superficial mention of various external-
ities, including air traffic control, queueing models, stack patterns, logistics, supply
chains, and so on. More specifically, this book deals with the analysis, simulation and
prediction of aircraft flight performance at several levels, including aerodynamics,
weight performance, flight mechanics, aircraft noise and environmental emissions.
Advanced aircraft performance analysis involves at least the following engineering
activities:
r verification that an aircraft achieves its design targets

r efficient operation of existing aircraft or fleet
r selection of a new aircraft
r modification and upgrade of the flight envelope
r upgrading and extension of the mission profile
r aircraft and engines certification process
r environmental analysis, including emissions on the ground and in flight
1
2 Prolegomena
r aircraft noise emissions

r design of a new aircraft with and without optimisation methods
Note that in this list of specifications, only the last one involves the design of
a brand-new aircraft, something often referred to as “conceptual design”. Safety in
aviation dominates over everything else. Thus, health monitoring, adverse weather
events and human factors become an integral part of operating an aircraft and
maintaining safety records.
The methods used for the evaluation of aircraft performance are based on
theoretical analysis and flight testing. The latter method is made possible by accurate
measurement techniques, including navigation instruments.
Performance flight testing involves the calibration of instruments and static tests
on the ground, testing at all the important conditions, gathering of data, data analysis,
calibration with simulation models, determination of charts for the certification and
the flight manual.
Performance analysis is based on the elaboration of flight data, either from flight
testing or from flight data recorders. It is normally done at the operational level
by commercial airlines in the attempt to determine how the airplane performs over
time, in comparison with other airplanes and in comparison with the manufacturer’s
claims. The complexity of the modern vehicle and the variability of all external factors
contribute to changes in performance that often do not correspond to the technical
specifications. Most manufacturers have their own flight performance codes tuned
for their own aircraft.
Performance prediction is at the base of any concrete aircraft design and opera-
tion. The estimation of weights, range and power plant size requires the calculation
of basic aircraft performance from a few input data. In this case the approximation is
generally good enough for parameter estimation and design. A manufacturer claims
(typically) that: its airplane has 20% lower direct operating costs (including −30%
fuel consumption) than competitor X. How are we going to find out that this is the
case? Do we buy one of its airplanes and one from competitor X? Do we spend 1,000
hours in flight testing and then decide? In the field of aircraft performance there is
no laboratory experiment that can be used for verification.
1.1 Performance Parameters

Most of the aircraft flight parameters are stored by the Flight Data Recorder (FDR),
commonly called the black box (although its colour is often orange). The set of data
recorded is now standardised and contains all of the parameters that may be useful
for accident investigation, systems monitoring, flight trajectory analysis and engine
performance. As of August 2002, the National Transportation Safety Board (NTSB,
United States) requires the records of at least 88 flight parameters for a transport
category airplane. 3 These data include values for all the degrees of freedom of the
aircraft, velocities, acceleration rates, controls positions and engine state (tempera-
tures, pressures, fuel flow, rotation speeds), and quantities as different as computer
failure, autopilot engagement, icing, traffic alerts and collision-avoidance systems.
The requirements for large helicopters include records for at least 26 different flight
1.1 Performance Parameters 3
and state parameters. An additional set of 30 parameters is recommended though

not required.
Modern FDR, based on solid-state memory, can record several hundred para-
meters for up to 24 hours. They can withstand accelerations of hundreds of g-s,
temperatures in excess of 1,000◦ , and survive in ocean depths of 6,000 m, whilst
emitting a locator beacon for about 30 days. If ditched into the ocean, they can be
located by satellite.
One of the most challenging recoveries was that of an FDR from an Airbus A330
that disappeared off the coast of Brazil on 1 June 2009. The FDR was eventually
found intact, nearly two years later, 4,000 metres below sea level, after searching
10,000 km2 of sea floor.
International regulations mandate that commercial operators of aircraft heavier
than 27,000 kg have their FDR data monitored. In practice, it is required to analyse
the flight parameters, particularly in cases when problems of any nature may occur
(too steep descent rate, hard landing, flight through clear sky turbulence, and so
on). The scope of this analysis is at least twofold: 1) to guarantee safety, and 2) to
understand whether the aircraft is operated efficiently.
Many performance indexes cannot be simply expressed by a single value; they
are presented with charts. Some performance data are readily available from the
manufacturer, other data can be inferred by appropriate analysis, others are clouded
by secrecy or confidentiality, and others are difficult to interpret because the condi-
tions under which the aircraft performs are not provided. Among the most common
data covered by secrecy are the drag data, the stability characteristics, the excess
power diagrams and the engine performance. Other examples are 1) the aircraft
range, when the payload is not supplied together with the range, 2) the altitude at
which this range is achieved, and 3) the radius of action of a military interceptor.
This radius, in fact, may lie in the field of enemy fire.
The maximum take-off weight (MTOW) and the operating empty weight
(OEW) are available for most aircraft. However, these data are not sufficient to
calculate the maximum payload because the difference between MTOW and OEW
must include the mission fuel. A weight advantage compared to heavier rivals trans-
lates into significant revenue-earning advantages, which in a competitive market is
the most important factor for choosing and operating an aircraft. It is not uncommon
to find manufacturers unhappy that their performance data and charts are published
in the public domain. Performance charts allow customers and competitors to look
at various options, to select the most competitive aircraft and to discover the flaws
of the competitors’ technology.
The operator of an aircraft is concerned that the performance parameters quoted
by the manufacturer match the actual performance, and therefore accuracy of per-
formance prediction methods is essential. Performance data in the Flight Crew
Operating Manual (FCOM) are not always accurate for a variety of reasons: 1)
the FCOM data are often extrapolated from a limited set of flight test data, 2) the
actual flight conditions are different from the conditions in the FCOM, and 3) ageing
effects on the airframe and the propulsion system may lead to performance sensibly
different from new aircraft. One aircraft type may have several FCOM because each
airline may have its own version.
4 Prolegomena
1.2 Flight Optimisation

Flight optimisation is at the heart of design and operation of all modern aircraft.
From the operational point of view, commercial aviation is driven by fuel prices,
and operations at minimum fuel consumption are of great relevance. Performance
optimisation requires notions of optimal control theory, a subject unfamiliar to
the aerospace engineers. In the past 30 years these optimal conditions have been
increasingly challenged by environmental concerns, including noise emission, air
quality near airports, global climate change and sustainability.
Computer solutions of aircraft performance are now routine jobs and have
reached a phenomenal level of sophistication to include the coupling among flight
mechanics, aerodynamics, structural dynamics, flight system control and differential
game theory.
There are two key types of optimisation: Optimisation of the aircraft perform-
ance during the design phase, and optimisation of the operational performance for
the given airplane. In the former case, one can investigate the alternative changes
in configuration that improve one or more performance parameters. This is more
appropriately the subject of aircraft design. We will consider some cases of oper-
ational optimisation. An excellent source for optimisation problems with aircraft
applications is the classical book of Bryson and Ho on optimal control 4 and Ash-
ley 5 for a variety of flight mechanics problems. Today there are programs that plan
optimal trajectory routes to minimise direct operating costs (DOC), whilst comply-
ing with several airline constraints. These programs have several types of input data:
weather conditions, route, aerodynamics, aircraft performance and flight-specific
information, such as payload, fuel cost, and so on. On output they provide the
amount of fuel for optimal cruise altitude, climb and descent points, optimal cruise
speed and flight path.
The flight controls themselves have reached a phenomenal level of sophistica-
tion, with several on-board computers and substantial software; they now represent
a key aspect of the aircraft system. The best airplane cannot fly without its embed-
ded software that satisfies the most stringent requirements. According to recognised
standards – for example, NASA 6 – software controlling vehicles for human flight
must have the highest level of quality assurance and is defined as Class A. Soft-
ware failure in this class may cause a loss of life, or catastrophic mission failure. Next
comes Class B software, which is designed for non-human flight: unmanned vehicles,
rockets and satellites. Any software failure can lead to a total or partial loss of the
vehicle but no loss of life (partial mission failure).
Figure 1.1 shows the flight control panel of the Gulfstream G550, with its pecu-
liar four-screen view. The monitor display at the bottom centre of the photograph
allows rapid performance calculation, including take-off speeds from basic input
parameters, that can be quickly programmed by the pilot.
1.3 Certificate of Airworthiness

The certificate of airworthiness is a document that grants authorisation to operate
an aircraft. It specifies the type of operations; the limits of operation of the vehicle
in terms of weights, take-off and landing requirements; and a number of other
1.3 Certificate of Airworthiness 5
Figure 1.1. Cockpit of the Gulfstream G550 with Honeywell DU-1310 visual displays (author’s
photograph).
parameters, such as maintenance records, service history and compliance with safety
regulations.
The certificate proves that the aircraft conforms to the type specified and it is
safe to fly. The certificate is valid as long as the aircraft meets the type specification
(commercial, commuter, utility, and so on), it is operated safely and all the airworthi-
ness directives are met. The aircraft may lose its certificate for a number of reasons,
including modifications upgrades and new directives approved by the international
organisations that make the aircraft obsolete, not just unsafe to operate. Other
documents are generally required, such as the type certificate data sheet, the air-
worthiness limitations, the flight log book, the certificate of maintenance and a list of
other papers. The airworthiness limitations contain specific data on the number of
flight hours, years of service or number of cycles for critical components and systems.
These data can be used as limiters in flight simulation programs.
The type certificate contains various technical data, limitations and cautions, as
well as reference to the appropriate technical manuals (operating manuals, mainten-
ance manuals, and so on). Separate type certificates are issued for aircraft, propellers
and engines. These certificates are issued by various national and international avi-
ation organisations and are sometimes a duplication of effort in different countries.
These certificates, which can run to several pages, contain at least some of the follow-
ing data: airworthiness category, engines, engine limits, fuel, limit speeds, centre of
gravity range, maximum certified weights, auxiliary power units, equipment, seating
capacity, all weather capabilities and so on.*
* For example: European Aviation Safety Agency, Type Certificate Data Sheet, Airbus A380. TCDS
A.110, Issue 04, Feb. 2009.
6 Prolegomena
Certificates of airworthiness and the type certificates are issued by the Federal
Aviation Administration (FAA) in the United States, the European Aviation Safety
Agency (EASA) in Europe, the Civil Aviation Authority (CAA) in the United
Kingdom and other national and international bodies around the world. Certification
is a complex legal and technical matter that falls beyond the scope of this book.
Note that the certificates issued by different bodies may contain different types of
information. Therefore, it is sometimes useful to obtain the different certificates,
particularly when mining for engine performance data.
1.4 The Need for Upgrading

Most aircraft performance applications focus on performance for design, perform-
ance optimisation and aircraft sizing. It seems naive to forget that more than 18,000
commercial airliners fly around the world every day. The same airlines will be flying
for the next 20 years, as businesses try to recover their capital investments.
How many aircraft are actually sized every year? Most aircraft are likely to be
upgraded and modified to fit the changing market and technological advances. The
technology that is fitted over the years can be phenomenally different from the first
design. The service time of a single aircraft is on the order of 20 to 25 years, and the
life of an aircraft family may exceed 50 years. A single airplane program consists
of several versions, derivatives, design improvements, weight configurations, power
plants, avionics and systems.
A life-time career can be devoted to a single airplane. In the early days of
aviation, a new aircraft could roll out of the factory in a few months. In 1936, it took
just one year for the German aircraft designer Kurt Tank 7 to get from concept to first
flight of the Focke Wulf Condor Fw-190, the first long-range passenger (and later
reconnaissance and bomber) aircraft to fly from Berlin to New York without en-
route stop (in 1938). By the 1960s, commercial airplane design and testing required
thousands of man-years. The Boeing B747-100, that first flew in 1969 8;9 , required
15,000 hours of wind-tunnel testing and about 1,500 hours of flight testing with five
aircraft 1 .
The B747-400 incorporated major aerodynamic improvements, including a more
slender wing with winglets to reduce drag. A weight savings of approximately 2,270
kg was achieved in the wing by using new aluminium alloys. The version B747-400ER
has an increased take-off weight of 412,770 kg. This allows operators to fly about 410
nautical miles (∼760 km) farther or to carry up to an additional 6,800 kg payload,
for a range up to 14,200 km (∼7,660 n-miles). A larger version of this airplane, the
B747-800, weighing about 443 metric tons, is due to carry on the 747-flag for many
years to come.
An even older airplane is the Lockheed Hercules C-130A. Its first model was
delivered to the U.S. military in 1956. The design of this aircraft actually started
several years earlier. By the early 1960s, a VSTOL variant was designed 10 . Since
then, the aircraft has progressed through at least 60 different variants. The version
C-130J is actually a new airplane. Compared to the earlier popular version C-130E,
the maximum speed is increased by 21%, climb time is reduced by 50%, the cruising
altitude is 40% higher, the range is 40% longer and its Rolls-Royce AE-2100DE
engines generate 29% more thrust, while increasing fuel efficiency by 15%. With
1.5 Military Aircraft Requirements 7
new engines and new propellers, the C130-J can climb to 9,100 metres (∼28,000
feet) in 14 minutes.
Technological advances in aerodynamics, engines and structures can be applied
to existing aircraft to improve their performance. Over time weights grow, power
plants become more efficient and are replaced, aerodynamics is improved by optim-
isation, fuselages are stretched to accommodate more payload and additional fuel
tanks are added. Within the first few years of the Airbus A380 rolling out of the
factory for the first time, new landing gear brakes design saved the weight of one
passenger. At the same time, the wing was made about 350 kg lighter.
1.5 Military Aircraft Requirements

Military aviation has been a key aspect of national security since the very beginning
of airplane engineering and is the source of many technological innovations. The
fighter aircraft has evolved from a reconnaissance airplane of the First World War
to the most complex aircraft of the modern day. Von Kármán 11 reported that they
first flew over the battle fields of Europe to spy on enemy lines. Then enemy aircraft
wanted to prevent this from happening, and hence their pilots started shooting at
enemy aircraft with a pistol.
From reconnaissance to air strikes the step was short. The first recorded air
bombing is attributed to the Italian Air Force during its Turkish Wars in North
Africa in 1911–1912. The Italians operated a number of aircraft for reconnaissance
(Taube, Deperdussin, Bleriot) but then started shooting at enemy forces on the
ground. Finally, on 1 November 1911, they decided to drop four grenades weighing
about 1.5 kg each. Lieutenant Giulio Gavotti carried out the air strikes in a Taube
airplane† . Because the activation of the grenade required two hands and one hand
was needed to steer the unstable airplane, the weapons were activated by snatching
the plug with the teeth; then they were thrown overboard. Although the first bombs
did not kill anybody (the blast was absorbed by the desert sand), the news was
sensational. The Hague Convention (1899) prohibited the launching of projectiles
and explosives from balloons. The Italians argued that the Convention did not apply
to the powered airplane‡ . Thus, the event was heralded as the start of a new era in
warfare.
General Giulio Douhet was the first to distinguish among indiscriminate bomb-
ing, carpet bombing and strategic bombing (1912). In due course, the Germans
used Zeppelin airships for the indiscriminate bombing of London in 1914 (the first
urban bombing). The Hague Convention did not apply to them either. In response,
the newly established Royal Air Force (RAF) performed the first air bombing of
land targets with airplanes transported by ship near the theatre of war (Cuxhaven
raid).
The psychosis of aerial bombing was sparked in the United Kingdom by H.G.
Wells, who in 1907 published a science fiction book, The War in the Air, in which a
† Some sources report that the grenades were used for the first time on 24 October 1911 and landed
on the desert; the day after the first casualty was recorded.
‡ The Hague Convention, signed on 29 July 1899, entered into force on 4 September 1900. Chapter
IV, in fact, limited the use of explosives from the air for five years, starting in 1900, by balloon or
other new methods of similar nature.
8 Prolegomena
fleet of airships attacked and bombed New York City. The Americans were not
impressed but years later, science fiction turned into reality, with great loss of
life.
Following the operations in North Africa in 1911, the scope of military recon-
naissance was considerably expanded: now it was possible to observe from above,
to provide aerial photography and update and improve maps with unprecedented
details. By the start of the First World War, reconnaissance aircraft had two seats,
one for the pilot and the other for an observer – without which the pilot would have
had difficulty returning to base. The observer’s equipment included a map, a pistol,
a watch, a pair of binoculars, a one-way radio system and a life vest. Typically, the
useful load was around 400 kg, including pilot, observer and their equipment. If we
allow for about 180 kg for all this weight, there was about 220 kg left, to be shared
between fuel and ordnance. The fuel flow was of the order of 32 kg per flight hour.
Without ordnance, an airplane would have been capable of an endurance of 6.5
hours. At an average speed of 100–120 km/h (∼50–60 kt), the radius of action would
have been 350 km (∼190 n-miles) in the most optimistic scenario.
In 1915, during the early days of the war, the Dutchman Anton Fokker 12 , work-
ing at the service of the German Army, invented a system that synchronised the
shooting of a machine gun through the propeller (interrupter gear) – mounted on a
single-seater monoplane. With the interrupter, pilots had hands free to manoeuvre
and fight at the same time. Occasionally the interrupter malfunctioned and claimed
the life of such accomplished pilots as Max Immelmann.
By the end of the Great War, the European powers had thousands of aircraft
at their disposal. It has been estimated that the total production of such aircraft
exceeded a staggering 75,000 units in four years, 32,000 of which were British; note
that only 15 years had passed since the invention of the airplane.
To follow the history of the development of the military aircraft means following
the development of key aeronautical technologies over the past 100 years. Two rel-
evant books on the birth of military airplanes are Driver 13 (on the British aviation)
and Opolycke 14 (on the French aviation). Stevens 15 and Weyl 16 report more gen-
eric historical details. Jackson 17 published a chronology of events recording aerial
warfare from the very beginning to the present.
The requirements for fighter aircraft now include multi-purpose missions, air-
craft with complex flight envelopes, several configurations (changeable in flight),
supersonic flight, combat capabilities, delivery of a wide range of weapon systems,
manoeuvrability, all-weather and night-and-day operations. The aircraft has become
a platform system of phenomenal complexity and cost. Yet history deserves another
revolutionary weapon system: the military aircraft of the future is unmanned. Nev-
ertheless, there will be a few decades before this happens. In the meantime, there
can be problems of excess capacity coupled with excessive costs, which in turn make
the military aircraft difficult to sell, operate and upgrade.
There are dozens of different mission scenarios 18 . Typical missions are basic,
assault, combat, retrieval, close support, transport, refuel and reconnaissance. For
each of these missions there is a specific take-off weight, mission fuel, payload, range,
maximum rate of climb and service ceiling. This field is now so advanced that engin-
eers use differential game theory and artificial intelligence to study the effectiveness
of a given aircraft and the tactical manoeuvrability to incoming threats 19 .
1.6 Review of Comprehensive Performance Programs 9
1.6 Review of Comprehensive Performance Programs

This review is limited to computer models that are documented in the open liter-
ature 20 . Both Airbus and Boeing have their own flight performance programs that
model their own aircraft by using a combination of first principles and extensive flight
data. These programs are not in the public domain. Versions of these programs are
provided to airline operators in order to facilitate their performance analysis and
their flight planning.
Computerised flight planning goes back to the late 1960s. One of the first
attempts was due to Simpson et al. 21 , after it was recognised that the optimal route
selection across the North Atlantic could lead to considerable fuel savings. The
method required to take into account the actual state of the atmosphere (specific-
ally, winds and temperatures). The analysis required some basic performance data
for the aircraft, such as climb and descent programs, cruise altitude, fuel consump-
tion and other parameters. Climate-optimal routing problems have remained topical
to this day 22 .
Roskam’s Advanced Aircraft Analysis (AAA) modules are based on Roskam’s
books 23 . These performance modules focus on aircraft design, from weight sizing to
aerodynamic prediction, control and stability analysis. Another code in this techno-
logy area is ACSYNT, whose origins go back to the 1970s. In recent years this code
has undergone major development to adapt it to aircraft conceptual design. The
code integrates various disciplines, including performance, design, costs, noise and
engineering process 24;25 . A flight optimisation system called FLOPS was developed
in the 1980s 26 to address detailed performance during preliminary design.
The program DATCOM 27 provides calculations on static stability, high-lift per-
formance, aerodynamic derivatives and trim conditions of the aircraft at subsonic
and supersonic speeds. The program has been used extensively for the rapid estim-
ation of the static and dynamic characteristics of high-performance aircraft in the
preliminary design stage. The approach followed by this method is accurate enough
for several applications 28 .
A comprehensive performance simulation that is used for the air-traffic man-
agement is the so-called BADA Model, developed at Eurocontrol. BADA uses the
lumped-mass approximation, a total energy model for the centre of gravity and a
basic performance model for the prediction of the aircraft trajectory 29;30 . Its main
application is the prediction of flight trajectories in terminal-area manoeuvre and
for a management of traffic at the current conditions and at forecast growth. The
basic equations used by BADA are ordinary differential equations for the centre of
gravity and the total energy equation (i.e., balance among kinetic energy, potential
energy and work done by the engines). Additionally, the model uses integral values
of the essential parameters of the aircraft, such as operational limits (from the flight
manual or other sources).
The total energy concept has been used 31 to predict the fuel consumption of
transport aircraft, without having to rely on statistical databases. This approach has
already been used in the 1980s by the FAA to provide a better automatic flight
planning for air traffic control. In principle, the method is quite powerful because
it does not require many details of the aircraft. In fact, it relies on the fundamental
energy balance, on the gross weight and on the path profile of the aircraft.
10 Prolegomena
ESDU 32 provides a suite of programs to calculate the performance of fixed-wing

aircraft, including flight performance, airfield performance and mission performance
(block fuel for a transport aircraft and radius of action for a military aircraft). The
program consists of an implementation of several derivations published as data units,
some of which are briefly discussed and cited in this book.
A comprehensive performance program of industry standards is PIANO 33 . This
program includes preliminary design options, a large database of aircraft models and
a detailed mission-performance analysis module. The program does a wide range
of performance calculations and it is shown to match closely the manufacturer’s
performance, although the technical details are not disclosed.
1.7 The Scope of This Book

The subject of this book is the operational performance of aircraft, with all of the
issues surrounding technology evolution (aircraft and engine upgrades) as well as the
environmental impact. In this context, operational performance falls between con-
ceptual design of aircraft and airport operations, which rely on limited information
on the aircraft.
We present relatively advanced cross-disciplinary topics. We aim to solve accur-
ately problems such as the best cruise altitude, the best cruise Mach number, and
the best climb and descent procedures. We deal with commercial aircraft powered
by either high by-pass turbofan engines or propellers-turboprop. We also discuss
high-performance aircraft (mostly for military use) that are capable of supersonic
flight. These topics are mapped into a computational framework that consists of a
number of comprehensive computer programs. One such program should be able to
accomplish at least some of the following tasks:
r trajectory optimisation and route planning

r mission analysis and field performance
r environmental emissions and fuel costs
r airframe–engine integration
r thermo-physics and structural dynamics
r aircraft noise trajectories
r noise impact around airports
r systems analysis
r verification of performance data
r competition analysis
r trade-off and parametric studies
A flowchart of the computer program is shown in Figure 1.2. This chart shows
the key aspects of the sub-models of the performance framework, as well as the
major items and sub-items of engineering analysis. Each of these items is discussed
in the following chapters.
The flowchart for a supersonic high-performance aircraft is similar to
Figure 1.2. However, the noise and emissions are not simulated, and there is more
1.7 The Scope of This Book 11
Figure 1.2. Flowchart of a comprehensive flight performance simulation code.
emphasis on the aerodynamics, which must contain models for transonics, super-
sonics and external stores. The thermo-structural models are limited to the aero-
thermodynamic heating at high supersonic speeds. The propeller model does not
apply to this type of airplane.
The propeller performance mind-map is shown in Figure 1.3. This model can
be fully integrated into the flowchart of the turboprop aircraft. The optimal design
block relies on the aerodynamics and geometry module and is not part of routine
performance calculations.
The key sub-topics addressed in this book are as follows:
1. Development of an accurate models of airplanes, based on partial information

available from the manufacturer. This model is validated against a number of
different metrics, including areas, volumes, dimensions and reciprocal positions
(Chapter 2).
2. Development of an accurate model for mass, inertias, centre of gravity of the
airplane, along with location of the mass of each sub-system (Chapter 3).
3. Development of an accurate and rapid model for aerodynamic performance,
including a variety of aerodynamic coefficients and derivatives at all flight con-
ditions (Chapter 4).
4. Development of an accurate performance model for high by-pass turbofan
engines, low by-pass turbojet engines for military use and turboprop engines.
The models are designed to provide at least 20 different engine parameters over
the full flight envelope (Chapter 5).
12 Prolegomena
Figure 1.3. Flowchart of a comprehensive propeller simulation code.
5. Development of an accurate propeller model, based on partial information of

its geometry, and on a specific aerodynamic model that simulates both axial and
yawed flow conditions over the full flight envelope (Chapter 6).
6. Development of a static stability model for both longitudinal and lateral sta-
bility, including the determination of the optimal centre of gravity and the
control required to establish equilibrium of forces and moments to the aircraft
(Chapter 7). This chapter does not deal with unsteady response and short-time
period of motion, which is a subject of flight dynamics.
All of the sub-models described previously are integrated into a flight mechanics
framework that is capable of simulating all of the flight conditions of a commercial
aircraft, as well as some selected conditions of a supersonic aircraft. Thus, there is
a separate presentation of ground operations, with focus on take-off (Chapter 9);
climb to initial cruise altitude (Chapter 10); descent, final approach and landing
(Chapter 11); and optimal and sub-optimal cruise procedures (Chapter 12). Chapter 8
deals with flight envelopes, atmospheric models and some flight performance at
supersonic Mach numbers, whilst Chapter 13 deals with turn and other manoeuvres
in the horizontal and vertical plane, powered and unpowered, in still air and in
downburst situations.
A number of special topics are presented in Chapter 14, which deals with thermo-
structural performance. The items discussed include in-flight icing, fuel-temperature
simulation, tyre heating and forcing at take-off and landing, as well as jet blast and
jet plume diffusion.
1.8 Comprehensive Programs in This Book 13
All of these models are applied to the simulation of commercial missions in

Chapter 15, where we deal with a number of diverse topics, such as fuel plan-
ning, optimal flight trajectories, payload-range analysis and other special cases. This
chapter also deals with risk analysis in aircraft performance.
The final chapters of the book deal exclusively with environmental performance.
Specifically, we have:
1. Development of sufficiently detailed sub-models for aircraft noise sources, for

the propulsion system (including engine, propeller and auxiliary power unit
[APU]) and the airframe (including landing gear) (Chapter 16).
2. Development of sub-models for noise interference, propagation, ground effect
and other wall effects (Chapter 17).
3. Mapping of noise models into a noise-flight-mechanics framework for the predic-
tion of noise signature on the ground for standard and arbitrary flight trajectories
(Chapter 18).
4. Finally, discussion of methods for the prediction of the main environmental
emissions from commercial aircraft, including carbon-dioxide, landing-and-
take-off (LTO) emissions, contrail avoidance strategies and optimal free flights
(Chapter 19).
Most computational procedures are described, although some of them require
considerable effort to be translated into a computer program.
1.8 Comprehensive Programs in This Book

Multi-disciplinary flight code validation in the field of fixed-wing aircraft is not
a well-developed discipline; the discipline, in fact, has not even been established.
This gap is in sharp contrast with the comprehensive aero-mechanics codes widely
used in rotorcraft engineering. Minimum validation requirements are required for
configuration, aerodynamic models, propulsion systems, flight mechanics integration
and aircraft noise. These major blocks are shown in Figure 1.4. At a deeper level,
we may be interested in various aspects of thermo-structural performance, as also
shown in Figure 1.2.
Within the configuration block, we address the geometry of the airplane, wetted
areas (§ 2.4), surface matching through the analysis of the cross-sectional areas
(§ 2.4.6), the mass distribution and the centre of gravity position (§ 3.8.2).
Within the aerodynamics block, we must verify the accuracy of the aerodynamic
coefficients, the profile drag, the aerodynamic polars, the transonic effects, and the
aerodynamic derivatives. With reference to the latter parameter we need some
indirect types of tests, such as the calculation of the control speed in air (§ 7.2).
Within the propulsion block, we need to address at least basic parameters, such
as fuel flows, turbine and combustor temperatures, rated thrust and power. If the
engine is powered by propellers, a separate set of tests must be carried out on the
propeller in isolation and with integrated mechanics (Chapter 6).
At the flight-mechanics level, we need to address the capability of the airplane
and compare it with official data published in the FCOM, in the type certificate and in
other documents. Typical examples of verification include the determination of the
14 Prolegomena
Wetted areas Pay-Range
Cross-S. area Ch. 2 Ch. 10-15 Envelopes
Configuration Flight Mech. WAT Charts

CG position
Cruise Mach
SAR
Validation
Noise
Ch. 16-18
Ch. 4 Aerodynamics Propulsion
VMCA Aero-Derivs. Engines Ch. 5

Aero Coeffs. Propellers Ch. 6
Polars
Figure 1.4. Flowchart of a verification and validation strategy for a comprehensive flight
performance code.
payload-range chart (§ 15.2), the flight envelope in the Mach-altitude plane (§ 8.6),
the weight-altitude-temperature charts for take-off performance, the optimal cruise
Mach numbers (§ 12.3) and cruise altitude (§ 12.7), and the specific air range.
Finally, within the aircraft noise block, there are several tests that must be
carried out at several levels: noise sources at the component level; noise sources
at the integration level; and noise propagation, scattering, reflection, atmospheric
effects and other effects arising from the operation of the airplane (Chapters 16–18).
Bibliography
[1] The Boeing Corporation, 2004.
[2] CIA Factbook. Continuously updated (www.cia.gov).
[3] NTSB. Flight Recorder Handbook for Aviation Accident Investigation, Office
of Research & Engineering, Washington DC, Dec. 2002.
[4] Bryson AE and Ho YC. Applied Optimal Control. Blaisdell, NY, 1969.
[5] Ashley H. Engineering Analysis of Flight Vehicles. Addison-Wesley, Reading,
MA, 1974.
[6] NASA. Software assurance standard. NASA Technical Standard STD-8739.8,
July 2004.
[7] Wagner W. The History of German Aviation: Kurt Tank, Focke Wulf’s Designer
and Chief Pilot. Schiffer Publ. Ltd, 1999.
[8] Sutter JF and Anderson CH. The Boeing model 747. J. Aircraft, 4(5):452–456,
Sept. 1967.
[9] Lynn Olason M. Performance and economic design of the 747 family of air-
planes. J. Aircraft, 6(6):520–524, 1969.
[10] Danby T, Garrand WC, Ryle DM, and Sullivan LJ. V/STOL development of
the C-130 Hercules. J. Aircraft, 1(5):242–252, 1964.
[11] von Kármán T. The Wind and Beyond: Theodore von Kármán Pioneer in
Aviation and Pathfinder in Space. Little Brown & Co., Boston, 1967.
Bibliography 15
[12] Dierikx M. Fokker: A Transatlantic Biography. Airlife Publishing Ltd, Shrews-

bury, UK, 1997. Chapter 2.
[13] Driver H. The Birth of Military Aviation: Britain 1903–1914. The Royal Histor-
ical Society, 1997. ISBN 0 86193 234 X.
[14] Opolycke LE. French Aeroplanes Before the Great War. Schiffer Publishing
Ltd, Agden, PA, 1999.
[15] Stevens JH. The Shape of the Aeroplane. Hutchinson & Co. Ltd, London, 1953.
[16] Weyl AR. Fokker: The Creative Years. Putnam & Sons, London, 1965 (reprint
1987).
[17] Jackson R. The Guinness Book of Air Warfare. Guinness, 1993.
[18] Gallagher GL, Higgins LB, Khinoo LA, and Pierce PW. Naval Test Pilot School
Flight Test Manual – Fixed Wing Performance, volume USNTPS-FTM-No. 108.
U.S. Navy Pilot School, Sept. 1992.
[19] Isaacs R. Differential Games: A Mathematical Theory with Applications to War-
fare, Pursuit, Control and Optimisation. John Wiley & Sons, 1965.
[20] Filippone A. Comprehensive analysis of transport aircraft flight performance.
Progress Aero Sciences, 44(3):185–197, April 2008.
[21] Simpson L, Bashioum DL, and Carr EE. Automated flight planning over the
North Atlantic. J. Aircraft, 2(4):337–346, 1965.
[22] Irvine E, Hoskins B, Shine K, Lunnon R, and Froemming C. Characterizing
North Atlantic weather patterns for climate-optimal aircraft routing. Meteoro-
logical Applications, 2012. DOI:10.1002/met1291.
[23] Roskam J. Design for minimum fuselage drag. J. Aircraft, 13(8):639–640, 1976.
[24] Jayaram S and Myklebust A. ACSYNT – A standards-based system for para-
metric computer aided conceptual design of aircraft. AIAA Paper 92-1268, Jan.
1992.
[25] Rivera F and Jayaram S. An object-oriented method for the definition of mis-
sion profiles for aircraft design. AIAA Paper 1994-867, Reno, NV, Jan. 1994.
[26] McCullers LA. Aircraft configuration optimization including optimized flight
profiles. In Experiences in Multidisciplinary Analysis and Optimization, NASA
CP-2327, pages 394–412. Jan 1984.
[27] Williams JE and Vukelich SP. The USAF stability and control digital DAT-
COM. Technical Report AFFDL-TR-79-3032, Vol. I, Air Force Flight Direct-
orate Laboratory, April 1979.
[28] Sooy TJ and Schmidt RZ. Aerodynmic predictions, comparisons and valida-
tions using missile DATCOM (97) and Aeroprediction (AP98). J. Spacecraft
& Rockets, 42(2):257–265, 2005.
[29] Nuic A. User Manual for the Base of Aircraft Data (BADA) – Revision 3.6.
Eurocontrol Experimental Centre, Bretigny-sur-Orge, France, July 2004. Note
10/04.
[30] Nuic A, Chantal P, Iagaru MG, Gallo E, Navarro FA, and Querejeta C.
Advanced aircraft performance modeling for ATM: Enhancements to the
BADA model. In Proceedings of the 24th Digital Avionics Systems Confer-
ence, Washington, DC, 30 Oct. to 3 Nov. 2005.
[31] Collins BP. Estimation of aircraft fuel consumption. J. Aircraft, 19(11):969–975,
Nov. 1982.
[32] ESDU. Aircraft Performance Program. Part 1: Introduction to the Computer
Programs for Aircraft Performance Evaluation. 00031. ESDU International,
London, Nov. 2006.
[33] Simos D. PIANO: A tool for preliminary design, competitor evaluation, per-
formance analysis. In ICAO Committee on Aviation Environmental Protection
Working Group 2 Meeting, Rome, Italy, May 2006.
2 Aircraft Models
Overview
In this chapter, we present methods for the determination of the geometrical config-
uration of a modern transport aircraft (§ 2.1). We use the concept of control points
and geometric wire-frames (§ 2.2). Thus, we establish a framework for the determin-
ation of linear dimensions, position of centroids, reference areas and volumes. The
methods shows are valid for any other type of aircraft. We use a stochastic approach
for more complex surfaces (§ 2.2.1). Section 2.3 deals with the geometry of the lifting
surfaces. We then show a method for the determination of the wetted areas (§ 2.4),
aircraft volumes (§ 2.5), including the wing tanks, and the mean aerodynamic chord
(§ 2.6). We produce a few examples of verification of aircraft configurations (§ 2.7).
Finally, we define some reference systems on the aircraft (§ 2.8).
KEY CONCEPTS: Transport Aircraft, Geometry Definition, Wetted Areas, Aircraft

Volumes, Fuel Tanks, Mean Aerodynamic Chord, Reference Systems.
2.1 Model for Transport Aircraft

The aircraft geometry represents the basis of all of the aerodynamic calculations, fuel
and payload capacity, centre of gravity position, stability and control requirements.
The construction of an accurate aircraft model is done in a number of steps that
require the choice of reference systems, the determination of the key sub-systems,
and an assembly procedure to determine the geometric relationships between parts
of the vehicle. Various reference systems are introduced to define geometrical rela-
tionships and operational conditions in flight. The procedures shown in this chapter
are required to calculate reference areas, reference lengths and reference positions
for a wide variety of calculations.
The aircraft is separated into clearly identified components, as shown in
Figure 2.1. There are the fuselage and its sub-systems, the wing and its sub-systems,
the tail surfaces and their sub-systems, the landing gear and its sub-systems, the
propulsion system and the auxiliary units, the landing gear groups and the wheels.
About two dozen different components are identified and then reconstructed. These
sub-systems are mapped into the comprehensive flowchart in Figure 2.2: the aircraft
16
Figure 2.1. CAD model showing assembly of aircraft systems, from Ref. 1 .
is decomposed into five major systems; each sub-system is decomposed into several
major sub-systems.
Each component must be considered in order to have a sufficiently accurate rep-
resentation of the aircraft. The components are assigned an identity card, a sequence
of control parameters, and a set of rules to manipulate the control parameters. The
reconstruction of the aircraft is done through the information read from the card
and the reconstruction algorithms, as shown in Figure 2.3. The model is a function
of the state parameters x = {x1 , x2 , · · · xn }, some of which are time-dependent.
The state parameters are divided into separate categories:
r Geometry: control points, shape parameters, angle settings
r Assembly: sub-systems and complete model
r Systems: power-plant, fuel systems, APU, landing gears
CONTROL POINTS. The whole procedure is based on a database of control points.

Each card corresponds to a component and a view (top, side and front). For example,
in this model, there is a wing-top, wing-side, and wing-front card. This
method can be generalised further to include the determination of the geometrical
configuration via a recognised computer-aided design (CAD) standard, such as DXF
and IGES. An example of a control-points file is reported in § A.0.1. The use of
CAD formats, whilst useful in some instances for illustration purposes (for example,
Figure 2.1), has limited value in this case.
Reliance on the few data published by the manufacturers (for example, wing
area, wing span) is not sufficient. Most of the aircraft drawings in the open domain
are for illustration purposes only; by themselves, they offer little accuracy. Further
18 Aircraft Models
Wing Engines
Fuel Tanks Pylons
Controls Wing Unit Propulsion

Nacelles
APU
Flaps
Propellers
Slats
Airplane
Ailerons Fuselage
Fore-body
Spoilers
Center
Winglets
Tail Surfaces Landing Gear After-body
H-Tail Seating
V-Tail
Cargo
Elevator Nose L-Gear
Fuel Tanks
Rudder Main L-Gear
Figure 2.2. Flowchart of aircraft systems, mapped from Figure 2.1.
checks are required to validate all of the data extracted. For example, the longitudinal
position of the control points is derived by using as a reference distance the fuselage
length or the overall aircraft length. On this scale, connecting two points having a
specified distance (for example, nose-to-engine-face) sometimes leads to a distance
that does not match the value specified by the manufacturer.
REFERENCE SYSTEM. It is necessary to define one reference system to calculate the

positions of each element. These reference systems will also be useful to calculate
the mass distribution, the position of the centre of gravity and the moments of
inertias (Chapter 3). All of the longitudinal positions (x) are calculated with respect
Ref. Points
Card 1
x ,y, z, ... Rule 1
Linear Dim.
Card 2
x ,y, z, ... Rule 2 Plan areas
Geometry
Wetted areas
Card n Volumes
x ,y, z, ... Rule n
Centroids
Figure 2.3. Construction rules for the geometry model.

Figure 2.4. Reconstruction of the aircraft geometry from the top view (selected control points
shown).
to the fuselage nose. All of the lateral positions (y) are calculated with respect to
the vertical symmetry of the aircraft. The vertical positions (z) are calculated either
with respect to the nose or to the ground, although the latter choice should account
for the possibility there is a small pitch-down configuration when the aircraft is on
the ground. The centre of this system has coordinates O = {0, 0, 0}; O is the nose
point.
Data provided by the manufacturers often include the range of some key dimen-
sions. Therefore, it is possible to find data that refer to an empty aircraft and an
aircraft at maximum ramp weight; in the latter case, it is possible to verify that some
quantities depend on the position of the centre of gravity.
Starting from the top view of the aircraft (Figure 2.4), each of the visible sub-
systems is defined by a sequence of points. The same is done from the side and front
views, Figures 2.5 and 2.6. The accuracy can be improved by zooming in to the items
of interest. The error increases as a component gets smaller, unless we are able to
zoom in to the area of interest.
Figure 2.5. Reconstruction of the aircraft geometry from the side view (selected control points
shown).
20 Aircraft Models
Figure 2.6. Reconstruction of the aircraft geometry from the front view (selected control
points shown).
2.2 Wire-Frame Definitions

The determination of the wire-frames of each aircraft component is done through
the rules shown in Figure 2.3, which allow the reconstruction of planforms and other
well-recognised geometrical shapes.
WING PLANFORM. A top view of the wing is defined through a set of control points
required to reconstruct the planform, the nominal span, the chord at the root and the
tip, the chord distribution and the sweep angles. A frontal view is used to reconstruct
the dihedral and the nominal height of the tip above the ground. A side view is not
useful for deriving any wing parameter, due to the presence of the nacelles and the
pylons.
The wing thickness and the twist distribution cannot be extracted from the draw-
ings; these parameters represent a rough estimate. For the thickness distribution, we
may have to refer to statistical data for known and similar aircraft. For the twist dis-
tribution, we can use aerodynamic analysis to derive an optimum planform geometry
that yields minimum induced drag (Chapter 4).
Figure 2.7 shows the wire-frame of the wing of the Boeing B737-900W (with
winglets). The dashed lines denote the approximate projection of the winglets onto
the horizontal plane. The wing span is better defined as “over-the-winglet”, that
is, with the outermost point of the wing at nominal configuration. A loaded wing
would bend downwards (whilst on the ground) or upwards (whilst in flight); hence
the actual span changes depending on the wing loading.
WING AREA. Some conventions are available for the definition of the wing area, but
there is not a universal agreement. The options are 1) exposed wing area plus the
fuselage section having a length equal to the root chord; 2) trapezoidal area plus
exposed glove area, plus the covered glove area; 3) exposed wing area, plus area
resulting from extension of the leading- and trailing-edge lines through the fuselage
centreline; and 4) exposed wing area, with the actual span and tip chord. Note that
none of these choices is a restriction or a limitation. Once a reference area is agreed,
all of the aerodynamic parameters will be referred to that area. We apply only the
first definition.
The wing span of very large aircraft is not uniquely identified, especially if the
wing is equipped with winglets. For example, at maximum fuel load, the wing span
2.2 Wire-Frame Definitions 21
20
x, m
25
span over winglet
0 5 10 15
y, m
Figure 2.7. Wire-frame of the Boeing B737-900W wing.
of the Boeing B747-400 over the winglets is ∼0.5 metres longer than the nominal
value jig position. Therefore, a decision is needed to define uniquely the“wing span”,
something that is not possible to do by inspecting line drawings. The wing area is
calculated with the methods explained in § 2.2.1 and is corrected for dihedral effects
with the following formula:
Axy
A= (2.1)
cos ϕ
where ϕ is the dihedral angle and Axy is the projection of the planform on the
horizontal plane.
TAIL SURFACES. The tail surfaces (horizontal and vertical stabilisers) are determined
with the same method as the wing. Figure 2.8a shows the wire-frame of the vertical
tail of a model Boeing B737-900, including the rudder. The full geometry is given by
the solid line. The dashed lines are the simplified wire-frame, with numbered control
points. The wetted area and the centroid are calculated for the full geometry. The
leading- and trailing-edge sweep angles, as well as the chord lengths and the mean
aerodynamic chord, are calculated on the wire-frame.
Figure 2.8b shows the wire-frame of the horizontal tail of the Boeing B737-900,
including the elevator. As in the case of the vertical tail, the rounding of some corners
makes the definition of some key parameters (tip chord, leading edge sweep, mean
aerodynamic chord) somewhat difficult. In both cases, the leading-edge sweep is
calculated from the control points 1 and 2; the trailing-edge sweep is calculated from
the control points 3 and 4.
2.2.1 Stochastic Method for Reference Areas

An essential aspect of our model is the determination of a general method for
calculating surface areas, position of reference points (leading-edges, centroids),
equivalent diameters and form factors. If the shape of the surface is trapezoidal, the
calculation of the geometry is straightforward. If there are five or six control points, a
22 Aircraft Models
2 3
2 3
12 6
10
4
z, m
y, m
8
2
6
1
1 4
4
4 0
32 34 36 38 40 36 38 40 42
x, m x, m
(a) V-tail (b) H-tail
Figure 2.8. Vertical tail of the Boeing B737-900.
convention is required to specify what these points denote. If the geometry is defined
by several points, the calculation of the area is done with a stochastic method. The
idea is explained in the following procedure:
1. Select the n points x, y that define the shape in a generic plane.

2. Define a closed curve with points x, y, with x1 = xn , y1 = yn .
3. Calculate the bounding box [dx, dy] and the bounding-box area Abbox .
4. Generate a random point xr , yr within the bounding box.
5. Check whether the random point xr , yr is inside or outside the curve.
6. Update the counter ki for points inside the curve.
7. After k shots, the current area Ak is defined by:
ki
Ak = Abbox . (2.2)
k
8. Check the convergence by using the criterion:
Ak − Ak−1
< , (2.3)
Ak−1
where is a tolerance that defines the relative change in area, typically = 10−4 .
9. If the convergence is not satisfied, continue from point 4.
The centroids are calculated from the average of the coordinates of internal
points. This method is particularly useful for the calculation of the fuselage nose
area, cross-sectional area of non-circular fuselages and the area of a wing section.
However, it is also useful for more complex wing shapes (including the filleting at
the leading and trailing edge). A typical result of this numerical model is shown in
Figure 2.9. This example is trivial because the wing is defined by only five points,
and the method would not be used in this case. A few thousand shots are required to
get the area within the specified tolerance, but the procedure will always converge.
The key aspect of this procedure is the algorithm that decides whether a point falls
inside or outside the contour.
2.3 Wing Sections 23
Bounding box
20
x, m cr
25
30
span
0 5 10 15 20 25
y, m
Figure 2.9. Stochastic calculation of wing area, aircraft shown in Figure 2.4.
2.3 Wing Sections

The exact geometry of the wing section of a modern transport aircraft is rarely
known. However, we know that in most cases these sections are supercritical airfoils.
If the spanwise thickness distribution t/c is available, then it is possible to produce
a model that is sufficiently accurate for performance calculations. Table 2.1 is a
summary of geometrical characteristics of selected supercritical wing sections. The
airfoils “SC” are NASA’s second-generation profiles 2 . The perimeter and the cross-
sectional areas are plotted in Figure 2.10. A linear fit of the tabulated data is the
following:
Ac = 4.128 · 10−3 + 6.452 · 10−1 (t/c), pc = 1.971 + 0.659 (t/c) (2.4)
where Ac and pc are the cross-sectional area and the perimeter for a unit chord,
respectively. Note that these equations are only valid in the thickness range reported
in Table 2.1.
Table 2.1. Cross-sectional areas of selected supercritical wing sections
Airfoil t/c Area Perimeter Notes
SC(2)-010 0.10 0.071047 2.03498 symmetric

SC(2)-012 0.12 0.085258 2.04624 symmetric
SC(2)-410 0.10 0.067538 2.03661
SC(2)-414 0.14 0.094633 2.06187
SC(2)-610 0.10 0.067510 2.03819
SC(2)-612 0.12 0.080860 2.05056
SC(2)-614 0.14 0.093931 2.06373
SC(2)-710 0.10 0.067424 2.03937
SC(2)-712 0.12 0.080638 2.05171
SC(2)-714 0.14 0.093730 2.06561
SC-1094-R8 0.094 0.066044 rotorcraft airfoil
SC-1095 0.095 0.065683 rotorcraft airfoil
24 Aircraft Models
0.11 2.07
Area
Perimeter
0.1 2.06
Perimeter / wetted area

Cross-sectional area
0.09 2.05
0.08 2.04
0.07 2.03
0.06 2.02
0.08 0.1 0.12 0.14 0.16
Relative thickness
Figure 2.10. Geometric characteristics of some NASA supercritical wing sections.
Because the section characteristics reported in Table 2.1 refer to a unit chord,
they must be rescaled to the actual chord:
Ac = Ac1 c2 , (2.5)
where Ac1 is the cross-sectional area of a wing section having unit chord. The analyt-
ical regressions show that the dominant parameter is the wing thickness. Therefore,
in absence of detailed data, we can use Equation 2.4 to perform various geometry
calculations, including wetted areas and volumes. Symmetric wing sections are to
model the thickness horizontal and vertical stabilisers.
2.4 Wetted Areas

The wetted areas of the aircraft and its components are required for a wide range
of aerodynamic and performance calculations. For the aircraft at landing or take-off
configuration we need to account for the high-lift surfaces and parts of the landing
gear. We consider the main elements separately.
2.4.1 Lifting Surfaces

The calculation of the wetted areas of all lifting surfaces require the planform (as
previously calculated) and some additional information, which can have the form of
average wing thickness, dihedral or anhedral angle. An alternative method consists
in using the thickness distribution t/c (for example, Equation 2.4), along with the
spanwise chord distribution c(y). The wetted area of a half-wing is then
b/2
Awet = pc (y)dy. (2.6)
o
where pc (y) is the perimeter at spanwise position y.

2.4 Wetted Areas 25
Side View
6
4
z, m
2
Interpolated
Section
Top view
0
0 2 4 6 8 10
x, m
Figure 2.11. Control points for reconstruction of the fuselage forebody of the Airbus A300-
600.
This equation is solved numerically, after discretisation of the wing in longitudinal

strips. The area is then corrected for the effect of dihedral or anhedral (Equation 2.1).
2.4.2 Fuselage
For convenience, the fuselage is sectioned in three parts: the forebody (nose), the
central section and the aft-body (tail), as illustrated in Figure 2.5. The central section
is cylindrical. However, we must also account for the wing-body junction. The wetted
area of the fuselage is then calculated from
Awet = Awetnose + Awetcentre + Awettail + Awetblend − Awetwb . (2.7)
We analyse separately the contributions in Equation 2.7.
FUSELAGE FOREBODY. The forebody is defined by a set of control points on the

vertical symmetry plane and by a set of control points in the horizontal plane through
the nose point. The procedure for interpolation consists in calculating the height dz
and the width of the airframe dy at a given longitudinal position x. The cross-
sectional shape of the airframe can be approximated by two half-ellipses (top and
bottom half), as shown in Figure 2.11. This operation requires finding the centre
of the ellipse and the vertical semi-axes b1 and b2 , with b1 + b2 = dz. If the cross-
section is not circular, it is possible to rescale the maximum cross-section along the
longitudinal axis.
In any case, unless the cross-section of the fuselage is circular, this type of
interpolation is not very accurate; in fact, there is the possibility that the forebody
and the central sections show a discontinuity in the cross-sectional area distribution,
as discussed further in § 2.4.6.
SEMI-ANALYTICAL METHODS. There is a simpler approach to the calculation of the

wetted areas – one that does not require numerical integrations. The cross-sectional
shape in the vertical plane is not too dissimilar from the fore section of a Sears-Haack
26 Aircraft Models
l/d = 0.543
4
z, m 0
-2
-4 l/d = 0.392
0 3 6 9 12 15
x, m
Figure 2.12. Forebody of the B747-400 fuselage, contour approximation through Sears-Haack
bodies.
body. From the aerodynamic theory, we know that this is a body of revolution of
minimum wave drag for given volume and length. The body is pointed at both ends,
and therefore the comparison must be done between a fuselage forebody of length
lnose and a Sears-Haack body of length l = 2lnose . It can be shown that the wetted
area of the fore section of the Sears-Haack body is
Awet = 0.715πlnose d. (2.8)
We will assume that the actual wetted area is
Awetnose 0.72π lnose d. (2.9)
The numerical method shown in Figure 2.11 for the Airbus A300-600 yields A =
123.5 m2 , whilst the Sears-Haack formula, Equation 2.9, yields A = 119.1 m2 .
The use of the Sears-Haack formula is not accurate for the Boeing B747-400,
as it is shown in Figure 2.12. Two Sears-Haack bodies have been used from the
nose point O ; the bottom one has a slenderness l/d = 0.392 and the top one has a
slenderness l/d = 0.543.
FUSELAGE AFT-BODY. The construction of the aft-body (tail) is done with the same
principle used for the forebody. In this case there is no clear waterline to follow. It will
suffice to do the interpolation with two semi-ellipses whose axes are the local width
and height of the airframe, as shown in Figure 2.13. Sometimes the cross-section
cannot be approximated by elliptical shapes. For example, if the aircraft has a rear
cargo door, it is likely that the shape of the tail cone is flat at the bottom. Thus, more
information must be extracted from the drawings to create suitable cross-sectional
shapes that make up the wire-frame of the rear fuselage.
FUSELAGE CENTRAL SECTION. In most cases the fuselage cross-section is cylindrical.

Thus, the wet area is calculated from
Awetcentre = π lcentre d, (2.10)

2.4 Wetted Areas 27
Side View
4
0
z, m
-4 Top view
35 40 45 50
x, m
Figure 2.13. Control points for reconstruction of the fuselage after-body.
where d is an average diameter. Additional control points may be required to derive

the complete shape, the perimeter, the cross-sectional area and the equivalent dia-
meter. In some cases, it is sufficient to take as an average diameter the quantity d =
height/width. We have verified that by using the average diameter on the Boeing
B747-400, the error on the circumference is less than 0.4%. In fact, the double-decker
section 5.50 m aft of the nose has a circumference of 19.54 m; by using the average
diameter, the circumference is 19.60 m.
WING-BODY BLENDING. The major difficulty is in the calculation of the fuselage-wing

intersection. We need to select sufficient control points from the drawings to define
interpolating functions. For smaller aircraft, the intersection is defined by the wing
section at the root. For aircraft such as the Airbus A320-200, the intersection can be
calculated from the usual drawings and numerical interpolation. The area enclosed
by the intersection will have to be mapped onto the central portion of the fuselage,
which is cylindrical. For many aircraft the wing box crosses through the lower fuselage
(notably, the B747 and the A380) and requires complex three-dimensional filleting.
We define some blending functions between the fuselage cross-section and the
maximum cross-section. Call l p the cross-sectional perimeter and l pmax the maximum
cross-sectional perimeter. If lwb is the longitudinal extension of the wing-body inter-
section, the perimeter function within this length is defined by

πx
p(x) = (l pmax − l p ) sin , (2.11)
lwb
where the coordinate x is calculated from the most forward position of the surface
blending. The sinusoidal function is arbitrary, but it serves well for this purpose. The
additional wetted area due to the wing-body blending is calculated by a numerical
solution of the integral

lwb
πx
Awet blend = (l pmax − l p ) sin dx. (2.12)
o lwb
28 Aircraft Models
8
8
6
6
z, m
z, m
Cylindrical
4 Cross-section
Fuselage 4
Cross-section
Maximum
Cross-section
2
2
0
0
-4 -2 0 2 4 -4 -2 0 2 4
y, m y, m
(a) B747-400 (b) A380-861
Figure 2.14. Wing-body intersection of very large aircraft (same scale).
For the B747, a further sectioning between centre-forward and centre-aft sections is
required, along with a blending function for the two sections.
WING-BODY INTERSECTION. Finally, we need to subtract from the fuselage wetted

area the contribution due to the wing-fuselage junction; see, for example, Figure 2.14.
A first-order approximation is derived by subtracting the cross-sectional area of the
wing root, Awetwb = Acr oot . To be more accurate, we need to map the wing root onto
the fuselage. This is done numerically, by using the information from the control
points.
2.4.3 Nacelles and Pylons

The engines are mounted on the aircraft with pylons and nacelles. The pylons are the
structural connections between the nacelles and the wing. The front diameter, the
maximum diameter, the aft diameter and the length of the nacelles can be derived
either from line drawings or from manufacturer’s data. With these data it is possible
to calculate the approximate surface area of a body of revolution. Torenbeek 3
provides formulas for a more accurate calculation, requiring about 10 geometric
parameters. A better accuracy is required because the wetted area of the engines
can exceed 5%. Our approach is to use a set of control points from three-view
drawings as shown in Figure 2.15, which has been elaborated from Airbus† .
If the nacelle can be approximated by a body of revolution, the wetted area is
estimated from
Awetnac = π Awetside − Ainlet − Aoutlet . (2.13)
† Airbus: Aircraft A300-600 Characteristics for Airport Planning, Blagnac, France, 1990.
2.4 Wetted Areas 29
Control points
z from ground plane, m
Pylon
4
C
B
engine axis
2
Nozzle plug
A D Aft suspension
Fore suspension
0
18 20 22 24 26 28
Distance from nose, m
Figure 2.15. Control points for the engine–pylon assembly for the Airbus A300-600 aircraft
with CF6 engines.
If the nacelle is not a body of revolution, as often is the case, from the side-view
wetted area, we define an average diameter
Awetside
d= . (2.14)
lnac
Then the first term of Equation 2.13 becomes πd lnac . The wetted area of the pylon
is calculated by multiplying the wetted area from a side view by ∼2.1. The inlet and
outlet areas are calculated from the respective diameters, which are derived from
points A–B and C–D in Figure 2.15.
2.4.4 Winglets
A conventional winglet is mounted at the tip of the wing pointing upward. In some
cases there can be a lower winglet. The essential parameters are the height from the
base, the base chord, the tip chord and the cant angle. Simple geometric formulas are
used to calculate the reference areas and the wetted areas. One detail worthy of note
is that the canted winglet increases the wing span. For cases where a winglet does
exist, we can define two wing spans: a nominal span to the root of the winglet and
an effective span “over-winglet”, that is, reaching the tip of the winglet. The wing
loading causes the upper winglet to rotate outward, thus increasing the effective
span. Figure 2.16 shows the side view of winglets from Airbus and Boeing aircraft.
The graph shows the control points used for reconstructing the geometry. In the case
of the B737 we also need control points from a frontal view.
Until recently, the development of the winglets was done as a retrofit technology
(Boeing B737-300, B737-900, B757-500, B767-300ER). The modern trend is toward
blended winglets (Gulfstream II; Boeing B787), which are better modelled by CAD
systems. Manufacturers claim that the improved aerodynamics of these winglets can
reduce fuel consumption at cruise by as much as 5%.
30 Aircraft Models
1 A380 B737-9 A320-2

z, m
-1
Control points
-2
0 2 4 6
x, m
Figure 2.16. Side view of some winglets (same scale).
2.4.5 Flaps, Slats and Other Control Surfaces

The shape, planform area and wetted area of these devices are required when the
aircraft is at take-off and landing. The exact shape of these surfaces cannot be
inferred from the line drawings because the aircraft is shown with its control surfaces
retracted. However, it is possible to calculate the span of each device (inboard and
outboard flaps, slats, ailerons, spoilers, elevators). For the remaining quantities, we
need to make assumptions or extract information from photographs. For example,
the chord of a Kruger slat cannot be determined unless the slat is deployed; likewise,
the chord distribution of the flaps cannot be determined from sources other than
digital photographs of aircraft at landing or take-off.
FLAP RACKS. The flap racks contain the mechanical linkages that operate the flaps.
Figure 2.5 shows five racks per side; the racks may not have the same dimension,
although they are shaped like long and narrow keels partially exposed at the trailing
edge. The wetted area of each of these racks is found by calculating the projected
area on the side view multiplied by the perimeter of the maximum cross-sectional
area. If some geometrical elements are missing, some further approximations are
required. For example, if the shape of the side view cannot be extracted, then the
wetted area of each rack is assumed to be
2
Arack p lrack , (2.15)
3
where p is the perimeter of the maximum cross-sectional area and lrack is the overall
length. The junction between the top of the racks and the bottom of the wing is
subtracted from the total wetted area.
2.4.6 Model Verification: Cross-Sectional Area

The distribution of cross-sectional area is an indication of whether some of these
interpolations have been done correctly. As mentioned, only the maximum cross-
sectional area can be extracted, and the cross-sectional shapes at the nose and tail are
50
25
Aft-body
Fuse Centre 40
Cross-sectional area, m 2
Cross-sectional area, m2
20 Nacelles
Forebody
30
15
Wing
10 20
V-Tail Tail
Nacelles
5 10
H-Stab
0 0
0 10 20 30 40 50 0 10 20 30 40 50
x, m x, m
(a) (b)
Figure 2.17. Distribution of cross-sectional area by component (left) and cumulative (right)
for the Airbus A300-600.
approximated by ellipses. The main consequence of this approximation is that the

cross-sectional area distribution shows a discontinuity at two points on the fuselage.
To avoid this inaccuracy, the cross-sectional areas at the nose and the tail are adjusted
iteratively. If Acs is the cross-sectional area at the end of the nose and Acs o is the
cross-sectional area of the fuselage, the correction to be introduced is
Acs = Acs o k1 , (2.16)
where k1 is the ratio between the two areas. Blending functions can be used to scale
the cross-sectional areas of the nose and the tail. The purpose of these blending
functions is to generate smooth variations in cross-sectional areas from the wire-
frame data given on the vertical and horizontal planes (data such as those shown in
Figure 2.13).
Figure 2.17 shows the cross-sectional area of the fuselage and other major com-
ponents of the Airbus A300-600 model. The fuselage area distribution is continuous,
as specified (graph 2.17a); the sum of all cross-sectional areas is shown in graph 2.17b.
Upon convergence, also the wetted areas of the forebody and the after-body are
corrected.
A similar analysis was carried out for the Boeing B747-400 aircraft model,
which has a more complex fuselage shape, Figure 2.18. This is due to the double-
deck architecture of the forebody, which is then blended with a single-deck aft
fuselage.
2.5 Aircraft Volumes

The calculation of the aircraft’s volume or the volume of some of its systems is
required for operational reasons (cargo and fuel capability), or for the estimation of
the wave drag at supersonic speeds. Cargo volumes are generally available from the
manufacturer and require no special treatment. To calculate the total volume of the
32 Aircraft Models
50
Fuselage
40
Cross-sectional area, m2
Wing
30
20
Wing-body V-Tail
Nacelles
10 H-Tail
0
0 10 20 30 40 50 60 70
x, m
Figure 2.18. Distribution of cross-sectional area for the Boeing B747-400.
aircraft, we use the method of components:
V = V f use + Vwing + Vht + Vvt + Veng + · · · ... (2.17)
The items not listed in Equation 2.17 are minor contributions. The fuselage is split
into nose, centre and tail section, as previously done in the wetted-area analysis
(§ 2.4).
Volumes of cylindrical, cubic and tetrahedral bodies can be calculated by stand-
ard formulas. More difficult is to estimate the volume of complex geometries, such
as a wing and a fillet in wing–body combination. For these cases, the use of CAD
systems is inevitable. The volume of a wing is
b/2
V=2 Ac dy, (2.18)
i
where Ac is the cross-sectional area of the wing. The integration is extended from
the wing root to the tip. An example of volume estimates is shown in Table 2.2 for
some commercial aircraft.
2.5.1 Case Study: Do Aircraft Sink or Float on Water?

A commercial aircraft lands “safely” on a body of calm water* , without any structural
damage and with the landing gear retracted. Will it sink or will it float?
The first question to ask is whether the aircraft is water-tight. If this is the case,
then the answer is: yes, for some time. If we consider the average aircraft density,
ρ/V, with the volume taken from Table 2.2, then we find ρ 70 to 130 kg/m3 .
Therefore, from the law of buoyancy, the aircraft would float. To be more precise,
* Landing on water is defined as “ditching”.
Table 2.2. Volume breakdown of selected aircraft; all volumes in [m3 ]
Item A300-600 B777-300 A380-861 B747-400
Fuselage nose 129.3 179.3 400.6 468.0

Fuselage centre 669.8 1,347.4 1,440.2 1,167.0
Fuselage tail 185.4 329.2 608.1 455.3
Wing–body blend 0.0 55.6 296.6 54.3
Fuselage 984.6 1,911.6 2,745.5 2,144.5
Wings 99.6 237.2 715.7 337.0
H-stabiliser 11.4 27.3 90.3 43.6
V-stabiliser 22.6 29.2 75.5 55.4
Nacelles 39.3 74.9 134.0 45.7
Total 1,157.5 2,280.1 3,759.1 2,826.2
the engine/nacelles volumes should not be taken into account because the engines
will be flooded with water and would drag the aircraft down. The aircraft is not
completely water-tight. A minimum float time must be guaranteed to evacuate the
aircraft from the doors above the wings.
AIRCRAFT SCALING. We make one further consideration on scaling down or up an

aircraft. If we simply rescale the aircraft’s dimension, the volume can be fitted by an
equation such as

V b
= exp a1 log + b1 . (2.19)
V ref b ref
For the Airbus A380 the parameters a1 and b1 are a1 = 2.9975, b1 = −3.8389.
2.5.2 Wing Fuel Tanks

The calculation of the theoretical fuel capacity of the aircraft is based on the volume
available for allocating the fuel tanks. The most common tanks are inside the wings.
Depending on the aircraft, there can be central tanks (within the fuselage) and tail
tanks (in the tail plane).
The calculation of the wing-tanks geometry relies on some assumptions regard-
ing the position of the spars. Usually there are two major spars, at the leading and the
trailing edge. Also, some information regarding the span of the tank is required. The
most important geometrical parameters are xspar 1 (position of the leading-edge spar,
as a fraction of the chord); xspar 2 (position of the leading-edge spar, as a fraction of
the chord); st (skin thickness); and bwt (span of the wing tank). With this information
and the shape of the wing section at relevant spanwise sections, it is possible to write
a simulation model for geometry of the tanks. The numerical model for calculating
the wing tanks relies on a function
Vtank = f (xspar 1 , xspar 2 , bwt , st , Wing Geometry). (2.20)
An example is shown in Figure 2.19; this graph displays the tanks of the Boeing B777-
300. The available volume is estimated at 63.37 m3 per side, giving a theoretical fuel
34 Aircraft Models
Figure 2.19. Reconstruction of the wing’s fuel tanks.
capacity of 126,720 litres (∼101,375 kg). Further discussion of the wing tanks is in
§ 3.7 and in § 14.3, where we present a numerical model that simulates the fuel
temperature in the wing tanks.
2.6 Mean Aerodynamic Chord

There is one special definition of wing chord that is often used in aerodynamics and
stability analysis. This is the mean aerodynamic chord (MAC, or c). This quantity is
defined mathematically as the average of the squared chord distribution:
b/2
2
MAC = c = c2 dy. (2.21)
A o
This definition is derived from the strip theory of an untwisted blade with a constant
wing section. Starting from the pitching-moment equation, the zero-lift pitching
moment of a generic strip is
dMo
Cmo(y) = , (2.22)
qc2 dy
where dMo denotes the pitching moment of the wing strip and q is the dynamic
pressure. Integration of Equation 2.22 leads to
b/2
Mo = Cmoqc2 dy. (2.23)
−b/2
Next, assume that there is no change in the zero-lift pitching moment along the span.
Thus, we can write
b/2
Mo = 2CMoq c2 dy = CMoqc A, (2.24)
o
from which we find the reference chord defined by Equation 2.21. For the simple
case of a trapezoidal wing with taper ratio λ, wing span b, root chord cr and
2.7 Geometry Model Verification 35
Table 2.3. Calculations of MAC for the Airbus A320-200

aircraft; graphs on the same scale
Wing MAC y(MAC) xLE (MAC) Comments
1 2.578 12.015 18.070

2 4.309 7.975 15.891
3 3.938 6.182 14.923 Present choice
4 5.009 6.629 15.164
Airbus 4.120 – 16.310 Reference chord
gross area A, the MAC is

2 1 + λ + λ2
c= cr . (2.25)
3 1+λ
The position of the CG is often given as a percent of the MAC. The position and
the dimension of the MAC are calculated from the wing design. Once this is done,
also the position of the point located at one quarter from the MAC’s leading edge is
known. The position of the CG with respect to the nose is called xcg . The conversion
of this datum to a percent of the MAC is

xcg − xleMAC
% c = 100 , (2.26)
c
where xLEMAC is the position of the leading edge of the MAC with respect to the
nose. The inverse formula, which allows for the calculation of the moment arm, is
%c · c
xcg = xle MAC + . (2.27)
100
In Figure 2.20 we show two wing models, both having the root at the wing-body
attachment. Wing No. 1 is the wire-frame of the Airbus A320-200 wing, projected
onto the horizontal plane; Wing No. 2 is a wing obtained by joining the trailing edges
of the root and the tip with a straight line and represents an “equivalent” trapezoidal
wing. Note the difference in mean aerodynamic chord.
If we further extrapolate this wing to the symmetry section of the aircraft
(y = 0), we have two additional wings: Wing No. 3 is representative of the true
wing, extending through the wing box; Wing No. 4 has a straight trailing edge. Note
again the difference in mean aerodynamic chord. Curiously, the MAC of Wing No.
3 occurs at the wing break point, although this is not always the case.
A summary of the MAC calculations is shown in Table 2.3 and compared with
some data reported by the manufacturer. For this aircraft, airbus reports a “reference
chord” that is relatively close to that of wing model No. 3; the difference is attributed
to interpolation errors from technical drawings. In conclusion, for the calculation of
the MAC, we consider a wing planform extending to the symmetry plane through
the wing box.
2.7 Geometry Model Verification

The validation of the geometric model is done against a test case for which a detailed
model exists. We have considered the wing-body combination F4 by DLR, which
36 Aircraft Models
xLE
15
y, m
xTE
20
Wing 1
Wing 2
0 5 10 15
x, m
(a)
xLE
15
y, m
xTE
20
Wing 3
Wing 4
0 5 10 15
x, m
(b)
Figure 2.20. Four different ways of calculating the MAC of a transport wing; reference wing
is that of the Airbus A320-200.
has been tested in a number of wind tunnels 4 . A finite-element version of the model
has been produced for CFD validation and has been used for the validation of our
model 5 . The aerodynamics is further discussed in Chapter 4. This geometry is a
wing-body, without nacelles, pylons or tail surfaces. Table 2.4 is a summary of the
areas and dimensions. The main difference is the estimation of the wing area and the
aspect ratio, which do not correspond to the reference area. A correct calculation of
this area relies on a clear definition of the wing planform.
2.7.1 Case Study: Wetted Areas of Transport Aircraft

We now apply the wetted-area method to the calculation of the wetted area to some
model aircraft. The wetted-area breakdown is given in Table 2.5. The contributions
are given in absolute value and in percentage of the total wetted area. The fuselage
is split into fore, central and aft sections and wing-body blend. In all cases the most
Table 2.4. Analysis of the geometry of the F4 aircraft model
Item Calculated Reference Error, %
fuselage length 1.1920 1.1920 0.0

fuselage diameter 0.1484 0.1484 0.0
fuselage width 0.1484 0.1484 0.0
nose length 0.2513 0.2600 3.3
tail length 0.5516 0.5600 1.5
cross area 0.4663 0.4663 0.0
up-sweep angle 12.4217 n.a.
area (exposed) 0.1397 n.a.
area (reference) 0.1772 0.1454 21.8
area (apex) 0.1789 n.a.
wing span 1.1754 1.1754 0.0
tip chord 0.0606 0.0606 0.0
root chord 0.2401 0.2401 0.0
taper ratio 0.2526 0.2526 0.0
aspect ratio 8.7948 9.35 5.9
MAC 0.1261 0.1412 0.8
sweep LE 27.3233 27.10 0.8
sweep QC 23.1920 25.00 0.7
dihedral 4.8000 4.8 0.0
average t/c 0.1258 n.a.
important contributions are the wing area and the central section of the fuselage.
The wetted areas of some control surfaces are approximate.
The wetted areas calculated for a number of transport aircraft are shown in
Figure 2.21 as a function of the manufacturer’s empty weight (MEW). The solid line
is a least-square fit of the data. In this case, the linear fit is a fairly good approximation
of the data. The equation of the curve fit, based on a sample of eight aircraft, is
Awet [m2 ] = 2.14 · 102 + 1.414W, (2.28)
where W denotes the MEW in metric tons.
2.8 Reference Systems

There are three reference systems to consider: the Earth system, the body system
and the wind system. Local reference systems may be required for specific reasons
(for example, wing aerodynamics). The aircraft is supposed to fly with respect to a
Cartesian system fixed on the ground (Earth Axes), which for our purposes is con-
sidered flat. In fact, most of the performance calculations will be done for relatively
short flight times and at relatively low altitudes. The Earth’s curvature and rotation
are important for inertial navigation systems and to take into account the Coriolis
effects (accelerations) over a rotating Earth. The Coriolis acceleration is estimated
at less 10−3 g in atmospheric flight mechanics. The gravitational field is characterised
by acceleration of gravity, equal to the standard value of g = 9.807 m/s2 . The Earth
system has the x axis pointing North, the z axis, normal to the ground and pointing
downward; the y axis points East, and makes a right-hand system with x and z.
38 Aircraft Models
Table 2.5. Wetted-area breakdown for the selected aircraft (calculated). All areas are in [m2 ]; ()* data
are approximate
A300-600 B777-300 B747-400 A380-861
Item A % A % A % A %
Fuselage 772.4 50.85 1,255.8 55.3 1,325.8 46.49 1,496.5 37.70

Nose 123.5 8.13 156.3 6.64 332.8 11.67 286.5 7.22
Centre 459.6 30.25 815.5 34.62 609.6 21.37 700.1 17.66
Tail 189.2 12.47 265.1 11.23 356.9 12.52 403.9 10.17
[Wing-Body] 18.8 0.80 26.5 0.93 105.0 2.65
Wing 417.1 27.46 726.3 30.83 909.8 31.90 1,477.8 37.23
Wing tips 1.2 0.43 0.02 – –
Winglets – – 14.6 0.51 10.0 0.25
Horizontal tail 110.0 7.22 175.6 7.45 227.7 7.98 352.6 8.88
Vertical tail 94.5 6.22 35.9 1.52 160.0 5.61 237.8 5.99
Nacelles 109.7 5.42 111.1 4.72 117.2 4.11 236.2 5.95
Pylons 22.6 1.49 32.0 1.36 74.0 2.60 114.8 2.89
Flap racks 19.4 1.27 18.8 0.80 22.8 0.80 44.2 1.11
Total ∼1,519 ∼100.0 ∼2,356 ∼100.0 ∼2,852 ∼100.0 ∼3,970 ∼100.0
Wing/Total area 17.0 18.6 20.0 22.2
Wet/Wing area 5.882 5.367 5.000 4.505
All flaps* 56.6 4.6 97.9 4.2 190.7 6.7 204.5 5.2
All slats* 41.0 2.8 79.9 3.4 110.0 3.9 137.2 3.5
All spoilers* 9.6 0.6 23.8 1.0 68.1 2.4 230.1 5.8
All elevators 17.4 1.1 21.9 0.9 29.7 1.0 49.4 1.2
Rudder 24.6 1.6 33.4 1.4 32.6 1.1 69.7 1.8
A380-861
4,000
2
Wetted area, m
3,000
2,000
1,000
0 100 200 300

Manufacturer’s Empty Weight, ton
Figure 2.21. Distribution of aircraft wetted areas versus MEW (calculated).
Figure 2.22. Body reference system and forces on the aircraft.
There are several ways to define reference axes on the aircraft; the choice will
be limited by the fact that there is always one plane of symmetry. Thus, there will
be a body-conformal orthogonal reference system, centred at the CG of the aircraft.
The position of the CG is a known parameter (see Chapter 3). The reference system
is shown in the three-dimensional view of Figure 2.22. The longitudinal axis x is
oriented in the direction of the forward speed (wind axis); the vertical axis z is
vertical (along the acceleration of gravity g) and the y-axis makes a right-hand
Cartesian system with x and z. The positive y-axis is at the starboard side of the
aircraft.
The correlation between body- and Earth-reference system is done by three
attitude angles. The pitch attitude of the aircraft is the angle θ between the longi-
tudinal axis and the horizontal plane (positive with the nose up). The heading is the
angle ψ between the aircraft’s speed and the North-South direction. It is positive
clockwise. The bank attitude φ is the angle between the aircraft spanwise axis yb and
the horizontal plane.
The side force would not normally be present on the aircraft (and its occurrence
should be avoided). It is mostly due to atmospheric effects (lateral gusts), asymmetric
thrust and centre of mass off the symmetry line (due, for example, to a differential
use of the fuel in the wing tanks). The presence of such forces may lead to a yawed
flight condition. The yaw angle β is the angle between the longitudinal axis and the
true air speed vector.
The velocity (wind) axis reference system indicates the direction of the flight
path with respect to the Earth system. At any given point on the trajectory the
aircraft has a track and a gradient. The track is the angle on the horizontal plane
between the flight direction and the North-South axis† . The gradient is the angle of
† The North-direction must be further specified: it can be magnetic North or geographical North.
40 Aircraft Models
xb yb
α FP
γ ϕ
x y
z z
zb zb
[vertical plane] [vertical plane]
North xb
FP
β
ψ
East
[horizontal plane]
yb
Figure 2.23. Relationships between angles and reference systems.
the velocity on the horizon. If V is the ground speed and Vw is the wind speed, the
air speed is found from
Va = V + Vw . (2.29)
The transfer of forces between one reference and the other is done through rotation
matrices. The order of these rotations is important for the correct development of
the flight-mechanics equations. For the full derivation of these equations we refer
the reader to Yechout et al. 6 .
2.8.1 Angular Relationships

We consider some simple flight cases to provide a correlation between angles in the
different reference systems, Figure 2.23. First, consider the pitch angles. From the
definitions, the pitch attitude θ is related to the angle of attack α and the flight-path
gradient γ by
θ = α + γ. (2.30)
Next, consider a yaw. The heading ψ is related to the track ξ and the side slip by the
equation
ψ = ξ − β. (2.31)
If there is no side slip, the heading and the track are the same angle. Finally, consider
a roll problem. The bank angle φ is the inclination of the spanwise body axis on the
horizontal plane. The bank attitude is the same as the bank angle on a level flight
path.
2.8.2 Definition of the Aircraft State

To conclude, we establish the set of parameters that fully describe the aircraft state
in flight-performance calculations. The aircraft state is given by the vector:
= {W, X, V, U, , , α, β; η, θ, ξ ; ISF , LG}. (2.32)
The meaning of the symbols in Equation 2.32 is as follows:
r W is the weight.
r X = {x, y, z} denotes the position of the CG with respect to a reference system
on the ground. Often we call the geopotential altitude h ≡ z.
r V = {u, v, w} represents the velocity vector. In most cases we work with two-
dimensional trajectories and assume v = 0. Hence, we call Vg ≡ u and vc ≡ w
the ground speed and the climb rate, respectively.
r U denotes the true air speed. In level flight and in absence of winds, U = V .
g
r = {φ, ϑ, ψ} is the vector of aircraft angles with respect to the ground reference
system; they are the bank angle, pitch attitude and heading angle, respectively.
If we deal with two-dimensional trajectories, the heading angle is neglected.
r = {φ̇, ϑ̇, β̇} is the vector of angular velocities. In most cases we work with
quasi-steady trajectories; this vector is set to zero and is eliminated from the
state equation.
r α, β are the effective mean inflow angle and the side-slip angle, respectively.
r The flight control systems are denoted by η (elevator angle), θ (rudder angle),
ξ (aileron angle).
r I denotes the state of the high-lift system and is expressed by an integer
SF
number associated to a set of flap and slat deflection angles.
r LG is a binary flag for landing-gear position: retracted or deployed.
For a complete simulation, we need to add the engine state (Chapter 5) and the
atmosphere.
Summary
We have shown a strategy for building the configuration of a transport aircraft by
careful use of technical documentation in the public domain. The strategy is based
on the use of control points along three views and some rules that are used to recon-
struct the aircraft geometry. By careful consideration of these rules, it is possible
to elaborate distances, positions, planform areas, wetted areas and volumes. Not all
of the geometrical characteristics are equally accurate; in fact, the procedure shown
may fall short of expectations when considering the details of the control surfaces
unless additional information is added to the database. In the absence of first-hand
information from the manufacturer, this strategy is sufficiently accurate for aircraft
performance. About two dozen different sub-systems have been identified and mod-
elled. The geometry information will be used at several levels in later chapters. A
verification example for a wing-body has been shown, as well as calculations for a
number of aircraft.
42 Aircraft Models
Bibliography
[1] Filippone A. Theoretical framework for the simulation of transport aircraft
flight. J. Aircraft, 47(5):1679–1696, 2010.
[2] Harris CD. NASA supercritical airfoils: A matrix of family-related airfoils. Tech-
nical Report TP-2969, NASA, March 1990.
[3] Torenbeek E. Synthesis of Subsonic Airplane Design. Kluwer Academic Publ.,
1985.
[4] AGARD. A selection of experimental test cases for the validation of CFD
codes. Technical Report AR-303, Volume II, AGARD Advisory Group, Aug.
1994.
[6] Yechout TR, Morris SL, Bossert DE, and Hallgren WF. Introduction to Aircraft
Flight Mechanics: Performance, Static Stability, Dynamic Stability and Classical
Feedback Control. AIAA, 2003.
Nomenclature for Chapter 2

a1 b1 = factors in Equation 2.19
A = area
AR = wing aspect-ratio, b2 /A
Abbox = area of bounding box
Ac1 = cross-sectional area of wing with unit chord
Ac = cross-sectional area of a wing
Acs = cross-sectional area
Awet = wetted area
Axy = projection of planform area on ground plane
b = wing span
bw = wing span over winglet
bwt = overall span of wing tanks
c = wing or airfoil chord
cb = chord at break-point
cr = root chord
ct = tip chord
c = mean aerodynamic chord (also MAC)
Cmo = sectional pitching moment
CMo = pitching moment coefficient of a wing
d = diameter
FP = flight path
g = acceleration of gravity
h = geopotential altitude
k = iteration count, Equation 2.3
k1 = ratio between areas, Equation 2.16
I = moment of inertia
ISF = integer operator that defines flap/slat setting
l = reference length
lcentre = length of central fuselage section
lwb = length of wing–body intersection
lnose = length of fore (nose) fuselage section

lp = perimeter of a fuselage cross-section
lrack = length of flap racks
ltail = length of aft (tail) fuselage section
LG = binary operator that defines state of landing gear
m = mass
Mo = pitching moment
O = nose point
p = perimeter, Equation 2.15
pc = perimeter, Equation 2.4
q = dynamic pressure
st = wing-skin thickness
t/c = relative wing thickness
tgap = dimension of tail as shown in Figure 2.5
V = volume
U = air speed or ground speed
V = vector velocity
vc = climb rate
Vg = ground speed
W = weight
x, y, z = Cartesian coordinates
xr ,yr = coordinates of random point
xcg = longitudinal coordinate of the CG
xle = longitudinal coordinate of the leading edge
xspar1 = position of leading-edge wing spar
xspar2 = position of trailing-edge wing spar
x = aircraft state parameters
X = position vector
Greek Symbols
α = angle of attack
β = yaw or side-slip angle
γ = flight-path angle
= tolerance quantity
η = normalised spanwise coordinate
η = elevator angle in Equation 2.32
λ = wing taper ratio
ξ = track angle; aileron deflection in Equation 2.32
θ = pitch attitude
ρ = density
ϕ = wing dihedral angle, Equation 2.1
φ = bank angle
= vector of aircraft angles in ground reference system
ψ = heading angle
= aircraft state vector, Equation 2.32
= vector of angular velocities
44 Aircraft Models
Subscripts/Superscripts
[.] = average value

[.]a = air
[.]b = body coordinates, see Figure 2.23
[.]le = leading edge
[.]nac = nacelles
[.]r = root, or random point
[.]r e f = reference conditions
[.]t = tip, or tail
[.]w = wing or wind
[.]wb = wing-body
[.]wet = wetted area
3 Weight and Balance Performance
Overview
The aircraft weight influences the performance of the airplane more than any other
parameter. Weight has been of concern from the earliest days of aeronautics (§ 3.1).
Therefore, we devote this chapter to the weight analysis and the relative aspects,
such as operational and design weights (§ 3.2); weight management issues (§ 3.3);
operational limits (§ 3.4); centre of gravity envelopes (§ 3.5); loading and certification
issues, operational moments and corresponding errors (§ 3.6); the use of the wing
tanks (§ 3.7); and, finally, mass properties, including determination of the CG and
the moments of inertias of the empty airplane (§ 3.8).
KEY CONCEPTS: Aircraft Size, Design Weights, Operational Weights, Mass Distri-
bution, Centre of Gravity Envelopes, Moments of Inertia, Fuel Tanks.
3.1 A Question of Size

By the start of the First World War, in the United Kingdom it was believed that
the limiting weight of the airplane could not exceed 2,000 lb (∼800 kg). Aircraft
engineers of that time ignored that in 1913–1914 the great Igor Sikorsky designed,
built and flew a four-engine aircraft known as Ily’a Murometz, or S-27, whose loaded
weight was 5,400 kg. The total installed power was about 400 hp (∼300 kW). The
aircraft could carry as many as 16 passengers in its uncomfortable quarters 1;2 .
Prior to this project, it was widely believed that a multi-engine aircraft would
not be capable of flying. The main concern centred around the question of how to
control the aircraft in the event of one engine failure – a likely occurrence in those
days. With the Ily’a Murometz, Sikorsky proved that 1) multi-engined aircraft can be
operated safely, even with two engines shut down; and 2) aircraft can get bigger for
passenger transport and military operations. On 30 June 1914, Sikorsky and three
crew members embarked on a record-breaking flight from St. Petersburg to Kiev and
return, a total distance of about 2,600 km (∼1,400 n-miles), covered in 26 hours.
Baumann 3 wrote a report in 1920, prompted by the construction of giant air-
planes in Germany toward the end of the First World War – airplanes weighing as
much as 15.5 tons and powered by 260 hp (∼195 kW) engines.
45
*
An-225
600
A-380
B-747-400ER
MTOW, tons
400
200
0
0 25 50 75 100
Wing span, m
Figure 3.1. Wing span versus MTOW for large commercial aircraft; An-225 at design point.
Cleveland 4 predicted aircraft weights of 2 million pounds (800 tons) and nuclear
power plants. The Lockheed Corporation persevered along the nuclear power plant
option for some years and in 1976 Lange 5 proposed aircraft concepts in the 900
metric-ton class, including a 275 MW nuclear power plant. By contrast, the design
office at Boeing proposed the span-loader concept 6 – a 1,270-ton aircraft that never
went beyond the design office (project 759).
A detailed review focusing on configuration alternatives and economic viability
of the big airplanes is available in Reference 7 . Cleveland’s analysis contains a dis-
cussion of historical growth in size that leads to a square-cube law. The argument
is that if technology had not improved, growth would have been halted by the fact
that the load on airplane structures increases with the linear dimension, when the
load is proportional to the weight. The load is proportional to the weight W; the
weight is proportional to the cube of the linear dimension 3 ; the cross-section of
the beam is proportional to the square of the dimension 2 . Therefore, in first approxi-
mation, load/cross-section ∼ . At some point this increase in load reaches the struc-
tural limits of the material, and the beam may collapse under the effect of its own
weight.
If the wing is scaled up while holding wing loading and structural stresses as
constant, its weight will grow roughly as W1.4 . However, when one looks at the
details of the components, they do not scale up with the same factor. Cleveland
showed that by doubling the gross weight and the payload of the aircraft, the wing
weight would have to increase by a factor ∼2.69; the airframe would grow by a factor
1.84; the electrical systems would grow by a factor ∼1.40.
Figure 3.1 presents a trend of aircraft maximum take-off weight (MTOW) and
corresponding wing span b. The analysis shows that the wing span increases faster
than the gross take-off weight, according to a power curve
W b2.417 , (3.1)
Table 3.1. Payload data for very large aircraft; X is the range at maximum payload
Aircraft MTOW PAY/MTOW X (km) Notes
Airbus A380F 560.0 0.268 n.a.

Airbus A-400M, Grizzly 142.0 0.260
Antonov AN-225, Mrya 600.0 0.370 4,500 design
Boeing C-17, Globemaster 264.5 0.288 4,700
Boeing B747-400F 396.8 0.284 8,240
Boeing B747-400ER F 412.8 0.290 9,200
Boeing B747-800 F 442.3 0.303 8,130
Lockheed C-5B, Galaxy 381.0 0.311 5,526
Lockheed C-130J, Hercules 79.4 0.245 5,400
Satic A300-600, Beluga 155.0 0.305 1,666
when the weight is given in [N] and the span is in [m]. From a productivity point of
view, the most important factor is not the absolute weight and size of the aircraft but
its useful payload. Historically, this has increased from about 10% to more than 30%
of the current generation of airplanes; this ratio has also increased with the increased
gross weight. The increase is driven by commercial requirements and by the need
to move bulky equipment and machinery. Table 3.1 summarises the weight-payload
data of the largest cargo airplanes currently in service; X indicates the maximum
range at maximum payload.
Freight is transported in standard containers called pallets. Loading and unload-
ing of pallets is done quickly with ground-based vehicles with conveyors. The loading
of pallets also rationalises the available space in the cargo hold. A suitable cargo
performance parameter is
PAY
E= X. (3.2)
MTOW
This parameter is a measure of how much payload can be carried over a given
distance, relative to the MTOW. It emphasises the fact that a certain payload can
be flown over a longer or shorter distance. This parameter has been estimated for a
number of aircraft and is shown in Figure 3.2.
The design of a new large airplane has often been met with external constraints
of a different nature. For example, the Airbus A380 was constrained in wing span,
overall length and overall height to fit the existing infrastructure as much as possible.
Even then, considerable investment was required on ground infrastructure, such as
increasing the width of runway, the gate capability, the ground handling, and the
evacuation procedures. With regard to the latter points, the appearance of the Boeing
B747 in the early 1970s caused a change in loading and unloading technology. At
that time the lower deck was too large to be loaded manually, so new pallet sizes (or
unit load devices [ULD]) were introduced.
3.2 Design and Operational Weights

There are several definitions of aircraft weights that are required for opera-
tions and certification. A comprehensive discussion of aircraft weights (including
2,500
B747-4F
KC-10A
2,000
Il-96T
C-5B
An-22
An-124
1,500
E, km
C-141B
C-17A
1,000
Il-76MD
G-222 An-70
500
A300-600
C-130J
0
0 100 200 300 400 500
MTOW, ton
Figure 3.2. Payload factor of selected transport aircraft.
historical trends) is given by Staton et al. 8 . The most important definitions are given
here.
MANUFACTURER’S EMPTY WEIGHT (MEW). This is the aircraft weight when the vehicle
is delivered to the operator. This weight includes non-removable items, that is, all
of the items that are an integral part of the manufacturer’s configuration. There
can be weight differences between airplanes in the same fleet, as discussed further
in § 3.4.
OPERATIONAL EMPTY WEIGHT (OEW). This is the MEW plus a number of additional
removable items carried for operational reasons, and inevitable causes, such as lub-
ricant oils and the unusable fuel. The unusable fuel is the fuel that cannot be pumped
into the engines under critical flight conditions, such as those specified by the aviation
authorities. The operational items include the catering and entertaining equipment,
flight and navigation manuals, life vests, passenger entrainment devices and emer-
gency equipment.
MAXIMUM ZERO-FUEL WEIGHT (MZFW). This is the maximum weight of the airplane
loaded only with bulk cargo. During flight, the wing fuel tanks generate a root
bending moment that partially overcomes the stress created by the wing lift. When
these tanks are empty and the aircraft is heavily loaded, there is a risk of exceeding
the design limits for the wing root bending moment. The root loading is shown in
Figure 3.3, points A and B. If the aircraft carries fuel only on the centre tanks, the
resulting fuel weight contributes to the wing bending moments. Thus, it is advised to
use first the centre tanks fuel and then the wing tanks fuel. For the same reason, the
wing tanks are filled first (see further discussion in § 3.5.1).
Lift Lift
Bending
moment
A B
W
Figure 3.3. Wing root loads.
MAXIMUM TAKE-OFF WEIGHT (MTOW). This is the certified maximum weight at lift-off;
that is when the front landing gears lose contact with the ground. The MTOW is
reached by a combination of payload and fuel, as discussed in § 15.2.
MAXIMUM RAMP WEIGHT (MRW). This is the weight of the aircraft “ready to go”,
when the doors are closed, the systems are started and the vehicle starts taxiing. The
difference between the maximum ramp weight and the maximum take-off weight
corresponds to the amount of fuel burned between leaving the air terminal and the
lift-off. This difference is only relevant for very large aircraft because it affects the
operations on the runway. For example, the Boeing B747-400 has a MRW = 398,255
kg and a MTOW = 396,800 kg. The difference of 1,455 kg (∼ 1,750 litres) is the
fuel that can be burned from the moment the aircraft starts taxiing and the take-off
point. A further analysis of taxi fuel is in § 9.10.
MAXIMUM STRUCTURAL PAYLOAD (MSP). This is the allowable weight that can be
carried by the aircraft. The sum OEW + MSP is less than the MTOW and should give
the MZFW. The payload can be virtually everything: passengers and their baggage,
bulk cargo, military equipment and other hardware. The MSP is never reached for
passenger aircraft. In general, there is a need to differentiate among MSP, volume
available (space-limited payload, such as number of pallets) or capacity (due to
seating limitations).
MAXIMUM BRAKE-RELEASE WEIGHT (MBRW). This is the maximum weight at the point
where the aircraft starts its take-off run. Lower values of the MBRW, as they often
occur, are called brake-release gross weight (BRGW). Some performance charts are
drawn at constant BRGW.
MAXIMUM TAXI WEIGHT (MTW). This is the certified aircraft weight for taxiing on the
runway and takes into account the loads on the landing gears and the CG envelope
of the airplane.
MAXIMUM LANDING WEIGHT (MLW). This is the weight of the aircraft at the point of
touch-down on the runway. It is limited by load constraints on the landing gear,
on the descent speed (and hence the shock at touch-down), and sometimes on the
strength of the pavement. Permissible loads on the pavement are regulated by the
ICAO 9 .
The difference between MTOW and MLW increases with aircraft size. For
example, the Boeing B777-200-IGW has an MTOW = 286,800 kg and a MLW =
208,600 kg. This yields a difference of 78,200 kg, corresponding to 27% of MTOW. In
extreme cases (landing gear or engine failure), fuel can be jettisoned for unscheduled
landing. On the Boeing B777 fuel is jettisoned through nozzle valves inboard of each
aileron. This operation is done more routinely by military aircraft. However, there
is the possibility of making an emergency landing with a weight above the MLW. In
fact, the certification of the aircraft allows for a certain margin above the MLW in
such cases† . In such an event, the aircraft must be inspected before being allowed to
restart its service.
WEIGHT RELATIONSHIPS. The relationship between the weights defined previously is

the following:
MRW ≥ MTW ≥ MBRW ≥ MTOW ≥ MLW, (3.3)
OEW + MSP + MFW > MTOW. (3.4)
There are a number of other weight definitions in the aviation jargon. For example,
the all up weight (AUW) generally refers to the weight of the aircraft in cruise
conditions. There are, of course, the weights of the various aircraft components,
such as the engines and the engine installation. For the engines, a dry weight is
sometimes reported, which refers to the weight of the engine without lubricants and
some ancillary systems.
The main weight limitations are reported in the Flight Crew Operation Manual
(FCOM). The whole process of keeping track of the operational weights and man-
aging the position of the CG is part of the Weight and Balance Manual (WBM) of the
airplane. The manual is procured by the manufacturer and is given to the operator
with all of the instructions needed to load the vehicle and to manage the position of
the CG. Other important documents include:
r Delivery Weight Report, that establishes the OEW, the MEW and the corres-
ponding CG positions.
r Weighing Check List, that contains a list of each item, its position and unit weight
(item, weight, moment-arm, moment).
Other operational weight restrictions exist, such as the maximum weight for bal-
anced field length, obstacle clearance, noise emission and available engine power.
Methods for estimating the structural empty weight are given in a number of text-
books on aircraft design, for example Roskam 10 , Torenbeek 11 , Raymer 12 , Staton 8
and more specialised publications 13 . Comparisons of weight breakdown for a num-
ber of aircraft were published by Beltramo et al. 14 .
† The relevant regulations for fuel jettisoning are FAR/JAR $25.1001.
3.3 Weight Management 51
Most commercial airlines carry freight (cargo) as part of their business model.
Some major airlines have a balanced split between passenger and cargo revenue.
Such an important source of business has been made available by the standard-
isation of the ULD. There is a recognised international code for all ULD* . For
each device, the information includes shapes, dimensions, volume and maximum
weight. Loading can be manual or semi-automatic. There are specific loading
rules, including weight limits, positions, tying-down, the effect of moving items
and so on.
3.3 Weight Management

The structural weight of almost all aircraft grows over time during their service life,
although there are exceptions. Weight grows due to a number of reasons, namely
new performance specifications, re-engineering of the power-plant, exploitation of
structural design margins, seating re-arrangement, and not least the correction of
design flaws, that may come after several years of service. One, often overlooked,
cause of weight increase is the collection of dust and dirt inside the airplane, along
with the effects of humidity that increase the water carried by the airplane† .
Weight and balance logbooks are maintained to keep a check on all of the
modifications done to the aircraft. The manufacturers also provide charts showing
the basic weights and position of the CG. Loading of commercial aircraft is done
according to the instructions provided. The airline flight management performs
basic calculations of passengers and baggage. There are models that provide rapid
solutions to the aircraft weight and balance as a function of passengers, baggage and
fuel.
Fuel consumption must obey special priorities to maintain the balance of forces
on the aircraft at different flight regimes. Fuel must be used from the inboard tanks
first. An optimal distribution reduces the requirements on aircraft trim and therefore
the drag associated with it, thereby maximising the profitability of the aircraft.
AIRCRAFT WEIGHING. Prior to weighing, the aircraft must be emptied of all its fuel.
Three basic steps are required: 1) defuelling, 2) inventory of items, and 3) weighing.
The first step consists of draining all of the fuel from the fuel tanks. The aircraft’s nose
is raised to the point where its pitch attitude is zero. Specific defuelling procedures
are available in the WBM of the airplane. When the fuel level reads zero, all of the
pumpable fuel has been drained. There remains the unpumpable fuel. Additional
quantities can be eliminated by using a water drain. The remaining fuel after this
operation is the undrainable fuel.
The second step is the inventory of all on-board items and equipment. This
check list is to be compared with the list of fitted equipment from the manufacturer
to determine the MEW as well as the OEW.
Finally, the weighing is performed, generally in a closed hangar, after measuring
accurately the pitch attitude. The weighing is done by placing each wheel-set on
* IATA ULD Technical Manual.

† The water carried by the Airbus A380 is 1,700 l (standard) or 2,500 l (optional). The latter value
corresponds to the mass of about 25 passengers and their luggage.
Table 3.2. Standard passenger weights

(rounded to full kg)
Summer Winter
Average adult 86 88
Average adult male 91 93
Average adult female 81 83
Child (2–13 years) 37 39
a separate scale. The accuracy of the scale* must be ±20 kg for a weight range
2, 000 < W < 20, 000 kg, or ±0.1% for a weight range W > 20,000 kg.
CARGO LOADS. Certified containers must be used. These containers have known
values of the dimensions, volume, empty weight and CG. The containers must be
weighted prior to loading in order to produce a loading chart. Loading of bulk
payload causes localised stresses on the structure and the floor panels. Values for
maximum distributed loads (kg/m2 or other units) are provided by the manufacturer.
PASSENGER AND PERSONNEL LOADS. The balance of the aircraft is done by calculat-
ing as accurately as possible the distribution of the loaded weights; these are the flight
crew, the cabin crew and the passengers. For each of these categories the moment
arm is known through the cabin and seats layouts. In particular, each seat row is
assigned a moment arm; the passenger weight is assumed to be acting at the centre
of the seat.
Passenger weights for weight management have been standardised† . These
weights are summarised in Table 3.2. The passengers’ baggage must be weighted
at check-in for a better estimation of the take-off weight.
3.4 Determination of Operational Limits

When a new aircraft is delivered by the manufacturer, it is weighted accurately in
order to determine the manufacturer’s empty weight and the corresponding position
of the CG. In principle, each aircraft is unique, so there will be an individual value
of the MEW and the corresponding CG. For example, there are at least 15 different
weight versions of the Airbus A320-200, with MRW variable between 66,400 kg and
77,900 kg‡ . The current regulations allow for the determination of the MEW from
a set of aircraft (a fleet). This operation, in particular, is done by the operator, who
determines the Dry Operating Weight (DOW) for a fleet in the same weight class 15 .
This determination relies on a weight limit of ± 0.5% of the MLW around the DOW
of the fleet. In mathematical terms,
DOWmin = DOW − 5 · 10−3 MLW, DOWmax = DOW + 5 · 10−3 MLW.

(3.5)
* JAR-OPS 1.605.
† Appendix 1 to JAR-OPS 1.620(g).
‡ Airbus: A320 Airplane Characteristics. Rev. 25, Sept. 2010, Airbus Customer Services, Blagnac,
France.
An aircraft can be part of the same fleet if

DOWmin ≤ DOW ≤ DOWmax . (3.6)
During the life of the aircraft, the DOW will change. However, the aviation
regulations prescribe that the DOW of the aircraft be contained within 0.5% of the
MLW for all purposes related to loading, balancing and trimming. For example,
the Airbus A320-200 has a certified MLW = 64,500 kg and a DOW = 39,400 kg.
The resulting variation of the DOW is 0.005 MLW, or 322.5 kg. Thus, the weight of
the aircraft can vary between 39,077.5 kg and 39,332.5 kg. Therefore, the 15 weight
configurations of the Airbus A320 cannot belong to the same fleet.
Similar considerations hold for the variation of the centre of gravity, which is
allowed to oscillate by 0.5% MAC (or the reference chord) above and below the
value indicated in the aircraft’s documents. The corresponding quantity is called Dry
Operating Centre of Gravity (DOCG). For the A320 the reference chord is 4.194 m.
Therefore, the CG is allowed to shift by ± 0.021 m. If larger shifts occur, the loading
and trimming instructions will generally not be valid.
The DOW and DOCG values are given for a configuration corresponding to
landing gear deployed, high-lift surfaces in stowed position and potable water in the
water tanks.
One of the first operators to perform this analysis understood that considerable
weight savings could be achieved by controlling the use of drinkable water and
catering to match the number of passengers on board 15 .
The determination of the operational weight and CG is affected by uncertainty
by weight contributions and their location (passengers, cargo, fuel and operational
items). Let us call E the moment error due only to the item location; the weight of
the item is known accurately. This error can be written as
E = Mr eal − M∗ , (3.7)
or
E = W(xr eal − x ∗ ), (3.8)
where x is the moment arm and the asterisk denotes the “assumed” location. Next,
assume that the weight has an inaccuracy of W, so that Wr eal = W∗ ± W. Such
inaccuracy causes an error Ew in both weight and CG position. The following cases
may occur:
r If W is forward of the CG, then a +W moves the CG forward; −W moves
the CG aft.
r If W is aft of the CG, then a +W moves the CG aft; −W moves the CG
forward.
None of these events should move the CG beyond the certified limits.
3.5 Centre of Gravity Envelopes

The CG must be contained within specified limits, both longitudinal and lateral, to
ensure that the aircraft is fully controllable and that the flight is safe. In this section
we address the longitudinal movement of the CG and its dependence on both weight
and loading. Thus, we must determine the CG envelopes of the airplane. A few
definitions are required:
r The term in-flight denotes any flight condition other than take-off and landing.
r The term aft denotes rear part, or toward the tail.
r The term go-around denotes the end of final approach, which occurs when the
aircraft is prevented from landing. In this case, the thrust is increased rapidly
and the airplane is forced to pitch up. The pitch-up has to be compensated by
the elevator. The elevator moment increases as the CG moves aft. There is the
possibility that for some CG positions, the elevator is not capable of providing
the necessary restoring moment. Go-around is further analysed in § 11.8.
r The term α-floor protection denotes a constraint in angle of attack and hence a
reduction in the risk of airplane stall.
The movement of the CG is expressed by an index depending on the CG position

in percent MAC. The index is defined by
∗ MAC
I = W(xCG% − xCG% ) + k, (3.9)
c
∗
where xCG% is the actual CG position in percent MAC; xCG% is the reference CG
position in percent MAC; c and k are two constant factors that are chosen to keep
∗
the index I within reasonable values. When xCG% = xCG% , the index is independent
of the weight, I = k. In all other cases, the index is proportional to both weight and
MAC. Thus, on a graph I-versus-W, each CG position is represented by straight
lines; these lines have a negative slope if the CG is forward of the reference CG;
they have a positive slope if the CG is aft of the reference CG. The iso-CG line is
calculated from Equation 3.9. On this diagram it is possible to draw the operational
limits of the aircraft at all flight conditions, as well as the zero-fuel limit. These limits
generally represent a compromise between flight performance and aircraft loading.
3.5.1 CG Travel during Refuelling

Airplane refuelling is done according to procedures specified by the manufacturer.
Refuelling is done without passengers, unless there are particular reasons to do
so with passengers already on board. The operation is closely regulated to avoid
incidents* .
The process of refuelling leads to a movement of the airplane’s CG, which is
sometimes called fuel vector, an example of which is shown in Figure 3.4. In this
graph, segment 1 refers to refuelling of inner tanks 1–4 up to 3,000 kg; segment
2 corresponds to refuelling of the outer tanks up to 4,500 kg per tank; segment 3
corresponds to more refuelling of the inner tanks 2 and 3 until they reach 18,200 kg
(the total fuel now being 81,200 kg). In segment 4 the inner tanks 2 and 3 are
refuelled to 25,700 kg each, the centre tank is fuelled to 17,000 kg and the trim tank
is filled to 2,400 kg. At this point, the total fuel load reaches 116,200 kg. In segment
5 the centre tank is filled to 40,000 kg and the trim tank reaches 5,900 kg (total
fuel load 142,700 kg). For the A340-500 the rear centre tank is filled to 14,956 kg
* Relevant regulations are JAR-OPS 1.305 and FAR 121.750 (airplane evacuation).
200
A340-600
A340-500
7
6
150
Fuel mass, tons 5 7
4
100
50 3
2
1
0
-10 -5 0 5 10 15 20
Index
Figure 3.4. CG movement during refuelling of the Airbus A340-500 and A340-600 (adapted
from the FCOM).
(segment 6). For both aircraft versions, the last segment of the fuel vector (segment
7) corresponds to all tanks full. The fuel vector depends on the fuel density and hence
on the fuel temperature. The largest differences in CG position occur toward the
high fuel loads. Note that the sequence indicated is common to other large aircraft
and must be controlled by a fuel computer. In Figure 3.4, the index starts from zero
when the fuel tanks are empty. The movement of the CG must be contained within
the certified limits.
3.5.2 CG Travel in Flight

There are three reasons why the CG moves during the flight: 1) fuel burn, 2) change
in configuration, and 3) passenger movement. Cargo movement should not occur.
FUEL EFFECTS. Due to the complex sequence of fuel pumping from the tanks, there is
a CG movement, which if uncontrolled would cause an increase in fuel consumption
(because of higher drag); in the worst case it may reach a point at which flight control
becomes difficult. Due to the sophistication of modern airplanes, CG movement
can only be controlled by the avionics systems, which are dedicated hardware and
software systems (flight control system [FCS]). The FCS calculates the position of
the CG and compares it to a target value. A control band is defined as a target CG
and 0.5% MAC. The FCS commands fuel transfers in order to maintain the CG
within the control band. Details of the operations done to maintain the CG position
depend on the specific airplane. An example of a CG target is shown in Figure 3.5
for the Airbus A340. Figure 3.6 shows an example of the movement of the CG due
to fuel depletion in flight.
300
270
Aircraft weight, ton

240
210
180
36 37 38 39 40 41 42
Aft CG target, %MAC
Figure 3.5. Aft CG target for the Airbus A340-300 (adapted from the FCOM).
EFFECTS OF CONFIGURATION. Such effects are due to landing-gear deployment and

retraction, high-lift surfaces deployment and retraction, water movement, and move-
ments in the cabin. The latter two examples are stochastic; therefore, a safety margin
must be included in the estimation of the CG. The deployment of the flaps moves
the CG position aft; the deployment of the slats moves the CG position forward.
Landing-gear retraction moves the CG position forward. At landing the potable
water will be less than at the start of the flight (some of it can be jettisoned) and part
of it is in the waste tanks. The latter contribution cannot be discounted because for
MTOW
Depletion Control
Vector Band
Weight (arbitrary units)
Fuel Vector
CG
Envelope CG
Control
ZFW
Index
Figure 3.6. Example of CG control movement during depletion.
large aircraft, it can add several hundred kilograms to the operational weight (for
the Airbus A330, the potable water capacity is 350 kg).
PASSENGER MOVEMENT. This is not an unusual situation when during a long-haul

flight, passengers move away from their seats and cabin crew operate services up
and down the aisle. The aircraft balancing is done with passengers at their seats, cabin
crew at their position, and trolleys stowed. There are various movement scenarios
used by the manufacturers (for design purposes) and the operators, including events
such as passengers moving to the lavatories and back, passengers moving from back
rows to front rows, trolleys moving a number of rows and so on. This is further
analysed in § 3.6.
3.5.3 Design Limits on CG Position

Several requirements, including legal ones, have to be considered in the design of
the CG envelopes and in the CG management during loading and in flight. The key
aspects of the problem are listed as follows:
r Take-off. The forward CG limit is governed by the strength of nose landing gear,
by the take-off rotation requirements, the manoeuvrability, and the deployment
of the high-lift devices. The aft CG limit is governed by the strength of the main
landing gear, by the nose gear adherence to the ground, the take-off rotation
(tail strike) and the stability in level and manoeuvre flight. Some of these effects
are further discussed in Chapter 9.
r Landing. The forward CG limit is set to the same value as the take-off limit; the
aft limit is either superimposed to the take-off limit or aft of the take-off limit.
r In-Flight. The forward limits are constrained by fuel consumption, efficiency of
the elevator and some aerodynamic requirements, including stall of the trim-
mable horizontal stabiliser. The aft limit requires stability in steady-state flight
and during manoeuvres, go-around and α-floor protection.
An example of CG limits at take-off is shown in Figure 3.7, where we have

indicated the major elements of constraint. The top end of the scale is the maximum
structural take-off weight, which is aft-limited by the structural integrity of the main
landing gears (MLG) and is forward limited by the nose landing gear. At lower
weights, the aft limit is labelled as handling qualities, which is a term that refers
to aircraft rotation and pitch-up, deployment of high-lift systems and general flight
stability. The forward limit is a compromise between performance requirements and
aircraft loading. There is a grey area at the top left of the graph where the CG is
unlikely to be placed. In fact, the compromise between performance and loading
tends to prefer the former over the latter one. At the very low end of the weight, the
aft CG limit is labelled as nose landing gear adherence. If the aircraft is improperly
loaded, it will tend to lean back onto the main landing gear, which causes the tail
strike.
Figure 3.8 shows the CG envelope for landing configurations. The aft and for-
ward limits are similar to the take-off case, but the weight is considerably reduced
for aircraft have MLW far below the MTOW.
400
NLG MLG
Limit MTOW Limit
350
Grey Area
CG not reached
Weight, ton
300
Handling
Quality
250
Perf &
Loading
200
NLG
Adherence
150
0 10 20 30 40 50 60
Index
Figure 3.7. CG limits for take-off configuration.
A map of all of the operational limits is shown in Figure 3.9. The curve at the
centre of the graph is the one shown in Figure 3.4. The solid lines represent the CG
envelope at various flight conditions: take-off, flight and landing. The inclined lines
represent loci of constant CG position. A forward CG limit is 17% MAC for a cruise
condition and 18% MAC for take-off and landing; the aft CG limit is 43% MAC for
cruise and 42% MAC for landing. Take-off at low gross weights is more restrictive.
400
350
Weight, ton
300
MLW
250 Go-Round
Perf. &
Loading
200 Handling
Quality
150
0 10 20 30 40 50 60
Index
Figure 3.8. CG limits for landing configuration.
400
MTOW
350
Weight, ton
300
ding CG
it
ise Limit
e Lim
MLW
Land
ruis
AFT Lan
250
MZFW
AFT Cru
C
.O &
FW D
CG T
200
.
O
FW D
T.
T
AF
150
0 10 20 30 40 50 60
Index
Figure 3.9. CG limits of the Airbus A340-600 at various flight conditions (adapted from the
FCOM).
3.5.4 Determination of the Zero-Fuel CG Limit

The calculation of the zero-fuel CG limit is required to ensure that the aircraft can
be controlled in all phases of flight, from take-off to landing. The reference axis for
the calculation of the moments is a percentage of the MAC (generally 25%). If an
item is loaded forward of the reference axis, it gives a negative (nose-down) pitching
moment; if it is loaded aft of the reference axis, the pitching moment will be positive
(nose up). The position of the zero-fuel CG is calculated from
Mo + M pax + M p
ZFCG = , (3.10)
ZFW
where Mo denotes the pitching moment of the empty airplane, M pax is the moment
due to passengers and M p is the payload moment. The take-off CG position (TOCG)
is
Mo + M pax + M p + M f
TOCG = , (3.11)
GTOW
where M f is the moment arm of the fuel. The solution of Equations 3.10 and 3.11
requires the determination of the exact position of each operational item. Some
of these components have a stochastic contribution. The operator can improve the
balancing scenario by calculating the exact weight of the cargo items and their
position on the airplane.
3.5.5 Influence of CG Position on Performance

This effect is best seen through a real-life example. Figure 3.10 shows the influence
of the aircraft weight, as well as the configuration (landing gear up or down; flap
deflection) on the stall speed of the ATR72-500 powered by Pratt-Whitney PW127M
120
0
110
up
LG
100 15
VS, kt
Figure 3.10. Influence of weight and

25 configuration on the stall speed of the
up 25 ATR72-500 turboprop airplane (adap-
90 LG
35 ted from the FCOM).
up
LG
wn
-do
LG down
-
80 LG
70
12 14 16 18
Weight, tons
engines. The weight is always important, but the flap deflection has a strong effect
on the stall speed as a result of the increased drag.
A forward CG increases the pitch-down moment, which commands the hori-
zontal stabiliser to produce a restoring pitching moment. The converse is true for an
aft CG. We conclude that the more forward the CG, the greater the stall speed.
3.6 Operational Moments

The calculation of the ZFW, the take-off weight (TOW) and the CG depend on 1)
the weight and CG position before loading any item (fuel, cargo, passengers, service
items and fuels); and 2) possible movements in the CG due to items moving during
flight.
Some inaccuracies are inherent to this process, due to uncertainty in the weight
components (for example, passenger weight distribution) and/or their position on
the loaded aircraft. Also, uncertainties in the DOW and DOCG, as previously noted,
require consideration of stochastic analysis.
By using the pitching moment, we define the operational margin as the differ-
ence between the actual aircraft moment and the assumed aircraft moment. The
inaccuracy of an item location causes a moment
Mi = Wi (xi ± xi ), (3.12)
where xi is the uncertainty in the moment arm xi of item i having a weight Wi . The
moment error is
Ei = ±Wi xi (3.13)
and increases with both the weight of the item and its uncertainty. The moment
inaccuracy due to both weight and location is
Mi = (Wi ± Wi )(xi ± xi ), (3.14)

3.7 Use of Wing Tanks 61
which can be expanded into
Ei = Wi xi + xi Wi + Wi xi . (3.15)
The latter term is of second order and could be neglected.
INACCURACY IN PASSENGER DISTRIBUTION. The passenger distribution on the air-

craft is one of the major sources of inaccuracy on both TOW and CG position.
With reference to the latter point, each seat row has a known moment arm. If
the allocation of seats is known exactly, then the balancing could be done rap-
idly on a spreadsheet. In practice, calculations are done by assuming an average
arm for a cabin section and considering the worst-case scenario. This case occurs
when all seats in front of the average moment arm are full and all seats behind the
average arm are empty. However, this case is unlikely to occur. More realistically,
most operators have various passenger-seat allocation procedures. Typically, win-
dow seats are filled first, aisle seats are filled next, and finally the middle seats (if
any) are filled. The worst-case inaccuracy is determined, following the specific cabin
layout.
With reference to the inaccuracy on the passenger weight, average weights are
considered, as shown in Table 3.2, following the recognised regulations* but, in
practice, there will be a distribution of weights that generates a passenger weight
inaccuracy:
Wpax = W pax ± Wpax . (3.16)
Assuming a normal distribution around the average value W pax and a standard
deviation σ , we assume Wpax = 3σ . Thus, if i = 1, . . . , n denotes the cabin index,
and Wpaxi denotes the total passenger weight in cabin i, we have
Wpaxi = ni W pax ± Wpaxi . (3.17)
The inaccuracy is the same for all cabins, so that

n 1/2
√
Wpax = Wpax
2
i
= n(3σ )2 = 3σ n. (3.18)
i
3.7 Use of Wing Tanks

The use of the fuel tanks is a complicated matter that depends on the specific aircraft.
An example of wing fuel tanks is shown in Figure 3.11, that refers to the Airbus A340-
600 airliner. Note that the inner tanks are further sectioned in two parts, one of which
is shaded in the graph. The tank of other airplanes may have a considerably different
layout.
Table 3.3 is a summary of fuel-tank capacities for some Airbus airplanes. Here
is an example of how the fuel is used on the Airbus A340-600. Fuel is fed to the
* Federal Aviation Advisory Circular AC 120-27D, Nov. 2004.

Centre Tank
Split Valve
Inner Tank 2
Surge Tank
Rear Centre Tank
(A340-500)
Inner Tank 1
Trim Tank
Outer Tank
Surge Tank
Figure 3.11. Fuel tanks of Airbus A340-600.
engines through the inner wing tanks only; therefore, there must be a supply system
from all other tanks. The sequence is the following‡ :
1. The ACT fuel (if applicable) is pumped to the centre tank.

2. The centre tank fuel (if applicable) is pumped into the inner wing tanks 1 and 4,
to maintain full tanks, until all inner tanks are balanced.
3. The rear centre tank fuel is pumped to the centre tank when the latter one falls
below 1,000 kg.
4. The centre tank fuel is pumped to the inner tanks, whilst maintaining the fuel
level of the latter tanks between 17,200 and 18,200 kg, until the centre tank is
empty.
5. The inner tanks are used down to 4,000 kg each.
6. The fuel from the trim tank is pumped to the inner tanks.
7. The inner tanks are used down to 2,000 kg each.
8. The outer tank’s fuel is pumped to the inner tanks.
9. The inner tanks are used until empty.
The outer wing tanks are not used until the sequence has reached point 8 –
a point relatively late in the cruise. The whole process is controlled by the flight
computer.
3.8 Mass and Structural Properties

The determination of the structural mass distribution is generally a problem of
aircraft design. However, because the position of the CG and the resulting moments
of inertia are crucial to the aircraft performance, we need to estimate the distribution
of these mass components on the airplane. The flowchart in Figure 3.12 indicates the
‡ Source: Airbus, Flight Operations Support Department.
Table 3.3. Fuel tanks of some Airbus airplanes. ACT = Additional Central
Tanks; Jet-A1 density at 15 ℃ = 0.804 kg/l
Airplane/Tank Usable [kg] Volume [l] Notes
A319/A320-200
Outer 691 860
Centre 6,476 8,095
Inner 5,436 6,795
ACT-1 2,349 2,936 optional
ACT-2 2,349 2,936 optional
Vent tank n.a.
Total 18,730 23,412 without ACT
A330-200/300
Outer ×2 2,865 2,303
Centre 32,625 26,231
Inner ×2 42,000 33,768
Vent tank n.a.
Trim 4,890 3,932
Total 109,185 135,802
A340-200/300
Outer ×2 2,865 2,303
Centre ×2 33,300 26,773
Inner 33,578 26,997
ACT-1 5,652 4,544
ACT-2 5,652 4,544
Vent tank n.a.
Trim 4,890 3,932
Total 111,076 89,305 without ACT
A340-600
Outer ×2 4,953 3,982
Centre ×2 43,279 34,796
Inner (1 & 4) 19,402 15,599
Inner (2 & 3) 27,322 21,967
Rear Centre 15,600 12,542
Vent tank n.a.
Trim 6,269 5,040
Total 168,503 135,476
process required to establish the structural properties of the airplane. First, we need
to build two separate databases: 1) the mass components, and 2) the position of the
centre of each component. With these databases, we can proceed to the calculation
of the airplane’s CG and thence its moments of inertia with respect to the principal
axes through the CG.
Therefore, we follow three steps to determine the relevant mass and struc-
tural properties: 1) mass distribution, 2) centre of gravity, and 3) moments of
inertia.
The disassembly of the airplane components must be done with care and con-
sideration of what data are available, what data can be inferred and what has to be
left behind. In this framework, the critical components of a transport airplane are
shown in Figure 3.13. Note that the exact distribution of the systems (hydraulics,
Mass Database
Centre / Gravity Moments of
Inertia
Mass Positions
Figure 3.12. Flowchart for calculation of CG and moments of inertia from structural masses
and their position.
pneumatics, air conditioning, navigation instruments, avionics, electrical systems,

de-icing systems, load handling systems) is unknown. Thus, the systems mass is the
difference between the manufacturer’s empty mass and the allocated components,
as shown in the flowchart in Figure 3.13.
3.8.1 Mass Distribution

The determination of the structural weight/mass distribution can be done with dif-
ferent methods, including empirical, statistical and semi-analytical methods. None
of these methods is completely satisfying, and in absence of hard data on the air-
plane, we need to make a choice as to the most suitable candidate for this purpose.
A method for the determination of fuselage and wing structural mass (or weight)
has been proposed by Ardema et al. 16 and implemented in the aircraft design code
Figure 3.13. Flowchart of airplane mass distribution. The fuselage shell is further split into
noise, centre and tail contributions.
ACSYNT. Various other methods exist for the estimation of these and other mass
components, including Torenbeek 11 , Raymer 12 and Roskam 10 . In this context, we
take an intermediate approach, to take advantage of some systems modelling presen-
ted in Chapter 2.
FUSELAGE MASS. We split this contribution into four components: 1) nose section, 2)
central section, 3) tail section, and 4) structural decks. We first calculate the structural
mass of the shell fuselage (without the decks). For this purpose, we use Torenbeek’s
formula 11 :

Lt
mshell = 0.23Sg VD ,
1.2
(3.19)
d
where Sg is the wetted area of the shell [m2 ], defined as the wetted area of the fusel-
age, including the portions taken by the wing–body intersection, the empennage-
intersection and the landing-gear bays. The tail arm Lt [m] is defined as the distance
between the root mid-chord of the wing and the root mid-chord of the horizontal
tail-plane; d [m] is the average fuselage diameter* . These quantities can be calculated
with the methods presented in Chapter 2.
The factor VD [kt] is the design diving speed, defined in terms of equivalent
air speed (see also § 8.3). The value of the VD depends on the reference altitude.
Torenbeek does not specify which altitude to consider. The airplane has a design
dive Mach number; its corresponding dive speed requires the design dive altitude,
which is not readily available.
A number of corrections have been proposed by Torenbeek himself, to include
a variety of cases: If the engine is a turbofan, the mass given by Equation 3.19
increases by 8%; if it is a turboshaft, it increases by 5%. If the airplane is a
cargo, the shell mass increases 10% and if the engines are mounted aft, it increases
by 4%.
In this context, the shell fuselage mass is estimated and its breakdown into nose-
central and tail contributions is done according to the relative wetted areas. Thus,
we assume that the shell mass per unit area is uniform across the airframe. For
example:

Awet nose
m f use nose = mshell . (3.20)
Awet f use
The mass estimation of the structural decks is more open to approximations. For
example, the decks of a transport airplane have a mass range of ρ f = 5 to 10 kg/m2 ;
the largest values apply to heavy cargo vehicles. The deck area can be estimated
from the airplane layout, and it should be straightforward for single-deck airplanes.
Hence, the mass of the deck is
mdeck ρ f Adeck. (3.21)
* Notice that Equation 3.19 is dimensionally incorrect. If the dive speed is in [kt], the linear dimensions
are in [m] and the areas are in [m2 ], Equation 3.19 gives a mass in [kg] with the factor 0.23. A different
factor must be used if the parameters are expressed in imperial units. Similar considerations hold
for Equation 3.23.
LIFTING SYSTEMS. The lifting systems include the wing, the horizontal and the ver-
tical tail plane. The control surfaces of each system are considered a part of the
corresponding system and will not be split into separate contributions. The struc-
tural mass of the wing is calculated from a least-squares regression of existing wings
in terms of the manufacturer empty mass, me [kg]:
mwing = −9.095 + 9.189 · 10−2 me . (3.22)
As for the fuselage case, Torenbeek makes use of equations that depend on the
design diving speed. For example, the gross wing mass in [kg] is expressed by

−2 VD A0.2
m = κ A 6.2 · 10 , (3.23)
( QC − 2.5)1/2
where Ais the reference area of the surface in [m2 ], QC is the quarter-chord sweep
angle in degrees, the design dive speed VD is in [kt] and κ is a factor equal to unity,
except for the trimmable horizontal stabiliser, in which case it is suggested to use
κ = 1.1.
PROPULSION SYSTEMS. The propulsion system consists of the engines, the APU, the
nacelles, the pylons and the propellers (if any). The dry engine, the APU and the
propellers’ mass can be extracted from the type certificate documents. The main
unknown of this problem is the contribution of the structural masses (nacelles and
pylons). We rely on statistical data and find that
Wnac+ pylon 0.6Weng . (3.24)
LANDING-GEAR MASS. The technical literature indicates that the average weight of
the landing-gear systems is of the order of 4% MTOW. Following the method
proposed by Torenbeek, if m = MTOW/g is the maximum take-off mass expressed
in kilograms, we calculate the nose landing-gear weight from:
mNLG = A1 + B1 m3/4 + C1 m + D1 m3/2 , (3.25)
with
A1 = 9.1, B1 = 0.082, C1 = 0.0, D1 = 2.97 · 10−6 . (3.26)
The landing-gear mass is split between rolling stock (tyres, wheels, brakes) and other
masses (struts, actuators, shock absorbers, systems and controls). The mass of the
tyres can be inferred from the tyre specifications, or calculated. The total mass of the
rolling stock is variable, but for the nose landing gear, it is about double the mass of
the tyres. Thus, the other masses can be calculated by direct difference. This split is
important in the determination of the moments of inertia. Within the landing-gear
group, the contribution of the brakes is relatively high. Although exact figures are
not available, Boeing estimates that replacing steel brakes with carbon brakes on a
Boeing 737 saves between 250 and 320 kg, depending on the aircraft version.
The main landing-gear mass is calculated in a similar fashion. The corresponding
semi-empirical expression is obtained from Equation 3.25 with coefficients
A1 = 18.1, B1 = 0.131, C1 = 0.019, D1 = 2.23 · 10−5 . (3.27)

In this case, the mass of the rolling stock is about 3.0 to 3.1 times the mass of the
tyres. The method is straightforward if the airplane has two main landing gears; it
is more elaborate otherwise. For example, some airplanes have three main landing
gears (Airbus A340) and others have four main landing gears (B747, A380). Not all
of these systems have the same number of wheels. Thus, in the case of the Airbus
A340, the single fuselage landing gear has two axles, whilst the wings’ landing gears
have three axles. In this case, the split between rolling stock is done according to the
total number of wheels. If mLG is the total mass of the landing gear, then
mMLG = mLG − mNLG. (3.28)
The mass distribution of the main landing gears is
m∗MLG1 = mMLG x1 , m∗MLG2 = mMLG x2 , (3.29)
with
(nw nLG)1 (nw nLG)2

x1 , x2 , (3.30)
(nw nLG)1 + (nw nLG)2 (nw nLG)1 + (nw nLG)2
where nw denotes the number of wheels. Note that x1 and x2 are non-dimensional
factors. To calculate the contribution of the structural masses, subtract the mass of
the rolling stock, which is proportional to the number of wheels of each unit.
FURNISHINGS. The mass of the furnishings is roughly proportional to the number of

seats. If n1 is the number of economy seats and n2 is the number of first-class seats,
then
W f ur n = n1 w1 + n2 w2 , (3.31)
where w1 and w2 are average industry values that take into account also the ancillary
systems, such as overhead closets, entertainment systems and so on. A typical seat
in economy class weighs about 20 to 30 kg, but there is considerable variation in the
data. For a first-class passenger, the seat mass is at least double, and the new ultra-
thin seat is about half of these values. No data are published to make this estimate
more accurate. The furnishings for each passenger are estimated as 1.5 times the
mass of their seat.
The mass of the furnishings is a key aspect in the overall financial viability of
an airline. Note, in fact, that a reduction of 10% in the mass of the furnishings on
an airplane having an average furnishings mass of 10% of the MEW would lead to
a decrease of 1% in the OEW. For example, a Boeing B737-900 operating over a
range of 2,000 km (1,079 n-miles) at full passenger load and 200 kg of cargo would
save about 40 kg of fuel with a reduction of 1% in the OEW.
The methods presented are not always accurate enough, and variations of sev-
eral percent around the correct values should be expected. Clearly, any substantial
inaccuracy will be reflected in the determination of the CG and the moments of iner-
tia. However, by using the method of components, there is scope for improvement
at a later time if more data become available.
3.8.2 Centre of Gravity

To begin with, we need to establish some reference points and axes. The roll (longi-
tudinal) axis of the airplane is through the centre of the central (cylindrical) section
of the airplane. The pitch axis y is along the span, pointing to the right, and the
yaw axis z makes a Cartesian coordinate system with x and y. The intersection of
these axes is initially indeterminate, and we assume that x = 0 and y = 0 corres-
pond to the fuselage nose position. The coordinates of each airplane component
xi , yi , yi are calculated with respect to this system. We then remove this indetermin-
ation by assuming that the pitch axis passes through the CG of the empty airplane.
The centre of gravity is calculated on the basis of the distribution of the structural
components:
1 1 1
xcg = mi xi , ycg = mi yi , zcg = mi zi , (3.32)
m i m i m i
where xi , yi , zi are the distances between the centroids of the components and the
reference zero point on the fuselage nose. For verification of the method used, we
should convert the xcg into a percent MAC. If this value falls within the CG range
of the airplane, the result is satisfactory.
3.8.3 Moments of Inertia

The determination of the airplane’s moments of inertia with respect to its principal
axes is important for several flight-mechanics calculations. The pitch moment of
inertia determines the pitch-up response of the airplane at take-off, and therefore it is
especially important. The determination of the moments of inertia is more elaborate
than the estimation of the mass components because we need to determine their
distribution on the airplane.
Let us start from the basics: the moment of inertia of a lumped mass m from an
axis is I = md2 , where d is the distance between the mass and the axis. The moment
of inertia is linear with respect to the mass, and therefore the moment of inertia of
any number of masses can be added.
The empty airplane has different moments of inertia, depending on its con-
figuration (for example, landing gear retracted or deployed). The accuracy of the
method is only limited by the amount of information available. To start with, if the
moments of inertia of the empty airplane are known, the moment of inertia of a
loaded airplane can be estimated by adding the contribution of fuel and payload
and any other item on board (assuming that their location is known). However, the
main problem is the inertia of the empty airplane. If it is possible to build a finite-
element version of the airplane, the problem of the determination of the moments
of inertia could be solved with a fair approximation. However, unless the reader
has access to commercial data, this is not quite possible. Therefore, we propose
an alternative method. The key aspects of the calculations include the following
steps:
r Calculate the position of the CG with respect to the reference point on the
aircraft.
r Calculate the position of the centres of mass of each component.
r Estimate the mass of each component.

r Calculate the principal moments of inertia from the following equation:

Ixx = mi (yi2 + zi2 ), Iyy = mi (xi2 + zi2 ), Izz = mi (xi2 + yi2 ),
i i i
(3.33)
where xi , yi , zi are the distances of the centroid of the sub-component from the centre
of gravity, xcg , ycg , zcg . Unless otherwise noted, the airplane is symmetric with respect
to the vertical plane x-z, or ycg = 0. For convenience, the centre of the axes is the
fuselage nose (manufacturers may use an alternative reference). The corresponding
radii of gyration are
1/2 1/2 1/2
Ixx Iyy Izz
rx = , ry = , rz = , (3.34)
m m m
which are sometimes written in normalised form:
r r
xx yy r zz
r̃ x = , r̃ y = , r̃ z = . (3.35)
b (b + )/2
In the following analysis, we will consider the main items, with the only scope of
identifying the largest contributors to the moments of inertia.
FIXED MASS COMPONENTS. The fixed components of the airplane are:
1. Fuselage System: This component is split between fuselage shell and structural
floors (decks). The fuselage shell is split in nose, centre and tail sections. For
each of these sections we calculate the position of the centroid (xc , yc , zc ) with
sufficient accuracy, except the central section, if there is a large wing–body
blending.
Nose and Tail Section. There are various levels of approximations that can
be used. However, due to the uncertainty on other components (see following
discussion), there is no need to be too clever. We calculate the position of the
centroid of these sections and consider the nose and tail contributions as lumped
masses.
Central Section. If the central section can be considered cylindrical, the following
equations are used for the component moment of inertia:
1 2 1 2
Ixx = m 3(r1 + r22 ) + l 2 ) , Iyy = Izz = mr , (3.36)
12 2
where r1 is the external diameter of the shell, r2 is the internal diameter and l
is the length of the cylindrical body. The shell size can be inferred from cross-
sections of the fuselage. In absence of data, the shell thickness is assumed to be
∼ 0.1 m for a wide-body aircraft. Equation 3.36 defines the moments of inertia
of the cylindrical body with respect to the cylinder’s principal axes. A shift of
the pitch axis is required to refer Iyy to the airplane’s pitch through the CG. The
transformation is the following:
2
l
Iyy = Iyy + m − xcg . (3.37)
2
Deck Floors. For a single-deck airplane, the deck has a known width wd and
an area Ad . The equivalent length of a rectangular plate is ld = Ad /wd . The
moments of inertia of this plate with respect to the principal axes are:
1 2 1 1
Ixx = ml , Iyy = mwd2 , Izz = m(ld2 + wd2 ), (3.38)
12 d 12 12
where in this case m denotes the mass of the whole deck. As in the previous
case, we need to transfer these estimates to the axes of the airplane. The transfer
functions are:
2
ld
Iyy ⇒ Iyy + m − xcg , Ixx ⇒ Ixx + m (zd − zcg )2 , (3.39)
2
where zd is the elevation of the deck with respect to the bottom of the fuselage
shell. The problem is more elaborate for double-deck aircraft. Because there
are not many of these aircraft, the analysis is done on an ad-hoc basis.
2. Wing system: This system is modelled with a lumped mass. The mass itself is
estimated with the methods discussed in § 3.8.1; the position of the centroid
is calculated from the computerised geometry model; the data required are
(m, xc , yc , zc )wing . This approximation is sufficiently accurate if the wing mass is
uniformly distributed on the planform.
3. Horizontal tail system: It is treated like the wing; the data required are
(m, xc , yc , zc )ht .
4. Vertical tail system: It is treated like the wing; data required are (m, xc , yc , zc )vt .
5. Engines: They are treated like full cylinders of known mass and dimensions,
which are inferred from the type certificate document. The data required are
(m, xc , yc , zc )eng . The formulas for the moments of inertia of the solid cylinder
are

1 2 1 3 2
Ixx = mdeng , Iyy = Izz = m d + leng ,
2
(3.40)
8 12 4 eng
where deng is the average diameter of the engine and leng is its overall length.
These moments are transferred to the principal axes of the airplane. Note also
that Ixx is proportional to the inertia of the rotating masses of the engine.
A comparison between the solid cylinder with the lumped-mass approxi-
mation indicates that there is a 1% difference in the moments of inertia.
6. Propulsion system: This contribution includes all of the propulsion mass except
the engines. In practice, we split this contribution into pylons (m1 ) and nacelles
(m2 ). The centroids of these components are known from the geometry model.
The split between masses is done according to the relative wetted area (no other
quantity is available). If m denotes the total mass contribution of the pylons and
the nacelles, then

Awet pyl Awet nac
m1 = m , m2 = m . (3.41)
Awet pyl + Awet nac Awet pyl + Awet nac
The corrected centre of mass for the nacelle+pylon combination is

m1 xc pyl + m2 xcnac m1 zc pyl + m2 zcnac
x= , z= , (3.42)
m m
FWD DOOR, RIGHT EMERGENCY EXITS AFT DOOR, RIGHT
G L
A
G
L A L A
A
FWD DOOR, LEFT EMERGENCY EXITS AFT DOOR, LEFT
Figure 3.14. Cabin-floor arrangement for the Airbus A320-200 (adapted from Airbus).
which causes a shift of the CG with respect to the centroid of the nacelles:
xshi f t = x − xCnac , zshi f t = z − zCnac . (3.43)
Note that for a conventional transport aircraft, this shift is backward and upward
from the CG of the engine.
7. Propellers: The mass of the propeller is known from the type certificate docu-
ment. The propeller location is in front of the engines; each propeller is treated
as a point mass; the data required are (m, xc , yc , zc ) pr op .
8. Auxiliary Power System: The APU data are known from the type certificate.
This component is generally located in the tail cone. Its position can be given
with sufficient accuracy. Although the mass is relatively small, its pitch moment
arm is large; the data required are (m, xc , yc , zc ) APU .
9. Furnishings: There can be considerable indetermination on this contribution.
In the first place, we split the furnishings between passenger seats and other
masses. The passenger seats have known position from the cabin architecture,
and therefore their moment arm is known. An example is shown in Figure 3.14
for the Airbus A320-200. However, to simplify the problem, we assume that
the seat mass is locally distributed along the deck. For the moments of inertia
we use Equation 3.38. A transformation of these contributions to the airplane’s
principal axes is required as in the previous cases.
The contribution of the other furnishings masses is where the real problem
is. We assume that this contribution is also uniformly distributed over the floor
deck, but its vertical position is higher than the seats themselves. From Fig-
ure 3.14 we see the position of two galleys at the fore and aft end of the cabin* ;
these items have large moment arms and contribute to the Ixx .
MOVABLE MASS COMPONENTS. The movable components of the airplane are the
landing-gear systems. Although the mass of the landing gear is relatively small in
proportion to the structural mass, these components are somewhat dislocated and
can yield very large moments of inertia with respect to the principal axes of the
* Adapted from Airbus: A320 Airplane Characteristics, Rev. Sept. 2010.

airplane. They also cause a movement of the CG by a few percent MAC on an

empty airplane (depending on the airplane).
1. Nose landing gear: This contribution is calculated as a sum of two compo-

nents: rolling stock and other masses. The former contribution is a solid cylinder
having the dimensions of a wheel centred at the bottom end of the vertical
strut; the latter component is a lumped mass placed at the centre of the vertical
strut.
2. Main landing gear: This contribution is treated like the nose landing gear. Units
with multiple axles are considered by dividing the rolling stock by the number
of axles. Axles are displaced by about one wheel diameter.
When the contribution of the rolling stock is calculated with the cylinder approx-
imation, it is important to determine the correct axes. When the landing gears are
deployed and retracted, these axes rotate by about 90 degrees. Thus, it is important
to determine if folding back into retraction is done longitudinally or laterally.
The control surfaces are associated to the wing system and the tail surfaces. They
are not calculated separately and are not allocated among the movable components
for practical reasons (there are too many unknown data). Any other contribution
(systems, avionics, hydraulics, air conditioning, and so on) will be classified under
other items. It may happen that the estimated structural mass is not the same as
the manufacturer’s empty mass. If the estimate is below the mark (as it is often
the case), the remaining mass is associated to generic “systems”, whose position is
indeterminate. Therefore, it is suggested to assume this residual mass as uniformly
distributed on the airplane.
The accuracy of the method proposed depends on the accuracy of its constitu-
ents. If the mass components and their position are inexact, one cannot expect great
accuracy in the moments of inertia.
OPERATIONAL MASSES. When the airplane is loaded with fuel, cargo and passengers,
the moments of inertia assume larger values. As in the previous cases, an accurate
estimate of the position of these loads is necessary. When these data are unknown,
some educated guess is required. For example, the passengers are assumed as uni-
formly distributed along the cabin (as it is done for the seats and other furnishings).
This contribution is calculated with Equation 3.38, via the transformation to the
principal axes of the airplane.
FUEL TANKS. The contribution of the fuel depends on which fuel tanks have been
loaded. Figure 3.11 shows the wing tanks of the Airbus A340. Looking at the data
in Table 3.3 for the Airbus A340-600, we note that about 90% of the fuel mass is
allocated to the wing tanks. The size and position of the trim tanks are crucial in
the overall flight management system because the corresponding fuel is pumped
between tanks for trim conditions.
For reference, we report some mass components of the Airbus A320-100/200 in

Table 3.4. The mass components are basically the same for the A318, A319-100 and
A320-100/200, but the moment arms are different.
Table 3.4. Weight breakdown of Airbus A320 airplanes; mass in [kg]
Wing System m Propulsion m
All slats 290 Inlet cowl 138

Inner flaps 228 Fan cowl (avg) 40
Outer flaps 244 Thrust reverser (avg) 202
All spoilers 132 Dry engine, CFM56 2,778
Wing tips & winglets 52 Complete engine 3,501
Fuselage System Pylon 593
Fwd pax door 98 Landing Gear
Aft pax door 96 NLG system 327
Fwd cargo door 121 Wheel + tyre 35
Aft cargo door 121 MLG system 1,010
Tail-cone + APU 460 Wheel + tyre 135
Brake 69
V-Stabiliser V-Stabiliser
Box 365 Box 418
Rudder 88 Elevator 97
Removable LE 48 Removable LE 84
3.8.4 Case Study: Moments of Inertia

Following the methods presented in the previous sections, we have produced com-
puter versions of a number of transport airplanes in configuration empty-landing-
gear-out. We calculated the moments of inertia and the corresponding radii of
gyration (Equation 3.34) with the FLIGHT code. Prior to these operations, we
calculated the position of the centre of gravity, both in absolute coordinates and as a
percent of the mean aerodynamic chord. These results are summarised in Table 3.5.
Figure 3.15 shows the estimated moments of inertia plotted as a function of the MEW.
Table 3.5. Airplane mass properties at take-off-empty (no fuel) configuration (calculated)
mass Ixx Iyy Izz rx ry rz

Airplane [kg] [kgm2 ] [kgm2 ] [kgm2 ] [m] [m] [m]
ATR72-500 12,980 0.615·106 0.430·106 0.548·106 6.88 5.75 6.50

A300-600 91,303 0.149·108 0.964·107 0.139·108 12.78 10.28 12.34
A319-100 39,838 3.163·106 1.901·106 2.908·106 8.91 6.92 8.54
A320-200 42,256 0.368·107 0.236·107 0.335·107 9.33 7.48 8.90
A340-600 171,832 0.667·108 0.359·108 0.475·108 19.70 14.45 16.62
A380-861 271,746 0.138·109 0.608·108 0.948·108 22.52 14.96 18.68
B737-900W 43,026 0.558·107 0.412·107 0.449·107 11.39 9.79 10.21
B747-400 179,800 0.696·108 0.374·108 0.471·108 19.72 14.42 16.19
B777-300 161,355 0.612·108 0.469·108 0.525·108 19.48 17.04 18.04
Dash8-Q400 17,222 0.107·107 0.803·106 0.105·107 7.90 6.83 7.80
G550 21,074 0.164·107 0.684·106 0.887·106 8.83 5.70 6.49
150
Ixx
Iyy
Izz
2
120
Moment of inertia, 10 kg m
6 90
60
30
0
0 100 200 300
3
Mass, 10 kg
Figure 3.15. Roll, pitch and yaw moments of inertia of some commercial airplanes (calcu-
lated); empty airplanes, take-off configuration.
Based on our analysis, the following regression curve is proposed:
I = a + bm + cm2 + dm3 , (3.44)
where the mass m is in metric tons and the moments of inertia are in 106 kgm2 . The
coefficients of this equation are given in Table 3.6. Equation 3.44 is valid for MEW >
10,000 kg. However, if the mass is restricted to MEW > 90,000 kg (large airplanes),
the moments of inertia grow linearly with the mass. The corresponding least-square
linear fit equation is:
Ixx = −49.597 + 0.681 m, Iyy = −0.298 + 0.269 m,

Izz = −26.11 + 0.441 m,
(3.45)
where the empty mass m is in tons (103 kg) and the resulting moments of inertia are
106 kgm2 .
Table 3.5 shows also the calculated radii of gyration, r x , r y , r z. A clear relationship
with the wing span and the mass cannot be established.
Table 3.6. Coefficients of Equation 3.44
a b c d
Ixx 3.2933 −1.6829 · 10−1 4.1838 ·10−3 −6.4162 · 10−6

Iyy 1.0628 −7.2817 · 10−2 2.8159 ·10−3 −6.4169 · 10−6
Izz 1.5154 ·10−1 −1.2659 · 10−2 2.4202 · 10−3 −4.0324 · 10−6
Bibliography 75
Summary
We have given a critical discussion of the aircraft size and its potential growth. We
have also shown how the growth has been made possible by several advances in
technology. Larger aircraft have higher cargo efficiencies, but consideration should
be given to other factors, such as the actual value of the payload and the range
required. We have given a definition of the key design and operational weights of
an aircraft and explained how they are calculated. We developed a method for the
estimation of the mass breakdown and for the CG; finally, we calculated the moments
of inertia and the radii of gyration of a number of modern transport aircraft. We have
emphasised the importance of estimating the position of the CG and the need to
control and guarantee its movement, both on the ground (whilst loading the aircraft)
and in flight. The position of the CG is critical to most flight conditions and will be
discussed in the following chapters.
Bibliography
[1] Finne KN. Igor Sikorsky: The Russian Years. Prentice Hall, 1988. ISBN
0874742749.
[2] Grant RG. Flight: 100 Years of Aviation. Dorling Kindersley Ltd, 2002.
[3] Baumann A. Progress made in the construction of giant airplanes in Germany
during the war. Technical Report TN 29, NACA, 1920.
[4] Cleveland FA. Size effects in conventional aircraft design. J. Aircraft, 7(6):483–
512, Nov. 1970.
[5] Lange RH. Design concepts for future cargo aircraft. J. Aircraft, 13(6):385–392,
June 1976.
[6] Whitener PC. Distributed load aircraft concepts. J. Aircraft, 16(2):72–75, Feb.
1979.
[7] McMasters JH and Kroo I. Advanced configurations for very large transport
airplanes. Aircraft Design, 1(4):217–242, 1998.
[8] Staton RN, editor. Introduction to Aircraft Weight Engineering. SAWE Inc.,
Los Angeles, 2003.
[9] Anon. Aerodrome design manual. Part 3. Pavements (Doc 9157P3), 2nd Edi-
tion, 1983 (reprinted 2003).
[10] Roskam J. Airplane Design Part V: Component Weight Estimation. Roskam
Aviation & Engineering, 1985.
1985.
[12] Raymer D. Aircraft Design: A Conceptual Approach. AIAA Educational
Series, 3rd edition, 1999.
[13] Udin SV and Anderson WJ. Wing mass formula for subsonic aircraft. J. Aircraft,
29(4):725–732, July 1992.
[14] Beltramo MN, Trapp DL, Kimoto BW, and Marsh DP. Parametric study
of transport aircraft systems cost and weight. Technical Report CR 151970,
NASA, April 1977.
[15] Galjaard ER. Real time related dry operating weight system. In 54th SAWE
Annual Conference, Paper 2278, Huntsville, AL, May 1995.
[16] Ardema MD, Chambers MC, and Patron AP. Analytical fuselage and wing
weight estimation of transport aircraft. Technical Report TM-110392, NASA,
May 1996.

a, b, c, d = coefficients of polynomial Equation 3.44
A = gross wing area
A1 , B1 , C1 , D1 = coefficients in Equation 3.25
Awet = wetted area
b = wing span
c = factor in Equation 3.23
d = diameter
E = cargo performance index, Equation 3.2; error, Equation 3.13
k = factor in Equation 3.23
I = index for CG position, Equation 3.9
Ixx = roll moment of inertia
Iyy = pitch moment of inertia
Izz = yaw moment of inertia
= linear dimension; length; fuselage length
Lt = horizontal tail arm
m = mass
me = manufacturer’s empty mass
M = flight Mach number
M = moment
n = cabin index
nw = number of wheels
n1 , n2 = number of seats, Equation 3.31 (economy, business)
r = distance or radius of gyration, or diameter
rx , r y , rz = radius of gyration: roll, pitch, yaw, respectively
Sg = wetted shell fuselage area
VD = design diving speed
VS = stall speed
w = width
w1 ,w2 = seats weight (economy and business)
W = weight
x1 , x2 = non-dimensional factors in Equation 3.30
xcg = position of the centre of gravity
∗
xcg = position of the centre of gravity (reference)
xi = moment arm of i-th item
xc , yc , zc = position of centroid
X = range
z = vertical coordinate/distance
zd = deck’s elevation with respect to the bottom of the fuselage
Greek Symbols
QC = wing sweep at quarter chord
κ = factor in tail mass Equation 3.23
ρf = specific mass (density)
σ = standard deviation
[.]∗ = corrected value
[.] = average value
˜
[.] = normalised value
[.]c = centroid
[.]cg = centre of gravity
[.]d = deck
[.]e = empty
[.]eng = engine
[.]i = item index
[.] f = fuel
[.] f ur n = furnishings
[.] LG = landing gear
[.]nac = nacelles
[.]o = reference value
[.] p = payload
[.] pax = passenger
[.] pyl = pylons
[.]ht = horizontal tail
[.]vt = vertical tail
[.]zf = zero fuel
4 Aerodynamic Performance
Overview
In this chapter, we present first-order methods, based on the principle of compo-
nents, to determine the aerodynamic properties of the aircraft. One is tempted
to develop methods of high accuracy, based on the physics. These methods are
inevitably complex and computer intensive. The requirement for methods of general
value inevitably limits the accuracy on any single aircraft. Aerodynamic data of
real airplanes are closely guarded. We show how these low-order methods, with a
compromise between physics and empiricism, yield results of acceptable accuracy.
We deal with aerodynamic lift in § 4.1. The wide subject of aerodynamic drag
(§ 4.2) is split into several sub-sections that present practical methods. The analysis
of the aerodynamic drag has a number of separate items, including a transonic model
for wing sections (§ 4.3) to be used in the propeller model (Chapter 6), transonic
and supersonic drag of wing–body combinations and bodies of revolution (§ 4.4),
and buffet boundaries (§ 4.5). We present a short discussion on the aerodynamic
derivatives in § 4.6. We expand the aerodynamic analysis to the estimation of the
drag of a float-plane in § 4.7. We finish this chapter with a simple analysis of the
vortex wakes and the aircraft separation distance (§ 4.8).
KEY CONCEPTS: Aircraft Lift, Aircraft Drag, Transonic Drag Rise, Supersonic Drag,
Bodies of Revolution, Buffet Boundaries, Aerodynamic Derivatives, Float-Planes.
4.1 Aircraft Lift

The determination of the lift characteristics of the wing ultimately depends on the
amount of information that is available. Often only the planform shape is known.
The twist distribution and the camber of the wing sections, essential for a realistic
calculation of the aerodynamics, are most likely unknown. We attempt to overcome
this problem with the following simplified analysis. The lift coefficient at cruise
configuration is given by
CL = CLα α e + (CLα α)ht , (4.1)
where α e denotes the mean effective inflow angle across the span. Equation 4.1
shows the separate contribution of wing and tail plane. The lift must be provided
78
with the fuselage nearly horizontal (zero attitude) to avoid passenger discomfort.
The contribution of the tail plane must be very small, although it is important for
longitudinal trim (Chapter 7). At cruise conditions we neglect the contribution from
the horizontal stabiliser. For transport aircraft, values of the CL at cruise are about
0.4 to 0.6.
The effective mean incidence αe is the sum of three contributions: 1) the mean
incidence in nominal conditions αo, that is, the mean wing angle with respect to the
aircraft’s longitudinal axis (a design parameter); 2) the pitch attitude θ; and 3) the
effect of climb rate, γ :
αe = αo + θ − γ . (4.2)
At cruise conditions γ = 0. Therefore, the lift coefficient can be expressed by the
following equation:
CL = CLα α o + CLα θ = CLo + CLα θ. (4.3)
The effect of wing thickness is of second order compared with the lifting properties
of a cambered lifting surface. For the NACA 00XX series the lift increases linearly
with the rate of 7 · 103 per 1% increase in relative thickness. Therefore, the angle of
attack is far more important in a wing-planform analysis.
4.1.1 Calculation of Wing Lift

We use a lifting surface method for the calculation of the wing loading. This is the
best method in terms of speed, flexibility and generality. It is flexible in terms of flight
conditions that can be prescribed; it is general in the sense that most wing parameters
can be added to the model, including winglets, controls surfaces and disjoint wings
(such as a wing–tail combination). The major limitations of this approach are the
linear aerodynamics underlying the models and the absence of viscous effects.
There are various numerical methods, but ultimately they are all equivalent
from a theoretical point of view, although they differ in the computer implement-
ation. A good summary of these methods is given by Katz and Plotkin 1 . Various
lifting-surface programs exist in the open domain. Due to the nature and extent of
the matter, we cannot devote the attention it deserves. In brief, the lifting-surface
theory takes into account the actual shape of the planform, the basic angle effects
(dihedral, angle of attack), the wing’s mean camber and the winglets. This model is
powerful enough to give results of engineering accuracy, including the aerodynamic
coefficients, the moments and the aerodynamic derivatives. The calculation of the
lift and lift distribution is closely related to the induced drag, discussed in § 4.2.1.
Results of the application of the method to the isolated wing and horizontal tail
plane of the A300-600 are shown in Figure 4.1. The data shown are the spanwise
distribution of lift coefficient and the product c Cl (local chord × local lift coefficient).
4.1.2 Wing Lift during a Ground Run

During the initial stages of a ground run, the effective mean aerodynamic incidence
α changes due to two important factors: 1) the pitch-down attitude of the airplane,
and 2) the ground effect. With reference to the former event, the pitch-down attitude
0.6 4 0.12 0.5

Cl Cl
c Cl c Cl
0.4
3
0.4 0.08
0.3
c Cl
c Cl
Cl
Cl
2
0.2
0.2 0.04
1
0.1
CL = 0.45, α = 1.65 degs CL = 0.10, α = 1.28 degs

0 0 0 0
5 10 15 20 25 0 2 4 6 8
y, m y, m
(a) Wing (b) H-tail
Figure 4.1. Spanwise lift distribution from the isolated wing and horizontal stabiliser of the
Airbus A300-600 (calculated).
is of the order of 1 degree, although its precise value depends on the airplane. In
this case the effective inflow consists of four contributions. In addition to the terms
appearing in Equation 4.2, there is the ground effect αg , which is an apparent angle.
Thus, we have:
αe = α o + θ − γ + αg . (4.4)
Figure 4.2 shows all of the contributions except the ground effect. On further analysis,
when the airplane is on the ground, then γ = 0; hence, the attitude is the result of two
contributions: the nominal pitch-down attitude and the effect of the shock absorber
of the nose landing gear. As the airplane gains speed, the nose landing gear is
relieved of its normal load; the shock absorber relaxes to its unloaded position and
contributes to the reduction of the nose-down pitch.
The ground effect is a more complex problem because it depends on the wing
configuration, the ground clearance and the value of the CL. In the lifting surface
method, the proximity of the ground is accounted for with a mirror boundary con-
dition that satisfies the constraint of no-through velocity at the ground.
Figure 4.3 shows examples of ground effect on some transport wings in their
full configuration with control surfaces retracted. The winglets are included in the
aerodynamic model. The ground effect is a linear function of the CL. However,
FL
Mean Chord
αo θ
U
γ Figure 4.2. Calculation of effective
angle of attack; FL = fuselage line.
Runway (horizontal)
4
Gulfstream G550, z = 1.80 m
Boeing B777-300, z = 2.45 m
Airbus A380-861, z = 3.10 m
αg = α∞ - Δ α
-2
-4
0.4 0.5 0.6 0.7 0.8
CL, plain wing
Figure 4.3. Change in mean incidence between OGE and IGE.
the general conclusion is that to achieve a specified CL, the mean incidence has to
decrease; in other words, for a given incidence, the wing provides a higher CL. This
difference vanishes as the aircraft becomes airborne.
The effect of ground clearance at fixed CL is shown in Figure 4.4 for four transport
airplanes. From this limited analysis, we infer that an empirical expression for the
ground effect on the effective mean inflow is
αIGE = a1 [1 + exp(a2 CL)] + a3 , (4.5)
where a1 , a2 , a3 are constant factors. The change in effective incidence for a given
configuration and CL is a function of the ground clearance z/b:
αg = αIGE − αOGE = f (z/b), (4.6)
where the mean OGE inflow angle is taken from Equation 4.2. The change in
mean incidence αg is shown in Figure 4.3. For example, if a wing has CLo 0.5
out of ground effect, then the increase in mean incidence on the ground is estimated
between 2.5 and 4 degrees. Assuming for the sake of discussion that αg 3 degrees,
with CLα 4.45, then we have CL 0.233, or CLg 0.73. This lift coefficient is
relatively modest and certainly not high enough to ensure that an airplane will be
able to take off.
4.1.3 Lift Augmentation

The term augmentation will be referred to all cases in which the zero-incidence lift
is either increased or decreased by the use of suitable control surfaces, placed at the
trailing edge or at the leading edge or both.
Height above ground, m Height above ground, m

0 10 20 30 0 10 20 30 40 50
5 5
Calculated Calculated
Exp fit curve Exp fit curve
Effective mean incidence α, degs

4 4
3 Min Clearance 3
Min Clearance
2 2
α = 3.8/ [1 - exp(-9 z/b)] α = 3.3/ [1 - exp(-9 z/b)]

1 1
CL = 0.6 CL = 0.6
0 0
0.1 0.3 0.5 0.7 0.1 0.3 0.5 0.7
Ground clearance, z/b Ground clearance, z/b
(a) A300-600 wing (b) A380-861 wing
Height above ground, m Height above ground, m

0 10 20 30 40 0 5 10 15
5 5
Exp fit curve Exp fit curve
4 4
Min Clearance
3 3
2 2
α = 4/[1 - exp(-9 z/b)] α = 4.3/ [1 - exp(-9 z/b)]
1 1
Min Clearance
CL = 0.6 CL = 0.6
0 0
0.1 0.3 0.5 0.7 0.1 0.3 0.5 0.7
Ground clearance, z/b Ground clearance, z/b
(c) B777-300 wing (d) G550 wing
Figure 4.4. Effect of ground clearance on the effective mean incidence for fixed CL
(calculated).
TRAILING-EDGE FLAPS. There are various types of flaps, identified by the number
of independent lifting surfaces (single-, double- and triple-slotted) and by their
mechanics (split-flap, Fowler flap, and so on).
Without going into such detail at this time, during take-off, climb out, final
approach, landing and manoeuvring, the aircraft will have at least part of its control
surfaces deployed. Most flight-operating manuals have limitations restricting the use
of flap above a certain altitude and air speed. The manufacturers do not demonstrate,
test or certify operations with flaps deployed at unconventional points in the flight
envelope.
There are two classes of methods to consider in the present context: 1) semi-
empirical methods, such as ESDU; and 2) first-order aerodynamic calculations based
on the lifting-surface theory. Each class of methods has its advantages and drawbacks.
For example, the ESDU has no physics and relies on the engineering interpretation
and interpolation of known wing performance. The essential aspect is the calculation
of the drag increase at zero angle of attack of the main wing for a flap or slat deflection
of a given angle. Each deflection causes a change in the effective wing chord (and
hence the effective wing area). However, the coefficients must be referred to the
original wing area.
From the ESDU method, the increment in lift at α = 0, due to a trailing-edge
flap deflection δ f is calculated from

c c
CL = CL = CLδ δ f , (4.7)
c c
where CL is the increment in lift coefficient at α = 0 based on the effective chord
c ; CLδ is the lift-curve slope due to flap deflection δ f . The term CLδ is calculated from
Glauert’s thin-airfoil theory:

CLδ = 2 π cos−1 cc + (1 − cc2 )1/2 , (4.8)
with

c
cc = 2 − 1. (4.9)
c
For a three-dimensional wing we need to consider a number of corrections, essen-
tially due to the finite span of the flap, its geometrical arrangement and the tip effects.
A useful expression is provided by ESDU 2 , on the basis of several published and
unpublished data:

CLα
CL(wing) = k f CL J po (1 − kto ) (o − i ), (4.10)
2π
where k f is flap-type factor; J po is the efficiency factor of a plain flap on an unswept
wing; kto is a correlation factor that depends on the geometry of the wing; i and
o are part-span factors depending on the inboard and outboard position of the
span and the wing’s planform. The limits of validity of Equation 4.10 are given by
ESDU 2 .
The deployment of flaps, in addition to leading-edge slats, increases the lift to
the highest level. If the spoilers are deployed, the lift decreases to an intermediate
level. The combined effect of leading-edge slats, trailing-edge flaps and inlay spoilers
is shown in Figure 4.5.
Along with a drastic change in lift characteristics, a spoiler deployment causes
a large increase in drag. Practical methods for calculating the spoiler characteris-
tics in the ground run are available from ESDU 3 . The lift is given from empirical
correlations

CL = − c1 CLs f + c2 CLflap OGE , (4.11)
where CLs f is the increment in CL due to the deployment of split flaps having the
same geometry as the spoilers; CLflap is the increment in CL due to flap deflection
only (spoiler retracted). The numerical factors c1 and c2 in Equation 4.11 are derived
empirically. The aerodynamic coefficients on the right-hand side of Equation 4.11
are calculated out of ground effect at the angle of attack corresponding to the ground
ed
ploy
de ye
d
ps
Fla e plo
& sd
ats oil
er
Sl p
S
s,
F lap
,
Wing C L δCL Sl
ats
ye
d
e plo
d
ats
Sl
δ CLF
CLsl α in ground roll
Wing incidence
Figure 4.5. Effect of high-lift devices and spoilers on wing lift.
attitude of the aircraft. The change in CL due to the deployment of flaps and spoilers
is

CLr = CLflap IGE + CL, (4.12)
where, in this case, CLflap denotes the change in CL due to flap deployment in ground
effect. If these data are unknown, it is suggested to use the following equation:

CLflap IGE 1.15 CLflap OGE . (4.13)
The net coefficient on the ground due to the deployment of all of the high-lift systems
and the spoilers is
CL = CLslat + CLr , (4.14)
where CLslat is the lift coefficient of the wing with leading-edge slats deployed and
CLr is taken from Equation 4.12.
Figure 4.6 shows the high-lift polar of a scaled Airbus A320 with flaps deployed
in landing and take-off configuration. The wind-tunnel data 4 are compared with the
calculated landing performance and show an acceptable agreement over a wide range
of lift coefficients, although there are stronger non-linear effects in the wind-tunnel
data at CL > 1.9.
4.1.4 Maximum Lift Coefficient

The models presented so far do not include provisions for the determination of the
maximum lift coefficient, CLmax . This calculation is generally not possible with low-
order methods, and even sophisticated aerodynamic methods may fail to provide
acceptable answers, with the possible exception of two-dimensional airfoils at high
Reynolds numbers. The theoretical value of the CL shown in Figure 4.6 would
Figure 4.6. High-lift polar (landing and take-off) of a scale model of the Airbus A320.
continue to grow, unless a limiter is set. Again, we need to refer to semi-empirical

methods, such as those by ESDU 5 , which cannot be reviewed here.
4.2 Aircraft Drag

There are two distinct ways of calculating the drag of an aircraft, although both
methods rely on the concept of components. These components are either the phys-
ical components of the aircraft or the physical components of the drag for a given
system 6 .
Figure 4.7 shows the logical connection between drag components in a clean
configuration. In Figure 4.8 there is the combination of components at take-off and
landing and manoeuvre. The remaining components are due to the atmospheric
conditions and the state of the runway.
4.2.1 Lift-Induced Drag

Even within the constraints of our geometry model (Chapter 2), the lifting surface
method described earlier is a better alternative for the calculation of the lift-induced
drag. This is done on the Treffz plane by integrating the downwash velocity through
the wing span. The lift-induced factor k is calculated from
CDi
k= . (4.15)
CL2
Equation 4.15 shows that the drag equation is parabolic. This equation includes all
of the effects of twist, dihedral and winglets. When such a computational method
is not available, we are forced to use some approximations from known closed-
form solutions in low-speed aerodynamics. For example, the lift-induced drag of a
Figure 4.7. Model for the aerodynamic drag of a transport aircraft, clean configuration.
Figure 4.8. Model for the aerodynamic drag of a transport aircraft, take-off, landing and/or
manoeuvre configurations.
1 0.1
Flight data
Lift-induced factor, k
0.8
0.08
0.6
CD / CL2
CL2
0.06
0.4
0.04
0.2
0 0.02
0 0.01 0.02 0.03 0.04 0.05
CD
Figure 4.9. Aerodynamic polar of model Douglas DC-10 airplane. Adapted from Callaghan 7 .
conventional wing system is

1 CL2
CDi = . (4.16)
e πAR
Suitable values of the factor e are e = 0.90 at take-off, e = 0.95 at cruise. From
Equation 4.16, the lift-induced drag factor k becomes
1 1
k= . (4.17)
e πAR
Figure 4.9 shows the correlation between experimental data points of the Douglas
DC-10 airplane 7 and the parabolic drag model in clean configuration
CD = CDo + kCL2 . (4.18)
This equation is verified by the linearity between CD and CL2 (black points); the
induced-drag factor is “nearly” constant, except at low values of the CD (white
points).
4.2.2 Profile Drag

The profile drag of the aircraft is calculated by summing the contribution of all of
the components. For a cruise configuration we have
CDo = CDfuse + CDwing + CDht + CDvt + CDpylon + CDnac . (4.19)
The engine’s drag is accounted for in the calculation of the net thrust; therefore, it
is not added to Equation 4.19. If the aircraft has a take-off or landing configuration,
then we need to add the effects of the control surfaces and the landing gear. There
is some physical justification in Equation 4.19 but, ultimately, we need to rely on

some form of empirical correlations. The key aspect of the calculation of the CDo
is the wetted area and its breakdown in components (which have been derived in
Chapter 2) and the average skin friction.
Most skin-friction components are calculated according to

1 l
cf = c f (x)dx, (4.20)
l o
where c f is the average skin-friction coefficient calculated from an equivalent flat
plate of length l; this length varies according to the component. The Reynolds
number based on the reference length is Re L = ρUl/μ. The c f can be calculated
from a variety of semi-empirical expressions that are dependent on Re L. For incom-
pressible boundary layers, semi-empirical equations for the c f (Re L) are given by
Prandtl-Schlichting, Shultz and Grünow, Kármán-Schoenerr and others 8 . At flight
Reynolds numbers these equations are equivalent to each other.
Wing and Tail Surfaces

The wing is the most highly engineered aerodynamic system. Again, one is tempted
to look at the details of the wing section, but these details are not readily available.
Nevertheless, we can take a generic supercritical wing section of moderate thickness
and apply it to all transport aircraft. We use a semi-empirical approach based on
the Eckert Reference Temperature for the laminar flow, as shown by White 9 and
Nielsen 10 ; we use the van Driest II theory 11;12 for the turbulent flow. The calculation
method will be based on 1) laminar skin friction, 2) turbulent skin friction, 3) laminar-
turbulent transition point, and 4) correction due to form factors.
1. LAMINAR SKIN FRICTION. The first aspect to consider is the flat-plate approximation.
If a suitable equivalent flat plate has been defined, the case is treated as a zero-
gradient compressible flow at a free stream Mach number M∞ and flight altitude.
The local skin-friction coefficient is based on the Blasius boundary layer, with an
additional factor C ∗ (Chapman-Rubesin constant)
0.664 √ ∗
cf = 1/2
C . (4.21)
Rex
The Chapman-Rubesin factor is related to the ratio between the reference temper-
ature T ∗ and the external temperature Te , or the temperature at the edge of the
boundary layer. Essentially, at the flight altitude this temperature is reduced to the
air temperature. The value of C ∗ resulting from the theory is
∗ 1/2
∗ T 1 + 200Te
C = , (4.22)
Te T ∗ /Te + 200/Te
with

T∗ Tw
= 0.5 + 2
0.039M∞ + 0.5 . (4.23)
Te Te
In Equation 4.23 Tw is the wall temperature. Due to the heat transfer between
flow and solid walls at compressible speeds, the wall temperature is different from
the external temperature. Due to further heat transfer between the plate and the
0.975
C*
0.95
0.925
h = 9,000 m
h = 10,000 m
Note: T w = Taw h = 11,000 m
0.9
0 0.5 1 1.5 2
Mach number
Figure 4.10. Chapman-Rubesin factor as a function of Mach number.
surrounding volume, the wall temperature can be different from the adiabatic tem-
perature, Taw . The latter quantity can be calculated from the compressible-flow
relation:
Taw γ −1 2
= 1+r M∞ , (4.24)
Te 2
where r = Pr1/2 is the recovery factor and Pr is the Prandtl number. Equation 4.24
can be rewritten to show the ratio Tw /Taw or its inverse, so that we can make
corrections whenever the wall temperature is known. In fact, multiply and divide
Equation 4.24 by Tw and re-arrange
Taw Taw Tw
= , (4.25)
Te Tw Te

Tw Tw γ −1 2
= 1+r M∞ . (4.26)
Te Taw 2
For the calculation of the average skin friction (Equation 4.20), the extent of the
laminar region must be estimated. In conclusion, the Chapman-Rubesin factor can
be interpreted as the ratio between the turbulent skin friction at compressible and
incompressible speeds. The computational procedure is the following:
1. Set the streamwise position and calculate Rex .

2. Calculate the ratio Tw /Te from Equation 4.26, with Taw /Tw = 1.
3. Calculate the ratio T ∗ /Te from Equation 4.23.
4. Calculate the Chapman-Rubesin factor C ∗ from Equation 4.22.
5. Calculate c f from Equation 4.21.
Figure 4.10 shows the Chapman-Rubesin factor as a function of the Mach num-
ber. The flight altitude determines the external temperature and has some effect at
high Mach numbers only.
-6
Rex x 10
0 2 4 6 8
1
0.9
Cf / Cf, inc
0.8
0.7
0.6
0 0.5 1 1.5 2
Mach number
Figure 4.11. Turbulent skin friction as a function of Mach number.
Figure 4.11 shows the ratio between compressible and incompressible c f as a

function of the Mach number. In this case, the flight altitude does not have an effect.
2. TURBULENT SKIN FRICTION. The turbulent skin friction is calculated from the van
Driest theory, in the implementation shown by Hopkins and Inouye 13;14 , to whom we
refer for the numerical details. In brief, the incompressible skin-friction coefficient
is calculated from the Kármán-Schoenerr formula
0.242
= log10 ( R̂e x ĉ f ), (4.27)
ĉ f
where the hat quantities denote incompressible conditions. The relationship between
incompressible and compressible Reynolds numbers is
R̂e x = Fx Rex . (4.28)
Equation 4.27 is implicit in ĉ f and must be solved iteratively, for example, by using a
Newton-Raphson method from a first-guess solution. The solution of Equation 4.27
is related to the compressible skin friction by
ĉ f
cf = , (4.29)
Fc
where the factor Fc and the factor Fx are the key to the whole method; Fc is
calculated from
⎧
⎨ m/(sin−1 α + sin−1 β)2 , Me > 0.1
Fc = √ 2 , (4.30)
⎩ 0.25 1 + ϑ , Me < 0.1
where ϑ is a relative temperature:

Tw
ϑ= ,
Te
and

γ −1 2
m=r 1+ M∞ ,
2
2A2 − B B
α= , β= , (4.31)
(4A2 + B2 )1/2 (4A2 + B2 )1/2
m 1/2 1 + m− ϑ
A= , B= . (4.32)
ϑ ϑ
The Reynolds factor Fx is a ratio between dynamic viscosity and is calculated from
Fμ μw
Fx = , Fμ = . (4.33)
Fc μe
These viscosities are calculated with Sutherland’s law (§ 8.1.1). In summary, the
computational procedure is the following:
1. Set the streamwise position and calculate Rex .

2. Calculate the factor Fc from Equation 4.30.
3. Calculate the factor Fx from Equation 4.33.
4. Calculate R̂e x from Equation 4.28.
5. Solve Equation 4.27 using the Newton-Raphson method.
6. Calculate the compressible c f from Equation 4.29.
This procedure is valid at a given streamwise position x. Integration over the

turbulent section of the wing yields the average value of the c f and hence the
turbulent CDo . To make the procedure more accurate, we can divide the wing in
streamwise strips and integrate both streamwise and spanwise.
3. LAMINAR-TURBULENT TRANSITION. Now we need an estimate of the laminar-

turbulent transition point. This aspect can be subject to much debate, and it is
in fact one of the most complex problems in aerodynamics. Again, we consider the
flow past a flat plate and use the Blumer–van Driest semi-empirical correlation 15 ,
which was verified by experimental data
√
1/2 132500 kt + 1 − 1
Retrans = 106
, (4.34)
39.2 kt2
where kt is an average level of free stream turbulence. For a value of kt = 0.2%,
we find Retrans 0.165 · 106 . This is a relatively high Reynolds number for turbulent
transition.
4. SHAPE EFFECTS. The skin friction calculated with this method applies only to a flat
plate without pressure gradients. To recover these gradients and the local curvature
effects of the boundary layer, it is customary to correct the c f with a form factor. A
suitable expression is the following:
f f = 1 + 2.7(t/c) + 100 (t/c)4 . (4.35)
For example, if the average wing thickness is t/c = 0.1, then f f 1.28. The calcula-
tion of the drag of the other lifting-surface units (horizontal and vertical tail) follows
the same method as the main wing.
Fuselage Drag
The fuselage contributes to the CDo through the skin friction and the base drag. For
convenience, the fuselage is divided into three main sections: forebody (or nose),
centre and after-body (or tail), as in Chapter 2. A first-order method for the nose
drag is the turbulent nose cone theory, derived first by van Driest and reported by
White 9 . The theory leads to a correlation between the flat-plate drag and the drag of
a nose cone at zero incidence. A necessary condition for this correlation is the same
Re L. For a turbulent flow
c f,cone
1.087 ÷ 1.176, (4.36)
c f, plate
depending on the state of the boundary layer; the approximation is rather poor. The
length of the plate is equal to the length of the√nose section. If the flow is laminar,
the ratio between skin-friction coefficients is 3. Clearly, this is not a good thing
because the turbulent c f calculated from Equation 4.36 is only 9 to 17% above the
reference value of the flat plate.
The drag of the central section can be calculated in at least two ways. First, by
using White’s semi-empirical expression for the drag of a very long cylinder
0.4
l −1/3
CD = 0.0015 + 0.30 + 0.015 Rel . (4.37)
d
Equation 4.37 is thought to be accurate to within 9%. The second option is to use
the turbulent skin friction from van Driest’s theory (as in the wing-drag analysis).
In this case, the calculation is done with a Reynolds number growing from the
value corresponding to the nose, Re = ρUlnose /μ. The same method is applied for
the calculation of the skin-friction drag of the tail section, with starting Reynolds
number Re = ρU(lnose + lcentr e )/μ.
BASE DRAG. A simple expression for the base drag of the fuselage is given by
Hoerner’s equation 16 :

0.029 Abase
CD = √ . (4.38)
cf Awet
However, this equation lacks accuracy, and more detailed methods are needed to
take into account effects such as the upsweep and the flow incidence. This is done,
for example, with ESDU 17 that provide technical details regarding the effects of the
upsweep and allow more precise calculations based on the actual geometry of the
aft fuselage. The drag increment due to upsweep is calculated from
Ap
CD = [G(α, βu , ) − G(α, 0, )] cd , (4.39)
Ar e f
where α is the angle of attack of the fuselage; βu is the mean upsweep angle; is
the downwash angle created by the wing half-way of the upswept section; Ar e f is the
cross-sectional area of the fuselage; Ap is the planform area of the upswept section
of the fuselage; and, finally, C D is the average drag coefficient of the rear fuselage
(empirical value). The function G is calculated from
sin(α − βu ) sin2 (α − βu − )
G(α, βu , ) = , (4.40)
cos βu
where G is a tabulated function of its arguments.
4.2.3 Wave Drag

The wave drag is the result of a system of shocks, weak as well as strong ones, and
shock-induced flow separation around the aircraft. In principle, all aircraft compon-
ents are subject to these effects. However, we will consider the separate contribution
from the lifting surfaces and from the fuselage forebody. Flight at higher Mach num-
bers, as is the case of high-performance aircraft, requires additional analysis from
the wing–body junction.
WAVE DRAG FROM LIFTING SURFACES. From a practical point of view, an essential
parameter is the divergence Mach number, defined by
t
Mdd = κ A − κCL − , (4.41)
c
with κ A variable from 0.87 (NACA 6-series airfoils) to about 0.95 (modern super-
critical wing sections) and κ is another factor that represents the effects of the
aerodynamic lift on the divergence Mach number
dMdd
= −κ. (4.42)
dCL
Average values for supercritical wing sections are κ = 0.1 − 0.14. Analysis of exper-
imental data for low-speed airfoils, such as NACA 0012 and 23012, indicates that
κ 0.2. A similar expression was reported by Malone and Mason 18 :
κA CL t/c
Mdd = −κ − . (4.43)
cos LE cos2 LE cos3 LE
Therefore, once the wing sweep is fixed and the CL is calculated at the flight condi-
tions, the divergence Mach number is only a function of the “technology level”.
There is a relationship between the wave drag and the critical Mach number.
This relationship is found in Hilton 19 :
CD = 20(M − Mc )4 , M > Mc . (4.44)

A relationship between Mc and Mdd can be calculated from Equation 4.44 and
Equation 4.43. In fact, derive Equation 4.44 and recall the definition of divergence
Mach number

dCD
= 80(Mdd − Mc )3 = 0.1, (4.45)
dM Mdd
Mc = Mdd − 0.108. (4.46)
The corresponding drag coefficient is now calculated from Equation 4.44. Note that
CDw = 0 at M ≤ Mc . This calculation must be repeated for the tail-plane and the fin.
FUSELAGE FOREBODY. An earlier example in the geometry analysis of the fuselage

of some commercial aircraft (§ 2.4.2) has highlighted that the forebody is a complex
shape that seldom can be associated to a recognised geometry. Thus, the determin-
ation of the wave drag should require higher-order methods for the solution of the
flow field. Due to the computational efforts required to carry out these simulations,
it is preferable in this context to outline criteria that allow the rapid estimation of the
required coefficients, on the basis of data sets produced by higher-order methods.
One such example is given in Ref. 20 , to which we refer for further details. The fore-
body drag is associated to a limited number of shape parameters, namely the fineness
l/d, the non-dimensional nose radius 2r/d and the free Mach stream number. For
conventional shapes, the wave drag of the forebody is small, up to M∞ 0.85.
4.2.4 Interference Drag

The interference drag is due to local flow disruption at the intersection between
major components, in particular at the fuselage–wing, fuselage–tail–plane, fuselage–
fin and fuselage–pylon intersections. To control separation, many airplanes have
fine-tuned details such as aft-fuselage fences, flow diverters on nacelles, wing fences,
vortex generators and blended wing-bodies. Many of these details are of second
order and are not easy to model within the present context. However, the problem
can be reduced to the determination of some interference factors, yielding a total
interference drag

CDint = CDint (i), (4.47)
i
where the sum is extended to all of the relevant interference cases. A number of
key cases are identified: intersection between a rectangular lifting wing and a wall;
intersection between a swept lifting wing and a wall. A basic case is a wing intersecting
a plane at 90 degrees. Hoerner 21 gives the following expression:
c2
CDint1 = c1 (t/c)3 + c2 . (4.48)
A
A number of semi-empirical corrections are then introduced to take into account
1) the effects of wing sweep; 2) the effects of the intersection angle (for example,
intersections at angles other than 90 degrees); and 3) the effects of wing lift. These
effects are given by the following equations, respectively:

c2
CDint2 = c3 2 + c4 , (4.49)
A
c2
CDint3 = c5 ϕ 2 + c6 ϕ , (4.50)
A
c2
CDint4 = c7 CL2 . (4.51)
A
For a lifting body intersecting another lifting body at 90 degrees, there is a further
term:
c2
CDint5 = c7 (t/c)4 + c8 (t/c)2 . (4.52)
A
The coefficients ci are determined by using best-fit of experimental data. Some of
these terms may be missing. For example, the fin produces no lift; hence Equa-
tion 4.51 yields zero. The effect of the dihedral ϕ is to reduce the interference drag.
The equations provided were derived for cases without fillets and represent the
worst possible scenario. The use of filleting at the junction can greatly reduce the
interference drag to about one-tenth. Therefore, before proceeding with the final
calculation of the interference drag, we need to check whether fillets are available.
Note that all of these quantities do not depend on the flight conditions (altitude
and Mach number), and therefore they are essentially constant values. More recent
studies on interference drag are available 22–25 but, ultimately, the accuracy of these
methods must be evaluated in the context of the accuracy achieved on the other
components.
4.2.5 Drag of the Control Surfaces

The control surfaces discussed in this section are “flat-panel” components: spoilers,
ailerons, elevators, rudders. Due to the limitation of the data that can be extracted
from the geometry model, also the flaps are effectively considered as flat panels. The
essential data of these components are the span, the average chord, the chordwise
and spanwise position and the deflection (five parameters per component). We dis-
cuss in particular the aerodynamic effects caused by the spoilers, which can be used
in conjunction with flaps and slats.
The term spoiler is used to denote a deployable panel mounted on the upper
surface of an aircraft wing. Spoilers are used mostly for braking the aircraft on
the ground upon landing. They disrupt the lift and create a down-force that adds
weight to the wheels. Added weight increases the rolling resistance and facilitates
the braking process. The effect on the drag is given by a combination of slats, flaps
and spoilers. The spoiler terms give rise to a drag increment

1 CL2
CD = CDo + CDout + K2 (CLr )2 + , (4.53)
2 π AR
where CDo is the increase in profile drag due to the deployment of slats and flaps for
those sections of the wing also affected by spoiler deflection; CDout is the increment
in profile drag due to the deflection of the flaps on those portions of the span not
affected by spoiler deflection. The last term in Equation 4.53 is the lift-induced
drag; in particular, this term denotes the lift-induced drag for the difference in lift
coefficient. The term K is a part-span factor. The complete calculation procedure,
the accuracy and the limitations are discussed by ESDU 3 .
A spoiler deflection above 80 degrees is equivalent to a flat plate in normal
flow mounted on a wall (the wing’s upper surface). For this case some drag data are
available as a function of the aspect-ratio of the spoiler. For a ratio span/chord above
3, the normal force coefficient is basically constant, CN 1.2, based on the spoiler’s
area. Hence, the contribution of one spoiler to the overall drag is

Aspoiler
CD CN . (4.54)
A
The contribution to the lift is more difficult to estimate because the spoiler effectively
blocks the flow on the upper side of the wing. In the context of the whole system, we
can assume that the lift is reduced to zero along the span with the spoiler. Therefore,
the residual lift is

bspoiler
CL = CLg . (4.55)
b
The lift in ground effect is to be calculated by considering the flaps fully retracted.
Like any other aerodynamic device, the spoilers are not effective at low speed. There-
fore, during the braking process, they can be retracted when the speed decreases
below a threshold. The remaining phase of the braking is done with the wheel
brakes.
In the absence of more accurate data, the drag due to flap deflection can be
estimated from the following equation 26 :
c 1.38 A
flap flap
CD kflap sin2 δ f , (4.56)
c A
where kflap = 1.7 for plain and split flaps; kflap = 0.9 for slotted flaps. The calculation
must rely on a reasonable value of the lift coefficient in ground effect, which provides
the lift-induced contribution.
4.2.6 Landing-Gear Drag

The calculation of the landing-gear drag is an elaborate process, not without uncer-
tainties, and relies exclusively on semi-empirical equations. The problem is split
into manageable parts by considering isolated landing-gear units, as indicated in
Figure 4.8.
The most comprehensive source of information on this subject is ESDU 79015 27 ,
who provide various derivations and include interference factors. Roskam 28 provides
semi-empirical relationships that are valid for smaller airplanes, based on earlier
research, mostly at NACA. Yet, there is a limited amount of data to be used for
validation and verification, despite considerable research over many years; most of
the research focuses on aerodynamic noise and structural optimisation. A further
analysis of the landing-gear assembly is in Chapter 16.
Figure 4.12. Multi-axle landing gear off the ground; wheels not aligned with incoming flow.
The present discussion is limited to retractable landing gears for transport air-
craft. Each unit is made of a number of wheels, generally arranged in tandem; a
main vertical strut; horizontal axles in numbers depending on the number of wheels;
and various other struts, obliquely placed against the incoming flow (laterally or
longitudinally), along with a series of complicating factors and surface roughness
due to various systems. The whole unit must retract into a bay inside the fuselage
or the wing. Therefore, there is a cavity bay of appropriate shape and bay doors.
When the doors are open, they are roughly aligned with the incoming flow in order
to offer least resistance to the air. In some modern airplanes, the doors are sectioned
in several parts, some of which are closed to minimise the cavity drag.
For a landing gear placed under the wing, there are additional complications
arising from the effects of flap deflection and wing thickness; both contribute non-
linear effects to the drag. A systems approach to the determination of the drag is
shown in the top right corner in Figure 4.8.
In the process of determination of drag coefficients, reference quantities must
be assumed, which are inevitably dependent on the specific sub-system.
1. WHEEL DRAG. We must distinguish between rotating and non-rotating wheels and
between wheel arrangements on one, two or three axles. When there is more than
one axle and the wheels are above the ground, there is an angle between the line
through the axles and the incoming flow. This arrangement is shown in Figure 4.12,
which clearly indicates that the front wheels faces the free stream and the aft ones
are partially shielded. In practice, the drag of the front-facing wheels is higher. The
drag of a non-rotating multi-wheel combination is

D CD bu dw − mn
= CDo , (4.57)
q A bogie CDo A
Inclined strut Vertical strut Vertical strut
Inclined strut
lu
lu
dw dw
m m
Ground Ground
n n
bu bu
(a) (b)
Figure 4.13. Typical landing gear of transport aircraft.
where CDo = 1.2 for sub-critical Reynolds numbers (Rec < 5 · 105 ) and CDo = 0.65
for supercritical Reynolds numbers. The Reynolds number is calculated by using the
wheel’s diameter and the free stream velocity. The definition of the other quantities
is given in Figure 4.13. The parameter bu is always the maximum width of the wheel
combination. If s = dw /bu , then the ratio CD/CDo is calculated from

CD 0.642 − 0.2660s + 0.0846s 2 − 0.0081s 3 ,
0.3 < s < 4.0, Re < Rec
= .
CDo 0.912 − 0.4850s + 0.1390s 2 − 0.0111s 3 ,
0.3 < s < 5.0, Re > Rec
(4.58)
Note that these equations are polynomial approximations of reference data and are
only valid within the specified range. To take into account the presence of a plate
wall near the wheels, Equation 4.57 is multiplied by an interference factor c, whose
value is 2.5 when wheel and plane are in contact (that is, wheel on the ground) and
equal to 1 when the diameter dw is equal to the distance from the plane.
2. LANDING-GEAR STRUTS. The drag of the vertical post is calculated as an isolated

cylinder with a rough surface (to take into account pipes, brackets and other mech-
anical details). This is acceptable for cases in which the other main strut is lateral, as
shown in Figure 4.13. For cases where there is a tandem condition, as in Figure 4.12,
the spacing between struts, the diameter ratio and the slenderness of the struts must
be taken into account. The Reynolds number in this case is based on the strut dia-
meter. There is a critical Reynolds number around which there is a “jump” in drag.
Methods based on semi-empirical analysis exist 29 to calculate the aerodynamics of
various arrangements of finite-length cylinders.
3. CAVITY BAY AND DOORS. The dimensions of the bay can be inferred from the
dimensions of the landing gear and from the retraction mechanics. In fact, for a main
landing gear retracting sideways to its bay under the wing, the depth of the cavity
must be of the order of the width bu to accommodate the wheels. The width must be
about dw nw /2 (nw = number of wheels) and the spanwise extension must be of the
order of the main strut.
The flow around and inside the bay is unsteady; any attempt to refer to steady-
state of average characteristics is a challenge. The aerodynamic phenomena within
the bay are complicated by the presence of the strut and other mechanical details. In
some cases, the bays are occluded by partly closed doors, which may cause resonance
phenomena and considerable acoustic emission. With these caveats in mind, the drag
of the bay is written as

D bulu
= CDb , (4.59)
qA A
where the coefficients CDb is the drag coefficient of the bay at the relevant Reynolds
number and given geometrical characteristics. Approximate empirical equations for
the drag of the bay are the following:
⎧ 2 3
⎪
⎪ dw dw dw
⎪ −0.0131 + 0.2363
⎪ − 0.4007 + 0.1924 , bu /dw = 1.0
⎪
⎪ l l lu
⎪
⎪
u u
⎪
⎨ 2 3
dw dw dw
CDb 0.0192 + 0.0586 − 0.1360 + 0.0738 , bu /dw = 2.0
⎪
⎪ lu lu lu
⎪
⎪
⎪
⎪ 2 3
⎪
⎪ d d d
⎪
⎩ 0.0132 + 0.0537 w − 0.1227 w + 0.0648 w , bu /dw = 4.0
lu lu lu
(4.60)
Equation 4.60 is only valid in the range 0.25 < dw /lu < 1. If dw /lu > 1, assume
CDb 0.015. The bay doors will be considered aligned with the free stream, although
this is not always the case. The drag is essentially a profile drag and can be calculated
with the formula
0.455
CD = . (4.61)
log10 Rel2.58
In this case the Reynolds number is based on the length of the strut, lu .
Computational Procedure for Landing-Gear Drag

1. Set the basic geometrical data required for the landing gear.
2. Calculate drag of the isolated wheel unit, Equation 4.57.
3. Calculate drag of the vertical strut.
4. Calculate drag of the bay cavity, Equation 4.59.
5. Calculate drag of the bay doors.
6. Sum all drag components to find uninstalled drag of landing gear.
4. INSTALLATION EFFECTS. The installation drag for landing gear under the wing is
of the order of 10 to 15%. The flap deflection is another factor that at take-off
and landing is of the order of 0.6 to 0.7. The total landing-gear drag is found from
summing the drag of all units. ESDU reports that this method may yield results that
are off by 50% from the correct value, even when full consideration is given to each
component. Therefore, the accuracy of the results must be evaluated with flight data.
The wing-thickness effect is included with the following correction, extrapolated
from ESDU data sheets 27 :
f1 = 1 + (2.15 − 2.90833 tm + 1.91667 tm2 − 0.64167 tm3 + 0.08333 tm4 )(t/c), (4.62)
where tm = 2 lv-strut /MAC − 0.2 is the ratio between the length of the main vertical
strut and the mean aerodynamic chord; t/c is the average wing thickness. For a
landing gear upstream of the flap, the effect of flap deflection is calculated from
2
f2 = 1 − (0.0186 − 0.018t + 0.0053 t 2 ) Eδ f , (4.63)
with
E = 3.65833cm − 5.373cm
2
+ 2.91667cm
3
, (4.64)
and cm = cflap /MAC. Finally, we have
CD = CD f1 f2 . (4.65)
When the geometry of the landing gear is not known, we can use Torenbeek’s
semi-empirical expression (Torenbeek 30 , Appendix G):
D m0.785
= ku , (4.66)
qA A
where ku is a factor depending on the amount of flap deflection and m is the air-
craft’s mass. It can be assumed that ku varies linearly with the flap angle δ f . The
corresponding function is

δf
ku = 0.28 − 0.13 10−3 , (4.67)
δ f,max
with δ f the flap deflection in degrees. The main difference between Equation 4.66
and the expanded ESDU method is that the latter is not dependent on the aircraft’s
weight. Hence, a comparison between the two approaches is necessarily contrived.
Nevertheless, a comparison was made for the main landing gear of the reference
transport airplane, and the result is shown in Figure 4.14, at the weights indicated.
Equation 4.66 yields values of the CD within ±13%.
4.2.7 Environmental Effects

Environmental effects on the aircraft’s drag are important at take-off and land-
ing. These effects include the precipitation drag, a term that usually refers to the
change in response due to a combination of displacement and spray. The displace-
ment component is the drag produced when the aircraft rolls on a surface at least
partially covered by standing water, snow, slush or ice. A practical method for the
displacement drag consists of using the following equation:
1 ∗
D= ρ Styre CD KU 2 . (4.68)
2
0.015
Present Calculation
Torenbeek, m = 170,000 kg
Torenbeek, m = 120,000 kg
0.0125
CD main carriage
0.01
0.0075
0.005
0 5 10 15 20 25
Flap deflection δf , degs
Figure 4.14. Main landing gear CD of reference transport airplane (calculated).
In Equation 4.68 Styre is the frontal area of the tyre; ρ ∗ is the density of the contami-
nant (water, snow, ice, and so on); CD 0.75 and K is a wheels coefficient, estimated
at 1.5 to 1.6. When referring to the whole aircraft, it might be more practical to use
the factor 0.75K, in addition to the other drag components, in order to have the
overall effect on the aircraft’s speed on the ground. A correction is required to refer
the coefficient to the wing area, so that the displacement CD is accounted for by
Styre
CD = 0.75K . (4.69)
A
A more detailed method was developed by van Es 31 for rolling on a snow-covered

runway. The method breaks down the displacement drag in components and grain-
density functions.
The spray drag is created by the impingement of the spray from the tyres onto
the lower parts of the aircraft (fuselage and wings). The spray drag can be estimated
from
CDspray 24lCDo , (4.70)
where l is the length of the fuselage behind the point at which the spray plume hits
the bottom of the fuselage; CDo is the skin-friction drag of the fuselage.
AQUAPLANING. Aquaplaning (or hydroplaning) depends on the aircraft’s speed and

on the tyre conditions. The critical speed required for aquaplaning can be estimated
from the following equation:
1/2
p
V = 17.5 , (4.71)
ρw
where p is the tyre pressure and ρw is the density of the water (or other contaminant);
p and ρw are in international units. Nevertheless, an opposite effect takes place
during aquaplaning: this is the drag due to the displaced water. The corresponding
drag coefficient can be estimated from Equation 4.68. More refined methods are
discussed in Chapter 9.
4.2.8 Other Drag Components

Additional drag arises mostly from unwanted effects, such as aerodynamic deterior-
ation and unavoidable configuration details. In the first category we have mis-rigging
of components (especially flaps and cargo doors), surface roughness, skin dents, loss
of seals and paint peeling. In the second category we have probes, antennas and
design gaps. These spurious components may add a few drag counts to the total
drag. They are always difficult to estimate and, in preliminary analysis, they can
be neglected altogether. The manufacturers generally provide data such as addi-
tional fuel burn over one year of operation (for average utilisation rates) for specific
details. For example, Airbus reports that the A300 burns 90 kg of additional fuel
over a 2,000 n-mile trip for a slat mis-rigging of 15 mm. A first-order estimate of the
corresponding drag is the following:
m f 1
D = T = = ρ ACDU 2 , (4.72)
fj 2
with translates into
m f 1
CD = 0.06 drag counts. (4.73)
f j ρ AU 2
This value is considerably lower than the accuracy that can be achieved with the
methods presented in this chapter. Equation 4.73 shows an example of coupling
between aerodynamics and propulsion. Thus, the accuracy on the CD depends also
on the accuracy of the specific fuel consumption used in the analysis.
4.2.9 Case Study: Aerodynamics of the F4 Wind-Tunnel Model

The geometrical model for this configuration was discussed in § 2.7. That geometry,
which is just a wing–body combination, was the subject of extensive wind-tunnel
testing 32;33 and is now used to verify the aerodynamic models shown in this chapter.
Figure 4.15 shows the calculated aerodynamic polars and their comparison with
selected wind-tunnel data. Figure 4.15b shows the drag polar plotted as CD versus
CL2 . If the induced-drag factor were constant, as also shown in Figure 4.9, this curve
would be a straight line. Neither the wind-tunnel data nor the calculations show that
this is the case. The correlation between the theoretical model and wind-tunnel data
is completely acceptable, considering the simplification used and other inaccuracies
which may arise in other parts of the comprehensive flight mechanics model. Fur-
thermore, even sophisticated CFD models have difficulties in providing fail-proof
results 34–36 .
0.8 0.8
ONERA S2 data ONERA S2 data
RAE 8ft x 8ft data RAE 8ft x 8ft data
Calculated polar Calculated polar
0.6 0.6
2
CL
CL
0.4 0.4
0.2 0.2
0 0
0.01 0.02 0.03 0.04 0.05 0.06 0.01 0.02 0.03 0.04 0.05 0.06
CD CD
(a) (b)
Figure 4.15. Aerodynamic analysis of the DLR F4 wing–body airplane model.
4.2.10 Case Study: Drag Analysis of Transport Aircraft

We now apply our method to the calculation of drag at cruise conditions to a typical
aircraft. Examples for validation are scarce, but one useful set of data is the large
report by Hanke and Nordwall 37 for the Boeing B747-100. This report includes
most aerodynamic derivatives and parametric effects, not easily found for other
airplanes.
Figure 4.16 shows the comparison between the present method and the data
interpolated from Hanke and Nordwall for the Mach number effects on the CD at
constant CL. For this aircraft, MMO = 0.85; therefore, calculations at higher speeds
0.06
Present Model
Hanke & Nordwall (1970)
Cruise
0.04
CL = 0.6
CD
CL = 0.5
CL = 0.4
0.02
0.7 0.75 0.8 0.85 0.9 0.95

Mach number
Figure 4.16. Mach number effects on the CD at constant CL for the Boeing B747-100.
2 0.5
Calculated CL (right scale)
Calculated CD (left scale)
0.4
1.5
0.3
CD x 10
CL
0.2
0.5
0.1
0 0
-5 0 5 10 15 20
Incidence, degs
Figure 4.17. Lift and drag coefficients of the ATR72-500. Dashed lines are reference data,
adapted from Ref. 38 .
are not of practical interest. The typical cruise Mach number is shown in the graph
for clarity.
4.2.11 Case Study: Drag Analysis of the ATR72-500

The next case refers to the turboprop aircraft ATR72-500. “Reference” lift and drag
data have been inferred from Ref. 38 and are further examined in § 14.1, where we
discuss the effects of icing on the aircraft. The drag coefficient in these reference
data has been divided by 10 because this operation would give the correct order of
magnitude of the aircraft drag. Therefore, these “raw” data represent some form of
reference for our calculations, which are shown in Figure 4.17. The result shows a
good prediction for the CL and a slight under-prediction of the CD, by about 8 to
10%. As in many other cases, one should be careful before assessing this result in
too much detail or even attempting to “improve” the result without full knowledge
of the nature of the “reference” data.
4.2.12 Case Study: Drag Analysis of the Airbus A380-861

Figure 4.18 shows the major drag components of the Airbus A380-861 versus cruise
altitude. In particular, Figure 4.18a shows the drag counts and Figure 4.18b shows
the overall CD and the glide ratio. From this analysis, it can be inferred that both the
induced-drag component and the wave drag are strongly affected by altitude, whilst
the profile drag of the lifting surfaces is practically constant. The glide ratio reaches
a maximum at a certain altitude, and then it decreases.
L/D
10 12 14 16 18 20 22
13 13
40 40
11 11
Altitude, 10 3 feet
Altitude, 10 3 feet
35 35
Altitude, km
Altitude, km
9 30 9 30
Lift-induced
Fuselage 25 25
Wing
7 H-tail 7 CD (bottom scale)
Wave L/D (top scale)
20 20
0 50 100 150 0.02 0.025 0.03 0.035 0.04
Drag counts CD
(a) (b)
Figure 4.18. Drag analysis of the Airbus A380-861 at cruise conditions, M = 0.850.
SENSITIVITY ANALYSIS. We now carry out a sensitivity analysis to understand how the
profile drag depends on inaccuracies in the estimation of the wetted-area (§ 2.4). We
assume a 2% inaccuracy (in excess) on each of these components: fuselage, wing-
body, wing system, horizontal tail, vertical tail, nacelles, pylons. The flight conditions
are fixed; in this example we have M = 0.80 at 33,000 feet, W = 420.0 tons. The result
is shown in Table 4.1 for a correction of 2% on each of the wetted-area components.
In other words, if we selectively introduce a correction of 2% in each of the wetted
areas, the CDo changes by less than one drag count at the specified flight conditions;
CDo is the total profile drag coefficient. The result is dependent on the flight altitude.
4.3 Transonic Airfoil Model

The accurate calculation of high-speed rotors (propellers and fans) relies on airfoil
data over a range of Mach numbers. Although there are plenty of aerodynamic data
at low speeds, along with sophisticated methods that allow an accurate replication
of the test data, the transonic effects are more difficult to address 39 . In this section
Table 4.1. Profile drag sensitivity for the Airbus

A380-861 resulting from Awet = 2%. All drag
coefficients are given as drag counts
Configuration CDo CDo
Nominal 189.73
Fuselage sensitivity 190.55 0.82
Wing sensitivity 190.69 0.96
H-Tail sensitivity 189.84 0.11
V-Tail sensitivity 190.07 0.34
Nacelles sensitivity 189.98 0.25
Pylons sensitivity 189.95 0.22
Nacelles sensitivity 189.89 0.16
we present a semi-empirical method to generate airfoil charts over a full range of

Mach numbers. These charts are used in the propeller model (Chapter 6).
Some aspects of wind-tunnel testing, including a critical assessment of the data,
are available in McCroskey 40 (for the NACA 0012 airfoil) and Bousman 41 . Both
authors investigated the accuracy of about a dozen experimental data sets and extra-
polated some correlation curves. Both authors provide conclusions regarding the
validity of each data set. Yamauchi and Johnson 42 analysed the effects of Reynolds
number of minimum drag coefficient and maximum lift coefficient. The method of
Beddoes 43 is powerful for extrapolating the airfoil performance over a range of
Mach numbers and angles of attack if the separation point on the suction side of the
airfoil is known. To start with, the lift coefficient is expressed as
CL = CLo (M) + CLα (α, M)α. (4.74)
The zero-lift angle of attack is only weakly dependent on the Mach number; we can
assume CLo constant, or otherwise a useful approximation is
CLo
CLo = CLo (M∗ ) + (M − M∗ ) , (4.75)
M
with CLo /M 0.04. The lift-curve slope depends on both the angle of attack and
the Mach number. A suitable correction for the lift-curve slope is done by using the
Kármán-Tzien second-order equation
CLα (α, M∗ )
CLα (α, M) = , (4.76)
β2
with

β1 = 1 − M2 , (4.77)
1 M2
β2 = β1 + . (4.78)
2 1 + β1
This correction is valid at compressible Mach numbers lower than Mdd . If the angle
of attack is fixed, then the CL produced at Mach numbers is higher than the value
at the reference Mach M∗ . If M > Mdd , the procedure is modified. In fact, the drag
coefficient can be expressed as
CD(M) = CD(α, M∗ ) + CD(M). (4.79)
Two corrections are required to Equation 4.79. First, we need a Reynolds number
effect, due to an increased Mach number. Second, we need a Mach number correction
to operate around the drag divergence point and beyond.
Assume that the airfoil polar is calculated at a Mach number M∗ (typically,
∗
M = 0.2). The Reynolds number effect is calculated from the definition of Mach
number. In fact,
Ul a Mc
Rel = = (4.80)
ν ν
where c is the chord; a is the speed of sound. If the atmospheric conditions are fixed,
then
Re(M) = Re∗ + Re∗ (M − M∗ ). (4.81)
The profile drag coefficient scales with the Reynolds number as
1
CD ∝ . (4.82)
(log10 Re)2.548
Hence, the Mach number effect on the profile drag coefficient becomes
CD log10 Re∗
∗ = . (4.83)
CD log10 Re
The CD can be calculated with reasonable accuracy over a range of angles of attack at
low speeds. Note that it is not possible to make a correction on the drag component
dependent on the pressure. It will be assumed that there is no appreciable change
in pressure distribution with the increasing Mach number, although this cannot be
correct at transonic speeds. The next step is to calculate the divergence Mach number
of the airfoil. This is done through the use of Equation 4.43 with LE = 0, with κ A
variable from 0.87 (NACA 6-series airfoils) to about 0.95 (modern supercritical
wing sections). The CL to be used in Equation 4.43 must be the corrected value,
obtained by a combination of Equations 4.76 and 4.74 (recall that the effective angle
of attack is fixed). The relationship between the drag rise and the critical Mach
number is calculated again from Equation 4.46. The corresponding drag coefficient
is now calculated from Equation 4.44. The additional data required include relative
thickness t/c, reference Mach number M∗ , and reference Reynolds number Re∗ . The
factor κ A is a free parameter that must be chosen carefully because the transonic
effects are strongly dependent on it. This problem can be overcome if the Mdd is
known at one value of the CL.
From the analysis of several experiments, McCroskey 40 concluded that the best
curve fit for the lift-curve slope of the NACA 0012 is

Re
β1 CLα = 0.1025 + 0.00485 log , (4.84)
106
with a maximum error of 0.0029. Equation 4.84 uses the Prandtl-Glauert compress-
ibility correction (Equation 4.77) rather than the Kármán-Tzien correction. A result
of this theory is shown in Figure 4.19. The comparison is done with selected wind-
tunnel data 41 . Observe that the Reynolds number effects are captured correctly. The
calculation was based on “reference data” at Re∗ = 2 · 106 , M∗ = 0.3, α = 0 degrees,
and κ A = 0.87.
It is possible to verify that the choice of the factor κ A is correct. First, we need to
find a best fit of the wind-tunnel data; then, we calculate the Mdd ; and finally, we cal-
culate κ A by solving Equation 4.43. There is some arbitrariness in fitting the reference
data, but we find that the best correlation is a polynomial of order 3. This gives Mdd =
0.818. Solution of Equation 4.43 yields κ A 0.91, a value higher than the one used.
The application of this method to the generation of tabulated data is shown in
Figure 4.20 for the rotorcraft airfoil SC-1095, which has a 10% thickness ratio. We
show the effects of Mach number on the lift-curve slope and the CLmax (Figure 4.20a),
the drag polar at Mach numbers up to M = 0.80 (Figure 4.20b), the effects of Mach
0.04
Present calculation
NASA TP-2003-212265
0.03
CDo
0.02
0.01
0
0.2 0.4 0.6 0.8 1
Mach
Figure 4.19. Transonic drag rise of the airfoil SC-1095 and comparison with wind-tunnel data.
number at increasing lift coefficients (Figure 4.20c) and the pitching-moment diver-
gence at transonic Mach numbers (Figure 4.20d).
The comparison with wind-tunnel data show that 1) the Reynolds number effects
can be captured with the same order of accuracy as the wind-tunnel data; 2) the lift-
curve slope can be calculated with engineering accuracy before the lift stall; and 3)
the divergence Mach number can be calculated accurately at zero angle of attack.
For angles of attack other than zero it was not possible to make a verification of the
method.
The method proposed allows to generate tabulated data of airfoil aerodynamics
(CL, CLα , CD) over a wide range of angles of attack, Reynolds numbers and Mach
numbers. The data required include the airfoil thickness, the reference Reynolds
and Mach numbers, and the corresponding aerodynamic polar. Rotor calculations
based on combined momentum and blade-element theory with extrapolated airfoil
data at transonic Mach numbers provide the correct estimation of the main rotor
parameters, including the Mach number effects.
4.4 Aircraft Drag at Transonic and Supersonic Speeds

The determination of the drag of aircraft flying at transonic and supersonic speeds
is a far more complicated matter. The complications arise from the wide range of
Mach numbers at which the aircraft operate, the complexity of the wing system
(even in absence of external stores), the wide range of angles of attack at which a
high-performance aircraft is designed to operate and the substantial non-linearities
that occur at transonic Mach numbers.
There are some classical methods available in the field of supersonic aerodynam-
ics that are not possible to review here. The principle of components is still a useful
2 2
1 1
CL
CL
0 M = 0.20 0 M = 0.20
M = 0.30 M = 0.30
M = 0.40 M = 0.40
M = 0.50 M = 0.50
M = 0.60 M = 0.60
M = 0.70 M = 0.70
M = 0.80 M = 0.80
-1 -1
-10 0 10 20 0 0.025 0.05 0.075 0.1
Angle of attack, degs CD
(a) (b)
0.06 2
α = -2.0
α = 0.0
α = 2.0
α = 4.0
α = 6.0
α = 8.0
0.04 α = 10.0 1
CD
CL
0.02 0 M = 0.20
M = 0.30
M = 0.40
M = 0.50
M = 0.60
M = 0.70
M = 0.80
0 -1
0.2 0.4 0.6 0.8 1 -0.15 -0.1 -0.05 0 0.05
Mach CM
(c) (d)
Figure 4.20. Calculated aerodynamic polars of SC-1095.
method, but the interference factors may have a stronger influence at high speed;
if these factors are not properly addressed, the sum will not lead to an acceptable
result. General aerodynamic methods are beyond the scope of this book. For the
transonic drag we will assume the following equation:
CD = CDo + ηCLα (α − αo)2 , (4.85)
where all of the coefficients are a function of the Mach number. A typical behaviour
of the drag coefficients is displayed in Figure 4.21. The data in this figure, along with
Equation 4.85, will be used for rapid calculations of transonic manoeuvres, including
specific excess power.
The principle of drag decomposition can be applied (with some care) at super-
sonic Mach numbers. In the latter case, we need methods for the calculation of the
1 4.5 0.06
CLα
η
4
0.8 0.04
3.5
CDo
CLα
η
3
0.6 0.02
2.5
0.4 2 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Mach number Mach number
(a) (b)
Figure 4.21. Transonic and supersonic drag characteristics of a model airplane in clean con-
figuration.
drag of several streamlined components (wing, tail surfaces) and non-streamlined

components (external stores and fuel tanks).
The choice of the calculation method depends on the amount of information that
is available, the computer resources and the amount of time that can be invested in
software development. Geometrical data for high-performance aircraft are scarce.
They can be obtained almost exclusively with the same method considered for the
transport aircraft. A suitable approach, adequate for flight-performance calculations,
is based on a similarity rule, as described by ESDU 44 .
Finally, we show a typical behaviour of the maximum lift coefficient as a function
of the Mach number, Figure 4.22, which will be used for high-speed manoeuvre
calculations of a supersonic jet airplane. The transonic dip at 1.1 < M < 1.3 is a
cause of strong non-linearities in the flight performance.
4.4.1 Drag of Bodies of Revolution

Most external stores and equipment on high-speed aircraft are bodies of revolution
with pointed forebodies and various after-body shapes, including boat tails and blunt
bodies. There is a variety of cases that is not possible to review in this context. We
limit our discussion to a representative case, which is the body shown in Figure 4.23.
This is a body of revolution having a forebody with “minimum wave drag” (to
be established), a central cylindrical body, a boat tail and a blunt base. The body is
characterised by its cross-sectional area A(or diameter d) and length l. The forebody
can have a number of shapes, but the most favourable from the point of view of low
wave drag include the spherically blunted tangent ogive, the spherically blunted
paraboloid and a modified ellipsoid. The boat tail can be conical (as shown) or
circular arc or parabolic. The main characteristics of this body are overall length l
and maximum diameter d; forebody fineness ratio l/d; radius of curvature of the
spherical nose r (and nose-blunting ratio 2r/d); boat-tail angle β and base diameter
db. The length of the forebody is fixed; the length of the boat tail can be derived
1.4
1.2
Maximum CL 1
0.8
0.6
0.4
0.2
0 0.5 1 1.5 2
Mach number
Figure 4.22. Transonic and supersonic behaviour of the maximum CL of a model aircraft.
from the boat-tail angle β. The length of the cylindrical body is arbitrary, but it must
be at least ∼ 3d to be able to consider the various sections independently.
Our problem is the determination of the aerodynamic drag over the full range
of Mach numbers, up to supersonic speeds, as in the previous case (see Figure 4.21).
ESDU has a number of practical methods on both wave drag and after-body drag
calculations 20;45 – 47 . The computational method available in Missile DATCOM 48
can estimate the drag of a variety of shapes, angles of attack and Mach numbers.
The analytical methods of high-speed aerodynamics are limited to fully developed
supersonic flow and to a few optimal shapes, such as the von Kármán ogive and the
Sears-Haack body 49 . Then there is a wide body of experimental work covering a
number of ogive-like geometries over the full transonic regime 50;51 . In general, no
extrapolations are possible due to the non-linear nature of the transonic flow and
Forebody Cylindrical body Afterbody
d db
r ro
l
Figure 4.23. Typical low-drag body of revolution.
0.3 0.3
CDw
Total C D
0.2 0.2
CD
CDβ
0.1 0.1
CDb
0 0
0.4 0.8 1.2 1.6 2 0.4 0.8 1.2 1.6 2
(a) Components (b) Total
Figure 4.24. Drag coefficient of slender body of revolution as function of Mach number.
various important effects, not least the Reynolds number 52 . A modern analysis would
be based on CFD computations, that would be used to generate tabulated data of
aerodynamic coefficients over the full range of Mach numbers for any fixed geometry.
Even in this case, the results must be thoroughly validated with experimental data.
At a basic level, the drag components of the body shown in Figure 4.23 are
1) forebody wave drag CDw ; 2) skin-friction drag CDo ; 3) boat-tail drag CDβ ; and
4) base drag CDb. The second contribution is considerably smaller than the other
contributions; in first analysis, it can be neglected (or calculated with the method
shown in § 4.2.2). Thus, we consider the drag build-up with the following sum:
CD CDw + CDb + CDβ . (4.86)
CASE STUDY. We consider a spherically blunted secant ogive forebody with a conical
boat tail and a blunted base. The main characteristics of the body are forebody
fineness ratio l/d = 2 and nose-blunting ratio 2r/d = 0.2. We have used the methods
of ESDU (as cited), which are based on tabulated data of representative cases
calculated with either the full potential equations or the Euler equations. Thus, we
estimated the three drag contributions in Equation 4.86; these contributions are
shown in Figure 4.24. The dashed lines are our own extrapolation at Mach numbers
beyond the validity of the model. Note that the CD is referred to the cross-sectional
area of the body, πd2 /4. Therefore, rescaling to the wing area is done according to

π d2 /4
CD → CD . (4.87)
A
For example, a wing tank having a diameter d = 0.51 m mounted on an aircraft

having a gross wing area A = 28.9 m2 would require a scale factor ∼ 0.01. In this
specific case, the maximum CD 0.34 at M = 1.18 in Figure 4.24b would be scaled
to CD ∼ 0.0034, which is equivalent to 34 drag counts.
4.5 Buffet Boundaries 113
4.5 Buffet Boundaries

Buffet is the structural excitation caused by separated air flow. This excitation is
represented by an amplitude, a frequency and a spectrum. The phenomenon is due
a number of different events, such as upstream flow separation (tail-plane in the
wake of the wing) and downstream separation (flow separation in the upsweep sec-
tion of the fuselage). The spectrum is characterised by a broadband energy content,
although occasionally there can be discrete frequencies due to coherent vortex shed-
ding. Buffeting is the aeroelastic response to the buffet excitation. Buffet onset is
determined from vibration measurements and is conventionally given as a vibra-
tion measuring at least 0.2g at the pilot’s seat 53 . For a wing, the buffet excitation
parameter is

2 mz̈
St G(St) = √ ζ, (4.88)
π qA
where St is the Strouhal number, z̈ is the RMS acceleration of the wing tip, and ζ
is the damping ratio, or the ratio between the aeroelastic damping and the critical
damping.
A considerable amount of research has been done on flow-induced separation
and its effects on the structural response of aero structures, although most of it relies
on wind-tunnel experiments and flight testing. Numerical simulation methods have
lagged behind.
The discussion in this section is limited to subsonic transport aircraft. For this
case, it is generally accepted that the aircraft must be able to perform at least a 1.3g
manoeuvre in cruise configuration in response to buffeting. In other words, a 0.3g
manoeuvre margin is imposed for any given flight altitude and true Mach number
when buffeting occurs.
From a performance point of view, the best indicator of wing buffet is the beha-
viour of the CLmax at increasing Mach numbers. Here the variation can be quite
striking, depending on the wing section, the wing planform and other geometrical
factors. No generalisation is possible. However, there exists a Mach number above
which the lift drops (transonic dip), before eventually recovering in the low super-
sonic regions.
A typical buffet boundary for a transport aircraft looks like the one in Figure 4.25,
which represents the performance of the Douglas DC9-80 series of airplanes. At very
low Mach numbers, the usable CL has relatively high values (point A). This point
occurs around 1.2VS . At high subsonic speeds, the usable lift decreases dramatically
and upon reaching the Mach number at point B, there is virtually no lift available.
Most buffet curves can be defined through points A and B.
The buffet speed and Mach number are given by the intersection between the
buffet boundary and the curve defined by the manoeuvre equation

2nW
VB = a MB = . (4.89)
ρ ACLmax−buffet
For a fixed weight and flight altitude, VB = f (n, CLmax−buffet ). The normal load factor is
set to n = 1.3 to account for the manoeuvre margin just described. In general, there
will be two intersections, giving low and high buffet speeds.
1.5
1.2
Maximum CL 0.9
0.6
0.3
B
0
0 0.2 0.4 0.6 0.8
Mach number
Figure 4.25. Buffet boundary of the Douglas DC9-80 series (elaborated from the FCOM).
4.6 Aerodynamic Derivatives

We limit the discussion to the main control surfaces of a modern transport airplane
because they can be realistically modelled with the methods presented in this book.
We assume that wings are equipped with ailerons and spoilers; horizontal stabilisers
are equipped with elevators and vertical stabilisers have a rudder. In many cases
these controls are sectioned in a number of parts. For example, the rudder and the
elevator can have two or more sections. Elevators can also have small deflection tabs.
The aerodynamic derivatives will be used for steady-state aircraft trim, stability and
control. For dynamic effects, the reader is referred to books that specifically address
flight dynamics, such as Stengel 54 .
If Ca denotes a generic aerodynamic coefficient that can be expanded in series
around an initial value with respect to a control angle α, then a linear approximation
is the following:

∂Ca
Ca = Cao + α, (4.90)
∂α
where Cao denotes the coefficient Ca evaluated at α = αo. The derivative in Equa-
tion 4.90 is an aerodynamic derivative and can be used to analyse the aerodynamic
response of the system to sufficiently small perturbations of the state parameter α
around its initial value. If the aerodynamics is linear, Equation 4.90 is in fact quite
useful; if it is not, there are limitations to the extent of the perturbations that can be
introduced.
For a simple wing, we can define perturbations with respect to the angle of
attack α and the side-slip β. There are six aerodynamic coefficients: three force
coefficients and three moment coefficients. These quantities can be referred either
to body-conformal axes {xb, yb, zb} or wind axes {x, y, z}. In the latter case, we have
4.7 Float-Plane’s Hull Resistance in Water 115
CD, CY , CL for the force coefficients and Cm, Cn , Cl for the moment coefficients. The
aerodynamic derivatives are
{CD, CY , CL, Cm, Cn , Cl }α {CD, CY , CL, Cm, Cn , Cl }β . (4.91)
If the wing has a dynamic motion such a pitch rate p, roll rate q or yaw rate r, then
we have the derivatives with respect to these quantities. If the wing has an aileron,
there are changes in the coefficients due to aileron deflection δ; the corresponding
derivatives are:
{CD, CY , CL, Cm, Cn , Cl }δ . (4.92)
Therefore, there are a large number of derivatives to deal with and, if the process is
extended to a fully configured airplane, there can be in excess of 100 derivatives to
account for, some of which are not always useful.
As long as the aerodynamics is linear (a reasonable assumption for lifting sur-
faces at moderate angles of attack and low subsonic Mach numbers), the derivatives
can be calculated with a lifting-surface method. If the lifting surfaces are calcu-
lated separately, they need a reference area; if this is not the same for all of the
components, the coefficients will have to be rescaled to the same reference area.
4.7 Float-Plane’s Hull Resistance in Water

Because the drag coefficient will have to be rescaled with the wing area, it is prefer-
able to calculate the ratio D/q = CD A.
CDhull = CDo + CDw + kw CL2 hull , (4.93)
where CDw is the wave drag, kw is a lift-induced drag factor and CLhull is the lift
produced by the immersed hull. The profile drag coefficient is calculated from the
wetted area and an estimated value of average skin-friction coefficient as shown in
§ 4.2.2.
GEOMETRY CALCULATIONS. The wetted area to consider is the immersed area A∗wet .
The volume of the immersed floats, Vf∗loat , can be calculated from the buoyancy law
at MTOW. If W is this weight, then
W
Vf∗loat = . (4.94)
2gρw
The actual volume of the float would have to be larger than this figure, for at least two
reasons: first, the float can serve as a stepping point for boarding the aircraft; second,
a temporary excess weight would sink the floats. Thus, we will assume (somewhat
arbitrarily) that the actual volume of the floats is Vf loat 1.5Vf∗loat . We can now
define an equivalent cross-sectional area Aeq for a float of length l and an equivalent
radius req
1/2
Vf loat Aeq
Aeq = , req = . (4.95)
l f loat π
The total wetted area is Awet 2πreq l f loat . The wetted area of the immersed portion
of the floats is
∗
Vf loat 1/2
A∗wet l f loat . (4.96)
πl f loat
WAVE DRAG. Calculation of the wave drag of a ship hull is generally a fairly complic-
ated problem. In this context, it will suffice to use a simplified expression of a theory
of thin-ship resistance (Michell-Havelock), which is given by an improper integral†
∞
4 x2
(1 − cos x)e−ax
2
CDw = dx (4.97)
πC p xo
2
x 2 − xo2
with
2
V Cp U 2δ Fr
Cp = , xo = , Fr = √ , a= . (4.98)
Aeq l Fr2 gl l Cp
In these equations, the parameter “Fr” denotes the Froude number; V is the sub-
merged volume; l is the length of the submerged hull; Aeq is an equivalent (or
maximum) cross-sectional area; and δ is the depth of the centroid of the maximum
cross section Aeq . Equation 4.97 contains three important terms: 1) an oscillating
√
factor (1 − cos x); 2) a damping factor exp(−ax 2 ); and 3) the improper factor x 2 /
that becomes singular at x = xo. The integral in Equation 4.97 is more specifically
called improper Riemann integral of the second kind and causes some difficulties.
However, the solution of Equation 4.97 shows that above a critical value of the
speed, the drag decreases uniformly.
In some cases, there can be interference between the internal leading-edge waves
because they intersect each other and travel to the opposite float. This problem is
similar to the Busemann biplane, in which mutual interference between the systems
of shocks created by two slender bodies can lead to a decrease in total wave drag,
if the system is properly designed (see, for example, Liepmann & Roshko 56 ). A
further analysis of the floats is shown in § 15.11, where we examine their effect on
the mission range of a propeller airplane.
4.8 Vortex Wakes

An important side effect of lift generation by an aircraft wing is the wake downwash
and the tip vortices. The downwash is the vertical component of the air speed behind
the airplane. The tip vortices are generated by the pressure differences between the
suction and the pressure side of the wing; they consist of a pair of counter-rotating
swirling vortices and their strength increases with the wing loading.
From low-speed aerodynamic theory we can derive a first-order estimate of the
downwash. The ideal condition, resulting from elliptic loading, yields the following
expression of the average normalised downwash, or induced inflow angle (see also
§ 7.1):
w 2L 2W
αi = = . (4.99)
U πρb U
2 2 πρb2 U 2
† Alternative expressions are available for this equation; see Newman 55 .
4.8 Vortex Wakes 117
0.04
CL = 0.50
CL = 0.60
Induced angle αi , rad

0
-0.04
-0.08
0 10 20 30 40 50
0 10 20 30 40
Spanwise coordinate, m
Figure 4.26. Calculated downwash of the Airbus A380-861 wing.
The induced inflow angle is associated with the local circulation. Therefore, in the
general case, it is possible to calculate a local induced angle and a local downwash.
This was done for the Airbus A380-861 and is shown in Figure 4.26 for two values of
the CL by using a lifting-surface method. The average downwash w can be calculated
from integrating the local downwash w(y). A rapid variation of the downwash occurs
at the tip, where the air flow is prevented from rolling up too quickly by the presence
of the winglet.
The key operational parameters in the determination of the strength of the
downwash is the air speed, the gross weight and the flight altitude, Equation 4.99.
If the downwash is too strong, it may cause a hazard to aircraft following behind
or crossing the flight path at a lower altitude. For this reason, there are regula-
tions in place that specify safe separation distances, according to the data given in
Table 4.2. Currently, there are only three weight categories: heavy refers to a weight
> 136,000 kg; medium refers to a weight 7,000 kg < W < 136,000 kg; light refers to
W < 7,000 kg. A special category exists for the Airbus A380 (super heavy). For the
cases not listed in this table, which refer mostly to light aircraft, the minimum radar
separation is enforced. The different weight classes and separation times are under
review and may change in the future. However, the scientific basis for the vortex
wakes is well established 57 – 59 .Considerable effort has gone into the analysis of vor-
tex decay, which appears to depend on several atmospheric parameters, including
the background turbulence, the aircraft-induced turbulence, the ambient stratifica-
tion and the wind shear. Vortex wakes are known to interact with the surrounding
environment, including the ground and bodies of water.
To have an idea about the effect behind the aircraft, consider the case of the
Airbus A380 on a climb-out at 3,000 feet at 220 KTAS and a gross weight of 380
tons. The average induced angle would be w/U ∼ 0.025, which corresponds to a
Table 4.2. Aircraft separation following ICAO rules
Xsep tsep
Flight Leader Follower [n-m] [min]
Heavy 2
Take-off Heavy Medium 2
Light 2
Medium Light 3
Heavy 2
Landing Heavy Medium 2
Light 3
Medium Light 3
Heavy 4
Cruise Heavy Medium 5
Light 6
Medium Light 5
w ∼ 3 m/s, or a vertical mass flow rate dm/dt = ṁ = ρwUb ∼ 31,100 kg/s, or 27,000
m3 /s. If this downwash is incurred by a following aircraft at a similar speed, a loss in
CL ∼ −0.1 should be expected.
Summary
We presented comprehensive methods to be used for the prediction of the aerody-
namic coefficients and their derivatives of a conventional airplane configuration. The
lift has been elaborated from cruise conditions, as well as from the (approximate)
effect of augmentation caused by flaps, slats and other devices. We have included
the effects of the ground on the effective inflow angle. The drag analysis is somewhat
more complex. It is calculated with the method of components. A separate model
is developed for the transonic drag rise. The model is used both for the lifting sur-
faces and the blade sections of a propeller. A number of verification cases have been
shown; these cases demonstrate the suitability of the models for aircraft performance
work. We have shown relatively simple models for bodies of revolution (to be used
in the drag estimation of the external tanks and other stores) and for a float plane in
water. Finally, we have elaborated on the role of the vortex wakes from the point of
view of the hazard created by trailing airplanes. We have explained that a minimum
separation is required to limit the changes in aerodynamic inflow conditions that
could cause the trailing aircraft to stall or lose control.
Bibliography
[1] Katz J and Plotkin A. Low Speed Aerodynamics. McGraw-Hill, 1992.
[2] ESDU. Wing Lift Coefficient Increment at Zero Angle of Attack due to Deploy-
ment of Plain Trailing-Edge Flaps at Low Speeds. Data Item 97011. ESDU
International, London, Nov. 2003.
[3] ESDU. Lift and Drag due to Spoiler Operation in the Ground Run. Data Item
76026. ESDU International, May 1977.
Bibliography 119
[4] Kiock R. The ALVAST model of DLR:. Technical Report IB 129 96/22, DLR,
Lilienthal Platz, 7, D-38018 Braunschweig, Germany, 1996.
[5] ESDU. Increments in Aerofoil Lift Coefficient at Zero Angle of Attack and in
Maximum Lift Coefficient Due to Deployment of a Double-Slotted or Triple-
Slotted Trailing-Edge Flap, with or without a Leading-Edge High-Lift Device,
at Low Speeds. Data Item 94031. ESDU International, London, Dec. 1994.
[6] ESDU. Estimation of Airframe Drag by Summation of Components: Principles
and Examples. Data Item 97016. ESDU International, London, Dec. 1996.
[7] Callaghan JG. Aerodynamic prediction methods for aircraft at low speeds with
mechanical high lift devices. In Prediction Methods for Aircraft Aerodynamic
Characteristics, volume AGARD LS-67, pages 2.1–2.52, May 1974.
[8] White F. Viscous Fluid Flow. McGraw-Hill, 1974.
[9] White F. Viscous Fluid Flow. McGraw-Hill, 1974. Chapter 7.
[10] Nielsen J. Missile Aerodynamics. McGraw-Hill, 1960. Chapter 9.
[11] van Driest ER. The problem of aerodynamic heating. Aeronaut. Eng. Rev.,
15:26–41, 1956.
[12] van Driest ER. On turbulent flow near a wall. J. Aero. Sci., 23(11):1007–1011,
1956.
[13] Hopkins EJ and Inouye M. An evaluation of theories for predicting turbulent
skin friction and heat transfer on flat plates at supersonic and hypersonic Mach
numbers. AIAA J., 9(6):993–1003, June 1971.
[14] Hopkins EJ. Charts for predicting turbulent skin friction from the van Driest
method II. Technical Report TN-D-6945, NASA, Oct. 1972.
[15] Blumer CB and van Driest ER. Boundary layer transition – Freestream turbu-
lence and pressure effects. AIAA J., 1(6):1303–1306, 1963.
[16] Hoerner SF. Fluid Dynamic Drag. Published by the Author, Bricktown, NJ,
1965.
[17] ESDU. Drag Increment due to Fuselage Upsweep. Data Item 80006. ESDU
International, London, Feb. 1988.
[18] Malone B and Mason WH. Multidisciplinary optimization in aircraft design
using analytic technology models. J. Aircraft, 32(2):431–437, March 1995.
[19] Hilton WF. High Speed Aerodynamics. Longmans & Co, London, 1952.
[20] ESDU. Forebodies of Fineness Ratio 1.0, 1.5 and 2.0, Having Low Values of
Wave Drag Coefficient at Transonic Speeds. Data Item 79004. ESDU Interna-
tional, London, June 1979.
[21] Hoerner SF. Fluid Dynamic Drag. Published by the Author, Bricktown, NJ,
1965. Chapter 8.
[22] Tétrault PA, Schetz JA, and Grossman B. Numerical prediction of interference
drag of strut-surface intersection in transonic flow. AIAA J., 39(5):857–864, May
2001.
[23] Kubendran L, McMahon H, and Hubbard J. Interference drag in a simulated
wing-fuselage junction. Technical Report CR-3811, NASA, 1984.
[24] Barber TJ. An investigation of wall-strut intersection losses. J. Aircraft,
15(10):676–681, Oct. 1978.
[25] Sakellaridis A and Lazaridis A. Experimental study of interference drag for
multi-element objects. Exp. Thermal & Fluid Science, 26:313–317, 2002.
[26] McCormick BW. Aerodynamics, Aeronautics and Flight Mechanics. John
Wiley, 2nd edition, 1995.
[27] ESDU. Undercarriage Drag Prediction Methods. Data Item 79015. ESDU
International, London, March 1987.
[28] Roskam J. Airplane Design, Part VI, Chapter 4. DARCorporation, 2000 (paper-
back edition).
[29] ESDU. Mean forces, pressures and moments for circular cylindrical structures:
finite-length cylinders in uniform flow. Data Item 81017, Amend. A. ESDU
International, London, May 1987.

1985.
[31] van Es GH. Method for predicting the rolling resistance of aircraft tires in dry
snow. J. Aircraft, 36(4):762–768, Oct. 1999.
[32] Redeker G. DLR-F4 wing-body configuration. In A Selection of Experimental
Test Cases for the Validation of CFD Codes, AGARD AR-303, Volume II,
pages B4–B21. Aug. 1994.
[33] Redeker G, Mueller R, Ashill PR, Elsenaar A, and Schmitt V. Experiments
on the DFVLR F4 wing body configuration in several European wind tunnels.
In Aerodynamic Data Accuracy and Quality: Requirements and Capabilities in
Wind Tunnel Testing, volume AGARD CP-429, July 1988.
[34] Langtry RB, Kuntz M, and Menter FR. Drag prediction of engine-airframe
interference effects with CFX-5. J. Aircraft, 42(6):1523–1529, Nov. 2005.
[35] O. Brodersen and A. Stürmer. Drag prediction of engine-airframe interfer-
ence effects using unstructured Navier-Stokes calculations. In 19th Applied
Aerodynamics Conference, AIAA Paper 2001-2414. Anaheim, CA, June 2001.
[36] Wurtzler KE and Morton SA. Accurate drag prediction using Cobalt. J. Air-
craft, 43(1):10–16, 2006.
[37] Hanke CR and Nordwall DR. The simulation of a large jet transport aircraft.
Vol. II: Modeling data. Technical Report D6-30643, N73-10027 Boeing Doc,
Sept. 1970.
[38] Caldarelli G. ATR-72 accident in Taiwan. In SAE Aircraft & Engine Icing
International Conference, ICE 13, Sevilla, Spain, Oct. 2007.
[39] Filippone A. Rapid estimation of airfoil aerodynamics for helicopter rotor
calculations. J. Aircraft, 45(4):1468–1472, July 2008.
[40] McCroskey J. A critical assessment of wind tunnel results for the NACA 0012
airfoil. Technical Report TM-100019, NASA, Oct. 1987.
[41] Bousman WG. Aerodynamic characteristics of SC1095 and SC1094-R8 airfoils.
Technical Report TP-2003-212265, NASA, Dec. 2003.
[42] Yamauchi GK and Johnson W. Trends of Reynolds number effects on two-
dimensional airfoil characteristics for helicopter rotor analyses. Technical
Report TM-84363, NASA, April 1983.
[43] Beddoes TS. Representation of airfoil behavior. Vertica, 7(2):183–197, 1983.
[44] ESDU. Similarity Rules for Application in Aircraft Performance Work. Data
item 97025. ESDU International, London, Sept. 2008.
[45] ESDU. The Wave Drag Coefficient of Spherically Blunted Secant Ogive Fore-
bodies of Fineness Ratio 1.0, 1.5 and 2.0 at Zero Incidence in Transonic Flow.
Data Item 89017. ESDU International, London, 1983.
[46] ESDU. Pressure Drag and Lift Contributions for Blunted Forebodies of Fine-
ness Ratio 2.0 for Transonic Flow (M∞ ≤ 1.4). Data Item 89033. ESDU Inter-
national, London, 1989.
[47] ESDU. Subsonic and transonic base and boat-tail pressure drag of cylindrical
bodies with circular-arc boat-tails. Data Item 96012. ESDU International,
London, 1996.
[48] Blake WB. Missile DATCOM: User’s manual – 1997 Fortran 90 revision. Tech-
nical report, US Air Force Research Lab, Air Vehicles Directorate, Wright-
Patterson Air Force Base, OH, 1998.
[49] Ashley H and Landahl M. Aerodynamics of Wings and Bodies. Addison-
Wesley, 1965.
[50] Harris RV and Landrum EJ. Drag characteristics of a series of low-drag bodies
of revolution at Mach numbers from 0.6 to 4.0. Technical Report TN-D-3163,
NASA, Dec. 1965.
[51] Wallskog HA and Hart RG. Investigation of the drag of blunt-noised bodies
of revolution in free flight at Mach numbers 0.6 to 2.3. Technical Report RM
L5314a, NACA, 1953.
[52] Bromm AF and Goodwind JM. Investigation at supersonic speeds of the vari-
ation with Reynolds number and Mach number of the total, base and skin-
friction drag of seven boattail bodies of revolution designed for minimum wave
drag. Technical Report TN 3708, NACA, 1956.
[53] ESDU. An Introduction to Aircraft Buffet and Buffetting. Data Item 87012.
ESDU International, London, 1987.
[54] Stengel R. Flight Dynamics. Princeton Univ. Press, 2004.
[55] Newman JN. Marine Hydrodynamics. The MIT Press, 1977.
[56] Liepmann H and Roshko A. Elements of Gas Dynamics. J. Wiley & Sons, 1983.
[57] Rossow V. Lift-generated vortex wakes of subsonic transport aircraft. Progr.
Aerospace Sciences, 35:507–660, Aug. 1999.
[58] Gerz T, Holzäpfel F, and Darracq D. Commercial aircraft wake vortices. Progr.
Aerospace Sciences, 38:181–208, 2002.
[59] Holzäpfel F. Probabilistic two-phase wake vortex decay and transport model.
J. Aircraft, 40, Mar. 2003.

a = speed of sound; term defined in Equation 4.98
a1 , a2 , a3 = factors in Equation 4.5
A = area; parameter defined in Equation 4.32
Ac1 = cross-sectional area of wing having unit chord
Aeq = equivalent area
Ap = planform area of the upswept section of the fuselage
Ar e f = maximum cross-sectional area of the fuselage
Aspoiler = planform area of a spoiler
Awet = wetted area
A∗wet = wetted area of immersed float
AR = wing aspect ratio, b2 /A
b = nominal wing span
bspoiler = lateral dimension of the spoiler
bu = width of retractable multi-wheel landing gear
bw = wing span over winglet
B = parameter defined in Equation 4.32
c = wing/airfoil chord
cc = corrected chord, Equation 4.9
c = effective chord
ci = constant coefficients, i = 1, 2, · · ·
cm = cflap /MAC, Equation 4.64
Ca = generic aerodynamic coefficient
CD = drag coefficient
CDb = drag coefficient of a landing-gear bay; base drag coefficient,
Equation 4.86
CDhull = hull-drag coefficient
CDi = induced-drag coefficient
CDint = interference-drag coefficient
CDo = profile-drag coefficient
CDw = wave drag; hull drag, Equation 4.97
CDβ = drag coefficient of boat-tail

cf = skin-friction coefficient
CL = lift coefficient
CLo = lift coefficient corresponding to α = 0
CLg = lift coefficient in ground effect
CLhull = lift produced by immersed hull
CLmax = maximum lift coefficient
CLα = lift-curve slope
CLflap = change in CL due to flap deployment in ground effect
CLs f = increment in CL due to the deployment of split flaps
CLr = change in CL due to flap and spoiler deflection
CN = normal force coefficient
cm = parameter in Equation 4.64, cm = cflap /MAC
Cm, Cn , Cl = roll/pitch/yaw moment coefficient
Cp = term defined in Equation 4.98
CY = side-slip force coefficient
C∗ = Chapman-Rubesin factor, Equation 4.22
d = diameter
db = base diameter, Figure 4.23
dw = wheels diameter
D = drag force
e = Oswald factor, Equation 4.16
E = parameter defined in Equation 4.64
f1 = installation effects, Equation 4.62
f2 = installation effects (effects of flap deflection), Equation 4.63
ff = form factor, Equation 4.35
fj = thrust-specific fuel consumption
Fc = parameter defined in Equation 4.30
Fx = parameter defined in Equation 4.33
Fμ = non-dimensional dynamic viscosity, Equation 4.33
√
Fr = Froude number, Fr = V/ gl
G = empirical function for fuselage upsweep drag, Equation 4.40
G = function of Strouhal number in buffet, Equation 4.88
J po = efficiency factor of plain flap, Equation 4.10
k = induced-drag factor
kf = flat-type factor in Equation 4.10
kflap = empirical parameter in Equation 4.56
kt = free stream turbulence level, Equation 4.34
kto = correlation factor in Equation 4.10
ku = factor in the flap deflection, Equation 4.67
kw = lift-induced-drag factor in hull-drag equation
K = part-span factor (Equation 4.53); wheels coefficient
(Equation 4.68)
lu = length of retractable multi-wheel landing gear on ground
lo = virtual forebody length, Figure 4.23

m = ground clearance of vertical strut, Figure 4.13
mf = fuel mass
M = Mach number
MB = buffet design Mach number
Mc = critical Mach number, Equation 4.44
Mdd = divergence Mach number
n = distance between wheels (Figure 4.13); normal load factor in
Equation 4.89
nw = number of wheels
p = pressure
p, q, r = pitch/roll/yaw rates, respectively
Pr = Prandtl number, Pr = c p μ/k
q = dynamic pressure, q = ρU 2 /2
r = recovery factor, Equation 4.24; radius of curvature in § 4.4.1
ro = radius of secant ogive, Figure 4.23
req = equivalent radius
Re = Reynolds number, Re = ρUl/μ
Rec = critical Reynolds number
Rel = Reynolds number based on length l
Retrans = Reynolds number for turbulent transition, Equation 4.34
Rex = local Reynolds number
s = ratio between wheel’s diameter and landing-gear width,
Equation 4.58
Styre = tyre’s frontal surface
St = Strouhal number
t/o = relative thickness
tm = geometric parameter in Equation 4.62
T = net thrust
T = temperature
Taw = adiabatic wall temperature
Te = external temperature
Tw = wall temperature
T∗ = reference temperature
U = air speed
V = volume; critical speed for aquaplaning, Equation 4.71
VB = buffet design speed
Vf loat = float volume
Vf∗loat = float volume, immersed
VS = stall speed
w = downwash velocity
W = weight
xo = factor defined in Equation 4.98
x,y,z = Cartesian coordinates
xb, yb, zb = body-conformal reference system
X = distance
Greek Symbols
α = angle of attack; parameter defined in Equation 4.31
αe = effective angle of attack
αg = wing angle of attack in ground effect
αi = induced angle of attack
αo = zero-lift angle of attack
β = side-slip angle; also β = (M2 − 1)1/2
β = parameter defined in Equation 4.31
β = boat-tail angle, Figure 4.23
β1 = Prandtl-Glauert factor, Equation 4.77
β2 = factor in Kármán-Tzien equation, Equation 4.78
βu = average upsweep angle of rear fuselage
γ = flight-path angle, § 4.1
γ = ratio between specific heats, γ = 1.4
δ = depth of the centroid of the maximum float cross-section,
Equation 4.98
δf = flap angle
η = normalised spanwise coordinate; induced-drag coefficient,
Equation 4.85
θ = aircraft attitude
ϑ = relative temperature in Equation 4.30, ϑ = Tw /Te
κ = lift factor in Equation 4.41
κa = airfoil technology factor, Equation 4.41
LE = leading-edge sweep angle
μ = dynamic viscosity
ν = kinematic viscosity
ρ = air density
ρ∗ = density of contaminant
ρw = water density
τ = wing’s volume parameter
ϕ = wing dihedral angle
i ,o = part-span wing factors, Equation 4.10
ζ = damping ratio in buffet, Equation 4.88
∞ = free-stream conditions
[.] = average value
ˆ
[.] = incompressible flow conditions
[.]e = external conditions
[.]IGE = in ground effect
[.]OGE = out of ground effect
[.]flap = relative to flap
[.]max = maximum value
[.]s f = split-flap
[.]sep = separation
[.]trans = turbulent transition
[.]w = wing
[.]α = derivative with respect to angle of attack
[.]β = derivative with respect to yaw angle
[.]δ = derivative with respect to flap deflection
(.)∗ = reference conditions
5 Engine Performance
Overview
In this chapter we present a basic analysis for three key gas-turbine engine architec-
tures: high by-pass turbofan, turboshaft and low by-pass turbojet with thrust aug-
mentation. We consider general gas turbine engine architecture (§ 5.1), the thrust
and power ratings (§ 5.2), the turbofan model (§ 5.3), the turboprop engine (§ 5.4) and
the low by-pass engine model (§ 5.5). We briefly mention the methods of generalised
engine performance (§ 5.6). We finally discuss the role of the auxiliary power unit
(§ 5.7). An important aspect of the presentation is the strategy for the determination
of the design point of the engine in absence of reliable data. In all cases we discuss
the calculation of the engine state as a function of the main operational parameters.
KEY CONCEPTS: Gas Turbine Engines, Thrust/Power Ratings, Engine Derating, Tur-
bofan Engines, Engine Design Point, Engine Simulation, Rubber Engines, Effects
of Contamination, Turboprop Engines, Turbojet Engines, Auxiliary Power Units.
5.1 Gas Turbine Engines

The term gas turbine is associated with a jet engine consisting of a compressor, a
combustion chamber, a turbine and an exhaust nozzle, although the name refers
to both jet-thrust engines and shaft-power engines. The main types of gas turbine
engines are the turbojet, the turbofan and the turboprop. The gas turbine is the core
of the engine. However, there are other parts whose function is essential (inlet, fuel
lines, fuel nozzles, sensors, collectors, thrust reverser).
The turbojet belongs to the first generation of gas turbine engines. It con-
sists of a single gas flow. The operation of the engine requires a number of aero-
thermodynamic stages: 1) compression of the inlet flow via a number of axial com-
pressor stages; 2) the transfer of the compressed air into the combustion chamber,
where it is mixed with fuel; 3) combustion in radially spaced combustion chambers;
4) discharge into a multi-stage turbine, rotating on the same shaft with the com-
pressor; and 5) ejection of all of the exhaust gases as a high-speed hot jet through
the nozzle. The air is captured by an inlet, whose other function is to provide
126
5.1 Gas Turbine Engines 127
pre-compression by an aero-thermodynamic mechanism called ram compression.

This is an adiabatic compression in the engine inlet due to flow deceleration.
The main function of the turbine downstream of the combustion chamber is to
operate the compressor. A considerable amount of power is generally required by
the compressor. The rest of the thermal and kinetic energy associated to the mass
flow is transformed into a high-speed, high-noise jet released from the nozzle.
The very first turbojets had centrifugal compressors, but as the engines became
better understood, they were replaced by more efficient axial compressors. As the
thrust requirements increased, the compressor’s architecture became more complic-
ated, with low- and high-pressure units, each with several rotor stages. The exhaust
gas leaving the combustion chamber has a high temperature (about 1,000◦ C). When
the compressor and the turbine are connected to the same shaft, their rotational
speed is the same. This coupling is referred to as spool.
The gas turbine can have an additional combustion stage (re-heat, or after-
burning). Fuel is injected after the primary combustion for the purpose of increasing
the engine thrust. After-burning uses the excess air that does not support the primary
combustion.
Gas turbines with this capability operate without re-heat most of the time
because the increase in thrust is derived at the expense of a considerable fuel con-
sumption. Their application is limited to some military jet aircraft.
A turbofan is a derivative of the turbojet. In this engine the excess air that does
not support the combustion is channelled through an external annulus and by-passes
the combustor. The by-pass ratio (BPR) is the ratio between the by-pass flow rate
and the core flow. This ratio has been increasing over the years, from about 1.1 to
values above 5 in modern engines. The General Electric GE-90 has a BPR = 8.4,
and the P&W GP-7000 series has a BPR = 8.
The other difference, compared to the basic engine, is a large-diameter fan placed
in front of a multi-stage axial compressor. The function of the fan is to increase the
capture area of the inlet and to channel the by-pass flow through the annulus of
the engine. The fan is powered by the engine itself, either on the same shaft as the
compressor or on a separate shaft (dual-compressor engine). The advantages of this
engine are that the exit flow has lower speed and lower average temperature and
produces far less noise.
A turboprop is an aero engine consisting of a gas turbine unit coupled with a
propeller. The thrust can be derived both from the jet engine and the propeller,
although in practice, most of the useful thrust is imparted by the propeller. Due
to the different speeds between the gas turbine and the propeller, these engines
have a reduction gear. A gas turbine rotates at speeds of the order of 10,000 rpm,
whilst the propeller’s speed is less than one third of this, as limited by the tip Mach
number.
Jet engines for helicopter applications are a variant of the turboprop and consist
of one or two gas turbines, a reduction gear and a rotor shaft. The reduction in
rotational speed is higher than the turboprop because helicopter rotor speeds do not
exceed 300 rpm.
Modern engines come with an FADEC (Full Authority Digital Engine Control),
which is the brain of the engine. A typical FADEC consists of a hardware unit
and a software unit. The hardware unit includes sensors to determine in real time
essential engine parameters and a digital computer (electronic engine controller) to

elaborate on the input data and generate a control response. One of the operations
that are possible with the electronic control is to balance the power from different
engines. For example, if there is a loss of power from one engine, the FADEC reacts
quickly to augment the power of the other engine. Other monitoring operations done
by the FADEC include cycle counting, cold start, engine stop and general engine
health.
The subject of aircraft engines is a specialised one and goes under the field of
aerospace propulsion or gas turbines. For a more in-depth and specialised presenta-
tion, the reader is invited to consult the specialised literature, for example, Mat-
tingly 1 , Oates 2 , Archer and Saarlas 3 , and the citations thereof.
Modern gas turbine engines must be defined by dozens of geometrical and
functional parameters. However, most critical data are hard to obtain because they
represent proprietary information by the engine manufacturers. Some of the aero-
engine programs are still considered a matter of national security. This aspect of
flight performance is a bottleneck in most simulation procedures.
A number of programs of industry standard are available to evaluate the essen-
tial static and transient performance of typical engines. However, both simulation
programs and engine manufacturers’ data cannot predict exactly the flight perform-
ance at altitude and the installation losses. These losses depend on the integration
between the engine and the airframe. The same airplane version can have different
engines, with performance slightly different from each other. This is a choice of the
airplane manufacturer, to reduce the dependency on any single engine manufacturer
and to drive down prices.
Initial testing of the engine is done on a test bed in experimental facilities that
allow the simulation of pressure altitude. Engine performance at different atmo-
spheric conditions is done by specific correlation rules. Therefore, there is a wide-
spread practice to work with normalised data from a mathematical model of the
engine.
5.2 Thrust and Power Ratings

Aircraft engines have various thrust and power ratings, based on operation time
and maximum allowable temperatures at selected sections of the engine. For a tur-
bofan engine, typical thrust ratings include maximum take-off, maximum climb and
maximum continuous thrust. For most ratings a time limit is associated to the thrust
output. The atmospheric conditions, the altitude and the speed must be specified.
The practice is to provide data for a standard day, although this is not always the
case. One example is shown in Table 5.1. The engine ratings* are for temperatures
below the values indicated in the table. Therefore, our interpretation is that these
engines should provide the quoted power rating at any “reasonable” atmospheric
temperature below the maximum temperature. Additional power ratings for this
type of engine include a maximum contingency power (2.5 min) and an intermediate
contingency power.
* Adapted from EASA: Type Certificate Data Sheet IM E041, PW100 Series, June 2008.
5.2 Thrust and Power Ratings 129
Table 5.1. Power ratings for PW127 turboprop engine variants, sea level;
maximum temperatures as indicated
MTOP[kW] MTOP[kW] T [◦ C] MCP T [◦ C]

Engine (5 min) (normal) (max) [kW] (max)
Basic 2,051 1,846 32 1,864 41

B 2,051 1,846 30 1,864 41
D 2,051 1,846 33 2,051 33
E 1,790 1,611 45 1,790 45
F 2,051 1,846 35 1,864 44
G 2,178 1,973 35 2,178 35
M 2,051 1,846 39 1,864 48
In ground operations, the manufacturers often refer to a breakaway thrust (or

power). This is the minimum thrust required to move the airplane. This thrust
depends on the brake-release gross weight.
5.2.1 Engine Derating

Often the thrust required is lower than the thrust available. This event occurs at take-
off weights lower than MTOW. Thus, to improve engine life and reduce maintenance
and fuel costs, there is the possibility of “derating” an engine. Alternatively, excessive
thrust may cause excessive structural loads, and derating is one way to prevent this
event from occurring. Finally, a derated engine finds applications on different aircraft
versions.
The term derating or flexible refers to the possibility of using the thrust appro-
priate to the take-off weight and atmospheric temperature. There are two important
differences between derated and flexible thrust:
r An aircraft operating with a derated thrust cannot revert to full thrust if the speed
is below the flap-retraction speed; an aircraft operating with flexible thrust can
revert to full thrust if conditions require it.
r The aviation regulations allow the use of derated thrust but not the use of flexible
thrust with contaminated runways.
An example of derated take-off thrust is shown in Figure 5.1. First, the engine is
flat-rated and provides a net thrust that is not dependent on the outside air temper-
ature (OAT) at temperatures below the flat rating. Above the derated temperature,
the net thrust is limited by the exhaust gas temperature (EGT) and decreases as
the OAT increases. If the actual TOW < MTOW, the thrust required is less than
the available thrust: Treq < Tav . Flexible thrust can be used only if the actual tem-
perature Tref is lower than the flexible temperature at the specified flexible thrust.
In practice, there are a number of discrete derated levels, given in percent of the
maximum thrust; each derated level must be certified. For example, D20 indicates a
20% derating, or 80% of the maximum thrust.
Opposite to derating is the engine “bump”. This jargon refers to a temporary
increase (5 to 10%) in the available thrust, to a value above the maximum take-off
thrust.
Thrust
Weight
Thrust
Tav Flat Rating Reduction
MTOW
EGT Limit
Treq TOW
OAT
Figure 5.1. Engine thrust derate at take-off.
There is also a derated climb. As in the case of derated take-off, a lower climb
thrust improves the engine life but, in general, causes an increase in fuel consumption
and an increase in the time to climb. However, from the point of view of the operators,
what counts most is the overall cost of operation, including items other than fuel. A
derated climb is considered independent from a derated thrust.
Finally, an important advantage of the derated thrust is represented by the
extended life of the engines. However, this gain must be evaluated against the
increase in take-off fuel, the longer take-off distance and the lower initial climb rate.
5.2.2 Transient Response

The response of the engine to an increase in fuel flow (throttle) is not instantaneous.
Times when sharp increases in thrust are required (take-off and go-around) are
critical. Figure 5.2 shows a typical turbofan-engine response curve. The transient
response depends on the specific engine. The engine must be protected against
compressor and turbine stall as well as flame-out. Specific regulations* require a
maximum of 5 seconds to accelerate from 15 to 95% of the go-around thrust. During
this time, it is likely that the airplane loses altitude (see § 11.8); on recovering the
go-around thrust, the engine must be able to ensure a minimum climb gradient
(again, as specified by the relevant regulations). Turbofan and turboprop engines
respond differently. The turboprop operates at nearly constant rpm and responds
to an increase in shaft power; the turbofan must respond with an increase in rpm in
order to increase its net thrust.
5.3 Turbofan Engine Model

The starting point of an engine model is the collection of essential data from the type
certificate and other documents. Photographs of the engine will help as well. These
data are indicated in the flowchart in Figure 5.3; they are split into two categories:
configuration and design limitations. The grey-shaded boxes indicate data that are
* Federal Aviation Regulations, FAR 33.73: Airworthiness Standards, Aircraft Engines.
100
80
Thrust recovery, %
60
40
Approach
Thrust
20
Time lag
0
0 2 4 6 8 10
Time, s
Figure 5.2. Typical turbofan thrust recovery in go-around manoeuvre.
generally unavailable; these include the number of blades and the diameters of all
compressor and turbine stages (LPC, HPC, LPT, HPC), as well as the rotor-stator
separation (RSS), the guide vanes and other smaller details, which may be needed
for either the aero-thermodynamics or the determination of the acoustic sources
(Chapter 16).
To study the transient performance, the inertias of the rotating parts are required
to define the rotational acceleration. The unavailable data are estimated from a
Dimensions
Fan
Geometry Design Lim.
# Blades
Diameter N1, N2
#LPC-HPC Engine
Max N1%, N2%
# Blades
TT4.5 or TT5
Diameter
Max EGT
Vanes & RSS Model
T/P Ratings
# LPT-HPT
Flat Rating
Nozzle
OPR, BPR
Diameter
Plug
Figure 5.3. Flowchart of an aero-engine model.

5
Fuel
Control
By−pass
11
2
1
Fan 8
7 10
3 4 9
Inlet 6
HP Compress Duct Exhaust

HP Turbine LP Turbine
LP Compress Combustion
Figure 5.4. Turbofan engine. Numbers indicate standard engine components.
limited number of engines for which a reliable cut-out drawing is available; the
actual number of compressor and turbine blades is one of the great mysteries of
aero-engine technology.
A thermodynamic diagram of the engine is shown in Figure 5.4. The graph shows
the logical connection between systems. Note that the inlet flow is split downstream
of the fan. The core flow goes through the axial compressor. The by-pass annular
flow is channelled to the duct and then mixed with the combustion gas. The numbers
beside each component represent standard nomenclature. The flow through the
engine is assumed as one-dimensional. This type of engine is typical of modern
commercial aircraft, characterised by very large diameters – in some cases, limited
only by the ground clearance. The BPR of these engines varies from about 4 to 6,
sometimes higher.
There is a standard nomenclature to define the aero-thermodynamic parameters
at all sections of the engine. For example, static temperatures are called TS; total
temperatures are TT; static pressures are PS; total pressures are PT; and the mass flow
rates are called W. Each quantity is further specified by a subscript, which denotes the
engine section. For example, the total temperature at the exit of the combustor is
TT4; the total temperature at the exit of the high- and low-pressure turbine is TT4.5
and TT5, respectively. The low- and high-pressure rotor rpm are called N1 and N2
(or N%1, N%2 if given in percent), respectively. A list of parameters used in our flight
mechanics and noise models is given in Table 5.2 for a jet-powered aircraft.
5.3.1 Aero-Thermodynamic Model

If we consider a vector of aero-thermodynamic parameters at the outlet of each
component (at least ρ, ṁ, T , p), then the operation of the engine is characterised
by about 40 parameters. However, we also need to consider the inlet parameters,
the rotational speeds and the overall performance (net thrust, SFC, fuel flow and
other parameters). The solution of the problem is simple in principle and quite
complicated in practice. In principle, we need to apply the fundamental laws of
conservation (momentum, energy, mass), as well as the equation of state of the gas.
In practice, these equations have to be applied to selected control volumes around
each of the engine sub-components (as shown in Figure 5.4). The model can be
Table 5.2. Turbofan-engine parameters used for flight and aircraft-

noise calculations
Item Symbol Description Text Symbol
1 W1 Mass flow rate ṁ

2 WC2.5 Core mass flow rate
3 Wf6 Fuel flow rate ṁ f
4 N%1 Low-pressure rotor rpm (%) N1
5 N%2 High-pressure rotor rpm (%) N2
6 TT2.1 Exit-fan temperature (core side)
7 TT2.2 Exit-fan temperature (by-pass side)
8 TT3 Combustor-inlet temperature
9 TT4 Exit-combustor temperature
10 TT5 Power-turbine temperature
11 TS9 Total static nozzle temperature
12 TT14 By-pass flow temperature, exit
13 FN Net thrust T
14 TSFC Specific fuel consumption fj
15 M9 Nozzle Mach number (core side)
16 TT2.5 LP compressor-exit temperature
17 PT2.5 LP compressor-exit pressure
18 PT3 HP compressor pressure, exit
19 PS9 Total static nozzle pressure
20 PT14 By-pass flow total pressure, exit
improved by further considering secondary effects, such as heat flux between the gas
and the external environment, bleed air at specified compressor sections, and the
use of secondary power for oil and fuel pressure.
The equations are then assembled together with matching conditions between
the outlet of one component and the inlet of the downstream component. The science
behind this thermodynamics is better described in books on gas turbine engines (as
cited). A number of computer programs of industry standard exist to help with this
demanding task. One of those is the GSP program, developed at NLR 4 , that we have
used for our engineering analyses.
5.3.2 Determination of Design Point

The flowchart in Figure 5.3 does not show one of the most important parameters: the
fuel flow (or the TSFC) at the design point. This datum is not available. Occasionally,
there are data on TSFC published in the literature; these data are not reliable because
the operation point is not fully specified. Notably, the ICAO databank does contain
this datum* . However, it can be used as a guideline. A simulation method by itself is
not capable of defining the design point. On the basis of the thermodynamic model,
a number of parametric runs can be carried out so as to minimise the fuel flow at the
nominal engine speed (100% rpm).
* ICAO Engine Data Bank. Updated regularly. Available in electronic form from the ICAO and
other aviation authorities.
1.4
1.3 Estimated
Design Pt
Wf6, kg/s
Iso-TT4.5
1.2
Iso-FN
1.1
380 400 420 440 460

W1, kg/s
Figure 5.5. Example of determination of design point at N%1 = 100, static conditions, sea
level, standard day.
The design point is defined as the engine state corresponding to a design net
thrust at sea level, static conditions, standard day, with the turbine temperature TT5
(or TT4.5) equal to (or slightly below) the value of the limit temperature in the type
certificate, with the engine running at 100% rpm. Now the procedure is as follows:
r Generate a table of performance data for variable mass flow W1 (at fixed fuel
flow Wf6) and variable Wf6 (at fixed W1), around the estimated design point.
r The design point is determined as the intersection between the limit TT4.5-curve
and the limit FN-curve. This point gives a unique value for W1 and Wf6.
The determination of the design point is shown in graphic form in Figure 5.5.
The iso-FN and the iso-TT4.5 curve refer to the certified net thrust at the specified
condition, that is, at sea level, static conditions, on a standard day.
There are some drawbacks in this procedure. In fact, on a hot day an engine
operating at N%1 = 100 (normal speed) would overheat; thus, the reference tur-
bine temperature could exceed the certified limit temperature, although it should
be able to operate normally. Alternatively, we can investigate the virtual design
point corresponding to a limit engine speed (for example, N%1 = 104) in standard
atmosphere.
5.3.3 Case Study: General Electric CF6-80C2

We consider the General Electric CF6-80C2, an engine that powers aircraft such
as the Airbus A300. Over the years, General Electric has developed at least 23
versions of this engine, each with a different pressure ratio and thrust rating. The fan
has 38 blades, a 4-stage booster and 80 composite exit guide vanes. The low-pressure
(LP) compressor consists of four stages and vanes mounted orthogonally. The high-
pressure (HP) compressor consists of 14 stages, with inlet guide vanes; the first 5
stator rows have variable incidence. The combustion chamber is annular, with rolled
Table 5.3. Selected engine data for the CF6-80C2A3;

data with an asterisk ∗ are estimated
Thrust Ratings
Maximum continuous (S/L, static) 24.853 kN
Take-off (5 min, S/L, static) 26.739 kN
Flat-rating temperature (continuous) 25◦ C
Flat-rating temperature (take-off) 30◦ C
Speeds
Design low-pressure rotor (N1) 3,280 rpm
Design high-pressure rotor (N2) 9,827 rpm
Low-pressure rotor (N%1), maximum 117.0%
High-pressure rotor (N%2), maximum 112.5%
Maximum Permissible Temperatures
Turbine exhaust gas (TS9)
Take-off (5 min) 1,233 K
Maximum continuous 1,198 K
Starting (40 s) 1,143 K
Bleeds
Compressor, stage 8, airflow (max) 8.8%
Compressor, stage 11, airflow (max) 1.5%
Gas Flows
Design fuel flow, take-off∗ 2.457 ks/s
Design mass flow not available
ring construction, aft-mounted, with film cooling. The high-pressure turbine has two
stages and the low-pressure turbine has five stages. The control system is FADEC.
The certification of the engine consists of a document that contains information
regarding 1) limitations, and 2) permitted fuels, oils and spare parts. The limitations
part is important in performance analysis because it contains data on rotor-speed
limits, temperature limits of some components and rated thrust (Table 5.3).
The solution of the engine problem across the full range of fuel flows, Mach
numbers and flight altitudes is shown in Figure 5.6. The graphs show the trends
of selected parameters: net thrust, mass flow rate, TSFC, LPT total temperature
on exit, nozzle Mach number and engine speed in percent. The graphs show the
engine limits, which are set as constraints from the type certificate of this engine.
The analysis is carried out at standard conditions, but it can be extended to include
other atmospheric models, including cold and hot temperatures.
At moderate to high fuel flow, the relationship between the fuel flow and net
thrust is roughly linear, which indicates that the TSFC is nearly constant. How-
ever, at low fuel-flow rates and idle conditions, the TSFC increases rapidly. An
engine running idle on the ground would have an infinite TSFC because the engine
generates no useful thrust. The use of a constant TSFC in aircraft performance ana-
lysis is incorrect. For an engine in idle mode, the fuel flow is a better indicator of
performance.
The effect of fuel temperature is generally very small and certainly negligible
when compared with other operational aspects. In practice, a colder fuel causes a
slight increase in TSFC and a negligible decrease in net thrust.
250
800
l
eve
aL
200 Se
2
Mass flow, kg/s

Net thrust, kN
600
4
150
6
14
100 400
8
IDLE IDLE
10
Temperature / RPM
50 12 Limit
200
Sea Level
14
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5

Fuel flow, kg/s Fuel flow, kg/s
(a) (b)
0.05
2 RPM Limit
Sea Level 800 2
Power turbine temperature, K
4
0.045 6
TSFC, kg / N h
8
10
14 12 Sea Level
4
0.04 700
6
0.035
600
8
0.03 10
500
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
(c) (d)
120
RPM Limit RPM Limit
14 12 10 8 6
1 4
14 12 10 8 6 4 2
2
Exhaust gas Mach number
100
Gas turbine rpm, %
0.8 el
Lev
el S ea
ev
aL 80
Se
0.6
60
0.4
40
0.2
20
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
(e) (f)
Figure 5.6. Simulation of General Electric CF6-80C2A3 turbofan engine; altitudes at 2,000 m
intervals (indicated as 2, 4, . . .); limit engine speed reduced to 105%; standard day.
ENGINE INSTALLATION EFFECTS. The thrust required for a flight condition (cruise or
manoeuvre) is called Treq . The thrust delivered by the power plant depends on a
number of factors. If T is the uninstalled thrust, then
Treq = (1 + κ)T, (5.1)
where the factor κ takes into account the installation losses. The term installation
loss refers broadly to all of the aero-thermodynamic effects that cause the engine to
function differently when it is installed. Tests are carried out by the engine manu-
facturer on the bench or uninstalled conditions. Installation losses include reduction
in mass flow rate (spillage drag), flow distortions, tail-pipe pressure drop, nozzle
drag, incomplete combustion, loss due to associated systems, and so on. The precise
estimation of each of these contributions is difficult because it depends on the flight
conditions (Mach, altitude, thrust required) and must be done with very detailed
engine models. An example of analysis is shown by Roth 5 on a turbojet engine.
5.3.4 Rubber Engines

A rubber engine is an engine that can be scaled to a lower or higher thrust. The scaling
applies to engines in the same family, which differ for relatively small details from
the point of view of comprehensive aircraft performance. For example, the engine
architecture of the CFM56-5C2 (basic engine) is similar to at least 30 engines in the
same family† . Differences may include redesigned fan blades, redesigned compressor
stages, reduced thrust, different combustors, compression rates, by-pass ratios, and
so on.
It is assumed that rubber engines perform identically, except for the thrust or
power rating. Therefore, they should be able to provide increased or decreased levels
of thrust and power with the same engine temperatures and specific fuel consump-
tion, although the overall compression rates may be different. With reference to the
SFC, at a given operation point we assume that a scaled turbofan engine provides
a scaled thrust T with a scaled fuel flow ṁ f = f j T, with TSFC equal to the “basic”
engine. The corresponding mass flow rate is scaled by assuming that the fuel–air
ratio is the same as the basic engine.
Because the compression rates of the rubber engine may be different, the pres-
sures in the relevant sections are updated. These data are to be used in some noise
sub-models, as described in Chapter 16.
There are some justifications for this type of operation: this level of detail of
one engine and its similar counterpart is either unknown, or difficult to model, or
unverifiable, or inconsequential from the point of view of flight simulation. The pre-
vious examples of redesigned fan blades and redesigned compressor stages are of
little interest because they are not associated to a known fan and compressor effi-
ciency that can be used in a one-dimensional aero-thermodynamic model. Likewise,
a redesigned combustor that has an unknown limit temperature cannot be used to
update the thermodynamic model.
† Federal Aviation Administration, Type Certificate Data Sheet E38NE: CFM International (CFM56
engines), Sept. 1996.
Figure 5.7. Volcanic ash particle from the Iceland eruption, March 2010, taken from a car in
South Yorkshire, England [photo courtesy of John Chandler].
5.3.5 Effects of Contamination

There are various environmental effects that can cause degradation of engine per-
formance and, in some serious cases, also engine shut-down within a few minutes.
One of the most serious problems centres around ash clouds. Although damage can
occur on all parts of the aircraft (airframe, windshields, and so on), this section deals
exclusively with engine-related problems. About 80 aircraft have been reported to
have been flown through volcanic ash between 1980 and 2000, with various degrees
of severity. Critical regions are around the Pacific Ocean, although events in North-
ern and Southern Europe routinely affect flight operations. These clouds can move
quickly around FL-300 (30,000 feet). They are made of very fine particulate (∼ 1µm)
and may carry electrostatic charges.
Ingestion of volcanic ashes causes, at various degrees, erosion of the fan blades
and compressor blades. If they manage to travel as far as the combustor, the ash
particles are melted; then, they either stick to the combustor walls and glaze or
travel further to the turbine stages (where they cause further structural damage), or
remain within the vanes, thus disturbing the high-pressure out-flow. Side effects on
the aircraft systems include damage to the Pitot tubes (essential for monitoring the
air speed) and the air filters (essential for maintaining cabin pressure). Melting of
volcanic ash is quite variable, from about ∼1,250 K to ∼1,600 K 6 , whilst combustor
temperatures can, in some cases, exceed this value.
Figure 5.7 shows an example of volcanic ash particle from the Eyjafjallajökull
eruption in the spring of 2010. The exagonal particle is most likely quartz.
For reference, consider the combustor temperatures of the engine CF6-80C2A3,
whose simulation is shown in Figure 5.6. The corresponding combustor temperat-
ure TT4 is shown in Figure 5.8. The shaded area indicates the temperature TT4 at
flight altitudes between 8,000 and 12,000 m (26,250 and 39,270 feet, respectively).
Therefore, it is very possible that melting occurs on the combustor, unless the engine
1,500
Temperature / RPM
Limit
2
1,400 4
6 v el
Le
8 a
Se
1,300 12
14
TT4, K
1,200
1,100
1,000
0 0.5 1 1.5 2 2.5
Fuel flow, kg/s
Figure 5.8. Combustor temperature of simulated engine CF6-80C2A3 (see also Figure 5.6).
operates at low thrust (low temperature) and the melting temperature of the ash is
high.
Monitoring of volcanic ashes is a complex matter beyond the realm of a single
aircraft performance‡ . Actions required in the presence of such an event include a
minimum clearance of 20 n-miles, possibly upwind of the affected area. If the cloud
cannot be avoided, the aircraft is to be commanded to make a U-turn or decrease
the thrust setting. This option limits the combustor temperatures and thus the risk
of particles melting in the combustor.
Another form of engine contamination is due to dust and ice intake on the
ground, either with the engines switched off or running. If the engines are to be
switched off for any length of time, the intakes must be closed-off to contamination
agents that would accumulate over time; the engine integrity could be compromised.
5.3.6 Performance Deterioration

Aircraft engines are subject to performance penalties that depend on the number
of work hours or cycles. The loss in performance is quantified by the increase in
TSFC. Reasons for such a deterioration include dirt accumulation in the first stages
of the compressor, increased tip clearances due to differential heating, blade erosion
and seal leakages. Most of these losses can be reduced by servicing the engine at
specified intervals; the service guarantees the best performance over a large number
of cycles. Typically, an improvement in excess of 1% in TSFC can be achieved. An
example is shown in Figure 5.9. This graph was drawn using data extracted from
Pratt & Whitney and refers to a typical high by-pass turbofan engine. The servicing
is done at 1,000-cycle intervals. The data clearly point toward a loss of more than
‡ There are nine Volcanic Ash Advisory Centres around the world that deal with information to the
civil aviation.
Δ TSFC, % Unwashed
1,000 cycle wash
0
0 1,000 2,000 3,000 4,000 5,000
Cycles
Figure 5.9. Loss of efficiency of turbofan engine with and without cycle wash (data adapted
from Pratt & Whitney).
1% in TSFC within the service schedule, which for an engine in this class means a
considerable amount of fuel burned to no benefit. Even with the service, after 4,000
cycles, there appears to be a permanent loss of TSFC of the order of 2%.
In the absence of detailed engine data, it will suffice to assume that the loss in
TSFC can be expressed as
f j (%) = 2.55 · 10−2 + 1.28 · 10−3 n + p, (5.2)
where n is the number of cycles and p is a permanent loss, which may depend on the
number of services already performed.
5.3.7 Data Handling

Because our engines have to be integrated into an airframe and thence with a
flight mechanics model, we need to take some further steps. Instead of solving the
engine problem at every step, we generate a complete flight envelope at selected
atmospheric temperatures. There are two reasons for this choice: 1) the solution of
the engine problem at each time step is a slow process; and 2) the solution of the
engine must be carried out in both direct and inverse mode. If the net thrust is a
constraint, then we need to calculate the fuel flow and the engine state corresponding
to the specified thrust (inverse mode). The functional relationship is
EngineState(T) = f (T, M, h, T ). (5.3)
If the fuel flow is specified, the engine is solved in direct mode to produce a thrust T.
The functional relationship is
EngineState(ṁ f ) = f (ṁ f , M, h, T ). (5.4)

5.4 Turboprop Engines 141
Over-speeding
Over-speeding
&
Over-heating
Temporary
Operation
N%1
Engine Envelope Over-heating

[Max Continuous Power/Thrust]
Temporary
Operation
TT5
Figure 5.10. Engine limits in the space TT5, N%1 (gas turbine inlet temperature and rpm).
The simulation of the engine’s envelope leads to a large matrix of data. Each envelope
is a multi-dimensional matrix in the space defined by the vector {M, h, T , ṁ f } for
each engine state parameter. There are several dozen state parameters, which arise
from the number of independent aero-thermodynamic parameters at all relevant
sections of the engine. For the purpose of the analysis carried out in this and following
chapters, we consider only a sub-set of these parameters, as listed in Table 5.2.
The envelope must be interpolated in either direct or inverse mode. For example,
in direct mode, if E is a generic engine parameter, then its functional dependence on
the operational conditions can be written as
E = f (dT , U, h, ṁ f ). (5.5)
We use a higher-order multi-dimensional interpolation of the matrix at the opera-

tional point defined by the vector {dT , U, h}. Thus, we construct a functional
E1 = f1 (). (5.6)
In the engine analysis, the state parameter can be greater than one, which corres-
ponds to an operational point that exceeds the design limits, either due to a too
high gas generator temperature TT5 or a too high turbine rpm N1%, or both. To
avoid engine operation beyond the design point, as shown in Figure 5.10, the state
parameter is calculated to match the worst possible condition. If the rpm is within
the design limits, Equation 5.6 is limited by the power turbine temperature to avoid
over-heating. If the engine is over-speeding (in general, about 104% the design rpm),
the power turbine temperature is the one corresponding to the limit rpm.
5.4 Turboprop Engines

A turboprop is a gas turbine engine coupled to a propeller. It is designed to deliver
shaft power instead of thrust force. For this reason, there are substantial differences
in the engine architecture, although there is a degree of commonality between the
two engines; the turboprop and the turbofan are both continuous mass flow engines
with compressor and turbine units.
1 2
Control Control
Bleed Load
Fuel 5
Control
3 7 8
4 6 9
1 2 3 4 4.5 5 9
Combustor
Compress. TRB Exhaust
Inlet Compressor
Power TRB
Figure 5.11. Aero-thermodynamic model of turboprop engine.
The turboprop is made of the engine itself, the cowlings, the engine mounts,
the firewalls and the drain lines. In its basic form, the engine is composed of an
axial compressor, a centrifugal compressor, an annular combustor, a centrifugal fuel
injection system, a multi-stage gas turbine and a single-stage free power turbine. The
gear shaft is considered part of the transmission system. The gas-generator turbine
and the power turbine operate at different speeds and are mounted on two co-axial
shafts (dual-spool engine). An aero-thermodynamic diagram of the engine is shown
in Figure 5.11, with the standard nomenclature for the components and the engine
sections. Only one compressor unit is modelled; the pressure ratio is provided by
this component.
Air is drawn in through the inlet; it is passed to a multi-stage compressor and
hence to a combustor. Fuel is added to the combustor through the fuel lines, with
the aid of fuel pumps. The hot gas from the combustor expands into a gas-generator
turbine, mounted on the same shaft as the compressor. The hot gas expands further
in the power turbine, that is coupled to the power shaft, and hence to the rotor shaft,
via a gear box. There are some conventions on the nomenclature that it is important
to clarify. The nominal gas-generator rotational speed is indicated by N1 and the
power turbine speed is called N2 . These parameters are found in the certification
documents of the engine.
A number of controls are indicated in the diagram. One control is the air bleed
from the compressor that can be used to run auxiliary systems. The air bleed
decreases the shaft power. Another control is the fuel flow into the combustor.
Finally, there is a control on the shaft power or torque that is used for transient
analysis of the engine.
The design point is determined in a similar fashion as explained in § 5.3.2.
However, in this case we need to match the design shaft power instead of a thrust;
there are no reference data to compare with because the ICAO databank is limited
to turbofan engines. The data handling follows the same strategy as shown in § 5.3.7.
The set of state parameters used is different. The net thrust is replaced by the shaft
power; the nozzle parameters include the residual thrust Fg ; the by-pass parameters
do not apply to this case.
The engine model must be developed in a way such that both the direct and
inverse solutions are possible, as in the case of the turbofan. In the direct mode the
5.5 Turbojet with After-Burning 143
"heavy" particles
dust
inlet
residual "small" particles
Engine centre line
Figure 5.12. Axi-symmetric, bifurcating intake with inertial particle separator.
fuel flow is the independent parameter; in the inverse mode, the shaft power is the
independent parameter.
5.4.1 Case Study: Turboprop PW127M

This turboprop engine is part of a family (see also Table 5.1). The engine has a two-
spool gas generator driving the six-bladed Hamilton propeller F568-1 (§ 6.2.4). The
main engine components are a centrifugal low-pressure compressor; a centrifugal
high-pressure compressor (HPC), diffuser pipes, an annular perforated combustion
chamber with 14 fuel nozzles; a high-pressure axial turbine (that drives the HPC); a
low-pressure axial turbine (that drives the LPC); and a two-stage free-axial turbine
that drives the reduction gearbox. The latter element is connected via a coupling
drive-shaft and is mounted offset of the main axis of the engine. On the gearbox,
there are a number of sub-systems, including the propeller brake, the over-speed
governor and a number of pumps (oil, propeller feather).
The intake is offset and has an “S” shape to provide a uniform intake into the
radial compressor. This intake also functions as an inertial particle separator that
filters out any large foreign objects, as shown schematically in Figure 5.12. The intake
air is also split between compressor and oil cooling.
A simulation of the Pratt & Whitney PW127M turboprop engine is shown in
Figure 5.13. The graphs show the mass flow rate, the shaft power, the turbine tem-
perature TT5 and the engine speed N%1 as a function of the fuel flow at the full range
of altitudes. The calculations have been carried out with a static engine in stand-
ard atmosphere and can be extended to any flight Mach number and atmospheric
temperature.
5.5 Turbojet with After-Burning

The turbojet is the basic gas turbine engine, Figure 5.14. All (or most) of the gas
flow is sent to the combustor. Because there is excess air that has not been combined
with the fuel, after-burning can be used to further augment the thrust.
18
e l
Lev 2,000
Sea
2
14
Mass flow, kg/s
4 1,500
PShaft, kW
10 6
8 1,000
6 10
500
v el
Le
10
a
Se
2
0
0 0.1 0.2 0.3 0 0.1 0.2 0.3
(a) (b)
900
10
100 l
e
Lev
800 Sea
Turbine Temp TT5, K
v el
10 Le
a
Se
N1%
700 80
600
60
500
0 0.1 0.2 0.3 0 0.1 0.2 0.3
(c) (d)
Figure 5.13. Simulation of PW127M turboprop engine; altitudes at 2,000 m intervals (indicated
as 2, 4, · · · ); static conditions; standard day.
The model for this jet engine is the Pratt & Whitney PW F-100. This is a two-
shaft augmented turbofan with small by-pass ratio, BPR = 0.7. The fan consists of
three stages, with rotor blades in titanium. Its rotational speed is 10,400 rpm. The
compressor consists of 10 stages with variable stators over the first three stages.
Its nominal speed is 13,450 rpm, with OPR = 24.5. The combustion chamber is
annular with film cooling and large-diameter air-blast fuel nozzles. Both the high-
pressure and low-pressure turbine are made of two stages. The maximum inlet
turbine temperature is about 1,340 ℃. The after-burner is made of five concentric
spray rings with downstream flame holders. Thrust vectoring is possible in later
versions of the engine, with up to 20 degrees’ deflection in all directions.
If there is no re-heat, the simulation of the engine is done in the same way as
the turbofan engine (§ 5.3). Figure 5.15 shows selected engine parameters at a flight
Mach number M = 0.8. If after-burning is used, we need to add a virtual combustor
downstream the low-pressure turbine.
5.6 Generalised Engine Performance 145
High-pressure High-pressure
Low-pressure compressor turbine Low-pressure
compressor
turbine
Nozzle
combustor
(a) Turbojet
Nozzle
Fuel spray Flame holder
After-burner duct
(b) Turbojet with after-burner

Figure 5.14. Diagram of a turbojet engine, with and without after-burner.
5.6 Generalised Engine Performance

There is a practical approach to engine performance which is based on generalised
parameters. These parameters are identified through a dimensional analysis, but
their determination requires some form of engine data in order to calculate best-fit
curves. To begin with, the main propulsion parameters are:
r thrust T
r fuel flow ṁ f
r rotational speed rpm
r air speed U (or Mach number M)
r air temperature T
r air pressure p
r engine diameter d
Because this is a relatively large number of parameters, the engine performance

is described by a reduced number of non-dimensional or engineering quantities. The
essential parameters are the air speed, the rpm, the net thrust and the mass flow
rate ṁ f . A normalisation of these parameters is performed by using the remaining
quantities in the previous list.
First, the speed U can be replaced by the Mach number with the definition
M = U/a. The speed of sound is derived from the flight altitude. Second, the rota-
tional speed of the engine is normalised with a/d, to yield the parameter rpm d/a.
Third, the thrust T is replaced by the parameter T/ pd2 (pressure × length2 = force).
Finally, the mass flow rate ṁ f is replaced by the dimensionless parameter
80 200
Δh = 2,000 m / 6,560 ft Δh = 2,000 m / 6,560 ft
60 150
S /L
Mass flow, kg/s

Net Thrust, kN
40 100
L
S/
20 50
0 0
0 0.5 1 1.5 0 0.5 1 1.5
(a) (b)
50 800
Δh = 2,000 m / 6,560 ft Δh = 2,000 m / 6,560 ft
Compressor pressure PT3, K
Static Nozzle temperature, K
40
700
S /L
L
30 S/
600
20
500
10
400
0 0.5 1 1.5 0 0.5 1 1.5
(c) (d)
Figure 5.15. Simulation of Pratt & Whitney turbojet engine F100, standard day, M = 0.8.
ṁ f /( pd2 /a) = ṁ f a/ pd2 . A further step is required for some parameters. By using
the relative pressure δ we have
T T 1
= , (5.7)
pd2 δ pod2
where po is the standard atmospheric pressure at sea level. With the definition of
√
speed of sound, a = γ RT , we have
rpm d rpm d rpm d

= √ √ = √ √ , (5.8)
a T γR θ γ RTo
√ √ √ √ √
ṁ f a ṁ f γ RT ṁ f T γ R ṁ f θ γ RTo
= = = . (5.9)
pd2 pd2 δ pod2 δ pod2
Instead of using dimensionless parameters, it is a practice in engine performance to

use the corrected thrust, rotational speed, and mass flow rate
√
T rpm ṁ f T
, √ , , (5.10)
δ T δ
along with the Mach number M. The remaining factors in Equations 5.7 to 5.9 are
all constant. The parameters in Equation 5.10 have dimensions. Thrust and mass
flow rate are the parameters most widely used in aircraft performance simulation.
The general dependence of the corrected parameters from the Mach number and
the rpm is
√
T rpm ṁ f θ rpm
= f1 M, √ , = f2 M, √ . (5.11)
δ θ δ θ
The technical literature uses various equations to approximate Equation 5.11, includ-
ing polynomials. The coefficients of these equations are extracted from databases.
For example, the first of Equation 5.11 is written as a function of the air speed, the
flight altitude and the rpm:

T N1 N1 2
= c1 + c2 M + c3 h + c4 h2 + c5 T + c5 √ + c6 √ , (5.12)
δ θ θ
where the ci are coefficients determined empirically from a database. Notice that
Equation 5.12 is not unique and that alternative forms can be found.
5.7 Auxiliary Power Unit

The APU is a gas turbine engine used to provide essential services to an aircraft in
flight and on the ground, such as electricity, compressed air, and air conditioning,
and to start the main engines. The first gas turbine APU was installed on the Boeing
727 (1963) and now is the standard technology on passenger-transport airplanes.
Modern airplanes operate at atmospheric conditions in which human bodies
cannot survive. Thus, the airplane must provide suitable cabin pressure, temperature,
humidity and air circulation. This is done through an Environmental Conditioning
System (ECS) powered by an APU.
The APU is generally mounted on the back of the fuselage. The inlet is through
the vertical stabiliser and the nozzle is through the tail-end of the fuselage. Typical
APU system architecture includes the following: a single-stage centrifugal impeller
with air-bleed valves, radial combustor with a variable number of fuel nozzles, a two-
stage axial flow turbine that drives the gearbox, an exhaust and a health monitoring
system. The gearbox is considered part of the system; it is used to reduce the rpm from
the high speed of the shaft power to the rpm required to run an electric generator.
Power must also be extracted for the fuel control units and the cooling system. Some
of these units are required to self-start at all altitudes up to the service ceiling of the
aircraft and provide critical service for ETOPS with one engine out.
An APU is required to be flexible in terms of services to be supplied. The
output power depends on specific requirements. There are several operation modes.
The term ECS refers to air bleed supplied to the air conditioning system, for typical
aircraft gate operation; then there is the maximum ECS, which denotes the maximum
aircraft environmental condition and may include some electric load. Other load
Table 5.4. Typical APU fuel flow [kg/s], depending on load

type and atmospheric conditions; RTL = Ready to Load;
ECS = Environmental Conditioning System; EL = Electrical
Load. All data are at sea level, standard day, given as kg/h
RTL Min ECS Max ECS

APU RTL Max EL Max EL Max EL
36-300 70 85 105 125

131-9A 75 95 115 125
331-350 120 140 175 210
331-600 160 180 225 290
conditions include 1) no load (the same as idle); 2) maximum shaft load (electric
shaft load only; no air bleed); 3) maximum bleed load (air extraction only; no electric
power); 4) maximum combined load (electric and air bleed load); and 5) main engine
start (air bleed supplied to engines turbines; some electric load possible). There are
various other engine ratings but no agreement on APU rating standards. Note that
the APU generate electrical power that is measured in kVA (kilo Volt-Ampere)
and mechanical power measured in kW (kilo Watt). The conversion between the
two units is 1 kW = 1.25 kVA.
Table 5.4 is a summary of basic APU performance and representative fuel
flows. At a reduced load, the APU burns considerably less fuel. For example, the
APU 331-600 uses 160 kg/h in ready-to-load condition and 225 kg/h at a maximum
ECS condition. Other important data include the output power and the electrical
power.
When the fuel consumption is unavailable, a rough estimate is still possible if
the maximum combined power is available. Take the example of the PW980A APU
used on the Airbus A380-800. This APU is rated for a maximum 1,800 hp (1,341
kW) power. Then take the best SFC available in the industry (about 0.35 kg/kW/h)
and multiply the two numbers. The result is a maximum fuel flow ∼470 kg/h.
Table 5.5 is a summary of estimated emissions from APU engines. To limit the
emissions, running costs and noise whilst on the ground, the use of APU is strictly
Table 5.5. Estimated APU power and emission database (adapted from several sources)
Aircraft P HC CO NOx
APU ṁ f [kg/hr] [kW] [g/kg] [g/kg] [g/kg]
36-150 BAE-146 68.0 0.61 6.45 5.10

36-300 A-320 100.0 0.15 2.05 10.10
36-4A Fokker F-28 61.0 0.36 13.47 5.10
85-129 B-737-100/300 106.5 1.03 17.99 4.75
131-9 A-320, MD-90 115.5 447 0.37 4.88 6.64
331-200ER B-757 121.5 0.43 4.13 9.51
331-350 A-330/340 186.0 0.23 1.86 9.90
331-500 B-777 243.0 750 0.20 1.89 11.41
331-600 A-340-600 290.0 894
PW901A B-747-400 391.3 1.50 16.78 3.15
PW980A A-380-861 470.0 1,340
140 700
P /δ = 44.7 kW P / δ = 44.7 kW
P /δ = 0 P/δ=0
600
Fuel flow / δ, kg/h
130
EGT, Celsius
500
120
400
110 300
-60 -40 -20 0 20 40 -60 -40 -20 0 20 40
o o
OAT, C OAT, C
(a) (b)
Figure 5.16. Estimated performance of the G550 APU system.
limited. The required services are provided via Ground Power Units (GPU), which
are run from the electricity network of the airport.
The APU does not require a separate certification; therefore, only few data are
released; compressor ratios are other key parameters that are not readily available.
Therefore, the aero-thermodynamics of these engines cannot be simulated. Addi-
tional considerations are in § 16.5 that deals with APU noise. Landing and take-off
emissions are discussed in § 19.3.
5.7.1 Case Study: Honeywell RE-220 APU

This APU powers the Gulfstream G550; this unit is capable of providing 40 KVA
or 45 kW electrical power, and bleed air at 4.48 bar for starting the main engine
and for environmental control. Its operating envelope extends to 45,000 feet, with
self-start up to 43,000 feet. Basic data are as follows: m = 108 kg; inlet area =
0.677 m2 ; exhaust area = 0.0613 m2 , rated EGT = 732 ℃ (maximum continuous);
rotor speed = 45,585 rpm; maximum rotor speed is 106%. Figure 5.16 shows some
graphs with the estimated performance of the unit. The ratio P/δ is the net shaft
power divided by the relative pressure.
Summary
We have given a brief description of the modern gas turbine engine. The problem
at hand is the determination of engine models suitable for aircraft performance
simulations; this is complicated by the lack of reliable data, especially the design
point. Nevertheless, it is possible to define a fairly accurate one-dimensional thermo-
dynamic model that provides most of the engine parameters required. The design
point is defined as the intersection of constant-thrust and constant turbine tem-
perature lines at the nominal engine rpm. The method is applied to high by-pass
turbofans, turboprops and low by-pass military engines. Even these simplified mod-
els are always insufficient. Therefore, there is widespread practice use of generalised
equations, which need the determination of some parameters, specific to the engine.
We have also considered the effects of engine contamination, such as the effects of
volcanic ashes. We finally discussed the APU and its role. The APU model is similar
to the other gas turbine engines, although in this case there are fewer data available
for a realistic simulation.
Bibliography
[1] Mattingly JD. Elements of Gas Turbine Propulsion. McGraw-Hill, 1996.
[2] Oates GC. Aerothermodynamics of Gas Turbine and Rocket Propulsion. AIAA
Educational Series, 1988.
[3] Archer RD and Saarlas M. An Introduction to Aerospace Propulsion. Prentice
Hall, 1996.
[4] Visser WPJ and Broomhead MJ. GSP: A generic object-oriented gas turbine
simulation environment. In ASME Gas Turbine Conference, number ASME
2000-GT-0002, Munich, Germany, 2000.
[5] Roth B. A method for comprehensive evaluation of propulsion system thermo-
dynamic performance and loss. AIAA Paper 2001-3301, 2001.
[6] Swanson SE and Beget JE. Melting properties of volcanic ash. In First Int. Symp.
on Volcanic Ash and Aviation Safety, Bulletin 2047, Seattle, WA, July 1991. US
Geological Survey.

Additional symbols are given in Table 5.2.
a = speed of sound
ci = constant coefficients
d = diameter
E = generic engine parameter
h = flight altitude
h = altitude
ṁ = mass flow rate (also called W1)
ṁ f = fuel flow rate (also called Wf6)
M = Mach number
n = number of wash cycles
N1 = nominal gas-generator turbine rpm (also called N1)
N2 = nominal power turbine rpm (also called N2)
p = pressure; permanent loss of engine performance (Equation 5.2)
P = engine (shaft) power
R = gas constant (dry air)
rpm = rotational speed
T = thrust (also called FN)
T = temperature
U = air speed
W = weight
Greek Symbols
γ = ratio between specific heats
δ = relative pressure
θ = relative air temperature

κ = installation losses, Equation 5.1
= engine state; throttle setting
ρ = air density
[.]av = available
[.]o = standard conditions
[.]ref = reference
[.]req = required
6 Propeller Performance
Overview
Aircraft propulsion by propeller is still the most widespread method of converting
engine power into useful thrust. We seek the propeller parameters required to
deliver specified thrust or power to the airplane (§ 6.1), depending on the flight
condition. The simplest method for calculating the propulsive performance is the
axial momentum theory (§ 6.2.1), which is useful when detailed data of the propeller
itself are unknown. When detailed data are available, the combined momentum
and blade element theory (§ 6.2.2), along with ancillary models for transonic flow,
offers a powerful and accurate method for propeller analysis. The integration of the
propeller with the flight mechanics is discussed in § 6.3. We show that when we trim
to a specified thrust or power, the propeller generally does not operate at its most
efficient point.
KEY CONCEPTS: Propeller Parameters, Propulsion Models, Momentum Theory,

Blade Element Method, Flight Mechanics Integration, Propeller Installation.
6.1 Propeller Definitions

The basic parameters used in the analysis of the propeller are the forward speed
U, the rotational speed rpm and the tip Mach number Mtip . In addition, there are
several geometrical quantities: the number of blades, the diameter d, the pitch ϑ, the
type of blade section, the chord distribution, the tip geometry and the hub geometry.
The pitch is a measure of the orientation of the propeller on a plane normal to
the axis of rotation, as shown in Figure 6.1. The reference line for the calculation
of the pitch is the chord. Figure 6.1 shows how the local inflow angleis calculated.
If the blade section is at a radial position y, the total inflow velocity is U 2 + (y)2 .
The direction of the resulting vector velocity is inclined by an angle α on the chord
line: this is the local inflow angle, or angle of attack of the blade section.
The net thrust is the resulting aerodynamic force in the direction of the forward
flight. The drag is parallel to the inflow speed; the lift is at a right angle with the drag
of the blade element. The resulting thrust is parallel to the axis of rotation.
152
L
T F
Ωy
U
Figure 6.1. Nomenclature of forces on
blade section. α D
θ
Rotor plane
Rotor axis
Ω
The solidity σ is the ratio between the blade area (projected on the rotor disk)
and the rotor disk* :
Nc
σ =2 , (6.1)
πd
where N is the number of blades and c is the mean chord. The solidity is an important
design parameter; in fact, the propeller coefficients can be normalised by σ , to express
the effective disk loading. To make the data more useful from an engineering point of
view, we need a measure of performance. This is the efficiency, or the ratio between
propulsive power and power at the shaft
TU
η= . (6.2)
P
The propulsive efficiency expresses the ability to convert power from the engine (or
shaft power) into useful power to fly at a speed U. The dependence of the efficiency
from the other parameters is expressed as
η = f (U, rpm, d, ϑ, · · · ). (6.3)
If the aircraft is stationary on the ground, the conversion efficiency is zero, and all
the shaft power generated from burning fuel is lost. The energy E is dissipated at the
blades and transferred to the slipstream, that has an axial and rotational velocity. In
general terms, the shaft power is
P = TU + E. (6.4)
By using the dimensional analysis, the forward speed, the rotational speed and the
diameter are replaced by another dimensionless group, the advance ratio
U
J= , (6.5)
rpm d
* Note that σ is the same symbol as the relative density of the air. We maintain the convention
currently used in propeller and helicopter analysis and use σ for the solidity. Therefore, to avoid
confusion, only ρ will be used for the air density.
or
U U
J1 = = . (6.6)
R Utip
The two values of the advance ratio are proportional to each other: J1 = J (60/π ).
The advance ratio is a measure of the advancement of the propeller in one revolution,
measured in number of diameters. The advance ratio is also a scaling parameter,
indicating that all the propellers with the same J , and geometrically similar, have
the same performance index. In other words, a propeller with diameter of 4 metres,
operating at 1,000 rpm, and a forward speed 70 m/s (J = 0.33; 136 kt) has the same
performance as a scaled-down propeller having a 2 m diameter rotating at 2,000 rpm
and advancing at the same speed.
For conventional aeronautic propellers, the tip speed Utip = R is restricted to
subsonic values to keep wave drag losses, noise and vibrations under acceptable lim-
its. If we introduce the Mach number, then the tip Mach number at flight conditions
is

Utip 1
Mtip = = U 2 + (R)2 . (6.7)
a a
Propeller charts, where the parameter is the pitch angle of the blades, are to be used
in the calculation of the range and the endurance of propeller-driven airplanes.
A number of other dimensionless coefficients are defined, and propeller charts
can be found with these parameters, that are the power, thrust and torque coeffi-
cients:
T P Q
CT = , CP = , CQ = . (6.8)
ρ A(R)2 ρ A(R)3 ρ A(R)2 R
A useful relationship between the efficiency and the advance ratio involves some of
the parameters in Equation 6.8:
CT
η=J . (6.9)
CP
Although of great importance for analysing the propeller performance, the coeffi-
cients given by Equation 6.8 are not useful for determining the engine power required
to fly at a given air speed altitude and gross weight. In fact, propeller operation is
analysed either at a constant rotational speed or a constant pitch. Operation with a
mix of rotational velocities and pitch angles is also possible. For this purpose, it is
necessary to have propeller data that do not depend on the rotational speed. If the
pitch setting is unique, then the efficiency is a single curve that can only be changed
with the advance ratio.
The essential propeller theory is available in Glauert 1 , Theodorsen 2 , and von
Mises 3 , among others. Data and propeller charts, for the purpose of basic per-
formance calculations, can be found in some old NACA reports, such as Hartman
and Biermann 4;5 and Theodorsen et al. 6 . More advanced concepts are available in
AGARD CP-366 7 . Other important aspects of propeller performance are the inter-
action with the wing and the fuselage 8;9 , compressibility effects 10;11 and the propeller
noise, discussed as a separate problem in Chapter 16.
Blade Sections Type Cert.
Mass Geometry Limitations Power/Torque
Rotor Params rpm
Installation Wind Direction

Propeller
Design
Transient
Aerodynamics Performance
Block
Transonics
Polars
Hub
Interference
Figure 6.2. Flowchart of a propeller simulation model.
The establishment of a computational model for an aircraft propeller relies on

the flowchart shown in Figure 6.2. There are three categories of data that must be
assembled:
r Limitations. This category includes data from the type certificate (variable or
constant pitch, variable or constant rpm, feathering and reversing modes, pitch
selection, braking and other quantitative and qualitative parameters). Then we
have power and torque limitations (both steady-state and transient), along with
the design power, wind direction and relative speeds* . With specific reference to
the power ratings, there are data on maximum continuous power (and torque),
maximum take-off power (and torque) and maximum reverse power.
r Geometry. This category includes the geometry of the blade sections at selected
radial stations, the rotor geometry (diameter, number of blades, twist, and so
on), the mass distribution and any other relevant parameter. With very few
exceptions, the wing-section details are unknown.
r Aerodynamics. This category includes the aerodynamic polars of the blade sec-
tions over a wide range of angles of attack and Mach numbers. A transonic
aerodynamic model may be required if the data do not extend to high Mach
numbers. For completeness, we need to account for the hub drag and the inter-
ference effects due to installation.
Some critical data are often missing: the blade-section geometry, the blade twist
and the three-dimensional characteristics, such as the axial and radial twist (scimitar
geometry). However, it is possible to derive some of these parameters on the basis
of optimum propeller design. The “Aerodynamics” block in Figure 6.2 relies on the
* There can be additional data, such as software requirements, airplane integration, and de-icing
equipment. See, for example, CAA: Type Certificate Data Sheet, Dowty R391 Propeller, No. 116,
Issue 7, Feb. 2003.
Table 6.1. Design limitations of

the Dowty propeller R391; w is
the wind direction
w [deg] P [kW] Vw [kt]
45–90 <750 60
270–315
45–90 >750 15
270–315
90–270 <750 45
90–270 750–3,000 15
90–270 >3,000 5
methods described in Chapter 4, in particular, on the aerodynamic polars generated

with the transonic airfoil model (see also Figure 4.20).
6.1.1 Propeller Limitations

The certification documents usually contain data such as wind speed and power-
output limits. An example is given in Table 6.1 for the Dowty propeller R391 (Lock-
heed 382J and C-130J, Alenia C-27J). This is a six-bladed propeller having a diameter
of 4.115 m, made of composite glass, carbon reinforced structure and nickel leading-
edge sheath for erosion protection. The propeller has a variable pitch, a constant
rpm, feathering, reversing and full hydro-mechanical control. Manual pitch selection
is available for ground manoeuvring and aircraft braking. The type certificate docu-
ments report that the electronic control software meets the most stringent standards.
Table 6.2 is a summary of data for some propellers and their application† .
Those propellers that operate at constant rpm may have different rpm for dif-
ferent flight conditions. One example is the Dowty R408 propeller, installed on the
Dash8-Q400 turboprop airplane. The propeller has speed ratings of 1,020 rpm (take-
off, as in Table 6.2), 900 rpm (maximum climb), and 850 rpm (cruise). These speeds,
as well as the collective pitch, are governed by the propeller electronic control (called
PEC in the technical jargon). The PEC governs the propeller parameters in response
to the flight conditions. At a given shaft power, a decrease in tip speed causes an
increase in CT and C P .
6.2 Propulsion Models

The aerodynamic analysis is carried out in two steps of increasing complexity and
accuracy by using the following theories: 1) axial momentum theory with constant
or variable inflow (§ 6.2.1); and 2) blade momentum theory (§ 6.2.2), which in
its complete extension includes elements of the momentum theory and allows the
detailed calculation of propeller performance also in non-axial flight conditions.
† Data compiled from type certificates of propellers and/or their aircraft, published by the CAA,
EASA and FAA. Note that there can be several propeller versions for each aircraft. Where several
versions exist, data have been averaged. Consult the appropriate documents for details.
Table 6.2. Some notable propellers and their applications; P is the engine’s maximum
take-off power; W is the propeller weight, including spinner; ∗ depending on aircraft version;
†
estimated
Propeller Airplane N d [m] rpm P [kW] W [kg]
Hamilton 14SF ATR42-300, CL-415 4 3.96 1,200 2,050 156

Hamilton F568 ATR72-500 6 3.93 1,200 2,050 164
Hamilton 247F ATR72-211 4 3.96 1,200 2,048 156
Dowty R408 Dash8-Q400 6 4.12 1,020 3,782 252
Dowty R391 C-27J, C-130J 6 4.12 1,020 3,505 326
Dowty R334 CASA 212-300 4 2.79 1,591 689 87
Dowty R175 Fokker F27* 4 3.66 1,170
Dowty R193 Fokker F27* 4 3.51 <1,672 203
Dowty R352 Fokker F50 (F27-050)* 6 3.65 1,200 1,624 172
Dowty R410 Fokker F50 (F27-050)* 6 3.65 1,200 1,624 172
Dowty R381 Saab-2000 6 3.81 1,100 2,786 227
†
Ratier-Figeac FH 386 Airbus A400M 8 5.31 850 8,195 360
Hartzell HC-E5N Piaggio Avanti 5 2.16 2,000 634 81
Hartzell HC-B3TN-3D DeHavilland DHC6 3 2.59 2,110 462 50
6.2.1 Axial Momentum Theory

This theory is useful for the estimation of propeller performance with only a few
data: diameter, rpm and number of blades. The theory is based on the most ele-
mentary theoretical basis for converting the rotating power of a propeller into useful
thrust (Rankine-Froude momentum theory). This theory is conceptually simple and
practically important because it provides a first-order estimate of the power and
thrust of an open airscrew.
According to the basic momentum theory, the propeller is reduced to a rotating
disk that imparts axial momentum to the air passing through it. For the purpose
of this discussion, the air is incompressible. For reference, consider the sections 0
(far upstream), 1 (just upstream the disk), 2 (just downstream the disk) and 3 (far
downstream), as shown in Figure 6.3. The subscripts will refer to quantities at these
sections.
The theory assumes that there is no flux along the limiting streamlines and
that the velocity is continuous through the propeller disk. The free stream velocity
is equal to the propeller’s speed U = Uo. The continuity equation written for the
incompressible mass flow rate within the stream tube is
Au = const. (6.10)
The propeller’s thrust is equal to the rate of change of axial momentum at the disk,
T = A( p2 − p1 ). (6.11)
Now we can apply the Bernoulli equation between sections 0–1 upstream and 2–3
downstream to the propeller (Figure 6.3),
1 1 1 1
po + ρU 2 = p1 + ρU12 , po + ρu23 = p2 + ρU22 . (6.12)
2 2 2 2
Rotor disk Propeller
U Streamtube
0 1 2
Figure 6.3. One-dimensional axial flow model of airplane propeller.
The difference between the two equations yields the pressure jump through the
propeller
1
p2 − p1 =
ρ(u23 − U 2 ), (6.13)
2
because from the continuity equation the velocity is continuous at the rotor disk, or
u1 = u2 . From Equation 6.11 the thrust generated by the propeller becomes
1
T= ρ A(u23 − U 2 ) = ρ A3 u3 (u3 − U). (6.14)
2
The total power is
1 1 1
P = T(u1 + U) = ṁ(u3 + U)2 − ṁU 2 = ṁu3 (u3 + 2U). (6.15)
2 2 2
The energy imparted to the slipstream is
1
E=A3 ρu3 (u23 − U 2 ). (6.16)
2
This energy is minimal when the slipstream velocity is equal to the propeller velocity.
The total power is found from summing Equation 6.16 and Equation 6.15. The
slipstream velocity far downstream, u3 , is related to the velocity at the disk u1 and
the free stream velocity U. The thrust absorbed by the propeller is also written as
T = ṁ(u3 + U) − ṁU = ṁu3 . (6.17)
By combination of Equation 6.15 and Equation 6.17, we have
1
u1 = u3 . (6.18)
2
In conclusion, the air speed at the rotor disk is the average between the propeller’s
speed and the speed far downstream. Also, the downstream velocity is twice the
induced velocity at the disk. For a static propeller this relationship provides the
value of the induced velocity at the rotor disk:
1
u1 = vi =u3 . (6.19)
2
The corresponding induced power is Pi = Tvi . The thrust becomes
T = 2ρ A(U + vi )vi , (6.20)
where ρ A(U + vi ) is the mass flow rate through the disk and 2vi is the total increase
in velocity. The corresponding power is the product between the thrust and the
velocity through the disk
P = T(U + vi ) = 2ρ A(U + vi )2 vi . (6.21)
This power expression contains two terms: 1) a useful power TU, and 2) an induced
power Tvi that is a loss due to the kinetic energy imparted to the flow. The propulsion
efficiency is
TU U
η= = . (6.22)
T(U + vi ) U + vi
Therefore, the propulsion efficiency decreases as the induced velocity increases. If
U = 0, the efficiency is zero, as anticipated. The induced velocity in terms of the
thrust is found from Equation 6.20, that is quadratic in vi . The only physical solution
of Equation 6.20 is a positive vi :

1 2T
vi = −U + U 2 + . (6.23)
2 ρA
Under static conditions, U = 0, we have

T T3
vi = , Pi = . (6.24)
2ρ A 2ρ A
An important aspect of this elementary theory is the behaviour of the pressure. For
points upstream or downstream of the propeller, there is a conservation of total
pressure (sum of the static pressure and dynamic pressure, q = ρU 2 /2). The total
pressure is given by the Bernoulli equation, Equation 6.12. The engine power is
used to increase the kinetic energy of the air passing through the disk. Therefore,
we conclude that the theory involves a sudden increase of pressure through the
propeller disk, whilst the velocity of the air in the stream tube is continuous. The flow
undergoes an acceleration, and therefore the slipstream must contract, according to
the continuity Equation 6.10.
VARIABLE INFLOW CONDITIONS. Various advancements and refinements can be

added. With respect to the non-uniform axial inflow, one needs to integrate the
change of axial momentum over the propeller disk, and the result is

T= ρu3 (u3 − U) d A. (6.25)
A
An essential hypothesis is that of axi-symmetric flow through the disk, that is quite
reasonable for a propeller in axial flight. This hypothesis leads to a slightly modified
theory. Consider an annulus of width dy, corresponding to an area d A = 2π ydy. The
element of thrust generated by the mass flow is
ṁ = ρ(U + vi )d A, (6.26)
through this annulus is
dT = 2ρ(U + vi )vi d A = 4πρ(U + vi )vi ydy. (6.27)
We define the induced velocity ratio

U + vi U + vi y Un
λ= = = r = tan φ r, (6.28)
R y R Ut
where Un and Ut are the velocity components normal and tangential to the rotor
plane, and φ is the inflow angle. In general, Un << Ut ; therefore we can make the
approximation
Un λ
tan φ φ = = . (6.29)
Ut r
If we introduce the inflow velocity ratio, Equation 6.28, we find

U + vi vi
dT = 4πρ (R)2 ydy = 4πρλλi (R)2 ydy, (6.30)
R R
with λi = vi / R. Note that λi is the induced velocity ratio in absence of axial flight
velocity. It can also be written as
vi vi U U U
λi = = + − =λ− = λ − λc , (6.31)
R R R R R
with λc = U/ R. The resulting thrust is found from integration of Equation 6.30,
R
T = 4πρ(R) 2
λλi ydy. (6.32)
o
The element of power is d P = dTvi , and therefore

R
P = 4πρ(R)3 λλi2 ydy. (6.33)
o
These integrations can only be done if the radial distribution of induced velocity is
known. A solution can be found by combining the results of the blade momentum
theory (§ 6.2.2). Various further advancements are available; it is now possible to
model the case of a propeller/rotor disk inclined by any angle on the free stream,
with almost any load distribution on the rotor 12 .
6.2.2 The Blade Element Method

The axial momentum theory provides integral quantities, such as thrust and power.
These characteristics do not seem to depend on the propeller’s geometry, which is
clearly a shortcoming. With the blade element it is possible to take into account all of
the geometrical information of the propeller. The resulting theory is more elaborate;
if efficiently programmed, it is a robust tool for engineering analysis in most flight

conditions.
In this framework, the blade sections operate like two-dimensional sections,
with the local inflow conditions derived by appropriate means in the rotating envir-
onment (see Figure 6.1). Although the definition of this inflow is not obvious, and the
interference between elements on the same blade and between blades is not taken
into account, the method is otherwise extremely powerful. It allows the calculation
of basic performance from the geometrical details and the two-dimensional blade
section aerodynamics.
The relationship among inflow angle, pitch angle and angle of attack is
α = ϑ − φ. (6.34)
The pitch is a geometrical setting, whilst the angle of attack of the blade section and
the inflow velocity are operational free parameters. The lift and drag forces on this
section are
1
dL = ρcCLU 2 dy, (6.35)
2
1
dD = ρcCDU 2 dy. (6.36)
2
These forces, resolved along the direction normal and parallel to the rotor disk, give
the contributions to the thrust, torque and power for the single blade
dT = dLcos φ − dDsin φ, (6.37)
dQ = (dLsin φ + dDcos φ)y, (6.38)
d P = (dLsin φ + dDcos φ)y. (6.39)

These elements can be written in non-dimensional form, using the definitions of
thrust, torque and power coefficients. The total thrust, torque and power for N
blades will require integration of these expressions from the inboard cut-off point to
the tip.
The integrals are, in fact, not solved directly but rather numerically. If we divide
the blades into a number n of elements, each having a radial width dy j , then

n
T=N (dLj cos φ j − dDsin φ j ) , (6.40)
j=1

n
Q= N (dLj sin φ j + dDcos φ j ) y j dy j , (6.41)
j=1

n
P=N (dLj sin φ j + dDcos φ j ) y j , dy j , (6.42)
j=1
with the forces evaluated from Equation 6.35 and Equation 6.36. All of the quantities
appearing in the aerodynamic forces change with the radial position, including the
chord, the air density (for high-speed flows), the Cl and the Cd . The aerodynamic
coefficients depend on the effective inflow angle α, Reynolds number Re and Mach
number M. The key problem is to find the inflow angle φ, the resultant inflow α and
the actual inflow velocity U at each blade section.
The aerodynamic coefficients of the blade section Cl and Cd are transformed
into propeller coefficients Cx , Cy by a rotation φ:
Cx = Cl sin φ + Cd cos φ, (6.43)
Cy = Cl cos φ − Cd sin φ. (6.44)

The next step is to calculate the interference factors. According to Glauert 1 , these
factors are:
k k
a=σ , a = σ (6.45)
F − σk F + σ k
with
Cy Cx
k= , k = . (6.46)
4 sin2 φ 4 sin φ cos φ
The inflow angle is obtained from
U 1+a
tan φ = . (6.47)
y 1 − a
The corrective factor F in Equation 6.45 is calculated from

2
F= cos−1 e− f (r ) , (6.48)
π
with
N1−r
f (r ) = . (6.49)
2 rφ
The inflow angle is unknown, and some iterations are required. A correction is
usually applied to this procedure because the aerodynamic flow around the blade
section does not follow that of the airfoil. Both the tip and the root are affected by
three-dimensional effects, which lead to loss of lift and hence loss of thrust.
The corrected thrust will be dT F(r ). A plot of F(r ) shows that it is equal to one
for most of the span, but it tends rapidly to zero near the tip, with a rate depending
on the number of blades and the inflow angle.
The iterative procedure for the inflow angle shows poor convergence properties,
unless a corrective action is taken. First, it is possible that the interference factors
a and a assume too large values. Following Adkins and Liebeck 13 , we set a limiter
a ≤ 0.7, a ≤ 0.7. Second, and most important, unless some form of under-relaxation
is applied, the procedure does not converge on a highly loaded propeller. If i is the
iteration counter at a generic blade section, the under-relaxation that we propose to
use is
1
φ̃ i+1 = (φi + φi+1 ) , (6.50)
2
where φ̃ denotes the corrected inflow angle at iteration i + 1 and φi+1 is the pre-
dicted inflow angle. The different behaviour in convergence properties is shown in
Figure 6.4. When we use Equation 6.50 the convergence is uniform and rapid; in
absence of under-relaxation, the solution is hopeless.
10 0.1 5
Inflow angle Inflow angle
Error Error
90
4
5 0.075
Inflow angle, α
Inflow angle, α
3
60
Error
Error
0 0.05
2
30
1
-5 0.025
0
0
-10 0
5 10 15 20 25 30 5 10 15 20 25 30
Iteration count Iteration count
(a) (b)
Figure 6.4. Effect of under-relaxation on inflow convergence of a highly loaded propeller,
y/R = 0.96.
The procedure described neglects a number of practical difficulties. These are

related to the format in which the aerodynamic data are available. Ideally, one would
have data like Cd = Cd (α, M), Cl = Cl (α, M) in a matrix form.
PROPELLER TRIM. The propeller can be trimmed to provide a required thrust, power
or torque. One way of doing this is to specify the flight conditions (h, TAS, rpm)
and to determine the collective pitch corresponding to the required parameter. A
suitable numerical method is the bisection. This method consists in bracketing the
solution (i.e., the collective pitch) with a very low and a very large value, ϑ1 and ϑ2 ,
respectively.
6.2.3 Propeller in Non-Axial Flight

Asymmetric flight occurs when the propeller axis is not aligned with the flight speed.
This situation generates variable inflow across the actuator disk and hence some
degree of unsteady flow and mechanical vibrations. The latter problem is typical of
several turboprop aircraft. From a propulsion point of view, there will be changes in
shaft power and net thrust. The resulting changes in in- and out-of-plane forces and
moments are important in the analysis of flight control procedures.
To understand how the asymmetric inflow can be generated, consider the aero-
dynamic effects of the fuselage, which change the effective angle of attack at the
disk, and the effects of aircraft attitude, especially at climb-out conditions.
To begin the analysis, let us consider a reference system based on the propeller.
The convention is that the x-axis coincides with the rotational axis of the propeller;
the y-axis is on the horizontal plane pointing to the right of the flight, and z makes a
right-hand Cartesian reference system with x and y; z points downward. A rotation
of the propeller disk around x, y and z is called roll, pitch and yaw, respectively. The
order in which these rotations take place is very important. In the following analysis,
we will be limited to combinations of pitch and yaw because roll effects are generally
Figure 6.5. Propeller reference system and rotation angles.
small unless the aircraft is performing a rapid manoeuvre. A sketch indicating pitch
and yaw angles is shown in Figure 6.5.
Figure 6.6 shows a field plot of the element of thrust coefficient dCT when the
propeller advances through the azimuth. The case refers to a yaw angle of 5 degrees,
a pitch angle of 10 degrees and typical cruise conditions. Although the blade sections
would be affected by time-dependent inflow conditions, this case has been calculated
in quasi-steady mode, whereby the blade sections move from one steady state to the
next.
dCT
1.00E-03
8.00E-04
6.00E-04
4.00E-04
2.00E-04
Ψ=0
Figure 6.6. Loads on disk for asymmetric flight of the F568 propeller; pitch = 10 degrees,
yaw = 5 degrees, U = 250 kt, ϑ = 40 degrees.
Figure 6.7. Reconstructed F568 propeller and hub, side and front views.
6.2.4 Case Study: Hamilton-Sundstrand F568 Propeller

The F568 is a modern propeller for turboprop-driven airplanes developed by
Hamilton-Sundstrand† for the ATR42, ATR72, CASA C295 and Ilyushin-114-100.
The propeller is six-bladed and has a nominal diameter of 3.93 m and a non-linear
twist with swept-back tips. The blades are made of a Kevlar shell with a graphite spar.
A titanium strip is applied at the leading edge to prevent blade erosion. The hub
is made of steel. The propeller system is protected against low pitch angle in flight,
over-speed and hydraulic pressure loss. The pitch is limited between −14 degrees
(reverse) and 78.5 degrees (feather).
The nominal rotational speed is 1,200 rpm with a design shaft power of 2,244 kW;
rotation is clockwise, looking forward. Electrical de-icing of the blades is available.
Some blades are interchangeable with each other. The geometry of the reconstructed
propeller is shown in Figure 6.7.
The propeller‡ is driven by a power turbine via a reduction gearbox. The collect-
ive pitch is controlled by a hydro-mechanic unit (propeller valve module, installed in
each engine nacelle). The pitch control is electronic with hydro-mechanical backup.
The propeller valve module is then controlled by an electronic control on each
engine; a propeller interface unit transfers control to the flight deck. The propeller
valve module has several important functions, including rpm control, over-speed
control, reversing, feathering, low-pitch protection and synchrophasing. The latter
term refers to the synchronisation of the phase between propellers, which may be
important in controlling interference noise between propellers (see propeller noise
in § 16.7).
† The manufacturer identifies the propeller with the numbers 568: 5 is the design number; 6 is the
number of blades; 8 is the blade shank size; F denotes the flange mounting. Additionally, there can
be another identifier, such as 568-x, where x denotes the airplane application.
‡ This information is adapted from the ATR72-500 Flight Crew Operating Manual.
KTAS
0 100 200 300
1 0.02
20
25
30
35
15
Propulsive efficiency, η
0.015
0.8
CT
10
0.01
5
0.6
35 0.005
30
25
20
15
0 5 10
0.4 0
0 0.2 0.4 0.6 0 0.002 0.004 0.006 0.008 0.01
Advance ratio, J CP
(a) (b)
7 1
5
6 10
0.9
Propulsive efficiency, η
5 15
0.8
CT / CP
20
4
25 0.7
3
30
0.6
2
35
1 0.5
0 0.002 0.004 0.006 0 0.002 0.004 0.006
CP CP
(c) (d)
Figure 6.8. Calculated performance of the Hamilton-Sundstrand propeller F568-1.
The propeller electronic control is a hardware–software box that controls the

pitch via a closed-loop feedback. The control input is the torque Q = P/ . At low
fuel flow, the control unit may be unable to adjust the pitch to match the torque;
therefore, it reduces the rpm. In fact, the rpm can take a number of values, which
are 82% (984 rpm) at climb-out and cruise, 100% at take-off and rejected take-off.
This is an important fact: the propeller can assume different rpm, depending on the
flight segment, but then it is maintained constant with an electronic pitch control.
Figure 6.8 shows the propeller performance (efficiency, shaft power, net thrust
and maximum sectional L/D) as a function of the advance ratio (and KTAS) for
selected pitch settings, as calculated with the FLIGHT code. Figure 6.8a shows that
by appropriate choice of the collective pitch, it is possible to ensure nearly constant
propulsive efficiency with increasing flight speed.
2000 25
20
Se
1500
a
Le
ve
Shaft power, kW
Net thrust, kN
Se
15
a
Le
ve
1000
l
10
7 ,0
00
m 7 ,0
00
500 m
5
0 0
0.1 0.2 0.3 0.4 0.5 0.6 0.1 0.2 0.3 0.4 0.5 0.6
Advance ratio, Jp Advance ratio, Jp
(a) (b)
Figure 6.9. Calculated altitude performance of the Hamilton-Sundstrand propeller F568-1.
The flight altitude has negligible effect on the propulsive efficiency. However,
the performance of the propeller (shaft power and net thrust) itself depends greatly
on the flight altitude, as shown in Figure 6.9.
The aerodynamic analysis presented earlier referred to the relatively simple case
in which the inflow is axial. However, it is not unusual that the propeller is required
to work in yawed conditions, for which there will inevitably be some limitations.
Another limitation is related to the matching between the propeller itself and the
engine. An example is shown in the propeller chart in Figure 6.10. This figure displays
0.02
NB: dots denote maximum efficiency
4,000
0.015
3,000
Shaft power, kW
CP
Engine
Design PWR
0.01
40 2,000
0.005
1,000
45
40
35
30
0 0
0 0.2 0.4 0.6 0.8 1
Advance ratio
Figure 6.10. Simulated F568 propeller; power coefficient and shaft power; lines of constant
pitch shown.
the C P and the corresponding shaft power as a function of the advance ratio. Note
that the propeller is capable of providing a C P considerably larger than the maximum
shaft power, indicated by the horizontal dotted line.
6.3 Flight Mechanics Integration

We have so far explained the performance of the propeller as an isolated propul-
sion system. We now discuss the problem of integration of a propeller into a flight
performance analysis. If Treq is the thrust required at a given flight condition, it will
be delivered by the propeller (in a large portion) and by the exhaust jet Fg of the
turboprop engine (in a smaller proportion):
Treq = ne (Tp + Fg ), (6.51)
where ne denotes the number of operating engines. The residual thrust Fg is thought
to be always in the direction of the propeller thrust Tp . The propeller thrust and the
corresponding shaft power are given by Equation 6.8. Each flight condition is now
discussed separately. In any case, once the propeller thrust Tp and the corresponding
CT have been calculated, we need to calculate the propeller C P and the propeller
power Pp . The power required by the power-plant will be
1 Pp
Preq = , (6.52)
ne ηg ηmηi
which accounts for losses in the gearbox, for the mechanical efficiency of the engine
and for the installation losses.
TAXIING. At this condition a relatively small thrust is required. The propeller is

trimmed at the prescribed ground speed to provide a thrust Tp that satisfies Equa-
tion 6.51, with Fg 0. The method is the following:
1. Trim the propeller to the requested thrust coefficient CT ; this operation provides
a power coefficient C P and hence a shaft power Preq , via Equation 6.52.
2. Solve the turboprop engine problem in inverse mode, in order to determine
the engine state E corresponding to the required power Preq . The solution will
provide, among other quantities, the residual thrust Fg , which is negligible.
This solution forces the propeller to work with a low efficiency. To overcome
the problem of high fuel consumption, it is often preferable to have one propeller
working at higher thrust than two propellers working at low thrust and low efficiency.
For this reason, many turboprop airplanes taxi in and out of the gate with one working
propeller, whilst the other one is feathered. Clearly, other problems intervene, such
as the lateral trim of the aircraft during ground roll, the start-up of the inoperative
engine before take-off and the operation of other systems.
Figure 6.11 shows the calculated performance of the F568 propeller at taxi
conditions for specified net thrust ranging from 2 kN to 8 kN. If the total net thrust
required for rolling on the taxi-way is 4 kN (2 kN × 2 propellers), the propulsive
0.4
0.3
Propulsive efficiency
0.2
0.1
0
0 2 4 6 8 10
Net thrust, kN
Figure 6.11. Propeller efficiency at taxi conditions; ground speed = 3.9 kt, altitude = 100 m.
efficiency is about 6% when two propellers are running and 17% when one propeller
is running.
TAKE-OFF. At take-off it is required to operate the propeller at maximum thrust:
max(T) = f (h, J1 , ϑo). (6.53)
This condition does not necessarily occur at maximum shaft power. Alternatively,
the propeller can be trimmed to maximum power or a fraction of the maximum
power to provide a net thrust. Again, the propulsive efficiency can be relatively low.
The numerical procedure is the following:
1. Set Fg = 0.
2. Trim the propeller to the thrust or power condition; this operation provides the
value of the collective ϑ and the corresponding shaft power or net thrust.
3. Solve the turboprop engine problem to determine the engine state E corres-
ponding to the required power Preq , corrected for all of the conversion losses
(Equation 6.52). The solution will provide the residual thrust Fg .
4. Calculate the corrected propeller thrust Tp = T − Fg and trim the propeller to
the net thrust Tp . Reiterate from point 3.
Sometimes there is the possibility of increasing slightly the rpm from its nominal
value, which is quite useful in improving the thrust capabilities.
EN-ROUTE CLIMB. The climb problem does not have a unique solution, particularly
because climb is performed in phases. There are at least two distinct climb segments
at constant CAS (or IAS), as discussed in § 10.3. For these climb segments, one
suitable computational solution is the following:
1. Specify the climb rate vc and calculate the required net thrust in accelerated
flight:

vc D 1 dU
T=W + + . (6.54)
U L g dt
2. The propeller is trimmed for a specified CT ; this operation yields the performance
parameters C P , ϑ and η.
3. Calculate the net shaft power Preq , corrected for all of the losses (Equation 6.52).
4. At the required shaft power, calculate engine state E; this operation will provide
the fuel flow ṁ f and the residual thrust Fg , among other parameters.
5. If the required climb rate is too high (engine state beyond flight envelope;
engine overheating or over-speeding), reduce the climb rate and reiterate from
point 1.
The level-flight acceleration from the lower to the higher CAS is treated in a
similar manner, except that an acceleration value is specified instead of a climb rate.
CRUISE. Equation 6.51 is to be solved at a fixed altitude and Mach number (or TAS)
because the (h,TAS) condition is determined in the computer simulation model as
an optimal operation point. Thus, the net thrust is equal to the total aerodynamic
drag:
1. Set Fg = 0.
2. Calculate the aerodynamic drag D = T at the current flight condition.
3. Trim the propeller to the required CT ; calculate the propeller’s parameters C P
ϑ, η; calculate the net shaft power, corrected for installation losses.
4. Calculate the engine state E corresponding to the required power; this operation
yields ṁ f , Fg and other engine parameters.
5. Correct the net propeller thrust according to Equation 6.51.
6. Trim the propeller as at point 2.
7. Iterate until convergence, that is, when there is negligible change in the net
propeller thrust (convergence is achieved in three or four iterations.)
EN-ROUTE DESCENT. This case is treated as the cruise problem, with the propeller
required to deliver the required thrust.
Figure 6.12 shows the flight envelope of the turboprop engine PW-127M (ATR-
72). The graph displays the behaviour of the residual thrust Fg as a function of the
shaft power over the full range of flight altitudes. At low power settings, this thrust is
clearly negligible, but it grows almost linearly. The propulsive efficiency is not used
in any of the flight conditions discussed. However, it is of some interest to calculate
the net propulsive efficiency of the propeller-gas-turbine engine system, which is
(Tp + Fg )
η= U. (6.55)
Pshaft
1.2
Residual thrust Fg9, kN 0.8 l

ve
Le
a
Se
0.6
0.4
0.2
0
0 500 1,000 1,500 2,000
Shaft power, kW
Figure 6.12. Estimated residual thrust Fg versus shaft power for the turboprop engine
PW127M; data at 1,000 m intervals, standard day.
Equation 6.55 would have to be corrected to take into account the gearbox efficiency
ηg (see § 6.4.1) and the shaft mechanical efficiency ηm. Thus, we have
(Tp + Fg )
η = η g ηm U. (6.56)
Pshaft
The foregoing analysis has shown that the propulsive efficiency is not a key parameter
in the airframe-propeller integration, as long as the propeller is trimmed to a specified
thrust or power. There are instances when the propeller efficiency is low and other
cases when it is high. The changes required in propeller pitch even within one flight
condition (for example, en-route climb) make it impossible to manage without an
appropriate flight control system. For this reason, modern variable-pitch propellers
are controlled by software and actuators (FADEC), like most other airplane sub-
systems.
6.3.1 Propeller’s Rotational Speed

Assume that the aircraft is required to fly with an air speed U; this condition also
requires T = D, that is, a fixed value of the required thrust. If the propeller rpm is
decreased, the following condition holds:
T
CT (R)2 = = constant. (6.57)
ρA
A decrease in rotational speed allows (in principle) the aircraft to fly faster by
virtue of an offset in the effective tip Mach number (Equation 6.7). Consider the
example shown in Figure 6.13, which refers to the Dowty R408 propeller that powers
the Bombardier Dash8-Q400 (see Table 6.2). This propeller has a take-off speed
0.04 0.04
850 rpm 850 rpm
1,020 rpm 1,020 rpm
θ = 43 degs θ = 43 degs
0.03 0.03
C
CT
CT
B
0.02 0.02
B A
A
0.01 D 0.01
0 0
0 100 200 300 400 0 0.2 0.4 0.6 0.8 1
KTAS Advance ratio
(a) (b)
0.02 0.02
850 rpm B
850 rpm
1,020 rpm 1,020 rpm
0.015 0.015
A A
CP
CP
0.01 0.01
0.005 0.005
0 0
0 0.2 0.4 0.6 0.8 1 0 0.01 0.02 0.03 0.04
Advance ratio CT
(c) (d)
Figure 6.13. Effect of rotational speed on the CT of the Dowty R408 propeller (calculated).
The curves are plotted at constant pitch values with a 2.5-degree interval.
equal to 1,020 rpm and a landing speed of 850 rpm. Assume that the aircraft cruises
at 280 KTAS. There are three trim options:
1. With a propeller speed 1,020 rpm, the aircraft is trimmed with a pitch ϑ 43
degrees and a CT 0.0167 (point A in Figure 6.13a). To maintain the air speed
whilst reducing the rpm, the CT must increase; this change can be obtained by
increasing the collective pitch (point A to C).
2. If the pitch is unchanged, reducing the rpm causes the aircraft to slow down
(point A to B in Figure 6.13a).
3. The final option shows the propeller moving to point D in Figure 6.13a as the
propeller rpm is reduced. This point corresponds to a lower CT . The only way
Equation 6.57 can be satisfied is by increasing the density altitude: in other
words, the aircraft must descend.
Figure 6.14. Side view of the propeller-nacelle-wing installation on the ATR72-500 (adapted
from the manufacturer’s documents).
Figures 6.13b and 6.13c show that when the CT and C P curves are plotted against
the non-dimensional speed J1 , they are basically overlapping. This is a useful result.
In fact, we calculate only one flight envelope, for example, at the take-off rpm rating
(1,020 rpm). To perform the analysis at landing conditions (850 rpm), we use the
chart in Figure 6.13b. For the case of a propeller trimmed to a required thrust,
we have the state vector S = {h, U, T, }. The corresponding CT is calculated from
Equation 6.57. For example, in Figure 6.13b, point A corresponds to a J A 0.67 and
CTA ∼ 0.0167 at 1,020 rpm. With the lower rpm, we have CTB (1,020/850) × 0.0167
0.02. The advance ratio is increased to the new value J B = U/ R 0.80. The
corresponding operation point moves to B, which requires an increase in collective
pitch from ϑ A 43 degrees to ϑ B 46 degrees. The power coefficient C P is calculated
at (J B, ϑ B), as shown in Figure 6.13d.
6.4 Propeller Installation Effects

The foregoing analysis has presented performance issues of an isolated propeller in
axial flight. Three key aspects must now be considered: one is the effect of installation,
such as the combined effect of propeller hub and downstream or upstream elements
(such as engine mounting and wing); the second aspect deals with the coupling of the
propeller to an engine shaft via a gearbox, which causes some losses in propulsive effi-
ciency; and the third is the issue of non-axisymmetric flight (yaw and angle of attack).
PROPELLER-WING INTERFERENCE. The propeller is mounted, via a nacelle, onto a

wing. The most common case is that of a tractor propeller, whereby the wing is
placed downstream of the rotor disk. Due to the combination of swirl effects and
increased air speed in the slipstream, the net effect on the wing is an increase in drag.
If the wing is placed at least one propeller radius downstream (Figure 6.14) and the
air-flow over the wing remains attached, the effect is minimal.
PROPELLER-INTAKE INTERFERENCE. The engine intake is generally within the work-

ing section of the propeller; thus, it collects some of the airflow and prevents the
development of the full slip-stream. One way of accounting for this effect is to
subtract from the mass flow rate the engine intake mass flow. This assumption is
reasonable because the intake flow is drawn axially. There is one second-order pos-
itive effect in this interference because the intake pressure increases slightly, and so
does the overall compression ratio. The installed net thrust is
Tp∗ = Tp − Ti (6.58)
where Ti is the absolute value of the loss due to the engine intake. This contribution
is written as

ṁvi if W1 < ṁ
Ti = (6.59)
0 if W1 > ṁ
because in the nomenclature shown in Figure 6.3, the intake is right behind the
propeller; ṁ is the mass flow rate through the intake area. We assume that if W1 < ṁ
then Equation 6.59 gives the correct estimate of the installation losses on the net
thrust; if W1 > ṁ, we assume that Ti ∼ 0 because the engine is drawing more mass
flow than it would normally be passing through the intake area if the engine were
not there.
The induced velocity must be calculated locally. The use of Equation 6.23 gives
a too high engine inlet mass flow. A more accurate estimation for the induced
velocity is vi d P/dT at the centre of the inlet. Thus, the average mass flow rate
passing through the inlet area Ai is ṁ ∼ ρ Ai vi . In summary, when the first instance
in Equation 6.59 is true, the thrust correction is
Ti = ρ Ai vi2 . (6.60)
For the purpose of this analysis, the induced velocity is taken at the centre of the
inlet and is given by vi = U(1 + a), where a is the local axial interference factor
calculated with the blade-element method (see § 6.2.2). It is not recommended to
use the basic momentum theory (Equation 6.51) because it grossly over-estimates
the value of the intake mass flow. In any case, the installation effects are introduced
iteratively, starting from Equation 6.51, according to the following procedure:
1. Specify the total thrust T and set Fg = 0; this condition specifies the CT (without
engine).
2. Trim the propeller to the CT ; calculate the C P and the uninstalled power P.
3. Solve the turboprop problem in the inverse mode to deliver the power P; the
solution also yields Fg and W1.
4. Calculate the installation thrust difference from Equation 6.60.
5. Assemble the total thrust from T = Tp + Fg − Ti .
6. If T > T, then reduce the fuel flow and restart from point 3; otherwise increase
the fuel flow and restart from point 3.
7. Iterate until the difference in Tp between iterations is negligible.
The increase or decrease in fuel flow is indeterminate. A more rational approach

is to build the computations method around a bisection method.
There are cases when the propeller is mounted aft, such as in the Piaggio Avanti.
In this case, we must account for the interference between the exhaust gas and the
propeller blades, which creates a complicated case of blade interaction and further
noise emission.
6.4.1 Gearbox Effects

The propeller requires a power train for reducing the large rpm of the gas turbine
engines to the more moderate rpm required to maintain the blade tips within a
moderate transonic Mach number. This is done through a gearbox that bears all the
torque transmitted by the engine to the propeller and reduces the rpm by a factor
of 5 to 10. There are various technological aspects related to the gearbox, but in
this instance we limit our discussion to performance problems, namely the power
losses. The loss of power can be monitored indirectly by the heat generation. As a
first-order analysis one can assume that this quantity is proportional to the square
of the pressure between gear wheels or to the square of the transmission torque. A
semi-empirical expression for gearbox efficiency is the following:
2
Q
ηg = 1 − gr , (6.61)
Q∗
where Q is the actual torque, Q∗ is the design torque, gr is the reduction ratio, or

gr = κ , (6.62)
N1
where N1 denotes the actual rotational speed of the power-turbine shaft and κ is an
empirical factor. The heat generated within the gearbox is estimated from
dQ
Q̇ = = (1 − ηg )Pshaft . (6.63)
dt
For example, at a nominal Pshaft ∼ 103 kW with ηg = 0.98, we have Q̇ 20 kJ/s,

which is a considerable amount of heat, especially considering that lubricating oils
have heat capacities of the order of 20 kJ/kg K.
Summary
We have defined the key performance parameters of an aircraft propeller. We then
proposed a method for modelling a realistic propeller based on a mix of manu-
facturers’ data, certified data, photographs and educated guesses. The propeller
performance is then computed by using a combined blade element and momentum
theory. The model is extended to include flight in yaw conditions. Propeller charts
are shown over the entire range of advance ratios, pitch settings and flight altitude.
The propeller model is fully integrated into the flight mechanics program by a com-
bination of the propeller thrust/power, the engine power and the residual jet thrust.
Various trim conditions are possible, including thrust and power. It is shown how the
efficiency of the propeller is greatly dependent on the flight conditions; in particular,
propellers operating at very low speed (taxi and ground manoeuvre) are inefficient.
We also demonstrated that the residual thrust from the power-plant can contribute
to the total thrust. Finally, we considered secondary effects, such as engine/propeller
installation and gearbox losses.
Bibliography
[1] Glauert H. Airplane Propellers, volume 4 of Aerodynamic Theory. Dover ed.,
1943.
[2] Theodorsen T. Theory of Propellers. McGraw-Hill, 1948.
[3] von Mises R. Theory of Flight. Dover Publications, 1959.
[4] Biermann D and Hartman EP. Wind-tunnel tests of four- and six-blade single-
and dual-rotating tractor propellers. Technical Report 747, NACA, 1942.
[5] Hartman EP and Biermann D. The aerodynamic characteristics of full-scale
propellers having 2, 3, and 4 blades of Clark Y and R.A.F. 6 airfoil sections.
Technical Report R-640, NACA, 1938.
[6] Theodorsen T, Stickle GW, and Brevoort MJ. Characteristics of six propellers
including the high-speed range. Technical Report R-594, NACA, 1937.
[7] Aerodynamics and Acoustics of Propellers, AGARD-CP-366, Feb. 1985.
[8] Wieselberger C. Contribution to the mutual interference between wing and
propeller. Technical Report TM-754, NACA, 1934.
[9] McHugh J and Eldridge H. The effect of nacelle-propeller diameter ratio
on body interference and on propeller and cooling characteristics. Technical
Report R-680, NACA, 1939.
[10] Delano JB. Investigation of the NACA 4-(5)(08)-03 and NACA 4-(10)(08)-
03 two-blade propellers at forward Mach numbers to 0.725 to determine
the effects of camber and compressibility on performance. Technical Report
R-1012, NACA, 1951.
[11] Stack J, Delano E, and Feldman J. Investigation of the NACA 4-(3)(8)-045 two-
blade propellers at forward Mach numbers to 0.725 to determine the effects of
compressibility and solidity on performance. Technical Report R-999, NACA,
1950.
[12] Conway JT. Exact actuator disk solutions for non-uniform heavy loading and
slipstream contraction. J. Fluid Mech., 365:235–267, 1998.
[13] Adkins CN and Liebeck RH. Design of optimum propellers. J. Propulsion &
Power, 10(5):676–682, Sept. 1994.

A = disk area
Ai = engine intake area
a = speed of sound
a = axial-interference factor, Equation 6.45
a = radial-interference factor, Equation 6.45
c = chord of a blade section
c = mean chord
Cd = sectional-drag coefficient
Cl = sectional-lift coefficient
CP = power coefficient
CQ = torque coefficient
CT = thrust coefficient
Cx = axial-force coefficient
Cy = normal-force coefficient
d = diameter
D = aerodynamic drag
E = energy
E = engine state
f = tip/root correction factor
F = Prandtl factor, Equation 6.48; total aerodynamic force (Figure 6.1)
Fg = residual thrust from turboprop engine (also Fg9)
gr = gear-reduction ratio
h = flight altitude
i = blade-element counter; iteration counter
j = blade counter
J ,J1 = advance ratios
k, k = parameters defined by Equation 6.46
L = aerodynamic lift
ṁ = mass flow rate (also W1)
ṁ f = fuel flow (also Wf6)
M = Mach number
Mtip = tip Mach number
N = number of blades
N1 = nominal gas generator turbine rpm (also called N1)
ne = number of operating engines
p = pressure
P = shaft power
Pi = induced power
Pp = propeller power
Preq = required power
q = dynamic pressure, ρU 2 /2
Q = shaft torque
Q∗ = design shaft torque
Q = thermal energy, Equation 6.63
r = radial position, y/R
R = rotor radius
t = time
T = net thrust
Tp = propeller thrust
Tp∗ = installed net thrust
Treq = required thrust
u = air speed (local; reference stations 1, 2, 3)
U = air speed
Un , Ut = normal and tangential velocity components
Utip = tip speed, R
vc = climb rate
vi = induced velocity at propeller disk
y = radial station
vc = climb rate
vi = induced velocity at propeller disk
Vw = wind speed
W = weight
x,y,z = Cartesian coordinate system
Greek Symbols
α = effective angle of attack
η = propulsive efficiency
ηg = gearbox efficiency
ηm = mechanical efficiency
ηi = effect of installation on propulsive efficiency
λ = tip-speed ratio (advancing)
λc = tip-speed ratio (static)
κ = empirical factor in Equation 6.62
ρ = air density
φ = inflow angle
φ̃ = corrected-inflow angle
ϑ = pitch angle
ϑo = collective-pitch angle
σ = rotor solidity
w = wind-direction angle
= rotational speed
7 Airplane Trim
Overview
The problem of airplane trim involves the determination of the control requirements
to maintain a stable flight or to perform specified manoeuvres. The position of the
centre of gravity (CG) is essential, and some consideration is given to this effect
on the cruise flight. In this chapter we consider the problem of static longitudinal
trim (§ 7.1), lateral trim and airplane control under asymmetric thrust (§ 7.2). We
consider only steady-state conditions. Transient conditions are the subject of flight
dynamics and therefore are not considered in this chapter.
KEY CONCEPTS: Longitudinal Trim, Trim Drag, Stick-Free Trim, Thrust Asymmetry,
Lateral Control.
7.1 Longitudinal Trim at Cruise Conditions

The airplane in free flight is subject to a number of forces that must be balanced to
ensure steady-state flight. In the following analysis, cruise condition is a term that
is extended to the airplane climbing and descending, subject to the airplane being
in clean configuration. We consider the role of the tail-plane and the elevator in
providing longitudinal control (longitudinal trim). Because the wing lift and the CG
are not at the same point, the airplane will have a nose-down or nose-up pitching
moment.
Although the CG is generally on the vertical symmetry plane, neither the
propulsive forces nor aerodynamic forces are on that plane. To begin with, we
assume that the contributions from engines and wings are symmetric, so that it is
possible to reduce the problem to a balance of forces and moments in one plane.
For longitudinal equilibrium, two equations must be satisfied: the equation for the
pitching moment and the momentum equation in the vertical direction.
The reference distances can be calculated in a number of ways. For example,
we can assign global longitudinal coordinates with respect to the nose and call x the
distances from this point. Note, however, that in most cases the manufacturers have
their own method of assigning longitudinal sections from a reference point that is
not at the nose.
179
180 Airplane Trim
Lw
Lt
xt
M ot
Mow
AC ηt AC
CG
ho c
hc
xLE (MAC) MAC

x LE (MAC)
xCG
Figure 7.1. Nomenclature for longitudinal trim, adapted from Ref. 1 ; c = MAC.
In stability calculations, we make reference to the mean aerodynamic chord

of the wing (MAC; see § 2.6); the MAC is itself referred to the coordinate system
centred at the nose. Another possible reference is the CG. However, this point
moves in flight, due to fuel burn, and with the airplane loading (see Chapter 3).
The nomenclature we use is the following: xLE is the root leading edge of the
wing, xLEt the root leading edge of the stabiliser; and xMAC and xMACt are the leading
edges of the MAC of wing and stabiliser, respectively. The distance between the CG
and xLE (MAC) is a fraction of the MAC and is called h; this quantity is not to be
confused with the flight altitude* . The distance between the aerodynamic centre AC
and xLE (MAC) and is fraction of the MAC and is called ho. These and other relevant
symbols are shown in Figure 7.1.
EQUATION FOR VERTICAL EQUILIBRIUM. This equation is:
Lw + Lt − W = 0. (7.1)
If we divide Equation 7.1 by the product q A, we obtain a normalised form, that is,
an equation containing only lift coefficients:
CLw + CLt − CL = 0. (7.2)
Note that all of the lift coefficients in Equation 7.2 are calculated with respect to
the wing area. This detail is quite important because in most cases the aerodynamic
coefficients of a wing are calculated with respect to the planform area of the wing
itself. The lift coefficients can be written as
CLw = CLα (α − αo),
(7.3)
CLt = CLαt αt .
* The symbol x is a dimensional quantity (a longitudinal coordinate), whilst h is non-dimensional.
The effective inflow angle of the horizontal stabiliser depends on the configuration
and the downwash created by the wing. With reference to the latter point, if ηt is the
angle between the neutral lines of the horizontal tail and the wing, and α denotes
the change in angle of attack due to the wing’s downwash, then the inflow angle of
the horizontal tail is
αt = α + ηt − α. (7.4)
The angle ηt depends on the configuration of the airplane; α depends on the

operational conditions, including weight, air speed and flight altitude. Next, assume
that the lift distribution on the wing is elliptical. This distribution generates a uniform
downwash w, such that
w 2Lw
= . (7.5)
U πρb2 U 2
A span efficiency factor can be introduced to take into account the departure from
elliptic loading; otherwise the downwash can be calculated from the effective wing-
loading distribution. The change in tail plane incidence as a result of this downwash
is
w CL
α , (7.6)
U π AR
where AR = b2 /A denotes the wing’s aspect-ratio. If a numerical solution of the
lifting surface is available, the equivalence in Equation 7.5 is not required: the norm-
alised downwash is calculated numerically (see also § 4.8). Finally, solve Equation 7.2
in terms of the wing’s angle of attack with the inflow angle for the horizontal tail
(Equation 7.4 and Equation 7.6). The result is the following:

w
CLα (α − αo) + CLαt α − + ηt − CL = 0, (7.7)
U
which solved in terms of the inflow angle yields

1 w
α= CL − CLα αo + CLαt + ηt . (7.8)
CLα + CLαt U
The additional condition for longitudinal trim is derived from the balance of the
pitching moments, which is derived next.
PITCHING-MOMENT EQUATION. With reference to Figure 7.1, the pitching-moment

equation is:
M = Mow + Mot + (h − ho)c W − Lt xt . (7.9)
The moment contributions in Equation 7.9 are (from left to right): moment of the
wing Mow and tail-plane Mot ; moment due to the CG offset (h − ho); and moment
of the tail-plane lift. If we divide Equation 7.9 by the quantity q Ac, we obtain a
non-dimensional form of the equation:
CM = CMow + CMot + (h − ho)CL − CLt Vt , (7.10)

182 Airplane Trim
where the factor Vt denotes the tail volume coefficient

At xt
Vt = . (7.11)
A c
The term CMo = CMow + CMot is the pitching moment of the airplane, interpreted as
the sum between the contribution of the isolated wing and horizontal stabiliser. The
pitching moment coefficients are defined by
Mow Mot Mow + Mot
CMow = , CMot = , CMo = . (7.12)
q Ac q Ac q Ac
With the inflow angle of the horizontal tail given by Equation 7.4, the pitching-
moment equation becomes:

w
CM = CMo + (h − ho)CL − Vt CLαt α − + CLαt ηt . (7.13)
U
The trim of the airplane can be done by using elevator and/or tab deflection. If
elevator and tab are present, the tail lift is then written in linearised form (thanks to
the small angles):
CLt = CLαt αt + CLδ δ + CLβ β, (7.14)
where δ and β denote the elevator and tab deflection, respectively. By including these
terms in Equation 7.13, the non-dimensional pitching-moment equation becomes:

w
CM = CMo + (h − ho)CL − Vt CLαt α − + CLαt ηt + CLδ δ + CLβ β . (7.15)
W
At least one control deflection is required to trim the airplane. The angle of attack of
the wing is taken from Equation 7.8. In the general case, Equation 7.15 has infinite
solutions. The best solution is the one that yields minimum trim drag (§ 7.1.1).
TRIM WITH LANDING GEAR DEPLOYED. The solution of the problem is similar when we
consider the airplane with landing gears deployed. The landing gear adds parasite
drag and has no effect on the induced drag. However, because of the location of
the landing gear, there can be an additional contribution to the pitching moment of
the airplane. The pitching moment Equation 7.15 will contain the additional term
CMlg for the landing gear. This term arises from the bluff-body drag of the landing
gear, which tends to cause a nose-down contribution. An estimate of this moment
contribution is as follows:
Mlg DNlg zN + DMlg zM , (7.16)
where zN and zM are the vertical coordinates (with respect to the CG) that define
the line of action of the nose- and main-landing-gear drag, respectively. To be
more precise, the drag contributions from the landing-gear system are split between
the following contributions: 1) the wheels drag acts at the centre of the wheel; 2)
the struts drag acts at the centre of the strut; and 3) the drag of the bay doors acts
approximately at the hinge points. The drag components are calculated in § 4.2.6.
The lines of action of these drag components with respect to the CG are calculated
with the methods shown in § 3.8.2 and the geometry model (see Chapter 2).
The corresponding pitching-moment coefficient CMlg is calculated by dividing

Equation 7.16 by q Ac. Therefore CMlg is added to the right-hand side of Equation
7.15 to yield a larger absolute value of the pitching moment.
7.1.1 Trim Drag

One aspect of the trim is that the deflection of the tail surfaces creates the trim drag.
The trim drag is calculated from
2
At
CDtrim = kCL − kCLw + kt CLt
2 2 2
. (7.17)
A
Equation 7.17 expresses the difference in induced drag between a condition of zero
tail-plane lift (no tail-plane trim required; all lift generated by the wing) and the
induced drag in the trimmed condition (lift split between wing and tail-plane). The
tail-plane term is multiplied by a corrective factor At /Aonly if the CLt is referred to
the tail-plane area. Unless we perform this rescaling, there is an inconsistent mix of
coefficients. In fact,

Lt At
= CLt = CLAt , (7.18)
qA A
where the right-hand term indicates that the CL is referred to the wing area A. Note
that there are cases in which the trim drag is negative.
7.1.2 Solution of the Static Longitudinal Trim

From Equation 7.13 we wish to find the elevator angle δ that trims the airplane. In
the absence of a trim tab, this angle is

CMo (h − ho)CL 1 w
−δtrim = + − CLαt α − + CLαt ηt , . (7.19)
Vt CLδ Vt CLδ CLδ U
For a given weight and air speed, δtrim depends linearly from the position of the CG.
Thus, the solution will have the form
δtrim = δo + CL f (h), (7.20)
with

CMo 1 w
δo = − CLαt α − + CLαt ηt , (7.21)
Vt CLδ CLδ U
h − ho
f (h) = . (7.22)
Vt CLδ
The value δ = δo is the elevator angle when the longitudinal position of the CG is
coincident with the aerodynamic centre of the wing. When h − ho > 0 (as shown
in Figure 7.1), the function f (h) is positive and the trim angle increases as the CG
moves aft. If the landing gears are deployed, the increase in pitching moment causes
an increase in control deflection δtrim .
184 Airplane Trim
The trim moment must be provided by the control system. If the control is
manual, the pilot would have to maintain the stick fixed at the position required for
trim. Alternatively, the elevator deflection can be set to zero if the tail is trimmable:
the horizontal tail can be rotated so that the angle ηt satisfies Equation 7.20 with
δtrim = 0:

1 CMo (h − ho)CL w
ηt = + − CLαt α − . (7.23)
CLαt Vt Vt U
The longitudinal trim can also be obtained with a fixed stabiliser and a δ = 0, provided
that the CG position satisfies the following condition from Equation 7.19:

CLαt w
h = ho − CMo + α− + ηt . (7.24)
CL U
7.1.3 Stick-Free Longitudinal Trim

Because the condition CM = 0 alone generates a hinge moment due to elevator
deflection, a more restrictive trim condition arises from forcing also the hinge
moment to be zero, that is: CH = 0. In this case, the trim is stick-free (hands-off ).
The hinge moment is a linear combination of the inflow angle of the tail-plane, the
elevator and the trim tab angles:

dCH dCH dCH
CH = αt + δ+ β. (7.25)
dαt dδ dβ
To simplify the equations, the derivatives in Equation 7.25 are replaced by the
following symbols:
dCH dCH dCH
b1 = , b2 = , b3 = . (7.26)
dαt dδ dβ
Thus, the hinge moment can be rewritten as

w
CH = b1 α − + b1 ηt + b2 δ + b3 β. (7.27)
U
The elevator deflection required to produce a zero hinge moment is:

b1 w b1 b3
−δ = α− + ηt + β. (7.28)
b2 U b2 b2
Equation 7.28 shows that the elevator is geared to the tab via the gear ratio −b3 /b2 . If
we insert the elevator angle from Equation 7.28 into Equation 7.10, we have a new
pitching moment
CM = CMo + (h − ho)CL − Vt H, (7.29)
with

b1 w b1 b3
H = CLαt − CLδ α− + CLαt − CLδ ηt + CLβ − CLδ β.
b2 U b2 b2
(7.30)
12 1
Elevator deflection, ηtrim, degs

10 0.6 0
CL = 0.4
CD trim, drag counts
8 -1 0.5
0.5
0.6
6 -2
CL = 0.4
4 -3
10 15 20 25 30 35 40 10 15 20 25 30 35 40
XCG %MAC XCG %MAC
(a) Trimmable H-stab (b) Untrimmable H-stab
Figure 7.2. Longitudinal trim of a model Airbus A320-200; trim drag and elevator deflection
at cruise configuration, M = 0.78, h = 10,000 m (∼32,800 feet); standard day.
To simplify this equation, assume

b1 b3
a1 = CLαt − CLδ , a2 = CLβ − CLδ . (7.31)
b2 b2
Russell 2 proves that these coefficients are the lift-curve slopes of the tail-plane.
Equation 7.29 is thus reduced to

w
CM = CMo + (h − ho)CL − Vt a1 α − + a1 ηt + a2 β = 0. (7.32)
U
Equation 7.32 is the stick-free pitching-moment equation. The hinge moment is zero,
but to trim the airplane, a tab angle must be found. This is done by solving the
equation in terms of the tab deflection β:

CMo CL a1 w a1
β=− − (h − ho) + Vt α− + ηt . (7.33)
a2 a2 a2 U a2
The solution depends on the lift coefficient and on the position of the CG. It can be
verified that, in general, there is only a restricted range of parameters that ensures
longitudinal trim.
NUMERICAL RESULTS FOR THE LONGITUDINAL TRIM. We consider again the case of
the model Airbus A320-200. The calculated trim drag is shown in Figure 7.2a. The
condition of trimmable stabiliser has been used, which leads to δtrim identically zero.
For a fixed value of the CL (or weight), the trim drag decreases as the CG moves aft;
for a given position of the CG, the trim drag increases with the CL (or weight).
Figure 7.2b shows the elevator’s angle required for longitudinal trim at cruise
conditions. In this case it was assumed that the horizontal stabiliser is not trimmable
(which is not the case for this type of airplane). The calculation is done iteratively
because the downwash from the wing (Equation 7.5) depends on the wing lift; the
effective angle of attack of the tail-plane depends on the split between Lw and Lt .
To avoid using relatively large elevator angles, it is necessary to maintain the CG
186 Airplane Trim
sufficiently close to the aerodynamic centre of the wing. During cruise, the weight
changes due to fuel burn, and so does the CG position. A decrease in weight requires
a movement of the xcg aft. One way to achieve this is to burn fuel from forward tanks
or, more generally, to manage the variation of the xcg by pumping fuel from aft tanks
to forward tanks. Only a flight-management computer can do so reliably.
7.2 Airplane Control under Thrust Asymmetry

If the airplane suffers an engine failure such that one engine must be shut down, the
airplane will tend to yaw toward the inoperative engine and to bank on one side.
When the operating engine’s CG is below the airplane’s CG, the airplane banks
down on the side of the operating engine. The amount of yaw depends on the engine
thrust, the moment arm of the thrust and the directional stability of the airplane 3 .
By reducing the air speed, the thrust required decreases and the control settings
become less restrictive.
An asymmetry situation may also arise in cases when the CG is off the vertical
symmetry of the airplane, with all engines functioning. This situation is not unusual
and includes cargo displacements, passenger location and fuel asymmetry. In fact,
most commercial airplanes are certified with lateral CG limits or with a maximum
load asymmetry. For example, the A300-600 was cleared to operate with a maximum
fuel asymmetry of 2,000 kg.
If the airplane is a turboprop, the drag created by the inoperative propeller
will contribute further to the yaw. With the airplane yawing toward the inoperative
engine, the asymmetry of the inflow increases the drag of the “advancing” wing; at
the same time, the lift decreases. The opposite effect takes place on the “retreating”
wing. As a result, the airplane rolls and the advancing wing banks down. This effect
can be compounded by a CG offset and by the drag of the propeller. Airplanes must
be able to overcome this potential loss of control by combined use of directional
stability, aileron and rudder deflections. In summary, a conventional airplane is
subject to the following events:
r destabilising effect of the working engine
r destabilising effect of the CG offset (due to banking or otherwise)
r stabilising effect of the side force on the fuselage and vertical tail
r stabilising effect of the side force on the rudder
For a turboprop airplane, other factors that may have to be included are 1) the
propeller’s torque, which tends to roll the airplane in the opposite direction of
the rotation; 2) the drag of the inoperative propeller (minimised by feathering the
blades); 3) the asymmetric blade effect, arising when the axis of rotation of the
propeller disk is not aligned with the inflow; and 4) the effect of the propeller’s
slipstream 4 . The first case may not be a concern if the torque-induced roll tends to
lift the wing with the inoperative engine.
7.2.1 Dihedral Effect

A yaw/side-slip generates a rolling moment. Consider a wing with dihedral ϕ oper-
ating with a small side-slip β. The wing has a higher angle of attack on the forward
Leeward Windward
β β
φ φ
Figure 7.3. Sketch of wing flow with side-slip.
side than on the rearward side. If the local chord is c, the fluid particles travel inward
or outward by an amount cϕ. These fluid particles travel toward a lower point on the
forward side and toward a higher point on the rearward side. The amount by which
the air is displaced depends on the geometry of the wing, as shown in the sketch in
Figure 7.3.
If the trailing edge is a straight line, then the spanwise offset does not depend
on the side of the airplane and is equal to cϕβ. Because the fluid particles have
fluctuated by an amount cϕβ over the chord c, the change in inflow conditions is
cϕβ
α = ± = ±ϕβ. (7.34)
c
This condition tends to generate different values of the lift on the forward and
rearward half-wing; hence a rolling moment. Furthermore, there will be a change
in wing drag that would tend to reduce the yaw. (This situation arises, for example,
when the airplane enters a downburst, as described in § 13.7.) The additional restoring
moments would be

bd bd bd
Cl = CL = −CLα α = −2CLα φβ , (7.35)
b b b

bd 2 bd
CN = CD = −4k CLα φβ , (7.36)
b b
where bd is the moment arm of the wing drag (an unknown quantity). If these effects
are not already included in the aerodynamic derivatives Clβ and CNβ , they should be
added to the equations for the rolling moment and the yaw moment, respectively.
PROBLEM FORMULATION. The following analysis deals with jet-powered airplanes.

For reference, consider a twin-engine airplane. The control of the airplane is achieved
by a combined rudder and aileron deflection, Figure 7.4. The fuselage in yaw has
an effect on the derivative of the pitching moment with respect to the lift: dCN /dCL.
There is little or no effect on the CL and the CLα , as demonstrated in Ref. 5 , and an
increase in drag, due to flow separation on the lee side. Assume that the response
of the airplane is linear; in other words, it can be calculated as a linear combination
188 Airplane Trim
Wing down
Aileron Down
β BL separation
Shut−down Aileron Up
V−Stab + Rudder Lift
Wing up
(a) Top view
Shut down
(b) Front view

Figure 7.4. Airplane configuration with asymmetric thrust.
of the rudder and aileron deflections, for any reasonable value of the side-slip and
bank angle. With these hypotheses, the static side-force coefficient is
W sin φ
CY = CYβ β + CYξ ξ + CYθ θ + = 0, (7.37)
ρ AU 2 /2
where
Y
CY = (7.38)
ρ AU 2 /2
is the side-force coefficient, β is the side-slip angle, φ is the roll angle, ξ is the aileron
deflection (assumed symmetrical) and θ is the rudder-deflection angle; CYβ , CYξ and
CYθ are the derivatives of the side-force coefficient with respect to the yaw angle β,
aileron angle ξ and rudder angle θ , respectively. The last term in Equation 7.37 is
the side-force coefficient created by the offset of the CG. It can also be written as
CL sin φ. The roll-moment coefficient required to stabilise the flight is
Cl = Clβ β + Clξ ξ + Clθ θ = 0. (7.39)

The linearised static yawing moment coefficient is

Tbt bt
CN = CNβ β + CNξ ξ + CNθ θ + + CDe = 0, (7.40)
ρ AU 2 b/2 b
where T is the thrust asymmetry, bt is the moment arm of the thrust asymmetry
and CDe is the idle engine’s drag coefficient. In the OEI case, assume that T is
the thrust of the functioning engine. Because Equations 7.37 to 7.40 reflect static
conditions for the airplane, they represent a limit case.
Because the thrust depends on the air speed, the problem has to be solved
iteratively. Furthermore, if the airplane has to maintain this air speed, the component
of the thrust along the vector velocity must balance the resulting drag. The latter
quantity cannot be derived from cruise conditions because the combination of yaw
and deployment of control surfaces contributes to an increase in the aerodynamic
resistance.
If we use the strip theory with the wing aerodynamics, the yaw does not cause
any difference in profile drag. There can be differences in induced drag due to
the difference in lift force. As far as the fuselage is concerned, there will be a
roughly parabolic behaviour of the drag with respect to yaw angles for −10 < β < 10
degrees 6 .
From Equation 7.37, the side-slip angle can be written in terms of the remaining
parameters as follows:

1 2W sin φ
β=− CYξ ξ + CYθ θ + . (7.41)
CYβ ρ AU 2
The side-slip angle increases with the weight and decreases with the increasing speed.
However, the value of the aerodynamic coefficients is important in limiting the yaw
effects. If the inoperative engine is at the starboard side, the airplane will veer to
the starboard. The rudder will have to deflect in the opposite direction to create
a restoring moment. The convention is that β < 0 and θ > 0. The bank angle φ
depends on the position of the thrust with respect to the centre of gravity of the
airplane.
NUMERICAL SOLUTION. Given the airplane data (weight, wing area, flight altitude)
and the stability coefficients, we need to calculate the minimum speed that guarantees
full control (longitudinal, lateral and directional). This is the minimum control air
speed, VMCA.
Calculation of the control derivatives is quite elaborate, but there are computer
programs designed for this purpose, for example, the U.S. Air Force Stability and
Control Digital Datcom 7 , ESDU 8 and a number of more recent vortex lattice aero-
dynamic programs. These coefficients are sensitive to the airplane configuration and
may change considerably from one airplane to another. Some stability derivatives for
the Boeing B747-100, Lockheed C-5A, Grumman F-104A and other representative
airplanes are available in Heffley and Jewell 9 .
For a commercial jet liner, FAR § 25.149 requires that control is assured in the
worst possible scenario, for example, with one engine at the take-off thrust, the
airplane at MTOW, the CG in the aft position and the flaps at take-off position and
the landing gear retracted. A maximum bank angle of 5 degrees is allowed. The
190 Airplane Trim
Table 7.1. Stability derivatives for calculation of airplane

response to asymmetric thrust; model Boeing B747-100
Clβ −0.2210 CNβ 0.1500 CYβ −0.9600

Clξ 0.0460 CNξ 0.0064 CYξ 0.0000
Clθ −0.0070 CNθ 0.1090 CYθ −0.1750
unknown parameters in Equations 7.37–7.40 are the minimum control speed U =

VMCA, the side-slip angle β, the aileron deflection ξ and the rudder deflection θ.
For a fixed rudder deflection θ = θmax , the solution system is written as
⎡ ⎤⎡ ⎤ ⎡ ⎤
Clβ Clξ 0 β Clθ θ
⎣ CNβ CNξ 2Tbt /ρ Ab ⎦ ⎣ ξ ⎦ = − ⎣ CNθ θ + CDe bt /b ⎦ , (7.42)
CYβ CYξ 2W sin φ/ρ A 1/U 2 CYθ θ
where the unknowns are β, ξ , 1/U 2 . For a fixed aileron deflection ξ = ξmax , the
solution system is
⎡ ⎤⎡ ⎤ ⎡ ⎤
Clβ Clθ 0 β Clξ ξ
⎣ CNβ CNθ 2Tbt /ρ Ab ⎦ ⎣ θ ⎦ = − ⎣ CNξ ξ + CDe bt /b ⎦ . (7.43)
CYβ CYθ 2W sin φ/ρ A 1/U 2
CYξ ξ
The value of CDe is calculated with the method described by Torenbeek 10 :

2 U U
CDe = 0.0785d2 + Aj 1 − , (7.44)
1 + 0.16M2 U j Uj
where U j is the velocity of the jet at the nozzle, d is the fan diameter and Aj is the
area of the nozzle. Typical values of U/U j are 0.92 (high by-pass turbofan), 0.42 (low
by-pass jet engine) and 0.25 (straight turbojet and turboprop). Equation 7.44 shows
that the CDe depends on the Mach number.
To maintain the linearity of the system, we can calculate the CDe a posteriori,
or iteratively. For example, guess the VMCA, calculate the Mach number, calculate
the CDe from Equation 7.44 and solve the system again. The solution of the system
requires the inversion of the matrix on the left-hand side. For certain values of the
aerodynamic derivatives, the solving system has no solution; hence it would not be
possible to control the airplane by any combination of aileron–rudder deflections.
This situation must clearly be avoided. From a computational point of view, the
non-existence of a solution (when a solution is known to exist) is an indication that
the derivatives are most likely incorrect.
A calculation is shown for the Boeing B747-100, whose aerodynamic coefficients
are known from Nelson 11 and summarised in Table 7.1. Figure 7.5 shows the side-slip
angle of the airplane for varying aileron deflection at a fixed rudder deflection, at
a fixed weight and air speed, as indicated in the graph. The side-slip is linear and
weakly affected by the rudder deflection.
Figure 7.6 shows the calculated VMC as a function of the airplane’s AUW. It
shows the VMC limited by the maximum rudder and aileron deflection. Both speeds
are above the stall speed, calculated with a CL = 1.8. With an AUW of 180 tons or
less, it is not possible to ensure full control of the airplane. In some cases the rudder
-2
θ = 15 degs
θ = 7.5 degs
Side slip angle, degs -4
-6
-8
0 5 10 15 20
Aileron angle, degs
Figure 7.5. Side-slip angle versus aileron deflection at maximum rudder for the Boeing B747-
100. The flight conditions are sea-level, standard day, MTOW, OEI, M = 0.53, φ = −5 degs.
deflection has an opposite trend to the one shown, and a branched solution can be
found.
CALCULATION OF AERODYNAMIC DERIVATIVES. The calculation of these quantities is

not trivial. If we have FDR data, with the airplane parameters U, W, φ, T, β, ξ ,
θ known quantities, then the system Equation 7.42 could be used to determine
250
Min control air speed, kt
200
150
Stall
100
Max aileron deflection

Max rudder deflection
Stall speed
50
150 200 250 300 350
Gross weight, ton
Figure 7.6. Calculated VMC versus gross weight for the Boeing B747-100. Bank angle φ = −5
degrees.
192 Airplane Trim
indirectly one set of derivatives if the other two sets are known. For example, we
can calculate (Cl , CN , CY )β if both (Cl , CN , CY )θ and (Cl , CN , CY )ξ are known. If the
derivatives are incorrect, the solution of Equation 7.42 or Equation 7.43 is either
impossible or unrealistic (e.g., large or negative values of the speed; values of the
rudder and aileron deflection beyond the limits).
Summary
We presented static stability conditions in the longitudinal and lateral directions.
In the former case, we have demonstrated that the position of the CG is crucial. If
the CG falls outside a limited range, it might not be possible to control the airplane.
Furthermore, the CG position governs the trim drag, which can cause additional fuel
consumption. All control strategies must minimise the trim drag. The lateral control
deals with the case of asymmetric thrust. Again, it is possible to control the airplane
within a limited set of operational parameters, including Mach number, altitude,
aileron and rudder deflections, the side-slip, and the bank angles.
Bibliography
[1] Filippone A. Encyclopaedia of Aerospace Engineering, volume 5, chapter 252.
John Wiley, 2010.
[2] Russell JB. Performance and Stability of Aircraft. Butterworth-Heinemann,
2003.
[3] ESDU. Loading on a Rigid Aeroplane in Steady Lateral Manoeuvres. Data
Item 01010. ESDU International, London, Oct. 2001.
[4] ESDU. The Influence of Propeller Slipstream on Aircraft Rolling Moment due
to Sideslip. Data Item 06012. ESDU International, London, Aug. 2006.
[5] Salmi R and Conner W. Effects of a fuselage on the aerodynamic characteristics
of a 42-degrees sweptback wing at Reynolds numbers up to 8,000,000. Technical
Report RM-L7E13, NACA, 1947.
[6] HH Page. Wind tunnel investigation of fuselage stability in yaw with various
arrangements of fins. Technical Report TN-785, NACA, Nov. 1940.
[7] Williams JE and Vukelich SP. The USAF stability and control digital DAT-
COM. Technical Report AFFDL-TR-79-3032, Vol. I, Air Force Flight Direct-
orate Laboratory, April 1979.
[8] ESDU. Computer program for prediction of aircraft lateral stability derivatives
in sideslip at subsonic speeds. Data Item 00025. ESDU International, London,
Oct. 2000.
[9] Heffley RK and Jewell WF. Aircraft handling qualities data. Technical Report
CR-2144, NASA, 1972.
1985. Appendix G-8.
[11] Nelson RC. Flight Stability and Automatic Control. McGraw-Hill, 2nd edition,
1998.

a1 , a2 = parameters defined in Equation 7.31
A = wing area
At = tail-plane area
Aj = nozzle area
AR = wing-aspect ratio
b = nominal wing span
bi = derivatives of the hinge moment, i = 1, 2, 3
bd = moment arm of wing drag force
bt = thrust arm
c = local wing chord
c = mean aerodynamic chord
Cl , CN , CY = roll/pitch/yaw moment coefficients
Clβ , CNβ , CYβ = roll/pitch/yaw moment derivatives with respect to side-slip
Clθ , CNθ , CYθ = roll/pitch/yaw moment derivatives with respect to rudder
deflection
Clξ , CNξ , CYξ = roll/pitch/yaw moment derivatives with respect to aileron
deflection
CDe = drag of inoperative engine
CDtrim = trim-drag coefficient
CH = hinge moment, Equation 7.25
CL = wing or total-lift coefficient
CLt = horizontal-tail lift coefficient referred to h-tail area
A
CLt = horizontal-tail lift coefficient referred to wing area
CLαt = lift-curve slope of the horizontal tail-plane
CLη = elevator’s lift-curve slope
CLδ = lift-curve slope of the elevator
CM = pitching-moment coefficient
CMo = pitching-moment coefficient (reference, or zero-lift)
CMlg = pitching-moment coefficient, landing gear
CMot = pitching-moment coefficient, tail-plane
CMow = pitching-moment coefficient, wing
CN = static yawing-moment coefficient, Equation 7.40
d = fan diameter; diameter
f (h) = function of the CG position, defined in Equation 7.22
L = total lift
H = function defined in Equation 7.30
h = position of the aerodynamic centre, fraction of the MAC
(non-dimensional)
ho = position of the centre of gravity, fraction of the MAC
(non-dimensional)
k = wing’s induced-drag coefficient
kt = tail-plane’s induced-drag coefficient
Lt = tail-plane lift
Lw = wing lift
M = Mach number
M = pitching moment
Mot = pitching moment of the horizontal tail
Mo = pitching moment of the wing
T = net thrust
q = dynamic pressure, q = ρU 2 /2
194 Airplane Trim
U = air speed
Uj = jet velocity
Vt = tail-volume coefficient, Equation 7.11
w = average downwash
W = airplane weight
xac = distance of wing’s aerodynamic centre from leading edge
(longitudinal)
xw = distance of wing’s aerodynamic centre from nose (longitudinal)
xt = distance between tail-plane’s aerodynamic centre and nose
(longitudinal)
xwt = distance between wing’s and tail’s aerodynamic centre
(longitudinal)
xcg = distance of centre of gravity from nose (longitudinal)
Y = side force
z = vertical coordinate
Greek Symbols
αe = effective angle of attack
αt = angle of attack, tail surface
β = tab deflection angle; side-slip angle
δ = elevator angle
δo = elevator angle when CG coincides with AC, Equation 7.21
δtrim = elevator angle that trims the airplane
η = elevator’s angle
ηt = angle between the neutral lines of tail and wing
θ = rudder deflection
φ = bank angle
ϕ = dihedral angle
ρ = air density
ξ = aileron deflection
[.]cg = centre of gravity
[.] j = jat quantity
[.]o = reference conditions
[.] M = main landing gear
[.] N = nose landing gear
[.]lg = landing gear
[.]t = tail
[.]w = wing
[.]β = derivative with respect to side-slip angle
[.]θ = derivative with respect to rudder deflection
[.]ξ = derivative with respect to aileron deflection
8 Flight Envelopes
Overview
This chapter deals with flight envelopes in the speed-altitude space. We present
various atmospheric models (§ 8.1), standard as well as non-standard. We give several
operating air speed definitions (§ 8.2), as well as design speeds (§ 8.3) and the
techniques required to measure them. For the steady-state level flight we derive
two optimal conditions: minimum drag and minimum power (§ 8.4). We discuss the
flight corridors at constant altitude and the ceiling performance airplanes (§ 8.5).
We discuss the flight envelopes of subsonic transport aircraft and their limitations
(§ 8.6), including the effects of the cabin pressure. We conclude the chapter with
flight envelopes at supersonic Mach number (§ 8.7), including dashspeed, supersonic
accelerations and propulsion limitations.
KEY CONCEPTS: International Standard Atmosphere, Atmosphere Models, Operat-

ing Speeds (EAS, CAS, TAS), Transition Altitude, Design Speeds, Optimum Level
Speeds, Ceiling Performance, Cabin Pressure, Flight Envelopes, Supersonic Dash,
Supersonic Acceleration.
8.1 The Atmosphere

Most performance calculations are done with a conventional atmosphere that has
been declared standard (§ 8.1.1), although the almost totality of flights take place
in far more complex atmospheres, with variation in the thermodynamic properties
in both the vertical and horizontal directions. To overcome the limitations of the
standard model, alternative atmosphere models are available to take into account
extreme temperatures (very hot and very cold), as explained in § 8.1.2. The effect
of air humidity is generally neglected in aerodynamics and engine performance,
although it is important for other aspects, such as noise propagation.
8.1.1 International Standard Atmosphere

Nearly all of the basic calculations of aircraft performance are done in International
Standard Atmosphere (ISA) conditions (standard day), whose parameters at sea
level are given in Table 8.1.
195
196 Flight Envelopes
Table 8.1. Sea-level data of the International

Standard Atmosphere
Parameter Symbol Sea-Level Value
Temperature To 15.15 ℃
Pressure po 1.01325 · 105 Pa
Density ρo 1.225 kg/m3
Viscosity μo 1.7894 · 10−5 N s/m2
Humidity H 0%
Observations on the state of the atmosphere at sea level go back hundreds of

years, but they have become systematic in the last century with aviation, rocket and
satellite data and perfect gas theory. A number of standard versions exist: NACA’s
atmosphere 1 , the ARDC 2 , the U.S. standard 3 (1962, amended in 1976) and the
ICAO standard 4 . These tables are basically equivalent to each other up to about 20
km (∼65,600 feet), which is the altitude of interest for most airplanes.
The atmosphere is divided into a number of layers. The atmosphere below
11,000 m (36,089 feet) is called Troposphere. It is characterised by a decreasing
temperature from sea level and reaches a standard value of −56.2 ℃. The altitude
of 11,000 m is called tropopause. The level above is called Lower Stratosphere and
covers an altitude up to 20,000 m (65,627 feet), in which the temperature remains
constant. The air density keeps decreasing with the increasing altitude. The upper
limit of this layer includes most of the atmospheric flight vehicles powered by
air-breathing engines. The Middle Stratosphere reaches up to an altitude h = 32,000
m (104,987 feet). In this layer the atmospheric temperature increases almost linearly
from the value of −56.2 ℃. The edge of space is generally considered to be at
an altitude ∼100 km. This altitude is somewhat arbitrary. At the edge of space, a
vehicle flying on aerodynamic lift alone would have to maintain a velocity larger
than the orbinal velocity. Therefore, the problem is to find the altitude at which
these velocities are the same.
A number of functions are used to approximate the ICAO data. In the tropo-
sphere a linear expression is used:
T = To + λh, (8.1)
where h is the altitude in metres and λ = −0.0065 K/m is the temperature lapse
rate. Interestingly, such a temperature gradient causes the atmosphere to be stable
to vertical movements. The limit gradient for vertical stability, corresponding to an
adiabatic atmosphere, is about −9 K/km, as demonstrated by Prandtl and Tietjens 5 .
The atmosphere is well described by the equation of the ideal gas:
p
= RT , (8.2)
ρ
where R = 287 J/kg K is the gas constant (air). Equation 8.2 written at two different
states leads to the equivalence
p ρ T
= . (8.3)
po ρo To
The relative density is called σ , the relative pressure is δ and the relative temperature
is θ . We call the altitude corresponding to a given air density density altitude. If,
instead, we relate the altitude to the local air pressure, then we have a pressure
altitude. To find the pressure-altitude and the density-altitude relationships, we use
the buoyancy law for still air along with Equation 8.1, to find the rate of change of
the pressure with altitude. The buoyancy law is
∂p
= −ρg. (8.4)
∂h
If we insert the differential form of Equation 8.1, the buoyancy law becomes
∂p ρg
=− . (8.5)
∂T λ
Finally, use Equation 8.2 to eliminate the density, rearrange the equation and integ-
rate from the sea-level altitude. The result is

p g T
ln = ln , (8.6)
po λR To
δ = θ (g/λR) . (8.7)
The value of the power coefficient is g/λR = 5.25864. If we insert Equation 8.1 in
Equation 8.7, we have a pressure-altitude correlation:

0.0065 5.25864 5.25864
δ = 1− h = 1 − 2.2558 · 10−5 h , (8.8)
To
with h expressed in [m]. Equation 8.8 is in good agreement with the ICAO data. A
more accurate expression is
δ = (1 − 2.2558 · 10−5 h)5.25588 . (8.9)
In Equation 8.9 the altitude is related to the pressure ratio, therefore it can be read
directly from the altimeter that is calibrated with the ISA reference value of po. To
find a density-altitude correlation in the lower atmosphere, we use the equation of
state; the relative pressure is taken from Equation 8.9. The result is
(1 − 2.25577 · 10−5 h)5.25588

σ = . (8.10)
1 − 0.0065 h/To
Figure 8.1 shows the ratios of density, pressure, temperature and speed of sound
from sea level to the altitude of 25,000 m. If γ denotes the ratio between specific
heats, the speed of sound in the atmosphere is calculated from

a = γ RT . (8.11)
The non-dimensional form of Equation 8.3 is δ = σ θ. The quantities can be

approximated by exponential functions in the troposphere. For a standard day,
these functions are
√
σ θ 4.25 , δ θ 5.25 , δ θ θ 4.75 . (8.12)
25
80
σ
θ
δ
20
a/ao
60
Altitude, 10 feet
Altitude, km 15
3
Tropopause 40
10
20
5
0 0
0 0.2 0.4 0.6 0.8 1
Relative value
Figure 8.1. ISA relative parameters as a function of altitude.
The relative temperature is readily calculated from Equation 8.1. The inverse prob-
lem (calculation of the altitude h corresponding to relative density σ ) is more elabor-
ate because it requires to solve a non-linear equation in implicit form. The solution
can be found with a bisection method. For the bisection method to work, one has
to choose two points at which the function has opposite values. It is safe to choose
σ1 = 0.01 and σ2 = 1, to make sure that the method converges to a solution.
Finally, the value of the air viscosity from the air temperature can be found from
Sutherland’s law
3/2
T To + C
μ = μo , (8.13)
To T +C
where C = 111 K is the Sutherland constant for air; μo is the viscosity at the reference
temperature (273.15 K). Note that the dynamic viscosity can be expressed as kg/ms
or N s/m2 . The two units are equivalent.
Other physical quantities that are sometimes used in performance calculations
include the thermal conductivity. The conductivity of dry air is
κd 0.023807 + 7.1128 · 10−5 (T − To) [W/mK]. (8.14)
The conductivity of moist air is

κv nv
κa κd 1 − 1.17 − 1.02 [W/mK], (8.15)
κd nv + nd
where nv and nd are the number of moles in the vapour and dry phases, respectively.
8.1.2 Other Atmosphere Models

The ISA model is useful to compare aircraft performance over all the range of atmo-
spheric altitudes. This is an idealised case that does not occur anywhere. Figure 8.2
Temperature, C
-60 -40 -20 0 20
15,000
NOTTINGHAM, 06 AUG 2008
RH
Temp
10,000
Altitude, m
5,000
0
0 20 40 60 80 100
Relative Humidity, %
Figure 8.2. Temperature and relatively humidity versus altitude detected by radiosonde above
central England.
shows a stratified atmosphere above central England in the dog-days of summer, as

reported by a radiosonde probe. There is a temperature inversion just above 1,000 m.
The distribution of humidity does not follow a recognised pattern. Clearly, these data
cannot be associated to a standard atmosphere. Therefore, some alternative models
are needed. The relative humidity is not used for flight-mechanics calculations but
it important in the propagation of acoustic sources (§ 17.2).
We recognise that a realistic analysis of aircraft performance requires considera-
tion of large deviations from the standard values to deal with extreme environmental
conditions: winters in the northern hemisphere, very hot weather on the ground. In
addition, airport altitude, humidity and precipitations, atmospheric winds, lateral
gusts and global air circulation have strong influence on flight and safety. Rain and
snow can be so heavy that take-off may have to be aborted. To simplify these matters,
the U.S. Department of Defense defines four non-standard atmospheres, referred
to as hot, cold, tropical and arctic (MIL-STD-210A). These profiles are shown in
Figure 8.3.
Three important classes of weather-related flight problems are icing (§ 14.1),
downbursts (§ 13.7) and atmospheric turbulence. Turbulence is a more familiar
weather pattern to the frequent flyer. It includes cases of free air and convective
air turbulence, atmospheric boundary layers and mountain ridge waves. No airplane
is immune to the powerful gusts of the atmosphere. The aviation industry refers
more often to wind shear rather than gust. Wind shear can occur in any direction,
although it is classified as vertical and horizontal wind shear. A vertical wind shear
causes turbulence that affects climb and descent; the amplitude of the wind speed
variation can reach 30 kt per 1,000 feet (305 m). A horizontal wind shear can cause
a shift from a head- to a tail-wind, and vice versa, with an amplitude up to 100 kt per
14
ISA 40
12
Tropopause
10
Altitude, 10 3 feet
30
Altitude, km
8
Te
Ho
20
m
6
Min
t/D
pe
e
r
im
se
at
Ar
rt
um
ct
Cl
Cl
ic
4
im
im
Arc
at
at
10
e
tic
e
2
1
0 0
-80 -60 -40 -20 0 20 40
o
Temperature, C
Figure 8.3. ICAO standard atmospheric temperature and reference atmospheres for flight-
mechanics calculations.
n-mile. Such a situation can arise in a downburst. A detailed discussion of turbulence

and flight is available in Etkin 6 and Houbolt 7 . An introduction to weather processes
and climatic conditions was published by Barry and Chorley 8 . A comprehensive
reference on aviation meteorology is published by HMSO 9 .
Among the global circulation effects there is the jet stream. This is an atmo-
spheric wind with an East-West prevalence, strongest around the troposphere, that
affects trans-Atlantic flights between Europe and North America. These streams
can be thousands of kilometers wide and affect the weather on a continental scale.
Two typical cases occur. First, in some regions of the jet, air converges to the stream.
As the jet accumulates air it becomes heavier and increases the air pressure at lower
altitudes down to sea level. Another effect of compression is due to downward
movement of the air, which in turns causes an increase in temperature and pre-
vents the formation of clouds. This case is characterised by high pressure and clear
skies.
The second case is when air diverges from the jet. The events are generally
opposite than the case described herein. There is a loss of atmospheric pressure as
the atmosphere above the jet becomes lighter. The reduction in pressure extends to
sea level and causes updrafts and faster evaporation. Rising air cools down, and the
humidity contained in the air freezes to form clouds. This event is also on a very
large scale (thousands of kilometers wide).
Figure 8.4 shows a snapshot of the North Atlantic jet stream. The shaded areas
show wind speeds in excess of 50 kt in the N-E direction. Particularly strong are
the winds above the British Islands, that eventually turn South over France. The
map refers to an altitude corresponding to a pressure of 300 mbar, or about 9,750 m
(∼32,000 feet). This is generally the altitude at which the winds are strongest. GFS
Figure 8.4. Jet stream over the North Atlantic [adapted from the U.S. National Center for
Environmental Protection].
is the Global Forecast System; this is a computer simulation of the weather made
available to the public over the Internet and other channels.
Overall, temperature variations in the atmosphere can be of the order of 80
degrees (−40 ◦ C to +40 ◦ C). It is not uncommon to encounter temperature inver-
sions, that is, cooler air at the ground level and warmer air at low altitudes, contrary
to the standard model. At the tropopause the temperature can fall below −70 ◦ C in
winter time at moderate to high latitudes.
Because the temperature does not appear explicitly in any of the performance
equations, a useful relationship with the pressure and density is required. If the
temperature has a constant deviation from the standard value, say a constant ±T ,
the method of § 8.1.1 can still be used because the temperature gradient is the
same (Equation 8.1). The only difference is that the symbol To denotes the sea-level
temperature, whatever that may be. A simple model for non-standard atmosphere
consists in adding or subtracting a constant value from the standard temperature
profile. Thus, the temperature distribution Equation 8.1 is replaced by
T = T I SA + dT , (8.16)
where dT is the temperature shift; a cold day is obtained with dT < 0 and a hot day
is dT > 0. The atmospheric pressure in the troposphere, at the tropopause and in
h - hp, km
-1 0 1
30
8
o
Geometric altitude h, 103 feet

-20 C ISA o
20 C
-30 oC -10 oC -10 oC o
30 C
Geometric altitude, km
6
20
10
0 0
-4 -2 0 2 4
3
h - hp,10 feet
Figure 8.5. Relationship between geometrical altitude and pressure altitude.
the stratosphere is, respectively:

−g/λR
T − T
p = po , (8.17)
To
−g/λR
Tt − T
pt = po , (8.18)
To

g
p = pt exp − (h − ht ) . (8.19)
RT I SA,t
In Equations 8.18 and 8.19 the subscript “t” denotes conditions at the tropopause.
These equations lead to a definition of two important concepts: the pressure altitude
and the density altitude. The flight altitude is not measured directly; the pressure
and the temperature are measured; thus, only a pressure altitude would be available
from Equations 8.17 through 8.19.
In this model, the temperature depends only on the altitude. Therefore, it is
possible to solve the pressure equation for a known pressure and unknown altitude.
If the temperature is below the standard value, the pressure altitude is lower than
the geopotential altitude; vice versa, if the temperature is above the standard value,
the pressure altitude is higher than the geopotential altitude.
The correlation between pressure altitude and geopotential altitude is shown in
Figure 8.5. This figure shows the difference between the geometrical altitude and
the pressure altitude at increasing geometrical altitude for specified temperature
shifts.
8.2 Operating Speeds

The ground speed Vg is the aircraft speed measured with respect to a fixed point
on the ground. The air speed U is the aircraft speed relative to the air and accounts
for the presence of atmospheric winds ±Vw . If an atmospheric wind is aligned with
the aircraft speed, the air speed is defined as
U = Vg ± Vw , (8.20)
where Vw is the wind speed (negative for a tail wind, positive for a head wind).
Measuring the air speed is of considerable importance. For example, an error in
defect may cause the aircraft to run into a stall condition. At incompressible speeds,
the air speed can be evaluated with the Bernoulli equation,
1
p + ρU 2 = p∗ , (8.21)
2
that gives the true air speed (TAS)

p∗ − p p 1
U = TAS = = √ . (8.22)
2ρ 2ρo σ
In Equation 8.22 p is the free-stream atmospheric pressure and p∗ is the stagnation

pressure and ρ is the density; the subscript “o” still denotes sea-level standard
conditions. An instrument that measures a difference in pressure between the free-
stream conditions and the static conditions is the Pitot probe. With the Bernoulli
equation the probe converts a pressure difference into air speed, as long as the
air density is known. Therefore, the instrument would not work without additional
readings. By taking a temperature reading and a static pressure reading at the same
time, we can find the value of the density ρ from the equation of ideal gases:
1 p
ρ= . (8.23)
R T static
If we refer to atmospheric conditions to sea level, we have an equivalent air speed:

p
EAS = . (8.24)
2ρo
The relationship between TAS and EAS is derived from Equation 8.22:
√
EAS = TAS σ . (8.25)
An Air Speed Indicator (ASI) based on this principle is of no use in high-speed

flight, where compressibility is important. From compressible aerodynamic theory
we know that the ratio between the stagnation pressure p∗ and the static atmospheric
pressure p corresponding to an isentropic deceleration from a Mach number M is
γ −1/γ
p∗ γ −1 2
=1+ M, (8.26)
p 2
6
Dynamic pressure
Impact pressure
5
Normalised pressure
4
0
0 0.5 1 1.5 2
Mach number
Figure 8.6. Impact pressure and dynamic pressure versus Mach number.
where γ is the ratio between specific heats at constant pressure and constant volume.
Solving for the Mach number, we find

2 p∗ γ −1/γ
M =
2
−1 . (8.27)
γ −1 p
The true air speed will be found from Equation 8.27, the definition of speed of sound
(Equation 8.11) and the equation of ideal gases (Equation 8.2):

∗ γ −1/γ
! 2γ p p −p
TAS = +1 −1 . (8.28)
γ −1 ρ p
Solving Equation 8.26 for the stagnation pressure p∗ , we find

γ − 1 2 γ /γ −1
p∗ = p 1 + M , (8.29)
2

∗ γ − 1 2 γ /γ −1
p − p= p 1+ M −1 . (8.30)
2
The pressure difference p∗ − p is the so-called impact pressure. At low Mach num-
bers, when the flow is practically incompressible, the impact pressure is equal to the
dynamic pressure, as shown in Figure 8.6. An instrument that measures the impact
pressure, the local static pressure and the local temperature provides the TAS (or
KTAS, if given in knots).
Sometimes it is more useful to calibrate the speed to sea-level conditions. The
local pressure p and density ρ are replaced by the sea-level values. The result-
ing speed is called Calibrated Air Speed; it is abbreviated CAS or KCAS (CAS
in knots):

γ −1/γ
2γ po p∗ − po
CAS = ! +1 −1 . (8.31)
γ −1 ρo po
It can be verified that the relationship between TAS and CAS is

" "
To TAS δ
CAS TAS = √ = TAS . (8.32)
T θ σ
In other words, the CAS is equal to the true air speed at standard sea level. If the
local static pressure can be measured and the local density is replaced with the sea
level ρo, then we find again the equivalent air speed. EAS is the TAS corrected for
changes in atmospheric density,

∗ γ −1/γ
! 2γ p p −p
EAS = +1 −1 , (8.33)
γ − 1 ρo p
a result equivalent to the case of incompressible flow. For a given TAS, the EAS
increases with the increasing flight altitude. The Indicated Air Speed (IAS) is the
aircraft speed indicated by the instrument that can be affected by errors (position,
time and pressure lag): IAS = CAS + error. Here, we assume that the error is
negligible, so that we can call CAS = IAS.
When the aircraft flies at supersonic speed, the instrument is unable to sense the
actual free-stream conditions. A normal shock establishes ahead of the instrument.
Two events must be taken into account: 1) a normal shock wave ahead of the
probe that produces a subsonic Mach number, and 2) an isentropic deceleration
from a subsonic Mach number to stagnation pressure p∗ in the probe. This pressure
is related to the static pressure p by the Raleigh equation that is found in most
textbooks dealing with high-speed aerodynamics 10 :
γ /γ −1
p∗ (γ + 1)2 M2 1 − γ + 2γ M2
= . (8.34)
p 4γ M2 − 2(γ − 1) γ +1
To find the Mach number, this equation must be solved for M.

Figure 8.7 shows the combination of CAS-TAS-altitude and flight Mach number.
An operation point is uniquely determined by two parameters, such as Mach-TAS
combination, Mach-altitude, or CAS-altitude and so on (for example, point A). An
aircraft climbing at constant CAS would have an increasing TAS if also the Mach
number increases; if the Mach number is kept constant, then the TAS decreases as
the aircraft climbs – a result that can be inferred also from Equation 8.32.
TRANSITION ALTITUDE. Finally, we introduce the transition (or cross-over) altitude,

which is defined as the geopotential altitude at which the CAS and Mach number
represent the same TAS. This altitude is defined by the equation:
103
htrans = [(To + dT )(1 − θtrans )] . (8.35)
λ
13
0.74 0.76 0.78 0.80 0.82 M = 0.84

40
12 7 0
=2
AS
KC
Altitude, 10 feet
Altitude, km 0
11 28
3
0 35
29
0
30
10 A
0
31
0.84
30
9
0.80
8
420 440 460 480 500 520
KTAS
Figure 8.7. CAS-TAS-Mach-Altitude chart.
In Equation 8.35 the transition altitude is in [m]; θtrans is the relative temperature at
the transition altitude
−λR/g
θtrans = δtrans . (8.36)
The relative pressure at the transition altitude is

[1 + κ MC2 AS ]γ /γ −1
δtrans = (8.37)
[1 + κ M2 ]γ /γ −1
with
γ −1 VC AS
κ= , MC AS = . (8.38)
2 ao
A plot of the transition altitude over a range of cruise Mach numbers is shown in
Figure 8.8. The transition altitude varies greatly with the air speed. For example, at
250 KCAS and M = 0.85 the transition altitude is ∼11,400 m (∼37,400 feet).
8.3 Design Speeds

Aircraft speeds have been increasing over the years. These speeds reached a point of
diminishing returns around 1970, when most of the commercial long-range airplanes
were powered by gas-turbine engines. Cruise speeds converged toward an average
M = 0.78 to 0.82. Some advances in aerodynamics have allowed a slight increase
in the cruise speed for the latest generation of commercial jet aircraft, which is
M 0.85. A number of other design speeds and Mach numbers are required, as
listed here:
1. Structural design Mach number, MC : This Mach number is specified by the

aircraft manufacturer. For operational flexibility it is assumed that MC is equal
8.3 Design Speeds 207
18
Cross-over altitude, 10 feet

15 50
Cross-over altitude, km
3
12 40
0.85 30
9
0.80
0.75
M
=0
.70
6 20
150 200 250 300

KCAS
Figure 8.8. Transition altitude versus KCAS at selected Mach numbers.
to the maximum operating Mach number, MMO. For a transonic transport

airplane, MMO Mcruise + 0.04. The corresponding speed is called structural
design speed, VC .
2. Design dive Mach number, MD: This Mach number is larger than the maximum
operating Mach number, with MD 1.07 MMO, or MD 1.25MC . Specific con-
ditions are set forth in FAR § 25.335, which includes clauses for verification and
demonstration. The corresponding speed (design dive speed) is VD 1.15VC ,
based on a 20-second dive on a 7.5-degree glide angle. The value of MD must be
selected such that design cruise Mach number does not exceed 0.8MD.
3. The Never-to-exceed speed VNE is determined by the structural limits of the
aircraft. For a given aircraft and given gross weight, this speed depends on the
flight altitude.
4. Design manoeuvre speed, VA: This is the minimum equivalent air speed at which
the airplane can produce lift with flaps retracted, at the design load factor and
design weight. If n = 1, we have the stall speed; otherwise, L = nW. According
√
to FAR § 25.335, VA ≥ VS1 n (VS1 is the stall speed with flaps retracted).
5. Design speed for maximum gust intensity, VB: This speed must satisfy the follow-
ing condition:

a1 Kg Uref VC 1/2
VB ≥ VS1 1 + , (8.39)
WL
where Uref is the reference equivalent air speed of the gust; WL = W/A is the
average wing loading, a1 is the normal-force curve slope and Kg is a factor, as
defined by FAR § 25.335. The parameters in Equation 8.39 must be given in
coherent units (the FAR documents give imperial units).
Table 8.2. Recognised international symbols for design air speeds and
Mach numbers
Symbol Alternative Definition FAR 25
VA VA Design manoeuvre speed

VB VB Design speed for maximum gust intensity
VC VC Design cruise speed
VD VD Design dive speed 1.15VC
VF VF Design flap speed
VSo VSo Design speed, flaps retracted
VS1 VS1 Design speed, flaps deployed
VNE VNE Not-to-exceed air speed
MMO MC Maximum operating Mach number
MD MD Design dive Mach number 1.07 MMO
6. Design flap speed, VF : This speed is a function of the flap angle, the flight altitude
and the stalling speed. It is in fact a restriction on flap deployment, which is given
as:
VF > 1.6VS1 with take-off flaps and MTOW.
VF > 1.8VS1 with approach flaps and MLW.
VF > 1.8VS0 with landing flaps and MLW, where VS indicates the flaps-up
condition.
A summary of these speeds is given in Table 8.2. Other operational and design
speeds are defined in Chapters 9 and 13.
8.4 Optimum Level Flight Speeds

We derive some closed-form solutions for steady-state flight that correspond to
minimum drag and minimum power. It is possible to demonstrate that a level speed
can be maintained by the aircraft if
T
≥ 2 CDo k. (8.40)
W
The limiting condition occurs at the absolute ceiling of the aircraft. The minimum
speed coincides with the maximum speed. At the absolute ceiling there is only one
possible speed
T
U= . (8.41)
ρ ACDo
The danger of this situation is that the aircraft cannot accelerate (because of insuffi-
cient thrust), it must not decelerate (because it could enter a stall condition) and it
must not perform a turn (because it could stall, as a consequence of increasing drag).
Theoretically, the only way out of this situation is a descent at constant air speed.
We now calculate the speeds relative to minimum drag and minimum power of
a generic aircraft, jet- or propeller-driven. Consider an aircraft whose drag equation
is parabolic. The drag force on the airplane is
1
D= ρ AU 2 (CDo + kCL2 ). (8.42)
2
8.4 Optimum Level Flight Speeds 209
The speed corresponding to minimum drag is found from the condition that

D D
D= L= W (8.43)
L L
is at a minimum. This implies that the glide ratio CL/CD is at a maximum. The
derivative of Equation 8.43 is to be calculated with respect to the CL. The condition
∂(CL/CD)/∂CL = 0 yields
"
CDo
CL = . (8.44)
k
This is the CL corresponding to minimum drag. The corresponding speed will be

2W k 1/4
Umd = . (8.45)
ρ A CDo
An alternative solution is to consider CL ∼ U 2 , that is, a minimum with respect to CL

will be a minimum with respect to the aircraft speed (and vice versa). The result will
be the same, although more elaborate. The conclusion of Equation 8.45 is that at a
given altitude, the speed of minimum drag increases with the wing loading and with
the profile-drag coefficient; it decreases with the increasing lift-induced factor. All
other parameters being constant, Umd increases with the flight altitude. The speed
corresponding to minimum engine power for the same aircraft is

CD 2W3
P = DU = 3/2 . (8.46)
CL ρA
At a given altitude the terms under square root are constant; therefore the speed of
3/2
minimum engine power is the speed that minimises the factor CD/CL . The condition
of minimum power is found from

∂ CDo + kCL2
= 0. (8.47)
∂U 3/2
CL
After working out the derivative in Equation 8.47, we find
−3/2 1/2
3CDo c1 U 4 − kc1 = 0, (8.48)
with c1 = 2W/ρ A. The solution of the latter equation is the speed of minimum power
1/4
2W k
Ump = . (8.49)
ρ A 3CDo
This speed corresponds to a lift coefficient
"
CDo
CL = 3 . (8.50)
k
√ relationship between the CL of minimum drag and minimum power is simply

The
3, whilst the ratio between corresponding speeds is
Umd √
= 3 Umd ∼ 1.32Ump .
4
(8.51)
Ump
KTAS KTAS
150 200 250 300 350 150 200 250 300 350
80 10,000
Profile drag Profile power
Induced drag Induced power
Total drag Total power
8,000
60
Power, kW
6,000
Drag, kN
40
4,000
20
2,000
0 0
200 300 400 500 600 700 200 300 400 500 600 700
TAS, km/h TAS, km/h
(a) (b)
Figure 8.9. Drag and power characteristics of generic subsonic airplane.
This proves that the speed of minimum drag is about 32% higher than the speed of
√
minimum engine power. Both optimal velocities change with altitude like 1/ σ .
The variation of drag and power with the aircraft speed is shown in Figure 8.9
for a generic subsonic jet aircraft. The induced component decreases with the speed;
the profile drag grows as U 2 ; the corresponding power grows as U 3 . The sum
beween the two contributions has a minimum at an intermediate value of the air
speed.
8.5 Ceiling Performance

The absolute ceiling is the maximum altitude at which an aircraft can keep a steady
level flight. Above this altitude, the engine power or thrust is not enough to overcome
the aircraft drag. The aircraft can zoom past the absolute ceiling by exchanging part
of its kinetic energy for potential energy (i.e., altitude). This can be done with an
inertial climb, even in absence of sufficient engine power.
The ceiling altitude depends on the type of aircraft. It is about 6,000 to 7,000 m
(∼19,700 to 23,000 feet) for turboprop aircraft; 10,000 to 12,000 m (∼32,800 to 39,400
feet) for commercial jet aviation; and it increases to 18 or 19 km (above 60,000 feet)
for high-performance military aircraft. It is seldom above this altitude. The Lockheed
SR-71 had a cruise altitude of 27,000 m (∼88,500 feet) and an estimated ceiling of
30,700 m (∼100,000 feet).
8.5.1 Pressure Effects on Human Body

The flight altitudes described could not be achieved without cabin pressurisation.
In fact, pressurisation is required at altitudes above 3,000 m (∼9,840 feet). The first
airliner to be fully equipped with a pressurised cabin was the Lockheed Constellation
in the 1940s.
At about 10,000 feet (∼3,050 m), pilots without an artificial oxygen supply begin
to suffer from hypoxia: their brains begin to show poor response and they speak
gibberish. Up to about 34,000 feet (∼10,360 m), the sea-level equivalent amount
of oxygen can be supplied by increasing the percentage of oxygen in the air. At
about 40,000 feet (∼12,200 m), all of the oxygen must be supplied artificially. At
about 50,000 feet (∼15,200 m), gases trapped in the body expand, swelling intestines
and rupturing lung tissues. In 1959, U.S. pilot William Rankin had to bail out of
his Vought F8U Crusader aircraft at 50,000 feet (∼15,200 m) and survived the fall
thanks to his oxygen mask. However, his abdomen stretched and bloated from the
expanding gas, air exploded through his ears, and he bled from his eyes, nose and
mouth.
8.5.2 Cabin Pressurisation

The cabin pressure is monitored by an automatic control system. The system allows
air to escape from the cabin by controlling an outflow valve. As the airplane changes
altitude, the valve is repositioned so that pressure changes are within the comfort
limits. Maintaining a sea-level pressure is not possible or even economical. A typical
cabin pressure is equivalent to about 2,000 m (∼6,560 feet). The relevant regulations
for cabin-pressure altitude are given by FAR, Part 25, § 841. A loss of cabin pressure
causes a rapid loss of consciousness. For this reason, oxygen masks are fitted as part
of the essential safety kit on passenger airplanes. Even then, the airplane is required
to descend rapidly.
The Gulfstream 550 (Appendix A) has an automatic emergency descent proced-
ure that, after a 90-degree left turn, can take the aircraft from 40,000 feet (∼12,190 m)
down to 15,000 feet (∼4,570 m) at the maximum descent rate (about 5,000 feet/min;
25.4 m/s). In some cases, for flight above 41,000 feet (∼12,500 m), the FAA regula-
tions also require that one pilot wear an oxygen mask to ensure that a loss of cabin
pressure does not hinder their physical capabilities. A typical cabin pressure as a
function of the flight altitude is shown in Figure 8.10.
Airplanes such as the Boeing B737 have two air conditioning systems. If one
system does not work, the aircraft can still fly but it has to maintain a cruise altitude
below 25,000 feet (7,620 m, δ = 0.3711). At this altitude, it is possible to breathe
only with difficulty. In case of failure of the air conditioning system, the aircraft
must descend to 14,000 feet (∼4,270 m, δ = 0.5875), for which about 4 minutes are
required. In these emergency situations, oxygen masks provide air supply to the
passengers and the crew.
8.6 Flight Envelopes

The set of all flight corridors from sea level to the absolute ceiling defines the flight
envelope. In other words, the flight envelope is the closed area in the M − h diagram
that includes all operating conditions for a particular aircraft configuration at a given
weight. The flight envelope depends on a large number of factors, including weight,
aerodynamics, propulsion system, structural dynamics, cabin-pressure limits and
atmospheric conditions.
3
Cabin pressure altitude, 10 feet
0 2 4 6 8
12 40
Flight altitude, 10 feet

Flight altitude, km
30
3
9
6 20
3 10
0 0
0 1 2 3
Cabin pressure altitude, km
Figure 8.10. Typical cabin pressure on a modern airliner.
Figure 8.11 shows a generic flight envelope for a commercial subsonic jet. The
graph shows lines of constant CAS and constant Mach numbers. The flight is limited
at high speed by the maximum operating speed (VMO) or the maximum operating
Mach number, Mmo. In general, the aircraft will cruise at a speed slightly lower than
this, at the recommended cruise speed (indicated by a thick dashed line).
15
0.4 0.6 0.8 1.0
Max Cabin Altitude
12 40
ll
Sta
Altitude, 10 feet
Altitude, km
9 30
3
MMO
KCAS = 250
300
KCAS =
= 450
6 20
KCAS
VMO
3 10
Constant CAS
Constant Mach
0 0
100 200 300 400 500 600
KTAS
Figure 8.11. Generic flight envelope of commercial subsonic jet aircraft.
True Mach number

0.2 0.4 0.6 0.8 1
500 B C 15
400 12
D
Altitude, km
Flight Level
300 E 9
200 6
S
CA
7K
36
100 3
A F
0 0
100 200 300 400 500 600
KTAS
Figure 8.12. Flight envelope of the Airbus A320-200-CFM; standard day, no wind; m = 58,800
kg (calculated).
8.6.1 Calculation of Flight Envelopes

The subject of this section is the determination of the flight envelope of a commercial
or transport aircraft. This envelope consists of three main segments:
1. Stall Line: This is the low-speed segment A-B from sea level (A) to the opera-
tional ceiling B (to be determined), in clean configuration. At low altitudes, there
can be an extension of the envelope thanks to the deployment of the high-lift
surfaces (Figure 8.12).
2. Operational Ceiling: This is the corridor B-C between the stall speed (or Mach
number) B and the maximum speed (or MMO).
3. High-Speed Line: This is the segment C-F from the operational ceiling (C) to sea
level (F), in clean configuration. This segment can be split into two sub-segments
if the operational ceiling is above the tropopause.
r The high-altitude segment C-D-E is a constant-Mach line; point D is at the
tropopause.
r The low-altitude segment E-F is constant-CAS line.
For the calculation of the stall line, the reference stall speed is
1/2
2W
VS = Ks , (8.52)
ρ ACLmax
where Ks is the stall margin and CLmax is the maximum (or usable) lift coeffi-
cient of the airplane in clean configuration. The stall speed depends on the flight
altitude; this altitude is increased up to a point B when one of the following events
occurs:
1. The climb rate falls below a threshold (for example, vc < 100 feet/min).
2. The maximum level acceleration falls below a threshold (for example, a/g <
0.2).
3. The cabin-pressure altitude falls below a threshold (for example, pc < 8,000
feet).
4. The engine surges (see also § 8.7.3).
The operational ceiling should be set at the most restrictive conditions among
these options. The third option is difficult to evaluate because the cabin pressure
depends on the aircraft systems, in particular the APU. For example, the A320
FCOM stipulates that the normal cabin-pressure altitude is pc ∼8,000 feet, with a
warning at pc ∼9,500 feet* . At standard conditions, the geometrical flight altitude
corresponding to this pressure differential would be
p = pc − pa (35.65 − pa ) kPa. (8.53)
The operational ceiling given in the FCOM is 12,500 m (∼41,000 feet), which corres-
ponds to a pressure pa 17.934 kPa at standard conditions. If the 8,000-feet cabin
pressure is maintained, the cabin-to-exterior difference would be p 17.71 kPa,
which is about one third of the limit pressure differential for the fuselage shell.
The fourth option requires a full engine simulation, with an analysis of the
compressor map. The first and second options are manoeuvre requirements (excess
power and excess thrust, § 10.2.1). For the purpose of the analysis carried out in
this context, we use the first two options. Therefore, we define the operational
ceiling as
# $
zop = min h(vc∗ ), h(a ∗ ) , (8.54)
where h(vc∗ ) denotes the flight altitude corresponding to a vc < vc∗ ; h(a ∗ ) denotes the
flight altitude corresponding to an acceleration a/g < 0.2. At the operational ceiling
zop , we need to establish the maximum speed of the airplane, C. We define this point
as the most restrictive among the following options:
1. The maximum level acceleration falls below a threshold (for example, a/g <
0.2).
2. The engine intake suffers a buzz (see also § 8.7.3).
The first option is to allow for some excess power to manoeuvre away from the
operational ceiling. Once this point is determined, we calculate the constant-Mach
limit of the envelope at high altitude, C-E. The limit flight Mach number is the MMO.
The lowest altitude E can be determined again as a minimum manoeuvre limit. At
the point E, we calculate the KCAS. At altitudes below E (segment E-F), the limit
of the envelope is determined by a constant KCAS, as previously calculated. The
segment E-F is sometimes split into two sub-segments, corresponding to KCAS1 at
* For this airplane, the maximum differential pressure (cabin-to-outside) is p = 8.6 psi (∼59.3 kPa);
this limit is never reached.
Predicted FL-510; M = 0.860

500 FCOM 15
400 MMO 12
Altitude, km
Flight Level
300 9
O
VM
=
200
AS
6
KC
0
34
100 3
S
CA
0K
30
0 0
0.1 0.3 0.5 0.7 0.9
Mach number
Figure 8.13. Flight envelope of the Gulfstream G550; standard day, no wind; m = 35,500 kg.
altitudes above 3,000 feet and KCAS2 at altitudes below 3,000 feet. Because the
manoeuvre limits can be arbitrary, further information may have to be gathered.
8.6.2 Case Study: Flight Envelopes of the A320 and G550

First, we consider the Airbus A320-200 with CFM engines. Figure 8.12 shows the
flight envelope of this airplane. The operational points A through F are indicated in
the graph. Without a limit on the cabin pressure, the operational ceiling is estimated
at 13,900 m (∼41,600 feet), well above the limit of 12,500 m given in the FCOM.
Once it has reached the point C, the airplane moves along the MMO line down to
a flight altitude 26,000 feet (E). Below this point we estimated a limit line at 338
KCAS down to sea level (F).
The second case refers to the Gulfstream G550. The present calculations are
compared with data available in the FCOM (Figure 8.13). We predict an operational
ceiling at 15,290 m (49,900 feet), just short of the value of 51,000 feet quoted by the
manufacturer. The stall line is not well predicted. However, if the FCOM is correct,
it would mean that the aircraft is capable of manoeuvring at low altitude with ∼100
KTAS, which seems extraordinary. We estimate that we need at least 135 KTAS to
fly at sea level.
On the high-speed segment, the upper portion of the FCOM indicates that the
aircraft must not operate above M = 0.86 at altitudes above 43,500 feet (see also
Figure A.1 in Appendix A). No such limit is found on the basis of excess thrust alone.
Therefore, we must conclude that this slightly reduced Mach number is related to
intake buzz and other engine issues.
The limit speed 340 KCAS is well predicted as a result of a good prediction
of the limit point E. However, the lower portion of the high-speed segment in the
FCOM indicates that the airplane must not exceed 300 KCAS below 9,000 feet.
8.7 Supersonic Flight

We present a limited number of problems in supersonic flight: dash speed, super-
sonic acceleration and limiting factors in flight envelopes. We use steady-state aero-
dynamic data for the calculation of an accelerated flight and make no allowance for
transient thrust, wing buffet and other flight-mechanics problems.
The flight-mechanics problems are exclusively the realm of military aircraft.
In practice, it will be difficult to obtain actual aircraft performance data because
manufacturers compete for lucrative contracts running for decades and worth billions
of dollars. The marketing headlines include statements such as a certain aircraft is
“uncompetitive and unworkable”, another one is “uncapable of carrying the same
weapons load” and yet another one “would regularly fly 700 miles at 40,000 feet
at just-less the speed of sound, and feels no different to when it does not carry
anything”. Statements like these will have to be critically evaluated before signing
such a contract.
8.7.1 Supersonic Dash

We begin with an aircraft whose engine thrust and transonic drag rise are given
in tabulated format from flight testing. We show how the solution of this problem
leads to a root finding of non-linear algebraic equation. The problem may lead to
non-physical and non-unique solutions. The conditions that must be satisfied by the
aircraft in level flight are:
T = D, L = W. (8.55)
The solution is given by the angle of attack that provides the maximum Mach number
compatible with Equation 8.55. From the equilibrium in the vertical direction we
find the angle of attack and the lift coefficient, respectively:
2W 1
α = αo + , (8.56)
ρ Aa CLα M2
2
CL = CLα (α − αo). (8.57)
We will consider αo 0, but the solution procedure is unaffected by αo. The equi-
librium in the horizontal direction is
1
T= ρ Aa 2 (CDo + ηCLα α 2 )M2 . (8.58)
2
By further simplification, we have
η 1
T = c1 CDo M2 + c2 (8.59)
CLα M2
with
1 2W2
c1 = ρ Aa 2 , c2 = . (8.60)
2 ρ Aa 2
In Equation 8.59 the aerodynamic coefficients are also a function of the Mach num-
ber. The solutions are

T2 1 ηCDo
M =
2
± T 2 − 4W 2 . (8.61)
ρ Aa 2 ρ Aa 2 CLα
Both solutions are positive. Although both transonic and supersonic speed are pos-
sible, Equation 8.61 does not give an indication as to how the aircraft can reach
these speeds. It is possible to demonstrate that Equation 8.61 has either no physical
solutions or a single solution or multiple solutions, depending on flight altitude and
gross weight.
8.7.2 Supersonic Acceleration

The next problem is to calculate the acceleration of the supersonic jet fighter from
a cruise Mach number (say, M = 0.8) to supersonic speed. There are different ways
to achieve this. One is an acceleration at constant altitude; another one is an accel-
eration at constant angle of attack and constant attitude. Finally, the aircraft can
accelerate to phenomenal speeds by doing a zoom-dive. Several flight-mechanics
studies exist in the technical literature to address acceleration problems 11;12 , includ-
ing transient and steady-state solutions. In this context, we will solve only the rel-
atively simple case of acceleration at constant altitude. Any other profile requires a
zoom-climb or a zoom-dive and is discussed in Chapter 10.
Acceleration at Constant Altitude

We now proceed to the calculation of such accelerations. The equation of motion in
the flight direction is
∂U
m = T − D. (8.62)
∂t
If we use the definition of speed of sound and rearrange the equation, we find
∂M 1T 1 ρa A
= − CDo + ηCLα α 2 M2 . (8.63)
∂t am 2 m
The angle of attack is established from the equilibrium condition in the vertical dir-
ection, L = W, Equation 8.56. As the aircraft accelerates in the horizontal direction,
it must decrease its lift coefficient in order to keep level flight. Solution of the prob-
lem requires integration of the ordinary differential equation. A typical simulation is
shown in Figure 8.14. All other conditions being the same, the weight is important in
setting the limits on maximum acceleration and supersonic dash. At m = 10,000 kg,
the drag nearly equals the available thrust; the aircraft is capable of going past the
M = 1.25 limit to reach an M 2.1. With a higher weight, the aircraft would not
be capable of performing this acceleration. A look at the function (∂U/∂t)/g shows
relatively low accelerations – compared with accelerations in high-speed turns and
pull-ups. The acceleration is about 0.35g at the start and goes down to nearly zero at
the transonic point T. It then grows again, thanks to the reduction in aerodynamic
drag; thus, the airplane can accelerate to its maximum speed (for this configuration
and altitude), as indicated by point M.
0.4
Available thrust
80 Drag
g-acceleration M
0.3
Thrust, Drag, kN
60
T
a/g
0.2
40
0.1
20
0 0
1 1.4 1.8 2.2
Mach number
Figure 8.14. Acceleration of supersonic aircraft model, constant flight altitude h = 8,000 m
(∼26,250 feet); m = 10,000 kg; standard day, no wind, clean configuration.
By comparison, Concorde’s best acceleration took the aircraft from M = 1 to

M = 2 in about 7 minutes. After-burning was used in the range M = 0.96 to 1.70.
In the military arena, the Vought F8U-3 with after-burning thrust was capable of
accelerating from M = 0.98 to M = 2.2 in 3 minutes and 54 s at 11,500 m (∼37,700
feet). The F-15 Strike Eagle is capable of accelerating from M = 1.1 to M = 1.8 in
56 s at about 32,000 feet (∼9,750 m).
8.7.3 Supersonic Flight Envelopes

Figure 8.15, adapted from Abercrombie 13 , shows the flight envelope of a high-
performance supersonic jet aircraft. The outer envelope is the limit of all operation
points from a statistical analysis. The graph shows a series of points that represent
operational flight conditions, including acceleration, deceleration, climb, descent
and cruise. The straight lines join extreme points of the acceleration or deceleration
envelope. The shaded area indicates the normal envelope. The maximum speed is
supersonic at all altitudes.
The envelopes discussed take into account only the basic aerodynamic and
propulsion characteristics but not other factors, such as structural limits of the air-
craft, aero-thermodynamic heating and other forcing at the extreme flight conditions.
Some of the limitations are described in this section and schematically shown in Fig-
ure 8.16. There are at least four phenomena that need consideration: intake buzz,
wing flutter, skin temperature limits and blade stall.
The intake buzz is the interaction between the oblique shock wave at the ramp
of the engine inlet and the ramp boundary layer (Seddon & Goldsmith 14 ). It creates
oscillating conditions inside the intake that assume the shape of forced vibrations.
20
Formal qualitative test
Qualitative performance data 60
15 50
Altitude, 10 feet
Altitude, km
40
3
10
30
20
5 Acceleration
Deceleration
3.5g turn 10
Descent
Climb
0 0
0 0.5 1 1.5 2
Mach number
Figure 8.15. Flight envelope of high-performance jet aircraft.
A typical frequency is about 10 Hz. If a shock occurs, it will be formed by two waves:
an oblique shock from the external compression flow and a normal shock. These two
shocks meet at a point and expand outward, as a λ-shock. When intake buzz starts,
the shock intersection point is inside the cowl lip and a shear flow is established.
For a given Mach number, if this shear flow is internal to the diffuser, it creates
flow separation that reduces the mass captured by the inlet because the separated
KTAS
0 250 500 750 1,000
15
45
Engine
Surge Intake
Altitude, km
Buzz
Altitude, kft
10
30
5
15
Wing
Stall Wing
Flutter
0 0
0 500 1,000 1,500 2,000
TAS, km/h
Figure 8.16. Limiting factors on high-performance flight envelope.
flow causes a blockage effect. In the meantime, the compressor runs at constant
r pm and tends to suck in all the air at the inlet. The pressure ratio in the engine
will decrease. With the inlet pressure decreasing below the stagnation pressure, the
shear flow disappears and so does the blockage effect; then the process starts again.
Some of these problems can be reduced or eliminated by proper intake design, in
particular, a variable geometry.
The engine surge is the result of compressor stall in the jet engine. As a result,
the complete engine may stall. This is a rare event, appearing first as a loud bang.
The air flowing over the compressor blades stalls just as the air over the wing of
an airplane. When airfoil stall occurs, the passage of air through the compressor
becomes unstable and the compressor can no longer compress the incoming air. The
high-pressure air behind the stall farther back in the engine escapes forward through
the compressor, and out of the inlet. This escape is impulsive, like an explosion.
Engine surge can be accompanied by visible flames around the inlet or in the tail
pipe. Often the event is so quick that the instruments do not have time to respond.
Generally, the instability is self-correcting. In modern engines there are surge valves
that pump the disturbed flow out of the engine and thus limit the instability.
The wing flutter is a dynamic coupling between elastic motion of the wing and the
unsteady aerodynamic loading. This dynamic response is at first stable but increases
with the Mach number. It depends on the geometry of the wing (aspect-ratio, sweep
angle), its stiffness, its moment of inertia and other parameters. This response (fre-
quency and damping) may depend on the acceleration rate of the aircraft. The
response can be quantified by the flutter number.
Summary
We have reviewed a number of aircraft design speeds and various design limitations.
We presented a model for standard atmosphere and some cases when there is devi-
ation from it; specifically, we have considered cases in which the atmosphere is colder
or warmer than the standard day; thence, we explained that it is important to under-
stand the difference between geopotential altitude and pressure altitude (which is
measured by an altimeter). From the atmospheric models we have defined a number
of operational speeds: calibrated air speed, equivalent air speed and indicated air
speed, all of which depend on altitude and pressure. We have calculated optimum
level flight speeds in a closed form and showed optimal conditions. Then we intro-
duced the concept of flight envelope at subsonic and supersonic Mach numbers and
discussed various limits, including operational ceiling, stall speeds, maximum speeds
and their causes: cabin pressure, engine surge or buzz, and excess thrust. We finally
addressed the problem of level acceleration past the speed of sound (supersonic
dash.)
Bibliography
[1] NACA. Standard atmosphere – Tables and data for altitudes to 65,800 feet.
Technical Report R-1235, NACA, 1955. (Supersedes NACA TN-3182).
[2] Minzer RA, Champion SW, and Pond HL. The ARDC Model Atmosphere.
Technical Report 115, Air Force Surveys in Geophysics, 1959.
[3] Anon. U.S. Standard Atmosphere. Technical report, U.S. Government Printing
Office, Washington, DC, 1962.
[4] Anon. Manual of the ICAO Standard Atmosphere, extended to 80 kilometres
(262,500 feet). Technical report, ICAO, 1993, 3rd edition.
[5] Prandtl L and Tietjens OG. Fundamentals of Hydro- and Aeromechanics.
Dover, 1957. Chapter 2.
[6] Etkin BE. Turbulent wind and its effect on flight. J. Aircraft, 18(5):327–345,
May 1981.
[7] Houbolt JC. Atmospheric turbulence. AIAA J., 11(4):421–437, April 1973.
[8] Barry RG and Chorley RJ. Atmosphere, Weather and Climate. Routledge,
London, 8th edition, 2003.
[9] Anonymous. Handbook of Aviation Meteorology. HMSO, London, 3rd edition,
1994.
[10] Kuethe AM and Chow CY. Foundations of Aerodynamics. John Wiley, 5th
edition, 1997.
[11] Miele A. Flight Mechanics. Vol. I: Theory of Flight Paths. Addison-Wesley,
1962.
[12] Bilimoria KD and Cliff EM. Singular trajectories in airplane cruise-dash optim-
ization. J. Guidance, Control and Dynamics, 12(3):303–310, May 1989.
[13] Abercrombie JM. Flight test verification of F-15 performance predictions. In
Performance Prediction Methods, CP-242. AGARD, 1978.
[14] Seddon J and Goldsmith EL. Intake Aerodynamics. Blackwell Science, 1999.

Symbols defined in Table 8.2 are not repeated here.
a = speed of sound; acceleration

a1 = normal-force curve slope, Equation 8.39
A = wing area
ci = constant factors, i = 1, 2 · · ·
C = constant in Sutherland’s law, Equation 8.13
CLmax = maximum-lift coefficient
h = altitude
hp = geopotential altitude
htrans = transition altitude
H = relative humidity
Kg = factor in Equation 8.39
Ks = stall margin
m = mass
M = Mach number
n = number of moles, Equation 8.15
p = pressure
pc = cabin pressure
po = sea-level standard pressure
p∗ = stagnation pressure
P = power
R = gas constant
t = time
T = net thrust
T = temperature
U = air speed
Umd = speed of minimum drag
Ump = speed of minimum power
Uref = reference (equivalent) gust speed, Equation 8.39
vc = climb rate
Vg = ground speed
Vw = wind speed
W = weight
WL = wing loading, W/L
zop = operational ceiling
Greek Symbols
αo = zero-lift α
δ = relative air pressure
δtrans = relative pressure at transition altitude
η = propeller efficiency; lift-induced drag coefficient
θ = relative air temperature
θtrans = relative temperature at transition altitude
λ = temperature-lapse rate
κ = parameter defined by Equation 8.38
κa = thermal conductivity of moist air
κd = thermal conductivity of dry air
κv = thermal conductivity of water vapour
ρ = air density
ρo = air density at sea level, standard conditions
σ = relative air density
[.]o = standard conditions
[.]a = air
[.]d = dry air
[.]g = ground
[.]t = troposphere
[.]v = water vapour
[.]w = wind
[.]∗ = reference conditions
9 Take-Off and Field Performance
Overview
This chapter deals with a realistic model of the aircraft at take-off conditions. After
an introduction to general take-off procedures, we deal with transport-type airplanes
(§ 9.1) and write the general take-off equations (§ 9.2). We carry out numerical solu-
tions with all-engines operating §9.3 and consider the case of one-engine inoperative,
both accelerate-and-go and abort-and-stop (§ 9.4). The takeoff of a turboprop air-
craft is solved in §9.5. Lateral control issues in both cases are dealt with in the
analysis of the minimum control speed (§ 9.6). We discuss the important cases of
aircraft braking (§ 9.7) and performance on contaminated runways (§ 9.8), including
extreme cases such as hydroplaning and other forms of contamination. For the sake
of completeness, we present more rapid closed-form solutions (§ 9.9). The chapter
ends with considerations on field performance, including taxiing and turning (§ 9.10)
and the effects of bird strike (§ 9.10.2).
KEY CONCEPTS: Take-Off Equations, Normal Take-Off, OEI Take-Off, Balanced

Field Length, Decelerate-Stop, Accelerate-Go, Minimum Control Speed, Contam-
inated Runways, Ground Manoeuvring, Bird Strike.
9.1 Take-Off of Transport-Type Airplane

The take-off phase is dependent on a vast array of aircraft parameters and external
circumstances – including the pilot’s skills. Take-off lasts little time and burns a
considerable amount of fuel.
There are several different types of aircraft take-off and landing. We will refer
exclusively to Conventional Take-Off and Landing (CTOL). This class includes most
civil, commercial and military vehicles, that is, subsonic jet transport, turboprop
airplanes, and cargo airplanes that take off and land horizontally. These airplanes
operate from normal airfields with a paved runway; they have moderate thrust ratios
or power loading.
We consider a take-off operation on a vertical plane. Due to the level of
approximations that it is possible to achieve, in the calculation of the take-off of a
224
Table 9.1. International symbols for take-off of a transport airplane
Symbol Math Definition FAR/JAR 25.107
VFE VF E maximum speed with flaps extended

VEF VEF engine failure speed VEF ≤ V1
VS VS stall speed
VMCG VMCG minimum control speed on ground < VLO
VMC VMC minimum control speed in air
VMU VMU minimum unstick speed ≥ VS
VR VR rotation speed > 1.05Vs
V1 V1 OEI decision speed ≥ VMCG
VLO VLO lift-off speed > 1.1VMU ; > 1.05VME (OEI)
V2 V2 speed over obstacle (take-off) > 1.2VS
VMBE VMBE maximum brake energy speed ≥ V1
VTIRE VT I RE maximum tyre speed VT I RE ≥ V1
conventional airplane, we will assume that the runway reference system and the
ground reference system are equivalent. The sequence of airplane speeds during
take-off are given in Table 9.1 and further explained graphically in Figure 9.1.
With reference to Figure 9.2a, take-off is performed with all engines operating.
The airplane accelerates, the nose landing gear lifts off and the airplane rotates with
a speed VR; the airplane lifts off the ground with a speed VLO and follows a rectilinear
flight path to clear a conventional screen at 35 feet from the ground. In Figure 9.2b an
engine failure occurs when the airplane has reached a speed VEF . The pilot decides to
continue with the remaining engine thrust (after stabilising the airplane) to take off
over a longer distance. In Figure 9.2c, following the engine failure, the pilot decides
to abort the flight and stop the airplane on the runway. The sequence of events in
Figure 9.2 is explained in further detail in the following sections. According to the
airworthiness requirements, there are four key distances on the ground:
A. All-engines-operating take-off distance from brake release to clearing of screen.

B. All-engines-operating accelerate-stop distance.
TODA
TOD
Take-off
VMCA
BR
VMBE
VMCG VEF V1 VR
VLO V2 Clearway
VTIRE
Figure 9.1. Sequence of reference speeds in a conventional take-off; TOD = take-off distance.
Airborne phase
Take−off
Brake release
Decision
Rotation
Lift−off
V V V
1 R LO
distance "E"
1.15E
distance "A"
1.15A
(a) All Engines Operating
Airborne phase
Take−off
Brake release
Engine failure
Decision
Rotation
Lift−off
V V V V
EF 1 R LO OEI
distance "F"
distance "C"
Shutdown
Spoilers
Brakes
STOP
VEF V1 Braking
Accelerate Decelerate & Stop

distance "D"
(b) OEI accelerate & Go/Stop

Figure 9.2. Take-off distances; adapted from Ref. 1
C. One-engine-inoperative take-off distance from brake release to clearing of

screen.
D. One-engine-inoperative aborted take-off distance from brake release to airplane
halt.
Clean Wing VMCG
High Lift VMCA
Ground Effect GC & Trim

Aerodynamics Dynamics
Prop-Trim
Propulsion Take-off Landing Gear
De-rating Dynamics
Risk Analysis External
Propeller Heat Loads
BFL ATC
AEO / OEI Human Fact.
Weather
Figure 9.3. Flowchart of take-off problem.
A number of critical distances are defined. For the all-engines-operating case

we have:
1
E= (xlo + xto) AEO . (9.1)
2
This is the AEO distance from brake-release to the halfway point between lift-off
and take-off. Similarly, for OEI we have:
1
F= (xlo + xto) OEI . (9.2)
2
The Take-Off Distance Required (TODR) is defined by
TODR = max {1.15A, C}. (9.3)
The take-off distance available (TODA) must be longer than the TODR. The Take-
Off Run Required (TORR) is defined by
TORR = max {1.15E, F}. (9.4)
The Accelerate-Stop Distance Required (BFL) is defined as
BFL = max {B,D}. (9.5)
Once the airplane becomes airborne and flies past the conventional screen, it
starts its initial climb. This flight segment is discussed separately in Chapter 10. The
main aspects of the take-off problem are illustrated in the flowchart in Figure 9.3, as
they are implemented in the FLIGHT program.
The forces and moments acting on the aircraft are divided into three separate
categories: 1) aerodynamics, 2) propulsion, and 3) tyre-runway contact. A realistic
model includes the following aspects:
r aerodynamics, including the effect of high-lift systems and the ground effect
r airplane dynamics, including minimum control speeds, longitudinal and pro-
peller trim
r landing-gear dynamics, including tyre loads, heating and ground-tyre inter-
action
r propulsion aspects, including engine derating and propeller performance
r risk analysis, including one engine shut-down, and balanced field length
r external factors, including weather, air traffic control and human factors
The aerodynamics has been presented in Chapter 4. Derated and flexible thrust
is discussed in § 5.2.1. The other aspects are discussed in this chapter, except air
traffic control and human factors, which are beyond the realm of the airplane itself.
9.2 Take-Off Equations: Jet Airplane

We consider the general case of a jet-powered transport-type airplane taking off from
a prepared runway. We write the equations for the centre of gravity of the airplane
moving on a vertical plane, that is, on a straight line whilst on the ground, and on
a straight climb. The computational model is built as a set of ordinary differential
equations that are solved from the brake-release point to the point of clearing a
conventional screen.
A detailed solution method for the take-off of transport airplanes is given by
ESDU 1– 3 . Powers 4 published a method for the estimation of the balanced field
length. An algorithm shown by Krenkel and Salzman 5 used such an integration
scheme to calculate the take-off performance and the balanced field length.
BASIC NOMENCLATURE. We need to specify the nomenclature before getting started:

V denotes the total speed with respect to a reference on the ground. If γ is the flight-
path angle, then the velocity components are u = V cos γ and v = V sin γ . In the
ground run, with all wheels on the runway, u = V and v = 0. The air speed is called
U. If there is no wind, then U = V. If the wind has only a horizontal component, the
relationship between air speed and ground speed is
1/2
U = (V cos γ ± Uw )2 + V 2 sin2 γ . (9.6)
The air speed U is also called TAS (or KTAS, if expressed in knots).
The main forces and moments are shown in Figure 9.4. The weight is split
between main and nose landing wheels. The wheels are subject to the reaction of the
roadway, which has two components: normal (which operates against the weight)
and parallel to the ground (which is effectively the rolling resistance). The forces
on the aircraft are reduced to weight (operating at the CG), aerodynamic lift from
the wing and the horizontal tail (operating on the aerodynamic centre of the wing),
and the corresponding aerodynamic moments. Finally, there is the propulsive thrust,
which is symmetric with respect to the vertical plane and therefore moved to the
xm
xn
Lw
xt
xw
Lt
+
CG Wing
H−tail −
zt
x
T
Main UC Nose UC
Fn
W
Fm
Figure 9.4. Aircraft forces on the ground.
symmetry plane. In Figure 9.4, we show a thrust vector that is horizontal and below
the wing.
Because the wing has a complex planform, it is useful to replace it with the
equivalent mean aerodynamic chord (MAC), which is calculated as shown in § 2.6.
The wings shown in Figure 9.4 must be interpreted as MAC.
9.2.1 Ground Run

To allow the aircraft to pitch up (nose under-carriage lift-off first), the aerodynamic
lift on the wing must be in the forward position with respect to the CG. The more aft
the CG, the higher the pitching moment. The dynamic equation that provides the
rotation of the airplane around the CG is
M y = Iy q, (9.7)
where M y is the moment of the aerodynamic forces, Iy is the polar moment of inertia
around the spanwise axis y (calculated in § 3.8.3) and q is the pitch rate, q = ω̇ y . With
reference to Figure 9.4, the total rigid-body moment around the CG is
M y = Mow + (Lt xt + Lw xw ) + (Fm xm + Fn xn ) + (μr m Fm zm + μr n Fn zn ) +Tzt (9.8)

% &' ( % &' ( % &' (
lift reaction friction
where Mow is the wing’s pitching moment and the underscores “t” and “w” denote
horizontal tail plane and wing, respectively* ; μr is the rolling coefficient; the remain-
ing quantities are indicated in Figure 9.4. Equation 9.8 does not contain the drag.
Its application point is not determined, so we assume that it lies on the line of the
* This analysis assumes that the aerodynamic coefficients are calculated by normalising with their own
wing area, CLt = Lt /q At , CL = Lw /q A.
xm xn
Lw
xt .
q
Lt xw
+ Wing
CG x
H−tail −
zt T
Nose UC
Main UC
W
Fm
Figure 9.5. Aircraft forces on the ground during rotation.
longitudinal axis through the CG. If we solve in terms of the angular acceleration,
we have
d2 α 1 ) *
= M ow + (L t xt + Lw xw ) + (F m xm + Fn xn ) + (μr m Fm zm + μr n Fn zn ) + Tz t ,
dt 2 Iy
(9.9)
dα
= ωy. (9.10)
dt
During the ground run, the longitudinal axis of the airplane has a slight negative
attitude. This contributes to the generation of a small downforce from the horizontal
stabiliser, unless a corrective action is taken. A small downforce from the horizontal
stabiliser improves the rotation of the airplane at lift-off. Due to the rotation of
the airplane around the main landing gears (Figure 9.5), the CG has two additional
rigid-body velocity components:
ur = −ω y xm sin α, vr = ω y xm cos α. (9.11)
Hence, the air speed around the main wing is augmented by ω y xm. The angle of
attack, instead, is increased by

vr
α1 = tan−1 . (9.12)
U − ur
The next problem is to resolve the reaction force on the wheels before rotation.
Because there is no rotation, the sum of all moments should be zero. The total
resistance can be split between main and nose landing gear:
R = μr m Fm + μr n Fn . (9.13)
With reference to Figure 9.4, the equilibrium of a rigid body with all of these vertical
forces is
Fm + Fn = Lw + Lt − W, (9.14)
(Lt xt + Lw xw ) + (Fm xm + Fn xn ) + (μr m Fm zm + μr n Fn zn ) + Tzt + Mow = 0. (9.15)

Shock absorber
Fore wheels
Rear wheels
Figure 9.6. Rotation of main under-carriage at take-off and landing.
The latter equation can be rearranged in a more useful form that shows the unknown
ground reaction forces Fm and Fn :
(xm + μr m zm)Fm + (xn + μr n zn )Fn = −(Lt xt + Lw xw ) − Tzt + Mow . (9.16)
The system of Equation 9.14 and Equation 9.16 can be cast into matrix form as
follows:

1 1 Fm Lw + Lt − W
= = .
(xm + μr m zm) (xn + μr n zn ) Fn −(Lt xt + Lw xw ) − Tzt + Mow
(9.17)
The solution of this system is straightforward, although for some load conditions
there is no solution. A few points are worthy of note. First, the contribution of the
wing pitching moment is normally “nose-down”. Therefore, an unusually large Mow
would prevent a load relief on the nose landing gear and thence would increase the
ground run. Second, if the line of action of the engine thrust is below the CG, it helps
with the rotation and thus contributes to relieving the load on the nose landing gear.
When the airplane starts rotating, Fn = 0; the total rolling resistance is reduced to
μr m Fm.
A more accurate model should be used if the main under-carriage has more than
one axle. In this case, at take-off the rear wheels are the last ones to lift off and the
first to touch down, as also shown in Figure 9.6. In both cases the rear wheels sustain
the largest loads.
9.2.2 Rolling Coefficients

The calculation of the rolling resistance in the take-off model requires the determin-
ation of the coefficients μr m and μr n (main and nose landing gear, respectively).
In general, there are three contributions to consider: 1) the decelerating force
created during the tyre roll, as a consequence of the mechanical deformation of
the tyre (μr R); 2) the effects of ground contamination (μr F ); and 3) the braking
forces (μr B). We assume that these effects are independent; thus, we can write
an equation such as:
μr = μr R + μr F + μr B. (9.18)
Practical methods for estimating the rolling resistance on a variety of road conditions,
including the effects of speed, are available from ESDU 05011 6 , which proposes
semi-empirical methods for dry, wet, slush and snow conditions. For example, the
rolling coefficient of a single tyre on a dry paved runway would be:

(1 − s)V 2 F 1/3
μr = ζo + ζ1 , (9.19)
2g p/ pa
where s is the braking slip ratio (s = 0 without braking), F is the normal load on
the single tyre, p/ pa is the inflation ratio: tyre pressure divided by atmospheric
pressure. The constant coefficients are ζo = 3.7699 · 10−3 N−1/3 , ζ1 = 4.60824 · 10−5
N−1/3 /s. Because Equation 9.19 shows an explicit relationship among normal load,
tyre pressure and rolling coefficient, the rolling resistance is a variable quantity that
depends on the aircraft loading as well as the runway conditions. For example, if
the main landing gears have nw wheels, then F = Fm/nw must be introduced in
Equation 9.19. The total resistance of the main landing gear is then found from
μr m Fm if the runway is dry or wet; if the runway is covered in snow, Fm must account
only for the leading tyres. Thus, if each under-carriage has four wheels in tandem,
then the rolling resistance would be μr m Fm/2. The second retarding effect, due to
contamination, is discussed separately in § 9.8.
9.3 Solution of the Take-Off Equations

The solution of the take-off problem is achieved by integrating the system of ordinary
differential equations (ODE). A suitable method of integration is a fourth-order
Runge-Kutta procedure with a constant time step. There are three segments to
consider: 1) ground run with all wheels on the runway to the point of nose wheels lift-
off; 2) ground run to the point of lift-off (all wheels off the ground); and 3) airborne
phase to the height of the screen. The system consists of eight differential equations:
du 1
= [T cos( + α + γ ) − Dcos γ − Lsin γ ] , (9.20)
dt m
dv 1
= [T sin( + α + γ ) − Dsin γ + Lcos γ − W] , (9.21)
dt m
dx
= u, (9.22)
dt
dh
= v, (9.23)
dt
dm
= −ṁ f , (9.24)
dt
dω y ) *
Iy = Mow + (Lt xt + Lw xw ) + (Fm xm + Fn xn ) + (μr m Fm zm + μr n Fn zn ) + Tzt ,
dt
(9.25)
dα
= ωy, (9.26)
dt
with
v
γ = tan−1 . (9.27)
u
Before lift-off, γ = 0. This system of equations is closed with Equation 9.17. When
the aircraft is airborne, the rolling resistance and the normal load terms disappear.
The aircraft can still be treated like a rigid body in the vertical plane. This body
has three degrees of freedom and is described by three differential equations
similar to the climb phase. Three more aspects of the take-off problem must be
considered:
1. AERODYNAMIC FORCES. The aerodynamic forces in the system of equations are

lift and drag. These forces are dominated by the position of the control surfaces
and the ground effect. Each flap/slat setting is associated to an index ISF = 0, 1, . . ..
For each value of the setting ISF , we have a well-defined flap and slat angle (δ f ,
δs , respectively), which depend on the aircraft. The calculation procedure for the
aerodynamic terms is the following:
r Set a tentative value of I and determine the angles δ , δ .
SF f s
r Calculate the lift and drag effects of δ , δ following the methods in § 4.1.3 and
f s
§ 4.2.5.
r Calculate the ground effects with the methods shown in § 4.1.2.
The aerodynamic forces are calculated at the air speed U, which is different
from the ground speed whenever there are atmospheric winds (Equation 9.6).
2. NET THRUST. The net thrust is delivered in equal amounts by a number of engines
ne . Cases when this does not occur include OEI (discussed separately) and small
imbalances in the fuel flow. The flight control system must be able to correct any
thrust imbalance; otherwise the consequences could be catastrophic. For the case of
all engines operating (AEO), we use
T = FN ne . (9.28)
At each time step, the net thrust FN is calculated numerically from the engine deck
(see Chapter 5).
At the brake-release point the thrust is increased rapidly. The normal procedure
is 1) brakes off, 2) stick forward about halfway, and 3) stick forward to full thrust
once the engines have stabilised. The engine thrust increases rapidly. Most airplanes
are not designed to operate with brakes on whilst the engines are spooling up. From
a flight simulation point of view it is sufficient to have a ramp function such as
FN = To tanh t, (9.29)
where To is the rated take-off at static conditions and t is the time, calculated from
brake-release. The thrust stabilises to the rated thrust within 3 seconds.
3. LONGITUDINAL TRIM. A major problem is the aircraft trim. If this is not done, it is
quite likely that the numerical solution causes the aircraft to pitch up too fast, and
the aircraft would enter a dangerous, if not impossible, trajectory. This happens if the
estimate of the pitching moment and/or the pitch moment of inertia is inaccurate. On
modern airplanes there is a control system called α-floor protection, which prevents
250 15 3 15
KTAS Ground roll
Climb rate Altitude from ground
Distance from brake release, km

2.5
200 12 12
Climb rate, m/s

2
Altitude, m
150 9 9
KTAS
1.5
100 6 6
1
50 3 3
0.5
Lift-off Lift-off
0 0 0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time from brake-release, s Time from brake-release, s
(a) (b)
600 5
Initial transient 4.9

500
Net thrust, kN
Fuel flow, kg/s
4.8
400
4.7
300
4.6
200 4.5
0 10 20 30 40 50 60 0 10 20 30 40 50 60
(c) (d)
Figure 9.7. AEO take-off analysis; BRGW = 128,500 kg; headwind = −2 m/s; h = 50 m;
take-off thrust (no bleed); flap δ f = 20 deg; xCG = 25% MAC; standard day (part I).
the aircraft from entering a flight trajectory with a too-high attitude. Many transport
airplanes have a rotation rate α̇ = ω y = 2 to 4 degrees/s. One possible solution is to
assume that the aircraft is trimmed at all times. Because
dα̇ My
α̈ = = (9.30)
dt Iy
if α̈ is above a limit value, the pitching moment is corrected. This is equivalent to

adding a damping function to Equation 9.25. At the end of the rotation we force the
airplane to proceed on a linear trajectory.
9.3.1 Case Study: Normal Take-Off of an Airbus A300-600 Model

In Figures 9.7 and 9.8 we show a calculation of the AEO take-off of a model Airbus
A300-600 powered by CF6-80C2 turbofan engines. The data plotted include the
distance from brake release, the ground speed, the true air speed (KTAS), the climb
0.1 1000
Main LG Main LG
Nose LG Nose LG
0.08 800
Landing Gear Loads, kN

Rolling coefficients
0.06 600
0.04 400
0.02 200
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
(a) (b)
Figure 9.8. AEO take-off analysis (part II).
rate, the rotation speed, the loads on the landing gears, the net thrust, the total rolling
resistance, the lift-over-weight ratio and the fuel burn. The calculation ends when
the aircraft has passed the screen. The graphs show the aircraft trajectory (x, h), its
velocity components (u, vc ), the net thrust, the fuel flow, the rolling coefficients and
the rolling resistance. Note the short time between nose lift-off and airplane lift-off.
Several parametric studies can be carried out to examine the effects of all of
the relevant parameters, including brake-release gross weight, airfield altitude, wind
speed, atmospheric temperature, runway gradient and runway conditions (six inde-
pendent parameters). This is in fact done in the W-A-T charts (weight-altitude-
temperature) that are part of the FCOM. One such chart is shown in Figure 9.9.
Calculations have been carried out over the full range of BRGW, at selected
altitudes on a standard day, and are compared with some reference data published
3.5
0 ft
0 ft
Airbus Data
4,00
Calculated
2,00
FAR takeoff runway length, km
S/L
2.5
Figure 9.9. Weight and altitude effects
on take-off of an Airbus A300-600
model; standard day; no wind; δ f = 20 2
degrees; CG at 25% MAC.
1.5
1
100 120 140 160 180
BRGW, ton
by Airbus. A fixed flap setting has been used. When all of the effects are taken into
account, these charts can become quite complicated. In some cases the FCOM shows
the take-off performance in a series of tables rather than charts. A further study of
take-off performance is in Chapter 15.
9.3.2 Effect of the CG Position on Take-Off

Each airplane has a certified CG range for ground roll and take-off, as already
discussed in Chapter 3. The CG position has an important effect on the aircraft
loads, during both the ground run and the airborne phase. The extreme cases are as
follows:
r Aft CG. There are at least four important reasons why a too-aft CG position
is inconvenient. First, it reduces nose-gear adherence and, in extreme cases
it can lift the nose and cause a tail strike. Second, for airplanes with engines
mounted below the wings, there is a further pitch-up effect caused by the thrust
force being located below the CG; the stalling speed is reduced and the take-off
speed V2 increases. Third, an aft CG reduces the ability to steer the aircraft
whilst taxiing. Finally, lateral control at low speeds must be secured by the nose
wheels because the vertical tail and rudder do not have the necessary dynamic
pressure to generate lateral control forces (see discussion in § 9.6).
r Forward CG. A too-forward CG position makes it more difficult for the airplane
to rotate (pitch up); thus, it increases the take-off run. Furthermore, this CG
position causes higher structural loads on the nose tyres, increased heating and
higher wear (as discussed in Chapter 14). When determining the aircraft take-
off performance, the calculation is always performed at the most forward CG
position.
An aircraft trimmed for take-off would have to take into account these effects.
A parametric study of the effect of CG position on the take-off performance of a
transport airplane is shown in Figure 9.10. This graph shows the change in take-off
distance as a function of CG position (in percent MAC) for three different BRGW.
This airplane has several versions with CG slightly different limits. For the correct
data, please refer to the FCOM* . The results clearly indicate that by moving the CG
aft (by increasing the % MAC), the take-off length decreases, by an amount that
depends on the BRGW. This analysis can be repeated at different altitudes, with the
effects of head and tail winds and cold and hot atmospheres.
9.3.3 Effect of Shock Absorbers

The effect of the shock absorbers (the nose under-carriage at take-off and the main
under-carriage at landing) can be introduced in the dynamic model of the airplane to
describe more precisely the performance before and after nose-gear lift-off. In fact,
as the dynamic pressure builds on the wing and the aircraft is trimmed to take off,
the load on the nose under-carriage gradually decreases, and the aircraft undergoes
* Airbus recommends an aft-shift of the CG by about 2% to improve the take-off performance, with
a CG variable between 25% MAC and 30% MAC for many of its aircraft.
1.2 m = 52,820 kg
m = 61,350 kg
m = 69,890 kg
-82 m
1.1
Take-off distance, km 1
0.9
-56 m
0.8
0.7
-35 m
0.6
20 25 30 35 40
CG position, % MAC
Figure 9.10. Calculated effect of CG position on the take-off run of an Airbus A320-200
model; standard day, no wind, sea level (do not use for flight planning).
a small upward rotation whilst on the ground. This increase in attitude contributes
to the increase in wing lift. Nose-gear lift-off occurs when the shock absorber has no
load. Thus, the main change in the model is the relaxation of the shock absorbers,
as indicated in Figure 9.11. The model is equivalent to a spring-damper in parallel.
The spring stiffness is κ and the damping is D. If m = F/g is the virtual mass arising
from the normal load F on the landing gear, then the system would have a natural
√ √
oscillation ωo = κ/m and a damping ratio ζ = D/2 mκ.
√
By using the critical damping Dcrit = 2 κm, the damped response is a return to
the undisturbed position in minimum time. The response of the system is
x(t) = xo(1 + ωot) exp(−ωot), (9.31)
where xo is the initial displacement under load F and ẋo = 0. The normal load is
generally known, and the maximum deformation of the unit can be found from
Figure 9.11. Shock-absorber model on nose landing

gear. k
landing-gear specifications or set to a reasonable limit. The stiffness is unknown, but

it can be inferred by simple analysis. For example, assuming a response time of 1
second (or another value) from Equation 9.31, we calculate the value of the stiffness
from the known limit deformation xo. It can be verified that the magnitude of the
stiffness is about one third of the maximum virtual mass. When the shock absorber
has been unloaded, the change in attitude is α xo/wb. Even in the worst possible
case, the increase in attitude is less than 1 degree.
A more accurate analysis of landing-gear behaviour in the ground run is a com-
plicated matter. Pacejka 7 produced comprehensive models of tyre mechanics that
can be used for analysis such as slip conditions under braking, torsional effects and
shimmy dynamics. The latter term refers to self-induced oscillations that resemble
a shopping trolley. These are oscillations of the wheels axis around the axis of the
main strut 8 – 10 . Shimmy dampers are sometimes used (Airbus A320, Boeing B737)
to overcome these vibrations.
9.4 Take-Off with One Engine Inoperative

The OEI take-off performance belongs to the category of risk analysis indicated in
Figure 9.3. Human factors are involved as well because a decision must be made to
either accelerate and take off or stop and abort the flight. The task is to carry out an
analysis for the case when one engine shuts down at some point during the ground
run. The scenario is limited to the loss of thrust, but it can be expanded into the
causes of the shut-down and what may happen in more extreme cases, such as fuel
leak.
The governing equations are the same as the AEO case, except that the inop-
erative engine creates asymmetric loads. With reference to the latter point, the
yawing moment must be restored by a combination of corrective actions. This is
only possible if the failure has occurred and has been detected at speeds above the
minimum control speed. The increase in aerodynamic drag is due to a combination
of 1) inoperative engine, and 2) rudder deflection:
CD = CDe + CDr udder . (9.32)
The engine contribution in Equation 9.32 is calculated from Equation 7.44. The
rudder deflection is required to stabilise the airplane when one engine shuts down.
The calculation of this quantity is made difficult by the interaction between tyre
performance on the ground 7 and the aircraft control system. However, if the rudder
deflection ξ is known, then the corresponding increase in induced drag is estimated
from

CDr udder = CLξ ξ ξ = CLξ ξ 2 , (9.33)
where CLξ is the lift-curve slope of the vertical-tail-plus-rudder combination. This

term is added only at speeds above the VMCG.
We now consider the case of decelerate-stop (abort) and accelerate-go (take-off)
and calculate the balanced field length.
9.4.1 Decelerate-Stop
When take-off is to be aborted following one engine failure, a sequence of events
takes place:
r Engine failure occurs at time t , when the aircraft has a speed V and has run a
1 1
distance x1 from brake release. These are called critical quantities.
r There is a time lag, t , between engine failure and the decision to abort the
lag
flight, due to the natural reaction of the pilot (human factors). This time is
usually ∼3 seconds. When the clock indicates a time t D = t + tlag , the aircraft
has reached a speed VD at a position xD from brake-release point.
r At the decision point, the thrust is cut, the brakes are applied and the aircraft is
prepared to a braking configuration by deployment of the inlay spoilers, which
also contribute to dumping the lift. Alternative braking methods are possible.
r The aircraft decelerates; when it is halted, it reaches the so-called stopping
distance.
A model is required for the transient thrust, both for the failed engine and the
running engine. It is not possible to make general statements because the transient
response depends on both engine and type of failure. For the running engine, we
need a model for the transient thrust when the throttle is cut back to zero. For the
running engine, we can define a cut function
Fcut (t) = ent , (9.34)
where n is an arbitrary power that defines the transient behaviour and t is the time.
Given the time to at which the net thrust has decayed to zero, the exponent can be
found from solving Equation 9.34 in terms of n. For modern high bypass turbofan
engines, n −1.5.
When one engine shuts down, the aircraft has an asymmetric thrust for the time
tlag + to. The yawing moment due to the combination of inoperative engine and
engine operating at full thrust is
Mz = Tyb + De yb, (9.35)
where yb is the engine’s arm, T is the thrust of the operating engine and De is the drag
of the inoperative engine. This yawing moment may be balanced by the tyre-runway
contacts and by the rudder. The total yawing moment is
Mz = T(t)yb + De yb − μsm Fm − μsn Fn − CLξ ξ, (9.36)
where μs is the friction coefficient between tyre and runway for sideways forcing
(when the tyres are in fact forced to skid laterally). It cannot be guaranteed that
during the transient the resulting moment is zero.
There remains the question of whether the pilot hits the brakes at the same time
as he or she cuts the thrust. A further time delay can be included in the model to
take into account this synchronisation. ESDU 1;2 considers the effect of the time lag
on brakes, spoilers and engine response. A summary of average time delays is given
in Table 9.2.
Table 9.2. Delay in response time after

activation for selected systems
System Delay[s] Comment
Pilot response ∼3
Throttle cut 3–8
Fuel cut 2–4
Prop pitch change 2–4
Spoiler deployment 0.5–1.0
Wheel brake 0.5 Manual
0.1–0.2 Automatic
9.4.2 Accelerate-Stop
The alternative possibility is to continue the acceleration with the remaining func-
tioning engine, provided that the aircraft has been trimmed and stabilised both
longitudinally and laterally.
It is possible that the specified flap setting is not capable of securing enough
aerodynamic lift. In this case, there is a need to increase the flap deflection. Two
cases are possible: 1) the lift is not enough for take-off, and 2) the lift is not enough
to climb with the minimum gradient specified by the international regulations. This
event is dependent on the gross take-off weight.
Computational Procedure for Accelerate-Stop

r Estimate the rotation speed, V = 1.1V .
R S
r Bracket the interval V < V < V with V = 0.25V , V = V .
1 crit 2 1 crit 2 crit
r Guess the critical speed from V = (V + V )/2.
crit 1 2
r Calculate the AEO take-off up to V .
crit
r When the aircraft has reached V , do the following:
crit
1. Continue acceleration with engine failure till the aircraft has cleared ref-
erence screen. The take-off distance is xto1 .
2. Decide to stop. Cut down thrust, hit the brakes, deploy the spoilers, dump
lift and calculate the stopping trajectory xto2 (these operations are done
with time delays).
r Compare the accelerate-stop distances. If the stopping distance x is greater
to2
than the acceleration distance xto1 , then set V1 = Vcrit ; or else set V2 = Vcrit .
r Repeat the iterations with new V till a satisfactory convergence is achieved.
crit
The result of this iterative procedure is shown in Figure 9.12. The right scale
shows the variation of the critical speed Vcrit . The convergence criterion is a difference
of 5 metres between OEI take-off and OEI abort. A smaller tolerance serves only to
increase the number of iterations and does not guarantee a better result. No reverse
thrust is used in the calculation.
Figure 9.13 shows the simulation for an Airbus A300-600 model. The data in
the graph are the ground speed and true air speed from the brake-release point to
the stopping point on the runway. The graph shows the AEO and OEI runs, the
3 80
Distance from brake release, km 2.5 OEI take-off 70

distance
V crit, m/s
2 60
Stopping
distance
1.5 50
Abort & Stop

Accelerate & Go
Critical Speed
1 40
1 3 5 7 9
Iteration count
Figure 9.12. Accelerate-stop analysis of ground speed for an Airbus A300-600 model; sea
level, standard day, no wind.
100 200
Ground speed, m/s
74
72
80
VD 150
Vc
Ground speed, km
70
28 30 32 34 36
60 time, s
KTAS
100
40
50
AEO OEI
20
Stop
Accelerate Brake
0 0
0 20 40 60 80
Time from brake release, s
Figure 9.13. Accelerate-stop analysis of ground speed.
accelerate and stopping sections. The detail on the top right of the graph is a closer
look at the ground speed around the critical time. A number of time delays have
been included in the model (delay for braking action, for spoiler deployment, and
so on), but the change in speed is still relatively sharp.
Figure 9.14 shows the behaviour of some key parameters as a function of time
from brake release for the aborted operation described in Figure 9.13. The data
shown are ground speed, distance from brake-release, drag coefficient, net thrust
and rolling resistance, normal load on the landing gears, lift as a fraction of the
weight and fuel burn. The sharp decrease in speed is due to the stopping method. In
fact, the response of the spoilers and the brakes are assumed to be instantaneous.
9.5 Take-Off of Propeller Aircraft

The take-off equations of the propeller-driven aircraft are formally the same,
although there are some important corrections to take into account. First is the
fact that propeller-driven aircraft are governed by shaft power rather than thrust.
Because the thrust is required in the dynamic equations, we need a correlation
between net thrust and shaft power. We start from
T = Tp + Fg , (9.37)
where Tp is the net propeller thrust and Fg is the residual thrust provided by the
engine exhaust. The engine throttle is set to take-off power. The propeller is trimmed
to the required shaft power. Once this is done, we have the state of the propeller,
which is defined by the set of non-dimensional parameters CT , C P and J :
C P = f (U, h, ϑ, rpm, Pshaft ), CT = f (U, h, ϑ, rpm, Pshaft ). (9.38)
With Equation 9.38, we have the required CT , which is then used to calculate the
propeller’s thrust. At the required Pshaft , the engine state provides the residual thrust
Fg . At take-off conditions, this quantity is of the order of 10 to 15% of the propeller’s
thrust.
If we use this approach, we do not need to make assumptions on the propeller’s
efficiency. In summary, the computational procedure is the following:
r Set the operational conditions.
r Set the engine to take-off shaft power.
r Trim the propeller at the required shaft power and calculate CT .
r Calculate the net thrust from the trimmed propeller, Tp .
r Solve the engine problem at the required shaft power and calculate Fg .
r Calculate the net thrust from Equation 9.37.
r Proceed as in the case of jet-powered aircraft.
Some numerical difficulties can be encountered in this process. For example, the
maximum thrust does not always correspond to maximum power, especially at low
speeds. Thus, we must decide whether to operate the propeller at maximum thrust
or maximum power. In most cases, the rpm is fixed and the propeller is required
to operate at its nominal speed. However, sometimes it is allowed to have a small
9.5 Take-Off of Propeller Aircraft 243
2.5
AEO Take-off
OEI Take-off
80
Distance from brake release, km

V
V 2
Stop
Accelerate-Go
Ground speed, m/s
60
Engine
Failure 1.5
Decision
40 to Abort
1
AEO OEI
20
Stop 0.5
0 0
0 0.5 1 1.5 2 2.5 0 20 40 60 80
Distance from brake release, km Time from brake release, s
(a) (b)
0.14 1.5 600 60

Total Drag Coeff. Net Thrust
Total Lift Coeff. Rolling resistance
50
Stop
0.135 1
Rolling resistance, kN
400 40
Net thrust, kN
CD
CL
Stop
0.13 0.5 30
200 20
Effect of
0.125 Idle Engine 0
10
Engine failure
0.12 -0.5 0 0
0 20 40 60 80 0 20 40 60 80
Time from brake release, s Time from brake release, s
(c) (d)
1200 1 200
Stop
Stop
Load on under-carriage, kN
0.8
150
800
Fuel burn, kg
0.6
L/W
100
0.4
400
50
0.2
Main UC Lift / Weight

Nose UC Fuel burn, kg
0 0 0
0 20 40 60 80 0 20 40 60 80
(e) (f)
Figure 9.14. Accelerate-stop analysis; BRGW = 141,000 kg; headwind = −2 m/s; altitude =
50 m; take-off thrust (no bleed); flap δ f = 20 degrees; xCG = 25% MAC; standard day.
15 180 15
Distance KTAS
1.5 Altitude Climb rate
10 120 10
Altitude , m
vc , m/s
1
X, km
KTAS
5 60 5
0.5
0 0 0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
(a) (b)
4.6 60 200
Shaft power Tyre load, MLG
Net thrust Tyre load, NLG
150
F = kN
4.5 50
FN , kN
P, MW
100
4.4 40
50
4.3 30 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
(c) (d)
Figure 9.15. Simulation of AEO take-off of the ATR72-500 with PW127M engines: wind =
−2 m/s; altitude = 50 m; δ f = 20 deg; xCG =25% MAC; m = 22,616 kg; standard day
(part I).
over-speed, which can have a considerable effect on the resulting thrust. Finally, the
trim process can be numerically slow; therefore the take-off of the propeller aircraft
is computationally more intensive.
Figures 9.15 and 9.16 show the solution of the take-off problem for the
ATR72-500 aircraft, powered with two PW127M turboprop engines and Hamilton-
Sundstrand F568-1 propellers. The data shown in Figure 9.15 are a) distance and
altitude above the airfield; b) air speed and climb rate; c) shaft power and net thrust;
d) normal loads on main and nose landing-gear tyres; e) total lift coefficient and
lift-to-weight ratio; and f) rolling coefficients for main and nose tyres. The case data
are indicated in the caption. Furthermore, in Figure 9.16d we show the residual
thrust from the turboshaft engines and its value relative to the total thrust. The data
clearly indicate that the residual thrust is about 10 to 15%; as anticipated, it is not
a negligible portion of the propeller thrust and must be accounted for in a proper
engineering analysis.
1.3 1.5 0.04

CL Rolling Coeff. MLG
Lift/weight Rolling Coeff. NLG
0.03
1.2 1
μr
L /W
CL
0.02
1.1 0.5
0.01
1 0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
(a) (b)
0.535 6
Fuel flow Residual thrust
Fuel burn 35 17
Residual thrust, % of total
0.53 5.9
25 15
mf, kg/s
mf , kg/s
Fg , kN
Fg , %
15 13
0.525 5.8
5 11
0.52 5.7
0 10 20 30 40 50 60 0 10 20 30 40 50 60
(c) (d)
Figure 9.16. Simulation of AEO take-off of the ATR72-500 with PW127M engines. Data as
in Fig 9.15 (part II).
RESPONSE OF A FAILED PROPELLER. The analysis carried out for a jet engine applies
in general terms also to the turboprop aircraft. However, following an engine failure,
the response of the aircraft is different. Unless an automatic corrective action is taken
on the failed engine-propeller, the drag on that side of the aircraft would greatly
increase. The procedure is to set the propeller to feathering position as quickly as
possible. This operation can take a few seconds. The rate of pitch change is about 15
to 25 degrees/s.
9.6 Minimum Control Speed

With reference to the twin-engine aircraft shown in Figure 9.17, the yaw moment
balance around the point C (centre of the main landing gear) is
Mvt = Tbt − Yn wb, (9.39)

Main
Yn wheels
bt
C
wb xvt
Nose
wheels
LVT
Engine Out
Figure 9.17. Lateral forces and moments for VMCG calculation.
where Mvt is the yaw moment created by the vertical tail as a consequence of a
rudder deflection ξ ; bt is the moment arm of the thrust asymmetry; wb is the wheel
base; and Yn is the side force on the nose wheels in the absence of lateral skidding.
The side force is equivalent to the cornering force if the plane of rotation is aligned
with the ground velocity. All forces and moments are a function of the ground speed.
Consider the rudder in its maximum design deflection ξmax . If the corresponding tail
moment Mvt > Tbt − Yn wb, then lateral control can be achieved with the rudder; if
the tail moment is Mvt < Tbt − Yn wb, then the rudder deflection can be decreased
and lateral control can be achieved by nose-wheel steering.
FAR § 25.149 defines the limits for the determination of VMCG on the ground.
In particular, the regulations require that the aircraft with OEI has a path on the
runway that does not deviate laterally by more than 30 feet (9.15 m) at any point.
The minimum control speed has to be lower than the lift-off speed. The VMCG must
be determined with the aircraft at the most critical condition, the most unfavourable
position of the CG and take-off weight. The moment generated by the vertical tail is
1
Mvt = Lvt xvt = ρU 2 Avt CLξ ξ xvt , (9.40)
2
where Avt is the vertical tail’s area and xvt is the distance between the line of action
of the rudder lift and the centre C of the main landing gear (Figure 9.17). Assuming
ξ = ξmax and solving for the ground speed U = VMCG, the yaw equation becomes
2(Tbt − Yn wb)
2
VMCG = . (9.41)
ρ Avt CLξ ξmax
The solution of Equation 9.41 depends on the limit side force on the nose tyres. A
semi-empirical method for calculation of this force is given by ESDU 11 , although it
requires the yaw angle of the tyre 7 . We will consider the case of an unbraked wheel.
For a given value of the normal load on the tyre, Fn , we are interested in finding
the limit side force (cornering force) on the tyre Yn before the tyre starts skidding
sideways. In practice, before such condition occurs, the tyre will operate in yaw
conditions. This tyre yaw introduces a major complication. In the first place, as
rotation axis
contact area
plane of rotation
Braking
Force
ψ
(side force) Y Yc (cornering force)

n
Figure 9.18. Cornering force on tyre.
the aircraft accelerates it gains lift and unloads the nose gear. As the nose gear is
unloaded, its cornering force decreases. Thus, unless the dynamic lift grows quickly,
the velocity VMCG will be inconveniently high. In fact, it is possible that there is a
range of ground speeds at which it is not possible to control the aircraft, either by
nose-gear steering or rudder deflection. It is necessary to decide whether the nose
tyres are operating in yaw because the cornering force depends on this angle.
Now assume that the tyre’s yaw angle ψ is sufficiently small (Figure 9.18). The
ratio between these two forces is the lateral friction coefficient
Yn
μs = . (9.42)
Fn
From the analysis of data published by Dreher and Tanner 12 on a number of tyres
under dry, wet and flooded runways, the sideways-friction coefficient can be written
as
μs = μs c1 (V, Fn )ψ + c1 (V, Fn )ψ 2 . (9.43)
A suitable computational procedure consists in iterating at increasing air speeds,

as follows:
r Set the maximum rudder deflection ξ = ξmax .
r Fix the true air speed, U = V ± Uw .
r Calculate the yaw moment Tbt − Yn wb.
r Calculate the rudder moment Mvt .
r If Mvt < Tbt − Yn wb, increase the true air speed.
r If Mvt < Tbt − Yn wb, the corresponding air speed is the VMCG; exit the loop.
An example of VMCG calculation is shown in Figure 9.19 for the conditions

specified in the caption. The result indicates that VMCG < Vlo, as expected. The
4,000 300
Tyre yaw = 4 degs V-tail & Rudder
Engine moment
Nose wheel
Nose wheel load
Normal nose gear load, kN

3,000
Moments, kNm 250

VMCG
VLO
2,000
200
1,000
0 150
40 60 80 100
KTAS
Figure 9.19. Calculated VMCG for a reference aircraft; m = 145,000 kg, standard day, no wind,
dry runway.
calculation assumed that the maximum rudder deflection is 10 degrees and that the
aircraft drifts laterally by 30 feet (9.15 m) over the distance of 1,000 m.
The result depends on several factors, including conditions of the runway, tyre
tread (ribbed or smooth), wind speeds and direction (headwind or tailwind), position
of the CG, tyre yaw angle, maximum rudder deflection and so on. The method
described does not take into account roll effects.
For a four-engine aircraft, we consider the case of only one engine inoperative.
The worst-case scenario is when one of the outer engines fails. The remaining steps
of the analysis would be the same.
EFFECTS OF DERATED THRUST. A few additional clarifications are required with

respect to the derated thrust. A derated thrust causes a reduction on the minimum
control speed on the ground and a change in take-off speeds. A reduced take-off
thrust must not result in loss of any functions and systems operations* . In the case
of one engine inoperative (OEI), a derated thrust reduces the destabilising yawing
moment. This moment has to be balanced by the vertical tail that is capable of
providing control only at speeds above the VMCG. A lower yawing moment leads
to a lower VMCG; a reduced VMCG leads to a reduced accelerate-stop distance.
For each derated level, there must be a certified VMCG.
9.7 Aircraft Braking Concepts

Aircraft braking on the ground is achieved by a combination of systems. The primary
system consists of wheel brakes; the secondary system is based on aerodynamics, in
* Federal Aviation Regulations: Reduced and Derated Takeoff Thrust (Power) Procedures, AC-25-
13, April 1988.
9.7 Aircraft Braking Concepts 249
V
V
Slip Contaminant
Normal load
(a) (b)
Figure 9.20. Tyre slip (normal operation) and aquaplaning.
particular spoilers. Finally, engine thrust reversers can be used. Each system can
be characterised by a mathematical equation in terms of the relevant parameters
(aircraft weight, position of the CG, ground speed, runway conditions).
MECHANICAL BRAKES. The aircraft can be brought to a halt by using appropriate

wheel brakes. The overall resistance is the sum of the rolling resistance of the
wheel–tyre combination. The larger the load on the wheel, the higher the braking
force. However, an important factor is the tyre–runway speed (or skid). A free-
rolling wheel has a rotational speed equal to the aircraft’s speed. A fully locked
wheel has no rolling speed. At intermediate condition the tyre skids on the runway.
The ratio r/V between the rotational speed of the tyre and the aircraft’s speed is
the so-called slip ratio.
The slip between the tyre and the roadway is shown in Figure 9.20. At normal
operating conditions, the tyre slip is equal to the ground speed (100% slip-ratio, or
free-rolling speed); when there is partial loss of contact, the tyre slip is lower and
causes skidding. The limiting case is when the tyre slip is zero; that is, the tyre is locked
(the slip-ratio is zero). The friction force depends on the slip-ratio: a free-rolling
tyre has zero-resistance and hence zero friction coefficient and zero braking action.
Clearly, this is not a good situation because it would be difficult to stop the aircraft
with mechanical brakes. Conversely, a skidding tyre has a poor braking performance
in addition to considerable risks of damage and explosion. Wheel-lock is prevented
with anti-locking (or anti-skid) systems. These systems compare the speed of each
wheel with the rolling speed of the aircraft. The maximum braking performance is
at slip-ratios between 10 and 15%. FAR § 23.109 recommends maximum braking
coefficients for wet conditions in the form
μb = c1 V 3 + c2 V 2 + c3 V + c4 , (9.44)
in which the coefficients are dependent on the tyre pressure p.
AERODYNAMIC SPOILERS. For a transport aircraft, braking is achieved by dumping

lift and mechanical brakes on the wheels. Lift is dumped from the wing by a rapid
deployment of the spoilers. When the spoilers are deployed, the wing lift becomes
negative and the centre of pressure moves aft; the net effect is to increase the appar-
ent weight and the rolling resistance by effectively creating a downforce. Thus, the
calculation procedure must include a correction for the distance between aerody-
namic centre and CG. It is convenient that this distance be reduced (see Figure 9.4
and Equation 9.16) because it forces more load on the main tyres rather than the
nose tyres. If this does not occur, the load on the nose tyres can be excessive and in
extreme cases these tyres can overheat and burst.
THRUST REVERSERS. Thrust reversers are engine systems that revert the direction
of the propulsive thrust. The resulting braking force is not dependent on the con-
ditions of the runway. Thrust reversers can create very large braking forces and
cause the aircraft to slow down with relatively high deceleration rates. Their use is
recommended on contaminated runways. A notable side effect is the considerable
acoustic emissions they generate. Consequently, their use is regulated; the aircraft
must demonstrate appropriate deceleration and stopping in the absence of thrust
reversers.
9.8 Performance on Contaminated Runways

There is a variety of conditions for paved runways. Each condition affects the rolling
resistance of the aircraft and the braking efficiency. The official classification in order
of increasing severity includes damp, wet, standing water, slush, wet snow, dry snow,
compacted snow and icy surface. Investigations on rolling-friction coefficients have
been carried out over the years. See, for example, Wetmore 13 , Harrin 14 , Yager 15
and Agrawal 16 .
The runway is contaminated when more than 25% of the runway surface area to
be used is covered by water more than 3mm deep, or by slush or loose snow equivalent
to at least 3mm of water. A damp runway contains small amounts of humidity. A wet
surface has a thin layer of water that makes it shiny; however, this water is not deep
enough to cause hydroplaning. Standing water is generally localised to areas where
drainage is poor after a rain storm. Slush is water mixed with snow at temperatures
above freezing. Wet snow is found after fresh precipitation; it is light-weight and
can be easily shaped. A dry snow is present at colder temperatures and dry weather;
compacted snow has been compressed by heavy machinery. In icy conditions, the
rolling resistance is lowest (μr < 0.005). Specifically, braking performance degrades
with slippery roadways; take-off acceleration and lateral control decrease.
Under wet conditions, the friction coefficient between the tyre and the road is
lower than normal because the contaminant cannot be pushed out by the tyre-road
contact; consequently there is a reduced contact and a reduction of the traction force.
The contact between tyre and contaminant does not generate a useful tractive force.
To increase the friction coefficient, it is necessary to have tyres capable of squeezing
the contaminant out of the contact area. In practice, this can be done at relatively
low speeds, but as the speed increases, the time available for the contaminant to be
squeezed out decreases and the roadway becomes more slippery, with potentially
catastrophic consequences.
An extreme case is hydroplaning also shown in Figure 9.20. Hydroplaning is
that condition in which the tyre is no longer in contact with the runway but travels
over a deep layer of water. The water film between the tyre and the runway reduces
the rolling resistance. Under these conditions, braking becomes ineffective and dir-
ectional control is lost. Additional problems that are incurred in this event include
the risk of spray ingestion into the engines, the flaps, the sensors and other delicate
aircraft systems.
A rule-of-thumb that is used to determine the friction coefficient under wet con-
ditions is to take half the value of the friction coefficients on dry runways. However,
this rule does not satisfy the requirements for safety, and new methods have been
developed that take into account the tyre inflation pressure, its wear state, the type
of runway and the anti-skid systems.
9.8.1 Contamination Drag

In the presence of loose runway contamination (water, slush and snow), an additional
form of drag appears. This drag is due to two factors: the contamination drag due to
the effects of displacement of the contaminant by the tyres (§ 9.8.1) and the resulting
spray impingement on the airframe (§ 9.8.2). The former contribution is always larger
than the latter. The maximum value of the contamination drag increases with the
roll speed up to a maximum, and then it decreases.
Figure 9.21 shows the effects of contamination on the airframe due to spray
plumes lifting from the nose and main tyres. In particular, at point A we have the
effect onto the fore fuselage due to the nose tyres. The same plume hits the lower
wing, inboard of the engine (B). It would be more damaging if this plume expands
over a wider angle to impinge on the engine intakes. The effects of the nose spray
plume on the central fuselage are shown by point C. Finally, the effect of the main
landing gear on the rear fuselage shows impingement of the spray on the horizontal
stabiliser (D) and on the lower fuselage (D).
For a single tyre, ESDU 17 gives the following expression for the displacement
drag:
⎧
⎨ GF1,1 = 12 ρw ς V 2 CD1,1 bs d, V/Vp ≤ 1
(9.45)
⎩
GF1,1 = fLGF1,1 (V/Vp = 1), V/Vp > 1
where bs is the tyre width at the intersection with the contaminant, d is the tyre
diameter, CD1,1 is the drag coefficient due to displacement of the contaminant, and
ς = ρc /ρw is the ratio between the density of the contaminant and that of the water
(ρw ). The factor fL accounts for the effects of ground speed V relative to a reference
speed Vp . The latter parameter is defined as
1/2
p
Vp = 34.28 , (9.46)
ς
where the pressure is given in bar and the reference speed Vp is in knots. The
displacement drag of twin tyres, GF2,1 , is given by
⎧
⎨ GF2,1 = f2,1 GF1,1 , V/Vp ≤ 1
. (9.47)
⎩
GF1,1 = fLGF2,1 (V/Vp = 1), V/Vp > 1
(a)
(b)
Figure 9.21. Effect of spray plumes from contaminated runways onto a transport aircraft.
The displacement drag of dual-tandem tyres GF2,2 (four tyres) is estimated from the
sum of two components:
GF2,2 = GF2,1 + GFi , (9.48)
where the first contribution is recognised as the displacement drag of twin tyres
(the leading pair and associated mechanical systems). The second contribution is an
impingement contribution on the trailing tyres, defined as
GFi = fi GFi (V/Vp = 1), (9.49)
with
GFi (V/Vp = 1) = ρw ς Vp2 bs d. (9.50)
Both factors fL and fi are given in Ref. 17 in graphical form. An example of calcu-
lation is shown in Figure 9.22, which shows the breakdown of the retarding forces
due to standing slush having a constant depth of 1 cm (0.01 m) for an Airbus
A320-200 model. The landing-gear contribution refers to the sum of two separate
15
Total drag
Nose L-gear
Main L-Gear
12
Contamination drag, kN
9
0
0 50 100 150
Ground speed, kt
Figure 9.22. Contamination drag of a transport airplane (calculated).
under-carriages. The tyre deformation was assumed as 2.5% of the tyre diameter
in both nose and main under-carriage (in lieu of a full aircraft trim), with tyre
inflation pressures equal to 190 and 110 psi (pounds-per-square inch), respectively
(∼1.31 and ∼0.76 MPa). The contamination drag increases to a maximum at about
120 kt, and then it decreases.
9.8.2 Impingement Drag

A first-order estimate proposed by the Joint Aviation Regulations (JAR) for the
displacement drag is
1
Ddispl = ρc Styre V 2 KCD, (9.51)
2
where ρc is the density of the contaminant (water, slush, snow); Styre is the tyre’s
frontal; area within the contaminant; K is the wheels coefficient (as discussed in
§ 4.2.7); and CD = 0.75 for both water and slush.
The impingement drag depends on the complex interaction between the geo-
metry of the airframe and the fluid dynamics of the displaced contaminant. The
current practice is to perform tests on the runway with various levels of contamin-
ation for certification purposes. However, a body of research is available to guide
in the estimation of this effect. ESDU 18 has basic methods for the estimation of
the shape of the plumes generated by the contact between tyre and contaminant,
the displacement drag 17 , as well as the estimation of the impingement drag 19 .
On the basis of a set of tests with a number of different aircraft with both standing
water and snow, Giesberts 20 reported that Equation 9.51 is not very accurate. In
particular, it is shown that the precipitation drag increases up to a ground speed, and
then it gradually decreases. Equation 9.51 predicts a uniformly increasing drag. An
engineering method based on particle dynamics has been proposed by Gooden 21

to predict the precipitation drag on runways contaminated by standing water. This
model includes the effects of tyre pressure and geometry, pool height and other
factors.
Following the ESDU method, the skin-friction drag due to the m−th impinge-
ment of the spray onto the airframe is
GFm = CFm qSwetm , (9.52)
where CFm is a skin-friction coefficient, q is a dynamic pressure, and Swet m is the area
affected by spray impact (fuselage, wing, nacelle, and so on). If k = Vspray /V denotes
the ratio between the spray speed and the ground speed, the dynamic pressure is
calculated from
⎧
⎨ q = 1 ρ ς V 2 centre and side plumes
o w
q= 2 . (9.53)
⎩
q = qo(1 − k2 ) forward spray
The calculation of the areas affected by the spray impact requires two ingredients:
an appropriate definition of the airplane geometry (CAD format or other useful
formats) and an estimate of the spray envelopes as they travel downstream. If these
data are available, it is possible to describe a numerical algorithm that calculates the
intersection between the airframe geometry and the spray envelope.
The last term required in Equation 9.52 is the average skin-friction coefficient
in the area affected by spray impact. Because the air flow is most likely turbulent
and the airframe is nearly flat relative to the size of the spray particles, we can
use well-known expressions for the turbulent skin friction over a flat plate. One of
these expressions is the equation of the skin-friction coefficient CF in incompressible
turbulent flat plate (Kármán-Schoenerr):
1
√ = 4.3 log10 (CF Rex ) . (9.54)
CF
In Equation 9.54 Rex denotes the Reynolds number based on the reference length
x: Re = ρUx/μ, with μ and ρ the dynamic viscosity and the density of clean water,
respectively. The reference length is taken as the maximum length of the m−th area
affected by spray impact. Simplifications can be applied at this point. For example,
the use of the root chord cr (which is readily available) is not too different from the
length of the spray impingement on the wing. Finally, note that the mean CF from
Equation 9.54 is added to the skin-friction drag occurring in normal conditions, that
is, due to air flow around the airplane.
9.9 Closed-Form Solutions for Take-Off

Closed-form solutions are obtained with a number of simplifications. They are useful
for a first-order investigation of the field performance and represent methods that
are useful in conceptual design. In fact, there is some value to be gained from
approximations such as lumped mass, parabolic drag equation, constant thrust and a
number of other assumptions. We consider the cases of the jet and propeller aircraft
separately.
Table 9.3. Average rolling coefficient

for some runway conditions
Runway Condition μr
Dry concrete/asphalt 0.02

Hard turf & gravel 0.04
Short & dry grass 0.05
Long grass 0.10
Soft ground 0.10–0.30
9.9.1 Jet Aircraft

The ground run is the distance between brake-release and lift-off of the forward
wheels. To calculate the ground run, we write the dynamics equations on the centre
of gravity of the aircraft in the horizontal and vertical directions, for a take-off from
a horizontal runway. The engine thrust is aligned with the vector velocity. Thus, we
have:
∂u
m = T − D − R, (9.55)
∂t
R = μr (L − W), (9.56)
where R is the ground resistance. Some average data for rolling-resistance coeffi-
cients are given in Table 9.3.
The acceleration of the aircraft can be written as

∂u ∂u ∂ 1 2
a= =u = V . (9.57)
∂t ∂x ∂x 2
With this definition, Equation 9.55 becomes
1 ∂ V2
m = T − D − R. (9.58)
2 ∂x
Integration of Equation 9.55, performed between brake-release (x = 0, t = 0) and
the lift-off point, yields
x
1
mVlo2 = (T − D − R) dx. (9.59)
2 o
Integration of the right-hand side of Equation 9.59 requires additional information

as to how the aircraft forces depend on the speed. For example, the front wheels raise
from the ground before the rear wheels. When this occurs, there is a step change in
the rolling resistance.
First, consider the lift-off speed. This speed can be estimated from the stall
speed. It is allowed to have Vlo 1.1 VS , which implies that CLlo 0.83CLmax ; this is
a reasonably safe margin. FAR Part 25 (§ 25.103) defines the stall speed as
VCLmax
Vstall ≥ √ , (9.60)
n
where n is the load factor in the direction normal to the flight path; VCLmax is
the calibrated airspeed obtained when the load-factor-corrected lift nW/q A, is a
maximum. The regulations prescribe the methods that must be used to calculate the
VCLmax . A stall warning must be clear at all critical flight conditions. Therefore, the
lift-off speed is
2W 1
Vlo2 = . (9.61)
ρ A CLlo
The take-off time is obtained by integration of the speed,
xlo
dx
tlo = . (9.62)
0 V
SIMPLIFIED SOLUTIONS. A first-order approximation is to take an average value of the

thrust, drag and rolling resistance over the ground run. Under these circumstances,
the result of the integration is
1
mVlo2 = T − D − R xlo. (9.63)
2
Suitable values of the parameters in Equation 9.63 are:
2
1 Vlo 1
T = To, D = ρ ACD , R = μr (W − L) μr W. (9.64)
2 2 2
By replacing the equivalences in Equation 9.63, we find

1 1 1
mV = To − ρ ACD Vlo − μr W xlo,
2 2
(9.65)
2 lo 8 2
that solved in terms of xlo yields
4mVlo2
xlo = . (9.66)
8To − ρ ACD Vlo2 − 4μr W
For a more refined solution, consider the thrust independent of the speed and equal
to the static thrust. Express the drag force as
1
D= ρ A(CDo + kCL2 )V 2 , (9.67)
2
where the CL and CD are relative to the ground configuration of the aircraft. At
this point we note that in the aircraft’s ground configuration the control surfaces are
extended.
Returning to the analysis of the momentum equation, the rolling resistance is

1
R = μr W − ρ ACLV , 2
(9.68)
2
with W = constant. The solution is
x
1 1
mVlo2 = (To − μr W)xlo − ρ A (CDo + kCL2 + μr CL)V 2 dx. (9.69)
2 2 o
The CL to be introduced in Equation 9.69 is a function of the angle of attack of the

aircraft (see § 4.1), but not the speed. A suitable approximation is that these surfaces
are set to take-off positions, and so the lift coefficient in ground effect is constant.
Therefore, Equation 9.69 becomes

x
1 1
mVlo2 = (To − μr W)xlo − ρ A(CDo + kCL2 + μr CL) V 2 dx. (9.70)
2 2 o
The last step in solving the dynamics equation is the knowledge of the speed V(x).
If an acceleration a(V) = ao + c1 V 2 is assumed, then it can be demonstrated that if
the total acceleration during the ground roll does not exceed 40% of its initial value
at lift-off, the “exact” and “constant” acceleration solutions for the ground distance
differ by about 2% – by all means an acceptable error from an engineering point of
view. By using a constant acceleration, the relationship between the speed and time
is quadratic. In fact,
∂u ∂u ∂ x 1 ∂ V2
a= = = = const. (9.71)
∂t ∂ x ∂t 2 ∂x
√
Integration of this leads to V 2 c x, bar a constant of integration. To match the
lift-off conditions {xlo, Vlo}, the relationship between speed and ground run must be
Vlo2
V 2 (x) = x. (9.72)
xlo
With this result, the solution of the integral term in Equation 9.70 is
mVlo2 /2
xlo = . (9.73)
(To − μr W) − ρ A(CDo + kCL2 + μr CL)Vlo2 /4
The final method of solution of the take-off equation consists in using the actual
thrust and drag. The engine thrust is expressed by polynomial functions, that is,
functions such as
T(U) = To(1 + c1 U + c2 U 2 ), (9.74)
with ci constant coefficients. Note that the thrust depends on the airspeed U, not the
ground speed V. The take-off Equation 9.59 will be written as

∂V 1 x
= (T − D − μr R) dx. (9.75)
∂t m o
The best method of integration relies on a fourth-order Runge-Kutta method with

variable time stepping, which is stable and accurate. The stopping criterion is the
lift-off point that we assume occurs when L = W.
ROTATION AND INITIAL CLIMB. The lift-off point is reached with at least one landing
gear on the ground. The rotation of the aircraft must be done with a small angle or
else there is a risk of a tail strike on the runway. Tail strike is not unusual with a wide-
body aircraft, and it has potentially serious consequences. Figure 9.23 shows that the
tail strike depends on the geometrical configuration of the aircraft, in particular the
landing-gear height hg and the distance between the centre of the wheels and
the strike point, l1 . Pinsker 22 proposed the solution to a number of problems that
Strike Point
Figure 9.23. Tail strike at the take-off flare; the limit rotation angle depends on the distance
between the point of contact with the ground and the main landing gear.
occur at lift-off, including conditions for tail strike and minimum and ground clear-
ance required for simultaneous banking and pitching.
At the start of the lift-off there is an abrupt change in rolling resistance. The
airborne phase starts with a further rotation of the aircraft. Then the aircraft climbs
along a straight path in order to fly past the reference screen. During flare the aircraft
has a centripetal acceleration
V2
n= , (9.76)
χ
where χ is the radius of curvature of the flight path. The distance on the ground run
during this phase will be x3 χ γ , and the height reached will be
γ2
h1 = χ (1 − cos γ ) χ , (9.77)
2
where γ is the climb angle. A first-order estimate for a commercial subsonic jet
indicates that h1 1 m (3 feet) and the distance covered is of the order of 12 m
(∼40 feet), which is negligible in most cases. The initial climb is done with a constant
climb angle, so that
h − h1 h
xa = . (9.78)
tan γ tan γ
The total take-off distance will be
xto xlo + xa . (9.79)
An alternative calculation of the airborne phase can be done with an energy method.
The change in total energy from lift-off to the point of clearing the screen is
1
Wh + m V 2 − Vlo2 = (T − D)xa , (9.80)
2
where u is the minimum airborne control speed. Therefore, the airborne distance is

Wh + m V 2 − Vlo2 /2
xa = . (9.81)
T−D
There are a number of other cases of interest. Take-off under icing conditions has
been studied extensively. One practical example is van Hengst 23 , who also provides
some lift curves with and without de-icing fluids for the Fokker 50 and Fokker 100.
The take-off performance with iced surfaces requires at least the knowledge of the
correct aerodynamic coefficients.
The take-off of seaplanes and flying boats is complicated by the additional
resistance of the water and the water waves. Essentially, we need to find a good
estimate of the resistance of the aircraft. This is the sum of the aerodynamic and
hydrodynamic drag (see also § 4.7). Relevant studies on this subject are those of
Perelmuter 24 and Parkinson et al. 25 DeRemer 26 showed take-off measurements
from shallow lakes of the Cessna 180.
9.9.2 Propeller Aircraft

The method of calculation follows closely that of the jet aircraft, although there are a
number of complications due to replacement of the jet thrust with a propeller thrust,
which depends on a number of operational parameters (see Chapter 6). A low-order
solution is found with the balance of forces along the horizontal direction is
∂V P
m = η − D − R, (9.82)
∂t V
or

∂ 1 2 1 P
V = η − D− R , (9.83)
∂x 2 m V
where both the engine power and the propeller efficiency depend on the aircraft
speed. We make a number of simplifying assumptions. If the power term in the
right-hand side of Equation 9.83 is much larger than the sum of the other two, then

∂ 1 2 η P
V . (9.84)
∂x 2 mV
In addition, if we take an average engine power and propeller efficiency, then we
have a first approximated value for the ground run by integrating Equation 9.84,

1 2 η P
d V = dx, (9.85)
2 mV
and finally*
1 m 1
xlo . (9.86)
4 Vlo2 η P
This equation can be improved by replacing the proper values of the aerodynamic
drag and the rolling resistance. The result is

∂ V2 1 P 1
= 2η − ρ ACDVlo2 − μr W . (9.87)
∂x m V 4
Equation 9.87 is solved for the lift-off distance xlo and yields
mVlo2
xlo = . (9.88)
2η P/V − ρ ACD Vlo2 /4 − μr W
+
* For the integration use the equivalence V 2 = z, V = z1/2 , which leads to z1/2 dz = z−1/2 /2.
We finally consider the case in which the drag and the rolling resistance change with
the speed. The governing equation can be written

1 ∂ V2 1 P
= η + c1 V 2 − μr W , (9.89)
2 ∂x m V
where the coefficient c1 is
1
c1 = − ρ A(CDo + kCL2 + μr CL). (9.90)
2
We consider CL a constant parameter in the ground run. Integration of Equation 9.89
yields

2 xlo η P
Vlo =
2
+ c1 V − μr W dx.
2
(9.91)
m o V
This equation can only be solved if the propeller’s efficiency and the engine power
as a function of the speed are known. The problem is to determine if the aircraft is
operated with a variable pitch/fixed rpm or vice versa.
9.10 Ground Operations

Ground operations include taxiing out of the gate and back to the gate once the
aircraft has landed. In a typical taxi operation, the aircraft moves at a low speed
along a paved ground, performs a turn and sits idle, whilst waiting for air traffic
control to give a go-ahead. Taxi times are greatly dependent on the airfield and
on airfield congestion. The fuel requirements for a taxi operation are calculated as
follows:
m f = m f1 + m f2 + m f3 , (9.92)
where the first term denotes the fuel burned during taxiing at a constant speed
utaxi ; the second term denotes the fuel burned in idle mode; and the last term is
the fuel required to accelerate the aircraft from rest to roll speed. If the aircraft is
halted more than once, m f3 must account for the number of accelerations. Solution
of Equation 9.92 requires the distance from the gate to the brake-release point and
the total taxi time – data that depend on the airport, the position of the aircraft
around the terminal building and other factors. However, if these data are known,
the calculation is done from
) x *
m f1 = ṁ f , (9.93)
V taxi
m f2 = (ṁ f t)idle (9.94)
with
) *
ṁ ftaxi = f j T = f j μr (W − L) − D , (9.95)
x
tidle = ttaxi − . (9.96)
V taxi
Table 9.4. Estimated fuel burn during a taxi-out: V = 5 m/s; X = 2.0

km from ramp to brake-release; all fuel burn in [kg]; times in minutes;
sea-level airfield; standard day; 6,500 km (3,510 n-miles) mission.
A380-861-GP B747-400-GE

Parameter 20 10 20 10
Fuel rest-to-roll 24.1 24.1 23.3 23.3

Fuel in idle mode 960.0 240.0 652.8 163.2
Fuel in roll mode 381.9 376.6 391.9 402.0
Total taxi fuel 1365.9 640.4 1,068.0 588.9
Time in idle 16.7 6.7 16.7 6.7
Time in roll 3.3 3.3 3.3 3.3
The acceleration fuel is calculated from

mV
m f3 = ṁ f , (9.97)
R
where R is the total resistance (rolling plus aerodynamic drag); the fuel flow ṁ f is
calculated by solving the engine problem with T R at the prescribed speed, airfield
altitude and atmospheric conditions.
The method proposed is only approximate because it does not take into account
effects such as one-engine-out taxi and asymmetric thrust during a turn. Some man-
ufacturers occasionally recommend operating the aircraft with one engine out for
part of the taxiing in order to conserve fuel. This recommendation comes with some
caution in case of heavy aircraft, severe weather conditions and large turns on the
taxiway. Important side effects of this procedure include 1) higher risk of loss of
braking capability and nose-wheel steering; 2) engine start-up failure (which would
require return to the gate); and 3) higher jet blast of the operating engine. The latter
problem is discussed in Chapter 14.
A summary of taxi-out fuel burn for some large aircraft is in Table 9.4 for two
different taxi times (10 and 20 minutes), for a fixed distance from the gate to the
brake-release point. We have assumed that the aircraft does not stop at intermediate
points, but the result can be expanded to include these effects, as well as the effects of
turning (see following discussion). In any case, the fuel burn required is staggering.
Therefore, large aircraft should be given queue priority at departure.
9.10.1 Ground Manoeuvring

Before reaching the point of brake-release and after completing the landing oper-
ation, transport aircraft must taxi out and back to the gate. In doing so, they must
manoeuvre slowly with the engines running at low efficiency. Ground manoeuvres,
including steering and turning, are the source of considerable fuel waste, as none of
these operations is done efficiently. An example of turning is shown in Figure 9.24
for the A380, adapted from Airbus* .
Other ground operations include moving the airplane in reverse from the gate
and moving the airplane to the gate. The operation is done by ground support
* Airbus A380 Airport Planning Manual, Oct. 2009.
R 45.7 m
45.7 m
7m
NOMINAL OFFSET
COCKPIT TRACK
R 25.9 m
6.62 m
NLG PATH
WLG PATH
TAXIWAY GUIDELINE
4.99 m
22.9 m
Figure 9.24. Airbus A380 turn from taxi-way to runway, 135 degrees turn.
equipment; the second one is often a manoeuvre done by the pilot. The manufac-
turers provide data and charts on ground manoeuvres, including minimum radius
of turn and procedures to steer the aircraft, including nose steering and the use of
asymmetric thrust.
9.10.2 Bird Strike

Birds and aircraft engines do not get along too well. It is not unusual that birds
enter into a collision course with an aircraft, something that is commonly defined
as a bird strike. Sometimes the problem is so serious that it leads to an accident or
aborted flight. Data from the Flight Safety Foundation 27 indicate that there have
been more than 52,000 bird strikes on commercial aircraft in the decade ending
in 1998. Analysis of these data indicates that the most likely strike points are the
engines (41%), the nose and the windshield (41%), the wing (7%) and the fuselage
(7%). Only 15 to 20% of these strikes cause damage and require a turn-back. In
the most severe cases, engine shut down is required at take-off. Approximately
50% of bird strikes into an engine cause damage, resulting at least in fan-blade
damage, which causes an immediate increase in vibration and increase in exhaust-

gas temperature. The damage increases with the engine thrust setting. There are
two important parameters to consider in the collision: the mass of the bird and the
relative speed U. The impact energy is E = mU 2 /2.
Bird-strike frequency increases as the airplane is close to the ground (take-off,
climb-out, final approach and landing). At least half of these strikes take place on
the ground and about 80% at altitudes below 500 feet (∼150 m) from the airfield.
Removing the birds from the airport area is the responsibility of the airport‡ .
Aside from the research on aircraft and engine response to a bird strike, con-
siderable research has gone into understanding birds’ behaviour in the proximity
of the airfield. In fact, most of the strikes occur at take-off and landing (90% of all
strikes), although strikes at higher altitudes are not uncommon. The introduction of
higher by-pass turbofan engines may cause higher accident rates, due to the large
mass flows required by these engines and their lower noise emission. As a result, the
birds do not have much advance notice to avoid the advancing aircraft.
The most widespread practice is to make the airfields unfriendly to birds. An
active protection method consists in landscaping the airfield with feeding areas (short
grass and water bodies) and areas where birds are unlikely to loiter (tall grass). By
contrast, sometimes lakes and ponds are covered with plastic balls that make birds’
landing impossible; vegetation that produces seeds is removed from the area.
These days, more technological help is already at hand. Bird radars have been
developed and tested at major airports and are due to be installed at control towers.
These radars are programmed to peg the birds’ flight parameters (altitude, direction,
speed) and the birds’ species. This is important because geese are recognised as more
dangerous than other birds (they are relatively big and fly in large flocks). Gulls seem
unable to avoid collisions, whilst red-tail hawks learn to stay away from airplanes’
flight paths. The radars are also programmed to identify soaring flight and flapping
wings. Birds that flap fast generally do so for a short time and then glide. To be
accurate, these radars have to work on the triangulation principle; three radars are
more effective than two. Bird-strike prevention strategies recommended by Airbus
include:
r To keep aircraft lights on at altitudes below 10,000 feet to help the birds locate
the aircraft.
r If bird strike is encountered on take-off and the speed is greater than the decision
speed V1 , the flight must continue.
r If bird strike is encountered on take-off and the speed is less than V but greater
1
than 100 kt, the take-off must be aborted, to allow inspection of the engines.
r If bird strike is encountered on landing, the aircraft is to fly through the flock
and land, whilst maintaining low thrust setting.
r If bird strike is encountered on landing, reverse thrust must not be used to stop
the aircraft because this operation would cause further damage to the engine.
‡ Various national organisations operate wildlife control and provide advice and recommendations
to the airports and crews. They can be found on the Internet with the keywords “bird strike”. The
first recorded bird strike is due to the Wright Brothers (1908).
Summary
In this chapter, we presented problems of field operations of a transport aircraft
powered by either turbofan or a turboprop engine. Specifically, we solved the prob-
lem of normal take-off and have identified the key parameters that affect the bal-
anced field length: the position of the CG, the weight, the airfield altitude, the wind,
the air temperature and other conditions. We have made a risk analysis with particu-
lar reference to the loss of one engine thrust/power and calculated the accelerate-go
and decelerate-stop distances. In the latter case, we have further considered the
problem of lateral control on the ground with asymmetric thrust. We demonstrated
that control can be achieved with the rudder and vertical stabiliser, provided that
the aircraft has reached a sufficient speed (minimum control speed). For some case
studies we have shown the behaviour of the key flight-mechanics parameters. The
atmospheric effects at the airfield are represented by the wind, by the runway con-
tamination and not least by the risk of bird strike. For completeness, we have shown
some approximate closed-form solutions.
Bibliography
[1] ESDU. Example of Take-off Field Length Calculations for a Civil Transport
Aeroplane. Data Item 87018. ESDU International, London, Oct. 1987.
[2] ESDU. Calculation of Ground Performance in Take-off and Landing. Data
Item 85029. ESDU International, London, Mar. 2006.
[3] ESDU. Force and Moment Components in Take-off and Landing Calculations.
Data Item 85030. ESDU International, London, Nov. 1985.
[4] Powers SA. Critical field length calculations for preliminary design. J. Aircraft,
18(2):103–107, 1981.
[5] Krenkel AR and Salzman A. Take-off performances of jet-propelled conven-
tional and vectored-thrust STOL aircraft. J. Aircraft, 5(5):429–436, Sept. 1968.
[6] ESDU. Comprehensive Method for Modelling Performance of Aircraft Type
Tyres Rolling or Braking on Runways Contaminated with Water. Data Item
05011. ESDU International, London, May 2005.
[7] Pacejka HB. Tyre and Vehicle Dynamics. Butterworth-Heinemann, 2002.
[8] Pritchard JI. An overview of landing gear dynamics. Technical Report TM-
1999-209143, NASA, 1999.
[9] Fallah MS and Bhat R. Robust model predictive control of shimmy vibration
in aircraft landing gears. J. Aircraft, 45(6):1872–1880, Nov. 2008.
[10] Gordon J. Perturbation analysis of non-linear wheel shimmy. J. Aircraft,
39(2):305–317, Mar. 2002.
[11] ESDU. Frictional and Retarding Forces on Aircraft Tyres, Part IV: Estimation
of Effects of Yaw. Data Item 86016. ESDU International, London, Oct. 1992.
[12] Dreher RC and Tanner JA. Experimental investigation of the braking and
cornering characteristics of 30 11.5 × 14.5 type VIII aircraft tires with different
tread patterns. Technical Report TN D-7743, NASA, Oct. 1974.
[13] Wetmore JW. The rolling friction of several airplane wheels and tires and the
effect of rolling friction on take-off. Technical Report R-583, NACA, 1937.
[14] Harrin EN. Low tire friction and cornering forces on a wet surface. Technical
Report TN-4406, NACA, Sept. 1958.
[15] Yager TS. Factors influencing aircraft ground handling performance. Technical
Report TM-85652, NASA, June 1983.
[16] Agrawal SK. Braking performance of aircraft tires. Progress Aerospace Sci-
ences, 23(2):105–150, 1986.
[17] ESDU. Frictional and Retarding Forces on Aircraft Tyres. Part V: Estimation
of Fluid Forces. Data Item 90035. ESDU International, London, 1990.
[18] ESDU. Estimation of spray patterns generated from the sides of aircraft tyres
running in water or slush. Data Item 83042. ESDU International, London, 1983.
[19] ESDU. Estimation of airframe skin-friction drag due to impingement of tyre
spray. Data Item 98001. ESDU International, London, 1998.
[20] Giesberts MKH. Test and evaluation of precipitation drag on an aircraft caused
by snow and standing water on a runway. Technical Report NLR-TP-2001-490,
NLR, Amsterdam, NL, Nov. 2001.
[21] Gooden JHM. CRspray – Impingement drag calculation of aircraft on water-
contaminated runways. Technical Report NLR-TP-2001-204, NLR, Amster-
dam, NL, Oct. 2001.
[22] Pinsker WJG. The dynamics of aircraft rotation and liftoff and its implication
for tail clearance especially with large aircraft. Technical Report ARC R& M
3560, Aeronautical Research Council, 1967.
[23] Van Hengst J. Aerodynamic effects of ground de/anti-icing fluids on Fokker 50
and Fokker 100. J. Aircraft, 30(1):35–40, Jan. 1993.
[24] Perelmuter A. On the determination of the take-off characteristics of a sea-
plane. Technical Report TM-863, NACA, May 1938.
[25] Parkinson J, Olson R, and House R. Hydrodynamic and aerodynamic tests of
a family of models of seaplane floats with varying angles of dead rise – NACA
models 57-A, 57-B, and 57-C. Technical Report TN-716, NACA, 1939.
[26] De Remer D. Seaplane takeoff performance – Using delta ratio as a method
of correlation. J. Aircraft, 25(8):765–766, Aug. 1987.
[27] Anon. Bird strikes found most common at low altitude in daylight. Flight Safety
Foundation Digest, 19(2):14–16, 2000.

Symbols listed in Table 9.1 are not repeated here.
a = acceleration
A = area or wing area
b = width of tyre contact on roadway
bs = tyre width in Equation 9.45
bt = moment arm of engine thrust
ci = constant factors
cr = root chord
c1 = factor in sideways friction coefficient, Equation 9.43
CDe = engine-drag coefficient
CDg = drag coefficient in ground effect
CF = skin-friction coefficient
CLg = lift coefficient in ground effect
CLmax = maximum lift coefficient
CLξ = lift-curve slope of vertical tail due to rudder deflection
CP = power coefficient of a propeller
CT = thrust coefficient of a propeller
d = diameter
De = drag of inoperative engine
D = damping, Equation 9.31
Dcrit = critical damping
De = idle engine’s drag
E = energy
f (..) = functional relationship
fi = factor in impingement drag of dual-tandem tyres
fL = factor in displacement drag of leading tyres
F = normal load on individual tyre
Fcut = thrust-cut function, Equation 9.34
Fg = residual thrust in turboprop engine (also Fg9)
Fm = normal load on main under-carriage
Fn = normal load on nose under-carriage
FN = net thrust from a single engine (also FN)
GFi = skin-friction drag due to i-th impingement on airframe
GF1,1 = displacement drag function, Equation 9.47
GF2,1 = displacement drag of twin tyres, Equation 9.47
GF2,2 = displacement drag of twin-tandem tyres, Equation 9.48
ISF = index denoting flap and slat setting
Iy = pitch moment of inertia of airplane
h = geometric altitude
h1 = height at the end of the flare
J = advance ratio of a propeller
K = wheels coefficient, Equation 9.51
k = ratio between spray speed and ground speed, Vp /V
k = ratio between spray velocity and airplane velocity, Vspray /V,
Equation 9.53
L = lift
m = mass
M = Mach number
Mow = pitching moment of a wing
My = pitching moment
Mz = yaw moment
n = power factor in Equation 9.34
n = normal-load factor in Equation 9.60
ne = number of engines
nw = number of wheels/tyres
p = pressure
q = pitch rate, q = ω̇ y
r = radius
R = rolling resistance
Re = Reynolds number
Rex = Reynolds number based on length x
s = braking-slip ratio
Swet = wetted area (in spray-contamination analysis)
Styre = frontal area of a tyre
t = time
tD = decision time
tlag = lag time
T = net thrust or total thrust
To = rated take-off thrust, static conditions
u, v = aircraft velocity components
ur , vr = rotational velocity components, Equation 9.11
U = air speed
VCLmax = airspeed corresponding to maximum CL, Equation 9.60
Vcrit = critical speed
V = ground speed
Vlo = lift-off speed
Vp = reference speed in spray analysis
Vspray = speed of contaminant spray
w = tyre width
wb = wheel base
W = weight
x = distance travelled
xa = airborne distance at take-off, Equation 9.81
xo = initial displacement of a spring
xCG = longitudinal position of the CG
xm = distance between main UC and CG on longitudinal axis
xn = distance between nose UC and CG on longitudinal axis
xo = spring displacement
xt = distance between aerodynamic centre of H-tail and CG
xw = distance between aerodynamic centre of wing and CG
y = vertical distance
yb = engine arm
Yc = cornering force
Yn = side force
z = dummy variable
zm = vertical distance between centre of main wheel and CG
zn = vertical distance between centre of nose wheel and CG
zt = vertical distance between line of thrust and CG
Greek Symbols
α = attitude or angle of attack
α1 = change in inflow angle of attack due to rotation, Equation 9.12
δs = slat angle
γ = flight-path angle
γr = runway incline
δf = flap angle
= thrust vector
ς = ratio between density of contaminant and water
ζ = damping ratio
ζo, ζ1 = coefficients in Equation 9.19
η = propeller efficiency
ϑ = propeller pitch
κ = spring stiffness
μr = rolling coefficient
μs = sideways rolling coefficient
μb = braking coefficient on runway
ξ = rudder deflection
ρc = contaminant density
ρ = density
ρw = water density
χ = radius of curvature of a trajectory
ψ = tyre’s yaw angle
ω = rotational speed
ωo = natural frequency of harmonic oscillator
ωy = dα/dt = angular velocity
[.] = average quantity

[.]a = atmosphere
[.] B = braking
[.]crit = critical quantity
[.] f = fuel quantity or flap
[.] F = effect of contaminant
[.]g = in-ground effect or gas quantity
[.]lo = lift-off
[.]m = main landing gear; spray impingement index
[.]n = nose landing gear
[.]o = reference quantity
[.] p = propeller or propulsive
[.]r = rolling
[.]t = tyre or tail
[.]to = take-off
[.] y = with respect to y-axis
[.]w = wing or wind
10 Climb Performance
Overview
The aircraft climb includes a variety of flight problems in which the airplane usually
(not always) gains altitude. We define the general governing equations for flight
in the vertical plane and thence some closed-form solutions for the propeller and
jet airplane (§ 10.2). We then present the general problem of climb by a transport
aircraft and show numerical solutions for both types of airplanes (§ 10.3), including
the case of one engine inoperative (§ 10.3.6). The case of the turboprop aircraft is
discussed in § 10.4. A powerful method for dealing with aircraft climb, particularly
at transonic and supersonic Mach numbers, is the total energy approach (§ 10.5).
There is a wide range of climb problems addressed with the energy methods. We
report a few of these cases in § 10.6.
Accelerated climb problems are exclusively the domain of numerical solutions.
Some of these methods are mathematically involved and will not be discussed in
sufficient detail. The assumption of quasi steady flight is valid for many conventional
aircraft.
KEY CONCEPTS: Closed-Form Solutions, Climb to Initial Altitude, Energy Method,

Specific Excess Power, Optimal Climb, Climb Trajectories.
10.1 Introduction
There are two methods for solving climb problems: by solution of the differential
equations that govern the motion of the centre of gravity and by the use of energy
methods. There is a difference in the climb characteristics of propeller- and jet-
driven aircraft. Although modern flight programs routinely include a turn during
climb-out and descent, we will restrict the discussion to a vertical plane. There is no
unique solution to a climb problem because there can be several free parameters
and different types of constraints.
The rate of climb is the aircraft’s velocity normal to the ground. As the aircraft
climbs, the power plant delivers less thrust. The aircraft will reach a point where it
can no longer climb; when this is the case, the aircraft reaches the absolute ceiling.
Rates of climb are often given in the technical literature as feet/minute. It is possible
269
for the manufacturer to provide the maximum instantaneous climb rate in such a
unit, although this represents a peak value that may not be maintained for a full
minute.
In some operations, the aircraft can zoom past the absolute ceiling by trading
its kinetic energy into potential energy (zoom climb). Past the absolute ceiling, the
aircraft might not be able to sustain controllable flight. Zoom-climb maximisation
of the F-4C and F-15 aircraft was carried out in the 1970s for stratospheric missions
reaching 27,000 m (∼88,600 feet). The maximum known rates of climb are around
18,000 m/min (300 m/s; ∼984 feet/s).
10.2 Closed-Form Solutions

Closed-form solutions are obtained by algebraic manipulation of the state equations
and are useful in a preliminary analysis. We introduce these methods before address-
ing the more realistic (and complicated) flight trajectories of a real-life airplane. We
consider separate cases for the jet- and propeller-driven airplane.
10.2.1 Steady Climb of Jet Airplane

Assume that the thrust angle is zero. The governing equation for the centre of gravity
of the aircraft in arbitrary flight is given by
∂U
T − D − W sin γ = m . (10.1)
∂t
Next, multiply this equation by the aircraft speed U, to find
∂U
TU − DU − Wvc = mU . (10.2)
∂t
The climb velocity in general accelerated flight is
T−D U ∂U
vc = U− . (10.3)
W g ∂t
We need to define three important concepts. First is the specific excess thrust
SET = (T − D)/W, which represents how much thrust is available for accelera-
tion relative to the weight of the aircraft. This quantity is a non-dimensional number
and depends on a number of operational factors, such as weight, altitude and atmo-
spheric conditions.
The second concept is the specific excess power (SEP), which represents how
much power is available for vertical climb, relative to the weight of the aircraft. The
SEP has the dimensions of a velocity and is equal to the climb rate:
T−D
SEP = vc = U. (10.4)
W
The corresponding climb angle is
vc T−D
sin γ = = (10.5)
U W
10.2 Closed-Form Solutions 271
and is equivalent to the SET. Finally, we define the normal load factor as the ratio
between the lift and the weight:
L
n= . (10.6)
W
If we use the normal load factor, the general expression of the climb angle is
T n
sin γ = − . (10.7)
W L/D
As a further simplification, if the thrust is not dependent on the flight speed, then
the maximum angle of climb is obtained with the speed corresponding to maximum
glide ratio, (L/D)max .
We now seek the optimal solution of a steady-state climb at subsonic speeds, for
an airplane whose drag equation is parabolic. We optimise the climb with respect to a
single parameter, the CL. Changes in CL can be achieved by change of configuration
(i.e., by deployment of the high-lift systems) and by changes in the angle of attack.
The fastest climb is ensured by maximum rate of climb. The mathematical condition
is found from
∂vc
=0 (10.8)
∂CL
or
"
∂ T CD 2 W 1
− √ = 0. (10.9)
∂CL W CL ρ A CL
If we replace the parabolic drag in Equation 10.9 and work out the derivatives, we
find
T −1
3CDo CL−2 − C − k = 0. (10.10)
W L
Equation 10.10 is a quadratic equation in the unknown CL−1 . The positive solution is
6CDo
CL = . (10.11)
T/W + (T/W)2 + 12CDo k
Equation 10.11 expresses the optimal CL, which is a function of both aerodynamics
and propulsion and structural factors.
10.2.2 Steady Climb of Propeller Airplane

Assume a propeller aircraft, whose longitudinal axis is aligned with the speed and the
thrust generated by the propellers. Start from Equation 10.2, with zero acceleration.
Recall that the effective power of the propeller aircraft is
TU = η P. (10.12)
The rate of climb of the aircraft is found by rearranging Equation 10.2 with Equa-
tion 10.12. The result is
η P − DU
vc = . (10.13)
W
From the definition of lift coefficient, replace the speed U with

2W
U= cos γ . (10.14)
ρ ACL
Equation 10.14 neglects the centrifugal acceleration but, in this case, it is an accept-
able approximation. By replacing Equation 10.14 into Equation 10.13, we find

η P − DU ηP CD 2 W
vc = = − cos γ 3/2
3/2
. (10.15)
W W CL ρ A
For a small climb angle (γ < 10 degrees), assume that cos3/2 γ ∼ 1. The approxi-
mation of a small climb angle is not required when solving the climb problem with
numerical methods.
FASTEST CLIMB OF PROPELLER AIRPLANE. The fastest climb is a flight program requir-
ing the climb rate to be at a maximum at all altitudes. At a given altitude h Equa-
tion 10.15 is only a function of the aircraft speed. The necessary optimal climb
condition requires that the derivative of the climb rate vc with respect to the relevant
flight parameters will be zero. Assuming that the throttle is at full position, for a pro-
peller engine, the free parameters are the flight speed U and altitude h, the advance
ratio J and the pitch setting θ of the propeller: vc = f (h, U, J, θ ). The climb rate can
be optimised on a point-to-point basis; therefore the altitude can be taken out of
the list. Because both the propulsive efficiency and the shaft power are dependent
on the air speed, a closed-form solution of this problem cannot be found, unless we
make major simplifying assumptions, such as η P = constant. In the latter case, the
propulsive effects are eliminated, and the condition of maximum climb rate is given
by

∂vc ∂ CDo + kc2L
= 0, or , (10.16)
∂U ∂U 3/2
CL
which can then be solved in closed form.
10.2.3 Climb at Maximum Angle of Climb

The angle of climb is given by Equation 10.5. The condition of maximum for this
angle is

∂ ∂ ηP − D
(sin γ ) = = 0. (10.17)
∂U ∂U W
If we assume again that the change in weight is negligible, then

∂ ηP D
− = 0. (10.18)
∂U W L
The optimisation problem proceeds in the same fashion as the preceding case. The
optimal equation is

∂ ∂ D
(η P) − W = 0. (10.19)
∂U ∂U L
If we assign
2W
c1 = , (10.20)
ρ A
then

∂ D CDo c1 k
=2 U− 3 , (10.21)
∂U L c1 U
and the optimal equation becomes

∂ CDo c1 k
(η P) − 2W U− 3 =0 (10.22)
∂U c1 U
or

∂η P ∂P CDo c1 k
+ η − 2W U− 3 = 0. (10.23)
∂J R ∂U c1 U
Propeller performance charts are required to solve this equation. Only by neglecting
the propulsive effects can a closed-form solution be found.
10.3 Climb to Altitude of a Commercial Airplane

The climb procedures of a modern commercial airplane are nowhere similar to
the cases presented so far. These procedures are sometimes called “optimal”, but
often they are not because they involved a series of manoeuvres defined to go around
obstacles, noise constraints and air-traffic management. We discuss the noise-related
strategy separately (Chapter 16). The climb to initial cruise altitude (ICA) of a
transport aircraft is done in several phases, some of which are considered “standard”,
as discussed herein.
10.3.1 Climb Profiles

We describe three typical profiles, named “Standard”, ICAO A and ICAO B. These
procedures apply below 10,000 feet (FL-100). The sub-steps are equivalent to each
other upon reaching a speed of 250 KCAS above 3,000 feet. Whenever accelera-
tion to 250 KCAS is required, it is assumed that the aircraft has not yet reached
that speed (this is not always the case). The thrust cut-back must be such that the
resulting climb gradient is not below the climb gradient obtainable with one engine
inoperative.
Standard Procedure
r Take-off is done at maximum thrust (or power), with an initial climb to 1,000
feet.
r Cut back thrust (or power) and accelerate past the flap-retraction speed.
r Climb out to 3,000 feet at constant KCAS.
r Pitch over and accelerate to 250 KCAS.
r Upon reaching 250 KCAS, climb to 10,000 feet.
2
LG Up AS
Flaps 2 KC
Take-off
C Accel.
Flaps 1
Flaps 0 D
ICA
1
B AS
3,000 ft
KC
800 ft
A
Figure 10.1. Typical climb schedule of transport aircraft.
ICAO A Procedure
r Take-off is done at maximum thrust (or power), with an initial climb to 1,500
feet.
r Cut back thrust (or power) at 1,500 feet.
r Climb at constant KCAS to 3,000 feet.
r Pitch over and accelerate, whilst retracting the flaps, to 250 KCAS.
ICAO B Procedure
r Take-off maximum thrust (or power), with an initial climb to 1,000 feet.
r Accelerate at full power, whilst retracting the flaps/slats.
r At clean configuration cut back thrust (or power).
r Climb at constant KCAS to 3,000 feet.
r Pitch over and accelerate to 250 KCAS.
The actual procedures followed by the airlines are not exactly as described. Further-
more, there are some variants. For example, in the ICAO B procedure the first target
altitude can be 800 feet, 1,000 feet or 1,500 feet. The main operational parameters are
the gross weight, the flap/slat angles and the thrust setting. In Chapter 18 we describe
the NADP1 and NADP2 procedures, which are used for noise-abatement strategies.
One example of climb profile is shown in Figure 10.1. The steps are as follows:
1. The aircraft starts at take-off (A) with state vector V A (using maximum take-off
thrust or appropriate derated thrust).
2. The aircraft performs an acceleration on a linear flight path. At least two options
are possible:
r To reach a target altitude (in this case, 800 feet) above the field with a take-off
thrust (B).
r To reach the target climb speed CAS1 with a take-off thrust. The airplane
state vector at the end of this segment is V B.
3. The aircraft performs a climb at constant CAS1 with a reduced thrust (or power),
to be determined, till it reaches a target altitude of 3,000 feet.
Table 10.1. Approximate limit speeds for selected commercial aircraft
Airplane VFE [kt] Setting VLO [kt] Type
ATR72-500 185 Flap 15 170 retract

150 Flap 30 160 deploy
A320-200 215 Flap 0
200 Flap 1+F
170 Flap 2
160 Flap 3
150 Full
A330-300 240 Flap 0
215 Flap 1+F
205 Flap 2
196 Flap 3
186 Full
B737-800 250 Flap 1
250 Flap 2
215 Flap 5
205 Flap 10
190 Full 15
185 Full 25
165 Full 30
160 Full 35
4. The aircraft performs a level acceleration at 3,000 feet to reach a target CAS2
(the thrust must increase).
5. The aircraft performs a climb with constant CAS2 to reach the target initial
cruise altitude and cruise Mach number (the thrust may decrease in this phase).
6. If a further altitude gain is required, it is done at constant Mach number; if an
increase in Mach number is required (although this is not normally the case),
the aircraft performs a level acceleration.
The cross-over altitude is defined as the altitude at which the constant CAS meets
the constant Mach number curve in the M − h envelope. This altitude depends on the
cruise Mach number and on aircraft parameters, such as gross weight and available
thrust. However, when this phase is required, the TAS and the CAS decrease as the
aircraft climbs to the tropopause.
Because the climb to the ICA shown in Figure 10.1 is done en-route, a longer
climb means a lower cruise segment. Vice versa, a faster climb would require a longer
cruise. The overall optimisation of the fuel burn has to be done in synchrony with
the cruise (see § 12.3).
CONFIGURATION CHANGES. Several events take place in the initial stages of the climb,
and these need further specification. In the first phase, from A to B, the aircraft
changes state several times, with a landing-gear retraction, and then a retraction,
in short sequence, of the high-lift devices. An example of flap and slat settings is
shown in Table 10.1 for selected aircraft. Landing-gear retraction is done as soon as
possible. When this event takes place, the drag decreases sharply by about 50%, and
the thrust required to maintain the TAS also decreases sharply. This event is not
associated to a change in lift. When a change in flap/slat setting takes place, there
is change in both lift and drag, although nowhere as dramatic as in the case of the
landing gear. Slat and flap operation takes place after landing-gear retraction, with a
time lag sufficient for stabilising the aircraft. The flight-mechanics equations include
switches such as
LGearPosition = {up, down} = f (h, U, W). (10.24)
A condition required for landing-gear retraction is that the aircraft has reached a
target speed. High-lift systems retraction is done after the retraction of the landing
gear. For the slat/flap position, we propose the following.
Computational Procedure
r The climb procedure starts with a C = C .
L Lto
r The flap/slat setting is given by an index I = 0, 1, 2 . . ., each index corres-
SF
ponding to a pre-defined configuration of the flaps and slats (depending on the
airplane).
r Within the climb-segment A-B determine the lift C∗ required to sustain a 1-g
L
flight. This approximation is acceptable if the aircraft is on a linear trajectory;
that is, normal load factor n = 1:
2W
CL∗ = . (10.25)
ρ AU 2
The lift is to be provided by the wing with the high-lift devices. The clean wing
lift is CL = CLo + CLα αe , where αe is the effective mean angle of attack of the
wing (presently unknown). The difference
dCL = CL∗ − CL (10.26)
is the difference in CL that must be obtained with the high-lift system. The
possible cases are:
1. If dCL 0, then ISF = 0 (no flap deflection).

2. If dCL < 0, then there is a need for a flap deflection.
3. Set tentatively ISF = 1; calculate the dCLf obtained with the flaps/slats and
check the difference Equation 10.26.
4. If dCL 0, then ISF = 1 is the correct setting; otherwise increase the setting
to ISF = 2 and repeat the procedure until dCL 0 is satisfied.
Each value of the index ISL corresponds to fixed angles δs and δ f (slat and flap
deflection, respectively). To each deflection there corresponds a change in wing lift
and drag, which will be calculated with the aerodynamic model. With this procedure,
we are able to calculate the impulsive change in aircraft configuration,* the corres-
ponding change in aerodynamics and finally the change required on the engine
settings.
There is the possibility that the best ICA cannot be reached, either for distance
limitations or for weight limitations. The first instance occurs when the airplane is
to perform a very short-range service: The en-route climb distance would not leave
time for cruise. In the second case, the large take-off weight would cause the airplane
* In practice, the change in flap/slat setting is not impulsive; it may take three to five seconds, both on
deployment and retraction.
TO distance E
LG Up
Flaps 1 Flaps 0
4
C D
3
1,500 ft
Take-off
BR 2
> 400 ft
A 1 B
VEF VLO V2 S/F retraction Vgreen & MCT

TOGA Clean
Figure 10.2. Flight path of OEI take-off and go-around.
to have a sluggish climb at intermediate altitudes and possibly insufficient climb rates
at higher altitudes. In both cases the target ICA must be decreased to a sub-optimal
value.
10.3.2 OEI Take-Off and Go-Around

Safety requirements determine the conditions that aircraft have to comply at take-
off if one engine fails in the most critical part of the flight. In case of engine failure,
following the procedures described in Chapter 9, if the option accelerate-go prevails,
the aircraft climbs out with the remaining engine power. The key events of this
procedure (also called TOGA, for take-off and go-around) are shown in Figure 10.2
and listed in Table 10.2. The graph shows various events from brake-release (BR) to
the point when the aircraft reaches a target altitude of 1,500 feet above the airfield.
Engine failure takes place with the aircraft on the ground with a speed VEF . The
go-around phase consists of a four-segments climb. Upon reaching the threshold
altitude of 35 feet, the aircraft initiates the climb at point A. The first segment
terminates at B, where the landing gears are retracted and the aircraft flies with
the take-off speed V2 . The second segment terminates at C, where the aircraft has
reached a minimum target altitude of 400 feet. The third segment terminates at D,
where the aircraft has fully retracted its high-lift surfaces whilst accelerating to a
target green-dot speed Vgreen . The final segment terminates at E, where the aircraft
has reached a target altitude of 1,500 feet with the green-dot speed and maximum
continuous thrust. The minimum climb gradients are given in Table 10.2.
10.3.3 Governing Equations

For a jet aircraft on an arbitrary flight path on the vertical plane, the equations of
motions are
∂U
m = T cos(α + ) − D − W sin γ , (10.27)
∂t
∂γ
mU = T sin(α + ) + L − W cos γ . (10.28)
∂t
Table 10.2. Key events in the OEI take-off and go-around procedure
1st 2nd 3rd 4th
Min. OEI gradient, ne = 2 0.0% 2.4% — 1.2%

Min. OEI gradient, ne = 4 0.5% 3.0% — 1.7%
Start VLO LG-retract h ≥ 400 ft Clean
Configuration Take-off Take-off S/F Retract Clean
Engine rating TOGA TOGA TOGA MCT
Speed VLO V2 → Vgreen Vgreen
Landing gear Down Up Up Up
The angle of climb γ is the angle between the flight direction and the horizontal
plane; the angle of attack α is the angle between the aircraft reference axis (zero-lift
axis) and the velocity vector; the thrust angle (vectored thrust) is the angle between
the reference axis and the direction of the engine thrust. This angle is generally small
and can sometimes be neglected. The flight path will be described by the differential
equations
∂x
= Vg cos γ , (10.29)
∂t
∂h
= Vg sin γ (10.30)
∂t
where Vg is the ground speed. The fuel flow is also part of the problem because it
affects the aircraft’s gross weight. The corresponding equation is
∂m ∂m f
=− = −ṁ f . (10.31)
∂t ∂t
The problem is to be closed with a set on initial conditions:
t = 0, U = Uo, γ = γo , x = xo, h = ho, m = mo. (10.32)
The aircraft can climb in an infinite number of ways, but a limited number of climb
programs deserve special word. There are climb programs that include local optimal
conditions and fixed starting conditions. These are typically initial-value problems.
These programs are 1) fastest climb, 2) climb at maximum climb angle, and 3) climb
at minimum fuel. The maximum-angle-of-climb problem is only important to clear an
obstacle in emergency situations, but it is not a normal way of operating the aircraft.
The minimum fuel to climb is also the most economical climb. There are also special
climb programs that require final conditions, such as speed or Mach number at a
given altitude. These problems are called two-value boundary problems.
10.3.4 Boundary-Value Problem

Once the configuration switches have been accounted for, the climb procedure still
depends on a number of parameters. The aircraft state at point E, VE = f (h, M, W)
can be pre-calculated by optimisation of the cruise conditions with the methods
described in Chapter 12. Thus, we are able to establish the final conditions of the en-
route climb. The initial conditions V A are also known from the take-off performance.
We now have a two-value boundary problem, with known conditions at the start and
5 300
Ground Run Trajectory
Take-off Thrust
250
4
Thrust
Flight altitude, km cut-back
C
200
Net thrust, kN
LGear
3 Retract D
150
2
KCAS1
Climb 100
Level
B Accel
1
50
KCAS2
A Climb
0 0
0 2 4 6 8 10
Flight time from take-off, min
Figure 10.3. Simulated en-route climb for the Airbus A320-200-CFM; standard day, no wind.
at the end (V A, VE ), and a number of functional constraints, such as specified thrust

on the first segment and constant KCAS climb on the second and fourth segments.
The free parameters are the thrust level of the first segment (A-B), the effective mean
angle of attack of the airplane in the first segment (A-B) and the second-segment
target climb speed CAS1.
The mean angle of attack will have to be selected in a way to minimise the flap
setting and hence the fuel consumption. The net thrust following take-off can be
maintained to the target altitude or CAS, but often there is no need for such a level
of thrust.
NUMERICAL SOLUTIONS. A typical result for the Airbus A320-200 model with CFM
engines is shown in Figure 10.3. The figure shows the trajectory, with the points A, B,
C, D previously indicated in Figure 10.1. The thrust schedule shows the decreasing
net thrust over the take-off run, the nearly constant thrust in the first segment of
the climb, the thrust cut-back at the start of the second segment and all of the other
events up to 5,000 m (∼16,400 feet) above the airfield. The sharp decrease in thrust
at point D is due to the fact that the airplane converts from acceleration to climb.
Another case, implementing the so-called NADP2 (Noise Abatement Departure
Procedure 2) (see details in Chapter 17), is shown in Figure 10.4. In this case, we
have a thrust cut-back at 3,000 feet with a minimum climb gradient of 4 degrees, up
to a flight altitude of 4,000 feet, when the aircraft restores its normal climb rate.
Table 10.3 is a summary of computed results. The table has five reference points:
0) airplane above the screen at 35 feet; 1) airplane at the end of the initial climb, when
it has reached the first target CAS; 2) airplane at the end of the first segment climb at
constant CAS; 3) airplane at the end of the level acceleration; and 4) airplane at the
top of climb (initial cruise altitude). At the bottom of the table there is a summary
of climb data.
Table 10.3. Climb report for the Airbus A320-200 with CFM56-5C4P turbofan engines and
331-9 APU; standard day, no wind
h t vc vc mf ṁ f
KCAS KTAS M [m] [min] [m/s] [ft/min] [kg] [kg/s]
0 165.1 165.5 0.251 61 [over screen]
1 266.0 307.6 0.482 3,098 2.51 20.25 3,986 506.1 3.362 [const CAS]
2 266.0 307.6 0.482 3,098 0.00 0.00 0 10.1 0.618 [accelerate]
3 266.0 436.9 0.751 10,038 22.77 5.08 1,000 701.3 0.513 [const CAS]
4 265.4 436.6 0.751 10,058 0.10 4.83 950 2.8 0.453 [const M]
Time to ICA 25.4 [min]
Fuel to ICA 1,220.3 [kg]
Distance to ICA 152.0 [n-m]
L-Gear retraction 26 [m]
Flaps retraction 100 [m]
L-Gear retraction time 3.5 [s]
CLIMB OPTIMISATION. The airplane model is the Airbus A320-200 with CFM engines.
The assumptions of this study include the following:
r NADP2 procedure (this is a low-noise procedure fully described in § 18.2)

r Specified BRGW (as indicated)
r Climb target: specified by the best ICA-Mach combination. In this instance we
have climb to FL-310 with a target M = 0.752, which was frozen for all cases.
r Free parameters are KCAS1 and first segment v ; the climb thrust is calculated
c
to satisfy the KCAS1-vc combination.
5 300
Ground Run Trajectory
Take-off Thrust
250
4
Thrust
cut-back
Flight altitude, km
200
Net thrust, kN
LGear C
3 Retract D
150
2
KCAS1
Climb 100
B B1 Level
1
Accel 50
Min Gradient KCAS2
A Climb
0 0
0 2 4 6 8 10
Flight time from take-off, min
Figure 10.4. Simulated en-route climb for the Airbus A320-200-CFM, with NADP2 proce-
dure; standard day, no wind.
25 8
20
51
6
55
52
15
Δmf, %
Δt, %
4
53
53
10
52 vc
50
51
54
2
5 51 52
50
vc
0
0
210 220 230 240 250 210 220 230 240 250
KCAS1 KCAS1
(a) Climb Fuel (b) Climb Time
Figure 10.5. Parametric study of climb to ICA of a model Airbus A320-200; m = 55,200 kg.
The iso-levels denote a constant cost index.
r Climb distance constrained to X

target = 170 km (∼91 n-miles). If the target ICA-
Mach is reached at a distance xc < Xtarget , we add the cruise fuel to the target
distance; if xc < Xtarget , the solution is discarded.
r Standard day; headwinds U = −4 kt measured 2 m above the airfield.
w
The cost index is a weighted average of the cost of fuel and the cost of time.
This index is discussed in § 12.9* . The result of our parametric analysis is shown
in Figure 10.5. The arrows point toward increasing climb rates. A minimum fuel
trajectory requires a high first-segment KCAS1 as well as a low climb rate; a minimum
climb time requires a high first-segment KCAS1 as well as a high climb rate. The
darker shades refer to lowest cost index, that is, most economical climb.
The second-segment KCAS2 is kept constant to the value corresponding to
the best ICA-Mach combination (§ 12.3). However, it is possible to introduce a
further parameter, the second-segment climb rate. Some manufacturers show climb
charts for increasing KCAS2 and various cruise Mach numbers. These constraints
are unlikely to satisfy the best cruise conditions.
10.3.5 Numerical Issues

From a computational point of view, when the landing gears are retracted there is an
impulsive change in drag. This causes an impulsive change in aerodynamics, climb
rates and flight path, unless a corrective action is applied. This action would require
a more detailed analysis of the retraction process, a method for estimating the drag
of a moving landing gear and the use of a much smaller time step than is required
for ordinary flight-path integration.
Another problem, perhaps more evident, is that the application of the conven-
tional climb strategy would lead to an aircraft suffering a jump in climb rate in
* This parameter is not the same as that in the FMS. Do not use for flight planning.
the second-segment climb (both at the start and the end; KTAS1 and KTAS 2 in
Figure 10.1). This problem can only be overcome by using a variable throttle setting.
10.3.6 Initial Climb with One Engine Inoperative

If following an engine failure at take-off the aircraft accelerates and continues its
flight, it must initiate the climb with a severely reduced thrust and with the aircraft
trimmed both laterally and longitudinally to ensure that the vehicle can stay airborne.
The OEI climb capabilities will have to be demonstrated at the certification level.
The initial climb of a jet-powered aircraft with one engine inoperative consists of
four conventional phases, in which the sequence of landing-gear retraction, high-lift
position and flight control actions is specified, along with the minimum climb gradi-
ents that must be guaranteed. This sequence of events, not reported here for brevity,
can be implemented into the flight-mechanics code by using impulsive switches at
trigger points (flight altitude, true air speed, climb rate, thrust setting).
10.4 Climb of Commercial Propeller Aircraft

We now carry out a numerical analysis for a commercial flight of a propeller airplane.
This analysis is similar to the case of the turbofan-powered airplane. The climb
segments are as described earlier (§ 10.3). However, there are a number of important
differences in the flight-mechanics model. The numerical procedure is described
here:
r After take-off the airplane maintains full power up to a point when the landing
gears are retracted. The engine-propeller matching is done at the limit engine
power. In other words, at the given speed-altitude the turboprop is capable
of providing a shaft power Pshaft . The corresponding power coefficient is C P =
Pshaft /qUtip
3
. The propeller is then trimmed to provide this value of the C P .
When this operation is done, the corresponding thrust coefficient CT is used to
calculate the net propeller thrust: Tp = qUtip
2
CT . Because there is a residual jet
thrust Fg from the engine, the total propulsive thrust with ηe operating engines
is
T = (Tp + Fg )ηe . (10.33)

r When the landing gears are retracted (numerically, this is an impulsive event),
the shaft power required to maintain speed or accelerate at the specified accel-
eration rate is considerably diminished. An acceleration and a climb rate are
set. This condition leads to a required thrust
dU
Treq = m + D + W sin γ = Tp + Fg . (10.34)
dt
The required thrust is provided by a combination of propeller and engine thrust.
Equation 10.34 is solved iteratively:
1. Set Fg = 0, Tp = Treq .
2. Trim the propeller to the required thrust CT = Tp /qUtip
2
.
3. Calculate the C P and Pshaft corresponding to the trimmed condition.
10.4 Climb of Commercial Propeller Aircraft 283
4. Solve the engine model in the inverse mode in order to determine the
engine state EngineState = f (W1, WF6, TT5, Fg9, . . .). Set Fg = Fg9.
5. Correct the required propeller thrust: Tp = Treq − Fg .
6. Reiterate from point 2 until the difference in residual thrust Fg is negligible.
r At the first reference altitude (1,000 feet or otherwise) there is a power cut-
back. Thus, the shaft power is reduced to a fraction determined by a minimum
climb rate and a minimum acceleration. The actual power required is calculated
iteratively; now the propeller is trimmed to a required power.
r Upon reaching the reference altitude with flaps and slats retracted, and with
a target KCAS = KCAS1, the airplane starts a constant-CAS climb. At each
altitude in the trajectory the TAS is specified by the h-CAS combination. The
thrust required for climb is
v
c dU
Treq = W + mvc + D. (10.35)
U dh
In this equation, the acceleration term dU/dt is converted to dU/dt =

vc (dU/dh). The derivative dU/dh depends only on the h-CAS combination.
Therefore, for a given gross weight, the required thrust Treq is a function of the
rate of climb. This parameter is normally set to a value recommended by the
manufacturer or the ATC.
Alternatively, a tentative climb rate is set, the propeller is trimmed to the
required Treq and the corresponding Pshaft is calculated; if this power is excessive,
the climb rate is reduced. In any case, Equation 10.35 is solved with the same
algorithm as Equation 10.34.
r Upon reaching a target altitude, for example 3,000 feet above the airfield, the
airplane performs a level acceleration to a target KCAS = KCAS2. The required
thrust is
dU
Treq = m +D (10.36)
dt
and depends critically on the acceleration rate, which can be the dominant term.
Once the acceleration is determined iteratively, Equation 10.36 is solved as
Equation 10.34.
r The final segment of the climb to initial altitude is done at constant KCAS =
KCAS2.
Figures 10.6 and 10.7 show the simulated climb trajectory from take-off of the
ATR72-500 with PW-127M engines and F568 propellers.
These graphs show a number of impulsive events, as described in the cap-
tions. The first impulsive event (A) is when the landing gear is retracted: the drag
decreases and the required thrust/power also decreases. The second impulsive event
is at the power cut-back (after C), then at the first constant-CAS climb (E), the
level-flight acceleration (F) and the start of the second-segment climb (G). Note
that there is also an impulsive change in the efficiency of the propeller, which is
maintained within the range η = 0.7 to 0.86. The fuel consumption increases rather
uniformly.
Distance from take-off, n-miles

0 10 20 30 40
6 5,000
A = Take-off
A B = L-Gear Up
F C = Initial Climb
D = Power Cut
E = KCAS1-Climb
F = Level Acceleration
C G = KCAS2-Climb
Flight altitude, km
Shaft power, kW
4 4,000
B
2 3,000
D Trajectory
0 2,000
0 5 10 15
Flight time, min
Figure 10.6. AEO climb to ICA of an ATR72-500 model with PW127M engines; standard
day, no wind.
Distance from take-off, n-miles

0 10 20 30 40
1 200
Efficiency
Fuel burn
0.9 150
Propulsive efficiency
Climb fuel, kg
G
D
0.8 100
E F
B
C A = Takeoff
A B = L-Gear UP
0.7 50
C = Initial Climb
D = Power Cut
E = KCAS1-Climb
F = Level Acceleration
G = KCAS2-Climb
0.6 0
0 5 10 15
Flight time, min
Figure 10.7. AEO climb to ICA of an ATR72-500 model; standard day, no wind.
Table 10.4. Climb report for the case shown in Figure 10.6
h t vc vc mf ṁ f
KCAS KTAS M [m] [min] [m/s] [ft/min] [kg] [kg/s]
0 161.9 162.3 0.246 61 [over screen]

1 196.4 228.0 0.358 3,098 7.96 6.39 1,257 93.8 0.197 [const CAS]
2 196.4 228.0 0.358 3,098 0.00 0.00 0 0.0 0.186 [accelerate]
3 196.5 288.2 0.479 7,615 14.82 5.08 1,000 144.8 0.163 [const CAS]
4 196.4 288.2 0.479 7,620 0.07 2.54 500 0.5 0.126 [const Mach]
Time to ICA 22.8 [min]
Fuel to ICA 239.1 [kg]
Distance to ICA 92.0 [n-m]
L-Gear retraction 26.0 [m]
Flaps retraction 100.0 [m]
L-Gear retraction time 8.7 [s]
Table 10.4 is a summary of climb calculations, with the various segment climbs,
the altitude climbed, the flight time, the fuel burned, the average fuel flow and other
quantities.
10.5 Energy Methods

The typical approach to the climb problems is done by using steady-state models.
However, this cannot be correct because as the climb rate and the optimal climb rate
change with increasing altitude, the aircraft must accelerate. The difference between
steady state and accelerated flight is particularly important for high-performance
aircraft. Bryson and Denham 1 proved that an optimal accelerated climb to a target
altitude requires about half the time of an optimum quasi-steady climb. Optimal
problems have been published by Rutowski 2 , Kelley 3 , Schultz and Zagalsky 4 ,
Calise 5 and others. Work on the subject includes optimum climb profiles of super-
sonic transport aircraft with noise minimisation 6 , and optimal trajectories, including
minimum fuel, minimum time or cost for fixed-range 7 .
Instead of considering forces on the centre of gravity of the aircraft, it is some-
times useful to write balance equations for the total energy of the aircraft in its
climbing flight. Methods of this nature are called energy methods. The first methods
based on the concept of total aircraft energy go back to the 1940s. A widely acclaimed
paper in this field is that of Rutowski 2 , although the original idea seems to belong
to the German engineer F. Kaiser, who developed the concept of “resultant height”
for the optimal climb profile of the Messerschmidt Me 262, the first-ever jet fighter.
Kaiser’s concept is reviewed by Merritt et al. 8 , and the climb technique is sometimes
called the “Kaiser procedure”. Kelley 9 contended that trading kinetic energy for
potential energy is essential in maximising aircraft performance. These methods,
although approximate, are extremely powerful, as evidenced by later research 10 .
These methods have been expanded to be defined as a total energy method; they
have become an industry standard for the air traffic control centres in Europe 11 and
have widespread use in the professional field 12 .
10.5.1 Total-Energy Model

The total-energy balance for the aircraft, reduced to a lumped mass, is the following:

dh dU dh
(T − D)U = W + mU , (10.37)
dt dt dt
where h is the flight altitude. Equation 10.37 contains separate terms for the vertical
and horizontal speed components. Solving for the vertical speed dh/dt, we have

dh T−D U dU −1
= U 1+ . (10.38)
dt W g dh
The rate of climb vc is the variation of the geopotential pressure altitude h p with
time, dh p /dt:

dh p T − T dh
= , (10.39)
dt T dt
where T denotes the temperature differential with respect to the standard atmo-
spheric model. On a standard day, the rate of climb is equivalent to the vertical
speed; on a hot day (T > 0) the rate of climb is lower than the vertical speed.
The term within square brackets in Equation 10.38 is a function of the Mach
number and is sometimes called energy share factor, E(M). In the stratosphere,
where the aircraft maintains a constant flight Mach number, we have
E(M) = 1 (10.40)
because dU/dh = d(a M)/dh = 0 (the temperature is constant, and so is the speed
of sound). In the lower atmosphere, where there is a temperature lapse rate λ, the
energy-share factor is
−1
γ Rλ 2 T − T
E(M) = 1 + M . (10.41)
2g T
An interpretation of this result is that a climb at constant Mach in the troposphere
causes a gain in climb rate, or a conversion of kinetic energy into potential energy –
thanks to the decreasing air temperature.
A third case is relative to a constant-CAS climb in the troposphere, a common
practice for commercial airplanes. It can be demonstrated that the energy share
factor can be written as
−1 ,−1
γ Rλ 2 T − T ρ po
E(M) = 1 + M + −1 (10.42)
2g T ρo p
with

ρ γ − 1 2 1/1−γ po γ − 1 2 γ /γ −1
= 1+ M , = 1+ M . (10.43)
ρo 2 p 2
A sub-case of this equation is a constant-CAS climb in the stratosphere. In this
case, the temperature is constant and the energy-share function, although formally
the same, has a lower value. The difference of E(M) in the troposphere and the
stratosphere for a constant-CAS climb is shown in Figure 10.8. In both cases, the
function is less than one. A constant-CAS climb requires an increase in true air
1.2
Energy share factor 1
0.8
Constant Mach, stratosphere

Constant Mach, troposphere
Constant CAS, stratosphere
Constant CAS, troposphere
0.6
0 0.3 0.6 0.9
Mach number
Figure 10.8. Energy-share factor for four different climb/descent rates; standard day.
speed, as a consequence of a decreasing air density; part of the climb energy must
be spent for acceleration.
Consider now Equation 10.37. This equation can also be written as

∂ 1 2
(T − D)U = m U + gh . (10.44)
∂t 2
The term within parentheses is the total energy E; this quantity, divided by the
acceleration of gravity g, has the dimensions of a distance and is called energy height
hE,
E U2
hE = = + h. (10.45)
g 2g
The energy E is the sum of the kinetic and potential energy of the aircraft per unit of
mass. The energy height Equation 10.45 represents the altitude at which the aircraft
would climb if it were to convert all of its kinetic energy to potential energy. The
time derivative of the total energy is the work done by the power plant
∂E TU − DU
= . (10.46)
∂t m
The time derivative of the energy height is equal to the specific excess power
∂h E TU − DU
= = SEP. (10.47)
∂t W
Some methods used for flight-path optimisation use Equation 10.45, with the addi-
tional assumption that the aircraft can instantaneously exchange kinetic energy with
potential energy, and vice versa. This approximation is a fairly good one if the short
period of motion is neglected. However, it leads to sharp changes in direction in the
flight path, which are unreasonable.
20
constant hE 60
15 50
Flight altitude, 10 feet

Flight altitude, km
3
40
constant hE
constant hE climb
10 dive
30
20
5 Max hE gain
10
0 0
0 0.5 1 1.5 2
Mach number
Figure 10.9. Constant energy levels on the plane M − h.
Figure 10.9 shows lines of constant energy height in the M − h plane for a unit
mass. These lines show a knee at a point located at altitude h = 11,000 m (36,089
feet). This is due to the change in atmospheric conditions at the tropopause. The
lines of constant energy are found from
E − U 2 /2 E − (a M)2 /2
h= = , (10.48)
g g
with E the assigned energy level. When the M = 0, then h = E/g. This means that all
the energy is potential energy. The geopotential height decreases with the increasing
speed.
10.5.2 Specific Excess Power Charts

Diagrams of the SEP summarize the total performance of an aircraft in the altitude-
speed plane and complement the flight envelope that is discussed in Chapter 8.
Lines of constant specific excess power, SEP, are only valid for a fixed configuration,
weight, load factor, angle of attack, engine throttle and atmospheric conditions. In
practice, only a limited number of SEP curves are drawn. An example is shown in
Figure 10.10. This figure shows the SEP distribution on the Mach-altitude plane.
Selected levels of SEP are shown. The line SEP = 0 denotes the limits of operation
stabilised level flight. There are two hills, at subsonic and supersonic speeds. At
subsonic speeds, maximum values of SEP are found at sea level; the SEP decreases
rapidly with the altitude. At supersonic speeds, maximum values of SEP are at
intermediate altitudes, which indicates that a supersonic aircraft is most effective
at altitude manoeuvring only within a limited altitude range. Outside this range, its
acceleration capabilities are severely impaired by high drag and reduced thrust.
The next plot, Figure 10.11, refers to the same aircraft without after-burning
thrust. This figure contains further information. For example, the excess power at
150
200
150 100
SEP, m/s
100
50
50
0
50 0
0 3
0.5
6
m
Ma 1 9 e ,k
ch
nu ud
m 12 tit
be
r 1.5 Al
15
2 18
Figure 10.10. Specific excess power plot with after-burning thrust at supersonic speed; mass
m = 12,000; standard day.
150 TT5[K]
200
950
930
910
150 100 890
870
SEP, m/s
850
100 830
50
50
0
50 0
0
3
0.5
6
Ma 1 9 , km
ch de
nu
m 12 t ti u
be 1.5
r Al
15
2 18
Figure 10.11. Specific excess power plot without after-burning thrust at supersonic speed;
colour map based on turbine temperature TT5; standard day.
18
A
15 A1 50
0
40
Altitude, 10 feet
Altitude, km 12 VS
3
9 30
50 50
6 B 20
100
150 10
3 0
0 0
0.5 1 1.5 2
Mach number
Figure 10.12. Energy levels and constant excess power lines; m = 12,000 kg; standard day.
supersonic Mach numbers is rather limited, and the aircraft would be unable to
manoeuvre. Second, the plot shows the distribution of temperature TT5 at the exit of
the turbine. The line corresponding to zero excess power is a limiting case because it
divides the flight envelope in two regions. If SEP < 0, the aircraft can only decelerate
because the thrust available is less than the thrust required to overcome the drag.
Therefore SEP = 0 is a stationary line for the aircraft speed; the condition SEP < 0
is outside the normal flight envelope of the aircraft. If we use the climb rate equation
and the definition of excess power, the line of zero SEP is found from the condition
U ∂U ∂h U ∂U
U sin γ + = + = 0. (10.49)
g ∂t ∂t g ∂t
Figure 10.12 is a two-dimensional plot in the Mach-altitude space. Lines of

constant SEP (0, 50, 100, 150 m/s) are shown, in addition to lines of constant energy
height and the wing stall limit, VS . Note that point A would be the absolute ceiling
if the condition on the stall speed were not as restrictive as shown in the graph.
However, because of this restriction the absolute ceiling would be at point A1 . In
any case, the aircraft should stay clear of this point.
10.5.3 Differential Excess Power Charts

A comparison between excess power plots of different high-performance aircraft
can provide valuable information regarding their manoeuvre capability. One such
example is shown in Figure 10.13, where two different configurations of the same
aircraft have been selected. From the performance point of view, these configurations
differ only in the transonic and supersonic drag characteristics; they have the same
engine and the same weight. At any point in the plane M − h we can define the
20
60
SEP = 0
15 50
Altitude, 10 feet
Altitude, km
40
3
10
30
Config. B
SEP = 100
20
5
10
Config. A
0 0
0.5 1 1.5 2
Mach number
Figure 10.13. Differential specific excess power for two configurations.
difference in excess power as
SEP = SEP(B) − SEP(A). (10.50)
Charts like those in Figure 10.13 tend to get complicated because one aircraft can
have a manoeuvring advantage in a region of the envelope and a handicap in another
region. In the present case, we have plotted only two SEP lines. One line is obviously
SEP = 0, that expresses the limits of the manoeuvre envelope. Another value is SEP
= 100 m/s. This is close to the maximum SEP for both aircraft at the given gross
weight.
These comparisons become less straightforward if the two aircraft are substan-
tially different. The choice of weights is essential, and only comparisons at similar
weights make sense. By reading this difference, the performance and design engin-
eers can improve some operations; the pilots learn to avoid the flight condition in
which they are likely to have a handicap.
10.6 Minimum Problems with the Energy Method

The energy method is most suitable for the analysis of the climb performance of
supersonic aircraft that accelerate past the speed of sound. A number of flight
programs are of interest: fastest climb (or minimum-time to climb), steepest climb
and minimum-fuel climb.
10.6.1 Minimum Time to Climb and Steepest Climb

The minimum time to climb to a target altitude is specified from the condition that
the gain in energy height is maximum with respect to the Mach number. This is
equivalent to maximising the climb rate (see Equation 10.4). The energy height is
introduced in place of h from Equation 10.45. A classical method for finding this
path is done by searching the curve that joins the points of maximum SEP at all
altitudes. It is essentially a graphical method that does not require the solution of
any equation. The solution can be found also numerically, by advancing along a
steepest ascent/descent direction. However, both methods break down when there
is no clear best-descent direction. To continue, the aircraft must pitch over and start
a zoom-dive along the constant-energy path.
The steepest climb condition is the maximum in the Mach-altitude plane of the
SEP with respect to the Mach number:
v T − D
c
maxM = . (10.51)
U W
10.6.2 Minimum Fuel to Climb

A fuel-climb problem is specified from Equation 10.31 that is divided by the rate of
change of total energy: Equation 10.46,

ṁ ∂m ∂t ∂m m
= = = −ṁ f . (10.52)
Ė ∂t ∂E ∂E (T − D)U
When we separate the differentials dm and dE, we find
dm dE
= −ṁ f . (10.53)
m (T − D)U
The ratio dm/m = dm1 is the specific change in aircraft mass, that is, the change in
mass due to fuel flow divided by the aircraft mass. A minimum-fuel-climb problem
is formulated by the condition that minimises dm/m for a given energy level; that is,
dm1 ṁ f
=− . (10.54)
dE (T − D) U
Therefore, a minimum-fuel-climb flight path requires the minimisation of the right-
hand side of Equation 10.54 or a maximisation of its inverse. We construct a
function
(T − D)U
f (h, M) = − , (10.55)
ṁ f
that is proportional to the excess thrust. Note that f (h, M) is a negative function in
the flight envelope of the aircraft because T ≥ D. The minimum fuel to climb is the
locus of the maximum energy increase per unit of fuel burned at a fixed energy E.
10.6.3 Polar Chart for the Climb Rate

The polar diagram shows the split between power required to keep the speed and
the power required to climb, in steady state. Diagrams of this type are calculated
at a constant mass, flight altitude and throttle setting. In general, only a few polars
in this family are of interest: the polar at maximum thrust, the power-off polar, the
polar at MTOW and the polar at absolute ceiling. The polar is obtained by joining
two segments: the climb and descent segments.
Ground speed, kt
0 200 400 600 800
200
Max climb rate

Max climb angle 30
150
Climb rate, 10 ft/min

C B
Climb rate, m/s
3
100 20
50 10
Stall
Max level speed
D
A
0 0
0 250 500 750 1,000 1,250 1,500
Ground speed, km/h
Figure 10.14. Positive climb polar of supersonic jet m = 11,000 kg, starting altitude h =
2,000 m (∼6,560 feet), steady flight, military thrust.
An example of calculation is shown in Figure 10.14. As the aircraft starts climb-

ing, the horizontal speed must decrease because a portion of the power must be used
to keep the air speed. The maximum vc is achieved at an intermediate forward speed,
as indicated. At forward speeds below this point, the climb rate decreases. The climb
at maximum climb angle is a yet lower speed vc /U. The speed of maximum climb
angle is lower than the speed of maximum climb rate; this corresponds to the tangent
point of the polar with the line through the centre of the reference system.
10.6.4 Case Study: Climb to Specified Mach Number

We use the energy method to calculate some climb profiles of a supersonic aircraft
model. The climb consists of the following segments:
1. Climb at constant flight-path angle from take-off. This segment starts from given
conditions at take-off (M, h, γ , W) and proceeds at full thrust on a given flight-
path angle until the combination of Mach-altitude reaches a value corresponding
to maximum SEP.
2. Pull-up at fixed normal load factor. The aircraft performs a pull-up at a fixed
load factor to assume a climb rate corresponding to the maximum SEP.
3. Subsonic climb at maximum SEP or maximum energy gain. The aircraft climbs
at maximum SEP until this parameter reaches a relatively low value (to be
specified); alternatively, it climbs between levels of constant energy.
4. Pull-out and zoom dive. The aircraft pulls out and starts a dive at a fixed value
of the flight path (to be specified) down to an altitude also to be specified.
18 18
15 50 15 50
0 0
Flight altitude, km
Flight altitude, km
Altitude, 103 feet
Altitude, 103 feet

12 40 12 40
4
A A
50 50 7
9 30 9 30
5 5
B B
6 20 6 20
6 3
100 4 50 100 50
3 0 0
3 10 3 10
150 150
1 2 2
1
0 0 0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
(a) (b)
Figure 10.15. Selected climb paths.
5. Pull-up at fixed normal load factor. This phase is similar to the second segment,
except that the aircraft is recovered from a negative flight-path angle.
6. Supersonic climb to target Mach number. The aircraft performs a supersonic
climb at maximum SEP until it reaches the target Mach number. The final
altitude is unspecified.
There are at least three free parameters: 1) the minimum-climb gradient before
the aircraft pulls out and starts a zoom-dive; 2) the flight path gradient of the zoom-
dive; and 3) the minimum altitude of the zoom-dive. We can specify secondary
parameters, such as the normal load factors in the pull-out manoeuvres, the initial
climb rate after take-off and other quantities. The constraints are the following:
r Engine: Operational limits set on N%1 = 104, TT5 = 970 K at maximum continu-
ous power, to avoid over-speeding and over-heating at the turbine section of
the engine.
r Aerodynamic Heating: Maximum skin temperature at stagnation T = 200
stag
(this temperature is calculated with an adiabatic model).
r Flight Mechanics: Speed sufficiently larger than the stall speed: U > V , and
S
below the wing buffet limit, U < Vbuffet .
The pull-up segments 2 and 4 are calculated by integration in the time domain
of the equations defined in § 13.6.
Figure 10.15 shows four different sub-optimal climb paths from take-off (M =
0.25; m = 12,000 kg; γ = 8 degrees; after-burning for supersonic flight; sea-level,
standard day) to a target M = 2.0 (without specified final altitude). In all cases, the
sequence of climb segments is as described previously and denoted by the numbers
1, 2, . . .. Both graphs show selected levels of SEP and the stall speed, calculated with
a stall margin KS = 1.15.
In Figure 10.15a the aircraft performs a maximum energy-gain climb after a 4g
pull-up. Then it performs a supersonic maximum-SEP acceleration. For the case
denoted by A, the aircraft is prevented from descending, whilst in case B it follows
Table 10.5. Climb time and fuel for the flight paths
shown in Figure 10.15
Flight path A B C D
Climb time [min] 3.12 4.12 4.22 4.41

Climb fuel [kg] 160.0 173.2 244.1 249.8
the path of maximum supersonic SEP. In Figure 10.15b the aircraft performs a
maximum-SEP climb after a 4g pull-up. It then performs a long zoom-dive to reach
about M = 1.4. For the case denoted by C, it is prevented from descending (with the
dive halted at a lower Mach), whilst in case D it follows the path of maximum SEP.
A summary of climb times and climb fuel is given in Table 10.5. The results indicate
clearly that the best climb (among those examined) is the flight path A, giving both
minimum time and minimum fuel.
10.6.5 Minimum Flight Paths

The case discussed herein was based on quasi-steady flight-mechanics analysis, which
in some cases is not of sufficient accuracy. Furthermore, there are cases when we
need to specify both initial and terminal conditions, subject to a minimum cost func-
tion (fuel, time, ground distance). The case of Figure 10.15 belongs to this class of
problems, known as trajectory optimisation, with or without terminal constraints.
The total-energy method is a viable approach to minimum flight paths, although it
assumes that the aircraft can convert instantaneously its kinetic energy into poten-
tial energy; therefore, it leads to flight paths with singularities (sharp turns) that
are unrealistic (see Figure 10.15). More general methods for the solution of these
problems are based on optimal control theory, a subject that falls beyond the scope
of aircraft performance. The methods are mathematically involved, but they can be
programmed on a variety of platforms.
The first relevant unsteady analysis of the aircraft climb is the one published
by Miele 13;14 . An example of three-dimensional climb-turn analysis is that of Neu-
man and Kreindler 15;16 , who derived climb-out and descent trajectories from and to
runways, including optimisation for minimum fuel consumption. These results show
that the velocity profiles for straight and turning flight are almost identical, except
for the final horizontal accelerated or decelerated turn.
CONTROL PROBLEM DEFINITION. The problem is to take the aircraft from its initial
state x o = {U, γ , h, m}o to its final state x e = {U, γ , h, m}e . If the engine is always
at full throttle, the problem is reduced to finding a control variable program α(t).
However, this option is unrealistic for most flight conditions, and the throttle must
be considered as a free parameter, to be found as a part of the solution.
The nomenclature of the problem is as follows: α(t) is the control variable pro-
gram (the angle of attack); x(t) = x(x1 (t), . . . , xn (t)) is the vector of state variables
programs (velocity, altitude, climb rate ...); ψ = (ψ1 , . . . , ψ p ) is the vector of terminal
constraints functions (ψ is a known function of the terminal point and the vector of
state variables); and, finally, φ is the cost function that depends on the final point
and the state variables x(te ) at the final point. Therefore, the problem is to take the
aircraft to the terminal state subject to the performance criterion being a minimum
or maximum. The objective function depends on the final state of the aircraft and
on the course of events between the start and the terminal point. The aircraft can
be subject to some terminal constraints. For example, one may want to have a final
Mach number, a final altitude or a final climb angle. The most important cases are
the following: minimum-time climb and minimum-fuel climb. To solve the problem,
we must use a form of the Lagrange multipliers to construct the so-called adjoint
equations. The derivation of the conditions for the Lagrange multipliers can be elab-
orate. Bryson and Ho 17 and Ashley 18 showed the entire derivation procedure. The
initial conditions are assigned. The terminal conditions can be assigned or left free.
The problem so formulated is a boundary-value problem. The stopping condition is
found from the altitude at the terminal point. The aircraft will reach that altitude
with a time te , which is assured to be a minimum.
The adjoint equations contain derivatives of the lift, drag, thrust and fuel flow.
These derivatives are calculated numerically. There are a variety of methods avail-
able to solve the optimal problem, including gradient methods, multiple shooting
algorithm and dynamic programming. Bryson and Denham 1 , in their solution of
the problem for the F-4 interceptor, used the steepest-descent method. The method
can be found in the technical literature (for example, Press et al. 19 ). The method
requires a first-guess of the solution; it then proceeds, by local linearisation around
the current point, in the direction of the steepest ascent or descent. The multiple
shooting method of Bulirsch and Stoer 20;21 has been used by Br̈uning and Hahn 22
for a variety of optimal-climb problems.
Summary
This chapter has discussed aircraft climb in the vertical plane. We have shown some
closed-form solutions, including fastest climb conditions. However, the climb to alti-
tude by commercial aircraft is done in steps, which are not always optimal. In fact,
there are constant-CAS segments with level acceleration. We have presented some
standard ICAO climb profiles. One important solution strategy is based on the use
of total energy (energy methods), which applies best to high-performance super-
sonic aircraft. These methods allow the rapid analysis of the specific excess power
in graphical form; thus, they are suitable for graphic interpretation rather than ana-
lytical treatment. The total-energy methods have been applied to a number of case
studies, including differential analysis between competing aircraft. We have shown
that even with the simplifications required by the instantaneous energy transfer, it
is possible to analyse problems such as air superiority and fastest climbs. However,
there are sophisticated methods in the technical literature. These methods are based
on control theory and allow the solution of climb problems with optimality conditions
and initial and terminal constraints. These methods have only been annotated; the
readers should consult the literature on control theory for theory and applications.
Bibliography
[1] Bryson AE and Denham WF. A steepest-ascent method for solving optimum
programming problems. J. Applied Mechanics, 29(2):247–257, 1962.
[2] Rutowski ES. Energy approach to the general aircraft performance problem.
J. Aero. Sci., 21(3):187–195, Mar. 1954.
[3] Kelley HJ and Edelbaum TN. Energy climbs, energy turns, and asymptotic
expansions. J. Aircraft, 7(1):93–95, Jan. 1970.
[4] Schultz RL and Zagalsky NR. Aircraft performance optimization. J. Aircraft,
9(2):108–114, Feb. 1972.
[5] Calise AJ. Extended energy management method for flight performance optim-
ization. AIAA J., 15(3):314–321, Mar. 1977.
[6] Berton JJ. Optimum climb to cruise noise trajectories for the high speed civil
transport. Technical Report TM-2003-212704, NASA, Nov. 2003.
[7] Ardema MD, Windhorst R, and Phillips J. Development of advanced methods
of structural and trajectory analysis for transport aircraft. Technical Report
CR-1998-207770, NASA, Mar. 1998.
[8] Merritt SR, Cliff EM, and Kelley HJ. Energy-modelled climb and climb-dash –
the Kaiser technique. Automatica, 21(3):319–321, May 1985.
[9] Kelley HJ. An investigation of optimal zoom climb techniques. J. Aero Sci,
26:794–803, 1959.
[10] Kelley HJ, Cliff EM, and Weston AR. Energy state revisited for minimum-time
aircraft climbs. Number AIAA Paper 83-2138. Aug. 1983.
[11] Anon. User Manual for the Base of Aircraft Data (BADA), Rev. 3.8. Eurocon-
trol Experimental Centre, April 2010. EEC Technical Rept. 2010-003.
[12] ESDU. Energy Height Method for Flight Path Optimisation. Data Item 90012.
ESDU International, London, July 1990.
[13] Miele A. Optimum flight paths of a turbojet aircraft. Technical Report TM-
1389, NACA, Sept. 1955.
[14] Miele A. General solutions of optimum problems of non-stationary flight. Tech-
nical Report TM-1388, NACA, Oct. 1955.
[15] Neuman F and Kreindler E. Optimal turning climb-out and descent of com-
mercial jet aircraft. Number SAE Paper 821468, Oct. 1982.
[16] Neuman F and Kreindler E. Minimum-fuel turning climbout and descent guid-
ance of transport jets. Technical Report TM-84289, NASA, Jan. 1983.
[17] Bryson AE and Ho YC. Applied Optimal Control. Blaisdell, New York, 1969.
[18] Ashley H. Engineering Analysis of Flight Vehicles. Addison-Wesley, 1974.
[19] Press WH, Teukolsky SA, Vetterling WT, and Flannery BP. Numerical Recipes.
Cambridge University Press, 2nd edition, 1992.
[20] Bulirsch R and Stoer J. Numerical treatment of ordinary differential equations
by extrapolation methods. Numer. Math, 8(1):1–13, Jan. 1966.
[21] Stoer J and Bulirsh R. Introduction to Numerical Analysis. Springer-Verlag,
2nd edition, 1993.
[22] Brüning G and Hahn P. The on-board calculation of optimal climbing paths.
In Performance Prediction Methods, volume AGARD CP-242, pages 5.1–5.15,
May 1978.

a = speed of sound
A = wing area
ci = constant coefficients, i = 1, 2 . . .
CL∗ = lift coefficient for 1-g flight
CLto = take-off lift coefficient
CP = power coefficient of a propeller
CT = thrust coefficient of a propeller

dm1 = specific change in aircraft mass, dm/m
D = drag force
E = energy level; energy-share factor
Fg = residual jet thrust (also Fg9)
h = geodetic altitude
hp = geopotential pressure altitude
hE = energy height
Ks = stall margin
ISF = index for slat/flap setting
J = advance ratio of a propeller
L = lift force
L = function of the course of events
m = mass
mf = fuel mass
M = Mach number
n = normal-load factor
ηe = number of engines
p = pressure
P = power
R = propeller radius
R = gas constant
t = time
T = net thrust
Treq = required thrust
T = temperature
Tstag = stagnation temperature
U = air speed
Uw = wind speed
Utip = tip speed of a propeller
vc = climb rate
Vg = ground speed
V2 = take-off speed
VEF = engine-failure speed
VLO = lift-off-speed
V = airplane state vector in flight
Vbuffet = buffet speed
Vgreen = green-dot speed
VS = stall speed
x = flight distance
x = state vector in control analysis
X = required distance
W = weight
Greek Symbols
γ = climb angle; ratio between specific heats in § 10.5.1
δf = flap angle
δs = slat angle
ε = thrust angle
θ = propeller pitch
λ = temperature lapse rate in the atmosphere, Equation 10.41
φ = cost function
ψ = terminal constraint
ρ = air density
= rotational speed
[.]e = end point
[.] f = fuel
[.]o = initial state
[.]γ = flight-path angle
˙
[.] = time derivative
¯
[.] = average value
[.]req = required quantity
11 Descent and Landing Performance
Overview
The descent deals with that segment of the flight when the airplane decreases its
flight altitude in a controlled mode. Landing still requires good pilot skills, which
are best appreciated in bad weather conditions. The descent can be a large portion
of the stage length, reaching in excess of 100 n-miles. Several distinct phases are
identified. We will consider separately the phase of en-route descent down to 1,500
feet altitude above the airfield (§ 11.1) and the final approach down to ∼50 feet
above the airfield (§ 11.2). We also discuss two unconventional flight procedures: the
continuous descent approach (§ 11.3), which has some advantages in terms of fuel
consumption and noise emissions, and the steep-descent approach, which is a more
complex manoeuvre (§ 11.4). We analyse the case of airplanes placed on holding
stacks (§ 11.6) and optimal performance issues. Landing consists of an airborne
phase and a landing run (§ 11.7). Effects of side gusts are considered, including
crab landing and wing strike. This chapter ends with considerations of go-around
trajectories, which are associated to aborted landing (§ 11.8).
KEY CONCEPTS:En-Route Descent, Continuous Descent Approach, Steep Descent,

Unpowered Descent, Holding Procedure, Landing Performance, Crab Landing, Go-
Around.
11.1 En-Route Descent

The aircraft starts descending from its cruise altitude well ahead of its destination.
The flight computer will indicate the distance to the airfield and the estimated
en-route descent, that is, the distance and time to landing, based on a number of
factors, including speed, altitude and winds. The en-route descent from the final
cruise altitude is done in stages. One such procedure is shown in Figure 11.1. First,
the aircraft will descend at a constant Mach number to an altitude where it reaches
a target CAS (or IAS). In the second stage, it maintains constant CAS (or IAS) to
an altitude of 10,000 feet (FL-100, 3,048 m). In the third stage it decelerates at a
constant altitude to reach a target CAS (below the previous one). The final stage is
a descent at a constant CAS to 1,500 feet. Below this altitude, the aircraft is in the
300
11.1 En-Route Descent 301
TOD
A
KCAS1
Decelerate C
B
KCAS2
10,000 ft
E
KCAS2
3,000 ft
D
o
3
Figure 11.1. Multi-stage en-route descent (graph not on scale).
terminal manoeuvre area. This is by no means a unique descent trajectories. There

are more complicated cases, involving more steps, constant-CAS specifications and
fixed descent rates.
The trajectory is characterised by a number of parameters: the final cruise alti-
tude (FCA) or initial descent altitude (IDA), the second-segment CAS, and the
fourth-segment CAS. For some flight conditions, the first-stage, constant-Mach des-
cent is absent because the aircraft is already within the specified CAS. An alternative
to this procedure is to avoid operation at FL-100 (KCAS2 in Figure 11.1) and start the
descent later. This procedure is meant to keep the aircraft at the optimum altitude
for a longer time.
If the engine is set to idle, the air speed is controlled by the flight altitude. A
higher weight increases the descent distance because it decreases the glide angle. In
fact, in stabilised flight we have
T−D D
γ − . (11.1)
W W
A high weight decreases the flight-path angle and causes a longer, more shallow
approach; the descent fuel increases because the engines have to run for a longer
time (even in idle mode they burn a considerable amount of fuel); a longer descent
shortens the cruise distance and hence the cruise fuel. A low weight combined with
high drag will cause a steeper descent path. However, a too fast descent would cause
a too rapid change in cabin pressure. To avoid the steep descent, power may be
applied, which decreases the value of γ in Equation 11.1. In any case, the flight
time is involved, and one must evaluate whether it is preferable to start a descent
from higher altitude and lower gradient or otherwise. Cruise and descent must be
considered jointly in a flight optimisation procedure.
A number of optimal problems can be defined; these problems include minimum
fuel for a given distance and initial altitude, minimum time for a given distance and
initial altitude, minimum time or fuel with unconstrained distance and a constrained
initial altitude, minimum cost index, and so on. We consider the fuel consumption
versus the descent time for a fixed weight at the FCA = IDA = TOD (Top of
Descent). The free parameter is the descent CAS (or IAS). The relationship between
descent time and descent fuel is evaluated at a constant cost index. Because the pilot
50
Net thrust
10
Flight altitude
40
8
Flight altitude, km
Net thrust, kN
30
6
20
4
2 10
0 0
0 5 10 15 20 25
Flight time, min
Figure 11.2. Simulated flight trajectory of the Airbus A320-200, standard day, no wind, from
FL-350 (∼10,670 m). Aircraft mass at the top of descent is 69,380 kg.
cannot be aware of all of these parameters, the Flight Management System (FMS)
calculates the optimum flight procedures. The numerical method is the following:
r The airplane is at the end of the cruise at the top of descent.
r If the top of descent is h > 11,000 m (∼36,000 feet), descend with the (constant)
cruise Mach number to the tropopause (FL-360). The engines are set to idle
mode, and the flight-path slope is calculated from Equation 11.1.
r Descend from FL-360 (or any other flight level) at constant KCAS . This speed
1
is defined from the cruise Mach number and the flight level (see § 8.2 and
Figure 8.7). The target descent altitude of this segment is 10,000 feet (FL-100).
The flight-path angle is constrained to −3 degrees.
r Set the aircraft to a flight level FL-100 and decelerate to KCAS . This speed is
2
calculated from the green dot speed Vgreen (defined in § 11.2) at the end of the
flight segment (1,500 feet above ground level). Assume that the third-segment
CAS is KCAS2 = Vgreen + 5 kt. Therefore, the calibrated air speed moves from
KCAS1 to KCAS2 . Again, we may choose to constrain the flight-path angle.
r Descend at constant KCAS to 1,500 feet and enter the terminal manoeuvre
2
area.
Occasionally, there is an intermediate step, with a level flight at 7,000 feet or

5,000 feet. Following this procedure, we calculated the en-route trajectory and the
corresponding net thrust for the Airbus A320-200 airplane model. The result is
shown in Figure 11.2. The engines are in idle mode during the descent segments but
are forced to spool-up in order to maintain speed in level flight.
A problem of interest in descent performance is the minimum-fuel or minimum-
time from the final cruise altitude, with or without a constraint on the en-
route distance. We consider the case of a descent with a distance constrained to
26.5
X d = 150 n-miles
IDA = 35,000 ft, FL-350
130 110 W = 90 tons
26
Descent time, min KCAS2 = 280

25.5
25
300
24.5
320
24
600 800 1,000 1,200
Descent fuel, kg
Figure 11.3. Parametric effects on descent performance for the Airbus A300-600 at a fixed
descent distance (calculated).
150 n-miles (Figure 11.3). We used a parametric value of the third-segment KCAS,
from 260 to 320 kt (these values of the KCAS do not satisfy the condition of green
dot speed at the terminal point). All of the cases refer to an initial descent from FL-
350. The results show how the descent fuel increases as the descent time decreases.
At this point, a proper choice must be done in terms of the cost index that relates
the cost of fuel to the cost of time.
11.2 Final Approach

The final approach is that flight segment below a threshold altitude (1,500 feet
above ground level), when the aircraft is preparing for landing. In the general case,
the aircraft may have to perform complex three-dimensional trajectories to avoid
obstacles and minimise noise on the ground. In this instance, we will consider only
the case of a rectilinear flight path, from the threshold altitude to the start of the
landing phase (∼50 feet above the airfield), as sketched in Figure 11.4. A complex
series of operations takes place in such manoeuvre, most of which are automated
(instrument landing). With reference to this figure, a few clarifications are required.
r “O”: Green Dot Speed: This is the speed of maximum L/D with engines inop-
erative and clean configuration.
r “S-speed”: This is the minimum speed at which the slats may be retracted at
take-off. On approach, it is used as a speed target when the aircraft is in CONF
1. The S-speed is equal to 1.22-1.25 Vstall at clean configuration.
r “F-speed”: This is the minimum speed at which the flaps may be retracted at
take-off. On approach, it is used as a speed target when the aircraft is in CONF
2 or CONF 3. The F-speed is 1.18-1.22 Vstall with the airplane in configuration
CONF 1+F.
Green Dot
Speed
Flaps 1 S-speed
Go Around
Flaps 2 Flaps 2
LG down LG down
Glide Intercept Rotate
< 2,000 ft Flaps 3 Flaps -1 step
2,000 ft LG Up
o
3
1,000 ft
CWY
Touch-down
Figure 11.4. Series of events in the terminal area manoeuvre (not on scale); H/L = high-lift.
How does the air speed change during this flight segment? Data provided by
Airbus indicate that typical deceleration rates are 10 to 20 kt per n-mile (10 to
20 km/h per km) on a descent slope of 3 degrees, with landing flaps and landing
gear deployed. Deceleration in level flight is higher, up to 30 kt per n-mile. One
recommendation is to use an approach speed Vapp that is approximately equal to
1
Vapp [kt] Ks Vstall + 5 + Vw , (11.2)
3
where all of the speeds are given in knots. Thus, the approach speed is equal to the
stall speed (corrected for the stall margin), an additional 5 kt and one third of the
headwind. We assume that the airplane starts from the green dot speed and ends with
the approach speed. Such a decrease in speed leads to a lower drag and, hence, lower
thrust setting and lower engine noise. Let us look first at the changes in configuration:
these are associated to the deployment of the high-lift devices and the landing
gear.
HIGH-LIFT SYSTEM DEPLOYMENT. The airplane is on a fixed glide slope, flying at

altitude h with true air speed U, with a weight W. The 1-g (unaccelerated) flight
condition requires a lift coefficient CL = 2W/ρ AU 2 . The lift coefficient in nominal
condition is derived from
CL = CLo + CLα α, (11.3)
where CLo is the wing’s zero-angle of attack; CLα is the lift-curve slope and α is the
effective mean angle of attack. The latter parameter is unknown unless the attitude of
the airplane is available. The two expressions for the lift must be the same; therefore,
one could calculate the effective mean angle of attack from Equation 11.3. However,
the wing might be unable to provide the lift required as the airplane descends. In
other words, in the first analysis, we would allow the airplane to increase its attitude
whilst decelerating. If such an increase in attitude, compatible with a condition
Table 11.1. Flap and slat settings for the Airbus A320-200. Angles in degrees
CONF Flap Slat Description Flight Segment
0 0 0 Clean Cruise
Hold
1 0 18 1
2 10 18 1+F
3 15 22 2 Take-off
Approach
4 20 22 3
Landing
5 40 27 Full
of flight stability, is not sufficient to produce the lift required, consider the lift
deficiency
2W
CL = − [CLo + CLα α] . (11.4)
ρ AU 2
Equation 11.4 expresses the difference between the lift required for a 1-g flight at
speed U and gross weight W and the lift that can be generated at the prescribed inflow
condition α. If CL < 0, the wing is unable to provide the lift required; therefore
the high-lift systems must be deployed. This is done in sequence, the exact nature
of which depends on the airplane. Consider, for example, the case of the Airbus
A320-200. The possible slat and flap settings are given in Table 11.1, which shows
that only a limited number of settings are possible.
This change in CL is associated to a combination of slat and flap deflection,
according to
CLhl = CLslat + CL f lap (11.5)
so that
2W
CL = − [CLo + CLα α] + CLhl 0. (11.6)
ρ AU 2
A numerical procedure in the flight-mechanics model is the following:
r If Equation 11.4 provides C 0, set configuration to zero (no slat/flap deploy-
L
ment).
r If C < 0, set configuration to 1. From Table 11.1: CONF 0; Flap 0; Slat 18.
L
Calculate the CLhl from Equation 11.5; if CL 0 in Equation 11.6, then the
correct setting is 1; otherwise, set configuration to 2 (CONF 2; Flap 10; Slat 18)
and repeat the procedure.
Each time a switch is activated from the procedure described, the airplane
changes configuration. This procedure generally works well, but it relies on the fact
that the attitude of the airplane is a known quantity. A small CLhl can be obtained
with a small increase in airplane attitude rather than a flap setting. The FMS could
be programmed to elaborate the best configuration, to control these switches and
to prevent changes in flap setting above a certain altitude and TAS. The FCOM of
each airplane gives information about these altitude-speed limits.
LANDING-GEAR DEPLOYMENT. This operation is a single switch from retracted to fully

deployed and, in practice it appears instantaneous. Landing gears are prevented from
1.6
CL
CD
0.14
1.4
0.11
1.2
3 4
CL
Cd
L-Gear 0.08
1
0.8 0.05
Flaps-2
0.6 0.02
0.25 0.5 0.75 1 1.25
Time, min
Figure 11.5. Lift and drag coefficients during the final-approach manoeuvre of the Airbus
A320-200 (calculated).
deploying above a threshold altitude and air speed. Deployment occurs over a narrow
range of altitudes above the airfield. The landing-gear deployment is prevented from
taking place before the Flap 2 setting (this is typically an Airbus procedure). By the
time the landing gears are deployed, the aircraft must be stabilised in its glide slope.
FLIGHT-MECHANICS MODEL. We set the airplane over a fixed glide slope. Landing on
a steeper gradient is constrained considerably by a number of factors, including stall
speed, descent rates and safety concerns 1 . The differential equation is
dU 1
= (T − D + W sin γ ) . (11.7)
dt m
The thrust is a free parameter and must be adjusted for minimum fuel burn, subject
to the airplane maintaining an air speed sufficiently higher than the stall speed:
U > Ks Vstall , (11.8)
where Ks is the stall margin (Ks 1.15). There is the possibility, at high landing
weights, relatively low air speed and decelerating aircraft, that the net thrust is
zero in Equation 11.7. This event would require the aircraft to fly in idle mode.
Although this procedure would save fuel, it is not completely safe: if an increase in
thrust is required to reconfigure the aircraft or adjust the flight path, the increase
(engine spool-up) may require several seconds, as explained in § 11.8. This delay in
taking control of the aircraft would compromise safety. Therefore, a final-approach
procedure should take place with at least some thrust in the final stages to allow full
control of the airplane by engine power if a go-around is required.
Figure 11.5 shows the change of the aerodynamic coefficients in the approach
trajectory. The CD increases as the aircraft descends. The small CD drop following
the increase in flap setting is due to a slight decrease in main under-carriage drag.
11.3 Continuous Descent Approach 307
CDA
A B
D C
7,000 ft
3,000 ft
Normal
o
3
~40 km ~25 km ~15 km
Touch-down
Figure 11.6. Continuous descent approach (CDA) versus conventional approach (not on
scale).
This effect is due to the blocking of the flow past the under-carriage caused by the
flap deflection.
11.3 Continuous Descent Approach

In a continuous descent approach (CDA), the aircraft descends on a fixed glide slope
of about 3 degrees from an altitude of about 7,000 feet above the airfield. In this
procedure, there is no level flight until the aircraft reaches the airfield. By contrast, a
conventional flight path is made up of three segments, the second of which is a level
flight at an altitude of 3,000 feet. The sketch in Figure 11.6 shows the differences
between the conventional flight path and the CDA. A number of advantages are
claimed with this new approach, including:
1. With the aircraft flying higher than normal, the distance to noise receivers on
the ground is larger and the noise impact is lower.
2. With the aircraft flying higher, without a level segment, less engine thrust is
required, which also contributes to lower noise emissions.
3. As a consequence of the previous point, also the fuel consumption is reduced.
First, a CDA does not affect the perceived noise EPNL (see Chapter 17) because
by the time the airplane switches from conventional to CDA procedures, it is way
too far to be picked up by the reference microphones (further discussion is in
Chapter 18). The level flight segment at 3,000 feet, taking place over a distance
of up to 10 km (∼5.5 n-miles), can affect some communities below the flight path.
On the CDA trajectory at ∼25 km, as indicated in Figure 11.6, the airplane main-
tains an altitude of ∼520 m (∼1,700 feet) above the conventional trajectory at point
D. Thus, the flight altitude point D has increased to ∼4,700 feet, which is a 57%
increase. Noise radiation is affected by the distance r as 1/r 2 , all other parameters
being the same. Hence, the CDA trajectory would have a noise signature ∼4dB less
than a conventional level flight at point D; this result partially justifies the claim of
reduced noise. Further discussion of noise performance is available in Chapters 16
and 17.
Table 11.2. Descent report for the A320-200, conventional descent. Aircraft mass at top of
descent: 69,380 kg; standard day, no wind (calculated)
h t X vd vd mf ṁ f
KTAS M [m] [min] [n-m] [m/s] [ft/min] [kg] [kg/s]
0 435.07 0.755 10,668

1 434.49 0.754 10,643 0.04 0.27 11.13 2,192 0.9 24.000
2 315.12 0.502 4,393 11.07 67.75 9.41 1,852 265.4 23.972
3 268.01 0.427 4,393 2.03 9.88 0.00 0 57.0 28.038
4 226.79 0.347 943 9.13 37.40 1.04 205 221.2 24.229
5 226.85 0.347 943 2.01 7.59 0.00 0 79.8 39.732
6 222.03 0.338 493 1.31 4.88 5.74 1,130 32.4 24.786
Totals 25.59 127.77 656.7
With regard to fuel consumption, point 3 is demonstrated by the results in

Tables 11.2 and 11.3, which are for the Airbus A320-200 with CFM engines. These
tables are summaries of aircraft performance, with the aircraft starting at the top of
descent at the same conditions (altitude, Mach number, weight). The CDA proce-
dure saves about 80 kg of fuel and takes place over a shorter distance.
A more extreme case is the one of continuous descent from the final cruise
altitude. The same aircraft burns the following amounts of fuel: ∼636 kg in a con-
ventional descent; ∼556 kg in a CDA descent; and ∼450 kg in a continuous descent
from the TOD. Thus, a saving of 180 kg of fuel could be achieved.
11.4 Steep Descent

The term steep descent refers to a glide slope greater than the conventional slope.
Steep descent trajectories can reduce noise by virtue of a larger distance from the
receiver on the ground but are more complicated to perform than conventional
trajectories. In this section we highlight some of the problems, mostly focusing on
the aerodynamics. By assuming a small angle of gliding and using the definition of
the CL, the descent rate is written as
"
2 W CD
vs = . (11.9)
ρ A CL3/2
Table 11.3. Descent report for the A320-200, continuous descent approach; aircraft mass at
top of descent: 69,380 kg; standard day, no wind (calculated)
MM h t X vd vd mf ṁ f
KTAS M [m] [min] [n-m] [m/s] [ft/min] [kg] [kg/s]
435.07 0.755 10,668

1 434.49 0.754 10,643 0.04 0.27 11.13 2,192 0.9 24.000
2 315.12 0.502 4,393 11.07 67.75 9.41 1,852 265.4 23.972
3 268.01 0.427 4,393 2.03 9.88 0.00 0 57.0 28.038
4 222.03 0.338 493 10.44 42.28 1.63 321 253.6 24.298
Totals 23.58 120.18 576.9

11.4 Steep Descent 309
9.5
L/D Max L/D 0.11
CD / C3/2
L
0.1
8.5
3/2
CL / CD
CD / CL
0.09
8
0.08
7.5
CD / C3/2
L min
7 0.07
1 1.5 2 2.5 3
CL
Figure 11.7. Aerodynamic parameters of the Airbus A310 scaled airplane.
The slope of the trajectory is γ = vs /U. For a fixed weight, a change in the descent
slope requires a change in the aerodynamic coefficients. If the descent is performed
at the same TAS as the conventional flight, an increase in γ requires an increase in
descent rate. Alternatively, if the descent rate is kept to the value of the conventional
trajectory, the TAS would have to decrease; in this case there is a risk of stalling the
aircraft. The TAS must decrease according to
U sin γ = const. (11.10)
If the underscore “r” denotes the reference flight conditions (or conventional tra-
jectory), the ratio between the steep-descent TAS and the reference TAS is
3/2
U γr CDr CL
= . (11.11)
Ur γ CD CLr
At a constant CL, the drag would have to increase as much as the flight-path angle
(in relative terms). However, a slower air speed would require a higher CL. There is
a combination of flap and slat settings (for example, Table 11.1) to achieve this. In
3/2
any case, the CD would have to grow faster than the CL ; in other words, the factor
CD
fd = 3/2
(11.12)
CL
must be larger in a steep-descent trajectory than in a conventional trajectory. A plot

of such a parameter is shown in Figure 11.7, which refers to a wind-tunnel model
of the Airbus A310 2 . High values of the aerodynamic factor fd are obtained with
either very low or very high CL. Low values of CL would increase the stall speed and
thence the risk of stalling the aircraft as it slows down. Figure 11.7 also shows that
the aerodynamic factor is at a minimum at relatively high CL values. Equation 11.11
Limit of
descent factor
γ=4
0.8
Possible operation γ = 4.5
V / Vr γ=5
0.6
Stall Limit
0.4
0.2
0.05 0.075 0.1 0.125 0.15
3/2
CD / C L
Figure 11.8. Parametric study of steep descent.
must be solved with the constraint

U Vstall
≥ Ks . (11.13)
Ur Ur
This limit is slightly more restrictive than the one required by FAR § 25.103. The
stall speed is defined as

2W 1
Vstall = , (11.14)
ρ A CLusable
where CLusable denotes a fraction of the maximum lift coefficient. Next, we need to
solve Equation 11.11 with the constraint Equation 11.13 and the definition Equa-
tion 11.14. With a fixed configuration, there is a small decrease in the Vstall as the
aircraft descends.
Now fix the flight-path slope. The aircraft is required to fly along this trajectory
from the initial altitude so that at the final altitude the descent rate is below a
threshold value considered safe:
vs fd
vs < vsr , = . (11.15)
vsr fdr
A parametric analysis of Equation 11.15 is shown in Figure 11.8. The horizontal
line denotes the stall constraint, Equation 11.14 for this case. A relative air speed
U/Ur below this line is not feasible. The other constraint is the band delimited by
the two vertical lines, 0.078 < fd < 0.115, which is the aerodynamic limit shown in
Figure 11.7. The wing system cannot provide a value of fd below or above this band.
It is evident that a descent with a slope γ = 5.0 degrees is not possible (see lower
right corner of the constraint box). A descent at γ = 4.5 degrees is only possible if
the aircraft operates at low CL, as implied by Figure 11.7. However, there is a narrow
band in the aerodynamic performance that allows a descent at γ = 4.0 degrees. To

extend the descent envelope, the aircraft will have to provide
3/2
fd = CD/CL > 0.12.
The aerodynamic limit will have to shift to the right of the graph for a very steep
descent to be possible. The next problem is to examine how this can be achieved.
There are four possibilities: 1) increase the CDo at given CL; 2) decrease the CL at
fixed CDo ; 3) increase the CDo and decrease the CL at the same time; and 4) increase
both CDo and CL, with the condition that fd increases. A detailed analysis is carried
out in Ref. 1 that also shows some control theory on trajectories with increasing
angles γ . In brief, the most effective method for increasing the glide slope is to
increase the non-lifting portion of the drag. This can be achieved with a combination
of technologies, such as clam shells (BAE RJ 146), conventional spoilers, or inverted
spoilers 3 .
11.5 Unpowered Descent

A fixed-wing aircraft is said to be gliding if it maintains a stable and controlled flight
with minimum or no engine power. In the following sections we discuss some gliding
flights and their optimal conditions.
11.5.1 Minimum Sinking Speed

The sinking speed is
"
2 W CD
vs = U sin γ = , (11.16)
ρ A CL3/2
having assumed that CD 2

CL2 . The best glide ratio is in the range of 14 to 18,
2
depending on the aircraft; hence CD is less than 0.5% of CL2 . If we maintain the
convention of positive γ clockwise from the horizontal plane, then the sinking speed
is a negative vertical velocity. From Equation 11.16 the condition of minimum sinking
speed is equal to the condition of minimum power for a powered aircraft:

2W k
Ump =
2
. (11.17)
ρA 3CDo
The corresponding value of the aerodynamic factor is

CD 4CDo
= . (11.18)
3/2
CL (3CDo /k)3/4
From Equation 11.18 the minimum sinking speed is

2 W 1/2 4CDo
vs = . (11.19)
ρ A (3CDo /k)3/4
Equation 11.19 shows that the minimum vs decreases as the aircraft descends –
which is a good thing. Also, the air speed decreases as the aircraft descends, which
is another good thing, as long as the speed can be safely maintained above stall (for
example, Equation 11.14) and the aircraft is fully controlled.
11.5.2 Minimum Glide Angle

The flight-path angle in absence of engine thrust is given by
T−D D D
sin γ = =− − . (11.20)
W W L
A minimum for sin γ is also a minimum for γ ; this occurs for an aircraft operating
at maximum glide ratio. Therefore, an optimal condition at a fixed altitude is

∂ D
= 0. (11.21)
∂U L
This condition was found previously (Equation 8.51). We proved that this was a
condition of minimum drag (or green-dot speed, Vgreen ):

2W k
2
Umd = . (11.22)
ρA CDo
We now compare two different descent conditions: minimum sinking speed, as pre-
viously discussed, and minimum glide angle. The minimum sinking speed is given by
Equation 11.17; therefore the ratio between the speed of minimum glide angle and
the speed for minimum descent speed is
Uvs 1
=√
4
= 0.7598 0.76. (11.23)
Uγ 3
Therefore, the speed of minimum descent rate is equal to 76% the speed of the
minimum glide angle. This ratio is the same as the ratio between the speed of
minimum power and minimum drag (Equation 8.51),
Uvs Ump
= . (11.24)
Uγ Umd
The flight path can be found from the steady-state equations
∂x ∂h
U, = vs , (11.25)
∂t ∂t
with the additional condition
vs
γ. (11.26)
U
The solution of this problem is found numerically. A typical result is shown in
Figure 11.9. An aircraft of the class of the Airbus A300-600 with a mass m = 145,000
kg was simulated without any meteorological effects. The initial altitude was h =
11,000 m (∼36,000 feet), with an initial Mach number M = 0.79.
The maximum gliding range is ∼225 km (∼120 n-miles); this is obtained with a
minimum-glide-angle program. This is enough to establish that the aircraft without
fuel should be able to glide the whole distance of 100 n-miles without any engine
power. The gliding time is ∼32 minutes. Head or tail winds will change the result.
Ground distance, n-miles

0 50 100
12 40
Min glide angle
Min glide speed
10
30
Altitude, 10 ft
Altitude, km
3
6 20
4
10
0 0
0 50 100 150 200 250
Ground distance, km
Figure 11.9. Glide range at two different flight conditions. Aircraft mass m = 145,000 kg.
11.5.3 General Gliding Flight

The results shown in the previous cases with the quasi-steady approximations can
also be obtained by integration of the flight dynamics equations. If some thrust is
available, the equations are
T cos α D
−
U̇ = − g sin γ , (11.27)
m m
T sin α L cos γ
γ̇ = + −g , (11.28)
Um Um U
having assumed that the thrust is aligned with the drag force. Because the glide slope
is small, we make an approximation:
vs
sin γ γ , cos γ 1, U V. (11.29)
U
Therefore, the drift-down equations become
T cos α D
U̇ =− − gγ , (11.30)
m m
T sin α L g
γ̇ = + − . (11.31)
Um Um U
These equations can be integrated as an initial-value problem. The use of the full
flight-path equations is required to study more general gliding problems, such as
gliding at constant CL, gliding with and without pitch damper. A glide at constant
CL may give rise to a large-amplitude oscillation (phugoid), if a pitch damping is
not used. The phugoid oscillation occurs when a powered or an unpowered vehicle
operates at a speed and attitude away from its equilibrium and when no adjustments
are made to its flight control systems.
Generally, one can look at the differential equations in the phase plane with
coordinates U, γ . This plane contains parametric plots of γ (t) versus U(t). Whereas
it may not be possible to solve the ODEs in closed form, it is always straightforward
to plot the tangent vectors to the solution trajectories in the phase plane. The traject-
ories in the phase plane themselves satisfy a first-order ODE. Thus, by eliminating
t between U and γ , we get the ODE
∂γ γ̇ T sin α/Um + L/Um − g cos γ /U
= = . (11.32)
∂U U̇ T cos α/m − D/m − g sin γ
It can be verified that this derivative is oscillating, and so is the solution, by an
amount that depends on the value of CL. A non-oscillating solution requires that
γ̇ /U̇ be uniform.
The stability problems related to the phugoid have been the subject of analyt-
ical investigation since Lanchester 4 ; they have been treated in virtually all books
on stability and control, with approximations, linearisations and closed-form solu-
tions (for example, Perkins and Hage 5 , Etkin 6 , McCormick 7 , Etkin and Reid 8 ,
Nelson 9 ). These books cite further references on the subject. In addition, Campos
et al. 10 discuss the speed stability of an aircraft in a dive.
11.5.4 Maximum Glide Range with the Energy Method

We solve the maximum glide range for a supersonic jet fighter with the energy method
and show how the solution to this problem can be substantially different from that
of subsonic flight. Take the definition of the horizontal velocity component, U = ẋ,
and divide by Equation 10.46
ẋ U
= . (11.33)
Ė (TU − DU)/m
In the absence of thrust, this equation can be simplified to
ẋ m
=− (11.34)
Ė D
or
∂x m ∂E D D
=− , = − = −g . (11.35)
∂E D ∂x m L
To maximise the gliding range, we must minimise dE/dx, subject to the constraint
L = W. In other words, the aircraft must descend by moving along a flight path
of minimum-drag or maximum-glide ratio. If we eliminate h in the drag force by
introducing the energy height, then the problem is a minimum drag with respect to
the flight speed:
min = −D(E, U), (11.36)

U
for a fixed energy level. It is possible that the initial conditions of the aircraft (speed
Uo and altitude ho for given energy level Eo) are not on the maximum range path.
This implies that the drag is not at a minimum, and the aircraft must zoom-dive or
zoom-climb to the optimal energy level. Once it has reached the point on the E = Eo
level, the aircraft glides along the optimal flight path. The flight-path angle is found
11.6 Holding Procedures 315
from the dynamic equation along the flight path with T = 0:

D 1 ∂U
sin γ = − + . (11.37)
W g ∂t
Because the derivative of the speed is negative, this term decreases the drag and thus
extends the glide path. The glide range is found from integrating Equation 11.35. It
consists of two terms:
2 E1 E2
m m
X= dx = dE + dE, (11.38)
o Eo D E1 D
where Eo is the initial energy; E1 is the energy at ground level; and E2 is the minimum
level speed at ground level. The first term in Equation 11.38 is due to the loss of
energy from the starting point to ground level; the second term is due to deceleration
at ground level, to a speed above stalling speed. Therefore, the aircraft can glide to
nearly ground level; then it performs another stretch by increasing the angle of attack
and deploying the high-lift control surfaces. If the initial speed is supersonic, the
aircraft will first dissipate its kinetic energy whilst keeping nearly constant altitude;
then it will glide, losing also its altitude (or potential energy).
11.6 Holding Procedures

It is 8:30 a.m. at London Heathrow. There are about 20 aircraft waiting to land. The
sky is speckled with their lights in the far distance. Each of these airplanes is burning
between 0.5 and 1.5 kg/s of fuel. The ATC is striving to manage this congestion.
Unfortunately, the aircraft may be required to “hold” on a “stack” before being
allowed to land, which causes considerable waste of time and fuel. The situation is
similar at major airports around the world every day.
When aircraft are forced on a “hold” flight path due to local air traffic congestion,
they are placed on a racetrack pattern made of two parallel segments joined by half-
circles (stacks). A typical hold altitude is 5,000 or 7,000 feet. Thus, the aircraft must
perform two U-turns for every lap. In a hold situation, range is not important but
endurance is 11 . Minimum-fuel turns are discussed in § 13.2.4 and show that minimum
turns are performed at the speed of minimum drag.
The minimum fuel flow corresponds approximately to the minimum drag (for jet-
powered airplanes) and to the minimum power (for propeller-powered airplanes).
In the former case, the best speed for a clean configuration is the speed of maximum
L/D (green dot speed). Thus, “green dot” is selected to hold the airplane.
There are instances when the green dot speed is too high, due to obstacle
proximity. If the aircraft is forced to maintain a lower speed, it must do so with
a slat/flap extended by one notch (CONF1, or “S” speed on Airbus). There are
also cases when the ATC requires the aircraft to maintain a speed that can be
quite different from the green dot. The hold altitude is not a free parameter; this
is enforced, but the rule is simple: the higher the flight altitude, the lower the fuel
flow.
An example of this situation is shown in Figure 11.10, which displays the sim-
ulated hold performance of the Airbus A320-200 with CFM engines. The green
dot speed is estimated at 248 kt at 10,000 feet (∼3,048 m), at the weight indicated.
Hold Mach number

0.3 0.4 0.5
0.85
MRM
0.8
Fuel flow, kg/s 0.75

LRM
0.7
0.65
Green Dot
0.6
Min Fuel
0.55
150 200 250 300 350
KTAS
Figure 11.10. Holding speed of the Airbus A320-200 at 10,000 feet (3,048 m), m = 69,380 kg
(calculated). Standard day, no wind.
The ICAO recommends that the hold speed does not exceed 230 kt at this altitude;
therefore the CONF1 configuration will have to be selected (see Table 11.1). A few
other points are indicated in the figure: the minimum fuel flow speed, the long-range
Mach/speed and the maximum-range Mach/speed. It is worthy of note that in this
case the green dot speed does not correspond to minimum fuel, a fact evidenced by
Airbus on some of their airplanes † .
If conditions allow, it is preferable to perform a linear hold by extending the
cruise at the green dot speed and then return to the airfield to join the stack for a
more limited amount of time. Each major airport has its own hold areas, variously
named for easy identification.
11.7 Landing Performance

Landing is the final flight segment of the aircraft when, upon reaching an altitude
of about 50 feet above the airfield, the aircraft proceeds to touch-down, rotation
and ground run, until the aircraft is stopped or taxied off the runway. Landing is a
nearly completely automated process but one that is fairly complicated by a variety
of factors due to atmospheric conditions such as precipitation, wind gusts, fog and
poor visibility at night. The first instrumented landing that took place in 1929 is now
an almost-forgotten piece of aeronautical history. There are two distinct phases: the
airborne phase and the ground run, exactly the reverse of the take-off. A sequence of
events is shown in Figure 11.11. Table 11.4 is a summary of relevant aircraft speeds
† Airbus, Flight Operations Support. Getting to Grips with Fuel Economy, Issue 3. Blagnac, France
(2004).
Airborne ground distance
Touch-down Stop
50 feet
VApp VTD VMBE Clearway

Actual landing distance
Landing distance available
Figure 11.11. Sequence of events in a landing procedure.
at landing. The required landing distance must be less than the landing distance
available‡ .
11.7.1 Airborne Phase

The airborne phase starts at a conventional altitude above the airfield (say, 50 feet),
with the airplane coming down at a relatively high-pitch attitude. The airplane slowly
pitches up as it descends. This phase ends when the main landing-gear wheels touch
the ground. This point is critical for the whole under-carriage unit because it can be
subject to extreme loads. Because there is no pre-spin of the tyres, at touch-down
there is an impulsive load that sets the wheels in motion to follow the speed of the
airplane. The impact energy E between main tyres and ground can be estimated
from

1 L
E= m− V 2 sin2 γ , (11.39)
2 g
where L is the aerodynamic lift at touch-down, V is the ground speed and γ is

the descent angle. This impact energy can be decreased by increasing the lift and
decreasing the approach angle.
Taking the aircraft to a halt on the runway requires finding ways to dissipate
the mechanical energy of the vehicle. This energy is roughly proportional to the
square of the velocity at the touch-down. Because all aircraft must take off and land
on the available runway length, the landing speed is somewhat independent of the
gross weight; therefore the energy that must be dissipated by the braking system is
proportional to the aircraft’s weight.
The length of the airborne phase is x1 ∼ href cos γ , where href is the altitude of
the conventional screen; the value of x1 is generally below the uncertainty of the
length of the ground run, as discussed next.
‡ The relevant regulations for transport aircraft are the FAR § 25.125 and § 121.195.
Table 11.4. Definition of landing speeds
Definition Symbol Description
Air speed at screen Vapp Steady approach, Vapp > VS

Touch-down speed VT D MLG touch-down, VT D/Vapp = 0.90–0.98
Minimum control speed, AEO VMCL Trimmed, most unfavourable CG & weight
Minimum control speed, OEI VMCL−1 Trimmed, φ < 5 degs, most unfavourable CG
11.7.2 Landing Run

With the airplane pitching up, touch-down is done with the main (rear) wheels. The
shock absorbers of the main landing gear suffer the vertical load of the airplane.
Upon deceleration, the aircraft pitches down until the nose wheels hit the runway
and then it is taken to a halt by various brakes. The runway conditions and the effects
of the brakes are discussed in Chapter 9.
There are several techniques to force the airplane to reduce its pitch attitude
upon landing. A combination of CG position, elevator setting and ground speed can
be used. Forward CG position and deceleration due to high drag work in combination
to produce a pitch-down moment.
The deceleration technique is similar to the case of aborted landing. The main
difference is that the airplane is lighter and should come to a stop in a shorter distance.
Brakes of a various nature can be used (thrust reversal, tyre brakes, ailerons). Thrust
reverse is to be used sparingly and only when the other means of bringing the aircraft
to a halt are not sufficient. At certification level, the landing distances have to be
demonstrated without thrust reversal. The differential equation is
∂U
m = D + μr (W − L) + W sin γr + B = R, (11.40)
∂t
where the right-hand side is the total resistance, B is the total braking force, and
γr is the slope of the runway. The CL and CD correspond to landing configura-
tion and account for ground effects. The landing run is obtained by integration of
Equation 11.40. The first step is to recast the equation as
∂U R
U = . (11.41)
∂x m
Integration of the last equation is performed between the touch-down point (sub-
script “o”) and the point at which the aircraft halts (subscript “1”). The result is the
landing distance
1
U
xl = m dU. (11.42)
o R
CLOSED-FORM SOLUTIONS. Assume that the runway has no inclination (γr = 0) and
that there is no braking force. The solution is found from
1
1
xl = m UdU. (11.43)
o D + μr (W − L)
By defining the constant coefficients c1 = ρ A(CD − μr CL)/2 (with constant aero-

dynamic coefficients) and c2 = μr W, the integral assumes the form
1
1 dU 2 1
= ln(c1 U12 + c2 ). (11.44)
2 o c1 U 2 + c2 2c2
Therefore, the landing run is
m
xl = ln(c1 U 2 + c2 ). (11.45)
2c2
Equation 11.45 is a simplification because in practice some form of braking is used
(aero-brakes, wheel brakes, and so on). The solution is similar to the one outlined
if the braking force is either constant or proportional to U 2 . The braking force is
expressed as
B = c3 T + c4 U 2 (11.46)
and contains the effects of the thrust reversal (via the coefficient c3 ) and the mech-
anical brakes (via the coefficient c4 ). A constant braking force −T is assumed in the
case of thrust reversal. The total resistance is
R = D + μr (W − L) + c3 T + c4 U 2 . (11.47)
The integral of Equation 11.42 becomes

1
1 1
xl = m dU 2 . (11.48)
2 o ρ A(CD − μr CL + c4 )U 2 /2 + μr W + T
In this case, the coefficients are
c1 = ρ A(CD − μr CL + c4 )/2, c2 = μr W + T, (11.49)
and the solution has the form of Equation 11.45.
NUMERICAL SOLUTIONS. Integration of Equation 11.40 is done numerically with a

time-marching scheme to provide the complete trajectory of the airplane as well
as the key performance parameters (airplane state, engine state). It is useful to
introduce some operators that define the state of the airplane and to calculate the
exact aerodynamic coefficients, which in the general case are not constant. In normal
circumstances, the change in state is defined by aileron, spoiler and flap settings.
The differential Equation 11.40 is solved with the relevant equations for the
rolling coefficients and normal loads on the wheels, as discussed in Chapter 9. Fig-
ure 11.12 shows the landing run of a model Airbus A320-200 with an initial mass
m = 59,900 kg at sea level with a head wind Vw = −2 m/s.
Landing is complicated by several factors, including atmospheric conditions.
Therefore, the data quoted by the manufacturers refer to a statistical analysis rather
than precise measurements or other predictions. However, abstracting from some of
these stochastic events, the main parameters reported in the FCOM are the landing
weight, the airfield altitude, the wind speeds and the flap setting. An example of
landing charts is shown in Figure 11.13. This figure displays the FAR runway length
0.2
2 150
Main LG
Distance
Nose LG
KTAS
120
0.15
1.5
Ground speed, kt
Rolling coeffs.
Distance, km
90
1 0.1
60
0.5 0.05
30
0 0 0
0 10 20 30 40 50 0 10 20 30 40 50
Time, s Time, s
(a) (b)
Figure 11.12. Landing run of the Airbus A320-200 model at sea level, Vw = −2 m/s, m =
59,900 kg (calculated).
requirements for the Boeing B747 (various versions, as indicated in the graph) for a
fixed-flap configuration (25 degrees), as elaborated from Boeing 12 .
11.7.3 Crab Landing

Strong cross winds at landing are a cause of hazard. Maximum wind speeds are recom-
mended and/or certified, along with procedures to control the aircraft. Adverse
weather conditions are encountered in about one-third of all landing accidents.
3
Landing weight, 10 kg
220 240 260 280 300
11
MLW
B-747-400F
FAR landing runway length, 10 feet
MLW
B747-400, -400C
FAR landing runway length, km
10
3
MLW 3
B747-400
domestic
9
m)
38 el 2.5
8 ( 2,4 al
ev
0 0 ft se
6,0
m)
38 el
(2,4 ev
7 0 ft al
6,00 se
2
6
Dry runway
Wet runway
FLAPS = 25 degs
5
500 550 600 650 700
3
Landing weight, 10 lb
Figure 11.13. Parametric effects on the landing field of the Boeing B747-400.
Table 11.5. Limit crosswind speeds coupled with runway conditions
Braking Runway Maximum Runway

action friction crosswind condition
Good > 0.40 Tested Dry or wet; no aquaplaning

Good/Medium 0.36–0.39 30 kt Slush
Medium 0.30–0.35 25 kt Dry snow
Medium/Poor 0.26–0.29 20 kt Standing water, wet snow
Poor < 0.25 15 kt Risk of aquaplaning
Unreliable — 5 kt
Runway contamination adds to the hazard* . Table 11.5 is a summary of limit cross-
wind speeds, runway-friction characteristics and braking-action effectiveness.
The final approach must be performed with wings level and with a steady side-
slip angle to align the longitudinal axis of the airplane with the centre of the runway.
This operation must be done with a combination of flight controls (aileron–rudder).
During the flare, any tendency of the airplane to roll downwind should be minimised
with an appropriate response of the side-stick. In any case, the airplane should not
have a crab angle or bank angle φ > 5 degrees.
The limitations involved in crab landing (Figure 11.14) can be summarised by
two conditions: 1) bank angle for a given crab angle, and 2) crab angle for a given
bank angle. Figure 11.15 shows one such chart with the approach speed Vapp as the
independent parameter. In Figure 11.15, point A denotes the condition of zero-crab
angle (at the given Vapp ), which corresponds to a bank angle φ 3; point B is the
condition of zero-bank angle, which corresponds to a crab angle of about 4 degrees. A
negative crab angle (aircraft pointing away from the wind) is a result of an excessive
rudder control. The data in Figure 11.15 change with the wind speed. There are a
number of limitations to be considered, including geometry (wing and tail strike)
and control authority (ability to maintain a steady side-slip under crosswinds). It
is possible that with high crosswinds, the aircraft is unable to perform a wing-level
landing. Even if the geometrical conditions would allow a large bank angle, there is
a risk of damage to the main landing gear.
Calculation of charts like the one shown in Figure 11.15 can be done with the
method explained in § 7.2 (control speed in air); the aerodynamic derivatives of the
airplane to side-slip and roll will have to be determined before setting up the system
of equations.
WING STRIKE. Aileron and rudder attitude are important for lateral control of the
aircraft. However, the geometric configuration is what determines the severity of
ground strike when the controls are ineffective. An example of wing strike is shown
in Figure 11.16. In this specific case, Figure 11.16a shows that the line between the
wheel contact (W) and the wing tip (T) runs above the lowest point of the engine
(E). Therefore, this would be a case of engine strike. If the engine’s diameter is
scaled down (Figure 11.16b), the engine is above the W-T line and is “protected” by
* Flight Safety Foundation, Flight Safety Digest, Vols. 17 (Nov. 1998) & 18 (Feb. 1999).
cross−wind
d. c. b. a.
reverse directional aircraft touch−down

thrust & control skidding
braking regained sideways
Figure 11.14. Crab landing in presence of strong side gusts.
the wing tip. In the latter case, the maximum bank angle φ that prevents wing strike
can be estimated from
2hg
tan φ = tan ϕ − tan θ tan , (11.50)
b − bt
where is the wing sweep, bt is the wheel track, hg is the height of the landing gear
and ϕ is the wing’s dihedral angle. There are cases in which the aircraft strikes the
ground with the engine instead of the wing tip.
B
4 φ=0
φ=2
Crab angle, degs
0 A φ=4
φ=6
-4
φ=8
-8
Vapp
-12
110 120 130 140 150 160
KIAS
Figure 11.15. Crab angle versus bank angle for a given wind speed (extrapolated from Airbus
data).
11.8 Go-Around Performance 323
(a) Engine strike
(b) Wing tip strike

Figure 11.16. Engine and wing strike.
11.8 Go-Around Performance

As illustrated in Figure 11.4, there is the possibility that the aircraft is unable to land
and thus it is forced to go-around. This means that landing is aborted, the throttle
is increased, the aircraft is set onto a climb flight path (after rotation), the landing
gears are retracted and the high-lift systems are retracted by one notch (or more,
depending on circumstances). The engines are unable to respond immediately to the
request of additional thrust (see also § 5.2.2 and Figure 5.2).
During the final approach the engine is set on idle, or nearly so. If a go-around
is required, the engine must spool up quickly to provide the required thrust.
CASE STUDY: GO-AROUND OF AN AIRBUS A319-100 MODEL. Figure 11.17 shows two
simulated trajectories of an Airbus A319-100 airplane model with CFM56-5B5
turbofan engines. The assumptions include standard day, maximum landing weight
Flight time, s
0 2 4 6 8 10
0.4 0
-5
Loss in flight altitude, m

Flight altitude, km
0.39 -10
-15
0.38 -20
-25
VAPP & IDLE Engines
VAPP - 10kt & IDLE Engines
0.37 -30
0 0.2 0.4 0.6 0.8
Flight distance, km
Figure 11.17. Go-around manoeuvre of an Airbus A319-100 model, starting with maximum
landing weight at 400 m (∼1,310 feet) above sea level, standard day.
at the start of the manoeuvre, 1-second time delay for flap reduced by one notch,
and 2-second time delay to landing-gear retraction. In both cases, the airplane starts
the manoeuvre from a 3-degree descent slope at 400 metres (1,310 feet), either with
the approach speed or approach speed reduced by 10 kt. The loss in flight altitude is
estimated at 12 m (40 feet) and 25 m (83 feet), respectively. In both cases the engine
is in idle mode at the initial point and requires in excess of 5 seconds to recover to
full thrust.
The result in Figure 11.17 has been obtained with an integration of the differ-
ential equations for the centre of gravity of the airplane. At the initial point, t = 0;
x = 0. The loss in flight altitude can be reduced if the airplane is operating in sta-
bilised mode, with the engine at a minimum setting (not in idle). In fact, it would
take less time for the engine to respond to the increase in fuel flow commanded
by the throttle. There is a combination of flight altitude and speed that will not
allow the safe recovery of the airplane; in that case the airplane must proceed to
landing. Parametric studies around this problem include the effects of initial altitude,
weight, atmospheric condition, time lag in pilot’s control actions, and engine response
modes.
Summary
In this chapter we presented the problem of powered and unpowered descent in
a fairly general form. We started from the en-route descent; we showed that the
aircraft is required to follow standard flight paths with two or more level flight seg-
ments, which may increase the flight time and the fuel burn. We have shown that
the CDA saves fuel. A more complex manoeuvre consists in setting the aircraft on
a larger glide slow (steep descent). We demonstrated that although there can be
some advantages, such as reduced noise at a receiver on the ground, the procedure
is seriously limited by the aerodynamics and that it is not possible to exceed 4.5 to 5
degrees. We then reviewed the holding procedures and stacking-patterns practices.
Finally, we considered landing problems, such as a conventional landing, crab land-
ing (in presence of side gusts), the risk of wing and engine strike and go-around
performance. In the latter case, we demonstrated that to be able to take control
of the aircraft in the case of an aborted landing, it is necessary to operate with a
minimum thrust. This procedure allows a more rapid recovery of thrust, which must
be used to go-around, that is, climb out of the airfield.
Bibliography
[1] Filippone A. Steep-descent manoeuvre of transport aircraft. J. Aircraft,
44(5):1727–1739, Sept. 2007.
[2] Kiock R. The ALVAST model of DLR: Technical Report IB 129 96/22, DLR,
Lilienthal Platz, 7, D-38018 Braunschweig, Germany, 1996.
[3] Filippone A. Inverted jet spoilers for aerodynamic control. J. Aircraft,
46(4):1240–1252, 2009.
[4] Lanchester FW. Aerodonetics. Constable, London, 1908.
[5] Perkins CD and Hage RE. Airplane Performance Stability and Control. John
Wiley, 1949.
[6] Etkin BE. Dynamics of Flight. John Wiley & Sons, New York, 1959.
[7] McCormick BW. Aerodynamics, Aeronautics and Flight Mechanics. John
Wiley, 2nd edition, 1995.
[8] Etkin BE and Reid LC. Dynamics of Flight: Stability and Control. John Wiley
& Sons, 1996.
[9] Nelson RC. Flight Stability and Automatic Control. McGraw-Hill, 2nd edition,
1998.
[10] Campos A, Fonseca L, and Azinheira R. Some elementary aspects of non-
linear airplane speed stability in constrained flight. Progress Aerospace Sci-
ences, 31(2):137–169, 1995.
[11] Sachs G and Christodoulou T. Reducing fuel consumption of subsonic aircraft
by optimal cyclic cruise. J. Aircraft, 24(5):616–622, 1987.
[12] Anonymous. 747-400 airplane characteristics for airport planning. Technical
Report D6-58326-1, The Boeing Corporation, Dec. 2002.

Symbols in Table 11.4 are not repeated here.
A = wing area
b = wing span
bt = wheel track
B = braking force
ci = constant coefficients
CLo = zero-α lift-coefficient

D = drag force
E = energy
fd = aerodynamic parameter, Equation 11.12
h = altitude
hg = height of landing gear, including wheel
Ks = stall margin
L = lift force
m = mass
mf = fuel mass
ṁ f = fuel flow
M = Mach number
P = power
R = rolling resistance
t = time
T = net thrust
U = air speed
vs = descent rate
V = ground speed
Vapp = approach speed
Vgreen = green dot speed
Vstall = stall speed
VT D = touch-down speed
VMBE = maximum brake energy speed
Vw = wind speed
x = distance
x1 = airborne distance at landing
xl = ground run at landing
X = flight distance
W = weight
Greek Symbols
γ = descent angle
γr = runway-inclination angle
= wing-sweep angle
φ = bank/roll angle
ϕ = wing’s dihedral angle
μr = rolling-friction coefficient
ρ = air density
[.]d = descent
[.]hl = high lift
[.]l = landing
[.]md = minimum drag
[.]r = relative value
12 Cruise Performance
Overview
In this chapter we introduce the concepts of distance flown by an aircraft with and
without stop for refuelling (§ 12.1). We discuss a number of cruise programs at sub-
sonic and supersonic speeds and some optimal problems in long-range cruise, with
or without constraints. First we present the analysis of the instantaneous cruise para-
meters (§ 12.2), including the specific range. We then provide numerical solutions of
the specific range for real aircraft (§ 12.3). For the sake of generality, we also present
the range equation, which is solved in closed form (§ 12.4), and deal with the separate
problems of jet aircraft at subsonic speed (§ 12.5) and propeller aircraft (§ 12.6). We
show a number of more advanced studies in cruise altitude selection (§ 12.7), cruise
performance deterioration (§ 12.8), cost index and economic Mach number (§ 12.9)
and various other effects, such as the effects of high-altitude winds and centre of
gravity position (§ 12.10). We deal briefly with supersonic aircraft (§ 12.11).
KEY CONCEPTS: Point Performance, Specific Air Range, Cruise Altitude Selection,
Range Equation, Long-Range Mach, Maximum-Range Mach, Cost Index, Economic
Mach Number, Effects of Winds, Centre of Gravity Effects, Supersonic Cruise.
12.1 Introduction
For most commercial aircraft the fuel consumed during the cruising phase of the
flight makes up the bulk of the fuel carried and is a key factor in the productivity
and direct operating costs of an aircraft. Since the early 1970s, with the price of fuel
soaring, both the airlines and the military have been concerned with energy-efficient
operations. For this reason, several cruise conditions have been studied. The fuel
costs of major international airlines are constantly reviewed. Whilst the DOC have
been reduced considerably, the fuel continues to be one of the major cost items* .
Jet fuel prices have grown about 350% from the year 2000 to the current day. On
average, they currently make up to 30% of the total operating costs and in excess
* Fuel costs are monitored in real time by IATA and other stakeholders, who provide real-time data
such as price per gallon, price variations, impact of fuel bill on commercial aviation and more.
328
of 40% per block hour. Considerable research now exists on issues such as costs,
emissions, efficiency, fuel hedging and trends in aviation.
OPTIMALITY ANALYSIS. It has been demonstrated that the steady-state cruise is not
always fuel-optimal. Speyer et al. 1;2 demonstrated that an oscillating control leads to
a decrease in fuel consumption. Gilbert and Parsons 3 applied periodic cruise control
on the McDonnell-Douglas F-4 interceptor, and Menon 4 showed mathematically the
conditions under which oscillating solutions occur in a long-range cruise. Sachs and
Christodoulou 5 analysed the benefits of cyclic flight, with dolphin-type climbs and
glides, and showed that (at least theoretically) the range can be increased with such
a flight program. Due to the number of free parameters involved (Mach number,
altitude, lift coefficient, angle of attack, gross weight, block fuel) and number of
external constraints (ATC, flight corridors, atmospheric conditions), there is a variety
of optimal and sub-optimal conditions. ATC issues and terminal-area constraints are
reviewed by Visser 6 . Various ESDU data items deal with cruise performance 7 – 9 .
DEFINITIONS. The aircraft range is the distance that can be covered in straight flight at
a suitable flight altitude. The aircraft cannot use all of the fuel, and allowance must be
made to account for the terminal phases (take-off and landing), manoeuvres (loiter,
holding at altitude) and fuel reserves for contingency. In practice, the different flight
sections are calculated separately and hence the range equations are limited to steady
flight at altitude.
The block fuel is the fuel weight required to fly a specified mission and includes
the fuel to taxi at the airport. The fuel reserve is a contingency amount of fuel that
takes into account the risk of not being able to land at the destination airport. The
mission fuel includes 1) the fuel required to take off, accelerate and climb to the
initial cruise altitude; 2) the cruise fuel; 3) the descent, terminal area approach and
landing fuel; and 4) manoeuvring and reserve fuel. For the determination of the
gross take-off weight (TOW), the taxi fuel at the departure airport is not included.
The taxi fuel after landing is extracted from the reserve fuel.
The endurance is the time on station, that is, the time the aircraft can be flown
without landing or in-flight refuelling. Maximum endurance performance applies to
search and surveillance operations. Refuelling has been traditionally a problem for
the military, but some commercial operations are also envisaged 10 , on the grounds
that the amount of fuel that has to be carried by a wide-body aircraft on a long-haul
flight can be four times the useful payload. Smith 11 has published a historical account
of military refuelling technology.
12.2 Point Performance

The instantaneous conditions in aircraft cruise are called point performance. The
basic point parameters are the glide ratio, the specific range, the figure of merit (F M),
the instantaneous endurance and the glide ratio. Other instantaneous parameters of
interest include the product CL M2 or W/ρ. These parameters are important because
the optimal flight path or flight program for a long-range cruise can be derived from
integration of the point-performance parameters. Generalised point-performance
optimisation is discussed in detail by Torenbeek and Wittenberg 12 .
12.2.1 Specific Air Range at Subsonic Speed

The specific range is the distance flown by burning one unit of weight (or mass or
volume) of fuel. For operational reasons, other units may be used. For example, to
evaluate the productivity of an aircraft, the airlines may quote a fuel consumption
in unit of volume or weight per unit of distance per passenger.
If the symbol f j denotes the thrust-specific fuel consumption (TSFC), the range
dX obtainable by burning a small amount of fuel dm f is
U
dX = Udt = dm. (12.1)
ṁ f
The specific air range (SAR) is the derivative of the range with respect to the mass:
∂X U U
SAR = = = . (12.2)
∂m ṁ f fj T
The physical dimensions of SAR are distance × unit mass of fuel. If ṁ f as a func-
tion of the speed and altitude is known, the SAR can be plotted directly from the
first equivalence of Equation 12.2. For a subsonic jet aircraft with a parabolic drag
equation, the specific range is
U 1
SAR = = = f (h, U, m), (12.3)
fj D c1 f j (CDo + kCL2 )U
with c1 = ρ A/2. Equation 12.3 satisfies the conditions D = T, L = W. In Equa-
tion 12.3 the speed can be replaced by U = a M, and the mass can be normalised
with the aircraft mass at the start of the cruise (ξ = m/mi ). The SAR for a subsonic
commercial jet with a parabolic drag is shown in Figure 12.1. It is demonstrated that
1) at a constant flight altitude and Mach number, the specific range increases with
the decreasing aircraft’s mass; and 2) at a constant aircraft weight and Mach number,
the specific range increases with the cruise altitude.
For a flight at constant speed, one can study the effects of an aircraft’s initial
mass mi on the specific range. The effect of the cruise speed on the specific range can
be found by studying the derivative ∂SAR/∂U. The condition that gives the speed
of maximum SAR is
∂SAR
= 0. (12.4)
∂U
Assuming that the cruise TSFC is constant, solution of this equation leads to the
speed for best range

2W 1/2 3k 1/4
U= . (12.5)
ρA CDo
This speed depends on the wing loading and on the flight altitude.
SUPERSONIC FLIGHT. A similar expression can be derived for supersonic flight after
replacing the correct drag equation in Equation 12.2. The result is
1
SAR = , (12.6)
c1 f j aσ (CDo + ηCLα α 2 )M
20
h = 10,000 m
M = 0.76
Specific Air Range, m/kg

0.80
18
16
h = 8,000 m M = 0.76
0.80
14
12
0.6 0.7 0.8 0.9 1
Relative mass, ξ
Figure 12.1. Estimated specific range for a subsonic commercial jet with parabolic drag; mass
m = 160,000 kg.
with c1 = ρo A/2. The angle of attack is calculated from the vertical equilibrium
L = W, which requires the following equation to be satisfied:
2W 1
α = αo + . (12.7)
ρ Aa CLα M2
2
12.2.2 Figure of Merit

The figure of merit (F M) is defined as the product between the flight Mach number
and the glide ratio. At subsonic speeds with Mach numbers M < Mdd , we have

L CL c1 m/a 2 Mσ
M =M = , (12.8)
D CDo + kCL 2
CDo + kc12 m2 /σ 2 a 4 M4
where
2g
c1 = . (12.9)
ρo A
Equation 12.8 allows to study the effects of flight altitude, aircraft mass and flight
Mach number. The analysis is limited to a fixed Mach number. A parametric study
of Equation 12.8 shows that 1) at constant Mach number, the FM increases with
the flight altitude; and 2) at a constant Mach number and constant altitude, the FM
decreases with the decreasing aircraft’s mass (i.e., it decreases during the cruise).
For the transonic and supersonic cruise of a reference supersonic aircraft, we
use the alternative drag Equation 4.85 that represents correctly the effects of Mach
number on the aerodynamic drag of the aircraft. The FM becomes

L CLα α
M =M . (12.10)
D CDo + ηCLα α 2
As for the case of the specific range, there is no local maximum with respect to the
angle of attack. The dominant parameter in the FM is the flight Mach number. The
angle of attack in Equation 12.10 is to be replaced with the equilibrium in the vertical
direction, Equation 12.7, and yields an involved expression of the FM, in which the
free parameters are the flight altitude and the Mach number.
The FM of this type of aircraft has two local maxima: at subsonic and super-
sonic Mach numbers. These maxima can be found by bookkeeping in a numerical
procedure rather than by derivation. The exact values of these maxima depend on
the flight altitude and on the relative weight of the aircraft. Optimal flight conditions
over a long-range cruise are found from the operational conditions that always give
maximum specific range or maximum FM.
12.2.3 Weight-Altitude Relationship

As fuel is burned during the cruise, the weight decreases. Starting from the lift
equation:

2 W 1 2 W
CL M2 = = . (12.11)
A ρ a2 A p
For a constant FM, the product CL M2 must be constant. In the stratosphere, where
the speed of sound is constant, the aircraft must climb as it burns fuel in order to
maintain a constant W/ρ. In the troposphere, where the speed of sound is variable,
the aircraft must climb to maintain constant W/ p.
Continuous changes in altitude and Mach number are not allowed by the ATC.
By convention, two flight levels are 1,000 feet apart and are specified by the flight
altitude in feet divided by 100. Thus, 31,000 feet corresponds to the flight level 310,
or FL-310* .
Current regulations for commercial aviation require aircraft to maintain restric-
ted flight corridors above a transition altitude. There is no universal agreement on
this altitude because it varies from 18,000 feet in North America to 13,000 feet in
New Zealand to 3,000 feet in Europe. Some rationalisation can be expected in the
future. Anyway, above the transition altitude, the aircraft is bound to assume a con-
ventional flight level or climb between flight levels. Over a long-range cruise there
can be two or more such flight levels, depending on local regulations and computer
flight procedures. Typical separations are 1,000 feet, 2,000 feet, and 4,000 feet.
12.3 Numerical Solution of the Specific Air Range

The solutions shown so far are simplified analyses derived with a closed-form expres-
sion of the aerodynamic drag and without proper consideration of the engine per-
formance in terms of air speed, flight altitude and throttle setting. This section deals
with a numerical solution of the SAR problem and it proposes a detailed and valid-
ated analysis method. We start with the case of a turbofan-powered aircraft.
* Additional information on UK practice is available in the CAA document CAP 410: “Manual of
Flight Information Services”, Part A. CAA, Gatwick, UK (2002). ISBN 0-86039 851 X.
Airplane
Engine
State W, CG
Flight M, h
Figure 12.2. Flowchart for the numerical cal- Atmosphere dTemp

culation of the SAR.
Aerodynamics
Propulsion
SAR
Our analysis will always refer to a unit of mass (kg) of fuel burned, so that
the SAR is expressed in m/kg, km/kg or n-miles/kg. Starting from Equation 12.2,
we have:
U aM aM
SAR = = = = f (h, M, CD, f j , W). (12.12)
ṁ f fj T fj D
In the presence of a wind Uw , we also define a specific range (SR):
U U ± Uw Uw
SR = = = SAR ± . (12.13)
ṁ f ṁ f ṁ f
Note that in any case we want to maximise the ground distance flown by burning
a unit of fuel. With the SAR, it will always be possible to calculate an optimal
flight condition; this is defined as the operation point in the flight envelope (M − h)
that yields the maximum SAR. The flight trajectory can be established from the
relationship between SAR and all-up weight. From the relationship established
by Equation 12.12, a full analysis of the SAR is multi-dimensional because the
parameter depends on a fair number of factors, as illustrated in the flowchart in
Figure 12.2.
Once the airplane-engine model is chosen, we need to establish the airplane
state (weight and CG position), flight conditions (altitude and Mach) and atmo-
spheric conditions (limited to a change in temperature with respect to the standard
day). This is the block of input parameters which is required by the aerodynam-
ics model and the propulsion model. The combination of these models yields the
SAR. There are some more practical ways of performing an SAR analysis; these are
based on:
r SAR-versus-Mach at fixed AUW and fixed altitude. This analysis leads to the
calculation of the optimal Mach numbers for a given h − AUW.
r SAR-versus-altitude at fixed Mach number and AUW. This analysis leads to
the calculation of the optimal cruise altitude at the given M − AUW.
0.12
ISA, no wind
0.11
FL-290
FL-310
SAR, n-miles/kg
FL-330
0.1
FL-350
FL-370
0.09
MMO
FL-410
0.08
Max-range Mach
Long-range Mach
0.07
0.7 0.75 0.8 0.85 0.9
Cruise Mach
Figure 12.3. Calculated Mach number effects on SAR for a generic airplane, m = 150,000 kg;
standard day.
Figure 12.3 shows the SAR versus cruise Mach number at selected flight levels.
The trend shown in this figure is typical of other aircraft in the same class. In
particular, at Mach numbers above the point of maximum SAR, the SAR itself
decreases rapidly, mostly as a result of the transonic effects. To the left of the
point of maximum SAR, the deterioration of performance is more moderate. This
result has important implications in the cruise analysis. The best choice of cruise
Mach number is obviously the Mach corresponding to the maximum SAR. For a
given all-up weight and flight altitude, the Mach number corresponding to max-
imum SAR is called the maximum-range Mach number (MRM). Flying at this Mach
number guarantees the maximum fuel efficiency. A few comments are in place
here:
r For a given altitude and Mach number, the SAR decreases as the weight
increases.
r The maximum SAR is virtually independent of cruise altitude – which is a
considerable result. In fact, the aircraft can maintain a constant Mach number
throughout the cruise.
r At altitudes below 33,000 feet (FL-330; ∼10,000 m), the same Mach number
leads to a higher ground speed; this would result in a shorter cruise time.
r Slowing down the airplane is not viable because the fuel consumption would
increase.
r Having to choose between flying slower or faster for a given loss in SAR, it is
best to fly faster to gain time.
The latter point leads to the definition of the long-range cruise Mach number
(LRM). The long-range cruise Mach number is defined as the Mach number higher
than MRM that corresponds to 99% of maximum SAR. In other words, flying at
a Mach number equal to the LRM causes a 1% loss in specific range whilst flying
reasonably faster than the MRM. This conclusion is of considerable benefit in the
operation of commercial aircraft; in fact there is a widespread practice in operating
the aircraft at the LRM instead of the MRM, unless external factors intervene. As
in the case of the MRM, the LRM is only weakly dependent on the cruise altitude
and sometimes on the all-up weight. This is another important advantage because
it allows the aircraft to fly at a constant Mach over a wide range of altitudes and
weights, with only a minimal loss in efficiency.
TURBOPROP AIRCRAFT. In this case it is necessary to trim the propeller at the prevail-
ing flight conditions at the required thrust. The trim procedure is explained in § 6.3.
This procedure provides the correct propeller settings and the shaft power, which
is then used to solve the engine problem in the inverse mode (see also Chapter 5).
Numerically the problem is far more involved because it requires a propeller model,
an engine model, an aerodynamic model and their full integration.
12.3.1 Case Study: Gulfstream G550

This airplane is described in Appendix A. The verification of the computational
method with the FLIGHT code is shown in Figure 12.4. A comparison is shown with
data from the FCOM over a range of altitudes and gross weights. A number of points
are made to clarify these charts:
r All weights are in thousands of pounds (for example, 50 = 50,000 lb = 22,680
kg) to comply with the way performance is shown in the FCOM.
r The SAR is given in both n-miles/lb (left scale) and n-miles/kg (right scale).
r The low end of each curve refers to the low-speed buffet Mach number (estim-
ated); the high end of the curves denotes the high-speed buffet Mach number
(also estimated).
The curves from the FCOM have been digitally interpolated up to the point
where they are clearly defined. The comparison is satisfactory, particularly in con-
sideration of the fact that the curves from the FCOM do not specify the position of
the CG (in all of the calculations it was assumed a CG at 25% MAC). The sharp
decrease in SAR after a relatively shallow and flat curve is due to the transonic
effects arising from the fuselage fore-body at incidence and the transonic effects on
the wing–body junction; neither case is modelled in the present case. Note, how-
ever, that lack of accuracy in this area is of small consequence. More important is
the prediction of maximum SAR and the corresponding flight Mach number. Both
quantities are overall well predicted, except in some cases when the aircraft is relat-
ively light (50,000 lb). The Mach number corresponding to maximum SAR (MMR)
is sometimes well predicted; at other times it can be as much as M = 0.1 off the
reference value.
This analysis can be repeated for the OEI case. It is possible to show how with
only one engine operating, the aircraft would have to fly at a much slower air speed
(M = 0.5 to 0.6) in order to maximise the SAR.
0.2 Flight Manual 0.44 0.2 Flight Manual 0.44
0.18 0.4 0.18 50 0.4
SAR, n-mile/kg
SAR, n-mile/kg
SAR, n-mile/lb
SAR, n-mile/lb
50 0.36 60 0.36
0.16 0.16
60
0.32 70 0.32
0.14 70
0.14
80 80
0.28 0.28
0.12 0.12
0.4 0.5 0.6 0.7 0.8 0.9 0.4 0.5 0.6

0.6 0.7
0.7 0.8
0.8 0.9
0.9
True Mach number True Mach number
(a) 30,000 feet (b) 33,000 feet
50
50 0.4 0.4
0.18 0.18
SAR, n-mile/kg
SAR, n-mile/kg
SAR, n-mile/lb
SAR, n-mile/lb
60
60
0.36 0.36
0.16 0.16
70
70
0.32 80 0.32
0.14 80 0.14
0.28 0.28
0.12 0.12
0.4 0.5 0.
0.6
6 0.7
0.7 0.
0.8
8 0
0.9
.9 0.4 0.5 0.6 0.7 0.8 0.9
(c) 35,000 feet (d) 37,000 feet
50
0.2 0.44 0.2 0.44
50
SAR, n-mile/kg
SAR, n-mile/kg
SAR, n-mile/lb
SAR, n-mile/lb
60
60
0.18 0.4 0.18 0.4
70
70
0.36 0.36
0.16 0.16
80
80
0.32 0.32
0.14 0.14
0.28 0.28
0.12 0.12
0.4 0.5 0.6 0.7 0.8 0.9 0.4 0.5 0.6 0.7 0.8 0.9
(e) 39,000 feet (f) 41,000 feet

Figure 12.4. Specific air range of Gulfstream G550 and comparison with the data in the flight
manual. Standard day; no wind.
SAR, n-miles/lb SAR, n-miles/lb

0.1 0.15 0.2 0.25 0.1 0.15 0.2 0.25
16 16
60 50 60 50
70 50 70 50
80 80
14 14
45 45
Altitude, 10 ft
Altitude, 10 ft
Altitude, km
Altitude, km
3
3
40 40
12 12
35 35
10 10
30 30
8 8
25 25
0.2 0.3 0.4 0.5 0.6 0.2 0.3 0.4 0.5 0.6
SAR, n-miles/kg SAR, n-miles/kg
(a) M = 0.80 (b) M = 0.84
Figure 12.5. Calculated altitude performance of Gulfstream G550, standard day, no wind;
weights in thousand of pounds (for example, 50 = 50,000 lb = 22,680 kg).
Figure 12.5 shows the altitude performance of the G550 at constant Mach num-
bers. The charts show the SAR versus the flight altitude at constant weight and
constant Mach number. The results indicate that as the aircraft becomes lighter, the
optimum altitude increases. In this particular instance, for a given all-up weight, the
optimum SAR decreases as the Mach number is increased. The SAR growth up to
the tropopause is essentially linear.
Finally, we show the figure of merit as a function of Mach number at two selected
flight altitudes, Figure 12.6. The weight is a parameter of the analysis. Specifically,
consider the case of the FL-350. The F M increases rapidly up to M ∼ 0.82, and then
it falls off. The Mach number corresponding to maximum F M varies greatly, from
about 0.65 to 0.82.
14 14
Figure of merit, M (L/D)
12 12
Figure of Merit
50 klb
60
50 klb
10 10
60 70
70
80
8 80 8
0.4 0.5 0.6 0.7 0.8 0.9 0.4 0.5 0.6 0.7 0.8 0.9
(a) 30,000 feet (b) 35,000 feet
Figure 12.6. Figure of merit of the Gulfstream G550; standard day (calculated).
0.55
FL-230
MRM LRM
0.5
SAR, n-miles/kg
14
16
0.45
18
20
0.4
0.35
175 200 225 250 275
KTAS
Figure 12.7. Specific air range of the ATR72-500 with PW127M turboprop engines and F568-
1 propellers. The dash-dot vertical line shows the cruise speed inferred from the FCOM;
standard day, no wind.
12.3.2 Case Study: ATR72-500

This turboprop airplane is described throughout this book: see § 6.2.4 for the pro-
pellers, § 5.4.1 for the engine model, § 14.1 for some aerodynamics. We now discuss
the calculation of the point cruise performance of the airplane. Figure 12.7 shows the
calculated SAR at flight level FL-230 (23,000 feet; ∼7,000 m). The graph displays the
estimated MRM and LRM with an AUW variable from 14 to 20 tons. The vertical
dash-dot line corresponds to the typical cruise air speed and has been inferred from
the FCOM. The simulation of the LRM is fairly close to the reference data for a
realistic range of weights.
12.3.3 Effects of Atmospheric Winds on SAR

Atmospheric winds can have strong effects on the fuel economy. For this reason, with
accurate meteorological data fully integrated into the flight management system, it
is possible to tackle adverse wind conditions. Houghton 13 showed that there are
benefits in intercepting and using major jetstream winds and the global weather
system to adjust the route and flight program. It was proved that fuel may be saved
regardless of the wind being a tail wind or head wind; savings of about 1% fuel have
been calculated.
An example of wind effects on the cruise performance of the G550 is shown in
Figure 12.8 for two different weights and a fixed altitude at FL-330. This chart shows
the specific range, or ground specific range, corrected for wind effects. The general
result is that tail winds increase the fuel mileage and decrease the optimal cruise
Mach number. The effect of a head wind is the opposite.
12.4 The Range Equation 339
0.4
0.18
Tailwind
Specific Range, n-miles/kg
Specific range, n-miles/lb

Headwind
0.16
0.35
60 klb
0.14
0.3
80 klb
0.12
0.4 0.5 0.6 0.7 0.8 0.9
Mach number
Figure 12.8. Wind effects on the specific range of Gulfstream G550, standard day, altitude =
33,000 feet; head wind = −10 kt; tail wind = +10 kt; weights as indicated.
The wind speed changes with altitude; therefore it is possible that switching
altitude is more economical. For example, it is possible that at a lower altitude
there is a favourable wind, which would increase the specific range. Such a decision
cannot be left to the pilot’s judgement; the FMS intervenes to optimise the flight
conditions.
Modern airplanes have wind-altitude trade charts that allow a determination
of the optimum cruise altitude. One such example is shown in Figure 12.9 for the
Airbus A320. For example, assume an AUW = 68,000 kg is flying at FL-350 with a
10 kt head wind. For this aircraft, a typical altitude switch is 4,000 feet, or 40 flight
levels. Thus, the airplane can switch from FL-350 down to FL-310 if the wind speed
difference is about 23 kt. This is equivalent to a tail wind of at least 13 kt (13 + 10 =
23 kt).
12.4 The Range Equation

The previous analysis has dealt with point performance, that is, main flight para-
meters that depend locally on the cruise vector state S = {h, M, W, Uw , dT }. The
next problem is to integrate some of these local parameters (specifically, the SAR)
in order to determine the cruise performance of the aircraft. A number of options
are possible, including the following:
r aircraft range for a given fuel mass
r mass required to fly a specified range
The range equation (also known as the Breguet equation) is the main tool
for the calculation of the cruise range of a generic aircraft, although it does not
provide directly any optimal cruise conditions and does not discriminate between
390
370 No
wi 11
nd
A
350
Altitude, km
Flight Level
330 10
10
kt
B
310 20
kt
30
kt 9
290
270
50 55 60 65 70
Gross weight, ton
Figure 12.9. Wind-altitude trade for the Airbus A320 at M = 0.78 (adapted from the FCOM).
flight programs. A number of approximations are used in practice to get first-order

estimates. The aircraft range is defined by the integral
t e e
dm f dm
X= U dt = U =− U (12.14)
o i ṁ f i ṁ
because the loss of weight is only due to fuel burn. The indexes “i” and “e” denote
“initial” and “end” conditions, respectively. For a jet aircraft the fuel flow is replaced
by the equation ṁ f = f j T. Now multiply and divide Equation 12.14 by the aircraft
weight. At cruise conditions the vertical forces are in equilibrium, L = W. Thus, we
have
i i
U U 1 i U L dm
X= dm = dm = . (12.15)
e fj T e fj D g e fj D m
In Equation 12.15 the fuel flow ṁ f must be expressed in units of mass per unit of
time. If other units are on input, the result would be wrong by one order of magnitude
or more. The product g f j is a fuel flow expressed in units of force per unit of time.
If a first-order approximation is sufficient, all the factors within the integral can
be considered constant, and the range equation is reduced to

U L mi U L 1
X ln = ln , (12.16)
g fj D me g fj D 1−ζ
where mi is the mass at the start of the cruise, me is the mass at the end, ζ is the block
fuel ratio, and ζ = m f /mi
mi mi 1
= = . (12.17)
me mi − m f 1−ζ
We conclude that at a fixed altitude the range increases with the flight speed, with
the glide ratio and with the block fuel ratio. More rigorously, the calculation of the
cruise range requires the solution of an integral that contains several parameters of
the aircraft: the initial weight, the block fuel, the flight altitude, the air speed, the
TSFC and the drag, as formulated by

L
X = f h, U, f j , , mi , m f . (12.18)
D
The speed can be replaced by the Mach number. The glide ratio is a function of the
aircraft’s angle of attack and the Mach number, L/D = f (α, M). Finally, the specific
fuel consumption depends on the flight altitude and the Mach number, f j = f (h, M).
The initial weight and the fuel mass can be combined into the non-dimensional block
fuel ratio, Equation 12.17. Therefore, Equation 12.18 becomes
X = f (h, M, α, ζ ) . (12.19)
The angle of attack is not a useful parameter and it is replaced with the CL because
CL = CLα α. Also, instead of the flight altitude, it is useful to have the relative air
density (density altitude), that appears in the aerodynamic terms. Thus, we arrive at
the expression
X = f (σ, M, CL, ζ ) , (12.20)
that is a function of non-dimensional parameters only. Each solution of Equa-

tion 12.20 is called a flight program. Global optima lie in a four-dimensional space.
Flight programs of interest are found with one or two parameters being constant.
12.4.1 Endurance
Endurance is the flight time for a unit of fuel (expressed in volume, mass or weight).
If the speed and flight altitude are constant, then the endurance is the ratio between
the speed and the range
i
1 1 1 m2 1 L dm
E= dm = . (12.21)
e fj D g m1 f j D m
For a propeller-driven aircraft instead, the corresponding equation is
i
1 η 1 i 1 η L dm
E= dm = . (12.22)
e fc DU g e fc U D m
It is useful to find a relationship between the range and the corresponding endur-
ance because the calculation of the former performance will lead to the latter.
Nevertheless, endurance problems are less interesting than range problems, with a
few exceptions, such as loiter, surveillance operations, flight within prescribed area
and so on. Some of these problems are analysed by Sachs 14 .
12.5 Subsonic Cruise of Jet Aircraft

We consider three flight programs: 1) cruise at constant altitude and constant Mach
number; 2) cruise at constant Mach number and constant lift coefficient; and 3)
cruise at constant altitude and constant lift coefficient. All of these cases require a
starting value of altitude, mass and speed (h, m, U), for which an optimal solution
can be found according to the methods presented in § 12.3. Another case is constant
altitude and constant throttle setting. The decreasing aircraft’s weight will result in
an increasing Mach number. The ATC rules do not favour variable flight speed;
therefore this flight condition is less interesting than the previous ones. The mass of
the aircraft is variable from an initial value mi to a final value me .
12.5.1 Cruise at Constant Altitude and Mach Number

During the cruise, the aerodynamic drag decreases because of the decreasing weight.
As the drag decreases, also the engine thrust must decrease and the throttle has to
be stepped down. This is done automatically with the modern FMS. From Equa-
tion 12.15 the range becomes

a M i L dm
X= (12.23)
g fj e D m
because the speed of sound depends only on the altitude, and the thrust-specific
fuel consumption depends on both altitude and Mach number. The maximum cruise
range is found from operating the aircraft at maximum glide ratio. If we use the
parabolic drag equation, the glide ratio becomes a function of the aircraft mass:
L c1 m
= , (12.24)
D CDo + kc12 m2
where
2g
c1 = (12.25)
ρ Aa 2 M2
is a coefficient depending on altitude and Mach number. The angle of attack of the
aircraft does not appear explicitly in Equation 12.24. From Equation 12.23 we find

aM i c1
X= dm. (12.26)
g f j e CDo + kc12 m2
The solution of the integral is from the tables of indefinite integrals 15 :

c1 1 k
dm = tan−1 c1 m . (12.27)
CDo + kc12 m2 kCDo CDo
The final expression for the cruise range is

aM 1 k k
X= tan−1 c1 mi − tan−1 c1 me . (12.28)
g f j kCDo CDo CDo
This equation can be reduced further by using some trigonometric equivalences (see
§ 12.5.5). From the known values of the mass at the start and end of the cruise, it
is possible to find the range from Equation 12.28. In particular, we find that at a
given altitude the range increases with the flight Mach number. Finally, note that
Equation 12.28 is a direct solution of the cruise problem, one that provides the cruise
range for a given fuel load. The inverse problem consists in the determination of the
cruise fuel corresponding to a fixed range X; this is a more involved problem from
the point of view of the mathematical manipulation.
12.5.2 Cruise at Constant Altitude and Lift Coefficient

An approach to the study of this flight program is to consider the balance of forces
in the vertical direction, L = W. With the expression of the CL we find that the air
speed and the Mach number are proportional to W1/2 :
1/2
2W
M= . (12.29)
aρ ACL
Therefore, the Mach number must be constantly reduced as the engines burn fuel.
With the substitution of Equation 12.29 the range equation is written as
1/2 i
a CL 2g
X= m−1/2 dm, (12.30)
g fj CD aρ ACL e
where we have assumed that the changes in fuel consumption due to changes in Mach
number can be neglected. Constant CL implies constant CD and constant CL/CD. By
further manipulation, the cruise range becomes
1/2
2 CL 2a 1/2 √ √
X= [ mi − me ] . (12.31)
f j CD gρ A
This flight program has a number of drawbacks, namely: 1) there is a loss in jet
engine efficiency with a reduced Mach number; 2) the cruise time is increased; 3) a
continuous variation of throttle setting is required; and 4) the cruise range is shorter
than other flight programs.
12.5.3 Cruise at Constant Mach and Lift Coefficient

The cruise of a jet aircraft with a constraint on Mach number and lift coefficient is a
cruise-climb flight. With these constraints the CD, the glide ratio CL/CD and the F M
are constant. The range equation becomes
i
1 L a dm
X= M . (12.32)
g D e fj m
The conditions for a constant F M were found in § 12.2.2. We proved that a climb is
necessary to compensate for the changes in mass. For flight in the stratosphere, also
the speed of sound is a constant. Then the cruise range can be written as

a L 1
X M ln . (12.33)
g fj D 1−ζ
The TSFC is not a constant but depends on both temperature and Mach number,
as explained in Chapter 5. Above 11,000 m the temperature is constant; therefore
cruise at a constant Mach number in the lower stratosphere is an optimal condition.
Flight distance, n-miles

0 1,000 2,000 3,000 4,000
1 12
11.5
Normalised mass 0.9
11
Altitude, km
0.8 10.5
10
0.7
9.5
Aircraft mass
Cruise altitude
0.6 9
0 1,500 3,000 4,500 6,000 7,500
Flight distance, km
Figure 12.10. Cruise-climb profile.
The global optimum range is found from the maximum of

1 L
f (h, M, CL) = M . (12.34)
fj D
Figure 12.10 shows how the flight altitude has to be increased as the aircraft flies.
When a climb is required, the aircraft moves from one flight level to another in
relatively short time. The increase in altitude shown in Figure 12.10 corresponds to
several step-ups. The fuel required for these short segments can be calculated from
energy considerations.
A further consideration is that for maximum range, L/D has to be optimal at
the start of the cruise. This maximum is a function of the aircraft mass; for a fixed
mass (as calculated from the mission analysis), it depends on the speed and altitude.
12.5.4 Comparison among Cruise Programs

We are now in a position to compare the various flight programs, assuming that the
aircraft has a fixed AUW, a fixed fuel mass and the same initial conditions – namely
the flight altitude, the Mach number and the lift coefficient. We study the best range
at the conditions summarised in Table 12.1.
Consider one aircraft model with an m = 145,000 kg and a fuel fraction ζ = 0.138.
The starting point is h = 11,000 m (∼36,000 feet), M = 0.80 for all of the cruise
profiles. The initial conditions require that the initial cruise lift by CL = 0.539 and the
glide ratio L/D 17.98, and F M 14.38. Assume an average TSFC 1.162 · 10−5
kg/Ns. The results are summarised in the right-hand column of Table 12.1. The
Table 12.1. Summary of subsonic cruise conditions, jet aircraft; starting

conditions: h = 11,000 m (∼36,000 feet), M = 0.80
Flight Program Constraints Range Equation Range (km)
A h, m Equation 12.28 5,314

B h, CL Equation 12.31 5,328
Cruise/Climb m, CL Equation 12.33 5,528
cruise-climb program achieves a range about 4% higher than the cruise at constant
altitude and Mach number. Therefore, the cruise-climb is the best flight program
among the cases considered.
12.5.5 Fuel Burn for Given Range

We now calculate the fuel burn of a jet aircraft for a given range (or segment) and
a constant altitude and Mach number. We start from the conclusions of § 12.5.1 and
write the range equation for the generic flight segment i- j.

aM 1 −1 k −1 k
Xi j = tan m j c1 − tan mi c1 . (12.35)
g f j kCDo CDo CDo
If the altitude and the Mach number are specified, to simplify Equation 12.35, intro-
duce the factors

aM 1 k
c3 = , c4 = c1 , (12.36)
g f j kCDo CDo
so that

Xi j = c3 tan−1 (c4 m j ) − tan−1 (c4 mi ) . (12.37)
Equation 12.37 can be further reduced to

c4 mi − c4 m j
Xi j = c3 tan−1 + π . (12.38)
1 + c42 mi m j
The term π only appears if c42 mi m j > 1. Now solve Equation 12.38 to find the value
of the fuel mass during the flight segment:

Xi j c4 mi − c4 m j
tan −π = . (12.39)
c3 1 + c42 mi m j
The unknowns are the aircraft mass at the end of the flight segment and the fuel
mass. We assume that
mi = m j − m f (12.40)
and solve Equation 12.39 for the unknown fuel burn. Use the known coefficient

Xi j
c5 = tan −π (12.41)
c3
to simplify the equation. Therefore:
c4 mi − c4 m j
c5 = , (12.42)
1 + c42 mi m j

c4 mi − c4 m j = c5 1 + c42 mi m j . (12.43)
Finally, the mass of fuel burned during the flight segment from i to j is
c5 + c42 c5 mi2
mf = . (12.44)
c4 + c42 c5 mi
The problem is not closed because the initial mass mi may not be known exactly.
Equation 12.44 represents the fuel burn over the segment i- j for a given altitude,
Mach number and mass at the start of the segment.
12.6 Cruise Range of Propeller Aircraft

We now get down to more mundane speeds and altitudes, to deal with propeller-
driven aircraft. These can be powered by gas turbine or reciprocating engines. The
range equation (Equation 12.14) is valid in the general case. For a propeller-driven
aircraft we need to replace the fuel-flow equation: ṁ f = fc P. The range becomes

1 U
X= dm. (12.45)
fc P
We replace the power equation, η P = TU, in Equation 12.45 to obtain

1 η L dm
X= . (12.46)
g fc D m
The latter equation contains the equilibrium conditions L = W, D = T. There is
a new variable: the propeller’s efficiency; this, as we have seen, depends on the
advance ratio J (see Chapter 6). If the engine power depends on the weight, then
the propulsive efficiency is also dependent on the weight by way of the shaft power.
The types of cruise conditions for the propeller-driven aircraft are similar to the
jet-engine aircraft. In particular, we note that the range is a function of this type:
X = f (σ, η, U, fc , ζ ) . (12.47)
We discuss the solution of the cruise flight at constant speed and altitude. The
remaining cruise programs can be found with an analysis similar to the subsonic jet
aircraft.
CRUISE AT CONSTANT ALTITUDE AND SPEED. The range equation for this flight condi-
tion becomes

η(J, θ ) i
L dm
X= , (12.48)
g fc e D m
12.7 Cruise Altitude Selection 347
where we have assumed that the propulsive efficiency η is a function of the propeller’s
advance ratio J and pitch setting θ . The SFC is only a function of h and U; therefore
it is taken out of the integral. If we write the glide ratio as a function of the aircraft
mass, as in § 12.5.1, then

η(J, θ ) i
c1 dm
X= . (12.49)
g fc e CDo + kc12 m2
The solution of the integral is

i
η(J, θ ) k
X= tan c1 m . (12.50)
g fc CDo
e
If the block fuel and the aerodynamics are fixed, then the range is maximised by the
function η(J, θ )/ fc . This factor is a combination of engine and propeller performance.
The solution of Equation 12.50 inevitably needs propeller charts.
12.7 Cruise Altitude Selection

From an optimisation point of view, the flight levels present a constraint. Cruise
optimisation is done according to a three-stage procedure. In the first stage we select
the best ICA and the best Mach number (§ 12.3); in the second stage we estimate
the final cruise altitude from the range equation (§ 12.4); and in the final stage we
select a sequence of constant-altitude and climb flight segments to reach the end
point.
Let us assume, without loss of generality, that we want to operate the aircraft at
the long-range Mach number. The range equation can be used to estimate the fuel
required and the altitude at the end of the cruise. The solution is

me
ρe = ρi . (12.51)
mi
The altitude can be found directly from the final air density ρe by using the atmo-
sphere model. Let us assume that the total climb required is h. However, this
altitude gain will have to be adjusted to the closest flight level. An item of concern is
that occasionally the aircraft is unable to reach the optimal condition, due to a num-
ber of factors, including available thrust, increased gross weight, ATC constraints
and adverse atmospheric conditions. The final stage in the cruise optimisation con-
tains a fairly large number of parameters. If n is the number of constant-altitude
steps, the number of climb steps is n − 1:
h
n−1 = , (12.52)
FLS
where “FLS” is the flight level separation. The final segment is always a cruise. For
example, if the adjusted climb is h = 4,000 feet (∼1,200 m) and FLS = 100 (or
1,000 feet), then n = 5 (four climb segments; five cruise segments). If FLS = 200,
then n = 3 (two climb segments; three cruise segments).
380 380
C C
FCA
370 370
360 360
Flight level
Flight level
B High
350 350
B High b
m
340 340 cli
ise
C ru
A Low
330 A 330
Low
320 320
0 1,000 2,000 3,000 4,000 0 1,000 2,000 3,000 4,000
Cruise range, n-miles Cruise range, n-miles
(a) FLS = 10 (b) FLS = 20
Figure 12.11. Effect of flight-level separation on the cruise trajectory of a Boeing 777-300 with
GE-92 turbofan engines. Adapted from Ref. 16 .
The distances flown in the cruise mode are called Xi and the distances covered
by the climb steps are called Xci . The constraints are

n
n−1
X= Xi + Xci , (12.53)
i i
h = (n − 1)dh. (12.54)
The total number of free parameters is 2n − 1, that is, n constant-altitude stage

lengths and n − 1 climb rates between flight levels. Mathematically, such a problem is
quite complex. Practical solutions deal with specified climb rates of equal magnitude
and cruise stages of equal magnitude, except the last one; this is calculated from the
difference between the required range and the distance covered in all of the previous
stages (cruise and climb). Thus, the number of free parameters is reduced to three
(two cruise stage lengths and one climb rate). An alternative cruise procedure relies
on the change of flight level based on the AUW of the aircraft. There are two possible
starting points of the cruise, as shown in Figure 12.11: a low profile starting at point
A and a high profile starting at point B. Once the cruise has started, there are three
possibilities:
1. Low profile: The step climb starts at the weight corresponding to the intersection
between the low flight level A and the theoretical optimal altitude, Figure 12.11a.
The flight profile will often be below the optimal cruise altitude; the aircraft will
have better manoeuvre margin, and if the FL is below 330, the TAS will be
higher. This procedure requires a limited number of intermediate calculations
with the range equation, to estimate the AUW at the switch point.
2. Middle profile: The step climb starts when the SAR at the higher flight level
is better than the SAR at the current flight level (Figure 12.11b). This proced-
ure requires the calculation of the SAR at two flight levels at each time step;
therefore, it is computationally more demanding.
12.8 Cruise Performance Deterioration 349
Table 12.2. SAR penalty due to non-optimal cruise altitude for some
Airbus airplanes
Airplane A300B4-600 A310-324 A320-211 A330-203 A340-642
+FL-200 2.0% 0.9% – 1.8% 1.6%

−FL-200 1.9% 1.4% 1.1% 1.3% 0.6%
3. High profile: The step climb starts at the weight (or distance) corresponding
to the flight level above the theoretical optimal altitude point B. The aircraft
moves to the next flight level when the flight level reaches the theoretical optimal
altitude, Figure 12.11a.
Table 12.2 is a summary of cruise-performance penalties arising from cruise at

the wrong flight level* . The penalty can be as much as 2%.
The minimum climb rate for transport aircraft is ∼300 feet/min (∼1.5 m/s). The
effect of the climb rate on the cruise fuel of the reference transport airplane is shown
in Figure 12.12 for a 2,000 km (1,079 n-miles) required range. The result indicates that
some fuel savings can be achieved by switching the flight level as quickly as possible.
The reference point for the calculation of the fuel savings is a climb with vc =
1.9 m/s (∼370 feet/min).
12.8 Cruise Performance Deterioration

Cruise performance tends to degrade over time. This inconvenient reality can be
the result of engine performance deterioration, airframe drag deterioration or a
combination of both. There are various (expensive) experiments that can be carried
out to verify how the deterioration affects cruise performance. For example, take
as a reference a new airplane with new engines; test flights can then be carried out
with old airframes and new engines, and new airframes with old engines. Four sets
of SAR data (S = SAR) can be generated to isolate the problem:
r new airframe with new engines, Snn
r old airframe with old engines, Soo
r new airframe with old engines, Sno
r old airframe with new engines, Son
The tests will have to be carried out with the same weight, with the same position
of the CG, over the same route and with sufficiently similar atmospheric conditions.
A further complication in the performance study arises from the fact that new and
old airplanes are not necessarily identical. There are always minor differences in con-
figurations, which must be carefully analysed before drawing any conclusions. With
these precautions, if there is no change in SAR resulting from a change of engines,
the loss in performance must be attributed to the deterioration in aerodynamic drag.
The relative loss in performance can be expressed in two ways:
Snn − Son Sno − Soo
S1 = , S2 = [change airframe]. (12.55)
Snn Sno
* Elaborated from Airbus: Getting to Grips with Fuel Economy, Issue 3, July 2004, Blagnac, France.
Climb rate, feet/min

500 1,000 1,500 2,000
0 0
-0.2
Cruise fuel saving, % -10
d mf / d vc, kg/ m/ s
-0.4
-20
-0.6
-30
-0.8
Cruise fuel saving

Fuel sensitivity
-1 -40
2 4 6 8 10
Climb rate, m/s
Figure 12.12. Effect of climb rate on cruise fuel for a 2,000-km (1,079 n-miles) required range;
standard day; no wind.
The first and second equations denote the effects of the engine on the new and old
airframes, respectively. Several causes could contribute to airframe drag deteriora-
tion, including panel mis-alignment, broken seals, leakages and surface roughness.
The other extreme case is when the performance deterioration is fully attributed to
the engines. This happens when the same airframe (regardless whether new or old)
is equipped with new and old engines. The relative loss in performance is:
Snn − Sno Son − Soo

S3 = , S4 = [change engines]. (12.56)
Snn Son
The deterioration in engine performance can be due to a number of causes, includ-

ing mis-alignment of fan and compressor blades, blade erosion, loss in pressure
ratio, over-heating, increased air bleed, loss of combustion efficiency, leakages and
so on.
The loss in performance can be a fraction of a percent or as high as a few
percent. Even in the most optimistic scenario, the costs arising from the operation
of a commercial airplane that is not kept as new can be phenomenally high.
12.9 Cost Index and Economic Mach Number

The economic Mach number, Mecon , is a performance parameter that takes into
account the cost of fuel and the cost of time. In fact, a higher Mach number
reduces the cruise time at the expense of increased fuel consumption; vice versa
12.9 Cost Index and Economic Mach Number 351
for a lower Mach number. The minimum direct operating costs (DOC) are some-
where in between (see § 15.8). In quantitative terms, we have
DOC = co + c1 m f + c2 t, (12.57)
where co is a fixed-cost term, c1 is a factor depending on the cost of fuel, and c2 is

a factor depending on the cruise time t. The DOC can be plotted against the cruise
Mach number.
Minimum fuel costs are obtained at the Mach number corresponding to max-
imum SAR. The time-related costs decrease with the increasing Mach number. These
considerations lead us to conclude that the Mecon is slightly higher than the MMR.
To characterise this result, a cost index (CI ) is used:
Cost of time c2 t c2 1
CI = = = . (12.58)
Cost of fuel c1 m f c1 ṁ f
The calculation of the cost index is dependent on the aircraft operator. Modern
flight-management systems scale the CI between 0 and 99 (or between 0 and 999).
Parametric studies of the economic Mach number can be done in terms of normalised
values of CI .
The physical dimensions of the CI are units of fuel per unit of time. CI = 0
means that c1 is large and c2 is small. If CI is maximum, c1 is small and c2 is large;
this occurs when the aircraft operates in minimum-time mode and maximum Mach
number (MMO). If the CI is high, the economic Mach number is insensitive to
weight. Conversely, if the CI is low, the economic Mach number is sensitive to both
weight and flight levels. Tax on aviation fuel, carbon trading schemes and other
environment-related costs will inevitably increase the cost of fuel compared to the
cost of time; therefore the value of this index is bound to decrease in the future. To
minimise the total trip cost we need to minimise the sum
c1 m f + c2 t.
Now consider the cost function Fc per unit distance:

c1 m f + c2 t c1 c2
Fc = = + , (12.59)
x SAR U
where U = a M + Uw is the true air speed and Uw is the wind speed. When the cost
associated to the fuel is equal to the cost associated to the flight time, then
c1 c2
= , (12.60)
SAR U
or
1 c2 1 CI
= ∼ . (12.61)
SAR c1 U U
It is not essential to know exactly the value of the factors c1 and c2 . It will be sufficient
to know that CI ∝ c2 /c1 . From this equivalence, we find the cost index
U
CIeq ∼ . (12.62)
SAR
12.0
ICW = 146 ton, X = 1,000 nm
380
A B
11.0 360
Time Priority
Altitude, km
Flight level
340
Fuel Priority
10.0
320
300
9.0
280
0.76 0.78 0.8 0.82
Economic Mach
Figure 12.13. Analysis of economic Mach number.
A value twice this CIeq corresponds to a time priority. A further scaling can be done
so that CImax = 99, as in a modern flight-management system. The minimum trip
cost is achieved by properly adjusting the Mach number, to take into account the
atmospheric conditions, aircraft performance, the cost of fuel and the marginal costs
because they are incurred by an additional flight time. An example of calculation
of the economic Mach number for a transport airplane is shown in Figure 12.13
against the cruise altitude. This graph shows the difference in cruise Mach at fuel
priority and time priority. This difference decreases as the aircraft flies at higher
altitudes.
The application of the cost index can be expanded to other flight conditions,
such as climb to initial cruise altitude (see Chapter 10) and en-route descent from
cruise altitude (see Chapter 11).
12.10 Centre of Gravity Position

In this section we study the effects on the cruise performance due to position of the
centre of gravity. We must always ensure that the loading of the aircraft gives a CG
position within the certified range (i.e., within the forward and aft limits). A forward
CG requires a nose-up pitching moment to establish a longitudinal trim (§ 7.1). This
trim can be achieved through a lower loading of the tail plane and/or a higher loading
of the wing. The result of the trim is to increase the induced drag (§ 7.1.1).
Most modern aircraft have an automatic flight-management system for the CG
that is based on pumping fuel in or out of a trim tank. The flight system optimises the
position of the CG, depending on the actual weight of the aircraft, so as to minimise
the total drag and hence the cruise fuel consumption. The position of the CG can be
calculated from the zero-fuel weight, the GTOW and the status of each fuel tank.
The software can be quite sophisticated, but it provides the position of the CG with
accuracy and celerity. For example, the Airbus A310 has CG limits between 20%
and 45%, with a mid-position at 27%. Flight data for this aircraft show that flight
with a CG in forward position can add up to 1.8% to the cruise fuel, compared with
the mid-position; conversely, a flight with the CG in aft position can save about 1.8%
of the cruise fuel, compared with the mid-position, irrespective of the flight level.
However, this performance is not general and must be studied for each aircraft.
Both Airbus and Boeing provide CG data and their effects on cruise on all of their
aircraft. For example, for the Airbus A318/A319/A320, the manufacturer requires
the following sequence of fuel use:
r additional centre tanks (ACT2, ACT1)

r centre tank, until empty
r inner wing tanks, until 750 kg are left in tanks
r outer wing tanks: fuel transferred into inner tanks
r inner wing tanks
In this sequence, only one fuel transfer is required: from the outer wing tanks
to the inner wing tanks, when the inner tanks’ level is below the threshold. The fuel
transfer is done by gravity.
CASE STUDY: CENTRE OF GRAVITY EFFECTS. The effect of the CG position on the SAR
has been calculated for a model Boeing B777-300 16 ; this is shown in Figure 12.14.
This graph displays two sets of curves, corresponding to a CG shift equal to ± 6%
MAC, respectively. For each position of the CG, the change in SAR depends on
both altitude and gross weight. For a fixed weight, moving the CG aft causes a
loss in SAR that increases with the flight altitude; moving the CG forward causes
an increase in SAR that improves with the flight altitude. Consider now the solid
lines (largest weight). The airplane would operate at FL-350 (point A or B); as
it burns fuel, it must climb to maintain the same SAR (or to avoid a larger SAR
penalty). After burning about 20 tons of fuel, the altitude must shift to FL-370
(point C or D).
12.11 Supersonic Cruise

We now deal with a more exotic flight performance. We study the optimal conditions
for cruise range at supersonic speeds for a given AUW and a given initial block fuel.
A number of cases are possible because the cruise conditions depend on the altitude,
Mach number and angle of attack. The range equation is given by Equation 12.14.
The general problem of supersonic cruise can be quite complicated. Windhorst
et al. 17 developed guidance techniques for minimum flight time, minimum fuel con-
sumption and minimum DOC for a fixed-range cruise. For example, a minimum fuel
trajectory consists of an initial minimum-fuel climb, a cruise-climb and a maximum
L/D descent. The minimum DOC trajectory is nearly the same as a minimum fuel
trajectory.
13
AFT FWD
Δ xCG = +0.06 MAC Δ xCG = -0.06 MAC 410
12 E F
C D 370
Altitude, km
Flight Level
11
A B
10 330
210
230
9
290
250 ton
8
-4 -2 0 2 4
Δ SAR, %
Figure 12.14. Effect of gross weight and the CG position on the optimal cruise altitude.
12.11.1 Cruise at Constant Altitude and Mach Number

For cruise in the stratosphere, the speed of sound is constant. In addition, if the
Mach number is also constant, the range equation becomes

a M i CL dm aM i CLα (α − αo) dm
X= = . (12.63)
g f j e CD m g f j e CDo + ηCLα (α − αo) m
2
The angle of attack at cruise condition is specified by L = W:
2W 1
α = αo + . (12.64)
ρ Aa CLα M2
2
When we replace this condition in Equation 12.63, we find
L co/M2 c1 (σ, M)m

= = , (12.65)
D CDo + ηco2 /M4 CDo + c2 (σ, M)m2
with
2
2mg 2g 2g η
co(σ, m) = , c1 (σ, M) = , c2 (σ, M) =.
ρ Aa 2 ρ Aa 2 M2 ρ Aa 2 M2 CLα
(12.66)
The glide ratio can be optimised with respect to the Mach number or the mass.
Equation 12.65 shows the dependence of the glide ratio from the Mach number and
the aircraft’s mass. When we use the coefficients defined by Equation 12.66, the
range becomes

aM i c1 (σ, M)
X= dm. (12.67)
g f j e CDo + c2 (σ, M)m2
The integral is rather involved, but it is similar to an expression seen before (Equa-
tion 12.28):
i " i
dm 1 −1 c2
= tan m , (12.68)
e C Do + c2 m
2 CDo e
CDo c2
Therefore, the solution is
" "
a M CLα c2 c2
X= tan−1 mi − tan−1 me . (12.69)
g f j CDo η CDo CDo
Equation 12.69 is similar to Equation 12.28, and is valid for the same flight program
at subsonic speeds. A specified range can be achieved by subsonic cruise at a low
altitude or a supersonic cruise in the lower stratosphere.
12.11.2 Cruise at Constant Mach Number and Lift Coefficient

Consider a high-speed cruise at a fixed Mach number and CL. The corresponding
angle of attack α is constant and is set at the start of the cruise, according to Equa-
tion 12.7. Thus, the figure of merit M(CL/CD) is also a constant. To maintain a
constant CL and α with a decreasing mass, the aircraft must climb, according to the
same principle illustrated in § 12.5.3. In fact, from Equation 12.7 α is constant if
m/σ = constant. Therefore, this flight program is a cruise-climb. With these consid-
erations, the range equation becomes
i
a CL 1 dm
X= M . (12.70)
g CD e f j m
The changes in f j must be considered; therefore we need to refer to an accurate aero-
thermodynamic model for the engine cycle, which provides the correct behaviour of
the fuel flow and hence the TSFC. For a given block fuel ratio, the range given by
Equation 12.70 is maximised if
M CL
R(σ, M) = = max. (12.71)
f j CD
If the TSFC were a constant, then the supersonic cruise would be optimal with a maxi-
mum glide ratio. The function R(σ, M) (the range factor) is shown in Figure 12.15:
there are two local maxima; one maximum can be at subsonic or supersonic speed,
depending on the altitude. The other maximum is at about M 1.6 and does not
vary much with the altitude. The TSFC is shown in the right scale. The two lines
denote the lowest and highest flight altitudes (for clarity, the TSFC at intermediate
altitudes is not plotted).
Summary
This chapter has dealt with cruise problems. We have shown that a large portion of
the fuel is burned in this flight segment, with the possible exception of very short-
haul flights. Therefore, numerical optimisation methods have traditionally addressed
cruise flight. In this context, the key performance parameter was the specific air
150 0.18
125
0.15
Range factor, Nh/kg

100
TSFC, kg/N h
75 0.12
16
50 14
12 0.09
10
25
8
0 0.06
0.6 1 1.4 1.8
Mach number
Figure 12.15. Range factor of supersonic aircraft model; flight altitudes are in [km].
range (SAR). We demonstrated that the study of the SAR is the basis of all cruise
optimality conditions. From this parameter we derived the maximum-range cruise
Mach number and the long-range cruise Mach number. Commercial operators use
the latter speed as a compromise between faster service and increased cost of fuel.
However, the problem is often more complicated and requires the determination of
the cost index function, which is a balanced approach to the cost of fuel and time.
On modern airplanes, the cost index is determined by the FMS.
Another key concept was the determination of the optimal cruise altitude. We
have shown that current regulations only permit flight at selected levels and that a
flight level switch (with a climb) is required to maintain the SAR as high as possible.
The combination of long-range Mach number and initial cruise altitude is a two-
parameter optimisation problem that depends on the weight. We solved a number
of problems, both for subsonic and supersonic flight. We also investigated the effect
of weight, CG position, winds and aerodynamic deterioration.
Bibliography
[1] Speyer JL. Non-optimality in the steady-state cruise for aircraft. AIAA J.,
14(11):1604–1610, Nov. 1976.
[2] Speyer JL, Dannenmiller D, and Walker D. Periodic optimal cruise of an
atmospheric vehicle. J. Guidance, 8(1):31–39, Jan. 1985.
[3] Gilbert EG and Parsons MG. Periodic control and the optimality of aircraft
cruise. J. Aircraft, 13(10):828–830, Oct. 1976.
[4] Menon PKA. Study of aircraft cruise. J. Guidance, Control and Dynamics,
12(5):631–639, Sept. 1989.
[5] Sachs G and Christodoulou T. Reducing fuel consumption of subsonic aircraft
by optimal cyclic cruise. J. Aircraft, 24(5):616–622, 1987.
[6] Visser HG. Terminal area traffic management. Progress Aerospace Sciences,
28:323–368, 1991.
[7] ESDU. Approximate Methods for Estimation of Cruise Range and Endurance:
Aeroplanes with Turbojet and Turbofan Engines. Data Item 73019. ESDU
International, London, 1982.
[8] ESDU. Introduction to the Estimation of Range and Endurance. Data Item
73018. ESDU International, London, 1980.
[9] ESDU. Estimation of Cruise Range: Propeller-driven Aircraft, volume 5, Per-
formance of Data Item 75018. ESDU, London, 1975.
[10] Bennington MA and Visser KD. Aerial refueling implications for commercial
aviation. J. Aircraft, 42(2):366–375, Mar. 2005.
[11] Smith RK. Seventy-five years of inflight refueling. Technical Report R25-GPO-
070-00746-1, Air Force History and Museum Program, Washington, DC, Nov.
1998. (Available on the U.S. Government online store; document out of print
as of 2005.)
[12] Torenbeek E and H Wittenberg. Generalized maximum specific range per-
formance. J. Aircraft, 20(7):617–622, July 1983.
[13] Houghton RC. Aircraft fuel savings in jet streams by maximising features
of flight mechanics and navigation. J. of Navigation, 51(3):360–367, Sept.
1998.
[14] Sachs G. Optimization of endurance performance. Progress Aerospace Sci-
ences, 29(2):165–191, 1992.
[15] Abramovitz M and Stegun I. Handbook of Mathematical Functions. Dover,
1972.
[17] Windhorst R, Ardema M, and Kinney D. Fixed-range optimal trajectories of
supersonic aircraft by first-order expansions. J. Guidance, Control and Dynam-
ics, 24(4): 700–709, July 2001.

a = speed of sound
A = wing area
b = wing span
ci = constant factors (i = 0, 1, 2, . . .)
CL = drag coefficient
CLα = lift curve slope
CI = cost index
CIeq = equivalence cost index
D = drag
E = endurance
fc = specific fuel consumption (also SFC)
fj = thrust-specific fuel consumption (also TSFC)
Fc = cost function
FM = Figure of Merit
h = flight altitude
J = propeller’s advance ratio
m = mass
ṁ f = fuel flow
M = cruise Mach number
Mecon = economic Mach number
Mdd = divergence Mach number
n = number of constant-altitude cruise stages
p = pressure
P = power; shaft power
R = range factor, Equation 12.71
S = cruise vector state
Sxy = specific air range, with x = n/o, y = n/o, Equation 12.55
SR = specific (ground) range
t = time
T = net thrust
T = air temperature
vc = climb rate
U = air speed
Uw = wind speed
W = weight
X = cruise range; flight distance
Xi j = range of flight segment i- j
Xi = distance flown in cruise mode
XCi = distance flown in climb mode
x, y, z = Cartesian coordinates
Greek Symbols
αo = zero-lift α
η = propulsive efficiency; lift-induced factor
θ = propeller-pitch angle
ζ = fuel-block ratio, m f /mi
ξ = relative change in aircraft mass; ξ = m/mi in § 12.2.1
ρ = air density
[.]c = constant portion
[.]e = end (or final)
[.] f = fuel
[.]i = initial (or start)
[.]i j = generic flight segment, § 12.5.5
[.]o = reference value
[.] p = payload
[.]w = wind
[.]nn = new airframe and new engines
[.]no = new airframe and old engines
[.]on = old airframe and new engines
[.]oo = old airframe and old engines
13 Manoeuvre Performance
Overview
This chapter deals with basic airplane manoeuvres, including flight in the vertical
plane and in the horizontal plane. After an introductory description of these man-
oeuvres (§ 13.1), we start with powered turns (§ 13.2) and unpowered turns (§ 13.3);
the latter turn essentially refers to glider-type vehicles, including soaring flight.
Next, we introduce the manoeuvre envelope of the airplane (§ 13.4) and discuss
pilot-related issues, such as sustainable g-loads. Roll performance (§ 13.5) is one
example of manoeuvre on a straight line by a high-performance aircraft, and pull-up
(§ 13.6) is one example of flight path in the vertical plane. Finally, we discuss aircraft
flight in a downburst (§ 13.7).
KEY CONCEPTS: Powered Turns, Banked Turns, Minimum-Fuel Turn, Turn Rates,
Cornering Speeds, Unpowered Turns, Manoeuvre Envelopes, V-n Diagram, Sus-
tainable g-Loads, Roll Performance, Pull-Up Manoeuvre, Downbursts.
13.1 Introduction
The term manoeuvre refers to any change of the flight path. There are at least
three basic types of manoeuvres: 1) a turn in the horizontal plane, when the aircraft
changes heading and banks on one side (lateral manoeuvre); 2) a turn in a vertical
plane, when the aircraft increases or decreases its altitude (pull-up and pull-out);
and 3) a roll around the aircraft’s longitudinal axis, when the aircraft rotates around
itself whilst following a straight flight path (longitudinal manoeuvre). Many complex
manoeuvres can be reduced to a combination of these.
During a turn on the horizontal plane, the aircraft gradually changes its bank-
ing angle from zero to a maximum and then back to zero. A pull-up occurs with
a variable radius of curvature and speed. Therefore, a turn manoeuvre is always
unsteady. Banking is always associated to some centrifugal accelerations that can
reach relatively high levels. In addition to the cases listed, there are a large number of
manoeuvres performed only by military aircraft, aerobatic aircraft and demonstrat-
ors. A concept of interest in high-performance aircraft is the agility, or the ability of
moving rapidly from one manoeuvre to another 1 .
360
φ Lift
y
b
Figure 13.1. Nomenclature for a banked turn. CG
zb
W
13.2 Powered Turns

A banked level turn (or coordinated turn) is a turn during which all of the forces
on the aircraft are balanced. Therefore, the aircraft flies with a constant air speed
and constant altitude. However, due to the change of heading, there is a centrifugal
acceleration. The nomenclature and the reference system for a banked turn are
shown in Figure 13.1. This is a view of the aircraft forces in the vertical plane {y, z};
φ is the bank angle, that is, the angle between the vertical plane and the aircraft’s
symmetry plane; χ is the radius of curvature of the flight trajectory.
We denote the quantities during a turn with a subscript “t”. In a correctly banked
turn the balance of forces along the horizontal and vertical direction are, respectively,
Lcos φ − W = 0, (13.1)
U2
Lsin φ = m . (13.2)
χ
The right-hand term of Equation 13.2 denotes the force due to the centrifugal accel-
eration of the CG. Combination of the two equations provides the bank angle φ,
1 U2
tan φ = . (13.3)
g χ
The bank angle increases with the speed. When the aircraft operates at the smallest
values of χ , it is said to perform a tight turn. During manoeuvres of this type, the lift
is higher than the aircraft’s weight. We define the normal load factor n as the ratio
L
n= . (13.4)
W
In a straight level flight n = 1. This value, also called 1-g flight, is the neutral value.
Now the bank angle, Equation 13.3, can be written in terms of n, by using Equa-
tion 13.1:
1
n= = sec φ. (13.5)
cos φ
From Equation 13.3, the radius of turn is a function of the load factor
U2 1 1 U2
χ= = √ , (13.6)
g tan φ g n2 − 1
where we have used the trigonometric equivalence sec2 φ = 1 + tan2 φ. In civil avi-
ation, the bank angle is kept within 25 degrees, which leads to a normal load factor
n ∼1.1; this acceleration can be tolerated by most passengers. From the definition of
turn radius χ, Equation 13.6, the centrifugal acceleration is
U2
a= = g n2 − 1. (13.7)
χ
This acceleration can be normalised with g, to provide the g-factor, or the relative
centrifugal acceleration during a banked turn
a
= n2 − 1. (13.8)
g
A normal load factor n = 1 corresponds to zero centrifugal acceleration, which
proves that n = 1 is a flight with no acceleration.
13.2.1 Banked Turn at Constant Thrust

We consider a case in which the turn is performed at constant thrust. The corres-
ponding drag coefficient is
2
2nW
CDt = CDo + kCL2 = CDo + k . (13.9)
ρ AU 2
There is an increase in lift-induced drag proportional to n2 and hence an increase in

overall drag, D = Dt − D > 0. If we write the conservation of total energy of the
aircraft, then
∂h E T−D
= U, (13.10)
∂t W
where h E is the energy height (see Equation 10.45); the right-hand side is the specific
excess power. This equation expresses the fact that an excess power can be used
to increase the speed or the altitude, or both. A change in aircraft drag dD can be
introduced in Equation 13.10, so that

∂h E T dD D
d = dU − U− dU, (13.11)
∂t W W W

dD ∂h U 2 T D ∂h U 2
U = −d + + dU − dU = −d + . (13.12)
W ∂t 2g W W ∂t 2g
If the flight altitude is constant (dh/dt = 0), an increase in drag causes a decrease in
speed, and vice versa:
dD U
U = − dU. (13.13)
W 2g
If, instead, the speed is maintained constant (dU = 0), then there must be a decrease
in altitude

dD ∂h
U = −d = vs < 0, (13.14)
W ∂t
where vs denotes the sinking speed. The magnitude of vs can be found from
Dt − D ρ AU 2 2W
vs = U= (CDt − CD) = k(n2 − 1). (13.15)
W 2W ρ AU 2
For a given speed, the rate of descent increases with n2 . The change in drag produces
a change in the speed of minimum drag, Umd . The CL corresponding to minimum
drag in a banked turn is found from the condition

∂ CLt
= 0. (13.16)
∂CL CDo + kCL2 t
The solution of Equation 13.16 is

"
1 CDo
CL = . (13.17)
n k
The ratio of minimum drag speeds between banked turn and level flight is
Umdt
= n1/2 . (13.18)
Umd
If the air speed before the turn is U > Umd , the drag increase due to the turn will
slow down the aircraft; this will contribute to a decrease in drag to the point when
the forces on the aircraft are balanced, and the aircraft can sustain a turn at constant
altitude, albeit at a lower speed. If, instead, the air speed before the turn is U < Umd ,
a decrease in speed will produce a further increase in drag. If the drag rise cannot
be matched by the engine thrust, the aircraft is forced to descend.
13.2.2 Turn Power and High-Speed Manoeuvre

Assume that the CL during a turn is constant and equal to the value in straight level
flight:
2W 2W
CL = CLt , or =n , (13.19)
ρ AU 2
ρ AUt2
which leads to the equation
√
Ut = U n. (13.20)
Equation 13.20 expresses the equivalence between aircraft speeds at constant lift
coefficient. If the turn is done at constant speed, then the relationship between lift
coefficients becomes
CLt = nCL. (13.21)
Next, consider a turn at constant power, P = Pt ; this condition requires
TU = Tt Ut . (13.22)
Equation 13.22 can be satisfied by increasing the net thrust as the air speed decreases.
The power in steady-state flight is calculated from the equilibrium of the forces in
the flight direction multiplied by the flight speed:
1
P= ρ ACDU 3 . (13.23)
2
Next, eliminate the velocity in Equation 13.23 by using the definition of CL. Equa-
tion 13.23 becomes

2 CD
P= (nW)3/2 . (13.24)
ρ A CL3/2
We conclude that the power required for a steady banked turn grows with the factor
(nW)3/2 ; at constant weight, the power for a banked turn increases with n3/2 . If also
3/2
the aerodynamic factor CD/CL is held constant, the ratio between power required
in a turn and the steady-state power is
Pt
= n3/2 . (13.25)
P
If the aircraft has a parabolic drag equation, Equation 13.23 becomes
1 2 (nW)2
P= ρCDo U 3 + k . (13.26)
2 ρA U
Equation 13.26 can be plotted versus the flight speed at different values of the load
factor n. Matching this power with the available engine power will give the operation
point. The minimum speed that the aircraft can sustain is the stall speed (within a
safety margin Ks ); hence, the lower limit of the curves will be given by the stall curve.
SUPERSONIC FLIGHT. We now consider the more complicated problem of high-

speed turn performed by a supersonic aircraft. Consider a turn at the flight altitude
h = 8,000 m (∼26,245 feet), with an aircraft mass m = 12,000 kg. The problem con-
sists in solving Equation 13.24 over a range of Mach numbers with the appropriate
aerodynamics and engine at selected throttle (for example, full thrust). The results
of the analysis are shown in Figure 13.2, for normal load factors up to n = 3. The
available engine thrust, with and without after-burning, is also shown. The power
required to make a supersonic turn is very large; a supersonic turn cannot be per-
formed at high load factors.
HIGH-SPEED MANOEUVRE. In § 10.5 we presented some concepts for aircraft climb

and acceleration by using the total-energy method. Our considerations were limited
to a 1g-flight. The same method can be applied to aircraft manoeuvring with an
apparent acceleration g. The drag is changed as follows:
CD CDo + n [ηCLα (α − αo)] . (13.27)
In Equation 13.27 the profile-drag coefficient is assumed to be unaffected by the

manoeuvre (which is not quite the case), and the induced drag is proportional to the
80
C
χ = 6.63 km
60
After-burning
Power, MW Thrust 4
χ - 9.95 km
A
40 B 3
χ = 16.0 km
20
n = 1.25
0
0.5 1 1.5 2
Mach number
Figure 13.2. Power for supersonic turn; aircraft mass m = 12,000 kg; α = 2 degrees (all cases);
altitude h = 8,000 m (∼26,245 feet).
load factor n. Therefore, the specific excess power is

T − nD
SEP . (13.28)
W
As usual, the weight is critical and will determine whether the aircraft is capable
of manoeuvring at supersonic Mach numbers. Figure 13.3 shows the predicted SEP
chart at n = 3.5. The result shows that the aircraft is not capable of sustaining this load
factor over a wide area of the envelope. The thick solid line shows the acceleration
limits for the airplane at the same weight in 1g-flight: the only way the aircraft can
move to a high supersonic manoeuvre area is to accelerate past the speed of sound
at a lower load factor.
13.2.3 Turn Rates and Corner Speed

The turn rate is the angular velocity of the aircraft. Consider again the case of a
coordinated turn. The turn rate is
U g 2
q= = n − 1. (13.29)
χ U
If we write the turn speed in terms of the load factor n, we find

2W
U= n . (13.30)
ρ ACL
Thus, the turn rate becomes
"
ρg ACL n2 − 1
q= . (13.31)
2W n
12 40
9 30
Altitude, 10 feet
Altitude, km
3
20
30
6
0
3 10
0
60
60
30
90
1g flight
0
0 0
0.4 0.8 1.2 1.6 2
Mach number
Figure 13.3. Specific excess power chart for the supersonic aircraft model in 3.5g flight, mass
m = 11,000 kg.
Equation 13.31 shows how the turn rate is dependent on altitude, CL and W/A.
When maximum turn rates are quoted, it is essential to specify at least the altitude
and weight.
The corner velocity is the minimum velocity at which the aircraft can achieve
its maximum load factor. It is found at the intersection between curve of maximum
constant n−factor with the curve corresponding to a manoeuvre at maximum CL.
Starting from the stall speed
1/2
2nW
VS = Ks , (13.32)
ρ ACLmax
the relationship between the load factor and the Mach number is
γp 1
n= CL M2 . (13.33)
2Ks2 W/A max
Equation 13.33 is obtained by replacing the equation of state of the ideal gases to
remove the speed of sound. The maximum CL is a function of the Mach number;
therefore Equation 13.33 is implicit. A numerical solution is shown in Figure 13.4 at
selected load factors. We estimate a corner Mach number M = 0.84. corresponding
to a turn rate q = 0.31 rad/s (17.5 deg/s) for a g−limit = 9. For a given load factor,
the effect of increasing flight altitude is to increase the corner Mach number. In the
stratosphere, there is virtually no effect.
We conclude by recalling that general three-dimensional turn problems for
minimum time and minimum fuel have been solved by Hedrick and Bryson 2;3 , using
the energy method. Kelley 4;5 solved the problem of differential turning with climb.
0.5
0.4
Turn rate, rad/s Corner Speed / Mach

0.3
STALL
CL = constant
0.2
Increasing CL
CLmax 8
9
6
4
0.1
n=2
0
0.6 1 1.4 1.8 2.2
Mach number
Figure 13.4. Corner speed for supersonic jet aircraft model; standard atmosphere.
13.2.4 Minimum-Fuel Turn

The banked turn is a common manoeuvre for a commercial aircraft. An example of
this occurrence is during take-off and landing, when the runway heading is different
from the assigned flight path, as shown in Figure 13.5. A simple analysis of the turn
fuel is possible if the aircraft’s drag is parabolic and if the turn is performed at a
constant altitude. A steady-state turn is defined by the following equations:
D = T, (13.34)
W = Lcos φ, (13.35)
mU ψ̇ = Lsin φ. (13.36)
These are the balance forces in the flight direction, in the vertical direction, and in
the direction normal to the flight path, respectively.
dm f = ṁ f dt = f j Tdt,
dt T
Tdt = T dψ = dψ.
dψ ψ̇
The fuel consumption during the turn is proportional to the time of the manoeuvre
or to the angle of turn ψ. Therefore, the total fuel is equal to the integral
ψ
dψ ψ
mf = fj T fj T . (13.37)
o ψ̇ ψ̇
For a fixed turn altitude, the problem is to minimise the ratio T/ψ̇. The solution must
be subject to the balance of forces. Divide Equation 13.34 by Equation 13.36 to find
T CD
= mU . (13.38)
ψ̇ CL sin φ
Eastbound
Take−Off Landing
Figure 13.5. U-turns at take-off and landing, for eastbound flights with prevailing westerly
winds.
We note that
T
= f (CL, φ, h). (13.39)
ψ̇
Therefore, at a fixed bank angle and fixed air speed, the condition of minimum will
be given by the condition of minimum drag, or maximum glide ratio,

∂f mU ∂ CD
= = 0. (13.40)
∂CL sin φ ∂CL CL
This equation shows that the condition for a minimum-fuel turn is to fly at minimum-
drag speed. The corresponding lift coefficient is
"
CDo
CL = . (13.41)
k

The inverse of the glide ratio becomes D/L = 2 CDo k , whereas the optimal condi-
tion becomes
T mU
=2 CDo k. (13.42)
ψ̇ sin φ
This quantity is not dependent on the flight altitude. The fuel required to perform a
U-turn at constant altitude is then
π
mU mU
mf = 2 fj CDo k dψ = 2π f j CDo k. (13.43)
0 sin φ sin φ
The result is that the turn fuel depends on the flight altitude. For example, if the
AUW = 140 tons, a turn normal load factor n = 1.1 at 200 KTAS would lead to a
turn fuel of the order of 120 kg. This is not a small amount of fuel: turning costs
13.3 Unpowered Turns 369
a considerable amount of fuel. The FLIGHT code has been programmed to always
perform minimum-fuel turns.
13.3 Unpowered Turns

We present the turn performance of unpowered airplanes and gliders. One important
consequence of a steady-state turn of an unpowered vehicle is a loss of altitude. The
calculation of the sinking speed vs , the radius of turn χ and the turn rate can be done
if the bank angle φ is assigned. The dynamics equations in the direction of the flight
path and in the vertical direction are, respectively:
D − W sin γ = 0, (13.44)
Lcos φ − W cos γ = 0. (13.45)
The sinking rate is

∂h DU
vs = =− . (13.46)
∂t W
If we replace the parabolic drag equation in Equation 13.46, we have

ρ ACDo 3 ρ Ak 2L 2 3
vs = − U − U . (13.47)
2W 2W ρ AU 2
The lift is found from Equation 13.45. Thus, we have
ρA 2W k
vs = − CDo U 3 − (n cos γ )2 . (13.48)
2W ρAU
Equation 13.48 has two contributions, both negative. For its solution, we require
the flight-path angle γ . The latter quantity can be calculated from the glide ratio of
the unpowered airplane, L/D, that is found by dividing Equation 13.44 by Equa-
tion 13.45,

D n
tan γ = , γ = tan−1 . (13.49)
Lcos φ L/D
Equation 13.49 requires the L/D, which is dependent on the speed. The turn rate is
found from
U g L g
= sin φ = cos γ tan φ, (13.50)
χ UW U
where we have used the combination of Equation 13.44 and Equation 13.45. The
radius of turn is found from the first of the equivalences in Equation 13.50. The
solution of this model is shown in Figure 13.6 for a sailplane starting a turn at an
altitude of 1,000 m (3,048 feet). The data used are CDo = 0.007, k = 0.022, m = 450
kg, and A = 17.0 m2 . For the same case and for a load factor n = 1.25, we have
calculated the best gliding speeds. These speeds are shown in Figure 13.7, along with
the turn radius. The minimum glide angle can be found graphically, as the tangent
to the sinking speed curve from the origin of the axes.
-0.5
Straight flight
n = 1.25 (φ = 36.9 degs)
n = 1.5 (φ = 48.2 degs)
Sinking speed, m/s

-1
Wing
Stall
-1.5
-2
10 20 30 40 50
TAS, m/s
Figure 13.6. Sinking speed for glider turning at selected load factors.
13.4 Manoeuvre Envelope: V -n Diagram

The velocity-load-factor (V-n) diagram is a flight manoeuvre envelope involving both
structural and aerodynamic limits. The speed axis can be either the Mach number
or the equivalent air speed (EAS). We consider separately the cases of subsonic and
supersonic flight.
KTAS
0 20 40 60 80
0 300
dvs/dU = γ
-0.5 Min glide angle

Sinking speed, m/s
200
Turn radius, m
-1
Min sinking speed
100
-1.5
Sinking speed
Turn radius
-2 0
0 10 20 30 40 50
TAS, m/s
Figure 13.7. Sinking speed for the glider of Figure 13.6; load factor n = 1.25.
13.4 Manoeuvre Envelope: V -n Diagram 371
A D
S/L 3 5 7 9 11
2 Flap 3 A2
Load factor S/L

1 D1
E
0
-1
H F
-2
0 0.2 0.4 0.6 0.8
Mach number
Figure 13.8. V-n diagram for a transport aircraft model; m = 58,800 kg, standard day
(calculated).
TRANSPORT AIRCRAFT. Figures 13.8 and 13.9 show the typical manoeuvre limits of
a transport aircraft (the Airbus A320). In Figure 13.8 the shaded area indicates the
normal area of operation. The Federal Aviation Regulations* prescribe a positive
limit load factor 2.5 ≤ n ≤ 3.8, although its value is determined by the gross weight.
The negative limit is n ≥ −1.
The limit condition with flaps extended has been calculated with Flap-3 (δ f = 15
degrees; δs = 22 degrees). The numbers 3, 5, . . . indicate flight altitudes in km. The
letters A, B, . . . denote the standard FAR points in the V-n diagram. In Figure 13.8
we have plotted the load factor versus the flight Mach number, from the solution
obtained with the FLIGHT code. The Mach limit is the dive Mach number, MD.
For reference, the cruise Mach has been assumed as MD − 0.1, although in practice
there can be small variations, as discussed in Chapter 12. An increasing flight alti-
tude pushes the left limit of the envelope to the right, both with plain and flapped
configurations.
The key operating points are shown in Figure 13.9, which is a different version
of the diagram, with the horizontal axis now being the equivalent air speed (KEAS)
instead of the Mach number. A number of observations are reported here:
r For both plain and flapped configurations, the left limit is virtually independent
of the KEAS.
r For clean configuration, the right limit depends on the flight altitude. An increas-
ing altitude causes the KEAS to decrease.
r Point A is the design manoeuvre point with flaps retracted. When n = 1 (1-g
flight), the corresponding EAS is VA.
r Point D corresponds to the design dive speed (or Mach number).
* Federal Aviation Regulations, §25.335 (Transport Category Airplanes), Subpart C (Structures).

No Flap A D
L A2
2 Flap 3
11 9 7 5 3 S/L
A1
Load factor
1
VD
E
0
VS1 VF VA VC
-1
H F
-2
0 200 400 600
KEAS
Figure 13.9. V-n diagram for a transport aircraft model; m = 58,800 kg, standard day
(calculated).
r Point F corresponds to the limit −1g at the design cruise speed.

r Point H corresponds to the limit −1g flight with flaps retracted (or −C
Nmax ).
r Point L corresponds to the limit load factor with flaps extended.
SUPERSONIC FLIGHT. The manoeuvrability of an aircraft depends on its ability to

perform rapid turns. This capability has some limits, which are imposed by the
aerodynamic characteristics, the propulsion system and the structural dynamics,
and – increasingly – on the resistance of the human body. The essential parameters
are the thrust ratio T/W, the wing loading W/Aand the maximum CL. There are two
types of envelopes: one is the load factor versus Mach number; the other is the turn
rate versus Mach number. First, consider the steady-state turn at constant altitude.
From the lift equation, we have
2nW
CL = . (13.51)
ρ Aa 2 M2
The relationship between normal load factor and Mach number becomes
1 a 2 M2 CL
n= = f (h, M, CLmax , W/A). (13.52)
2 W/A
From Equation 13.52 we draw the following conclusions:
r For a flight in the troposphere, the load factor decreases with increasing altitudes.
A turn in the lower stratosphere is not affected by altitude.
r At a given altitude, the maximum load factor is limited by C . This parameter
Lmax
depends on the Mach number, as explained in Chapter 4. However, also a
manoeuvre at CL < CLmax may be of concern because of buffeting associated
to the unsteady separated flow or because of control and stability problems.
13.4 Manoeuvre Envelope: V -n Diagram 373
Transonic Dip
8
A D
5
S/L 7
6 A2 9
Maneuver 3 11
Flap
Load factor
4
S/L MD
2
D1
0 E
-2
H F
-4
0 0.5 1 1.5 2 2.5
Mach number
Figure 13.10. V-n diagram for a supersonic jet aircraft; m = 15,800 kg, standard day.
Therefore, the CL for turning must be a fraction of CLmax (M); we call this fraction
the usable lift (see Figure 4.22 in Chapter 4).
r The third limit is due to the maximum engine thrust. One can think of moving
along the line of structural limit and increase the Mach number as much as the
engine thrust allows. The maximum Mach number during a turn also depends
on the angle of attack. The governing equation is
1
nT = ρ A CDo + ηCLα α 2 a 2 M2 , (13.53)
2
or
n T(M)
M2 = , (13.54)
c1 CDo + ηCLα α 2
where c1 = ρ Aa 2 /2 is a constant coefficient at a constant turn-altitude. Equa-

tion 13.54 is an implicit relationship between the mass m and the angle of
attack α.
r Finally, at a critical speed the load factor can exceed the structural limit. There-
fore, if other factors do not intervene, the manoeuvre envelope has a flat top.
A qualitative example of manoeuvre envelope for a high-performance aircraft

is shown in Figure 13.10, as obtained with the SFLIGHT program. For a given gross
weight, the envelope depends on the flight altitude. The lift is limited by the subsonic
buffet and the horizontal tail deflection limit. The maximum g−factor depends on
the structural limits. The maximum level speed depends on the altitude; therefore,
the speed limit in the flight envelope can move backward or forward, depending
on the aircraft. The manoeuvre envelope at sea level is indicated by the shadow
area.
13.4.1 Sustainable g-Loads

We have previously discussed the manoeuvre limits from the point of view of the
aircraft. There is one more factor that limits the centrifugal accelerations of the
aircraft: the pilot. The large accelerations usually encountered by an aircraft in a
tight turn can reach values beyond human sustainability. An important reference is
NASA’s Bioastronautics Data Book 6 . LeBlaye 7 discussed aspects of aircraft agility
and sustainable loads. Jaslow 8 studied the problem of loss of spatial orientation in
a coordinated turn and its effects on accident rates on various classes of aircraft. All
of the studies confirm that posture is critically important. Pilots of high-performance
aircraft can tolerate forces of +9g with the use of special suits, as discussed in Ref. 9 .
Humans lose consciousness when the acceleration is in the +g direction, when blood
is drawn away from the brain toward the feet.
A sustainable g-load is an acceleration that does not affect the pilot in any
serious way (orientation, vision, heart beat, blood pressure). At a high g-load even
at a relatively low Mach number, at the highest load factors, the pilot’s breathing
must be aided by oxygen reserve. A load factor n 4 sustained for more than 5
seconds is potentially dangerous. A load factor n ∼ 2 can be endured for a long
time. Unconsciousness sets in at variable times, depending on the g−level (0.1 to 1
second).
13.5 Roll Performance

In this section we consider the one-degree of freedom roll of a high-performance
aircraft. During a roll the aircraft rotates around its longitudinal axis whilst main-
taining a straight course. The free parameter is the roll angle φ. Roll rates and roll
accelerations for these aircraft can be relatively high and are a key factor to their
overall agility. Roll performance and requirements, as from U.S. Military Standards,
generally refer to 1) maximum roll rate, 2) time to roll to 90 degrees, and 3) time to
execute a 360-degree roll.
The roll discussed below is difficult to achieve at high speeds due to a number of
unwanted effects. One such effect is the yaw response. If the aircraft is turning left in
roll, it will cause a yaw in the right direction. This effect is called adverse yaw. Another
important effect is the inertial coupling (or roll coupling). When an aircraft rotates
around an axis that is not aligned with its longitudinal axis, the inertial forces tend to
swing the aircraft out of the rotating axis, with potentially lethal consequences. The
problem is compounded by low-aspect-ratio wings, high speeds, high altitudes and,
consequently, high inertial forces compared to the aerodynamic forces. This problem
has been treated theoretically by Phillips 10 , before it was actually encountered, and
is a complex subject for stability and control. Phillips pointed out that there is a
precise relationship among the moments of inertia that defines the limits of inertial
stability and instability in roll. The critical flight conditions for yaw divergence,
pitch divergence and auto-rotational rolling have been theoretically established 11 .
Seckel 12 reported that a steady-state roll is a manoeuvre that cannot be performed
but must involve cyclic variations of angles of attack α and side-slip β.
In the present case, we consider an aircraft that does not have problems with
inertial coupling. When the aircraft starts a roll (φ = 0), its longitudinal axis is not
inflow
Roll angle φ = 0
Ωr wing going up
U
aileron down ξ
wing going down

U
Ωr Roll angle φ = 180
inflow
Figure 13.11. Inflow conditions on starboard wing, assuming an anti-clockwise roll in the flight
direction.
aligned with the rotation axis (though by a small angle); this generates a side-slip β.
When the aircraft is on its side (φ = 180 degrees), side-slip force has to be generated
to balance the weight. The combination of yaw and roll can lead the aircraft into
inertial coupling instability.
Consider now the situation illustrated in Figure 13.11. After 180 degrees, the air-
craft will find itself upside-down. The inflow on the wing is made of two components:
one is due to the forward speed U; the other is due to the roll itself. If we assume
that the former component does not change during the aircraft rotation, then the
starboard wing will find itself with the same angle of attack αo of the wing-level flight.
This angle is generally small.
The rotational component of the speed changes is shown in Figure 13.11. The
maximum angle of attack is at the tip and is estimated from
U 2U
tan α = = , (13.55)
pb/2 pb
where p = φ̇ is the roll rate. It can be verified that in many cases the forward speed
is much higher than the rotational speed; therefore: tan α α. Figure 13.12 shows
the arrangement with exaggerated angles. In general, the higher the roll rate and
the lower the forward speed, the higher the risk of stall. The problem is complicated
by the presence of a spanwise component of the velocity. Delta wings with various
aerodynamic arrangements can improve the stall behaviour at high angles of attack.
Wings for high roll rates must be able to operate at high angles of attacks without
stalling.
A rolling moment can be created by operating on the control surfaces. For
reference, the ailerons are set at an angle ±ξ on either side of the symmetry plane
(yb = 0), although in practice the deflection angle can be different. In the latter case,
the angle ξ to consider is a mean value. A linear response cannot be guaranteed
because the suction and pressure on the wing tips distort the flow. The vertical tail
creates additional flow separation and damping.
2b
Portside Starboard
UP pb/2
p DOWN
φ
U
pb/2
α
pb/2
Figure 13.12. Inflow conditions on rolling wing, seen on plane normal to flight velocity. Anti-
clockwise roll in the flight direction.
In practice, a deflection speed of about 60 deg/s is achieved. With a maximum

aileron authority of 30 degrees (up and down), about half a second is required for
the aileron to reach its limit position. In most roll analyses, a step change in aileron
angle is considered. To calculate the variation of angle of attack during the rotation
around the x-axis, we have to add the effect of the rotational speed. Therefore, the
inflow direction at spanwise position y has the following expression:
p y p y
α(y) tan−1 . (13.56)
U U
The one-degree-of-freedom aileron roll is described by the differential equation
Ix ṗ = Lξ + L p , (13.57)
where Ix is the principal moment of inertia around the roll axis x, ṗ = φ̈ is the angular
acceleration, p = φ̇ is the role rate, Lξ is the rolling moment due to aileron deflection,
and L p is the aerodynamic damping moment due to roll rate p = φ̇. Therefore, the
one-degree-of-freedom aileron roll is governed by a forcing moment (created by
the aileron) and a damping moment (created by the aerodynamics). We rewrite the
moments in the following form:

∂L ∂L
Lξ = ξ, Lp = p. (13.58)
∂ξ ∂p
These derivatives express the moment curve-slopes, that is, the moment response due
to a small change in aileron angle and roll speed, respectively. With these definitions
the roll Equation 13.57 becomes

1 ∂L 1 ∂L
ṗ = ξ+ p. (13.59)
Ix ∂ξ Ix ∂p
The parameter defined by

1 ∂L
Lξ = (13.60)
Ix ∂ξ
is the aileron effectiveness. The effectiveness Lξ denotes the scale of lateral response
to a unit input in aileron deflection. Likewise, we define the damping in roll parameter
Lp :

1 ∂L
Lp = . (13.61)
Ix ∂p
This quantity is the derivative of the rolling moment with respect to the roll rate,
normalised with the moment of inertia. By using the definitions of effectiveness and
damping in roll, the roll equation finally becomes
ṗ = Lξ ξ + Lp p. (13.62)
The parameters Lξ and Lp are integral properties of the aircraft and depend on the
Mach number, density altitude and several geometrical details. If Cl is the rolling-
moment coefficient, we have:

∂L ∂ 1 ρ AU 2 b ∂Cl ρ AU 2 b
= ρ AbU Cl =
2
= Clξ , (13.63)
∂ξ ∂ξ 2 2 ∂ξ 2
where Clξ denotes the derivative of the rolling-moment coefficient with respect to
the aileron deflection. In other words, Clξ is a non-dimensional aileron effectiveness.
The response to the aileron deflection ξ is
ρ AU 2 b
Lξ ξ = Clξ ξ, (13.64)
2Ix
from which we find the relationship between Clξ and Lξ :
2Ix
Clξ = Lξ . (13.65)
ρ AU 2 b
The damping in roll can be written as

1 ∂L 1 ∂ 1 ρ AU 2 b ∂Cl ρ AUb2 ∂Cl
Lp = = ρ AbU Cl =
2
= .
Ix ∂p Ix ∂p 2 2Ix ∂p 4Ix ∂(pb/2U)
(13.66)
The damping moment is associated to the damping-in-roll coefficient

∂Cl ∂Cl 2U ∂Cl
Cl p = = = . (13.67)
∂(pb/2U) ∂α b ∂p
The values of Cl p are dependent on the speed regime (subsonic or supersonic),
on the wing geometry (taper ratio, aspect-ratio, wing sweep) and aeroelastic effects.
Therefore, a generalisation cannot be made. The order of magnitude for the damping
coefficient is Cl p = −0.5 to −0.1. However, this value generally decreases when
approaching the speed of sound and increases again at low supersonic speeds. From
Equation 13.67 and Equation 13.66, we find
4Ix
Cl p = Lp . (13.68)
ρ AUb2
With substitution of Equation 13.65 and Equation 13.67 in the roll equation, we have

ρ AUb2 U 1
ṗ = Clξ ξ + Cl p p . (13.69)
2Ix b 2
Equation 13.69 is a differential equation that must be solved with the initial condi-
tions
p(t = 0) = 0. (13.70)
The factor
2Ix
τ= (13.71)
ρ AUb2
has the dimension of a time; it is a time constant depending on the aircraft. The
solution of Equation 13.69 with the boundary conditions Equation 13.70 is
2U Clξ
p= ξ 1 − e−t/τ . (13.72)
b Cl p
The asymptotic value of the roll rate is
2U Clξ
pmax = ξ. (13.73)
b Cl p
The moment response due to aileron deflection is not always linear. The damping
and aileron derivatives depend on a fairly large number of parameters, including the
geometry of the aircraft (wing, tail, and aileron) and operational conditions (altitude,
Mach number and aileron deflection). The functional relationships are
U = a(h) M, Clξ = f (M, h, geometry), Cl p = f (M, h, geometry). (13.74)
These laws depend on the aircraft, as discussed next.
13.5.1 Mach Number Effects

The ESDU reports some useful charts for roll analysis 13;14 ; in particular, it publishes
charts to interpolate the values of Lp and Lξ for a given wing geometry, aileron
deflection and Mach number. Some experimental data are available in Sandhal 15
(delta wings of rocket-propelled vehicles at supersonic speeds up to M = 2.0), Myers
and Kuhn 16;17 (swept-back wings at M < 0.8) and Anderson et al. 18 (straight wings
at transonic speeds). The Mach number effect is marginal for most wings up to the
transonic divergence, when there is a drop in the aileron effectiveness.
The effects of Mach number on the roll rate depend on the behaviour of the
damping-in-roll and on the aileron effectiveness. These parameters have specific
values for each aircraft but can also be extrapolated from existing general tables, such
as ESDU 14;13 . Roll performance, with particular reference to damping characteristics
as a function of Mach number, has been investigated by Stone 19 , Sanders 20 and
others. Theoretical derivations of the damping-in-roll coefficients have been done
by Malvestuto et al. 21 and Jones and Alskne 22 at supersonic speeds.
Some roll data for early military airplanes were collected by Toll 23 . These data
proved that the roll rate reaches a maximum value before decreasing more or less
13.6 Pull-Up Manoeuvre 379
2 0
CLp
CLβ
1.5 -0.5
3
CLξ x 103
CLβ x 10
1 -1
0.5 -1.5
0 -2
0.5 0.6 0.7 0.8 0.9
Mach number
Figure 13.13. Mach number effects on aileron effectiveness and rolling-moment derivative
for airplane with a straight wing; wing section; airplane Republic R-4, wing 10% thick, α = 0
degs, twist = −2 degs. Elaborated from Anderson et al. 18 .
sharply – depending on the airplane and with some exceptions. The best roll rates up
to the end of World War II were the Spitfire with clipped wings (150 deg/s) and the
Focke Wulf Fw-190 (160 deg/s). A Spitfire with clipped wings had a lower radius of
gyration, a lower roll moment of inertia and hence a higher roll rate. It is reported 24
that the original version of the Spitfire-I was unable to compete with the agility of the
Focke Fw-190 because this had a higher roll rate. The highest roll rates of acrobatic
aircraft are in the range of 300 deg/s.
The aileron may be necessary at high speeds for roll control. In fact, as the speed
increases, there is another unwanted response: an increase in effective dihedral
angle (wing dropping, or uncommanded roll). The rolling moment starts appearing
at transonic speeds in steady level flight from an asymmetry in the airplane. A right
side-slip (in the flight direction) causes a left roll-off, and vice versa. “Wing drop”
refers to an abrupt and irregular lateral motion of the aircraft, due ultimately to an
asymmetric wing stall. If uncontrolled, it can cause an unstable roll. This is shown
for a case of a straight wing in Figure 13.13, elaborated from Anderson et al. 18 ; Clβ
is the rolling-moment derivative with respect to the side-slip angle β: Clβ = ∂Cl /∂β.
This problem appeared in the first generation of transonic aircraft in the 1950s.
Wing drop has affected most high-performance military aircraft to the present day.
Research on the subject is available in Rathert et al. 25 , Chambers and Hall 26 , and Hall
et al. 27 .
13.6 Pull-Up Manoeuvre

The pull-up manoeuvre is a turn in the vertical plane and flight path inclined by a
variable angle γ . The dynamic equations of the aircraft are the same as in the case
of climb flight (see Chapter 10), that we rewrite here for convenience:
∂h
= U sin γ , (13.75)
∂t
∂U
m = T − D − W sin γ . (13.76)
∂t
If the load factor n = L/W is introduced in Equation 13.76, we find
∂γ ∂γ ∂γ g
= vc = U sin γ = (n − cos γ ) . (13.77)
∂t ∂h ∂h U
By eliminating the speed U from Equation 13.77 with the definition of CL, we find

∂γ gρ ACL n 1
= − . (13.78)
∂h 2nW sin γ tan γ
Equation 13.78 expresses the change of heading with respect to time. This rate
of change is dependent on the wing loading, on the flight altitude, and on the lift
coefficient. A maximum of the turn rate is found by deriving Equation 13.78 with
respect to n and CL.
When the aircraft pulls up, there is a change in tail-plane incidence, αht , which
depends on several factors. This change in incidence is
q gn
tan αht = xht = xht , (13.79)
U U2
where xht is the distance between the aerodynamic centre of the tail-plane and the
CG, q; is the turn rate. Thus, the tail-plane incidence increases with the load factor
and with the distance of the tail-plane. If the resulting incidence αht is too high, there
is a risk of tail-plane stall.
13.7 Flight in a Downburst

Localised thunderstorms near the ground are a serious hazard to aircraft flight,
particularly in the terminal-area manoeuvre. A thunderstorm with a strong vertical
component of the wind (downdraft) associated to heavy precipitations is called
downburst. Its discovery happened only in the 1970s. Exceptionally, the downdraft
has speeds of up to 30 m/s. As the winds hit the ground, they spread radially; thus they
create hazards to aircraft either approaching the runway or taking off. A downburst
of 5 km radius (2.7 n-miles) easily fits into an airport. Such a downburst is sometimes
called a “microburst” and lasts between 2 and 5 minutes. Larger downbursts can last
up to 20 minutes and are sometimes called “macrobursts.”
The wind shear has two components: horizontal and vertical; the former com-
ponent is basically a head or a tail wind. It can change by as much as 100 kt per n-mile;
the latter component generally results in turbulence that may affect the stability of
the aircraft through the wind-shear layer.
The effect of a microburst is highlighted in Figure 13.14. There are three import-
ant effects to note. First, there is the increase in IAS as the aircraft approaches the
downburst and enters the outflow area. The second effect is the loss of altitude as
the aircraft traverses the column of downward air. Finally, as the aircraft exits the
downburst through the outflow region, the IAS drops.
downburst
Gust front
Gust front
rain
Figure 13.14. Effect of microburst on aircraft (rotorcraft) descent and landing.
DOWNBURST PHYSICS. For a downburst to be considered wet it must be accompanied

by more than 0.25 mm of rain; in all other cases the downburst is to be considered
dry. For a downdraft to be considered a downburst, the vertical wind must exceed
20 m/s (Srivastava 28 ). It has also been noted that a key difference between the
downburst and other downdraft phenomena is that a downburst continues to gain
strength and intensity as it descends below the cloud base and that the maximum
intensity is realised close to the Earth’s surface. The downdraft has also been called
a reverse tornado because rather than drawing air up through a short column into a
storm cloud, the downburst creates an intense, narrow downward flow of air below
the cloud which then spreads outwards radially. However, in the tornado the airflow
has a large rotational component, whereas the downburst does not have a rotating
wind component.
The formation of a downburst begins with a mild downdraft in a region with
strong precipitation, for example in a storm cloud (rain and/or ice). The downdraft
is created by an air mass having higher density (negative buoyancy) with respect to
the surrounding atmosphere; this causes the air mass to descend. The negative buoy-
ancy that drives the downdraft is the result of both cooling of the air mass through
melting and evaporation of precipitation and the precipitation loading present in
the air mass. It is possible that the downdraft’s intensity increases significantly as it
descends, to the point where its descent velocity exceeds ∼20 m/s; at higher descent
speeds a downdraft is considered to have a magnitude sufficient to be considered
a downburst 28 . The main condition required to cause this effect is a deep layer of
air with a significant temperature lapse rate. There is a variety of combinations of
temperature lapse rates (also referred to as temperature stratifications) and pre-
cipitation rates that can cause sufficient vertical velocity to create a downburst. A
dry downburst typically has a temperature lapse rate approaching the dry adiabatic.
In contrast with more stable temperature stratifications, such as approaching the
wet adiabatic, heavy precipitation rates are required to drive the downdraft and
increase its intensity. It is also possible for a downburst to be created and driven
entirely below the cloud base. In such circumstances the precipitation loading and
evaporative cooling below the cloud are sufficient to drive the downburst.
The problem of storm penetration must be addressed for safety reasons and
for certification of the aircraft for operation under realistic weather conditions.
A number of accidents due to downbursts have now been documented. Zhu and
Etkin 29 produced a model for the wind field in a downburst, based on a distribu-
tion of dipoles, and provided analytical formulas for the head-wind, side-wind and
downdraft components. Since then, a number of developments have been produced.
Zhao and Bryson 30;31 proposed models for optimal guidance of an aircraft through
the downburst. In addition, a number of papers have addressed escape strategies 32 – 34
and avoidance 35 .
DOWNBURST WIND MODEL. We assume the wind-field model proposed by Zhu and
Etkin 29 . The model is inviscid and is based on a circular distribution of doublet σ
at height h. A mirror image of the singularity distribution is assumed to force the
boundary condition of no through-flow at the ground. According to the model, the
velocity induced at a point x, y, z in a ground-based reference system by a source
distribution of intensity σ per unit area is

3 (z − h)(x − ξ )
u=− σ (ξ, η) dξ dη, (13.80)
4π S r5

3 (z − h)(y − η)
v=− σ (ξ, η) dξ dη, (13.81)
4π S r5

1 3(z − h)2 1
w=− σ (ξ, η) − 3 dξ dη, (13.82)
4π S r5 r
where r is the distance
r 2 = (x − ξ )2 + (y − η)2 + (z − h)2 , (13.83)
and ξ, η is the coordinate of the source σ . To force the boundary condition on

the ground, these velocities are augmented by the velocity induced by a source
distribution −σ at altitude −h. The value of these surface integrals depends on
the source distribution. However, the solution is assumed to be axi-symmetric, and
the source distribution σ (ξ, η) is only a function of the radial distance from the centre
of the downburst
σ = σm [1 − sin(r/R)] . (13.84)
In Equation 13.84 σm is the maximum value of the momentum source. The para-
meters of the model are 1) the height of the source distribution, 2) the area of the
area distribution, and 3) the source-strength distribution. These parameters can be
chosen to fit reasonable downbursts. For the integration of Equations 13.80 to 13.82,
we perform a transformation to cylindrical coordinates. The value of the source σm
is calculated iteratively, so that the condition of downburst velocity w is enforced at
a specified altitude below the cloud base. We follow this procedure:
r Set the downwash speed w (a negative value) at the core of the downburst, at
an altitude equal to half the cloud base.
r Use the bisection method with a starting guess of σ = 1 and σ = 105 . These
m1 m2
values of the source strength should bracket the solution. Upon convergence,
we have forced the wind speed to be equal to w at the specified point.
Y X
Figure 13.15. Downburst field on the plane y = 0; cloud base at 1,000 m, R = 1,500 m.
Figure 13.15 shows the downburst field on the vertical plane y = 0 through the
downburst. The graph shows the streamlines from the cloud base. The contour lines
represent the magnitude of the wind velocity. Note that with this wind model the
velocities are unrealistically high just below the cloud base; the wind components
are singular as r → 0. Alternative wind models exist, including vortex rings 36 and
general axi-symmetric models 37 .
Another example of a field solution is shown in Figure 13.16a. This case refers to
an aircraft flying level with KTAS = 250 at an altitude 1,000 m (3,048 feet) through
the centre of the downburst. The figure shows both the horizontal and the vertical
velocity components created by the wind model. The maximum downwash is right
below the cloud base (centre of the downburst), as expected. The corresponding
change in effective angle of attack is shown in Figure 13.16b.
13.7.1 Aircraft Manoeuvre in a Downburst

The position {x, y, z} of the aircraft is calculated with respect to a reference system on
the ground. We assume that the aircraft moves on the vertical plane and will have a
position {x, z}, where the x-axis is in the direction of advancing flight and z is a vertical
20 4
R = 2,000 m R = 2,000 m
h = 1,500 m w
h = 1,500 m
w = -5 m/s @ 500 m u
wmax = -5 m/s @ 500 m
downburst velocities, m/s
10
0
Δ α, degrees
-4
-10
-20 -8
-3 -1.5 0 1.5 3 -3 -1.5 0 1.5 3
x/R x/R
(a) (b)
Figure 13.16. Wind field in a downburst and change in wing’s angle of attack in a downburst.
horizon
θ
γ FL
Vg
FP
U α
Vgx u
Vgz Vg
w
U
Wtas u
Utas
(a) head-wind
horizon
θ
γ FL
Vg
U α FP
Vgx u
Vgz Vg
w
Wtas U
u
Utas
(b) tail-wind
Figure 13.17. Velocity components and angles in the ground reference system.
axis pointing up. At this position the wind-speed components are {u, v, w}. These
speeds are calculated from the downburst model. The velocity components with
respect to the ground system are {Vgx , Vgy , Vgz}. We call γ the flight-path angle (i.e.,
the angle between the ground velocity and the horizon) and θ the pitch angle (i.e.,
the angle between the fuselage line and the horizon). We also call α the angle bet-
ween the fuselage line and the true air speed. These angles and velocities are shown
in Figure 13.17. The head wind causes the inflow angle α to increase; the TAS
decreases; the opposite occurs with a tail wind. In the absence of winds, the inflow
angle is α = γ − θ.
The velocity components in the ground reference system are:
Vgx = Vg cos γ , Vgy = 0, Vgz = Vg sin γ . (13.85)

The true air speed has the following components

Utas = Vgx + u, Vtas = v, Wtas = Vgz + w (13.86)
with
2
2 1/2
TAS = U = Utas + Vtas
2
+ Wtas . (13.87)
The flight-path angle is

−1 Vgz
γ = tan . (13.88)
Vgx
The inflow angle α is

Wtas
α = tan−1 − θ. (13.89)
Utas
If the attitude is θ = 0 in absence of atmospheric winds, the aircraft provides a lift
CLθo > 0, this is equivalent to having a zero-lift angle
CLθo
αo = − . (13.90)
CLα
Thus, the lift coefficient can be written as
CL = CLα (α − αo). (13.91)
As the aircraft enters the downburst, there is a change in inflow angle, as illustrated
in Figure 13.17, and thence a change in lift and drag. In particular, an increase in
lift would cause the aircraft to decrease the descent rate and possibly even climb.
If the inflow angle α is to be maintained, whilst maintaining the pitch angle (pitch
guidance), a change in speed is required. From Equation 13.89, we must set the
constraint

Wtas Vgz + w Vgz
tan−1 = tan−1 = tan−1 ; (13.92)
Utas Vgx + u Vgx t=0
where t = 0 denotes the time before entering the downburst event. Equation 13.92
contains two unknowns: Vgx and Vgz; if we attempt to maintain the same ground
speed schedule as required in a conventional approach, we only need to control the
descent rate Vgz. In a typical downburst, w > u (as shown by the Etkin wind model),
along with Vgz > Vgx . Upon entering the downburst, the numerator of the argument
increases in absolute value, whilst the denominator decreases. Therefore, to keep
the effective inflow angle α constant, the true descent rate must decrease whilst the
aircraft accelerates. Past the centre of the downburst, the opposite event takes place,
and the aircraft must slow down whilst increasing its descent rate, although a too
rapid loss in altitude would compromise safety at a very critical time.
An alternative guidance mode is a constant flight-path angle (γ -guidance). This
mode is specified by the following equation:

Wtas − w
γ = tan−1 = const. (13.93)
Utas − u
Upon entering a downburst, the numerator of the argument decreases (Wtas < 0;
w < 0), whilst the numerator increases (the aircraft encounters a head wind, u < 0).
0 6 800 200
α KTAS
γ h
Angle of attack α, degrees

Flight path angle γ, degs
600 190
-2
KTAS
5
Altitude, m
400 180
-4
200 170
-6
3 0 160
0 10 20 30 40 50 0 10 20 30 40 50
Flight time, s Flight time, s
(a) (b)
Figure 13.18. Flight in a downburst with deceleration to a target approach speed.
The flight-path angle is maintained if the aircraft slows down, something that is
achieved by reducing the thrust. Another form of control is based on acceleration.
Figure 13.17 shows that the airspeed tends to decrease upon entering the downburst
and it increases on the way out. A control law that would ensure constant TAS would
require an acceleration and then a deceleration. Examples of these types of controls
are found in the technical literature 38 .
13.7.2 Case Study: Flight in a Downburst

The analysis of the downburst relies on several parameters, including the height of
the cloud base, the diameter, the downdraft at the core of the downburst and the
initial conditions of the aircraft (including whether the aircraft flies through the axis
of the downburst or sideways). The overall performance depends on the control
laws, which include at least a TAS-control, flight-path control and angle-of-attack
control, with various constraints (on descent rate, true air speed, normal load factor).
We provide an example here, which was simulated for the Airbus A320-200-
CFM landing in the presence of a downburst. The case refers to an initial mass m =
56,400 kg, with an initial TAS = Vgr een + 5 kt at 2,500 m above sea level. The airplane
is forced to maintain a fixed attitude, equal to the value in absence of winds. The
characteristics of the downburst are 2,500 m diameter with a cloud base at 1,000 m
and a maximum downdraft of −10 m/s on the axis at 500 m below the cloud base.
Figure 13.18 shows some characteristics of the trajectory when the airplane is
guided to a target TAS, equal to an approach speed on the ground. Figure 13.19
shows a similar example, but with the airplane attempting to maintain a TAS within
95% of the initial value. In both cases, the airplane sinks below the downburst, and
unless it is guided by optimal control laws, it may end up on the ground before
reaching the runway. The sinking effect is shown by the flight-path angle becoming
steeper as the airplane enters the core of the downburst. The results have been
obtained with the flight-mechanics model of the FLIGHT program.
Bibliography 387
0 6 800 200
α
γ
Angle of attack α, degrees

Flight path angle γ, degs
600 190
-2
KTAS
5
Altitude, m
400 180
-4
4
200 170
-6
KTAS
h
3 0 160
0 10 20 30 40 50 0 10 20 30 40 50
(a) (b)
Figure 13.19. Flight in a downburst with TAS-control.
Summary
In this chapter we considered simple aircraft manoeuvres. We discussed both
powered and unpowered turns. We demonstrated that at constant thrust the aircraft
would descend, due to an increase in drag and the effects of apparent centrifugal
acceleration. We further demonstrated that these accelerations can exceed the res-
istance of the human body (sustainable g-loads). Thus, there are limits imposed
not only by the structures and the dynamics of the airplane but also by the pilot.
Clearly an unmanned vehicle would not have this problem, and it is easy to imagine
new high-performance aircraft systems capable of even more rapid turns. Other
high-performance manoeuvres include roll (with limitations imposed by inertial
coupling). At a more mundane level, we have discussed the manoeuvre envelopes of
transport-type aircraft (V-n diagram). We concluded this chapter with a discussion of
flight in a downburst, which is an adverse weather condition sometimes encountered
at low altitude near an airfield.
Bibliography
[1] AGARD WG 19. Operational Agility, volume AR-314. AGARD, April 1994.
[2] Bryson AE and Hedrick JK. Minimum time turns for a supersonic airplane at
constant altitude. J. Aircraft, 8(3):182–187, 1971.
[3] Hedrick JK and Bryson AE. Three-dimensional minimum-time turns for super-
sonic aircraft. J. Aircraft, 9(2):115–121, Mar. 1972.
[4] Kelley HJ. Differential-turning optimality criteria. J. Aircraft, 12(1):41–44,
1975.
[5] Kelley HJ. Differential-turning tactics. J. Aircraft, 12(2):930–935, Feb. 1975.
[6] Webb PMD. Bioastronautics Data Book, volume SP-3006. NASA, 1964.
[7] LeBlaye P. Human Consequences of Agile Aircraft, RTO-TR-015, chapter Agil-
ity: Definitions, Basic Concepts and History. NATO, Jan. 2001.
[8] Jaslow H. Spatial disorientation during a coordinated turn. J. Aircraft, 39(4):
572–576, 2002.
[9] AGARD. Current Concepts on G-Protection Research and Development, LS-

202, May 1995.
[10] Phillips WH. Effect of steady rolling on longitudinal and directional stability.
Technical Report TN 1627, NACA, June 1948.
[11] Pinsker WJG. Critical flight conditions and loads resulting from inertia cross-
coupling and aerodynamic stability deficiencies. Technical Report ARC CP-
404, Aeronautical Research Council, 1958.
[12] Seckel E. Stability and Control of Aircraft and Helicopters. Addison-Wesley,
1964. (Appendix I contains aircraft performance and stability data).
[13] ESDU. Rolling moment derivative Lξ for plain aileron at subsonic speeds. Data
Item 88013. ESDU International, London, Oct. 1992.
[14] ESDU. Stability derivative Lp , rolling moment due to rolling for swept and
tapered wings. Data Item Aero A.06.01.01. ESDU International, London,
Mar. 1981.
[15] Sandhal CA. Free-flight investigation of the rolling effectiveness of several delta
wings – Aileron configurations at transonic and supersonic speeds. Technical
Report RM-L8D16, NACA, Aug. 1948.
[16] Myers BC and Kuhn RE. High subsonic damping-in-roll characteristics of a
wing with quarter-chord line swept 35◦ and with aspect-ratio 3 and taper ratio
0.6. Technical Report RM-L9C23, NACA, May 1949.
[17] Myers BC and Kuhn RE. Effects of Mach number and sweep on the damping-
in-roll characteristics of wings of aspect-ratio 4. Technical Report RM-L9E10,
NACA, June 1949.
[18] Anderson SB, EA Ernst, and RD van Dyke. Flight measurements of the wing-
dropping tendency of a straight-wing jet airplane at high subsonic Mach num-
bers. Technical Report RM-A51B28, NACA, April 1951.
[19] Stone DG. A collection of data for zero-lift damping in roll of wing-body
combinations as determined with rocket-powered models equipped with roll-
torque nozzles. Technical Report TN 3955, NACA, April 1957.
[20] Sanders CE. Damping in roll of models with 45◦ , 60◦ , 70◦ delta wings determined
at high subsonic, transonic and supersonic speeds with rocket-powered models.
Technical Report RM-L52d22a, NACA, June 1952.
[21] Malvestuto F, Margolis K, and Ribner HS. Theoretical lift and damping in roll
at supersonic speeds of thin sweptback tapered wings with streamwise tips,
subsonic leading edges, and supersonic trailing edges. Technical Report R-970,
NACA, 1950.
[22] Jones AL and Alksne A. A summary of lateral-stability derivatives calculated
for wing plan forms in supersonic flow. Technical Report R-1052, NACA, 1951.
[23] Toll TA. Summary of lateral-control research. Technical Report R-868, NACA,
1947.
[24] Skow AM. Agility as a contributor to design balance. J. Aircraft, 29(1):34–46,
Jan. 1992.
[25] Rathert G, Rolls L, Winograd L, and Cooper G. Preliminary flight investigation
of the wing dropping tendency and lateral control characteristics of a 35◦ swept-
back airplane at transonic Mach numbers. Technical Report RM-A50H03,
NACA, Sept. 1950.
[26] Chambers JR and Hall RM. Historical review of uncommanded lateral-
directional motions at transonic conditions. J. Aircraft, 41(3):436–447, May
2004.
[27] Hall RM, Woodson SH, and Chambers JR. Accomplishments of the abrupt-
wing-stall program. J. Aircraft, 42(3):653–660, May 2005.
[28] Srivastava RC. A model of intense downdrafts driven by the melting and evap-
oration of precipitation. J. Atmospheric Sciences, 44(13):1752–1773, 1987.
[29] Zhu S and Etkin B. Model of the wind field in a downburst. J. Aircraft,
22(7):595–601, July 1985.
[30] Zhao Y and Bryson AE. Optimal paths through downbursts. J. Guidance,
Control and Dynamics, 13(5):813–818, Oct. 1990.
[31] Zhao Y and Bryson AE. Control of an aircraft in a downburst. J. Guidance,
Control and Dynamics, 13(5):819–823, Sept. 1990.
[32] Bobbitt RB and Howard RM. Escape strategies for turboprop aircraft in
microburst windshear. J. Aircraft, 29(5):745–752, Sept. 1992.
[33] Dogan A and Kabamba PT. Escaping microburst with turbulence: Altitude,
dive and guidance strategies. J. Aircraft, 37(3):417–426, May 2000.
[34] Mulgund SS and Stengel RF. Target pitch angle for the microburst escape
maneuver. J. Aircraft, 30(6):826–832, Nov. 1993.
[35] de Melo DA and Hansman RJ. Analysis of aircraft performance during lateral
maneuvering for microburst avoidance. J. Aircraft, 28(12):837–842, Dec. 1991.
[36] Ivan M. A ring-vortex downburst model for flight simulations. J. Aircraft,
23(3):232–236, Mar. 1986.
[37] Visser HG. Optimal lateral-escape maneuvers for microburst wind-shear
encounters. J. Guidance, Control & Dynamics, 14(6):1234–1240, 1994.
[38] Miele A, Wang T, and Bowles RL. Acceleration, gamma and theta guidance for
abort landing in a windshear. J. Guidance, Control, & Dynamics, 12(6):815–821,
Nov. 1989.

a = speed of sound
A = wing area
AR = wing aspect-ratio
b = wing span
bt = thrust arm
c = wing chord
c1 = constant coefficient
Cl = rolling-moment coefficient
CLθo = lift coefficient at zero attitude, Equation 13.90
Cl p = damping-in-roll coefficient
Clξ = non-dimensional aileron effectiveness
Clβ = rolling-moment derivative with respect to side-slip
CLmax = maximum-lift coefficient
CN = normal-force coefficient
CM = pitching moment
D = drag force
f = generic function, Equation 13.39
h = altitude
ho = arbitrary starting altitude
Ix = roll moment of inertia
k = lift-induced drag factor
Ks = stall margin
L = lift force
L = rolling moment
Lp = aerodynamic moment due to roll rate
Lξ = rolling moment due to aileron deflection
Lp = damping in roll
Lξ = aileron effectiveness
m = mass
M = Mach number
MD = design dive Mach number
M = yawing moment
n = normal load factor
N = pitching moment
ṁ f = fuel flow
p = pressure
p = roll rate (or angular velocity)
pmax = maximum value of roll rate (asymptotic)
P = power
r = distance; distance in polar coordinates
R = radius of downburst
t = time
T = net thrust
u,v,w = wind-velocity components in downburst
U = air speed
vc = climb rate
VD = design dive speed
VF = maximum design speed with flaps extended
Vg = ground speed
vs = descent rate
VS = stall speed
VS1 = stall speed, 1-g flight, flaps retracted
W = weight
u, v, w = wind-velocity components in wind model
w = average downwash
xac = distance of wing’s aerodynamic centre from leading edge
xcg = distance of centre of gravity from nose
x, y, z = Cartesian coordinates (ground-based)
Greek Symbols
α = angle of attack; inflow angle
αo = zero-lift angle of attack/inflow
β = side-slip angle
γ = flight-path angle; ratio between specific heats, Equation 13.33
δ = flap or slat angle
η = induced-drag coefficient for supersonic model
η = coordinate in downburst-wind model
θ = pitch attitude
ϑ = angle in polar coordinates

ξ = aileron deflection; coordinate in downburst-wind model
ρ = air density
σ = doublet distribution in wind model
σm = maximum value of σ
τ = time constant, Equation 13.71
φ = bank angle
χ = radius of curvature
ψ = turn angle
[.]ail = aileron
[.]b = body-conformal
[.] f = fuel
[.]md = minimum drag
[.]t = turn
[.]tas = true air speed
[.]w = wind
[.] p = reference respect to roll rate
[.]ξ = reference to aileron deflection
14 Thermo-Structural Performance
Overview
This chapter deals with thermodynamic and structural models of various aircraft
sub-systems. Two main aero-thermodynamic problems are discussed: cold-weather
operation (§ 14.1), including aircraft icing (§ 14.1.1), and fuel temperature in the
aircraft tanks (§ 14.3 and § 14.4). Basic characteristics of aircraft fuels are discussed
in § 14.2. A lumped-mass model for thermo-structural performance of aircraft tyres
is presented in § 14.5 and aims at the rapid prediction of tyre temperatures. This is
useful at high speeds, heavy aircraft and rejected take-off. The final problem is that
of jet blast (§ 14.6), which is of relevance during ground operations, for safety of
infrastructure and ground personnel.
KEY CONCEPTS: Cold-Weather Operations, Aircraft Icing, Aviation Fuels, Fuel

Temperature, Tyre Heating, Jet Blast.
14.1 Cold-Weather Operations

Cold-weather operations cause a number of problems on the ground, at take-off, at
landing and in flight. There are a variety of issues of relevance, including icing, in-
flight ice protection, de-icing on the ground, fuel characteristics at low temperatures
and aircraft performance on contaminated runways (including breaking perform-
ance, precipitation drag, aquaplaning). We will not deal with ice protection, de-icing
and the atmospheric causes of icing because they are beyond the scope of flight
performance. Our presentation will focus on icing effects on performance charac-
teristics.
The problem of aircraft icing is exemplified by a number of fatal crashes that
over the years have plagued commercial aviation. Aircraft icing has its beginning in
World War II, with bombers flying at high altitudes and at all-weather conditions.
Ever since, a large database of icing cases has been collected. Due to the complexity
of icing phenomena, it became clear that aircraft could not be certified for all icing
conditions. Today, icing conditions for certification must make reference to the
Federal Aviation Regulations FAR § 25, Appendix C (Continuous Maximum Icing
and Maximum Intermittent Icing).
392
Clean Wing
1.5
Wing (iced)
Airplane CL
Right Wing (iced)
0.5
Left Wing (iced)
0
-5 0 5 10 15 20
True angle of attack, degs
Figure 14.1. Icing effects on the wing lift of the ATR72; data adapted from Ref. 4 .
Icing can occur on various parts of the vehicle, although icing on the wing, control
surfaces and propeller blades (if present) is particularly severe. Typical consequences
include an increase in drag and a decrease in lift – not necessarily by the same amount
on right and left wings. Differential changes in aerodynamic parameters can lead to
flight conditions beyond the normal flight envelope of the airplane.
Modern aircraft are certified to operate in icing conditions, within reasonable
limits. Icing effects can be traced by the Flight Data Recorder (FDR), either directly
or indirectly. There is abundant documentation of aircraft icing in the technical
literature, from the physics of ice formation to atmospheric effects and aircraft
operations. It is not within the scope of this chapter to review the whole subject,
which is addressed in the appropriate literature 1;2 . Ice formation on the ground and
in flight are discussed by Asselin 3 .
To clarify the icing severity on airplane lift and drag, we show the recorded data
for a cargo ATR72-500, adapted from Caldarelli 4 , in Figure 14.1 and Figure 14.2.
This case refers to a fatal crash. A peculiar aspect of this incident is the difference
between left- and right-wing lift, alongside a rapid decrease in drag. For the ATR, a
CL = −0.5 causes an increase in stall speed by 10 kt with flaps at 15 degrees. Most
critically, this loss in lift is caused by very small ice accretions.
As the aircraft begins to stall, its lateral and longitudinal stability are comprom-
ised. With regard to the former problem, wing stall occurs first at the inboard sections
of the wing and proceeds outwards, thereby minimising the destabilising effects of
rolling moment. However, this scenario holds only in case of symmetric ice accretion,
which, as we have seen, might not be the case. With regard to the latter problem,
there can be an abrupt change in pitching moment as well as tail-plane icing.
Figure 14.3 shows some typical wing conditions, with and without icing. Figure
14.3a displays the normal operation of the wing, when the airflow is smooth and the
boundary layer is attached. Figure 14.3b shows a case of leading-edge icing, whose
0.04
Wing Drag (iced)
0.03
Airplane CD
0.02
0.01
Clean Wing
0
0 5 10 15 20
True angle of attack, degs
Figure 14.2. Icing effects on the wing drag of the ATR72; data adapted from Ref. 4 .
effect is to change the shape of the wing section and hence to displace the air flow
away from the solid wall. The boundary layer thickens and eventually separates in
the aft portion of the chord. Figure 14.3c displays the case of a landing configuration
in which icing is present on the leading-edge slat. Although this is potentially a
dangerous condition, if there is no icing on the wing itself, it is possible that the wing
boundary layer remains attached as a result of the flow through the slat-wing gap;
this flow energises weak boundary layers on the suction side of the wing. Finally,
Figure 14.3d shows the effect of morning frost, which typically extends through most
of the suction side. The surface is modified, particularly in roughness, and boundary
layer separation is likely.
The case of Figure 14.3b is used for certification purposes. The effect of aft
boundary layer separation does not affect the circulation by a great deal, unless the
wing is forced to operate at large angles of attack. The aircraft would be certified to
land under conditions displayed in Figure 14.3c because the wing lift is not severely
affected by icing on the slat. The final case is more problematic and can result in
large-scale separation and consequent loss in circulation, that is, lift. This case is not
certified for safe flight.
14.1.1 Aircraft Icing

Icing occurs with various degrees of severity, defined by the aviation authorities as
trace icing, light icing, moderate icing and severe icing. Severe icing occurs when de-
icing systems fail to remove the hazard. In this case, immediate flight diversion is
required.
The atmosphere contains water vapour in variable proportions. At each tem-
perature and pressure, the maximum water vapour is limited by the saturation
point (or dew point) condition. At saturation, the water condenses in rain or
(a) Normal conditions
(b) Leading-edge icing
(c) Slat icing
(d) Morning dew icing

Figure 14.3. Some typical wing-icing conditions.
ice crystals. Both rain and ice form clouds. The water vapour at saturation point
increases with the air temperature. Saturation is attributed to two important phe-
nomena: lifting of warm air (due to weather instability or ground topography)
and air cooling at night with a clear sky. The lower temperature causes at least
part of the vapour to precipitate into water or ice. However, there are condi-
tions when the condensation or freezing point may be exceeded without precip-
itation taking place. This phenomenon is called supercooling. This event is known
to be unstable, and once a perturbation is introduced, condensation or freezing
is almost instantaneous. This is the basic mechanism that causes aircraft icing. In
practice, the range of air temperatures that cause icing is −40◦ C < T < −10◦ C, at
altitudes from ground level to ∼3,048 m (10,000 feet, FL-100). The parts of the
aircraft most exposed to icing are the protruding parts, including lifting surfaces,
leading-edge slats, hinges, engine intakes, vertical and horizontal stabilisers and
so on.
A rather large number of parameters are known to affect the formation of the
ice. These include the air temperature, water-drop-size distribution, water content,
air speed, flight altitude, type of cloud, aircraft size, local skin temperature, anti-icing
system and more. The ice shapes are at least as varied. Two of the most dangerous
ones are the double horn at the leading edge and runback ice, that is, a flat lump of
ice behind the leading edge. Larger leading-edge radii and slotted slats are known
to be effective in reducing the risk of wing icing in flight. The air speed is a critical
parameter because it affects the so-called aerodynamic heating. A higher speed can
cause an increase in stagnation temperature and thus a decreased risk of wing icing.
Thus, icing at temperatures above −10◦ C is unlikely.
One atmospheric phenomenon that is known to cause the formation of large
droplets of ice is the so-called temperature inversion. Normally, the temperature
gradient is negative (i.e., the temperature decreases with increasing altitude), but
there are cases when the temperature gradient becomes positive. Around the inver-
sion point there is a pocket of relatively cold air. Rain originating from the warmer
layer above the inversion turns into freezing rain, freezing drizzle and ice. It is not
uncommon that frozen droplets collide and coalesce, thereby forming larger droplets
that are heavier and fall down faster.
A freezing drizzle is made of uniform precipitation made of very small drops,
d < 500 μm, that freeze upon impact with a solid surface or the ground. A freezing
rain is made of larger droplets or small droplets that are widely separated. Both
freezing rain and freezing drizzles are a form of supercooled droplets, which can
grow to sizes 100 times larger than droplets considered in the icing certification
standards. For certification purposes, the droplets must have mean diameters in the
range of 5 to 135 μm, whilst supercooled droplets can reach a diameter of 2,000 μm
(2 mm).
Droplet trajectories near the solid surfaces of the aircraft are strongly depend-
ent on their size. Smaller particles tend to follow the streamlines and will travel
downstream, except near the stagnation areas, where they can be forced to impact
with the airplane. Larger particles are less subject to aerodynamic forces and more
susceptible to inertia forces; hence, they will travel on more straight trajectories and
upon impact will generate larger areas of ice coverage. Inertia forces grow with the
droplet diameter d3 , whilst aerodynamic forces grow as d2 . Thus, as the droplets
become larger, their dynamics is dominated by their inertia. This scenario assumes
that upon impact the droplets crash on the surfaces and essentially coalesce around
their impact area. The in-flight icing process itself is a result of a balance between
the freezing of supercooled air on the aircraft and the removal of ice due to erosion,
evaporation and sublimation.
Another icing phenomenon is the cold soak, in which low skin temperature
comes into contact with moist air at air temperatures that would normally be above
freezing. For example, it has been found in the past that the practice of tankering
requires the aircraft to land at the destination airport with so much fuel that parts
of it are in contact with the upper skin of the wing. Because the temperature of this
fuel is below freezing at normal landing conditions, the wings become heat sinks and
freeze the moisture in the air. Hence, wing icing may occur at air temperatures well
above 0◦ C. Icing conditions are more frequent than aircraft icing. Two important
NO ICING
Air temperature, Celsius

-10
POSSIBLE ICING
-20
ice
-30
-40
0 100 200 300 400
Air speed, kt
Figure 14.4. Typical-ice accretion limits.
changes in aircraft operations are increase in air speed and change of flight altitude,
as shown in Figure 14.4. Ground icing, including the thin layer of morning frost,
must be cleared before any aircraft operation. Even a small ice thickness can cause
unwanted aerodynamic degradation during the critical ground roll, take-off and
initial climb.
14.2 Aviation Fuels

In this section we deal briefly with the main characteristics of aviation fuel, including
types of fuel, specifications, composition and fuel properties. Aviation fuel is a refined
product of crude oil; about 12% of aviation fuel is derived from each barrel of crude.
There are two classes of fuels within this category: 1) aviation gasoline (also called
AvGas), used for reciprocating engines; and 2) turbine fuels, used for gas turbine
engines. We will refer only to the latter category.
There are various commercial grades of turbine fuels, including Jet-A, Jet-A1,
Jet-B, JP-4, JP-5, JP-7, JP-8 and JP-9. The most widespread fuels are Jet-A and Jet-
A1; Jet-B (also called wide-cut) has a lower freezing temperature. The remaining
specifications are U.S. military grades. These fuels are fully specified by international
standards* , which define the full array of physical properties and characteristics
(density, viscosity, volatility, composition, combustion, corrosion, additives, toxicity,
and so on).
Technical information on aviation fuels is available in Goodger and Vere 5 ,
in addition to some international standards. Table 14.1 shows some basic data of
aviation fuels. The most important properties for performance calculations are the
* United States: ASTM D1655; IATA: ADD76-1; United Kingdom: DERD.2494; France: AIR 3405/C
(for Jet-A1).
Table 14.1. Characteristics of aviation fuels, at 15◦ C; data are averages
Fuel Wide-cut Kerosene AVgas
Specific weight 0.762 kg/l 0.810 kg/l 0.715 kg/l

Specific combustion heat 43.54 MJ/kg 43.28 MJ/kg 43.71 MJ/kg
fluid density as a function of temperature, the freezing point, the combustion heat
and the stoichiometric ratio. Table 14.2 is a summary of characteristics for Jet-A and
Jet-A1 provided by the company BP. These data, in particular the density and the
freezing points, are different from the ones quoted by other manufacturers.
FUEL COMPOSITION. There are four main groups of hydrocarbons in a turbine fuel:
1) paraffins, iso-paraffins, and cyclo-paraffins; 2) aromatics; 3) naphthalenes; and 4)
olefins. Other smaller components include sulphur (which has corrosive properties
and is a harmful combustion by-product), naphthenic acid (which is also corrosive),
water and additives. With reference to the latter category, there are a number of
approved chemicals, such as corrosion inhibitors, icing inhibitors, static dissipators,
anti-oxidants and so on.
TEMPERATURES. At the altitudes of normal flight operation the air temperatures can
reach values below −60◦ C, especially on cross-polar routes via the North Pole. The
aviation fuels currently used, mostly Jet A and Jet-A1, have a freezing point at a
higher temperature, which poses serious safety risks. In particular, the freezing point
of Jet-A1 is −47◦ C, the freezing point of the Jet-A is −40◦ C and Jet-B (wide-cut) has
a freezing temperature of −50◦ C. The Russian equivalent of Jet-A1 is called TS1,
although this fuel has a slightly lower freezing temperature. These are average values
that do not reflect the actual freezing phenomenon. At these low temperatures there
Table 14.2. Characteristics of turbine fuels Jet-A and

Jet-A1; data are adapted from BP
Properties Jet-A Jet-A1
Composition
Aromatics (volume) 23.4% 19.5%
Sulphur (mass) 0.07% 0.02%
Volatility
Boiling point 280◦ C 252◦ C
Flash point 51.1◦ C 42◦ C
Density, 15◦ C 820 kg/m3 804 kg/m3
Fluidity
Freezing point −51.0◦ C −50◦ C
Viscosity, −20◦ C 5.2 mm2 /s 3.5 mm2 /s
Combustion
Specific energy 43.02 MJ/kg 43.15 MJ/kg
Smoke point 19.5 mm 25.0 mm
Naphthalenes (volume) 2.9% 1.5%
can be a precipitation of waxy material that results in blocking of the fuel feed lines
and the fuel filters. As a consequence, some routes cannot be flown in very severe
conditions. For example, winter flight operations across Siberia can easily encounter
atmospheric temperatures of −70◦ C. For increased safety there is a warning system
ahead of the inlet into the engine’s combustors, although it is not always clear how
much capability for fuel heating there is. A safety margin quoted by Airbus is that
the limiting fuel temperature should be the freezing temperature, plus the engine’s
heating margin, plus 2◦ C for safety (Boeing requires 3◦ C above freezing point). The
addition of anti-freeze additives affects the water content in the fuel. The change in
temperature is more rapid toward the end of the flight, as a result of a lower heat
capacity of the remaining fuel. Stirring of fuel can prevent freezing. For example,
recirculating fuel with the booster pump can lower the freezing point by several
degrees. Other techniques to control the fuel temperature may include waste heat
from the power systems.
The properties of the fuel are somewhat variable, so we are bound to find
average values of the relevant physical properties. Considerable research exists in
this area, and different figures are given for certain properties. The heat capacity
of the liquid fuel at temperatures between 200 and 500 K is calculated from the
following semi-empirical expression (elaborated from Faith et al. 6 ):
C p [kJ/kg K] = 1.4103 − 4.8877 · 10−6 T f + 7.1575 · 10−6 T f2 . (14.1)
Simple calculations can be carried out with C p 2,000 J/kg K. The thermal conduct-
ivity is approximated by the following polynomial expression:
k [W/m K] = 2.6936 · 10−1 − 4.2273 · 10−4 T f + 2.0251 · 10−7 T f2 . (14.2)
In Equations 14.1 and 14.2 the temperature is in Kelvin.

Similar expressions could be derived for the fuel vapour. However, the data for
this phase are available mostly at high temperature. Some equations exist for the
fuel density of Jet-A and Jet-A1, such as:
ρ f 1.0490 − 8 · 10−4 T [kg/litre], (14.3)
where the temperature must be given in Kelvin. This equation provides a constant
gradient dρ f /dT f = −8 · 10−4 kg/K. A nominally empty volume is actually occupied
by fuel vapour.
The vapour pressure of the jet fuels is related to the absolute temperature
through the following approximate semi-empirical expression‡ :
p∗ [bar] = 2.256 · 103 exp(−3899/T ∗ ). (14.4)
From Equation 14.4 we can apply the equation of state (§ 14.4.1) to calculate the
corresponding vapour density ρ ∗f and hence the derivative of this parameter with
respect to the vapour temperature, Equation 14.21.
‡ Alternative expressions are found in the technical literature.

KTAS
100 200 300 400 500
1.2
50
Temperature recovery, C
1.15
T / To 40
30
1.1
B
20
1.05
A 10
1 0
50 100 150 200 250 300
TAS, m/s
Figure 14.5. Relationship among TAT, SAT, and recovery temperature.
14.3 Fuel Temperature in Flight

The fuel temperature varies greatly, as the fluid is subject to heat transfer with the
tanks and the atmosphere; the temperature may reach very low values, to a point at
which it crystallises and freezes. However, also high temperatures are to be avoided.
For example, the FCOM of the Boeing B737-300, the B747-400 and B767 specify
that the maximum tank fuel temperature cannot exceed +49◦ C (Jet-A and Jet-A1
only). Low temperatures cause difficulty in fuel pumping, transfer between tanks and
fuel moving through the engine feed lines and valves. This is particularly dangerous
if freezing occurs in and around the fuel feed system, although the presence of
small water particles in the fuel will certainly compound the problem. Such an event
has been attributed to the crash landing of a Boeing B777 at London Heathrow in
January 2008. For this reason, the fuel temperature is one of the many parameters
recorded by the FDR. Data elaborated‡ on more than 140,000 flights of the Boeing
B777 with three engine types (Rolls-Royce, General Electric and Pratt & Whitney)
indicate that the lowest recorded fuel temperature was −39◦ C, although in about
17% of the cases the temperature was never below −20◦ C.
Extended investigations on fuel-temperature profiles were carried out in the
1970s. Ref. 7 reports fuel temperature data for all the U.S. Air Force aircraft of the
time (including the B-52G, C-141, KC-135, XB70a, B-1 and others). Both Airbus
and Boeing produce fuel-tank-temperature statistics for their aircraft.
Statistical data indicate that the outer wing tanks usually have a fuel temperature
3 degrees below that of the inner tanks. Hence, under severe conditions the outer
tanks should be used first. The fuel temperature depends on the flight time. An
example is shown in Figure 14.6, which was elaborated from Airbus’s flight test data.
‡ Unofficial data of the Air Accident Investigation Branch (AAIB), 2008.
14.3 Fuel Temperature in Flight 401
30 0
Inner Tank
Outer Tank
20 Total Air Temp. -10
Static Air Temperature, C

Total Air Temperature, C
10
-20
0
-30
-10
-40
-20
-50
-30
0 1 2 3 4 5
Flight time, hours
Figure 14.6. Fuel cooling during a long-range flight (elaborated from Airbus A300/A310).
The outer tanks reach a stabilisation temperature after 3 flight hours. There are a
number of temperatures used in this type of analysis. The Static Air Temperature
(SAT) is the temperature of still air. The Total Air Temperature (TAT) is the SAT
plus a term called kinetic heating; the corresponding increase in temperature is called
temperature recovery. The relationship between these temperatures is given by the
energy equation
U2
T = To + . (14.5)
2C p
In Equation 14.5, TAT = T , SAT = To; the kinetic heating (or recovery temperat-
ure) is U 2 /2C p . Equation 14.5 is the compressible form of the Bernoulli equation; it
expresses the fact that during an adiabatic transformation, the loss in kinetic energy
is transformed in heat by an equal amount. For a flow reaching a point of stag-
nation, such as the leading edge of a wing, the recovery temperature is maximum.
Equation 14.5 is equivalent to the following:
T γ −1 2
=1+ M, (14.6)
To 2
where γ is the ratio between specific heats of the atmosphere and M is the flight
Mach number. The relationship among TAT, SAT and recovery temperature is
shown in Figure 14.5 as a function of the air speed; we have assumed a reference
SAT = To = 288 K (15◦ C). For example, at a speed of 250 kt the temperature
recovery is ∼9 degrees (point A); at 400 kt it is ∼21 degrees (point B). A total
temperature of −30◦ C at M = 0.78 corresponds to a SAT of about −57◦ C.
Why is there heat transfer between the fuel and the atmosphere? The fuel
tanks are not insulated for practical reasons (added structural weight and cost).
Due to the size of the tanks, the minimum fuel temperature is at the contact with
the tank surfaces; a positive temperature gradient exists from the surfaces inwards.
Fuel pumping, fuel mixing and tank rocking all contribute to lowering the freezing
temperature. A risk of fuel freezing will always exist. Because the fuel is at a tem-
perature below the freezing point of the water, it is possible (though unlikely) that
some crystals of water ice exist within the fuel. If these crystals are drawn toward
the fuel filters, there is the risk of reducing the fuel flow and clogging the fuel lines.
14.4 Fuel-Temperature Model

We now proceed with developing a low-order model to predict the movement of the
fuel temperature in flight. The method is based on an isothermal fuel (lumped mass).
This assumption greatly simplifies the problem. A more accurate method must rely
on a finite-element analysis of the fuel-tank system. However, the method presented
here can be implemented in a general flight-performance program to run alongside
other types of calculations.
Assume that the tank has an n−sided geometry. Its faces do not have to be
necessarily flat. In general, it is possible to calculate the exact volume of the geo-
metry and the contact surface S for any level of the fuel between a minimum and a
maximum. It is possible to define a numerical function
S = f (V) (14.7)
that gives the total wetted (or contact) area S for any fuel volume V. The heat
transfer from the fuel to the outer system takes place through contact. The free
surface of the fuel, S, is shared by the liquid and vapour phases. The differential
equation for this surface in terms of the fuel flow is

dS dS dVf dS
= ṁ f ρ f = ṁ f ρ f f1 (V), (14.8)
dt dVf dt dVf
where we have neglected the change in fuel density with respect to time; ṁ f is the
fuel flow out of the tank; f1 (V) denotes the numerical relationship between the
convection surface S and the fuel volume V. If the overline (.) is used for quantities
through the free contact surface and the starred quantities (.)∗ denote the fuel vapour,
the problem is formulated by the following ordinary differential equations:
dVf ṁ f
, (14.9)
dt ρf
dS
= ρ f ṁ f f1 (Vf ), (14.10)
dt
dS
= ρ f ṁ f f2 (Vf ), (14.11)
dt
dQ
= hc (T f − T ) S, (14.12)
dt
dQ
= hc T f − T f∗ S, (14.13)
dt

dT f 1 dQ dQ
= + , (14.14)
dt C pm f dt dt
dρ f dρ dT
f f
= . (14.15)
dt dT f dt
Equation 14.10 and Equation 14.11 are the rate of change of the wetted and free sur-
face, respectively; these are functions of the actual fuel volume Vf ; the latter quantity
depends on the time via the fuel flow rate, given by Equation 14.9. Equation 14.12
expresses the heat transfer rate by contact through all of the wetted surfaces of the
tank; Equation 14.13 is the heat transfer by natural convection through the free
surface of the fuel; Equation 14.14 is the rate of change of the fuel temperature.
Equation 14.15 is used to close the system; its right-hand side uses the chain rule
of differentiation, which allows a decoupling of the thermal properties of the fuel
(dT f /dt) from the time derivatives. The derivative dρ f /dT f is a physical property of
the fuel, as discussed in § 14.2.
An improvement to the model consists in writing a similar set of equations for
the vapour phase. Specifically, we would need to calculate the temperature and
density (or pressure) of the fuel vapour. The coupling of the differential equations is
made through the temperature T , that now represents the fuel-vapour temperature
rather than the static atmospheric temperature. The differential equations for the
vapour phase are:
dV ∗ dV
=− , (14.16)
dt dt
dS∗ dS
=− , (14.17)
dt dt
∗
dS dS
= , (14.18)
dt dt
dQ∗
= h∗c T f∗ − T S∗ , (14.19)
dt
∗
dQ dQ
=− , (14.20)
dt dt
∗

dT f∗ 1 dQ∗ dQ
= + , (14.21)
dt C p m∗f dt dt

dρ ∗f dρ ∗f dT f∗
= . (14.22)
dt dT f∗ dt
It can be observed that four of these equations are straightforward. In fact, Equa-
tion 14.16, Equation 14.17 and Equation 14.18 can be calculated directly from the
liquid fuel and the characteristics of the tank; Equation 14.20 is also calculated by
direct inference. Equation 14.21 and Equation 14.22 are more problematic. The
problem is to calculate the mass of the vapour (m∗f ) in thermal equilibrium with the
liquid fuel. We assume that m∗f = ρ ∗f V ∗ .
14.4.1 Fuel-Vapour Model

The fuel vapour is described with the non-linear gas equation of Peng and Robinson 8 .
The relationship among pressure, absolute temperature and molar volume Vm is
RT aα
p= − 2 , (14.23)
Vm − b Vm + 2bVm − b2
where R is the gas constant. The other terms are:
Tc2 Tc
a = 0.45724R2 , b = 0.07780R , (14.24)
pc pc
2
α = 1 + (0.37464 + 1.54226ω − 0.26992ω2 )(1 − Tr1/2 ) , (14.25)
T
Tr = . (14.26)
Tc
The quantities denoted with the subscript “c” are critical quantities, whilst ω is called
the acentric factor; Vm is the molar volume. For the Jet-A fuels, on average 9 we have
pc = 2, 235 kPa, vc = 0.564 l/mol, Tc = 537.9 K. (14.27)
The values of the acentric factor and the molar volume are, respectively,
ω = 0.457, Vm = 0.129 kg/mol. (14.28)
14.4.2 Heat-Transfer Model

To calculate the heat-transfer rates, we need to estimate the transfer coefficients.
This is done through the Nusselt number
hc L
Nu = , (14.29)
kc
which is the ratio between the convective heat transfer and the conduction heat
transfer through a solid surface. In Equation 14.29, L represents the characteristic
length of the process, which is L S/btank, that is, the ratio between the wetted area
and the spanwise extent of the tank btank. The thermal conductivity kc is assigned
from Equation 14.2; hc is the convection-heat-transfer coefficient. Thus, if a suitable
expression for the Nusselt number is found, then the convective heat transfer is
calculated from Equation 14.29.
From heat-transfer textbooks we can find various expressions for the Nusselt
numbers, depending on whether the flow is laminar, transitional or turbulent; specific
research addresses heat-transfer rates in jet fuels. In the present case, we use the
relationship
Nu L = 0.54Ra1/4 104 < Ra < 107 ,
(14.30)
Nu L = 0.15Ra1/4 Ra > 107
where “Ra” represents the Rayleigh number; Ra = Gr Pr; Gr is the Grashof num-
ber; and Pr = C p /μ/kc is the Prandtl number. This equation is appropriate for a
natural convection problem from a warm lower flat surface 10 and hence, with some
dxw
hc4 hc3
Vapour Fuel
Figure 14.7. Nomenclature for fuel tank

heat exchange.
hc2 hc1 Liquid Fuel
extrapolation, to the cooling of the liquid fuel. The Nusselt number for the vapour,
assuming that the top surface of the tank is being cooled, is nearly the same. For
the calculation of the Nusselt numbers we need a reference length. This is taken as
S/btank, although alternative choices are possible. The total heat-transfer coefficient
hc depends on the tank’s material construction (including its thickness) and can be
written as
1 1 dxw 1
= + + , (14.31)
hc hc1 kc hc2
where hc is the convection-heat-transfer coefficient (fuel- and air-side), dxw is the
wall thickness and as shown in Figure 14.7. The right-hand side of Equation 14.31
represents the total transmittance of the system. The heat transfer is dominated by
the conduction-convection on the air-side, with some contribution from the fuel-
side.
A typical value of thermal conductivity for aluminium is 220 to 250 W/m2 K,
whilst the convective heat transfer of the air is between 10 and 100 W/m2 K. The
natural convection-heat-transfer coefficient of air and fuel is another parameter that
does not respond to a simple estimate.
14.4.3 Numerical Solution

In summary, the system is made up by 13 ordinary differential equations (Equa-
tion 14.17 is redundant). For each phase (liquid and vapour), the unknowns are
the fuel temperature T f ; the corresponding density ρ f ; the contact and free-surface
areas S and S, respectively; the fuel volume V; and the heat-transfer rates by contact
and convection. The next step is to define the initial conditions:
r Initial temperature. On a hot day, T can exceed the OAT if the aircraft spends
f
time loading and taxiing. If the tank is partially full, there will be considerable
evaporation of fuel.
r Fuel flow. We need to know how the aircraft has been loaded and how the fuel
is to be used. Thus, we have an initial value of the fuel volume (or weight) in
the reference tank and a rule for fuel used.
1 1 1
1
B
0.8 0.8 A 0.8

0.8
Normalised contact area
Normalised free surface

Normalised fuel level
Normalised fuel level

A
0.6 0.6 0.6
0.6
B
0.4 0.4 0.4 0.4
0.2 0.2 0.2 0.2
Fuel-tank contact area Free surface

Fuel level Fuel level
0 0 0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Normalised fuel volume Normalised fuel volume
(a) (b)
Figure 14.8. Example of wing fuel-tank calculations: normalised fuel-tank wetted areas, fuel
level and fuel volume.
r Total air temperature and the fuel flow. The best way to acquire these data is to
perform a full mission calculation and store these two quantities as a function
of flight time.
An example of wing-tank characteristics is shown in Figure 14.8a; this figure

displays the fuel level and fuel-tank contact area for a given fuel volume; all data are
normalised. If the fuel level is 70%, the fuel volume is ∼40% of the tank’s capacity
(point A) and the fuel-tank contact area is ∼87% of the total area enveloped by the
tank (point B). The free surface shown in Figure 14.8b starts from zero (with the
tank empty); it reaches a maximum at a fuel level around 30% and goes down to
zero in the limiting case of full tank. The lack of smoothness is due to interpolation
errors.
Some simplification is required before proceeding. First, we neglect the heat
transfer between the liquid and vapour through the free surface, which causes Equa-
tion 14.11, Equation 14.13, Equation 14.18 and Equation 14.20 to be redundant.
A combination of the remaining equations is now possible. In particular, Equa-
tion 14.12 combined with Equation 14.14 leads to a simple equation for the liquid
fuel temperature:
dT f hc S
= (T f − T ) . (14.32)
dt C pm f
A similar combination for the vapour fuel leads to an equation for the vapour
temperature:

dT f∗ hc S ∗ ∗
= Tf − T . (14.33)
dt C pm f
The factor
C pm f
τ= (14.34)
hc S
represents a time constant in similar thermodynamic problems. The higher this factor,
the more slowly the fuel responds to changes in temperature. A too rapid fuel cooling
is not a good thing; therefore, one may want to find ways to increase the time constant.
From the definition of Equation 14.34, it can be gleaned that for a given fuel mass,
τ increases as the contact surface decreases. However, it is also clear that τ cannot
be constant because fuel flows in or out of the tank, the contact surface changes and
the heat-transfer coefficient changes as a result of variable outside conditions. In
this context, the fuel will cool more rapidly (within a given time) when the fuel mass
is low and the contact area is large. As an example, a fuel tank containing a fixed
amount of 20,000 kg of fuel would have an estimated τ 105 seconds, or about 168
minutes.
Toward the end of the flight, there is relatively little fuel left in the tanks, and
the OAT increases rapidly as the aircraft descends. Because the time constant is now
greatly decreased, the fuel responds more rapidly to a change of outside temperature,
as demonstrated in Figure 14.6.
14.4.4 Numerical Solution and Verification

In this section we illustrate the steps required for a numerical solution of the fuel-
temperature model. The sequence of operations is the following:
r Calculate a flight trajectory: altitude, OAT, fuel flow, as a function of flight time.
r Set the fuel-tank characteristics (fuel-tank volume, initial fuel and fuel flow).
r Set the initial conditions for the ODE: liquid/vapour fuel, geometric character-
istics.
r Integrate the system of Equation 14.32 and Equation 14.32 in the time domain.
We now consider the case of a fuel tank whose characteristics are shown in
Figure 14.8a. The flight trajectory is shown in Figure 14.9a, which indicates that
there are four step climbs at cruise from FL-280. There is a corresponding step
change in the OAT.
The system of differential equations is solved in the time domain by using a
fourth-order Runge-Kutta method with the time-step set by the flight-trajectory
data. Figure 14.9b displays the fuel volume and the liquid-fuel temperature during
the flight. The OAT is also shown for reference. The high fuel rates during the
climb cause the fuel level to go down faster (see left side of the graph). By the end
of the flight, the fuel tanks are relatively empty, and the remaining fuel increases
its temperature rapidly as the aircraft descends. Crucially, the method presented
neglects the effects of the thermal-layer thickness near the wall, where there are
considerable temperature gradients.
VERIFICATION OF THE METHOD. The proposed thermodynamic model is compared to

the FDR data of a Boeing B777-236ER that suffered a crash upon landing at London
Heathrow in January 2008 11 . This airplane, at an altitude of ∼700 feet (∼210 m),
suffered a reduction in thrust from the right engine, due to reduction in fuel flow; a
few seconds later the left engine was also reduced. The flight control system detected
this loss in fuel flow and commanded a full valve opening. No change in fuel flow
12 300
LIQ fuel Temp
OAT
Tank fuel Volume 0.3
10 280
280
Fuel volume (normalised)

Flight altitude, km
8 Altitude
Temperature, K
260 260
0.2
OAT, K
6
OAT 240 240

4 Jet A1 freezing
0.1
220
2 220
0 200 0
0 100 200 300 400 0 100 200 300 400
Flight time, min Flight time, min
(a) (b)
Figure 14.9. Flight trajectory used for fuel temperature calculations.
resulted, as a consequence of the fuel valves being clogged with ice within the system.
The remaining thrust was not sufficient to maintain control of the airplane, which
eventually crashed 1,000 feet (∼300 m) from the runway.
Figure 14.10 shows the simulated fuel temperature for this case; the simulation
is compared with the temperature data extracted from the FDR 12 . The graph shows
the liquid and vapour fuel temperature (as calculated), the SAT (from the FDR),
the tank fuel volume (from the FDR) and the fuel temperature (from the FDR).
LIQ fuel Temp (calc) 0.6

VAP fuel Temp (calc)
*
SAT
Tank fuel volume *
260 *
Fuel Temperature
Normalised fuel volume
0.4
Temperature, K
240
Jet A1 freezing
0.2
220
200 0
0 180 360 540
Flight time, min
Figure 14.10. Calculated fuel temperature for the flight recorded by the AAIB 11 . The data
with the asterisk ()* are FDR data.
Fuel level
Figure 14.11. Example of cold fuel
temperature distribution in box-shaped
tank.
-70 -60 -50 -40 -30 -20 -10 0

Fuel temperature, C
For reference we have added the conventional freezing temperature of the Jet-A1
fuel. The fuel temperature itself never reaches the freezing threshold. However,
it is possible that local temperatures are lower, especially if the fuel is contamin-
ated with water. In fact, the AAIB investigation revealed the presence of water
in the fuel, whose temperature reached a minimum of −34◦ C; the minimum TAT
was −45◦ C.
Because the average fuel temperature in the tank is higher than the freezing
temperature, a more detailed investigation may be needed. There exist computer
programs for the fuel-temperature prediction (for example, at the Boeing Com-
pany); these programs are more accurate than the model presented here and rely on
field solutions of the heat-transfer problem, with a finite-element or finite-volume
analysis. These programs establish that, contrary to the assumption of uniform tem-
perature, there is a positive temperature gradient from the bottom of the tank to the
free surface (the fuel at the bottom is the coldest). Furthermore, there is generally
a step change in temperature at the liquid-vapour interface, the vapour phase being
considerably colder, though not as cold as the fuel at the bottom of the tank. Thermal
stabilisation is reached within 2 to 3 hours into the flight. Figure 14.11 shows a typical
temperature distribution in a box-like fuel tank. The colder temperature is achieved
after about 3 hours.
14.5 Tyre-Heating Model

The history of aviation is littered with catastrophic failures that in hindsight turned
out to be design shortcomings† . In 1963, a Caravelle III jetliner crashed soon after
take-off from Durrenasch, Switzerland, due to a tyre explosion during landing-
gear retraction. Official investigations discovered that the tyre had exploded due to
† An exhaustive database of aircraft accidents around the world is available at the website of the
National Transportation Safety Board, www.ntsb.gov.
di dw
θ r
δt
compression b
lc
Figure 14.12. Aircraft tyre nomenclature.
over-heating during a missed take-off; tyre explosion damaged a fuel line, which in
turn started an explosion; no lives could be saved. The pilot, in fact, decided to clear
the morning fog with the aircraft wings, by taxiing the aircraft down to the end of
the runway and back. By the time the aircraft left the ground, it had run three times
on the runway. At the time there was ignorance of the abnormal tyre heating due to
heavy loads, friction effects on the ground and brake operation. Nowadays, a missed
take-off requires additional performance controls on the tyres and landing gear, what
are commonly called load-speed-time charts. The relevant regulations prescribe the
installation of appropriate thermal-sensitive devices. Furthermore, tyres must be
stopped before assuming their folded position in the hold.
Modelling the heating of a tyre under heavy load is an exceedingly difficult
problem. We propose a simple analysis that uses a limited set of data, such as those
that can be found for a conventional transport aircraft. As usual some assumptions
are required.
Figure 14.12 shows the nomenclature that will be used for deriving the ther-
modynamics and the pressure loading on an aircraft tyre. Some practical formulas
for tyre deflections are available in ESDU 13 . The tyre dimensions are given accord-
ing to some international standards. We will consider the specifications given as
tyre-diameter/tyre-width/tyre-rim in imperial units. For example, a marking
49 × 19.0−20 indicates a tyre with a diameter dw = 50 inches (1.270 m), a width
w = 19.0 inches (0.483 m), and a rim (dw − di )/2 = 20 inches (0.508 m). These
data are converted into SI units in order to be consistent. Additional data that are
required include at least the tyre pressure, the type of gas and the mass of the
tyre. Aircraft tyres are generally inflated with inert gases such as helium or nitro-
gen to minimise the effects of expansion and contraction during extreme changes in
temperature.
The front view of Figure 14.12 shows two possible deflections on the roadway,
which depend on the tyre pressure. We will assume that the width of the contact b is
equal to the width of the tyre, b w. This assumption is a valid one in the context of
the overall approximations that are required to estimate the heat loads. The length
of the contact is lc . This is conveniently found in terms of vertical deflection δt from
the formula 13 :
2
δt δt F
7.2 + 0.96 − = 0, (14.35)
dw dw ( p + 0.08 pr )w(wdw )1/2
where F is the external vertical load, p is the inflation pressure of the unloaded tyre
and pr is the tyre rated inflation pressure. This quadratic equation is solved in terms
of δt /dw . Only the positive solution makes sense. From the vertical compression δt ,
we can calculate the compression volume dV, which is equal to
1 2
dV = w θr − lc cos(θ/2) , (14.36)
2
where θ is the sector angle corresponding to the road contact lc (see Figure 14.12).
One of the key performance parameters of a tyre is the deflection, conventionally
given by
δt
δtyre % = · 100. (14.37)
dw /2
Aircraft tyres are designed for a deflection of 32%; this value is considerably higher
than tyres for other applications.
TYRE MASS. Often the mass of the tyre is unknown. It can be estimated from the
volume equation and a weighted specific mass. The volume is:

dw di
V = π wdw τ1 + 2π − τ1 − τ2 − (nwd)g . (14.38)
% &' ( % 2 2
&' ( % &' (
tread sides grooves
In Equation 14.38 di is the internal diameter (the rim), τ1 is the thickness of the
tread, τ2 is the average thickness of the sides, ng is the number of grooves, and wg
and dg are their width and depth, respectively. The resulting mass is the weighted
average of the tyre components:

m= V xi ρi , (14.39)
i
where xi is the proportion of material i and ρi is the corresponding specific mass.

The volume occupied by the inflating gas is a fraction of the above, and it can be
derived directly from Equation 14.38.
The rotational speed of tyre when aircraft has ground speed U is ω = U/r . The
frequency of the loading is f = ω/2π . With a tyre diameter of 1 m and a speed of
50 m/s, the frequency is about 16 Hz (a relatively low value), but the centrifugal
acceleration is a staggering 254g!
NUMERICAL METHOD. The increase in tyre pressure due to loading F on the wheel is
calculated iteratively. ESDU 13 suggests using the formula
2
w δt
pt = 1.5 p . (14.40)
dw dw
In our numerical method, this formula is inserted into a loop that allows the calcula-
tion of compression volume, contact length, pressure and temperature rise in static
conditions:
1. Calculate unloaded tyre pressure using the current gas temperature Tg : pt =

RTg ρg .
2. Calculate the vertical compression δt from Equation 14.35.
3. Calculate the change in pressure pt from contact (Equation 14.40).
4. Calculate the compression volume dV from Equation 14.36.
5. Recalculate the tyre pressure.
6. If the residual on the latest tyre pressure is below a tolerance, exit loop.
Before entering this loop, it is necessary to calculate the mass of the inflating
gas. This calculation is done under unloaded conditions, by using the equation of the
ideal gas. The next problem is to calculate the heat loads on both tyre and inflating
gas. It is convenient to write the energy equation separately for tyre and gas.
The tyre is loaded by the external force; thus, it deforms with a frequency f .
Part of the energy is lost due to hysteresis. The hysteresis H is defined as the ratio
between the energy lost and the deformation energy. The hysteresis depends on
tyre temperature and the frequency. Following some research published by Lin and
Hwang 14 , it is reasonable to assume that H varies between 0.1 and 0.3. The hysteresis
is a local parameter that changes within the tyre. The highest values correspond to
the lowest temperatures. Furthermore, the frequency effect seems to appear only
at these temperatures. A secondary effect, albeit an important one at high rolling
speeds, is due to the centrifugal forces. One consequence of the centrifugal loads
is the generation of a standing wave that causes a distortion on the side of the
tyre leaving the contact with the roadway. This condition is attributed to under-
inflation and should not normally be encountered during a take-off operation. The
deformation work done on the tyre during a full rotation is

2π
W1 = Fδt . (14.41)
lc
The work done during a time step dt is taken from

ωdt
dW = W1 . (14.42)
2π
The work actually lost in the deformation is dW H. This work is partly dissipated by
the tyre through external convection (Ec ), through internal convection (Eg ), through
friction with the inflating gas (E f ) and by physical contact with the roadway (Er ).
The remaining part (Et ) will serve to increase the thermal energy of the tyre itself.
Combination of Equation 14.41, Equation 14.42 and an appropriate value of the
hysteresis is the key to the whole method.
Brakes Load on wheel
Flange
Bead
heat
GAS
Figure 14.13. Model for thermo-structural External convection
loads on aircraft tyre. Internal
Convection
Structural model
Friction
Road surface
Reaction
HEAT LOADS. Figure 14.13 illustrates how the tyre is modelled. There is interaction
between the tyre and the roadway, the inflating gas and the external environment.
The gas interacts with the inner walls of the tyre. We need to find a way to estimate
the heat loads, which are expressed by the equation
dW = dEt − dEc − dEr − dEg − dE f . (14.43)
Additional terms that may have to be included in Equation 14.43 arise from the
internal heat that is generated during the deformation and eventually by specific
cooling systems. The radiation heat transfer to the gas is about three orders of
magnitude lower than the other contributions; thus, it is not considered in our
analysis.
The forced convective heat transfer from the tyre to the air can be calculated
by using a suitable expression of the Nusselt number as a function of the Reynolds
number. This relationship can be written as
Nu = a Reb. (14.44)
The coefficients a, b depend on the problem; the Reynolds number is calculated from
the aircraft speed and the rotational speed. For a thin rotating cylinder (w/dw
0.017), the equation is
Nu = 5.88 · 10−3 Re0.925 . (14.45)
There is the problem of extending this correlation to the typical widths of an aircraft
tyre (w/d 0.25). However, once this problem is solved, the convective heat flux is
dEc
Q̇c = = hc Ac (Tt − T ), (14.46)
dt
with hc and kc denoting the convective-heat-transfer coefficient and the thermal

conductivity. These quantities are calculated with the following equations:

kc
hc = Nu, (14.47)
dw
0.9
T
kc = 0.0241 . (14.48)
273.15
The convective area Ac is approximated by

dw − di
Ac ≈ 2π + π w(dw + di ) − wlc . (14.49)
4
The heat transfer to the inflating gas can be written with a similar expression, except
that the Nusselt number and the heat-transfer coefficient kc are different and depend
on the type of inflating gas. A suitable correlation is found in the technical literat-
ure 15 . These data are relative to heated rotating cylinders, with an axial gas flow.
The problem is governed by the rotating Reynolds number Rer = ωdw /2ν and the
axial flow Reynolds number Rea . The data indicate that for Rer > 2.77 · 105 , the
Nusselt number is insensitive to the axial flow, which shows that the heat transfer is
dominated by the rotation. At lower Rer the axial flow has a role that increases as
the rotation decreases. Data are available for Rea = 0, which corresponds broadly to
the present case, although there would be the problem of accounting for the end-wall
effects. The Nusselt number is calculated from

2.85 · 105 , Rer ≥ 2.77 · 105
Nu = −4 −9 (14.50)
15.77 + 6.52 · 10 Rer + 3.86 · 10 Rer , Rer < 2.77 · 105 .
The heat transfer with the roadway occurs through the contact surface Ac , which has
an elliptic shape, with axes lc and w; the resulting formula is:
dEr
Q̇r = = kct Ac (Tt − Tr oad ), (14.51)
dt
where kct is the thermal conductivity of the tyre. To simplify the matter, we assume
that the runway is at the same temperature as the atmosphere (in the presence of
direct sunlight it can be considerably higher). Finally, the energy accumulated by the
tyre itself is described by
dEt
Q̇t = = C pt mt (Tt − T ). (14.52)
dt
Equation 14.52 does not differentiate between parts of the tyre and will only provide
an average tyre temperature. Local differences in temperature have been measured
by McCart and Tanner 16 , who concluded that the temperatures along the inner walls
are lower than the temperatures around the outer surface. This result also implies
that the road contact is a more important factor. Data from Goodyear* indicate
that the difference in temperature between tread centreline and tread should be
* Manufacturers publish technical data to back up performance claims; often these are not official
reports.
negligible at normal taxi speeds. The temperature of the bead is higher at taxi speeds
above 30 km/h and may reach 30 degrees in excess of the tread. Thus, the method is
only applicable to taxi conditions.
The heat coefficient can be estimated by the weighted average of its components.
These are rubber (∼80%), steel chords (∼5%), and textile fabrics (∼15%), but
some manufacturers quote different proportions. The exact composition of a tyre
is proprietary information. Average heat-transfer coefficients are rubber: C p 1.5
kJ/kg K; steel: C p 0.46 kJ/kg K; and nylon-based fabrics: C p 1.7 kJ /kg K. The
weighted average is 1.478 kJ/kg K, although there can be considerable differences
with real tyres; furthermore, there can be special mechanisms that minimise tyre
heating by means of cooling.
The change in energy for the gas is derived from the convective term specified
by Equation 14.46 and by the friction between the inner tyre and the gas. The
movement of the inflating gas inside the tyre is not amenable to simple modelling.
For the purpose of the present analysis, some approximations are required. A few
key physical aspects are the following:
r The inflating gas moves with the same speed as the tyre walls.
r The gas is affected by centrifugal forces that press against the outer radius.
r The gas is lagging behind the walls elsewhere.
With a leap of faith we now approximate the relative motion of the gas with
respect to the tyre to a turbulent shear flow in a pipe of equivalent diameter and
length. In this pipe there is no relative speed between the walls and the gas. The set
of assumptions is the following:
r The tube’s diameter d is the average between the rim and the width.
1
r The Reynolds number of the internal flow is
ρg u1 d1
Rei = , (14.53)
μg
where u1 is the maximum rotational speed difference between the tyre and the
gas (slip speed). This is a fraction of the tyre’s rotational speed.
The friction coefficient f is calculated from the equation

1 d1 4.67(d1 /)
√ = 4 log + 2.28 − 4 log √ , (14.54)
cf Re c f + 1
where is the average roughness of the internal walls. Equation 14.54 is in implicit
form and must be solved iteratively. The average resistance between the gas and the
tyre is

1 π
R ρg (dw − di )(dw + di ) c f u21 (14.55)
2 4
and the energy dissipated in the time step dt is
E f Ru1 dt. (14.56)

380 380
360 360 ABORT
ABORT
Temperatures, K
Temperatures, K
340 340
NLG NLG
GO
GO
320 320
ABORT
ABORT
300 300
MLG MLG
GO GO
280 280
0 10 20 30 40 50 0 10 20 30 40 50
Time, s Time, s
(a) Tyres (b) Gas (air)
Figure 14.14. Estimated tyre and temperatures for accelerate-stop of airplane model Boeing
B777-300; standard day; 2 m/s headwind; airfield at 50 m above mean sea level.
The differential equations that describe the tyre and gas temperature are,
respectively:
dTt 1
= (dW − dEc − dEr − dEg ) , (14.57)
dt C pt
dTg dEg
= . (14.58)
dt C pg mg
These equations are integrated with the initial conditions
t = 0, Tt = T , Tg = T , Tr = T . (14.59)
Equation 14.57 and Equation 14.58 can be included in the solving system for take-off
and landing and solved simultaneously with the corresponding group of equations
(for example, § 9.2) by using a fourth-order Runge-Kutta integration.
14.5.1 Numerical Simulations

The integration of the tyre-heating model with the take-off equations is shown in
Figure 14.14 for a case of accelerate-go and decelerate-stop of a model Boeing
B777-300 aircraft. Figure 14.14a shows the temperature distribution for the tyres
and Figure 14.14b shows the corresponding gas temperature. The thermo-structural
loads are higher for the nose landing gear (NLG) due to the position of the CG.
These high loads on the NLG are not unusual. In any case, the temperatures decrease
uniformly for the accelerate-go event because of the cooling effect of the air around
the tyres. For the decelerate-stop event this cooling is absent, and the additional
heating is derived from the extended tyre-runway contact, as well as internal friction
with the gas and the effect of the mechanical brakes. Notice that this procedure is
fully integrated with the take-off equations, which are discussed in Chapter 9.
The validation of the model proposed cannot be done directly because the few
available data are not in a usable format. Some tyre-heating data for typical taxi
120
Bead
Tread shoulder
Tread centerline
Temperature rise, K
80
Figure 14.15. Estimation of tyre temper-

atures. Data adapted from Goodyear;
U = 48 km/h (∼29 kt); deformation =
0.32. Tyre dimensions and load condi-
tions are unknown. 40
0
0 5 10 15
Distance, km
operations are given by Cavage 17 for the Boeing B727. The results indicate that
the heating depends on the loading, the taxi speed and the type of tyre. Data are
available for taxi speeds up to 70 mph (110 km/h). A qualitative comparison can be
drawn with some data published by Goodyear (although the actual load conditions
and tyre specifications are not disclosed), Figure 14.15. The bead temperature is
always the highest. In this example, an airplane was forced to run on a long runway
at a specified speed, and the temperature was measured at selected points on the
tyres.
By way of comparison, Figure 14.16 shows some predicted tyre temperatures on
the main under-carriage units for a model Airbus A300-600 forced to run at various
speeds. The predicted temperature is an average within the tyre, coherently with the
150
X = 40,000 feet (12.192 km)
120
64 km/h
Temperature rise, K
50 km/h
90
Figure 14.16. Estimation of tyre tem-
peratures from taxiing; m = 150.0 tons;
headwind = −2 m/s (−3.9 kt); altitude = 32 km/h
60
50 m; xCG = 30% MAC; standard day.
30
0
0 30 60 90 120
Time from brake release, s
32
21
200 C
100 C
11
50 C
distances in [m]
0 23 46
Figure 14.17. Temperature levels for the A380-861 on the ground with idle engines (side
view).
lumped mass model. The position of the CG is important because it determines the
load distribution on the main and nose tyres and therefore can affect the heating of
both groups of tyres.
14.6 Jet Blast

If a jet can propel a Boeing 747 into the sky, imagine what it can do to you. It can flat-
ten buildings, lift ground vehicles up into the air and create other damage at ramps
and taxiways. A numerical study by Melber-Wilkending 18 illustrated the problem
of an Airbus A380, whose jet blast hit the corner of a hangar. The wind generated
around the structure creates its own weather on a length scale of at least 100 m.
Most airplane manufacturers provide thermal and velocity maps of jet blast with the
vehicle on the ground, with at least two operating conditions: idle and full thrust.
These conditions are important in ground operations for the safeguard of infra-
structure, protection of nearby vehicles and the safe operation of ground personnel.
Regulations are in place for safe ground operation 19 . An accurate determination of
jet-blast characteristics includes maps of both speeds and temperatures. In nearly
every FCOM of modern airplanes, there is a section containing these types of data
at low and high power settings. Determination of these maps can be done by a
combination of testing and numerical simulation. Numerical simulations are used
at increasing frequency and require the use of unsteady Reynolds-averaged Navier-
Stokes equations with primary and secondary flows (by-pass), as well as ground
plane and engine tilt and toe-in. Thermal maps can be determined with various
thermo-graphic techniques, including infrared thermography.
Figure 14.17, adapted from Airbus* , shows the temperature distribution from the
jet engines of the Airbus A380 in idle setting. The plumes are lifted from the ground
and can be considered as “free” jets. At higher thrust settings there is an interaction
of the plumes with the ground. Velocity contours indicate that the 56 km/h wind
(maximum safe jet blast) extends about 50 m behind the airplane with idle engines.
With the engines operating at breakaway thrust, the jet plume of this airplane extends
* Airbus, A380 Airplane Characteristics, Issue Nov. 2008, Airbus Customer Services, Blagnac, France.
Bibliography 419
to about 90 m. When the engines are at take-off thrust, the plume extends about 0.5
km behind and 50 m up in the sky.
There are no easy extrapolations that can be made from this or other cases due
to the complex interaction plumes between plumes and the ground at the relevant
engine settings. The scientific literature contains a number of semi-empirical rela-
tionships to predict the jet expansion and velocity distribution 20 . However, these
equations are only applicable when there is no interaction with boundaries. In prac-
tical terms, because the jet-blast area can be very large, some airports are equipped
with blast-deflection barriers. A recent analysis of experimental data at some major
airports was published by Slaboch 21 and aimed at providing guidelines for reducing
hazards on manoeuvring aircraft.
Summary
In this chapter we presented a number of performance problems related to thermo-
dynamics and structural mechanics. In the former class of problem, aircraft icing is
considered of utmost importance, having been the source of some prominent air-
craft accidents. There is a serious degradation of aerodynamic characteristics of the
wing, which can even be asymmetric. Thus, aircraft control becomes impossible.
Fuel-temperature problems arise both at high and low temperatures. We addressed
the problem of very cold fuel and its behaviour in flight. We developed a lumped-
mass model that is capable of predicting the fuel temperature change in flight. Fuel
freezing is unlikely, except in the most extreme circumstances, in combination with
other factors related to the fuel pipelines. In the category of structural problems, we
addressed tyre loads and the response of tyre to extreme loads, which may cause
abnormal heating. Also in this case we developed a lumped-mass model for tyre
loads and heating, which is fully integrated with the flight mechanics of the field
performance (taxi, take-off and landing).
Finally, we mentioned briefly the problem of jet blast. Jet exhausts at high power
can have damaging consequences on infrastructure and ground personnel.
Bibliography
[1] Kind RJ, Potapczuk MG, Feo A, Golia C, and Shah AD. Experimental and
computational simulation of in-flight icing phenomena. Progr. Aerospace Sci-
ences, 34:257–345, 1998.
[2] Lynch FT and Khodadoust A. Effects of ice accretions on aircraft aerodynam-
ics. Progress Aerospace Sciences, 37(8):669–767, 2001.
[3] Asselin M. Introduction to Aircraft Performance. AIAA Educational Series,
1997.
[4] Caldarelli G. ATR-72 accident in Taiwan. In SAE Aircraft & Engine Icing
International Conference, ICE 13, Sevilla, Spain, Oct. 2007.
[5] Goodger E and Vere E. Aviation Fuels Technology. Macmillan, London, 1985.
[6] Faith LE, Ackerman GH, and Henderson HT. Heat sink capability of a Jet-A
fuel: Heat transfer and coking studies. Technical Report CR-72951, NASA,
July 1971.
[7] Gray CN and Shayeson MW. Aircraft fuel heat sink utilization. Technical
Report TR-73-51, US Air Force Aero Propulsion Labs, Wright-Patterson AFB,
Ohio, July 1973.
[8] Peng DY and Robinson DB. A new two-constant equation of state. Industrial
& Eng. Chemistry: Fundamentals, 15:59–64, 1976.
[9] Huber ML and Yang JC. A thermodynamic analysis of fuel vapor characterist-
ics in an aircraft fuel tank ullage. Fire Safety Journal, 37:517–524, 2002.
[10] Incropera F and De Witt DP. Introduction to Heat Transfer. John Wiley, 1985.
Chapter 9.
[11] AAIB. Accident to Boeing B-777-236ER, G-YMMM at London Heathrow
Airport on 17 January 2008. Interim Report EW/C2008/01/01, Sept. 2008.
Aldershot, Hampshire, UK.
[12] Filippone A. Theoretical framework for the simulation of transport aircraft
flight. J. Aircraft, 47(5):1679–1696, 2010.
[13] ESDU. Vertical Deflection Characteristics of Aircraft Tyres. Data Item 86005.
ESDU International, London, May 1986.
[14] Lin YJ and Hwang SJ. Temperature prediction or rolling tires by com-
puter simulation. Mathematics and Computers in Simulation, 67:235–249,
2004.
[15] Seghir-Ouali S, Saury D, Harmand S, Phillipart P, and Laloy O. Convective heat
transfer inside a rotating cylinder with an axial flow. International J. Thermal
Sciences, 45:1066–1078, 2006.
[16] McCarty JL and Tanner JA. Temperature distribution in an aircraft tire at low
ground speeds. Technical Report TP-2195, NASA, 1983.
[17] Cavage WM. Heating comparison of radial and bias-ply tires on a B-727 air-
craft. Technical Report DOT/FAA/AR-TN97/50, US Department of Trans-
portation, Nov. 1997.
[18] Melber-Wilkending S. Aerodynamic analysis of jet-blast using CFD consider-
ing as example a hangar and an Airbus A380 configuration. In New Results in
Numerical and Experimental Fluid Mechanics, volume 92 of Notes on Numerical
Fluid Mechanics. Springer, 2006.
[19] Anon. Transport Canada Aeronautical Information Manual (TC AIM): AIR
Section 1.7: Jet and Propeller Blast Danger. Transport Canada, Oct. 2011 (con-
tinuously updated). TP 14371.
[20] Witze P. Centerline velocity decay of compressible free jets. AIAA Journal,
12(4):417–418, April 1974.
[21] Slaboch PE. An operational approach for the prediction of jet blast. In
AIAA Aerospace Sciences Conference, AIAA Paper 2012-1225, Nashville, TN,
Jan. 2012.

a, b = factors in Peng and Robinson equation of state, Equation 14.24;
factors in Nusselt number, Equation 14.44
b = width of tyre contact
Ac = area for convective heat transfer
btank = wing-tank span
CD = airplane-drag coefficient
cf = friction coefficient, Equation 14.54
CL = airplane-lift coefficient
Cp = heat-capacity coefficient
d = diameter
di = internal diameter of the tyre (rim), Figure 14.12
d1 = tube diameter; average tyre diameter
dg = depth of the groove

dxw = wall thickness
dw = tyre diameter
E = energy
f = frequency
f1 , f2 = generic function
F = normal load on tyre
Gr = Grashof number
hc = convection-heat-transfer coefficient
H = mechanical hysteresis
kc = thermal conductivity
lc = length of tyre-road contact
l f use = fuselage length
L = characteristic length
m = mass
M = Mach number
Nu = Nusselt number
ng = number of grooves
p = pressure
pr = tyre-rated pressure
Pr = Prandtl number
Q = heat transfer
R = rolling resistance; resistance between gas and tyre
R = gas constant
Re = Reynolds number
Ra = Rayleigh number
Rea = axial-flow Reynolds number
Rer = rotating-flow Reynolds number
S = contact surface
S = free fuel surface
t = time
T = temperature
Tr = reduced temperature in Peng-Robinson equation, Equation 14.26
u1 = maximum slip speed between tyre and gas
U = ground speed or air speed
Vc = critical molar volume, Equation 14.27
V = volume
Vm = molar volume
V1 = normalised volume
w = tyre width, Figure 14.12
wg = width of the groove
W = mechanical work
W1 = mechanical work in one full rotation
xi = mass fraction of component i
Greek Symbols
α = factor in Equation 14.25
= wall roughness
δt = vertical deflection of the tyre
γ = ratio between specific heats (air)
θ = sector angle for tyre-road contact
ρ = density
ρw = water density
τ = time constant
τ1 = thickness of the tread
τ2 = average thickness of the tyre sides
ω = acentric factor, Equation 14.28
ω = rotating speed of the wheels
˙
[.] = time derivative
[.]c = critical gas parameter or convection
[.] f = liquid fuel quantity, due to friction
[.]g = gas quantity
[.]i = internal
[.] L = with respect to reference “L”
[.]r = runway quantity; thermal quantity in Equation 14.43
[.]t = tyre quantity
[.]∗ = vapour fuel quantity; FDR data in Figure 14.10
15 Mission Analysis
Overview
Mission analysis refers to the complete set of assumptions regarding the operation
of an aircraft and the resulting weight and fuel planning. Several parametric con-
siderations can be added, including mission optimisation, environmental emissions
and various commercial or military trade-offs. This chapter illustrates how a detailed
mission analysis can be carried out. We start by defining mission scenarios (§ 15.1)
and the payload-range charts (§ 15.2). We then make a detailed qualitative and
quantitative mission analysis (§ 15.3), including the problems of mission range and
mission fuel (§ 15.4), the fuel-reserve policies (§ 15.5) and the short-range perform-
ance at MTOW (§ 15.6). In § 15.7 we discuss more advanced problems, such as service
with an intermediate stop (§ 15.7.1), tankering (§ 15.7.2) and the point-of-no-return
(§ 15.7.3). We proceed with the calculation of the direct operating costs (§ 15.8). We
show some case studies: an aircraft selection problem (§ 15.9), a fuel mission planning
for a transport aircraft (§ 15.10) and the effect of floats on the payload-range chart
of a propeller airplane adapted with floats for water operations (§ 15.11). We close
this chapter with an introduction to risk analysis in aircraft performance (§ 15.12).
KEY CONCEPTS: Mission Profiles, Flight Scheduling, Payload-Range Charts, Mission

Analysis, Reserve Fuel, Fuel Planning, Tankering, En-Route Stop, Direct Operating
Costs, Risk Analysis, Route Selection.
15.1 Mission Profiles

A mission profile is a scenario that is required to establish the weight, fuel, payload,
range, speed, flight altitude, loiter and any other operations that the aircraft must be
able to accomplish. The mission requirements are specific to the type of aircraft. For
high-performance aircraft they get fairly complicated and require some statistical
forecasting. The deterministic approach is not sufficient or even recommended.
Over the years many commercial aircraft operators have specialised in niche
markets that offer prices and services to selected customers. These niches include the
executive jet operators between major business centres, operators flying to particular
destinations (oil and gas fields), the all-inclusive tour operators to holiday resorts
423
424 Mission Analysis
32
32
28
Local Time at Departure

Inb
Local Time at Arrival

ou 28
nd
24 , M=
Date Ch 0.8
ange 2
20 24
16
20
12 M= 0.80
Outbound,
16
8
Figure 15.1. Scheduling of trans-Atlantic flight.
and the no-frills airlines, flying to minor and under-used airports. These operators
have different schedules and cost structures.
Let us start with long-range passenger operations, which are serviced by subsonic
commercial jets. The basic principle is that the airplane takes off from airport A and
flies to airport B along a recognised flight corridor; then it returns to A. The main
parameters of the mission planning are the distance between the airports, the flight
time, the down time at the airport for getting the airplane ready (also called time-on-
station), the flight speed, the local air traffic and the departure times at both ends.
Back at the airport of origin, the day is not over; the operator may wish to utilise
the airplane for another flight to the same destination or to another destination. The
key is the departure time and the minimisation of the curfew. Figure 15.1 shows a
typical schedule of such an airplane over a trans-Atlantic route from a major airport
in Europe to an airport on the East Coast of the United States.
Due to the time-zone effect, a late-morning departure from Europe arrives to
the United States in the mid-afternoon. An early-evening departure arrives back in
Europe in the early hours of the day after. Over a 24-hour period the airplane will
have done a return flight and worked about 14 to 16 hours. For a flight arriving late
in the evening, a return may not be possible until early morning on the next day.
This adds to the operational costs because of the need of maintaining the crew away
from the home port.
The schedule shown in Figure 15.2 shows a typical schedule between a UK
airport and mainland Europe. Due to the relatively short distance of 600 n-miles and
one time zone, it is possible to operate two return services and one outbound service,
whilst maintaining the night-time curfew (shaded area). The final outbound flight
must remain at the destination airport till the following morning. This aircraft will
be the first to make an inbound flight. To run efficiently, this service would require
two aircraft, one at each airport. If this is not possible, then the last outbound flight
must be cancelled from the schedule.
15.1 Mission Profiles 425
24 24
Local time, Destination Airport

20 20
Local time, Origin Airport

16 16
2
12 12
1
8 8
4 4
NO Night Flights
0 0
600 n-miles, +1 Time Zone

Figure 15.2. Scheduling of short–medium flight.
Another key factor in the utilisation of the airplane is the turn-around time,
that is, the time required to prepare the airplane for the next flight. Considerable
logistics are required to ensure all services to passengers, cargo and airplane within
a target time frame. Up to six different categories of services are involved: passenger
boarding/disembarking, cargo loading/unloading, refuelling, cleaning, catering, and
ground handling. The time needed to get the aircraft ready for the next inter-
continental flight may require up to 3 hours. However, Airbus reports that the
turn-around time of the A380 is 90 minutes with standard servicing on two decks
and 120 minutes with servicing from the main deck* .
15.1.1 Operational Parameters

Block time is the time required to travel from the departure gate to the destination
gate. If the block time of each flight is added, at the end of one year the airplane will
have a total number of block hours, which airline operators may want to maximise.
There are three parts in the block time: taxi out, airborne time and taxi in. An airplane
flying a day-time shorter route, and returning in the mid-afternoon, should be able
to complete another return flight to the same destination or otherwise. For an airline
company operating several airplanes, scheduling and optimal operation of the fleet
is a complex problem. Events such as bad weather can lead to dozens of airplanes
and flight crews out of position for several days. Scheduling and operation of aircraft
is a subject for operations research and is addressed by specialised publications, for
example Gang Yu 1 , who deals with demand forecasting, network design, route plan-
ning, airline schedule planning, irregular operations, integrated scheduling, airport
traffic simulation and control, and more. Coy 2 presents a method for the statistical
* Airbus, A380 Airplane Characteristics, Rev. Oct. 01/09 (2009), Blagnac Cedex, France.
prediction of the block time, which, unlike the deterministic scenario shown in Fig-
ure 15.1, relies on a number of stochastic parameters, including congestion, weather
conditions and delays accumulated on an earlier leg.
In the following discussion, we present only a deterministic scenario and focus
on performance* . For a transport-type airplane, there are two basic types of mission
analysis:
r Range Planning: For given atmospheric conditions, given the payload and the
fuel load, calculate the distance that can be flown. The problem is to establish the
functional relationship between the flight plan and the operational parameters:
X = f (Wp , W f , Atm). (15.1)
r Fuel Planning: For given atmospheric conditions, given the required distance
X and the payload weight Wp , calculate the mission fuel W f and the weight
breakdown. The problem is to establish the functional relationship between the
fuel planning and the key operational parameters:
W f = f (X, Wp , Atm). (15.2)
The operator Atm is a vector containing all of the atmospheric conditions encountered
during the flight. For convenience, these conditions are split between 1) take-
off and climb-out, 2) cruise, and 3) final approach and landing. More specifically,
the parameters are atmospheric temperatures, wind speed, wind direction and rel-
ative humidity. The latter parameter is only relevant for noise calculations (see
Chapters 16 to 18).
15.2 Range-Payload Chart

The payload-range chart belongs to the range flight-performance analysis (Equa-
tion 15.1) because the distance that can be flown is unknown. The flight distance
depends not only on the amount of fuel but also on the weight of the payload. In
most cases the combination of maximum payload and maximum fuel load exceeds the
MTOW. If we know how to calculate the flight range, then it is possible to construct a
chart showing how some of the aircraft weights are related to the range. Because dif-
ferent cruise techniques are available, we could construct different weight-payload
diagrams corresponding to each flight program. Furthermore, there are effects of
atmospheric conditions, climb and descent techniques, and reserve fuel policy that
can change the weight-payload performance considerably. Unless all of these con-
ditions are specified, it is not possible to compare the range-payload performance of
two aircraft.
There are several ways in which the payload-range performance can be analysed.
The first approach is shown in Figure 15.3, which displays the range-payload chart
for three commercial subsonic jet aircraft of the Airbus family. The manufacturer
generally specifies the range at maximum passenger load (including baggage) and
some value of the bulk payload; the latter item is important for most commercial
airlines. However, passenger-carrying airplanes seldom operate at MTOW.
* We refer to “mass” and “weight” interchangeably, although they are physically different. The
aviation jargon refers to weights rather than mass; our computer models work with “mass”.
60
A-300-600
295 pax + 28 T cargo (A-340) A-330-200
A-340-300
253 pax + 24 T cargo
MTOW = 276.5
Payload, ton 40
MTOW = 233.0
295 pax + baggage

273 pax + baggage
266 pax + baggage
20
MTOW = 170.5
0
0 2,000 4,000 6,000 8,000 10,000
Range, n-miles
Figure 15.3. Maximum payload range for Airbus airplanes.
Another way of showing the payload-range performance is presented in Fig-

ure 15.4. The data have been elaborated from Boeing 3 and refer to the B747-400.
This diagram is considerably different and more detailed than Figure 15.3. The
oblique lines denote constant-BRGW values. A trade between an increase in range
and the corresponding decrease in payload shall be made at constant brake-release
gross weight:

d ZFW
.
dX BRGW
3
Range, 10 km
6 8 10 12 14
600
260
MZFW, 747-400 combi. 256,279 kg
550 MZFW, 747-400, 248,073 kg

250
BR75 l
8
GWb (
85
Weight, 10 kg
Weight, 10 lb
75
65
0
55
0(
x 97)
0(
3
(3 8
3
10
0(
34
3
29
3
8
24
240
0)
5)
)
9)
500
420 pax + baggage
394,000 lb, 178.76 t
Fue
230
l
450
220
400
0 2 4 6 8
3
Range, 10 n-miles
Figure 15.4. Payload-range chart for the Boeing B-747-400 and -400 Combi (CF6-80C2B1F
engines); standard day, M = 0.85, cruise-climb profile; FAR international reserves.
90
120 60
747-ERF
113ton payload
Seoul
V’ver
150 30
Bejing
Dallas
NYC Moscow
Miami Mumbai
180 London 0
0 0.2 0.4
210 330
727-200F J’sburg
113ton payload 747-400F
240 300 113ton payload
270
Figure 15.5. Range of Boeing 747-family of aircraft. Data elaborated from Boeing.
The ZFW is the weight of the aircraft except the loaded fuel; it includes the bulk pay-
load, the crew and all of the operational items. At a constant BRGW, the range must
decrease and, therefore, this derivative must be negative; its value is the sensitivity
of the range to changes in ZFW at constant BRGW.
The key parametric on the flight-range effect is the wind. Figure 15.5 shows
some destinations that can be reached with the Boeing 747. The data have been
extrapolated from the manufacturer and include typical mission rules, 85% annual
winds and air-traffic allowances. This type of chart is more useful from a marketing
point of view than an engineering point of view because it gives very few details on
the bulk payload, the passenger load, the flight procedures, the availability of flight
corridors and so on.
Another example of a payload-range chart is shown in Figure 15.6 that refers
to the business jet airplane Gulfstream G550† . The data are presented in the range-
GTOW plane at selected values of the payload. The flight time corresponding to a
flight distance is useful in general flight planning (such as in Figure 15.1). The fuel
load has to be inferred from the data. For example, for a specified payload Wp and
required range X, the diagram returns a weight W = GTOW. The corresponding
fuel weight W f is found from the difference:
W f = W − We − Wp . (15.3)
The operating empty weight We is given in the airplane documents (FCOM, type
certificate).
† Gulfstream G550. Flight Crew Operating Manual (2005).
Flight time, hours

0 3 6 9 12 15
90
Max Payload 40
3,000 lb (1,360 kg)
80 1,600 lb (725 kg)

GTOW, 10 lb
GTOW, ton
35
3
70
30
60
No payload 25
50
0 2,000 4,000 6,000
Range, n-miles
Figure 15.6. Payload-range of Gulfstream G550. Data elaborated from Gulfstream Aero-
space; M = 0.80; 200 n-miles diversion fuel, standard day, no wind.
From Equation 15.3 the fuel fraction ξ = W f /W (fuel weight over gross weight)
can be written as
We Wp
ξ =1− − . (15.4)
W W
Because the empty weight is fixed, Equation 15.4 is a linear relationship between
payload fraction and fuel fraction, with the gross weight W being a parameter.
15.2.1 Case Study: Range Sensitivity Analysis

For a flight range of 5,000 n-miles (point X in Figure 15.7), the difference in GTOW
between an airplane loaded to its maximum payload weight (point A) and an airplane
loaded with 1,360 kg (3,000 lb) of payload (point B) is 2,052 kg (4,520 lb). The certified
maximum payload is 2,812 kg (6,200 lb). This means that the difference in payload
between the two cases is 760 kg (1,680 lb):
GTOW = 2,052 kg, Wp = 760 kg, W f = 2,052 − 760 = 1,292 kg.
This means that an increase of the payload by 760 kg requires an additional ∼1,300
kg of fuel to cover the same distance, or

dW f
= 1.7. (15.5)
dWp X
This is by any account a modest value, especially when compared to the overall
operating costs of the airplane. Alternatively, consider the case of two flights at the
same payload (points B and C in Figure 15.7).
X = 719 n-miles, GTOW = 0, W f = −Wp = 760 kg.
Flight time, hours

0 3 6 9 12 15
90
40
B
Δ GTOW
C A
80
GTOW, 10 lb
35
GTOW, ton
Max Payload
3
3,000 lb (1,360 kg)
70
30
60
25
X
50
0 2,000 4,000 6,000
Range, n-miles
Figure 15.7. Analysis of payload-range sensitivity of Gulfstream G550.
Thus, the decrease in payload is offset by an increase in fuel, which leads to the
following equivalences:
d ZFW = d (We + Wp ) = dWp = −dW f , (15.6)

d ZFW dW f
=− = −1.057 kg/n-mile. (15.7)
dX GT OW dX
15.2.2 Case Study: Payload-Range of the ATR72-500

We used the ATR72-500 powered by PW127M turboprop engines and Hamilton
F568-1 propellers. This configuration has been discussed in other sections of the
book. We calculated the payload-range performance for two different weights and
compared our results to the “official” performance quoted by the manufacturer.
These results are shown in Figure 15.8. The calculation consists of four key points:
TOW-limited range, maximum-payload range, maximum-fuel range and ferry range.
15.2.3 Calculation of the Payload-Range Chart

We now proceed to a practical method for the calculation of the payload-range
chart. To avoid computer-intensive calculations, some simplifications are required.
We need to define three critical operational points: 1) the maximum-payload range
Xp1 ; 2) the maximum-fuel range Xp2 ; and 3) the ferry range Xp3 . The operation at
ranges between Xp1 and Xp2 requires to consider the aircraft at constant BRGW.
The difference is only marginal, but the additional computational burden would be
considerable.
6
Payload, tons
Calculated
ATR Data, basic
ATR Data
0
0 500 1,000 1,500 2,000
Range, n-miles
Figure 15.8. Payload-range performance of the ATR72-500.
1. Maximum-Payload Range: Assume that there are no passengers; assume a bulk

payload equal to the MSP. Then use a bisection method to calculate the fuel
planning for a tentative range X with the constraint GTOW = MTOW. In other
words, if the fuel planning corresponding to the tentative range X leads to a
GTOW < MTOW, then increase the range; otherwise decrease the range. The
maximum-payload range is bracketed by X1 = Xdesign /2 and X2 = Xdesign . This
procedure converges to Xp1 in 7 to 10 iterations, not without difficulties (it is
possible that the aircraft is unable to climb to the estimated ICA, or that the
GTOW corresponding to the range X is considerably larger than the MTOW).
2. Maximum-Fuel Range: Assume GTOW = MTOW. In this case, as the range
increases, payload has to be traded for fuel. If Wmf w is maximum fuel capacity,
the bulk payload must be
Wp = GTOW − We − Wmf w .
It is not useful to add passenger weight at this time because passengers require
consideration of on-board services. This is a fixed quantity for the maximum-
fuel range Xp2 . After bracketing the range Xp1 < Xp2 < Xdesign , use the bisection
method to calculate the maximum-fuel range. For a given iteration, if GTOW
< MTOW, increase the tentative range, otherwise decrease the range. The
procedure should converge to Xp2 in less than 10 iterations unless numerical
difficulties intervene (such as at the previous point).
3. Ferry Range: There is no payload in this case. Thus, the take-off weight must be
GTOW = We + Wmf w .
After bracketing the solution with Xp2 < Xp3 < 1.1Xdesign , use the bisection
method to calculate ferry range. For a tentative range if the mission fuel is lower
than the fuel capacity, the range must increase; otherwise it must decrease.
4. Maximum-Passenger Range: Along with the three points specified previously,
a passenger aircraft must show the design range at maximum passenger load.
Figure 15.9. Flowchart of a mission-fuel analysis.
This range is usually between Xp1 and Xp2 . For this point, the payload weight
is the sum of all passengers, their baggage, average weight of service items and
average number of crew members (and associated weight). After bracketing the
solution with Xp1 < X < Xp2 , the solution procedure follows the same steps as
in the previous scenarios.
The calculation of a payload-range must rely on a number of assumptions, some

of which can be arbitrary. The fuel reserve (§ 15.5) and the atmospheric conditions
must be specified.
15.3 Mission Analysis

A realistic flight planning requires an optimisation of the route (track, vertical pro-
file), Mach number and altitude, with the constraints imposed from the air traffic
control and the aviation regulators. The flight planning must be based on real-
time data, such as winds, temperatures, payload and other parameters. Therefore, a
detailed analysis involves several parameters. A flowchart of the computer program
is shown in Figure 15.9.
The boxes with an asterisk refer to the sub-models: the airplane model is
described in Chapter 2; the aircraft noise model is discussed in Chapters 16 and 17
(see also flowchart in Figure 16.3). The fuel contingencies are discussed in § 15.10.
An interface is required to associate a flight trajectory to a noise calculation (§ 18.3).
15.3 Mission Analysis 433
required mission diversion
3
FCA
ICA
altitude
2 4
7
9
LTO < 3,000 ft
1 climb to ICA stepped cruise enroute descent

5
6
Airfield
Airfield
4
Figure 15.10. Standard mission profile of passenger aircraft, from Ref. .
A typical mission profile for a commercial jet is shown in Figure 15.10. It shows a
climb to cruise altitude, a cruise and a descent. It also includes the case of an aborted
landing and an extension of the flight at a lower altitude. Diagrams of this nature
can be made more precise by considering a quantitative scale for the distance and
altitude, and for each flight segment. More specifically, there are two sub-problems:
r Mission range for given fuel and payload without constraint on BRGW, § 15.3.1.
r Mission fuel for given range and payload with constraint on BRGW, § 15.4.
The weight constraint can be specified on the BRGW or the GTOW. The second
case is only constrained by the fact that GTOW ≤ MTOW.
Calculation of the exact fuel required for a mission requires consideration of

several objectives, some of which can be conflicting. Safety comes first. After that, the
airline must consider its profitability via its own accounting system and punctuality
of the service. Other items include actual payload, weather phenomena, operational
constraints, route and altitude constraints. The cost of fuel is in most cases the largest
cost item of a commercial airline. Price volatility makes planning difficult and risky.
Fuel planning itself requires three phases:
1. Pre-flight: It refers to measures in terms of route planning, ground operations,

usage of APU versus ground services, turn-around processes and other adminis-
trative tasks.
2. Flight: It refers to all of the flight procedures to be programmed in the flight com-
puter, such as planned route, climb and descent schedules, flight level selection
and so on.
3. Post-flight: It includes the collection and analysis of flight data that are used to
implement strategies for reduction of fuel consumption. These measures include
monitoring of the engines, the airframe and the systems from the FDR.
Calculation of the mission fuel (point 2) requires adding up the fuel required
by each segment of the flight. For a passenger operation, the mission requirement
is straightforward. The fuel for the APU must be considered as well because for a
large aircraft, this means a considerable amount of extra weight. The APU fuel flow
can be of the order of 150 kg/h; for a 7-hour intercontinental flight of a wide-body
commercial airliner, this means 1,050 kg, or the weight of 12 paying passengers.
Our numerical procedure is based on a combination of optimal fuel planning
and optimal flight trajectories. Each segment is treated with some degree of inde-
pendence. For example, the initial cruise condition is established on the basis of
the optimal SAR with the estimated AUW at the top of climb. Likewise, the des-
cent trajectory is based on initial conditions established from the cruise. Alternative
methods have been proposed, including the use of multi-variate gradient optimisa-
tion 5 . These methods, aided by the use of constraints, offer some advantages in
global optimisation.
15.3.1 Mission Range for Given Fuel and Payload

The problem is to establish how far the aircraft can fly, with the mandatory fuel
reserves and under specified atmospheric conditions. The numerical procedure is
more elaborate than the calculation of a cruise range because of the implicit relation-
ship between airplane trajectories (climb, cruise, descent) and BRGW. We propose
the following algorithm:
1. Calculate the fuel from ramp up to the ICA. This operation includes taxi-out,
take-off, climb to ICA and APU fuel to that point. Call this fuel W fIC A; xc is the
distance to ICA.
2. Calculate the extended all-out fuel W fext to take into account holding and diver-
sion procedures. This sub-task is fully specified in § 15.5.
3. Using the procedures indicated in Chapter 11, calculate the descent fuel W fd and
distance xd from the top of descent (final cruise altitude), with an initial weight
W = BRGW − (W fext − W fres ), (15.8)

% &' (
fuel remaining
where W fres is the reserve fuel, which is analysed separately in § 15.5. The weight
at the top of descent is equal to the BRGW minus the fuel remaining at the
end of the service; this fuel includes the contribution from the diversion and
the reserve. If this is the first iteration, start the descent from the ICA (the
FCA is indeterminate); otherwise, estimate the FCA from the density altitude.
In the latter case, the FCA is found from the condition that W/ρ is a cons-
tant, or
σend = σstart (Wstart − W fcruise ) . (15.9)
The step from σend to the FCA requires the inverse solution of the atmospheric
model. The FCA is then adjusted to the nearest possible flight level (§ 12.7). In
any case, add the APU fuel relative to this flight segment.
4. If this is the first iteration, neglect the APU and estimate the fuel available for
cruise from
W fcruise = W fusable − (W fIC A + W fd + W fext ) . (15.10)

In any other case, add the APU fuel for the duration of the flight.
5. If this is the first iteration, estimate the cruise range with

U L W fcruise
xcruise log 1 + . (15.11)
fj D W f − W fc
In any other case, calculate the cruise fuel according to the procedures explained
in Chapter 12 (stepped-climb).
6. The estimated mission range is
X1 = xc + xd + xcruise . (15.12)
7. The calculations are repeated from point 2. The stopping criterion is based on
the change of X1 between iterations; when the updated mission range does not
change by any meaningful amount (for example, 1 n-mile), the procedure has
converged.
Note that the convergence analysis is done on the mission range. An important
sub-case is when
We + W f + Wp = MRW. (15.13)
When Wp is maximum and W f is the maximum allowable fuel weight compatible with
the MRW, the corresponding range is the maximum-payload range. When the fuel
weight is maximum and the payload is the maximum allowable weight compatible
with the MRW we have the maximum-fuel range. Finally, if Wp = 0, the ramp weight
is most likely below the MRW. The corresponding range is called maximum-ferry
range.
15.4 Mission Fuel for Given Range and Payload

The problem is to establish the amount of fuel to load onto the aircraft in order
to transport a specified payload over a specified distance Xr eq , under specified (or
forecast) atmospheric conditions. At the end of the process, we will have a detailed
weight breakdown. There are two important steps: 1) make a first estimate of the
mission fuel, and 2) improve the first estimate with a suitable numerical method.
15.4.1 Mission-Fuel Prediction

We define the computational steps required in a procedure called MissionFuel(..),
which returns the fuel required for a given range, payload, atmospheric conditions
and other external factors. The computational procedure is as follows:
1. Estimate the ramp mass (or weight or the BRGW) by making a guess of the fuel
required by the mission, W∗f . This weight is
W We + Wp + W∗f . (15.14)
2. Calculate the fuel from the ramp (or gate) up to the ICA. This operation includes
taxi-out, take-off, climb to ICA and APU fuel to that point. Call this fuel W fIC A;
xc is the distance to ICA.
3. Calculate the extended all-out fuel W fext to take into account holding and diver-
sion procedures. This sub-task is fully specified in § 15.5.
4. Using the procedures indicated in Chapter 11, calculate the descent fuel W fd
and distance xd from the top of descent. Follow the same procedure described
at point 3 on page 434.
5. Calculate the cruise distance from
xcruise = Xr eq − xc − xd . (15.15)
6. Calculate the cruise fuel required to fly the distance xcruise from the conditions
at ICA (see point 2). This is done with step climbs, following the procedure
discussed in § 12.7, by integration of the SAR. Add the APU fuel for this
segment.
7. Calculate the ramp weight. This is derived from the sum of all segment fuels.
With reference to point 1 in this procedure, the initial fuel estimate is either
provided externally or is specified by the following equation:

Xr eq
Wf c cus f , (15.16)
Xdesign
where cus f < 1 denotes the amount of usable fuel as a portion of the full tanks; the
coefficient c = 0.90 to 0.95 is a factor that improves the prediction and is found from
extensive numerical analysis.
15.4.2 Mission-Fuel Iterations

There are at least three different methods for calculating iteratively the mission fuel:
a) Method based on updating the initial guess until convergence.

b) Method based on updating the latest guess with an under-relaxation.
c) Method based on a predictor-corrector of the initial guess.
These methods are listed from the less robust to the most robust. In some
circumstances all of these methods converge, but there is no guarantee that they
will always converge. The difficulty increases with the increasing required range and
with the size of the airplane because of the increased non-linearity between the fuel
weight and the mission fuel.
A) MISSION-FUEL UPDATE. Compare the weight Wi calculated at iteration i with the

earlier estimate Wi−1 . Now use the updated ramp weight to recalculate the mission
fuel. The residual is defined as
Wi
E=1− . (15.17)
Wi+1
There is no guarantee that this procedure converges or that it converges to the
correct weight. It depends both on how good is the first estimate and on the required
range; it gets worse as the required range is close to the design range. In the latter
case, it is possible than on the second iteration, the ramp weight exceeds the MRW
and thence the climb becomes sluggish (not enough excess power is available), the
0.06
0.05
Weight Update
Residual 0.04
0.03 Under-relaxation
0.02
0.01
0
2 4 6 8 10
Iteration count
Figure 15.11. Error monitoring, Equation 15.17, in iterative mission-fuel analysis.
best ICA cannot be reached and, ultimately, the computations fail. Therefore, it may
be useful to add a constraint so that in no case the ramp weight exceeds the MRW.
B) MISSION-FUEL UPDATE WITH UNDER-RELAXATION. A method that partially over-

comes this convergence problem consists in using an under-relaxation. If Wi is the
initial weight estimate, and Wi+1 is the new weight estimate, then the iteration i + 2
is started with
1
Wi+2 = (Wi + Wi+1 ) . (15.18)
2
This method does converge but it is considerably slower. A comparison between the
two methods is shown in Figure 15.11. Neither option is satisfactory.
C) PREDICTOR-CORRECTOR METHOD. Given an estimate W∗f of the mission fuel, a

mission calculation should predict that the mission fuel is W f = W∗f . However, if this
were to be the case, it would be a lucky outcome. In fact, we have either W f > W∗f
or W f < W∗f . The difference E = W f − W∗f is the error (or residual). The problem is
how to improve this prediction. The algorithm is the following:
r Estimate the mission fuel W∗ and call the function MissionFuel(..), which
f
returns the fuel W f . This is the first iteration, i = 1.
r If W < W∗ , then the initial guess was an under-estimate. Therefore, we perform
f f
a new mission calculation with MissionFuel(..) and an increased fuel
W f = (1 + )W∗f , (15.19)
where denotes a small fraction of W∗f . Typical values are = 0.02 to 0.04,
although this depends on how good the first estimate W∗f is.
r If W > W∗ , then the initial guess was an over-estimate. Therefore, we perform

f f
a new mission calculation with MissionFuel(..) and a decreased fuel
W f = (1 − )W∗f . (15.20)
r At the end of the second iteration, the error should have decreased because we
moved the prediction W f toward its correct value.
r We keep correcting the mission fuel W in either direction (Equation 15.19 or
f
Equation 15.20) until we move from a positive residual Ei to a negative residual
Ei+1 (or vice versa). This event can be verified by the condition Ei Ei+1 < 0.
r The correct fuel W is then found by linear interpolation between E and E .
f i i+1
r A final iteration is carried out to calculate the fuel for each flight segment and
the weight breakdown.
For the sake of discussion, assume for example that the initial guess W∗f is an
under-estimate of the solution. Use the underscore “p” to denote the “predicted”
fuel, W f p . The algorithm is the following:
W f pi = W∗f + W∗f i, → W fi , Ei = W fi − W f pi > 0 (15.21)
W f pi+1 = W∗f + W∗f (i + 1), → W fi+1 , Ei+1 = W fi+1 − W f pi+1 < 0 (15.22)
W f pi+2 = Interpolation(Ei , Ei+1 , W f pi , W f pi+1 ). (15.23)

Equations 15.21 and 15.22 give a prediction of the mission fuel at iteration i, i + 1,
respectively. Equation 15.23 is the correction and the solution of the problem; the
corresponding residual is Ei+2 0. Note that the residual is calculated between the
fuel W fi returned by the procedure MissionFuel(..) and the predicted fuel W f pi .
The minimum number of iterations required for convergence is three.
Figure 15.12 shows the convergence of this method for two values of the required
range. This case refers to a Boeing B747-400 with CF6 engines on a standard day, with
full passenger load. In graph 15.12a convergence is achieved after four iterations.
At the second iteration the residual moves below zero; therefore, the interpolation
Equation 15.23 returns a corrected mission fuel. Due to the non-linearity around the
solution, another iteration is required. In graph 15.12b convergence is achieved in
three iterations.
15.5 Reserve Fuel

A fuel reserve is required by all of the regulators (FAA, ICAO, AEA, and so on).
This reserve depends also on the choice of alternate airport and company policies.
For example, the Association of European Airlines (AEA) specifies a 200 n-miles
(370 km) diversion flight for short- and medium-range aircraft and 250 n-miles (463
km) for long-range aircraft. In addition, a 30-minute holding at 1,500 feet (457 m)
altitude and a 5% mission fuel reserve for contingency are required. For domestic
flights in the United States it is specified 130 n-miles (∼240 km) diversion and 30
minutes holding at 1,500 feet. However, a number of alternatives are allowed, some
of which are subject to prior approval. These include a 20-minute extension of the
0.04 4 0.04 4
1
0.02 2 0.02 2
Residualf
Residualf
Error, %
Error, %
0 34 0 0 3 0
-0.02 2 -2 -0.02 -2
-0.04 -4 -0.04 -4
71.5 72 72.5 73 73.5 84.5 85 85.5 86
Mission fuel, ton Mission fuel, ton
(a) X = 6,000 km (3,240 n-miles) (b) X = 7,000 km (3,780 n-miles)
Figure 15.12. Predictor-corrector analysis of mission fuel for a Boeing B747-400-CF6; stand-
ard day, no wind.
trip or a 3% trip fuel if an alternate airport can be found en route. The ICAO
International* rules prescribe that the reserve fuel should take the aircraft to the
alternate airport specified in the flight plan and then hold for 30 minutes at 1,500
feet before landing at standard atmospheric conditions. Upon landing, the remaining
fuel must be a minimum 3% of the trip fuel.
From an operational point of view, the range may be limited by the absence of
diversion airports over a predefined route. The ICAO has a particular rule (ETOPS)
permitting twin-engine aircraft to fly longer routes (previously off-limits) that have
no diversion airports within a 60- or 120-minute flight. This rule allows several
modern twin-engine aircraft to fly over the oceans and in remote parts of the world.
A discussion of ETOPS performance with OEI is available in Martinez-Val and
Perez 6 .
HOLDING STACKS. When holding is required at the destination airport, the aircraft
is placed on a racetrack pattern that is made of two straight legs and two U-turns
(§ 13.2.4). The key aspect of the holding pattern is maximum endurance rather
than maximum range. The maximum endurance is achieved at minimum fuel flow
or maximum glide ratio. However, at some airports the aircraft may be subject to
holding at a specified air speed (or Mach number) that is sub-optimal. For the model
Airbus A300-600, the holding speed is 210 kt with a partial slat extension and 240
kt in clean configuration. An alternative is to place the aircraft on a linear holding
for part of the time (if the pilots are advised in time). This solution extends the
cruise time and reduces the racetrack holding; potentially, it can save a considerable
amount of fuel. For example, a 15-minute linear holding of the reference aircraft at
h = 35,000 feet and M = 0.80 can save ∼100 kg of fuel.
Optimisation of the holding-pattern performance is critical for short-segment
flights because it is not unusual to have a holding time of the same order as the cruise
* ICAO Annex 6: §4.3.6.3.

time. In the following discussion we fix both the holding Mach number and holding
altitude. For additional theoretical aspects of holding optimisation, see Sachs 7 .
NUMERICAL METHOD. We carry out the analysis in two steps. First, we find the contin-
gency range. This includes the specified range and the additional range as prescribed
by the aviation policies. Then we calculate the fuel corresponding to this range.
r The diversion distance X flown with the maximum landing weight MLW yields
div
an increase in mission range estimated by the equation
Wmlw
Xdiv = (c R)div , (15.24)
Wto
where the coefficient cdiv accounts for all of the factors in the diversion flight
that are sub-optimal (lower speed, altitude, engine efficiency); R (or Rdiv ) is the
diversion distance.
r The holding time (or loiter) flown at the maximum landing weight MLW yields
an increase in mission range estimated by the equation
Wmlw
Xhold = (cUt)hold , (15.25)
Wto
where the coefficient chold accounts for loss of efficiency as in the diversion flight.
The holding speed is generally half of the cruising speed.
r The extended duration of the flight at the cruise speed for a time t yields an
exd
increase in mission range estimated by the equation
X = Utexd . (15.26)
r The contingency fuel is a percentage of the mission fuel m /m = 0.05–0.10,
fres f
depending on actual policies.
The equivalent all-out range is then the sum of all contributions

) *W
m fres mlw
Xout = Xr eq 1 + + (c R)div + (cUt)hold + Utexd , (15.27)
mf Wto
where Xr eq is the mission range required. The exact calculation of this parameter is
essential. Torenbeek 8 proposed some practical values for the coefficients in Equa-
tion 15.27:
cdiv = chold 1.1 + 0.5 η M , (15.28)
where η M is the log-derivative of the propulsive efficiency with respect to the Mach
number. For a modern high by-pass turbofan engine this quantity is estimated at
η M 0.225 for M = 0.8; it increases with the decreasing Mach number, so that
η M 0.325 at M = 0.4. Equation 15.27 can be further simplified if the reserve fuel
is used for extension of the range at the cruise conditions.
An alternative method consists in adding separately the contributions for hold
and diversion. Clearly, a number of parameters are required for the solution of
this problem: the hold time, altitude and speed; the diversion distance, altitude and
speed. The fuel required from these two contingency segments will be added to the
mission fuel.
A B C
Figure 15.13. Fuel-reserve planning with redispatch.
Finally, we need to consider the unusable fuel, that is, the fuel that cannot
be pumped into the engines. Although this is a small portion of the fuel capacity
(typically 0.4 to 0.8% of the capacity), it is a constant amount of fuel that adds to the
operating empty weight.
15.5.1 Redispatch Procedure

The reduction of the reserve fuel relies on the existence of an alternate airport, as
shown in Figure 15.13. The graph indicates the departure point (origin), the alternate
airport (initial destination) and the final destination. The graph shows also a waypoint
(redispatch). The redispatch procedure is regulated by the relevant authorities* .
The reserve fuel must be sufficient to take the airplane from the origin to the
initial destination B. In normal circumstances, having passed the point B, the airplane
continues its flight; upon reaching the waypoint a decision must be made whether to
divert to the initial destination or to continue to the final destination; this is done on
the basis of a comparison between the predicted fuel burn and the available fuel. In
this scenario, the reserve fuel required is limited to the flight segment AB. By the
time the airplane has reached the initial destination, none of the reserve fuel is used,
unless the atmospheric conditions turn out to be worse than forecast.
Now consider a more complex network of alternatives, as shown in Figure 15.14,
which indicates a refuelling station, an alternative airport beyond the final destination
and a second alternate after refuelling.
It is possible to show that the minimum mission fuel with the redispatch proced-
ure is
m f = m ftaxi,A + m f AC + 0.1m fBC + m fC D + m fhold,D + · · · (15.29)

%&'( % &' ( % &' (
trip reserve altern.
With the standard procedure, the mission fuel is
m f = m ftaxi,A + m f AC + 0.1m f AC + m fC D + m fhold,D + · · · (15.30)

%&'( % &' ( % &' (
trip reserve altern.
The difference between the two procedures is in the reserve fuel. If the redispatch
point B is chosen appropriately, a considerable amount of reserve fuel can be saved.
Redispatch is not considered in the parametric fuel-reserve analysis carried out in
§ 15.10.
* See, for example, FAR § 121.631: Original dispatch or flight release.

Destination Alternate
D
Redispatch
A B
Origin C
Final Destination
E F
Refueling 2nd Alternate
Figure 15.14. Fuel-reserve planning with redispatch for a network of alternatives.
15.6 Take-Off Weight Limited by MLW

Flight over short distances requires a limitation of the TOW due to the relatively
low maximum landing weight (MLW). The problem is the following: calculate the
distance that the airplane must fly if it is loaded with the maximum payload, takes off
with the maximum take-off weight (MTOW) and lands with the MLW. This range
is calculated from the following numerical procedure:
r The aircraft starts with MRW, a payload W = MSP, and whatever fuel W
msp f
can be allocated, so that
GRW = We + Wmsp + W f . (15.31)

r The fuel required to taxi out, take off and climb to a cruise altitude (to be
determined) is calculated and added. By the time the aircraft has reached the
“cruise” altitude, it will have burned a fuel W1 and travelled a distance x1 .
r Now consider the descent phase. Estimate the weight at the top of descent from
Wd Wmlw (1 + r ), (15.32)
where r denotes the fuel reserve, as percent of the initial fuel. Calculate the
descent fuel W3 and the en-route descent distance x3 .
r Finally, consider the cruise range x (unknown). The aircraft weight at the start
2
and end of the cruise should be
Wstart = GRW − W1 , Wend = Wstart − W fc , (15.33)
where the cruise fuel W fc is
W fc = Wmtow − Wmlw − W1 . (15.34)

r Estimate the cruise range from the range equation

U L Wend
x2 = log . (15.35)
g fj D Wstart
r The MLW-limited range is the following:
xmlw x1 + x2 + x3 . (15.36)
The range estimate given by Equation 15.36 can be improved by further itera-
tions, but the result does not change a great deal.
As a practical example, consider the Airbus A-380-861 with GP-7270 engines.
This aircraft has the following characteristics: MSP = 90.7 tons; OEW = 270.3 tons,
MLW = 386.0 tons. If we load the airplane with the MSP, we have ZFW = 361.0
tons, which is exactly the certified MZFW. In this case, the amount of fuel required
to reach the MLW is ∼27 tons. With such an amount of fuel, the airplane can
quickly return to the airfield of origin if it was to abort the mission. Because in a
typical climb the aircraft burns ∼9 tons of fuel, in addition to ∼ 2 tons for taxi and
take-off, the aircraft will have to dispose of about 16 tons of fuel before landing.
15.7 Mission Problems

There are quite a variety of mission scenarios. Even in the relatively simple case
of a transport aircraft, important decisions may have to be taken when planning
a flight and during the flight. In this section we deal with a number of practical
problems. We begin with the mission analysis with an intermediate stop (§ 15.7.1);
in this case we want to understand whether there are benefits in terms of fuel
consumption if we split the mission and call at an intermediate destination. The
second problem deals with fuel tankering, which is a practice sometimes used to
save money at destination airports if the fuel is more expensive there (§ 15.7.2).
Finally, we consider a contingency scenario in which the pilot may have to abort the
flight and decide whether to return to the origin airport or continue to its planned
destination (§ 15.7.3).
15.7.1 Cruise with Intermediate Stop

Is it more economical to fly a long distance no-stop or with an intermediate stop?
Some airlines occasionally offer cheaper seats to a destination via an off-route stop,
which is counter-intuitive to both the customers and the airplane designers. The
aviation industry has developed aircraft for very long range to serve distant points
without en-route stops. Some of the current-generation airlines can fly, in some
limited scenarios, in excess of 10,000 n-miles. This is close to the ultimate global
range, 20,000 km (∼10,790 n-miles), advocated by Küchemann 9 . The global range is
half the Earth’s circumference at the equator and would allow an aircraft to fly from
any point to anywhere around the world – at least in principle.
The problem is one of long-range cruise in which a subsonic commercial jet has to
travel from airport A to a destination airport B. The distance along the flight corridor
is within the certified maximum range of the aircraft. However, here we consider the
problem of an en-route stop at airport C. The aircraft carries enough fuel to reach
airport C, refuels at this base, takes off again and flies to its final destination B. We
do not consider the costs of landing at C or the direct operating costs incurred by
the operator for increasing the time of the return journey. If x is the distance to be
flown in a direct flight, and x1 , x2 are the flight segments, we assume that
x1 + x2 = x. (15.37)
Table 15.1. Fuel use for mixed long- and short-range service of
the Boeing B777-300 (calculated)
m f [kg] m f [kg] m f [kg] ṁ f [kg/s]

X[n-m] total climb cruise cruise
300 6,674 2,553.9 955.0 2.04

5,000 106,287 3,056.3 94,864.6 2.63
5,300 112,396 3,056.6 100,662.7 2.62
In other words, the en-route stop C is on the route to the final destination B. A stop
with a diversion clearly increases the total range. The method consists in calculating
the mission fuel for three flight segments: x, x1 , x2 . We then calculate the relative
fuel cost. This is the ratio between the fuel required by the flight with en-route stop
and the direct flight:
m f (x1 ) + m f (x2 )
mfr = . (15.38)
m f (x)
This quantity can be plotted as a function of x1 /x (the ratio between the first flight
segment and the total range). Consider the case in which an airline operates an
intercontinental service between A and B separated by 5,000 n-miles. Another
important destination is C, which is 300 n-miles from B. The results of the calculation
performed with the FLIGHT program are shown in Table 15.1. The data show that
stopping at en-route airport at 5,000 n-miles and proceeding to the final destination
300 n-miles away causes an increase in fuel consumption by 0.5%.
15.7.2 Fuel Tankering

The aircraft has to carry the minimum required amount of fuel to perform its mission.
However, this type of analysis does not take into account any contingencies arising
from the costs of purchasing fuel. Some commercial operators use the practice of
“tankering”, that is, loading more fuel than required by a mission profile to offset
the costs of purchasing fuel at a higher price at destination. As a rule, the longer
the flight, the less capacity there is for tankering. Figure 15.15 shows the tankering
performance of the Airbus A320-200, adapted from the manufacturer’s data* . The
optimum weight is not fully specified; that is, it is not clear what payload is carried.
Assume that the required range is Xr eq for a total payload weight Wp . The fuel
required for this mission is W f (including fuel reserves). To establish the optimum
tankering, we need to forecast the payload and the weather for the return journey. At
this point the matter can become complicated; to simplify, we assume that the aircraft
performs the return journey with the same payload and atmospheric conditions.
Hence, the amount of fuel required is the same.
From the point of view of the calculation, there is no difference between fuel
weight and payload weight. The increase in fuel burn due to an increase in payload
(or fuel itself) is the derivative
dW f
m= . (15.39)
dWto
* Airbus: Getting to Grips with Fuel Economy, Issue 3, Blagnac, France, July 2004.
1.1
0
1.08 50
1,
Fuel price ratio

0
1.06 1,00
750
1.04
500
250 n-miles
1.02
1
50 60 70 80
Optimum weight, ton
Figure 15.15. Tankering performance of the Airbus A320 relative to cruise FL-330. Data
adapted from Airbus.
This derivative is always positive. Therefore, the additional fuel required to carry
the tanker fuel W ftanker is mW ftanker ; the total increase in take-off weight (or increase
in fuel weight) is
Wto = W f = (m + 1)W ftanker . (15.40)
The price to be paid at departure for this additional fuel is
Pd = mpd W ftanker + pd W ftanker . (15.41)
The first contribution is due to the cost of carrying the tanker fuel; the second term
is the cost of the tanker fuel itself. The cost saved at arrival is
Pa = pa W ftanker . (15.42)
The difference Pd − Pa must be negative (which corresponds to a saving):
Pd − Pa = mpd W ftanker + pd W ftanker − pa W ftanker < 0, (15.43)
which implies
pa
> 1 + m. (15.44)
pd
The tanker fuel in Equation 15.43 is indeterminate. Note, however, that the fuel
derivative (Equation 15.39) increases with the tanker fuel. Therefore, for a given
favourable fuel price ratio, it becomes progressively less convenient to tanker any
fuel; alternatively, for a marginally favourable fuel price ratio, an increase in tanker
fuel can be offset by the additional cost of carrying it. Mathematically, the optimum
tanker fuel is W∗ftanker such that
pa
m= − 1. (15.45)
pd
For our calculation, we fix the price ratio pa / pd > 1 (or pd / pa < 1). The computa-
tional procedure is the following:
1. Specify the required range, payload and fuel price ratio.

2. Establish a minimum and a maximum tanker fuel W ftanker −1 , W ftanker −2 .
3. Use the bisection-iteration method between these two values of W ftanker to cal-
culate the fuel derivative m that satisfies Equation 15.45. This procedure should
converge to W∗ftanker (with some caveats).
With reference to point 3 in the previous procedure, there is an inner loop

that calculates the fuel sensitivity by using central differences, that is, two mission
calculations with the tanker fuel W ftanker ± dW f , where dW f is a small number.
For certain values of the fuel price ratios, the procedure does not converge. In
fact, if pa / pd is high, the procedure attempts to calculate the tanker fuel correspond-
ing to a high value of the fuel derivative (Equation 15.45), which is incompatible
with the flight systems. For example, if the fuel price at arrival is double the cost
at departure, the fuel derivative that solves Equation 15.45 should be m = 1, a far
higher value than actually required. A similar difficulty exists for pa / pd 1, which
implies that for marginal differences in fuel costs, it is uneconomical to tanker any
fuel. Realistic values of the fuel derivative are m = 0.01 to 0.07, depending on air-
craft, range, payload and other factors. We conclude that in these circumstances it
is always convenient to use tankering.
For the Airbus A320-200, we have carried out the following analysis: Xr eq = 750
n-miles; bulk payload Wp = 2,000 kg; passenger load equal to 80%; standard day,
with a price fuel ratio pa / pd = 1.05. The optimum tanker fuel is W ftanker = 1,273 kg,
corresponding to a ramp weight of ∼65,400 kg. This result does not match the value
quoted by the manufacturer, which in any case is not fully specified. The iterative
analysis carried out with the FLIGHT program is shown in Figure 15.16. The error on
Equation 15.45, that is, the difference between left- and right-hand sides, is ∼0.002
after four iterations.
Because fuel tankering causes additional fuel burn, it also causes additional
environmental emissions. If environmental costs are added to the analysis, the incent-
ive to tanker may disappear altogether. In any case, to be precise we would need
to account for secondary costs, such as increased maintainance costs on engines,
tyre and brakes wear, discounted for the shorter turn-around time at destination.
Thus from an operational point of view, it may be more convenient to formulate the
problem as a break-even fuel price ratio.
15.7.3 Equal-Time Point and Point-of-No-Return

If diversion is required due to external events, landing must be planned at an airport
different from the planned destination. Such events may include engine failure, loss
in cabin pressure, adverse weather, air traffic control and more. A contingency plan
is formulated on the basis of the equal-time point and the point-of-no-return. The
situation is shown in the sketch of Figure 15.17. These quantities are defined as
follows:
0.06 1,400
Tanker fuel
Error = 0.002
Fuel derivative
0.05 1,200
Fuel derivative
Tanker fuel, kg
0.04 1,000
0.03 800
0.02 600
1 2 3 4
Iteration
Figure 15.16. Fuel-tankering calculation for the Airbus A320-200.
r The equal-time point P is the point during the flight when the time required
ET
to return to the airport of origin and the time required to continue the flight to
the planned destination are equal.
r The point-of-no-return P is the point during the flight when the fuel remaining
NT
is not sufficient to return to the airport of origin.
The point-of-no-return comes after the point of equal time. The wind direction
is critical. These two points must be converted to a geographical position and must
be assigned an estimated time of arrival.
Assume that the aircraft has reached the point PET . Due to external events, it
must be decided whether to return or continue. At this point, the fuel required for
the inbound flight (return to origin) is less than the fuel required for the outbound
flight. However, the flight time is the same. In this case, there is a time constraint (or
priority) rather than a fuel constraint.
PET PNR
Wind
Take-off
Vreturn Vcontinue
Figure 15.17. Point of equal time and point-of-no-return.

Next, assume that the aircraft has reached the point PNT . The fuel required to
return would be more than the fuel required to continue. In this case, there is a fuel
constraint rather than a time constraint.
If we use our FLIGHT program, the calculation of PET and PNT is relatively
straightforward if there is no wind. We calculate a full mission with the procedure
MissionFuel(..), which returns a mission time t and a mission fuel m f ; we then
calculate a posteriori PET as the point where the flight time is t/2; PNT is the point
where the fuel burned is m f /2. However, this is not a satisfactory solution for two
reasons. First, the wind cannot be neglected. Second, we cannot wait until the end of
the flight to decide what to do in an emergency situation. When time is critical and
the aircraft must return to the origin, we must consider the time required to make a
U-turn (§13.2.4). A procedure to take into account these effects is the following:
1. Specify the mission parameters, including range, wind and wind direction.
2. Perform the mission analysis with function MissionFuel(..).
3. Estimate PET as a point somewhere in the cruise segment, determined by a
distance XET from the origin airport and a flight time t ET .
4. Perform the mission analysis with the function MissionFuel(..) up to XET ;
the distance to continue would be Xc = X1 − XET ; the remaining time would be
t = t1 − t ET .
5. Perform a U-turn at the cruise altitude (§ 13.2.4).
6. Switch wind direction.
7. Continue the mission to the destination (e.g., for the remaining distance).
8. Compare the time to return t ET with the time to continue t: if t ET > tc , then the
point PET was over-estimated; otherwise, it was under-estimated.
The last point of this procedure can be implemented into a bisection method
that iterates between an under-estimate and an over-estimate of the solution. The
procedure can be computationally demanding because it requires two calls to
MissionFuel(..) to bracket the solution and five to seven bisection iterations.
Because each mission analysis requires about four to five iterations to converge,
there are up 40 to 50 calls to MissionFuel(..), unless the converge criterion is
relaxed. The aircraft will have some mandatory fuel reserve (§ 15.5). In any case, we
assume that the aircraft maintains the fuel reserve.
An example is the following. A model Boeing B747-400 with CF6 engines is
on a mission from London Heathrow to New York Kennedy airport. The nominal
required range is 5,700 km (3,078 n-miles), with 1 metric ton of cargo and 80%
passenger load. It encounters headwinds of 39 kt at cruise altitude. For this specific
case, the equal-time point is estimated at 1,615 n-miles (52.4% of the required
distance) at a flight time of 233.5 minutes from take-off.
15.8 Direct Operating Costs

Direct operating costs (DOC) are the costs incurred in the ownership and operation
of an aircraft. They include the cost to operate scheduled and unscheduled flights,
to insure and to maintain the aircraft as airworthy. The aircraft costs money even if
it stays on the ground. The parametrisation of the fuel costs is aleatory, due to the
costs of aviation fuel on the international markets and to the cost of fuel at different
airports around the world.
An example of DOC analysis is given by Beltramo et al. 10 , who developed cost
and weight estimating relationships for commercial and military transport aircraft.
Kershner 11 has demonstrated that whilst DOC have consistently decreased over
time, the impact of the fuel cost has remained high and in some historical circum-
stances has increased to more than 50%. The prediction of the operational costs due
to fuel consumption can be calculated with the methods explained in this chapter.
Other studies in this area focus on the fuel consumption for subsonic flight 12 and for
supersonic transport 13 .
The appearance of budget airlines on the major world markets has contributed to
a substantial change in the structure of the DOC. It is sometimes exciting to learn that
we can buy a ticket to an international destination within Europe or the continental
United States at a cost comparable to the cost of this book. Cost items such as tick-
eting, customer service, seat allocation, baggage handling, ground services, in-flight
catering, airport taxes, leasing contracts and so on have been dissected, reduced or
removed altogether from the DOC. However, the cost of the fuel remains essentially
the same. This cost is an essential part of the aircraft performance.
CALCULATION METHOD. The calculation of the DOCs of an airplane is a contentious

issue. Data from major airlines are published by the IATA and are closely monitored.
On the design side, there are unsubstantiated claims made by the manufacturers. On
the operational side there are uncertainties arising from market conditions, financial
arrangements and a long list of externalities (inflation, variable interest rates, fuel
costs, tax liabilities, labour relations, adverse weather, and so on). Thus, if we want
to be realistic, we need to build these uncertainties in the costs analysis. With the
foregoing cautions in place, we provide a simplified analysis that is based on about
half a dozen major cost items. These are:
1. Ownership costs, which are a combination of cash payments on acquisition, loan

repayment over a fixed length of time and residual value of the airplane at the
end of the financing contract.
2. Insurance costs, which will be considered a fraction of the value of the airplane,
although this might not be the case (the third-party liabilities do not decrease
over time).
3. Fuel costs, based on average costs between departure and arrival at today’s price,
adjusted for inflation, or an agreed deviation from it.
4. Crew costs, consisting of pilots and flight attendants, based on full-time rates.
5. Spare parts costs, consisting in parts for the propulsion system, airframe, landing
gear and any other items.
6. Labour costs, based on recognised rates (in today’s prices, adjusted for inflation)
for propulsion, in-house and contracted-out operations; part of the problem is
to gather reliable data for these items, the man-hours required to carry out the
service and the down-time for the airplane.
7. Other costs, including landing fees, ground handling, pilot training, on-board
service items, costs for off-station residence and so on. None of these costs can
be neglected because they make up the difference between being profitable and
making a loss.
We start with the ownership costs, which are the most onerous. These costs are
estimated from the equation
1
C1 = [P − F(1 + I)n − Rn ] , (15.46)
n
where P is the aircraft price agreed at the acquisition date; F is the financing (i.e.,
the fraction of the acquisition price requiring a loan); I is the interest rate on the
loan, assumed to be constant over the life-time n (in years) of the loan itself; and
Rn is the residual value after n years. The latter item is uncertain and must be based
either on today’s price or perhaps corrected for inflation at the end of the loan. To
be on the safe side, we could assume that the aircraft will be written off at the end
of the period, Rn = 0. The insurance cost is estimated from
I
C2 = Rj , (15.47)
100
where Rj is the residual value at year j. The fuel cost over one year is
C3 = NC f uel m f , (15.48)
where N is the number of cycles, C f uel is the average fuel cost per kg, and m f is the
fuel burn per cycle over the specified stage length. This quantity is calculated with a
mission analysis and will depend on a realistic estimate of the passenger-load factor
and gross take-off weights. Fuel-price oscillations are accounted for in the averaging.
Increases in fuel consumption due to engine age are accounted for with a separate
model (see, for example, § 5.3.6). The deterioration-adjusted fuel burn is

D f (x) = α1 1 − e−α2 x + (nw − 1)/c, (15.49)
where x = N/1,000 is the number of thousand cycles since the last wash, α2 a deteri-
oration lapse (α2 1 to 1.5) and α1 is the deterioration effect on the fuel burn after
1,000 cycles (α1 2 to 3); nw is the number of total engine washes and c is the
irretrievable loss of performance over time (c 20). The crew cost is

Tb Tb Tb
C4 = n1 S1 + n2 S2 + n1 S3 , (15.50)
To 1 To 2 To 1
with n1 and n2 the number of pilots and average flight attendants required to operate
the aircraft year round; S1 and S2 are the corresponding salary costs based on full-time
service, Tb is the total number of utilisation hours per year, and To is the contractual
number of working hours. The number of flight attendants will be estimated from
the passenger load and the type of services (about 1/30 in economy and 1/10 in
premium/business class). Other major crew-related costs include pilot training S3 ,
which takes place ahead of the acquisition and at scheduled intervals during the
lifetime of the airplane. The costs of the spare parts is

C5 = SPk, (15.51)
k
where SPk (k = 1, · · · ) denotes spares for the propulsion system, APU, airframe,
landing gear, tyres and any other items. The maintenance schedule depends on
the number of cycles and airplane age; it is divided into three broad categories:
propulsion and associated systems, in-house and contracted out. Maintenance costs
increase with ageing aircraft and cannot be oversimplified. As a result, the labour
costs associated to maintenance are likely to go up as the number of hours required
increases. The associated maintenance costs are:

C6 = LkMHk, (15.52)
k
with Lk the labour rates (any currency per hour) for the categories listed herein
and MHk the number of man-hours required in each category to carry out the
maintenance schedule. The number of man-hours involved depends on the airplane
(gross weight, number of engines, landing gear) and the number of cycles. These costs
suffer a jump as soon as the airplane comes off the warranty period. In practice, this
means that part of the costs are transferred from the manufacturer to the operator.
This transfer is called burden and is not necessarily an ageing cost.
The landing fees depend on the airport, the weight class of the aircraft and
local congestion; it may be subject to additional levies, such as CO2 , NOx emissions
and noise emissions. We assume that there are two major contributions: from the
gross weight (or number of seats) and the overall emissions, with weights c1 and c2 ,
respectively:
C7 = 2(c1 W + c2 E)N. (15.53)
As anticipated, the maintenance costs increase over time. Figure 15.18 shows such
effects. These charts have been elaborated from a statistical analysis carried out by
the Rand Corporation 14 . The data have been generated from a limited sample of
airplanes analysed; their limitations are discussed in the report cited. The charts
provide some weights to account for the costs of major items, such as airframe,
engine and burden effects. In all cases, it is shown that new aircraft have rapidly
increasing costs, which eventually flatten out with age. Apart from having to update
the coefficients, depending on type of aircraft, technology level and other factors,
the conclusions are generally valid. If we use the form
C56 = C5,1 + C6,1 (15.54)
to denote the combined parts and man-hour costs for the engine, the function C56
should follow the trend of Figure 15.18b. Even then, we need to establish values
for labour rates L and man-hours MH at year zero. The DOC model, as presented,
requires the evaluation of about 30 independent parameters. A summary of these
parameters is in Table 15.2 for clarity.
On a final note, we report that there has been an introduction of leasing contracts
for the propulsion system. These contracts allow the airline operator to simply pay
a cost proportional to the number of flight hours. When this is the case, the analysis
proposed here will have to be adjusted to take into account this different cost
structure.
1.8 1.8
1.5 1.5
Airframe cost / flight hour
Engine cost / flight hour

1.2 1.2
0.9 0.9
0.6 0.6
0.3 0.3
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Airplane age, years Airplane age, years
(a) (b)
1.8 1.8
1.5 1.5
Total cost / flight hour
Burden / flight hour
1.2 1.2
0.9 0.9
0.6 0.6
0.3 0.3
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Airplane age, years Airplane age, years
(c) (d)
Figure 15.18. Estimated ageing effects on maintenance costs.
Table 15.2. Summary of parameters for DOC model
Item Parameters Reference
Ownership P, F, i, n, Rn Equation 15.46

Insurance I Equation 15.47
Fuel N, C f uel , m f Equation 15.48
Crew n1 , n2 , S1 , S2 , To Equation 15.50
Spare parts SP1 , SP2 · · · Equation 15.51
Labour rates L1 L2 · · · MH1 , MH2 · · · Equation 15.52
Landing fees c1 , c2 , E Equation 15.53
Other costs To be specified
Ageing effects on fuel α1 , α2 , nw Equation 15.49
Ageing effects on maintenance Figure 15.18
15.9 Case Study: Aircraft and Route Selection 453
Table 15.3. Calculated payload fuel efficiency for long-haul commercial flight; all
flights at M = 0.80; all weights are in metric tons
Aircraft OEW MTOW PAY X [n-m] Fuel GTOW
B 129.9 275.0 35.0 7,560 74.62 244.1

A 90.1 165.0 35.0 3,780 55.69 156.1
B 129.9 275.0 50.0 7,560 77.94 263.1
PRODUCTIVITY MEASURES. There are specific parameters used by commercial oper-

ators to define their productivity. One of the most important is the Available Seat
per kilometre (ASK) or nautical mile (ASM). This parameter expresses the number
of seats available for sale, flown 1 km (or 1 n-mile). Then there is the Cost per avail-
able seat-km (CASK) or seat-n-mile (CASM). The CASK is equal to ASK/DOC
and determines the actual cost incurred by the operator of making a seat available
on a certain airplane over a certain route. Because the DOC changes as a function
of the various parameters, so does the CASK. Sometimes a cost analysis is carried
out by excluding the cost of fuel; this operation requires the ex-fuel CASK. Finally,
there is the revenue per ASK (RASK). To make a profit, a commercial operator
must have RASK > CASK.
15.9 Case Study: Aircraft and Route Selection

How do we select a new airplane from two or more competing airplanes? We will
have to face a mountain of marketing documentation about the benefits of one
airplane with respect to the other; each airplane is inevitably 20% more fuel efficient
than the other one, with the costs of the competing airplane being calculated by
the competitor, not the manufacturer. This is not an easy decision to make and
even between two competing airplanes of one manufacturer, a business decision
will have to be made on the basis of hard evidence; the process can be costly and
time-consuming.
In the next analysis we will refer to the relatively simple problem of deciding
which airplane to dispatch for a certain mission. We assume that we have to connect
two airports separated by a distance of 14,000 km (∼7,560 n-miles) and that we have
to deliver a payload of 35,000 kg. We have airplanes A and B, whose OEW and
MTOW are given in Table 15.3.
From the payload-range chart, we conclude that aircraft B can make a no-
stop flight to destination. For this aircraft, we estimate a fuel burn of ∼77.62 tons.
The service can also be done with aircraft A with an en-route stop half-way (7,000
km; 3,780 n-miles). The total fuel required (including the two legs) in this case
is ∼55.69 tons, which is a savings of about 25%. Therefore, the second option is
more economical. If, instead, we are required to deliver a 50-ton payload, aircraft A
is not large enough; aircraft B instead can easily accommodate that payload, and the
increase in fuel consumption over the previous case is only ∼4.5%. The summary of
our calculations is in Table 15.3.
Figure 15.19 shows the result of a parametric analysis carried out with a model
Boeing B777-300 airplane with GE-92 engines. The purpose of this analysis was to
No-Stop
0.06 En-route Stop at X/2
Fuel/pax/n-mile, kg
0.055
0.05
0.045
XA XB
0.04
0 1,000 2,000 3,000 4,000 5,000 6,000
Required range, n-miles
Figure 15.19. Analysis of B777-300 airplane model flight with and without en-route stop;
standard day, no wind.
understand how the en-route stop affects the fuel cost. The results indicate that for
a distance above ∼3,000 n-miles, an en-route stop could save fuel. The reason for
this is that the longer the required range, the heavier the airplane, as a result of
having to carry a larger amount of fuel. The solution shown in Figure 15.19 is not as
straightforward as it may look. First, the parametric analysis is carried out with the
assumption that a suitable airport is available at exactly half the distance. Second,
there can be considerable costs in landing at an intermediate destination. Third, for
some values of the range, it would not be possible to stop anyway. For example, for a
3,000 n-miles flight it would make no sense to stop 500 n-miles to the final destination
unless there is a commercial reason to do so. Refer also to the case illustrated in
Table 15.1. The range X between XA and XB is the most efficient and corresponds
to a fuel burn (per passenger per n-mile) that is within 1% of the minimum. The
results shown refer to full passenger load. Therefore, the fuel/pax/n-mile is equi-
valent to fuel/seat/n-mile. The result would be slightly different for a less-than-full
aircraft.
The solution can be further refined to include other factors and possibly an
estimate of the DOC. Green 15 showed an analysis for cargo aircraft and recommen-
ded that from the point of view of fuel efficiency, a flight segment shall not exceed
7,500 km (∼4,000 n-miles) at the current level of technology. In-flight refuelling is an
option that can be considered for either extending the range or increasing the pay-
load for a fixed range. However, the calculation must include the cost of operating a
tanker aircraft and to maintain a loiter trajectory for some time 16 .
The next analysis focuses on the fuel use as a function of the required range.
Figure 15.20 shows the fuel fraction (with respect to the fuel embarked) used for
climb to the ICA and for cruise. The cruise fuel converges toward 80% of the total
1
Cruise fuel
Climb fuel
0.8
Fuel fraction
0.6
0.4
0.2
0
0 1,000 2,000 3,000 4,000 5,000 6,000
Range, n-miles
Figure 15.20. Analysis of fuel use by the Boeing B777-300 with GE engines; standard day, no
wind; full passenger; no bulk cargo.
fuel embarked, whereas the climb fuel decreases to a few percent. Shorter jour-
neys require relatively higher portions of the fuel to be used for climb. Specifically,
on a 250-n-mile journey the climb fuel exceeds 25% of the total fuel embarked.
The remaining fuel is used in the other flight segments and includes the reserve
fuel.
15.10 Case Study: Fuel Planning for Specified Range, B777-300

We apply the methods presented earlier to the calculation of the mission fuel and
gross take-off weight for the Boeing B777-300 with General Electric GE092 tur-
bofan engines and 331-500 APU. A summary of mission calculations is shown
here, which are computer outputs from the program FLIGHT. Looking at a more
detailed level, we have reports for the various segments of flight. Table 15.4 shows
a list of flight conditions and assumptions. Table 15.5 is a summary of fuel/weight
planning resulting from the calculations. We discuss separately the different flight
segments.
Taxi-Out Report
It is not unusual that the aircraft must remain on hold waiting for a departure slot,
due to local traffic or bad weather. Fuel consumption on the taxi-way can be so high
that the flight must be aborted. The jet engines’ performance is optimised for cruise,
and considerable efficiency penalties are faced on the taxi-way. Therefore, taxi-times
must be estimated with accuracy.
Some economy can be gained by taxiing the aircraft with OEI or by towing the
aircraft to and out of the gate, or by parking the aircraft away from the gate. The
Table 15.4. Operational data for mission analysis in

case study
Item Value Unit
Required range 2,698 [n-miles]

Required bulk payload 0.0 [ton]
Required pax 74.6% full capacity
Outbound taxi time 10 [min]
Inbound taxi time 8 [min]
Wind speed at cruise −5.0 [m/s]
Change in temperature 0.0 [K]
Take-off airfield altitude 50.0 [m]
Landing airfield altitude 50.0 [m]
Derating none
Hold altitude 507 [m]
Hold time 30.0 [min]
Diversion distance 200.0 [n-miles]
Baggage allowance/pax 15 [kg]
Average pax weight 80 [kg]
OEI procedure is not recommended in a number of cases, including high GTOW,

slippery ground, nose-wheel steering performance, limitations on jet blasts, safety
for ground personnel and other reasons. It can be more efficient to taxi at a high
speed (although the limit is set to 25 to 30 kt) with a burst of engine power, and then
run the engine with idle power.
A first-order estimate of the taxi fuel is found with the method shown in § 9.10
(Chapter 9). A summary of the taxi-out performance for this specific case is shown
in Table 15.6. This table indicates that a considerable amount of fuel, half a ton, is
required to reach the start of the runway! Thus, we face an important operational
and environmental problem that must be addressed by more efficient procedures.
Take-Off Report
This report is shown in Table 15.7. There is a large set of airplane and operational
data required for the complete take-off analysis. The wind speed and runway state
are specified on input. The remaining data are calculated by the computer program.
The flap setting is also part of the solution, as the numerical algorithm attempts to
perform a take-off with the minimum flap setting compatible with a minimum initial
climb gradient. Included in the analysis is the result of the thermo-structural model
for the tyres (§ 14.5) and the minimum control speed on the ground (§ 9.6).
En-Route Climb Report

Key data are the time to climb, the distance to climb and the fuel to climb. The various
climb segments are identified and report data such as average climb rate, fuel use
and fuel flow. Landing-gear retraction and flaps retraction are done instantaneously.
Climb is done in four segments, the last one being a constant-Mach climb to the
Table 15.5. Summary of flight-planning analysis
Fuel Use Report
Segment m f [kg] m f [%] X[n-m] t [min] Notes

Climb 4,230.2 8.05 191.76 28.4
Cruise 37,415.5 71.16 2,317.72 299.7
Descent 5,037.1 9.58 210.44 37.3
Take-off 214.6 0.41
Taxi-out 504.1 0.96 1.12 10.0
APU/ECS 4,674.9 8.89
Taxi-in 415.9 0.79 from reserve
Fuel burn 52,582.0
Fuel Breakdown Report (all data [kg])
Taxi-out 504.1
Take-off 214.6
Climb 4,230.2
Cruise 37,415.5
Descent 5,037.1
Approach 89.5
APU/ECS 4,674.9
Diversion/Hold 2,498.1
Reserve 2,754.0
Total 57,834.1 42.1% of full tanks
Final Weight Report (all data [kg])
OEW 161,355
Bulk payload 0.000
Passengers 23,520
Flight crew 1,140
Useful payload 24,520
Service items 0.873
Fuel 57,834
Non-usable fuel 824
Ramp weight 243,658
BRGW 243,658
Zero-fuel weight 187,887
Take-off weight 249,816
Final Cruise weight 239,096
Landing weight 193,075
optimal (or sub-optimal) ICA. This report is shown in Table 10.3 in Chapter 10 and
is not repeated here for brevity.
Cruise Report
This report is shown in Table 15.8. The cruise flight is done in constant-altitude
segments at suitable flight levels, followed by constant-climb rate between flight
levels (§ 12.7). In this case, the aircraft operates between FL-330 and FL-370 and
does two step climbs. Each constant-altitude cruise segment is approximately 750
n-miles. The climb rate between flight levels is specified.
Table 15.6. Taxi-out report of fuel/

weight-planning analysis
Parameter Value Unit
Fuel rest-to-roll speed 14.52 [kg]

Fuel in idle mode 121.60 [kg]
Fuel in roll mode 368.02 [kg]
Total taxi fuel 504.14 [kg]
Time in idle 6.7 [min]
Time in roll 3.3 [min]
Roll speed 5.0 [m/s]
Roll distance from ramp 3.0 [km]
Taxi procedure AEO
Table 15.7. Take-off report of fuel/weight-planning analysis
Parameter Value Unit Notes
Gross take-off weight 243.658 [ton]

Airfield altitude 50.00 [m]
Air temperature 0.00 ± ISA
Wind speed −2.00 [m/s] −3.9 [kt]
Runway state dry
Thrust angle 0.000 [degs]
Stall margin 1.150
Max lift coefficient 2.435
Stall speed 69.64 [m/s] 135.3 [kt]
xCG 35 [% MAC]
Pitch moment of inertia 104.70 [106 kgm2 ]
Flap setting, δ f 20 [deg]
Rotation velocity, VR 63.46 [m/s] 123.3 [kt]
Rotation distance, xR 822.56 [m]
Rotation time, t R 29.00 [s]
Lift-off velocity, VLO 71.46 [m/s] 138.8 [kt]
Lift-off distance, xLO 1,038.88 [m]
Lift-off time, t LO 32.85 [s]
Velocity over screen 71.64 [m/s] 139.2 [kt]
Mach over screen 0.211
Distance to screen, xT O 1,150.18 [m]
Time over screen, tT O 34.65 [s]
Climb angle over screen, γT O 12.48 [deg]
Fuel burn 214.60 [kg]
VMCG 55.88 [m/s] 108.6 [kt]
Max (main) tyre temperature 293.2 [K] +0.6 [K]
Max (nose) tyre temperature 373.7 [K] +26.3 [K]
Main tyre temp. over screen 292.6 [K]
Nose tyre temp. over screen 347.4 [K]
Brake-release net thrust 877.873 [kN]
Brake-release fuel flow 7.947 [kg/s]
Table 15.8. Cruise report of fuel/weight-planning analysis; Xr eq = 2317.7

[n-miles]; ICW = 239.096 [ton], M = 0.803
h FL X t mf ṁ f vc
[n-m] [min] [kg] [kg/s] [ft/min]
10,058 330 772.57 99.26 12,837.1 2.156

19.68 2.54 238.1 1.562 787
10,668 350 772.57 99.73 12,435.0 2.078
19.52 2.54 230.3 1.510 787
11,277 370 733.37 95.58 11,674.9 2.036
Totals 2,317.72 299.65 37,415.5 2.081
Descent Report
En-route descent is performed according to the principles described in Chapter 11.
From the final cruise altitude (FCA), the airplane follows some steps down to a
reference altitude, for example 1,500 feet (457 m) above the airfield, with a gradual
decrease in air speed. A typical descent report is shown in Table 11.2 in Chapter 11;
this is not repeated here for brevity.
Analysis of Contingency Fuel

There are various reserve fuel policies, some of which are discussed in § 15.5. A
combination of various contingency policies as a function of the required range is
shown in graphical form in Figure 15.21. The 20-minute extension of the flight is
assumed to take place at the cruise Mach and the final cruise altitude. This fuel
reserve is not dependent on the required range because the all-up mass at the
start of the flight extension is virtually a constant parameter. The holding cases
6,000
20 min extension
15 min hold; 1,500 ft
5,000 5 min hold; 1,500 ft
5% Mission Fuel
3% Mission fuel
Reserve fuel, kg
4,000
3,000
2,000
1,000
0
1,000 2,000 3,000 4,000 5,000
Required range, n-miles
Figure 15.21. Effect of cruise range on contingency fuel; standard day, no wind; full passenger
load, 300 kg of bulk cargo (calculated).
Table 15.9. Basic performance data of model float-plane
Data Value Unit
Cruise TAS 163.4 m/s (592 km/h, 319.5 kt)

Cruise altitude 6,000 m (19,685 feet)
Specific fuel consumption 0.3567 kg/h/kW
Fuel capacity 27,200 kg
(15 or 5 minutes) are performed at 1,500 feet above the airfield at the hold M =
0.350. Other flight conditions are specified in the caption. For the holding cases,
we assumed that the initial mass corresponds to the all-up mass before the final
approach, which yields a nearly constant reserve fuel.
If an alternate en-route airport is available, the best option is a 3% reserve fuel
below 2,200 n-miles. If this is not possible, a 5% reserve fuel is less restrictive than
the 20-minute cruise extension for stage lengths below 2,600 n-miles. This conclusion
is quite general: each aircraft has a critical stage length beyond which a change in
contingency policy may yield a weight advantage. This stage length depends on the
actual conditions, in particular weather patterns and payload.
15.11 Case Study: Payload-Range Analysis of Float-Plane

For this study we consider the Lockheed C130J with floats. This is one of the many
derivatives of the Hercules aircraft family, designed to perform special operations
such as troop transport, anti-submarine warfare, mine counter-measures, search and
rescue, fire-fighting, inter-island transport, oil spill response and more. Some basic
performance data for the calculation of the floats’ drag are given in Table 15.9.
The estimated floats’ dimensions are summarised in Table 15.10. These data have
been calculated from three-view drawings of the airplane, which can be found in
the public domain. The wetted area and the volume are more difficult to calculate
because these floats are not bodies of revolution.
15.11.1 Estimation of Floats Drag from Payload-Range Chart

One useful application of the payload-range chart is the estimation of the drag
difference between different configurations of an airplane. Figure 15.22 shows the
estimated payload-range performance of the aircraft with and without floats. The
difference in ferry range (zero payload) is about 570 n-miles. This difference is a
reflection of a considerable drag created by the floats, the bracing and the increased
structural weight.
Table 15.10. Estimated floats’ dimensions
View A [m2 ] x [m] y [m] z[m]
Top 49.530 21.20 3.39

Front 4.584 2.97 2.00
Side 26.384 20.55 1.89
15.11 Case Study: Payload-Range Analysis of Float-Plane 461
20
15
Payload, tons
10
Without floats
With floats
0
0 500 1,000 1,500 2,000
Range, n-miles
Figure 15.22. Payload range chart of Lockheed C130J, with and without floats.
A first-order estimate of the drag increase is done by assuming that 1) the

structural weight of the aircraft is unchanged; 2) the cruise speed is unchanged;
3) the cruise altitude is unchanged; and 4) the drag caused by the bracing is negligible.
These effects will be re-evaluated at a later time.
Assume that the nominal performance (C130J without floats) is indicated with
the subscript “1”, whilst the new configuration (with floats) is indicated with sub-
script “2”. The work done to fly the airplane over the distance x1 is W1 = Tx1 . The
difference in work done to fly the airplane from x1 to x2 is
W = Tx = (T1 − T2 )(x1 − x2 ). (15.55)
In steady level flight T = D. The power relationship for a propeller aircraft is
η
T= P, (15.56)
U
where T is the net propulsive thrust, η is the propulsive efficiency, U is the true
air speed and P is the shaft power. The relationship between shaft power and fuel
flow is
ṁ f = SFC P. (15.57)
At constant TAS, constant propeller efficiency and constant SFC, we have
η η ṁ f
T = D = P = . (15.58)
U U SFC
The difference in fuel flow required is
ṁ f = ṁ f2 − ṁ f1 . (15.59)
The total mass of fuel use by the aircraft is about the same (if we exclude the effects
on climb-descent performance). Thus,
ṁ f1 t1 ṁ f2 t2 = m f . (15.60)
The condition of constant TAS is quite useful at this point because x = Ut.

t1 x1
ṁ f2 = ṁ f1 = ṁ f1 . (15.61)
t2 x2
Thus, the difference in fuel flow is

x1 x
ṁ f = ṁ f1 − ṁ f1 = −ṁ f1 . (15.62)
x2 x2
If we introduce Equation 15.62 into Equation 15.58, we find

η ṁ f1 x
D = . (15.63)
U SFC x2
Note that x < 0 and, therefore, the change in drag is positive. Equation 15.63 is
valid only for constant speed, constant cruise altitude and constant SFC. Calculation
of the drag increase would require the value of x. In first analysis, this value can be
taken directly from the payload-range chart (for the ferry condition). However, we
must also consider that the aircraft must take off, climb to cruise altitude, descend
and land with a minimum of fuel reserves. Therefore, not all of the fuel capacity can
be used for cruise. We assume that the fuel capacity is cleared for the mandatory
reserves (say 5% of the usable fuel for a ferry operation); an additional 13% is
needed for climb to altitude, descend and land. Thus, the usable fuel is about 82%
of the fuel capacity.
The cruise range is lower than the mission range because the airplane performs
an en-route climb and descent. In first analysis the discount is about 100 n-miles.
Using the basic performance data shown in Table 15.9, we estimate that
m f1 1.042 kg/s, D 24,840 kN, CD 0.00174. (15.64)
In conclusion, the analysis indicates that the floats add about 174 drag counts to the
cruise-drag coefficient. This is a considerable amount of drag: it corresponds to a
decrease in ferry cruise range by 570 n-miles.
A more accurate calculation is possible, starting from the result just achieved.
The first step forward is to consider the optimal air speed of the aircraft with floats.
The optimal speed (minimum-power speed) is
1/2
2W 1
Ump = , (15.65)
ρ A CLmp
with
1/2
3CDo
CLmp = . (15.66)
k
The difference in profile drag is CD = CDo . The aircraft with the floats must fly
faster, as shown by the equation

Ump2 CLmp2 1/2 CDo + CDo 1/2 CDo 1/2
= = = 1+ > 1. (15.67)
Ump1 CLmp1 CDo CDo
For a 20% increase in profile drag, the optimal TAS would have to increase by about
10%. Thus, if we take into account these new flight conditions, the new fuel flow is
−1/2
U1 x1 x1 CDo
ṁ f2 = ṁ f1 = ṁ f1 1+ , (15.68)
U2 x2 x2 CDo

x1 CDo −1/2
ṁ f = ṁ f2 − ṁ f1 = ṁ f1 1+ −1 . (15.69)
x2 CDo
Equation 15.69 can be used to re-iterate and calculate a new value for the increase
in profile drag.
15.12 Risk Analysis in Aircraft Performance

Virtually all of the problems in aircraft performance presented so far are based on
deterministic analysis. In other words, given a set of initial conditions and airplane
state, it is possible to evaluate almost exactly how the airplane will respond. This is far
from true. There are two issues at stake: 1) the risk of failure and the corresponding
consequences; and 2) the uncertainty on some state parameters. An example of
statistical performance analysis is available in Ref. 17 .
As in any other engineering field, there is a risk associated to the operation of
an airplane. Safety in air transportation is paramount but risk-free operation, like
any other pursuit in life, is virtually impossible. It is perhaps an unfortunate fact
that a considerable proportion of the technological progress achieved in aerospace
engineering is attributed to accidents of various levels of severity. We presented
some examples, but the literature on the subject is vast.
The meaning of risk itself must be clearly defined, particularly the relationship
between the risk itself and its severity (or its consequences). Consider the case of
ETOPS, which allows twin-engine airplanes to fly over land for a prescribed amount
of time (60, 120 or 180 minutes) with one engine failure. The risk analysis associated
to this type of operations has allowed twin-engine airplanes to operate across North
Atlantic routes. ETOPS safety criteria correspond to less than 3 · 10−9 fatalities per
flight hour arising from a total loss in engine thrust from independent causes. The
event of one engine failure on a twin engine is more severe than an engine failure
on a three- or four-engine airplane† .
The analysis shall consider at least the “unit of risk”, which is the probability of
occurrence of a certain event (incident, engine failure, flight diversion, accident) per
flight hour or other relevant measure, for example number of cycles 18 . Most current
regulations have been derived from the works of the ICAO and adapted by national
aviation regulators, with some local variations. Figure 15.23 shows the notional
relationship between the unit of risk and the severity of failure. This severity is
mapped onto a failure consequence. The shaded area corresponds to an unacceptable
level of risk. The ETOPS risk level is shown for reference.
† For operational issues, see Airbus: Getting to Grips with ETOPS, Flight Operations Support and
Line Assistance, Issue V, Oct. 1998, Blagnac, France.
Catastrophe Loss of Aircraft
Hazard Injuries
Damage
Failure Consequence
Major
UNACCEPTABLE RISK
Failure Severity
Emergency
Abnormal
Minor
ETOPS Safe Ops
10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 100
Failure Probability
Figure 15.23. Risk analysis in aircraft performance.
Aside from the unit of risk, there is the problem of uncertain parameters (due
to stochastic events) and/or imprecise evaluation of parameters. Performance pre-
diction, then, can only be probabilistic. The performance charts produced by the
manufacturers (as part of the FCOM) are a case in point. Although the graphs asso-
ciated to the FCOM show clean lines, the predicted performance must be interpreted
in a statistical sense with a high level of confidence.
Some of the performance calculations that are most subject to parameter uncer-
tainty, stochastic external factors and risks of mechanical failure include take-off
and landing 19;20 . ESDU has a number of data items on this subject, covering basic
statistical analysis and practical methods for evaluating the performance risk 21;22 .
Within the context of this discussion, we aim to carry out a “sensitivity” analysis
for selected parameters. Because the number of system parameters is quite large,
the investigation space is inconveniently large. Therefore, it is necessary in the first
instance to isolate the dominant parameters, to evaluate their standard deviation
and to establish the effects of such parameters. We give a few examples here.
SPECIFIC AIR RANGE GUARANTEE. The guarantee usually carries a penalty if the per-
formance cannot be matched to the specifications. An example of variability was
shown in § 12.8, where we examined the effects of some parameters on cruise per-
formance deterioration. In this instance, we simply assume that a number of tests
are carried out; the tests indicate some variability in the SAR; the result is given in
the following array.
SAR, % = {1.21, 1.35, −0.59, 0.23, −2.01, 1.05, 1.45, −1.27, −0.97, 0.45}.
(15.70)
Equation 15.70 is a 10-point array containing the percentage change in SAR calcu-
lated with 10 flight tests. The statistical properties of this array are: sample standard
deviation = 1.21%. On a normal distribution, the mean value has a confidence level
of 68.2% within one standard deviation (±1.21%), 95.4% confidence within two
standard deviations (±2.42%). This level of confidence will have to be considered
carefully before signing any delivery contract.
TAKE-OFF FIELD LENGTH. A preliminary analysis was carried out in Chapter 9, when
we considered the effects of an engine failure at a critical point. We then defined
the balanced field length and determined the actions to be taken to either stop the
airplane or continue the take-off run. Take-off charts contain a number of parametric
events that take into account brake-release gross weight, atmospheric temperature,
longitudinal wind speed, airfield altitude, runway gradient and flap setting. This type
of analysis is in fact deterministic, in the sense that a parameter variation is assumed
and the response of the aircraft is measured or simulated. The actual statistical
analysis is based on “risk”, that is, the probability that certain combinations of winds,
airfield altitude, engine failure and so on can conspire (independently or otherwise)
to cause an unacceptable level of risk, as indicated in Figure 15.23. Additional
considerations may be necessary if the airplane has changed configuration and its
condition has deteriorated since the last inspection. The exercise of bringing all these
effects together is a major project.
NOISE PERFORMANCE GUARANTEE. A detailed analysis of aircraft noise is carried out

in Chapter 16. Noise measurements are also based on statistical values* . The effects
of various sources on overall noise can be assessed by calculating the noise response
to a set of random variations in parameter settings. This analysis is strongly non-
linear because of the log-sum between components and because the noise metric is
an integral over the full trajectory.
Summary
We addressed several problems in aircraft-mission analysis, with focus on trans-
port operations. We first examined issues of flight scheduling to maximise the
use of the aircraft, and demonstrated that departure and arrival are time-critical.
Next, we illustrated the importance of the payload-range chart, which is one of the
main drivers in aircraft utilisation. We presented several numerical methods for
the computation of payload-range charts and for mission planning; in particular,
we demonstrated how the predictor-corrector scheme applied to the mission fuel is
the most robust method for calculating the mission fuel. We presented a detailed
case study of transport-mission analysis and discussed each flight segment separately.
A number of other transport of practical interest include flight with en-route stop
and fuel tankering. We highlighted the importance of the corresponding choices,
which respond to market conditions. The DOC depend on the fuel consumption but
also on a considerable number of parameters in, around and external to the aircraft.
* An example is available in the certification of the Airbus A318, A319, A320, A321. See EASA,
Type Certificate for Aircraft Noise, TCDSN EASA.A.064, Issue 10, June 2011.
Calculation of the costs is made difficult by the lack of data that can be generalised.
We showed a case study of a float-plane mission and the influence of the floats’ drag
on its payload range chart. We concluded the chapter by illustrating the importance
of risk analysis in aircraft performance.
Bibliography
[1] Gang Yu, editor. Operations Research in the Airline Industry. Kluwer Academic
Publishers, 1998.
[2] Coy S. A global model for estimating the block time of commercial passenger
aircraft. J. Air Transport Management, 12:300–305, 2006.
[3] Anonymous. 747-400 airplane characteristics for airport planning. Technical
Report D6-58326-1, The Boeing Corporation, Dec. 2002.
[5] ESDU. Examples of Flight Path Optimisation Using a Multi-Variate Gradient-
Search Method. Data Item 93021. ESDU International, London, Mar.
1995.
[6] Martinez-Val R and Pérez L. Extended range operations of two and three
turbofan engined airplanes. J. Aircraft, 30(3):382–386, 1993.
[7] Sachs G. Optimization of endurance performance. Progress Aerospace Sci-
ences, 29(2):165–191, 1992.
[8] Torenbeek E. Cruise performance and range prediction reconsidered. Progress
Aerospace Sciences, 33(5-6):285–321, May-June 1997.
[9] Küchemann D. The Aerodynamic Design of Aircraft. Pergamon Press, 1978.
[10] Beltramo MN, Trapp DL, Kimoto BW, and Marsh DP. Parametric study
of transport aircraft systems cost and weight. Technical Report CR 151970,
NASA, April 1977.
[11] Kershner ME. Laminar flow: Challenge and potential. In Research in Natural
Laminar Flow and Laminar-Flow Control, volume NASA CP-2487. NASA
Langley, Mar. 1987.
[12] Isikveren AT. Identifying economically optimal flight techniques of transport
aircraft. J. Aircraft, 39(4):528–544, July 2002.
[13] Windhorst R, Ardema M, and Kinney D. Fixed-range optimal trajectories of
supersonic aircraft by first-order expansions. J. Guidance, Control and Dynam-
ics, 24(4):700–709, July 2001.
[14] Dixon M. The maintainance costs of aging aircraft. Technical Report Rand
MG456, RAND Corporation, 2006.
[15] Green JE. Greener by design – The technology challenge. Aero J., 106(1056):
57–113, Feb. 2002.
[16] Bennington MA and Visser KD. Aerial refueling implications for commercial
aviation. J. Aircraft, 42(2):366–375, Mar. 2005.
[17] Anon. Statistical loads data for the Boeing B777-200ER aircraft in commercial
operations. Technical Report DOT/FAA/AR-06/11, U.S. Dept. of Transpor-
tation, Office of Aviation & Research, Washington, DC, Nov. 2006.
[18] Wagenmakers J. A review of transport airplane performance requirements
might benefit safety. Flight Safety Foundation Digest, 19(2):1–9, 2000.
[19] ESDU. Example of Risk Analysis for an Aircraft Subject to Performance Errors.
Data Item 08006. ESDU International, London, Aug. 2009.
[20] ESDU. Example of Risk Analysis Applied to Aircraft Landing Distance. Data
Item 08005. ESDU International, London, June 2008.
[21] ESDU. Statistical Methods Applicable to Analysis of Aircraft Performance Data.
Data Item 91017. ESDU International, London, Oct. 1991.
[22] ESDU. Variability of Standard Aircraft Performance Parameters. Data Item

91020. ESDU International, London, Oct. 1991.

A = wing area
c = irretrievable loss of engine performance; constant
cus f = usable fuel ratio, Equation 15.16
C1 = ownership cost
C2 = interest rate
C3 = fuel cost
C4 = crew cost
C5 = spare-parts cost
C6 = maintenance cost
C7 = landing fees
C56 = combined spare-parts and man-power cost, Equation 15.54
C f uel = fuel cost per kg
CLmp = lift coefficient for minimum power
E = fees due to environmental emissions; residual in fuel update
F = financing
i = iteration counter
I = interest rate on loan
k = induced-drag coefficient
kw = counter in DOC analysis (§ 15.8)
L = labour rate (currency units per hour)
r = fuel reserve, in percent
m = mass
mf = fuel burn
mfr = fuel-mass ratio, Equation 15.38
ṁ f = fuel flow
M = Mach number
MH = man-hours
n = lifetime of loan
n = number of iterations/bisection steps
nw = total number of engine washes
n1 ,n2 = number of pilots/flight attendants, Equation 15.50
N = number of cycles for engine wash
pa , pd = fuel price (arrival, departure)
P = price
P = power
PET = point of equal time
PNT = point-of-no-return
r = fuel-reserve ratio
Rdiv = diversion distance
Rn = residual value after n years
S = salary (wages) based on full-time rates
SP = spare parts
t = time
T = net thrust
Tb = utilisation time (block time plus turnaround time)
To = full-time working hours per year
U = true air speed (also TAS)
Ump = speed of minimum power
W = mechanical work
W = weight
Wd = weight at the top of descent
We = empty weight
Wf = fuel weight
Wland = landing weight
Wmf w = maximum fuel weight
Wmsp = maximum structural payload weight
Wmlw = maximum landing weight
Wp = payload weight
Wto = take-off weight
x = number of cycles in thousands, N/1,000; distance
xc = en-route climb distance
xd = en-route descent distance
xi = flight-distance segments (climb, cruise, descent . . . )
Xdesign = manufacturer’s design range
X = stage length
Xdiv = diversion distance
Xhold = holding distance
Xout = equivalent all-out range
Xp = payload range
Xr eq = required range
Greek Symbols
α1 = deterioration, Equation 15.49
α2 = fuel-efficiency-deterioration-lapse coefficient
ηM = log-derivative of the propulsive efficiency
= fraction of the estimated mission fuel
μr = rolling resistance
ξ = fuel fraction, Equation 15.4
ρ = air density
[.]c = climb
[.]d = descent
[.]e = empty
[.] ET = equal-time
[.]ext = extended range
[.] f = fuel
[.]k = index for maintenance items or spare parts
[.]i = iteration counter
[.]mlw = maximum landing weight
[.] NT = no return time
[.]msp = maximum structural payload
[.] p = payload; predicted value
[.]div = diversion
[.]hold = hold
[.]res = reserve
[.]to = take-off
[.]∗ = estimated quantity
16 Aircraft Noise: Noise Sources
Overview∗
The subject of aircraft noise is too vast to be covered in a single chapter; thus, we focus
on those issues that are required for a low-order simulation of aircraft noise emission
and propagation. The chapter is organised as follows: We first discuss the importance
of the subject (§ 16.1) and introduce basic definitions that are used to characterise
aircraft noise (§ 16.2). We then establish the framework for the development of an
aircraft-noise program that is synchronised with the flight mechanics (§ 16.3). The
sources of noise are split into propulsive (§ 16.4) and non-propulsive components
(§ 16.6). We include a discussion of the APU noise as a separate item (§ 16.5). The
propeller-noise model (§ 16.7) is discussed in isolation from airframe and engines.
Chapter 17 deals with propagation effects. Chapter 18 deals with noise trajectories.
KEY CONCEPTS: Sound, Noise Metrics, Engine Noise, Airframe Noise, Propeller
Noise, APU Noise, Aircraft-Noise Model.
16.1 Introduction
In the beginning there was silence, or so the story goes. Then came human activities
and jet engines. This is when the noise started to hurt. Low-flying aircraft cause
nasty physiological effects and are bad for your well-being 1 . To be fair, noise pol-
lution is just one of many environmental issues that face the aviation industry; it is
part of the larger context of local air quality, combustion emissions, environmental
compatibility, policies and regulations and public health.
In this chapter we shall identify the main causes of aircraft noise and some
methods for noise reduction. Aircraft certification must meet increasingly tough
international regulations (FAR Part 36 and ICAO Annex 16) and restriction of
operations at several airports around the world. It will be explained how excessive
noise emission can make an aircraft obsolete. In the current world of regulations,
Concorde would not be allowed to fly; a first-generation jet airplane would be
immediately grounded.
∗ Dr. Zulfaa Mohamed-Kassim contributed to this chapter.
470
Because a considerable number of models still rely on empirical and semi-

empirical evidence, at times the physics is replaced by a direct approach to noise,
which is currently the most straightforward solution. This is a rapidly expanding area
of research, with new methods being proposed on a regular basis. The use of physics-
based models is computationally more demanding and, for some types of aircraft
performance study, it is not practical. Engineering applications include aircraft noise
from conventional and unconventional configurations.
16.2 Definition of Sound and Noise

The essential parameters for defining noise are the sound pressure level (SPL), the
acoustic power P, and the acoustic intensity I. The sound pressure p is calculated
with respect to a reference value pref = 20 μPa; this value corresponds to the sound
pressure at the threshold of hearing. The sound intensity is defined as the sound
pressure multiplied by the speed of propagation, or the acoustic power per unit
area:
a2 pp p2 Fp r E1 P
I = pa = p = = = = = , (16.1)
a aρ ρa At t A A
where F p is the force created by the acoustic pressure p, ρ is the air density, E is the
energy, t is the time, r is the distance from the source and Ais the propagation area.
For a point source, the propagation area is spherical, so P decays as 1/r 2 ; for a line
source, the propagation front is cylindrical, and P ∼ 1/r . A measure of the loudness
of the sound is given in Bel. Because this unit is generally a too large quantity,
a decibel (10−1 Bel = 1 dB) is used in practice. The sound loudness in decibels is
defined from the ratio between the actual sound intensity I and the sound intensity
at the threshold of hearing, Io

I p 2 p
SPL(dB) = 10 log = 10 log = 20 log , (16.2)
Io pref pref
where “log” is the symbol for logarithm with base 10. The threshold of hearing
is set at pref = 20 μPa at which the corresponding SPL is defined as 0 dB. From
Equation 16.2, about 12% increase/decrease in sound pressure corresponds to ±1
dB change in relative sound intensity; ±1 dB is the minimum perceptible sound
difference by a sharp ear (just noticeable difference); 6 dB corresponds to a doubling
of the sound pressure; and +10 dB is required before sound is perceptibly louder.
Generally, 120 dB is the threshold of pain, 110 dB is a very noisy jet aircraft, 80–90
dB is street noise in a busy city, 65 dB is a noisy office in which verbal communication
is disrupted, and 50 dB is the average office environment. In a sound-proof room the
SPL is of the order 10–20 dB.
The noise is also characterised by its frequency. The range of frequencies that can
be heard by a human ear is between 20 and 20,000 Hz, and the range of intensity is 4
to 120 dB, although this range depends on the person. Human hearing is less sensitive
at very low and high frequencies. To account for this, some weighting filters can be
applied. The most common frequency weighting in current use is “A–weighting” that
conforms approximately to the response of the human ear. The A–weighted sound
level, called dBA, accounts for the fact that the middle to high frequency range (500
Table 16.1. Summary of integral noise metrics
Metrics Acronym ANSI 3 Full Name Reference
A-Weighted Noise
Exposure-based SEL L AE Sound exposure level § 16.2.2
LAEQ L AeqT Equivalent sound level Equation 16.11
Maximum-level LAMAX L ASmx Maximum sound level
Tone-Corrected Perceived Noise

Exposure-based EPNL L EPN Effective perceived noise level § 16.2.1
Maximum-level PNLTM L PNLSmx T-corrected max perceived noise p. 474
to 5,000 Hz) is more annoying to human ears than low frequencies. However, this is
partly compensated by the fact that high frequencies show larger decay rates as they
propagate through the atmosphere; therefore the most important frequency range
from the point of view of annoyance is the 200–2,000 Hz range. A full description of
the basic noise metrics is given in many acoustics textbooks, for example, Pierce 2 .
The frequencies used in aircraft noise are a 1/3 octave band series. When two
frequencies f1 and f2 are such that f2 = 2 f1 , it is said that they are separated by
an octave. This would normally be the size of a band. However, in aircraft noise,
the band is 1/3 octave, and the audible spectrum is divided in 1/3 octave bands.
Therefore, two consecutive frequencies are
f2 = 21/3 f1 . (16.3)
This means, for example, that if f1 = 20 Hz, then f2 = 25 Hz, f3 = 31.5, and so on
(the frequencies are approximated to the nearest integer).
The application of these concepts to aircraft noise is somewhat more complicated
than a single acoustic pressure in dB. In fact, the problem of noise is addressed in
term of annoyance and its related causes, such as peak, duration, frequency of events,
and so on. Thus, other measures of noise have been developed to characterise the
complete aircraft trajectory. These new metrics are integral, in the sense that a
segment of the flight trajectory is associated to a single noise level. Integral noise
metrics are discussed next. Basically, there are three types of metrics: 1) exposure-
based, 2) maximum noise level, and 3) time-based metrics. Most of these metrics
have been standardised. A partial list of metrics is given in Table 16.1. A full list of
terms is published by the ANSI 3 .
16.2.1 Integral Metrics: Effective Perceived Noise

Because the noise heard on the ground is better characterised by a single scalar
parameter, we need a method to transform the SPL of the individual component
noise, at each frequency and time step, into such a quantity. This is currently done
with a quantity called Effective Perceived Noise Level (EPNL). The EPNL is a
weighted value that takes into account loudness and duration. The mathematical
120
100 Phon
100
90
80
80
70
SPL, dB
60
60
50
40
40
30
20
20
10
0 1 2 3 4
10 10 10 10
Frequency, Hz
Figure 16.1. Perceived noise with selected Noy levels. Adapted from ISO 226.
definition of this quantity is
EPNL = PNLmax + C + D, (16.4)
where PNLmax is the maximum perceived noise level (in dB), C is a correction (in
dB) for tonal noise and D is a correction for duration of the noise (also expressed
in dB). The main procedure to calculate the EPNL is done in a number of steps
(readers are referred to Smith 4 for detailed calculations of the EPNL).
1. The SPL signal is converted into a PNL (Perceived Noise Level) through a
conversion chart that is shown in Figure 16.1. This chart shows levels of constant
Noy, which is a measure of equal loudness. Take, for example, the frequency f ,
corresponding to SPL( f ). The combination of an SPL( f ) for a single f gives one
value of the Noy (directly or by interpolation). A total NoyTotal can be derived
by integrating each Noy value across the entire one-third octave-band frequency
spectrum where

24
NoyTotal = 0.85Noymax + 0.15 Noyi . (16.5)
i=1
The relationship between NoyTotal and PNL at each time-step in the trajectory
is
PNL(dB) = 40 + 10 log2 (NoyTotal ). (16.6)
2. Next, we have to define a tonal correction C for each time-step. This correction
depends on the frequency band and most critically on a level difference F,
whose calculation can be laborious (see Edge and Cawthorn 5 and Smith 4 ).
Once the level difference is determined, the tonal correction is given in the chart
Tone correction, dB
0.5 kHz < f < 5 kHz
f < 0.5 kHz and f > 5 kHz
0
0 5 10 15 20 25
Level difference F, dB
Figure 16.2. ICAO tone-correction chart.
of Figure 16.2 for various frequency bands. The maximum tone correction is 6.7
dB. Thus, the tone-corrected PNL is
PNLT = PNL + C [dB]. (16.7)
3. Calculate the maximum PNLT, defined as PNLTM, which is a scalar value.

4. Calculate a correction for duration of noise. If t = t2 − t1 is the time of the flight
path analysed, the duration is
t2
1
D = 10 log10 10(PNLT/10) dt − PNLTM. (16.8)
t t1
5. The effective perceived noise is calculated by summing the PNLTM and its
duration in the flight path
EPNL = PNLTM + D. (16.9)
In this sense, the signal has been cleared of frequency effects, but it takes into
account other aspects, such as tone and duration.
The method described indicates that the EPNL cannot be measured directly;
this parameter has to be calculated using information of the entire flight path, SPL,
spectrum level and shape. The value of the EPNL can be considerably different from
the loudness.
When using the EPNL metrics, noise reduction assumes a different meaning. In
fact, an EPNL reduction can be achieved by a combination of changes in the flight
path, changes in the acoustic signature that would limit the effects of tonal correction
and other factors. Furthermore, because the EPNL metric is an integral quantity, by
itself it is not representative of the accuracy of the noise-prediction method.
16.2.2 Integral Metrics: Sound Exposure Level

The sound exposure level (SEL) is also called LAE (dBA). It is calculated as follows:
r Calculate the maximum loudness along the trajectory, Lmax .

r Define the reference sound level Lref = Lmax − 10 dB.
r Calculate the time interval [t1 , t2 ] of the trajectory when L > Lref .
r Calculate the SEL from the integral:
t2
LAE = 10 log10 10 SPL(t)
dt. (16.10)
t1
In the A-corrected SPL spectrum, the loudness is defined as the maximum

SPL(dBA). Another integral parameter related to the SEL, is the Equivalent Con-
tinuous Noise Level, called LAeqT . This is defined as
LAeqT = LAE − 10 log10 to, (16.11)
where to is the time duration of interest.
16.3 Aircraft Noise Model

The noise produced by the aircraft is modelled with the method of components. This
method consists in reconstructing the noise from the contribution of all of the sub-
systems. Inevitably, there will be approximations and assumptions, some of which
are not completely rigorous. The second important aspect is that the noise model
for each component can be approximated, and better models may be developed
in the future. The advantage of this approach is that if one component provides
unsatisfactory answers, then it can be improved independently from the remaining
systems.
Figure 16.3 shows the breakdown in noise-source components. These are given in
three broad categories: propulsion, airframe and interference. Among the propulsion
sub-components, the propeller is not relevant if the airplane is powered by turbofan
engines. The APU contribution is generally not considered, although its effect is
substantial when the aircraft is on the ground (§ 16.5). The propeller noise will be
discussed separately from the propulsive noise (§ 16.7).
Among the airframe components, the effects of the fuselage itself is not included
because it is not significant. The interference contributions that are considered are
the jet-by-jet noise shielding (although this is not important on turboprop engines),
the fuselage scattering effects and the wing reflection of the noise sources; the latter
item is particularly important if the engines are mounted above the wing. The flow-
chart indicates that the aircraft-noise simulation can be built in a modular fashion: it
is possible to replace one sub-model with another one without having to reconsider
the rest of the framework. A full review of sub-models for aircraft noise is available
in Casalino et al. 6
System analysis indicates that approximations of the same order are needed to
either reduce noise from different systems or to calculate the noise emission with a
Figure 16.3. Noise-sources breakdown in a conventional airplane.
satisfactory accuracy. The far-field noise SPL is calculated from the sum:

SPLk /10
SPL = 10 log10 10 . (16.12)
k
Note that the resulting noise is calculated by first converting the various con-
tributions into decimal scale; the final result requires a decimal logarithm. Due to
the arbitrary definition of SPL such that it is 0 when p = pref , Equation 16.12 yields
a strange result on the summation of the total SPL. For example, the sum of three
contributions SPLk = 0 dB yields a total SPL= 4.77 dB, not zero as one would expect
in a conventional sum.
An airplane that is 50% quieter is an aircraft whose overall sound pressure is
50% lower. Because the noise metric is logarithmic, this figure translates into an
OASPL that is reduced by 6 dB. This result is derived from

p p
OASPL = 20 log10 − 20 log10
2 pref pref

p p (16.13)
= 20 log10 − log10
2 pref pref
= 20 log10 (1/2) −6 dB.

180
Visual Retarded
position position
Retarded emission angle, degs

M
θ θe
135
r
re h
Receiver
90
4
=
0.
M
=
M
45
0
0 45 90 135 180
Visual emission angle, degs
Figure 16.4. Corrected polar emission angle as function of flight Mach number. Curves are
shows at intervals M = 0.1.
If we invert Equation 16.13, the change in acoustic pressure p corresponding to a

known change in OASPL is
p
= 10(OASPL/20) − 1. (16.14)
p
In practice, this means that at sea level and standard day, the decrease in acoustic
pressure corresponding to a reduction of OASPL by 6 dB is only 480 Pa. However,
if the metric used is a “weighted” one, such as the perceived noise level (§ 16.2.1),
then noise reduction has a different meaning.
16.3.1 Polar-Emission Angle

In general, the aircraft or the noise source is found at some point above the receiver.
In the most basic case, the source-receiver position is determined by the distance r
and the emission angle θ; alternatively, it is defined by a distance and an altitude.
However, most noise-radiation effects depend on the emission angle and thence on
the directivity. The polar-emission angle is a function of the aircraft’s speed (or Mach
number). This is a result of the nature of wave propagation at a finite speed when
the source itself travels at a comparable speed.
With reference to Figure 16.4, if r is the “visual” distance source-to-receiver
and θ is the corresponding “visual” emission (or radiation) angle of a noise source
travelling with a Mach number M, the “corrected” (or retarded) emission angle θe
is calculated from

cos θe = cos θ 1 − M2 sin2 θ + M sin2 θ. (16.15)
The retarded distance is

sin θ
re = r . (16.16)
sin θe
Figure 16.4 shows the effect of flight Mach numbers up to M = 0.4, which is the
range of interest in perceived aircraft noise. At M = 0.4 the maximum correction
in the emission angle is θ −12 degrees and occurs when the noise source is right
above the receiver. A further elaboration of the emission angle is presented in § 16.6,
where we consider the case of a receiver away from the flight track. We will use the
symbol θ to denote the visual and corrected emission angle, as applicable.
16.4 Propulsive Noise

The determination of the propulsive noise still relies mostly on semi-empirical meth-
ods that contain a very limited amount of physics. A notable exception is the calcula-
tion of a single-jet noise, whose physical basis has been first established by Lighthill,
which is insufficient to model high by-pass turbofan engines anyway.
Even with the simplifications required by the semi-empirical models, a com-
prehensive approach to engine noise requires a considerable amount of engine
data, divided into two categories: 1) geometry and configuration data, and 2) aero-
thermodynamic data. In the former category we have parameters such as fan, com-
pressors and turbine diameters, number of blades, rpm. Once the engine model
has been developed, as explained in Chapter 5, the aero-thermodynamic simulation
provides the operating data required.
16.4.1 Noise-Propulsion System Interface

The numerical determination of the propulsive noise relies on a reliable estimate of
the engine state for a specified flight condition. We begin with the turbofan-powered
aircraft. Our model for a turbofan engine relies on the engine architecture shown in
Figure 16.5. There are two possibilities:
r The fuel flow rate is specified; the aero-thermodynamics of the engine is solved
to provide net thrust, pressures, temperatures and speeds at the relevant engine
sections (direct mode).
r The net thrust is specified; the aero-thermodynamics of the engine is solved
to provide mass-flow rates, fuel-flow rates, pressures, temperatures and speeds
(inverse mode).
In the first instance, we rely on externally generated noise trajectories such as,
for example, FDR data. The fuel flow is directly measured and monitored, whilst the
net thrust can only be determined through a flight-mechanics analysis.
The case of the turboshaft engine coupled to a propeller is similar, but it has
some important differences, which are shown in Figure 16.6. In the inverse mode,
the shaft power is determined via flight-mechanics equations. The problem is then
reduced to the determination of the fuel flow corresponding to the specified shaft
power. This calculation must be done iteratively to take into account the losses at
N%1 TT5
by-pass flow
N%2
Combustor
(FN)
T VJ
HP Compr
LP Compr
HP Turbine
LP Turbine
Fan
Wf6 Fuel Feed
[Engine control]
[to fuel tanks]

Figure 16.5. Turbofan engine model for noise simulation.
the gear-box and the effects of the residual thrust. The engine state is calculated
from the fuel-flow rate.
16.4.2 Fan and Compressor Noise

A semi-empirical method for compressor and fan noise was developed by
Heidmann 7 on the basis of an earlier method at Boeing-NASA. The method relies
on an experimental database and applies to fan and single-stage compressors, with
N%pt N%gg TT5

Propeller
8
7
6
2 3 9
4 -F J
T
HPC
LPC Nozzle
Comb.
HPT
LPT
Gearbox 5
Fuel Feed
[Reverse control]
[Direct control]
[to fuel tanks]
[Fuel Temperature]
Figure 16.6. Turboshaft-propeller model for noise simulation.
or without guide vanes. For a multi-stage compressor, it would be necessary to con-

sider each stage separately. In practice, this is a nearly impossible task because most
of the data required by the method (temperature rise across each stage, num-
ber of rotor/stator blades, rotor/stator blade chords, rotor-stator spacing) are not
available.
The comparison with noise data is shown to be generally acceptable at frequen-
cies above 1 kHz but considerably off the mark at lower frequencies, where the
model has a cut-off. This was seen as a limitation at the time. However, the method
was implemented in the NASA code ANOPP 8 and later slightly refined. The ESDU
method 9 follows the same type of analysis as Heidmann’s, with some minor modi-
fications. This method is not demonstrated as being valid in the case of centrifugal
compressors, such as those found in turboshaft engines.
The model includes a combination of broadband noise, discrete-tone noise and
combination-tone noise. The broadband and discrete-tone noise components arise
from the inlet and the discharge sides of the fan or compressor; the combination
tone occurs only on the inlet side. The power of the method derives from the
relatively limited number of parameters that are requested. These parameters are
mass-flow rate, temperature rise through the stage, design, and operating tip Mach
numbers. Some corrections can be introduced to take into account the presence of
guide vanes, rotor-stator spacing, inlet flow distortion tone cut-off and blade-row
attenuation. The method involves the prediction of the spectrum shape and level
and the noise directivity for each component.
BROADBAND NOISE. This noise component is associated to turbulence in the flow

through the blade rows, arising from inlet turbulence, wall boundary layers, and vor-
tex interaction between blades and vanes. The latter item depends on non-periodic
wakes. The formula for the broadband noise is

T ṁ
SPL( f ) = 20 log + 10 log + F1 + F2 + F3 + F4 . (16.17)
To ṁo
In Equation 16.17, F1 is a function of the tip Mach number, F2 is a function of the

rotor-stator spacing, F3 is a function of the directivity and F4 is a function of the blade
passing frequency; ṁ is the fan or compressor mass-flow rate, and ṁo is a reference
mass-flow rate; T is the temperature rise across the fan or compressor and To is
a reference temperature. The first term in the right-hand side of Equation 16.17
represents a contribution due to a temperature rise in the compressor stage; the
second term is a contribution due to mass-flow rate. This equation has been validated
for a compressor with a small number of stages.
The effect of blade passing on the broadband noise is given by a log-normal
distribution that has a maximum at 2.5 times the blade passing frequency fb:
2
ln( f/ fb)
F4 = 10 log10 (ec ), c= . (16.18)
ln 2.2
In practice, fans and compressors exhibit a more complex broadband spectrum than
the one predicted by Equation 16.18. The directivity, for both the inlet and discharge
sides, is shown in Figure 16.7.
0 0
Directivity, dB
Directivity, dB
-20 -20
-40 -40
-60 Inlet -60 Inlet

Discharge Discharge
0 45 90 135 180 0 45 90 135 180

Emission angle, degs Emission angle, degs
(a) Broadband (b) Discrete Tone
Figure 16.7. Single-stage fan and compressor noise directivities.
DISCRETE-TONE NOISE. The formula for the discrete-tone noise is

T ṁ
SPL( f ) = 20 log + 10 log + F1 + F2 + F3 . (16.19)
To ṁo
In Equation 16.19, the functions F1 , F2 , and F3 depend on tip Mach number, the
rotor-stator spacing, and the directivity, respectively. However, these functions are
different from those defining the broadband noise (see Equation 16.17).
COMBINATION-TONE NOISE. This is generated by supersonic tips at 1/2, 1/4 and 1/8
of the blade passing frequency and is given by a similar equation:

T ṁ
Lp = 20 log + 10 log + F1 + F2 . (16.20)
To ṁo
Again, the functions F1 and F2 depend on the tip Mach number and the directivity,
respectively. Formulas are given for the 1/2, 1/4 and 1/8 blade passing frequencies.
A blade row attenuation correction is applied for multiple-stage compressors but not
for a single-stage fan.
The temperature rise across the stage can be estimated from the adiabatic com-
pression
T 1 ) (γ −1)/γ *
= pr −1 , (16.21)
T ηc
where ηc is the thermodynamic efficiency of the stage, pr is the compression ratio
of the stage, and γ is the ratio between specific heats (γ = 1.4 will suffice for most
calculations).
The method described applies to a single-stage fan or compressor, although
the method has been tested successfully for two-stage compressors. The use of
multi-stage compressors in modern turbofan engines complicates the analysis. One
approach is to approximate the multi-stage compressor as a single stage by taking a
temperature and pressure rise between the first and the last stages. Unfortunately,
this approximation yields unrealistically high noise for a large number of stages.
Figure 16.8. Example of acoustic liner sheet for duct-fan noise reduction.
Another option is to consider only the effects of the first stage of LP and HP
compressors as representative of the far-field noise 7 .
The equations providing the noise level (Equation 16.17, Equation 16.19 and
Equation 16.20) are valid for an observer located 1 metre away from the source.
Therefore, for an observer at a distance r the result must be corrected for spherical
spreading, atmospheric absorption and other dissipative and amplification factors.
For the sake of generality, we assume that the noise spreads along a spherical front
without encountering any obstacles. In this case, the SPL varies inversely to the
square of the distance from the source (i.e., the inverse square law). Assume that at
a distance r1 the sound pressure level is SPL1 ; at a distance r2 > r1 , this quantity will
be reduced to SPL2 . The relationship between the two SPLs is

r1
SPL2 = SPL1 + 20 log . (16.22)
r2
Note that if the distance increases, the “log” is negative; therefore, there must be a
+ sign to the right of SPL1 . Equation 16.22 shows that a doubling of the distance
causes the SPL to be reduced by 6 dB.
With reference to the fan, which is one of the largest causes of engine noise, we
must address some additional aspects, which involve the presence of the inlet duct
and the so-called acoustic liners, both contributing to some noise reduction 10 . The
acoustic liners are passive or adaptive systems consisting of a variety of perforated
plates, honeycomb structures, and woven materials, whose purpose is to increase
the noise absorption in the duct over a specified range of frequencies. An example
of such liners is shown in Figure 16.8. A vast amount of research exists in this
area 11 – 13 . ESDU has some practical methods for estimating the acoustic impedances
and absorption 14;15 . The effectiveness of these systems depends on a large number
of factors, including the length of the duct (long or short), the type of liner, the
impedances and the directivity. The inlet ducts of modern turbofan engines are
relatively large, up to one-third of the fan diameter* . This means that the fans are
buried deep into the nacelles and allow the installation of a variety of technologies
for further noise reduction, including the liners.
Modern turbofan engines have wide-chord fans, with a greatly reduced number
of blades (typically 22 to 26 blades); this design contributes to the reduction of the
blade-passing frequency. The broadband noise signature is similar to the narrow-
chord fans. The reduction of the fan noise level requires appropriate strategies that
address the tonal components and lower frequency bands (600 to 4,000 Hz).
In many cases it is possible that the fan blades operate in supersonic flow. This
condition is typical of the tip regions at high engine speeds and moderate to high
aircraft speed (especially at take-off). The resulting noise is referred to as “buzz-
saw”. This event is characterised by acoustic pressures at frequencies multiple of the
engine rotational speed 16;17 .
16.4.3 Combustor Noise

The semi-empirical theory proposed by ESDU 18 (derived from earlier methods by
the Society of Automotive Engineers [SAE]) relies on an equation that correlates
the sound pressure level, the noise power level and the directivity:

ρoao Wref
SPL( f, θ ) = PWL( f ) + D(θ ) − 20 log r + 10 log10 2
, (16.23)
4π pref
where PWL is the one-third octave band noise power level in dB, and θ is the
angle between the engine axis and the receiver (emission angle); the subscript “o”
denotes ambient conditions outside the engine. The PWL is derived from a spectral
correction of the A-scale overall PWL:
PWL = OAPWL + S( f ), (16.24)
with the overall sound power level defined by:

aoṁ22 T 2 p3 2 Ttur b −4
OAPWL = 10 log10 + 10 log10 − C,
2
Wref T3 po To
(16.25)
where ṁ2 is the core-mass-flow rate, T is the temperature rise in the combustor, T3
is the combustor inlet temperature, and p3 is the corresponding pressure; Ttur b is
the temperature drop across the whole turbine system downstream of the combustor;
and C is an assigned constant, which in this case is −60.5dB. The item provides the
directivity D(θ ) and the spectral function S( f ) to be used in Equation 16.23 and
Equation 16.24. The SPL given by Equation 16.23 must be corrected for Doppler
effects and atmospheric attenuation. The approximation of the method is estimated
between −3 dB and +5 dB, although the method lacks a rational validation and
verification.
* For example, the length of the inlet duct of the GE90-115B is about 1.5 m; the total liner area of the
PW4048 engine powering the B777-300 is about 6.2 m2 .
16.4.4 Turbine Noise

A method to predict the tonal and broadband noise from a single axial-flow turbine
is presented here following the semi-empirical approach of the General Electric
Company 19;20 . The method relies on empirical relations that express the depend-
ence of sound pressure level on frequency and polar directivity angle. Relative to
other noise-producing engine components, fast first-order predictive methods and
empirical data for turbine noise are still lacking and the available ones are rather
crude 21 . Recent efforts focus more on developing physics-based modelling to pro-
duce robust predictions without any reliance on empiricism 22 and thus widen their
implementations on various types of engines, but these are unlikely to be favourable
options for fast real-time applications in the near future.
Turbine noise, particularly from low-pressure turbines, can be substantial at
approach 23 . Its importance on the overall engine noise increases with increasing tur-
bofan engine by-pass ratio 22 . The sound radiated from a turbine is highly dependent
on the polar angle and peaks with an emission angle of 110 to 130 degrees relative to
the intake axis 4 . This aft-radiated sound is influenced by sound refraction through
the turbulent jet mixing region. Pure tones dominate the sound spectra with the
fundamental harmonic set at the turbine-blade-passing frequency. The turbulent
mixing layer may also influence the tonal noise by diffusing it into more broadened
or “hay-stacked” signals 24 .
The empirical modelling used in the Aircraft Noise Prediction Program 19 is
described here to calculate the far-field SPL from an axial-flow turbine, which is
given as

p2
SPL = 10 log10 2
. (16.26)
pref
The mean-square acoustic pressure p2 is formulated as
ρ∞ a∞ Ai ∗ D(θ )S(η)
2 4
p2 = , (16.27)
4πr 2 [1 − M∞ cos θ ]4
with pref = 2 × 10−5 Pa as the reference acoustic pressure. Equation 16.27 takes into
account the effects of spherical spreading and forward flight via the Doppler term
[1 − M∞ cos θ ]. In addition, it uses two empirical relations to quantify the effect of
polar angle θ and frequency f on the turbine noise. The dimensionless acoustic
power parameter ∗ is detailed in § 16.4.4.
The first empirical relation is a polar directivity function D(θ ) that is given as two
sets of dimensionless datasets, each for the broadband and tonal noise, respectively;
these are tabulated as functions of the polar directivity angle θ in Table 16.2.
The second empirical relation is a spectrum function S(η) that is given as a
tabular dataset for the broadband noise and an empirical function for the tonal
noise; S(η) is a function of the dimensionless frequency parameter η, defined as
f
η = [1 − M∞ cos θ] , (16.28)
fb
Table 16.2. Polar directivity levels
Broadband Tonal
θ [deg] log10 D(θ) log10 D(θ)
0 −0.789 −1.911
10 −0.689 −1.671
20 −0.599 −1.471
30 −0.509 −1.261
40 −0.409 −1.061
50 −0.319 −0.851
60 −0.219 −0.641
70 −0.129 −0.431
80 −0.029 −0.231
90 −0.071 −0.021
100 0.151 0.189
110 0.221 0.389
120 0.231 0.589
130 0.211 0.259
140 0.111 −0.191
150 −0.029 −0.591
160 −0.229 −0.931
170 −0.549 −1.271
180 −0.869 −1.611
where fb = NB is the turbine-blade-passing frequency, with N being the turbine

rotational speed (in Hz) and B the number of rotor blades for the single-stage
turbine.
For the broadband noise, the spectrum S(η) is given in Figure 16.9. This dataset is
extrapolated to extend the range of the broadband spectrum levels beyond log10 (η) <
−0.903 and log10 (η) > 0.602. With these parameters, the broadband SPL for any
-0.6
-0.9
-1.2
log10 S(η)
-1.5
-1.8
-2.1
-1 -0.5 0 0.5
log10 η
Figure 16.9. Spectrum function for broadband noise.
Table 16.3. Empirical constants for

turbine acoustic power
Broadband Pure Tones
K 8.589 × 10−5 1.162 × 10−4

a 1.27 1.46
b −1.27 −4.02
specific frequencies can be calculated. To obtain the 1/3 octave band SPL, each
band is divided into smaller sub-bands where the sub-band SPLs are computed
using the sub-band centre frequencies. These are then integrated and normalised to
produce the 1/3 octave band SPL for the broadband noise.
For the tonal noise, the spectrum function is given by
S(η) = 0.6838 × 10(1−n)/2 , (16.29)
where n is the harmonic number of each tone. The fundamental harmonic, that is,
n = 1, is determined by the turbine-blade-passing frequency fb. In this method, the
tonal SPLs are combined with the 1/3 octave broadband SPL to obtain the total 1/3
octave band SPL. This is done by locating the number of harmonics as well as their
harmonic numbers that are contained within each band. Given a 1/3 octave band
centre frequency fc , the lowest (nl ) and highest (nu ) harmonic numbers within the
band can be found as follows:

nl = 1 + int 10−1/20 η
(16.30)

nu = int 101/20 η .
From this, a band would contain nu − nl + 1 number of tones only if nl ≤ nu . The

tonal SPL can then be calculated from Equations 16.26 and 16.27 and combined with
the broadband SPL to obtain the total SPL.
Acoustic Power
The empirical formulation of the dimensionless turbine acoustic power ∗ in Equa-
tion 16.27 is given as
a
∗
h∗t,i − h∗s, j
=K (UT∗ )b , (16.31)
h∗t,i
where the empirical constants K, a and b depend on the type of the noise source,
either broadband or pure tones. These values are provided in Table 16.3; h∗ denotes
the specific enthalpy: the index “i” denotes inlet condition; “j” denotes exit condi-
tions; “s” is the static condition and “t” is the total. The dimensionless rotor tip speed
UT∗ depends on the rotational speed N and rotor-blade diameter d as follows:
Nd
UT∗ = π . (16.32)
a
Table 16.4. Spectrum function for

broadband noise
Constituent Gas Molecular Mass
Nitrogen, N2 28.01340
Oxygen, O2 31.99880
Carbon Dioxide, CO2 44.00995
Water Vapour, H2 O 18.01534
Argon, Ar 39.94800
The ratio of specific enthalpies in Equation 16.31 requires the input of the total
temperature at the turbine inlet (Tt,i ) and the static temperature at the turbine exit
(Ts, j ). Following the documentation of Ref. 25 , the specific enthalpy h∗ (T ) for any
arbitrary temperature T is defined as

∗ 1 ∗ T ∗ T∞
h (T ) = h − hr , (16.33)
T∞ r To To
where T∞ is the ambient temperature and To is a reference temperature taken as the
sea-level temperature. The reference-specific enthalpies (hr∗ ) depend on a number
of parameters for the constituent gases of air, namely nitrogen (N2 ), oxygen (O2 ),
carbon dioxide (CO2 ), water vapour (H2 O) and argon (Ar). For a given temperature
ratio T /To, hr∗ is calculated from
5
∗ T ∗ ∗ T
hr = Rk xkhr,k . (16.34)
To To
k=1
Here, R∗k,xk, and ∗

hr,kare the dimensionless gas constant, the mass fraction and the
component specific enthalpy, respectively, for the k-th constituent gas; R∗k and xk∗
are defined as follows:

∗ 1 Ru
Rk =
R mk

mk
xk = (16.35)
mT

5
mT = mk.
k=1
Here, Ru = 8, 314.32 kgm3 /Ks2 is the universal gas constant and R is the dry-air
gas constant. The molecular masses mk of the gases are shown in Table 16.4. The
masses of the component gases xk, which depend on the engine fuel-to-air ratio and
the absolute humidity of the ambient air, are tabulated in Ref. 25 . The component-
∗
specific enthalpies hr,k are also tabulated as a function of the temperature ratio
T /To.
Turbine Noise Simulation

The turbine-noise model for a single engine is simulated to study the effect of
various parameters on the turbine SPL level received 500 m away from the engine
100 1,000 10,000
Figure 16.10. Effect of polar directivity angle θ on 1/3 octave band SPL.
with a flight Mach number M = 0.5. As mentioned previously, only the acoustic
and attenuating effects of spherical spreading and forward flight are incorporated
in the formulation; attenuation due to ground reflection, atmospheric absorption,
refraction and turbulence scattering are not accounted for.
In Figure 16.10, SPL is plotted against the 1/3 octave band centre frequencies
for various polar directivity angle θ ranging from 0 to 180 degrees. Generally, θ
influences the SPL via the empirical polar directivity function D(θ ) and the Doppler
shift term.
The shape of the SPL across the frequency spectra is governed by the spectrum
function S(η). The results show that the turbine noise generally peaks at frequencies
between 5 kHz and 10 kHz, depending on the shift from the effect of forward flight.
The peaks are mainly due to the pure tones which have been averaged into the
1/3 octave band scale. With respect to changes in the polar angle, the noise is most
attenuated aft of the engine at θ = 180 degrees. The overall noise peaks for θ of 0
and 180 degrees differ by about 20 dB.
The turbine rotational speed N can have a significant influence on turbine noise
because it determines the turbine-blade-passing frequency fb and thus the tonal-
peak frequencies. In Figure 16.11, N is varied from 3,000 rpm to 12,000 rpm. In
these simulations, the turbine noise is loudest for the smallest rotational speed with
a significant contribution of about 24 dB from the fundamental tonal noise level
relative to the highest broadband noise level. The overall 1/3 octave band SPL
generally shifts downward in the dB scale across the entire frequency spectra when
the rotational speed is increased. In addition, the first tonal peak shifts toward larger
frequencies with increasing rotational speed. This tonal contribution toward the
turbine SPL relative to those of the broadband noise also decreases with increasing
120
100
80
60
40
N = 3,000 RPM
20 N = 6,000 RPM
N = 9,000 RPM
N = 12,000 RPM
0
100 1,000 10,000
Figure 16.11. Effect of turbine rotational speed (rpm) on 1/3 octave band SPL.
rotational speed. At N = 12,000 rpm, the first tonal-peak noise level is about 8 dB
above that of the highest broadband noise level.
The effects of blade number B on the SPL are depicted in Figure 16.12. Because
the blade number directly determines the blade-passing frequency, part of the blade-
number effect on the noise level is to shift the tonal peaks in proportion with the
number of blades, similar to the effect of turbine rotational speed as shown in
Figure 16.11. This is clearly seen in the results as the fundamental harmonic is
shifted from f = 500 Hz to 10 kHz when the blade number is increased from 25
to 100. The SPLs of these fundamental harmonics remain unchanged at about 90
dB. The second effect of increasing the blade number is seen in the attenuation of
the broadband noise levels particularly at frequencies away from the blade-passing
frequencies.
16.4.5 Single-Jet Noise

The single-jet noise model applies only to pure turbojet engines. This type of engines
has virtually disappeared due to the high noise level and low efficiency. Fundamental
studies in this field were done by Lighthill 26;27 . Lighthill’s analysis shows that the
acoustic power of a high-speed jet in stationary surrounding is
ρ j Aj Vj8
P ∼ ρ j Aj Vj3 M5 = , (16.36)
a 5j
where Aj is the jet area, M = Vj /a j the Mach number, and a j the average speed of
sound in the jet. This equation is also known as Lighthill’s eighth power law. The
100 1,000 10,000
Figure 16.12. Effect of blade number B on 1/3 octave band SPL.
mass flow of the jet is ṁ = ρ j Aj Vj ; therefore

ṁVj7
P∼ . (16.37)
a 5j
The far-field noise produced by a high-speed jet behaves as if generated by a mono-
pole; hence, the intensity I decays like the inverse of the distance, 1/r ,
P ṁVj7 ṁVj7
I∼ ∼ ∼ . (16.38)
A 2πra 5 a 5r
The corresponding SPL is
7

I c1 ṁVj
OASPL = 10 log = 10 log , (16.39)
Io Wref a 5j r
where c1 is the constant of proportionality in Equation 16.38 and Wref is the reference
acoustic power. The importance of the jet speed is evident from Vj7 . It is verified that
if we reduce the jet speed by half, Equation 16.39 gives a reduction in OASPL of 21
dB to an observer at the same distance r . If, conversely, the distance of the observer
is doubled but the jet speed is maintained to the original level, then the reduction
in OASPL is only 6 dB. This result emphasises once again the importance of using
high by-pass turbofan engines instead of pure jet engines.
If the jet is derived from an engine flying at a speed V, then it is convenient to
introduce the engine thrust in Equation 16.38. A rough approximation for the thrust
is T ṁ(Vj − V). Thus, we have
T Vj7 TVj6
I∼ , I∼ . (16.40)
(Vj − V) a 5j r a 5j r
Secondary/Ambient
Shear Layer
Primary/Secondary
Initial Shear Layer
region
Mixed Flow
Vs Interaction
Vp
Primary/Hot Jet
Secondary/Cold Jet
Figure 16.13. Flow features of a coaxial hot-cold jet from turbofan engine.
This equation shows the importance of cutting back the thrust, something that is
routinely done in take-off operations. A simple calculation from Equation 16.40
shows that a reduction of the thrust by 50% leads to a reduction of the OASPL by
∼3 dB.
16.4.6 Co-Axial Jet Noise

The prediction of a single-jet noise relies on the theory of Lighthill 26;27 and the
subsequent developments, including the method of Balsa and Gliebe 28 and Stone
et al. 29 for coaxial jets. In modern jet engines the jet is made of a hot core and a cold
by-pass flow at a lower speed. The problem is considerably more complicated than
a single-stream jet. The additional parameters are at least the velocity ratio between
jets, the area ratio and the temperature ratio.
Experimental noise data exist 30;31 for a wide range of jet velocity ratios, nozzles
area, Strouhal numbers and polar angles, although some limitations exist as well,
including the interpolation of data. Other methods have been published by the SAE
on the basis of methods developed by NASA, Boeing and others. A somewhat more
comprehensive study, involving a reasonable degree of physics, is available from
Fisher et al. 32;33 . A full description of the model is given in those references. We
will limit the discussion to the acoustic-model equations and to the implementation
strategy within our comprehensive model.
Figure 16.13 shows a sketch of a typical co-axial jet, made of a hot-core (primary)
and a cold by-pass (secondary) jet. The initial region near the nozzle is character-
ised by a turbulent mixing between jets of different velocities and temperatures.
Two shear layers are identified: a shear layer between the secondary jet and the
atmosphere, and an internal shear layer between the primary and secondary jets.
The mixing of the secondary jet with the atmosphere generates high-frequency
noise.
Following Fisher et al., the noise from a co-axial jet flow is calculated by summing
three contributions: 1) the fully mixed jet, 2) the secondary jet, and 3) the effective
jet. Some further notation is required at this point: the subscripts “p” and “s” denote
the primary and secondary jets, respectively; the subscript “m” denotes the fully
mixed jet.
1. FULLY-MIXED JET. The first contribution is essentially due to an isolated jet of

appropriate diameter, velocity and temperature. If f is the frequency and θ is the
polar emission angle (as defined earlier), then this contribution is written as
SPLm(θ, f ) = SPL(θ, f, Vm, Tm, dm) + 10 log10 Fd ( f1 , f ), (16.41)
where the velocity Vm, temperature Tm and jet diameter dm are calculated from

Vm ψ2 dm ψoψ1 1/2 Tm ψo
= , = , = . (16.42)
Vp ψo dp ψ2 Tp ψ1
The parameters in the right-hand side of Equation 16.42 are
ψo = 1 + Vr Ar , ψ1 = 1 + Vr Ar σr , ψ2 = 1 + Vr2 Ar σr , (16.43)
where Vr = Vs /Vp is the velocity ratio, Ar = As /Ap is the area ratio, and σr = ρs /ρ p
is the gas density ratio. Equation 16.41 gives the SPL of the fully mixed jet at
frequencies above the cut-off frequency f1 = Vm/dm. The factor Fd is a function that
represents the fraction of the energy radiated from positions downstream of the axial
coordinate x1 :

1 1
Fd (x1 , f, f1 ) = 1 + fˆ + fˆ2 + fˆ3 exp( fˆ), (16.44)
2 6
where

f
fˆ = m . (16.45)
f1
The factor m is a shape parameter (typically, m = 4). The frequency f1 is such that
the energy radiated upstream and downstream of the position x1 is about the same.
The first term on the right-hand side of Equation 16.42 is calculated by using the
method of ESDU for subsonic circular nozzles 34 . This term is discussed separately
in § 16.4.7.
2. SECONDARY JET. The secondary-jet contribution is

SPLs (θ, f ) = SPL(Vs , Ts , ds , θ, f ) + 10 log10 Fu ( f1 , f ). (16.46)
In Equation 16.46 the function Fu is the complement of Fd and represents the fraction
of the energy radiated upstream of the axial position x1 :
Fu (x1 , f1 , f ) = 1 − Fd (x1 , f1 , f ) (16.47)
and denotes the fraction of the energy radiated downstream of the position x1 .
3. EFFECTIVE JET. The effective-jet contribution is

SPLe (θ, f ) = SPL(Vp , T p , de , θ, f ) + dB, (16.48)
with the effective diameter de calculated from
de = dp (1 + Ar2 Vr )1/2 . (16.49)
Figure 16.14. Nomenclature for single

jet.
The last term on the right-hand side of Equation 16.48 is a noise reduction that
depends on the level of relative turbulence. This is in fact the ratio between the
average turbulence level in the co-axial jet and the primary jet and was found to be
in the range of ∼2/3. The actual SPL reduction derived in Ref. 33 is

1 + 16y
dB = 10 log10 0.2 , (16.50)
1 + 7y
with
(τ − 1)2 M2j Vj Tj
y= , Mj = , τ= . (16.51)
1 + 0.65(τ − 1) ao To
16.4.7 Far-Field Noise from a Subsonic Circular Jet

The method of ESDU 34;35 , based on semi-empirical data, is used in this context for
the determination of the noise from a single-stream circular jet. This term appears
in the right-hand side of Equation 16.41, Equation 16.46 and Equation 16.48. The
OASPL corresponding to a frequency f at a receiver position as indicated in Fig-
ure 16.14 is
ω
ρj Aj po
SPL = SPLn + 10 log10 + 10 log10 + 20 log10 , (16.52)
ρo r2 pI SA
where SPLn denotes a normalised SPL, Aj is the area of the jet, and ω is a variable
density index; po is the atmospheric pressure and pI SA is the atmospheric pressure
under standard conditions. Thus, the second term of Equation 16.52 is a density
correction, the third term is a form factor and the fourth term in the right-hand side
is a correction for atmospheric conditions. The normalised SPL, the variable density
index, the normalised spectra and other quantities are calculated from charts. The
data are limited to polar-emission angles between 30 and 120 degrees; therefore,
calculation of noise from a distant source at a small or large emission angle must be
done by careful extrapolation. The reader is referred to the original documents for
details.
One further important point in the formulation of Equation 16.52 is that only
the SPL is given. To calculate the spectrum pressure levels, the method relies on
an experimental database. This database allows the calculation of the 1/3 octave
band SPL for a given set of Strouhal number, jet temperature and speed. We have
used the database through a complex interpolation method consisting in three main
steps: 1) a cubic spline through the Strouhal number; 2) a cubic spline through the
emission angle; and 3) a bilinear interpolation through the temperature ratio and
the jet velocity.
The theory presented was developed for a static engine. In our case, the engine
is almost never static but flies with a true air speed V. The velocity ratio Vr in the
co-axial jet model should not be affected by the flight speed to any large degree.
However, the speed of the single jet with respect to a still air must be corrected to
Vj − V if Vj > V. If Vj < V, the model does not provide a reliable estimate of the
noise and the SPL is set to zero. In fact, Vj − V < 0 causes a singularity in all of the
log terms; cases in which this event may occur include en-route descent with idle
thrust.
The aero-thermodynamic data for the jet noise calculation are (T , p, V) for both
primary and secondary jets. If the area ratio is not known, use the approximation
dcore dcore
0.46, 0.48, (16.53)
dbypass dfan
that has been inferred from the analysis of large engines with by-pass ratios BPR > 5.
The fan diameter is generally a known parameter, so it is possible to calculate the
remaining quantities.
Co-axial Jet-Noise Verification

The verification of the model is done with the data available in Fisher et al. 32;33 .
The data were taken by arrays of microphones placed at 12 m from the nozzle and
at polar-emission angles from 30 to 120 degrees. Data are available for a range of
primary jet temperatures and for a range of speed ratios. These data have been
interpolated from Ref. 33 and are affected by digital interpolation error of up to 0.5
dB (the original datasets were not available).
The nozzle dimensions of the test cases are dp = 33.2 mm; ds = 58.2 mm. We
have considered the case having a primary-jet temperature T p = 800 K and have
further assumed that the secondary jet has an atmospheric temperature Ts = 288
K (this datum was not readily available from the published data). The primary and
secondary jet velocities are 265 m/s and 170 m/s, respectively. The reference polar
angles are 40, 90 and 120 degrees.
The predictions have been extrapolated to the lowest frequencies, where the
database does not provide useful information on the jet noise. In many real cases
the temperature ratio T p /Ts is greater than 2.5, which is above the limit of the
temperature database. The results indicate that the extrapolation is not very good,
with errors in excess of 5 dB at frequencies below 100 Hz. The comparisons shown
by Fisher et al. have a lower cut-off frequency of about 300 Hz, depending on the
test case.
Figure 16.15 shows the spectral analysis of the calculated co-axial jet noise and
the comparison with the reference data. The squares have approximately the same
size as the error bars on the reference SPL.
16.4.8 Stone Jet Noise Model

The co-axial jet noise described previously is accurate enough for implementation
into a comprehensive flight-mechanics program but lacks generality due to the lim-
itation of the experimental database on which it is founded. This database has a
90 90
80 80
70 70
SPL, dB
SPL, dB
60 60
50 50
Reference data Reference data
40 40
2 3 4 5 2 3 4 5
log10 f log10 f
(a) θ = 40 degs (b) θ = 90 degs
90
80
70
SPL, dB
60
50
Calculated
Reference data
40
2 3 4 5
log10 f
(c) θ = 120 degs
Figure 16.15. Co-axial jet noise; Tp = 800 K, Vp = 265 m/s, Vs = 170 m/s.
relatively limited range of velocity ratios Vp /Vs and jet velocity ratios. At take-off
and high-power climb, it is likely that the jet operation point exceeds by far the
matrix of the database. An alternative model is described next to overcome these
limitations (Stone et al. 36 ). The model is a semi-empirical prediction method for
both single- and dual-stream nozzles derived from an analytical approach using real
physical scaling laws; its empirical database covers a wide range of engine size.
The dual-stream model will be used in our applications, which are on commercial
transport aircraft with high-bypass-ratio turbofan engines.
In this model, jet noise is decomposed into multiple components of subsonic
and supersonic noise sources based on their flow-length scales and source locations.
The subsonic noise sources come from large-scale, small-scale and transitional (or
intermediate) turbulent mixing structures. In addition, a jet nozzle with a plug would
generate noise as the inner-stream flow separates from the plug. When at least one of
the exhaust streams is supersonic, shock noises are generated from either or both of
the inner and outer streams as well as downstream of the plug. In our implementation,
only the subsonic noise sources are needed.
The model has been validated over a wide range of cases, including applications
on supersonic business jets 37 . Issues pertaining to quality experimental facilities to
obtain accurate jet-noise measurements and minimise upstream noise have been
addressed in the model following the work of Ahuja 38 . This contaminated upstream
noise typically comes from upstream sources unrelated to the jet noise (i.e., valves,
elbows, obstructions or the combustor).
Theory
In this section, models for each of the noise components for subsonic exhaust con-
ditions are presented. Overall sound levels uncorrected for refractions (UOL) are
presented for the large-scale mixing regions (denoted with underscore [.] L), for
the small-scale mixing regions, for the transitional and intermediate scale regions
(denoted with underscore [.]T and underscore [.] S ) and the plug-separation noise
(denoted with underscore [.] P ). SPL measured in one-third octave-band scales are
extracted from the overall sound levels through empirical correlations as a function
of an effective directivity angle θ and the logarithmic Strouhal number log(St), a
function of the frequency f . Examples of the correlation procedures are given in
§ 16.4.8. Finally, these SPLs are corrected based on the actual locations of the noise
sources.
LARGE-SCALE MIXING NOISE. The large-scale mixing noise consists of low-frequency

noise signatures; it is the dominant noise component for the composite jet noise if
shock noise components are not present. Its uncorrected overall sound level UOL L
and the corresponding normalised value UOLL are obtained with the following
relation:

UOL L = UOLL − KL + 10 log p̂2 + log ÂL + ω L log ρ̂ L . (16.54)
The UOL normalisation is done with respect to the flow and geometry parameters
and follows a lengthy formulation. The term KL is a function of a corrected Mach
number and a corrected emission angle; ÂL is a corrected nozzle area. These and
other parameters in Equation 16.54 are defined through a dozen algebraic equations.
Their derivation would require a lengthy discussion. The reader is referred to the
original paper 36 ; all of the terms within the parentheses in the right-hand terms are
normalised quantities and are denoted with a hat [.]. ˆ
Once the UOL L is obtained, the SPLs for each one-third octave-band centre
frequencies are correlated from an empirical database for the large-scale mixing
noise using the following two parameters:
0.1
VL
θ L = θcor,L , (16.55a)
a∞
0.4(1+cos θL)
f dL Tt,L
log St = log . (16.55b)
Ve,L Tt,∞
Corrections to the directivity angle θcor,L and the source-to-receiver distance rcor,L
requires the source location Xs,L of the region generating the noise, where

θ
Xs,L = L1 + 4 + dL (PR1 PR2 )−1/4 /LSF, (16.56)
30
with θ in degrees; PR1 and PR2 are noise-suppression parameters for the inner and
outer stream nozzles, respectively; LSF is the linear scale factor (ratio of full-scale
to model-scale dimensions). L1 is a length scale. These parameters are defined as
the ratios of the wetted perimeters for the suppressed to the unsuppressed nozzles.
Jet nozzles can be modified with triangular-serrated chevrons at the trailing edges
to suppress mixing noise 39;40 .
SMALL-SCALE MIXING NOISE. The small-scale mixing noise, also termed the outer
shear layer mixing noise, occurs at relatively high frequencies, generated near the
nozzle exit. Its uncorrected overall sound level UOL S is given by

UOL S = CS − KS + 75 log V̂e,S + 10 log p̂2 + log ÂS + ω L log ρ̂ S , (16.57)
where CS is a noise-component coefficient, given in dB. The other parameters are

defined in the original paper 36 by twelve algebraic equations.
Similar to the procedure for the large-mixing noise, the SPLs for the small-scale
mixing noise for each one-third octave-band centre frequencies are correlated from
a relevant empirical database mapped from the following two parameters:
0.1
VS
θ S = θcor,S , (16.58a)
a∞
0.4(1+cos θS )
f dS Tt,S
log St = log . (16.58b)
Ve,S Tt,∞
Corrections in the SPL, directivity angle and distance are based on the following
definition of the source location:

θ
Xs,S = dS /LSF. (16.59)
45
TRANSITIONAL/INTERMEDIATE-SCALE MIXING NOISE. The transitional or inter-

mediate-scale mixing noise is primarily composed of middle-to-high frequency noise.
Unlike the large and small-scale mixing noise, the effect of the density ratio for the
transitional-scale mixing noise was found to be negligible. The uncorrected sound
level UOLT is defined as

UOLT = CT − KT + 75 log V̂e,T + 10 log p̂2 + log ÂT , (16.60)
where the parameters in the right-hand side of the equations are calculated through
nine algebraic equations (see again the original paper).
Correlations with an empirical database to obtain the one-third octave-band

SPL are mapped using the following parameters:

Ve,T 0.1
θT = θcor,1 , (16.61a)
a∞

f d2,1,th Tt,1 0.4(1+cos θT )
log St = log , (16.61b)
Ve,T Tt,2
where d2,1,th is the outer physical throat diameter of the inner-stream nozzle. Correc-
tions in the SPL, the directivity angle and the source-to-receiver distance are based
on the following definition of the source location:

θ
Xs,T = L1 + dh,1,th /LSF, (16.62)
45
where dh,1,th is the hydraulic diameter of the inner-stream throat.
INNER-STREAM PLUG-SEPARATION NOISE. A jet nozzle with a blunt-tipped plug would

produce a noise when the inner-stream separates from the plug. This noise is rel-
atively important for an engine with high by-pass ratio and low mixed-jet velocity.
The following relationships determine the uncorrected sound level UOL P for the
plugged-tip nozzle:

UOL P = UOLP − KP + 10 log p̂2 + log ÂP + log ρ̂ P , (16.63)
where the right-hand parameters are calculated via a number of algebraic equations,
like in the previous cases 36 . Correlations with an empirical database to obtain the
one-third octave-band SPL are mapped using the following parameters:

V1 0.1
θ P = θcor,P , (16.64a)
a∞
) *
log St = log 0.5 f dP / V1 a∞ (1 − M cos θcor,P ) , (16.64b)
where r P is the physical plug-tip radius of the inner-stream nozzle. Finally, the SPL
correction is derived from the following source location:

θ
Xs,P = L1 + LP + 2 r P /LSF. (16.65)
45
Empirical Correlations
Once the uncorrected sound levels have been determined for each of the noise
components, the one-third octave-band SPL can be derived through correlations with
the empirical database mapped using two parameters: the effective polar directivity
angle θ and the logarithmic Strouhal number log(St). The SPL obtained is the
free-field SPL at the receiver location. The data are stored as SPL − UOL. Selected
plots of the empirical data for the large-scale, small-scale, transitional and plug-
separation noise are shown in Figure 16.16. In the detailed report by Stone et al. 36 ,
θ covers a range from 0 to 250 degrees only, mostly at intervals of 10 degrees
except for larger angles. The logarithmic values of the Strouhal number for all of the
noise components run from −3.6 to 3.6. Extrapolations are necessary for conditions
beyond the recorded data.
(a) Large-scale mixing noise (b) Small-scale mixing noise
(c) Transitional-scale mixing noise (d) Plug-separation noise

Figure 16.16. Database of SPL − UOL for selected θ between 0 and 250 degrees, as shown
in the legend (data extracted from Stone et al. 36 ).
Sound-Level Corrections
The SPL obtained from the empirical database must be corrected due to the varying
source locations of the individual component noise. With this regard, the source-to-
receiver location and the polar- directivity angle must be corrected from their initial
values taken relative to the centre of the nozzle exit plane. The corrected, or actual,
values are derived from the following relationships:

2
Xs Xs
rcor = r 1 + +2 cos θ , (16.66a)
r r

−1 Xs + r cos θ
θcor = cos . (16.66b)
rcor
100 1,000 10,000 100,000 100 1,000 10,000 100,000
(a) θ = 90◦ (b) θ = 120◦
100 1,000 10,000 100,000 100 1,000 10,000 100,000
(c) θ = 135◦ (d) θ = 150◦

Figure 16.17. Comparison between the total jet noise SPL from actual data and the predicted
component jet noise SPL with V1 = 0.98 and M = 0.05 (data extracted from Stone et al. 36 ).
From these corrected parameters, the actual free-field sound pressure level (SPLa )
is computed as
r
cor
SPLa = SPL − 20 log . (16.67)
r
Model Verification
Limited validations of the Stone jet model have been performed on model-scale
dual-stream nozzles 36 . Important results from this study are shown in Figures 16.17
and 16.18. The experimental data highlight the dominance of the large-scale mixing
noise over the other noise components when the exhaust conditions are subsonic.
Higher jet velocity was found to increase the contributions of the transitional and
small-scale mixing noise; in one case, the peaks of both the large-scale and the
transitional-scale noise are comparable, as shown in Figure 16.18(a).
In these limited datasets, predictions from the Stone jet model are found to
provide satisfactory results relative to the actual data, particularly when the jets
are near the overhead locations, that is, for θ between 90 and 120 degrees. For
larger values of the polar angles, the actual peak SPL are over-predicted by about
1–2 dB.
100 1,000 10,000 100,000 100 1,000 10,000 100,000
(a) θ = 95◦ (b) θ = 139◦

Figure 16.18. Comparison between the total jet noise SPL from actual data and the predicted
component jet noise SPL with V1 = 1.74 and M = 0.0 (data extracted from Stone et al. 36 ).
Sensitivity Analyses
Sensitivity analyses are performed to identify critical input parameters for the simu-
lation of the jet noise. In these analyses, only the parameters relevant to the primary-
and secondary-jet properties are studied: these are the jet velocities V1 and V2 , the
nozzle areas A1 and A2 , the total temperatures Tt,1 and Tt,2 , and the mass-flow rate ṁ1
and ṁ2 . The nominal input parameters for this case study are taken from that of an
Airbus A320 aircraft at its closest encounter with a receiver located at a FAR point
during take-off. The secondary stream is thus the exhaust flow from the turbofan.
The overall jet noise is found to be most sensitive to variations in both the
primary- and secondary-jet velocities. If the primary-jet velocity is changed by ±10%
of its normal operating value, the overall sound level OASPL changes by at most
2 dB. A 20% change to V1 would increase the OASPL by 4.6 dB. Most of these
changes occur in the high-frequency regime of the noise spectra. Conversely, the
secondary-jet stream influences the lower-frequency noise, with an increase in the
OASPL of 3.3 and 6.7 dB if V2 is increased by 10% and 20% of its initial value,
respectively. Figure 16.19 illustrates the effect of both jet streams on the SPL in the
one-third octave-band scale.
The other input parameters have only minor effects on the overall sound level
of the jet noise. Of these inputs, the primary-jet nozzle has the most influence, with
a change of OASPL equal to 0.4 or 0.9 dB if A1 is enlarged by 10% or 20% of its
original size, respectively. A closer inspection of its effect on the one-third octave-
band noise spectra reveals that the changes influenced by the primary nozzle area
occur mostly on the high-frequency noise. Variations in the turbofan area also affect
the high-frequency noise with minor but noticeable influence on the low-frequency
noise. However, the overall effect on the OASPL by the turbofan area is small, with
an increase of 0.2 dB if the area is enlarged by 20%. Figure 16.20 shows the sensitivity
analysis on the SPL due to the variations of the nozzle areas.
16.4.9 Jet-Noise Shielding

When multiple jets are present, an interference event takes place called jet shielding.
When propagating through a high-temperature and high-speed jet, an acoustic wave
10 100 1,000 10,000 10 100 1,000 10,000
(a) Varying V1 (b) Varying V2

Figure 16.19. Changes in the one-third octave-band SPL due to variations in the input speeds
V1 and V2 , whose initial values are 400 m/s and 275 m/s, respectively.
encounters a different impedance that would tend to absorb, reflect and diffract the
incident wave. The jet-shielding mechanisms are the following:
r Sound waves from a jet may be reflected, refracted, diffracted, scattered (due to
flow turbulence) or fully transmitted upon encounter with another jet that acts
as a shielding jet.
r The acoustic shielding level of the source depends on the location of the receiver
with respect to the dual-jet system. Because the total acoustic power is conserved
within a sphere enclosing the source, a noise reduction at one receiver location
as a result of jet shielding would result in a noise increase at some other location.
For example, a receiver standing in the jet axes plane might be completely or
partially shielded from the noise of the jet farthest away by the adjacent jet
10 100 1,000 10,000 10 100 1,000 10,000
(a) Varying A1 (b) VaryingA2

Figure 16.20. Changes in the one-third octave-band SPL due to variations in the input speeds
A1 and A2 , whose initial values are 0.6 m2 and 1.3 m2 , respectively.
Figure 16.21. Nomenclature for jet-shielding noise model.
located in between. However, these shielded sound waves are actually deflected
toward another location by the shielding jet; these deflected sound waves would
then superimpose with other sound waves to increase the noise level at the new
location.
r Constructive or destructive phasing between multiple coherent sound waves
(i.e., between direct and deflected sound waves) may occur to increase or
decrease the noise level at a receiver.
If P1 and P2 denote the sound power at a receiver in absence of noise shielding,

and PS is the resulting sound power due to shielding, the difference in noise is

PS2
SPL = 10 log10 . (16.68)
P12 + P22
An analytical formulation of noise shielding is due to Gerhold 41 . Simonich et al. 42

performed experiments on dual jet noise and provided a refined mathematical model.
ESDU 43 (and associated documents) give a practical implementation that is affected
by numerical errors at large frequencies.
Consider the situation shown in Figure 16.21. The two jets are identical and
have a nominal diameter d j ; their axes are separated by a distance ro. The distance
between the observer and the centre of the farthest jet at the nozzle is called r ;
r denotes the distance from the closest jet. The angle β in this case denotes the
polar angle measured from the plane normal to the jet axis; likewise, denotes the
azimuthal angle on a plane normal to the jet axis.
Note that when the receiver is sufficiently far from the jets, the difference in
angles and distances becomes negligible. In the following analysis, the subscript “o”
denotes the conditions outside the jet (atmospheric conditions) and “1” denotes the
conditions inside the jet (high temperature).
Some complications arise if the airplane performs a three-dimensional trajectory
whilst banking. In this case, we consider the vector r between the mid-point between
the nozzles and the receiver. If the jet axes are aligned with the air speed vector V
(i.e., the thrust line is parallel to the air speed and tangent to the flight trajectory),
then the plane normal to the jets is defined uniquely by V. Thus, the vector r projected
onto the plane is r n and the angle β is defined by

r · r1
β = cos−1 , (16.69)
|r| · |r n |
with
r n = (V × r) × V. (16.70)
The angle ψ is

−1 zr − z j
ψ = cos ± ϕ, (16.71)
|r n |
where the effect of the bank angle ϕ depends on whether the banking/turning is done
toward the receiver zr (closest jet down) or away from it (closest jet up).
The key idea of the model is to express the acoustic pressure p in terms of a
velocity potential , which is done through the relationship

d
p = −ρo . (16.72)
dt
Because the potential is a complex number (as discussed herein), also the acoustic
pressure from this equation is a complex number. Expanding from this result, the
difference in SPL is estimated from
2
p ||
SPL = 10 log10 = 20 log10 . (16.73)
po |o|
Equation 16.73 is only valid to compute the SPL of a single jet shielded by another
jet with reference to the single unshielded jet. To compute the realistic effect of
a dual-jet system with shielding effect on a receiver at any specific location, the
sound power of each jet shielded by the adjacent jet has to be treated separately.
The SPL is then calculated by superimposing the sound power of both jets with
shielding operative and normalising it with the total power of both jets operating
individually without any shielding effect. If jet “1” is the jet closest to the receiver
and jet “2” is the one farther away, we call
r p12 the power spectral density for jet 1 shielded by jet 2.
r p22 the power spectral density for jet 2 shielded by jet 1.
r 2
po1 2
, po2 the total sound power of unshielded jet 1 and jet 2, respectively.
r For a discrete tone, we have:

p12 + p22
SPL = 20 log . (16.74)
2
po1 + p02
2
The quantity p2 is the square of the RMS pressure p2 , which must be computed
if one were to physically record the pressure signal of a discrete tone in real time. In a
numerical computation, the ratio of the squares of these RMS pressures for discrete
tones is equivalent to the ratio of the squares of the pressure magnitude | p|2 . In
the formulation, each power spectral density term ( p2 ) is written by acknowledging
that it is the square of the pressure magnitude | p|2 . This SPL is only for a single
frequency because each sound power is actually the power spectral density (i.e.,
sound power for a single frequency, or discrete tones). To obtain the one-third-
octave bands sound power or SPL, the results associated with the frequencies
of smaller sub-bands within each one-third-octave band have to be integrated and
normalised appropriately.
Because the undisturbed air density ρo is a known quantity, the problem now
focuses on the determination of a suitable potential function. The theory gives the
velocity potential due to a unit source at a distance r , corresponding to a wave
number ko = ω/ao:
∞
1 −iωt
= e m cos(m) exp(−imπ/2) exp(ikor ) Fm(β). (16.75)
4πr
m=0
This is a cumbersome expression that requires further specification. First, the factor
m has the following values: = 1 for m = 0, = 2 for m > 0. The complex function
Fm of the polar angle β is
Fm1
Fm(β) = Jm(xo) − (16.76)
Fm2
with

Fm1 (β) = Hm(xo) ρ1 a12 T12 cos β Jm(x1 )Jm (x2 ) − ρoao2 T2 Jm(x1 )Jm (x2 )
(16.77)
Fm2 (β) = ρ1 a12 T12 cos β Jm(x1 )Hm (x2 ) − ρoao2 T2 Hm(x1 )Jm (x2 )
where, in order, we have:
r J = m-th order Bessel function of the first kind
m
r H = m-th order Hankel function of the first kind
m
Properties of these complex functions are given in most textbooks of mathem-

atical special functions. Furthermore, we have the following definitions:
T1 = ao/a1 − M sin β, with M = Vj /ao (16.78a)
T2 = (T12 − sin2 β)1/2 (16.78b)
xo = koro cos β (16.78c)
x1 = kor j T2 (16.78d)
x2 = kor j cos β (16.78e)
Jm (x2 ) = ∂ Jm/∂ x2 = ∂ Jm/∂β (16.78f)
Hm (x2 ) = ∂ Hm/∂ x2 = ∂ Hm/∂β (16.78g)

The velocity potential due to a unit source in absence of the shielding jet is
∞
1 −iωt
o = e m cos(m) exp(−imπ/2) exp(ikor ) Jm(xo). (16.79)
4πr
m=0
Both velocity potentials are complex numbers in the form
z(t) = |z|y.
Such complex numbers describe a signal in the time domain. Now all of the elements
are in place for the calculation of the shielding effects, according to the procedure
discussed here.
The jet-shielding formulation is valid for all azimuthal angles (0 < β < 360
degrees) and for all polar-emission angles θ ±20 degrees away from jet axis (i.e.,
20 < θ < 160 or equivalently, −70 < β < 70).
Computational Procedure
r Set jet characteristics, geometrical arrangement and position of the receiver.
r Set atmospheric conditions.
r Set frequency (or the wave number).
r Calculate the parameters given in Equation 16.78; the derivatives are calculated
numerically* .
r Calculate the factor F (β) from Equation 16.76.
m
r Calculate the shielding potential from Equation 16.75 via a fast Fourier trans-
form.
r Calculate the isolated potential from Equation 16.79 via a fast Fourier trans-
form.
r Calculate the change in SPL from Equation 16.73.
r (The calculation can be repeated for other wave numbers and receiver posi-
tions.)
There are a number of methods to simplify the integration and normalisation of

the results. The exact approach would be to calculate the total sound power within
each one-third-octave band by integrating the power spectral densities of its smaller
sub-bands multiplied by the widths of the sub-bands. The total sound power (with
shielding operative) of each band is then normalised with the bands’ total sound
power of both jets without shielding. The resultant SPL formulation would be as
follows:

Ps
SPL = 10 log . (16.80)
Pt

ns

Ps = wb p12 + p22 i , (16.81)
i=1

ns
2
Pt = wb po1 + po2
2
i
, (16.82)
i=1
* There are a variety of open-source programs for the calculation of the Bessel functions and their
derivatives.
f = 0.5 kHz 90 r =3d f = 1.0 kHz 90 r =3d

β = 50 degs r =5d β = 50 degs r =5d
120 60 120 60
θ = 40 degs θ = 40 degs
150 30 150 30
Ψ Ψ
180 0 180 0
-2 -1 0 1 -2 -1 0 1
Δ SPL, dB Δ SPL, dB
210 330 210 330
240 300 240 300

270 270
(a) f = 0.5 kHz (b) f = 1.0 kHz

Figure 16.22. Jet-shielding effects.
where Ps and Pt are the shielded and unshielded total sound pressure power, respect-
ively, for each 1/3 octave band, obtained by summing the power spectral densities
of both jets multiplied by their corresponding sub-bandwidths across the 1/3 octave
bandwidth. Here, ns is the number of sub-band partitions within each 1/3 octave
band, i is the i−th sub-band partition within a 1/3 octave band, and wb is the sub-
bandwidth.
Typical values of the ratio d j /ro are as high as 0.215 for the Boeing MD-90 and as
low as 0.145 for the Boeing B777-300. Thus, it is reasonable to assume that practical
values of this parameter are included within 0.15 and 0.22. Figure 16.22 shows the
jet shielding effects at increasing azimuthal angles φi , the effects of frequency, polar-
emission angle and distance between jet axes are shown. Figure 16.23 shows the jet
shielding as a function of the polar-emission angle.
4 4
0 0
Δ SPL, dB
Δ SPL, dB
-4 -4
-8 -8
Theory Theory
ro = 3dj Invalid
ro = 3dj Invalid
ro = 5dj ro = 5dj
-12 -12
90 60 30 0 90 60 30 0
θ, degs θ, degs
(a) f = 0.5 kHz (b) f = 1.0 kHz

Figure 16.23. Jet-shielding effects.
16.5 APU Noise

The auxiliary power unit (APU) causes cabin noise as well as noise on the ground,
with airplanes at the gate or taxiing in and out of the ramp; thus, it affects personnel
on the ground as well as communities around airports. At some airports, regulations
require that APU and the environmental control systems be shut down 5 minutes
after arrival and be started no earlier than 5 minutes before the estimated time of
departure; exceptionally (in cold or hot weather), the use of the APU is extended to
15 minutes* . Consequently, the airline operator is required to hook into an electrical
power-supply system (as part of the ground services).
The APU itself is a gas-turbine engine located in a compartment inside the
fuselage tail. The APU provides electrical power as well as compressed air and works
like a turboshaft engine (refer to § 5.7). Among the various APU noise sources,
exhaust noise and combustion noise are dominant for most airplanes. A study in
APU combustion noise was done by Tam et al. 44 . These authors demonstrated that
the peak combustion SPL occurs in a narrow band, 250 to 350 Hz. However, the
exact noise level of the measurements was unpublished.
For receivers beyond the immediate vicinity of the aircraft, the estimation of the
noise signature must rely on an accurate model for noise scattering and reflection,
as well as a detailed description of the horizontal and vertical obstacles between
the source and the receiver. For receivers in the immediate vicinity, the use of a
ground-reflection model should be sufficient.
The aero-thermodynamic data needed by the model include inlet and exit com-
bustor temperatures, inlet pressure, mass-flow rate and atmospheric conditions.
Even this limited set of data is difficult to gather. One of the few options avail-
able to realistically collect these data is to stand directly behind the airplane and
take measurements† . As an alternative, we propose the following method:
r Fuel flow: To estimate the APU fuel flow, use the data in Table 5.5. If the APU
is not in the table, extrapolate the fuel flow on the basis of the number of seats
in the airplane.
r Mass-flow rate: Estimate the APU mass-flow rate by assuming an average value
of the fuel-to-air mixing ratio. In absence of accurate data, use ∼ 0.025 to 0.028.
Thus, an APU burning ṁ f ∼ 100 kg/h of fuel would have a mass-flow rate ṁ ∼
0.9 to 1.0 kg/s.
r Combustor inlet pressure: This quantity should be derived by the overall com-
pression ratio, which is generally unknown. In absence of other data, assume
OPR ∼ 14 to 16.
r Combustor temperatures: The inlet temperature can be calculated from the
OPR. The output temperature is calculated with an energy equation in the
control volume defined by the combustor.
* Each airport has its own regulations. Specific and updated data can be checked on the Internet; key
words “airport noise and emissions”.
† According to the FAA, the APU shall not have its own type certificate, because it must be considered
part of the equipment in the aircraft. Therefore, there is no obligation to disclose any data of these
power systems. The European Safety Agency (EASA) issues standards (CS-APU) for “certification
specifications”, which is what is needed for basic engineering simulation.
Figure 16.24. Reference systems for airframe noise models.
r Exhaust gas temperature (EGT): This is given in the Type Certificate Data
Sheet. If this document is not available, use EGT ∼ 850◦ C (starting) and EGT
∼ 550◦ C (continuous).
r Exhaust gas velocity: If d is the nozzle’s diameter, then V ∼ 4ṁ/d2 ρ . Assume
j j j j
a fully expanded jet (ρ j = ρ) to simplify the analysis.
With the quantities listed here, it is possible to estimate the APU noise.
16.6 Airframe Noise

We now proceed with the analysis of selected airframe-noise components. Some
models that rely on flow physics are available to predict the noise from simple
geometries. However, the problem is still dealt with by using a considerable amount
of empiricism. In our case, we generally rely on approximate geometries that include
few details beyond the planform of the component (by itself approximate) and the
deflection angle.
The basic nomenclature is shown in Figure 16.24, which displays the position of
the source (interpreted as the position of the aircraft’s CG), the generic position of
the receiver, the polar-emission angle θ and the azimuth angle φ. The polar-emission
angle is

−1 r · V
θ = cos (16.83)
Vr
and must be corrected for Mach number effects as explained in § 16.3.1.
The azimuth angle φ is calculated by first taking the minimum distance r1
between the flight path and the receiver. The altitude of the source at this point
is h. The uncorrected azimuth angle is

h
φ = cos−1 . (16.84)
r1
When the receiver is just below the flight path, as in the FAR/ICAO take-off and
landing reference positions, then φ = 0 at all times.
The method proposed by Fink 45 , implemented in the noise code ANOPP 46 and
adopted by ESDU 47 , is based on the following semi-empirical equation for the noise
from lifting surfaces:

ρ2a4 p
SPL = 10 log p + 10 log
2
2
− 20 log . (16.85)
pref po
The first contribution in Equation 16.85 is due to the mean-square acoustic pressure
p2 . This quantity is written as
Pb2 DF
p2 = , (16.86)
4πr 2 (1 − M cos θ)4
where P is the acoustic power, b is the wing span (or a cross-wise characteristic
length), D is the acoustic directivity function, and F is a spectrum function of
the Strouhal number. At the denominator of Equation 16.86, there is a spherical-
propagation factor 4πr 2 and a source-amplification factor. The product Pb2 is a
function of the flight Mach number and several empirical parameters and has the
form:
Pb2 = k1 k3 Mk2 , (16.87)
where the coefficient k2 depends on the geometrical configuration of the item. The
directivity function in Equation 16.85 depends on the polar-emission angle θ and the
azimuth angle φ. The Strouhal number involved in the method is in fact a “corrected”
Strouhal number, given by the following expression:
fl
St = (1 − M cos θ), (16.88)
V
where f is the frequency, l is a characteristic length and V = Ma is the true air speed.
For all the lifting components, the corrected Strouhal number is

S V S −1/5 (1 − M cos θ)
St = 0.37 f , (16.89)
b ν b V
where S and b are the surface and the spanwise characteristic length of the item.
Hence, the ratio S/b is a streamwise length.
The third term in the formulation of Equation 16.85 is a correction due to
the difference in atmospheric pressure between source and receiver locations ( p =
pressure at source; po = pressure at receiver).
16.6.1 Wing Noise

Equation 16.85, in addition to being empirical, gives the value of an SPL that is not
dependent on whether or not the lifting surface produces a lift. This is in contrast
with Lilley’s theory 48;49 of airframe noise, which is discussed next.
In many analyses, following the models of Ffowcs-Williams and Hall 50 and
Howe 51 , the wing is modelled as a semi-infinite flat plate in turbulent flow. For a
wing of high-aspect-ratio, the problem is associated with the scattering of turbulent

kinetic energy from the trailing edge. The noise intensity just below an aircraft flying
at an altitude h above the ground, with a speed V, is found from

1.7 ρ AV 3 M2 uo 5 δ 1.7 ρ AV 5 uo 5 δ
I= = , (16.90)
2π 3 h2 V δ ∗ te 2π 3 a 2 h2 V δ ∗ te
where uo is the characteristic speed for the turbulent flow; δ and δ ∗ are the bound-
ary layer thickness and boundary-layer displacement thickness at the trailing edge,
respectively. Equation 16.90 ignores the presence of other systems (fuselage, tail
plane, engines) and the sweep of the trailing-edge line, which would require a factor
cos3 te . Finally, the noise intensity at aircraft positions other than the fly-over must
account for a corrective factor sin2 (θ/2). The corrected sound intensity is

1.7 ρ AV 5 uo 5 δ
I= sin2 (θ/2) cos3 te . (16.91)
2π 3 a 2 h2 V δ ∗ te
The ratio (δ/δ ∗ )te depends on the state of the boundary layer at the trailing edge.
For a turbulent boundary layer, this ratio is at least equal to 8. We take the value
(δ/δ ∗ )te 10, which should account for turbulent flow with some separation at rel-
atively high CL. This ratio, along with the turbulent characteristics uo/V, can be
calculated more precisely with modern computational fluid dynamics programs for
the three-dimensional wing. The air speed V in Equation 16.90 is found from the
definition of CL,

1.7 ρ AV 3 2W uo 5 δ
I= , (16.92)
2π 3 a 2 h2 ρ ACL V δ ∗ te
17 V 3 W uo 5
I , (16.93)
π 3 a 2 h2 CL V te
3
V W
I c1 , (16.94)
h2 CL
with
17 uo 5
c1 = . (16.95)
π 3 a 2 V te
This coefficient will depend on the local atmospheric conditions (speed of sound,
altitude), the Reynolds number and the geometry of the aircraft; c1 has a nearly
constant value that fits remarkably most experimental data collected in the past 30
years, for systems as diverse as wide-body aircrafts and birds. Typical values of the
parameters in Equation 16.95 are uo/V 0.1, a 340 m/s (sea level). With these
values we find c1 4 · 10−11 .
What happens if the CL is increased? High lift is generally associated to large
suction peaks on the upper surfaces of the wings, which trigger instability in the
boundary layer. This issue is a complex matter for computational aerodynamics.
CFD analysis, as discussed by Lockard and Lilley 49 , indicates that the average CL is
related to the turbulent quantities by the approximate equation
u 5 δ
1
4
o
= 1 + CL2 , (16.96)
V ym te 4
100
90
80 B-747-400
W = 300 tons
Theoretical lower bound CL = 0.6
OASPL, dB of airframe noise h = 150 m
M = 0.32
70
60
50
40 2 3 4 5 6
10 10 10 10 10
2 2
W U M / CL h
Figure 16.25. Theoretical lower bound of fly-over wing noise in level flight.
where ym is the length scale of eddies. Thus, the parameter ym replaces the displace-
ment thickness δ ∗ in Equation 16.92. In areas of adverse pressure gradients ym/δ ∗
0.2 to 0.3. Operating this substitution in Equation 16.92, these authors found

1.7 WV M2 1 2 4
I= 3 1 + CL , (16.97)
π CLh2 4
where CL is a mean value on the wing. Equation 16.97 does not take into account
the presence of partly deployed flaps. Equation 16.90 and Equation 16.97 show that
the sound intensity I depends on the factor
WV M2
F= . (16.98)
CLh2
Equation 16.97 represents the lower bound of aircraft noise, such as if it is created by
the airframe alone, without engines, under-carriage, tail plane and high-lift devices.
Figure 16.25 shows the trend of this lower bound in terms of the factor F. This factor
can be obtained for an infinite combination of aircraft weight, speed, lift coefficient
and flight altitudes.
The range of the factor F, Equation 16.98, covers five orders of magnitude. A
medium-size aircraft, such as the Airbus A300-600, with a fly-over altitude of the
order of 100 m (348 feet), will have F 6 · 103 . A drawback of this formulation is
that it does not provide the spectral content of the acoustic pressures.
16.6.2 Landing-Gear Noise

The determination of the landing-gear noise is one of the most difficult task in the
analysis of aircraft noise because the acoustic response results from the complicated
(a) Nose (b) Main

Figure 16.26. Typical landing-gear geometry showing main struts and other details.
geometry of the unit. In fact, there are various non-aerodynamic components (struts,
wheels) directly exposed to the airflow, along with the under-carriage bay (a cavity
with open doors). An example of landing gears is shown in Figure 16.26.
The only realistic option within a flight-mechanics model is to use validated
semi-empirical models. One of these has been proposed by Guo 52 and validated
with experimental data 53;54 . Experimental data on a 6.3%-scale-model landing gear
of the Boeing B777 (spectra and directivity) have been published by Humphreys
and Brooks 55 .
Following this method, the spectral characteristics of the landing gears depend
on three contributions: a low-frequency contribution, mostly due to the wheels; a
mid-frequency contribution, mostly due to the main struts; and a high-frequency
contribution, due to the small geometrical details of the unit. The overall acoustic
emission is then calculated by summing the incoherent noise energy of the spectral
components. Such a decomposition arises from a detailed analysis of experimental
and flight-test data. Each of the contributions has a different spectral response
and directivity. For each component, the method requires to define scaling factors,
reference lengths, Strouhal numbers and normalised spectral functions.
The mean-square acoustic pressure is written as a sum of contributions at low,
medium and high frequencies:
p2 = p2L + p2M + p2H . (16.99)
The landing-gear noise is
ρoao2 M6 A(r )D(θ )

p2 = {PL + PM + PH } , (16.100)
r 2 (1 − M cos θ )4
which includes the contributions at low, medium and high frequency; the corres-
ponding SPL in dB is calculated from

p2
SPL = 10 log10 2
. (16.101)
pref
Each of these acoustic pressure components (given within the brace brackets)
depends on the term:
P = β SD(θ )F (St), (16.102)
where St is the Strouhal number; β is a radiation-efficiency factor; D is a directivity
function of the polar-emission angle θ ; and S is the size effect of the component.
The model defined by Equation 16.100 and Equation 16.102 contains elements
of atmospheric conditions (density ρo, speed of sound ao, absorption A), a convective
amplification factor (1 − M cos θ)4 , a spherical propagation factor (1/r 2 ), a directiv-
ity D, a spectrum function F, a radiation efficiency β and a component dimension S.
It is convenient to remove the effect of atmospheric absorption, which is considered
separately (see § 17.2).
For the low-frequency component, the size effect is due mainly to the wheels.
The Strouhal number and the size effect are, respectively,
fd
St L = , SL = π Nw wd. (16.103)
V
For the medium-frequency component that is due mainly to the vertical struts, the
Strouhal number and the size effect are
fs
St M = , SM = si li , (16.104)
V i
where si is the perimeter of the cross-section of the i−th strut, li is its length, and s is
the average cross-section of the struts. The sum is extended to all of the main struts
of the assembly. The contribution at high frequency is
fl
St H = , SH = ηl 2 , (16.105)
V
where l is the length scale of the high-frequency noise. This length scale must be
of the same order as the geometrical features in the landing-gear assembly that are
responsible for turbulence noise (hydraulic hoses, wires, gaps, and so on). Clearly,
this quantity depends on the landing gear. However, an average value given by Guo
is l = 0.15s. The factor η is a complexity factor that is related to the high-frequency
noise. This factor is defined by the following equation:

Nw lt W Nw − 2
η = 1 + 0.028 −1 1+2 sin 2γt . (16.106)
Nref lref Wref Nref
In this equation, the reference values have been chosen as Nw = 2, Wref = 150,000
lb (68,040 kg) and lref = 300 inches (7.62 m); lt is the total length of the struts of the
under-carriage unit; γt is the wheel-track-alignment angle. The normalised spectral
functions are given by
Stς
F(St) = A , (16.107)
(B + Stμ )q
where A, B, σ , μ and q are semi-empirical coefficients. In particular, ς , μ and q define

the shape of the normalised spectrum and are calculated by fitting experimental data;
A and B normalise the function F = 1 at some value of the Strouhal number, Sto.
The directivity function is
D = (1 + hd cos2 θ )2 , (16.108)
where hd is another empirical parameter. A summary of all the parameters for this
model is given in Guo 52 . In absence of detailed geometrical data, the following
expressions can be used for a large airplane (OEW > 100 ton):
lmain = 3.71(lvstrut + laxle )main , s main = 0.01lmain
(16.109)
lnose = 2.85(lvstrut + laxle )nose , s nose = 0.024lnose .
For a smaller airplane, use
lmain = 3.45(lvstrut + laxle )main , s main = 0.015lmain
(16.110)
lnose = 3.71(lvstrut + laxle )nose , s nose = 0.027lnose .
In all cases, following the analysis shown by Guo 52 , the length scale of the small
details (responsible for the broadband noise) is
l = 0.15w. (16.111)
Finally, we must account for the installation effects of the landing gear. These effects
are essentially due to the reflection of acoustic waves from the wing and the fuselage.
Detailed models do not exist, but Guo provides a semi-empirical estimate for some
airplanes (Boeing B777) that correspond to a maximum of 0.8 dB increase in SPL
when the aircraft is overhead (θ = 90 degrees). This is achieved through a corrected
directivity function
D = 1.2(1 − 0.9 cos2 θ)2 . (16.112)
ESDU 47 uses the following expression for the mean-square acoustic pressure:
l 2 D(θ )F(St)
p2 = P. (16.113)
4πr 2 (1 − M cos θ )4
Apart from the inclusion of the atmospheric absorption, the two models are similar.
The calculation is done by using an additional empirical equation that correlates the
reference length l, the acoustic power P and the flight Mach number:
Pl 2 = c1 c3 Mc2 , (16.114)
where the coefficients c1 , c2 and c3 are dependent on the component.
The method presented is applicable to most isolated under-carriages, whether
they are main or nose units. A correction for the main under-carriage is suggested,
to take into account the fact that the inflow Mach number at the under-carriage
location is generally lower than the flight Mach number. A value of M = 0.75 to 0.80
should be sufficient to take into account the effects of the flow around the airplane,
in absence of more detailed data. However, more accurate information about the
flow around the landing gear could further improve the noise prediction.
100
M = 0.18
M = 0.24
90
SPL, dB
80
70
2 2.5 3 3.5 4
log10 f
Figure 16.27. Landing-gear noise of the the Boeing B737; polar-emission angle = 90 degrees.
Another point of concern is that although the installation effects are accounted
for by the term given in Equation 16.112, the model does not include the effects of the
under-carriage bays and their relative cavities and cavity doors. For some modern
airplanes, such as the Airbus A380, the cavities are partially closed by repositioning
the doors. This technique limits the exposure of the bays to the airflow and thus
reduces both aerodynamic drag and acoustic emission.
Landing-Gear Noise Verification

The method described has been validated for a number of cases. We now refer to the
Boeing B737 main landing-gear system, whose data are available in Ref. 53 . A full
breakdown of systems components and dimensions is given in the aforementioned
paper. We report here only the summary of data relevant for the noise calculation.
The noise measurements were taken on an isolated landing gear, with an array
of microphones placed on a line at about 10 feet (∼3 metres) from the centre of
the unit. Data are available for a range of Mach numbers (from M = 0.18 to M =
0.24) and a range of polar-emission angles. The comparison between simulated
acoustic emission and reference data is shown in Figure 16.27 for two Mach numbers
and polar-emission angles equal to 90 degrees. The noise-level dip at frequencies
f ∼ 500 Hz is attributed to a spurious vortex shedding that is not captured by the
theory.
Figure 16.28 shows the noise directivity for two different aircraft as predicted
by this theory. The comparison with the test data 56 – 58 is fair; however, note that the
data are somewhat scattered.
16.7 Propeller Noise

The propeller noise has two fundamental components: the harmonic (or rotational)
noise, arising from the blade-passing frequency, and the broadband noise, arising
6
A320, 2-wheel LG Calculated
B-737, 2-wheel LG
5
Directivity, dB 4
0
60 90 120 150
Polar emission angle, degs
Figure 16.28. Predicted landing-gear noise directivity and comparison with data for the Airbus
A320 and Boeing B737.
from turbulence-related acoustic excitation. The latter contribution is dominant at

relatively high frequency and provides an acoustic signature that is continuous in
the frequency spectrum. By contrast, the former contribution is a signal at discrete
frequencies depending on the blade-passing frequency. It turns out that the propeller
noise is dominated by the harmonic content; therefore our analysis will focus on this
contribution. We will present a low-order numerical model for harmonic noise based
on an implementation of the model proposed by Hanson and Parzych 59 . This method
is suitable for implementation in a comprehensive flight-mechanics code.
16.7.1 Propeller’s Harmonic Noise

Refer to Figure 16.29. Suppose that the centre of the propeller is A; the propeller
is moving toward the observer O along the x–direction. By the time the acoustic
waves emitted at Areach the observer, the propeller will be at position B. The initial
distance to the observer is r ; the remaining distances and angles are clearly indicated
in Figure 16.29. {r, θ, φ} is a spherical coordinate system when the x-axis is in the
flight direction; {r, θ , φ } is a spherical coordinate system when the x-axis is in the
propeller axis.
There are two distinct acoustic-power contributions, one due to lift and the other
one due to volume (or thickness), described as follows:
r The lift (or loading) contribution P is generated by steady-state blade loading
L
that rotates with the blade. As such rotation takes place, there is a phase velocity
between blade and receiver. When the flow is not axi-symmetric (a condition
that arises when there is a yaw or a pitch), there is also unsteady blade loading.
The lift noise is represented by a dipole distribution.
r The volume (or thickness) contribution P arises from the displacement of the
T
air flow when it passes through the blade. The higher the flow deflection, the
Figure 16.29. Reference system and nomenclature for propeller-noise model; r, θ, φ is a spher-
ical coordinate system.
higher the thickness effect. An unsteady contribution arises, as in the previous

case, due to varying incidence and inflow as the blade performs a full revolution.
The thickness noise is represented by a monopole distribution.
The thickness and loading noise are represented by the monopole and dipole
distribution, respectively. Quadrupole noise induced by non-linear effects only arises
when the propeller blades rotate at high speeds such that the blade tips’ Mach
number Mti p is within the transonic or supersonic regimes, as reported by Hanson
and Fink 60 . The broadband or random component is mainly induced by the turbulent
flow which only becomes significant when the blades are subjected to high angles of
attack.
The dominant noise for a subsonic propeller can be predicted using the
frequency-domain computational approach developed by Hanson and Parzych 59 .
Both noise components are computed at the blade-passing frequency as well as at
its higher harmonics. Unsteady flow effects are accounted for in the model when
non-co-axial, or angular, flow enters the propeller. The angular inflow produces the
unsteady periodic effect because it causes the relative blade section velocity to vary
within a blade revolution.
Hanson and Parzych reported that results from the model compare well with
experimental propeller data for various parametric conditions and inflow angles.
The model validation is not satisfactory, however, when compared to prop-fans data;
they reasoned that this is probably due to the aerodynamic loading estimation for
the prop-fans and not the acoustic radiation because it is based on exact equations.
The starting point for the Hanson and Parzych propeller noise model is the
exact theory of Goldstein’s acoustic analogy applied to moving surfaces embedded
within a moving medium. It describes the acoustic pressure disturbance p = ρ ao2 of

a source as the sum of integrals involving three source terms: the thickness noise due
to displaced volumes of air, the loading noise due to surface forces and the quadru-
pole noise due to the non-linear effects accounted for in Lighthill’s stress tensor. The
model takes into account the various coordinate systems involved, that is, the pro-
peller and flight coordinate-systems for moving and stationary observers; a careful
treatment of the coordinate system conversions has been detailed elsewhere 61;59 .
Of interest to the application presented here is the prediction of far-field noise
radiation received by a stationary observer on the ground. This is achieved through
expansion of each of the source terms in Goldstein’s formula. Computationally
efficient forms are obtained following normalisation into non-dimensional variables,
discretisation of the blade-area elements and conversion using Fourier series into
the frequency domain. Assuming that the quadrupole noise is negligible for the
subsonic propeller application, the non-dimensional acoustic pressure disturbances
for the thickness noise pTm and the loading noise pLm are shown in Equations 16.115
and 16.116, respectively. These pressures are normalised with respect to the ambient
pressure po. Evaluations of the Bessel function Jn (x) at various orders are required
for the assessment of the non-dimensional acoustic pressures:
−iγ Bkmei[kmr+mB(±φ −π/2)] −i[km Sc xμ,ν +mBφμ,ν ]

pTm = e
4π Sod f μ ν

1
× V0μ,ν JmB(km Ss ro) + (Vcμ,ν − i Vsμ,ν )ei(±φ −π/2) JmB+1 (km Ss ro)
2
-
1 −i(±φ −π/2)
+ (Vcμ,ν + i Vsμ,ν )e JmB−1 (km Ss ro) (16.115)
2
−i Bkmeikmr i[(mB−k)(±φ −π/2−φμ,ν )+kφr e f ] −ikm Sc xμ,ν

pLm = e e
4π So μ ν

mB − k
× i Ss Fr kμ,ν JmB−k (km Ss ro) + Fφkμ,ν JmB−k(km Ss ro)
kmro
-
+ Sc Fxkμ,ν JmB−k(km Ss ro) , (16.116)
where the counters μ and ν refer to the radial and chordwise elements; m is the
harmonic blade-passing frequency, k is the index for the loading harmonic, and km is
the wave number in the harmonic Green function, km = mBMti p ; d f is the Doppler
shift, B is the number of blades, r0 = r/di . The factors Sc and Ss are defined by:
cos θ sin θ
Sc = Ss = . (16.117)
1 − M cos θ 1 − M cos θ

Finally, So denotes the amplitude radius: So = x 2 + β 2 (y2 + z2 ). Equation 16.115
describes the thickness-source terms V0μ,ν , Vcμ,ν and Vsμ,ν . The loading-source terms
for the tangential and axial loading, Fφkμ,ν and Fxkμ,ν respectively, are given in
Equation 16.116. The radial loading Fr kμ,ν is found to have a negligible contribution
of only about 0.6 dB 59 toward the loading noise and therefore is not included in
the total noise prediction. The lift and drag coefficients, CL and CD, both for steady
Frequency, Hz Frequency, Hz
50 100 150 200 50 100 150 200
120 120
Axial inflow (calculated) Axial inflow (calculated)
NASA data (1993) Yaw = 5 degrees
110 110
100 100
SPL, dB
SPL, dB
90 90
80 80
70 70
1 2 3 4 5 6 7 1 2 3 4 5 6 7
Harmonic number, m Harmonic number, m
(a) (b)
Figure 16.30. The effects of angular inflow on propeller SPLh for various harmonics of the
blade-passing frequency. Left graph shows comparison with the data in Ref. 59.
and unsteady flows, are treated as inputs to the noise model. These characteristics
are calculated with the propeller model discussed in Chapter 6, although alternative
methods can be used, from tabulated data to computational fluid dynamics methods.
The unsteady loading for the analysis presented here utilises the ESDU method
developed specifically for subsonic propeller blades in non-axial inflow 62 .
The parameter of interest for the total propeller noise, that is, the sound pressure
level, in Equation 16.118 is calculated by summing the square of the thickness and
loading acoustic pressures normalised with the reference acoustic pressure pref .
These noise metrics are calculated at the blade-passing frequency and its higher
harmonics and presented as SPLh :
2 2
po po
SPLh = 10 log10 PTm + PLm . (16.118)
pref pref
Due to the Doppler effect, a stationary observer would receive the noise with the
same SPLh but shifted to frequencies different from the source frequencies ( fs )
on the moving aircraft. The SPLs must be shifted to these receiver frequencies via
the Doppler relation f = fs /(1 − M cos θ ), then re-scaled into the one-third octave-
band scales where the SPL of the shifted harmonics that lie within the same band
are integrated accordingly.
This propeller-noise model is fully coupled with the aerodynamic-performance
model described in Chapter 6 to generate the correct distribution of loads. The
following results illustrate the changes in the propeller SPLs when the blades are
subjected to non-zero angular inflows. In Figure 16.30, the SPLs are plotted against
the harmonics for the conditions without and with a yaw angular inflow of 5 degrees.
The noise level is measured at a location 1.34d away from the propeller hub, where
d is the blade diameter. The SPL decreases with increasing harmonics for both
120 120 Axial inflow

Yaw = 5 degrees
100 100
80 80
SPL, dB
SPL, dB
60 60
40 40
20 20
Axial inflow
Yaw = 5 degrees
0 0
0 30 60 90 120 150 180 0 30 60 90 120 150 180
Polar angle, degs Polar angle, degs
(a) 1st harmonic (b) 5th harmonic
Figure 16.31. Effect of angular inflow on propeller SPLh for the 1st and 5th harmonics.
conditions; this value increases by about 5 dB for each harmonic when the blades
are subjected to unsteady flows.
The SPL variations with the polar directivity angle θ for the 1st and 5th harmonics
are shown in Figure 16.31. The azimuthal directivity angle φ is set at 90 degrees,
equivalent to a position below the propeller. The unsteady loads experienced at an
inflow setting of 5 degrees cause the noise level to increase; the increase is both a
function of the polar angle and the harmonics. For all configurations presented, the
peak noise is achieved forward of the propeller at a polar angle between 60 and
80 degrees. In Figure 16.32, the azimuthal or circumferential directivity angle for
the plane perpendicular to the flight path is varied with a fixed polar angle of 70
degrees. When the inflow is co-axial with the propeller axis (i.e., no angular inflow),
the SPL around the propeller is uniform at about 118 dB, unaffected by any unsteady
loading. Subjecting the blades to unsteady loading with a pitch or yaw angular inflow
of 5 degrees generally increases the SPL but non-symmetrically, due to the blade
rotational direction and the specific angular shift of the inflow.
16.7.2 Propeller’s Broadband Noise

In many cases, the broadband noise is characterised by lower SPL than the tonal
noise. However, whilst the tonal noise is limited to a few frequencies, the broadband
noise contribution is distributed through the full spectrum of audible frequencies.
There are several analytical, semi-analytical and semi-empirical methods to
account for the broadband noise. The methods are applied with various degrees
of accuracy to a number of different systems, including propellers and helicopter
rotors. Without making a full review of the subject, which is quite extensive, we con-
sider a method described by Magliozzi et al. 63 , where further technical details and
references can be found. This method has the advantage of providing a sufficiently
accurate estimate of the broadband noise due to the scattering of the turbulent
boundary layer at the trailing edge.
Figure 16.32. The effects of angular inflows on propeller SPLh for the first harmonic and
various azimuthal directivity angles.
If ω is the angular frequency corresponding to the frequency f , then ω = 2π f .

This frequency is normalised with the boundary-layer thickness at the trailing edge,
δ ∗ , and the inflow velocity U:
∗
δ
ω̃ = ω . (16.119)
U
The one-third octave band SPL is then estimated from the following equation:
⎧ −4 ⎫
⎨ 4 3/2 ⎬
ω̃ ω̃
SPL = C + 10 log10 0.613 + 0.5 , (16.120)
⎩ ω̃max ω̃max ⎭
where ω̃max is the maximum value of ω̃. It is reported that ω̃max 0.1. The log term in
Equation 16.120 is always negative, with a maximum around ω̃/ω̃max 1. The factor
C denotes an overall SPL and is given by

δ∗b
C = 10 log10 M5 2 D + 141.3. (16.121)
r
The remaining factors in Equation 16.121 include a lateral length scale b (span or
element width) and a directivity D, which is calculated from
2 cos2 (θ/2)
D= . (16.122)
(1 − M cos θ )[1 − (M − Mc ) cos θ]2
Equation 16.122 is the only function of the corrected emission angle, the flight Mach
number and the critical Mach number Mc ; it further contains a Doppler correction.
When the corrected emission angle is θ = π , D = 0, irrespective of Mach number.
This causes Equation 16.121 to become singular; Equation 16.121 is singular also in
static conditions (M = 0).
The displacement thickness δ ∗ required to normalise the angular frequency is

calculated from semi-empirical formulas in turbulent flow; one such equation is
δ∗
= 0.047 Re−1/5
c , (16.123)
c
where Rec is the Reynolds number based on the chord c. Equation 16.123 only
applies to a flat plate; therefore, some additional corrections must be introduced to
account for the effects of effective angle of attack αe , camber and thickness. From
extensive aerodynamic analysis we find that a good correction is δ ∗ (α) = Fδ ∗ . The
correction factor is
exp(c1 αe + c2 )
F(α) = , (16.124)
c3
where ci are factors depending on the pressure and suction side; the parameter c3 is
a normalisation factor that forces F = 1 at αe = 0.
This basic method is to be applied to the calculation of broadband noise from a
propeller. To do so, we apply the usual blade-element theory. For a generic element,
we are able to calculate the inflow conditions and the aerodynamic coefficients by
following the methods presented in Chapter 6. In particular, we have the chord c,
the element width dy = b, the inflow velocity U, the Reynolds number Rec and the
displacement thickness at the trailing edge. The critical Mach number is calculated
as shown in § 4.2.3. If the propeller is sufficiently far from the receiver, the corrected
emission angle θ and distance r are virtually unaffected by the propeller’s rotation.
This assumption greatly simplifies the calculation. For a given frequency f we need
to perform an integration in the radial direction to account for the contribution of
all of the blade elements. Then we need to account for all of the blades. Again, if
the propeller is sufficiently far from the receiver, we assume that all of the blades
contribute equally to the broadband noise.
A typical result of this numerical method is shown in Figure 16.33, which displays
the noise spectrum for the Dowty R408 six-bladed propeller trimmed to provide a
power P = 1,328 kW at a flight speed 115 KTAS at 200 m above sea level. The
receiver is at a visual distance r ∼ 165 m. There are only three dominant tonal
contributions, but overall the broadband noise is very low.
Summary
This chapter has reviewed some of the low-order acoustic methods that can be used
for the estimation of the noise sources from a variety of aircraft systems. We have
illustrated how a noise-prediction program can be configured and proceeded with
splitting the noise sources between propulsive and non-propulsive sources. Among
the propulsive sources there is the propeller and the APU, which is modelled in
a fashion similar to a gas-turbine noise. A number of limitations are evident. First
of all, the modelling is carried out on a narrow basis, with relatively few geomet-
rical and operational parameters. Second, the models themselves rely on limited
understanding of very complex phenomena. Nevertheless, the framework for noise
prediction is a robust one; if more advanced submodels do become available, they
can be plugged into the noise program to improve the predictions. All of the inac-
curacies that may arise from any of the models presented in this chapter must be
80
Tonal component
Tonal + Broadband
70
60
SPL, dB
50
40
30
20
10
102 103 104
Frequency, Hz
Figure 16.33. Noise spectrum for the R408 six-bladed propeller model trimmed at a specified
power. Result uncorrected for atmospheric absorption.
viewed in the wider context of noise-propagation effects, which can substantially

change the nature of noise, in terms of both levels and frequency content. These
statements will be demonstrated in the next chapter.
Bibliography
[1] Franssen EAM, van Wiechen CMAG, Nagelkerke NJD, and Lebret E. Aircraft
noise around a large international airport and its impact on general health and
medication use. Occup. Environ. Med., 61(5):405–413, 2004.
[2] Pierce AD. Acoustics: An Introduction to Its Physical Principles and Applica-
tions. Acoust. Soc. America, 1989.
[3] Anon. Acoustical terminology. Technical Report S1.1-1994, American
National Standards Institute, June 1994.
[4] Smith MTJ. Aircraft Noise. Cambridge University Press, 2004.
[5] Edge PM and Cawthorn JM. Selected methods for quantification of community
exposure to aircraft noise. Technical Report TN-D-7977, NASA, 1976.
[6] Casalino D, Diozzi F, Sannino R, and Paonessa A. Aircraft noise reduction
technologies: A bibliographic review. Aero Sci. & Techn., 12(1):1–17, 2008.
[7] Heidmann MF. Interim prediction method for fan and compressor noise source.
Technical Report TM X-71763, NASA, 1979.
[8] Hough JW and Weir DS. Aircraft noise prediction program (ANOPP) fan
noise prediction for small engines. Technical Report CR-198300, NASA,
1996.
[9] ESDU. Prediction of Noise Generated by Fans and Compressors in Turbojet
and Turbofan Engines Combustor Noise from Gas Turbine Engines. Data Item
05001. ESDU International, London, June 2003.
[10] Bielak GW, Premo JW, and Hersch AS. Advanced turbofan duct liner concepts.
Technical Report CR-1999-209002, NASA, Feb. 1999.
Bibliography 525
[11] Ko SH. Sound attenuation of acoustically lined circular ducts in the presence
of uniform flow and shear flow. J. Sound & Vibration, 22(22):193–210, 1972.
[12] Tam CWK, Kurbatskii KA, Ahuja KK, and Gaeta RJ. A numerical and exper-
imental investigation of the dissipation mechanisms of resonant acoustic liners.
J. Sound & Vibration, 245(3):545–557, 2001.
[13] Astley RJ, Sugimoto R, and Mustafi P. Computational aero-acoustics for fan
duct propagation and radiation. Current status and application to turbofan liner
optimisation. J. Sound & Vibrations, 330(16):3832–3845, 2011.
[14] ESDU. The Acoustic Attenuation of Absorbent Linings in Cylindrical Flow
Ducts. Data Item 00012, Amend. B. ESDU International, London, June 2010.
[15] ESDU. The Acoustic Attenuation of Absorbent Linings in Rectangular Flow
Ducts with Application to Annular Flow Ducts. Data Item 00024. ESDU Inter-
national, London, Mar. 2011.
[16] McAlpine A and Fisher MJ. On the prediction of “buzz-saw” noise in aero
engine inlet ducts. J. Sound & Vibration, 248(1):123–149, 2001.
[17] McAlpine A, Fisher MJ, and Tester BJ. “Buzz-saw” noise: A comparison of
measurement with prediction. J. Sound & Vibration, 290(3-5):1202–1233, 2006.
[18] ESDU. Prediction of Combustor Noise from Gas Turbine Engines. Data Item
05001. ESDU International, London, Feb. 2005.
[19] Zorumski WE. Aircraft noise prediction program (ANOPP) theoretical
manual. Technical Report TM-83199, Part 2, NASA, Feb. 1982.
[20] Matta RK, Sandusky GT, and Doyle VL. GE core engine noise investigation –
Low emission engines. Technical Report FAA-RD-74, FAA, Feb. 1977.
[21] ESDU. Aircraft Noise Prediction. Technical Report Data Item 02020, ESDU
International, Sept. 2009.
[22] van Zante D and Envia E. Simulation of Turbine Tone Noise Generation Using
a Turbomachinery Aerodynamics Solver. Technical Report TM-2010-216230,
NASA, Mar. 2010.
[23] Chien E, Ruiz M, Yu J, Morin B, Cicon D, Schweiger P, and Nark D. Com-
parison of predicted and measured attenuation of turbine noise from a static
engine test. 13th AIAA/CEAS conference, AIAA paper 2007-3533, Rome,
Italy, May 2007.
[24] ESDU. An Introduction to Aircraft Noise. Data Item 02020. ESDU Interna-
tional, London, 2002.
[25] Zorumski WE. Aircraft noise prediction program (ANOPP) theoretical
manual. Technical Report TM 83199, Part 1, NASA, Feb. 1982.
[26] Lighthill MJ. On sound generated aerodynamically – Part I. Proc. Royal Soc.
London, Sect. A, 211:564–587, 1952.
[27] Lighthill MJ. On sound generated aerodynamically – Part II. Proc. Royal Soc.
London, Sect. A, 222:1–32, 1954.
[28] Balsa TF and Gliebe PR. Aerodynamics and noise of coaxial jets. AIAA J.,
15(11):1550–1558, Nov. 1977.
[29] Stone JR, Groesbeck DE, and Zola CL. Conventional profile coaxial jet noise
prediction. AIAA J., 21(3):336–342, Mar. 1983.
[30] Olsen W and Friedman R. Jet noise from coaxial nozzles over a wide range
of geometric and flow parameters. In AIAA Aerospace Meeting, AIAA Paper
1974-0043, Washington, DC, Jan. 1974.
[31] ESDU. Computer-based Estimation Procedure for Coaxial Jet Noise. Data Item
01004. ESDU International, London, May 2001.
[32] Fisher MJ, Preston GA, and Bryce WD. A modelling of the noise from coaxial
jets. Part 1: With unheated primary flow. J. Sound & Vibration, 209(3):385–403,
Jan. 1998.
[33] Fisher MJ, Preston GA, and Mead WD. A modelling of the noise from coaxial
jets. Part 2: With heated primary flow. J. Sound & Vibration, 209(3):405–417,
Jan. 1998.
[34] ESDU. Estimation of Subsonic Far-Field Jet-Mixing Noise from Single-Stream

Circular Nozzles. Data Item 89041. ESDU International, London, Feb. 1990.
[35] ESDU. Computer-based Estimation Procedure for Single-Stream Jet Noise.
Data Item 98019. ESDU International, London, Nov. 1998.
[36] Stone JR, Krejsa EA, Clark BJ, and Berton JJ. Jet Noise Modeling for Sup-
pressed and Unsuppressed Aircraft in Simulated Flight. Technical Report TM-
2009-215524, NASA, Mar. 2009.
[37] Stone JR, Krejsa EA, and Clark BJ. Jet Noise Modeling for Supersonic Business
Jet Application. Technical Report TM-2004-212984, NASA, Mar. 2004.
[38] Ahuja KK. Designing Clean Jet Noise Facilities and Making Accurate Jet Noise
Measurements. Technical Report AIAA-2003-0706, AIAA, Jan. 2003.
[39] Bridges J and Brown CA. Parametric testing of chevrons on single flow hot
jets. In 10th AIAA/CEAS Conference, Manchester, UK, Sept. 2004.
[40] Rask O, Kastner J, and Gutmark E. Understanding how chevrons modify
noise in a supersonic jet with flight effects. AIAA J., 49(8):1569–1576, Aug.
2011.
[41] Gerhold CH. Analytical model of jet shielding. AIAA J., 21(5):694–698, May
1983.
[42] Simonich JC, Amiet RK, and Schlinker RH. Jet shielding of jet noise. Technical
Report CR-3966, NASA, 1986.
[43] ESDU. Jet-by-Jet Shielding of Noise. Data Item 88023. ESDU International,
London, Mar. 1992.
[44] Tam C, Pastouchenko N, Mendoza J, and Brown D. Combustion noise of
auxiliary power units. AIAA Paper 2005-2829, May 2005.
[45] Fink MR. Noise component method for airframe noise. J. Aircraft, 16(10):659–
665, 1979.
[46] Zorumski WE. Aircraft noise prediction program theoretical manual, Part 1.
Technical Report TM-83199, NASA, Feb. 1982.
[47] ESDU. Airframe Noise Prediction. Data Item 90023. ESDU International,
London, June 2003.
[48] Lilley GM. The prediction of airframe noise and comparison with experiment.
[49] Lockard DP and Lilley GM. The airframe noise reduction challenge. Technical
Report TM-213013, NASA, 2004.
[50] Ffowcs-Williams JE and Hall LH. Aerodynamic sound generation by turbu-
lent flow in the vicinity of a scattered half plane. J. Fluid Mech, 40:657–670,
1970.
[51] Howe MS. A review of the theory of trailing edge noise. J. Sound & Vibration,
61(3):437–466, 1978.
[52] Guo YP. Empirical prediction of aircraft landing gear noise. Technical Report
CR-2005-213780, NASA, 2005.
[53] Guo YP, Yamamoto KJ, and Stoker RW. Experimental study on aircraft land-
ing gear noise. J. Aircraft, 43(2):306–317, Mar. 2006.
[54] Guo YP. A component-based model for aircraft landing gear noise prediction.
J. Sound & Vibration, 312:801–820, 2008.
[55] Humphreys WM and Brooks TF. Noise spectra and directivity for a scale-
model landing gear. In 13th AIAA/CEAS Aeroacoustics Conference, AIAA
2007-3458, Rome, Italy, May 2007.
[56] Guo YP, Yamamoto KJ, and Stoker RW. Component-based empirical model
for high-lift system noise prediction. J. Aircraft, 40(5):914–922, Sept. 2003.
[57] Stoker RW. Landing gear noise report. Technical report, NASA Report NAS1-
97040, 1997.
[58] Dobrynski W and Buchholtz H. Full-scale noise testing on Airbus landing gears
in the German-Dutch wind tunnel. AIAA paper 1997-1597, May 1999.
[59] Hanson DB and Parzych DJ. Theory for noise of propellers in angular inflow
with parametric studies and experimental verification. Technical Report CR-
4499, NASA, Mar. 1993.
[60] Hanson DB and Fink MR. The importance of quadrupole sources in prediction
of transonic tip speed propeller noise. Journal of Sound and Vibration, 62(1):19–
38, Jan. 1979.
[61] ESDU. Prediction of Near-Field and Far-Field Harmonic Noise from Subsonic
Propellers with Non-Axial Inflow. Data Item 95029. ESDU International, 1996.
[62] ESDU. Estimation of the Unsteady Lift Coefficient of Subsonic Propeller Blades
in Non-Axial Flow. Data Item 96027. ESDU International, London, 1996.
[63] Magliozzi B, Hanson DB, and Amiet RK. Propeller and propfan noise. In
Hubbard HH, editor, Aeroacoustics of Flight Vehicles: Theory and Practice,
RP 1258. NASA, Aug. 1991.

a = speed of sound
a,b = parameters in Equation 16.31; Table 16.3
A = area
A = attenuation function
Ar = area ratio in co-axial jet noise
A, B = parameters in landing-gear model, Equation 16.107
b = wing span in Equation 16.86; spanwise reference length
B = number of propeller blades
c = exponent defined by the second Equation 16.18
c1 , c2 , c3 · · · = constant factors
C = factor in broadband noise, Equation 16.121
C = correction of SPL for tonal noise in Equation 16.4;
constant in Equation 16.25
CS ,CT = parameters in UOL equations for small-scale,
transitional regions
d = wheel diameter; nozzle diameter; other diameter
de = effective jet diameter
df = Doppler factor
dh,1,th = hydraulic diameter of the inner-stream throat,
Equation 16.62
d2,1,th = outer physical throat diameter of the inner-stream nozzle
D = directivity function
D = correction of SPL for duration, Equations 16.4 and 16.8
E = acoustic energy
G = Green function
F = correction factor for displacement thickness,
Equation 16.124
f = frequency
f1 = frequency corresponding to equal downstream/
upstream propagation
fˆ = corrected frequency in co-axial jet noise, Equation 16.45

fb = blade-passing frequency
fc = centre frequency in a one-third octave band
F = noise-level difference; factor in wing-noise model,
Equation 16.98
Fd = fraction of energy radiated downstream, Equation 16.44
Fm = complex function in shielding model, Equation 16.76
Fp = force created by acoustic pressure
Fu = fraction of energy radiated upstream, Equation 16.47
F = normalised spectrum, Equation 16.107
Fd = downstream spectral-energy function, Equation 16.44
Fi = sound-pressure correction in engine noise, i = 1, . . . , 5
Fu = upstream spectral-energy function, Equation 16.47
F1 = a function of the tip Mach number
F2 = function of the rotor-stator spacing
F3 = function of the noise directivity
F4 = function of the blade-passing frequency
Fφkμ,ν , Fxkμ,ν , Fr kμ,ν = tangential, axial, radial loading in propeller noise
h = flight altitude
hd = parameter in directivity function, Equation 16.108
h∗ = specific enthalpy, § 16.4.4
hr∗ = reference specific enthalpy
∗
hr,k = component-gas specific enthalpy
Hm = m-th order Hankel function of the first kind
i = imaginary unit
I = noise intensity (dB)
Jm = m-th order Bessel function of the first kind
k1 , k2 , k3 = factors in airframe noise model, Equation 16.87
km = wave number in harmonic Green function,
km = mBMti p , Equations 16.115 and 16.116
kp = sinusoidal period of oscillation for the pressure
transform
k, ko = wave numbers
K = factor in Equation 16.31, Table 16.3
KL, KS , LT = parameters in UOL equations for large-scale,
small-scale, transitional regions
li = length of i–th landing-gear strut
lt = total length of struts in landing gear
L = sound pressure level; turbulence length scale
LAE = sound exposure level
LAeqT = equivalent continuous sound level
Lp = peak level of fundamental tone in fan noise,
Equation 16.20
L1 = length scale in jet-noise model
LSF = linear scale factor (ratio of full-scale to

model-scale dimensions)
m = mass or molecular mass
m = jet-shape parameter, Equation 16.44; wave number in
Equation 16.75; harmonic number in propeller noise
mT = total molecular mass of a gas
ṁ = mass-flow rate
M = Mach number
Mc = critical Mach number
Mti p = tip Mach number
n = wave number ratio in ground reflection
n = harmonic number in spectral function, Equation 16.29
nk = number of computational points
nl , nu = lowest and highest harmonic numbers, Equation 16.30
ns = number of sub-band partitions, Equation 16.81
N = rotational speed of turbine stage
Nν = number of propeller elements in chordwise direction
Nw = number of wheels
p = atmospheric or acoustic pressure
pr = pressure ratio
pref = reference acoustic pressure, 20 · 10−6 Pa
p = acoustic pressure disturbance
P = acoustic power
p2 = mean square acoustic pressure
pL, pT = acoustic pressure due to loading/thickness in propeller
noise
Ps ,Pt = shielded and unshielded sound pressure level,
Equation 16.81 and Equation 16.82
PS = acoustic power resulting from jet shielding,
Equation 16.68
PR = suppressed-to-unsuppressed wetted perimeter ratio
q = semi-empirical parameter in landing-gear noise
Q = image source strength
r = distance source-to-receiver
r = vector distance source-to-receiver
rn = projection of r onto plane normal to jet
rj = jet radius
ro = distance between jets; radial coordinate normalised
with blade diameter (propeller noise)
r = distance between receiver and closest jet, Figure 16.21
r1 = minimum distance, Equation 16.82
R∗k = constituent gas dimensionless gas constant
Rec = Reynolds number based on chord c
Ru = universal gas constant
R = dry-air gas constant
s = perimeter in Equation 16.104
s = average diameter of landing-gear struts

So = amplitude radius
S = reference area for airframe noise components,
Equation 16.89
S = spectral function
Sc , Ss = factors in Equations 16.113 and 16.114
St = Strouhal number
t = time
to = reference time
T = atmospheric temperature
Tr = reference temperature
Ti,i = total inlet-turbine temperature
Ts, j = static exit-turbine temperature
To = water triple-point temperature
T = thrust in jet-noise model, Equation 16.38
T1 , T2 = factors in jet-shielding noise, Equations 16.78a
and 16.78b
uo = isotropic turbulence level, Equation 16.90
U = inflow velocity
UT∗ = dimensionless tip speed, Equation 16.32
UOL = overall sound level, uncorrected for refraction, dB
UOL’ = normalised UOL
V = true air speed
Vr = velocity ratio in co-axial jet noise
V = vector velocity with components u, v, w
V0μ,ν , Vcμ,ν , Vsμ,ν = thickness-source terms in propeller noise
w = wheel width
wb = sub-band width in jet-shielding noise
W = weight
Wref = reference acoustic power, 10−12 W in § 16.4.3
Wref = reference weight in landing-gear model, Equation 16.106
WF = complex error function
x1 = reference position for co-axial jet noise
x1 , x2 , x3 = factors in jet-shielding model,
Equations 16.78c, 16.78d, 16.78e
xk = constituent-gas mass fraction
Xs = axial source location relative to outer stream
nozzle exit
y = co-axial jet parameter defined by Equation 16.51
ym = length scale of turbulent eddies in Equation 16.96
z = height from the ground
Greek Symbols
αe = effective inflow angle
β = radiation efficiency factor, Equation 16.102
β = polar angle between receiver and plane normal to jet axis,
Equation 16.67
δ = boundary-layer thickness; relative pressure, p/ po

δ∗ = boundary-layer displacement thickness
= factor of the wave number, Equation 16.75 and Equation 16.79
η = landing-gear complexity function, Equation 16.106
η = dimensionless frequency, Equation 16.28
ηc = thermodynamic efficiency of a compressor stage
θ, θ = polar-emission angle, polar angle in spherical coordinates
(propeller noise), see Figure 16.28
γ = ratio between specific heats, γ = C p /Cv
γt = wheel-track-alignment angle, Equation 16.106
m = factor in potential, Equation 16.75
λ = wavelength
= sweep angle of a lifting surface
μ = semi-empirical parameter in landing-gear noise
ν = kinematic viscosity in Equation 16.89
μ, ν = index for radial/chordwise elements in propeller noise,
Equations 16.115 and 16.116
∗ = acoustic power parameter, Equation 16.31
ρ = air density
τ = temperature ratio, last of Equation 16.51
σ = semi-empirical parameter in landing-gear-noise model
σr = gas-density ratio in co-axial jet
ς = power coefficient in normalised spectra, Equation 16.107
ϕ = bank angle, Equation 16.71
φo = shielding velocity potential due to a unit source
φ = azimuth angle in flight path, Figure 16.24
φ, φ = azimuthal angle in spherical coordinate system (propeller noise)
= velocity potential
ψ = angle between source-to-receiver and plane normal to jet,
Figure 16.21, Equation 16.71
ψo, ψ1 , ψ2 = parameters in co-axial jet noise, defined by Equation 16.43
= azimuthal angle on plane normal to jets
ω = variable density index, Equation 16.52
ω = angular frequency
[.]a = referred to airframe or atmosphere
[.]cor = corrected for source location
[.]e = referred to engines
[.]e = corrected or retarded value
[.]h = harmonics in propeller noise
[.] H = high frequency
[.]i = inlet condition
[.] I SA = International Standard Atmosphere
[.] j = jet quantity; exit condition in turbine-noise model
[.] L = low frequency; large-scale mixing noise in § 16.4.8
[.] M = medium frequency

[.]m = mixed jet
[.]main = relative to main landing gear
[.]n = projection of vector to a plane
[.]nose = relative to nose landing gear
[.] p = primary jet flow
[.] P = inner-stream plug-separation noise in § 16.4.8
[.]o = standard atmospheric conditions; other reference conditions
[.]r = primary/secondary ratio in co-axial jet noise
[.]ref = reference quantity
[.]s = secondary-jet flow; static or source
[.] S = small-scale mixing scale, in § 16.4.8
[.]t = total or stagnation condition in co-axial jet noise, § 16.4.8
[.]te = trailing edge
[.]T = transitional separation noise in § 16.4.8
[.] = derivative in jet-shielding model
[.] = average value
ˆ
[.] = normalised quantity in jet model
[.] = mean-square value
[.]∞ = far-field conditions
[.]1 = inner stream (i.e., the core jet flow)
[.]2 = outer stream (i.e., the by-pass flow or the ambient)
[.]1 , [.]2 = jet number in jet-shielding model (1 is closest jet)
˜
[.] = normalised value
17 Aircraft Noise: Propagation
Overview∗
Noise propagation involves all of the physical events that take place between the
noise source and the receiver. Due to the relatively large distance between source and
receiver (from a minimum of ∼100 m to several kilometers), atmospheric absorption
is particularly important – no less important than the determination of the noise
sources themselves. If propagation takes place over an unbounded medium, the use
of the atmospheric absorption model would be sufficient to estimate the effect at
the receiver. However, there are cases where the topography of the terrain and the
short distance between the airplane and the ground cause effects such as reflection
and scattering to become important. We begin by considering briefly the airframe
noise shielding (§ 17.1); we continue with the standard absorption model in the
atmosphere (§ 17.2). We describe ground-reflection problems in § 17.3. We finally
discuss the combination of wind shear and temperature gradients in the propagation
of aircraft noise over long distances (§ 17.4).
KEY CONCEPTS: Airframe Noise Shielding, Wing Scattering, Ground Effect, Atmo-
spheric Absorption, Wind Effects, Turbulence Effects.
17.1 Airframe Noise Shielding

A number of interference effects take place around the airframe due to the reciprocal
position between noise sources and solid surfaces. The effect is frequency-dependent.
The investigation of new blended airframes, which actually benefit from the noise
shielding 1 – 3 , has increased the interest in this acoustic effect.
One method is called the equivalent source method 4;5 , which uses a finite number
of sources to satisfy the boundary condition on the airframe to simulate the acoustic
scattering. Based on this method, a fast-scattering model was developed to calculate
the sound reflected and scattered by the airframe, as shown by Dunn and Tinetti 6 .
Although some researchers pointed out that a small number of sources may cause a
∗ Dr. Zulfaa Mohamed-Kassim contributed to this chapter.
533
ΔSPL (dB)
3.67
3.63
3.59 Figure 17.1. Simulated scattering noise on the
3.55 Airbus A320 wing; level flight, one span below
3.51 the wing.
3.47
3.43
poor accuracy, a series of investigations have shown that this accuracy is acceptable
when the number of sources is carefully chosen 7 .
In the equivalent scattering method, we replace a vibrating body with a system
of sources (equivalent sources) in the interior of the body. Hence, the sound field is
expressed by a superposition of an incident sound field (caused by the real acoustic
source) and a scattering sound field (caused by the equivalent source).
An example of calculation is shown in Figure 17.1, which displays the change in
SPL due to engine noise in the presence of wing scattering; the field plot refers to a
plane parallel to the wing, placed one span below.
Although the equivalent source is a good choice for airframe design and noise
diagnosis at specific frequencies, it still needs considerable time to compute. The
most serious problem is that the method needs the numbers of equivalent sources
and points on the boundary to be larger than a certain value to keep the accuracy.
Methods based on ray-tracing theory 8 have been developed to solve the sound
propagation in a non-uniform atmosphere. With the ray-tracing method the sound
wave front emitted from a source is divided into many small pieces. The reflection
from wing and ground can be calculated if the wing and ground are treated as
boundaries. The smaller the sound piece and the time interval, the more accurate
the result. The number of time steps depends on the source-to-receiver distance.
To overcome the computational difficulty and satisfy the requirements of a fast
comprehensive noise model, we propose a simplified method, which assumes that the
reflection from the wing obeys the rule of reflection from a flat surface. This assump-
tion is acceptable when 1) the lower wing surface has a small curvature, and 2) the
receiver has a distance larger than the curve’s length scale. Based on these assump-
tions, only one or two reflections need to be calculated for each source. We call this
method the reflection-only method. The computational procedure is the following:
1. Divide the wing surface into quadrilateral elements.

2. Calculate the normal vector n of each element.
3. Compare these normal vectors with the source and observer positions; check if
they satisfy the reflection condition: incident angle = reflected angle.
4. If the reflection condition is satisfied, a reflected wave will be calculated and

superposed to the direct acoustic wave at the observer position. Otherwise, the
observer will be marked as non-reflection.
The scattering from the fuselage is modelled only when the aircraft is close to
the ground. In this instance, we use the model of a sufficiently long cylinder over an
impedance boundary 9 . However, this model can only be applied when the receiver
point is sufficiently far from the axis of the airframe; otherwise it would require a
fictitious extension of the fuselage in order to account for this interference. The basic
idea of this model is to express the total sound field as a sum of four components:
the incident field, the reflected wave, and the scattered fields from the cylinder and
its images. Theoretical details can be found in the references cited.
17.2 Atmospheric Absorption of Noise

Sound propagating through the atmosphere is absorbed by an amount depending on
frequency, distance and atmospheric conditions. Atmospheric absorption essentially
means that the noise heard by the receiver is lower than the noise due to distance
propagation effects. The key parameters in the atmosphere are the changes in its
temperature, pressure and relative humidity. With so many parameters at play,
the result tends to be complicated. In aircraft certification, the knowledge of the
atmospheric attenuation is used to derive the noise at the source from measurements
on the ground. In the present case, where we attempt to predict the noise at the
ground, we need the atmospheric attenuation as a correction of the noise propagating
through an ideal medium in thermodynamic equilibrium (isothermal and isobaric).
If we abstract for a moment from atmospheric winds, then we define the absorp-
tion function
A = F(h, r, dT , H, f ),
where h is the altitude of the source with respect to the receiver, r is the distance
from the receiver, dT is the change in temperature around the standard value,
H is the relative humidity and f is the frequency of the acoustic emission. The
attenuation function is generally given in dB per meter. For computational reasons,
it is convenient to have an explicit relationship between the attenuation rate and
its physical parameters. Therefore, the attenuation at a given frequency f over a
distance r due to an aircraft flying at an altitude h is found from the integral
r
A( f, h, H, r ) = F(dT , H, r ) dr. (17.1)
o
Values of the absorption have been found experimentally over several years (for
example, Zuckerwar and Meredith 10 ). The American Institute of Physics (AIP)
published a standard approved by ANSI 11 . Another recognised standard is the
ISO 9613-1 12 , which is substantially equivalent to the ANSI standard. ESDU 13
has published a model that includes the effects of ground refraction, atmospheric
turbulence and wind shear. Morfey and Howell 14 review the theory of acoustic wave
propagation from aircraft.
From the ANSI model, the sound attenuation rate at a frequency f is

1/2
−11 pr T
A [dB/m] = 8.686 f 2 1.84 · 10 +
pa Tr
−5/2
−2.2391 · 103 fr o T
+ 0.01275 exp + (17.2)
T fr o + f
2 2 Tr
−5/2 ,
−3.3520 · 103 fr n T
+ 0.10680 exp .
T fr2n + f 2 Tr
In this equation pa is the atmospheric pressure at the reference point, pr is the

reference pressure (101,325 Pa), and Tr is the reference temperature (293.15 K, or
20 ℃); the terms fr o and fr n are given by

pa 0.020 + hc
fr o = 24 + 4.04 · 104 hc , (17.3)
pr 0.391 + hc
−1/2 −1/3 ,
pa T T
fr n = 9 + 280hc exp −4.170 −1 , (17.4)
pr Tr Tr
where hc is the molar concentration of water vapour in the atmosphere. If the relative
humidity H is given in percent, the molar concentration is given by

psat pr
hc = H , (17.5)
pr pa
and the saturation pressure psat is
psat = pr 10V . (17.6)
In Equation 17.6, the exponent V is calculated from

To1 T
V = a1 1 − − b1 log10 +
T To1
(17.7)
) * ) *
c1 1 − 10c2 (T /To1 −1) + d1 −1 + 10d2 (1−T o1 /T ) − e1 .
In Equation 17.7, To1 is the triple-point isotherm temperature (273.15 K). The numer-
ical coefficients for this equation are given in Table 17.1. At a given position in space
(h, T , p), it is possible to calculate the saturation pressure psat from Equation 17.6
(through the power V given by Equation 17.7). The saturation pressure is used to
calculate the molar concentration of water at the same point from Equation 17.5.
The latter parameter is used by the terms fr o and fr n and, finally, the attenuation
A from Equation 17.2. The cumulative effect of attenuation over a long distance
requires the integration of Equation 17.1. The calculation must be repeated over
the full spectrum of noise frequencies. In practice, it is convenient to use the same
reference frequencies as the dB-A scale. The main difficulty consists in accounting
for inhomogeneous humidity across different layers of the atmosphere. This effect is
sometimes evident to the receiver when the aircraft crosses areas of different cloud
thickness.
Table 17.1. Numerical coefficients for Equation 17.7
a1 b1 c1 d1 e1 c2 d2
10.79586 5.02809 1.50474 ·10−4 0.42873 ·10−3 2.21960 8.29692 4.76955
FAR Chapter 36, “Noise Standards: Aircraft Type and Airworthiness Certific-
ation”, requires that the noise be corrected for a fixed humidity of 70%, standard
conditions on the ground, no winds and for the spectrum from 50 Hz to 10 kHz.
The attenuation function calculated with this model is greatly dependent on the
frequency. In fact, a 10-kHz frequency is absorbed 104 times more than a 10-Hz
frequency. Most of the attenuation effects occur above 1 kHz. Consequently, high-
frequency noise is less likely to be heard by the receiver at a long distance, whilst
low-frequency noise is persistent and is affected more by distance than atmospheric
conditions. The atmosphere acts like a low-pass filter. However, unless the distance
of the aircraft is considerable, the SPL corrected for atmospheric absorption is
generally lower than the accuracy of most aircraft-noise predictions.
Sample calculations of the atmospheric absorption are shown in Figure 17.2 for
variable altitude and relative humidity, as indicated. The data show that the relative
humidity is a stronger factor than the flight altitude.
The values of the absorption A, integrated over a long distance, may lead to a
large reduction in noise level, as shown in Figure 17.3. The reduction in noise level
is evident after a few hundred metres, with the high frequencies being cut first.
Because in practice the aircraft noise propagates through layers of the atmo-
sphere that are not isobars, not isothermal and not even isohumidity, we need to
carry out the integration of Equation 17.2. Several methods, of variable accuracy,
have been proposed. However, due to the uncertainty on the relative humidity,
we consider the propagation through piecewise segments of constant atmospheric
conditions.
There are various other aspects of sound attenuation that we may need to
consider, such as the geometric divergence, the effects of winds and the ground
1 H = 20% 1 H = 20%
10 H = 50% 10 H = 50%
H = 80% H = 80%
Attenuation, dB/m
Attenuation, dB/m
0 0
10 10
10-1 10-1
10-2 10-2
102 103 104 102 103 104

(a) Sea level, ISA (b) h = 500 m, ISA
Figure 17.2. Atmospheric absorption model, according to ISO 9613-1.
f = 100 Hz
-10
Δ SPL 500 Hz
-20
1kHz
2kHz
-30 1
10 102 103 104
Propagation distance, m
Figure 17.3. Atmospheric absorption versus distance at selected frequencies, H = 70%.
reflection. The latter case is important, when noise reflected from the ground propag-
ates between the source and the receiver.
17.3 Ground Reflection

The analysis carried out so far assumes that the acoustic waves propagate through
an unbounded medium – an assumption that is reasonable in free flight. However,
when the aircraft is relatively close to the ground, the acoustic-wave propagation is
impeded by the presence of the ground. Typical effects are refraction and absorption
of acoustic waves.
There exist theoretical models that allow the prediction of such effects. Theoret-
ical studies on this problem are well established starting with the work of Rudnick 15
and Ingard 16 in the early 1950s. More recent work on the subject was done by
Attenborough 17;18 . A detailed method of calculation is given by ISO 9613-2 19 .
Consider the source-receiver arrangement shown in Figure 17.4. The distance
R1 denotes the distance of a direct wave from the source to the receiver; R2 denotes
the distance from source to receiver along an indirect wave reflected by the ground.
The incident angle φ is equal to the reflection angle. This condition gives a unique
travel path of the reflected acoustic wave.
The model presented here is based on Attenborough 20 . The ground is considered
as an infinite boundary with a given impedance. If Z is the ground impedance,
H = 1/Z is the admittance. The impedance is a function of a number of parameters,
namely the flow resistivity σe and the inverse effective depth of the surface layer, α.
The resultant sound field at the observer’s location due to a single-frequency source
is represented by the following total velocity potential:
eik1 R1 eik1 R2
tot = + Q, (17.8)
4π R1 4π R2
Figure 17.4. Ground reflection of acoustic waves.
where k1 denotes the acoustic propagation coefficient in the air. The second term in
Equation 17.8 includes the relative image source strength Q, which is a proportion-
ality factor of the reflected wave being superposed with the direct wave. This factor
can be interpreted as the spherical-wave-reflection coefficient. From Equation 17.8,
the resulting pressure field is

∂tot
ptot = −ρ . (17.9)
∂t
Assume that β denotes the normalised specific admittance, equivalent to the inverse
of the specific impedance Z; M is the density ratio, n is the propagation coefficient
(or wave number) ratio, and w is a numerical distance. If subscripts “1” and “2”
denote the upper medium (air) and the lower medium (ground), respectively, these
quantities are defined as
ρ1 k1
M= , n= , β = nM. (17.10)
ρ2 k2
Furthermore, to simplify the algebra, set the following equivalences:

1 sin θ 2
n1 = 1 − 2 , s1 = 1 − , β1 = 1 − β 2 , M1 = 1 − M2 .
n n
(17.11)
With this nomenclature, the reflection coefficient is
cos θ − β(1 − s1 )1/2
Rp = . (17.12)
cos θ + β(1 − s1 )1/2
The image source strength is
Q = [Rp + B(1 − Rp )] F(w). (17.13)
The boundary-loss factor is

√
F(w) = 1 + i π we−w erfc(−iw).
2
(17.14)
In Equation 17.13, the factor B is defined by
) *
1/2 1/2
! n 1 M 1 + β cos θ s 1 + sin θβ1 cos θ + βs1
B= × √ . (17.15)
s1 M13 2 sin θβ1 cos θ + β n1 /M1
The propagation coefficient on the ground k2 can be approximated by using the
following equation 18 :
−0.7 −0.595
k1 f f
= 1 + 0.0978 + i0.189 . (17.16)
k2 σe σe
The previous formulation follows the same approach as radio propagation; it utilises
a numerical distance parameter w, re-derived 20 as
√ √
β cos θ n1 − sin θ n1
w = ik1 R2 1 + √ . (17.17)
M1
Earlier studies on acoustic ground reflection used asymptotic expansions as a means
to rapidly evaluate F(w). Given that w = w 2 , two different series are used for
different values of w divided at |w | = 10:
⎧
⎪
⎪ √
w n
⎪
⎪ 1 + i πw e −w
− 2w 1 + e−w if |w | < 10
⎪
⎨ (2n + 1)(n − 1)!
n
F(w ) =
⎪
⎪
⎪
⎪ 1 1×3 1×3×5
⎪
⎩− + + + · · · if |w | ≥ 10
2w (2w )2 (2w )3
(17.18)
This approach has been shown to produce errors at small values of w and is unstable
particularly with |w | close to 106 . To overcome this problem, part of the boundary-
loss factor is evaluated instead as the complex-error function, WF (w), defined as
WF (w) = e−w erfc(−iw).

2
(17.19)
The complex-error function, also known as the Faddeeva function, can be efficiently
computed to within machine precision using an algorithm developed by Poppe and
Wijers 21 . The effect of air turbulence is incorporated into the model using a coher-
ence parameter . It takes into account the effect of turbulence on the coherence
between the direct wave and the reflected wave:
= e−0.25β P , (17.20)
where
⎧ √
⎨ 0.5 if D/k1 >
β = (17.21)
⎩ √
1.0 if D/k1 <
and
√
P =< μ2 > k12 D π . (17.22)
The factor < μ2 > is the fluctuating index of refraction, and is the turbulent length
scale 22 .
Following Chessel 23 , the difference between the sound pressure levels for a
single wave frequency measured with the presence of the ground relative to that
obtained without the ground (i.e., the free-field SPL) is given by
2
R1 R1 2R2 − R1
SPL = 10 log10 1 + |Q| + 2 |Q| cos +α , (17.23)
R2 R2 k1
with Q = |Q|eiα . A positive SPL value indicates that the sound due to the direct
wave is enhanced due to the presence of the ground, whereas a negative value means
that the actual SPL is lower than that of the free-field.
SIMPLIFIED MODEL. Most other first-order models typically assume a number of sim-
plifications to arrive at more compact solutions that are relatively easier to arrive at in
the derivation and are attractive for fast numerical computation. Typical simplifying
assumptions are |n| > 1, |M| << 1, |β| < 1 (hard-boundary case), and φ 0 (small
grazing angles). In addition to these, the ground impedance can be further simplified
by treating the ground as a locally reacting boundary, where the wave propagation
in the ground is independent of the incident angle that leads to |n| >> 1. Following
these simplifications, one would arrive at the reduced definitions for some of the
parameters above, with B 1:
cos θ − β
Rp = , (17.24)
cos θ + β
"
ik1 k2
w= (cos θ + β) . (17.25)
2
This solution predicts the ground-reflection correction, SPL, for a single-frequency
sound wave. In practice, most measurements are recorded in the 1/3 octave band
levels, which integrate the single-frequency sound levels within each band. To obtain
an octave band reading from the SPL calculated herein, each band is subdivided
into smaller bands where the SPL value for each sub-band centre frequency is
calculated. These sub-band SPL are then multiplied with their sub-band widths and
integrated within each 1/3 octave band. The integrated values are then normalised
with the corresponding 1/3 octave band-widths.
17.3.1 Ground Properties

The acoustic properties of the ground are represented by the complex impedance.
Earlier research in this field modeled the ground impedance with a single parameter,
the specific flow resistance per unit thickness, based on the work of Delaney and
Bazley 24 . The model relies on curve fitting of various experimental measurements of
actual ground impedances. Attenborough 25 shows that the single-parameter model
is unreliable, particularly outside the range of the fitted data; he then introduced
a more accurate model using two parameters: the effective flow resistivity, σe , and
the rate of exponential decrease of porosity with depth, α. The specific normalised
Table 17.2. Typical values for flow resistivity

and inverse effective depth
Ground σe [kNs/m4 ] α [m−1 ]
Snow 5 – 20 0
Short grass 30 – 50 20 – 45
Sandy soil 60 – 100 0
Dirt road 40 10
Wet compact soil 4,000 0
Tarmac 4,500 0
impedance Z based on the two-parameter model is given by the following equation:

1/2 1/2
σe σe aα
Z= +i + . (17.26)
γ πρ f γ πρ f 4γ π f
Practical values for σe and α are given in Table 17.2.

Comparisons between the two approaches with experimental data are shown in
Figure 17.5. Evidence from these comparisons shows that the two-parameter model
is better at accurately predicting the corrected sound levels. A weakness of the single-
parameter model is that it cannot accurately predict the primary-dip amplitude; it is,
however, able to match the primary-dip frequency. The predictions are better with
the two-parameter model given that the correct values of both σe and α are used.
17.3.2 Turbulence Effects

Most noise calculations rely on the assumption of still atmosphere (no winds, no
turbulence). This is in fact far from true and, to be realistic, even a simple turbulence
model would be of considerable benefit. There are several models, but the simplest
10 10
5 5
0 0
Δ SPL, dB
Δ SPL, dB
-5 -5
-10 -10
Exp. Data Exp. Data

-15 2 params, α = 70/m -15 2 params, α = 70/m
2 params, α = 8/m 2 params, α = 8/m
1 param 1 param
-20 -20
102 103 104 102 103 104
(a) (b)
Figure 17.5. Comparison between present method and data from Rasmussen 26 .
one is based on the evaluation of the fluctuating index of refraction < μ2 >:
σv2 σt2
< μ2 >= + , (17.27)
a2 4T 2
where σv is the root mean square (RMS) of the wind-velocity fluctuation and σt is
the RMS of the temperature fluctuation. The effect of turbulence on the acoustic
wave refraction is more pronounced at frequencies below 5kHz.
17.4 Wind and Temperature Gradient Effects

The effects of atmospheric winds on the propagation of aircraft noise are essential.
For example, receiver positions downwind experience more annoyance than receiver
positions upwind. The differences can be several dB. There is a combination between
wind and ground effects, which cannot be separated. The first step in a practical
analysis is to assume an average wind-shear profile, in absence of turbulence and
local topographical effects, which is
ζ
Vw (z) z
= , (17.28)
Vwo zo
where Vw (z) denotes the horizontal wind speed at height z, Vwo is the wind speed at
height zo, and ζ is the wind-shear exponent; this exponent depends on the type of
terrain, with 0.1 < α < 0.25. In practice, zo will be an altitude above the ground at
which wind speeds are available. A typical reference altitude is zo = 10 m.
The degree of the frequency-dependent attenuation or enhancement depends
on a number of factors, that is, horizontal propagation distance, wind speed and
direction (Vw and θw ), temperature variation with height (dT /dh), and source and
receiver heights (Hs and Hr ). The presence of temperature variations produces a
speed-of-sound gradient γc . An increase in the speed of sound from source to receiver
would refract the sound downward and generally enhance the sound received. This
is due to a positive sound-speed gradient resulting from downwind propagation
or temperature inversion (i.e., an increase of temperature with height). An upwind
propagation or a normal decrease of temperature with height (i.e., negative γc ) results
in upward-refraction that produces a shadow region downstream; sound levels are
mostly attenuated in this case. Figure 17.6 illustrates these sound-refraction effects
on sound rays emanating from a single source.
A theory on the combination of wind, temperature gradient and ground reflec-
tion has been developed by Rasmussen 27 , who expressed the acoustic pressure field
with a Hankel transform
∞
p = −2 Jo(κ D)P(z, κ) κdκ, (17.29)
o
where Jo is the Bessel function of the first kind of zeroth order, and P(z, κ) is the
transform of p, D is the horizontal distance, r is the propagation distance (see also
Figure 17.4). In terms of pressure ratio relative to the free stream, the acoustic
pressure is
∞
p
= −2r Jo(κ D)P(z, κ) κdκ. (17.30)
po o
Figure 17.6. (a) Upward refraction due to upwind sound propagation or temperature inversion
(γc < 0), and (b) downward refraction due to downwind propagation or normal temperature
variation (γc > 0).
Following derivations by Pierce 28 and Rasmussen, with the assumptions that the
sound speed varies linearly with the height and that the speed-of-sound gradient
γc z << 1, two expressions for the pressure transform are obtained, depending on
the sound-speed gradient. The vertical-density variation is assumed to be negligible
in the derivation; for γc > 0 and γc < 0, we have respectively:

w (τ ) + qw(τ )
P+ (z, κ) = −ν(τ + yL) w(τ + yS ) − w(τ − yS ) L (17.31)
ν (τ ) + qν (τ )

w (τ ) − qw(τ )
P+ (z, κ) = −w(τ − yL) − w(τ − yS ) − w(τ − yS ) L, (17.32)
ν (τ ) − qν (τ )
where
−1/3
L = 2|γc |ko2 , τ = k2 − ko2 L, (17.33)
min(Hs , Hr ) max(Hs , Hr )
yS = yL = q = ikoβL, (17.34)
L L
and ko is the wave number of the sound propagation on the ground; i.e., ko = 2π f/a.
The pressure-transform solutions utilise the Airy-Fock functions 29 , ν(τ ), w(τ ) and
Figure 17.7. Downwind and upwind

sound propagation.
Ai(τ ). These functions are expressed by

√
ν(τ ) = πAi(τ ), (17.35)

√ iπ 2π
w(τ ) = 2 π exp Ai τ exp i . (17.36)
6 3
For values of |τ | > 3, the Airy-Fock functions can be substituted by some
approximations2 , restricted to small speed-of-sound gradients – a condition eas-
ily met in outdoor noise propagation. The gradient itself is calculated from a set of
temperature, wind speed and wind direction measurements on the ground at heights
z1 and z2 , respectively.
Figure 17.7 illustrates the wind vectors relative to propagating rays at the meas-
uring height; the wind speed on the ground is zero. The linear sound-speed gradient
is
a(z2 ) − ao
γc = , (17.37)
ao z2
where ao is the speed of sound at ground temperature and
a(z2 ) = a(z2 ) + Vw (z2 ) sin θw . (17.38)
For a given wind direction, during a typical trajectory an aircraft can change its
position from downwind to upwind (or vice versa) with respect to a fixed receiver
on the ground. How this may take place is best explained with the aid of Figure 17.8.
The aircraft is travelling from left to right (or West to East) with a head wind. In the
first instance, the aircraft is to the West of the receiver; thus, the receiver is placed
upwind. When the aircraft has moved to the East of the receiver, the receiver itself
is placed downwind.
17.4.1 Numerical Solution

To obtain the difference in sound pressure levels SPL relative to the free-field,
a modified expression for pressure ratio in Equation 17.30 can be integrated
Wind
Flight R
S
Wind
R
S
Figure 17.8. Receiver position (R) moving from upwind to downwind at fly-over. Aircraft (S)
travelling from West to East.
numerically. The efficiency of the integration depends on a number of factors, one

of which is the truncated terms for the lower and upper limits of the integral, a and
b, such that
b
p
SPL = 20 log10 = 20 log10 −2r Jo(κ D)P(z, κ) κdκ . (17.39)
po a
Rasmussen provided a recommendation on the limits of integration as well as the

discretisation size dκ for use on his datasets whose frequencies range from 100 Hz
to 2 kHz only. The rules break down for wider and higher frequency ranges. It is
found that a new set of rules can be applied to optimise the integration parameters
by analysing the behaviours of the Bessel function Jo and the pressure transform
P(z, κ); Jo governs the oscillations of the functions to be integrated for the upwind
propagation. In the downwind propagation, the oscillations are dominated by the
pressure transform P(z, κ). The discretisation size is optimised by setting
dκ = kp /3D (17.40)
with

1 if γc < 0
κp = . (17.41)
ko2 + (6π|γc |ko2 )2/3 − ko if γc > 0
The lower limit of integration is a = 0 for all values of γc ; the upper limit is

b = ko2 + 1/Pave 2 (17.42)
with Pave = 0.5 for γc > 0 and Pave = 0.04 for γc < 0. The number of computational
points is set to nk = b/dκ. Following this approach, the computational cost increases
with the horizontal propagation distance and the frequency of propagation, particu-
larly for the downward-refracting case where the discretisation size is much smaller
as imposed by P(z, κ). The computational efficiency can be improved significantly by
reducing the terms to be computed through algebraic simplification of the Airy-Fock
functions and various approximations.
This solution describes the sound correction for a single frequency. To obtain
a 1/3 octave band reading of the SPL, each band is subdivided into smaller bands
where the SPL value for each sub-band centre frequency is calculated. These
sub-band SPLs are then multiplied with their sub-band widths and integrated
15
10
ΔSPL 5
-5
V = 0 m/s
-10 V = 1 m/s
V = 3 m/s
V = 5 m/s
-15 1
10 102 103 104
Frequency, Hz
Figure 17.9. SPL for downwind propagation over grassy field with σe = 200 · 103 Nsm−4 .
Input data are equivalent to those in Rasmussen 27 ; wind speeds as indicated.
within each 1/3 octave band. The integrated values are then normalised with the
corresponding 1/3 octave bandwidths.
MODEL VERIFICATION. In Figure 17.9, the SPL received at a location 120 m down-
stream of a source for various wind speeds are plotted for a wide frequency range
of 10 Hz < f < 10 kHz. The input data are identical to those used by Rasmussen 27 .
The corresponding results below match his numerical solutions, which are presented
within a narrower frequency band of 100 Hz to 2 kHz, with the exception for slightly
different primary-dip magnitudes. These small variations can be attributed to the
different approaches of setting the integration parameters, as described in § 17.4.1.
Comparisons with experimental data in Figures 17.10 and 17.11 show that the
trends of the simulated attenuation curves are similar to those measured. However,
their magnitudes vary considerably to within 7 dB for the downward-refracting case
and within 8 dB for the upward-refracting case. This discrepancy can be largely
attributed to two factors. First, there is a possible mismatch between the simulated
ground, wind, and atmospheric parameters and the actual parameters at the time the
real data were recorded. For example, only the temperature-gradient regime (i.e.,
lapse, neutral or inversion) and the actual temperatures were reported by Parkin
and Scholes 30 but not the actual temperatures at various heights. In addition, the
data were recorded in various seasons of the year. The ground parameters, which
may vary across the field, were not reported either. Reported uncertainties of the
measurements are within the range of the differences observed in the simulated
values. The uncertainty on the data varies from ±0.6 dB to ±10.3 dB, depending
on the atmospheric conditions, measurement sites, time of the year and receiver
distance.
20
Fine scale simulation
1/3 Octave band
Parkin & Scholes
10
ΔSPL
0
-10
-20 1
10 102 103 104
Frequency, Hz
Figure 17.10. Comparison of the SPL attenuation with the experimental data of Parkin and
Scholes 30 . The main inputs are D = 109.73 m (360 feet), Hs = 1.83, Hr =1.52 m, Vw = 4.57
m/s downstream, and σe = 200 · 103 Nsm−4 .
0
ΔSPL
-10
-20 Fine scale simulation

1/3 Octave band
Parkin & Scholes
1 2 3 4
10 10 10 10
Frequency, Hz
Figure 17.11. Comparison of the SPL attenuation with the experimental data of Parkin and
Scholes 30 . The main inputs are identical to those in Figure 17.10; wind speed Vw = −4.57 m/s.
Summary
This chapter has introduced some numerical models for the prediction of import-
ant effects in sound propagation. Because the distance source-to-receiver can be
of the order of a few kilometers (or n-miles), the noise recorded at the receiver
can have a completely different acoustic content from the noise sources. The most
prominent effect is the atmospheric absorption in still and unbounded atmosphere
Bibliography 549
with a constant level of humidity. The standard model indicates that high frequencies
are cut more rapidly than low frequencies. A distant airplane can be identified by
the acoustic pressures at low frequencies. Yet, this model is not sufficient to take
into account all of the effects, which include ground reflection (or absorption), tem-
perature differentials and most critically the wind. A model that combines wind and
temperature gradients has been implemented in a numerical form and extended to
the prediction of far-field noise. This model is computationally more expensive than
the noise sources themselves and requires serious computer hardware even with the
approximation of wind shear not dependent on the topography.
Bibliography
[1] Rawlins AD. The engine over wing noise problem. J. Sound & Vibration,
50(4):553–569, 1977.
[2] Gerhold CH. Investigation of acoustical shielding by a wedge-shaped airframe.
[3] Agrawal A and Dowling A. The calculation of acoustic shielding of engine
noise by the silent aircraft airframe. In 11th AIAA/CEAS Aeroacoustics Conf.,
Monterey, CA, May 2005.
[4] Ochmann M. The source simulation technique for acoustic radiation problems.
Acustica, 81:512–527, 1995.
[5] Dunn MH and Tinetti AF. Aeroacoustic scattering via the equivalent source
method. In 10th AIAA/CEAS Aeroacoustics Conference, AIAA Paper 2004-
2937, May 2004.
[6] Dunn MH and Tinetti AF. Application of fast multipole methods to the NASAx
fast scattering code. In 14th AIAA/CEAS Aeroacoustics Conference, AIAA
Paper 2008-2875, Vancouver, BC, May 2008.
[7] Pinho MEV. On the use of the equivalent source method for nearfield acous-
tic holography. ABCM Symposium Series in Mechatronics, pages 590–599,
2004.
[8] Schroeder H, Gabillet Y, Daigle GA, and L’Experance A. Application of the
Gaussian beam approach to sound propagation in the atmosphere: Theory and
experiments. J. Acoust. Soc. Am., 93(6):3105–3116, 1993.
[9] Lui WK and Li KM. The scattering of sound by a long cylinder above an
impedance boundary. J. Acoust. Soc. Am., 127(2):664–674, 2010.
[10] Zuckerwar AJ and Meredith RW. Low-frequency sound absorption measure-
ments in air. Technical Report R-1128, NASA, Nov. 1984.
[11] Anon. Method for the calculation of the absorption of sound by the atmosphere.
Technical Report S1.26-1978, American National Standards Institute, June
1978.
[12] Anon. Attenuation of sound during propagation outdoors – Part I: Calculation
of the absorption of sound by the atmosphere. Technical Report ISO-9613-1,
International Standards Organisation, Geneve, CH, 1993.
[13] ESDU. Prediction of Sound Attenuation in a Refracting Turbulent Atmosphere
with a Fast Field Program. Data item 04011. ESDU International, London,
May 2004.
[14] Morfey CL and Howell GP. Nonlinear propagation of aircraft noise in the
atmosphere. AIAA J., 19(8):986–992, Aug. 1981.
[15] Rudnick I. The propagation of an acoustic wave along a boundary. J. Acoust.
Soc. Am., 19:348–356, 1947.
[16] Ingard U. On the reflecton of a spherical sound wave from an infinite plane. J.
Acoust. Soc. Am., 23:329–335, 1951.
[17] Attenborough K. Review of ground effects on outdoor sound propagation from
continuous broadband sources. Applied Acoustics, 24:289–319, 1988.
[18] Attenborough K. Sound propagation close to the ground. Ann. Rev. Fluid
Mech., 34:51–82, Jan. 2002.
[19] Anon. Attenuation of sound during propagation outdoors – Part II: General
method of calculation. Technical Report ISO-9613-2, International Standards
Organisation, Geneve, CH, 1996.
[20] Attenborough K. Propagation of sound above a porous half-space. J. Acoust.
Soc. Am., 68:1493, 1980.
[21] Poppe GPM and Wijers CMJ. More efficient computation of the complex error
function. ACM Trans Math. Software, 16, 1990.
[22] Daigle GA, Piercy JE, and Embleton TFW. Line-of-sight propagation through
atmospheric turbulence near the ground. J. Acoust. Soc. Am., 74:1505–1513,
1983.
[23] Chessel CI. Propagation of noise along a finite impedance boundary. J. Acoust.
Soc. Am., 62:825–834, 1977.
[24] Delaney M and Bazley E. Acoustical properties of fibrous absorbent materials.
Appl. Acoustics, 3:105–116, 1970.
[25] Attenborough K. Ground parameter information for propagation modeling. J.
Acoust. Soc. Am., 92(1):418–427, Jan. 1992.
[26] Rasmussen KB. Sound propagation over grass covered ground. J. Sound &
Vibration, 78(2):247–255, 1981.
[27] Rasmussen KB. Outdoor sound propagation under the influence of wind and
temperature gradients. J. Sound & Vibration, 104:321–335, 1986.
[28] Pierce AD. Acoustics: An Introduction to Its Physical Principles and Applica-
tions. Acoust. Soc. America, 1989.
[29] Hazewinkiel M. Encyclopaedia of Mathematics, chapter Airy Functions,
page 65. Kluwer Academic, 1995.
[30] Parkin PH and Scholes WE. The horizontal propagation of sound from a jet
engine close to the ground, at Hatfield. J. Sound & Vibration, 2, 1965.

Symbols listed on Table 17.1 are not repeated here.
a = speed of sound; lower limit of integration in Equation 17.39
A = atmospheric absorption
Ai(τ ) = Airy function
b = upper limit of integration, Equation 17.42
B = parameter in ground-reflection model, Equation 17.15
D = ground distance between source and receiver in ground reflection
f = frequency
fr n = factor given by Equation 17.4
fr o = factor given by Equation 17.3
F = absorption function
F(w) = boundary-loss factor, Equation 17.18
k = wave number
k1 , k2 = acoustic-propagation coefficient of air/ground
h = flight altitude
hc = molar concentration
H = ground admittance, H = 1/Z
Hr , Hs = height of receiver/source above ground
i = imaginary unit
Jo = Bessel function of first kind, zero-th order
k = wave number
ko = wave number of the sound propagation on the ground
= turbulence length scale
L = parameter defined in the first Equation 17.33
M = ratio between air densities, M = ρ1 /ρ2
M1 = parameter defined in Equation 17.11
n = propagation coefficient
n1 = parameter defined in Equation 17.11
nk = number of computational points
p = pressure; acoustic pressure
pa = atmospheric pressure
pr = reference pressure
psat = saturation pressure
ptot = total acoustic pressure
P = acoustic power
Pave = average values of the pressure transform for convergence
P(z, κ) = pressure transform in wind-propagation effects
q = function of the wave number ko, Equation 17.34
Q = velocity potential source strength
r = distance
R = gas constant
Rp = reflection coefficient, Equation 17.12
R1 ,R2 = distances in ground-reflection model, Figure 17.4
s1 = parameter defined in Equation 17.11
t = time
T = temperature
T01 = triple-point temperature Equation 17.7
V = exponent in Equation 17.6, given by Equation 17.7
Vw = wind speed
Vwo = wind speed at altitude z = zo
w = numerical distance parameter, Equation 17.25
w(τ ) = Airy-Fock function
w = square of the numerical distance, w = w 2
WF = complex-error function
x = sampling point on surface boundary
yL,yS = functions of receiver/source height, Equation 17.34
z = vertical coordinate
zo = reference altitude in wind-shear model, Equation 17.28
Z = ground impedance
Greek Symbols
α = inverse effective depth, Equation 17.26
β = specific normalised admittance of the ground
β1 = parameter defined in Equation 17.11

β = function of coherence parameter , Equation 17.21
θ = polar-emission angle
θ = angle between incident wave and normal to ground, θ = 1 − φ
ζ = wind-shear exponent, Equation 17.28
γ = ratio between specific heats, γ = C p /Cv
γc = speed-of-sound gradient
κ = variable of integration, Equation 17.29
κp = parameter defined by Equation 17.41
dκ = discretisation size
= coherence parameter, Equation 17.20
μ = semi-empirical parameter in landing-gear noise
μ2 = fluctuating index of refraction, Equation 17.27
ν(τ ) = Airy-Fock function
ρ = air density
φ = angle between indirect acoustic wave and the ground, φ = 1 − θ
= total velocity potential, Equation 17.8
σe = effective flow resistivity
σt = root mean square of temperature fluctuation
σv = root mean square of wind-velocity fluctuation
τ = argument of Airy-Fock functions, τ = (k2 − ko2 )L
χ = angle between wind vector and source-receiver line
ω = angular frequency
[.]r = receiver
[.]s = source
[.]sat = saturation
[.]tot = total
[.]w = wind
[.]o = reference or ground conditions
18 Aircraft Noise: Flight Trajectories
Overview
In Chapter 16 we discussed various methods that can be used for the rapid prediction
of aircraft-noise sources. In Chapter 17 we described the effects of atmospheric
propagation, interference and scattering. In this chapter we deal with aircraft-noise
trajectories. In particular, we discuss noise certification (§ 18.1) and noise-abatement
procedures (§ 18.2). The implementation of the various sub-models requires a flight-
mechanics integration (§ 18.3) and a data-handling structure, to process the large
matrices produced by several noise sources. A noise sensitivity analysis is presented
(§ 18.4) to show the relative importance of the various noise sources to explain
the most useful strategies for noise reduction. We discuss two case studies: the
ICAO/FAR noise trajectories of a jet-powered aircraft (§ 18.5) and the ICAO/FAR
noise trajectories of a turboprop aircraft (§ 18.6). We also present a number of
noise applications, such as steep descent and wind effects (§ 18.7). To complete our
discussion of the noise modelling, we provide an example of verification with flight
data (§ 18.8). The last item of investigation is the noise footprint around the airfield
(§ 18.9).
KEY CONCEPTS: Aircraft Noise Certification, Noise-Abatement Procedures, Flight-

Mechanics Integration, Noise Sensitivity, Noise Trajectories, Wind Effects, Foot-
prints, Sonic Boom.
18.1 Aircraft Noise Certification

Unlike a Beethoven symphony, aircraft noise is not welcome music to the ear.
The second movement of the Ninth Symphony does not start with a landing-gear
deployment. Communities that live near large airports know this fact rather well. In
fact, they have been making noise on their own by voicing their concern.
Aircraft noise has been increasing over time, but the really loud noise started
with the jet-powered aircraft in the 1950s. Their presence prompted noise limitations
at Heathrow Airport in London as early as 1959. London Gatwick followed suit in
1968, and the restrictions expanded worldwide, as shown in Figure 18.1. Today
commercial aircraft operations include limits on night flights and restrictions on
553
1,000
Airports with Noise Restriction

800
600
400
200
0
1960 1970 1980 1990 2000 2010
Year
Figure 18.1. Worldwide trend in airports with noise restriction.
flight paths. Considerable research is available on health effects 1 . As a result, taxes

are levied on the noisiest aircraft on most airports; charge models have been in place
since the 1980s 2 to compensate for marginal damage.
Research into aircraft noise started in earnest in the late 1960s. In 1965,
Gebhardt 3 reported on the noise problems of the Boeing B727 that had entered
service one year earlier. Crighton 4 estimated that the noise level of a first-generation
Boeing B737 (1965) was the same as the world population shouting at once, whilst
a second-generation aircraft would only produce the same noise as the city of New
York shouting at once. Concorde was the loudest of all aircraft; its noise perform-
ance did not comply with even the early noise regulations; exceptions were made
for its case, and the aircraft was eventually prevented to fly at supersonic speeds
over land. Its flight operations sparked mass protests in the 1970s 5 . For military
vehicles, where detection is a critical issue, lower frequencies are also important
because low frequencies have lower attenuation rates (§ 17.2) and therefore can
be heard over long distances. There is also concern about operations from military
airports 6 .
The technical literature is rich in information regarding noise-certification issues,
regulations, data and procedures, for example Refs. 7;8 . Various national and inter-
national organisations are involved in one form or another in setting up working
groups, regulations and strategies. It is not possible to review these issues in any
detail at this point. It suffices to mention that the problem of aircraft noise involves a
number of factors beyond the aircraft itself, such as the level of traffic at an airfield.
Smith 5 is a fairly good source of historical information. It includes some discussion on
noise sources (power plants, fuselage, propellers, gas turbines), data acquisition and
performance prediction, sonic boom, relevant historical notes and a very extensive
bibliography. Several aviation organisations have considerable amounts of technical
publications (continuously updated) that cover any aspect of aircraft noise.
It has now been established that the best approach to aircraft-noise reduction
must be achieved through a balanced approach* that takes into account all of the
aspects of the operation of the aircraft, including:
r noise reduction at the source (or aircraft design)

r improvements in land use and urban developments
r establishment of optimal flight procedures
r restriction of operations (as shown in Figure 18.1)
At the operational level, there is an increasing need for assessing noise perform-
ance for new and extended runways, changes in operational procedures, increase in
traffic at airports and changes in air-space allocations.
The certification of aircraft noise relies on measurements at recognised points
around the airfield. Microphones record the noise signals over the complete flight
trajectory and a software-hardware interface calculates the effective perceived noise
EPNL. Specifically, the ICAO has established three measuring points, which provide,
respectively, the take-off noise, the lateral noise, and the final approach (landing)
noise. For this purpose, the standard position of the microphones is defined as
follows:
r Take-Off Noise: The microphone is placed at the centre of the runway, 6,500
m down from the brake-release point. The aircraft must have the retracted
landing-gear configuration, with a speed greater than V2 + 10 kt. Furthermore,
thrust cut-back must take place at altitudes above 210 m (∼690 feet); after a
thrust cut-back, the aircraft must maintain a flight-path angle greater than 4
degrees (Figure 18.2a).
r Lateral Noise: The microphone is placed 450 m (∼1,480 feet) off the centre of the
runway, where the aircraft noise at take-off is maximum. The aircraft must have
configuration with landing gear retracted, with a speed greater than V2 + 10 kt
(Figure 18.2b) and maximum thrust. Note that in this case several microphones
parallel to the runway are required to establish the maximum noise level.
r Landing (Approach) Noise: The microphone is placed 2,000 m (∼6,560 feet)
upstream of the runway threshold, with the aircraft on a −3 degree slope flight
path ending 300 metres beyond the threshold (Figure 18.2c). On a level ground,
the microphone will be 120 metres below the aircraft. The aircraft must have its
landing gear out, the flaps deployed and an air speed of VS + 10 kt.
The case of lateral noise requires further discussion. In fact, the regulations
only specify the lateral position of the microphone; the longitudinal position is
indeterminate. A set of microphones is placed along a line parallel to the runway,
as shown in Figure 18.2b. The noise level is recorded at all of the microphones,
and the resulting sideline noise level is the maximum value among the recordings
(interpolation of data may be required).
Table 18.1 shows the position of the microphones for community noise at
London Heathrow. The data are, respectively, runway number, reference point,
* ICAO: Assembly Resolution A35-5, Appendix C, Doc 9848 (2001).

Thrust cut-back
o
4
Brake-Release
6,500 m
(a)Take-off
Brake-Release
45
0
m
45
0
m
R
(b)Sideline
120 m
3o
R
ld
2,000 m
s ho
re
th 300 m
(c)Landing
Figure 18.2. Standard noise-measuring stations according to ICAO Chapter 3.
elevation above sea level, elevation above airfield (EAA), latitude, longitude, dis-
tance from brake-release, limit correction, maximum day-time (07:00-23:00) and
maximum night-time (23:00-07:00) noise.
The actual flight paths are affected by atmospheric conditions, gross weight and
local air traffic procedures. Therefore, there is always a dispersion in the flight paths,
both vertically and on the ground tracks. More realistically, when assessing real flight
trajectories, we must consider flight-corridor boundaries.
Table 18.1. Microphone positions for aircraft-noise measurements at London Heathrow
RWY Ref. Elev EAA Lat Long Dist Limit Day Night
[m] [m] (N) (W) [km] [dB] [dB] [dB]
6 18 −6 51 27 56 00 03 15 7 6.58 0.3 94.4 87.4

A 20 −4 51 28 57 00 03 12 4 6.25 2.3 97.0 90.0
B 20 −4 51 28 40 00 03 11 7 6.00 4.8 99.5 92.5
27L/R C 18 −6 51 28 14 00 03 14 8 6.58 −0.3 94.4 87.4
D 17 −7 51 27 45 00 03 20 1 6.83 −0.6 94.1 87.1
E 17 −7 51 27 24 00 03 21 6 7.20 −1.0 93.7 86.7
F 21 −3 51 28 21 00 02 34 5 6.40 0.9 95.6 88.6

G 21 −3 51 28 06 00 02 33 8 6.50 −0.1 94.6 87.6
09R H 21 −3 51 27 45 00 02 34 0 6.37 1.2 95.9 88.9
I 20 −4 51 27 14 00 02 34 1 6.60 −0.3 94.4 87.4
The reference atmospheric conditions are pressure po = 101,325 Pa; air temper-
ature 25◦ C; relative humidity H = 70%; zero wind. Actual tests must be carried out
under the following conditions: no precipitation; air temperature between −10 and
+35◦ C; relative humidity 20% < H < 95%; wind speed less than 12 kt; cross-wind
speed less than 7 kt (measured 10 m above the ground); no anomalous conditions
that may affect the measurements.
There are a variety of operational conditions that can affect the measurements.
For this reason, the certification requires several sets of measurements, data reduc-
tion and corrections.
Sound measurements are taken with microphones placed 1.2 m above the
ground. Comparisons between measured and estimated data are made difficult by
the great variability in atmospheric conditions as well as conditions on the ground
(terrain shape, airfield altitude, presence of large obstacles). Some of these diffi-
culties can be overcome if the microphones are placed on the ground because the
ground reflection is then removed.
The maximum noise levels allowed by the ICAO depend on the number of
engines and on the maximum take-off weight. The rules for Chapter 3 (valid until
2006) for take-off noise were as follows:
⎧⎧
⎪
⎪ ⎨ 89 m < 48.1
⎪
⎪
⎪
⎪ 66.65 + 13.29 log m 48.1 < m < 385 2 engines
⎪
⎪ ⎩
⎪
⎪ 101 m > 385
⎪
⎪
⎪
⎪ ⎧
⎪
⎨ ⎨ 89 m < 28.6
EPNL(dB) = 69.65 + 13.29 log m 28.6 < m < 385 3 engines (18.1)
⎪
⎪ ⎩
⎪
⎪ 104 m > 385
⎪
⎪
⎪
⎪ ⎧
⎪
⎪ ⎨ 89 m < 20.2
⎪
⎪
⎪
⎪ 71.65 + 13.29 log m 20.2 < m < 385 4 engines
⎪
⎩⎩
106 m > 385
The maximum noise is 89 dB at weights below the thresholds (all engines); the
maximum noise is 101 dB (2 engines), 104 dB (3 engines), and 106 dB (4 engines) at
110 110
4 Engines
Landing
3 Engines
ICAO Chapt 3 limit EPNL, dB

ICAO Chapt 3 limit EPNL, dB
105 105
Sideline
2 Engines
100 100
95 95
90 90
NOTE: for ICAO Chapt. 4 subtract 2dB Note: for ICAO Chapt. 4 limits subtract 2dB
85 85 4
10 4
10 5
10 6
10 105 106
Mass, kg Mass, kg
(a)Take-off (b) Sideline & Landing
Figure 18.3. ICAO Chapter 3, noise limits at take-off, sideline and landing/approach.
weights above the thresholds (385 tons). The sideline noise limits are:
⎧
⎨ 94 m ≤ 35
EPNL(dB) = 80.87 + 8.51 log m 35 < m < 400 (18.2)
⎩
103 m ≥ 400
The landing (approach) noise limits are:

⎧
⎨ 98 m ≤ 35
EPNL(dB) = 86.03 + 7.75 log m 35 < m < 280 (18.3)
⎩
105 m ≥ 280
In Equation 18.1, Equation 18.2 and Equation 18.3, the mass m is expressed in
1,000 kg. A graphic form of these equations is shown in Figure 18.3.
Chapter 3 of the ICAO regulations (now expired) allowed a trade-off in case of
non-compliance at one reference point. The more stringent regulations now enforced
with the ICAO Chapter 4 require that none of the noise levels can exceed the levels
defined in Chapter 3. Furthermore, the aircraft must demonstrate a cumulative
margin of at least 10 dB at all reference points. This means that for a given weight,
the sum of the noise levels (take-off, sideline, approach) should be at least 10 EPNL
(dB) below the sum of the maximum noise levels calculated with Equation 18.1,
Equation 18.2, and Equation 18.3. The margin on any single point (take-off, sideline,
approach) over Chapter 3 regulations must be at least 2 EPNL (dB).
We could argue whether the regulations have fostered progress in noise-
abatement technology or progress in aircraft engineering has led to noise abatement
which was eventually endorsed by the regulations 5 . However, if we look at a time-
line showing the progress in noise reduction and plot the data against the regulations,
we find a striking correlation, which is shown in Figure 18.4. This figure shows a trend
of the certified noise levels on average for selected commercial subsonic jets. The
graph shows how, as a combination of more efficient engine and airframe design,
the noise levels have been reduced considerably from the first generation of jet-
powered aircraft. By the time the ICAO Chapter 4 was introduced, most aircraft
100
ICAO Stage 3 ICAO Stage 4
Certified noise EPNL, dB

95
90
85
ICAO Stage 2
80
1960 1970 1980 1990 2000 2010 2020
Year
Figure 18.4. Trends in certified noise of commercial aircraft at year of introduction.
were already compliant with the more stringent regulations. As a result, one may
argue that even more restrictive limits will be introduced (at least another 3 dB cumu-
lative cut) to drive noise emissions down. Further data and comparison are shown by
Waitz et al. 9 .
A Case for a Noise Tax. In the past there has been considerable debate on
forcing compliance with noise regulations via taxation. Non-compliance with noise
regulation can be structural or accidental. Structural causes are the result of more
stringent regulations that make existing aircraft obsolete, as shown in Figure 18.4
(technology obsolescence). Accidental events are those that occur infrequently due
to bad manoeuvre, late departure or other reasons.
Consider the case of the technology obsolescence. One of the options pro-
posed includes a tax per landing (or movement), which can be offset by retrofitting
newer technologies. Assume that there exists a noise-reduction technology capable
of bringing down the noise metrics below a threshold. Assume that this technology
can be retrofitted on the aircraft at a cost C. This cost is an acquisition item that can
be described with Equation 15.46. The cost per movement would be
1 1
C= [P − F(1 + i)n − Rn ] , (18.4)
n Mn
where Mn is the number of movements in the n−th year of operation; Rn is the
residual value in the n−th year. To simplify, assume that no financing is required
(F = 0). The retrofitting cost can be offset if the tax is such that
1 1
tax > (P − Rn ) . (18.5)
n Mn
For example, if the price is P = 1 (in whatever currency), Mn = 2,000 and the
expected life-time of the technology is 20 years, then the noise tax that would offset
Climb Thrust
3,000 ft
Climb Thrust
V2 + 10 / 20 kt
800 ft
Take off Thrust

V2 + 10 / 20 kt
Retract flaps
Brake-Release Take-off
(a) NADP 1
Climb Thrust
3,000 ft C
Climb Thrust
accelerate
to flaps-up
800 ft B
Take off Thrust

V2 + 10 / 20 kt
A
Flaps retraction
Brake-Release (start)
Take-off
(b) NADP 2
Figure 18.5. ICAO noise-abatement procedures at departure.
the new technology is
tax > 0.25 · 10−4 [1 − Rn ] . (18.6)
18.2 Noise-Abatement Procedures

For both departure (take-off) and arrival (approach and landing), there are some
recommended flight procedures aimed at minimising noise. We limit our presenta-
tion to a few representative cases. For the departure case, ICAO recommends two
procedures for noise abatement, one called close-in (or NADP 1) and the other
called distant (or NADP 2). These trajectories are shown in Figure 18.5.
The noise-abatement departure procedure NADP1 (see Figure 18.5) requires a
power reduction at or above a prescribed minimum altitude above the airfield (in
this case, 800 feet). The retraction of the flaps is delayed until a prescribed maximum
altitude is reached (in this case, 3,000 feet). Above this altitude, accelerate to normal
en-route climb speed. The noise-abatement procedure is started at 800 feet. On

reaching this altitude, the engine thrust is adjusted whilst maintaining a climb speed
V2 + 10 kt < V < V2 + 20 kt, with flaps and slats in the take-off configuration. The
landing gear is retracted.
The noise-abatement departure procedure NADP2 (Figure 18.5b) requires a
flap/slat retraction on reaching a minimum altitude (in this case, 800 feet), whilst
maintaining a minimum climb rate. As in the previous case, the noise-abatement
procedure is started at 800 feet, with a climb speed V2 + 10 kt < V < V2 + 20 kt.
Starting at this altitude, the aircraft must accelerate to the speed for flaps/slats
retraction; the attitude is reduced; thrust is reduced as soon as the flaps are retracted.
In this phase, a positive climb rate must be maintained with a speed V2 + 10 kt
< V < V2 + 20 kt. Upon reaching 3,000 feet, the aircraft accelerates to normal en-
route climb speed (as specified by the FCOM).
Similar procedures exist for approach and landing trajectories. For a commer-
cial airliner, the standard procedure is a 3-degree descent trajectory in the terminal
areas, although in some cases the aircraft is forced to perform complicated three-
dimensional manoeuvres to go around topography and ATM constraints. The engine
power setting is minimal, although there is scope for tuning on the throttle. Instead,
special care must be placed toward the operation of the high-lift devices and the
landing gear, which are fully deployed. Two things can be done: 1) reduce the air
speed, and 2) increase the distance between the aircraft and the noise receiver on
the ground. When possible, other procedures can be attempted, such as best aero-
dynamic configuration (optimal flap setting), continuous descent and steep descent.
The latter procedures are neither standard nor possible with some aircraft 10 . The
term steep approach is used to denote any trajectory steeper than 3 degrees, although
in practice it is not possible to exceed 4 degrees. A noise reduction of up to 2 dB has
been reported.
A continuous descent requires no level flight below 7,000 feet. This procedure is
applied at some airports (for example, London City) with some aircraft (BAE 146).
A reduction of up to 5 EPNL(dB) is claimed with this procedure.
18.2.1 Cumulative Noise Index

One of the issues with aircraft noise is its persistence due to continuous movements,
that is, take-off and landings at short time intervals during the day and night. Thus,
there is a cumulative effect on annoyance, which is described by further metrics. If
there are n movements in the reference time and each movement causes a noise level
LAE , then the equivalent continuous sound level LEQ is given by equations such as:

1 n
LEQ[24h] = 10 log 10 LAE (i)/10
(18.7)
24 × 3, 600
i=1

1 n
LEQ[day] = 10 log 10 LAE (i)/10 (18.8)
(23 − 7) × 3, 600
i=1
for the full day and day-time only, respectively. Additional metrics are available
from the technical standards 7 . In any case, these cumulative metrics are used to
Figure 18.6. Flowchart of aircraft-noise model.
determine noise maps around airports, typically with noise levels between 57 and 72
LEQ(dBA).
18.2.2 Noise-Program Flowchart

The analysis of about a dozen independent components is required to establish a
rational basis for aircraft noise. A flowchart of the model is shown in Figure 18.6.
The sources are split in two categories: 1) propulsive noise, arising from the power
plants; and 2) non-propulsive noise (or airframe noise).
In principle, we should add also the noise generated by the APU, which can
be a large gas-turbine engine on its own. However, this is generally not done. One
justification for this shortcoming is that the mechanical details of the APU are
undisclosed. However, this contribution is of considerable importance when the
aircraft is on the ground with the APU running. To understand this effect, we can
perform a sensitivity analysis on the same vein as described by Equation 16.13.
Another noise component not included in the analysis is the tyre noise, which occurs
during the ground roll: this is of no relevance for certification purposes. Viewed in
a larger context, this example indicates that the detailed knowledge of one noise
component is of limited use if other components are ignored.
Once the noise sources are identified and modelled (following the examples in
Chapter 16), the acoustic perturbations have to be tracked in the travel from the
source to the receiver. For this purpose there is a propagation module that takes
Figure 18.7. Distances from noise sources to receiver (selected items).
into account the effects of the atmosphere and the presence of boundaries (the
ground or other obstacles). The combination of reflection, scattering, propagation
and absorption is sent to another module called “Signal Analysis”. The function of
this module is to elaborate the SPL data and provide a suitable quantitative measure
of the aircraft noise.
The aircraft is usually tracked over a long trajectory. During landing and take-
off the aircraft changes configuration. For example, in the terminal-area manoeuvre,
the flaps are extended at about 750 m (∼2,500 feet) above the airfield; the landing
gears are deployed at about 500 m (∼1,600 feet) above the airfield; the flaps are at
landing setting at about 450 m (∼1,500 feet); and so on.
The noise program operates on the basis of a known flight trajectory, previously
calculated with the flight-mechanics program. These trajectories can be defined as
flight data (from radar tracking or flight recorder data systems), information on the
state of the aircraft and the engine, and a detailed knowledge of the atmospheric
conditions around the aircraft and the airfield.
The distance between the receiver and the noise source depends on the source
itself, as indicated in Figure 18.7, although when the aircraft is far enough, the noise
model is unable to discriminate between points A, B, C. . . . It is interesting to note
that the effect of such components as the landing gear is affected by their relative
distance to the receiver. For a Boeing B777-300 on approach, a signal from the main
landing gear reaches the receiver below the flight track about 0.09 second later than
the nose landing gear. In any case, the exact position of the noise source can be
determined from the aircraft geometry model (see Chapter 2). If r is the vector
position of the CG with respect to the receiver at the airfield, and r a is the position of
the noise source A with respect to the CG, then the correct source-receiver distance
r̃ is
r̃ = r + r a . (18.9)
A number of aircraft-noise-simulation methods have been developed over the

years, most notably NASA’s code ANOPP 11 – 14 and its successor ANOPP2 15 ; the
FOOTPR Footprint/Radius Code, developed by NASA Glenn Research Center

since 1981 16 ; INM (Integrated Noise Model), developed by the U.S. Federal Aviation
Administration since 1978 17 ; ANCON (Aircraft Noise Contour Model), developed
by the UK Civil Aviation Authority since 1992 18 ; and PANAM, developed at DLR 19 .
Some of these methods (INM, ANCON) are based on recorded databases and offer
very limited simulation capabilities.
In broad lines, there are models that are developed specifically for airport oper-
ations and planning (for example, INM and ANCON) and models for aircraft design
(ANOPP and its evolutions). The model we present here is designed to address
system-design issues on existing aircraft.
The purpose of the noise program is to predict aircraft noise with a reasonable
accuracy, preferably ±1 EPNL(dB), in a reasonable amount of time, which is the
same time as the duration of the flight trajectory or less. Thus, if the aircraft is
tracked for 2 minutes, the calculation time must be ∼2 minutes (or less) with the
best available hardware. A noise map will have enough information if there are at
least 200 receivers on the ground over at least 10 km2 (10 km × 1km). To calculate
such a map, we would need ∼400 minutes, or 6.7 CPU hours. The calculation of a
noise map due to 100 aircraft movements (50 take-off and 50 landings) will require
weeks of calculation, and it would clearly be unmanageable unless the numerical
methods are improved. If atmospheric winds are included, the computational times
will increase by at least a factor 10. Thus, it is clear that as the complexity of the
problem increases, accuracy and details must be sacrificed in order to obtain a
solution. Therefore, the value of integrated methods such as INM becomes evident
when complete noise maps around airports have to be established.
18.3 Flight-Mechanics Integration

Integration between the flight-mechanics and the noise sources array is done through
an operator that provides the vector of parameters necessary to define the state of
the aircraft:
⎛ ⎞
⎜ ⎟
S = f ⎝t, Lat, Lon, zg , θ, φ, ψ, IAS, V, Vg , N1 , ṁ f , LG, SF, W, T , H, Vw , w ⎠
% &' ( % &' ( % &' ( % &' ( % &' (
position mechanics engine conf ig atmospher e
(18.10)
Equation 18.10 defines the state of the aircraft state by a set of parameters in the
categories of position, flight mechanics, engine state, configuration and atmospheric
conditions. This is a standard format that can be used in other applications such as
flight simulation.
The parameters Lat and Lon are the GPS longitude and latitude, respectively;
zg is the geometrical altitude; θ , φ, and ψ are pitch, bank and yaw angles, respectively;
IAS is the indicated air speed; Vg is the ground speed; N1 is the engine’s rotational
speed; ṁ f is the fuel flow; V is the true air speed; LG denotes the state of the landing
gear (0 = retracted; 1 = deployed); SF denotes the state of the high-lift systems
(see, for example, Table 11.1); T is the outside air temperature, H is the relative air
humidity, Vw is the magnitude of the wind speed and w is the wind direction.
18.3 Flight-Mechanics Integration 565
North
ξ
y
[GPS coords]
z x Runway East
Approach
Microphone
Ground
Track
Departure
Figure 18.8. Typical departure trajectory, with a ground track and a local reference system.
Figure 18.8 shows a typical situation at departure. In our convention, there is a

reference system on the ground, whose longitudinal axis is aligned with the runway.
If the trajectory is taken from FDR data, the coordinates are GPS. A transformation
is carried out, so that all of the GPS coordinates are transferred to the conventional
reference system on the ground. The transformation is done with the Vincenty
inverse distance method 20 . This method only gives the distance ro between two
points on the ground. The distance is corrected to take into account the altitude of
the aircraft, according to
r ro2 + z2g . (18.11)
18.3.1 Noise Data Handling

For a specified noise trajectory, such as Equation 18.10, referred to a single receiver
on the ground, there will be a number of sampling steps x1 , x2 , · · · . The aircraft’s noise
is produced by a number of airframe components a1 , a2 , · · · . For each component,
there will be a spectrum of frequencies f1 , f2 , · · · , up to a cut-off frequency. The
airframe noise is characterised by a noise level La . A similar matrix exists for the
propulsive noise components. Thus, we have the matrices
La (xi , a j , fk), Le (xi , a j , fk).
The way these data are handled is key to the determination of the noise level at
the receiver position. It is also important to remember that each component is
contributed by more than one unit (for example: engines, flaps, landing gear).
Let us assume that the SPL has already been corrected for Doppler effects, dir-
ectivity and spectral effects because these corrections are generally already part
of the noise models, as discussed in Chapter 16. The first data handling is the
correction for sound-pressure attenuation (atmospheric absorption and other dif-

fusive effects):
La (xi , a j , fk) → La (xi , a j , fk) − A(xi , fk, H, dT )
(18.12)
Le (xi , a j , fk) → Le (xi , a j , fk) − A(xi , fk, H, dT ).
If the noise source is very far, the attenuation can be higher than noise level at
the source location; this would lead to a negative noise level, which is clearly not
possible. In practice, it means that the noise has been entirely dissipated through
absorption, diffusion and other processes. A negative noise level is reset to zero.
The next operation is to calculate the noise level for each position xi of the
aircraft. This is done through Equation 16.12. We need to sum the SPL for the
airframe and the propulsion system for each position of the aircraft and for each
spectral frequency; then we sum the SPL of the airframe with that of the propulsion.
The resulting noise level at position xi and frequency fk is

L(xi , fk) = 10 log10 10 La (xi , fk) + 10 Le (xi , fk) .
From this equation, we calculate the OASPL, the PNL and other metrics, following
the standard methods of acoustics.
FLIGHT TIME AT RECEIVER. Another important correction arises from the distance
source-to-receiver, which can be considerably higher than r/a. For example, in iso-
thermal atmosphere, a signal travels at a constant speed and the distance covered
after a time t is r = at; in other words it will take x/a seconds to hear the sound
emitted at the source; thus, there is a time delay equal to r/a. The problem is more
complicated in a non-isothermal atmosphere. In this case, the acoustic signal emitted
at time t by the source is heard at the receiver at a retarded time tr :

dr
tr = t + . (18.13)
a
Assuming that there is no boundary effect, the direct acoustic wave from the source
reaches the receiver at time
hs
dz r
tr = t + , (18.14)
hr a tan ϕ a
where ϕ is the angle between the acoustic wave and the ground and a is the
average speed of sound. There is no guarantee that the receiver time tr is a non-
decreasing function. If the atmospheric temperature varies with altitude according to
T = To − λh, then there is a closed-form solution of Equation 18.14. The time delay
is
2 ) *
t = √ (To − λhr )1/2 − (To − λhs )1/2 . (18.15)
λ tan ϕ γ R
18.4 Noise Sensitivity Analysis

A noise sensitivity analysis is important for understanding which noise sources are
dominant and where modelling and simulation efforts should be concentrated. As
18.4 Noise Sensitivity Analysis 567
Table 18.2. Noise sensitivity matrix for a Boeing 777-300 for ± 2 dB on take-off and landing
trajectories (simulated data); APUJ = APU jet; APUC = APU compressor; the receivers are at
standard ICAO/FAR position (§ 18.1)
Landing Take-Off
Rank Item EPNLc +dB −dB Item EPNLc +dB −dB
1 MLG 97.37 1.508 −1.354 FAN 99.55 1.685 −1.590

2 SLAT 88.80 0.291 −0.221 JET 88.18 0.404 −0.325
3 FAN 84.25 0.072 −0.087 LPT 83.30 0.056 −0.034
4 NLG 83.38 0.020 −0.053 COMB 78.42 0.030 −0.012
5 COMB 82.47 0.019 −0.050 Wing 63.52 0.005 −0.003
6 FLAP 77.50 −0.005 −0.039 MLG 63.30 0.000 0.000
7 HPC 77.18 0.086 −0.123 HSTAB 62.81 0.002 −0.001
8 Wing 74.29 −0.037 −0.015 LPC 61.16 0.001
9 HSTAB 74.24 −0.047 −0.009 HPC 58.46 0.001
10 LPC 64.65 −0.060 −0.001 NLG 48.17
11 JET 46.01 −0.062 SLAT 47.09
12 VSTAB 44.82 −0.062 APUC 19.53
13 APUC 38.37 −0.062 APUJ
14 LPT −0.062 VSTAB
15 APUJ −0.062 HPT
16 HPT −0.062 FLAP
discussed next, it is futile to embark into precise simulation of noise sources that have
little impact; such is the case of a high-lift system or a landing gear in a standard take-
off trajectory. On approach and landing these specific noise sources are dominant.
A noise sensitivity analysis is shown in the data of Table 18.2. EPNLc is the
perceived noise level of the noise source “c” (as named in the table). The column
denoted by +dB shows the change in total EPNL due to an increase of 2 dB in
the spectrum of that component at all points in the trajectory. In other words, at
each position in the aircraft trajectory, the SPL of the component “c” is increased
by 2 dB across the spectrum, from 20Hz to 10 kHz. In Table 18.2, on the take-off
column, an increase by 2 dB in the dominant component (FAN, rank 1) causes an
increase in total EPNL by about 1.69 dB. Likewise, a decrease by 2 dB in the same
component causes a decrease in aircraft EPNL by ∼1.59 dB. As we move down the
table, to follow noise sources that are less and less dominant, the contribution of
any change to the aircraft SPL becomes negligible. In the empty boxes, this change
is effectively zero. At landing the contribution of the main landing gear (MLG) is
dominant. This result explains why landing-gear noise reduction is so critical in this
phase of flight. An increase of 2 dB in this component causes an ∼1.5 dB increase in
the overall noise level. However, a decrease by a similar amount causes a decrease
by ∼1.35 dB. In both cases, the role of the APU in generating noise is negligible. We
conclude that for a take-off trajectory we should focus on the accurate prediction of
the jet noise and the fan noise. Landing is more complicated by the fact that many
source components are contributing, coupled with the fact that the aircraft is closer
to the receiver. The role of each source changes if we change the position of the
receiver.

3 4 5
Main LGear
90 Fan
HPT
LPT
JET
Total
80
OASPL, dBA
70
60
50
5 6 7 8 9
Flight distance, km
Figure 18.9. Calculated noise trajectory of the A320-200 at landing; ICAO/FAR reference.
18.5 Case Study: Noise Trajectories of Jet Aircraft

We consider the case of a model Airbus A320-200 with CFM56-5C4 turbofan
engines. The trajectories are generated by the flight-mechanics model. The inter-
face is given by the vector state in Equation 18.10. We consider the cases of ICAO
reference points for take-off and approach noise. The reference conditions are sea-
level airfield, standard day, no wind and a relative humidity level of 70%.
LANDING. Figure 18.9 shows the calculated noise trajectory at approach/landing.

The graph shows the time-dependent overall noise level (in dB) at the receiver.
Selected contributions are named in the graph, such as the main landing gear, the
fan, the HPT, the LPT and jet noise. The largest single contribution is the fan, with
the landing-gear noise being relatively high as well. The symbols on the OASPL
indicate the time sampling rate of this trajectory.
A more detailed analysis of the contribution is done with the component bar
chart, which is shown in Figure 18.10. Only the largest contributions are named. The
result shows that landing-noise prediction must rely on accurate methods for several
components. The bar charts indicate the change in overall EPNL due to a change by
±2 dB in each of the components. The sensitivity analysis highlights the importance
of a correct prediction of the fan noise.
TAKE-OFF. Figure 18.11 shows the calculated noise trajectory at take-off.

Graph 18.11a tracks the first three minutes of the flight from the brake-release.
The shaded area at the bottom-left corner represents the first 75 seconds, which
are shown in graph 18.11b. Only selected noise sources are indicated in the caption.
Specifically, graph 18.11a shows the four largest contributions, which are jet, fan,
18.5 Case Study: Noise Trajectories of Jet Aircraft 569
100
Fan
90
Slat MLG
EPNL(dB)
80
NLG
Comb.
LPC
Flaps
70
60
Component
Figure 18.10. Calculated bar chart contribution of the noise trajectory shown in Figure 18.9.
combustor and LPT. The airframe contributions are smaller, for reasons evident in
graph 18.11b: at the flight time indicated by A, there is a small thrust reduction, from
take-off level to initial climb. At the flight time indicated by B, the landing gears are
retracted; thus, the MLG and NLG contributions are zero from this point onwards.
At point C, the slat is retracted, and its contribution is discarded. At the flight
time D, there is a thrust cut-back, which contributes to a large reduction in fan and
jet noise.
The noise trajectories discussed herein refer to the case of ICAO/FAR certi-
fication points. These receiver points must be verified first to secure noise certifica-
tion by the relevant authorities. However, the community is interested in the noise
levels at points well outside the airfield, where the effects can be just as annoying
70
B
OASPL OASPL D
90 JET SLAT
FAN NLG
COMB MLG
LPT
80
60
A
OASPL, dB
OASPL, dB
70
C
60
50
50
40 40
0 60 120 180 0 15 30 45 60 75
(a) (b)
Figure 18.11. Take-off noise trajectory of an Airbus A320-200 aircraft model (calculated).
100
OASPL Fan
90 MLG (1)
Slat (2)
NLG (3) 90
Fan (4)
Component EPNL, dB
Flap (5) Slat MLG
80 Jet (6)
80
SPL, dB
NLG
Comb.
LPC
Flap Wing
70 HPC
70
60 2
60
3
4
5
6
50 50
60 70 80 90 100 0 2 4 6 8
Flight time, s Component
(a) (b)
Figure 18.12. Landing-noise trajectory of an Airbus A320-200 aircraft model; reference point
1,000 m upstream of the ICAO/FAR landing; standard day.
as at the runway threshold. We now consider the case of a receiver point placed
1,000 metres upstream of the ICAO/FAR landing point and right below the flight
path of the aircraft. The analysis of the noise signature of the aircraft is shown in
Figure 18.12.
This analysis shows that there is a relatively long exposure to noise, with sev-
eral components contributing on a comparable level. Graph 18.12a shows selected
contributions (numbered 1 to 6 and named in the legend). The maximum EPNL
occurs at about 80 seconds in the aircraft flight time. The fan is the largest source
of concern, but the landing gear and the slat are also of considerable importance.
Graph 18.12b shows the bar chart of the component contribution to the EPNL. The
error bars indicate the noise sensitivity to a ±2 dB change in the noise level of each
component, as previously discussed. Aside from the fan’s contribution (easily the
largest noise source), the airframe components are particularly high. Noise reduction
in this case requires a balanced approach at the airframe and engine levels.
18.6 Case Study: Noise Trajectories of Propeller Aircraft

We consider the case of a model ATR72-500 with PW-127M turboprop engines and
Hamilton-Sundstrand F568-1 propellers. The power-plant is discussed in § 5.4.1. The
propeller is discussed in § 6.2.4. The case refers to a sea-level airfield, on a standard
day, without wind and a relative humidity level of 70%.
LANDING. Figure 18.13 shows the calculated noise trajectory at landing. There are
several contributions, but the propellers represent the dominant source of noise.
This result indicates that the turboprop operates differently from a turbofan
aircraft, and noise-reduction strategies must address propeller issues. However, for
completeness, we show the sensitivity analysis for this case, Figure 18.14, which again
refers to the change in EPNL(dB) corresponding to an error ±2dB in each separate
noise-source component. The sensitivity is shown by the error bars. The amplitude
18.6 Case Study: Noise Trajectories of Propeller Aircraft 571
90
OASPL
Propellers
Wing
FLAP
80 NLG
SPL, dB MLG
70
60
50
80 90 100 110
Flight time, s
Figure 18.13. Calculated noise trajectory of the ATR72-500 at landing; ICAO/FAR reference;
selected noise components shown.
of the error bars denotes the change in total EPNL(dB). In other words, an error of
±2 dB on the propeller noise causes an error on the EPNL that has an amplitude
equal to +1.551 − (−1.428) = 2.979 dB. The contribution of the other components
to the overall error decreases with the component’s EPNL.
TAKE-OFF. Figure 18.15 shows a similar case for the take-off trajectory. Again, the
propeller noise is the dominant contribution – by far.
A summary of numerical results is in Table 18.3. The data include maximum
loudness [dBA], the distance to the receiver corresponding to maximum noise (r )
75
Props
MLG
Flaps
70
Comb.
EPNL(dB)
NLG
Wing
HPC
65
60
Component
Figure 18.14. Calculated landing-noise sensitivity, case of Figure 18.13.
Table 18.3. ATR72-500 noise trajectories; All noise levels are in dB (calculated)
m r h θ ICAO4 ICAO4
[ton] Loud [m] [m] [deg] EPNL SEL LAeqT Limit Margin
Landing 22.67 85.87 124 121 76.8 82.18 77.20 57.10 96.2 −14.0
Take-off 20.61 75.26 730 715 78.0 77.23 71.35 48.23 85.0 −7.8
and the corresponding flight altitude (h), the polar-emission angle (θ), the effective
perceived noise level, the sound exposure level, the equivalent continuous noise
level, the ICAO Chapter 4 noise limits and the corresponding margin. A negative
value indicates that the noise performance complies with the ICAO Chapter 4
regulations.
18.7 Further Parametric Analysis of Noise Performance

There is a variety of problems that can be analysed with the aircraft noise model.
In this section we present the effect of a steep approach and the atmospheric winds
on approach and take-off. All of the calculations have been done with the FLIGHT
code.
STEEP DESCENT. The flight mechanics of steep descent was discussed in § 11.4. We
now consider the consequences of a steep glide slope on the noise metrics. The
example deals again with the Airbus A320-200 with CFM engines. We have con-
sidered a conventional trajectory with γ = 3.01 degrees and a steep trajectory with
γ = 3.775 degrees. The flight takes place in standard atmosphere with negligible
wind speeds (the wind is included in the trajectory simulation but not in the noise
Distance, n-miles
2.5 3 3.5 4
OASPL
80 Wing
Prop
70
SPL, dB
60
50
REC
40
4 5 6 7 8
Distance from brake-release, km
Figure 18.15. Calculated noise trajectory of the ATR72-500 at landing; ICAO/FAR reference.
18.7 Further Parametric Analysis of Noise Performance 573
Table 18.4. Calculated noise metrics (in dB) over a

conventional and steep landing trajectory at a ICAO/FAR
landing point and point 1,000 m upstream (3,300 m from
touch-down)
Metrics ICAO/FAR FAR − 1,000 m
Normal Steep Normal Steep
EPNL 95.2 92.4 91.9 89.3

SEL 91.3 89.5 89.3 87.5
LAeqT 73.0 71.9 70.7 69.5
propagation). The resulting noise level is in shown Figure 18.16. Note that the flight
times of these trajectories are not synchronised; therefore, the shift in the peak noise
level does not correspond to actual conditions. However, the peak is reduced by ∼2
dB, and the consequences on the integral noise metrics are of a similar magnitude. A
summary of calculated performance is in Table 18.4, which also contains the results
for a microphone placed 1,000 m upstream of the ICAO/FAR reference point. We
conclude that some noise reduction can be achieved by flying a steep trajectory.
WIND EFFECTS. Two sub-cases are considered: 1) ICAO/FAR microphone at landing

with head wind or tail wind, and 2) sideline microphone at landing with a 90-degree
wind.
Figure 18.17a shows the calculated noise trajectories with a tail-wind, a head-
wind and a no-wind condition. The wind speed is Vw = ±4 m/s, measured 1.2 m
from the ground. The difference in EPNL is contained within 0.7 dB, which is
due to a combination of increased OASPL peak and duration. However, there is
90
γ = 3.01 degrees
γ = 3.75 degrees
80
OASPL, dB
70
60
50
20 30 40 50 60 70 80
Flight time, s
Figure 18.16. Simulated landing-noise trajectories of an Airbus A320-200, conventional and
steep approach. The noise metrics at the ICAO/FAR point are given in Table 18.4.
Headwind
U Wind
90 90
Tailwind Flight
runway
D
80 80
OASPL, dB
OASPL, dB
70 70
60 Headwind 60 Downwind
50 50
Tailwind Upwind
40 40
40 50 60 70 80 90 40 50 60 70 80 90
(a) (b)
Figure 18.17. Wind effects on perceived noise at landing; aircraft model is the A320-200-CFM
(calculated).
a considerable wind effect over long distances, which cuts the noise level at the
receiver. One point worthy of note is that the notion of head and tail wind depends
on the position of the microphone with respect to the aircraft (see also Figure 17.8).
Figure 18.17b shows the calculated noise trajectories with a fixed wind at 90-
degrees, as indicated in the top-right box. The central point is the ICAO/FAR
reference point; the side-line points are placed 450 metres off the runway centreline.
With a fixed wind direction, one sideline microphone is upwind (U) and the other
is downwind (D). The calculations show an upward shift in the OASPL, which
contributes to an increase in perceived noise by about 1.5 EPNL. The downwind
microphone measures an increased noise level over the full trajectory. The upwind
microphone measures a noise level that drops rapidly once the aircraft has passed
through its minimum distance.
The analysis of the wind effects at take-off is shown in Figure 18.18, which shows
the noise level at the ICAO/FAR reference point for three cases: no wind, tail wind,
and head wind. The wind speed is Vw = ±4 m/s, measured 1.2 m from the ground.
The calculations show that there is no difference in OASPL peak between head and
tail wind. However, there is a difference between wind and absence of it (top right
box), which leads to an increase of ∼ 0.7 EPNL (the EPNL increases from 90.9 dB
to 91.6 dB). Strong wind effects take place over a long distance, as shown to the left
of the main graph.
We conclude that the wind effects are important at all conditions. When the
aircraft is closest to the receiver, the peak noise level can be sensibly affected so as
to cause an increase in perceived noise by 0.7 to 1.5 EPNL for the cases calculated.
Over long distances, the noise level can be either cut or increased by several dB.
18.8 Verification of the Aircraft-Noise Model

The ultimate test of an aircraft-noise model is based on real flight data that are
synchronised with noise measurements on the ground. Data have been gathered
85
headwind
90 80
tailwind
80
75
80 85 90 95 100
OASPL, dB
70
Tailwind
60
50
Headwind
40
50 100 150
Flight time, s
Figure 18.18. Wind effects on perceived noise at take-off; aircraft model is the A320-200-CFM
(calculated).
from a campaign of measurements 21;22;23 carried out on the Airbus A319-100 flown
by Lufthansa* .
Figure 18.19 shows two views of selected flight trajectories and the relative
position of the microphones on the ground. With reference to graph 18.19a, the
vertical scale has been expanded and shows lateral movements of the order of one
wing span. The change in direction at the start or end of the trajectory is the effect
of the aircraft moving along the taxi-way.
Two sets of microphones are available: one set is at 1.2 metres from the ground;
the other set is on the ground and is used to remove the effects of ground reflection
and diffraction. Microphones 1–12 are on the take-off/departure side; microphones
13–25 are on the approach/landing side. Only selected microphones are shown in
Figure 18.19. Measurements were taken at the Schwerin-Parchim airfield (IATA
code: SZW; ICAO code: EDOP). The ground track shows an apparently rough
flight trajectory, with lateral oscillations of the order of one wing span. None of
the microphones is directly below the airplane except microphone 13. Graph 18.19b
shows the position of the landing gear (LG) deployment and retraction.
Selected results obtained with the FLIGHT program are shown in Figure 18.20
and Figure 18.21. In both cases we have the microphones as named in Figure 18.19.
The flight data that have been used are a sub-set of the noise-state vector given by
Equation 18.10. The detailed atmospheric conditions were not available. Standard
values were assumed.
Overall, the results show a good comparison with the flight data. The peaks are
generally well predicted, although there are cases where the peak OASPL are missed.
* Selected raw data have been kindly provided by the Institute of Aerodynamics and Flow Technology
of DLR; Braunschweig, Germany, Nov. 2011.
Approach / Landing Departure

400
200 16
1
13
Y, m
24 18 6
0 3
15 10
21
-200
-400 14
-30,000 -20,000 -10,000 0 10,000 20,000

X, m
(a) Horizontal plane
Approach / Landing Departure

2,000
16
24 18 13 6 3 1
21 15 10
1,500 14
Z, m
1,000
500
LG LG
0
-30,000 -20,000 -10,000 0 10,000 20,000
X, m
(b) Vertical plane
Figure 18.19. Landing and take-off flight trajectories with microphone positions on the
ground.
There are also cases where there is a large difference in OASPL upstream of the
peak. However, note that the flight data are affected by high-frequency fluctuations,
possibly due to localised gusts and turbulence. Although small wind speeds at the
microphone location do not make a large difference (see also Figure 18.17), sudden
gusts at altitude affect the propagation characteristics and result in sudden changes
in acoustic pressure; these changes can be of the order of ±10 dB.
QUESTIONS ON NOISE-PREDICTION ACCURACY. The complexities of the noise signal

point to an important question: What can be achieved with a low-order aircraft noise
model? The answer lies in the type of metrics we wish to examine. Because the key
metrics for certification, trajectory optimisation and land planning are of the integral
type, then we could restrict the accuracy requirements to those characteristics that
allow an accurate prediction of these metrics. With reference to the EPNL, we need
to be able to predict at least the following quantities: 1) peak noise level; 2) flight
time corresponding to the peak level; and 3) noise level over the 10 dB-down time.
Additionally, we would need some accuracy in the spectral content of the noise,
100
Flight prediction
Experimental data
90
13
15
OASPL, dBA
80
24
21 18
70
Airframe
60 noise
50
30100 30200 30300 30400 30500
Time, s
Figure 18.20. Comparison between prediction and experimental data for an approach/landing
trajectory. Microphone numbers are the same as in Figure 18.19; microphones on the ground.
which is more difficult to establish because its nature depends on the combination
of the various contributions.
The sensitivity analysis has highlighted the components that need addressing for
the noise reduction. Therefore, the parameters listed here would certainly apply to
those components.
100
Flight prediction
10 Experimental data
90
OASPL, dBA
80
Airframe 6
noise
70 3
1
60
50
50700 50800 50900
Time, s
Figure 18.21. Comparison between prediction and experimental data for a take-off/departure
trajectory. Microphone numbers are the same as in Figure 18.19; microphones on the
ground.
2,000
1,000 90
80
70
80 0
y, m
0 10
70
110
80
100
-1,000 90
-2,000
-2,000 0 2,000 4,000 6,000
x, m
Figure 18.22. Calculated noise footprint at take-off for the Airbus A320-200 with CFM tur-
bofan engines. Standard day, no wind, H = 70%; mission: 2,000 km (1,080 n-miles), 100%
pax. Indicated levels are EPNL(dB). Aircraft travelling from left to right.
18.9 Noise Footprint

Once the relevant noise metrics have been defined, several zones around the airfield
are defined for the purpose of land use and noise management, including Zone A,
which is in the proximity of the runways and subject to the most disruptive noise;
and Zone B, subject to more moderate noise exposure. Other zones (such as C and
D) are sometimes defined. These areas can be several square kilometres around
the airport, although their extent has shrunk over the years. The 90-dB-level noise
footprint of the first jet was ∼30 km2 , whilst a modern regional jet has a 90-dB-level
footprint of ∼5 km2 .
A straightforward extrapolation of the noise trajectory analysis is the noise
footprint on the ground (for a single aircraft movement) and the stacking patterns
(for several aircraft movements in and out of the same airfield). First, a noise footprint
is determined as the EPNL or SEL values over a grid centred around the airfield.
Stacking patterns are obtained from a sum of single-event footprints. Analyses of this
type can be done with the true GPS coordinates of the aircraft and the receiver on
the ground. The transformation from GPS coordinates to actual source-to-receiver
distances is done through an inverse solution of the geodesics 20 .
Figure 18.22 shows the calculated noise footprint of the Airbus A320-200 at take-
off. The aircraft travels from left to right. The brake-release point is at (0,0). The
trajectory is straight. The area within the 90-dB contour is 7.12 km2 . The maximum
lateral extension of this contour exceeds 1 km. The sideline noise can be found from
the analysis of this footprint. In fact, consider the line parallel to the centre of the
runway (dash-dot line), placed 450 m laterally. The maximum noise level on this line
is ∼97 EPNL.
The landing-noise footprint for the same case is shown in Figure 18.23. The
aircraft travels from left to right. Notice that the 90-dB contour extends about 3
km upstream of the landing point, but it is very narrow, ∼0.1 km in the lateral
direction. This result is very important because it shows that receivers below the
flight path are subject to high noise levels, whilst receivers away from the flight
path are comparatively unaffected. If atmospheric winds are taken into account,
18.9 Noise Footprint 579
1,000 70
70
80
y, m
90
0
90
80
-1,000
70
-6,000 -4,000 -2,000 0

x, m
Figure 18.23. Calculated noise footprint at landing for the Airbus A320-200 with CFM tur-
bofan engines. Standard day, no wind, H = 70%. Mission: 2,000 km (1,080 n-miles), 100%
pax. Indicated levels are EPNL(dB). Aircraft travelling from left to right.
receivers upwind of the flight path will be affected by lower noise levels and receivers
downstream will suffer higher noise.
The noise footprint of a propeller aircraft such as the ATR72-500 shows con-
siderable differences from the turbofan engines discussed so far. Figure 18.24 shows
the calculated footprint at landing. The equal-noise curves envelop a much smaller
area. To be fair, we would need to consider the weight effects.
REAL-TIME METRICS. The previous examples of noise footprint illustrate the “steady-
state” noise performance of the aircraft because we have used the EPNL. However,
if we want to look at the real-time effect of a passing aircraft (approaching or taking
off), the EPNL cannot be used. We then consider an instantaneous noise level
described by the OASPL. One such example is shown in Figure 18.25. The noise
carpet spreads far and wide, particularly after take-off. Whilst on the ground, the
noise is strongly affected by the ground reflection.
60
500
65
70
70
85 75
y, m
0
75
70
-500
60
65
-3,000 -2,000 -1,000 0

x, m
Figure 18.24. Calculated noise footprint at landing for the ATR72-500 with PW127M turbo-
prop engines; standard day, no wind, H = 70%. Mission: 1,000 km (540 n-miles), 100% pax.
Indicated levels are EPNL(dB). Aircraft travelling from left to right.
2 2
50
50 50
0 0
50
-2 -2
0 5 10 0 5 10
(a) (b)
2 2
50
0 50 0
-2 -2 50
0 5 10 0 5 10
(c) (d)
2 2
0 50 0
50
-2 -2
0 5 10 0 5 10
(e) (f)
2 2 50
50
50
0 0
50
50
-2 50 -2
0 5 10 0 5 10
(g) (h)
2 50 2 50
0 0
50
50 50
-2 -2
0 5 10 0 5 10
(i) (j)
Figure 18.25. Simulated noise carpet of the Airbus A320-200 at take-off. Graphs show iso-
level OASPL[dB] at selected time steps; dimensions are [km]; standard day; no wind; OASPL
levels are at 10 dB intervals.
18.9.1 Noise Maps Refinement

The number of microphones that must be used to achieve an acceptable resolution
of the noise maps is a fairly large number, possibly several hundred. Even from a
computational point of view, the use of hundreds of microphones may require several
hours of calculation. Therefore, we propose to refine the noise map by increasing
the resolution of the grid after the noise footprint is calculated.
75 75
70
70
600 600
75
75
75
75
80 80
400 400
y, m
y, m
80
80
85
75
200 200
80
80
85
85
75
80
85
90
90
0 0
0 2,000 4,000 6,000 8,000 10,000 0 2,000 4,000 6,000 8,000 10,000
X, m X, m
(a) Coarse grid (b) Refined grid
Figure 18.26. Effects of grid refinement on a generic EPNL map.
If the grid is defined by the arrays x = x(1, · · · , nx ) and y = y(1, · · · , n y ),

we double the size of these arrays, e.g. we halve the resolution of the grid.
The transformed grid will be defined by arrays x2 = x2 (1, · · · , 2nx − 1) and y2 =
y2 (1, · · · , 2n y − 1). Once the grid has been refined, we apply a cubic spline inter-
polation in two directions to calculate the function value at the new grid points 24 .
A typical result of this operation is shown in Figure 18.26. The refined map gives a
more accurate estimation of the area affected by noise.
18.10 Noise from Multiple Aircraft Movements

We discuss the calculation of the noise carpet due to the operation of one airplane
type from one airfield (noise stacks). This is a simplified case because, in reality, a
commercial airfield will operate with several aircraft types. We have two noise car-
pets: approach/landing and take-off/climb-out. The reason for two separate carpets
is that the area affected is relatively large.
OVERLAP OF LANDING AND TAKE-OFF GRIDS. The arrays xL(..), yL(..) represent the
landing-carpet grid coordinates; the arrays xT (..), yT (..) represent the take-off/climb-
out noise carpet. These carpets are partially overlapping, as shown in Figure 18.27.
The black dots are landing grid points which fall within the take-off carpet. At these
points a sum between the noise level LL and LT is required. Because the noise LT
is not available at this black grid point, an interpolation is carried out to determine
its value. We use a bilinear interpolation within the top cell (in dashed line). The
white dots denote grid points in the take-off carpet which overlap the landing carpet.
Again, a bilinear interpolation within the relevant grid cell is used to calculate the
noise level LL at this point.
For the overlapping grid points the noise level is

L = 10 log10 10 LT /10 + 10 LL/10 . (18.16)
Landing
Take-off
Figure 18.27. Landing and take-off noise carpets for calculation of noise stacks.
SYNCHRONISATION OF FLIGHT TIME. Before proceeding with this interpolation, we

need to establish the relationship between flight times.
Let us consider first all operations in the same category (approach or take-
off). At a given time, there can be an arbitrary number of aircraft, depending on
separation and carpet size. If the flight trajectories have n time steps, which are
separated from each other by ns time steps, then the correlation between the flight
movement j and its time step i is found from the following relationships:
t(i, j) = n, n = i1 , · · · , i2 , i 1 = 1 + ns ( j − 1), i 2 = i 1 + n − 1. (18.17)
In other words, Equation 18.17 is a matrix with sub-vectors {1, 2, · · · , n} shifted by
ns steps; all other entries in the matrix are zero. The number of airplanes within the
carpet is determined by the relationship between i, ns and n; if ns > n, there will be
one airplane at the time.
We use a Universal Time (UT). The raw trajectory of an approaching flight i L
has a time ti L , with ti L = 0 at the start. Likewise, the trajectory of a take-off flight
i T has a time ti T , with ti T = 0 at the start (brake-release point). There is a specified
separation between landing flights. To synchronise these flights, the time sequences
are transferred into a universal time.
Next, consider the approach/landing carpet as a reference; if ti L is the current
time step, the take-off carpet must be calculated exactly at this time. The sequence
of time steps ti T generally does not coincide with the sequence ti L , but it should be
possible to find an index i T such that ti T < ti L < ti T +1 . If this is the case, we carry out
an interpolation of the SPL in time, between i T and i T + 1.
Figure 18.28 shows the noise signature predicted with the program FLIGHT by
a sequence of Airbus A320-200s landing and taking off at one airfield, along a con-
ventional straight trajectory. The aircraft movements are separated by 75 seconds.
The touch-down point is at x = 0; the aircraft travels from the left to the right
of the page. For the case shown in Figure 18.28 the calculation of the cumulat-
ive metrics (§ 18.2.1) is straightforward because the noise level is the same for all
flights. The symbols L1, L2 denote approach/landing flight numbers 1 and 2, respect-
ively; the symbols T1, T2 denote take-off flight numbers 1 and 2, respectively. In
frame 18.28h there are two approaching flights and in frame 18.28i there are two take-
off flights. The noise pattern on the ground is strongly dependent on the separation
time.
L1 T1 L1 T1
2 2
0 0
50 50
-2 -2
x x
-5 0 5 10 -5 0 5 10
(a) (b)
L1 T1 L1 T1
2 2
50
0 0
50
-2 -2
x x
-5 0 5 10 -5 0 5 10
(c) (d)
L1 T1 L1 T1
2 2
0 0
-2 -2
x x
-5 0 5 10 -5 0 5 10
(e) (f)
L1 T1 L2 L1 T1
2 2
0 0
-2 50 -2
x x
-5 0 5 10 -5 0 5 10
(g) (h)
L2 T2 T1 L2 T2 T1
2 2
0 0
-2 -2
x x
-5 0 5 10 -5 0 5 10
(i) (j)
L2 T2 T1 L2 T2 T1
2 2
0 0
-2 -2
x x
-5 0 5 10 -5 0 5 10
(k) (l)
Figure 18.28. Simulated noise stacks for multiple landings and take-offs of the Airbus A320-200; ts =
75 s. Graphs show iso-level OASPL[dB] at arbitrary time steps; OASPL intervals are 5 dB; carpet
dimensions are [km].
The minimum separation time depends on several factors, including the airport
and the local ATC. For take-off, the guidelines in Table 4.2 must be used. For landing,
the separation can be shorter, due to the fact that trailing aircraft are above leading
aircraft and thus are not affected by any downwash. However, as the separation time
is changed, the noise signature on the ground changes considerably. More complex
algorithms exist to analyse these events in more detail.
18.10.1 Noise Reduction and Its Limitations

Due to the logarithmic scale, the noise reduction achieved in the past few decades is
better than it appears in Figure 18.4. Further noise reduction is still possible. By how
much? Some specialists believe that technological advances will eventually lead to
the development of a quiet aircraft – one that can be seen but not heard 25 . However,
completely innovative configurations are required, such as a blended wing-body
aircraft with aft-mounted engines. This configuration would exploit the scattering
and reflection properties of the aircraft to minimise the engine noise at a receiver on
the ground.
The airframe noise is somewhat different and depends on the scattering of the
boundary layer from lifting surfaces. Problems of minimum theoretical noise levels
have received attention in recent years, after revisiting research done in the 1970s
on bird flight, in particular on the barred owl (Kroeger et al. 26 ). The problem has
been reviewed by Lilley 27 in a paper titled “The Silent Flight of the Owl”.
The barred owl (Strix varia) is the only bird known to fly silently at frequencies
above 2 kHz. In some experiments, the owl was let to fly in a closed chamber from
a 3-metre-high perch to a pile of food on the ground. The flight was observed and
the noise measured. The average speed was estimated as 8 m/s with a CL 1 and
a gross weight 0.6 kg. It was demonstrated that the upper surface of the wing is
stalled, or nearly so, and that the wing air flow is nearly turbulent. The removal of
the trailing-edge fringe and the leading-edge comb increased the noise to the level
produced by other birds. Thus, evolution has created efficient physical mechanisms
that cut off the perceived noise to 2 kHz, by means of natural control of the flow
separation, its instability and turbulence transition.
Summary
In this chapter we made several applications of the noise-performance code, includ-
ing the flight-mechanics integration and the noise-matrix handling. We explained
the basis for aircraft noise certification and various procedures that can be followed
to reduce the noise at the official certification points. Through a sensitivity analysis,
we demonstrated that not all noise sources are equally important. By looking at the
EPNL(dB) integral metrics, we made a detailed analysis of the noise-level contribu-
tions at take-off and landing. The latter case is demonstrated as more complicated
because there are several noise sources that contribute to the overall EPNL at a
similar level. At take-off, the most important noise components are the engine fan
and jet/nozzle. For a propeller aircraft we demonstrated that the dominant acoustic
effect is due to the propeller itself. Wind effects are always important. Over short
distances, they may contribute to a change in EPNL of the order of 1.5 dB. Over
Bibliography 585
long distances, they can affect the sound level by several dB. A comparison with
flight-noise data has shown the complexities of the acoustic signals that must be
simulated.
The analysis of the noise trajectories, especially approach and landing, has
demonstrated that the highest noise level is concentrated below the flight path.
It is safe to assert that in absence of atmospheric winds, at a lateral distance of
∼1.5 km from the flight path, the noise level is relatively low (with respect to other
sources of noise). In the presence of atmospheric winds, it is best to be placed at an
upwind position with respect to the flight path. Having the option between living in a
mansion below/downwind the flight path or in a modest dwelling, the choice should
be the latter – every time.
Bibliography
[1] Franssen EAM, van Wiechen CMAG, Nagelkerke NJD, and Lebret E. Aircraft
noise around a large international airport and its impact on general health and
medication use. Occup. Environ. Med., 61(5):405–413, 2004.
[2] Alexandre A, Barde J-Ph, and Pearce DW. The practical determination of a
charge for noise pollution. J. Transport Econ. & Policy, 14(2):205–220, May
1980.
[3] Gebhardt GT. Acoustical design features of Boeing Model 727. J. Aircraft,
2(4):272–277, 1965.
[4] Crighton DG. Model equations of nonlinear acoustics. Ann. Rev. Fluid Mech.,
11:11–33, 1979.
[5] Smith MTJ. Aircraft Noise. Cambridge University Press, 2004.
[6] Shahadi PA. Military aircraft noise. J. Aircraft, 12(8):653–657, 1975.
[7] ECAC. Report on standard method of computing noise contours around civil
airports. Vol. 1: Applications guide. Technical Report Doc. 29, European
Civil Aviation Conference, Dec. 2005.
[8] ECAC. Report on standard method of computing noise contours around civil
airports. Vol. 2: Technical guide. Technical Report Doc. 29, European Civil
Aviation Conference, Dec. 2005.
[9] Waitz I, Kukachko S, and Lee J. Military aviation and the environment: His-
torical trends and comparison to civil aviation. J. Aircraft, 42(2):329–339, 2005.
[10] Filippone A. Steep-descent manoeuvre of transport aircraft. J. Aircraft,
44(5):1727–1739, Sept. 2007.
[11] Fink MR. Noise component method for airframe noise. J. Aircraft, 16(10):659–
665, 1979.
[12] Fink MR and Schlinke RH. Airframe noise component interaction studies.
J. Aircraft, 17(2):99–105, 1980.
[13] Zorumski WE. Aircraft noise prediction program theoretical manual, Part 1.
Technical Report TM-83199, NASA, Feb. 1982.
[14] Kontos K, Janardan B, and Gliebe P. Improved NASA-ANOPP noise pre-
diction computer code for advanced subsonic propulsion systems – Volume 1.
Technical Report CR-195480, NASA, Aug. 1996.
[15] Lopes LV and Burley CL. Design of the next generation aircraft noise pre-
diction program: ANOPP2. In 17th AIAA/CEAS Aeroacoustics Conference,
AIAA 2011-2854, Portland, June 2011.
[16] Clark BJ. Computer program to predict aircraft noise levels. Technical Report
TP-1913, NASA, 1981.
[17] Boeker ER, Dinges E, He B, Fleming G, Roof CJ, Gerbi PJ, Rapoza AS,
and Hemann J. Integrated noise model (INM) Version 7.0. Technical Report
FAA-AEE-08-01, Federal Aviation Administration, Jan. 2008.
[18] Ollerhead JB, Rhodes DP, Viinikainen MS, Monkman DJ, and Woodley AC.
The UK civil aircraft noise contour model ANCON: Improvements in Version
2. Technical Report R&D 9842, Environmental Research and Consultancy
Dept., Civil Aviation Authority (CAA), June 1999.
[19] Bertsch L, Dobrzynski W, and Guerin S. Tool development for low-noise
aircraft design. J. Aircraft, 47(2):694–699, Mar. 2010.
[20] Vincenty M. Direct and inverse solutions of geodesics on the ellipsoid with
application of nested equations. Technical Report XXIII, No. 176, Survey
Review, 1975.
[21] Pott-Pollenske M, Dobrzynski W, Buchholz H, and Guerin S. Airframe
noise characteristics from flyover measurements and predictions. In 12th
AIAA/CEAS Aeroacoustics Conference, AIAA 2006-2567, Cambridge, May
2006.
[22] Guerin S and Michel U. Prediction of aero-engine noise: comparison with A319
flyover measurements. In Tech. Rep., DLR, DLR IB 92517-04/B3, 2007.
[23] Bertsch L, Guérin A, Looye G, and Pott-Pollenske M. The parametric air-
craft noise analysis module – status overview and recent applications. In 17th
AIAA/CEAS Aeroacoustics Conference, AIAA-2011-2855, Portland, Oregon,
USA, 5–8 June 2011.
[24] Press WH, Teukolsky SA, Vetterling WT, and Flannery BP. Numerical Recipes.
Cambridge University Press, 2nd edition, 1992.
[25] Crichton D, de la Rosa Blanco E, Law T, and Hileman J. Design and operation
for ultra low noise take-off. In 45th AIAA Aerospace Sciences Meeting, AIAA
2007-0456, Reno, NV, Jan. 2007.
[26] Kroeger RA, Gruska HD, and Helvey TC. Low speed aerodynamics for ultra
quiet flight. Technical Report TR-971-75, US Air Force Flight Directorate
Laboratory, 1971.
[27] Lilley GM. A study of the silent flight of the owl. AIAA Paper 1998-2340,
1998.

a = speed of sound
A = wing area
A = atmospheric attenuation of sound
C = cost, Equation 18.4
d = diameter
f = frequency
F = financing
h = flight altitude
i, j = counters for grid arrays (x, y)
i1 , i2 = index for start/finish of noise trajectory in noise stacks, Equation 18.17
L = noise level in dB
LAE = sound exposure level
LEQ = equivalent continuous sound level (prescribed time)
LG = binary value for landing-gear state
m = mass
ṁ f = fuel flow
Bibliography 587
M = Mach number
Mn = number of aircraft movements on n−th year of service
n = number of steps in flight trajectory, Equation 18.17; number of years;
number of aircraft movements in Eq. 18.7 and 18.8
ns = number of separation time steps, Equation 18.17
nx , n y = number of grid points in footprint calculation
N1 = engine’s rpm
p = pressure; acoustic pressure
pa = atmospheric pressure
P = price
r = distance
r = position vector
R = gas constant
Rn = residual value of noise technology at n−th year, Equation 18.5
S = vector operator, Equation 18.10
SF = binary value for flaps/slats state
t = time
tr = time at receiver location
T = atmospheric temperature
V = true air speed
Vg = ground speed
VS = stall speed
Vw = wind speed
V2 = take-off speed
W = aircraft weight
x = distance on the ground
x, y = grid coordinates for footprint calculation
z = vertical coordinate or altitude
zg = ground elevation
Greek Symbols
δ = relative pressure
θ = pitch angle/attitude
λ = temperature-lapse rate
ρ = air density
φ = bank angle
ϕ = angle of incidence of acoustic wave
w = wind direction
ψ = heading angle, Equation 18.10
[.]a = airframe
[.]c = component
[.]e = engine
[.]i = position index; array index

[.] j = component index; array index
[.] L = landing, in § 18.10
[.]k = frequency index
[.]o = reference or sea-level conditions
[.]r = receiver
[.]s = source
[.]T = take-off, in § 18.10
˜
[.] = corrected value
[.] = average value
19 Environmental Performance
Overview
The previous chapters have shown several aspects of environmental performance,
including aircraft noise and carbon emissions. This chapter elaborates further on the
impact of aviation. We discuss the effects of commercial aviation on the formation of
condensation trails (§ 19.1) and possible methods for mitigation, including altitude
flexibility. We discuss briefly the controversial issue of radiative forcing of various
forms of pollution (§ 19.2) and a method for calculating the landing and take-
off emissions (§ 19.3). We show a case study for a transport aircraft to illustrate
key parametric effects on carbon-dioxide emissions (§ 19.4). We then propose an
example of “perfect flight” (§ 19.5), that is, a flight that is not constrained by air
traffic regulations. One of the key aspects that is due to dominate emissions (the
trading scheme) is briefly reviewed in § 19.6.
This chapter does not attempt to enter the debate of environmental perform-
ance and its relationship with the climate change, as many stakeholders and inter-
dependencies are involved. Furthermore, these deeper issues now overlap those
of atmospheric physics, biochemistry and policy-making 1 , which are beyond our
context.
KEY CONCEPTS: Aircraft Contrails, Contrail Factor, Radiative Forcing, Altitude

Flexibility, Carbon Emissions, LTO Emissions, Emissions Trading, Perfect Flight.
19.1 Aircraft Contrails

The blue skies! – What a rare pleasure at our latitudes. One cold afternoon, a clearing
in the clouds brings crispy skies. The air traffic, which is normally invisible above the
clouds, reveals the full scale of its impact, as shown in Figure 19.1. The photo shows
at least 14 distinct vapour trails, from less than 1 minute old to about 30 minutes.
The oldest contrails evolve into cirrus clouds.
The contrail cover is a well-known side effect of commercial flights. Thus, it is
natural to ask: What can we do to prevent them from appearing? Several aspects
must be considered. First is the understanding of the atmospheric conditions that lead
to extensive contrail formation. Second is the understanding of how the gas-turbine
589
Figure 19.1. Contrail cover at 16:00 GMT, 22nd January 2011 (sun-down); geo-coordinates
53◦ 37 N; 2◦ 36 W (approximate), looking SW.
engines operate and also how their exhaust gases contribute to these artificial clouds.
There is the question of whether a change in operational parameters (flight altitude,
speed and routing) can minimise the impact. Finally, to be taken seriously, we
should also ask whether the elimination of the contrails addresses more important
environmental aspects. With regard to the latter concern, there are academic studies
that have highlighted considerable effects, beyond the pollution of the blue skies.
For example, a study by Travis et al. 2;3 showed that during the unprecedented four-
day halt in commercial flights above North America following the 9/11 events, the
absence of contrails caused an increase in daily temperature change by more than
1℃. This result is attributed to persisting contrails, which can reduce the radiative
heat transfer between the ground and the higher atmosphere, thus trapping heat
at lower altitudes at night time and reducing radiative heat from the sun during
day time. Such an experiment could have been repeated over Northern Europe in
Spring 2010, when most of the air traffic was shut down due to unprecedented (again)
ash-cloud fallout from a volcanic eruption in Iceland.
Not all contrails are clearly visible with a naked eye. For a given viewing angle,
the visibility is determined by light scattering in the direction of the observer. If the
scattering generates a luminance contrast above a threshold value, the contrail is
visible with a naked eye. The threshold value is determined by the optical thickness
(or optical depth), defined as 4

3D ρ
τ= ml Qext , (19.1)
2 d ρp
where D is the geometrical thickness of the contrail in the direction of the observer, d
is the average diameter of the particles in the contrail, ρ p is the mean particles density,
ml is the mass fraction of liquid or iced water, and Qext is a parameter called extinction
efficiency, which depends on the wave-length of light in the visible spectrum. A
detailed analysis of Equation 19.1 indicates that the optical depth is dependent on
the emission index for ice particles and varies from zero to approximately 3.84 for a
particle diameter d ∼ 6 μm. Review of data in the technical literature indicates that
a contrail having a physical depth of 50 m is visible if made of water droplets with
d ∼ 2 μm and water content ∼102 kg/m3 . As far as viewing angles are concerned, the
most favourable conditions are when the contrails are at a small angle with respect
to the sun.
The speculation that contrails affect the climate is not new. A short article
on contrails appeared in the magazine Popular Mechanics in March 1943. At that
time there was only a limited effect, due to constraint on both altitude and speeds;
airplanes of that time were powered by internal-combustion engines.
A study published in 1970 5 indicated that the ice crystals in the contrail caused
an inadvertent cloud seeding, thus leading to increased cloud cover and modification
of precipitation conditions. The research on contrail formation is now mature and
extensive. The research on mitigating options is still lagging behind. A work pub-
lished by Fichter et al. 6 addressed the impact of the flight altitude and the possibility
of changing flight levels. This paper demonstrated that a descent in 2,000 feet
(∼ 610 m) steps decreases the likelihood of contrail formation. A descent by 6,000
feet (∼1,830 m) would decrease the contrail coverage by 45%. Ström and Gierens 7
carried out a simulation of a liquid-hydrogen–powered airplane and found that
although the amount of water vapour in the exhausts is considerably larger than in
the exhaust plumes of a conventional airplane, the absence of aerosol particles (soot)
decreases the formation of ice crystals; hence the contrails are optically thinner.
Consideration of alternative fuels such as hydrogen is important, but it does
not consider the real constraints of the aviation industry. The only available power
plants are gas-turbine engines running on aviation kerosene. This is not going to
change anytime soon. Commercial airplanes of the current generation will still be
in service 25 years thence, with some airplanes operating for as many as 40 years.
New airplanes being rolled out now (Airbus A380, Boeing 787, Airbus A350) will
be operating beyond 2050. No one will scrap these airplanes with the argument
that a new, more environmental airplane is on the way. Although power-plant re-
engineering is not uncommon, as discussed in Chapter 1, the question is to what
extent a revolutionary power plant (that yet does not exist) can be integrated into
an existing airframe and ground infrastructure.
19.1.1 Cirrus Clouds

Aside from the visual aspects of the contrails moments after formation, there are
several – more serious – aspects that must be considered. We are specifically con-
cerned with the evolution of the contrails into cirrus clouds and the effects of the
radiation balance within the lower atmosphere 8;9 .
Cirrus clouds are established at relatively high altitudes, between the high tro-
posphere and the lower stratosphere. They are composed exclusively of ice particles.
These types of cirrus clouds have a long life in supersaturated air. The cirrus clouds
expand over a period of several minutes to a few hours, aided by high-altitude winds;
eventually they transform into cirrus clouds that are almost natural, although their
origin is not. One such example is shown in Figure 19.2 where a cirrus cloud, several
minutes old (top right), is seen near a fresh contrail (left).
Figure 19.2. Example of degeneration of a contrail into a cirrus cloud.
The origin of the contrail takes place within a fraction of a second from the
emission. In some instances it forms rings that become unstable and eventually
break up and disappear, thanks to the evaporation of the ice crystals. One such
example is shown in Figure 19.3.
The technical literature reports also cases of aircraft dissipation trails named
distrails 10 . These events are much less common, less visible and of shorter dura-
tion. They occur when an aircraft passes through a supercooled altocumulus cloud.
Because these clouds have altitudes below the normal cruise altitude (below 20,000
feet), the aircraft is likely to encounter this situation when changing altitude.
Figure 19.3. Example of unstable contrail; break-up of rings, dissipation and evaporation.
14
M = 0.85 M = 0.83
450
E
430
13
Flight altitude, km A
410
Flight Level
B
C
12
390
D
370
11
350
ΔSAR = -1.5%
10 330
0.045 0.05 0.055
SAR, n-miles/kg
Figure 19.4. Altitude flexibility analysis for the Airbus A380-861; W = 323.70 tons; standard
day, no wind.
19.1.2 Cruise Altitude Flexibility

We investigate the changes in operational conditions at the current technology level
that would lead to a decrease in contrail impact. When we look at these options,
we soon find out that they must be discarded. For example, a trade in flight alti-
tude cannot be done at the expense of fuel consumption. As we demonstrated in
Chapter 12, commercial airliners operate at maximum SAR over a restricted range
of altitudes. The optimal cruise conditions for a given all-up-weight are established
by a unique combination of Mach-altitude parameters. One way of overcoming this
limitation is to increase the altitude flexibility so that the optimal cruise conditions
become independent of altitude. We have demonstrated 11 that this is not possible
in the general case because of limits in the flight physics; however, a ±2,000-foot
altitude flexibility can be obtained by trading in the cruise Mach number.
ALTITUDE-SAR RELATIONSHIP. One example is shown in Figure 19.4, which shows the
simulated altitude performance of the Airbus A380-861 with GP-7270 engines. There
are two curves showing the SAR at M = 0.85 and M = 0.83. Assume the airplane is
cruising at M = 0.85, which is close to the long-range Mach number (LRM). The best
flight level would be FL-420 (point A), unless the aircraft is forced to a lower level
by the ATC. A descent by 2,000 feet to FL-400 (point B) would cause a loss in SAR
equal to about −1.5%, which is relatively high. Alternatively, the airplane could fly
at FL-400 (point C) and at the lower Mach M = 0.83 (close to the maximum-range
Mach number). This procedure has the advantage of actually increasing the SAR,
although the airplane would have to sacrifice some speed. It can do even better by
descending to FL-380 (point D) at M = 0.83. This option requires again a loss of
about 1.5% SAR. One more option is to move higher to FL-430, either at M = 0.85
or M = 0.83, which can perhaps avoid some high-altitude weather conditions without
sacrificing speed or air range. In conclusion, a flexibility of about 5,000 feet can be
achieved, as long as the airline operator accepts a loss in SAR equivalent to about
1.5%. For an airplane such as the A380, a 1% loss in air range means an extra 0.22 kg
of fuel per n-mile (at the weight indicated in the graph), or 220 kg per 1,000 n-miles,
which easily reaches 2,000 kg per day. In short, at the end of the day, this procedure
will cost a phenomenal amount of capital and whilst avoiding one environmental
problem it would compound another one – the increase in carbon-dioxide emissions
from fuel burn.
If we accept this 2,000-foot flexibility by using a combination of Mach-altitude
flight management, we still need to negotiate with the ATC that we intend to descend
from the flight level previously allocated. An alternative could be a go-around, i.e.,
contrail avoidance by making a detour rather than a change of altitude. Extending
the cruise by just going around is not workable, either in commercial or in practical
terms. In any case, for these options to be possible, we would need a completely new
ATC system, one that is able to handle unscheduled changes of flight level, possibly
from all directions.
What is a 1% of fuel savings? As of 2007* , worldwide use of aviation fuel stood
at ∼515,000 tons/day (188 million ton/year). If we assume that only 80% of the fuel
is burned at cruise (on average), then a 1% savings would lead to a savings of ∼4,100
ton/day of aviation fuel, or ∼13 million kg of CO2 /day.
DESIGN ISSUES. Assume that an airplane is to operate at constant TAS over the
altitude range suitable for cruise. To make the SAR insensitive to altitude, we would
need to make the drag insensitive to altitude. The TSFC is an engine-performance
indicator and is temporarily excluded from the analysis. The drag analysis shown in
§ 4.2.12 indicates that the dominant effect is that of the induced drag, which grows
rapidly with the altitude; this effect is ultimately attributed to the air density. Let
us set dD/dh = 0. This condition determines the optimum altitude with respect to
aerodynamic configuration alone 11 . After carrying out the algebra, this equation
takes the following form:

dρ d 1
+c = 0, (19.2)
dh dh ρ
where the factor c is defined

2
k 2W 1
c= < 1. (19.3)
CDo AV V2
A quick analysis of Equation 19.2 shows that the optimal cruise altitude decreases
when the factor c increases. Such an effect can be achieved with a number of options,
most of which are unreasonable. One reasonable option, albeit a challenging one, is
the reduction in profile drag.
If we look closer at the TSFC, we find that it is not constant; in fact, it may
suffer large variations at cruise altitude as a result of increasing or decreasing fuel
flow (i.e., increasing or decreasing thrust or drag). If the drag increases, so does the
TSFC. In other words, the TSFC and the drag move in the same direction as the
* ICAO Environmental Report 2010.

altitude increases or decreases. Therefore, the optimum design problem involves

aerodynamics as well as propulsion.
A number of studies 12;13 proposed a cruise altitude shift. This shift was found
to be dependent on the latitude and the time of year. These papers also refer to a
number of studies that address the consequences of shifting cruise altitude above
and below the current levels. In particular, Mannstein et al. 13 , following extensive
measurements, contend that a ± 2,000-foot shift in altitude would be sufficient to
avoid 50% of the contrails. Measurements shown by Sussmann 14 behind a Boeing
B747 show that the vertical dispersion of the vortex wake determines the vertical
dispersion of the contrail. An increased diffusion would limit the effectiveness of
any cruise-level shift.
19.1.3 The Contrail Factor

Having excluded large changes in cruise levels, reduction in cruise Mach number,
changes of aircraft routings and major upgrades in the air traffic control, we are
left with the engine technology. What can be done? First of all, let us examine the
likelihood of contrail formation. There is a relatively simple rule that can be used:
this is the Appleman-Schmidt criterion, which is generally written as
e q C p
G= =p , (19.4)
T h j
where G is the ratio between changes in water vapour pressure (e) and the corres-
ponding change in temperature during mixing; p is the atmospheric pressure at the
flight level; C p is the specific heat capacity of air; q is the change in water content in
the plume; and h p is the corresponding change in enthalpy. The contrail parameter
is
q C p
C= . (19.5)
h j
The contrail parameter is proportional to the contrail factor: G = p C. The function
e(T ) is called mixing line; thus, Equation 19.4 gives the slope G of the mixing line.
The threshold temperature T M for H = e/esat = 100% (saturation) can be fitted
by the following equation 4 :
) *2
T M = −46.46 + 9.43 ln(G − 0.053) + 0.720 ln(G − 0.053) , (19.6)
with T M in ℃ and G in Pa/K. The threshold temperature corresponding to a relative

humidity H < 100% is
1
TC = T M − [esat (T M ) − Hesat (TC )] . (19.7)
G
Equation 19.4 can be written in a form that contains the propulsive efficiency of the
engines. This form of the equation was further demonstrated to be extremely import-
ant because it links the probability of contrail formation to the engine performance.
THE APPLEMAN-SCHMIDT CRITERION. An important consequence of the criterion is

the relationship among flight altitude, humidity level, and contrail existence. An
example is shown in Figure 19.5a. This graph shows levels of constant relative humid-
ity in the temperature-altitude space. In particular, it shows levels corresponding to
100 100
RH = 0% RH = 0%
RH = 100% RH = 100%
T for persistence T for persistence
400 400
200 200
Pressure, 102 Pa
Pressure, 10 Pa
350 B 350
Flight Level
Flight Level
2
300 300
300 300
C A
No Contrail No Contrail
Formation C A
400 Contrail 250 400 250
Formation
Possible Possible
Contrail Contrail Contrails
ISA Formation
ISA - 5o
200 200
500 500
-60 -50 -40 -30 -20 -60 -50 -40 -30 -20
o
Temperature, C Temperature, oC
(a) Standard day (b) Cold day
Figure 19.5. Application of the Appleman-Schmidt criterion to contrail prediction; threshold
temperature at selected levels of relative humidity.
dry air (H = 0%) and saturated air (H = 100%). Assume that an aircraft is flying at
ISA conditions at FL-340. Its operation point is indicated by point B in Figure 19.5a.
At this point the atmosphere is dry. Therefore, there is a high probability of creating
a persistent contrail. To avoid this situation, the aircraft should descend to FL-290 to
point A. At this point, in fact, the air is saturated with water vapour and the aircraft
will not create a contrail. A descent to an intermediate altitude cannot guarantee that
a contrail will be avoided. Thus, the descent required is 50 flight levels, that is, 5,000
feet or more. In a standard atmosphere, contrails will form at altitudes above 8.2 km.
If the atmosphere is colder, for example ISA −5 ℃, then the situation is shown
in Figure 19.5b. Assume again that the aircraft is flying at FL-340. Its operational
point is B. This point is in the region of dry air; therefore, a contrail will certainly
form. To avoid the contrail, the aircraft would have to descend to FL-250.
19.1.4 Effects of Propulsive Efficiency

Let us begin from the propulsive efficiency of the jet engine, which is
FN V
η= , (19.8)
ṁ f Q
where FN is the net thrust, V is the true air speed, ṁ f is the fuel flow, and Q is
the combustion heat of the aviation fuel (Q ∼ 43.5 MJ/kg; see § 14.2). At cruise
conditions this efficiency is between 0.27 and 0.33; it decreases slightly as the aircraft
burns fuel. The enthalpy of the plume* is
1
h p = h + (Vj − V)2 , (19.9)
2
where Vj − V is the relative speed of the plume with respect the true flight speed V.
We now write the energy balance in a frame of reference moving with the aircraft.
* There is a difference between jet and plume: a turbulent jet is a flow produced by a pressure
drop through a nozzle. A turbulent plume is a fluid motion whose primary source of energy and
momentum is body forces derived from density variations.
The change in total energy (kinetic and thermal) is equal to the flow of combustion
energy due to fuel burn. If fuel is burned at a rate ṁ f , the input energy rate is ṁ f Q.
The change in total energy of the plume is the product between the total mass-flow
rate and the change in enthalpy

1 2
ṁ f Q = ṁ j h j − he + Vj − V 2 , (19.10)
2
where he denotes the enthalpy of the external gas. The same energy-budget equation
written in a reference fixed on the ground is

1 2
ṁ f Q − FN V = ṁ j h j − he + Vj − V 2 . (19.11)
2
The product FN V between the net thrust and the true air speed V is the useful
work done on the aircraft, which is moving by reaction. Thus, this quantity must be
subtracted from the inflow of energy. Insert the definition of propulsive efficiency
(Equation 19.8) into Equation 19.11 to find

1 2
(1 − η)ṁ f Q = ṁ j h j − he + V −V 2
(19.12)
2 j
ṁ j
(1 − η)Q = h p , (19.13)
ṁ f
where

1 2
h p = h j − he + Vj − V 2 (19.14)
2
denotes the change in enthalpy of the plume. The mass budget for water vapour is
ṁ f
q = q j − qe = EIH20 , (19.15)
ṁ j
where EIH20 is the emission index of water, that is, the amount of H2 O produced by
the jet for a unit mass of fuel burned; ṁ f is the fuel flow, and ṁ j is the mass-flow rate
of the jet. If we insert Equation 19.15 and Equation 19.13 into the contrail parameter
(Equation 19.5), we finally get the following expression
pC p
G = EIH2O . (19.16)
Q(1 − η)
The relationship between the contrail threshold temperature and the overall propul-
sive efficiency is shown in Figure 19.6 for selected values of the relative humidity. The
shaded area indicates the range of propulsive efficiency of commercial aircraft; the
highest values of η correspond to the most modern engines. This is an inconvenient
result. In fact, as engines become more efficient, they become more likely to create
the conditions for contrail formation. This result was demonstrated by Schumann
et al. 15 on the basis of flight tests with an Airbus A340 and a Boeing B707 flying side
by side: the A340 turned out to produce contrails, whereas the B707 did not, under
comparable conditions (same altitude, same air speed).
Graph 19.6b shows that the vertical corridor for contrail formation increases
with the increasing efficiency. This result would point to the option of flying higher
rather than lower.
-35 11
Contrail Threshold Temperature, C
Contrail Threshold Alititude, km

o
-40
80% 10
60% RH = 20%
40%
-45 40%
9 60%
80%
-50
RH = 20%
8
-55
-60 7
0.1 0.2 0.3 0.4 0.5 0.2 0.3 0.4 0.5
Propulsive efficiency Propulsive efficiency
(a) (b)
Figure 19.6. Threshold contrail factor and altitude as a function of propulsive efficiency.
A further application of the altitude flexibility concept is shown in Figure 19.7.

The shaded area indicates the vertical displacement of the contrail and must be
interpreted as a region where there is a high contrail factor. The aircraft travels
from the left of the graph to the right. The normal procedure indicates that at
some point the aircraft is going to enter contrail conditions at FL-330. However, if
sufficient information is available well ahead of time, the flight control system can
be programmed to avoid the grey area. This is done with two stepped descents from
the initial FL-310 and one stepped climb until the altitude stabilises at FL-290. Aside
from ATC issues, there is the problem in gathering enough atmospheric information
ahead of the aircraft. The scenario shown in Figure 19.7 cannot be generalised.
15,000
410
Normal Procedure
370
12,000
330
Flight Level
Altitude, m
290
9,000
250
Contrail Path Contrail
Avoidance
210
6,000
170
0 500 1,000 1,500 2,000
Cruise distance, n-miles
Figure 19.7. Example of contrail-avoidance strategy with shift in cruise altitude.
19.2 Radiative Forcing of Exhaust Emissions 599
However, there are software models capable of modelling the contrail frequency over
a large region from a mix of weather forecast, atmospheric parameters and aircraft
characteristics 16 . The work of Irvine et al. 17 shows an analysis of the effect of weather
patterns, including winds and contrail factors, on the optimal flight altitude of a trans-
Atlantic flight: the weather-optimal trajectory is often outside the acceptable flight
levels (FL-290 to FL-410).
19.1.5 Heat Released in High Atmosphere

We consider the problem of combustion heat released in the high atmosphere during
a conventional cruise. The heat released is calculated from
Qc ∼ (1 − η)Qṁ f tc ne , (19.17)
where η is the average propulsive efficiency, Q is the specific combustion heat of

the aviation fuel, tc is the total cruise time, and ne is the number of engines. We
have verified that the propulsive efficiency changes very little during the cruise;
therefore, the approximation is acceptable. This means that a commercial airplane
in the same class as the Airbus A320-200 releases about 140 GJ of heat during the
cruise segment of a 2,000 km (1,089 n-miles) flight. This is equivalent to burning
∼ 3,200 kg of aviation fuel in the open air at the flight altitude, corresponding to
an average propulsive efficiency η 0.28. If we expand from this result and apply
the calculation to the whole of commercial aviation, we conclude that heating of
the upper atmosphere is not only a result of the persistence of greenhouse gases
and persistent contrails but also a direct heating of the atmosphere. It has been
demonstrated that the diffusion of the exhaust plume has a complex behaviour 18;19 .
There are several implications, including, for example, the dilution of the plume and
the vertical size of the plume.
19.2 Radiative Forcing of Exhaust Emissions

There are several contributions to the exhaust emissions, some of which are direct
and others are indirect, that is, by-product of a pollutant with the atmosphere. In
addition to contrails and cirrus clouds, as previously discussed, we have:
r Carbon-dioxide, CO . This is by far the largest contribution because it is approx-

2
imately equal to 3.1 times the weight of the fuel burn; CO2 has long been con-
sidered the main culprit of environmental pollution.
r Nitrogen oxides, NO . This contribution includes both NO and NO ; they are a
x 2
by-product of the combustion of aviation fuel; and they affect the production of
ozone O3 and methane CH4 . Their radiative forcing is investigated by Gardner
et al. 20 and Stevenson et al. 21 . The problem is compounded by high combustor
temperatures in modern gas-turbine engines.
r Ozone, O . At flight altitudes below 13 km (∼42,650 feet), the emission of NO
3 x
causes an increase in O3 , whilst at higher altitude (16 to 20 km), it causes the
destruction of O3 22 . Aircraft-induced ozone can have a long-term contribution
to the atmospheric temperature.
r Methane, CH . Methane is a powerful greenhouse gas. This contribution is not

4
by itself relevant because it is not a by-product of gas-turbine combustion, but
NOx emissions could contribute to the reduction of CH4 , which would have a
cooling effect.
r Water vapour, H O. Emission of water vapour from gas-turbine engines is relat-
2
ively small, compared to the weather effects in the troposphere, and is removed
by precipitation. However, at very high altitudes, where the atmosphere is dry
(see also Figure 8.2), an increase in water content acts as a greenhouse effect.
Due to the limited air circulation at stratospheric altitudes, the water content
has a long residence time. A side effect of stratospheric vapour is to create
stratospheric clouds that may have direct and indirect effects on the ozone.
r Soot and Sulphates. Soot is made of extremely small particles, less than 30 μm
in diameter. Even smaller unburnt liquid particles are removed from the exhaust
plumes; these particles combine with the atmospheric components to produce
sulphuric acid, hydrocarbons, and water vapour, as well as electrically charged
molecules.
One of the key aspects of the entire pollution problem is the long-term effect
of the various forms of pollutants. This effect is measured by a parameter called
radiative forcing. In simple words, the radiative forcing is a means of estimating by
how much the atmospheric temperature increases as a result of a pollutant, assuming
that the pollutant effect is similar to a greenhouse. A mathematical expression often
given for the radiative forcing is
Ts
RF = , (19.18)
λ
where Ts is the change in mean global surface temperature and λ is a sensitivity
parameter. The radiative forcing is measured in [W/m2 ] (or more practically in
[mW/m2 ]); therefore, the sensitivity parameter λ is given in [K m2 /W]. The problem
with Equation 19.18 is that the sensitivity parameter varies greatly depending on
the pollutant. Despite these caveats, Equation 19.18 is used to compare different
emissions.
The value of the radiative forcing is still a matter of scientific debate. We report
data based on the study of Sausen et al. 23 , which gives an indication of the various
effects and their level of uncertainty. We draw the reader’s attention to the fact that
the warming effects caused by carbon dioxide (CO2 ), ozone (O3 ) and contrails are
of similar magnitude (RF ∼ 20 mW/m2 ). Soot and water vapour are also of the same
magnitude but considerably smaller (RF ∼ 2 mW/m2 ). The radiative factor of the
cirrus clouds is presently indeterminate. Various estimates indicate this contribution
to be higher than any other contribution, with values of ∼25 mW/m2 , Minnis et
al. 8 . The upper bound of the uncertainty is considerably higher, up to ∼80 mW/m2 .
Finally, the radiative factor of methane (CH4 ) is negative (RF ∼ −14 mW/m2 ), which
would contribute to a cooling rather than a warming.
19.3 Landing and Take-Off Emissions

The landing and take-off (LTO) emissions are used to estimate the environmental
impact of commercial aircraft at altitudes below 3,000 feet (∼915 m). This altitude
Figure 19.8. Landing and take-off operations of a commercial flight.
corresponds approximately to the mixing layer: pollutants emitted below this average
altitude can have effects on air quality at lower altitudes. Emissions at ground level
will have a shorter-term impact than emissions at higher altitudes.
Figure 19.8 shows the typical flight modes of a commercial flight at a large airport.
The airplane is on a final approach (A), proceeds to landing (B) and then performs
a roll to the gate (C), eventually stopping for other traffic, turning at several points,
before eventually reaching the gate (D). When it is ready for departure, it rolls out
(E) and reaches the start of the runway to a brake-release point (F). It then waits
for the go-ahead, performs a ground run (G), a take-off (H), an initial climb (I) and
reduces the thrust (J). All of these phases are included in the LTO emissions, as long
as the airplane is below 3,000 feet.
The ICAO defines five different modes for the purpose of certification and for
dispersion calculations. These modes are 1) final approach, 2) taxi-in and ground idle,
3) taxi-out and ground idle, 4) take-off, and 5) climb-out. The taxi-in and out are
essentially the same mode. ICAO goes on to propose an average time in each mode
and an average thrust setting on each mode, as given in Table 19.1. During an actual
flight, neither the times nor the thrust settings are the same as those in Table 19.1.
The pollutants considered in the analysis are NOx (nitrogen oxides), HC (hydro-
carbons), CO (carbon-monoxide), SOx (sulfate oxides), particulate matter (or soot),
in addition to carbon-dioxide CO2 , which is proportional to the fuel burned. With
the exception of the latter component, the other pollutants depend non-linearly on
the thrust rating. To avoid complicated analysis that involves combustion modelling,
these emission indexes can be inferred from a databank published by the ICAO
itself* . The emission indexes are generally given as grams of pollutant per kilogram
of fuel burn. An example is shown in Figure 19.9 for the CO, NOx and HC. Thus,
the emission E j of pollutant j would be

Ej = EIi m fi , (19.19)
i
* ICAO Engine Emissions Data Bank. Periodically updated. Available from the ICAO and other
aviation regulators. This databank does not contain information on turboprop engines and APU.
Table 19.1. ICAO flight modes, times and thrust

rating as % of maximum thrust
Flight Mode Time [min] % Thrust
Final approach 4.0 30

Taxi-in & ground idle 7.0 7
Taxi-out & ground idle 19.0 7
Take-off 0.7 100
Climb-out 2.2 85
Total 32.9
where the fuel-in-mode i is calculated from m f = ṁ f t; this equation requires the

knowledge of both the fuel flow and the time in mode; the fuel flow would be
proportional to the thrust-in-mode. However, there is a problem here because this
assumption implies that the TSFC is a constant. This assumption is only approxim-
ately true at high thrust settings (take-off and climb-out), as illustrated previously
(see Chapter 5).
A more accurate estimation of the LTO emissions relies on the actual fuel burn
for all the flight modes below 3,000 feet. The emission indexes are dependent on
the thrust rating, as shown in Figure 19.9. Therefore, we still need to approximate
the ICAO emission indexes with interpolation functions. The emission of species j
during the time step dt is calculated from

dm j = ṁ f (N1 )dt EI j (N1 ). (19.20)
This equation expresses the fact that the emission dm j of species j is proportional
to the fuel flow ṁ f and the emission index EI j . Both the fuel flow and the emission
50 4
Idle
CO
NOx
HC
40
3
Emission index, gr/kg
Emission index, gr/kg
30
Approach 2
Take-off
Climb-out
20
1
10
0 0
0 20 40 60 80 100
N%1
Figure 19.9. Emission indexes for the CFM56-5C4 as a function of the engine rpm. Data
elaborated from the ICAO databank.
Table 19.2. LTO emissions summary for Airbus A320-200 with CFM56 engines; standard
day, no wind, 2,000 km (1,079 n-miles) mission; all data in [kg], as calculated
Segment CO NOx HC t[min] Notes
1 Taxi-out, roll 0.942 0.358 0.065 6.7

2 Taxi-out, idle 1.885 0.716 0.131 3.3
3 Takeoff 4.288 1.727 0.195 0.4
4 Climb-out 5.022 2.929 0.057 1.5 to 3,000 feet
5 Approach 1.337 2.391 0.026 3.1 from 3,000 feet
6 Landing 0.150 0.149 0.008 0.8 from 35 feet
7 Taxi-in, roll 0.410 0.109 0.029 6.7
8 Taxi-in, idle 2.049 0.546 0.146 1.3
9 APU 0.018 0.025 0.001 23.9 total
Total LTO 16.084 8.925 0.658 23.9
index are a function of the actual engine speed N1 (or N%1). The actual emission
index is interpolated from the reference values in the ICAO database. The LTO
emissions are calculated by integration of Equation 19.20, that is, by integration of
the fuel flow when the airplane is below 3,000 feet:
b b
E j = ne dm j = ne ṁ f (N1 )EI j (N1 )dt, (19.21)
a a
where ne is the number of operating engines. In Equation 19.21 “a” denotes the
moment when the airplane leaves the gate; “b” denotes the time when the airplane
reaches the gate at destination. In summary, the computational procedure is the
following:
1. At the current flight condition, solve the engine problem: Wf6, N%1.
2. Calculate the emission index corresponding to N%1 by interpolation of the ICAO
data (see Figure 19.9).
3. Calculate the emission dmi within the time step dt from Equation 19.20.
4. Advance the flight time by dt and proceed from point 1, as long as h < 3,000
feet.
An example of calculated LTO emissions is given in Table 19.2, which refers

to the Airbus A320-200 with CFM56-5C4 turbofan engines; the flight planning was
done for a 2,000-km (1,079-n-mile) mission, on a standard day, without winds, full
passenger load and 300 kg of bulk cargo. The flight scenario involves a 10-minute
taxi-out and an 8-minute taxi-in. The engine warm-up is not included in the table.
The LTO emissions for this engine quoted by the ICAO databank can be quite
different from the calculated values. In fact, considerable variations can be detected
among different types of flights, and caution must be exercised when taking averages.
Actual data can differ by a factor 2. Note that the LTO time is 23.9 minutes instead of
32.9 minutes prescribed in Table 19.1. The main difference comes from the average
taxi-out roll.
A sub-case of the LTO emissions is the ground-operations segment. At conges-
ted airports, some aircraft spend up to one hour between pushing back from the
gate and starting the take-off run (segment D to F in Figure 19.8); for most of this
1,500 0.5 1,500 0.5

CO2/pax CO2/pax
CO2/pax/nm CO2/pax/nm
1,200 1,200
0.4 0.4
CO2 / pax / n-mile, kg
CO2 / pax / n-mile, kg

CO2 / pax , kg
CO2 / pax, kg
900 900
0.3 0.3
600 600
0.2 0.2
300 300
0 0.1 0 0.1
0 1,000 2,000 3,000 4,000 5,000 0 1,000 2,000 3,000 4,000 5,000
Stage length, n-miles Stage length, n-miles
(a) 75% pax load (b) 100% pax load
Figure 19.10. Calculated CO2 emissions of the Boeing B777-300-GE, standard day, 9 kt
head wind.
time, the engines run in idle mode and consume large amounts of fuel whilst emit-
ting harmful exhaust gases. Reduction of these emissions is beyond the domain of a
single airplane and must be examined in the context of airport operations, queueing
strategies and flight distributions.
19.4 Case Study: Carbon-Dioxide Emissions

We now set the task of studying the carbon emissions of a commercial airplane,
using as a parameter the stage length and the passenger load. The latter parameter
is expressed as a percent of filled seats; 100% means full aircraft. These airplanes
generally carry also some form of cargo. However, we have excluded this case in
order to highlight the emissions as a function of passenger load only.
Figure 19.10 shows the calculated CO2 emissions over a range of required stage
lengths. The data are shown in terms of emissions per passenger and per n-mile. The
results illustrate that the emissions per travelling passenger increase almost linearly
at intermediate stage lengths; over long-haul distances the increase is super-linear,
due to the fact that a considerable portion of the aircraft’s weight is made up of fuel.
The fuel-use efficiency is better shown by the quantity CO2 /pax/n-mile.
A similar analysis is carried out with the Airbus A320-200, as shown in Fig-
ure 19.11. Again, it appears that the emissions per passenger are considerably higher
when the aircraft operates at only three quarters of full capacity. Additional analyses
are presented in Ref. 24 .
The final case refers to a parametric analysis of a typical transport mission. We
have included the effects of nine different parameters, as listed in Figure 19.12. The
sensitivity analysis consists in changing the parameters listed over a nominal value,
which correspond to a specified mission. The mission is a London–New York flight
(5,570 km; 3,078 n-miles), with 80% passenger load, 15 tons of bulk cargo, 18 minutes
of taxi-out time at London, 8 minutes taxi-in at New York, and 20 kg/pax baggage
allowance, on a standard day, with an average cruise wind speed equal to an 18-kt
head wind.
19.4 Case Study: Carbon-Dioxide Emissions 605
600 0.24 600 0.24

CO2/pax CO2/pax
CO2/pax/nm CO2/pax/nm
CO2 /pax/nm, kg
CO2 /pax/nm, kg
400 0.2 400 0.2
CO2 /pax, kg
CO2 /pax, kg
200 0.16 200 0.16
0 0.12 0 0.12
1,000 2,000 3,000 1,000 2,000 3,000
Stage length, n-miles Stage length, n-miles
(a) 75% pax load (b) 100% pax load
Figure 19.11. Calculated CO2 emissions of the Boeing Airbus A320-200-CFM, standard day,
9-kt head wind.
Variation of the nominal parameters is as follows: bulk cargo ± 5 tons (effect

of commercial payload); passenger load ±10% (effect of passenger revenue); flight
route −25 n-miles and +75 n-miles (effect of routing); extra fuel load +1 or +2
tons (error in fuel loading); taxi-out time ±6 minutes (effect of airport congestion);
outside air temperature ±20 ℃ (weather effects); and wind speed of 9 kt or 36 head
wind (weather effects). The final parameter is a continuous descent approach (CDA)
at the final destination; this parameter can be true or false; it denotes an accurate
flight trajectory, although it is subject to certification. The data in Figure 19.12 show
the response to a change of a single parameter. By far the most important parameter
is the wind, which can add or subtract several percentage points in fuel burn and
10 1. Air Temperature
2. Mission Range
3. Cargo weight
8 4. Extra fuel weight
5.Cruise winds
Change in Fuel burn, %
6 6. Passenger load
7. Bag allowance
4 8. Taxi-out time
9. Continuous descent
2
-2
-4
-6
0 1 2 3 4 5 6 7 8 9 10
Sensitivity parameter
Figure 19.12. Fuel-burn sensitivity to key mission parameters for the Boeing B777-300-GE
(simulated).
Destination
Restricted
Air Space
Departure
Departure Destination
(a) Vertical (b) Horizontal

Figure 19.13. Vertical and horizontal profile of a typical commercial flight.
exhaust emissions. Air temperature is also an important contributor, but there is little
that can be done. Interestingly, an error in excess in fuel load does not contribute
much to the total fuel consumption, whilst baggage allowance and taxi-time are
operational items that can be controlled. A CDA can have some benefits, as also
shown in § 11.3.
The issues highlighted in this sections are applied to a single aircraft flight. It
is possible to scale up the investigation to include one type of aircraft, one route,
and one airport pair. Thence, it is possible to extrapolate to a larger or even global
air transportation system, to include all types of emissions, which can then be clas-
sified by region, by season, by aircraft movements and other parameters. One such
framework is available with the SAGE initiative 25 .
19.5 The Perfect Flight

The previous chapters have explained how commercial flights operate under ATC
and a considerable number of constraints, Figure 19.13. Climb to initial cruise altitude
is done in steps, including at least one level acceleration; cruise is done in steps,
with constant-altitude segments followed by climbs, which are required to move
the airplane between recognised flight levels. Descent and final approach are also
done in steps, and when not constrained by the topography, airplanes are set on
standard flight levels and eventually on holding patterns. The flight path from origin
to destination is not a straight one. Air traffic conditions, weather patterns and
temporary restrictions due to external factors may cause the airplane to take winding
routes to the final destination.
These apparently complex procedures simplify the tracking of airplanes and
allow the current ATC to manage commercial traffic. However, all of these proced-
ures force the airplane to follow sub-optimal flight paths and waste further time and
fuel at terminal areas. Keeping the airplane on a hold is perhaps the principal source
of wastage. Data from London Heathrow airport (2011) indicate that airplanes spend
about 55 cumulative hours on holding patterns every day, corresponding to about
190 tons of fuel and 600 tons of carbon-dioxide emissions. The perfect flight should
not have to comply with any flight level; the airplane would be free to move to the
best altitude. To be fair, the idea is not completely new. Flight-path optimisation
was applied as early as the 1970s to the DC-10 and Boeing B727 models 26 – 28 ; it was
demonstrated that a fuel savings of up to 10% could be achieved.
19.5 The Perfect Flight 607
If we exclude considerations on taxi times, on take-off run (which is essentially

unaffected by the perfect flight requirements) we would have the following flight
segments.
CONTINUOUS CLIMB. The most economic climb will be based on the cost index, which
is a weighted parameter that takes into account the cost of fuel as well as the cost of
time (§ 12.9).
CRUISE-CLIMB. We have demonstrated that the optimal cruise profile is a continu-

ous climb (§ 12.5.3). This strategy corresponds to minimum fuel burn for a given
initial weight. We establish the cruise condition from the optimal ICA-Mach at the
estimated initial cruise weight.
CONTINUOUS DESCENT. We assume that the airplane can perform a continuous des-
cent from the end of the cruise (top of descent) to the airfield (§ 11.3). Most of this
descent can be done with a nearly idle engine.
THE OPTIMAL TRAJECTORY. Following the previous discussion, there will be no hold-
ing on a stack and the airplane can proceed directly to landing. We assume the
standard international reserves (§ 15.5). We consider a “nearly” perfect flight, to
allow for the practical constraints in the real world: the flight will be as conventional,
except the cruise (which will be a continuous climb) and the descent, which will
be a continuous descent to 3,000 feet, with a short level flight, and descent on a 3-
degree slope. The climb is constrained by two constant-CAS segments, with a level
acceleration in between.
The flight-performance model has been run with the Airbus A320-211 powered
by two CFM56-5C4 turbofan engines. The summary of flight conditions is:
• standard day; no wind

• required range: 2,500 km; 1,349 n-miles
• passenger load: 100%
• cargo load: 300 kg
• taxi-out time: 10 minutes
• taxi-in time: 8 minutes
• departure/destination airports at sea level
• average passenger mass: 75 kg; average baggage mass: 15 kg
• climb schedule: conventional
• diversion flight: 2,000 feet below the FCA
• APU fuel: included in the analysis (ECS mode)
• fuel reserve: minimum between 20-minute extension and 200 n-mile diversion
• holding: 30 minutes at 1,500 feet
• U-turn at 5,000 feet
• engine state: new
• aerodynamics: clean
Note that the fuel reserve in these cases is always the 20-minute extension of
the cruise because the combination of 200 n-mile diversion and holding on a stack
requires more fuel.
Table 19.3. Analysis of a perfect flight with an Airbus A320-200 model; the range is 2,500 km
(1,349 n-miles)
Total Climb Climb Descent

Constraints Fuel Fuel Fuel Fuel m f %
Conventional 8,804 951 5,917 657 0.8
Scenario I Cruise-Climb 8,697 951 5,875 598 −1.20

Scenario II & No Restriction on ICA 8,351 1,062 5,350 655 −5.14
A summary of simulations is in Table 19.3; we also show the contributions of

the climb to ICA, cruise and descent; m f is the change in fuel burn, in percent,
compared to the conventional flight.
Figure 19.14 shows the trajectory of our nearly perfect flight. Note that there are
two flight segments, at climb and descent, that are level flights (A and B, respectively).
Flying at such high altitudes may have a strong side effect: this is the increased
environmental impact of emissions at altitude, including the risk of generating more
contrails and the effects on atmospheric ozone.
In Scenario I the aircraft is forced to set its initial cruise altitude to a recognised
flight level (FL-350), but then it is free to perform a continuous climb, followed by
a continuous descent from the top of climb. These conditions on cruise and descent
are maintained in Scenario II.
In Scenario II the aircraft is not constrained by a flight level and climbs to
a considerably higher altitude before starting its cruise (11,820 m; 38,870 feet).
Consequently, the aircraft burns more fuel to climb and to descend. However, the
reduced cruise fuel at a higher altitude (up to 40,000 feet) more than compensates
for these segment fuels. The fuel saving is estimated at about 430 kg, or 5% of the
fuel burned in a conventional trajectory; the “perfect” trajectory produces a savings
of about 1,300 kg of carbon dioxide.
19.6 Emissions Trading

The European Emission Trading Scheme (ETS) will require the industry to purchase
certificates that allow the emission of a specified amount of CO2 (measured in tons).
Due to the price volatility of both aviation fuel and certificates (which are traded
like commodities), it is not possible to make an accurate assessment of the additional
cost that airlines will incur, although it is likely to exceed 10% of the cost of fuel.
Emissions beyond the level specified in the certificates will incur a fine. Thence,
the argument goes, airlines will have to drive their emissions down. Any emissions
savings, with respect to the certificates, can be sold in the open market at commercial
rates.
The ETS by itself does not contribute to cutting emissions and may contrib-
ute to diverting profits away from the industry. However, because a price tag is
associated to these emissions 29 , according to the free market, there should be
financial incentives in keeping the emissions as low as possible. Initially, these
certificates are designed to apply to all flights to and from European airports.
However, unless there is worldwide agreement for this implementation, we will
19.7 Other Aspects of Emissions 609
12 40
35
10
Flight altitude, km
Altitude, 10 3 feet
30
8
25
6 20
15
4
A B 10
2
5
0 0
0 250 500 750 1,000 1,250 1,500
(a) Perfect flight
12 40
35
10
Flight altitude, km
Altitude, 10 3 feet
30
8
25
6 20
15
4
10
2
5
0 0
0 250 500 750 1,000 1,250 1,500
(b) Conventional flight

Figure 19.14. “Nearly” perfect trajectory (a) and conventional trajectory (b) for a commercial
flight.
see a distortion of the market that will actually cause an increase in aviation
emissions.
19.7 Other Aspects of Emissions

The environmental aspects of commercial aircraft can be summarised with the head-
ings shown in the flowchart of Figure 19.15. On the operation side, we have the
aspects discussed in this book: noise, fuel burn, exhaust emissions and contrails (and
contributions to the climate over short, medium and long terms) and the role of the
ATC.
Operation Design
Noise Energy
Airplane
Fuel Burn Materials
Emissions Resources
On the Ground
Contrails
ATC
Recycling
Land Use
Chemicals
Lubricants
Figure 19.15. Environmental challenges in commercial aviation.
The latter item should not be under-estimated; in fact, the ATC role has
increased with the expansion in air traffic. Today the ATC provide services such as
weather forecast, surveillance, communications, navigation, mission support, and all
traffic and congestion management. The system has outgrown itself, which shows in
sub-optimal flight paths, diversion routes, stacking patterns and so on. Major efforts
are ongoing to modernise the system, which should improve the environmental
performance of aircraft even in absence of improved aircraft technology. At the
very least, improved ATC should transfer flight delays from the air to the ground 30 .
Aircraft emissions not considered in our analysis include the following items:
r fuel dumping in emergencies
r fuel handling
r maintenance of aircraft engines
r painting of aircraft
r service vehicles (catering, technical service, passenger shuttles)
r aircraft deicing.
On the design side (which is beyond our scope) there are aspects of energy
use, raw materials and general resources, including land use and airport expansions.
Finally, on the ground side there are issues of recycling at the end-of-life and the use
of chemicals and lubricants (acquisition and disposal). What shall we do with about
300 aircraft due to retire from service every year?
Summary
We have introduced a number of environmental issues in aircraft flight perform-
ance. We started with the problem of aircraft contrails and cirrus clouds, which
are expanding rapidly and may cause further damage than the emission of CO2
from gas-turbine-engine combustion. We analysed the problem of altitude flexibility
Bibliography 611
and identified a potential source of improvement by using appropriate operational

changes. Among the major contributors of the exhaust gases, NOx , CO and hydro-
carbons HC have been discussed in the context of landing-and-take-off emissions.
We introduced a calculation method, which must rely on a reliable database. The
database is of limited use if the airframe and engines are poorly maintained. To offset
part of the emissions specifically associated to CO2 , we illustrated the problem of a
“perfect flight”, that is, a flight along the shortest trajectory that is not constrained by
a flight level or by climb and descent procedures. This procedure could lead to con-
siderable fuel savings, provided that the air traffic control will be able to cope with
the changes required. Finally, we briefly reviewed some elements of environmental
pollution around the aircraft itself.
Bibliography
[1] Penney J, Lister D, Griggs D, Dokken D, and McFarland M, editors. Aviation
and the Global Atmosphere. Cambridge University Press, 1999.
[2] Travis DJ, Carleton AM, and Lauritsen RG. Contrails reduce daily temperature
range. Nature, 418:601, 8 August 2002.
[3] Travis DJ, Carleton AM, and Lauritsen RG. Regional variations in U.S. diurnal
temperature range for the 11–14 September 2001 aircraft groundings: Evidence
on jet contrail influence on climate. J. Climate, 17:1123–1134, 2004.
[4] Schumann U. On conditions for contrail formation from aircraft exhausts.
Meteorologische Zeitschrift, 5(1):4–23, Jan. 1996.
[5] Murcray WB. On the possibility of weather modification by aircraft contrails.
Monthly Weather Review, 98(10):746–748, Oct. 1970.
[6] Fichter C, Marquat S, Sausen R, and Lee DS. The impact of cruise altitude on
contrails and relative radiative forcing. Meteorologische Zeitschrift, 14(4):563–
572, Aug. 2005.
[7] Ström L and Gierens K. First simulation of cryoplane contrails. J. Geophys.
Res., 107(D18):4346, Sept. 2002.
[8] Minnis P, Ayers PK, Palikonda R, and Phan D. Contrails, cirrus trends, and
climate. J. Climate, 17(8):1671–1685, April 2004.
[9] Spichtinger P and Gierens KM. Modelling of cirrus clouds – Part 1: Model
description and validation. Atm. Chem & Physics, 9(2):685–706, 2009.
[10] Duda DP and Minnis P. Observations of aircraft dissipation trails from GOES.
Monthly Weather Review, 130:398–406, Feb. 2002.
[11] Filippone A. Cruise altitude flexibility of jet transport aircraft. Aero. Science &
Technology, 14:283–294, 2010.
[12] Gierens K, Lim L, and Eleftheratos K. A review of various strategies for contrail
avoidance. Open Atm. Science J., 2:1–7, Jan. 2008.
[13] Mannstein H, Spichtinger P, and Gierens K. A note on how to avoid contrail
cirrus. Transp. Res. D, 10:421–426, Sept. 2005.
[14] Sussmann R. Vertical dispersion of an aircraft wake: Aerosol lidar analysis of
entrainment and detrainment in the vortex regime. J. Geophys. Res., 104(D2),
1999.
[15] Schumann U, Busen R, and Plohr M. Experimental test of the influence of
propulsion efficiency on contrail formation. J. Aircraft, 37(6):1083–1087, Nov.
2000.
[16] Schumann U. A contrail cirrus prediction tool. In Int. Conf. on Transport,
Atmosphere and Climate, Maastricht, the Netherlands, June 2009.
[17] Irvine E, Hoskins B, Shine K, Lunnon R, and Froemming C. Characterizing
North Atlantic weather patterns for climate-optimal aircraft routing. Meteoro-
logical Applications, 2012.
[18] Schumann U, Konopka P, Baumann R, Busen R, Gerz T, Schlager H, Schulte P,

and Volkert H. Estimate of diffusion parameters of aircraft exhaust plumes near
the tropopause from nitric oxide and turbulence measurements. J. Geophysical
Res., 100(D7):147–162, July 1995.
[19] Schumann U, Schlager H, Arnold F, Baumann R, Haschberger P, and Klemm
O. Dilution of aircraft exhaust plumes at cruise altitudes. Atmospheric Envir-
onment, 32(18):3097–3103, 1998.
[20] Gardner R, Adams K, Cook T, Deidewig F, Ernedal D, Falk R, Fleuti E,
Herms E, Johnson C, Lecht M, Lee D, Leech M, Lister D, Mass B, Metcalfe
M, Newton P, Schmitt A, Vandenbergh C, and van Drimmelen R. The ancat/ec
global inventory of NOx emissions from aircraft. Atmospheric Environment,
31(12):1751–1766, 1997.
[21] Stevenson D, Dohery R, Sanderson M, Collins W, Johnson W, and Derwent
R. Radiative forcing from aircraft NOx emissions: Mechanisms and seasonal
dependence. J. Geophysical Res., 109:D17307, 2004.
[22] Wuebbles DJ and Kinnison DE. Sensitivity of stratospheric ozone and possible
to present and future aircraft emissions. Technical Report JCRL-JC-104730,
Lawrence Livermore National Labs, Aug. 1990 (presented at DLR Seminar on
Air Traffic and Environment, Bonn, Nov. 1990).
[23] Sausen R, Isaksen I, Grewe V, Hauglustaine D, Lee D, Myhre G, Kohler M,
Pitari G, Schumann U, Frode S, and Zeferos C. Aviation radiative forcing in
2000: An update on IPCC (1999). Meteorologische Zeitschrift, 14(4):555–561,
Aug. 2005.
[24] Filippone A. Analysis of carbon-dioxide emissions from transport aircraft. J.
Aircraft, 45(1):183–195, Jan. 2008.
[25] Kim B, Fleming G, Balasubramanian S, Malwitz A, Kee J, Ruggiero J, Waitz
I, Klima K, Stouffer V, Long D, Kostiuk P, Locke M, Holcslaw C, Morales
A, McQueen E, and Gillette W. SAGE: System for Assessing Global Aviation
Emissions, Sept. 2005. FAA-EE-2005-01.
[26] Lee HQ and Erzberger H. Algorithm for fixed-range optimal trajectories. Tech-
nical Report TP-1565, NASA, July 1980.
[27] Erzberger H and Lee HQ. Constrained optimum trajectories with specified
range. J. Guidance, Navigation & Control, 3(1):78–85, Jan.-Feb. 1980.
[28] Ashley H. On making things the best – Aeronautical uses of optimization. J.
Aircraft, 19(1):5–28, 1982.
[29] Scheelhaase JD and Grimme WG. Emissions trading for international aviation
– An estimation of the economic impact on selected European airlines. J. Air
Transport Management, 13:253–263, 2007.
[30] Vranas DJ, Bertsimas PB, and Odoni AR. The multi-airport ground-holding
problem in air traffic control. Operations Research, 42:249–261, Mar. 1994.

a, b = limits of integration, Equation 19.21
A = wing area
c = factor defined by Equation 19.3
C = contrail parameter, Equation 19.5; concentration
Cp = specific heat capacity of air
d = particle diameter
D = contrail thickness; aerodynamic drag
e = vapour pressure
esat = vapour-saturation pressure
Ej = emission of pollutant species j
EI = emission index
EI H2 O = emission index of water
FN = net thrust
G = contrail function defined by Equation 19.4
h = flight altitude
h, h j = enthalpy; enthalpy in the jet
he = enthalpy of external flow
m = mass
M = Mach number
ml = mass fraction of liquid, Equation 19.1
ṁ f = fuel-flow rate (also Wf6)
ne = number of engines
N1 = engine rotational speed (also N1)
p = atmospheric pressure
q = specific water-vapour mass concentration
Q = combustion heat content of aviation fuel
Qc = heat released at high altitude, Equation 19.17
Qext = extinction efficiency
RF = radiative forcing
R = gas constant
t = time
tc = cruise time
T = temperature
TC = threshold temperature for humid air
TM = threshold temperature for saturated air
Ts = average global surface temperature
V = true air speed
Vj = speed of the jet
W = weight
Greek Symbols
= ratio between gas constants water vapour and air
λ = sensitivity parameter in radiative forcing
ρ = air density
ρp = particle density
τ = optical depth, Equation 19.1
[.]e = external (outside the jet/plume)
[.] f = fuel
[.] j = jet; counter for emission calculations
[.] p = plume
[.]sat = vapour-saturation conditions
20 Epilogue
Commercial aviation has matured rapidly since the introduction of gas-turbine-

engine technology. The investments required to develop new aircraft have increased
exponentially. These investments include capital, human resources across a wide
spectrum of competences, infrastructure and time. In this framework, the import-
ance of operational aircraft performance is crucial because the existing technology is
locked into a large fleet that is due to operate for several decades into the future. Air-
plane evolution is therefore inevitable and is proved in several case studies, both in
the commercial and in the military world. The issues raised in this book will continue
to dominate the debate; we expect considerable changes in the future, from the ATC,
to aero-engine technology, to aircraft design, operations and demand management.
Each chapter has illustrated major issues in aircraft flight performance; conclu-
sions have been drawn in each separate instance. To conclude, we wish to mention
some key aspects of wider interest.
ROLE OF COMPREHENSIVE ANALYSIS IN AIRCRAFT PERFORMANCE. Performance is

what sells an airplane; the ability to predict and verify performance parameters
is key to the whole aircraft industry. We have emphasised the fact that the field of
comprehensive analysis is not well developed (unlike rotorcraft engineering); the
amount of reference data is relatively modest, as reported in the flight manual and
in some type certificate documents. However, these data are often insufficient for an
accurate engineering analysis. In this book we have attempted to make advancements
in this area and have proposed several numerical methods, validation strategies and
sensitivity analyses that can improve the prediction of aircraft performance.
THE KNOWLEDGE BASE. As technology progresses, the knowledge base in aircraft

engineering is bound to increase – at least theoretically. The knowledge base is
interpreted as the ensemble of written communications and personal experience.
The written communications include papers, technical reports, technical specifica-
tions, drawings, graphs, databases, experimental data, flight data, regulations, engin-
eering standards and so on. All of this written communication has to be studied and
interpreted. Increasingly, the answer is given by the computer or the Internet. The
computer calculates, on the basis of a software system that is taken for granted, and
produces results that often are not critically assessed. However, decisions cannot
614
Epilogue 615
be delegated to computers, particularly when they involve human factors, safety,

compliance with regulations and large procurement contracts. With this in mind,
industrial and professional expertise is crucial; maintaining expertise over time has
become increasingly challenging.
IMPACT OF AVIATION. With regard to the commercial aspects of air transport, we

have hit various limits in terms of airport capacity, air traffic control, airport secur-
ity, consequences of local disruption, ground transport integration, and so on. In this
book, we have focused on the airplane itself, which is at the heart of this infrastruc-
ture. The airplane may be small in comparison to an international airport or the
sky, but it has become increasingly entangled in a complex web of ground services
expanding way beyond the runway; in the sky, there are enough airplanes to poten-
tially cause collisions on a regular basis. Thus, they cannot navigate on their own.
Traffic congestion is no longer a term used to denote ground vehicles.
AVIATION EMISSIONS FORECAST. Considerable research already exists on the effects

of aviation on atmospheric physics, as briefly mentioned in this book. A vast amount
of technical work is available in the field of aggregate aviation emissions* .
The emissions from commercial aviation are indeed high. However, their con-
tribution must be viewed in the larger context of emissions from other sources, in
particular the effects of deforestation, the loss of natural habitats, the increase in
world population and other threats of various magnitude. As we write this section,
the world’s population has reached the 7-billion mark; surely, it will continue to rise.
The effect of population increase alone (about 1.2% per year minimum since 1965)
leads to an increase of about 84 million people per year† . Each of these new citizens
of the world will need to use energy. If they were to emit only 1 ton of carbon-
dioxide per year (about 1/10 of the UK; 1/20 of the United States and Canada),
the total additional worldwide emissions would be about 84 million tons/year. This
effect would be compounded in the following year by an additional 85 million tons,
and so on. To this effect, we would need to add the loss of natural environment to
make room for more urban infrastructure. The total amount of direct carbon-dioxide
emissions from aviation is estimated at about 500 million tons per year; thus, the
effect of population increase in one year alone corresponds to one-sixth of these
emissions, at the conservative rates that we have assumed. At the current growth
rate, the population increase offsets any realistic cut in aviation emissions.
The “one CO2 -ton society” that has been suggested in some quarters will not be
possible if we keep flying. However, we cannot imagine a modern society without
air transportation. Furthermore, if we stop deforestation for one day only, we save
enough CO2 to transport about 4 million passengers from London to New York.
This estimate is based on average CO2 emissions of 810 million metric tons‡ per year
(based on 2000–2005 rates), and 550 kg CO2 /passenger over a 3,100-n-mile stage
length with a 394-seat Boeing B777-300.
* See, for example, EMEP/CORINAIR Emission Inventory Guidebook; available on the Internet,
periodically updated.
† Source: United Nations; World Bank.
‡ Harris NL, Brown S, Hagen SC, Saatchi SS, Petrova S, Salas W, Hansen MC, Potapov PV, and
Lotsch A. Baseline map of carbon emissions from deforestation in tropical regions. 336(6088):1573–
1576, 2012.
APPENDIX A
Gulfstream G-550
The Gulfstream G-550 is a long-range business/executive jet powered by two Rolls-

Royce BR710 turbofan engines. It has advanced cockpit avionics (see Figure 1.1) and
a variety of seating configurations. The G-550 is a transport-type airplane, certified
for day and night, extended over-water and polar navigation. A few examples of
calculations are shown throughout the book (see, for example, the SAR analysis in
§12.3.1 and the payload-range in § 15.2).
A summary of weights, capacity and limitations is given in Table A.1. A summary
of geometrical dimensions is given in Table A.2. The wing is swept back 27 degrees at
the quarter-chord location, with a 3-degree dihedral, and with a tip-mounted upper
winglet. The wing does not have a leading-edge slat. The main flight controls are
a single Fowler-type flap, ailerons and trim tabs. There are three spoilers per side.
There are some drag-reduction devices, such as vortex generators that wrap around
the winglet junction, vortex generators on the upper side of the wing and blunt
trailing edges. Furthermore, there are seals installed around rudder, elevator, and
thrust reverser and fairings around antennas. A summary of operational limitation
is given in Table A.3.
Power Plant
The engine is a two-shaft high-by-pass ratio Rolls-Royce BR710 C4-11 turbofan
engine with a single-stage low-pressure compressor, a ten-stage high-pressure com-
pressor, a two-stage high-pressure turbine and a two-stage low-pressure turbine. The
engine is equipped with FADEC. Basic data are shown in Table A.4. The engines
are mounted high in the aft fuselage through horizontal pylons.
Airplane Dimensions
Table A.2 shows the main dimensions of the airplane. Where data are indicated as
n.a., it means that they cannot be conventionally defined. For example, the vertical
stabiliser has several sides, and although a root chord could be defined, it is more
difficult to decide what is the leading-edge sweep.
617
618 Appendix A: Gulfstream G-550
Table A.1. Weights and capacities of the G-550
Weights & Capacity
Maximum take-off weight 41,277 kg

Maximum ramp weight 41,458 kg
Operating empty weight 20,960 kg
Maximum zero-fuel weight 27,721 kg
Maximum landing weight 34,155 kg
Maximum payload 2,812 kg
Maximum fuel capacity 18,734 kg
(1 left, 1 right tank)
Maximum allowable fuel asymmetry (take-off) 454 kg
Maximum allowable fuel asymmetry (flight) 908 kg
Seating capacity 14–18
The flight controls include the ailerons, Fowler flaps (one per side) and spoilers
(two flight spoilers per side and one ground spoiler per side). The tail surfaces consist
of a fixed-geometry vertical stabiliser with a rudder and a trimmable horizontal sta-
biliser with flight controls (right/left elevators). The elevators incorporate adjustable
trim tabs.
Table A.5 is a summary of landing-gear characteristics. The noise gear is steer-
able and the main landing gear has an anti-skid system. The main landing gear has
a main strut with a shock absorber, a structural post at the back, an axle fitting assy
Table A.2. Basic dimensions of the G-550

Parameter Value Unit Parameter Value Unit
Airplane Horizontal Tail Plane
Overall length 29.38 m Wing span 10.72 m

Overall height 7.90 m Wing sweep at L.E. 33.0 degs
Fuselage length 26.16 m Wing area 23.24 m2
Cabin length 15.27 m Aspect-ratio 4.96
Tail upsweep angle 13.8 degs Root chord 3.084 m
Fuselage height 2.44 m Taper ratio 0.456
Fuselage width 1.88 m Dihedral angle 0.3 degs
Main Wing Vertical Tail Plane
Wing span (reference) 28.48 m Height 3.58 m

Tip chord 1.620 m Wing sweep at L.E. n.a.
Root chord 6.130 m Wing area 15.1 m2
Taper ratio 0.265 Aspect-ratio 0.85
Wing sweep at L.E. 27.0 degs Root chord n.a. m
Wing sweep at Q.C. 30.5 degs Taper ratio n.a.
Wing area (reference) 90.89 m2
Mean aerodynamic chord 3.795 m
Dihedral angle 3.0 degs
Table A.3. Operational limits of the G550
Speeds
Long-range Mach number 0.80
Maximum operating Mach number 0.885
Minimum control speed, landing, S/L 110 KCAS
Minimum control speed, take-off, S/L 107 KCAS
Minimum control speed, air 112 KCAS
Tail wind limit, take-off 10 kt
Tail wind limit, landing 10 kt
Aerodynamic Controls
Maximum UP/DOWN aileron deflection 11 degs
Maximum UP/DOWN aileron trim tab deflection 15 degs
Maximum UP/DOWN elevators deflection n.a. degs
Maximum spoiler deflection
clean wing +30 degs
full aileron deflection +55 degs
Maximum right/left rudder deflection 22 degs
Maximum flap deflection 39 degs
Operating Limitations
Maximum cruise altitude (level) FL-510
AEO service ceiling 13,015 m
OEI service ceiling 7,870 m
Maximum approved take-off altitude 3,048 m
Maximum runway slope ±2 degs
Maximum payload 5,700 n-miles
3,000 lb (1,300 kg) payload 6,450 n-miles
1,600 lb (0.725 kg) payload 6,700 n-miles
ferry range 6,900 n-miles
BFL at MTOW, ISA, sea level 1,800 m
Final approach speed at MLW, sea level 105 kt
Maximum fuel temperature 54 ℃
Minimum fuel temperature (red alert) −37 ℃
Minimum turning radius 34.14 m
Table A.4. Selected data of the Rolls-Royce BR710

C4-11 gas turbine engine
Item Value Unit
Dry weight 1,818 kg

Overall length 4.660 m
Fan diameter 1.785 m
Number of fan blades 24
Maximum continuous thrust (S/L, ISA) 64.3 kN

Maximum take-off thrust (S/L, ISA) 68.4 kN
Turbine gas temperature 860
maximum continuous thrust 860 ℃
take-off (5 minutes) 900 ℃
By-pass ratio 4.2
Compressor ratio 24
Minimum fuel temperature −40 ℃
Maximum oil temperature +160 ℃
620 Appendix A: Gulfstream G-550
Table A.5. Landing gear of the G550
Dimensions
Number of wheels 6
Landing-gear groups 3
Main gear wheels 2×2
Nose gear wheels 6×1
Wheel base 13.72 m
Wheel track 4.37 m
Main gear tyre H35 × 11.0-18
Nose gear tyre 21 × 7.25-10
Main tyre pressure 12.8 bar
Nose tyre pressure 15.2 bar
Distance of nose gear from nose 2.1 m
Distance of main gear from nose 15.8 m
Total length of main struts 2.70 m
Average diameter of main struts 0.16 m
Tyre speed limit 195 kt
Fuse plugs on each wheel 4
Brake warning temperature 650 ℃
and a side-brace actuator. The nose landing gear has a main strut with a shock
absorber, an inclined drag brace, a downlock actuator and two head lamps.
Performance
Figure A.1 shows the flight envelope of the airplane, as elaborated from the FCOM.
This envelope is also discussed in § 8.6.2.
FL-510; M = 0.860
500 15
43.5 kft, M=0.885

400 MMO 12
Altitude, km
Flight Level
32.2 kft, M=0.885

300 9
27,8kft, M=0.85
O
VM
=
200 6
AS
KC
0
34
100 3
S
CA
0K
30
0 0
0.1 0.3 0.5 0.7 0.9
Mach number
Figure A.1. Flight envelope of the Gulfstream G550, standard day, no wind.
Appendix A: Gulfstream G-550 621
A.0.1 Geometry Model

The following table is an example of our control-points model for this aircraft. See
Chapter 2 for details.
"Gulfstream G550" airplane name

"1.0.0" airplane model version (can be updated!)
0.5d0 accuracy of data (percent, variable, ** estimated **)
"TOP" view
"wing" 5 notes
10.9137 1.1867 wing attachment (leading edge), wrt NOSE point
18.1895 13.4489 wing tip (leading edge)
19.8107 13.4489 wing tip (trailing edge)
16.9241 2.2151 wing break (trailing edge), close to fuselage
17.0428 1.1867 wing attachment (trailing edge)
"flaps" 4 COMPONENT, number of control points

16.9241 2.2547 generic points
16.3706 2.4524
18.5058 9.7306
18.8617 9.7306
"aileron" 4 COMPONENT, number of control points

18.8617 9.7700 generic points
18.1895 9.7700
18.9803 12.1435
19.4549 12.1435
"tailplane" 5 COMPONENT, number of control points

24.5558 0.1582 inboard leading edge
27.8378 5.1027
28.2728 5.3600
29.3405 5.3600
27.6401 0.0800 inboard trailing edge
.......
APPENDIX B
Certified Aircraft Noise Data
Table B.1 and Table B.2 summarise the certified noise performance of selected
aircraft, in various categories. A full database of certification data is published by
the Federal Aviation Administration and is periodically updated. The FAA database
provides noise levels of hundreds of aircraft configurations, including the effects of
different engine installation. Military aircraft are excluded. The tables also report
the MTOW, the by-pass ratio of the engines and the position of the flaps, both at
take-off (TO) and approach/landing (AP). The noise data are provided at take-off,
sideline (SL) and approach.
Table B.1. Certified noise levels for commercial aircraft
Flap Flap
Aircraft MTOW Engines BPR (TO) (AP) TO SL AP
ATR72-500 22.50 2 × PW-127M 15 25 79.0 80.7 92.3
A-300B4-622R 149.70 2 × PW-4158 4.85 0 40 88.0 98.3 101.3

A-310-324 150.00 2 × PW-4152 4.85 15 40 90.6 97.2 100.2
A-320-211 68.00 2 × CFM56-5A1 6.00 10 35 85.3 94.4 96.4
A-330-301 180.00 2 × CF6-80E1A2 5.05 14 32 87.0 97.9 98.5
A-330-321 230.00 2 × PW4164 4.85 8 32 95.6 97.5 98.0
A-340-312 220.00 4 × CFM56-5C3 6.60 17 32 88.0 95.8 97.3
A-340-312 270.00 4 × CFM56-5C3 6.60 17 32 96.2 95.3 97.2
A-380-842 391.00 4 × Trent 972 94.6 94.5 98.0

A-380-861 391.00 4 × GP7270 8.30 94.8 94.5 97.1
BAE 146-RJ100 46.00 4 × LF 507-1F 5.10 18 33 86.1 88.1 97.6
B-717-200 64.58 2 × BR700-715A1 4.66 5 40 84.0 89.0 91.6

B-737-300 56.47 2 × CFM56-3 5.00 1 40 82.4 89.7 98.5
B-737-300 63.28 2 × CFM56-3 5.00 1 40 83.9 90.9 97.6
622
Appendix B: Aircraft Noise Data 623
Table B.2. Certified noise levels for commercial aircraft (part 2)
Flap Flap
Aircraft MTOW Engines BPR (TO) (AP) TO SL AP
B-737-500 49.00 2 × CFM56-3 5.00 5 40 81.0 89.3 98.4

B-737-500 63.05 2 × CFM56-3-B1 5.00 5 40 87.3 90.0 100.0
B-737-700 60.33 2 × CFM56-7B22 5.40 1 40 82.6 92.5 95.8
B-737-700 70.08 2 × CFM56-7B26 5.10 1 40 84.6 94.7 95.9
B-737-700/IGW 72.12 2 × CFM56-7B24 5.30 1 40 86.6 92.9 96.1
B-737-700/IGW 77.57 2 × CFM56-7B27 5.10 1 40 86.6 95.2 96.1
B-737-800 79.01 2 × CFM56-7B27 5.10 1 40 87.0 94.7 96.5
B-737-900 74.39 2 × CFM56-7B24 5.30 1 40 86.6 92.0 96.4
B-737-900 79.01 2 × CFM56-7B27 5.10 1 40 86.7 94.2 96.4
B-747-100 322.06 4 × JT9D-3A 4 5.10 10 30 105.4 102.1 104.6
B-747-100 332.94 4 × JT9D-3A (*) 5.10 10 30 109.4 99.6 107.2
B-747-400 394.60 4 × CF6-80C2B1F 5.20 25 99.7 98.3 101.4
B-747-400 394.60 4 × PW 4056 4.80 10 30 101.5 99.7 104.7
B-747-400 396.90 4 × RB211-524G 4.30 10 30 99.2 98.0 103.8
B-757-300 124.74 2 × RB211-535E4B 4.10 5 30 88.4 94.8 95.4
B-757-300 107.05 2 × RB211-535-E4 4.10 5 30 84.8 93.9 95.2
B-767-200 160.06 2 × CF6-80C2-B2 5.00 1 30 89.5 93.7 96.4
B-767-200 175.54 2 × CF6-80C2-B4 5.00 1 30 90.6 95.0 96.4
B-767-200/ER 136.08 2 × CF6-80C2B2F 5.00 1 30 85.1 93.8 95.8
B-767-300 184.61 2 × PW 4056 4.80 5 30 94.2 95.7 100.2
APPENDIX C
Options for the FLIGHT Program
The key performance options of the FLIGHT program described in this book are
given here. After loading the airplane, the following user-menus are available:
1. Performance Charts
(a) Aerodynamics (Chapter 4)
(b) Specific Air Range (Chapter 12)
(c) Engine Charts (Chapter 5)
(d) Flight Envelopes (Chapters 8 and 13)
(e) Propeller (Chapter 6)
(f) WAT (AEO Take-Off) (Chapter 9)
(g) Balanced Field Length (Chapter 9)
(h) Payload-Range (Chapter 15)
(i) Economic Mach Number (Chapter 12)
(j) CG Effects (Chapter 12)
(k) Buffet Boundary (Chapter 4)
(l) Specific Excess Power (Chapter 13)
(m) Jet Thermo Charts (Chapter 14)
(n) Atm-Speed Charts (Chapter 8)
(o) Holding Speeds (Chapter 15)
(p) V-n diagram (Chapter 13)
2. Mission Analysis (Chapter 15)
(a) Fuel Planning
(b) Aircraft Range
(c) Matrix-Fuel-Plan
(d) Equal-Time Point
3. Aircraft Noise (Chapters 16–18)
(a) Take-Off at FAR/ICAO Point
(b) Landing at FAR/ICAO Point
(c) Sideline at FAR/ICAO Point
(d) Arbitrary Trajectory
(e) Noise Footprints
(f) Stacking Patterns
624
Appendix C: FLIGHT 625
4. Exhaust Emissions
(a) Exhaust Emissions versus Range (Chapter 15)
(b) Contrail Analysis (Chapter 19)
5. Flight Optimisation
(a) Minimum Climb-Fuel (Chapter 10)
(b) Optimum Climb between Flight Levels (Chapter 12)
(c) Fuel-Tankering Analysis (Chapter 15)
6. Manoeuvre Analysis (Chapter 13)
(a) Landing in Downburst
(b) Take-Off in Downburst
7. Trim Analysis
(a) Minimum Control Speed on Ground, VMCG (Chapter 9)
(b) Minimum Control Speed in Air, VMCA (Chapter 7)
8. Direct Operating Costs (Chapter 15)
Each sub-option requires entering of a set of operational data. All of the geo-
metry, mass and inertia properties (Chapter 2 and Chapter 3) are given automatically
on output.
The propeller option points to another program, with user-input as herein. After
loading the propeller model, the following sub-menus are available:
1. Propeller Geometry
2. Performance at Design Point
3. Performance in Oblique Flight
4. Performance Charts
5. Propeller Design
6. Propeller Trim
7. Propeller Noise
8. Performance of Ducted Propeller
Index
Note: “t” after page number denotes a table; “f” after page number denotes a figure
wing, 375 climb, 279f

α-floor protection, 54, 57, 234 CO2 emissions, 604, 605f
λ–shock, 219 continuous descent, 308t
g–load, 362, 374 descent, 308f
331-600 APU, 148 elevator deflection, 185f
flap angles, 305t
Accelerate-go, 240 flight envelope, 213f
Accelerate-stop, 227 flight in downburst, 386
Acoustic power, 486 fuel tanks, 62t
Acronyms, xxv–xxviii heat release, 599
Admittance, 538, 539 holding, 315
Advance ratio, 154 L-gear noise, 516f
Aerodynamic derivatives, 114, 191 landing, 320f
Aerodynamic heating, 294 LTO emissions, 603t
After-burning, 127, 143 MAC, 35t, 36f
Agility, 360 mass components, 73t
Aileron, 376, 379 mass properties, 73t
authority, 376 noise guarantee, 465
effectiveness, 377 noise stacks, 582
yaw control, 187, 190 noise trajectory, 568f, 573–575f
Air conditioning, 147, 211 perfect flight, 607–608
Air traffic control, see ATC, 332 tankering, 445f, 447f
Airbus, 263, 303, 316, 322f trim drag, 185f
A300, 25f, 134, 303, 349t, 440 weight versions, 52
Beluga, 47t A330, 349t
CG asymmetry, 186 fuel tanks, 62t
cross-sectional area, 31f gliding range, 312
deterioration drag, 102 A340, 349t, 460
fly-over, 512 CG limits, 58f, 59f
fuel temperature, 401f CG position, 56f
ground effect, 82f contrail experiment, 597
tyre temperature, 417 fuel tanks, 62f, 62t
wetted area, 33t, 38t mass properties, 73t
wing lift, 80f A380, 7, 67, 425
A319 altitude flexibility, 594
fuel tanks, 62t APU, 148
go-around, 323 downwash, 117
mass properties, 73t drag, 104
A320, 27, 52–53, 349t drag sensitivity, 105t
cabin floor, 71f ground effect, 82f
627
628 Index
Airbus (cont.) Argon, 487t

ground manoeuvre, 262f Armstrong’s line, 211
jet blast, 418 ATC, 315, 329, 332, 584, 598, 606,
mass properties, 73t 609
span constraint, 47 ATR72, 173f, 338
taxi-out fuel, 261t drag polar, 104
water load, 51 icing, 393
wetted area, 33t, 38t mass properties, 73t
wing-body, 28f noise footprint, 579f
winglet, 29 noise trajectories, 571f, 570–572
A400, 47t payload-range, 430, 431f
propellers, 157t SAR charts, 338f
payload-range, 427f take-off, 244
Aircraft accidents Available-seat-km (ASK), 451
Airbus A330, 3 Aviation fuel, 397
B777 Heathrow, 400 Aviation records
bird strike, 262 Ily’a Murometz (1914), 45
tyre explosion, 409 Axial momentum theory, 157
volcanic ash, 138
Aircraft size, 33, 45 BAE RJ 146, 311, 561
Aircraft state, 564 Bank angle
Airfoils turn, 361
icing, 395f with thrust asymmetry, 186
NACA 0012, 108 Banked level turn, 361
NACA 00XX, 79 Barred owl, 584
NACA 23012, 93 Bernoulli equation, 157
NACA 6 series, 93, 107 Bessel functions, 505, 519, 546
perimeter, 23, 23t Bird strike, 262–263
SC-1095, 107 Black box, see FDR
Airlines, 2, 423, 425, 449, 594, 608 Block fuel, 329, 340
Airport, 419, 555, 601 Block time, 425
Gatwick, 554 Blumer-van Driest theory, 91
Heathrow, see Heathrow Boeing, 399
London City, 561 B727, 147, 554
New York JFK, 448 tyre temperature, 416
noise restriction, 554f B737, 554
Schwerin-Parchim, 575 cabin pressure, 211
Airworthiness, 4 L-gear noise, 516f
Airy-Fock functions, 545 landing gear, 516
Alenia C-27J, 156 mass properties, 73t
Altitude stabiliser, 22f
flexibility, 593, 598 wing, 20
Altitude effects wing-body, 28f
on atmosphere, 198f B737-900
on SAR, 337f, 337 winglet, 29
ANSI B747, 6
noise absorption, 535–537 -100 stability derivatives, 190
noise metrics, 472t control speed, 191f
Antonov AN-225, 47t cross-sectional area, 32f
Appleman-Schmidt criterion, 595 drag, 103f
APU, 71, 147–149, 434, 562, 567 equal-time point, 448
emissions, 148t forebody, 26
fuel flow, 148f jet blast, 418
in fuel planning, 435 lateral control, 191f
mass, 71 mass properties, 73t
noise, 475, 508 mission fuel, 439f
Aquaplaning, 101, 249f, 321 payload-range, 427f
Area range, 428f
cross-section, 23, 30 taxi-out fuel, 261
reference, 20 ULD, 47
wetted, 24–29 vortex wake, 595
Index 629
WAT charts, 320f range sensitivity, 429

wetted area, 33t, 38t spherically blunted ogive drag, 112
B747, ramp weight, 49 Turboprop PW127M, 143
B777, 453, 507, 563 wetted areas, 36
-200-IGW, 50 Case study: Do aircraft float?, 32
CO2 emissions, 604f CASK, 453
flight levels, 348 Ceiling
fuel tanks, 33 absolute, 208, 210
fuel temperature, 408f operational, 213
ground effect, 82f service, 211
mass properties, 73t Centre of gravity, 51, 68, 189
mission sensitivity, 604 envelope, 53
SAR effects, 353 in flight, 55
tyre temperature, 416 percent of MAC, 35
wetted area, 33t, 38t range, 352
C-17 Globemaster, 47t travel, 54
MD-90, 507 Certificate
Boeing projects, 46 airworthiness, 4
Bombardier Dash8-Q400, 73t, 156, noise, 553
171 safety, 381
Brakes, 66, 73t, 249 Cessna 180, 259
Braking, 248, 321 CF6-80C2, 138, 139f
Breguet range equation, 346 engine data, 135t
Broadband noise, 480, 485, 515, 521 CFM56, 134f, 137, 323, 607
Buffet emission indexes, 602f
boundaries, 113 Chapman-Rubesin factor, 88
speed, 113 Cirrus clouds, 591
Busemann biplane, 116 Climb
Buzz-saw, 483 continuous, 607
en-route, 169
Cabin pressure, 210, 211, 214 ICAO A profile, 273
CAD formats, 17f, 17, 254 ICAO B profile, 274
Calibrated Air Speed, 205 ICAO Standard profile, 273
Caravelle, 409 minimum-fuel, 296
Carbon-dioxide, CO2 , 487t minimum-time, 296
Carbon-monoxide, CO, 601 polar diagram, 292
Cargo load, 52 supersonic, 294
Cargo performance index, 47 Climb angle, 278
Case study Clouds, 138, 200, 380, 591
AEO take-off, 234 Cockpit, 4
aircraft drag, 103 Cold soak, 396
ATR72 payload-range, 430 Combination tone noise, 481
ATR72, SAR, 338 Combustor temperature, 138
CG effects on SAR, 353 Computer models
Climb to specified Mach, 293 configuration change at climb, 276
CO2 emissions, 604 accelerate-stop, 240
drag of Airbus A380-861, 104 equal-time point, 448
drag of ATR72-500, 104 fuel tankering, 446
F4 wind-tunnel model, 102 laminar skin-friction drag, 89
F568 propeller, 165 landing-gear drag, 99
flight envelopes, 215 propeller integration, 168–170
fuel planning, 455 propeller trim at take-off, 242
General Electric CF6-80C2, 134 stochastic area, 22
go-around, A319, 323 turbulent skin-friction drag, 91
Gulfstream G550, SAR, 335 Concorde, 127, 218, 554
heat released in atmosphere, 599 Contamination, 250, 321
Honeywell RE-220 APU, 149 Contrail, 589–591
moments of inertia, 73 factor, 595
noise trajectories, A320, 568 Control points, 17, 621
noise trajectories, ATR72, 570 Coriolis acceleration, 37
payload-range, float-plane, 460 Corner velocity, 366
630 Index
Cost index, 351–352 trim, 183

Crab landing, 320, 321 wave, 93–95, 116
Cruise wheels, 97
– climb, 343 wing, 88
in fuel planning, 435, 436
propeller aircraft, 346–347 EASA, 508
subsonic, 341–346 Eckert reference temperature, 88
supersonic, 353–355 Efficiency, 153
Cruise-climb, 607 propeller, 153
CTOL, 224 propulsive, 596
Elevator, 73t, 182, 183
Data handling Emission angle, 477–478
engine, 140 Emissions, 615
noise, 565 CO2 , 604
Decelerate-stop, 239 LTO, 600–604
decibel, 471 Emissions Trading, 608
Density altitude, 197 En-route stop, 443
Derating, 129, 248 Endurance, 329, 341
Descent Energy
continuous, 307–308, 607 acoustic, 471
en-route, 170 impact on landing, 263, 317
steep, 308–311 method, 285–288
unpowered, 311 potential, 287
Dihedral, 21, 25, 95, 322 Energy height, 287
Dihedral effect, 186 Engine
Direct Operating Costs, 353 contamination, 138
Directivity, 480, 481f, 481, 483, 485t, 488f, 488, design point, 133, 134f
497, 498, 510, 514, 515, 517f, 521, 522 failure, 186
Discrete-tone noise, 480 installation effects, 137
Displacement thickness, 522 rubber, 137
Distrails, 592 spool-up, 306
Diversion distance, 440 state, 140
DOC, 448–453 strike, 321
Doppler effect, 483, 484, 488, 520, 565 surge, 220
Douglas DC-10, 87 wash cycle, 140
Douglas DC9-80, 113, 114 Engine strike, 321
Douhet, Giulio, 7 EPNL, 472, 472t
Downwash, 116, 181, 185, 383, 584 Equivalent Air Speed, EAS, 205
Dowty propeller, 156, 157t Equivalent all-out range, 440
R381, 157t Equivalent continuous sound level, 561
R391, 157t Equivalent Noise Level, 475
R408, 157t Equivalent Source Method, 533
Drag, 85 Estimated time of arrival, 447
body of revolution, 110 ETOPS, 147, 439, 463
cavity bays, 98 Exhaust-gas temperature, 127, 129, 135t, 508, 509
contamination, 100, 251
displacement, 100 F4 wing-body, 35, 37t
fuselage, 92 F568 propeller, 157t
fuselage upsweep, 93 altitude effects, 167f
hull, 115 disk loads, 164f
idle engine, 190 geometry, 165f
impingement, 253 shaft power, 167f
interference, 94 taxi conditions, 169f
landing gear, 96 Faddeeva function, 540
landing-gear struts, 98 FADEC, 127, 135, 171, 617
lift-induced, 85 Fastest climb, 291
profile, 87, 99 FCOM, 3, 50, 214, 235, 236, 319, 335, 418, 464, 561
sensitivity analysis, 104 FDR, 2–3, 400, 407, 408f, 408, 564
spoilers, 95 Figure of merit, 331, 337
spray, 101, 254 Fineness, 94, 112
Index 631
Flaps, 73t, 81, 83 cross-section, 30–31

Fowler, 617 forebody, 25
Kruger, 30 inertia, 69
racks, 30 mass, 65
Flare, 257, 321 tail, 26
Fleet, 52 wetted area, 25
FLIGHT code, xxiv, xxix, 73, 166, 227, 335, 369,
371, 386, 444, 446, 448, 572, 575, 582 Gas turbine, 126
Flight Data Recorder, see FDR Gavotti, Giulio, 7
Flight envelope, 211–215 Gearbox, 143, 165, 171, 175
supersonic jet, 218 Global Forecast System, 201
Flight levels, 332, 332, 347, 348f, 348 Go-around, 323
Flight management system, see FMS thrust recovery, 131f
FLIGHT program, 624–625 GP-7000 engine, 127, 443
Flight stacks, 315 GPS coordinates, 564
Flight testing, 2 Grashof number, 404
Float-plane, 115, 259 Gravity suit, 374
payload-range, 460 Green dot speed, 277, 303, 304, 315
Flowchart Green function, 519
aerodynamic drag, 86f Ground effect, 84, 256
aircraft systems, 18f Ground operations, 260
CG calculation, 64f Ground reflection, 538
engine model, 131f Grumman F-104, 189
environmental challenges, 610f Gulfstream G550, 215, 617–621
FLIGHT code, 11f, 14f APU, 149
mass distribution, 64f cabin pressure, 211
mission fuel, 432f cockpit, 4
noise sources, 476f figure of merit, 337
propeller code, 12f flight envelope, 215f, 620f
propeller simulation, 155f ground effect, 82f
turbofan engine noise, 479f landing gear, 618
FMS, 302, 305, 352 mass properties, 73t
Focke Fw-190 (Wulf Condor), 6, 379 operational limits, 619t
Fokker payload-range, 428–429
Anton, 8
F-100, 259 Hague Convention, 7
F-50, 259 Hamilton-Sundstrand, 143, 165, 166f, 167f
F27, 157t Hankel functions, 505
F50, 157t Hartzell propellers, 157t
Fuel, 397 Heathrow airport, 1, 315, 448, 554, 555, 557t, 606
contingency, 440 High-lift devices, 30
cost, 4, 433 History
density, 399 aircraft icing, 392
diversion distance, 440 military aircraft, 7–8
heat capacity, 399 Holding, 439, 440
jettisoned, 50 Humidity, 196t, 426, 557, 595, 597
loiter, 440 Hydrocarbons, HC, 601
mission, 444 Hydrogen propulsion, 591
planning, 426 Hydroplaning, see Aquaplaning
price, 328 Hysteresis, 412
temperature, 398
temperature in flight, 400 IATA, 328
undrainable, 51 ICAO, 196, 316, 439
unusable, 440 atmosphere, 197
vector, 54 climb profiles, 273–274
Fuel tanks, 33, 61, 72 databank, 601, 602f
use, 63t, 353 environment, 594
Furnishings, 67 ISA, 200f
mass, 67, 71 LTO emissions, 600, 601
Fuselage LTO flight modes, 602t
632 Index
ICAO (cont.) Constellation, 210

noise, 510, 572 SR-71, 210
noise certification, 555, 556f, 558 Loiter, 315, 440
pavement, 50
unit of risk, 463 MAC, see mean aerodynamic chord
wake separation, 118t, 584 Mach number
Iceland volcanic eruption, 138 design dive, 65, 207
Icing, 392, 394 economic, 350–352
propeller, 165 long range, 334
Ideal gas equation, 196, 203 maximum range, 334
Ily’a Murometz, S-27, 45 structural design, 206
Immelmann, Max, 8 tip, 152
Impact pressure, 204 McDonnell-Douglas
Impedance, 535, 538, 539, 541 F-15, 270
Indicated Air Speed, 205 F-15E, 218
Inertial coupling, 374 F-4, 270, 296, 329
Inertial particle separator, 143 Mean aerodynamic chord, 34, 180
Initial value problem, 278 Messerschmidt Me-262, 285
Inlet, 126, 132, 480, 482, 486, 487, 508 Methane, 600
INM program, 564 Microburst, 380
Installation effects, 137 Military Standards, 199
Intake, see Inlet Mission
Intake buzz, 218 failure, 4
Interrupter gear, 8 planning, 424
Italian Air Force, 7 reserve fuel, 438
Molecular mass, 487t
Jet Moments of inertia, 68–72
co-axial, 491 MTOW, 3, 48
Fisher model, 491–494 versus payload, 47
fully mixed, 492 versus wing-span, 46
plug, 496
Stone model, 494–500 Nacelles, 28–29
Jet blast, 418 NADP, 274
Jet shielding, 501–507 Nitrogen oxides, NOx , 601
Jet stream, 200, 338 Nitrogen, N2 , 487t
Jet-A fuels, 398, 398t, 399, 400, 404, 408 Noise
absorption, 535
KLOC, xxiv certification, 553, 555
certified levels, 622t, 623t
Landing data handling, 565
flare, 321 footprint, 578, 580
instrument, 300, 303 ground reflection, 538
Landing gear, 96–100 microphones, 557
Gulfstream G550, 620t night curfew, 556
installation, 99 shielding, 533
mass, 66 tax, 559
noise, 512 Noise directivity, see Directivity
Lift, 78 Normal load factor, 271, 276, 293, 361, 372
augmentation, 81 1g flight, 362
curve slope, 83, 106, 107, 108 turning, 369
ground effect, 79 Noy, 473
spoilers, 95 Nozzle, 126, 127, 135, 142, 190, 501, 503, 504
wing, 79 APU, 147, 509
Lockheed, 45 fuel, 143, 144
382J, 156 NTSB, 409
C-130, 6 Nuclear power, 45
C-130J, 47t Nusselt number, 404, 413, 414
C-130J with floats, 461f
C-5A, 189 Ogive, 110–112
C-5B, 47t Optical thickness, 590
Index 633
Optimisation, 4, 278, 347 Redispatch, 441

Owl technology, 584 Reference system, 18, 37
Oxygen, 487t noise trajectory, 564
Reflection-only method, 534
Passengers, 45, 52, 54, 57, 61, 67, 72, 431 Republic R-4, 379f
Payload-range, 426–429 Riemann integral, 116
calculation, 430 Risk analysis, 463–465
float-plane, 460–463 Roll, 374–379
Peng-Robinson equation, 404 Roll rates, 379
Perfect flight, 606–608 Rolling
Phugoid, 313 coefficient, 255, 255t
Piaggio Avanti, 157t, 174 damping, 377
Pitching moment, 60, 181, 229, 234 moment, 186, 375
Pitot probe, 203 Rolls-Royce
Point performance, 330 AE-2100DE engine, 6
Polar diagram, 292 BR710 engine, 617, 619t
Power, 130, 137 Olympus-593, 127
APU, 147 Rubber engine, 137
coefficient, 154 Rudder, 73t, 187–191
ratings, 128–129 Runge-Kutta methods, 257, 407
Predictor-corrector, 437 Runway, 50, 101, 261, 317, 321
Pressure conditions, 250, 321t
acoustic, 471 temperature, 414
impact, 204
stagnation, 203 Saab-2000, 157t
standard atmosphere, 196t Safety, 277, 320, 456, 463
Pressure altitude, 202 SAR, 330
Profile drag, see Drag ATR72-500, 338
Propeller Gulfstream G550, 335–337
247F, 157t numerical solution, 332–334
advance ratio, 153, 154 risk analysis, 464
coefficients, 154 SAT, see static air temperature
Dowty R391, 156 Scale effects, 46
efficiency, 153 Scheduling, 424, 425
electronic control, 156, 165 Sears-Haack body, 26f, 26, 111
F568, 165–168 SEL, see sound exposure level
failure, 244 Sensitivity analysis
installation effects, 173 drag, 104
intake, 173 mission parameters, 604
mass, 71 noise, 567
pitch, 152, 165 noise components, 501
wing interference, 173 range, 429
Pull-out, 293 Shimmy dampers, 238
Pull-up, 293 Shock absorbers, 236
PW F-100, 143, 146f Side-slip, 40, 186, 375
PW127M, 60, 143, 244 Sikorsky, Igor, 45
residual thrust, 171f Skin friction, 88, 91
PW980A APU, 148 coefficient, 88, 90, 92
Pylons, 28–29 laminar, 88
turbulent, 90
Radar, 263 Slats, 73t, 83, 102, 276
Radiative forcing, 599–600 retraction, 283
Radiosonde, 199 setting, 276
RAF, 7 Slush, 100, 232, 250, 253
Raleigh equation, 205 Software, xxiii, 4, 127, 171
Ram compression, 127 Solidity, 153
Range equation, 339 Sound
Ray tracing, 534 hearing threshold, 471
Rayleigh number, 404 intensity, 471
RE-220 APU, 149 loudness, 471
634 Index
Sound exposure level, 475 asymmetry, 186

Sound pressure level, 471 coefficient, 154
Specific excess power, 270, 287 cut-back, 555
charts, 288 reversal, 318
differential, 290 reversers, 250
load-factor effect, 364 Time
Specific excess thrust, 270 equal-time point, 446
Specific range no return, 447
jet aircraft, 330 noise stacks, 582
supersonic, 330 Time delay
Speed noise, 566
dash, 216 take-off, 239
design dive, 207 Time-on-station, 424
design flap, 208 Time-zone effect, 424
design manoeuvre, 207 Tip
ground, 40, 203 losses, 162
maximum gust intensity, 207 Mach number, 154
minimum control, 189 speed, 154
minimum drag, 209 vortex, 116
minimum power, 208 TOGA, 277
never-to-exceed, 207 Torque, 154, 155
sinking, 311 coefficient, 154
stall, 213 Total air temperature, 401
Spitfire, 379 Trajectory, 309
Spoilers, 73t, 95, 249, 311 continuous descent, 307
Spray, 101, 252f curvature, 380
Square-cube law, 46 final approach, 303
Stability derivatives, 189 NADP, 560, 561
Stall, 379, 380 optimal, 607
line, 213 perfect flight, 606
warning, 256 steep descent, 308
Stall speed, 255, 290, 294, 364 Transition altitude, 332
Standard day, 195 Transonics, 105, 108
Static air temperature, 401 Trim
Statistics, 20, 423, 426, 451, 464–465 drag, 183
Steep descent, 308–311, 561, 572 longitudinal, 179
Steepest climb, 272 propeller, 168
Stochastic analysis, 21 stick-fixed, 183
Stratosphere, 196 stick-free, 184
Strouhal number, 491, 493, 496, 498, 510, 513, 514 tank, 352
Supercooling, 394 with landing gear, 182
Supersonic acceleration, 217–218 Trimmable H-stabiliser, 57, 66, 184, 185, 618
Sutherland’s law, 198 Tropopause, 213, 288
Synchrophasing, 165 Troposphere, 196
Turbofan, 127
Tail strike, 57, 257, 258f Turbojet, 126, 143
Take-off Turboprop, 127, 141
CG effects, 236 Turbulence, 117, 254, 510, 541, 542, 576, 584
risk analysis, 465 Turn radius, 362
shock absorbers, 236–238 Turn rate, 365
time, 256 Turn-around time, 424
Tank, Kurt, 6 Type certificate, 5, 128, 130, 134, 137, 428, 509
Tankering, 444–446 APU, 508
Tanks, see Fuel tanks propeller, 156
TAT, see total air temperature Tyre, 101
Taxiing, 168, 261–262 cornering force, 246
fuel, 261t dimensions, 410
Temperature inversion, 198, 396 explosion, 409
Thermal conductivity, 399, 405, 414 heating, 409–416
Thrust hysteresis, 412
Index 635
inflation gas, 410 maximum structural payload —, 49

load-speed charts, 410 maximum take-off —, 48
material composition, 415 maximum taxi —, 49
slip ratio, 249, 415 maximum zero-fuel, 48
standing wave, 412 operating empty —, 48
useful load, 49
U-turn, 139, 367, 448, 607 Wells, H.G., 7
Under-relaxation, 162, 437 Wetted area, 24, 88, 460
Unit load device (ULD), 50 breakdown, 33t
fuselage, 25
V-n diagram, 372–373 nacelles, 28
Vectored thrust, 278 pylons, 28
Velocity potential, 506, 538 shell, 65
Vincenty inverse distance, 565 wing, 24
VMCA, 189 Wheel lock, 249
VMCG, 245, 248 Wind, 203, 321
Volcanic ash, 138, 590 at landing, 321t
Volume, 31 downburst, 380
cargo, 31 effect on noise, 543, 573–574
fuel tanks, 33–34 effect on propeller, 156t
scaling, 33 effects on payload-range, 428
von Kármán ogive, 111 effects on SAR, 338
von Kármán, Theodore, 7 Wind shear, 199, 380
Vought F8, 211, 218 Wind tunnel, 106, 107, 108
Wing, 20, 88
Wakes, 116–118 area, 20
WAT charts, 235 buffet, 113
Water vapour, 198, 394, 487t, 600 dropping, 379
Wave drag flutter, 220
airfoil, 105–108 fuel tanks, 33, 401
forebody, 94 geometry, 20–21
ogive, 110 loading, 46
wing, 93 sections, 23
Weather, 199 span, 20, 20
global circulation, 200 stall, 375
icing, 199 strike, 322
turbulence, 199 tanks, 61
Weight Winglets, 29, 117
limits, 45 Wright Brothers, 263
management, 51–52
passenger, 52 Yaw, 40
Weight effects adverse, 374
on control speed, 191f angle, 39
Weighting, 51 attitude, 39
Weights in roll, 186
all up —, 50 propeller, 163
manufacturer’s empty —, 48
maximum brake-release —, 49 Zeppelin airships, 7
maximum landing —, 49 Zoom climb, 210, 270
maximum ramp —, 49 Zoom dive, 217, 290, 292, 294, 295, 314

Advanced Aircraft Flight Performances

Uploaded by

Copyright:

Available Formats

Advanced Aircraft Flight Performances

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

Advanced Aircraft Flight Performances

Uploaded by

Copyright:

Available Formats

ADVANCED AIRCRAFT FLIGHT PERFORMANCE

This book deals with aircraft flight performance. It focuses on commer-

Dr. Antonio Filippone’s expertise is in the fields of computational and

1. J. M. Rolfe and K. J. Staples (eds.): Flight Simulation

© Antonio Filippone 2012

This publication is in copyright. Subject to statutory exception

First published 2012

Printed in the United States of America

Library of Congress Cataloging in Publication Data

ISBN 978-1-107-02400-7 Hardback

Cambridge University Press has no responsibility for the persistence or accuracy of

Tables page xvii

2.7 Geometry Model Verification 35

3 Weight and Balance Performance . . . . . . . . . . . . . . . . . . . . . . . . . 45

4.2.9 Case Study: Aerodynamics of the F4 Wind-Tunnel Model 102

5 Engine Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126

6 Propeller Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152

6.3 Flight Mechanics Integration 168

7 Airplane Trim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179

8 Flight Envelopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

9 Take-Off and Field Performance . . . . . . . . . . . . . . . . . . . . . . . . . 224

9.3 Solution of the Take-Off Equations 232

10 Climb Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269

10.6.4 Case Study: Climb to Specified Mach Number 293

11 Descent and Landing Performance . . . . . . . . . . . . . . . . . . . . . . . . 300

12 Cruise Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328

12.6 Cruise Range of Propeller Aircraft 346

13 Manoeuvre Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360

14 Thermo-Structural Performance . . . . . . . . . . . . . . . . . . . . . . . . . 392

15 Mission Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423

16 Aircraft Noise: Noise Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . 470

16.4.6 Co-Axial Jet Noise 491

17 Aircraft Noise: Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533

18 Aircraft Noise: Flight Trajectories . . . . . . . . . . . . . . . . . . . . . . . . 553

19 Environmental Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589

Appendix A: Gulfstream G-550 . . . . . . . . . . . . . . . . . . . . . . . . . . 617

Appendix B: Certified Aircraft Noise Data . . . . . . . . . . . . . . . . . . . 622

Appendix C: Options for the FLIGHT Program . . . . . . . . . . . . . . . 624

2.1 Cross-sectional areas of selected supercritical wing sections page 23

8.1 Sea-level data of the International Standard Atmosphere 196

This book is a derivative of an earlier textbook on flight performance. This new

AAIB Air Accidents Investigation Branch, United Kingdom (www.aaib.gov.uk)

Acronyms Used in This Book

APU Auxiliary Power Unit

KTAS True Air Speed in knots

TOW Take-off Weight

The U.S. Department of Defense and NATO publish a dictionary of acronyms

r verification that an aircraft achieves its design targets

r aircraft noise emissions

1.1 Performance Parameters

and state parameters. An additional set of 30 parameters is recommended though

1.2 Flight Optimisation

1.3 Certificate of Airworthiness

1.4 The Need for Upgrading

1.5 Military Aircraft Requirements

1.6 Review of Comprehensive Performance Programs

ESDU 32 provides a suite of programs to calculate the performance of fixed-wing

1.7 The Scope of This Book

r trajectory optimisation and route planning

Figure 1.2. Flowchart of a comprehensive flight performance simulation code.

Awetnose 0.72π lnose d. (2.9)