Ernst Heinrich Hirschel

Arthur Rizzi
Christian Breitsamter
Werner Staudacher

Basic Principles
and Unit Problems
Separated and Vortical Flow in Aircraft Wing
Aerodynamics
Ernst Heinrich Hirschel Arthur Rizzi
• •

Christian Breitsamter Werner Staudacher

•

Separated and Vortical Flow

in Aircraft Wing
Aerodynamics
Basic Principles and Unit Problems

123
Ernst Heinrich Hirschel Arthur Rizzi
Institute of Aerodynamics Department of Aeronautical
and Gasdynamics and Vehicle Engineering
University Stuttgart Royal Institute of Technology
Zorneding, Germany Stockholm, Sweden

Christian Breitsamter Werner Staudacher

Chair of Aerodynamics Institute of Flightsystems
and Fluid Mechanics Bundeswehr University Munich
Technical University of Munich Zorneding, Germany
Munich, Germany

ISBN 978-3-662-61326-9 ISBN 978-3-662-61328-3 (eBook)

https://doi.org/10.1007/978-3-662-61328-3
© Springer-Verlag GmbH Germany, part of Springer Nature 2021
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer-Verlag GmbH, DE part of
Springer Nature.
The registered company address is: Heidelberger Platz 3, 14197 Berlin, Germany
Preface

This book is devoted to a presentation and discussion of the fluid mechanical

aspects of separated and vortical flows in aircraft wing aerodynamics. The main
focus is on large aspect-ratio wings and on small aspect-ratio delta-type wings. In
general, we do not deal with aerodynamic design issues of these wings.
Our aim is to foster an understanding of the basic properties of the wing vortex
systems and their behavior. We further want to demonstrate the capabilities of
numerical simulation methods for such flows and the interpretation of the results
from a fluid-mechanical point of view.
The economical and ecological pressures on, in particular, transport aircraft are
progressively increasing. Military aircraft, on the other hand, faces large challenges
with regard to agility and maneuverability, but also to economical operation.
This all requires a refinement of the aerodynamic design and consequently an
increasingly better handling of separation and vortical flow phenomena. We see, on
the other hand, that since about two decades disciplinary and multi-disciplinary
discrete numerical simulation methods play a progressively larger role in all aircraft
design and development processes. Their potential is very large, but to realize it
fully, a very good understanding of the flow phenomena, that govern the aerody-
namic properties of an aircraft, is necessary. A tool is only as good as the person,
who selects and wields it.
Accordingly, we consider in the introduction the different mathematical models
that underlie the aerodynamic computation methods from the linear panel method
up to the most sophisticated Reynolds-averaged Navier–Stokes methods and the
recently emerging scale-resolving methods. Particular methods are the discrete
Euler methods, which as rather inexpensive methods take into account compress-
ibility effects and without an explicit Kutta condition also permit to describe
lifting-wing flow.
The concept of the kinematically active and inactive local vorticity content of a
shear layer is introduced. It gives important insight into a number of flow phe-
nomena, but also, with the second break of symmetry—the first one is due to the
Kutta condition—an explanation of essential phenomena, which are connected to
lifting-wing flow fields. The prerequisite is an extended definition of separation,

v
vi Preface

which besides the classical—ordinary—separation introduces the concept of

flow-off separation at sharp trailing edges of large aspect-ratio wings and sharp
leading edges of small aspect-ratio delta-like wings.
The concept of the kinematically active and inactive vorticity content, with a
compatibility condition for the flow-off separation at sharp edges, permits to
understand the properties of the evolving trailing vortex layer of large aspect-ratio
wings. This vortex layer then leads to the pair of trailing vortices behind the aircraft.
The concept, on the other hand, gives proof that discrete Euler methods at sharp
delta or canard leading edges indeed lead to an exact simulation of the evolution
of the primary lee-side vortex pair.
The book treats three main topics.
1) Basic Principles are considered in introducing chapters. They concern rele-
vant boundary-layer properties, vortex theory, the local vorticity content of shear
layers, the matter of discrete Euler solutions for lifting wings, the Kutta condition in
reality and an introduction to the topology of skin-friction and velocity fields.
2) Unit Problems concentrate on isolated flow phenomena, respectively con-
figuration parts. In our case, these are the flow phenomena present at large
aspect-ratio and small aspect-ratio delta-type wings. Capabilities of panel methods
and discrete Euler methods are investigated. One Unit Problem is the flow past the
wing of the Common Research Model. We demonstrate besides others that the
tip-vortex system leads to a very small non-linear lift and that a bridge can be
spanned over to the small aspect-ratio wing. Other Unit Problems concern the
lee-side vortex system appearing on this type of wings with sharp and fully or partly
round leading edges and also the vortical flow past a blunt-edged configuration at
hypersonic speed.
3) Particular Flow Problems of large aspect-ratio and small aspect-ratio
delta-type wings. In short sections, practical design problems are discussed.
The treatment of separated and vortical flow past fuselages, although desirable,
was not possible in the frame of this book.
The authors of the book are from the aerospace field and were for many years—
in teaching, research, as well as in industrial aircraft design—deeply involved in
phenomenological, mathematical, computational and flight-vehicle shaping issues
of separated and vortical flow. They wish to give the student and his teacher, the
researcher, and in particular, also the practical aerospace engineer an in-depth
knowledge about separated and vortical flow in aircraft wing aerodynamics.

Zorneding, Germany Ernst Heinrich Hirschel

Stockholm, Sweden Arthur Rizzi
Munich, Germany Christian Breitsamter
Zorneding, Germany Werner Staudacher
September 2020
Acknowledgements

The authors are much indebted to many colleagues, whose input was crucial for the
book. First of all, we wish to thank S. Pfnür, whose master thesis work was the
essential input to Chap. 8. Our doctoral students S. Crippa, J. Fischer, R. Hentschel, A.
Hövelmann, B. Schulte-Werning and S. Riedelbauch generously made available large
parts of their theses for Chap. 10. Substantial input was received from A. Büscher and
A. Schütte, too. Many thanks to all of them.
Directly and thankfully received was data, material, critical and constructive
comments, input of all kind and advise from A. A. Allen, K. Becker, N. Bier, O.
Brodersen, M. Drela, R. Friedrich, H. Fütterer, S. Görtz, W. Heinzerling, M. Herr,
S. M. Hitzel, D. Hummel, R. Konrath, H.-P. Kreplin, J. M. Luckring, D.
Niedermeier, R. Rudnik, and M. Tomac. C. Weiland made ad hoc computations and
illustrations for the book. G. Simeonides and C. Weiland were reading all the
chapters and gave critical and constructive comments.
Our spouses deserve special thanks for their patience, which was tested over
quite a long time.
Lastly, special thanks go to Thomas Ditzinger, the Editorial Director of
Interdisciplinary and Applied Sciences and Engineering from Springer, because for
many years, he very effectively and helpfully supported the publication of the
works of the first author of the present book.

September 2020 Ernst Heinrich Hirschel

Arthur Rizzi
Christian Breitsamter
Werner Staudacher

vii
Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 The Content of the Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 The Application Background . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 What Is Separation? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.1 The Classical View . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.2 The View Taken in This Book . . . . . . . . . . . . . . . . . 10
1.3.3 The Definition of Separation in This Book . . . . . . . . . 15
1.4 A Wider View at Separation . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.5 Flow Physical and Mathematical Models of Separated
and Vortical Flows . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 18
1.6 Basic Principles, Unit Problems and the Contents
of the Chapters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 24
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 25
2 Separation: Some Relevant Boundary-Layer Properties,
Interaction Issues, and Drag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1 Relevant Boundary-Layer Properties . . . . . . . . . . . . . . . . . . . . 30
2.2 Interaction Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.3 A Special Issue: The Locality Principle . . . . . . . . . . . . . . . . . 36
2.4 Aspects of Drag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.4.1 Drag Components . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.4.2 The Drag Divergence . . . . . . . . . . . . . . . . . . . . . . . . 39
2.5 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3 Elements of Vortex Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.1.1 Models of Finite-Wing Theory of Flight . . . . . . . . . . 46
3.1.2 Benefits in Studying Vortex Dynamics . . . . . . . . . . . . 47

ix
x Contents

3.2 The Concept of Vorticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

3.2.1 General Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.2.2 Designations and Definitions . . . . . . . . . . . . . . . . . . . 49
3.3 Stokes’ Theorem and the Concept of Circulation . . . . . . . . . . 50
3.3.1 The Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.3.2 Circulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.4 Origins of Vorticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.4.1 The Kutta Condition . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.4.2 The Cambridge School—No-Slip Viscous
Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 53
3.4.3 The Göttingen School—Inviscid Vortex-Sheet
Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 55
3.4.4 Other Origins . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 56
3.4.5 Short Conclusion . . . . . . . . . . . . . . . . . . . . . . . . ... 58
3.5 Entropy and Total Enthalpy Gradients and Vorticity:
Crocco’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.6 Equations of Transport of Vorticity . . . . . . . . . . . . . . . . . . . . 59
3.7 Helmholtz’s Vorticity Theorems . . . . . . . . . . . . . . . . . . . . . . . 60
3.8 Kelvin’s Circulation Theorem . . . . . . . . . . . . . . . . . . . . . . . . 61
3.9 Law of Biot-Savart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.10 Vortex Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.11 Structure of Trailing Vortices . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.11.1 SAAB 39 Gripen Wake Model . . . . . . . . . . . . . . . . . 68
3.11.2 Trailing Vortex Instability . . . . . . . . . . . . . . . . . . . . . 68
3.11.3 Crow Instability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.12 Vortex Layers and Vortices . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.12.1 Roll-Up of Shed Vortex Layers . . . . . . . . . . . . . . . . . 69
3.12.2 Vortex Stretching . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.12.3 Vortex Pairing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.13 Vortex Breakdown, Vortex Re-configuration . . . . . . . . . . . . . . 75
3.13.1 Fundamental Studies . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.13.2 Computed Vortex Breakdown Over a Delta Wing . . . 77
3.13.3 Vortex Re-connection . . . . . . . . . . . . . . . . . . . . . . . . 78
3.14 Separation and Vortex Flow Control . . . . . . . . . . . . . . . . . . . 78
3.15 Vortex Flows and Dynamic Structural Loads . . . . . . . . . . . . . 80
3.16 Basic Quantities of Trailing-Vortex Flow Fields . . . . . . . . . . . 82
3.17 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4 The Local Vorticity Content of a Shear Layer . . . . . . . . . . . . . . . 93
4.1 Definition and Derivation of the Local Vorticity-Content
Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Contents xi

4.2 Kinematically Active and Inactive Vorticity Content:

Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.2.1 Rankine Vortex . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.2.2 Two-Dimensional Boundary Layer . . . . . . . . . . . . . . 97
4.2.3 Near Wake of a Lifting Airfoil . . . . . . . . . . . . . . . . . 98
4.2.4 Bound Vortex of a Lifting Airfoil . . . . . . . . . . . . . . . 102
4.2.5 Three-Dimensional Boundary Layer . . . . . . . . . . . . . . 104
4.2.6 Near Wake (Trailing Vortex Layer) of a Lifting
Finite-Span Wing . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.2.7 Summary of the Results . . . . . . . . . . . . . . . . . . . . . . 107
4.3 Lift and Induced Drag: Two Breaks of Symmetry . . . . . . . . . . 108
4.3.1 First Symmetry Break: The Lifting Airfoil . . . . . . . . . 108
4.3.2 Second Symmetry Break: The Lifting Finite-Span
Wing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
4.3.3 The Symmetry Breaks in the Reality of Aircraft
Wings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.4 Flow Pattern at the Trailing Edge of Large-Aspect Ratio
Lifting Wings: A Compatibility Condition . . . . . . . . . . . . . . . 113
4.5 Final Remark . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
4.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
5 The Matter of Discrete Euler Solutions for Lifting Wings . . . . . . . 119
5.1 Vorticity Creation in Euler Solutions of Lifting-Wing
Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.2 Vorticity and the Related Entropy Rise . . . . . . . . . . . . . . . . . . 121
5.3 Critical Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6 About the Kutta Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6.1 Decambering Phenomena . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6.1.1 Boundary-Layer Decambering . . . . . . . . . . . . . . . . . . 128
6.1.2 Shock-Wave Decambering . . . . . . . . . . . . . . . . . . . . 129
6.2 Kutta Condition and Kutta Direction in Reality . . . . . . . . . . . . 131
6.3 Geometric Properties of Trailing and Leading Edges of
Actual Wings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
6.4 When Can an Edge Be Considered as to Be Aerodynamically
Sharp? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
6.5 Implicit and Explicit Kutta Condition, Modeling and Grid
Generation Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
6.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
xii Contents

7 Topology of Skin-Friction and Velocity Fields . . . . . . . . . . . . . . . . 147

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
7.1.1 General Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
7.1.2 Three-Dimensional Attachment and Separation . . . . . . 148
7.1.3 Detachment Points and Lines . . . . . . . . . . . . . . . . . . 150
7.1.4 Lighthill’s Separation Definition and Open-Type
Separation and Attachment . . . . . . . . . . . . . . . . . . . . 151
7.2 Singular Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
7.2.1 Flow-Field Continuation and Phase Portraits . . . . . . . 152
7.2.2 Off-Surface Flow-Field Portraits . . . . . . . . . . . . . . . . 156
7.2.3 Singular Points in Off-Surface Velocity Fields . . . . . . 161
7.3 Singular Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
7.4 Topological Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
7.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
7.4.2 Surface Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
7.4.3 Off-Surface Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
7.5 Structural Stability and Changes of Flow Fields . . . . . . . . . . . 170
7.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
8 Large Aspect-Ratio Wing Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
8.1.1 Five Flow Domains at and Behind the Lifting
Wing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
8.1.2 Influence of the Trailing Vortex Layer and Vortices
on the Wing’s Performance . . . . . . . . . . . . . . . . . . . . 185
8.2 Panel Method (Model 4) Solutions—Proper and Improper
Results in Flow Domain 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
8.2.1 The Kolbe Wing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
8.2.2 The Forward-Swept Wing . . . . . . . . . . . . . . . . . . . . . 190
8.3 Creation of Lift in an Euler Solution (Model 8) for a Lifting
Large-Aspect Ratio Wing—Proof of Concept in Flow
Domain 0 and 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
8.3.1 The Computation Case and Integral Results . . . . . . . . 194
8.3.2 Details of the Computed Flow Field
of Domain 0 and 1 . . . . . . . . . . . . . . . . . . . . . . . . . . 194
8.3.3 The Circulation and the Kinematically Active
Vorticity Content . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
8.4 RANS/URANS Solution (Model 10) for the CRM Case:
Flow Domains 0, 1, and 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
8.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
8.4.2 Computation Method and Grid Properties . . . . . . . . . 200
8.4.3 Flow Domain 0: The Flow over the Wing, . . . . . . . . . 202
8.4.4 Excursion: The Wing-Tip Vortex System and
Non-linear Lift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
Contents xiii

8.4.5 Flow Domain 0 Contd.: Trailing-Edge Flow

and the Compatibility Condition . . . . . . . . . . . . . . . . 214
8.4.6 Flow Domain 1: The Trailing Vortex Layer
in the Near Field . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
8.4.7 Flow Domain 2: The Trailing Vortices Appear . . . . . . 223
8.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
8.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
9 Particular Flow Problems of Large Aspect-Ratio Wings . . . . . . . . 233
9.1 Supercritical Airfoil—Shock-Wave/Boundary-Layer
Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
9.2 Flow Past a High-Lift System . . . . . . . . . . . . . . . . . . . . . . . . 239
9.3 The Wing in High-Lift Condition . . . . . . . . . . . . . . . . . . . . . . 244
9.4 The VHBR Engine and the Nacelle-Strake Vortex . . . . . . . . . 248
9.5 Wing-Tip Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
9.6 The Wake-Vortex Hazard: The Problem and Means
to Control It . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
9.7 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
10 Small Aspect-Ratio Delta-Type Wing Flow . . . . . . . . . . . . . . . . . . 269
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
10.1.1 Flight Vehicle Classes . . . . . . . . . . . . . . . . . . . . . . . . 270
10.1.2 Notes on the History . . . . . . . . . . . . . . . . . . . . . . . . . 277
10.2 Non-linear Lift Exemplified with the Plain Delta Wing . . . . . . 285
10.2.1 Geometrical and Flow Parameters . . . . . . . . . . . . . . . 285
10.2.2 The Non-linear Lift as Aerodynamic Phenomenon . . . 286
10.2.3 The Matter of Secondary and Higher-Order Lee-Side
Vortex Pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
10.2.4 Basic Influences of Sharp and Round
Leading Edges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
10.2.5 Limits: Vortex Breakdown and Vortex
Overlapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
10.2.6 Correlations of Lee-Side Flow Fields . . . . . . . . . . . . . 294
10.2.7 Flow-Physical Challenges . . . . . . . . . . . . . . . . . . . . . 297
10.2.8 Manipulation of Lee-Side Flow Fields,
an Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
10.3 Vortical Flow Past the Sharp-Edged VFE-1 Delta
Wing—Different Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
10.4 Creation of Lift in the Euler Solution (Model 8) for the
Sharp-Edged VFE-1 Delta Wing—Proof of Concept . . . . . . . . 305
xiv Contents

10.4.1 The Computation Case and Integral Results . . . . . . . . 306

10.4.2 Details of the Computed Flow Field . . . . . . . . . . . . . 308
10.4.3 The Circulation and the Kinematically Active
Vorticity Content in the Euler Simulation . . . . . . . . . 312
10.5 Vortical Flow Past the Round-Edged VFE-2 Delta Wing . . . . . 315
10.5.1 The Wing and the Subsonic Computation Case . . . . . 315
10.5.2 Two Pairs of Primary Vortices . . . . . . . . . . . . . . . . . 317
10.5.3 Vortex Breakdown . . . . . . . . . . . . . . . . . . . . . . . . . . 323
10.6 Partly Developed Swept Leading-Edge Vortices, the
SAGITTA Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
10.7 Laminar Hypersonic Flow Past a Round-Edged
Delta Wing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
10.8 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
10.9 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
11 Selected Flow Problems of Small Aspect-Ratio
Delta-Type Wings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
11.1 Lift and Stability Problems Connected to Lee-Side Vortex
Systems of Delta-Type Wings . . . . . . . . . . . . . . . . . . . . . . . . 360
11.2 Wing-Planform Shaping and Optimization . . . . . . . . . . . . . . . 364
11.2.1 Effects of the Wing Geometry . . . . . . . . . . . . . . . . . . 364
11.2.2 Wings of High-Speed Vehicles . . . . . . . . . . . . . . . . . 366
11.3 Wing Sections and Leading-Edge Flaps . . . . . . . . . . . . . . . . . 368
11.3.1 Wing Sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
11.3.2 Leading-Edge Flaps . . . . . . . . . . . . . . . . . . . . . . . . . 371
11.4 Fuselage Forebody Strakes . . . . . . . . . . . . . . . . . . . . . . . . . . 373
11.5 Spanwise Blowing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376
11.5.1 Blowing at a Pilot Model Without Strake and
Forebody Fins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
11.5.2 Blowing at the Pilot Model with Strake and Without
Forebody Fins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
11.5.3 Summarizing Remarks . . . . . . . . . . . . . . . . . . . . . . . 381
11.6 Design Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
11.6.1 Basic Configuration . . . . . . . . . . . . . . . . . . . . . . . . . 383
11.6.2 Modifications of the Wing Geometry . . . . . . . . . . . . . 385
11.6.3 The Final Configuration . . . . . . . . . . . . . . . . . . . . . . 387
11.7 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
12 Solutions of the Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
Contents xv

Appendix A: Useful Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417

Appendix B: Constants, Atmosphere Data, Units, and Conversions . . . . 431
Appendix C: Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
Appendix D: Abbreviations, Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
Permissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
Subject Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 445
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
Chapter 1
Introduction

Vortex layers and vortices, although not occupying much space in the flow field
around and behind an aircraft, can be seen as “the sinews and muscles of the fluid
motion”, as D. Küchemann was putting it [1].
Separated and vortical flow is present at any aircraft. The largest part of the flow
past a flight vehicle, however, can be considered as inviscid. Viscous effects are
restricted to the boundary layer, which is a bound vortex layer, in contrast to the
free or trailing vortex layer, which is the lifting wing’s wake. Despite of these vortex
layers, the flow over a lifting wing, for instance, can well be described in the frame
of inviscid theory, even only in the frame of potential flow theory. This all is the topic
of the classical text books.
The aerodynamic phenomena appearing in reality today are described with more
or less fidelity by flow realizations in ground-simulation facilities or by advanced
discrete numerical simulation methods (Computational Fluid Dynamics, CFD). This
in particular also holds for the vortex layers and vortices which are present in any
aircraft’s flow field.
A book with a concise presentation of the basics of fluid mechanics, and in partic-
ular of vorticity fields, vortex dynamics, etc., with applications to airfoils and wings
is that of M.J. Lighthill [2].
Newer books on vortex phenomena usually cover the whole range of phenomena
down to laminar-turbulent transition and turbulence, which basically all are related
to the phenomenon of vorticity, see, e.g., the books of H.J. Lugt (1996) [3] and of
J.-Z. Wu, H.-Y. and M.-D. Zhou (2006) [4].
D.J. Peake and M. Tobak in 1980 published their NASA Technical Memoran-
dum, later AGARDograph, on three-dimensional interactions and vortical flows [5].
A book specifically devoted to vortex flows in high angle-of-attack aircraft aerody-
namics is that of J. Rom (1992) [6]. The topology of three-dimensional separated
flow fields is treated by J. Délery (2013) [7]. Relevant conference proceedings are
beside others those of AGARD symposia held in Göttingen, Germany in 1975 [8],

© Springer-Verlag GmbH Germany, part of Springer Nature 2021 1

E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_1
2 1 Introduction

and in Rotterdam, the Netherlands in 1983 [9], and that of an IUTAM Symposium
in Novosibirsk, USSR in 1990 [10].
Monographs on boundary-layer and viscous flow in general are, for example, those
of F.M. White (2005) [11], T. Cebeci and J. Cousteix (2005) [12], H. Schlichting and
K. Gersten (2006) [13], and E.H. Hirschel, J. Cousteix and W. Kordulla (2014) [14].
Regarding aircraft aerodynamics in general we point to the classical works—still
very rewarding to study—of H. Schlichting and E. Truckenbrodt (1959/1979) [15]
and D. Küchemann (1978/2012) [16].
Recent publications are from J.D. Anderson, Jr. (2011) [17], M. Drela (2014) [18]
and the relevant chapters in C.-C. Rossow, K. Wolf, P. Horst (eds.) (2014) [19]. To
appear in 2020 is the book of A. Rizzi and J. Oppelstrup, who in particular present
an approach to aerodynamics with discrete numerical methods—computational fluid
dynamic—[20].
Since the 1990s large advancements were made in both discrete numerical sim-
ulation methods and experimental techniques. They led to new insights and also to
new views at the topic of separated and vortical flow.
About ten years ago economical and ecological pressures anew began to ask
for more efficient transport aircraft. This not only holds for the cruise phase (fuel
efficiency), but also for take-off, climb, approach and landing (reduction of noise and
exhaust emission). In general also a better and more reliable determination of static
and dynamic mechanical loads on the airframe and its components became desirable
in order to reduce airframe mass and to increase the structural life time.
Military aircraft basically underlie the same demands. Here, however, agility,
stalled flight and post-stall flight lead to additional requirements. Unmanned fighter
aircraft permit very high maneuverability which puts even stronger demands on the
airframe design.

1.1 The Content of the Book

For the aircraft designer and developer not only the numerical and experimental tools
govern his or her success, but first of all the knowledge of and the insight into the
aerodynamic phenomena present at the aircraft which is to be defined and developed.
This holds the more, if in the flight-envelope opening process of the aircraft, problem
diagnosis becomes necessary.
Therefore the book concentrates on separated and vortical flow past aircraft con-
figuration elements and total configurations. The background of application basically
is given with large aspect-ratio wing aircraft on the one hand and small aspect-ratio,
high leading-edge sweep, i.e., delta-type wing aircraft on the other hand. The former
are characterized by a leading-edge sweep small enough, so that no lee-side vortices
appear, the latter by a leading-edge sweep so high, that lee-side vortices will appear
at higher angles of attack.
The physical and mathematical basics and concepts of attached, separated and
vortical flow are presented to the needed degree. The concept of kinematically active
1.1 The Content of the Book 3

and inactive vorticity content of a vortex layer is introduced. It permits on the one
hand to connect the singularities of potential theory with the viscous reality of vortex
sheets and vortices.
On the other hand it allows to show that discrete numerical solutions of the poten-
tial equation (panel methods) and the Euler equations in principle are viable. Such
methods today still have their place in conceptual and pre-design work. In particular
the decades old question of how vortices appear in Euler solutions is answered.
Basically the applicability of computational methods is considered. In much detail
discussed are examples of flows past high and low aspect-ratio wings, and that, if
given, also in view of the whole aircraft with its performance, flyability and control-
lability demands.
In the following sections we first sketch the application background, then aspects
of separation are considered, a wider view at it is given. The introduction closes with
a classification of flow-physical and mathematical models of in particular separated
and vortical flow and an overlook over the following chapters.

1.2 The Application Background

The application background of the book are fixed-wing aircraft of all kind. However,
we concentrate on the phenomena present at large transport aircraft and likewise at
fighter aircraft, manned and unmanned. The reason for this is that much and detailed
information and data are available for such aircraft from a host of former and present
research programmes.
Typical large transport aircraft are shown in Fig. 1.1. In cruise flight the angle of
attack nominally is α = 0—as well as the flight-path angle γ—but the wing has a
rigging angle of incidence, as well as an appropriate spanwise distribution of camber
and twist of the airfoil sections, such that the demanded lift is obtained.

Fig. 1.1 Two typical transport aircraft. Left: Airbus A350 [21], right: Boeing 787 [22]
4 1 Introduction

Fig. 1.2 Schematic of the influence of configuration elements on the circulation distribution in
span direction Γ (y), its change in y-direction dΓ (y)/dy and local discrete vortices [23]. Left side:
clean wing, b0 indicates the distance between the fully developed trailing vortices, the tip vortex is
not indicated. Right side: wing with deflected flap

While the aircraft is in cruising mode, the wing is clean, i.e., slats and flaps are not
activated. A vortex sheet, the trailing vortex layer, separates from the trailing edge
and rolls up downstream into the two trailing vortices, left side of Fig. 1.2.
At the horizontal stabilizer, which as a rule exerts a downward directed trim force,
a secondary counter-rotating vortex layer and trailing vortices emerge. These and the
wing and stabilizer tip vortices are merged into the wing’s primary trailing vortex
layer, respectively the trailing vortices. Tertiary vortices emerging, for instance, at the
wing’s root, the flap tracks, the fuselage and at the nacelles of the turbo-jet engines,
are also merged into the primary vortex layer/trailing vortices.
If landing flaps, short flaps, and slats are activated, the situation changes drasti-
cally. The right side of Fig. 1.2 shows the effect for an isolated trailing-edge flap.
Lift, hence circulation increases, but a flap vortex appears. Its strength is connected
to the gradient dΓ (y)/dy at the spanwise end(s) of the flap.
Important, however, is to note that flap deflection as a rule goes together with
significantly high angles of attack. In order to get an idea what is to be expected,
we look at a typical flight path of a transport aircraft with the major flight segments,
Fig. 1.3.
In this figure, 1 denotes the segment of climb-out from take-off to thrust reduction,
2 that above it. In between is the possible segment 1 , which we disregard. Segment
3 is the cruise segment, which ends with the approach 4 to the airport. 4 is a possible
hold segment, when the aircraft has to wait for landing. We disregard that, too. 5, 6,
and 7 are the segments immediately ahead of touch-down—where also the ground
effect is an issue—8 is the special case of go-around, when the final approach of the
aircraft is aborted.
For medium-range and medium-size aircraft, like Boeing 737 and Airbus A320,
we collect data in Table 1.1.
1.2 The Application Background 5

Fig. 1.3 Schematic of the

flight path of a transport
aircraft with different flight
phases, after [24]

Table 1.1 Typical transport aircraft: flight-path segments and angle of attack α, flight-path angle
γ, settings of flaps η f lap , and slats ηslat , the latter two downward deflected with positive sign. H
and v are the characteristic altitude and speed. All numbers are approximate, in reality depending
on a number of parameters
No. Segment H u ∞ /M∞ α γ η f lap ηslat
(m) (m/s)/(-) (◦ ) (◦ ) (◦ ) (◦ )
1 Take- <120 80/- 5 12 15 22
off/climb-
out
1
2 Climb 2 650 130/- 2 7.5 0 0
3 Cruise 10,000 -/0.8 0 0 0 0
4 Approach 700 70/- 5.5 −3 15 22
5 Final 270 65/- 5 −3 40 27
approach
6 Flare 2.5 60/- 5.5 −2 40 27
7 Landing 1 60/- 4.5 −1 40 27
8 Go around 180 75/- 3 13 20 22

The data are typical ones, however depending much on any given case, and here
are only meant to give the reader an idea what has to be expected for the different
flight segments. The angle of attack α = 0 is related to the nominally horizontal
longitudinal axis of the aircraft. Not indicated is the power setting of the engines in
the different segments.
We see that indeed the flap settings partly go together with markedly increased
angles of attack. This must be kept in mind when considering Fig. 1.4. At the left the
influence of a fuselage vortex is shown, which leads to dΓ (y)/dy < 0. At the right
the full picture with the primary and the tertiary vortices is given. Indicated too is
the engine-nacelle vortex.
6 1 Introduction

Fig. 1.4 Schematic of the influence of configuration elements on the circulation distribution in
span direction Γ (y), its change in y-direction dΓ (y)/dy and local discrete vortices [23]. Left side:
influence of the fuselage. Right side: full configuration with propulsion system

The influence of the engine jets is restricted to the near field of the vortex-layer
system. Zones with high excess velocity during roll-up of the trailing vortex layer
are transported into the adjacent vortices. This can lead to changes of the vortex-core
diameters and to a reduction of the axial vorticity.
Of particular concern are separation and vortex phenomena at the nacelle and the
wing. The ever-growing by-pass ratios of modern engines lead to large airframe-
engine integration problems. Wing and nacelle stall in the presence of highly elastic
structural elements and an inhomogeneous atmospheric environment, in particular
in the final flight segments, pose large simulation and design problems. Chapter 9 is
devoted to short presentations of particular problems of large aspect-ratio wings.
The situation is quite different when we look at fighter aircraft like shown in
Fig. 1.5. These aircraft can have wings with highly swept leading edges and canards
(right part of the figure) or hybrid wings, i.e., a trapezoidal wing in combination with
strakes (left part of the figure). A common characteristic is that they partly operate
at very high angles of attack.
In Table 1.2 some flight data are collected. The data are typical ones, depending
much on the given aircraft and its mission. Again they are only presented in order
to give the reader an idea what has to be expected on the different flight segments.
Also not indicated is the power setting of the engines in the different segments.
We observe in particular a very broad range of the angle-of-attack values. In the
two combat segments the aircraft is maneuvering in the stall, in particular also in the
super-stall domain. Resulting are complex primary lee-side vortex systems, which
interact with a number of vortices and vortex layers stemming from configuration
1.2 The Application Background 7

Fig. 1.5 Two typical fighter aircraft [25]. Left: McDonnell Douglas F/A-18 at high angle of attack
with strake vortices. Right: Eurofighter Typhoon

Table 1.2 Typical fighter aircraft: flight-path segments and angle of attack α, flight-path angle γ,
settings of flaps η f lap , and slats ηslat , the latter two downward deflected with positive sign. Possible
foreplane (canard) settings are not given. H and v are the characteristic altitude and speed. All
numbers are approximate ones, in reality depending on a number of parameters (n.a.: not available.,
vari.: variable)
Phase H u ∞ /M∞ α γ η f lap ηslat
(m) (m/s)/(-) (◦ ) (◦ ) (◦ ) (◦ )
Take-off 0 65/- 15 10 10–20 20
Climb-out 0–9,000 150/- 5–10 3–10 15 20
Cruise 9,000 -/0.8–0.9 2–8 0 0 0
Dash 9,000– -/<2 2–8 0 0 0
15,000
Combat conventional 1,500– -/0.4–1.4 −5–25 vari. n.a. n.a.
13,000
Combat thrust vectoring 1,500– -/0.4–1.4 −10–70 vari. n.a. n.a.
13,000
Descent 10,000 150/0 5–10 < − 10 15 20
Landing 0 70/- 10–15 −10 0 20

elements and weapons. If a canard is present, its trailing vortex layers and tip vortices
interact with the general flow field. Similar interactions occur, if strakes are employed.
Figure 1.6 schematically shows as an example the flow structure over the
Eurofighter Typhoon configuration. It was determined from images generated with
the laser-light section technique [26].
We observe the lee-side leading-edge vortex, with an intersection where the slat
ends at its fuselage side (red and blue). The wing-tip vortex turns in the same direction.
The foreplane (canard) has an effective angle of attack of 17◦ . This means that its
wake-vortex layer and the wing-tip vortex turn in the same sense as the main lee-side
vortex turns. Vortices arising at the inlet, the fuselage strake and at the externally
carried weapons interact with the main structures.
8 1 Introduction

Fig. 1.6 Schematic of vortex layers and vortices at the Eurofighter at α = 27◦ , ηslat = 19.5◦ and
ηcanar d = −10◦ [26]

Fig. 1.7 A special case: unmanned combat air vehicle (UCAV) in the fuel-receiving mode [27]

For even larger angles of attack, the vortex field may become asymmetric to the
x-z plane and a side force ensues, although the aircraft nominally flies with the
side-slip angle β = 0.
Unmanned combat air vehicles—also called drones—permit much higher g-loads
than manned fighter vehicles. The flight parameters basically will be those of manned
vehicles. As example of such a vehicle we show in Fig. 1.7 a photo of the Northrop
Grumman X-47B demonstrator of an unmanned combat air vehicle (UCAV). It is
designed for aircraft carrier-based operations.
The particular wing shape—basically a delta wing with large strakes—together
with the signature reducing placement of engine inlet and nozzle on the upper (lee)
1.2 The Application Background 9

side and the serrated trailing edge are resulting in very complex lee-side flow fields
at large angles of attack.

1.3 What Is Separation?

1.3.1 The Classical View

The boundary layer was discovered by Ludwig Prandtl. He presented his findings at
the Third International Congress of Mathematicians in August 1904 in Heidelberg
with the paper “Über Flüssigkeitsbewegung bei sehr kleiner Reibung (On Fluid
Motion with Very Small Friction)” [28]. With his work he also settled the question
whether the fluid sticks to a body surface (no-slip wall boundary condition), which
still was not known in the second half of the 1800s. A slip of the flow along the
surface was considered to be possible [29].
What generally is not mentioned, is that in [28] the concept of boundary-layer
separation was laid out, too. Actually it was Prandtl’s work as an engineer on flow
problems of an exhaust system—the intended pressure recovery in a diffusor did
not happen—that later led to the discovery of the boundary-layer and the separation
phenomenon, see, e.g., also [30].
Prandtl’s original sketch of the streamlines in the vicinity of the separation point,
Fig. 1.8, shows from the left to the right the incoming two-dimensional boundary layer
(the primary boundary layer, Sect. 2.1), the appearance of the point of inflection due
to the adverse pressure gradient [14], the uplifting of the boundary layer (incipient
separation) ahead of the separation point, the separation point (the small circle at the
wall), the separation streamline (the full line leaving the separation point upwards,
its elevation angle λ is exaggerated in this illustration), and the streamlines beneath
the separation line. These streamlines are directed upstream, and turn around into
the direction of the separation line. The turning location is indicated by the broken
line emanating from the separation point. The existence of the secondary boundary
layer, Sect. 2.1, is not indicated.

Fig. 1.8 Prandtl’s sketch of

separation [28]
10 1 Introduction

Originally Prandtl’s boundary-layer concept simply regarded viscous flow past a

semi-infinite flat plate. The matter of boundary-layer separation can not be treated
with his boundary-layer equations. The most general approach employs the Navier–
Stokes equations.
Of course the problem of separation found more attention soon. What happens
at the separation point and downstream of it? In a general manner, the question first
has been raised by S. Goldstein: how to determine the solution of the boundary-layer
equation downstream of a station where the velocity profile is given [31].
Goldstein did show, amongst others, that generally the solution of the boundary-
layer equation is singular, if the velocity profile has a zero derivative at the wall, i.e.,
the wall shear-stress is zero. In such a case it is not possible to continue a boundary-
layer calculation downstream of the point of zero wall shear-stress. This can be
considered as one of the roots of interaction theory, see the sketch of this field in
[14], and also the overview of flow physical and mathematical models of separated
and vortical flows in Sect. 1.5 of the present book.
With regard to the mentioned singular behavior we would like to emphasize that
this is a property of the mathematical model in question. The singularity does not
appear in reality.1 The skin friction at the separation point actually does not vanish,
it just changes its sign! Of course the boundary-layer assumptions—small thickness,
small wall-normal velocity, hence “impressed” pressure field [14]—are no more
valid.
At the separation point a streamline leaves the surface as indicated in Fig. 1.8.
(In flow-topology considerations the separation point is called a singular point. Only
a few of such points exist on a body surface, Chap. 7. However, at these points no
singularity of the wall shear stress is present.) The elevation angle λ of the separating
streamline is small. K. Oswatitsch has given it as a function of the local streamwise
gradients of the skin friction and the wall pressure [33]. We discuss his approach in
Chap. 7.

1.3.2 The View Taken in This Book

We are dealing with flow past bodies of finite length and volume. These basically
are wings and fuselages. There we find separation and vorticity—the latter concen-
trated in vortex layers and vortices—at, above, below, and behind the bodies. In this
sub-section we look at the basic phenomena, first at the flow past an airfoil as a
canonical two-dimensional configuration, then at the flows past several canonical

1 Inthe context of flow singularities the singularity of the Blasius solution at the “leading edge”
(x = 0) of the flat plate with zero thickness should be mentioned. In the hypothetical reality of
such a flat plate—the zero thickness of course cannot be realized in an experiment—no singularity
exists. However, a very small initial region is present, where the assumption of continuum flow is
not valid. We point in this regard to the discussion of—although hypersonic—leading-edge flow in
[32].
1.3 What Is Separation? 11

three-dimensional configurations. In Sect. 1.3.3 finally the definition of separation

employed in this book is given.2
For his 2000 Lanchester Lecture A. Elsenaar had chosen the provoking title “Vor-
tex Formation and Flow Separation: the Beauty and the Beast in Aerodynamics”
[34]. The Beauty clearly are trailing vortex layers and the pairs of trailing vortices.
The Beast is separation.
Regarding the beast “separation” in [35, 36] globally a distinction was made
between two kinds of separation: ordinary separation—or simply put separation—,
which is a squeeze-off separation, and flow-off separation. The two appear in three-
dimensional as well as in two-dimensional viscous flow, Chap. 7.3
Ordinary separation, the classical separation, is the separation form which usually
is considered. The phenomena range from steady to highly unsteady ones (vortex
shedding), depending on the body shape and the free-stream situation.
Figure 1.8 above shows the canonical example of ordinary separation of two-
dimensional flow. In three-dimensional flow it appears, for instance, at round body
flanks where two boundary-layer streams meet and squeeze each other off the surface.
Of course, both ordinary and flow-off separation can be present at the same time,
see the sketch in the middle of Fig. 1.9 below. The primary separation there is a
squeeze-off or ordinary separation, at the trailing edge we have flow-off separation.
Flow-off separation happens at trailing edges of wings, trim and stabilization
surfaces, as well as at sharp leading edges of delta wings. Steady motion of the
aircraft assumed it basically is a steady phenomenon, but at trailing edges of finite
thickness, small unsteady flow portions may be present.
The concept of flow-off separation at least partly overcomes the beastly nature
of separation. It permits to connect the properties of the boundary-layer flow at the
trailing edge of lifting wings with the properties of the ensuing trailing vortex layer,
or at sharp leading edges of delta wings with the ensuing feeding layers of the primary
lee-side vortices. These connections are the main topic of Chap. 4 and there especially
of the Sects. 4.3 and 4.4. Of course behind the flow patterns of the boundary layers
are the flow patterns of the corresponding external inviscid flow fields.
• Flow Past a Lifting Airfoil as Canonical Two-Dimensional Configuration
The flow past a lifting airfoil can have several faces. In Fig. 1.9 schematics of separa-
tion are given. The airfoil moves steadily at subsonic speed, i.e., at a speed below the
critical Mach number (subcritical case). The angle of attack in each case is a virtual
one.
We see in the upper part of the figure the classical picture of a lifting airfoil.
Usually this mode of flow past an airfoil is considered to be flow without separation.
However, in our view, flow-off separation happens. Kinematically inactive vorticity
leaves the trailing edge. This only holds for steady motion of the airfoil, i.e., with

2 In Chap. 2 we consider some relevant properties of boundary layers, because they are what separates

from a body surface.

3 In these two kinds of separation always two boundary layers are involved in the separation process.

This situation is the rule, but there are singular points at the body surface where the situation is
different, Chap. 7.
12 1 Introduction

Fig. 1.9 Schematics of flow past a lifting airfoil as canonical two-dimensional configuration. Steady
motion at subcritical speed. Upper part: airfoil with flow-off separation. Middle part: airfoil with
steady ordinary and flow-off separation. Lower part: airfoil with unsteady ordinary and flow-off
separation

time unchanging lift. If the lift changes, during that process kinematically active
vorticity leaves the trailing edge, Sect. 4.2.3.
In the middle part of the figure the two kinds of separation are present. Shown is a
case with a large steady double recirculation configuration.4 The question is, whether
and in what angle-of-attack and Reynolds-number range such a steady separation
pattern can exist. Beyond such a (possible?) range we find the situation indicated in
the lower part of the figure: unsteady separation with vortex shedding. Regarding
these considerations we note that in the steady case only kinematically inactive
vorticity enters the far wake. This also happens in a time-averaged sense in the case
with vortex shedding. The lift of the airfoil time-averaged is constant.
• Three-Dimensional Flow Past Canonical Configurations We consider three
canonical configurations, a lifting large aspect-ratio wing, a lifting small aspect-ratio,
high leading-edge sweep wing (delta wing), and a lifting fuselage-like body, all in
steady motion at subcritical Mach number, Fig. 1.10. The flows past the first two
configurations are treated in detail in the Chaps. 8 and 10, flows past fuselage-like
bodies are not considered.
Consider the flow cases shown in Fig. 1.10. They are based on the assumption
of small to moderate angles of attack. The configurations move steadily at subsonic
speeds.
Flow-off separation happens at acute edges, in the case of a large aspect-ratio wing
at its trailing edge, Fig. 1.10a. The boundary layers coming from the upper and the
lower side of the wing flow off the trailing edge, Chap. 6, and merge, forming a wake.
This wake is called trailing vortex layer. In the lifting case it contains both kinemat-

4 The case of a single separation bubble—example in Fig. 7.2a in Sect. 7.1.4—is found, if a suction
peak is present shortly behind the airfoil’s nose (peaky pressure distribution) and laminar-turbulent
transition happens across this bubble (bubble transition) [14].
1.3 What Is Separation? 13

Fig. 1.10 Three canonical configurations with three-dimensional separation and vortex phenomena
[35, 37]. a a typical large aspect-ratio wing with small leading-edge sweep, b a slender wing with
large leading-edge sweep, c a fuselage. Small or moderate angles of attack, subcritical speed. The
situation at the afterbody of the fuselage is not considered. LE: leading edge. Flow-off separation
is indicated as type a, ordinary separation as type b
14 1 Introduction

ically inactive and active vorticity, Chap. 4. This vortex layer further downstream of
the wing curls up and forms the two discrete trailing vortices, Sect. 8.
A separate phenomenon is the wing-tip vortex.5 It is formed due to the flow
around the wing tip, Chap. 8. The tip vortices merge into the trailing vortex layer
respectively into the trailing vortices. If the wing tip is sharp edged (acute edge),
the tip vortex comes into being by flow-off separation. If the wing tip is rounded,
ordinary separation happens. At the wing tip also secondary and higher separation
phenomena can be present, Sect. 8.4.3.
The situation basically is similar at small aspect-ratio wings with highly-swept
leading edges, Fig. 1.10b. There we have at the trailing edge the trailing vortex layer
like in the case of the large aspect-ratio wing. At the swept leading edges lee-side
vortices appear.6 These vortices come into being either by flow-off separation, if the
leading edge is sharp, or by ordinary separation, if the leading edge is round. The
lee-side vortices downstream interact with the trailing vortex layer.
At the lee side of the small aspect-ratio wing the flow field is governed by the
(primary) lee-side vortex pair. Secondary and even tertiary separation and attach-
ment phenomena can be induced. At high angle of attack, vortex breakdown can
happen, Sect. 10.2.5. Note that for the delta wing in Fig. 1.10b such phenomena are
not indicated.
At a fuselage-like body at small to medium angle of attack, Fig. 1.10c, we see
ordinary separation appearing at the body flanks. At larger angles of attack the sit-
uation can become very complex with an unsteady von Kármán vortex-street like
appearance.
Such a separation pattern can lead to side forces, although the configuration nom-
inally moves without side-slip. The effect typically can appear at rounded forebodies
of fighter aircraft and missiles at large angle of attack, as well as at delta-like wings,
see the example in Fig. 7.19 in Sect. 7.5.
At the afterbody of the fuselage also ordinary separation happens. Quite complex
flow topologies are possible, depending on the configuration details. In Fig. 1.10c
the situation at the afterbody of the fuselage is not indicated.7
Of course at a real aircraft configuration with fuselage, wing, aerodynamic sta-
bilization, trim, and control surfaces, as well as propulsion-system elements and
high-lift devices a multitude of vortex layers and vortex configurations is possible.
This in particular holds for delta-wing configurations.

5 Trailingvortices and tip vortices often are mixed up. They are separate phenomena. They only in
a sense fall together in Prandtl’s lifting-line model.
6 The appearance of the lee-side vortex pair depends not only on the angle of attack, but also on the

leading-edge sweep and the free-stream Mach number, Sect. 10.2.

7 The reader should note that we do not treat fuselage-flow problems in this book.
1.3 What Is Separation? 15

1.3.3 The Definition of Separation in This Book

In the preceding sub-section we presented an overall picture of separation phenomena

as we find them in aircraft aerodynamics. Necessary, however, is a general look at the
involved flow fields. This general look will be given in Chap. 7, where the topology
of skin-friction and velocity fields is treated.
Above we have noted that generally always two boundary layers are involved
in a separation process. In the following Chap. 2 therefore we sketch some relevant
boundary-layer properties. In addition we discuss the diverse modes of interaction,
which can happen in the flow past a body, as well as the locality principle.
Also above we have introduced the terms ordinary separation and flow-off sepa-
ration. We will see in Chap. 7 that this implies singular points (attachment and sepa-
ration points) as well as singular lines (attachment and separation lines) at the body
surface as well as singular points and to a degree also singular lines above/behind
the body.
In this book we always deal with separated and vortical flow past finite-length
bodies, i.e., wings of all kinds, separately or as whole flight-vehicle configurations.
At finite-length bodies, separation naturally always occurs: the attached viscous flow
past the body finally leaves the body, at a blunt afterbody via ordinary separation.
The situation in three-dimensional flow generally is very complicated compared to
that in two-dimensional flow, which is a special case.
The classical definition of separation in two-dimensional flow, which is based on
the observation that the wall-shear stress vanishes at the separation point, τwall = 0,
is not sufficient.8 (We do not give here an overview over the criteria found in the
literature, see, e.g., [5].)
In view of this discussion it readily can be imagined that the definition of flow
separation as we have it in two-dimensional boundary-layer flow—vanishing of the
skin-friction—is useless.
We state that a possible definition reads like [14]:
Separation is present, if vorticity is transported away from the body surface by
convection, and subsequently vortex sheets and vortices are formed. The vortex lay-
ers and vortices carry both kinematically active and inactive vorticity. On the body
surface the boundary-layer equations are not valid at the respective separation loca-
tions.
In a sense, of course, this definition is trivial, because at bodies of finite extent
separation always happens.
Remains to consider what happens with the vorticity carried away from the body
surface and how the picture of separation evolved since its inception by Ludwig
Prandtl. This is sketched in the following section.

8 This is one of three criteria discussed, for instance, by E.A. Eichelbrenner [38].
16 1 Introduction

1.4 A Wider View at Separation

A remarkable fact is that the concepts of vortex motion and vorticity were developed
quite a time before the concept of the boundary layer came into being.9
In 1752 d’Alembert proved that in incompressible, inviscid, and irrotational, i.e.,
potential flow, no net force acts on a body moving with constant velocity relative to
the surrounding fluid, d’Alembert’s paradox.10 H. von Helmholtz many years later
rejected the general potential flow assumption. Important progress was made with
his introduction of vortex discontinuities and vortex motion, his formulation of the
vortex laws, then with Kelvin’s theorem, and the theorem of Stokes—all treated in
Chap. 3.
The concept of vortex discontinuities and vortex motion of course did not solve
the problem of how to understand and describe the lift of a wing. The concept of
circulation—Stokes’ theorem connects the vorticity in a domain with the circulation
around it—was pointing into the right direction. Of the two rivaling theories, the
circulation theory and the discontinuity theory—see the great study “The Enigma of
the Aerofoil” by D. Bloor [41]—the former, i.e., that of circulation around an airfoil
finally prevailed. The circulation is determined by the Kutta–Joukowski condition—
in short Kutta condition—at the sharp trailing edge.
In 1902 M.W. Kutta published his lift theory, which is a circulation theory [42].
The amount of circulation around the airfoil is determined by the condition that the
flow smoothly leaves the trailing edge. N. Joukowski independently developed his
theory of the lift of an airfoil in the years 1902–1912 [43].
The step from the lifting airfoil to the lifting finite-span wing was made by F.W.
Lanchester [44] and by L. Prandtl [45]. Regarding the still ongoing debate of the
priority of the two researcher’s work we refer the interested reader to the historical
note of J.D. Anderson, Jr. [17].
At the lifting finite-span wing a trailing vortex layer leaves the trailing edge.
Several chord lengths behind the wing this layer has rolled up to the pair of trailing
vortices. Prandtl in his lifting-line model postulates the lifting—or bound—vortex
to lie within the wing, the trailing vortices at the wing tips—the trailing vortex layer
and the tip vortex are not a topic—and at infinity behind the wing the starting vortex.
This vortex-filament arrangement hence forms a closed path. In this way the first
theorem of the vortex laws of H. von Helmholtz is fulfilled.
The lifting-line model hence is a strong, however very practical simplification of
the situation in reality. With it the flow downwash at the location of the wing and
the induced drag was described. The lifting-line model was the starting point for
a succession of more and more refined models, see, e.g., [15, 17], and in a larger
context also [16].
The Kutta condition actually means that flow-off separation happens at the trailing
edge. However, when Kutta published his theory, the existence of the boundary layer

9 We give only a very brief sketch of the developments. The interested reader will find detailed
material in, for instance, [39, 40].
10 Below we will see that also compressible potential flow is possible.
1.4 A Wider View at Separation 17

was not yet known. Hence also a connection between circulation, vorticity and certain
boundary-layer properties was not established.
In 1963 Lighthill introduced the wall-normal integral of the vorticity across the
boundary layer as the total vorticity per unit area of (a) boundary layer [46]. In
the mid 1980s he connected the circulation around an airfoil with the vorticity in
the boundary layers at the upper and the lower side of it [2]. He now called the
wall-normal integral the local strength of a vortex layer.11
Also in the mid 1980s Lighthill’s original concept was generalized to three-
dimensional flow and the concept of kinematically active and inactive vorticity con-
tent Ω of a shear layer was introduced [35, 47]. This concept, Chap. 4, allows
1. to connect the vortex singularities of potential flow theory to the boundary lay-
ers, the vortex layers and the vortices of real airfoil and wing flows in the large
Reynolds-number limit Re → ∞,
2. to understand why the linear and non-linear potential flow methods (lifting surface
methods, panel methods) permit to determine lift of airfoils and lift and induced
drag of wings,
3. to understand why Euler methods in principle permit to determine lift of airfoils
and lift and induced drag of wings, and in addition also to model trailing vor-
tex layers of large aspect-ratio wings as well as the (primary) lee-side vortices
appearing above wings with high leading-edge sweep.
Item 2 and 3, however, only cover lifting airfoil and wing cases where the ‘real’
flow leaves the surface via flow-off separation (implicit Kutta condition). An explicit
Kutta condition in order to mimic ordinary separation is not viable. This aspect of
the Kutta condition—the implicit one as well as the hypothetical explicit one—is
treated in Chap. 6.
Regarding d’Alembert’s paradox, the flow-off separation at an airfoil’s sharp trail-
ing edge—modeled as Kutta condition in potential flow methods—can be considered
as to produce a break of symmetry (first symmetry break), Sect. 4.3.1. (This must not
be seen geometrically, but with regard to the pressure distribution.) Hence in such
cases a force—lift—can be determined also with inviscid computation methods.12
At a finite-span wing another break of symmetry happens (second symmetry
break). Hence also with an inviscid computation model lift, the flow downwash at
the location of the wing, and the induced drag can be determined. The reason for that
is that in reality the shear of the upper and the lower inviscid flow part at the trailing
edge is present already in the basic inviscid flow field: the wing upper and lower side
flow-fields shear, Sect. 4.3.2.

11 The reader should note that we do not use the term strength for the vorticity integral as Lighthill
does. We call that integral the intensity of the boundary layer or vortex layer, and more generally,
the local vorticity content of a shear layer.
12 The drag of the airfoil then is a purely viscous drag, composed of the skin-friction drag and the

viscosity-effects induced pressure or form drag due to the displacement properties of the boundary
layers at the upper and the lower side of the airfoil, Sect. 2.4. In supercritical compressible flow
other effects come into play, also Sect. 2.4.
18 1 Introduction

1.5 Flow Physical and Mathematical Models of Separated

and Vortical Flows

For the description of separated and vortical flow physical and mathematical models
have been developed and are in use. We give now—following [35]—a characteri-
zation of these models, summarized in Table 1.3. This characterization is necessary
to remove ambiguities in the considerations, investigations and discussions in this
book.
We generally assume that the flow is rather well ordered, i.e., that only the basic
features are present, as sketched in Figs. 1.9 and 1.10. Secondary features, like sec-
ondary and tertiary separation phenomena and unsteadiness, however, implicitly are
covered, too.
Preliminary remarks: (1) airfoils and wings of all kind (lifting surfaces, but also
aerodynamic stabilization, trim and control surfaces) summarily are called wings.
(2) An implicit Kutta condition is what naturally is given at a sharp (acute) trailing
or leading edge and which leads to flow-off separation. This concerns Model 1, 2, 8,
9, 10 and 11. (3) An explicit Kutta condition is a condition, which in a mathematical
model or computation method would have to be prescribed. It concerns Model 4, 5
and 7.
We now come to the characterization of the models given in Table 1.3. In this
book we mainly deal with Model 1, 2, and 4 as well as with Model 8, 9, 10 and 11.
– Model 1: Real flow
This is the flow—a physical phenomenon—as perceived in the frame of continuum
mechanics. It is to be prescribed partly or completely with the help of the following
mathematical models.
– Model 2: Modeled viscous flow
This is a qualitative/quantitative mathematical model, which is discussed in
Chap. 4. It assumes high Reynolds numbers and neglects strong and global inter-
action, Sect. 2.2. Thus it permits to construct the link between the boundary layers
on a wing and the external inviscid flow via the concept of the kinematically active
or inactive vorticity content of a shear layer. This means that this model gener-
ally allows to connect the singularities of potential flow theory with properties of
boundary layers, shear layers and vortices. With this model results of Models 4
and 8 can be understood.
– Model 3: Potential flow
A pure potential (incompressible, irrotational and inviscid) flow model (Laplace
equation), i.e., a model without a Kutta condition, is not suited for the description
of lifting wing flows. In such a flow the global force on a body is zero (paradox of
d’Alembert).
– Model 4: Modeled Potential Flow
The decisive feature of this model is the explicit Kutta condition at a wing’s trailing
edge. This condition introduces into the solution the flow feature observed in reality
(Model 1), that at the trailing edge of a wing—an acute corner—the flow smoothly
leaves the wing’s surface (flow-off separation). The Kutta condition introduces the
1.5 Flow Physical and Mathematical Models of Separated and Vortical Flows 19

Table 1.3 Flow physical and mathematical models

No. Model name Description
1 Real flow As perceived in reality, with boundary
layers, vortex layers, vortices, transport of
vorticity by diffusion and convection
2 Modeled viscous flow Distinction between inviscid flow and
boundary layer, kinematically active and
inactive vorticity content of boundary
(shear) layer is related to the magnitude of
the inviscid external flow
3 Potential flow (Laplace eq.) Incompressible irrotational inviscid flow,
no global force acts on a body (paradox of
d’Alembert)
4 Modeled potential flow Incompressible irrotational inviscid flow, in
particular on wings with an explicit Kutta
condition and hence vortex layers and also
vortices, both in the form of vortex
singularities
5 Compressible potential flow Compressible irrotational inviscid flow, in
particular on wings with an explicit Kutta
condition and hence vortex layers and also
vortices, both in the form of vortex
singularities
6 Euler flow Inviscid flow, vorticity is created in curved
shock waves only, convective vorticity
transport
7 Modeled Euler flow Inviscid flow, in particular on wings with
an explicit Kutta condition and hence
vortex layers and vortices, both regarded as
vortex singularities
8 Discrete modeled Euler flow Inviscid flow, in particular on wings with
an implicit Kutta condition and
consequently vortex layers and also
vortices of finite thickness because of
diffusive transport due to the inherent
viscosity of the numerical scheme
9 Navier–Stokes flow Most completely modeled laminar, in
principle also transitional flow (DNS),
creation and transport of vorticity, inherent
numerical viscosity may introduce error in
diffusive transport of all kind
10 RANS, URANS flow Reynolds averaged Navier–Stokes
turbulent flow, also unsteady (URANS)
flow, creation and transport of vorticity,
inherent numerical viscosity
11 Scale-resolving and hybrid flow LES etc. modeled turbulent flow, creation
and transport of vorticity, coupling with
RANS/URANS (hybrid methods), inherent
numerical viscosity
20 1 Introduction

needed symmetry breaks, Sect. 4.3, which lead to net global forces. In the case of
the airfoil it is the first symmetry break which leads to the lift. In the case of the
finite-span wing the second symmetry break then leads to the trailing vortex layer,
the downwash at the location of the wing, and the induced drag.
How far the Kutta condition can be generalized to mimic ordinary separation—
implicit Kutta condition—is an open question. Attempts in that direction, however,
indicate that this is not a viable option.
For lifting wings, the implementation of the Kutta condition introduces, in the
simplest form, potential vortices as bound (lifting) and trailing vortices. Many
models have been proposed and are/were in use, from the simplest one, Prandtl’s
lifting-line model, to lifting surface theories of all kind, see, e.g., [15, 17].
The most developed Model 4 methods today are the panel methods, see, e.g., [48].
In panel models source and vortex singularities are employed. Trailing vortex
layers behind wings or vortex layers emanating from highly swept sharp leading
edges of delta wings can be geometrically modeled as discontinuity sheets, see,
e.g., the examples in [6].
– Model 5: Compressible potential flow
The general equation describing inviscid, compressible, however irrotational flow
is the velocity potential equation, see, e.g., [17].13 Variations of it range from
the small perturbation equation down to the classical potential (Laplace) equation
(Model 3 and 4). The compressible potential flow approach (full potential equa-
tion) formerly did find its applications in airfoil flow. Today due to its inherent
restrictions it does no more play a role, see also Sect. 6.1.2.
– Model 6: Euler flow
It is to surmise that a pure Euler flow model does not allow—like the pure potential
flow model, Model 3—to compute a global force on a body, except for the wave
drag in transonic and supersonic flow. A proof, however, like that of d’Alembert
seems not to exist. We note that in compressible flow only finite velocities can be
obtained with this model, in contrast to the potential flow model, where infinitely
large velocities are possible.
– Model 7: Modeled Euler flow
It is to expect that the introduction of an (explicit) Kutta condition into Model
6 leads to the symmetry breaks like in Model 4. Consequently lift, induced drag
and discontinuity sheets should evolve in a—hypothetical—solution. This would
amount to analytical solutions, which are not known, or to ideal discrete modeled
Euler solutions (Model 8), which, however, are also not known. See in this regard
also the study [50].
– Model 8: Discrete modeled Euler flow
General solutions of the Euler equations obviously can only be found by employ-
ing discrete numerical solution algorithms. These allow, due to their nature, only
a finite resolution of the computation domain. Moreover, they introduce diffusive
properties into the solution, because of either inherently (implicit) and/or explicitly

13 Actually this equation is derived from the continuity equation. For a generalized derivation see,

e.g., [49].
1.5 Flow Physical and Mathematical Models of Separated and Vortical Flows 21

added, artificial (also called numerical) viscosity, which ensures/enhances numer-

ical stability.14 These diffusive properties, although unwanted in principle, lead
to a—computed—flow behavior with an implicit Kutta condition that reproduces
sufficiently well important aspects of the reality. Moreover it leads to a finite thick-
ness of vortex layers and vortices, and in them to the appearance of an entropy
rise, Chap. 5.
Typical for Model 8 solutions, in particular for flow past lifting surfaces, are the
following features:
(a) No explicit Kutta condition, at least in finite volume methods, is necessary at a
trailing edge or at a highly swept leading edge, where in reality flow-off separation
occurs.
(b) No stagnation point is reached at the sharp trailing or leading edges, respectively,
as is the case in potential flow (Model 4).
(c) Vortex layers and vortices are not discontinuities or singularities, as it would be
the case in Model 6 and 7, but resemble the real phenomena. Particularly the
kinematically active vorticity in principle is correctly simulated.
(d) At round longitudinal edges (non-swept round wing tips, fuselage flanks), where
in reality ordinary separation occurs, vortex layers may evolve, which are ren-
dered physically incorrect.
For all these features no rational explanations exist so far. A sharp trailing edge—
an acute edge with finite opening angle—is a geometrical singularity, where in an
ideal inviscid flow (Model 4) a stagnation point exists. This is because the flow
cannot change direction at a concave corner with a finite velocity. This is the flow
situation at the trailing edge, where the flow passes into the bisector of the acute
edge. The stagnation point is an exact result of potential theory. (We point in this
respect to the question of finite-angle or cusp trailing edge of supercritical airfoils
[51], but note that in the reality of aircraft the trailing edges are blunt, Sect. 6.3.)
Potential theory, Model 2, on the other hand, allows for the flow around a con-
vex corner, though with infinite velocity. That is not possible for an compressible
Euler solution due to the finite total enthalpy of compressible flow. Maybe the
inviscid flow solution (Model 6) contains both cases, and the explicit Kutta con-
dition (Model 7) must suppress the convex-corner part of the solution.
If, however, an implicit Kutta condition is used, i.e., if one simply makes use of
the fact that in the solution of the Euler equation the flow breaks away from the
surface at acute corners, that flow may contain in a way both solutions, even if no
infinite velocity is possible for the Euler flow.
At sharp trailing edges or swept leading edges the implicit Kutta condition gen-
erally works well, at least in the sense that for flight-vehicle design investiga-
tions useful and reliable results can be obtained—though to a certain degree with
restricted accuracy. At round swept leading edges and at round longitudinal edges,
as noted under (d), a pseudo implicit Kutta condition may appear, and ordinary
separation is present. That generally is unphysical. The application of an explicit

14 An ideal discrete modeled Euler solution (Model 7) would not have such diffusive properties.
22 1 Introduction

Kutta condition, which mimics ordinary separation of the cross-flow, is not a viable
option.
At this point it must be mentioned that in a mathematical model with any kind
of Kutta condition—here Models 4, 5, 7, 8—the flow, i.e., the numerical solu-
tion scheme must have sufficient freedom to evolve, Fig. 1.11. In particular in
three-dimensional cases the shear between the separating flows must correctly be
described.
If in cases with supercritical Mach number a shock wave is present, say at the
upper (suction) side of an airfoil or wing, the Kutta direction does not lie in the
bisector of the trailing edge, but is deflected toward the upper side, Sect. 6.1.2.
This shock-wave decambering, which appears in addition to the boundary-layer
decambering, must be allowed for in numerical solutions. This holds especially
for explicitly prescribed Kutta conditions.
We conclude this discussion of Model 8 with the statement that apparently no
rational theory exists regarding this model. Our description is based on countless
observations from the satisfying application of Euler codes in research and aircraft
design work. The observed behavior, however, can be explained with the help of
the concept of the local kinematically active and/or inactive vorticity content of a
shear layer, as discussed in Chap. 4.
– Model 9: Navier–Stokes flow
The Navier–Stokes equations as continuum model of laminar fluid flow yield the
most complete description also of—steady and unsteady—separated and vorti-
cal flow. This description of course only holds for discrete numerical models of
these equations together with the continuity and the energy equation. A practical
employment of such models for transitional and turbulent flow is not possible
because of the computer power and storage which would be needed. The direct
numerical simulation (DNS) of laminar-turbulent transition and turbulent flow is
a research tool, but today by no means a tool of design work.
– Model 10: RANS, URANS flow
For attached turbulent flow a derivative of the Navier–Stokes (NS) equations are
the Reynolds-Averaged Navier–Stokes (RANS) equations, which employ statis-
tical turbulence models, see, e.g., [52]. For unsteady flow problems the URANS
equations are utilized, but not for massively unsteady separated flow. For such
flows scale-resolving methods are an option, see below Model 11.
The RANS approach is in wide use, also for the numerical simulation of the
flow past lifting wings, with both large and low aspect ratio. Recent reviews and
overviews of the capabilities of the RANS approach can be found in [53, 54].
A basic problem can be laminar-turbulent transition. Its prediction and simula-
tion today relies on empirical and semi-empirical criteria and models, see, e.g.,
[14]. This works with well-known flow families, but not in general. Non-empirical
transition models and methods have been proposed and developed, but for several
reasons so far they did not come into practical use.
For the flow past large aspect-ratio wings—discussed in Chap. 8—the RANS
approach is the established approach, with transition usually being a background
problem only. For the low aspect-ratio, high leading-edge sweep wing flows—
1.5 Flow Physical and Mathematical Models of Separated and Vortical Flows 23

Fig. 1.11 Kutta condition: necessary freedom to evolve [35]. a flow-off separation at a wing’s
trailing edge, b ordinary separation at the flank of a body

treated in Chap. 10—also the RANS approach is the established approach, although
certain observations point to deficiencies, see, e.g., [55]. For the flow past highly
swept round leading edges the prediction of laminar-turbulent transition poten-
tially is a problem.
First and second-order boundary-layer methods, [14], can be considered as sub-
sets of NS/RANS methods. However, treated with them can only weak interaction
problems, Sect. 2.2, hence neither ordinary nor flow-off separation.
– Model 11: Scale-resolving and hybrid flow
Massively unsteady separated flow (generally flow with vortex shedding) cannot
be treated with RANS or URANS methods. For such flows scale-resolving meth-
ods, e.g., large-eddy simulation (LES) is an option. Several different versions are
in use, for instance detached-eddy simulation, DES [56]. We note in passing that
DNS, see above Model 9, is scale-resolving, but not counted as Model 11 method.
Recent developments are hybrid RANS/LES models. Here the changeover from
the RANS solution to the LES solution poses a particular problem. We only cite
one recent European publication treating the gray area problem [57].
Another problem lies in the very nature of turbulent attached and separated flow.
A laminar boundary layer, or shear layer (vortex layer) in general, has a smooth
edge. (With laminar boundary layers this is chosen to be the 99 per cent thick-
ness.) For turbulent boundary layers and shear layers, as well as vortices, RANS
or URANS solutions suggest the presence of smooth edges, too. But the reality
(Model 1) is different. The mentioned flow phenomena in reality have rugged
unsteady edges, which cannot be described with RANS or URANS methods.
However, in general this is not necessary. But when problems like impingement of
shear layers or vortices on (downstream located) configuration elements are to be
treated (dynamic loads and/or vibration excitation), scale-resolving LES or hybrid
RANS-URANS/LES approaches become necessary.
We close this section with the observation that nowadays in aircraft design work
the influence of the aeroelastic deformation of, for instance, a wing under load more
and more needs to be taken into account. A compact introduction to aeroelasticity
of large-aspect ratio wings can be found in [51]. Aeroelasticity must be taken into
account, too, if flight-experimental data are to be used for comparison purposes, the
same holds in connection with wind-tunnel data, see e.g., [58].
24 1 Introduction

1.6 Basic Principles, Unit Problems and the Contents of the

Chapters

When treating issues of separated and vortical flow in aircraft wing aerodynamics in
this book, our goal is to obtain an understanding of the problems at hand, to show
the simulation capabilities of computation methods of different modeling level, and
to demonstrate and to discuss the relevant flow phenomena.
Times are bygone, in which the configuration finding, at least of military aircraft,
throughout was made with modular wind-tunnel models. Today the shape defini-
tion basically is made on the computer. However, behind this statement lies the
understanding of the complexity of the vehicle aerodynamics as well as the concept
of verification and validation of the discrete numerical methods in use, see, e.g.,
[59]. Helpful, even necessary, is the establishment and consideration of the problem-
complexity hierarchy. We refrain from a detailed discussion and point instead to
[60].
Figure 1.12 gives an example of a system decomposition, in this case for a fighter
aircraft [61]. The figure shows at the basis Unit Problems with isolated flow physics.
Below this level of course a level—not shown—is present with Basic Principles, for
instance boundary-layer theory and vortex theory.
In this book we basically have a structure, which first gives an introduction to Basic
Principles. These consist of (a) an overview of boundary-layer properties, interaction
issues and drag, Chap. 2, (b) an introduction to elements of vortex theory, Chap. 3, (c)
the introduction of the concept of the local vorticity content of shear layers, Chap. 4,
(d) the matter of discrete numerical solutions of the Euler equations, Chap. 5, (e) a

Fig. 1.12 Hierarchy of system decomposition [61]. PAI stands for Propulsion-Airframe Integration
1.6 Basic Principles, Unit Problems and the Contents of the Chapters 25

consideration of the Kutta condition in the real flight-vehicle environment, Chap. 6,

and (f) a short introduction to flow-field topology, Chap. 7.
Above this level with Basic Principles we discuss Unit Problems—the lowest
level in Fig. 1.12—in terms of the flows past large aspect-ratio wings, Chap. 8, as
well as small aspect-ratio delta-type wings, Chap. 10. For each of these two wing
classes we eventually consider particular flow problems, which can be considered to
belong to the Subsystems level in the figure, Chap. 9 for large aspect-ratio wings, and
Chap. 11 for small aspect-ratio delta-type wings. Solutions of the problems posed in
Chaps. 2–11 are given in Chap. 12.
The book closes with a collection of useful relations, Appendix A, air proper-
ties, atmosphere data, units and conversions in Appendix B, symbols in Appendix C,
abbreviations, acronyms in Appendix D, and, following the acknowledgement of
copyright permissions, the author and the subject index.

References

1. Küchemann, D.: Report on the IUTAM symposium on concentrated vortex motion in fluids. J.
Fluid Mech. 21, 1–20 (1965)
2. Lighthill, J.: An Informal Introduction to Theoretical Fluid Mechanics. Clarendon Press, Oxford
(1986)
3. Lugt, H.J.: Introduction to Vortex Theory. Vortex Flow Press, Potomac (1996)
4. Wu, J.-Z., Ma, H.-Y., Zhou, M.-D.: Vorticity and Vortex Dynamics. Springer, Berlin (2006)
5. Peake, D.J., Tobak, M.: Three-dimensional interactions and vortical flows with emphasis on
high speeds. NASA TM 81169 (1980) and AGARDograph 252 (1980)
6. Rom, J.: High Angle of Attack Aerodynamics. Springer, Heidelberg (1992)
7. Délery, J.: Three-Dimensional Separated Flow Topology. ISTE, London and Wiley, Hoboken
(2013)
8. N.N.: Flow separation. In: Proceedings of AGARD Symposion, Göttingen, Germany, May
27–30, 1975. AGARD-CP-168 (1975)
9. N.N.: Aerodynamics of vortical type flows in three dimensions. In: Proceedings of AGARD
Symposion, Rotterdam, The Netherlands, April 25–28, 1983. AGARD-CP-342 (1983)
10. Kozlov, V.V., Dovgal, A.V.: Separated flows and jets. In: Proceedings of IUTAM-Symposium,
Novosibirsk, USSR, July 9–10, 1990. Springer, Berlin (1991)
11. White, F.M.: Viscous Fluid Flow, 3rd edn, revised. McGraw-Hill Series in Mechanical Engi-
neering (2005)
12. Cebeci, T., Cousteix, J.: Modeling and Computation of Boundary-Layer Flows, 2nd edn. Hori-
zons Publishing, Long Beach and Springer, Berlin (2005)
13. Schlichting, H., Gersten, K.: Boundary Layer Theory, 8th edn. Springer, Berlin (2000)
14. Hirschel, E.H., Cousteix, J., Kordulla, W.: Three-Dimensional Attached Viscous Flow.
Springer, Berlin (2014)
15. Schlichting, H., Truckenbrodt, E.: Aerodynamik des Flugzeuges, vol. 1 and 2. Springer, Berlin
(1959), also: Aerodynamics of the Aeroplane, 2nd edn (revised). McGraw Hill Higher Educa-
tion, New York (1979)
16. Küchemann, D.: The Aerodynamic Design of Aircraft. Pergamon Press, Oxford: also AIAA
Education Series, p. 2012. AIAA, Reston (1978)
17. Anderson Jr., J.D.: Fundamentals of Aerodynamics, 5th edn. McGraw Hill, New York (2011)
18. Drela, M.: Flight Vehicle Aerodynamics. The MIT Press, Cambridge (2014)
19. Rossow, C.-C., Wolf, K., Horst, P. (eds.): Handbuch der Luftfahrzeugtechnik. Carl Hanser
Verlag, München (2014)
26 1 Introduction

20. Rizzi, A., Oppelstrup, J.: Aircraft Aerodynamic Design with Computational Software.
Cambridge University Press, Cambridge (2020)
21. Körner, H.: Einleitung. In: Rossow, C.-C., Wolf, K., Horst, P. (eds.) Handbuch der Luft-
fahrzeugtechnik, pp. 25–43. Carl Hanser Verlag, München (2014)
22. Jefferys, D.: Personal communication (2019)
23. Breitsamter, C.: Nachlaufwirbelsysteme großer Transportflugzeuge - Experimentelle Charak-
terisierung und Beeinflussung (Wake-Vortex Systems of Large Transport Aircraft—
Experimental Characterization and Manipulation). Inaugural thesis, Technische Universität
München, 2007, utzverlag, München, Germany (2007)
24. Sachs, G.: Flugmechanik. In: Rossow, C.-C., Wolf, K., Horst, P. (eds.) Handbuch der Luft-
fahrzeugtechnik, pp. 255–309. Carl Hanser Verlag, München (2014)
25. Krämer, E.: Kampfflugzeuge. In: Rossow, C.-C., Wolf, K., Horst, P. (eds.) Handbuch der Luft-
fahrzeugtechnik, pp. 113–150. Carl Hanser Verlag, München (2014)
26. Hitzel, S.M.: Personal communication (2019)
27. With permission. https://uhdwallpapers.org/wallpaper/us-navy-502-aircraft44559/(2019)
28. Prandtl, L.:Über Flüssigkeitsbewegung bei sehr kleiner Reibung. In: Proceedings 3rd Interna-
tional Mathematicians Congress, Heidelberg, pp. 484–491 (1904)
29. Goldstein, S.: Fluid mechanics in the first half of the century. Annu. Rev. Fluid Mech. Palo
Alto 1, 1–28 (1969)
30. Eckert, M.: The Dawn of Fluid Dynamics. Wiley-VCH, Weinheim (2006)
31. Goldstein, S.: On laminar boundary-layer flow near a position of separation. Q. J. Mech. Appl.
Math. 1, 43–69 (1948)
32. Hirschel, E.H.: Basics of Aerothermodynamics, 2nd edn, revised. Springer, Cham (2015)
33. Oswatitsch, K.: Die Ablösebedingungen von Grenzschichten. In: Görtler H. (ed.), Proceedings
of IUTAM Symposium on Boundary Layer Research, Freiburg, Germany, 1957. Springer,
Berlin, pp. 357–367 (1958). Also: The Conditions for the Separation of Boundary Layers. In:
Schneider, W., Platzer, M. (eds.) Contributions to the Development of Gasdynamics, pp. 6–18.
Vieweg, Braunschweig Wiesbaden, Germany (1980)
34. Elsenaar, A.: Vortex formation and flow separation: the beauty and the beast in aerodynamics.
Aeronaut. J. 615–633 (2000)
35. Hirschel, E.H.: On the creation of vorticity and entropy in the solution of the Euler equations
for lifting wings. MBB-LKE122-AERO-MT-716, Ottobrunn, Germany (1985)
36. Hirschel, E.H.: Evaluation of results of boundary-layer calculations with regard to design
aerodynamics. AGARD R-741, 5-1–5-29 (1986)
37. Eberle, A., Rizzi, A., Hirschel, E.H.: Numerical Solutions of the Euler Equations for Steady
Flow Problems. Notes on Numerical Fluid Mechanics, vol. 34. Vieweg, Braunschweig Wies-
baden (1992)
38. Eichelbrenner, E.A.: Three-dimensional boundary layers. Annu. Rev. Fluid Mech. Palo Alto
5, 339–360 (1973)
39. von Kármán, T.: Aerodynamics—Selected Topics in the Light of Their Historical Development.
Cornell University Press, Ithaka (1954)
40. Rizzi, A., Hirschel, E.H.: General developments of numerical fluid mechanics until the middle
of the 20th century. In: Hirschel, E.H., Krause, E. (eds.) 100 Volumes of Notes on Numerical
Fluid Mechanics and Multidisciplinary Design, NNFM100, pp. 61–76. Springer, Berlin (2009)
41. Bloor, D.: The Enigma of the Aerofoil-Rival Theories in Aerodynamics, 1909–1930. The
University of Chicago Press, Chicago (2011)
42. Kutta, M.W.: Auftriebskräfte in strömenden Flüssigkeiten. Illus. Aeronaut. Mitt. 6, 133–135
(1902)
43. Joukowski, N.: Über die Konturen der Tragflächen der Drachenflieger. Z. Flugtech. Motor-
luftschiffahrt 1, 281–284 (1910); 3, 81–86 (1912)
44. Lanchester, F.W.: Aerodynamics, London (1907) and Aerodonetics, London (1908)
45. Prandtl, L.: Tragflügeltheorie, I. und II. Mitteilung. Nachrichten der Kgl. Ges. Wiss. Göttingen,
Math.-Phys. Klasse, 451–477 (1918) and 107–137 (1919)
References 27

46. Lighthill, M.J.: Introduction boundary-layer theory. In: Rosenhead, L. (ed.) Laminar Boundary
Layers, pp. 46–113. Clarendon Press, Oxford (1963)
47. Hirschel, E.H.: Vortex flows: some general properties, and modelling, configurational and
manipulation aspects. AIAA-Paper 96–2514 (1996)
48. Hess, J.L.: Panel Methods in Computational Fluid Dynamics. Annu. Rev. Fluid Mech. Palo
Alto 22, 255–274 (1990)
49. Weiland, C.: A comparison of potential- and Euler-methods for the calculation of 3-D super-
sonic flows past wings. In: M. Pandolfi (ed.), Proceedings of the 5th GAMM-Conference on
Numerical Methods in Fluid Mechanics, Rome, October 5–7, 1983. Notes on Numerical Fluid
Mechanics, vol. 7, pp. 362–369. Vieweg-Verlag, Braunchweig Wiesbaden (1983)
50. Pulliam, T.H.: A computational challenge: Euler solution for ellipses. AIAA-Paper 89–0469
(1989)
51. Vos, R., Farokhi, S.: Introduction to Transonic Aerodynamics. Springer Science+Business
Media, Dordrecht (2015)
52. Leschziner, M.: Statistical Turbulence Modelling for Fluid Dynamics—Demystified. An Intro-
ductory Text for Graduate Engineering Students. Imperial College Press, London (2016). ISBN
978-1-78326-660-9
53. Durbin, P.A.: Some recent developments in turbulence closure modelling. Annu. Rev. Fluid
Mech. Palo Alto 50, 77–103 (2018)
54. Bush, R.H., Chyczewski, T.S., Duraisamy, H., Eisfeld, B., Rumsey, C.L., Smith, B.R.: Rec-
ommendations for future efforts in RANS modeling and simulation. AIAA Paper 2019–0317,
(2019)
55. Rizzi, A., Luckring, J.M.: Evolution and use of CFD for separated flow simulations relevant to
military aircraft. Paper presented at the AVT-307 Symposium on Separated Flow: Prediction,
Measurement and Assessment for Air and Sea Vehicles, Trondheim, Norway 07–09 October
2019. STO-MP-AVT-307-11 (2019)
56. Spalart, P.R.: Detached-eddy simulation. Annu. Rev. Fluid Mech. Palo Alto 41, 181–202 (2009)
57. Mockett, C., Haase, W., Schwamborn, D. (eds.): Go4Hybrid: Grey Area Mitigation for Hybrid
RANS-LES Methods. Notes on Numerical Fluid Mechanics and Multidisciplinary Design, vol.
134. Springer, Cham (2018)
58. Bier, N., Keye, S., Rohlmann, D.: Advanced design approach for a high-lift wind tunnel model
based on flight data. In: Radespiel, R., Niehuis, R., Kroll, N., Behrends, K. (eds.) Advances in
Simulation of Wing and Nacelle Stall. Proceedings of Closing Symposium of the DFG Research
Unit FOR 1066, Dez. 1–2, 2014, Braunschweig, Germany. Notes on Numerical Fluid Mechan-
ics and Multidisciplinary Design, NNFM131, pp. 337–350. Springer International Publishing
Switzerland, Cham (2016)
59. Oberkampf, W.L., Trucano, T.G.: Verification and validation in computational fluid dynamics.
Prog. Aerosp. Sci. 38, 181–274 (2002)
60. Luckring, J.M., Boelens, O.J.: A unit-problem investigation of blunt leading-edge separation
motivated by AVT-161 SACCON Research. RTO-MP-AVT-189, 27-1–27-27 (2011)
61. Tomac, M.: Towards automated CFD for engineering methods in aircraft design. Doctoral
thesis, KTH Royal Institute of Technology, Rep TRITA-AVE 2014:11, Stockholm, Sweden
(2014)
Chapter 2
Separation: Some Relevant
Boundary-Layer Properties, Interaction
Issues, and Drag

This chapter is devoted to an introduction to some of the concepts used when dealing
with separated and vortical flow.
We note first that the flow problems we are dealing with in general are Galilean
invariant [1]. Therefore we can consider a flight vehicle in our mathematical models
and in ground simulation (computational simulation, ground-facility simulation) in
a fixed frame with the air-stream flowing past it. In the reality the vehicle moves
through the—quasi-steady and quasi-uniform—atmosphere.
In passing we also note that usually the Eulerian description of the flow is
employed, which in contrast to the Lagrangian description considers the flow at
every fixed point in space as steady or unsteady flow. For the different possible
formulations of the governing equations see, e.g., [2, 3].
The flow field in our perceived reality (Model 1 in Sect. 1.5) is a one-domain
entity. In aerodynamics and fluid mechanics, however, that entity is differentiated
into several domains: the uniform far field, the inviscid flow field past the vehicle,
the boundary layer flow at the vehicle’s surface, the separated flow, the wake.
The classical one-domain models to describe inviscid flow fields are the modeled
potential flow (Model 4) and the discrete modeled Euler flow (Model 8). General
one-domain approaches to describe viscous flow fields are the Navier–Stokes (NS)
methods (Model 9), the Reynolds-Averaged Navier–Stokes (RANS) methods (Model
10), and the scale-resolving methods (Model 11).
Applying a boundary-layer method (modeled viscous flow, Model 2) means work-
ing with a two-domain model, because always an inviscid flow model—first—has
to be employed. The recently evolved hybrid RANS/LES methods are two-domain
models, too. Strong interaction models are three-domain approaches, see, e.g., [4,
5].
Considering the two-domain model “boundary layer flow and inviscid flow” leads
to the topic of interaction. In this chapter we discuss the most important types of
interaction between boundary-layer flow and inviscid flow.

© Springer-Verlag GmbH Germany, part of Springer Nature 2021 29

E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_2
30 2 Separation: Some Relevant Boundary-Layer Properties, Interaction Issues, and Drag

First, however, we treat some boundary-layer properties, relevant in particular

regarding separation. The chapter closes with a consideration of drag. The most
important drag generating mechanisms are sketched as well as the vehicle shapes
employed to handle them.

2.1 Relevant Boundary-Layer Properties

We first introduce the concepts of intensity and strength of a boundary layer:

– With intensity we name the amount of the local vorticity content |Ω| of the bound-
ary layer under consideration. In our case this means the local vorticity content of
the boundary layer at separation. The reader will find the details of the concept of
the local vorticity content in Chap. 4.
– The strength of a boundary layer has nothing to do with the local vorticity content.
As strength we define the average surface-tangential momentum < ρ u 2 >, which
a boundary layer carries. The average momentum of each of the two involved
boundary layers comes into play in particular in squeeze-off (ordinary) separation.
Flow-off separation happens at sharp wing trailing edges, but also at sharp leading
edges in the case of highly swept wings.1 Always two boundary layers are involved.
Generally their strengths are not very different from each other. Note however that
locally the flow-off directions at the upper and the lower side of a finite-span lifting
wing’s trailing edge are not the same, Sect. 6.2.
Ordinary or squeeze-off separation is what happens at regular surface portions,
when an adverse pressure gradient leads to separation. That too always involves two
boundary layers, which squeeze each other off the surface.
In two-dimensional flow the boundary layer in main flow direction, the primary
boundary layer, in general has a larger strength than the other, the secondary one,
Fig. 1.8. The secondary boundary layer is that in the separation zone, which usually
has a low vorticity content and a low strength. This boundary layer in a sense is an
induced one, induced by the separation of the primary boundary layer.
The situation is different in three-dimensional flow, in particular, if open-type sep-
aration is present, Chap. 7. Then the two boundary layers, which squeeze each other
off the surface—creating the primary separation line—can have similar strengths.
The situation becomes more complicated, if secondary and even tertiary sepa-
ration and attachment patterns, as well as flow unsteadiness are present. Then this
characterization of the involved boundary layers may become questionable.
The effect of the average surface-tangential momentum—the strength of the
boundary layer—is explained best with the separation disposition of the primary
boundary layer in two-dimensional flow. This boundary layer can negotiate an
adverse pressure gradient the better, the more surface-tangential momentum it carries
[5].

1 Regarding the geometry of actual trailing and leading edges see Chap. 6.
2.1 Relevant Boundary-Layer Properties 31

Fig. 2.1 Schematic of two-dimensional tangential velocity velocity profiles u(y) of a laminar and
b turbulent flow [5]

The classical example is that a laminar boundary layer will separate earlier with a
given adverse pressure gradient than a turbulent one. The different tangential velocity
profiles u(y) in Fig. 2.1 point to the different surface-tangential momentum flux
properties of laminar and turbulent boundary layers, provided that the density profiles
are similar in both cases.
A measure of the surface-tangential momentum flux is the momentum thickness
δ2 —better termed the momentum-loss thickness—
 y=δ  
ρu u
δ2 = 1− d y. (2.1)
y=0 ρe u e ue

Here u = u (y) is the surface-tangential flow component in the boundary layer, u e

that at the edge (e) of it, ρ = ρ (y) the density, and ρe the density at the edge, y is
the surface-normal coordinate.2
Equation (2.1) indicates that the fuller a tangential boundary-layer velocity pro-
file u(y) is—with a constant or nearly constant density profile—the smaller is the
momentum deficit compared to the inviscid boundary-layer edge flow. The density
profile ρ(y) matters in high-speed flow with actively or passively (thermal radia-
tion cooling [6]) cooled vehicle surfaces. It matters also, if (surface) heating of the
boundary layer is present, for instance due to a surface heat exchanger.
We consider now what is influencing (a) the shape of the tangential boundary-layer
velocity profile, and (b) the shape of the boundary-layer density profile.
– (a) Of course, the tangential velocity profile of a turbulent boundary layer is fuller
than that of a laminar one, Fig. 2.1. But both the laminar and the turbulent velocity
profile can be influenced by several means. This can be explained with the help of

2 This momentum thickness concept does not hold for three-dimensional boundary layers. There
the momentum-flow displacement thickness is the physically meaningful concept [5]. Nevertheless,
the two-dimensional momentum thickness concept suffices for our considerations, as long as the
treated flow is not too strongly three-dimensional.
32 2 Separation: Some Relevant Boundary-Layer Properties, Interaction Issues, and Drag

Fig. 2.2 Schematic of a boundary-layer velocity profile u(y) without point of inflection: 1 and 2,
and with point of inflection: 3 [5]

the wall compatibility condition [5]. We do not show the derivation but give the
main results.
The wall compatibility condition basically indicates via the second derivative of
the tangential velocity profile at the wall, ∂ 2 u/∂ y 2 | y=0 , whether the tangential
boundary-layer velocity profile has a point of inflection or not, Fig. 2.2.
This usually is seen only in view of the stability and transition properties of a
laminar boundary layer. A point of inflection is present if ∂ 2 u/∂ y 2 | y=0 > 0. This
happens if an adverse pressure gradient is given (the classical consideration), or
surface-normal injection (blowing), or heating of the boundary layer [5].
If the three entities are simultaneously present, they can enlarge or reduce the
magnitude of ∂ 2 u/∂ y 2 | y=0 . The wall temperature via the inverse of the viscosity
reduces the magnitude, if the wall is hot, and enlarges it, if the wall is cold.
The second derivative on the other hand becomes negative with a favorable pres-
sure gradient, with suction, and with surface cooling.
Here we take a broader view and say that the presence of a point of inflection of
u(y) means a less full tangential velocity profile and hence a reduction of the tan-
gential momentum flux, profile 3 in Fig. 2.2. This effect is the stronger the larger
∂ 2 u/∂ y 2 | y=0 is.
If the second derivative is negative, the velocity profile is fuller, profiles 2 and 1 in
Fig. 2.2. The momentum flux is the more enlarged, the more negative the second
derivative at the wall is. (Profile 2 is the classical zero pressure gradient Blasius
boundary-layer profile.)
All this holds for both laminar and turbulent boundary layers, even if in Fig. 2.2
the tangential velocity profiles have been sketched only for the former.
However, we have to note that we tacitly assumed hydraulically smooth surfaces,
i.e., surfaces with no roughness, respectively roughness of sub-critical size. Super-
critical roughness of a surface influences the stability and the transition behavior
of laminar boundary layers. (Turbulence tripping, for instance, is applied in low-
Reynolds number wind-tunnel experiments, see, e.g., [7].)
Turbulent boundary layers are strongly affected by super-critical roughness, show-
ing a sizeable increase of the skin friction, and, where it applies, also of the surface
2.1 Relevant Boundary-Layer Properties 33

heat transfer. For attached turbulent boundary layers much knowledge is available
[4, 8]. Regarding the separation behavior of boundary layers on surfaces with
super-critical roughness, the picture is not unambiguous, see for instance the cor-
responding discussions in [9].
– (b) Regarding the density profile ρ(y), we also follow the discussion in [5]. In
attached viscous flow past a flat or nearly flat surface the gradient of the static
pressure normal to the wall is small and in the large Reynolds number limit of
flat-plate flow goes to zero.3 This means that in the direction normal to the wall
the pressure is constant—or nearly constant—and equal to that of the external
inviscid flow4

p = pe . (2.2)

Consequently, with the equation of state p = ρ R T , we have in the boundary layer

ρ T = ρe Te = constant, (2.3)

and hence it holds the proportionality

1
ρ∝ . (2.4)
T
This indicates that a hot wall leads to a small density at and above the wall. At
a cold wall the density is large. In this case the average tangential momentum
flux < ρ u 2 > is larger than in the hot-wall case. The density profile ρ(y) hence
must be regarded in Eq. (2.1), if a wall-normal temperature gradient exists in the
boundary layer.
Regarding the separation inclination of a boundary-layer, the magnitude of the
Reynolds number is important, too. This can qualitatively be understood with the
help of the empirical separation criterion of B. Thwaites [10]:
1 du e 2
δ = −0.09. (2.5)
ν dx 1
A deceleration of the external inviscid flow—negative du e /d x, equivalent to a
pressure rise, positive dp/d x—leads to separation, once that criterion is fulfilled.
Because the displacement thickness δ1 is inversely proportional to the Reynolds
number, Appendix A.5.4, the Reynolds number counteracts the flow-deceleration
effect. Increasing the Reynolds number hence reduces the separation tendency.

3 If locally the inverse of the boundary-layer thickness δ is of the order of magnitude of the largest of
the principal surface curvatures radii Ri , the pressure gradient is no more small due to the centrifugal
forces induced by the surface curvature [5]. This is the best known boundary-layer higher-order
effect. It can be treated with second-order boundary-layer theory.
4 External means at the outer edge of the boundary layer.
34 2 Separation: Some Relevant Boundary-Layer Properties, Interaction Issues, and Drag

For turbulent flow the separation criterion of B.S. Stratford, [11], see also [12],
actually shows the same effect, i.e., that an increasing Reynolds number counteracts
the flow-deceleration effect and reduces the separation tendency.
Finally we note that all the above also holds for the main-flow direction, i.e.,
streamwise, velocity profile of three-dimensional boundary layers, at least if the
three-dimensionality is not too strong. Moreover, the three-dimensionality—with a
given cross-flow pressure gradient—is directly influenced by the streamwise pressure
gradient, suction, blowing, cooling and heating. A number of examples is discussed
in [5], notably also a virtual boundary-layer fence due to distributed surface suction.

2.2 Interaction Issues

When we consider separation, either flow-off or ordinary separation, we do that in

boundary-layer terms. That means that we treat a two-domain problem: the boundary
layer flow and the (external) inviscid flow.
If the boundary layer is to serve as phenomenological model of attached vis-
cous flow, and further, if a boundary-layer method is employed for the numerical
simulation of this flow, one has to ask, under what conditions this is permissible.
The boundary-layer concept considers the flow past a body as to be composed of
the external inviscid flow and the boundary layer flow. The inviscid flow transmits its
pressure field to the boundary layer. Two kinds of interaction between the two flow
fields are usually distinguished: weak and strong interaction. Classical boundary-
layer theory mainly treats two-dimensional flow at infinite or semi-infinite bodies,
and if three-dimensional flow is considered, this also usually is not done for bodies
of finite dimensions [4].
Our definition of separation in Sect. 1.3.3 brings in vortex layers and vortices
which carry away from the body surface kinematically active and inactive vorticity.
Hence we have to consider a third kind of interaction, that of the vortex layers and
the vortices with the flow past the body. This kind of interaction we call global
interaction. In this context we also have the issue of the locality principle, which we
discuss in Sect. 2.3.
The three kinds of interaction are characterized now, following the discussion in
[5]:
– Weak interaction: The presence of attached viscous flow virtually changes the con-
tour of the body (displacement effect of the boundary layer). In two-dimensional
flow the displacement δ1 thickness is given by
 y=δ  
ρu
δ1 = 1− d y. (2.6)
y=0 ρe u e

The displacement of the inviscid flow in general is positive (possible exceptions

are cold-wall cases), i.e., the displacement thickness δ1 > 0, which represents a
2.2 Interaction Issues 35

streamline, virtually enlarges the thickness of the body.

In three-dimensional boundary layers δ1 is not found so easily. One has to solve
a first-order differential equation over the body surface [5]. δ1 then represents a
stream surface, which virtually enlarges the volume of the body.
If the displacement effect is small—the Reynolds number is an important factor—,
one speaks of weak interaction between the viscous flow—the boundary layer—
and the inviscid flow past the body. The inviscid flow is only weakly, even negli-
gibly, affected by the presence of the boundary layer.
In boundary-layer computations the virtual thickening of the body can be taken
into account—regarding the external inviscid flow—by adding the displacement
thickness to the body contour, or, more elegantly, by introducing on the body sur-
face an equivalent inviscid source distribution, e.g., the transpiration velocity [5].
This holds for both first-order and second-order boundary-layer computations.
– Strong interaction: Separation, either flow-off separation at trailing edges of air-
foils or wings (also at sharp leading edges of delta wings) or ordinary separation—
the classical separation—, leads to a strong interaction between the original, but
now separating boundary-layer flow and the external inviscid flow. This means
that inviscid flow and viscous flow can no more be treated independently of each
other. This holds locally and downstream of the separation location.
A particular case of strong interaction is shock wave/boundary-layer interaction,
which appears if supercritical flow past the wing is present with a pre-shock Mach
number larger than M f oot ≈ 1.3−1.35 [13]. Then the boundary layer separates
below the shock foot (in two dimensions: bubble separation). This is discussed in
Sect. 9.1. Below that Mach number weak interaction is present without separation.
The separation locations in two or in three dimensions usually can be determined
to a sufficient approximation by boundary-layer computations. The separation
process itself—strong interaction—cannot be described in the frame of classical
boundary-layer theory, it cannot be treated by means of boundary-layer methods
alone. The triple-deck theory, for instance, was devised for such flow situations. It
led to a significant progress in the understanding and the description of separation
and other strong flow interaction problems [5].
NS (Model 9) or RANS (Model 10) methods as one-domain computation
methods—which in practice today are the methods of choice—treat flow sepa-
ration inherently without any regard to the strong interaction which happens there,
if one looks at these phenomenon from the side of boundary-layer theory. However,
if the flow is turbulent, the turbulence modeling problem remains.
– Global interaction: Attached viscous flow on realistic body shapes of finite extent
separates from the body surface basically either by ordinary or by flow-off separa-
tion. Regardless of how and where this happens, kinematically active and inactive
vorticity leaves the surface and then is present in the wake flow.
At lifting large aspect-ratio wings, for instance, this wake is initially a trailing
vortex layer, which soon rolls up to a pair of trailing vortices, the wing vortices.
The kinematically active vorticity in this wake causes a global interaction, which
results in the well understood downwash at the location of the wing and the induced
drag of the lifting wing. The upstream and over-the-wing changes of the flow field,
36 2 Separation: Some Relevant Boundary-Layer Properties, Interaction Issues, and Drag

however, usually are rather small.

This is in contrast to wings with large leading-edge sweep, i.e., delta wings. For a
certain combination of sweep angle and angle of attack a pair of lee-side vortices,
usually accompanied by secondary and even higher-order vortex phenomena, is
present, Sect. 10.2. At the leeward side of the wing, a global interaction happens,
too. The flow field there is altered completely. At the windward side, however, the
flow field generally is not changed much.
Global interaction is described with one-domain methods. For large aspect-ratio
wings these can even be inviscid (Model 4 and Model 8) methods but now in
general they are NS/RANS (Model 9/10) methods. In particular for delta wings
and for fuselages NS/RANS methods are to be employed, even if for pre-design
purposes discrete modeled Euler methods (Model 8) still are in use. Examples of
both are discussed in Chaps. 8 and 10.
This discussion of the three kinds of interaction is supported by a large number of
observations from experimental and theoretical/numerical flow field investigations.
The observations point to the fact that to a certain extent a ‘locality principle’ exists.
The concept of the locality principle was put forward in [14] and later independently
in [15].

2.3 A Special Issue: The Locality Principle

The locality principle means that a local change in body shape, or the separation
of flow—with or without kinematically active and inactive vorticity in the wake—
changes the flow only locally or downstream of that region, respectively the sep-
aration region. Upstream of that region, the changes generally are small, typically
at the lifting wing the downwash at the location of the wing.5 This holds also for
subsonic flows, although mathematically their characteristic propagation properties
are of elliptic nature, such that always a global interaction occurs.
In the one-domain NS/RANS methods (Model 9/10) all three interaction kinds
are described correctly, problems may arise regarding turbulence modeling (RANS)
in the very separation region, and if the wake flow is unsteady. Due to the math-
ematically parabolic character of the boundary-layer equations, the Model 2 level,
however, implicitly assumes semi-infinite or infinite bodies. Therefore with Model
2 approaches one must ask, whether and how the interaction must be taken into
account, either for phenomenological considerations or for simulation purposes. Of
course on that level only the attached part of the flow can be described.
The locality principle is observed to hold in general, except in cases, where massive
separation occurs, or where a structurally unstable topology of the velocity field is
present, Sect. 7.5. If the principle holds, it is possible to study attached viscous flow

5 This contradicts van Dykes statement that a wake exerts a first-order influence even in the flow
upstream [16]. Of course, the integral forces and moments, which the flow exerts on the body, are
affected by separation.
2.3 A Special Issue: The Locality Principle 37

phenomena at bodies of finite length on the boundary-layer level without taking into
account the wake flow. Then it is also permitted, as usually is done, to employ a rather
coarse discretization of the wake domain behind a wing, see in this regard Sect. 5.3.
However, if computations on the Model 2 level are performed, if possible the
global interaction should be taken into account. In the case of large aspect-ratio
lifting wings, for instance, this is done in most methods automatically with the com-
putation of the inviscid flow. In linear (potential, Model 4) methods—today usually
panel methods—an explicitly imposed Kutta condition at the trailing edge serves
this purpose. In Euler methods (Model 8), it is the implicit Kutta condition, which
is present at sharp trailing edges. In all these cases, kinematically active vorticity in
one or the other form is present in the wing’s wake.

2.4 Aspects of Drag

In this section we shortly discuss some aspects of the aerodynamic drag, which
aircraft experience. At the center of the discussion is the need to overcome the
so called drag divergence, which occurs when the vehicle enters the transonic flight
regime.6 This need has led to some distinctive configurational aircraft features. These
in turn govern separated and vortical flow at and behind the aircraft. Of course this
strongly depends on the respective configuration.

2.4.1 Drag Components

The drag of an aircraft is a particular topic. Different concepts exist to break it up

and to analyze it. We look at the aerodynamic effects and list the most important
components of the aerodynamic drag:
– skin-friction drag (Dskin f riction ),

6 When in the late 1930s the potential of airbreathing jet propulsion emerged—recognized at that
time predominantly in Germany [17]—the drag divergence was a matter of very high concern.
In Germany up to 1945 frantic work at research organizations and in aircraft industry was
conducted in order to shift in particular the drag divergence to as high as possible (sub-sonic) flight
Mach numbers. Connected to the supercritical airfoil are the names K.H. Kawalki and B. Göthert,
to the swept wing the names A. Busemann, H. Ludwieg, A. Betz, to the delta wing A. Lippisch,
and to the area rule O. Frenzl. The area rule later was confirmed with the equivalence theorem by
F. Keune and K. Oswatitsch (1953). An overview of the work is given in [17] and a very detailed
account in [18].
When the war ended, the outcome of this work, which in Germany barely came to an application,
found its way to the Allies. In [17] a short section “Transfer of the German Aeronautical Knowledge
After 1945” is included, details regarding the American acquisition of data are given in [19]. Large
research and development efforts in a short time then changed the geometrical appearance of all
kind of aircraft. For the USA see in this respect the publication of J.D. Anderson, Jr., “The Airplane:
A History of Its Technology” [20].
38 2 Separation: Some Relevant Boundary-Layer Properties, Interaction Issues, and Drag

Fig. 2.3 Theoretical and experimental pressure distributions at the surface of a Joukowsky airfoil
[21]

– viscous-effects induced pressure drag, usually called form drag (D f or m ),

– induced drag (Dinduced ),
– wave drag (Dwave ),
– interference drag (Dinter f ).
The first two components together constitute the viscous drag Dvisc =
Dskin f riction + D f or m . In cruise it generally amounts up to about half of the whole
aerodynamic drag of an aircraft.
Regarding the drag divergence, the form drag is what needs to be considered
in detail. The point of departure of our considerations is the drag of an airfoil at
subcritical Mach number, hence without supersonic flow pockets and terminating
shock waves. To understand it, consider Fig. 2.3. The theoretical pressure distribution
was found with linear potential theory (Model 4). The airfoil has a trailing edge with a
non-zero trailing-edge angle. The flow leaves the trailing edge in bi-sector direction.
This means a corner flow on both sides and therefore a stagnation point each. That is
clearly indicated with the pressure coefficient being c p = 1 at x/L = 1. The airfoil
has lift but no drag.
The agreement between the theoretically and the experimentally obtained pressure
coefficient is very good up to x/l ≈ 0.8. That means that a weak interaction only is
present between the boundary layer and the external inviscid flow. The disagreement
at 0  x/l  0.3 is not explained. Laminar-turbulent transition over a separation
bubble is the likely cause.
2.4 Aspects of Drag 39

At x/l  0.8 the theoretical and the experimental result strongly disagree. The
boundary layers at the upper and the lower side in the reality of the experiment due
to their displacement properties prevent the recompression to the stagnation point.
At the trailing edge we now observe c p ≈ 0.15. The result is the form drag, which
always is present also in flow-off separation. We too see that this kind of separation
means that a strong interaction between the boundary layers and the external inviscid
flow occurs.
This result holds for the flow at the aft end of any body. At a round end of a
fuselage of course ordinary separation occurs. Topologically it can be very complex
compared to the flow-off separation. This also holds for the separation at the in reality
slightly blunted trailing edge of a wing, Sect. 6.3.

2.4.2 The Drag Divergence

The drag coefficient of a wing or fuselage is almost constant over the Mach number
range from small to rather large subsonic free-stream Mach numbers. This is in
contrast to the lift coefficient of an airfoil or wing, where a clear dependence on the
Mach number is present. In linear potential theory (Model 4) that is modeled with
the Prandtl–Glauert rule [3, 21].
Figure 2.4 over a large Mach-number span shows the almost constant drag coeffi-
cient C D of airfoils of different thickness. We observe, however, that above a certain
∗
Mach number M∞ the drag coefficient strongly begins to rise. The rise begins the
earlier, the thicker the airfoil is. The phenomenon is called the transonic drag diver-
gence. Accordingly we get a dip in the lift coefficient and also in the pitching moment
coefficient (not shown).
The cause of the drag divergence, which directly concerns the zero-lift drag D0
only, can be sketched as follows.7 (For details see Sect. 9.1.) Above the critical
Mach number a pocket of supersonic flow appears over the airfoil’s surface. This
pocket generally is terminated by a shock wave. (The shock-free airfoil is a special
topic.) The resulting shock-wave/boundary-layer interaction—a strong-interaction
phenomenon—causes an extra thickening of the down-stream boundary layer.8 The
enlarged displacement thickness enlarges the form drag. This and the wave-drag
increment due to the shock wave, the boundary-layer decambering and the associated
shock-wave decambering combined, Sect. 6.1, cause the transonic drag divergence
and the lift divergence.

7 The zero-lift drag is the drag of the aircraft without the induced drag.
8 The shock wave terminates orthogonally to the airfoil’s surface, although at the boundary-layer
edge particular phenomena can be present. For all pre-shock Mach numbers this leads to a reduction
of the unit Reynolds number and hence to a thickening of the boundary layer.
40 2 Separation: Some Relevant Boundary-Layer Properties, Interaction Issues, and Drag

Fig. 2.4 Drag coefficients C D of symmetric airfoils (NACA—00xx—1, 13 30) of different thick-
ness ratios (6–18 per cent) at α = 0 as function of the free-stream Mach number M∞ [21]. Mea-
surements of B. Göthert, DVL 1941 [22]

The drag-divergence Mach number Mdd is higher than the (lower9 ) critical Mach
∗
number M∞ . This means that the effect in a sense is delayed. Different definitions
are in use of the drag-divergence Mach number, we just point to [23].
One means to shift the drag-divergence Mach number Mdd to Mach numbers as
high as possible—and also to cap the drag—is evident when looking at Fig. 2.4. The
thinner the airfoil, the lower is the excess speed past it (relative to the free-stream
speed), and the higher is Mdd . This leads to the concept of the thin wing. (Usually
this is a small aspect-ratio wing.)
The thin-wing concept is limited to high-performance aircraft, the F-104
Starfighter being a prominent example. For transonic transport aircraft, which carry
the engines at the wings, as well as the fuel in them, the thin wing is no option. The
wing needs enough spar height in order to bear forces and moments. The now com-
mon solution is to back-sweep the wing. This in effect means that the free-stream
“sees” a thinner wing. (The same effect is present at the forward-swept wing.)
The wing section, the airfoil, moreover can be shaped in a way, that with a given
thickness ratio the excess speed is reduced. This is the concept of the supercritical

9 In the literature often the lower critical Mach number∗ is distinguished from the upper critical
M∞,l
Mach number M∞,u ∗ . At M ∗ supersonic flow begins to appear at the body, at M ∗
∞,l ∞,u the flow past
the body is fully supersonic, at a blunt-nosed body except for the subsonic region at the nose.
2.4 Aspects of Drag 41

airfoil/wing. Since long the combination of the swept wing and the supercritical
airfoil is the rule.
For supersonic flight the rule today is a wing with subsonic leading edge, i.e., the
delta wing with its different shapes. Subsonic leading edge means that the apex half
angle is smaller than the angle of the free-stream Mach cone μ∞ (sinμ∞ = 1/M∞ ).
This amounts to a sweep angle ϕ0 < 90◦ − μ∞ .
An aircraft is composed of subsystems. Geometrically these are the wing, the
fuselage, the tail unit and the nacelles/engines. In supercritical flight interference
effects between these subsystems are present, which go beyond the low-speed effects.
To minimize these interference effects, the distribution of the cross-section area of
the whole aircraft in the main longitudinal axis (free-stream) direction has to be as
smooth as possible. This is the so called area rule.
What have these shape particularities of transonic and supersonic aircraft to do
with our topic, separated and vortical flow in aircraft aerodynamics? We give very
short sketches only. The reader will find the details in the later chapters.
– The wing sweep, either backward or forward, leads to distinctly different properties
of the upper and lower boundary layers and the near-wake flow, the trailing vortex
layer, behind the trailing edge. Even if earlier problems with flow separation,
which led to the application of boundary-layer fences, have been overcome, the
stall behavior is still a topic.
– Regarding delta wings or canards of hybrid wings with their highly swept leading
edges, the appearance of the lee-side vortices, Fig. 1.10b in Sect. 1.3.2, is a very
prominent topic of this book. The provision of general flight quality and maneuver-
ability demands a deep understanding of the lee-side flow, as well as high-fidelity
aerodynamic design and verification tools.
– The area rule finally demands appropriate engine-nacelle locations at the wing as
well as special geometrical features of the fuselage. In particular at high angles of
attack separation phenomena at the nacelles affect the wing flow and separation
behavior.

2.5 Problems

Problem 2.1 Assume an airfoil at a flight altitude of H = 10 km as to be a flat

plate with a chord length of c = 5 m. Determine the wall shear stress at the locations
x/c = 0.5 and 1.0 and also the displacement thicknesses δ1 at the two locations. At
x/c = 1 strong interaction effects are neglected. Employ the approximate relations
given in Appendix A.5 for the Mach numbers M∞ = 0.5 and 0.8 and at the wall the
recovery temperature. Assume laminar flow throughout. The boundary-layer edge
values can be approximated with the free-stream values.

Problem 2.2 Repeat Problem 2.1 for turbulent flow.

42 2 Separation: Some Relevant Boundary-Layer Properties, Interaction Issues, and Drag

Problem 2.3 Repeat Problem 2.1 for a wall temperature twice as high as the recovery
temperature.
Problem 2.4 Repeat Problem 2.2 for a wall temperature twice as high as the recovery
temperature.
Problem 2.5 Give a short summary of the results of the Problems 2.1–2.4.
Problem 2.6 The drag coefficient is not the same as the drag. Get from Fig. 2.4 for
the section drag coefficient Cd for the Mach numbers M∞ = 0.2, 0.4 and 0.8. Assume
Cd to be constant up to the critical Mach number. Assume flight at H = 10 km, a
reference area Ar e f = 1 m2 , and compute the drag for the three Mach numbers. How
does the drag behave?
Problem 2.7 The 6 per cent thick airfoil at α = 2◦ at M = 0.3 has a section lift
coefficient Cl = 0.2. How large is the section lift coefficient at the Mach numbers
0.2, 0.4, 0.8? Remember the Prandtl–Glauert rule.
Problem 2.8 How large is the actual lift of the airfoil from Problem 2.7 at the three
Mach numbers. Use the flight parameters from Problem 2.6.
Problem 2.9 Give a short summary of the results of the Problems 2.6–2.9.

References

1. Shapiro, A.H.: Basic equations of fluid flow. In: Streeter, V.L. (ed.) Handbook of Fluid Dynam-
ics. McGraw-Hill Book Company, New York, pp. 2-1–2-19 (1961)
2. Bird, R.B., Stewart, W.E., Lightfoot, E.N.: Transport Phenomena, 2nd edn. Wiley, New York
(2002)
3. Anderson Jr., J.D.: Fundamentals of Aerodynamics, 5th edn. McGraw-Hill Book Company,
New York (2011)
4. Schlichting, H., Gersten, K.: Boundary Layer Theory, 8th edn. Springer, Berlin (2000)
5. Hirschel, E.H., Cousteix, J., Kordulla, W.: Three-Dimensional Attached Viscous Flow.
Springer, Berlin (2014)
6. Hirschel, E.H.: Basics of Aerothermodynamics, 2nd edn, revised. Springer, Cham (2015)
7. N.N.: Boundary-layer simulation and control in wind tunnels. AGARD-AR-224 (1988)
8. Cebeci, T., Cousteix, J.: Modeling and Computation of Boundary-Layer Flows, 2nd edn. Hori-
zons Publishing, Long Beach and Springer, Berlin (2005)
9. Knopp, T., Eisfeld, B., Calvo, J.B.: A new extension for k-ω turbulence models to account for
wall roughness. Int. J. Heat Fluid Flow 30, 54–65 (2009)
10. Thwaites, B.: Approximate calculation of the laminar boundary layer. Aeronaut. Q. 1(3), 245–
280 (1949)
11. Stratford, B.S.: The prediction of separation of the turbulent boundary layer. J. Fluid Mech.
5(1), 1–16 (1959)
12. Cebeci, T., Mosinskis, G.J., Smith, A.M.O.: Calculation of separation points in incompressible
turbulent flow. J. Aircr. 9(9), 618–624 (1972)
13. Délery, J.: Transonic shock-wave boundary-layer interactions. In: Babinsky, H., Harvey, J.K.
(eds.) Shock-Wave Boundary-Layer Interactions, pp. 5–86. Cambridge Universtity Press, Cam-
bridge (2011)
References 43

14. Hirschel, E.H.: On the creation of vorticity and entropy in the solution of the Euler equations
for lifting wings. MBB-LKE122-Aero-MT-716, Ottobrunn, Germany (1985)
15. Dallmann, U., Herberg, T., Gebing, H., Su, W.-H., Zhang, H.-Q.: Flow-field diagnostics: topo-
logical flow changes and spatio-temporal flow structure. AIAA Paper 95–0791, (1995)
16. Van Dyke, M.: Perturbation Methods in Fluid Mechanics. Academic, New York (1964)
17. Hirschel, E.H., Prem, H., Madelung, G. (eds.): Aeronautical Research in Germany–from Lilien-
thal until Today. Springer, Berlin (2004)
18. Meier, H.U. (ed.): German Development of the Swept Wing–1935–1945. Library of Flight.
AIAA, Reston (2010)
19. Samuel, W.W.E.: American Raiders: The Race to Capture the Luftwaffe’s Secrets. University
Press of Mississippi, Jackson (2004)
20. Anderson Jr., J.D.: The Airplane: A History of Its Technology. AIAA, Reston (2002)
21. Schlichting, H., Truckenbrodt, E.: Aerodynamik des Flugzeuges, vol. 1 and 2, Springer, Berlin
(1959), also: Aerodynamics of the Aeroplane, 2nd edn (revised). McGraw Hill Higher Educa-
tion, New York (1979)
22. Göthert, B.: Hochgeschwindigkeitsuntersuchungen an symmetrischen Profilen mit verschiede-
nen Dickenverhältnissen im DVL-Hochgeschwindigkeits-Windkanal (2.7 m ∅) und Vergleich
mit Messungen in anderen Windkanälen. ZWB/FB 1506 (1941)
23. Vos, R., Farokhi, S.: Introduction to Transonic Aerodynamics. Springer Science+Business
Media, Dordrecht (2015)
Chapter 3
Elements of Vortex Theory

This chapter is devoted to a brief introduction to elements of vortex theory, which

are of relevance for the topic of the book. We shortly outline each item, give basic
mathematical descriptions, illustrating sketches, and provide supporting references.
General literature recommended for a deeper study of the topic as such are the
monographs by, e.g., P.G. Saffmann (1992) [1], H.J. Lugt (1996) [2] and J.-Z. Wu,
H.-Y. and M.-D. Zhou (2006) [3].
Regarding monographs on aerodynamics in view of potential-flow theory the
reader will find ample material in, e.g., H. Schlichting and E. Truckenbrodt
(1959/1979) [4], J.D. Anderson, Jr. (2011) [5], M. Drela (2014) [6] and in the Hand-
buch der Luftfahrzeugtechnik by C.-C. Rossow, K. Wolf and P. Horst (2014) [7].

3.1 Introduction

It is interesting to note that the circulation theory, i.e., the vortex theory of the lifting
airfoil, evolved with the work of M.W. Kutta and N. Joukowski in the years 1902 to
1912, see the sketch of its general development in Sect. 1.4. Their work regarded the
lifting airfoil, and that at a time, when human heavier-than-air flight barely was at
its beginning [8, 9].
For quite a time then all aerofoil data still had to be obtained from experimental
work and fitted to wings with different aspect ratios, planforms, etc., by empirical
formulae based on past experience with other airfoils and wings.

© Springer-Verlag GmbH Germany, part of Springer Nature 2021 45

E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_3
46 3 Elements of Vortex Theory

Fig. 3.1 Various models of finite-span wing theory consisting of different vortex line arrangements.
a lifting-line model b wing-sheet model, and c vortex-lattice model, after [2]

3.1.1 Models of Finite-Wing Theory of Flight

The Laplace equation describes incompressible, irrotational and inviscid (potential)

flow around the wing, but surprisingly no global force acts on it, because in a very
basic sense no interesting motions can take place in incompressible homogeneous
unbounded fluids without vorticity, Sect. 1.4. This is the Laplace equation, Model 3
of Table 1.3, and the d’Alembert paradox.
With the independent work of F.W. Lanchester and L. Prandtl, which culminated
in Prandtl’s lifting-line model, a theoretical model of the lifting wing was created.1
It consisted of a system of vortices that imparted to the surrounding air a motion
similar to the actual flow, and that yielded a force equivalent to the lift known to be
created. The vortex system consisted of four main parts: the bound vortex, the two
trailing vortices and the starting vortex. Each of these may be treated separately but
they are all component parts of one whole acting together, see Fig. 3.3.
We use this theory to illustrate the connection of lift to vorticity, its creation and
the dynamics that vortices undergo in the various Models 4, 6, 8 and 10 of lifting
wing flow.
Figure 3.1 displays three variants of Model 4 for finite-span wing theory consisting
of different vortex line arrangements. The first is Prandtl’s lifting-line model that
forms the basis for an analytical finite-wing theory, the second, also analytical, is a
wing-sheet or lifting-surface model, and the third, a numerically discrete description
of the wing sheet, is the vortex-lattice model.
Since they do not account for vortices shed from side edges, the first two models in
Fig. 3.1a, b are valid only for the flow sketched in Fig. 1.10a, while the vortex-lattice
model, Fig. 3.1c is applicable for the flows in both Fig. 1.10a, b.
As background to understanding how the Models 4 to 10 of Table 1.3 perform,
the current chapter briefly summarizes some relevant fundamentals of vortex theory,
discussing among others:
1. vortex sheets associated with lifting surfaces,
2. their creation and dynamics including roll-up into vortices associated with the
flows shown in Fig. 1.10a, b,

1 In the literature often the term “Lanchester–Prandtl-Theory” is found, in Germany it reads “Prandtl–

Lanchester-Theory”, see also the remark on Sect. 1.4.

3.1 Introduction 47

3. stability and decay of vortices in the wake behind a wing,

4. vortex-flow technology applied in aircraft design.

3.1.2 Benefits in Studying Vortex Dynamics

The first benefit is overcoming the d’Alembert paradox as Kutta and Joukowsky did.
Model 3, the Laplace equation is linear. Hence vortex singularities can be superposed
on a given solution to obtain desired features, as described further in this chapter.
Secondly the motion of an incompressible fluid can be described either by the
Navier–Stokes equations for the velocity and the pressure (primitive variables), or
by the vorticity equation for the velocity and the vorticity. The two versions are
equivalent.
Historically it has been preferred to describe the flow field in terms of velocity
and pressure, i.e., to solve the Navier–Stokes equations, because the velocity vector
and the pressure are more tangible quantities than the vorticity vector. However, the
alternative description using the vorticity concept has its advantages. In particular
before numerical solutions of the compressible Navier–Stokes equations, Model 9,
became so prevalent in the 1990s, it was thought that solving the vorticity transport
equation offered a simpler approach.
Admittedly, the concept of vorticity is less intuitive than that of the pressure, but
we can argue that the grasping of a physical concept is to a large extent a matter
of usage. Vorticity must be understood from its collective properties, and to convey
such an understanding is the purpose of this chapter.
For instance, vortex patterns can be described and analyzed better by vorticity, or
in the limit by point vortices, since the vorticity field—also—is invariant with respect
to Galilean and rotational transformations (except for a constant). Three-dimensional
vorticity patterns reveal stretching and twisting processes, bringing about local inten-
sification of vorticity. In general vortices serve as a paradigm for the development of
dynamic flow patterns, since their theory is so well advanced.
Perhaps Dietrich Küchemann’s well-known metaphor that ‘vortices are the sinews
and muscles of fluid motion’ captures best the essence of what we mean by pattern
recognition. Vortex theory has become a paradigm for pattern development even in
general.
A better understanding of vortical phenomena helps us to understand and judge the
validity of our computed results, from Models 8 to 10 (Euler, RANS, URANS)—and
of course also Model 11—that we present in later chapters. Here one must caution that
our solutions are to the compressible equations, while the vortex dynamics discussed
in this chapter derive from the incompressible equations.
Keep in mind also that the most complete and elegant description of vorticity,
however, exists for inviscid fluids, since vorticity is here attached to fluid elements
and is not lost through diffusion. In such cases the Lagrangian description is the
natural one and gives the best insight into the essence of vorticity, because it is
conserved and can be traced back to its initial state.
48 3 Elements of Vortex Theory

This chapter therefore provides us with a bridge in our discussion from the fluid-
flow models outlined in Chap. 1 to the computed results of these models presented
and analyzed as Unit Problems in later chapters.

3.2 The Concept of Vorticity

3.2.1 General Remarks

The understanding of vortex theory requires the clarification of two central concepts,
namely, what is a vortex and what is vorticity? Both are elusive concepts, but in
opposite ways. A dictionary easily defines a vortex as the rotation of fluid elements
around a common center, but these are just words without mathematical precision.
In contrast to the vortex concept, the vorticity vector ω has an exact mathematical
definition as the curl of the velocity vector ω = curl v, but its general physical
meaning is difficult to grasp.
The standard explanation is that vorticity is a measure for the angular velocity of a
fluid element, and we see that a vortex requires vorticity for its existence. Metaphor-
ically speaking vorticity is the building agent for vortices, but the reverse statement
that a vorticity field represents a vortex, however, is not always true, since a straight
shear layer, for example a boundary layer, has vorticity but is not a vortex.
However, the vorticity field is a continuum, and to relate the vorticity to the angular
velocity of a fluid element in this continuum, from which it can diffuse to neighboring
elements, causes imaginative difficulties inherent in the meaning of a continuum.
The way to proceed is to make the vortical flow field compatible with the vorticity
by relating the vorticity concept (or any other mathematically defined quantity) to
that of the vortex.
Flow fields with three-dimensional vortices in general have complex structures,
and they can be understood to a large extent by the movement, stretching, and inter-
action of vortex filaments (‘strings’).
Although these processes are nonlinear and, hence, capable of abrupt, drastic
changes within a flow field, they involve solenoidal-field properties along with mea-
sures like circulation that give us relations between vortices and vorticity, which
then makes it easy to study mathematically the generation, behavior, and decay of
vortices.2
In this book we are dealing with vortical patterns seen in computed solutions of
Models 8, 9 and 10, which include the generation, behavior, and the decay of a single
vortex as well as a system of vortices, vortex sheets, shear layers etc. of high and low
kinematically active and/or inactive vorticity content, and we need to make some
physical sense of what we observe. Vorticity, in short, is a crucially important feature
of any shearing motion and provides a path towards further understanding.

2 Solenoidal means source-free, or zero divergence in all points of a vector field.

3.2 The Concept of Vorticity 49

3.2.2 Designations and Definitions

As stated above, a vorticity vector field, ω(x, y, z), is defined as the curl of the
velocity vector
ω = curl v = ∇ × v. (3.1)

Vorticity appears in the decomposition of the velocity field as the rotation term
in the fundamental theorem of kinematics, which states that the velocity of a fluid
element v + dv in a continuum is composed of the rates of translation v, deformation
D · d r and rotation 21 curlv × d r [10].
The vorticity represents a measure of the local rate of rotation of a fluid element.
A vortex in a viscous fluid rotates like a solid body at the axis and thus requires
vorticity. There is no vortex without vorticity.
This seems to be a contradiction, when looking at a potential (point) vortex or a
vortex sheet (slip line or surface). However, the concept of kinematically active and
inactive vorticity content, Chap. 4, permits us to overcome this contradiction.
A vorticity line is a line tangent to the vorticity vector—analogous to the streamline
in a velocity vector field. If the whole field is irrotational except for an isolated
vorticity line, this line is then called a vortex line.
In three-dimensional flows a bundle of vortex lines is enclosed in a vortex tube,
which is a surface comprising all the vortex lines passing through points of a particular
loop as shown in Fig. 3.2.
An isolated vorticity tube is a vortex tube, which incloses a vortex filament [11].
The Rankine vortex is an infinitely long straight vortex filament.

Fig. 3.2 A vortex tube is a surface comprising all the vortex lines passing through points of a
particular loop [3]
50 3 Elements of Vortex Theory

3.3 Stokes’ Theorem and the Concept of Circulation

3.3.1 The Theorem

From the definition ω = curl v it follows that the vorticity field is solenoidal—
div ω = 0—since for any vector field u we have div curl u ≡ 0. The solenoidal
property means that vorticity tubes (like stream tubes) cannot end inside the fluid.
Next, for a general vortex tube, as sketched in Fig. 3.2, composed of the vortex
lines passing through a particular loop C, the solenoidal property div ω = 0 implies
that all along the vortex tube the product of its cross-sectional area with the magnitude
of the vorticity remains constant. To see this, we apply the Divergence Theorem to
this property over the volume V of the fluid, bounded by two different surfaces S1
and S2 , spanning the tube and by the part of the tube between them, and obtain
 
ω · n dS = div ω dV = 0 (3.2)
∂V V

since div ω ≡ 0.
The bounding surface ∂V of the volume comprises not only S1 and S2 , but also
the part of the vortex tube between them, the latter making no contribution to the
left-hand side of Eq. 3.2 since, by the definition of a vortex tube, the vorticity vector
is tangential to it, thus giving ω · n = 0.
It follows that 
ω · n dS = constant, (3.3)
∂V

which gives the useful result that the inner product of S with the magnitude of the
vorticity remains constant along the tube.

3.3.2 Circulation

The concept of circulation was introduced by W. Thomson (Lord Kelvin) in 1869

and is defined in the following way.
The quantity stated in Eq. 3.3 to take everywhere a constant value along a general
vortex tube, is a characteristic property of that vortex tube, called its strength Γ , i.e.,
its circulation.
It is the famous mathematical theorem due to Stokes, which gives a special inter-
pretation of the left-hand side of Eq. 3.3. Stokes’s theorem applies to any surface
S for which we are able to make a consistent, continuously varying selection of a
direction for the unit normal vector n at each point of the surface and for which the
boundary ∂S consists of one or more closed curves C, Fig. 3.2. Stokes’s theorem
states that, for any vector field, and thus in particular for the velocity field v
3.3 Stokes’ Theorem and the Concept of Circulation 51
  
curl v · n dS = ω · n dS = v · d, (3.4)
S S C

where  is the unit vector tangential to the closed curve C. Thus from Eq. 3.3 we
conclude  
ω · n dS = v · d = Γ, (3.5)
S C

that the circulation Γ is the strength of the vortex tube.

The Kutta–Joukowski Theorem, underlying the foundation of the circulation the-
ory of lift for the models in Fig. 3.1, states that an isolated two-dimensional wing
section in an incompressible inviscid flow feels a lift force lu per unit width

lu = ρv ∞ Γ, (3.6)

see also Sect. 3.16.

3.4 Origins of Vorticity

The basic foundation of the finite-span wing theory of flight is the creation of lift
through flow circulation around the wing, that translates into a mathematical model,
in the simplest case into the lifting-line model. In this model a ring vortex is formed,
where one leg is placed inside the wing—the bound vortex—another leg—the start-
ing vortex—is far downstream, connected on both ends by two trailing-vortex legs,
illustrated in Fig. 3.3.
The figure is the epitome of Prandtl’s lifting-line model, which permits to describe
the flow downwash at the location of the wing and the induced drag.
The figure demands a closer inspection. Globally the lifting-line model is a Model
4 (Table 1.3) theory for lifting wings with large aspect ratio. The perceived reality,
Model 1, is reduced in the following way. The trailing vortex layer, Sects. 4.3 and
4.4, has undergone an immediate roll-up. The instantly present trailing vortices are

Fig. 3.3 The bound vortex, the two trailing vortices, and the starting vortex form a closed vortex
ring, which is the simplest model of a lifting wing of finite span, after [2]
52 3 Elements of Vortex Theory

located at the wing tips at a chosen chord location and there they connect with the
bound or lifting vortex. At the wing tips the tip vortices, or more in general the tip-
vortex systems, Sect. 8.4.3, are not present. Hence also not their relevance in view of
the non-linear lift of wings with small aspect ratio, Sect. 8.4.4.3 The reader should
be aware that often in the literature verbally and even conceptually the tip vortices
are mixed up with the trailing vortices.
For the purpose of discussing origins of vorticity, it suffices to explain the flow
mechanism present in the lifting-line model that creates the bound vortex leading to
circulation and lift.
Let us look at various models for wings within the framework of potential-flow
(Model 4) theory. The essence of two-dimensional wing theory contains the idea of
the bound vortex or alternatively a circulation around the wing. In the mathematical
formulation, the bound vortex is represented by a point, i.e., potential vortex located
inside the wing section. In Sect. 4.2.4 we show that the boundary layers at the upper
and the lower side of the airfoil contain the kinematically active vorticity connected
to that vortex.
In an unsteady two-dimensional flow, or in a steady or unsteady three-dimensional
flow, also vortices outside of the wing must be considered. Because of Helmholtz’s
law for persistence of circulation, the vortex lines must be closed by connecting the
starting vortex with the trailing vortices.

3.4.1 The Kutta Condition

In several papers published in the first decade of the 20th century, Kutta and
Joukowsky independently proposed that the lift on an airfoil at incidence in a steady
unseparated flow is given by potential-flow theory with the unique value of the circu-
lation that removes the inverse-square-root velocity singularity at the trailing edge.
This proposal—tantamount to saying that in the unsteady start-up phase the action
of viscosity is such that, in the ultimate steady motion, viscosity can be explicitly
ignored, but is implicitly incorporated in a single edge condition—is known as the
Kutta–Joukowsky hypothesis, commonly called the ‘Kutta condition’.4
Although this suggests the ‘action of viscosity’ as the responsible mechanism,
there is no fully universal convincing physical explanation for the production of the
starting vortex and the generation of the circulation around the wing section. This
issue has been debated for over a hundred years, as described in book-length by D.
Bloor [12].
The most usual explanation, represented following Bloor as The Cambridge
School, is based on the large viscous forces in the boundary layer, associated with
the high velocities round the trailing edge invoked by the no-slip condition on the
surface.

3 The simple lifting-line model anyway does hold only for wings with large aspect ratio, say Λ  3.
4 Regarding the Kutta condition in the reality of aircraft see Chap. 6.
3.4 Origins of Vorticity 53

This line of reasoning infers that circulation cannot be generated—and no aerody-

namic lift can be produced—in an inviscid fluid, because the first Helmholtz theorem,
Sect. 3.7, says that it is not possible to endow a fluid particle with vorticity. Moreover,
Kelvin’s circulation theorem, Sect. 3.8, shows that the circulation around a material
circuit remains zero if initially zero.
The question then arises whether vorticity can be created without violating these
theorems, and without appeal to viscous or barotropic effects.5 Another line of rea-
soning, which we call The Göttingen School, however, argues that local flow accel-
eration is equally important and that this is sufficiently high to account for the failure
of the flow to follow round the sharp trailing edge, without invoking viscosity.
The following two sections present arguments on both sides of the debate whether
the vortex-sheet creation is essentially viscous or inviscid. We note that in Chap. 4
we do not discuss the formation of the starting vortex and the development process
of the circulation around the wing.

3.4.2 The Cambridge School—No-Slip Viscous Mechanism

Fundamentally unsteady, Fig. 3.4 illustrates in more detail the formation process of
the starting vortex [13]. Initially at rest the wing starts to accelerate, and in the first
instant Fig. 3.4a presents a qualitative picture of the streamlines around any cross
section of the wing.
The Cambridge School argues that instead of irrotational flow around the trailing
edge, as in Fig. 3.4b, the no-slip condition in the boundary layer generates vorticity
causing the flow to break away in the starting vortex, Fig. 3.4f.
• Lighthill’s Concept—Viscous Pressure Gradient M.J. Lighthill has devel-
oped a hypothesis for a surface-pressure gradient that determines the precise amount
of vorticity that must be created in the boundary layer, consistent with the circula-
tion around the wing [14]. He considered the two-dimensional unsteady flow over
the airfoil and started from the Navier–Stokes equations in streamline coordinates
(n, s), where s and n are the unit tangent and normal vectors of any streamline.
At t = 0 the airfoil impulsively moves from rest. Since pressure waves travel with
infinite speed in an incompressible fluid, an infinitely thin shear layer develops along
the surface, which reduces the velocity to zero to satisfy the no-slip condition. The
Navier–Stokes equations along the airfoil surface then instantly reduce to (s being
the streamwise, n the surface-normal direction)

1 ∂p ∂ω
=ν ≡ σ, (3.7)
ρ ∂s ∂n
1 ∂p ∂ω
= −ν , (3.8)
ρ ∂n ∂s

5 The barotropic fluid is a fluid, whose density is only a function of pressure.

54 3 Elements of Vortex Theory

Fig. 3.4 Phenomenological sketch of streamlines of inviscid potential and rotational flow around
a wing during its first instants of acceleration from rest [13]: a flow around sharp trailing edge S; b
potential model, flow around S allowed; c compressible Euler equations, flow expands to vacuum in
the limit; d separates with a tangential discontinuity; e rolls up into a vortex; and f is shed drawing
the separation point B to the trailing-edge point S

referred to as Lighthill’s relations. σ is the vorticity flux in chord direction along the
wing’s surface. Note that ν ∂ω = ν ∂∂nu2 , so the right hand side is the normal gradient
2

∂n
of the viscous shear force.
The amount of vorticity that is diffused into the fluid next to the boundary is the
flux of vorticity, σ, being equivalent to the vorticity source strength on the boundary.
According to Lighthill, to fulfill the no-slip condition and the solenoidality of
vorticity, the vorticity flux σ depends only on the pressure change along the wall.
Not everyone, however, agrees with Lighthill’s conclusion. For example H.J. Lugt
[2] disagrees with this creation of the vorticity flux by the pressure gradient, and
argues instead that it are the shear stresses that drive the mechanism.
3.4 Origins of Vorticity 55

3.4.3 The Göttingen School—Inviscid Vortex-Sheet

Mechanism

Let us turn now to the Göttingen School and its arguments for the opposing concept—
an inviscid mechanism. This line of reasoning begins with F. Klein in 1910, and was
nicely described and summarized more recently by P.G. Saffmann [1].
• Klein’s Kaffee-Löffel Experiment In 1910 F. Klein addressed the origin-of
vorticity problem with his Kaffee-Löffel (coffee spoon) experiment [15], further
studied in 1950 by A. Betz [16].
In Klein’s experiment a two-dimensional plate (a line segment) is set in motion
through a perfect incompressible fluid with velocity normal to the plate.
Klein argues that if the plate is removed (by pulling out the coffee spoon) or
dissolved (certainly possible in a gedankenexperiment), then a vortex sheet is left in
the fluid. Non-zero circulation has been created around any of the fluid circuits and
there is no contradiction with the Helmholtz-Kelvin laws, because no fluid particle
has acquired vorticity, and no circulation has been generated about closed contours
lying entirely within the fluid.
The new contours with circulation were not closed beforehand; they intersected
the body (spoon). This is an example in which vorticity has been created as a sheet—
i.e. a singular distribution—by changing the topological properties of the flow.
• Early Computational Experience with Model 8 In 1981 the second author of
this book obtained one of the first numerical solutions of the Euler equations (Model
8) for transonic flow around the ONERA M6 wing, [13], remarkably without the
implementation of a Kutta condition, which is essential in all existing computational
methods solving the potential equations, Methods 4 and 5.
In explaining why his method did not require a Kutta condition, Rizzi applied
arguments, similar to those for the coffee spoon, to what happens when a wing
begins to accelerate from rest.
Consider again Fig. 3.4c at the instant when the wing is first set in motion. The air
over the rear part of the wing section is required to change direction suddenly while
still moving at high speed, the pressure around the trailing-edge point S will drop
and draw towards it air from the upper surface, that then meets the airstream from
the lower surface.
If these two streams adjoin with differing velocities along their interface, they
form a tangential discontinuity, i.e. a vortex sheet, that by self-induction rolls up into
a vortex, Fig. 3.4d. The sense of the shed vorticity acts to impede the velocity around
the trailing edge drawing B towards S in a self-correcting process, Fig. 3.4e, and the
vortex is convected downstream by the surrounding stream, Fig. 3.4f.
The shed vortex is equal in strength and opposite in sense to the circulation
round the wing and is called the starting vortex. Subsequent computations with a
highly refined grid confirmed that these results were adequately resolved and not
grid dependent [17].
• Shear between Two Inviscid Flow Fields Boundary-layer computations of
the first author of this book for finite-span wings led to observations of flow field
56 3 Elements of Vortex Theory

properties at the lower (pressure) and the upper (suction) side of lifting wings. These
and the observations of the second author in his numerical solutions of the Euler
equations for lifting wings led to the concept of the kinematically active and inactive
vorticity content of two-dimensional and three-dimensional shear layers, Chap. 4.
That chapter also presents a thorough and more general analysis of the origin
of vorticity, which for finite-span wings essentially is seen to be due to the second
break of symmetry, with the Kutta condition representing the first break of symmetry,
Sect. 4.3. Examples in the form of Unit Problems are given in Chap. 8 for large aspect-
ratio wings and in Chap. 10 for small aspect-ratio delta-type wings.

3.4.4 Other Origins

In support of counter arguments to the mechanism of no-slip condition, other non-

viscous mechanisms for creating circulation can be mentioned.
• Inviscid Weis-Fogh Mechanism Insect flight usually is thought of as low-
Reynolds-number movement, hence with strong viscous effects, but surprisingly, the
Klein mechanism may work also in an inviscid fluid.
The work of T. Weis-Fogh, [18], on the hovering flight of certain wasps indicates
that it is possible to generate circulation and lift in the complete absence of viscosity
by the alternate separating and folding of two plates, see Fig. 3.5.
Although the total circulation is zero for irrotational flow, according to Kelvin’s
theorem, the two plates have, after separation from each other, the same amount of
circulation but with opposite sign.
Lighthill calls the process clap and fling as the hovering insect moves its wings
in a horizontal plane, and computes the strength of the rolled-up vortex [19]. Here,
in subscribing to this inviscid mechanism, Lighthill plants his feet in both sides of
the controversy, Cambridge versus Göttingen schools of thought.
• Across Shock Waves It must be mentioned, too, that vorticity is generated in
compressible flow, when it passes through a curved shock wave in either the transonic,
supersonic or hypersonic flow regime, producing an entropy-vorticity wake behind

Fig. 3.5 The Weis-Fogh mechanism of circulation generation producing lift, after [2]
3.4 Origins of Vorticity 57

shock
s = const.
ω=0
s = const. shock wake
ω=0

viscous region

Fig. 3.6 Airfoil at supercritical speed [6]: vorticity-free flow ahead of the shock wave, vorticity
field behind the shock wave

it [20]. The connection between entropy and vorticity is the topic of Crocco’s law.
This law is discussed in the following Sect. 3.5.
Here we consider shortly two typical flow situations with shock waves, where
vorticity is created. Figure 3.6 shows an airfoil at supercritical speed. Ahead of the
shock wave, which terminates the supersonic flow pocket at the upper side of the
airfoil, the entropy s is constant, also the vorticity ω. The flow is homentropic.
Behind the shock wave, which is curved, the entropy is not constant, hence vortic-
ity is present in the shock wake. The flow downstream of the shock wave, however,
is isentropic, i.e., the entropy is constant along the streamlines, but differs from
streamline to streamline.
In the figure the boundary layers at the upper and the lower side of the airfoil, as
well as the airfoil’s wake, are marked as viscous region. We argue in Sect. 4.2.3 that
the shock wave’s curvature is such that the produced vorticity flux compensates the
difference of the vorticity fluxes in the boundary layers. In this way the whole wake
of the airfoil, the vorticity or entropy wake in Fig. 4.6, does not carry a kinematically
active vorticity content, as is demanded for the steady flight situation.
We note in addition that in the inviscid picture of the flow, the shock wave terminat-
ing the supersonic flow pocket on the upper side of the airfoil impinges orthogonally
on the airfoil’s surface, Sect. 4.2.3. In the viscous reality a shock-wave/boundary-
layer interaction is present, with potentially large consequences for the airfoil’s or
wing’s performance, Sects. 2.4.2 and 9.1.
The other flow situation is that at a blunt body at supersonic or hypersonic speed,
Fig. 3.7. High supersonic or hypersonic flight vehicles as a rule have blunt noses in
order to cope with the high thermal loads. Re-entry vehicles moreover enlarge the
total drag with the blunt nose, since they essentially fly a braking mission [20]. The
figure shows the bow-shock wave of a blunt body at angle of attack. At the windward
side the shock wave is lying almost parallel to the body surface. The strongly curved
bow shock leads to a distinctive entropy layer, hence a vorticity layer.
The figure shows that the vorticity layer, in terms of the surface tangential velocity
v(n)—n being locally the direction normal to the body surface—has different profiles
at the windward and the leeward side. At the windward side the profile is wake-like,
whereas it is slip-flow boundary-layer like at the leeward side. The different shapes
58 3 Elements of Vortex Theory

Fig. 3.7 Blunt body at angle of attack at high supersonic or hypersonic speed [20]: entropy layers
at the lower and the upper side. Note that the boundary layers are not indicated

are due to the fact that the stagnation point S1 lies below the nose point of the body
at the windward side. At zero angle of attack, the profile is slip-flow boundary-layer
like everywhere.
One effect of interest in this case is the so-called entropy-layer swallowing of the
boundary layer, which leads to a decrease of the boundary-layer thickness, hence an
increase of the thermal load, and moreover influences the laminar-turbulent transition
process [20].

3.4.5 Short Conclusion

In conclusion, the debate about whether the vorticity-origin mechanism is viscous

or inviscid driven has become moot.
Today the overwhelming majority of studies are computed with Methods 9 and 10,
and also 11, because many other boundary-layer phenomena, including separation
are involved and need to be considered. Method 8 is used only for certain specialized
inviscid flow cases.

3.5 Entropy and Total Enthalpy Gradients and Vorticity:

Crocco’s Theorem

Combining the first law of thermodynamics with the momentum equations yields the
Crocco theorem that connects vorticity, a kinematic flow property, to thermodynamic
properties, the specific entropy s and the specific total enthalpy h o :
3.5 Entropy and Total Enthalpy Gradients and Vorticity: Crocco’s Theorem 59

T ∇s = ∇h o − v × ω, (3.9)

valid for steady inviscid adiabatic flow.

Crocco’s theorem shows that whenever an enthalpy or entropy gradient is present,
the flow must be rotational. This explicitly confirms the equivalence between isen-
tropy and irrotationality. The theorem is particularly important for transonic flow
and supersonic/hypersonic blunt-body flow, where a shock wave either terminates a
supersonic region, as indicated in Fig. 3.6, or envelops the body, Fig. 3.7.
Of importance is that the entropy rise over a shock wave goes together with a drop
of the total pressure, see, e.g., [20]. If locally the subscript ‘1’ stands for the flow
ahead of the shock wave, and ‘2’ behind it, we obtain the total-pressure change over
the shock wave to

pt2 = pt1 e−(s2 −s1 )/R , (3.10)

with R being the gas constant.

The total-pressure loss in the context of the topic of our book concerns for instance
the shock-wave decambering effect of airfoils, respectively wings, Sect. 6.1.2, and
also the prerequisites for the compatibility condition at the trailing edge of a lift-
ing wing, Sect. 4.4. Also the performance of aerodynamic control surfaces is to be
considered with regard to a possible total-pressure loss occurring ahead of such a
surface.

3.6 Equations of Transport of Vorticity

M. Drela, [6], derives the Helmholtz transport equation for the behavior of vortic-
ity by formally taking the curl of the momentum equation, using vector identities,
combining and rearranging terms:
     
D ω ω ∇ρ × ∇ p 1 ∇·τ
= ·∇ v+ + ∇ × . (3.11)
Dt ρ ρ ρ3 ρ ρ

The baroclinic source term ∇ρ × ∇p can cause vorticity to appear wherever there
are density and pressure gradients present. However, in isentropic flow, where the
viscous term is negligible, the isentropic f ( p, ρ) = 0 relation holds, the ρ and p
gradients are parallel, and the baroclinic term vanishes.
The simpler incompressible form of Eq. 3.11, resulting from ρ and μ being con-
stant, is
Dω
= (ω · ∇) v + ν∇ 2 ω. (3.12)
Dt

The term (ω · ∇)v on the right-hand side represents vortex tilting and vortex
stretching. Stretching causes a rotating fluid’s vorticity to intensify—think of a pirou-
60 3 Elements of Vortex Theory

etting ice-skater stretching his arms up—when stretched by the components of the
velocity gradient matrix ∇ v, which are parallel to ω itself. However, if ω = 0 to
begin with, then this term is disabled, since there is no initial vorticity to stretch or
tilt.
The Helmholtz vorticity equation 3.11 or 3.12 simplifies greatly for most aerody-
namic problems. These typically have uniform flow and hence ω = 0 upstream, and
their viscous stresses are negligible outside of viscous layers and outside of shock
waves.

3.7 Helmholtz’s Vorticity Theorems

The vorticity theorems of H. von Helmholtz (1858) concern the motion of fluid at
and in the vicinity of vortex filaments [1–3]. The theorems hold for inviscid flow
and flow, where the influence of viscosity can be ignored. In real flows all vorticity
phenomena are affected by the dissipative effects of viscosity.
• First Helmholtz Vorticity Theorem The First Helmholtz Vorticity Theorem
states:
The integrated vorticity flux over a cross surface of a vorticity tube, or the cir-
culation of the tube, is constant, independent of the shape and location of the cross
surface or its boundary over which the integrals are estimated.
In other words the vortex strength is constant along the vortex tube (or filament),
and therefore, Helmholtz’s first theorem simply reflects the same solenoidal nature
of the vorticity field as the velocity field for an incompressible fluid.
Since the vortex strength is constant along the vortex filament, its strength cannot
suddenly go to zero. Thus, a vortex cannot end in the fluid. It can only end on
a boundary or extend to infinity. Of course in a real, viscous fluid, the vorticity
is diffused through the action of viscosity and the width of the vortex filament can
become large, until it is hardly recognized as a vortex filament. For example a tornado
has one end on the ground, but at the other end, the vortex diffuses over a large
area with distributed vorticity. A ring vortex, which forms a closed path, is another
example.
• Second Helmholtz Vorticity Theorem The Second Helmholtz Vorticity The-
orem states:
If and only if the flow is circulation-preserving, then a material vorticity tube
moves with the fluid.
Singularities such as vortices in the flow move along with the local flow velocity.
An example here is the way vortices in the trailing vortex layer (wake) of a wing
interact and curve around each other forming a non-planar wake as they are convected.
• Third Helmholtz Vorticity Theorem The Third Helmholtz Vorticity Theorem
states:
If and only if the flow is circulation-preserving, a material vorticity tube has
constant strength.
3.7 Helmholtz’s Vorticity Theorems 61

As we shall see in the next section, this is quite similar to what Kelvin derived
independently eleven years later in his theorem.

3.8 Kelvin’s Circulation Theorem

Previously we have seen how the circulation Γ is related kinematically to the vorticity
through Stokes’ theorem. In 1869 Kelvin derived his circulation theorem on the
persistence of circulation showing how it is dynamically governed:
If and only if the acceleration is curl-free, the circulation along any material loop
is time invariant: 
D D
Γ = v · d = 0, (3.13)
Dt Dt C

where the closed curve C consists always of the same particles of fluid moving with
the flow velocity.
This condition defines a special class of flows of significant interest, known as
circulation-preserving flows. It is similar to Helmholtz third law which, historically,
had been obtained otherwise and earlier.
• Movement of Vortex Lines One especially valuable deduction from Kelvin’s
theorem concerns the movement of vortex lines. This is Helmholtz’s third theorem,
an exact consequence of the Euler equations, which states that vortex lines move
with the fluid.

3.9 Law of Biot-Savart

The definition of vorticity ω = curlv provides us with the vorticity field through the
differentiation of the velocity field.
What does the reverse function look like, if the vorticity field is given, and the
velocity field is sought?
The answer requires an integration process, and Lugt provides the answer [2], p.
90.
62 3 Elements of Vortex Theory

Fig. 3.8 Sketch for the

variables in the
vorticity-induction equation,
after [2]

The dynamic equation of vorticity transport, Eq. 3.12, describes the convection
of the vorticity with the velocity induced by the distribution of vorticity itself. A
distribution of vorticity ω(r  ) induces the velocity v(r) by the relation

1 s
v(r) = − ω(r  ) × dV, (3.14)
4π V s3

where the variables are indicated in Fig. 3.8. This is the vorticity-induction equation.
The vorticity field may occupy only a part or the whole of the space over which
the velocity field is nonzero. For instance, in a potential vortex, the vorticity field
is confined to the singular point with infinite ω. A confined vorticity field, which
exponentially decays away from its boundary, is called a compact field.
If the vorticity is concentrated to a single line filament L of circulation Γ , its
strength is constant by Kelvin’s theorem, and Eq. 3.14 reduces to the Biot-Savart law

Γ dr  × s
v(r) = − , (3.15)
4π L s3

where the vector dr  is tangential to the vortex filament. Note that the integral itself
is purely geometric, and can be evaluated without knowing the filament circulation
Γ a priori.
• Self-Induced Motion of Thin Vortex Filaments In general, the prediction of
the free motion of vorticity is a difficult unsolved problem governed by unsteady
nonlinear equations. The solutions that have been obtained involve one or more of
the following simplifications: steady or quasi-steady flow, small perturbations on a
straight filament, small vortex-core size, and simple vorticity distributions and/or
flow configurations.
It is well known that if self-induced motion of the line filament is calculated
by evaluating the velocity from Eq. 3.15 on the filament itself, the result will be
logarithmically infinite, if the filament is curved, and zero, if it is straight. Thus,
self-induced motion occurs only for curved filaments, but to obtain the correct value
for the velocity, further considerations of the finite size of the vortex core as well as
the vorticity distribution are required.
3.9 Law of Biot-Savart 63

• Vortex-Core Cutoff Method The Biot-Savart law is a valid description of the

velocity field induced by a thin vortex filament, if proper account is taken of the flow
within and near the vortex core.
One of the important interpretations of this is—under the restrictions mentioned
previously—that the concept of a cutoff distance is asymptotically valid. Very often,
the motion of curved vortex filaments has been calculated by integrating the Biot-
Savart law, but excluding a small segment of the filament of length  on either side
of the point on the filament itself to avoid the logarithmic singularity.
If  is chosen appropriately, the correct self-induced motion will be predicted by
the cutoff method. Another important result of this work is the concept of effective
core size.
A vortex of core size a with a particular distribution of vorticity and axial velocity
is kinematically—i.e. for motions well away from it—equivalent to a vortex of core
size a  with constant vorticity. This requires that the size of the vortex core be small
in comparison with the vortex-core radius of curvature and with the scale of axial
variation along the filament. The next section makes use of the concept of effective
core size.

3.10 Vortex Models

Figure 3.9 characterizes the structure of trailing vortices in a cross-section of a wing’s

wake. Relevant are two radii, which are defined on the basis of their velocity, respec-
tively vorticity distributions, see, e.g., [1].

Fig. 3.9 Trailing vortices: definition of the viscous core radius rc and the vorticity or outer core
radius rv [21]. s is the span-wise load factor
64 3 Elements of Vortex Theory

1. Viscous or inner core radius rc . The viscous core radius rc defines the inner vortex
core. The circumferential velocity vθ increases linearly from 0 at r = 0 up to vθmax
at r = rc , so the inner core rotates like a rigid body.
2. Vorticity or outer core radius rv . The distribution of the circumferential veloc-
ity of a vortex in reality is not that of the Rankine vortex, see below and also
Sect. 4.2.1. For r > rv the flow can be considered free from viscosity and vortic-
ity effects, therefore the radial distribution of circumferential velocity is that of
the potential vortex. rv is typically several times larger that rc . The region between
rc and rv contains the transition from the viscous core of the vortex to the potential
vortex. It contains a large part of its circulation, and viscosity effects are small.
The following discussion indicates some of the requirements that determine these
radii as well as other parameters, and some of the commonly used vortex models are
then presented.
• Finite-Core Vortex As mentioned above, in theoretical studies and numerical
computations the singularity of a point or potential vortex is often replaced by a
core of finite vorticity called a vortex blob or a vortex patch. A vortex with such a
non-singular center is called a finite-core vortex. The Rankine vortex with a core of
finite uniform vorticity is an example. Other vorticity distributions of the core are
possible, which do not necessarily satisfy the Euler equation but which may simulate
viscous flow properties or other core behavior.
• Wake Trailing Vortices: Model Requirements The direct way to determine the
structure of trailing vortices in the wake behind the wing is to calculate accurately the
roll-up process of the trailing vortex layer as it convects downstream. This however
is an extremely demanding computation, and other approximate methods are sought
without attempting to follow the roll-up.
One such process assumes a final configuration for the flow and requires conser-
vation of some chosen flow quantities between the initial and final configurations,
both modeled as two-dimensional flows.
For example, if the trailing vortex layer from each half of the wing rolls up
into a single vortex, Stokes’ theorem gives the magnitude of the circulation. Then
conservation of momentum requires that the impulse of the vortex layer equals the
impulse of the two trailing vortices, thus locating each vortex at the center of gravity
of the shed vorticity in its half of the vortex layer.
Consider a finite core vortex with radius rc . As rc → 0, while Γ0 = const., the
kinetic energy increases without bound, therefore implying a wing with infinite
induced drag. This is not an adequate flow model; the vorticity in trailing vortices
is finite and distributed in some manner. The following are some examples fulfilling
conditions for adequate models. For further details the reader should consult S.E.
Widnall [11] and the publications [21, 22] of the third author of the book.
• Rankine Vortex The Rankine vortex is an infinitely long straight vortex filament
whose interior rotates like a rigid body, thus with constant vorticity. Most early
discussions of the far-field structure of trailing vortices were based on the vortex-
wake model of J.R. Spreiter and A.H. Sacks [23], in which after roll-up the wake
3.10 Vortex Models 65

consists of two Rankine vortices: the vorticity being uniform in two circular vortex
cores and zero outside. The circumferential velocity vθ is

Γ0 r
r  rc : vθ = ,
2πrc rc
Γ0
r > rc : vθ = . (3.16)
2πr
This vortex has a clear definition of core size rc : all of the vorticity is in the
core, the peak tangential velocity is at the edge of the core, and the total circulation
is proportional to the product of peak velocity and core size. The core radius rc is
determined by requiring kinetic energy to be conserved in the roll-up process. For a
wing with elliptic circulation distribution the core radius rc is easily determined.
• Lamb–Oseen Vortex The Lamb–Oseen vortex analytically satisfies the unsteady
Navier–Stokes equations. It has been used extensively for initializing large-eddy sim-
ulations and in the design of LIDAR (light detection and ranging) matched filters.
The only non-zero velocity component, the tangential velocity vθ as a function of
the radial distance r from the vortex center, is

Γ0
[1 − e−β( rc (t) ) ],
r 2
vθ (r ) = (3.17)
2πr

where rc (t) = rc,02
+ 4βν(t − t0 ) is the vortex core radius, which is defined as the
radius, where the tangential velocity is maximum. Γ0 is the vortex circulation, and r
is the distance from the vortex center, ν the kinematic viscosity, t the time, β = 1.256
usually.
Note that the Lamb–Oseen vortex has a core that grows with time, so it is never
steady. It may be employed however by making use of the two-dimensional unsteady
analogy to steady three-dimensional wake flow.
• Burgers’ Vortex Burgers’ vortex is an exact incompressible Navier–Stokes flow.
It demonstrates the existence of extrema of vorticity in a steady three-dimensional
flow:
κ0 ar 2
vθ (r ) = [1 − e− 2ν ],
r
vr (r ) = −a r,
vz (z) = 2 a z, (3.18)

where κ0 = νa = 2π Γ
, hence the parameter a is determined by the circulation and the
kinematic viscosity ν.
The tangential velocity, vθ , has a finite value in the vortex core, which reduces
to the potential vortex as ν → 0. The extremum in three-dimensional flow is due to
stretching of the vorticity lines.
66 3 Elements of Vortex Theory

The constant a can have only positive values and is a measure of the radial influx.
The meridional flow, vr , vz is a potential flow that is meaningful only for ν > 0.
In other words, it is an example of a laminar potential flow. If however ν → 0, the
tangential velocity represents a potential flow for an inviscid fluid. The meridional
flow, vr , vz , may be considered the lower part of a circulatory motion, called a cell.
Notice that the vorticity in Burgers’ vortex is in the z-direction. As in two-dimensional
flows, all vorticity lines are parallel to each other.
• Batchelor Vortex The Batchelor vortex was obtained by G. Batchelor (1964)
as an approximate solution to the Navier–Stokes equations under a boundary-layer
type approximation. It has been used extensively as a typical mathematical model
of vortices, for example in vortex stability studies and also as models for trailing or
jet-like vortices.
The Batchelor vortex can be represented in the cylindrical coordinates (z, r, θ) as
q
[1 − e−r ],
2
vθ (r ) =
r
vr (r ) = 0,
vz (z) = a + e−r .
2
(3.19)

Non-dimensionalized velocities and lengths, and time scaled by the ratio of length
to velocity scales give
q
[1 − e−r /(1+ Re ) ],
2 4t
vθ (r, t) =
r
vr (r ) = 0,
1
e−r /(1+ Re ) ,
2 4t
vz (z, t) = a + (3.20)
1 + Re
4t

where Re is the Reynolds number and q is the swirl strength as the ratio of the
maximum tangential velocity to the core velocity.
The parameter a here designates the free stream velocity. It has been noted that
the translation and inversion of the axial velocity vz do not affect the instability of
the Batchelor vortex, so that one can set a = 0. Batchelor found this vortex suitable
to describe a trailing vortex far downstream of an aircraft. We note again that this is
an unsteady vortex with growing core.
• Burnham-Hallock Vortex The Burnham-Hallock vortex is the most widely
used model for wake vortex applications, which include processing of LIDAR obser-
vations, initialization of large-eddy simulations and modeling of aircraft response to
wake encounters. The tangential velocity field is given by

Γ0 r2
vθ (r ) = , (3.21)
2πr r 2 + rc 2
3.10 Vortex Models 67

where rc is the vortex core radius. Also numerical vortex-lattice models often use this
formulation for the calculation of the influence coefficients, e.g., to avoid infinities,
where trailing vortices could intersect bound vortices.

3.11 Structure of Trailing Vortices

Flight through trailing wing vortex wakes is by no means a new experience having
occurred since WW I, when pilots encountered vortex wakes trailing from maneu-
vering aircraft, or sometimes even flew into their own wakes.6
Figure 3.10 illustrates the flight environment of an aircraft with large aspect-ratio
wing, see also Figs. 8.1 and 9.25. The control of the so-called wake-vortex hazard is
the topic of Sect. 9.6, where the latter figure is discussed in some detail.
For several reasons, the subject is of great importance today. With increasing
air traffic around airports and in flight corridors, there are safety aspects, ranging
from dynamic flight behavior to concern about structural loads or time separation for
takeoffs and landings. Modern fighters with relaxed longitudinal static stability are
also extensively relying on active flight control systems. This trend is seen also with
modern civil aircraft. All this together has given new dimensions to the problem.
Flight control systems process information from sensors like accelerometers,
angle-of-attack sensors and pressure gauges, and the system responds on this infor-

Fig. 3.10 Wake-vortex hazard encountered by aircraft flying behind another aircraft [21]

6 Regarding the development and the structure of the wake of a large aspect-ratio wing see the
discussion of the corresponding Unit Problem in Sect. 8.4.
68 3 Elements of Vortex Theory

mation. The pilot also is acting, interacting and reacting in this scenario. How this
all will take place in wake encounters, and what response there will be, is not easy
to foresee, so simulation models are required. The SAAB Gripen fighter is a good
example of this.

3.11.1 SAAB 39 Gripen Wake Model

In 1999 a SAAB 39 Gripen aircraft was lost due to vortex-wake encounter while
practicing air combat below 3,000 ft. Thus the SAAB company launched a project
to develop a model for dynamic analysis and real-time simulation of flight through
wake vortices.
The aerodynamic model consists of both the wake flow model behind the gener-
ating aircraft as well as the computational model for estimating incremental forces
and moments induced by the wake on the incoming aircraft.
Desktop as well as real-time models for flight simulation of vortex-wake encoun-
ters were developed. Simulations were demonstrated with a six degrees of freedom
aircraft model. The model has been used for development of flight control sys-
tems, control laws, flight safety investigations as well as for demonstrations of wake
encounters. The reader is referred to Sedin et al. for details [24].

3.11.2 Trailing Vortex Instability

Recent concern over the hazard presented by the trailing vortices produced by large
aircraft has stimulated the study of flows with concentrated vorticity in free motion.
Since these vortices can be strong and persistent enough to pose a safety hazard to
other aircraft, it is clearly desirable to be able to predict the structure, position, and
persistence of such vortices as well as to understand the mechanisms by which vortex
wakes are dissipated, see also Sect. 8.4.
Under most conditions, trailing vortices undergo a natural sinusoidal instability,
the Crow instability, that eventually causes them to touch and break into a series
of crude vortex rings. This process destroys the initial wake structure more rapidly
than viscous or turbulent decay of the individual filaments. An example of practical
interest is the case of two counter-rotating vortices. Means to control the wake-vortex
hazard are discussed in Sect. 9.6.

3.11.3 Crow Instability

The Crow model considers the decay of a vortex pair in an inviscid linearly stratified
atmosphere. The model treats the generation of vorticity both in the wake due to
3.11 Structure of Trailing Vortices 69

nonuniform density and at the interface between the oval of fluid carried by the pair
and the surrounding fluid.
The first quantitative analysis of the three-dimensional instability of a vortex pair
in an ideal homogeneous fluid was given by S.C. Crow [25]. The results of his
analysis are in good agreement with observations of general features of aircraft wake
instability and measurements of amplification rates.
This instability is due to the mutual inductance of the sinusoidally perturbed pair.
A single sinusoidally perturbed vortex filament will rotate with angular velocity Ω
about its own axis. Instability occurs when the velocity field induced at the filament by
the presence and perturbation of the other filament annuls the self-induced rotation.
The perturbations then diverge on two planes tipped symmetrically at an angle
of about 45◦ with the horizontal. Both anti-symmetric and symmetric modes are
predicted by theory, but only the symmetric modes have been observed in flight.
General features of the modes are discussed in [11], see also [22]. The self-induced
rotation Ω depends on the details of the vorticity distribution in the core.

3.12 Vortex Layers and Vortices

We look first at the roll-up process of vortex layers. These are the trailing vortex layers
of lifting large aspect-ratio wings, which roll up into the pair of trailing vortices,
but also the so-called feeding layers, which lead to the lee-side vortices of small
aspect-ratio delta-like wings. We only give an overview over some items, the reader
interested in the theoretical details of roll-up of vortex layers should consult, e.g.,
[1–3].
Once the trailing or the lee-side vortices are established, vortex stretching and
vortex pairing are phenomena, which can occur. These are shortly considered, too.

3.12.1 Roll-Up of Shed Vortex Layers

Given the difficulty to state a precise definition of a vortex due to its elusive nature, we
focus here—where we are most secure in our understanding—on the development
of the trailing vortex layer, which is present behind a finite-span lifting wing.
We discuss the behavior of aircraft wakes following the generation of this initial
vortex layer: the roll-up of the sheet to form regions of concentrated vorticity, the
trailing vortices, and their motion. Figure 3.11 presents the overall problem; the left
side shows the spiraling of the sheet in a plan-wise cut through the sheet, the right
side shows a numerical model for the spiral discussed below.
Under most conditions the shed vortex sheet will roll up, and the wake structure
will be established within a few wing spans. A step-by-step calculation of the roll-up
process then is the way to predict the initial wake structure.
70 3 Elements of Vortex Theory

Fig. 3.11 Overall vortex-sheet roll-up. Left: roll-up of a wing’s trailing vortex layer, after [2].
Right: a wake or free-vortex sheet in a computational method (Model 4 of Table 1.3) [26]

For lightly loaded wings at high Reynolds numbers, roll-up takes place gradually
enough to justify a local two-dimensional model of the wake flow, and yet rapidly
enough to use an inviscid model of the flow. Unfortunately, even this simplification
results in a problem that has no known solutions.
Moreover in a remarkable paper from 1951, appreciating the high kinematically-
active vorticity content in leading-edge vortex sheets, J.R. Spreiter and A.H. Sacks,
[23], estimated that these sheets roll up 18 times more rapidly, in terms of chord
lengths, behind a low-aspect-ratio triangular wing than behind a large aspect-ratio
rectangular wing, see also Sect. 8.4.4. The problem for the low-aspect-ratio wing
therefore is even more difficult.
The complete three-dimensional determination of the shape of the trailing vortex
layer throughout the rolling-up process presents a problem of extreme difficulty, and
amenable only to computational solutions using Model 8 and higher models.
• Initial Computational Models for Sheet Roll-Up During the early 1980s two
computational approaches to the roll-up problem were feasible: either Model 4 or
Model 8 methods.
Use of panel representations of high-order make the vortex-sheet model amenable
to calculation with Model 4 methods, since it reduces or eliminates singular behavior
of the velocity field at the edges of panels, so that unrealistic disturbances do not
arise from the close approach of two turns of a spiral or of a sheet to a wall.
The representation usually adopted comprises a line vortex of varying strength,
and a nearly planar feeding vortex element that maintains continuity of circulation
between the free edge of the finite, outer part of the sheet and the line vortex along
the axis.
In this formulation the leading-edge vortex consists of a tightly wound spiraling
vortex sheet of infinite extent. In order to simulate the flow field outside of this vortex,
the vortex sheet is cut as it has reached some angular extent, while the remaining
inner region is modeled by an isolated line vortex connected to the sheet by a so-
called feeding sheet. This leads to the free-vortex sheet potential flow model depicted
3.12 Vortex Layers and Vortices 71

Fig. 3.12 Vortex sheet

tracked with two methods for
the flow past a flat 70◦ swept
delta wing, M∞ = 0, α =
20◦ . Upper part: The Model
4 (VORSEP) result [26].
Lower part: The captured
sheet in Model 8 [27].
Vorticity magnitude contours
were computed with a
medium mesh with 80 × 24
× 40 cells

on the right in Fig. 3.11 showing how the wing (or wake) and the free vortex sheet
are discretized by panel elements.
Both the line vortex and the feeding element experience forces, but the position
of the line vortex is chosen so that the total transverse force on each lengthwise
element of the combination is zero. With the core representation described above,
a complete discretization of the vortex sheet is possible—in the framework of the
Prandtl–Glauert approximation—in many different ways.
The potential flow problem is solved by employing a surface doublet distribution
μ(x) on both Sw —representing the wing and the trailing vortex layer—and Sv — rep-
resenting the feeding vortex layers shed from the swept leading edges. The velocity
field defined by the surface doublet distribution is an exact solution of the incom-
pressible flow equations and satisfies the far-field boundary condition. Further details
of this Model 4 computational method, VORSEP, are found in [27].
The second method computes the numerical solution to the incompressible Euler
equations, Method 8, and captures the roll-up process in the discrete solution.
H.W.M. Hoeijmakers and the second author of this book carried out an important
comparison of results computed with these two methods for the flow past a flat 70◦
swept delta wing, M∞ = 0, α = 20◦ [26, 27].
The comparison in Fig. 3.12 shows a three-dimensional view of the computed
tracked vortex sheet in VORSEP as well as the vorticity magnitude contours in three
planes of the medium resolution grid for the captured vortex in the Euler solution.
The computed geometry of the tracked vortex sheet, Fig. 3.12, is very nearly
conical, up to 70 per cent of the root chord.
72 3 Elements of Vortex Theory

Figure 3.13 superimposes the shape of the tracked vortex sheet onto the vorticity
contour in the spanwise cross-flow plane x/c = 0.6 in both the coarse and fine-grid
Euler solutions [28].
The vortex-sheet shapes are cross-sections with planes x/c = constant, whereas
the vorticity contours are projections on these planes of the surface of the constant
mesh coordinate, which intersects the wing at the corresponding stream-wise station.
Due to the artificial viscosity implied in the numerical method, the vortex sheet
is spread over a number of cells in the medium grid solution, and as expected,
substantially less in the fine-grid solution.
This means that in general the vortical flow region as computed by the Euler code
on the medium grid occupies a larger region than enclosed by the vortex sheet tracked
by VORSEP. But as the truncation error as well as the numerical viscosity decreases
due to decreasing grid size, the agreement is much better for the finer grid.
It follows from Fig. 3.13 that in this respect the two solutions agree quite well.
The strength and position of the vortex features of both computation models are in
good agreement.
Figure 3.13 is central to our discussion, because it correlates two world views,
namely the view from Model 4, where the natural element is that hard-to-define
concept of a vortex, with the view from Model 8, where the natural element is the
mathematically-precise concept of vorticity over a compact field, and shows that, in
this case, they are equivalent. The linking of these two concepts is very reassuring
for all our discussions in this book, see in this regard also Chap. 5.
We note concluding that present-day discrete numerical computation meth-
ods, Model 8 and higher, need to have a sufficiently fine discretization—grid
independence—of the domain where the roll-up happens, see, e.g., Sect. 8.4.2 and
Fig. 8.50, where beginning at x ∗ = 9 the original grid resolution is shown to be
insufficient.

Fig. 3.13 Spanwise section x/c = 0.6: comparison of captured vorticity contours, solid line, [28],
and tracked vortex sheet in the panel method VORSEP solution, dashed line. Left part: medium
mesh with 80 × 24 × 40 cells. Right part: fine mesh with 160 × 48 × 80 cells
3.12 Vortex Layers and Vortices 73

3.12.2 Vortex Stretching

Vortex Stretching was mentioned above in Sect. 3.6. The term (ω · ∇)v on the right-
hand side of Eq. 3.12 can be interpreted physically as the stretching and twisting of
vortex lines. The appearance of this nonlinear term in the vorticity-transport equation
is unique in the sense that it does not occur in the other forms of the Navier–Stokes
equations.
Consider an example. If in three-dimensional flow only the ωz component is
nonzero, that is, ω = ωz k, the vorticity-transport equation Eq. 3.12 simplifies to

Dωz ∂ω
= ωz . (3.22)
Dt ∂z

The vortex line is stretched when ω changes with z. Stretching a vortex line makes
the fluid spin faster.

3.12.3 Vortex Pairing

A single vortex filament does not move in an infinite fluid, since the velocity at the
singularity is cutoff to zero. Two or more vortices, however, move due to their mutual
interaction. The problem of describing such interaction will be discussed briefly here.
In an infinite, inviscid fluid, individual vortex pairs and rings move with constant
speed. In reality, friction causes the velocity to decay. We call two vortices of opposite
rotation in a two-dimensional flow a vortex pair. The strengths of the two vortices
need not be the same, and the core can be either singular, as in the case of the point
or potential vortex, or a blob.
Examples of practical interest are, for the case of two counter-rotating vortices, the
trailing vortices behind the wings of large aspect-ratio airplanes. Another example is
the combination of leading-edge and trailing vortices on delta wings, which we look
at more closely below.
• Vortex Pairing Behind a Delta Wing In his doctoral thesis, Y. Le Moigne, a
doctoral student of the second author of this book, studied vortex interactions above
and behind a delta wing at an angle of attack of α = 20◦ in low subsonic flow at
M∞ = 0.2 [29]. He solved the Euler equations—Model 8—using an automatic mesh
refinement procedure with a vortex sensor based on the eigenvalue analysis of the
tensor of the velocity gradients to improve the mesh resolution in the vortical regions.
Figure 3.14 shows the usual strong leading-edge vortex pair and a second vortex
pair, the trailing vortices, formed by the roll-up of the trailing vortex layer shed from
the trailing edge. The vortices are visualized with streamlines. The trailing vortices
are seen to interact with the leading-edge vortices and to roll around them, a particular
case of vortex pairing.
74 3 Elements of Vortex Theory

Fig. 3.14 Streamline visualization of the full-span solution using the vortex-adapted refined grid
[29]

The interesting point to notice is that the streamlines in the trailing vortices spiral
in the opposite direction to the one in the leading-edge vortices, corroborating D.
Hummel’s observation, [30], which is sketched in Fig. 3.15.
The flow field at the upper surface of the wing is restructured by the lee-side
vortices in such a way that finally at the trailing edge the general flow direction is
symmetrically outward, as sketched for the closed lee-side flow field in Fig. 7.12.
The effect of course is present also with an open lee-side flow field, Fig. 7.11, but
there it is restricted to the immediate vicinity of the outer edges, see also the figures
in Sect. 10.4.2. At the lower side of the wing the flow field more or less is not affected
by the lee-side phenomena.
All this is in contrast to the flow-field patterns always observed on lifting large
aspect-ratio wings. In Sect. 4.3.2 it is shown in a paradigmatic way that the flow at
the lower side is symmetrically outward and inward at the upper side. The resulting

Fig. 3.15 Schematic of the interaction of the lee-side vortex pair with the trailing vortex pair
originating from the trailing vortex layer behind the wing’s trailing edge, after [28]
3.12 Vortex Layers and Vortices 75

trailing vortex layer, hence the trailing vortices have the same sense of rotation as
the tip vortices.
In Fig. 3.15 the outward sense of rotation in the trailing vortex sheet is indicated
as being opposite to the lee-side vortices. The whole picture then results in double-
branched inner vortex cores (marked as ‘trailing edge vortex’) and the usual single-
branched outer vortex cores (marked as ‘leading edge vortex’).
Coming back to Fig. 3.14 one can see in addition that just behind the trailing edge
the trailing vortex is linked to the leading-edge vortex, forming somewhat like a
vortex pair. Further downstream it seems to be further separated from the leading-
edge vortex. These results are an example of the interactions between two vortex
pairs.

3.13 Vortex Breakdown, Vortex Re-configuration

A spectacular phenomenon of vortex flows, termed vortex breakdown or burst, was

first visualized by H. Werlé in a water tunnel for the leading-edge vortex over a delta
wing [31], and was later confirmed by D.H. Peckham and S.A. Atkinson, [32], and
also B.J. Elle [33].
The vortex breakdown appears, when at some point along its axis, a well-ordered
vortex suddenly becomes chaotic, as Fig. 3.16 indicates schematically.
Initially high-momentum jet-type vortex-core flow is present. An adverse axial
pressure gradient over the wing causes the breakdown, which can be characterized
as the development of a stagnation point on the vortex axis. It follows a region
of reversed axial flow encapsulated by a greatly swollen stream surface. A low-
momentum wake-type vortex-core flow ensues.
Experiments reveal essentially two basic types of vortex breakdown: the axisym-
metric bubble-type and the non-axisymmetric helical (spiral) type. Most experiments
were carried out with delta wings but also in pipes.

Fig. 3.16 Schematic illustration of vortex breakdown over a delta wing [34]
76 3 Elements of Vortex Theory

Characteristic features of vortex breakdown over delta wings are:

• The bubble type is characterized by a stagnation point on the vortex axis followed
by an oval-shaped recirculation zone. The bubble is single-celled with the recircu-
lation flow moving upstream along the vortex axis. The bubble is usually two or
three (upstream) core diameters in length. Downstream of the bubble the vortex is
turbulent and diffuses rapidly with distance.
• In the spiral type of breakdown a rapid deceleration of the core flow takes place.
Immediately downstream, the vortex core filament abruptly kinks and starts to
spiral around the axis of the structure, forming a corkscrew-like distortion of the
vortex core. The spiral structure can persist for one or two turns before breaking
up into large-scale turbulence. For leading-edge vortices the sense of the spiral
winding is opposite to the direction of rotation of the upstream vortex, however,
the rotation of the winding is in the same direction as the rotation of the upstream
vortex.
• The flow upstream of breakdown is steady, whereas numerous researchers have
observed unsteady flow downstream of the breakdown point. In Sect. 10.5.3 we
discuss vortex breakdown in the frame of Unit Problems.
Finally we note that vortex re-configuration happens, if the vortex structure after
breakdown enters an area with favorable pressure gradient. This was observed in
pipe flow. Whether it can occur with delta wings, is not known.

3.13.1 Fundamental Studies

E. Krause has carried out a number of fundamental studies for incompressible flow
and established some basic conclusions about vortex breakdown [35–37].
In 1990 a time-dependent numerical solution of the incompressible Navier–Stokes
equations described the process of vortex breakdown of isolated slender vortices
[35]. It showed the transition from bubble-type to spiral-type breakdown, when side-
boundary conditions compatible with stable breakdown were implemented.
In another incompressible study of a slender columnar vortex with its axis parallel
to the oncoming axial flow, Krause shows that in inviscid flow the azimuthal velocity
component near the axis vanishes with the axial velocity component [37].
This close coupling between the axial and the azimuthal flow motion, referred to
as the onset of vortex breakdown for vanishing axial flow, is described by the balance
of the convective acceleration and vortex stretching terms in the vorticity transport
equation, formulated by von Helmholtz in 1858, and represents a fundamental result
for inviscid incompressible flow. It can be formally stated as:
In axially symmetric, incompressible, inviscid flow of a columnar vortex with its
axis parallel to the main flow, the angular velocity near the axis vanishes, when the
axial velocity component vanishes and a stagnation point is formed.
However, in laminar flow, with viscous forces acting near the stagnation point,
the angular velocity does not necessarily vanish with the axial velocity going to zero.
3.13 Vortex Breakdown, Vortex Re-configuration 77

Nevertheless this result links the occurrence of breakdown with the presence of a
stagnation point, and this can be checked in computations of more complex setting.

3.13.2 Computed Vortex Breakdown Over a Delta Wing

In his doctoral dissertation S. Görtz, a doctoral student of the second author of this
book, carried out a comprehensive and exhaustive study of vortex breakdown over
delta wings using Models 9 and 10 [38]. The detailed discussion of lee-side vortex
flow past delta wings, including the vortex-breakdown phenomenon, is given in
our Chap. 10. Presented here is one case that supports Krause’s claim of a stagnation
point occurring even in the practical situation of compressible high Reynolds-number
turbulent flow over a delta wing.
Figure 3.17 shows a three-dimensional view of breakdown for a turbulent com-
putation of Model 10 type at 35◦ angle of attack using the one-equation turbulence
model of Spalart and Allmaras.
The left-hand side of the picture presents an iso-surface of total pressure colored
in magnitude of axial velocity.
The feeding shear layer can be seen to emanate from the sharp, highly swept
leading edge. The primary vortex itself emanates from the apex of the wing. It
curves slightly inboard when intercepting the flat upper surface but curves back

Fig. 3.17 Vortex breakdown: computed total pressure iso-surface and vortex core filament [38].
M∞ = 0.16, Re = 1.97 × 106 , α = 35◦ , 1-eq. turbulence model
78 3 Elements of Vortex Theory

further downstream. The fluid entrained in the core accelerates gradually along the
core until the magnitude of the axial velocity reaches a value of about M = 0.4, i.e.
2.5 times the freestream velocity.
Notice the following expansion of the core and the preceding deceleration of the
axial fluid.
Although it is difficult to identify complete stagnation here, the color contours
definitely show that the axial flow component is strongly reduced, becoming even
negative (i.e. flow upstream) inside the spiral core, in line with the Krause result [37].
On the right hand side of the wing a recirculation zone is seen downstream of
the breakdown point. The spiraling nature of the core behind the recirculation zone
suggests that a bubble type of breakdown with a “spiraling tail” has been predicted.

3.13.3 Vortex Re-connection

Reconnection encompasses merging and splitting of vortex filaments, but since a

precise and rigorous definition of a vortex eludes us, a commonly agreed upon and
clear definition of vortex reconnection is lacking, as well as a consensus on relevant
physical mechanisms.
Nevertheless, reconnection has been observed in laboratory experiments and in
the sky in condensation trails of airplanes after the Crow instability occurred. The
two trailing vortices touch, split, and rearrange themselves into a row of irregularly
shaped vortex rings, see, e.g., Fig. 8.1.
The study of vortex reconnection is of particular importance in understanding the
evolution of vortical structures in turbulence.

3.14 Separation and Vortex Flow Control

In modern aircraft technology control of boundary-layer separation and of vortex

flow is an important engineering tool. An aircraft today is unthinkable without such
control. The benefits of separation and vortex control include drag reduction, lift
increase, stabilization and performance. Control can be a direct design means or a
“repair” solution, after test flights have shown a need for correction.
We list and comment on a few significant control means without going into details.
• Vortex generators: control of boundary-layer separation Vortex generators are pas-
sive devices, which are embedded in a boundary layer in order to prevent sepa-
ration. They can differ in size and shape and can be denoted by different names,
however their mechanism is usually the same—to generate longitudinal vortices
in order to reduce or even avoid separation.
The effect of the longitudinal vortices is the transport of high momentum flow from
the outer domain of the boundary layer into wall-near flow portions. In this way
3.14 Separation and Vortex Flow Control 79

the momentum of the wall-near boundary-layer flow is enhanced and its proneness
to separation is reduced.
Recently sub-boundary-layer vortex generators (SBLVGs) found much attention.
They are fully submerged in the boundary layer. In Sect. 9.2 we give a short account
of them.
For a more recent and thorough account of the modeling and use of vortex gener-
ators in the design and performance of aircraft, the interested reader is referred for
instance to A. Jiràsek [39], a doctoral student of the second author of this book.
General discussions of means to influence boundary-layer flow in particular in
three dimensions, can be found in [40].
• Geometrical shaping of configuration elements A passive means to avoid the sep-
aration at the wing-root/fuselage intersection is the wing-root fairing. Today it can
be found on almost all airplanes. Without this fairing separation in the form of
a horse-shoe vortex occurs, which causes a drag increment and leads to buffet at
high angles of attack. We give a short discussion of this issue in Sect. 8.4.3.
Another topic is the wing planform shaping and optimization of small aspect-ratio
delta-like wings. We sketch this topic in Sect. 11.2. A particular problem of such
slender wings is the pitch-up behavior at higher angles of attack. We present a
design example in Sect. 11.6.
• Suction and Blowing Suction through a surface can be used to control, for instance,
laminar-turbulent transition—hybrid laminar flow control (HLFC)—or to suppress
flow separation over an aerodynamic control surface. In that case just ahead of the
hinge line boundary-layer flow is removed by suction and attached flow is ensured
over the flap. Another application is at supersonic and hypersonic engine ramp
inlets, where at the ramp corners shock/boundary-layer separation is controlled
by suction. In [40] results of an old study are shown regarding suction as a virtual
boundary-layer fence.
Blowing is used at a number of separation problems. We do not go into details.
We note, however, a particular means to achieve separation-free flow over high-
lift systems, the “tangential blowing” at the slat and the Fowler flap. The effect
is that the flow through a gap, driven by the pressure difference between the
lower (pressure) and the upper (suction) side of the wing, prevents boundary-layer
separation at the upper side of the wing behind the slat and at the Fowler flap,
Sect. 9.2.
Regarding small aspect-ratio delta-like wings spanwise blowing was an important
topic in the 1970s. In Sect. 11.5 we give an account of some work in this regard.
Noteworthy also is the topic of jets in cross flow, which regards vehicle flight
control issues.
• Strakes Strakes are devices to generate vortices. These are used, for instance, to
generate side forces on forebodies and to stabilize delta-type aircraft at high and
even post-stall angles of attack in the lateral/directional motion, see Sect. 11.4. A
now prominent example is the nacelle strake, which behind the nacelles of high
bypass-ratio engines prevents separation when the wing is in high-lift condition.
In Sect. 9.4 the topic of the nacelle-strake vortex is discussed in some detail.
Figure 3.18 shows a combination of nacelle-strakes and vortex generators, which
80 3 Elements of Vortex Theory

Fig. 3.18 Boeing 707 during high-lift test: flow control with nacelle strakes and vortex generators
[41]. Note the visualized nacelle-strake vortices

re-energizes the flow over the upper surface of the wing behind the nacelles, the
two together preventing flow separation there.
Nacelle strakes were possibly for the first time used on the DC-10 as a solution to
avoid a significant loss in maximum lift coefficient, when flaps and landing slat
extensions are deployed [39]. They can lower stall speed in approach configuration
and reduce the required take-off and landing runway length, a reduction which for
example in the case of the DC-10 aircraft was about 6 per cent.

3.15 Vortex Flows and Dynamic Structural Loads

Turbulent boundary layers, separation regions, vortex sheets and vortices have rugged
edges and large sub-structures, which are unsteady and can cause buffeting, structural
fatigue, aileron buzz (vibration of control surfaces), etc. All of these phenomena
warrant a brief general problem discussion of dynamic structural loads.
At delta wings at high angle of attack, the phenomenon of leading-edge vortex
breakdown over the wing planform is of specific interest. The transition from stable
to unstable core flow, evident by the rapid change in the axial velocity profiles from
jet- to wake-type with increasing angle of attack, leads to extremely high turbulence
intensities at the breakdown position and to increased turbulence levels further down-
stream. Hence, the buffet excitation level increases strongly above a certain angle of
attack, and wing and fin normal-force spectra may exhibit narrow-frequency band
peaked distributions.
3.15 Vortex Flows and Dynamic Structural Loads 81

Figure 3.19 schematically indicates the problem of vertical tail buffet for the F/A-
18.
The unsteady aerodynamic loads excite the vertical tail structure or even the wing
structure in their natural frequencies, resulting in increased fatigue loads, reduced
service life and raised maintenance costs.
For example, the fin buffeting problem plagues twin-fin configurations like the
F-15 and F/A-18, but single-fin aircraft are also affected. Therefore, comprehensive
research programs have been undertaken aimed at understanding the buffeting loads
and reducing the structural response.
The related vortex-flow features are carefully analyzed using wind tunnel tests
on small- and full-scale models supplemented by flight tests and detailed numerical
flow simulations. In addition, design and analysis methods have been developed to
describe the fin-buffet environment and to predict buffet loads for use in aircraft
design.
The buffet loads do not only decrease the fatigue life of the airframe, but may, in
turn, limit the angle of attack envelope of the aircraft. To counter buffeting problems,
several methods have been suggested. They deal with alterations of the structural
properties like stiffness and damping, and aerodynamic modifications for passive or
active control of vortex trajectories to avoid a direct impact of the burst vortex flow.
The structural dynamic loads can be reduced to increase the service life and to
enhance the maneuverability by allowing to extend the angle of attack envelope.

Fig. 3.19 Schematic illustration of the vertical tail buffet problem on the F/A-18 aircraft [34]. LEX:
leading-edge extension
82 3 Elements of Vortex Theory

The interaction with a fin involves the following phenomena:

• the time-averaged vortex breakdown location depending on the adverse pressure
gradient set by the recompression at the wing trailing edge and/or by the blockage
of the fin,
• the helical mode instability of the breakdown flow,
• quasi-periodic oscillations of the breakdown location, distortion of structural
dynamic properties.
For a thorough account of the aero-structural modeling of the F/A-18 the reader
should turn to [42, 43]. In this regard also the work reported in [44] should be
mentioned, and in a wider context the publications [45–47].
In closing this section we note that the buffeting problem also exists for any
other aircraft and not only in relation to the vertical fin. When the undercarriage is
extended, wheel-well doors and the wheel well (as large cavity) are exposed to the
air flow. On fighter aircraft, like bombers, the bays for internal weapon carriage pose
the same problem during weapon release.
In all cases buffeting means dynamic structural loads, which are of high concern.

3.16 Basic Quantities of Trailing-Vortex Flow Fields

In order to get basic quantities and relations of trailing-vortex flow fields potential
flow theory (Model 4 of Table 1.3) is useful, even if in principle it is restricted to
the low subsonic flight domain. For general applied aerodynamic purposes see, e. g.,
[4–7].
The relation between the spanwise lift L(y) and circulation Γ (y) distributions is
given by the Kutta–Joukowsky theorem:

d L = ρ∞ u ∞ Γ (y)dy. (3.23)

The lift coefficient is

L
CL = , (3.24)
q∞ A

where A is the reference area, usually the wing’s surface projected into the x-y plane,
and ρ∞ 2
q∞ = u . (3.25)
2 ∞

the dynamic pressure.7

The wing loading Ws is defined as

7 Regardingthe reference area we note that Airbus uses the ‘Airbus Gross’ definition and Boeing
the ‘Wimpress’ definition (after its inventor at Boeing), www.lissys.demon.co.uk/pug/c03.html.
3.16 Basic Quantities of Trailing-Vortex Flow Fields 83

Fig. 3.20 Local chord

element of a finite-span
wing [4]

Mg L
Ws = = [N /m 2 ]. (3.26)
A A
The dimensionless circulation is, with b being the wing span

Γ (y)
σ(y) = . (3.27)
u ∞ b/2

Consider now the local chord element in Fig. 3.20.

The local normal force coefficient C z is found by the integration of the Δ-pressure
coefficient along the chord section8 :

1
C z (y) = Δc p dy, (3.28)
c(y) c(y)

with Δc P = c pu − c pl being the pressure-coefficient difference between the upper

and the lower chord side and c(y) the local chord length.
For small angles of attack we have Cl ≈ C z , and find the lift distribution in span
direction to be
dL
= Cl c(y) q∞ . (3.29)
dy

Generally the local lift coefficient Cl can be written with Eq. (3.23) as

1 dL 2Γ (y)
Cl (y) = = . (3.30)
c(y) q∞ dy u ∞ c(y)

If at a spanwise location y the local lift coefficient Cl is given, the local value of
the circulation reads
1
Γ (y) = Cl (y)c(y)u ∞ . (3.31)
2
The dimensionless circulation σ(y) then is

8 Local force and moment coefficients are denoted with lower-case letter indices.
84 3 Elements of Vortex Theory

Fig. 3.21 Schematic of

finite-span lifting wing [4].
Upper part: Circulation
distribution. Lower part:
trailing vortex layer behind
the wing

Cl (y)c(y)
σ(y) = . (3.32)
b

For a lifting wing the circulation distribution in spanwise direction, Γ (y), is

sketched in the upper part of Fig. 3.21. The lower part shows in planform view the
vortex layer with the local strength being dΓ = (dΓ /dy)dy  , see also the compati-
bility condition in Sect. 4.4.
The lift L of the wing then is expressed as
 b/2  b/2
L = q∞ Cl c(y) dy = ρ∞ u ∞ Γ (y) dy. (3.33)
−b/2 −b/2

Consider Fig. 3.22. It schematically shows the roll-up process of the trailing vortex
layer which leaves the lifting wing. At the end of the extended near field, x/b ≈ 10,
Sect. 8.1, the fully established trailing vortices are present.
The bound or root circulation at the wing in Fig. 3.22, Γ0 , can be written in terms
of the lift coefficient C L of the wing, the spanwise load factor s, and the aspect ratio
Λ = b2 /A, as
C L u∞b L
Γ0 = = , (3.34)
2sΛ ρ∞ u ∞ b0

with the load factor s: 

1 b/2
Γ (y) b0
s= dy = . (3.35)
b −b/2 Γ0 b
3.16 Basic Quantities of Trailing-Vortex Flow Fields 85

Fig. 3.22 Idealized

circulation distribution and
roll-up process of the trailing
vortex layer, resulting in the
trailing-vortex pair [21]

The quantity b0 denotes the lateral distance of the barycenters of the two trailing
vortices. If the circulation distribution is elliptic

Γ (y) = Γ0 1 − (2y/b)2 , (3.36)

the spanwise load factor s is (the superscript ∗ referring to the elliptical distribution)

b0 π
s = s∗ = = . (3.37)
b 4
The result is that the lateral distance b0 of the rolled-up vortices is smaller than
the wing span b. Only for wings with a very large aspect ratio we get b0 /b → 1 [4].
With an elliptic circulation distribution the root circulation becomes
2C L u ∞ b
Γ0∗ = . (3.38)
πΛ
The normalized circulations G is
Γ
G= . (3.39)
Γ0
86 3 Elements of Vortex Theory

These relations are helpful to compare wake-vortex properties due to a given

circulation distribution with those of an elliptical distribution.
The induced downward velocity w0 on the vortex centers (law of Biot-Savart,
Sect. 3.9) is
Γ0 CL u∞
w0 = = . (3.40)
2πb0 4πΛs 2

For the elliptic circulation distribution we obtain

4C L u ∞
w0∗ = . (3.41)
π3 Λ
With the induced velocity w0 and the lateral distance b0 a timescale t0 can be
defined:

b0 2π(sb)2 Λb
t0 = = = 4πs 3 . (3.42)
w0 Γ0 CL u∞

The time scale t0 defines the time interval at which the vortex pair moves downward
by the distance equal to the vortex spacing b0 . For the elliptic circulation distribution
it reads

π4 Λ b
t0∗ = . (3.43)
16 C L u ∞

The time scale can be seen as a measure of the trailing vortex age. It depends
strongly on the load factor s: t0 ∝ s 3 .
The dimensionless timescale related to a downstream location x = u ∞ t, respec-
tively dimensionless x ∗ = x/b = u ∞ t/b, is

t x 1 CL
τ= = = x∗ 3 . (3.44)
t0 u ∞ t0 s 4πΛ

For the elliptic circulation distribution this reads

t 16C L
τ∗ = = x∗ 4 . (3.45)
t0∗ π Λ

The dimensionless time τ permits an analysis of the trailing vortex pair at the
location x ∗ = x/b, taking into account the load factor s, the lift coefficient C L and the
aspect ratio Λ. For the elliptical circulation distribution as aerodynamic reference
distribution, C L and Λ are the only relevant configuration parameters.
In the design of large aspect-ratio wings generally the elliptical circulation distri-
bution is the target distribution, because it means minimum induced drag. Assuming
such a distribution, we obtain for the lift L ∗ and for the induced drag Di∗ the relations
3.16 Basic Quantities of Trailing-Vortex Flow Fields 87

π
L∗ = ρ∞ b u ∞ Γ0∗ (3.46)
4
and π
Di∗ = ρ∞ Γ0∗ 2 , (3.47)
8
respectively
L∗ 2
Di∗ = , (3.48)
πq∞ b2

q∞ being the dynamic pressure of the freestream.

The coefficient of the induced drag at elliptical circulation distribution reads

C L∗ 2
C D∗ i = , (3.49)
πΛ
with the lift coefficient C L being

L∗
C L∗ = . (3.50)
q∞ A

In design aerodynamics the drag coefficient usually is composed of the zero-lift

drag coefficient C D0 and the induced drag coefficient C Di .
The latter then reads

C L∗ 2
C Di = , (3.51)
πeΛ
with e being span-efficiency factor, if the isolated wing is the topic, and e being the
Oswald efficiency factor, if the whole aircraft is considered.
In the literature the matter appears to be presented somewhat varying [5–7]. In
any case the Oswald efficiency factor usually is given with 0.7  e  0.85 for the
subsonic/transonic aircraft with large aspect-ratio wings, which we consider here,
whereas the span-efficiency factor is given with 0.9  e  1.
These relations permit quick estimates of aerodynamic properties of large aspect-
ratio wings. We note that in cruise at optimal lift-to-drag ratio the induced drag is
less than half of the total drag.

3.17 Problems

Problem 3.1 Consider Fig. 3.3. The lifting-line model is the simplest model of a
lifting wing in the frame of circulation theory. (a) What flow model is behind it. (b) Is
the circulation Γ0 constant and the same on all four legs? (c) What vorticity theorems
does the lifting-line model obey? (d) How is the situation in the two-dimensional
88 3 Elements of Vortex Theory

Fig. 3.23 Sketch of the airfoil

case of an airfoil? (e) How is the situation in the quasi-three-dimensional case of the
infinite swept wing?

Problem 3.2 Consider steady level flight of a large transport aircraft with a mass
of m = 230,000 kg, a wing span of b = 60 m and an aspect ratio of Λ = 9 flying
at the Mach number of M∞ = 0.82 at an altitude of H = 10 km. Consider further
an elliptical circulation distribution. By how many meters moves the wake vortex
system approximately downward within a flight path length of x = 50 km?

Problem 3.3 Consider now for this aircraft configuration an approach flight with
u ∞ = 80 m/s at a lift coefficient of C L = 1.4, assuming again an elliptical circulation
distribution. Which maximum circumferential velocity is attributed to the trailing
vortices applying on the one hand the Lamb–Oseen vortex model and on the other
hand the Burnham-Hallock vortex model? How can the difference be judged? The
viscous core radius rc is estimated to about 3 percent of the wing span b.

Problem 3.4 Calculate the lateral distance b0 of the rolled-up wake vortex system
for a parabolic circulation distribution with respect to the wing span b. How may this
affect the wake vortex decay scenario? The spanwise parabolic circulation distribu-
tion is given by Γ (y) = Γ0 (1 − (2y/b)2 ).

Problem 3.5 Simulate the incompressible flow around a 10 per cent thick biconvex
circular arc airfoil at zero incidence with three constant strength source distributions
of equal length on the cord line. What is the pressure coefficient in the most upstream
collocation point?

Problem 3.6 Consider a flat plate placed in an air stream at a small angle of attack,
α, as sketched (Fig. 3.23).
Model the flow with a discrete vortex at x/c = 0.25. What is the strength Γ of
the vortex?

Problem 3.7 Place two plane airfoils in a tandem arrangement separated by a dis-
tance .
Use the lumped vortex method, Appendix A.2.2, to determine the lift on each
airfoil and the variation with the spacing .
3.17 Problems 89

Problem 3.8 Consider an aircraft flying at constant speed. The flight Mach number
is M∞ = 0.8, the flight altitude H = 10 km. The wing span is b = 60 m. The mass
of the aircraft is 200,000 kg. Assume an elliptic circulation distribution, an Oswald
efficiency factor e = 0.7 and compute the coefficients of lift and induced drag as well
as the lift-to-drag ratio, assuming that the induced drag is about one third of the total
drag.

References

1. Saffmann, P.G.: Vortex Dynamics. Cambridge University Press (1992)

2. Lugt, H.J.: Introduction to Vortex Theory. Vortex Flow Press, Potomac, Maryland, USA (1996)
3. Wu, J.-Z., Ma, H.-Y., Zhou, M.-D.: Vorticity and Vortex Dynamics. Springer, Berlin Heidelberg
New York (2006)
4. Schlichting, H., Truckenbrodt, E.: Aerodynamik des Flugzeuges, Vol. 1 and 2, Springer-
Verlag, Berlin/Göttingen/Heidelberg, 1959, also: Aerodynamics of the Aeroplane, 2nd edition
(revised). McGraw Hill Higher Education, New York (1979)
5. Anderson Jr., J.D.: Fundamentals of Aerodynamics, 5th edn. McGraw Hill, New York (2011)
6. Drela, M.: Flight Vehicle Aerodynamics. The MIT Press, Cambridge, MA (2014)
7. Rossow, C.-C., Wolf, K., Horst, P. (eds.): Handbuch der Luftfahrzeugtechnik. Carl Hanser
Verlag, München, Germany (2014)
8. Anderson Jr., J.D.: The Airplane: A History of Its Technology. AIAA, Reston Va (2002)
9. Hirschel, E.H., Prem, H., Madelung, G. (eds.): Aeronautical Research in Germany–from Lilien-
thal until Today. Springer-Verlag, Berlin Heidelberg New York (2004)
10. Serrin, J.B.: Mathematical Principles of Classical Fluid Mechanics. In S. Flügge (ed.): Hand-
buch der Physik, Band VIII/1, Strömungsmechanik 1, Springer-Verlag (1959)
11. Widnall, S.E.: The Structure and Dynamics of Vortex Filaments. Ann. Rev. Fluid Mech. Palo
Alto, CA 7, 141–165 (1975)
12. Bloor, D.: The Enigma of the Aerofoil-Rival Theories in Aerodynamics, 1909–1930. The
University of Chicago Press, Chicago and London (2011)
13. Rizzi, A.: Damped Euler-Equation Method to Compute Transonic Flow Around Wing-Body
Configurations. AIAA J. 20(10), 1321–1328 (1982)
14. Lighthill, J.: Introduction Boundary-Layer Theory. In: Rosenhead, L. (ed.) Laminar Boundary
Layers, pp. 46–113. Clarendon Press, Oxford (1963)
15. Klein, F.: Über die Bildung von Wirbeln in reibungslosen Flüssigkeiten. Zeitschrift für Math.
und Physik 59, 259–62 (1910)
16. Betz, A.: Wie ensteht ein Wirbel in einer wenig zähen Flüssigkeiten? Die Naturwissenschaften
9, 193–96 (1950)
17. Rizzi, A.: Three-Dimensional Solutions to the Euler Equations with one Million Grid Points.
AIAA Journal 23, 1986–1987 (1985)
18. Weis-Fogh, T.: Quick Estimates of Flight Fitness in Hovering Animals, Including Novel Mech-
anisms for Lift Production. Journal Exp. Biol 59, 169–230 (1973)
19. Lighthill, J.: On the Weis-Fogh Mechanism of Lift Generation. Journal Fluid Mech. 60, 1–17
(1973)
20. Hirschel, E.H.: Basics of Aerothermodynamics. 2nd, revised edition. Springer, Cham Heidel-
berg New York (2015)
21. Breitsamter, C.: Nachlaufwirbelsysteme großer Transportflugzeuge - Experimentelle Charak-
terisierung und Beeinflussung (Wake-Vortex Systems of Large Transport Aircraft—
Experimental Characterization and Manipulation). Inaugural Thesis, Technische Universität
München, 2007, utzverlag, München, Germany (2007)
90 3 Elements of Vortex Theory

22. Breitsamter, C.: Wake Vortex Characteristics of Transport Aircraft. Progress in Aerospace
Sciences 47(1), 89–134 (2011)
23. Spreiter, J.R., Sacks, A.H.: The Rolling Up of the Trailing Vortex and its Effect on the Down-
wash Behind Wings. Journal Aero. Sci. 18, 21–32 (1951)
24. Sedin, Y.C.J., Grasjo, I., Kullberg, E., Larsson, R.: A Model for Simulation of Flight Passages
through Trailing Tip Vortices. Paper ICAS 9(3), 2002–7 (2002)
25. Crow, S.C.: Stability Theory for a Pair of Trailing Vortices. AIAA Journal 8(12), 2172–2179
(1970)
26. Hoeijmakers, H.W.M.: Computational Aerodynamics of Ordered Vortex Flows. Doctoral The-
sis, TU Delft, Rept. TR diss 1729, Delft, The Netherlands (1989)
27. Hoeijmakers, H.W.M., Rizzi, A.: Vortex-Fitted Potential and Vortex-Captured Euler Solution
for Leading-Edge Vortex Flow. AIAA Journal 23, 1983–1985 (1985)
28. Rizzi, A.: Multi-cell Vortices Computed in Large-Scale Difference Solution to the Incompress-
ible Euler Equations. Journal Comp. Phys. 77, 207–220 (1988)
29. Le Moigne, Y.: Adaptive Mesh Refinement and Simulations of Unsteady Delta-Wing Aero-
dynamics. Doctoral Thesis, KTH Royal Institute of Technology, Rep. TRITA-AVE 2004:17,
Stockholm, Sweden (2004)
30. Hummel, D.: On the Vortex Formation over a Slender Wing at Large Angles of Incidence. In:
High Angle of Attack Aerodynamics, Conference Proceedings AGARD CP-247, 15-1–15-17
(1978)
31. Werlé, H.: Quelques résultats expérimentaux sur les ailes en flèches, aux faibles vitesses,
obtenus en tunnel hydrodynamique. La Recherche Aéronautique 41, (1954)
32. Peckham, D.H., Atkinson, S.A.: Preliminary Results of Low Speed Wind Tunnel Tests on a
Gothic Wing of Aspect Ratio 1.0. Report CP-508, Aeronautical Research Council (1957)
33. Elle, B.J.: An Investigation at Low Speed of the Flow Near the Apex of Thin Delta Wings with
Sharp Leading Edges. Reports and Memoranda 3176, Aeronautical Research Council (1958)
34. Breitsamter, C.: Unsteady Flow Phenomena Associated with Leading-Edge Vortices. Progress
in Aerospace Sciences 44(1), 48–65 (2008)
35. Krause, E.: The Solution to the Problem of Vortex Breakdown. Lecture Notes in Physics Vol
371, Springer-Verlag, Berlin, 35–50 (1990)
36. Krause, E.: On an Analogy to the Area-Velocity Relation of Gasdynamics in Slender Vortices.
Acta Mech. 201, 23–30 (2008)
37. Krause, E.: Stagnant Vortex Flow. Acta Mech. 209, 345–351 (2010)
38. Görtz, S.: Realistic Simulations of Delta Wing Aerodynamics Using Novel CFD Methods.
Doctoral Thesis, KTH Royal Institute of Technology, Rep TRITA-AVE 2005:01, Stockholm,
Sweden (2005)
39. Jiràsek, A.: Vortex Generator Modeling and its Application to Optimal Control of Airflow
in Inlet. Doctoral Thesis, KTH Royal Institute of Technology, Rep TRITA-AVE 2006:66,
Stockholm, Sweden (2006)
40. Hirschel, E.H., Cousteix, J., Kordulla, W.: Three-Dimensional Attached Viscous Flow.
Springer, Berlin Heidelberg (2014)
41. Campbell, J.F., Chambers, J.R.: Patterns in the Sky–Natural Visualizations in Aircraft Flow
Fields. NASA SP-514 (1994)
42. Guillaume, M., Gehri, A., Stephani, P., Vos, J.B., Mandanis, G.: F/A-18 Vertical Tail Buffeting
Calculation Using Unsteady Fluid Structure Interaction. Aero J., Vol 115, No. 1166 (2011)
43. Vos, J.B., Charbonnier, D., Ludwig, T., Merazzi, S., Gehri, A., Stephani, P.: Recent Devel-
opments on Fluid Structure Interaction Using the Navier-Stokes Multi Block (NSMB) CFD
Solver. AIAA-Paper 2017–4458, (2017)
44. Breitsamter, C.: Turbulente Strömungsstrukturen an Flugzeugkonfigurationen mit Vorderkan-
tenwirbeln. (Turbulent Flow Structures at Aircraft Configurations with Leading-Edges Vor-
tices). Doctoral Thesis, Technische Universität München, 1996, utzverlag, München, Germany
(1997)
45. Mabey, D.G.: Some Aspects of Dynamic Loads Due to Flow Separation. AGARD-R-750 (1988)
References 91

46. Luber, W., Becker, J., Sensburg, O.: The Impact of Dynamic Loads on the Design of Military
Aircraft. AGARD-R-815, 8-1–8-27 (1996)
47. Breitsamter, C., Schmid, A.: Airbrake Induced Fin Buffet Loads on Figther Aircraft. Journal
of Aircraft 45(5), 1619–1630 (2008)
Chapter 4
The Local Vorticity Content of a Shear
Layer

In the mid 1980s computer speed and storage had developed to a degree that Euler
methods (Model 8 in Table 1.3) became a viable tool for aerodynamic design work. At
that time very much discussed was the fact that at delta wings with sharp leading edges
lee-side vortices resulted in the relevant angle of attack and Mach number regime.
The question was, where is the apparent vorticity coming from, and accordingly the
associated entropy rise.
Oddly enough this was not an issue for Euler solutions—like for modeled
potential-flow (Model 4) solutions—for lifting large aspect-ratio wings. In these
cases at that time Euler methods produced reasonable results regarding lift, pitching
moment and induced drag. The involved trailing vortex layers and vortices, however,
remained unobserved.
Back then the first two authors of this book argued that the vorticity over delta
wings was produced in the same way as over large aspect-ratio wings. Accordingly at
the Military Aircraft Division of Messerschmitt–Bölkow–Blohm (MBB) the concept
of the local vorticity content of a shear layer was developed and tested. With this
concept it could be shown how the problem may be understood.
In the present chapter the concept of the local vorticity content of a shear layer is
defined and explained. The concept permits to connect in a simple and descriptive way
viscous phenomena of reality in the form of Model 2 in Table 1.3 to singularities of
the classical potential theory, Model 4, which analogously holds for Euler solutions,
too, Chap. 5.1 This means that in general viscous phenomena can be explained in
terms of potential theory and vice versa. In particular the circulation theory of lifting
airfoils and finite-span wings can be connected to viscous phenomena, which carry
either kinematically active or inactive vorticity.
To any vortex-flow singularity a unique amount of vorticity can be assigned. This
mainly is a question how the limit Rer e f → ∞ is obtained. Behind this are the pre-
requisites of Model 2: high Reynolds numbers, and no strong or global interactions.

1 See also the introductory discussion in Sect. 1.4.

© Springer-Verlag GmbH Germany, part of Springer Nature 2021 93
E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_4
94 4 The Local Vorticity Content of a Shear Layer

The presentation and discussion of the local vorticity content of a shear layer in
this chapter follows the publications [1–6].
The chapter has four sections. In the first two the concept of the local vorticity
content of shear layers is introduced. The next section gives an interpretation of lift
and induced drag as breaks of symmetry, followed in the last section by a discussion of
the flow pattern at the trailing edge of large aspect-ratio lifting wings. A compatibility
condition is introduced for this purpose.
In Chap. 5 the matter of discrete numerical solutions of the Euler equations (Model
8) for lifting-wing flows is treated. Discussed in particular are the appearance of
vorticity and the entropy rise in such solutions.
Applications regarding the different aspects of the local vorticity content regarding
large aspect-ratio wings then are given in Chap. 8 and regarding small aspect-ratio
delta-type wings in Chap. 10.

4.1 Definition and Derivation of the Local

Vorticity-Content Vector

Figure 4.1 shows a generalized shear layer, which may be either a laminar or a
turbulent one. The curvilinear orthogonal x, y, z-coordinate system has been placed
with its origin P0 onto the skeletal surface of the shear layer. The x, y-coordinates lie
in this surface. The coordinate z is rectilinear and normal to the surface. The velocity
components are u, v, w. The x  , y  , z  ,-system is the reference coordinate system [7].
The external inviscid velocity vector V eu is located at the upper edge of the shear
layer at z = δu , and the vector V el at the lower edge at z = δl . (We keep the vector
notation of the original publications.) Note that generally the upper and the lower
edge of the shear layer are not stream surfaces [7].
The following considerations assume a negligible curvature of the shear layer, so
that for convenience the considerations can be made in Cartesian coordinates. For
an exact approach see, e.g., [5].
We assume high Reynolds-number flow and consequently a very thin shear layer,
i.e., a small extent in the direction normal to the surface, the z-direction, and a small
velocity w in that direction. Hence we introduce the boundary-layer stretching [7],
 
z̃ = z Rer e f , w̃ = w Rer e f . (4.1)

z̃ and w̃ are dimensional entities. The reference Reynolds number is

u r e f ρr e f L r e f
Rer e f = . (4.2)
μr e f

Like in boundary-layer theory now all velocities and lengths are non-
dimensionalized with vr e f and L r e f , respectively, and the density with ρr e f and the
viscosity with μr e f .
4.1 Definition and Derivation of the Local Vorticity-Content Vector 95

Fig. 4.1 Element of a generalized shear layer in orthogonal curvilinear coordinates [3]. The coor-
dinate x generally denotes the main flow direction

The vorticity vector in Cartesian coordinates reads

 
∂w ∂v ∂u ∂w ∂v ∂u
ω = r ot V = [ωx ; ω y ; ωz ] = − ; − ; − . (4.3)
∂y ∂z ∂z ∂x ∂x ∂y

Applying the stretching leads to

  
1 ∂ w̃ ∂v ∂u 1 ∂ w̃ 1 ∂v ∂u
ω= Rer0.5 − ; − ; 0.5 − . (4.4)
ef
Rer e f ∂ y ∂ z̃ ∂ z̃ Rer e f ∂x Rer e f ∂x ∂y

We introduce now the local vorticity content of a shear layer, the vorticity-content
vector Ω, as—at a given location x—the integral of the vorticity across the shear
layer in z-direction [3]:
z̃=δ̃u
Ω = [Ωx ; Ω y ; Ωz ] = ω d z̃. (4.5)
z̃=δ̃l

With Eq. (4.4) we finally obtain for the high Reynolds-number limit Rer e f → ∞
(without the stretching indications at the lower and the upper bound)

Ω| Rer e f →∞ = [−v; u; 0]δδlu . (4.6)

96 4 The Local Vorticity Content of a Shear Layer

Note that the shapes of the functions u(z̃) and v(z̃) do not play a role. However,
the functions must be continuous and they must continuously blend into the external
inviscid flow u e (δl ), ve (δl ) and u e (δu ), ve (δu ).

4.2 Kinematically Active and Inactive Vorticity Content:

Examples

The vorticity content of a shear layer in terms of the components of Ω can be non-
zero or zero. In the first case we call the vorticity content kinematically active, in the
second case kinematically inactive.
Kinematically active vorticity globally influences the flow field (vortex dynam-
ics), whereas kinematically inactive vorticity locally cancels out. We illustrate these
two kinds of vorticity content with several examples. We consider boundary layers
and airfoil wakes. First the Rankine vortex is discussed, then the two-dimensional
boundary layer and the lifting airfoil together with its near wake both in subcritical
and supercritical steady motion. The discussion of the three-dimensional boundary
layer and the trailing vortex layer of the lifting finite-span wing follows. A summary
closes the section.

4.2.1 Rankine Vortex

We begin with the seemingly trivial application of the vorticity content concept
to the Rankine vortex, Fig. 4.2. Actually it is the application of Stokes’ theorem,
Sect. 3.3. The vortex is idealized in the sense of Model 2, Table 1.3. The vorticity
content vector—now without the high Reynolds-number limit and in cylindrical
coordinates—reads

r0 2π r0 2π
1 1 d(r v)
2 ω r dϕ dr = 2 r dϕ dr = Γ = 2πr0 v0 . (4.7)
0 0 0 0 2 r dr

The vorticity content hence is equal to the circulation Γ . That of course is the line
integral around the vortex core: Γ = 2πr0 v0 .
The result means that with a given circulation Γ , i.e., a given amount of vorticity
in the vortex core, the core can have any diameter, because r0 v0 = constant.
To a given potential flow singularity with circulation Γ , Model 4, thus belongs
a finite vorticity content. Although the diameter of the singularity is zero and v0
→ ∞, the above vorticity content can be assigned to it. In the reality, Model 1,
however, the core can have any diameter because due to diffusion the diameter may
grow, Sect. 3.10, and under certain circumstances the vortex even may break down,
Sect. 3.13.
4.2 Kinematically Active and Inactive Vorticity Content: Examples 97

Fig. 4.2 Schematic of the Rankine vortex and its radial velocity (v) and circulation (Γ ) distribution
[3, 4]

We can suppose already that a connection exists to the bound vortex of a lifting
airfoil, as it is found in the frame of potential theory. Before we treat this case in
Sect. 4.2.4, we look at the vorticity content of the boundary layer, Sect. 4.2.2, and of
the airfoil wake, Sect. 4.2.3.

4.2.2 Two-Dimensional Boundary Layer

In Fig. 4.3 the profile u(z) of a two-dimensional boundary layer is shown. The external
inviscid streamline is oriented along the x-coordinate of the element of the general-
ized shear layer shown in Fig. 4.1.
The skeletal surface of that element (z = 0) now constitutes the solid body surface
with the no-slip condition u (z = 0) = 0. The upper bound δu is the boundary-layer
thickness δ, where u (z = δ) = u e .2
The vorticity content vector reads

Ω| Rer e f →∞ = [0; u e ; 0] . (4.8)

The result is that the boundary layer has a finite local vorticity content. The local
vorticity vector points normal to the plane of the two-dimensional boundary layer in
the y-direction.

2 Regarding the definition of the boundary-layer thickness see Appendix A.5.4.

98 4 The Local Vorticity Content of a Shear Layer

Fig. 4.3 Schematic of the

tangential velocity profile
u(z) of a two-dimensional
boundary layer [3, 4]

4.2.3 Near Wake of a Lifting Airfoil

Figure 4.4 shows two idealized airfoil near wakes.3 Case (a) stands for an airfoil
in steady subcritical motion, and case (b) for an airfoil in steady, but supercritical
motion, with a supersonic flow pocket and a terminating shock wave at the upper
(suction) side. At the trailing edge in both cases the static pressure at the lower and
the upper side of the wake is the same, regardless whether the airfoil is lifting or
not. This also holds for both wakes in the limit Rer e f → ∞, even if they are slightly
curved.
In case (a) the equal pressures lead to the same velocity of the external inviscid
flow at the lower and the upper side of the wake. The vorticity content hence is

Ω| Rer e f →∞ = [0; 0; 0] . (4.9)

Fig. 4.4 Schematic of

two-dimensional wakes [3,
4]: a wake of an airfoil in
steady subcritical motion, b
wake of an airfoil in steady
supercritical motion

3A discussion of the flow at the trailing edges of real airfoils and wings is given in Chap. 6.
4.2 Kinematically Active and Inactive Vorticity Content: Examples 99

Fig. 4.5 The wake of Fig. 4.4b in the limit Rer e f → ∞ as the two-dimensional vortex sheet (slip
line) of potential theory [3, 4]

Wake (a) therefore is considered to be kinematically inactive, i.e., the vorticity

leaving the trailing edge cancels out. This holds for both the lifting and the non-lifting
airfoil. Only during lift changes, due for an angle of attack change or for a speed
change, does the wake carry kinematically active vorticity.
For the lifting airfoil this is in accordance with potential theory, Model 4. (The
non-lifting case is trivial.) Once the steady motion of the airfoil has been established,
the bound vortex of the airfoil and the opposite starting vortex at infinity behind the
airfoil do no more change the strength of their circulations. The wake by no means
must be symmetric, it is only the two edge values of the external inviscid velocity
which are symmetric.
In case (b) the total pressure loss due to the terminating shock wave at the upper
side of the airfoil leads to different external inviscid velocities at the lower and the
upper side of the wake, as the static pressure has to be the same. In this case a finite
local vorticity content is present:


Ω| Rer e f →∞ = 0; u eu − u el ; 0 . (4.10)

In the high Reynolds-number limit thus the wake collapses into the two-
dimensional vortex sheet (slip line) of potential theory, Fig. 4.5. The kinematically
active vorticity content Ω y is hidden in this layer.
This case needs a closer inspection, because also in the steady supercritical lifting
case no kinematically active vorticity is to leave the airfoil via its wake. This must
be seen in the following way. Consider Fig. 4.6. A supersonic flow pocket is present
at the suction side of the airfoil. The terminating shock wave leads to a total pressure
loss. Hence we find at the trailing edge of the airfoil a smaller external inviscid
velocity at the upper side than at the lower side of the wake. The result is a wake
flow situation like that of case (b) in Fig. 4.4.
100 4 The Local Vorticity Content of a Shear Layer

Fig. 4.6 Schematic of the flow at the trailing edge of a lifting airfoil in steady supercritical motion
[5]

Table 4.1 Parameters of the CAST 7 airfoil case [8]

M∞ Re L ,∞ L (m) T∞ (K) x/L|trans. α (◦ )
0.7 4·106 1 300 0.07 2

If we only consider this wake, we would conclude that kinematically active vor-
ticity leaves the airfoil’s trailing edge, which, however, is not permitted, because we
consider the lift as being unchanging. In this case the effect of the shock wave must
be taken into account. Actually not only the boundary-layer wake is to be considered,
but also the shock wake. The two together constitute a vorticity or entropy wake,
which indeed does carry only kinematically inactive vorticity, Fig. 4.6.
Hence we postulate that the vorticity created by the shock wave must cancel the
kinematically active vorticity of the boundary layers, which leaves the trailing edge4 :
z=δs

Ω|s = r ot v dz = 0; −(u eu − u el ); 0 . (4.11)
z=δu

For kinematic reasons thus for lifting wings at transonic (and supersonic) speeds,
wake considerations must include the shock wake. Obviously the bounds of the
boundary-layer wake are adjusted to constitute the much thicker vorticity or entropy
wake behind the airfoil or wing. The mechanism which brings about the necessary
shock form and its change of strength to achieve the cancelation, is not known.
In the doctoral thesis on self-adaptive grid refinement by J. Fischer, a student of
the first author of this book, besides others the viscous flow past a supercritical airfoil
was studied [8]. The considered airfoil was the so-called CAST 7 airfoil. A large
number of experimental data is available for this airfoil [9]. The parameters chosen
in the thesis are given in Table 4.1.
Figure 4.7 shows the airfoil with the generated iso-Mach lines (viscous solu-
tion). Despite the presence of the boundary layer, the terminating shock wave at

4 Theamount of vorticity which must be canceled is small compared to the amount of vorticity of
each of the involved boundary layers, usually only a few per cent.
4.2 Kinematically Active and Inactive Vorticity Content: Examples 101

Fig. 4.7 Cast 7 airfoil, viscous case [8]: Mach number isolines, Mmin = 0, Mmax = 1.3, M =
0.02. The broken line is the sonic line M = 1

the upper side is seen to impinge orthogonally on the airfoil’s surface, as inviscid
theory demands [10]. Away from the surface the shock wave is slightly curved, the
pre-shock Mach number diminishes away from the surface.5 The shock wave finally
tapers off and at its upper end it blends into the sonic line.
We discuss now the vorticity wake at the trailing edge of the CAST 7 airfoil, x =
L, see Fig. 4.6. Because the velocity difference between the lower and upper external
inviscid flows at the trailing edge is small, in [8] only the vorticity wake of the inviscid
case was considered. Nevertheless, the result illustrates well the situation.
We look at the y-component of the vorticity-content vector Ω. The simplification
of the vorticity content integral, Eq. (4.6), is not adequate in this case, because we
do not consider a thin boundary layer. Instead the full y-component of the vorticity
must be taken into account:
x=L ,z=0.55L x=L ,z=0.55L  
∂u ∂w
Ωy = ω y dz = − dz. (4.12)
x=L ,z=−0.55L x=L ,z=−0.55L ∂z ∂x

The lower and the upper bound of the vorticity integral were placed at z = −0.55
L, and z = 0.55 L. Figure 4.8 shows the evolution of the integral Ω y . We look only at
the solution on the fully refined grid, represented by the broken line. Between z/L =
−0.55 and the trailing edge (z/L = 0) the value of the integral is zero. At the lower

5 These properties are important, because behind a straight shock wave with constant pre-shock
Mach number, a total pressure loss is present, but no vorticity. See in this regard Crocco’s theorem,
Sect. 3.5.
102 4 The Local Vorticity Content of a Shear Layer

Fig. 4.8 Cast 7 airfoil [8]:

development of the integral
Eq. (4.12) at the airfoil’s
trailing edge,
x=L

side of the airfoil no shock wave is present, hence no vorticity. At z/L = 0 a steep
descent to Ω y ≈ −7.5 m/s occurs.6
For z/L > 0 the integral rises and finally reaches Ω y ≈ 0. Although the solution
exhibits wiggles and Ω y does not exactly reach zero, this result can be seen to prove
that the vorticity wake is kinematically inactive. A new investigation with today’s
available algorithms and computer power appears to be desirable.
In summary we state that the embedded wake profile, Fig. 4.4b, does not change
the conclusion. The vorticity content of the boundary-layer wake Ω y = u eu − u el is
compensated by the vorticity content of the wake of the shock wave. The whole wake,
i.e., the vorticity or entropy wake—indicated in Fig. 4.6—in steady supercritical flow
is kinematically inactive, as is demanded.

4.2.4 Bound Vortex of a Lifting Airfoil

Consider the flow situation over a lifting airfoil in steady subcritical motion, Fig. 4.9.
The flow is two-dimensional throughout. As we have seen above, a two-dimensional
boundary layer has a finite kinematically active vorticity content, which locally is—in
the high Reynolds-number limit—Ω| Rer e f →∞ = [0; u e ; 0]. The wake downstream
of the trailing edge, on the other hand, does carry only kinematically inactive vorticity,
as was discussed in the preceding sub-section. Hence it is no more considered.
We determine the kinematically active vorticity content of the two boundary lay-
ers over the length of the airfoil L in the high Reynolds-number limit by taking
the double integral over the boundary-layer thicknesses (z-direction) and along the

6 Note that a discrete (Model 8) Euler solution close to a solid surface always exhibits a thin
total-pressure loss layer, casually called Euler boundary layer. That is an artifact due to the finite
discretization of the computation domain and the flow variables used.
4.2 Kinematically Active and Inactive Vorticity Content: Examples 103

Fig. 4.9 Schematic of the boundary layers and the wake at an lifting airfoil in steady subcritical
motion [3]

boundary/shear-layer edges (s-direction). The integration is made in the sense of

Model 2, not in the sense of Model 1 or 9, because the trailing-edge interaction—and
its upstream influence—between the boundary layers and the external inviscid flow
is not taken into account:
s=L z=δl ,δu s=L
u(z, s) ds dz| Rer e f →∞ = u e ds = Γ. (4.13)
s=0 z=0 s=0

The conclusion is that the double integral of the vorticity in the boundary layers
for Rer e f → ∞ reduces to the line integral of the external inviscid velocity of the
boundary layers which gives the circulation Γ of the bound vortex of the lifting
airfoil.
This is an application of Stokes’ theorem via the concept of the local vorticity
content of a boundary layer. The result was given first in 1986 by M.J. Lighthill [11],
as we mentioned it in Sect. 1.4.
The interpretation is that the two boundary layers in terms of their vorticity content
over the length of the airfoil constitute the rotational core of the bound vortex.
Actually it is the excess of the vorticity content of the upper over that of the lower
boundary layer. This is also evident in terms of the line integral. At the upper side of
the airfoil the external inviscid velocity is larger than at the lower side. The excess
accounts for the actual amount of circulation.
For a better understanding it may help to consider the case of a symmetric airfoil
at zero angle of attack and hence with zero lift. Then the vorticity content at the lower
side of the airfoil has the same amount as that at the upper side, but with an opposite
sign. It cancels out. The same holds for the external inviscid velocity. Consequently
the circulation and hence the lift are zero.
104 4 The Local Vorticity Content of a Shear Layer

Fig. 4.10 Schematic of the

profile of a
three-dimensional boundary
layer in external inviscid
stream-line oriented
coordinates [7]

4.2.5 Three-Dimensional Boundary Layer

In Fig. 4.10 the profile v(z) of a three-dimensional boundary layer is shown.7 The
profile is given in the external inviscid streamline-oriented (t, n) coordinate system
[7]. (The t-coordinate locally lies tangential to the external inviscid streamline, the n-
coordinate normal to it. This coordinate system is kind of a natural coordinate system
for three-dimensional boundary layers. Other systems, of course, can be employed.
The result below will be same.) The external inviscid streamline (t-direction) is
oriented along the x-coordinate of the element of the generalized shear layer shown
in Fig. 4.1.
The velocity profile v(z) can be decomposed into the main-flow profile (stream-
wise) vt (z), which resembles the profile of a two-dimensional boundary layer, and
into the cross-flow profile vn (z), which can have quite different shapes [7]. In any
case vn is zero at the body surface (z = 0) and at the outer edge of the boundary layer
(z = δ).
The skeletal surface of the element of the generalized shear layer in Fig. 4.1 (z
= 0) again constitutes the solid body surface with the no-slip condition vt (z = 0)
= 0, vn (z = 0) = 0. The upper bound δu is the boundary-layer thickness δ, where
vt (z = δ) = |V e | = vte (main-flow profile), vn (z = δ) = 0 (cross-flow profile).
The vorticity content vector in the high Reynolds-number limit reads


Ω| Rer e f →∞ = 0; |V e | = vte ; 0 . (4.14)

The result is that the three-dimensional boundary layer has a finite local vorticity
content which is that of the main-flow profile. The cross-flow profile’s vorticity

7 Note that the stream surface of the boundary layer does not lie in a plane, as in the two-dimensional

case, Fig. 4.3. Even more complex forms of the stream surface are possible [7].
4.2 Kinematically Active and Inactive Vorticity Content: Examples 105

content is zero. The local vorticity-content vector Ω points in the n-direction, i.e.,
the direction normal to the t-z-plane.

4.2.6 Near Wake (Trailing Vortex Layer) of a Lifting

Finite-Span Wing

Consider the flow past the canonical lifting wing in Fig. 1.10a. Intuition tells us—
because of the finite span of the wing—that a pressure relaxation takes place between
the high pressure on the lower side and the low pressure on the upper side. This
relaxation leads at the lower side to a general flow deflection toward the wing’s tip
and at the upper side away from it toward the wing’s root, see Sect. 4.3.2. The result
is that the inviscid wing upper and lower flow fields are shearing.
We look at the flow as it locally leaves the trailing edge of a back-swept finite-
span lifting wing in steady subcritical motion, Fig. 4.11. (The shown s-t-coordinate
system is the local wake coordinate system. It may not lie in the datum plane of
the wing. In particular near the wing tip this must be regarded.) Because we discuss
on the basis of Model 2, we disregard strong interaction phenomena and a possible
finite, if small, thickness of the trailing edge (Sect. 6.3). The flow over the lower side
generally has a direction toward the wing tip, that over the upper side away from the
wing tip. We assume that the velocity vectors in the figure lie in the skeletal plane of
the trailing vortex layer, which leaves the trailing edge.
Because we have assumed subcritical motion, the total pressure, like the static
pressure, is considered to be the same on the upper and the lower side of the trailing
edge. The magnitudes of the external inviscid velocity vectors at the upper side (V eu )
and at the lower side (V el ) of the trailing edge are the same: |V eu | = |V el |.
We define the angle between the two external inviscid velocity vectors V eu and
V el as the (local) trailing-edge flow shear angle ψe . Locally the magnitude of the

Fig. 4.11 Local wake

(trailing vortex layer)
coordinate system (view
from above onto the upper
surface of the wing near the
trailing edge) [7]: idealized
situation of the flow at the
trailing edge of a back-swept
finite-span lifting wing in
steady subcritical motion.
The coordinate y points in
span direction, the
coordinate x in free-stream
or chord direction
106 4 The Local Vorticity Content of a Shear Layer

Fig. 4.12 Schematic of the decomposed near-wake structure of a backward-swept finite-span lifting
wing [3, 4]. Left: local wake coordinate system; right: wake in idealized reality

shear angle is a measure of the strength of the trailing vortex layer leaving the trailing
edge, Sect. 4.4.
The angle ε between the s-direction and the chord direction x, i.e., the (local)
vortex-line angle, is small but not necessarily zero. Its sign is governed by the local
sweep of the trailing edge [12]: positive (toward the wing tip) for back-swept trail-
ing edges, negative (toward the wing root) for forward swept trailing edges. The
magnitude of ε at given lift depends on the thickness of the wing.8 We note that
in potential-flow theories of lifting wings, for instance lifting-surface theories, the
vortex-line angle ε is assumed to be zero, see, e.g., [13].
The structure of the wake between the sheared external inviscid flows at the lower
and the upper side directly at the trailing edge is complex. We show it schematically
in a decomposed manner in Fig. 4.12.
The two vectors of the external inviscid flow V eu and V el are decomposed in such
a way that they have the components u eu = u el in s-direction (bi-sector direction) and
veu = −vel in t-direction (normal to the bi-sector direction). The angles ψeu and ψel
are found from
veu ve
tan ψeu = , tan ψel = l , (4.15)
u eu u el

where the u e are the components of the two velocity vectors in s-direction and the
ve those in n-direction. Note that ψeu = −ψel and, Fig. 4.11, ψe = ψeu + ψel .

8 See the examples in Chap. 8.

4.2 Kinematically Active and Inactive Vorticity Content: Examples 107

Fig. 4.13 The wake of

Fig. 4.12 right in the limit
Rer e f → ∞ as vortex sheet
of potential theory [3, 4]

The velocity component u(z) in s-direction resembles the (two-dimensional) wake

of a lifting or non-lifting airfoil in steady subcritical motion. Locally we can relate
to this 2-D wake the viscous drag, i.e., the skin-friction drag plus the form drag.
The velocity component v(z) in t-direction, on the other hand, resembles a vortex
with the axis in s-direction. To this we can relate the induced drag. This is a topic of
Sect. 4.3.2.
The vorticity content vector in the high Reynolds-number limit reads



Ω| Rer e f →∞ = Ωs ; Ωt ; Ωz = −(veu − vel ); 0; 0 , (4.16)

and with veu = −vel because of the bisector orientation



Ω| Rer e f →∞ = 2vel ; 0; 0 = −2veu ; 0; 0 . (4.17)

This means that the 2-D wake-like profile u(z) indeed has a zero vorticity content,
whereas the vortex-like profile v(z) has a finite vorticity content.
Figure 4.13 shows the decomposed wake, Fig. 4.12 right, in the high Reynolds-
number limit collapsed into the vortex sheet of potential theory.
The kinematically active vorticity content Ωs is hidden in this layer. The s-
coordinate as a streamline obviously represents a vortex line [11], because there
with u eu |t=0,z=0 = u el |t=0,z=0 being considered as the flow vector V e |t=0,z=0

Ω| Rer e f →∞ × V e |t=0,z=0 = 0. (4.18)

4.2.7 Summary of the Results

We summarize the results:

1. In the frame of Model 2 the concept of the local vorticity content of a shear layer
yields in the limit Rer e f → ∞ that this content is equal to the external inviscid
108 4 The Local Vorticity Content of a Shear Layer

velocity in the case of boundary layers, or to the difference of these velocities in

the case of wakes. The thickness of the shear layer goes to zero for Rer e f → ∞.
2. The shapes of the velocity profiles in the shear layer do not play a role, provided
the profiles are continuous up to the second derivative.
3. The concept of the local vorticity content permits to distinguish between kinemat-
ically active and inactive vorticity. The former influences the surrounding flow
field (vortex dynamics), the influence of the latter locally cancels out.
4. Wakes of airfoils—non-lifting or lifting—in steady motion carry no kinematically
active vorticity.
5. In the case of an airfoil or a finite-span wing in steady supercritical motion the
concept of the local vorticity content leads to the conclusion that the vorticity
wake behind the—by necessity variable in strength—shock wave must be taken
into account regarding the wake properties.
6. Wakes, i.e., trailing vortex layers, of lifting finite-span wings in steady motion
locally can be decomposed into a kinematically active and a kinematically inactive
part. The inactive part along the main-flow direction can be considered to locally
carry the viscous drag, and the active part in the cross-flow direction the induced
drag.
7. The concept of the local vorticity content permits to connect a singularity of
potential theory with a finite amount of vorticity, which is hidden in the singularity.
This holds for vortices, boundary layers and wakes, which in the high Reynolds-
number limit become vortex singularities and vortex sheets, respectively.

4.3 Lift and Induced Drag: Two Breaks of Symmetry

Preliminary remark: symmetry in this section has two meanings. First it is meant in
the sense that the surface pressure distribution is such that no net force acts on the
airfoil or wing, and secondly it is also meant in the classical geometrical sense.

4.3.1 First Symmetry Break: The Lifting Airfoil

In 1752 d’Alembert did prove that no net force is exerted on a body, which steadily
moves through an incompressible and inviscid fluid. This situation is described by
potential flow theory (Model 3, Table 1.3).
An airfoil in such a potential flow schematically is shown in Fig. 4.14. The
stagnation-point streamline impinges vertically on the airfoil’s surface. At the stag-
nation point a half-saddle point (S1 ) is present, as well as at the rearward stagnation
point (S2 ), where the flow leaves the surface, again normal to the surface, see Chap. 7.
The flow turns around the sharp trailing edge. In potential flow this means an infinitely
large speed, which, however, is permitted in such a flow model.
4.3 Lift and Induced Drag: Two Breaks of Symmetry 109

Fig. 4.14 Schematic of steady incompressible and inviscid flow past an airfoil. No net force is
exerted on the airfoil

Fig. 4.15 Schematic of

steady incompressible and
inviscid flow past an airfoil
with a Kutta condition. A lift
force is exerted on the airfoil

The flow far behind the airfoil has the same direction as that ahead of the airfoil,
i.e., no momentum flux is deflected downwards. This indicates that no airfoil lift is
present. That no drag is present, is not indicated. That would have required to show
that no wake exists.
We consider the flow situation in Fig. 4.14 as being symmetrical. A break of
that symmetry happens, if a Kutta condition is inserted at the airfoil’s trailing edge,
Fig. 4.15. In 1902 Kutta introduced this feature into potential lift theory. The Kutta
condition reflects what is observed in reality: the flow leaves smoothly the sharp
trailing edge of an airfoil, flow-off separation.
The basic topology of the surface flow field is as in Fig. 4.14. Now, however, the
rearward stagnation point (S2 ) lies at the trailing edge. If the trailing edge has a finite
opening angle and the flow leaves the surface in bi-sector direction, the velocity at
S2 —in the frame of potential theory—indeed is zero.
That a lift force is present, is indicated by the downward deflection of the flow
behind the airfoil. However, no drag is present. (That there is no wake is not indicated.)
Circulation theory, Model 4 in Table 1.3, models the flow past an airfoil with
a Kutta condition by, for instance, combining the undisturbed free-stream with a
potential vortex, the bound vortex, see, e.g., [13]. The bound vortex has a circulation
Γ with a strength such that the induced velocity shifts the rearward stagnation point
S2 from its location on the upper side of the airfoil, Fig. 4.14, to the trailing edge,
Fig. 4.15. In this way flow-off separation at the trailing edge is enforced. The resulting
lift force per unit span l is

l = ρ∞ u ∞ Γ.
110 4 The Local Vorticity Content of a Shear Layer

4.3.2 Second Symmetry Break: The Lifting Finite-Span Wing

Consider the surface stream-line pattern of the inviscid incompressible flow past the
wing-like 3:1:0.125 ellipsoid shown in Fig. 4.16. The angle of attack is α = 15◦ .9
We look from above at the right-hand side half-span of the ellipsoid, 0  y 
b/2. The forward stagnation point is located on the lower side (the pressure side)
below—in this view—the apex (x = 0, y = 0). Also located on the lower side is the
attachment line. From this line the streamlines diverge on the lower side (pressure
side) in almost chord direction and around the leading edge on the upper side (suction
side) also in almost chord direction.
At the trailing edge we see—on the upper side of the ellipsoid—the convergence
of the streamlines into the detachment line, Sect. 7.1.3. The streamlines on the upper
side converge directly into that line, those from the lower side also, around the
trailing edge. The detachment line enters the rear stagnation point, which lies—in
this view—on the upper side above the antapex (x = c (y = 0), y = 0).
We observe in the view from above that overall a shear is present in the streamline
pattern—the wing upper and lower side flow-fields shear—, which at the trailing
edge manifests itself in the shear angel ψe between the streamlines over the lower
and the upper side of the ellipsoid. This shear angle is zero in the symmetry plane (y
= 0) and increases toward the wing’s tip.

Fig. 4.16 Surface streamlines of the inviscid incompressible flow past a wing-like 3:1:0.125 ellip-
soid at angle of attack α = 15◦ (Model 3) [5]. The z-coordinate lies normal to the x, y-coordinates

9 The figure is based on a computation case in [14]. The flow field was computed with exact potential

theory (Model 3 of Table 1.3) [15, 16].

4.3 Lift and Induced Drag: Two Breaks of Symmetry 111

The origin of the shear can be imagined in the following way. Because we deal
with a linear potential flow problem, we find the streamline pattern in Fig. 4.16 by
superimposing the solution for the free-stream component v∞,x = cos α v∞ with that
for the component v∞,z = sin α v∞ , see also Sect. 7.4.3. For zero angle of attack no
shear at all is present. With lift at a given span location y > 0 the shear increases
with increasing angle of attack, attains a maximum, and for 90◦ angle of attack again
it becomes zero.
We now come back to Model 1. Results from experimental and theoretical/numeri-
cal flow-field investigations on several configurations suggest the existence of a local-
ity principle, Sect. 2.3. In our context it says that a local change of a body shape and
the resulting flow separation affects the flow only at that location and downstream
of it. Of course the flow is changed also upstream—in subsonic flow fields due to
their elliptical properties—but these changes are small in general. However, although
being small, they can be non-negligible, if the ensuing flow wake carries kinemati-
cally active vorticity, which leads, for instance, to the downwash at the location of
the lifting wing and to the induced drag.
If the trailing edge of our wing-like ellipsoid is sufficiently thin and the flow
is sufficiently real, the latter undoubtedly will break away from the surface at the
trailing edge in the general form of flow-off separation.10
Because of the locality principle the flow upstream of the trailing edge will retain
the general shear between the lower and upper (inviscid) streamlines in terms of the
(local) trailing-edge flow shear angle ψe (= ψe,u + ψe,l ), Sect. 4.2.6. That shear—i.e.
prior to the convergence to the potential-flow detachment line in Fig. 4.16—then
extends downstream into the wing’s wake—the trailing vortex layer—leading to a
flow geometry like that sketched in Fig. 4.17. This is the effect of the second break
of symmetry.
In Fig. 4.17 also indicated is the local wake coordinate system, Sect. 4.2.6,
Fig. 4.11. The indicated (local) vortex-line angle  is negative, because the trail-
ing edge at that location is swept forward. The skeleton surface of the trailing vortex
layer is assumed to lie initially in the wing plane, not in the Kutta direction, Sect. 6.2.
We summarize our observations. They have been obtained by discussions of a
wing-like ellipsoid, however, they hold for every finite-span steadily moving lifting
wing:
1. The inviscid streamlines (Model 3) on the surface of the wing-like ellipsoid at
angle of attack without a Kutta condition are (odd) symmetrical with regard to the
x ≈ 0.5 c, y, z - plane. This also holds for realistic wing shapes, then however the
symmetry is not a strictly geometric one. No net force acts on the configuration.
2. If the flow is a realistic one (Model 1), it does not turn around the trailing edge
but breaks away from it (flow-off separation). This is in accord with the Kutta
condition, which represents the first break of symmetry, leading to a lift force. At
the same time also the (odd) symmetry of the inviscid streamlines with regard to
the x ≈ 0.5 c, y, z - plane is broken: second break of symmetry.

10 Trailing edge properties as well as the Kutta condition in reality are treated in Chap. 6.
112 4 The Local Vorticity Content of a Shear Layer

Fig. 4.17 Surface streamlines of the inviscid incompressible flow past a wing-like 3:1:0.125 ellip-
soid at angle of attack α = 15◦ : expected real inviscid flow pattern at and downstream of the trailing
edge (Model 1) [5]. The “vortex layer” is the wing’s trailing vortex layer. The z-coordinate lies
normal to the x, y-coordinates

3. The second break of symmetry, however, retains an important property of the

inviscid surface streamline pattern of Model 1, viz. the shear of the inviscid
streamlines, the wing upper and lower flow-fields shear. The shear in terms of
the (local) trailing-edge flow shear angle ψe in Model 1 is approximately that of
Model 3 ahead of the rear convergence to the detachment line shown in Fig. 4.16.
4. The shear due to the second break of symmetry introduces into the wake—due
to the flow-off separation—a kinematically active vorticity content, Sect. 4.2.6.
The result is the induced kinematically active wake, the trailing vortex layer, and
hence the downwash at the location of the wing and the induced drag. The induced
drag is a generic property of the inviscid (!) flow past the lifting wing, not of the
separating three-dimensional boundary layers at the upper and the lower side of
the wing, which constitute the wake flow. Hence it can be found with any properly
formulated inviscid flow model, in particular with the methods of potential flow
theory (Model 4) and with discrete numerical Euler methods (Model 8) as long
as the Kutta condition implicitly or explicitly is present.

4.3.3 The Symmetry Breaks in the Reality of Aircraft Wings

Regarding the situation in the reality of lifting wings we make two observations:
1. The Kutta condition is very complex, Chap. 6. Present are (a) the in spanwise
direction varying trailing-edge flow shear angle ψe , (b) the vortex-line angle ,
4.3 Lift and Induced Drag: Two Breaks of Symmetry 113

which may be constant only over a portion of the trailing edge, (c) decambering
effects, (d) a finite thickness of the trailing edge, and (e) at delta-type wings swept
round leading edges.
2. The shear of the inviscid flow fields at the upper and lower side of the lifting wing
is not so obvious as in the case of our flat ellipsoid. This in particular holds for
small aspect-ratio delta-type wings. There the lee-side vortex system, which is
present at higher angels of attack, Chap. 10, completely changes the picture. At
the lifting large aspect-ratio wing, on the other hand, we get the tip-vortex system,
Sect. 8.4.4, which has its own influence, which becomes larger with smaller aspect
ratio.

4.4 Flow Pattern at the Trailing Edge of Large-Aspect

Ratio Lifting Wings: A Compatibility Condition

We study the connection of the local kinematically active vorticity content, which
leaves the trailing edge, with the spanwise circulation distribution of the lifting wing.
This connection is described by a compatibility condition. Our objective is to link
aspects of Model 4 to aspects of Model 2 of Table 1.3.
In Sect. 4.2.6 we locally decomposed the flow at the trailing edge into a 2-D wake-
like profile u(z) and a vortex-like profile v(z), Fig. 4.12. We found with Eq. (4.17)




Ω| Rer e f →∞ = Ωs ; Ωt ; Ωz = 2vel ; 0; 0 = −2veu ; 0; 0

that the vortex-like profile has a finite—kinematically active—vorticity content,

whereas the 2-D wake-like profile has zero vorticity content—hence is kinemati-
cally inactive.
Figure 4.13 shows the decomposed wake of Fig. 4.12 right, in the high Reynolds-
number limit collapsed into the vortex sheet of potential theory. The kinematically
active vorticity content Ωs is hidden in this layer.
We ask now how the kinematically active vorticity content Ωs locally is connected
to the span-wise circulation distribution Γ (y) of the wing. The overall situation over
a lifting wing is sketched in Fig. 4.18. Here Γ0 is the so-called root circulation.
Prandtl in his lifting-line wing model located the trailing vortices, fully developed,
at the wing tips [18], see the discussion of this matter in Sect. 3.4. In reality, however, a
vortex sheet leaves the trailing edge, with in span-wise direction increasing strength—
the trailing vortex layer in Fig. 4.18.
This sheet rolls up to the two discrete trailing vortices. The (initial) horizontal
distance b0 of the axes of the fully developed trailing vortices is smaller than the
wing span b. For a wing with elliptical circulation distribution along the wing span,
the spanwise load factor, Sect. 3.16, reads
114 4 The Local Vorticity Content of a Shear Layer

Fig. 4.18 Schematic of a lifting wing: depiction of span-wise circulation distribution Γ (y), trailing
vortex layer (wake) behind the wing, and the two counterrotating trailing vortices with circulation
Γ0 [17]

b0 π
s= = . (4.19)
b 4
The ratio b0 /b → 1 is only reached for a wing with very large aspect ratio.
The location, where the roll-up process of the trailing vortex layer toward the
pair of trailing vortices is completed, can not be defined exactly. The process is
an asymptotic one, depending on the magnitude of the lift, the span loading and
the wing’s aspect ratio. In the case of large aspect-ratio swept or unswept wings
the approximate location can be a few half-span distances downstream of the wing,
[17], in the case of low-aspect ratio wings the location can be much closer to the
trailing edge at one chord length or less behind the wing [19], see in this regard also
Sect. 8.4.4.
Above we have noted that locally in the trailing vortex layer the s-coordinate,
Fig. 4.13, is both a streamline and a vortex line. The kinematically active vortic-
ity leaving the trailing edge, the trailing vorticity, therefore can be considered as
streamwise vorticity [11].11
Hence, locally the connection of the kinematically active vorticity Ωs and the
circulation Γ (y) obviously is via the span-wise gradient of the latter. If the vortex-

11 Lighthill calls the trailing vorticity also residual vorticity.

4.4 Flow Pattern at the Trailing Edge … 115

line angle  is small, Ωs at a location y must be equal to the change of the circulation
in the wing’s spanwise direction:
dΓ
= −Ωs = −(veu + vel ) = 2 |V eu | sin ψeu , (4.20)
dy

which in [2] was introduced as a compatibility condition.

In principle this is a discussion of the lifting wing’s discontinuity surface of
potential theory, Model 4, from the view of viscous modeled flow, Model 2. However,
it is not necessary to assume  = 0, as in most of the potential-flow wing theories.
Equation (4.20) shows that with increasing dΓ /dy either |V eu | must increase or
the trailing-edge shear angle ψeu . As we have seen above, the second break of sym-
metry essentially and approximately leaves the increase of ψeu in spanwise direction
unchanged. Hence it basically is the change of ψeu , which leads to the fulfillment of
the compatibility condition. This is another aspect of the locality principle, Sect. 2.3.
Regarding the compatibility condition, Eq. (4.20), we shortly look at the situation
at the lifting infinite swept wing. The infinite swept wing and the locally infinite
swept wing are simple and convenient quasi-two-dimensional local approximations
of the geometry of swept wings [7].
Because in the infinite swept wing approximation all flow parameters and in
particular the circulation Γ do not change in spanwise direction, the trailing-edge
shear angle ψeu must be zero. This means that the flow field close to and at the wing’s
trailing edge is different from that of the approximated lifting finite-span wing. No
kinematically active vorticity leaves the trailing edge. Nevertheless, the infinite swept
wing is useful in, for instance, investigations of boundary-layer stability and laminar-
turbulent transition at and close to the leading edge.

4.5 Final Remark

The concept of the kinematically active vorticity permits to connect fundamental

singularities of potential theory (Model 4) with the viscous and vortical flow phe-
nomena found in the perceived reality (Model 1). Partly we have shown that with
very simple examples. The elliptical wing shape of Sect. 4.3.2, for instance, is far
away from that of present-days large aspect-ratio wings. Nevertheless, the concept
permits to demonstrate important flow properties. These are present on realistic wing
shapes, too, as we show in Chap. 8.
The concept moreover permits to assess the capabilities of computation methods
of several model levels. For flows over small aspect-ratio delta-type wings with sharp
leading edges, for instance, it permits to prove that discrete numerical solutions of
the Euler equations, Model 8, indeed lead to an in principle correct presentation of
the primary lee-side vortex pair, Chap. 10.
The introduction of the concept of flow-off separation at sharp trailing or leading
edges on the other hand is the precondition for some of these findings, even if in
116 4 The Local Vorticity Content of a Shear Layer

reality the edges may not be sharp in the exact sense of the word, Chap. 6. The Kutta
condition in the reality of lifting wings, too, is a much more intricate concept as it
usually is perceived. It is important to be aware of these facts when treating flow
problems of lifting wings.

4.6 Problems

Problem 4.1 At the beginning of Sect. 4.1 it is assumed that the considered shear
layer is very thin. Check this for a boundary-layer flow. Assume a flat wing surface
with a chord length of L = 5 m and flight at H = 10 km altitude with the Mach number
M∞ = 0.8. Assume at the wing’s surface the recovery temperature. How thick is the
boundary layer for (a) laminar and (b) turbulent flow at x/L = 0.5 and at x/L = 1,
there with the strong interaction being disregarded. Is the assumption justified?

Problem 4.2 Derive the vorticity-content integral of the Rankine vortex, Sect. 4.2.1.
The z- component of vorticity in cylindrical coordinates reads:
1 d(r vθ )
ωz = .
r dr
Problem 4.3 The upper side of an airfoil often is called the suction side, and the
lower side the pressure side. What actually is the situation? Is there suction in the
sense of the word at the upper side, how comes the lift into being?

Problem 4.4 Prove that the s-direction in Fig. 4.12 is a vortex line.

Problem 4.5 Circulation theory, basically belonging to potential theory (Model 4),
finds the lift of an airfoil with the help of a circulation Γ . How can this be interpreted
physically when looking at Figs. 4.14 and 4.15 in Sect. 4.3. Remember that potential
theory permits to superimpose different flow fields/solutions.

Problem 4.6 In Fig. 4.18 the spanwise circulation distribution is shown with the
circulation Γ0 at y = 0 and a diminishing Γ (y) to the left and the right side. Why
is Γ (y = 0) smaller than Γ0 ? This is reflected mathematically in a condition at the
trailing edge. What is its name?

Problem 4.7 Consider Fig. 4.18. How must the trailing vortex layer be sketched in
the frame of the lifting-line theory, steady movement of the wing assumed? What
kind of vorticity content does it carry? Kinematically active or inactive or both of
them? Where does it come from?

Problem 4.8 In Fig. 4.18, lower part, at the symmetry line y = 0 the vortex layer is
indicated with a finite extent. How can this be interpreted?
References 117

References

1. Hirschel, E.H.: Considerations of the vorticity field on wings. In: Haase, W. (ed.) Recent
Contributions to Fluid Mechanics, pp. 129–137. Springer, Berlin (1982)
2. Hirschel, E.H., Fornasier, L.: Flowfield and vorticity distribution near wing trailing edges.
AIAA-Paper, 1984–0421 (1984)
3. Hirschel, E.H.: On the creation of vorticity and entropy in the solution of the Euler equations
for lifting wings. MBB-LKE122-AERO-MT-716, Ottobrunn, Germany (1985)
4. Hirschel, E.H., Rizzi, A.: The mechanisms of vorticity creation in Euler solutions for lifting
wings. In: Elsenaar, A., Eriksson, G. (eds.) International Vortex-Flow Experiment on Euler
Code Validation. FFA, Bromma (1987)
5. Eberle, A., Rizzi, A., Hirschel, E.H.: Numerical Solutions of the Euler Equations for Steady
Flow Problems. Notes on Numerical Fluid Mechanics, Vol. 34. Vieweg, Braunschweig Wies-
baden (1992)
6. Hirschel, E.H.: Vortex flows: some general properties, and modelling. Configurational and
manipulation aspects. AIAA-Paper, 96–2514 (1996)
7. Hirschel, E.H., Cousteix, J., Kordulla, W.: Three-Dimensional Attached Viscous Flow.
Springer, Berlin (2014)
8. Fischer, J.: Selbstadaptive, lokale Netzverfeinerung für die numerische Simulation kompress-
ibler, reibungsbehafteter Strömungen (Self-Adaptive Local Grid Refinement for the Numerical
Simulation of Compressible Viscous Flows). Doctoral Thesis, University Stuttgart, Germany
(1993)
9. Stanewsky, E., Puffert, W., Müller, R., Batemann, T.E.B.: Supercritical Airfoil CAST 7—
Surface Pressure, Wake and Boundary Layer Measurements. AGARD AR-138, A3-1–A3-35
(1979)
10. Zierep, J.: Der senkrechte Verdichtungsstoß am gekrömmten Profil. ZAMP, vol. IXb, pp. 764–
776 (1958)
11. Lighthill, J.: An Informal Introduction to Theoretical Fluid Mechanics. Clarendon Press, Oxford
(1986)
12. Mangler, K.W., Smith, J.H.B.: Behaviour of the vortex sheet at the trailing edge of a lifting
wing. Aeronaut. J. R. Aeronaut. Soc. 74, 906–908 (1970)
13. Anderson Jr., J.D.: Fundamentals of Aerodynamics, 5th edn. McGraw Hill, New York (2011)
14. Schwamborn, D.: Laminare Grenzschichten in der Nähe der Anlegelinie an Flügeln und
flügelänlichen Körpern mit Anstellung (Laminar Boundary Layers in the Vicinity of the Attach-
ment Line at Wings and Wing-Like Bodies at Angle of Attack). Doctoral thesis, RWTH Aachen,
Germany, 1981, also DFVLR-FB 81-31 (1981)
15. Zahm, A.F.: Flow and force equations for a body revolving in a fluid. NACA Rep. No. 323
(1930)
16. Maruhn, K.: Druckverteilung an elliptischen Rümpfen und in ihrem Außenraum. Deutsche
Luftfahrtforchung, Jahrbuch 1941, SI, pp. 135–147 (1941)
17. Schlichting, H., Truckenbrodt, E.: Aerodynamik des Flugzeuges, vol. 1 and 2, Springer,
Berlin/Göttingen/Heidelberg, 1959, also: Aerodynamics of the Aeroplane, 2nd edn (revised).
McGraw Hill Higher Education, New York (1979)
18. Prandtl, L.: Tragflügeltheorie, I. und II. Mitteilung. Nachrichten der Kgl. Ges. Wiss. Göttingen,
Math.-Phys. Klasse, 451–477 (1918) und 107–137 (1919)
19. Spreiter, J.R., Sacks, A.H.: The rolling up of the trailing vortex sheet and its effect on the
downwash behind wings. J. Aeronaut. Sci. 18, 21–32 (1951)
Chapter 5
The Matter of Discrete Euler Solutions
for Lifting Wings

With the development in particular of fighter aircraft with delta-type wings the phe-
nomenon of lee-side vortices became a topic in the field of aerodynamics. (Configu-
rational, fluid-mechanical and operational topics related to aircraft with such wings
are treated in Chaps. 10 and 11.)
For the description of the lee-side vortices analytical and semi-empirical meth-
ods, as well as panel methods (Model 4 in Table 1.3) were developed, which made
necessary the geometrical modeling of the vortex sheets (feeding layers) emerging
from the (sharp) leading edges of the delta wing [1]. These methods at the beginning
of the 1980s gave way to discrete numerical solutions of the Euler equations (Model
8 of Table 1.3). Now since long of course methods based on Model 10 and higher
govern the field. Nevertheless it is still appropriate to look at the Euler methods in
view of the general problem of lifting-wing flow simulations.
Here we study the topics, which emerged when the first Euler solutions came into
use, viz. the creation of vorticity and the entropy rise in the flow field. These topics
are treated in the following three sections using the concept of kinematically active
vorticity. Examples are given then in Sect. 8.3 for a large aspect-ratio wing and in
Sect. 10.4 for a delta wing.

5.1 Vorticity Creation in Euler Solutions of Lifting-Wing

Flow

The discrete modeled Euler solution, Model 8, for lifting-wing flow must include
the wake (trailing vortex layer) of reality (Model 1) in one or the other form. We
have discussed in Sect. 4.2.6—with the help of Model 2—how this wake for high
Reynolds numbers shrinks into the discontinuity layer—or trailing vortex layer—of
potential wing theory (Model 4).

© Springer-Verlag GmbH Germany, part of Springer Nature 2021 119

E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_5
120 5 The Matter of Discrete Euler Solutions for Lifting Wings

Fig. 5.1 View of the three-dimensional wake of a lifting finite-span wing in steady sub-critical flow
[4]. Left: the wake in the limit Rer e f → ∞ as discontinuity layer of potential theory, see Fig. 4.13.
Right: the discontinuity layer widened up by numerical diffusive transport of vorticity: the Euler
wake

Hidden in this discontinuity layer is the kinematically active vorticity of reality.

Analog to potential wing theory this layer should appear also in an ideal Euler solution
(Model 7). That would be a weak solution of the Euler equations (Model 8) as was
shown in [2].
In a discrete modeled conservatively formulated Euler solution the discontinuity
layer is smeared out over some grid points. That is in analogy to the captured shock
wave in supersonic flow [3]. For subcritical flow we show this situation schematically
in Fig. 5.1.1 At the left side we see the discontinuity layer as discussed in Sect. 4.2.6.
Hidden in this layer is the kinematically active vorticity content Ωs . At the right side
we see the by the diffusive transport widened discontinuity layer resulting from the
discrete Euler solution.
The diffusive transport in the strict sense is a “false” diffusive transport. Never-
theless, the “Euler wake” in principle has the right properties. The u(z)-profile now
is uniform, because in an inviscid flow model no skin-friction drag and no viscosity
induced pressure or form (profile) drag is present, hence no kinematically inactive
vorticity.
Actually the u(z)-profile in Fig. 5.1 should be shown as being non-uniform,
because of the properties of the Euler solution upstream of the wake on the wing’s
surface.2 There we have on both the upper and the lower side what we loosely call

1 The following discussion also holds for supercritical flow. There the whole vorticity wake including

the entropy-layer wake must be considered, Sect. 4.2.3.

2 Seein this regard also the discussion in [5].
5.1 Vorticity Creation in Euler Solutions of Lifting-Wing Flow 121

an Euler boundary layer. Usually that is seen as a total-pressure loss of the flow
along the body surface, and also of a total-temperature loss [4]. In regions of flow
deceleration this can be rather pronounced.3 However, as long as at the trailing edge
the external inviscid velocities u eu and u el are the same, the vorticity content of the
u(z)-profile is kinematically inactive, which means that the u(z)-profile can have
any form.
The v(z)-profile is the vortex-like profile shown in Fig. 4.12, right. The precise
form of this profile also does not matter. What matters is that the kinematically
active vorticity content, which was hidden in the discontinuity layer, has reappeared,
Fig. 5.1, right. This vorticity content “carries” the induced drag, and its appearance
is compatible with a discrete inviscid model of finite-span lifting-wing flow (Model
8).
Provided a Kutta condition is present, the Euler wake appearing in discrete numer-
ical solutions of the Euler equations for lifting-wing flow—past either large or small
(delta) aspect-ratio wings—thus is a necessary and sufficient property of such solu-
tions.
The examples presented in Sect. 8.3 (large aspect-ratio wing) and in Sect. 10.4
(delta wing with sharp leading edges) give proof of the applicability of Model 8
methods, see also Sect. 3.12.1. The appearing vorticity content in principle has the
right amount. To it belongs an entropy rise, or equivalently, a total pressure loss. This
is treated in the following section.

5.2 Vorticity and the Related Entropy Rise

The vorticity present in the Euler wake is accompanied by an entropy rise, respec-
tively a total pressure loss. We investigate that now with the help of Crocco’s theorem,
Sect. 3.5.
This law for steady inviscid iso-enthalpic flow reads

v × r ot v = v × ω = −T grad s, (5.1)

where v is the velocity vector, ω the vorticity vector, T the temperature and s the
entropy.
We introduce ω in the stretched form, Eq. (4.4), and for Rer e f → ∞ (note that w
→ 0 in this case) and find4
 
1 ∂ u 2 + v2 ∂s
= −T . (5.2)
2 ∂z ∂z

3 The total-pressure loss depends on the chosen variables, the discretization scheme, and the grid-
fineness. Depending on the simulation problem at hand, one has to be aware of that.
4 z is the stretched coordinate, however, we have omitted the tilde.
122 5 The Matter of Discrete Euler Solutions for Lifting Wings

Assuming a perfect gas, we relate the temperature T to the total temperature Tt

and the velocities u and v:

u 2 + v2
T = Tt − . (5.3)
2c p

We return to the element of a generalized shear layer, Fig. 4.1, integrate Eq. (5.2),
beginning at the lower edge z = δl and obtain

−2 c p Tt + u 2 (z) + v 2 (z)
s(z) − s(z = δl ) = c p ln . (5.4)
−2 c p Tt + u 2el + ve2l

At the lower edge the external inviscid velocity can be written as

u 2el + ve2l = |V el |2 . (5.5)

The Mach number Mel of the external inviscid flow then reads:
|V el |
Mel = . (5.6)
(γ R Tel )0.5

We introduce these two expressions into Eq. (5.4) and finally arrive at

s(z) − s(z = δl ) Δs(z)

= =
cp cp
  (5.7)
γ−1 2 u 2 (z) + v 2 (z)
= ln Mel (1 − )+1 .
2 |V el |2

The equivalent total pressure loss Δpt (z) is, see, e.g. [3]
γ Δs(z)
− γ−1
Δpt (z) = ptel − pt (z) = ptel (1 − e cp
). (5.8)

Thus the entropy rise and the equivalent total pressure loss are functions of the
squares of the velocities u and v in the wake and the edge velocity |V el |. For the
thermodynamically singular case of zero Mach number—incompressible flow—no
entropy rise happens. A total pressure loss, however, occurs, but it cannot be pre-
scribed in the frame of our consideration.
We look now at four wake cases, the third one being the Euler wake:

1. Two-dimensional wake, Fig. 4.4a: Eq. (5.7) simply reduces to

 
Δs(z) γ−1 2 u 2 (z)
= ln Mel (1 − 2 ) + 1 , (5.9)
cp 2 u el
 5.2 Vorticity and the Related Entropy Rise 123

Although the wake is kinematically inactive, we observe, as expected, an entropy

rise throughout the wake and hence a total pressure loss. Regarding the wake in
Fig. 4.4b, which is present in steady supercritical motion, we have to remember
that in that case the entropy wake includes the wake of the shock wave, Fig. 4.6.
2. Three-dimensional lifting-wing wake, Fig. 4.12 right: In the decomposed view
both the kinematically active wake v(z) and the kinematically inactive wake u(z)
are contributing to the entropy rise and hence to the total pressure loss. Equa-
tion (5.7) explicitly contains these wake parts.
3. Euler wake, Fig. 5.1 right: In this case we have assumed u(z) = u el = const. and
Eq. (5.7) becomes
 
Δs(z) γ−1 2 u 2el + v 2 (z)
| Rer e f →∞ = ln Mel (1 − )+1 . (5.10)
cp 2 |V el |2

We see that with u(z) = u el = const. the entropy rise is reduced compared to that
of the real wake, but because v(z) < vel , it does not vanish. We note that here the
remark in [5], see also above, Sect. 5.1, regarding the uniformity or non-uniformity
of u(z) has a point. Although the u-component vorticity content still remains to
be zero, the entropy production is affected.
4. Euler wake of the infinite swept wing: On Sect. 4.4 we have noted that no kine-
matically active vorticity leaves the trailing edge of the infinite swept wing. The
trailing-edge shear angle ψe is zero. This means, Fig. 5.1, that vel and veu as well as
v(z) are zero. Because we assume for the Euler wake u(z) = u el = const., Eq. (5.7)
becomes
 
Δs(z) γ−1 2 u 2el
| Rer e f →∞ = ln Mel (1 − 2 ) + 1 = 0. (5.11)
cp 2 u el

The result is that in this case no entropy rise happens, except for the upper and
lower Euler boundary layer.

5.3 Critical Evaluation

Euler codes, i.e., Model 8 methods, Table 1.3, now are for a long time in use, originally
even for data-set generation, currently at most only in the earlier phases of flight-
vehicle shape definition. Nevertheless, a look at their capabilities and shortcomings
is appropriate. We basically follow the discussion, which is given in [4].
In Sect. 5.1 of this chapter we have argued that, in analogy to potential theory
(Model 4), an Euler solution for a finite-span lifting wing exhibits a discontinuity
surface downstream of the trailing edge, a computational realization of the trailing
vortex layer. At a wing with highly swept leading edges—the main geometrical
property of delta wings—above a critical combination of the “normal angle of attack
124 5 The Matter of Discrete Euler Solutions for Lifting Wings

α N ” and the “normal leading-edge Mach number M N ” also a discontinuity surface is

leaving each leading edge, Sect. 10.1. These discontinuity surfaces are the “feeding
layers” of the lee-side vortices.
We note in passing that over both wing types also secondary and even tertiary
vortex phenomena can be present, Model 1. In addition we note that also at the trailing
edge of the delta wing a trailing vortex layer is present, similar to the situation of
large aspect-ratio wing, Fig. 1.10b.
The said discontinuity surface is the vorticity wake, Model 1, in reality, but now
in the limit Rer e f → ∞. True discontinuity surfaces—vortex sheets, as in potential
theory—treated in Euler methods would be exact, but would hamper very much the
solution methods for general wings and aircraft. The reason is that they would have
to be “fitted”, i.e., presented by coordinate surfaces, like shock fitting. Therefore the
appearance of Euler wakes of finite thickness, which are “captured” by the solution—
similar to shock capturing—is acceptable.
Indeed, to capture Euler wakes is the only practicable approach, and that in view of
the fact that the kinematically active vorticity content appearing in them in principle
is described exactly.
A study of numerical Euler solutions, however, shows at least two problem areas
regarding the phenomenon “Euler wake”:
1. In the ideal case the thickness of the Euler wake (Model 8) should be that of the
real wake (Model 1). It is evident that, if the Euler wake is smeared out over,
say, four cells, it depends on the size of the cell, i.e., the local fineness of the
discretization of the computation domain, how thick the wake is.
In the case of an isolated wing, the thickness of the Euler wake does not matter,
as long as the wake transports the right vorticity content.
If, however, vortex sheets or vortices interact with each other or with parts of the
configuration—typical, for instance, for the lee-side vortices of delta wings—the
solution will be erroneous, if the thickness of the domain of the vortex sheets or
the vortices is inadequately represented.
2. Due to the kind of discretization of the computation domain behind a wing, a part
of the kinematically active vorticity may be lost and hence also a certain amount
of the circulation in the wake. This in general holds for all discrete numerical
solutions of such flow cases, see, e.g., the CRM wing wake case in Sect. 8.4.
However, in both panel (Model 4) and in RANS (Model 10) and higher model
solutions usually a rather coarse discretization is applied in the flow domain
behind the wing. Despite this, at the wing forces and moments as well as flow
properties can be computed to a sufficient degree of accuracy. This also holds
for Euler methods, of course only within the inherent limitations of the Model 8
approach.
This concerns an isolated large aspect-ratio wing. But if the trailing vortex layers
or vortices interact with each other or with the empennage, the solution will have
deficits, if the discretization is inadequate. This problem of course is present also
with low aspect-ratio wings, in particular with delta wings. If lee-side vortex
5.3 Critical Evaluation 125

phenomena are present, a suitable lee-side discretization or an automatic mesh

adaptation is of utmost importance.
The first problem area has to do with the diffusive properties of Model 8 Euler
solutions, as to a degree also has the second one. In addition it must be ensured that the
far-field boundary conditions, which in most of the methods in use are characteristic
boundary conditions, do not influence the conservation of circulation in the far wake,
see, e.g., [4]. Even for lifting airfoils in some codes the circulation is found to vanish
near the outer boundary of the computation domain. All this is a topic also for Model
10 (RANS/URANS) and for Model 11 (Scale-resolving) simulations.

5.4 Problems

Problem 5.1 Why can u(z) in Fig. 5.1 have any shape as long as u(δu ) = u(δl ).

Problem 5.2 Why are discrete numerical solutions of the Euler equations (Model 8)
of interest in design work? What are the differences in terms of capabilities compared
to those of Model 4 and Model 10/11 methods?

Problem 5.3 What are the differences in terms of costs? Look at the whole processes
of the different methods.

Problem 5.4 If u(z) is not constant, what wake case regarding the entropy rise
holds? Why does the function u(z) matter?

Problem 5.5 Why is a proper grid resolution necessary in discrete numerical solu-
tions? Which flow models in Table 4.1 are concerned.

References

1. Rom, J.: High Angle of Attack Aerodynamics. Springer, Heidelberg (1992)

2. Powell, K.G., Murman, E.M., Perez, E.S., Baron, J.R.: Total pressure loss in vortical solutions
of the conical Euler equations. AIAA J. 25, 360–368 (1987)
3. Hirschel, E.H.: Basics of Aerothermodynamics, 2nd edn (revised). Springer, Cham (2015)
4. Eberle, A., Rizzi, A., Hirschel, E.H.: Numerical Solutions of the Euler Equations for Steady Flow
Problems. Notes on Numerical Fluid Mechanics, vol. 34. Vieweg, Braunschweig Wiesbaden
(1992)
5. Hoeijmakers, H.W.M.: Modelling and numerical simulation of vortex flow in aerodynamics.
AGARD-CP-494, 1-1–1-16 (1991)
Chapter 6
About the Kutta Condition

The Kutta condition in all physical and mathematical flow models defined in Sect. 1.5
is generally understood as appearing at acute edges of aerodynamic surfaces. That
can be the “zero thickness” trailing edge of a large aspect-ratio wing or the “sharp”
leading edge of a delta wing. Usually it is assumed that the flow leaves the trailing
edge in bisector direction.
The reality, however, demands a closer look, because the Kutta direction—we
call it this way—is not necessarily the same as the bisector direction. The closer
look is necessary for an understanding of the related flow phenomena in general.
And then it holds for the application of the physical and mathematical models of
Sect. 1.5 to the real shapes of trailing (and leading) edges and hence also for the grid
generation for discrete numerical methods. Accordingly we treat two different topics
in the following sections: decambering phenomena and the Kutta condition and the
matter of the Kutta condition and direction in reality.

6.1 Decambering Phenomena

In the subcritical case (subsonic flight regime) inviscid flow-off separation from an
acute edge happens in the bi-sector direction. However, at the trailing edge of a lift-
ing wing in reality, the actual flow-off direction—the Kutta direction—is slightly
deflected upward. This leads to a reduction of the lift. This effect is called decamber-
ing. There are two basic mechanisms: boundary-layer decambering and shock-wave
decambering, the latter appearing in the supercritical case (transonic and supersonic
flight regime). We discuss and illustrate these kinds of decambering at airfoils, i.e., in
two dimensions. The results hold in three dimensions, too, and on wings and aerody-
namic trim, stabilization and control surfaces of all kinds, as well as lift enhancement
surfaces.

© Springer-Verlag GmbH Germany, part of Springer Nature 2021 127

E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_6
128 6 About the Kutta Condition

6.1.1 Boundary-Layer Decambering

Boundary-layer decambering is due to the influence of the boundary-layer displace-

ment properties [1].1 The boundary-layer decambering effect at an airfoil in subcrit-
ical motion manifests itself in a slight reduction of the lift slope dC L /dα, which
(inviscid) thin airfoil theory obtains as dC L /dα = 2π [3]. The effect decreases with
increasing Reynolds number.
To understand the effect, we consider the situation at a symmetric airfoil. We
look first at the situation, which we find at zero angle of attack. There the boundary-
layer displacement thicknesses—small, but non-negligible—at the upper and the
lower side of the airfoil are the same at the trailing edge of the airfoil.2 The flow-off
direction is that of the bisector direction of the trailing edge, which is the same as the
freestream direction, decambering is not present. (However, with a non-symmetric
airfoil, boundary-layer decambering already is an issue at the nominal non-zero lift
situation.)
At angle of attack, we observe that the displacement thickness at the trailing edge
is larger on the upper side, the suction side, than on the lower side, the pressure side.
Hence the flow-off direction shifts somewhat toward the upper side—decambering
happens. The reasons for the different displacement thicknesses are the following:
(a) With positive angle of attack, the forward stagnation point moves away from
the nose point to the lower side of the airfoil. This means an increase of the boundary-
layer running length over the upper side and hence a thicker boundary layer—and
larger displacement thickness—at the upper side of the trailing edge.
(b) The boundary layer over the suction side—because of the larger external
inviscid velocity there—is higher loaded with an adverse pressure gradient than
that over the pressure side. This means an additional increase of the displacement
thickness of the upper-side boundary layer at the trailing edge compared to that of
the lower-side boundary layer.
(c) Depending on the actual surface pressure and hence the external inviscid
velocity distributions, the locations of the laminar-turbulent transition on the upper
and the lower side are different. If, and that generally is the case, the transition
location on the upper side is more forward than that on the lower side, this results
in a longer running length of the turbulent boundary layer over the upper side, and
hence a further increment of the displacement thickness at the upper side of the
trailing edge.

1 The concept of boundary-layer decambering seems to have been developed by M. J. Lighthill [2].
2 Flow-off separation, like ordinary separation, locally is characterized by strong interaction of the
external inviscid and the boundary-layer flow, Sect. 2.2. Nevertheless we can speak here simply of
the boundary layers at the trailing edge.
6.1 Decambering Phenomena 129

Regarding boundary-layer decambering we summarize:

1. Boundary-layer decambering occurs in subcritical and supercritical cases, in the

latter case together with shock-wave decambering (see next sub-section). It is
due to the displacement effect of the boundary layers in the vicinity of the trailing
edge.
2. The magnitude of the displacement effect depends on the shape of the airfoil
and the angle of attack, with that on the surface pressure distribution and then
on the location of laminar-turbulent transition. The primary flow parameter is
the Reynolds number. The higher the Reynolds number, the smaller is the dis-
placement thickness and hence the boundary-layer decambering.
3. This all holds for both airfoils and wings as well as for stabilization and trim
surfaces, see above.

6.1.2 Shock-Wave Decambering

Shock-wave decambering is an effect, which appears—in addition to boundary-

layer decambering—in the supercritical case, i.e., in the transonic flight regime. To
understand shock-wave decambering, we have to look at the flow situation at the
trailing edge of a lifting airfoil. Consider as example the case illustrated in Fig. 4.6
on Sect. 4.2.3. There on the suction side of the airfoil a supersonic flow pocket with
a terminating shock wave is present. We treat the inviscid case, however in a very
simplified way.
For all steady motion cases, which we consider here, at the upper and at the lower
side of the trailing edge the (static) surface pressures p are equal, if we disregard
possible near-wake curvature effects. In the subcritical case also the total pressures
pt are the same on both sides of the trailing edge, also the velocities V u and V l , and
hence also the absolute values of the momentum-flux vectors |Q|.
To define the latter we consider the flow in a small domain n at and parallel
to the surface. If a x is the unit vector of a surface-oriented curvilinear grid at the
surface in downstream direction, [1], we approximately obtain the momentum-flux
vector as Q = ρ|V 2 |n a x .
In the subcritical case the absolute values of the momentum-flux vectors are the
same at the upper (u) and the lower (l) side of the trailing edge: |Q u | = |Q l |. This
situation is present in all subcritical cases, but also in the supercritical case of a
symmetric airfoil at zero angle of attack (non-lifting case).
In the supercritical case shock waves are present, (a) either only one at the suction
side of the airfoil as shown in Fig. 4.6, or, (b) at both the upper and the lower side,
that on the suction side having a larger strength.
130 6 About the Kutta Condition

Fig. 6.1 Inviscid flow at the trailing edge of a lifting supercritical airfoil with an embedded shock
wave at the suction side: deflection of the flow (Kutta direction) out of the bi-sector (shock-wave
decambering) [4]

In case (a) the ensuing total pressure loss leads to a reduction of the suction side
velocity and hence of the momentum flux |Q u | at the trailing edge (this reasoning
also holds for case (b)): |Q u | < |Q l |. (The magnitude of the total pressure loss at
the surface can be found [4], if the pre-shock Mach number has been determined,
because the terminating shock wave impinges vertically on the surface, inviscid flow
[5].)
The consequence in both cases is the upward deflection of the flow out of the
bi-sector of the trailing edge, the shock-wave decambering, as illustrated in Fig. 6.1.
Shock-wave decambering came into view in the late 1970s, early 1980s, when
discrete numerical methods based on the compressible full potential equation, Model
5 of Table 1.3, were developed. These methods, employing the conservative formula-
tion of the potential equation, were able to describe embedded shock waves, although
with problems regarding their strength and location. Hence, the predictions of the
aerodynamic properties of airfoils were erroneous.
The problem with the full potential equation methods is that, although the shock
waves are captured, the in reality occurring total pressure loss does not arise. (Poten-
tial methods are not able to describe a total pressure loss.) Hence the flow leaves
at the trailing edge in bi-sector direction, i.e., the shock-wave decambering is not
described. Although schemes were developed to overcome this problem, see, e.g.,
[6–8], full potential equation methods finally gave way to discrete numerical Euler
methods (Model 8) and Navier–Stokes/RANS methods (Model 9/10), which both
inherently describe shock-wave decambering.
Regarding shock-wave decambering we summarize:

1. Shock-wave decambering occurs in supercritical cases. The effect is the larger,

the larger the pre-shock Mach number is. Transonic airfoils may have supersonic
flow pockets and hence terminating shock waves on both sides. Then the effect
becomes weakened and in extreme cases may even be canceled.
6.1 Decambering Phenomena 131

2. The terminating shock wave has another effect. Below a pre-shock Mach number
at the shock foot of M f oot ≈ 1.3−1.35 the boundary layer does not separate, a
weak shock-wave/boundary-layer interaction is present, Sect. 9.1. In the inviscid
picture the shock terminates at right angle to the surface, see above. For all super-
sonic Mach numbers the unit Reynolds number decreases over a normal shock
wave [9]. The boundary layer behind the shock hence gets a thickness increment
compared to that in front of the shock.3 This causes an extra thickening of the
downstream boundary layer.
At M f oot  1.3–1.35 a strong shock/boundary-layer interaction with a sepa-
ration bubble happens. The in both cases also enlarged displacement thickness
enlarges the boundary-layer decambering. (At even higher Mach numbers we get
shock stall, i.e., a complete separation behind the shock foot and in consequence
transonic buffet, Sect. 9.1. Then our considerations become invalid.)

For an example of the combined effect of boundary-layer and shock-wave decam-

bering see below Fig. 6.8 in Sect. 6.5.

6.2 Kutta Condition and Kutta Direction in Reality

The Kutta condition and the Kutta direction in the perceived reality (Model 1 of
Table 1.3) are different from that considered in Model 4 and 7. There the inviscid
flow leaves smoothly in the bi-sector direction of the trailing edge, Fig. 4.15.
We first consider the two-dimensional case. Regarding the external inviscid flows
over the upper and the lower side of the trailing edge, we have in viscous modeled
flow, Model 2—and also in Navier–Stokes flow and derivatives flow (Model 9/10)—
in a sense two Kutta directions. They are, moreover, not in bisector direction due to
the decambering effects. Regarding the actual Kutta direction, it helps to define it as
the direction of the near-wake center line, Fig. 4.6 in Sect. 4.2.3.
In the three-dimensional case we also have two external inviscid Kutta directions
due to the decambering effects. Moreover, in the lifting case of the finite-span wing
they are sheared by the trailing-edge flow shear angle ψe , which is a function of the
span-wise location. Between them the viscous flow has a complex pattern, Fig. 4.12
in Sect. 4.2.6. Also here we define as Kutta direction the direction of the near-wake
center line, which in that figure is the s-coordinate. This line is a vortex line, which is
deflected from the x-coordinate (chord direction) by the positive or negative vortex-
line angle .
In the following section we will see that in practice on operational aircraft the
situation is even more complex. Trailing edges in reality have a finite thickness.

3 All
boundary-layer thicknesses depend on the inverse of some power of the Reynolds number,
Appendix A.5.4.
132 6 About the Kutta Condition

6.3 Geometric Properties of Trailing and Leading Edges of

Actual Wings

Trailing edges usually are considered as having acute angles or being sharp edges. In
reality they are not sharp in the sense of the word.4 Manufacturing demands lead to
a certain bluntness, but also practical demands do so. Leading edges of subsonic and
transonic transport aircraft are well rounded. This also holds if aircraft have swept
wings in order to shift the transonic drag divergence to Mach numbers as high as
possible. This shift can also be achieved with the unswept “thin wing”. Then the
leading edge has to be sharp. Highly swept leading edges are found on supersonic
aircraft. The leading edges of their wings or strakes usually have a rather small
bluntness. However, often they are considered as being sharp.
We look at several aspects of geometric properties of trailing edges and leading
edges: (a) what wing trailing-edge properties are found on existing aircraft, (b) how
are the aerodynamic properties of an airfoil or wing affected by a given trailing-edge
thickness, (c) when can an edge be considered as to be “aerodynamically” sharp in
the sense of Küchemann (Sect. 6.4).
The literature is vast. However, we do not intend to give a systematic survey. For
airfoil and wing layout see, e.g., [11, 12]. We note that NACA in 1945 published
airfoil shapes with finite trailing-edge thickness [13]. They have, for instance, ratios
‘trailing-edge thickness’ to ‘local chord length’ h/c = 0.126 per cent (NACA 0006
Basic Thickness Form), or h/c = 0.252 per cent (NACA 0012 Basic Thickness
Form).
A matter is also the question of finite-angle versus cusp trailing edges of super-
critical airfoils, the latter also for practical reasons with finite thickness. The reader
is referred in this regard to [11].
In the following considerations we assume the typical slender airfoil or chord
section of a wing. Of course the airfoil shape plays a role, in particular also the
trailing-edge angle.5 Detailed airfoil and wing design parameters cannot be treated
in the frame of this book.
(a) Trailing and leading edge properties on existing aircraft.
Wing trailing-edge thicknesses found on subsonic transport aircraft range from
h ≈ 0.5 to 1 cm.6 The ratio h/c lies between 0.1 and 0.5 per cent.7 On transonic
transport aircraft (moderate leading-edge sweep) one finds h ≈ 1–1.5 cm, with local

4 Although D. Küchemann with the term “aerodynamically sharp edge”, [10], to some degree cir-
cumvented the topic of geometrical properties of trailing edges, we believe that a closer look is
justified.
5 A large trailing-edge angle θ may even lead to a negative dC /dα at small angles of attack [14].
L
6 The authors are grateful to the Deutsches Museum in München, where it was possible to investigate

the wings of a large number of aircraft. Thanks go also to colleagues who made data available.
7 The actual shapes of blunt trailing edges vary from sharply cut off to well rounded. Although the

detailed shapes locally may play a role, we do not take them into account.
6.3 Geometric Properties of Trailing and Leading Edges of Actual Wings 133

ratios h/c ≈ 0.2–0.5 per cent. It is intended in the future to come down to h ≈ 0.5 cm
in order to reduce the drag penalties, which are associated with finite trailing-edge
thicknesses.
We also note the data of the Common Research Model (CRM), see Sect. 8.4, which
is a test configuration of the international AIAA CFD Drag Prediction Workshop.
At the location of the Yehudi break of the trailing edge, the thickness is h = 1.31 cm
and the ratio h/c ≈ 0.15 per cent. On the CRM wind-tunnel model near the wing tip
the ratio is h/c ≈ 0.48 per cent. This thickness is needed in order to place a gauge
there.
The situation is similar on modern fighter aircraft with highly swept leading edges.
Trailing-edge thicknesses vary between h ≈ 0.5 and 2 cm. The thinnest trailing edge
was found at a MiG 21 fighter with h ≈ 0.15 cm. However, the delta wing shapes of
all aircraft lead to small local values h/c. At the root region of the wings h/c ≈ 0.1
per cent is found, at the tip region h/c ≈ 0.3–0.5 per cent.
With wings possessing highly swept leading edges also the leading-edge radius
r L E is of interest. That usually varies from the root region to the tip region. In the
root region we find r L E ≈ 2–3 cm, and in the tip region r L E ≈ 0.7–1.5 cm. The ratio
r L E /c accordingly ranges from 0.5 to 0.1 per cent.
The drag divergence problem, which arises in the transonic flight regime, has
led to the supercritical airfoil design in combination with leading-edge sweep. An
alternative is to use the already mentioned “thin wing”. The leading edge of the
trapezoidal wing of the F-104 Starfighter—the only example of a thin wing for which
it was possible to obtain data—has a radius r L E < 0.1 cm. The ratio ‘maximum wing
thickness’ to ‘chord length’ of the wing is d/c = 3.36 per cent. The trailing-edge
thickness is h ≈ 0.25 cm. With chord lengths of c ≈ 3.2 m (root) and c ≈ 1.45 m
(tip) we have trailing-edge values h/c ≈ 0.1–0.2 per cent.
A special situation arises with hypersonic flight vehicles. There thermal loads and
issues of structural integrity may demand quite blunt leading and trailing edges. A
notable example is the trailing edge of the wedge-like vertical stabilizer of the X-15,
which, however, was needed in order to obtain at hypersonic speeds the necessary
directional stability.8
(b) Trailing-edge thickness and aerodynamic performance of airfoils.
An early systematic investigation of the influence of a blunt trailing edge of an
airfoil was reported in 1927 [15]. At the Aerodynamische Versuchsanstalt (AVA)
Göttingen wind tunnel experiments were made with the asymmetrical airfoil No.
508 and with the symmetrical airfoil No. 460. Both airfoils had the length c =
20 cm. The airfoils were systematically truncated in 2 cm steps, Fig. 6.2. Lift-drag
coefficient curves and lift-moment coefficient curves were measured.

8 In [3] a detailed discussion of the fluid mechanical aspects of this case can be found.
134 6 About the Kutta Condition

Fig. 6.2 Truncation of the

airfoils No. 508 and 460 [15]

We discuss the results for the first 2 cm truncation. The airfoil No. 508 then has
the ratio ‘trailing-edge thickness’ to ‘local chord length’ h/c ≈ 2.75 per cent, and
the No. 460 airfoil h/c ≈ 5.5 per cent. We note that both values are much larger than
those given above.

The experimental data show an increase of the drag at zero lift, C D0 , of about
20 per cent for the No. 508 airfoil, and about 63 per cent of the No. 460 airfoil.9

The maximum lift, C L max , is reduced by about 9 percent for the No. 508 airfoil, and
about 3.5 per cent for the No. 460 airfoil. The center of pressure is moved forward
as to be expected. Regarding the drag increase, generally the symmetrical airfoil is
more affected by the truncation than the asymmetrical one. This is due to the larger
trailing-edge angle θ of the airfoil 460.
The measurements were made in a low-speed wind tunnel. Both airfoil models had
a span of 1 m. In the report the Reynolds number and the state of the boundary layer
are not given, but laminar-turbulent transition can be expected to have occurred at or
shortly downstream of the maximum thickness locations of the airfoils. Coefficients
also are given related to the actual shortened airfoil lengths. Some of the conclusions
in [15] are not acceptable, because then not the coefficients must be considered, but
the actual forces.
Nevertheless, the data show the expected changes of the aerodynamic properties,
if a trailing edge has a finite thickness. Even if manufacturing and practical demands
lead to a certain thickness, it should be as small as possible in order to avoid, in
particular, drag penalties.
The topic of finite trailing-edge thickness of course must be seen in the general
frame of airfoil and wing design, see, e.g., [10–12, 16].
The influence of the trailing-edge properties on the aerodynamic performance of
an airfoil or wing is one topic. The other topic is the flow at and behind a blunt
trailing edge. In general it can be assumed that the two boundary layers, which flow
off the surface at the respective edge, have thicknesses much larger than the edge

9 The prime indicates that the coefficient is related to the original airfoil length of 20 cm.
6.3 Geometric Properties of Trailing and Leading Edges of Actual Wings 135

thickness. Nevertheless, an embedded wake will be present, which, due to the vortex
shedding, has an unsteady character. For two-dimensional cases several authors report
the presence of a von Kármán vortex street, for instance [17]. Respective data are
not known for three-dimensional lifting-wing cases, viz., cases with a shear of the
upper and the lower external inviscid flow, the wing flow-fields shear, see Fig. 4.17
on Sect. 4.3.2.
Related to the vortex shedding is noise generation due to a blunt trailing edge.
This in general appears not to be a very critical issue. The reason is that the related
frequencies are quite high. This can be seen from a consideration with the help of the
Strouhal number. Vortex shedding behind a cylinder (two-dimensional case) has—
Reynolds number dependent—Strouhal numbers Sr ≈ 0.2 (Re ≈ 104 ) to Sr ≈ 0.3
(Re ≈ 107 ). These numbers are generally observed for the vortex shedding behind
blunt bodies, see, e.g., [18]. (The Reynolds number is defined as Re = ρuh/μ.)
We write
fh
Sr = , (6.1)
u
f being the shedding frequency, h the diameter of the cylinder, in our case the
thickness of the trailing edge, u the free-stream velocity, in our case the external
inviscid velocity at the trailing edge (upper and lower side).
The relation for the frequency then reads
Sr u
f = . (6.2)
h
With typical aircraft flight data we get frequencies in a narrow band at the upper
bound of the human hearing domain. This “beep” is clouded by the noise coming from
the airframe and its components, and the propulsion system. The blunt trailing-edge
noise concerns less the cruise configuration of an aircraft, but mainly the take-off
and the landing configuration. The low-noise aircraft is a big topic and the blunt
trailing-edge noise is part of that. For a more recent study on this noise and possible
reduction concepts see [19].
The situation is similar with wind turbine rotors, but also helicopter rotors. There
potentially so called blunt “flatback” airfoils can be employed at the innermost part
of the rotor blades. The benefit would be lighter and stronger blade structures. The
flatback airfoil is found by opening the trailing edge, not by simply truncating the
airfoil as is the rule with trailing edges of aircraft wings. The structural benefit,
however, is bought with a possible extra noise generated by the vortex shedding at
the blunt trailing edge, see, e.g., [20, 21]. Also here usually only the two-dimensional
case is studied. This is permitted because the flatback airfoil is only employed at the
inboard section of the rotor.
136 6 About the Kutta Condition

6.4 When Can an Edge Be Considered as to Be

Aerodynamically Sharp?

This question regards modeling issues in both theoretical/numerical and experimental

work. One can imagine that the ratio ‘trailing-edge thickness’ to ‘upper and lower
boundary-layer displacement thicknesses’ h/δ1 plays a role. Detailed investigations
in this regard and—more so—criteria, are not known to the authors.
The approaches of numerical aerodynamics today, Model 10 of Table 1.3, permit
a proper grid resolution at blunt trailing or leading edges. The turbulence modeling
problem, of course, remains. For pre-design activities Euler methods, Model 8, are
in use. With them, a high resolution of, for instance, trailing edges, is not the rule.
The situation is different at highly swept leading edges, where it is to ask whether
for a given case a somewhat rounded edge can be considered and modeled as being
sharp, see below.
Experimental work in high Reynolds-number ground facilities should aim for a
true representation of the geometrical properties of trailing and leading edges of the
aircraft model. This is necessary, because then the ratio h/δ1 can be met. However,
the resolution of geometrical details is a general problem, because usually in such
tunnels the aircraft models are not very large. In low Reynolds-number facilities the
ratio h/δ1 will not be met and the obtained drag data, for instance, will be flawed.
Regarding the question when the swept round leading edge of a delta wing can
be regarded as aerodynamically sharp, VFE-2 work has shown that the cross-flow
bluntness parameter pb at least permits qualitative considerations, Sect. 10.2.4:
rle (x)
pb (x) = , (6.3)
b (x)

where locally rle is the leading edge radius and b (x) the span width, both being
defined in the wing cross-section at the wing’s location x.
In order to support the matter of the cross-flow bluntness parameter, we try a
gedankenexperiment. We assume that it is possible, like for linear (Model 4) potential
flow, to split the flow field past the delta wing into one part parallel to the wing’s x-
axis and into one part normal to the x-y-plane, parallel to the z-axis. The free-stream
component normal to the wing (z-direction) then is vn = sin α u ∞ , that in x-direction
vt = cos α u ∞ . Accordingly we get Mn = sin α M∞ and Mt = cos α M∞ .
We consider the flow normal to the wing in the form of the flow past the elliptical
cylinder sketched in Fig. 6.3. This cylinder, sufficiently flattened, is to represent the
cross-section of the delta wing. We define δ = b/d as the inverse thickness ratio of
the cylinder. The circular cylinder is given, if δ = 1, the cross section of a delta wing
approximately, if δ is very large.
6.4 When Can an Edge Be Considered as to Be Aerodynamically Sharp? 137

Fig. 6.3 Geometrical parameters of the elliptical cylinder. The inverse thickness ratio is δ = b/d.
ϕ is the polar angle

If the flow is incompressible, linear potential theory, [22], yields for the flow
velocity vc around the cylinder contour
(1 + δ)vn
vc =  . (6.4)
1 + δ 2 cot 2 ϕ

The surface pressure coefficient then reads

vc 2
cp = 1 − ( ) . (6.5)
vn

For the circular cylinder we have δ = 1. Hence we obtain the familiar result at ϕ =
90◦ and 270◦ with the velocity vc,max = 2 vn , and the pressure coefficient c p,min =
−3.
If the elliptical cylinder is flattened more and more, the maximum velocity at
the edges ϕ = 90◦ and 270◦ rises accordingly and with it the favorable pressure
coefficient ahead of these locations and the adverse one behind them Fig. 6.4, left
part, shows this for small δ numbers at ϕ = 90◦ .

Fig. 6.4 Surface pressure coefficients c p (ϕ) [23]. Left part: small inverse thickness ratios δ, c p in
the interval 0◦  ϕ  180◦ . Right part: large inverse thickness ratios δ, c p in the interval 80◦  ϕ
 100◦
138 6 About the Kutta Condition

Fig. 6.5 Gradients of the surface pressure coefficient dc p /dϕ(ϕ) [23]. Left part: small inverse
thickness ratios δ, dc p /dϕ in the interval 0◦  ϕ  180◦ . Right part: large inverse thickness ratios
δ, dc p /dϕ in the interval 80◦  ϕ  100◦ . The locations of the maxima of the adverse gradients
are marked and their magnitudes are given

Note that the pressure distribution is symmetric around the “leading edges” at
ϕ = 90◦ and 270◦ . For large δ numbers we obtain very steep pressure peaks, Fig. 6.4,
right part.
For the small thickness ratios the ϕ-gradient of the pressure coefficient as function
of the polar angle ϕ is given in Fig. 6.5. We see that the maximum of dc p /dϕ rises
with increasing δ and that it moves closer to the very leading edge at ϕ = 90◦ , left
part of the figure. For us of interest is the maximum of the adverse gradient, which
lies above the “leading edge”, i.e. at ϕ > 90◦ .
For the large inverse thickness ratios very high gradients are present, which lie
very close to the “leading edge”, right part of the figure. It appears that at least for δ
≈ 200 the leading edge comes close to act as an aerodynamic sharp one.10
If we now consider the inverse thickness ratio δ as being representative for the—
along the leading edge varying—bluntness parameter pb of a round-edged delta
wing
rle d/2 1
pb |loc = |loc =  |loc ≡ , (6.6)
b b 2δ
we possibly could derive from the above results criteria where ordinary flow sepa-
ration would occur and where the leading edge indeed could be considered as to be
aerodynamically sharp, which would imply flow-off separation.

10 Wegratefully acknowledge that the results shown in the following figures were provided by
C. Weiland [23].
6.4 When Can an Edge Be Considered as to Be Aerodynamically Sharp? 139

Concluding we have to note that it is the gradient of the surface pressure, which
is the most deciding parameter regarding separation, not the gradient of the surface-
pressure coefficient. With
dp ρ∞ dc p
= sin2 α u 2∞ , (6.7)
dϕ 2 dϕ

we see that the angle of attack plays a role, too. The higher this angle, the higher is
the actual maximal pressure gradient.
At this point we finish our gedankenexperiment. To continue we would have
to introduce a separation criterium. For two-dimensional flow the empirical cri-
terion of B. Thwaites for laminar flow, and that of B. S. Stratford for turbulent
flow, see Sect. 2.1, would be candidates. Our flow fields, however, are highly three-
dimensional, such that these criteria cannot simply be employed.
In addition we must remember that our considerations so far were based on linear
potential theory. When looking at the flow of reality, we at least must take into account
that the surface pressure coefficient cannot become negative beyond all limits. The
limit is given by the vacuum pressure coefficient c pvac , Appendix A.1. Figure 6.6
shows this for the general free-stream situation.
In our case, however, with Mn = sin α M∞ the vacuum pressure coefficient
can reach quite large—negative—numbers. Anyway, we must not overstress our
gedankenexperiment.

Fig. 6.6 Vacuum pressure coefficient c pvac as function of the free-stream Mach number M∞
140 6 About the Kutta Condition

6.5 Implicit and Explicit Kutta Condition, Modeling and

Grid Generation Issues

At acute edges, where flow-off separation is present, we speak of an implicit Kutta

condition. If the flow is forced away from the surface by an adverse pressure field,
we have an explicit Kutta condition.
The explicit Kutta condition has been employed—with mixed success—in order
to enforce secondary separation in Euler solutions (Models 7 and 8) of inviscid flow
past delta wings at angle of attack, see, e.g., [24, 25]. In general such approaches are
not to be recommended.
An explicit Kutta condition—enforced by a Kutta panel at the trailing edge of the
airfoil—was employed successfully in order to model the shock-wave decambering
in solutions of the conservative full potential equation (Model 5). In this approach
the purpose of the Kutta panel was to force the trailing-edge flow into the Kutta
direction, Fig. 6.1, Sect. 6.1.2.
A particular approach is that with a free-floating and deformable, i.e., force-free
Kutta panel. With the help of such a panel the grids of discrete numerical methods
(Models 8 and 9) can locally be adapted to the flow pattern. That leads in wakes
of all kinds to a more appropriate grid orientation and also to a more convenient
implementation of turbulence models.
We show two examples, which made use of such a panel [26]. However, the
prerequisite for this approach is the self-organization of the grid with the help of
suitable sensors [27].
The basic situation is given in Fig. 6.7. The left side shows the typical O-grid,
which can be a structured or an unstructured grid. On the right side the grid adapted
to the Kutta direction is given. It is a typical C-grid, in this case a quasi-prismatic
grid. Important is the smooth transition of the grid from the wing surface into the
near wake.

Fig. 6.7 Generic trailing edge discretization [26]. Left side: O-grid. Right side: adapted C-grid
6.5 Implicit and Explicit Kutta Condition, Modeling and Grid Generation Issues 141

Fig. 6.8 Flow past the

RAE-2822 airfoil (M∞ =
0.73, α = 2.79◦ , Re =
6.5·106 ) [26]. Upper part:
initial grid oriented at the
bi-sector of the trailing edge.
Lower part: final grid
oriented at the free-floating
deformable Kutta panel

A realistic example is given in Fig. 6.8. The configuration is the RAE-2822 airfoil
in a transonic (supercritical) condition. The airfoil has a supersonic flow pocket on
the suction side, which is terminated by a shock wave, like shown in Fig. 4.6 on
Sect. 4.2.3.
The grid is a hybrid Cartesian grid, with an embedded quasi-prismatic grid for the
resolution of boundary layers and wakes [27]. The upper part of Fig. 6.8 shows the
typical embedded C-grid, oriented along the bisector of the airfoil’s trailing edge.
At the lower part with the help of a free-floating Kutta panel the embedded grid
is oriented along the center line of the wake. In this case the wake is much better
captured.
We point in particular to the effect of the combined shock-wave and boundary-
layer decambering, Sect. 6.1. The former results in the deflection of the Kutta direc-
tion out and upwards of the bi-sector of the trailing edge, the latter in the thicker
boundary layer at the upper side of the airfoil and also at the upper side of the
wake. The result is reflected in the lower part of Fig. 6.8 by the (self-organized) grid
resolution with a thicker quasi-prismatic grid.
The final example, shown in Fig. 6.9, demonstrates the versatility of the approach.
In the upper part the initial grid for a three-element airfoil in high-lift state is given.
At the sharp trailing edges of both the slat and the main airfoil straight free-floating
deformable Kutta panels are located, oriented at the bi-sectors of the edges. The
Kutta panel at the trailing edge of the main airfoil is short in order not to interfere
with the possible boundary-layer separation over the flap, Sect. 9.2. The flap has a
blunt trailing edge. No Kutta panel is placed there.
142 6 About the Kutta Condition

Fig. 6.9 Flow past a

three-element airfoil (M∞ =
0.2, α = 16◦ , Re = 9·106 )
[26]. Upper part: initial grid
with straight free-floating
deformable Kutta panels.
Lower part: grid after three
iterations

The lower part of the figure shows the grid and the two Kutta panels after three
iterations. The Kutta panels now indicate the center lines of the respective wakes.
The results of the converged solution agrees well with the available experimental
data. For details see [26, 27].

6.6 Problems

Problem 6.1 What is the difference between boundary-layer decambering and

shock-wave decambering? When does which of them appear, what are the reasons?
Consider the airfoil case.

Problem 6.2 Consider an aircraft flying with M = 0.1 at the altitude H = 1 km.
Assume that at the wing’s trailing edge with the thickness h = 10 mm the velocity
is u = u ∞ and also density and viscosity have the free-stream values. How large
approximately is the Reynolds number and hence the Strouhal number? What shed-
ding frequency is to be expected? Is it at the upper bound of the human hearing
domain?
6.6 Problems 143

Fig. 6.10 Equivorticity lines of a wing with a no-slip boundary conditions and b with perfect-slip
boundary conditions

Fig. 6.11 Starting vortex pair with secondary vortices behind a suddenly moved cylinder for Re =
550

Problem 6.3 Why is the angle of attack α a factor regarding dp/dϕ in Eq. 6.7?

Problem 6.4 Derive the formula of the vacuum pressure coefficient.

Problem 6.5 A hypothetical glider with perfect-slip boundary conditions on the

wing can produce a starting vortex and lift just as a glider does with nonslip condi-
tions. Discuss the reason for this behavior with the aid of Fig.6.10.

Problem 6.6 Explain why at higher Reynolds numbers secondary vortices can
develop in the wake of circular cylinders and spheres as shown in Fig. 6.11.

Problem 6.7 The Kutta condition is a hypothesis, necessary in potential-flow

(Model 4) theory, to determine airfoil circulation. In viscous fluid flow, such a con-
dition is superfluous. Discuss the starting process for viscous flows and compare it
with classical airfoil theory.
144 6 About the Kutta Condition

References

1. Hirschel, E.H., Cousteix, J., Kordulla, W.: Three-Dimensional Attached Viscous Flow.
Springer, Berlin (2014)
2. Lighthill, J.: On displacement thickness. J. Fluid Mech. 4, 383–392 (1958)
3. Anderson Jr., J.D.: Fundamentals of Aerodynamics, 5th edn. McGraw Hill, New York (2011)
4. Hirschel, E.H., Lucchi, C.W.: On the Kutta Condition for Transonic Airfoils. MBB-UFE122-
AERO-MT-651, Ottobrunn, Germany (1983)
5. Zierep, J.: Der senkrechte Verdichtungsstoß am gekrümmten Profil. ZAMP, vol. IXb, pp. 764–
776 (1958)
6. Lucchi, C.W.: Shock correction and trailing edge pressure jump in two-dimensional transonic
potential flows at subsonic uniform Mach numbers. In: 6th Computational Fluid Dynamics
Conference, Danvers, MA. Collection of Technical Papers (A83-39351 18-02), AIAA, pp.
23–29 (1983)
7. Lucchi, C.W.: Ein Subdomain-Finite-Element-Verfahren zur Lösung der Konservativen
vollen Potentialgleichung für Transsonische Profilströmungen (A Sub-Domian Finite-Element
Method for the Solution of the Conservative Full Potential Equation for Transonic Airfoil
Flow). Doctoral Thesis, Technical University München, Germany (1984)
8. Klopfer, G.H., Nixon, D.: Non-Isentropic Potential Formulation for Transonic Flows. AIAA-
Paper 83–0375 (1983)
9. Hirschel, E.H.: Basics of Aerothermodynamics, 2nd, revised edition. Springer, Cham (2015)
10. Küchemann, D.: The Aerodynamic Design of Aircraft. Pergamon Press, Oxford: also AIAA
Education Series, p. 2012. Va, AIAA, Reston (1978)
11. Vos, R., Farokhi, S.: Introduction to Transonic Aerodynamics. Springer, Dordrecht (2015)
12. Obert, E.: Aerodynamic Design of Transport Aircraft. IOS Press, Delft (2009)
13. Abbott, I.H., Von Doenhoff, A.E., Stivers, Jr., L.S.: Summary of Airfoil Data. NACA Report
No. 824 (1945)
14. Hoerner, S.F., Borst, H.V.: Fluid-Dynamic Lift. Hoerner Fluid Dynamics, Bricktown (1975)
15. Ackeret, J.: Versuche an Profilen mit abgeschnittener Hinterkante. Vorläufige Mitteilungen der
Aerodynamischen Versuchsanstalt zu Göttingen, Heft 2, 18ff. (1924), also NACA Technical
Meomorandum No. 431 (1927)
16. Henne, P.A.: Innovation with computational aerodynamics: the divergent trailing-edge airfoil.
In: Henne, P.A. (ed.), Applied Computational Aerodynamics. AIAA Educational Series. pp.
221–262. AIAA, Washington, D.C. (1990)
17. Lawaczeck, O., Bütefisch, K.A.: Geplante Untersuchungen über v. Kármánsche Wirbelstraßen
als eine mögliche Ursache für Buffet-Onset. In: Probleme der experimentellen transsonischen
Aerodynamik, W. Lorenz-Meyer (ed.). DFVLR Bericht 251-77 A45, Göttingen, Germany,
6-1–6-8 (1977)
18. Wu, J.-Z., Ma, H.-Y., Zhou, M.-D.: Vorticity and Vortex Dynamics. Springer, Berlin (2006)
19. Herr, M.: Trailing-Edge Noise—Reduction Concepts and Scaling Laws. Doctoral Thesis, Tech-
nical University Braunschweig, Germany, also DLR-FB 2013-32 (2013)
20. Berg, D.E., Zayas, J.R.: Aerodynamic and Aeroacustic Properties of Flatback Airfoils. AIAA-
Paper 2008–1455 (2008)
21. Bangga, G.S.T.A., Lutz, Th., Krämer, E.: Numerical investigation of unsteady aerodynamic
effects on thick flatback airfoils. In: Proceedings of the 12th German Wind Energy Conference
DEWEK, May 2015, Bremen (2015)
22. Schlichting, H., Truckenbrodt, E.: Aerodynamik des Flugzeuges, Vol. 1 and 2, Springer, Berlin/
(1959). Also: Aerodynamics of the Aeroplane, 2nd edition (revised). McGraw Hill Higher
Education, New York (1979)
References 145

23. Weiland, C.: Personal Communication (2018)

24. Newsome, R.W.: A Comparison of Euler and Navier-Stokes Solutions for Supersonic Flow
Over a Conical Delta Wing. AIAA-Paper 85–0111 (1985)
25. Eberle, A., Rizzi, A., Hirschel, E.H.: Numerical Solutions of the Euler Equations for Steady
Flow Problems. Notes on Numerical Fluid Mechanics, vol. 34. Vieweg, Braunschweig Wies-
baden (1992)
26. Deister, F., Hirschel, E.H.: Self-Organizing Hybrid-Cartesian Grid/Solution System for Arbi-
trary Geometries. AIAA-Paper 2000–4406 (2000)
27. Deister, F.: Selbstorganisierendes hybrid-kartesisches Netzverfahren zur Berechnung von Strö-
mungen um komplexe Konfigurationen (Self-Organizing Hybrid-Cartesian Grid System for the
Computation of Flows Past Complex Configurations). Doctoral Thesis, University Stuttgart,
Germany, Fortschrittsberichte VDI, Reihe 7, Strömungstechnik, Nr. 430 (2002)
Chapter 7
Topology of Skin-Friction and Velocity
Fields

The topological analysis of velocity and skin-friction fields potentially is a useful tool
for flow-field interpretations and also for problem diagnosis. In practice, however,
normally not much use is made of topological analysis, maybe because it is often
treated in a rather formalistic way. In view of the topic of this book we intend to
accentuate the practical rather than the theoretical side of the field of topology. Even
this will be made only in a rather sketchy way.
In the introductory Sect. 7.1 we first have a short look at the pertinent literature. It
follows a general examination of the phenomena of interest in the flow fields, which
pose the background of our book: flow attachment and separation, Sect. 7.1.2, as well
as flow detachment, Sect. 7.1.3. In Sect. 7.1.4 finally Lighthill’s separation concept
will be confronted with the concept of open-type separation and also open-type
attachment.
An introduction to singular points of skin-friction fields follows in Sect. 7.2. The
classical approach to that topic is not considered in detail, the objective is the discus-
sion of practical issues. Phase portraits—the flow patterns around a singular point—
are addressed as well as off-surface flow-field portraits.
In Sect. 7.3 the topic of singular lines, i.e. attachment and separation lines, is
introduced. Classical flow topology only looks at the pattern—the phase portrait
(see Sect. 7.2)—of the velocity, respectively the skin-friction field, at and around a
singular point. Regarding singular lines, we take a broader view and concentrate on
the particular flow-field properties connected to attachment and separation lines.
Section 7.4 treats topological rules, again with emphasis on application. The last
section, Sect. 7.5, is devoted to the structural stability of flow fields. Also here we do
not look at it in the classical way. Our objective is to demonstrate practical issues.
In the whole chapter we throughout assume steady flow—though with a few
exceptions. Body surfaces may be flat or curved, the flow may be compressible
or incompressible, laminar or—time-averaged—turbulent. Regarding the coordinate
convention we note that the surface-parallel coordinates are x and z, with the velocity

© Springer-Verlag GmbH Germany, part of Springer Nature 2021 147

E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_7
148 7 Topology of Skin-Friction and Velocity Fields

components accordingly u and w, the surface-normal coordinate is y and the surface-

normal velocity component v. Exceptions are noted explicitly.
In the Chapters devoted to the discussion of results of simulations of the flow
fields past large aspect-ratio wings, Chap. 8, and small aspect-ratio delta-type wings,
Chap. 10, we will use results of the present chapter in order to further elucidate the
findings there.

7.1 Introduction

7.1.1 General Remarks

Classical flow topology mainly treats singular points and in their vicinity the pattern
of skin-friction lines, the phase portraits.
When studying flow separation issues at body surfaces, we also have to look at
flow attachment issues. We follow the classical approach, but also treat singular lines,
i.e. attachment and separation lines.
The reader is warned that we do not go into the details of definitions and derivations
as can be found in the literature. We note in particular the classical NASA Technical
Memorandum, later AGARDograph, of D. J. Peake and M. Tobak [1], the proceedings
of a IUTAM-Symposium [2], two reports of U. Dallmann [3, 4], and the more recent
book of J. Délery [5]. All works should be consulted, if a deeper study of flow
topology is desired.

7.1.2 Three-Dimensional Attachment and Separation

Flow attachment is generally understood as the impingement of the inviscid free

stream on a body surface. The body surface itself is covered by a thin sheet of
viscous flow, the attached viscous flow, which generally is of boundary-layer type
(weak interaction with the external inviscid flow, Sect. 2.2). However, in a separation
region (strong interaction with the external inviscid flow, see Sect. 2.2, too) also
separated viscous flow can (re-)attach on the body surface. A prominent example is
the separation bubble which can appear in two-dimensional form—for instance the
separation bubble over an airfoil with a “peaky” pressure distribution—and also in
three-dimensional form.
We first consider the attachment process assuming inviscid flow throughout. The
free stream impinges in the form of a streamline at, for instance, the forward attach-
ment point, i.e., the nose point, of a fuselage. At this point—the primary stagnation
point—the surface velocity is zero. The flow at the body surface then evolves—
with non-zero velocity along the inviscid surface streamlines—exclusively from this
7.1 Introduction 149

attachment point. No other streamlines than the stagnation-point streamline impinge

on the body surface.1
An attachment line is a location along the body surface where the arriving flow
diverges to the left and the right side of it without impinging on the body surface. The
attachment line itself is a streamline which may originate at the forward stagnation
point, or elsewhere, see Sect. 7.1.4, and is part of a dividing surface which separates
the two diverging flow parts.
Examples are the attachment lines at the leading edge of swept wings and at
the lower symmetry line of (round or nearly round) fuselages at angle of attack,
Fig. 1.10. The attachment point and the attachment line in inviscid flow can well be
prescribed by solutions of the Euler equations, the potential equation or the linearized
potential equation (panel methods), Models 8, 4, 5 in Table 1.3. We loosely call such
attachment lines primary attachment lines.
If the flow on the body surface is viscous, the inviscid flow, impinging or arriving
at the body surface, impresses its pressure field and streamline pattern almost fully
on the boundary layer (weak interaction). The role of the surface streamlines of the
inviscid flow is now taken over by the skin-friction lines. (At the attachment point
the boundary layer has a finite thickness [6].)
A (primary) attachment line usually begins at a singular point. At a round fuselage
at angle of attack only one primary attachment line appears along the lower symmetry
line, Sect. 7.5. However, the primary attachment line can be split into two primary
attachment lines. This is typical for the windward side flow past a flat or almost flat
surface at angle of attack, being the characteristic of a delta wing, Sect. 7.5. However,
a primary attachment line must not necessarily begin at a singular point, then it is an
open-type attachment line, Sect. 7.1.4.
Attachment lines can appear also at other locations of a body surface, usually in
connection with separation phenomena. Then we call them embedded attachment or
reattachment lines. If they appear in a regular pattern, we may call them secondary,
tertiary and so on attachment lines. Examples can be found in Chaps. 8 and 10. Often
these lines are open-type attachment lines.
The separation line in three-dimensional flow is the location, where the viscous
flow, actually two converging boundary layers, separates from the body surface. This
is the case of ordinary separation. The separation line is part of a dividing surface—a
separatrix—, which separates the converging boundary layers on each side of it.2
The classical theory assumes that such a separation line always begins at a singular
point, Sect. 7.1.4. Since long, however, it is accepted that open-type separation is a
reality. In general very complex separation patterns can be present. Like in the case
of attachment lines, secondary, tertiary and so on separation lines can be observed.
Somewhat different is the case of flow-off separation which happens at acute
(sharp) edges. Note that both “sharp” trailing edges and swept leading edges in

1 Note, however, that more than one primary attachment point can be present, depending on the
overall shape of the aircraft.
2 A very detailed discussion of the distinction between attached and separated flows is given in [1].
150 7 Topology of Skin-Friction and Velocity Fields

reality are not sharp in the exact sense of the word. They usually have a finite, though
small thickness, and can only be aerodynamically sharp, Sect. 6.4.
In general it can be assumed that the two boundary layers, which flow off the
surface at the respective edge, have thicknesses much larger than the thickness of
the edge. In reality hence flow-off separation with a thin embedded wake due to
ordinary separation can be assumed to be present. The thin wake may even have an
unsteady character, resembling a von Kármán vortex street. How the geometry of
such a trailing edge is to be modeled in a discrete numerical computation method,
has not yet been satisfactorily established. See in this regard the considerations of
the Kutta condition, Chap. 6.
Of interest when considering separated and vortical flow past bodies of finite
extent are the situations at attachment points and lines, at separation points and lines,
and also at detachment points and lines. The latter are a useful concept in order to
distinguish between the viscous and the inviscid picture of flow leaving a finite body,
see the next sub-section. This is to be recommended in view of the fact that separation
of three-dimensional flow is not defined as simply and unambiguously as that of two-
dimensional flow, Sect. 1.3.3. However, in the literature the term detachment can have
different meanings.
The concept of limiting streamlines was introduced by W. R. Sears [7]. We mention
it here, because it is often used in the literature and point to [6].

7.1.3 Detachment Points and Lines

Regarding detachment points and lines we note that in (sub-critical) inviscid flow—
other than in viscous flow—the flow leaves the surface of a finite body without
transporting kinematically active or inactive vorticity away from the body surface,
see also [6]. The surface streamlines, however, form similar patterns as we find them
in viscous flow for the skin-friction lines. This all holds as long as no Kutta condition,
either explicit or implicit is present.
The term detachment line was introduced in [8]. We consider the inviscid flow
past a wing-like thin ellipsoid at angle of attack, Fig. 7.1. The flow field and the
streamlines were computed with exact potential-flow theory [9] (Model 4).
We look from above at the ellipsoid.3 The forward stagnation point and the forward
dividing streamline—the attachment line—lie, because of the positive angle of attack,
at the lower side of the ellipsoid. The streamline pattern is symmetrical around the
lateral axis (which is the major axis). Hence we find both the rear stagnation point—
which we designate detachment point—and the rear dividing streamline—which we
designate detachment line—at the upper side of the ellipsoid.
In applied aerodynamics less pronounced detachment lines typically appear at the
lee side of fuselages at small to moderate angles of attack. There a thickening of the

3 See also Sect. 4.3.2.

7.1 Introduction 151

Fig. 7.1 Streamlines of the inviscid velocity field past a wing-like 3:1:0.125 ellipsoid at angle of
attack α = 15◦ [8]. View toward the upper side

viscous layer happens without separation of the flow. This might be, for instance, of
interest in design considerations.

7.1.4 Lighthill’s Separation Definition and Open-Type

Separation and Attachment

Before we come to the topic of singular points, we note a matter, which for a time
was much discussed.
In [10] M. J. Lighthill treats the definition of separation. He states as necessary
condition for separation that the particular skin-friction line to which the other skin-
friction lines converge, must originate from a singular point, which is a saddle point,
see, e.g, S1 in Fig. 8.26 in Sect. 8.4.3. Likewise an attachment line must begin at a
singular point, a nodal point, see, e.g, N2 in Fig. 8.26.
Consider the two-dimensional flow past a wing section in Fig. 7.2a. The sepa-
ration streamline emerges from a half-saddle point and then re-attaches, forming a
closed separation bubble. The separation streamline decomposes the flow field into
the bubble flow and the flow above it. The separation streamline hence is called a
separatrix or separator. It separates portions of the flow field. The separatrix emerges
from the half-saddle point on the surface and also ends in such a point.
Two-dimensional separation is structurally unstable, Sect. 7.5. A small perturba-
tion then leads to the flow field shown in Fig. 7.2b. Lighthill’s separation condition,
however, is not violated.
152 7 Topology of Skin-Friction and Velocity Fields

Fig. 7.2 Topological structure of two-dimensional flow separation at a wing section [3]: a structure
with a closed separation bubble, b structure after a disturbance

In three-dimensional flow fields with separation the separatrices are stream sur-
faces which separate flow portions with different origin and behavior. On the body
surface accordingly both attachment and separation lines, i.e., skin-friction lines with
different origin, are separatrices. In Sect. 7.3 we treat them as singular lines.
In any case, Lighthill’s statement, which J. Délery puts it as “A flow is separated
if its surface flow pattern contains at least one saddle point” [5], for a long time was
considered to be the valid definition.
However, in the 1970s K. C. Wang did show that this is not necessarily true.
He described the phenomenon of ‘open-type separation’ [11]. In such a case the
separation line does not begin at a singular point, but at some location in the skin-
friction field, see for example Fig. 8.32 in Sect. 8.4.3. Peake and Tobak call such a
separation pattern ‘local separation’ [1].
However, not only open-type separation can be observed, but also open-type
attachment [6]. This holds for both primary and higher-order attachment lines. An
attachment line may not only begin without a singular (saddle) point, Fig. 8.27 on
Sect. 8.4.3, it also may end without a singular point, Fig. 8.31 in Sect. 8.4.3. Whether
the latter also holds for separation lines has not yet been established. In some instances
they are seen as simply tapering off.
In some of the following chapters we demonstrate and discuss both open-type
attachment and separation lines. We note, though, that open-type attachment to our
knowledge so far has been observed only in numerical realizations of flow fields.
For them holds that reservations are due regarding turbulent flow because of the
employed turbulence models. Regarding laminar flow, this does not hold.

7.2 Singular Points

7.2.1 Flow-Field Continuation and Phase Portraits

The pattern of the skin-friction lines—also that of inviscid streamlines—on the sur-
face of a body can be considered as a continuous vector field. Orthogonal to the
skin-friction lines are surface vortex lines [10].4 Important is that through each point
on the body surface passes one and only one skin-friction line.

4 H.J. Lugt differentiates between vorticity lines and vortex lines [12]. What is called here vortex
lines is in his nomenclature vorticity lines.
7.2 Singular Points 153

But on the body surface always locations are present, where singular points and
singular lines (attachment and separation lines) exist.5 Singular points are locations in
the skin-friction field where the skin friction as well as the surface vorticity become
zero. The number of skin-friction lines passing through such a point generally is
different from one.
Along attachment and separation lines the skin friction is non-zero. Along attach-
ment lines an infinite number of skin-friction lines diverges from them. In contrast
to this along separation lines an infinite number of skin-friction lines converges to
them.
Of interest for us is what happens at singular points, i.e., what is the flow pattern
in the vicinity of a singular point. (Singular lines will be treated in Sect. 7.3.)
The classical approach begins with a flow-field continuation around a surface point
P0 (x0 , y0 , z 0 ). We do not show this, but refer the reader instead to, for instance, [6],
where a very general formulation of the wall compatibility conditions is employed.
The basic assumptions are steady flow and a flat or curved body surface. The flow
may be compressible or incompressible, laminar or—time-averaged—turbulent, heat
flux may be present, the body surface is non-permeable and the no-slip condition
holds. Important is that the flow-field continuation is made with the Navier-Stokes
equations and not with the boundary-layer equations.
For convenience, however, usually the derivation is made in Cartesian coordinates
on a flat surface and for incompressible flow. This does not impair the general validity
of the results.
Our coordinate convention is as follows: x and z are the surface-parallel coor-
dinates, y is the surfaces-normal coordinate, accordingly u and w are the surface-
parallel velocity components and v is the surface-normal component.
With the wall shear-stress components
∂u ∂w
τw x = μ | y=0 , τwz = μ | y=0 , (7.1)
∂y ∂y

we obtain with a Taylor expansion for the velocity components at the point P(x, y, z)
in the vicinity of the surface point P0 (x0 , y0 , z 0 )
 
1 ∂τwx ∂τwx 1 ∂p 2
u| P = τw x y + xy + yz + y + ... , (7.2)
μ ∂x ∂z 2 ∂x P0

 
1 ∂τwz ∂τwz 1 ∂p 2
w| P = τw z y + xy + yz + y + ... , (7.3)
μ ∂x ∂z 2 ∂z P0

    
11 ∂τwx ∂τwz 1 1 ∂p
v| P = − + y 2 + ... = y 2 + .... (7.4)
2μ ∂x ∂z P0 2 μ ∂ y P0

5 We do not always mention detachment points and lines. It is self-evident what also holds for them

in the following discussion.

154 7 Topology of Skin-Friction and Velocity Fields

The systematic identification of singular points then is made with the help of
the phase-plane analysis [13]. That analysis permits to find possible geometrical
configurations of the patterns of the skin-friction lines—or the streamlines—in the
neighborhood of the singular point.6
The final result is a vector and matrix arrangement of the first-order terms of the
equations, see, e.g., [3, 5, 15, 16]
1 1
V (P) = A(P0 ) X + B(P0 ). (7.5)
y μ

Here X = (x, y, z)T is the location vector of point P and A a Jacobian matrix
which contains the expansion terms of lowest order at point P0 :
⎛ ∂τ 1 ∂ p ∂τwx ⎞
wx
⎜ ∂x 2 ∂x ∂z ⎟
⎜ ⎟
⎜ 1 ∂p ⎟
A=⎜
⎜ 0 0 ⎟ ⎟ . (7.6)
⎜ 2 ∂y ⎟
⎝ ∂τ 1 ∂ p ∂τwz ⎠
wz
∂x 2 ∂z ∂z P0

The matrix B contains small terms of higher order.

The eigenvalues of the matrix A read:


   2
1 ∂τwx ∂τwz 1 ∂τwx ∂τwz ∂τwx ∂τwz
λ1,3 = + ± − +4 , (7.7)
2 ∂x ∂z 2 ∂x ∂z ∂z ∂x

1 ∂p
λ2 = . (7.8)
2 ∂y

The matrix A is further analyzed by investigating the trace T , the Jacobian deter-
minant J , and the discriminant  of it, see, e.g., [13].
The trace, the Jacobian determinant, and the discriminant read:

∂τwx 1 ∂ p ∂τwz
T = + + , (7.9)
∂x 2 ∂y ∂z

∂τwx 1 ∂ p ∂τwz ∂τwz 1 ∂ p ∂τwx

J= − , (7.10)
∂x 2 ∂ y ∂z ∂x 2 ∂ y ∂z

 = T 2 − 4J. (7.11)

6 In [14] these are called “phase portraits” of the surface shear-stress vector field.
7.2 Singular Points 155

Fig. 7.3 Patterns of skin-friction lines—phase portraits—in the neighborhood of P0 (x0 , y0 , z 0 )

[6]: basic singular points in the chart of trace T and Jacobian determinant J , with the discriminant
 of the Jacobian matrix A as parameter

The combination and signs of these parameters determine the pattern of the skin-
friction lines in the immediate neighborhood of a singular point P0 (x0 , y0 , z 0 ).
The resulting singular points are collected in Fig. 7.3. These are the basic singular
points. Other singular points or combinations of them (merged points) are also pos-
sible. They are not discussed here. The interested reader instead is referred to, e.g.,
[3, 5, 16].
Two classes of singular points are distinguished: saddle points S (J < 0) and nodal
points N (J > 0,   0), both for attaching flow: T > 0, right-hand side of Fig. 7.3
and separating/detaching flow: T < 0, left-hand side of Fig. 7.3. In topological rules
focus points F (J > 0,  < 0) are counted as nodal points. This also holds for center
points C (J > 0, T = 0).
156 7 Topology of Skin-Friction and Velocity Fields

A nodal point is the common point of an infinite number of skin-friction lines,

which for attaching flow—upper right quadrant of Fig. 7.3—are directed away from
it, and for separating/detaching flow toward it—upper left quadrant. At that point
all skin-friction lines except one—which is normal to it—are tangential to a given
skin-friction line. The star node is an exception.
An infinite number of skin-friction lines is also associated with a focus point,
although without a common tangent line, upper quadrants of Fig. 7.3. No skin-friction
line is associated with the singular point of a center point (T = 0, J > 0).
For both attaching and separating flow two single skin-friction lines are associated
with a saddle point, each toward and away from it—lower left and right quadrant
of Fig. 7.3. All other skin-friction lines in the neighborhood of a saddle point are
deflected from that point in the directions of the single skin-friction lines.
Important is the fact that only the phase portraits lying in the quadrant J > 0, T <
0 are stable ones, see, e.g., [17]. For the phase portraits lying in the other quadrants it
holds that small changes of flow or geometrical parameters change them. This is the
matter of structural instability of singular points to which we come back in Sect. 7.5.

7.2.2 Off-Surface Flow-Field Portraits

With the term off-surface flow-field portrait we denote the off-surface flow pattern in
the vicinity of a singular point P0 (x0 , y0 , z 0 ). This of course is a matter of the pattern
of the attaching or separating flow. We discuss the flow portraits of a few cases,
which we deem to be most important. But we do this also in order to demonstrate
the versatility of the flow-field continuation approach. Much more and more detailed
material can be found, e.g., in [5, 16, 18].
• Two-Dimensional Separation We ask for the angle λ, the separation angle,
at which the separation line leaves the surface.7 This concerns both plane and axis-
symmetric flow, assuming zero or negligible streamwise surface curvature.
The obvious criterion for separation is the vanishing of the wall shear stress τwx ,
actually it is the change of sign of τwx in the separation point, Fig. 1.8 on Sect. 1.3.1.8
If in flow (x−) direction both the functions τwx (x) and pw (x) are known, we can
determine the angle λ, under which the separation streamline leaves the surface.9 In
Fig. 1.8 this streamline is the full line emanating from the separation point, which is
the singular point P0 (x0 , y0 , z 0 ). (That point is indicated in the figure by the circle at
y = 0.) We follow the derivation given by K. Oswatitsch in 1957 [18].

7 The reader should note that λ is not one of the eigenvalues of the matrix A, Eq. 7.6. The designation

λ for the separation angle is used here, because it was used in [6].
8 Experimental evidence of separation usually is given by changes of the wall-pressure distribution

compared to that of the unseparated case. A pressure plateau may be formed, or at the aft of a
two-dimensional body the recompression is severely suppressed.
9 We keep the coordinate convention of the preceding sections.
7.2 Singular Points 157

With Eqs. (7.2) and (7.4) reduced to two-dimensional flow, we find at the separa-
tion point P0 , where τx = 0, with
y v
tanλ = = (7.12)
x u
after some manipulations
∂τwx
∂x
tanλ = −3 ∂p
| P0 . (7.13)
∂x

The pressure gradient ∂ p/∂x is positive, because it is the adverse gradient which
leads to separation. In such a situation of course the gradient of the wall shear stress
∂τwx /∂x is negative. Hence λ is positive, as is to be expected.
In Fig. 1.8 on Sect. 1.3.1 the broken line emanating from the separation point—
the circle—is the location of the turning of the streamlines in the region beneath the
separation line. This line is defined by the disappearance of the tangential velocity
component: u(x, y) = 0.
The elevation angle λ∗ of this line is found from Eq. 7.2 to be
2
tanλ∗ = tan λ. (7.14)
3
Remains to ask how the separation line behaves away from the surface. In general
it is to be expected that it turns more or less strongly toward the body surface.
Moreover, it is obvious that the above result cannot be obtained by employing the
boundary-layer equations. That is possible only with the help of solutions of the
Navier-Stokes or RANS equations (Models 9 and 10 in Table 1.3).
• Stagnation Point We consider the primary stagnation point at a body surface.
There we find in the axis-symmetric case the star nodal point as an attachment node
as shown in the upper right quadrant of Fig. 7.3. This is an isotropic node. At a general
stagnation point, for instance at a blunt body at angle of attack, the star nodal point
is non-isotropic [5]. A cut through the node, orthogonally to the surface, gives the
picture like shown in Fig. 7.8b for the axis-symmetric case: the nodal point becomes
a half-saddle point.
At the primary attachment point we have an absolute pressure maximum, the
stagnation pressure. The lateral pressure gradient is zero (we write it for the positive
x-direction only): ∂ p/∂x = 0. Downstream of the attachment point we have a negative
pressure gradient: ∂ p/∂x < 0, i.e., the flow gets accelerated away from the stagnation
point.
The skin-friction has an absolute minimum at the primary attachment point: τx =
0. The lateral gradient is zero: ∂τx /∂x = 0. Downstream of the attachment point we
have a positive skin-friction gradient: ∂τx /∂x > 0, i.e., the skin-friction increases.
We emphasize that at the stagnation point the boundary layer has a finite thickness
δ [6].
158 7 Topology of Skin-Friction and Velocity Fields

Fig. 7.4 Sketch of a

streamline assumed to leave
the body surface along a
separation line [6]

How does the stagnation-point streamline impinge on the surface? The relation
for the angle λ at the two-dimensional separation point, Eq. (7.13), also holds for this
case. With ∂ p/∂x negative, and ∂τx /∂x positive in its vicinity downstream, when
approaching the attachment point, tanλ approaches infinity.
The result is that the attaching streamline at right angles impinges to the surface
in that point. This is the same result which potential theory gives for the stagnation
point. Generalizing this we state that both in inviscid and viscous flow the streamline
impinges on the primary stagnation point at a right angle.
In the upper left quadrant of Fig. 7.3 a separation node is shown as star node. Such
a nodal point is shown at the aft end of the body in Fig. 7.8a. In reality such a node
will not appear there, because in viscous flow we find separation upstream of the
aft end of the body. Instead we may have a picture like shown in Fig. 7.9. There we
have an attachment nodal point at the aft end of the body, which in the cut shows up
as a half-saddle point. The separation star node of Fig. 7.3 generally appears to be
possible only in the inviscid flow picture (detachment).
• The Three-Dimensional Separation Line Above it was stated that a stream-
line only leaves the body surface at a singular point. What happens along a three-
dimensional separation line? Indeed, no streamline leaves that line. We show this
like it was done in [18].
We assume that locally the direction of a hypothetical separating skin-friction line
is the same as that of the separation line itself, Fig. 7.4.
The angle λ2 , under which the streamline is assumed to leave the surface, is
y v
tanλ2 = = . (7.15)
x u
Putting Eqs. (7.2) and (7.4) into Eq. (7.15) we obtain at the surface (z = 0):

1 ( ∂τ
∂x
x
+ ∂τ
∂z
z
)y
tanλ2 = − |P . (7.16)
2 τx + ∂τx x + 1 ∂ p y 0
∂x 2 ∂x

With tanλ2 → y/x for y → 0, x → 0, this equation can be rearranged to yield

7.2 Singular Points 159

Fig. 7.5 Illustration of a

vortex filament which leaves
the body surface in a focus
point [18]. Thin lines are
skin-friction lines, full lines
streamlines

∂τz
3 ∂τ
∂x
x
+ ∂z
+ 2 τxx
tanλ2 = − ∂p
| P0 . (7.17)
∂x

We see immediately that no meaningful result can be obtained from this equation
for λ2 unless we have τx | P0 = 0. This would mean, because in P0 by definition τz =
0, that P0 is a singular point.
The result is that indeed along the separation line no streamline can leave the
surface. That is possible only in the singular point P0 . In the two-dimensional case
∂τz /∂z ≡ 0, and we are back to the result of Eq. (7.13).
• The Focus Point A very particular singular point is the focus point, which in
the upper quadrants of Fig. 7.3 is indicated for both separating and attaching flow. In
the first case a vortex filament leaves the surface, in the other case such a filament
impinges on the surface.
Figure 7.5 shows such a vortex filament, which was obtained by flow-field contin-
uation [18]. Neither a separation line nor a separation surface is involved, instead an
infinite number of skin-friction lines and streamlines converges toward the axis of the
filament. Oswatitsch in [18] notes that it may be a question of definition, whether to
call this separation or not. In any case, vortex filaments of this type can be observed
in reality, see, e.g., Fig. 8.29 in Sect. 8.4.3.
• Flow-Off Separation At the sharp trailing edge of an airfoil potential theory
(Model 4) yields a singularity. Actually in viscous flow we have flow-off separation
there. From the very edge a streamline leaves the airfoil, constituting the center line
of the wake. Even if it may be against our intuition we must consider the edge point
as a singular point. Actually it is a half-saddle point, Sect. 7.4, like the stagnation
point at the nose of the airfoil.
In the reality of an aircraft’s wing the flow situation at the trailing edge is complex,
Chap. 6. The trailing edge has a finite thickness, we have decambering effects and,
moreover, in the case of the lifting wing, we have a shear between the upper and
160 7 Topology of Skin-Friction and Velocity Fields

Fig. 7.6 Sketch of a

separation line and the
separating stream surface
leaving the body surface [6]

lower side flow, Sect. 4.4. This all holds too for the flow-off separation at the sharp
leading edge of a delta wing.
• The Separating Stream Surface We consider now ordinary separation, and
there the—dividing—stream surface between the two boundary-layer streams, which
squeeze each other off the surface, Sect. 1.3.3. Its intersection with the body surface
is the separation line, Fig. 7.6.
With the help of Eq. (7.3) it is possible to estimate the angle λ1 of the separating
surface with respect to the z-direction at the point P0 on the separation line. At that
point the separating surface is defined by τz = 0. For z  0, if the elevation angle of
the flow at the surface is small, we have w ≈ 0.
Thus we find for the angle λ1 with w = 0 from Eq. (7.3):
∂τz
∂z
tanλ1 = −2 | .
∂ p P0
(7.18)
∂z

Our result so far is that streamlines do not depart upwards from a separation line
(see above “The Three-Dimensional Separation Line”). The separating stream sur-
face, which emanates from it, is formed by the two boundary-layer streams squeezing
each other off the body surface. The question now is, what is the situation at attach-
ment lines?
• The Attaching Stream Surface With the reasoning which we used for the
separation line, it can be shown that no streamline impinges on the attachment line
along its length. From Eq. (7.18) we deduce also that the attachment stream surface
stands at a right angle to the surface, if ∂ p/∂z → o for z → 0, whereas ∂τz /∂z is
finite.
• The Boundary-Layer Flow Being Parallel to the Body Surface? Above it
was stated and also shown that only in a few—singular—points on the body sur-
face streamlines actually impinge on or leave the body surface. This implies that in
attached viscous flow very close to a non-permeable wall the boundary-layer flow is
parallel to the body surface.
To prove this we consider a two-dimensional boundary layer. The result also holds
for three-dimensional boundary layers.
7.2 Singular Points 161

We again use the result of the Taylor expansion in terms of Eqs. (7.2) and (7.4).
We find for small distances y from the surface

u ∼ y, v ∼ y 2 , (7.19)

and hence the streamline elevation angle θ in relation to the surface

v
tan θ = ∼ y. (7.20)
u
The result is that when the surface is approached in attached boundary-layer flow,
the flow in the limit becomes parallel to it:

y → 0 : θ → 0. (7.21)

We keep in mind that in a boundary layer y and v are small compared to x and
essence of boundary-layer theory that y and v have the order of
u. Actually it is the
magnitude O(1/ Rer e f ) [6].
Hence the above result tells us further that the whole viscous flow in the boundary-
layer flow limit Rer e f → ∞ becomes parallel to the surface. A purely inviscid flow
at the body surface of course anyway is parallel to that.
Our result plays also a role in hydrodynamic stability theory. The derivation of
the Orr-Sommerfeld equation assumes that the flow is exactly parallel to the surface,
see, e.g., [6].10 For not so large Reynolds numbers “non-parallel effects” as well as
“surface-curvature effects” come into play and are a topic of stability theory.

7.2.3 Singular Points in Off-Surface Velocity Fields

On a body surface singular points can change their character, for instance in a plane
normal to the surface through a singular point. Singular points on the surface then
become, for instance, half-nodes N
or half-saddles S
, Sect. 7.4.
Away from the body surface the patterns of streamlines around singular points
can be very different. This holds for both viscous and inviscid flow. (This implies
that singular points are not necessarily connected only to skin-friction line patterns.)
In certain cases we can clearly observe streamlines, which connect singular points
on the body surface with singular points in the flow field away from the surface. This
even holds if we have unsteady flow due to vortex shedding. Then instantaneous
patterns can be observed.
In other cases the situation needs to be considered with care. This holds, for
instance, when we consider the flow pattern in a surface orthogonally to the center line
of a delta wing with lee-side vortices. Here, like above, streamlines appear to connect

10 The Orr-Sommerfeld equation describes when a two-dimensional laminar boundary layer

becomes unstable, triggering laminar-turbulent transition, see, e.g., [6].

162 7 Topology of Skin-Friction and Velocity Fields

singular points on the body surface with singular points away from the surface.
However, the streamlines are only apparent ones. Actually they are the streamlines
traces, which are projected into the observation plane, the Poincaré surface [19]. In
[5] they are called pseudo-streamlines.
We will meet these cases again in Sect. 7.4, where we treat topological rules.

7.3 Singular Lines

Attachment and separation lines generally are not a big topic in flow-field topology.
They connect—in Lighthill’s picture, Sect. 7.1.4—singular points, or—a geometrical
view—they separate flow domains, hence they are called also separators (see above).
Here we call these lines singular lines, because certain geometrical and flow-field
properties along them behave in a singular way compared to the properties of the
main flow domains over the body surface. Attachment and separation lines are always
present on an aircraft’s surface and some of the flow properties to be observed along
these lines are of considerable practical interest.
In order to avoid misconceptions we note that an attachment line is not a stagnation
line. The only exception is the case of an unswept wing of infinite span, and that is the
two-dimensional case. The same holds for a separation line. Along both attachment
and separation lines the flow velocity and the wall-shear stress are non-zero.
We study now these properties following closely the presentation given in [6].
However, we do that in a descriptive way only, without the proofs given there. The
reader interested in the proofs is referred to that publication.
Singular lines in the frame of this section are primary, secondary etc. attachment
and separation lines. Primary attachment lines as a rule are attachment lines also of
the inviscid flow past a body. An attachment line may have its origin in a singular
point, Fig. 7.8a. That can be a nodal point—forward (primary) stagnation point—or
a saddle point. However, like open-type separation, also open-type attachment is
possible, Sect. 7.1.4. In later chapters examples are given.
At an ordinary airplane configuration with or without swept wings, only one
primary stagnation point is present. That is located at the nose of the fuselage.11 If
the airplane has a forward swept wing, three primary attachment points are found.
We remember, see Sect. 7.2.2, that only in a singular point—attachment point—a
streamline impinges on the body surface. Along an attachment line this does not
happen. This holds for both viscous and inviscid flow. It also holds that a streamline
never becomes a skin-friction line.
Separation lines are present only in viscous flow past a finite body, either as
ordinary or as flow-off separation lines. Only in a singular point—separation point—
a streamline leaves the body surface. Along a separation line this does not happen.
A skin-friction line never becomes a streamline which leaves the body surface. In
inviscid flow the picture is similar with detachment lines and detachment points.

11 We neglect possible forward stagnation points at the propulsion units and at antennas and the like.
7.3 Singular Lines 163

Fig. 7.7 Schematic of

general singular lines of both
inviscid and viscous flow: a
attachment line, b ordinary
separation/detachment line
[6]

The reader is asked to contemplate these statements. Open attachment or separa-

tion lines do not have a singular point at their beginning, partly also not at their end.
We indeed have a singular behavior of the flow in such cases.
The above holds quite in general for every configuration and the flow past it.
The results of our investigations in principle apply to any kind of attachment and
separation lines.
Attachment and separation/detachment lines appear in two canonical forms as
shown in Fig. 7.7. Typical for an attachment line (a) is that an infinite number of
skin-friction lines diverges from it. This holds also for the surface streamlines of the
related inviscid flow.
However, the inviscid and the viscous attachment lines generally do not coin-
cide with each other. The reason is that the skin-friction lines of three-dimensional
attached viscous flow are more strongly curved than the surface streamlines of the
related (external) inviscid flow [6]. The two attachment lines coincide only if they
lie, for instance, on the surface generator of an infinite swept wing (ISW).12
Typical for a ordinary separation line (b) is that an infinite number of skin-friction
lines converges toward it. This holds also for the streamlines at a detachment line of
inviscid flow.
We look now at the interesting flow-field properties of attachment and ordinary
separation/detachment lines and single out the following five items:
1. Relative maximum of the surface pressure At a curved inviscid attachment line,
the surface pressure has a relative maximum in direction normal to the inviscid
attachment line, 1 - · - 1 in Fig. 7.7a. In general, the pmax -line lies close to the
attachment line (also to the viscous one), and only on it, if the ISW situation is
given. The location of the relative pressure maximum can be of interest in view
of laminar flow control.
2. Points-of-inflection line On the convex sides of the singular lines shown in
Fig. 7.7, the stream lines or the skin-friction lines have a point of inflection. A

12 This surface generator is a geodesic, the boundary layer is a quasi-two-dimensional one [6].
164 7 Topology of Skin-Friction and Velocity Fields

points-of-inflection line is present at every curved attachment or separation line.

In general it lies close to the respective attachment or separation line. In the ISW
situation it disappears. The point-of-inflection line is a purely geometrical prop-
erty, interesting as visual indicator of three-dimensional attachment or separation
lines.
3. Characteristic thickness of the viscous layer At an attachment line the charac-
teristic thickness of the viscous layer c has a relative minimum in the direction
orthogonal to it, 1 - · - 1 in Fig. 7.7a.13 At a separation line c has a relative
maximum in the direction orthogonal to it, 2 - · - 2 in Fig. 7.7b. This thickness
property is determining in view of property 5.
4. Relative minimum of |τw | Along attachment and separation lines the skin-friction
is non-zero. Normal to a viscous attachment line, 1 - · - 1 in Fig. 7.7a, or to a
separation line, 2 - · - 2 in Fig. 7.7b, the absolute value |τw | of the skin-friction
vector has a relative minimum. The minimum lies close to the respective line.
This property is interesting also as indicator of an attachment or separation line.
5. Extrema of the thermal state of the surface Along attachment and separation lines
the wall temperature Tw and the heat flux in the gas at the wall qgw are non-zero.
Normal to an attachment line, 1 - · - 1 in Fig. 7.7a, a relative maximum, and
normal to a separation line, 2 - · - 2 in Fig. 7.7b, a relative minimum exist. These
extrema lie close to the respective line. Extrema of the thermal state are of utmost
importance for hypersonic flight vehicles. There along attachment lines hot-spot
situations are present and along separation lines cold-spot situations [20].
We emphasize that at attachment lines the distances of the extremum lines or the
points-of-inflection lines to the attachment lines as such in general are very small.
The reason for this is that an attachment line usually is only weakly curved. In the
ISW situation, the pmax -line, the points-of-inflection line and the |τw |-minimum line
lie on it.
At separation lines, the situation is different. At ordinary separation lines the dis-
tances between the separation line and the extremum lines of the skin-friction vector
and the thermal state of the surface are not necessarily very small. This holds also for
points-of-inflection lines. The reason for this is that ordinary separation in general
is not related directly to curvature maxima of the surface. This does not hold for
flow-off separation lines at sharp edges, for instance wing trailing edges, Fig. 1.10a,
or sharp leading edges of highly swept wings, Fig. 1.10b. There the distances can be
very small.
The proofs of the listed properties are given in [6]. The results in principle are
valid for incompressible and compressible, laminar and turbulent flow. However, for
convenience, the presentations are sometimes simplified. Detachment lines are not
treated as separate topic. What applies for them of the above five items is more or
less self-evident.
In closing this section, we ask how to recognize three-dimensional attachment
and in particular separation lines on a body surface. Visually the respective patterns

13 For the meaning and the definition of the characteristic thickness c see Appendix A.5.4.
7.3 Singular Lines 165

are easily to recognize, see the many examples in the book. But there is no simple
separation criterion like in two-dimensional flow with τw = 0.
In [21] the following indicators have been proposed to detect separation in com-
puted data:
1. Local convergence of skin-friction lines.
2. Occurrence of a |τw |-minimum line.
3. Bulging of the boundary-layer thickness and the displacement thickness.
For attachment lines we note accordingly:
1. Local divergence of skin-friction lines.
2. Occurrence of a | pw |-maximum line.
3. Occurrence of a |τw |-minimum line.
4. Indentation of the boundary-layer thickness and the displacement thickness.

7.4 Topological Rules

7.4.1 Introduction

Topological rules give the relationship of singular points on a body surface or in

a general observation surface. We give a short introduction to them. We basically
follow the presentation given in [6]. For a deeper study see, e.g., [1] or [5].
General assumptions are that the flow is steady, the body is simply connected, that
the velocity field and the skin-friction field past the body are continuous, and that the
body is immersed in a uniform upstream flow field.14 Nodal points and focus points
are topologically equivalent, hence are counted together. Only a few illustrations are
given, more are to be found in the following section and in later chapters.

7.4.2 Surface Rules

Two basic rules are considered:

• Rule 1, due to A. Davey [22] and also M. J. Lighthill [10], concerns the inviscid
surface-velocity field or the skin-friction field on a three-dimensional body. It says
that on the body surface the number of nodal points N (focus points are counted as
nodal points) is larger than the number of saddle points S by two:

14 Time-dependent flow can be treated by applying the rules to the instantaneous flow field [5].
166 7 Topology of Skin-Friction and Velocity Fields

Fig. 7.8 Schematic of

steady inviscid flow past an
axis-symmetric body. a
Streamlines on the surface:
two nodal points N . b
Streamlines seen in the
two-dimensional plane
cutting the body through its
axis: two half-saddle points
S

 
N− S = 2. (7.22)

This rule holds for simply connected body surfaces. For non-simply connected
surfaces the complexity p is taken into account by
 
N− S = 2 − 2 p. (7.23)

For simply connected surfaces we have p = 0. If one hole is present in the surface
p = 1, for two holes p = 2 and so on [5].
• Rule 2, due to J. C. R. Hunt et al. [23], concerns skin-friction lines and stream-
lines in a two-dimensional plane cutting a three-dimensional body. The sum of nodal
points N plus one half of the number of half-nodal points N
is one less than the sum
of saddle points S plus one half of the number of half-saddle points S
:
 1
 1

( N+ N )−( S+ S ) = −1. (7.24)

2 2
We illustrate these two rules with the flow past an axis-symmetric body. Figure
7.8a shows the streamlines of the inviscid flow at the body surface with the forward
(attachment) and the rearward (detachment) stagnation point. Both are nodal points,
Rule 1 is fulfilled because saddle points are absent. In Fig. 7.8b the body is cut by a
two-dimensional plane through its axis. The nodal points of Fig. 7.8a now become
half-saddle points. Rule 2 is fulfilled.

7.4.3 Off-Surface Rules

If the flow near the surface is viscous, we see separation at some aft location of the
body, Fig. 7.9. We assume a steady separation region.15 The forward and the rearward

15 In reality this is given only for very small Reynolds numbers, see, for instance, Fig. 7.17 and also

the flow visualizations in [24].

7.4 Topological Rules 167

Fig. 7.9 Schematic of steady viscous flow past an axis-symmetric body. Skin-friction lines and
streamlines seen in the two-dimensional plane cutting the body through its axis: one saddle point
S, four half-saddle points S
, two focus points F, counted as nodal points

stagnation point are half-saddles. Although inviscid flow attaches at the forward
stagnation point, S1
, the flow there is viscous, i.e. the viscous layer or boundary layer
at that point has a finite thickness.
Note that the rearward stagnation point, S2
, now is an attachment point, too. The
shear layers emanating from S3
and S4
are merging and split in S1 and then move
partly, with a wake-like appearance, toward S2
. Actually it is the circumferential
separation line which shows up in the cutting plane as the two half-saddle points
S3
and S4
. The separation region is a toroid, which shows up as two center points
(degenerated focus points) F1 and F2 . The separation region is closed by a saddle
point, S1 . Rule 2 is fulfilled.
Quarter-saddle points were introduced in [25] in order to treat the topology of the
flow past a delta wing, independent of the type of leading edge, sharp or rounded. If
the lower side of a delta wing—or of a fuselage—is flat or nearly flat, two primary
attachment lines lie near the side edges of the lower surface, Sect. 7.5. Between them
the flow is nearly or fully two-dimensional.
Consider now Fig. 7.10. The skin-friction lines at the lower side of the wing leave
the primary attachment lines like sketched in Fig. 7.21 on Sect. 7.5. In a plane two-
dimensional cut A-A, the Poincaré surface, then only the traces of the outward flow
are seen and the primary attachment lines appear as quarter-saddle points S

.16
• Rule 2 then changes into Rule 2’:
 1
 1 
1 

( N+ N )−( S+ S + S ) = −1, (7.25)

2 2 4
which in our case results in
1 1
4 − (1 + 7 + 2) = −1. (7.26)
2 4
If the lower side of the body or wing has a convex shape—like shown in Fig. 7.20
on Sect. 7.5—the attachment line lies at the lower apex of it. Then, instead of the

16 Note that with the attachment lines—the two primary lines, and also (!) with the two secondary

lines and the tertiary line—inviscid flow attaches. However, the attachment-line flows themselves
are viscous, i.e. the viscous layers or boundary layers at those lines have finite thicknesses, Sect. 7.3.
168 7 Topology of Skin-Friction and Velocity Fields

Fig. 7.10 Sketch of steady viscous flow past a delta wing with primary and secondary lee-side
vortices [25]. Skin-friction lines and streamlines seen in the plane two-dimensional cut A-A, i.e. the
Poincaré surface: one saddle point S (•), seven half saddle points S
(◦), two quarter-saddle points
S

(), four focus points F (×), counted as nodal points

Fig. 7.11 Sketch of the open

lee-side flow field in the
Poincaré surface. Possible
secondary and higher-order
vortical structures were
omitted

two quarter-saddle points present at the lower side of the delta wing in Fig. 7.10, one
half-saddle point is present and Rule 2 is fulfilled.
Finally we note two principally different lee-side flow topologies, which can be
observed in a Poincaré surface. The first one, which we call open lee-side flow field,
is sketched in Fig. 7.11.
It appears at the lee side of delta wings, depending on the shape of the wing and
the angle of attack, Sect. 7.5. However, we observe this kind of topology also at large
aspect-ratio wings, there regarding the wing-tip vortices. In later chapters examples
are shown.
The other lee-side flow topology is the closed lee-side flow field, Fig. 7.12. That
is observed primarily above delta wings at high angles of attack.
A problem, which sometime occurs, is the interpretation of a particular flow
pattern. Consider in Fig. 7.13 the inviscid flow streamlines at the upper surface of the
thin ellipsoid from Sect. 7.1.3. The angle of attack now is α = 30◦ . The flow solution
again was made with potential theory (potential flow, Model 3).17

17 We gratefully acknowledge that the following figures were provided by C. Weiland [26].
7.4 Topological Rules 169

Fig. 7.12 Sketch of the

closed lee-side flow field in
the Poincaré surface.
Possible secondary and
higher-order vortical
structures were omitted

Fig. 7.13 Streamlines at the

upper surface of a 3:1:0.125
ellipsoid at the angle of
attack α = 30◦

We look from above at the flow field and see the detachment point and the detach-
ment line. (The streamlines coming from the lower side to these entities are not
plotted.) Attachment point and attachment line lie at the lower side of the ellipsoid.
Along the centerline the flow appears to be completely in x-direction.
What we expect to see in the Poincaré surface is the trace of the center stream-
line toward the lower side of the ellipsoid and away from the upper side, Fig. 7.14.
This would demand—at the upper side—in Fig. 7.13 a recognizable flow component
toward the centerline.
This more or less hidden flow pattern can be visualized by considering two par-
ticular solutions of the potential equation, one for α = 0◦ and one for α = 90◦ . In
[26] this was achieved by setting in the first case u ∞ = sin 30◦ and in the second
u ∞ = cos 30◦ . Because we deal with potential theory, the computed velocity fields
can be superimposed. The resulting flow field is that shown in Fig. 7.13.

Fig. 7.14 The Poincaré surface at x = 0.5

170 7 Topology of Skin-Friction and Velocity Fields

Fig. 7.15 Streamlines of the two particular solutions at the upper surface of the ellipsoid. Left side:
solution for α = 0◦ . Right side: solution for α = 90◦

Indeed the α = 90◦ solution clearly shows the flow turning toward the centerline.
Moreover, the dot in the right picture indicates the upper detachment point where
the flow leaves the surface.
With all due reservations this is a possible explanation. Of course with that we
have a special case, which cannot be generalized. In the reality of three-dimensional
compressible viscous flow other considerations may be helpful.

7.5 Structural Stability and Changes of Flow Fields

Above we have noted that only the phase portraits lying in the quadrant J > 0 and T <
0 of Fig. 7.3 are stable. The flow field past a body depends on the parameters Reynolds
number, Mach number, thermal state of the surface, see, e.g., [20], as well as on the
angles of attack, of side-slip, and on the body geometry and its aero(thermal)elastic
deformation.
If one or more of these quantities are changed, the flow field past the body changes.
If then the topological structure of the flow field, i.e., the number and the kind of the
singular points and their connections remain unchanged, the structure is topologically
stable.
If, however, small changes of one or more of the above quantities lead to a new
topological structure of the flow field, the original topological structure is unstable.
In that case (a) new singular points may appear, original ones may disappear, and
(b) the connections of the singular points may change. In case (a) we have a local
bifurcation, that is the flow field is changed only locally. In case (b) we speak of a
global bifurcation, because the whole flow field is changed.
Global bifurcation is to be expected, if, for instance, saddle point-saddle point
structures are given, like shown in Fig. 7.2a. The separation bubble there is beginning
and ending with a half-saddle point each.
7.5 Structural Stability and Changes of Flow Fields 171

Fig. 7.16 Structural change of the flow at the aft of a blunt body [27]. Local bifurcation: A → B,
steady separation; global bifurcation: B → C, periodic separation (vortex shedding)

Consider the situation shown in Fig. 7.16. Schematically the flow field at a blunt
afterbody is given with the Reynolds number Re as bifurcation parameter (from the
thesis of B. Schulte-Werning, a doctoral student of the first author of this book [27]).
From case A, attached viscous flow at very small Reynolds number Re (two half-
saddle points, one is located at the forward stagnation point, Rule 2 is fulfilled), the
flow field changes with increasing Re to case B, where we see at the rear stagnation
point a steady symmetric separation pattern. This is considered to be a local bifurca-
tion. We count now one saddle point, four half-saddle points, and two focus (nodal)
points. Also with that Rule 2 is fulfilled, i.e., here, too, we have a valid topological
structure.
If Re is increased further, global bifurcation leads to a break-up of the connections
between the original saddle points and a periodic vortex shedding is present, case C.
At a real body a further increase of Re then will lead to three-dimensional unsteady
separation and vortex shedding.
We illustrate the above and in particular the latter observation with the flow past
a sphere as function of increasing Reynolds-number domains, Fig. 7.17.
The figure is based on a compilation in [27] of experimental and theoretical data
given in the literature. We observe the following separation and wake phenomena as
function of the Reynolds number Re:
(1) Re < 20. Creeping flow, no separation behind the sphere.
(2) 20 < Re < 500. Steady annular vortex behind the sphere. For Re > 130 the free
stagnation point behind the sphere begins to oscillate and the wake becomes
unstable.
(3) 500 < Re < 1.5·103 . Periodic shedding of annular vortices and appearance of
a “vortex chain”.
(4) 1.5·103 < Re < 104 . Discrete annular vortices are forming an “annular vortex
street”, which, however, disintegrates shortly behind the sphere. Periodic vortex
clusters appear.
(5) 104 < Re < 3.7·105 . Rotating separation of two vortex filaments at the back
side of the sphere and formation of a helix-like wavy wake.
(6) 3.7·105 < Re. Turbulent boundary layer over the sphere (the figure indicates
a transition triggering device). Separation happens now far behind the spheres
172 7 Topology of Skin-Friction and Velocity Fields

Fig. 7.17 Structures of wakes of a sphere in steady motion as function of the Reynolds number
[27]

equator. A strong drag decrease results. In the time-average a horse-shoe like

separation of a vortex sheet appears.

It must be noted that this classification is a very crude one [27]. It is not unam-
biguous whether the wake development always is an unsteady three-dimensional
phenomenon and whether the structural changes of the flow topology are as shown.
Much depends on the setup of an experiment or a numerical investigation. Therefore
the different experimental and theoretical/numerical investigations may have led to
different flow-field realizations.
The sphere is geometrically a very simple body. The flow past it can be very
complex as we just have seen, with several structural changes. A strong tendency to
structural changes of the separation topology, however, is seen also on hemisphere-
nosed cylindrical configurations, which represent the simplest form of an aircraft’s
fuselage. In [28] the interested reader will find experimental results obtained with
such a configuration for Mach numbers in the subsonic and the transonic regime,
with laminar, forced and natural turbulent flow, at angles of attack up to α = 40◦ .
The flow topology is analyzed, also cases with open-type separation are identified.
Further discussions of these experimental data are contained in [3, 4].
Structural changes do not necessarily happen instantly. We look at Fig. 7.18 where
the topology of the skin-friction lines on the upper side of a delta wing and the pseudo-
streamlines in the Poincaré surface are sketched for different angle-of-attack ranges
[29].
First we study the lee-side picture, looking at the left side of the wing, see also
the discussion of Fig. 10.15 on Sect. 10.2.2.
7.5 Structural Stability and Changes of Flow Fields 173

Fig. 7.18 Possible flow structures at a slender delta wing with sharp leading edges [29]. Upper
part: small angle of attack, accordingly only a small separation bubble along the leading edge, open
lee-side flow field. Middle part: medium angle of attack, lee-side vortices, secondary vortices, open
lee-side flow field. Lower part: large angle of attack, lee-side vortices, secondary vortices, closed
lee-side flow field
174 7 Topology of Skin-Friction and Velocity Fields

• Leeward side
At the beginning of our consideration we note that in general over a slender delta-
wing leading-edge separation phenomena begin to appear at the rear of the wing.
With increasing angle of attack they move forward in direction of the wing tip.

• Small angle of attack At small α a small three-dimensional separation bubble is

present along the leading edge. We see a kind of an open lee-side flow field with
parallel or conical flow at the center domain.
• Medium angle of attack A fully developed lee-side vortex system—with secondary
vortices—is present. The flow field is of open lee-side type. At the center domain
again parallel or conical flow is present.
• High angle of attack The lee-side flow field is of closed type. The vortex system is
like that at medium angle of attack. Along the center line now lies an attachment
line. Skin-friction lines diverge from it to both sides.

How the change from one structure to the other happens, is an open question.
Criteria for the appearance of either flow structure are not known to the authors of
this book. Deciding parameters are the wing shape, the leading-edge shape, angle
of attack, free-stream Mach and Reynolds numbers. The structure changes basically
also appear at delta wings with round leading edges. At transonic and supersonic
free-stream Mach number embedded cross-flow shocks are present.
Additionally we note that at very slender wings, in general slender configurations,
the lee-side vortex system with increasing angle of attack becomes asymmetric with
a strong influence on forces and moments, Fig. 7.19, see also Sect. 10.2.5. This phe-
nomenon is discussed in detail in [1] from the side of structural stability/instability.
• Windward side Whereas the lee-side flow field with their vortex phenomena has
found widespread interest, the flow at the windward side almost none. As mentioned
in Sect. 7.4.3 we can distinguish between two classes of configurations: (1) more or
less round fuselage, (2) flat or nearly flat lower surface of a wing or a fuselage, in
particular a delta wing with round or sharp leading edges.
For class (1) we get the picture shown in Fig. 7.20: at angle of attack the primary
attachment line lies on the windward surface centerline of the forebody. There it will
remain with increasing angle of attack unless at very high angle the whole flow field
becomes asymmetrical and even unsteady.
But what happens, if we continuously flatten the body until we have the shape of
a delta wing? Obviously at some station of the process a bifurcation will occur and
we get the two attachment lines typical for the flat delta wing.
In class (2) the situation is quite different. A gedankenexperiment shows us that at
zero angle of attack of for instance a symmetrically round leading-edged delta wing
the primary attachment lines are located on the generatrices of the leading edges.
This is exactly the situation, which we have at a large aspect-ratio wing at zero angle
of attack. If the angle of attack is increased, the attachment lines move a little distance
below the generatrices. Even at high angles of attack this distance will be small. At
very high angles of attack the two attachment lines probably will merge into a single
attachment line at the center line of the wing, which can be considered as a reverse
bifurcation.
7.5 Structural Stability and Changes of Flow Fields 175

Fig. 7.19 Asymmetric flow past a slender delta wing at high angles of attack at M∞ = 2.8 [1]. At
the right interpreting sketches from D. J. Peake and M. Tobak of the vapor-screen photographs at
the left from [30]

Fig. 7.20 Sketch of the

skin-friction line pattern at
the lower side of a circular
forebody at angle of attack
[31]
176 7 Topology of Skin-Friction and Velocity Fields

Fig. 7.21 Sketch of the skin-friction line pattern at the flat lower side of a forebody at angle of
attack [31]

What we see in the lower part of Fig. 7.18 probably in reality happens only if
already an asymmetrical lee-side vortex arrangement is present.
All this will depend to a degree on the actual shape of the windward side, it
also holds for sharp leading edges. In the angle-of-attack range considered here,
the primary attachment line will never lie in the center line of the lower surface, as
indicated in Fig. 7.18. For all angles of attack we get the picture shown in Fig. 7.21.
Important is that between the two primary attachment lines the flow is more
or less two-dimensional, depending on the degree of flatness of the surface. The
two-dimensionality is a very welcome flow property, because it leads to a favorable
onset flow, for instance of that approaching an engine air inlet or an aerodynamic
control surface [31]. Precise angle of attack ranges as discussed with Fig. 10.15 on
Sect. 10.2.2 are not known, because class (2) has sofar found only little attention in
the field of flow topology.
• Concluding Remarks We have discussed structural stability and flow-field
changes by looking at the flow past a sphere and past a slender delta wing. We have
seen that a host of possible topological structures can evolve, depending on the range
of flow parameters and geometrical properties. However, we did hardly cover the
wide range of actual flow and geometrical parameters, which we may encounter in
real flight vehicle operation.
Our aim was to show qualitatively the high complexity, which can be present and
with which the flight vehicle designer is confronted. No general theory and no criteria
exist to predict what one gets topologically in reality. Of course, basic flow struc-
tures and their changes can be described with the help of topological considerations,
structural stability and bifurcation theory as is shown, for instance, very detailed in
a DLR report by U. Dallmann [4].
In design work and in problem diagnosis, however, one has to be aware of the
possibility of multiple phenomena, which have a more or less large influence on the
separated and vortical flow field past the flight vehicle and hence on the behavior
and the performance of the latter. It helps to study—actually that is necessary—
what was found in earlier ventures and in experimental and in theoretical/numerical
research, investigations and vehicle shape development. Even then surprises, either
unwelcome or welcome ones, cannot be avoided.
7.6 Problems 177

7.6 Problems

Problem 7.1 Open-type attachment and separation lines are present in some of the
figures in Sect. 8.4. Identify these figures and give short descriptions.

Problem 7.2 Which topological rules apply for the two flow fields sketched in
Fig. 7.2?

Problem 7.3 Consider the flow past the circular forebody of Fig. 7.20. If you look
at the cross section ahead of the inlet, what topological structure is present, if the
flow would be inviscid?

Problem 7.4 The eigenvalues of the matrix A, Eq. (7.6), are found with det (A −
λI), where I is the n × n identity matrix, and λ (n = 1, 2, 3) are the eigenvalues.
Derive the eigenvalues of the matrix A.

Problem 7.5 What is the prerequisite for plane-of-symmetry flow? Give a verbal
definition.

Problem 7.6 Assume a curved inviscid attachment line. Inflection points appear in
the approaching streamlines on one side of the attachment line. On what side do they
appear and why?

Problem 7.7 What are off-surface flow portraits? Give a verbal account.

Problem 7.8 In Sect. 7.3 several flow characteristics associated with singular lines,
i.e., attachment and separation lines, are discussed. Give a summary of these char-
acteristics.

Problem 7.9 How is separation defined in two-dimensional flow? What are the
indicators of separation in three-dimensional flow?

References

1. Peake, D.J., Tobak, M.: Three-dimensional interaction and vortical flows with emphasis on
high speeds. NASA TM 81169 (1980) and AGARDograph 252 (1980)
2. Moffat, H.K., Tsinober, A. (eds.): Topological Fluid Mechanics. Proceedings of the IUTAM
Symposium, Cambridge, GB, 1989. Cambridge University Press (1990)
3. Dallmann, U.: Topological structures of three-dimensional flow separations. DLR Rep. 221-82
A 07 (1983)
4. Dallmann, U.: On the formation of three-dimensional vortex flow structures. DLR Rep. 221-85
A 13 (1985)
5. Délery, J.: Three-Dimensional Separated Flow Topology. ISTE, London and Wiley, Hoboken
(2013)
6. Hirschel, E.H., Cousteix, J., Kordulla, W.: Three-Dimensional Attached Viscous Flow.
Springer, Berlin (2014)
178 7 Topology of Skin-Friction and Velocity Fields

7. Sears, W.R.: The boundary layer of yawed cylinders. J. Aeronat. Sci 15, 49–52 (1948)
8. Hirschel, E.H., Fornasier, L.: Flowfield and Vorticity Distribution Near Wing Trailing Edges.
AIAA-Paper 84–0421 (1984)
9. Schwamborn, D.: Boundary layers on finite wings and related bodies with consideration of the
attachment-line region. In: Viviand, H. (ed.) Proceedings of the 4th GAMM-Conference on
Numerical Methods in Fluid Mechanics, Paris, France, October 7 - 9, 1981. Notes on Numerical
Fluid Mechanics, vol. 5, pp. 291–300. Vieweg, Braunschweig Wiesbaden (1982)
10. Lighthill, M.J.: Attachment and separation in three-dimensional flow. In: Rosenhead, L. (ed.),
Laminar Boundary Layers, pp. 72–82. Oxford University Press, Oxford (1963)
11. Wang, K.C.: Boundary layer over a blunt body at high incidence with an open type of separation.
Proc. R. Soc., Lond. A 340, 33–55 (1974)
12. Lugt, H.J.: Introduction to Vortex Theory. Vortex Flow Press, Potomac (1996)
13. Kaplan, W.: Ordinary Differential Equations. Addison-Wesley Publishing Company, Reading
(1958)
14. Andronov, A.A., Leontovich, E.A., Gordon, I.I., Maier, A.G.: Qualitative Theory of Second-
Order Dynamic Systems. Wiley, New York (1973)
15. Hornung, H., Perry, A.E.: Some Aspects of Three-Dimensional Separation, Part I: Streamsur-
face Bifurcations. Z. Flugwiss. und Weltraumforsch. (ZFW) 8, 77–87 (1984)
16. Bakker, P.G., de Winkel, M.E.M.: On the Topology of Three-Dimensional Separated Flow
Structures and Local Solutions of the Navier-Stokes Equations. In: Moffat, H.K., Tsinober,
A. (eds.), Topological Fluid Mechanics. Proceedings of the IUTAM Symposium, Cambridge,
GB, 1989, pp. 384–394. Cambridge University Press, Cambridge (1990)
17. Tobak, M., Peake, D.J.: Topology of three-dimensional separated flows. Ann. Rev. Fluid Mech.,
Palo Alto 14, 61–85 (1982)
18. Oswatitsch, K.: Die Ablösebedingungen von Grenzschichten. In: H. Görtler (ed.), Proceedings
of the IUTAM Symposium on Boundary Layer Research, Freiburg, Germany, 1957, pp. 357–
367. Springer, Berlin (1958). Also: The Conditions for the Separation of Boundary Layers. In:
Schneider, W., Platzer, M. (eds.) Contributions to the Development of Gasdynamics, pp. 6–18.
Vieweg, Braunschweig Wiesbaden, Germany (1980)
19. Dallmann, U., Hilgenstock, A., Riedelbauch, S., Schulte-Werning, B., Vollmers, H.: On the
footprints of three-dimensional separated vortex flows around blunt bodies. In: Attempts of
Defining and Analyzing Complex Vortex Structures. AGARD-CP-494, 9-1–9-13 (1991)
20. Hirschel, E.H.: Basics of Aerothermodynamics. 2nd, revised edn. Springer, Cham (2015)
21. Hirschel, E.H.: Evaluation of Results of Boundary-Layer Calculations with Regard to Design
Aerodynamics. AGARD R-741, 5-1–5-29 (1986)
22. Davey, A.: Boundary-layer flow at a saddle point of attachment. J. Fluid Mech. 10, 593–610
(1961)
23. Hunt, J.C.R., Abell, C.J., Peterka, J.A., Woo, H.: Kinematical studies of the flows around
free or surface-mounted obstacles; applying topology to flow visualization. J. Fluid Mech. 86,
179–200 (1978)
24. Van Dyke, M.: An Album of Fluid Motion. The Parabolic Press, Stanford (1982)
25. Hirschel, E.H.: Viscous Effects. Space Course 1991, RWTH Aachen, Germany, 12-1 to 12-35
(1991)
26. Weiland, C.: Personal Communication (2017)
27. Schulte-Werning, B.: Numerische Simulation und topologische Analyse der abgelösten Strö-
mung an einer Kugel (Numerical Simulation and Topological Analysis of the Separated Flow
Past a Sphere). Doctoral Thesis, Technical University München, Germany (1990)
28. Bippes, H., Turk, M.: Oil flow patterns of separated flow on a hemisphere cylinder at incidence.
DLR Rep. 222-83 A 07 (1983)
29. Werlé, H.: Apercu sur les Possibilités Expérimentales du Tunnel Hydrodynamique a Visual-
ization de l’O.N.E.R.A. ONERA Tech. Note 48 (1958)
References 179

30. Fellows, K.A., Carter, E.C.: Results and Analysis of Pressure Measurements on Two Isolated
Slender Wings and Slender Wing-Body Configurations at Supersonic Speeds. Vol. 1, Analysis,
ARA Rep. 12 (1969)
31. Hirschel, E.H., Weiland C.: Selected Aerothermodynamic Design Problems of Hypersonic
Flight Vehicles. In: Progress in Astronautics and Aeronautics, AIAA, Reston, Va., vol. 229.
Springer, Heidelberg (2009)
Chapter 8
Large Aspect-Ratio Wing Flow

This chapter is devoted to the discussion of the flow—mainly the trailing vortex
layer and the pair of trailing vortices—past lifting large aspect-ratio wings in view
mostly of the results and insights gained in Chap. 4. Considered always is the clean-
wing situation, selected topics of the real-wing situation (high-lift system, integrated
engines etc.) are presented in Chap. 9.
We begin in the introduction with general considerations of wing-wake properties
as well as wing shapes and their influence on the wing flow.
Unit Problems are treated in the following three sections. In Sect. 8.2 it is shown
that—and how—the flow-vorticity properties are present in panel method solutions
(Model 4 in Table 1.3) of lifting wing flow. The same is made in Sect. 8.3 for Euler
solutions (Model 8). In Sect. 8.4 finally the flow past the Common Research Model
(CRM) is studied with the help of a RANS/URANS solution (Model 10). Besides
the wake flow and its vorticity properties also issues of the flow at the wing root and
at the wing tip are considered.

8.1 Introduction

8.1.1 Five Flow Domains at and Behind the Lifting Wing

The topic of the chapter is separated and vortical flow past clean lifting large aspect-
ratio wings.1 Generic wing forms and also the wing of a modern transonic transport
aircraft configuration are considered.
We first have a look at the separation and vortex phenomena being present over
and behind a real aircraft in cruise condition, Fig. 8.1. The aircraft is considered to be
in straight and level flight, in steady longitudinally stable motion and in a trimmed
state.

1 Regarding differentiation aspects of large and low aspect-ratio wings see Sect. 8.4.4.
© Springer-Verlag GmbH Germany, part of Springer Nature 2021 181
E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_8
182 8 Large Aspect-Ratio Wing Flow

Fig. 8.1 Schematic of the development of the wake system of a large aspect-ratio aircraft in cruise
flight [1]. Five Flow Domains: Domain 0: Wing Flow (not indicated), Domain 1: Near field, Domain
2: Extended near field, Domain 3: Mid field/Far field, Domain 4: Decay region

Four flow domains of the wake are indicated in the figure [1]. There are no sharp
boundaries between them. Depending on the view one takes considerable overlaps
are present. We aim for a qualitative description of the separated and vortical flow
over and behind the aircraft. We add a fifth domain, Domain 0, to also treat the flow
phenomena found over the wing, in particular at its trailing edge.
– Flow Domain 0 Wing: On the wing the upper and lower side flow-fields shear
is present. At the upper wing side the flow generally is directed slightly toward
the wing’s root, at the lower side toward the wing’s tip. Engine and flap track
installations locally influence the flow fields. Because the trailing edge is swept
backward, except maybe for a kink,2 the combined overall flow direction is out-
ward, resulting in a positive vortex-line angle (y), Fig. 4.11. The trailing-edge
flow shear angle ψe (y) increases from the wing’s root toward its tip. The resulting
near-wake structure, Fig. 4.12, however is somewhat distorted due to the actual
shape of the trailing edge, which generally has a finite thickness, Sect. 6.3.
The change of ψe (y) in y-direction is connected to the spanwise distribution of the
circulation Γ (y), reflected by the compatibility condition Eq. (4.20) in Sect. 4.4.
Accordingly the wake—the trailing vortex layer—carries kinematically active vor-
ticity Ωs , whose strength increases toward the wing tip. The kinematically inactive
vorticity Ωt locally reflects the chordwise boundary-layer properties.
The flow leaves the wing’s trailing edge in the form of flow-off separation. To be
added then are three further—ordinary—separation phenomena, Fig. 8.2.
Present in any case are the tip vortices which come into being by the flow around
the wing tips and the subsequent separation. Their sense of rotation is that of

2 This kink near the wing root is called a Yehudi break after its inventor at Boeing [2].
8.1 Introduction 183

Fig. 8.2 Schematic of ordinary and flow-off separation at the CRM configuration, [3], topic of
Sect. 8.4

the trailing vortices, into which they finally are merged. Usually at the wing tip
a secondary separation is present, which leads to an additional weak, but coun-
terrotating secondary vortex, tertiary vortices are possible, too. Hence we have a
tip-vortex system.
At engine pylons and nacelles separation phenomena may lead to discrete vortices,
which interact directly or indirectly with the near wake. No such vortices should
be due to the flap tracks.
At the wing roots finally horse-shoe vortices are present, if no proper wing-root
fairing is applied.3 In general a wing-root fairing is present, with different layout
at the upper and the lower side of the wing root.
– Flow Domain 1 Near field: At the beginning of Domain 1 the near wake is estab-
lished as a trailing vortex layer, containing kinematically active and inactive vortic-
ity. Domain 1 extends from the wing’s trailing edge downstream over a distance of
approximately the reference chord length of the wing lμ or the half span b/2. Often
it is considered to extend up to the beginning of the afterbody of the fuselages. The
roll-up process of the trailing vortex layer is in the beginning, which eventually
leads to the pair of counterrotating trailing vortices at the end of the domain, see
the schematic in Fig. 4.18. Besides the roll-up of the trailing vortex layer, other
effects occur, in particular the appearance of the wing-tip vortex system, as well
as possible wing-root vortices.
– Domain 2 Extended near field: Due to self-induction, Sect. 3.12.1, the trailing
vortex layer begins to curl up into the pair of counterrotating trailing vortices.
The tip-vortex system and possible other vortices are merged into the vortex layer

3 These vortices sometimes are called necklace vortices.

184 8 Large Aspect-Ratio Wing Flow

Fig. 8.3 The wake of an aircraft in flight [4]

during the curl-up process.

Because the aircraft flies in a trimmed state, a weaker trailing vortex layer with a
kinematically active vorticity of opposite sign leaves the horizontal stabilizer. It
also curls up into a pair of trailing vortices which each counter-rotate to the main
trailing vortex pair. They finally are merged into that pair, however with particular
mechanisms.
From the fuselage as well as from the vertical stabilizer mainly kinematically
inactive vorticity joins the aircraft’s wake. The influence of the propulsion jets
is more or less restricted to the extended near field and usually without larger
importance. Figure 8.3 gives an impression of the aircraft’s wake in the extended
near field.
– Flow Domain 3 Mid field/Far field: The far field sees the fully established trailing
vortex pair more or less unchanged for a distance of up to one hundred wing
spans. The vortex diameters increase slightly due to diffusion processes. The ratio
of vortex spacing to vortex length of course shrinks and the susceptibility increases
for reciprocal induction processes.
– Flow Domain 4 Decay region: In the decay region finally the vortex pair breaks
up due to the so-called Crow instability, Sect. 3.11.3. Atmospheric disturbances
are taken up, their intensities governing the extent of the decay process.
8.1 Introduction 185

8.1.2 Influence of the Trailing Vortex Layer and Vortices on

the Wing’s Performance

Of interest now is the influence of the trailing vortex layer/vortices on the wing’s lift
and drag. L. Prandtl described with his lifting-line model, Fig. 3.3, the situation in
the frame of what now is called circulation theory (Model 4) [5]. The basic result
is that the kinematically active vorticity in the trailing vortex layer, respectively the
trailing vortices, induces an extra wing drag, the induced drag.
Besides the skin-friction drag and the form drag this is the third part of the total
drag, which can amount to be the dominant part of it. In the transonic and the
supersonic regime the wave drag and the interference drag appear, Sect. 2.4.
We do not reproduce Prandtl’s theory, instead we point to the literature, e.g., [6].
We state the three important results for a finite-span lifting wing without aerodynamic
or geometrical twist at a subsonic flight speed such that the flow can be considered
as being incompressible:
– (1) If the wing has an elliptical planform, the circulation distribution Γ (y) is
elliptical, the induced drag Di is minimum,
– (2) If the circulation distribution is elliptical, the induced downwash velocity wi
at the wing is constant in span direction, as is the induced angle of attack αi .
– (3) If an elliptic circulation distribution is given, the induced drag coefficient is
inversely proportional to the aspect ratio Λ, Sect. 3.16:

C L2
C Di = . (8.1)
πΛ
With this result the amount of induced drag Di of two wings at equal lift L 2 =
L 1 and equal reference area Ar e f2 = Ar e f1 , but different aspect ratios Λ, can be
compared in the frame of potential theory (Model 4). Hence neglecting viscous
forces and compressibility effects we obtain in terms of the force coefficients
 
C2 1 1
C Di,2 − C Di,1 = L − . (8.2)
π Λ2 Λ1

Aircraft wing design, also in the transonic flight regime, always makes use of these
insights, however, not by following them exactly. Only a few aircraft actually had or
have a wing with elliptical planform. A wing with such a planform is more expensive
to manufacture than a wing with straight leading and trailing edge. Examples were
seen, where the wing originally had an elliptical planform, which was changed to a
tapered planform, once quantity production was demanded.
An approximation to an elliptical planform to a degree results from wing tapering.
The taper ratio of a wing is λt = ct /cr , where ct is the wing-tip chord length, and cr
that of the wing’s root.
The aspect ratio Λ—in the literature also denoted with A R—always is an impor-
tant parameter, because it governs the lift-to-drag ratio L/D—the aerodynamic effi-
186 8 Large Aspect-Ratio Wing Flow

ciency of the wing—and hence the flight performance. Values range from Λ ≈ 6 for
small general aviation aircraft and up to Λ ≈ 50 for high-performance gliders.
In general the aspect ratio should be as large as possible. Limiting are aerodynamic
shape demands, structural issues, weight, maneuverability, sub-system issues, and
airfield restrictions.
Special aerodynamic shape demands arise beyond the subsonic flight regime.
Regarding transonic aircraft we have seen in Sect. 2.4 that due to drag issues either a
swept wing or a thin wing must be employed. The thin wing concept is not a choice
for transonic transport aircraft, because of structural and fuel carrying demands. The
rule now is the backward swept wing with supercritical chord sections, although in
principle the wing also can be swept forward. Modern backward swept-wing aircraft,
for example the Airbus A350 and the Boeing 787 have an aspect ratio Λ = 9.5.
In the supersonic flight regime, Chap. 10, we generally find aircraft with small
aspect-ratio wings, for instance the unswept but tapered thin-wing F-104 Starfighter
with Λ = 2.45, the delta wing (ogive wing) Concorde with Λ = 1.55, the Eurofighter
Typhoon fighter aircraft with Λ = 2.205, the F-35A with Λ = 2.663, and in the
hypersonic flight regime the Space Shuttle Orbiter with a strake-delta planform and
Λ = 2.265.
Wing design still strives for a spanwise elliptical circulation distribution. The
circulation around a wing chord section of course is not the line integral of the
velocity along a streamline. The inviscid streamlines already at an elliptic wing do
not lie in chord direction, see, e.g., Fig. 7.1. Instead of the span-wise distribution of
the circulation usually that of the equivalent local lift coefficient Cl is considered,
Sect. 3.16.
The desired local lift distribution is found—for the given design lift of the wing—
with the help of changes of the wing’s taper, the (static) geometrical twist distribution,
and the camber distribution.4 Also playing a role is the aerodynamic twist due to the
elastic deformation of the wing under load, in particular if the wing is a swept wing,
see, e.g. [7].
In the frame of this book we cannot go into the details of wing design. We point to
the publications of D. Küchemann 1978/2012 [8], H. Schlichting and E. Truckenbrodt
1959/1979 [9], E. Obert 2009 [2], R. Rudnik in the Handbuch of Luftfahrzeugtechnik
(in German) 2014 [10], R. Vos and S. Farokhi 2015 [7], and of A. Rizzi and J.
Oppelstrup 2020 [11].
In the following sections we demonstrate with a few Unit Problems the capabilities
of three kinds of computational flow models of Table 1.3. In Sect. 8.2 results of several
panel methods (Model 4) are checked with flow computations past lifting wings
(Flow Domain 0). It is shown that the overall results are in reasonable agreement.
Regarding the flow fields near the trailing edges of the wings, however, most of the
methods yield improper results. Only a higher-order formulation gives the proper
result, fulfilling at the trailing edge the compatibility condition, Eq. (4.20).

4 Thedecades old concept of in-flight camber changes now seems to approach reality thanks to the
potential of carbon-fiber structures.
8.1 Introduction 187

An Euler solution is applied to a large aspect-ratio wing in Sect. 8.3 (Flow Domains
0 and 1). Demonstrated is that the near-wake properties are computed in a proper way
with a Model 8 method. At the trailing edge the compatibility condition is fulfilled.
In Sect. 8.4 finally the flow past the wing of the Common Research Model (CRM)
is investigated (Flow Domains 0, 1 and 2). Applied was a Model 10 method, i.e., a
method with the highest modeling level. All the considered lifting large aspect-ratio
wings have in common a trailing vortex layer, which originates at the trailing edge
of the wing. Tip-vortex systems play a certain role, too. This is shown only for the
CRM case. There also happens an interaction with the trailing vortex layer of the
horizontal tail plane.
Our aim is to provide and foster the understanding of the basic properties of
the trailing vortex systems and their behavior. We further want to demonstrate the
capabilities of numerical simulation methods of such flows and the interpretation
of the results from a fluid-mechanical point of view. Generally we do not deal with
detailed configurational issues of large aspect-ratio aircraft. In the following chapter,
Chap. 9, we sketch some practical problems.

8.2 Panel Method (Model 4) Solutions—Proper and

Improper Results in Flow Domain 0

This section is devoted to a short presentation of results of potential-flow theory

for two different lifting wings in Flow Domain 0. Results are shown regarding the
trailing-edge flow shear angle ψeu , the kinematically active vorticity content Ωs , the
vortex-line angle , the circulation Γ and its derivative in spanwise direction dΓ /dy,
as well as skin-friction line patterns of boundary-layers, which for one of the wings
were obtained with boundary-layer computations.
We discuss examples of flows found with panel methods, i.e., Model 4 methods.
The wings are the Kolbe wing, [12], and a forward swept wing, which was studied
in [13].
The results were first published in [14]. All figures in this section are from that
paper. In [14] also properties of panel methods were treated, which are of importance
when considering Model 4 methods. In the following figures we give references to
panel method 1 and to other panel methods (2, 3, etc.).
Panel method 1 is the HISSS method of L. Fornasier [15].5 This higher-order
panel method has linear source distributions and quadratic doublet distributions in
both chord-wise and span-wise directions. We will see that only this method yields
proper trailing-edge flow results.
The other methods are lower-order methods. They have stepwise constant dou-
blet distributions—or equivalent—in both chord- and span-wise directions. Such
approaches lead to an erroneous determination of the v-component of the velocity
vector near the wing’s trailing edge. A similar problem was reported in [16], where

5 HISSS method = Higher-Order Subsonic-Supersonic Singularity method.

188 8 Large Aspect-Ratio Wing Flow

Table 8.1 Geometrical and flow parameters of the Kolbe wing example [14]
ϕ0 [◦ ] Λ λt M∞ α [◦ ] Recm cm [m] Boundary-layer
state
45 3 0.5 0.25 8.5 18·106 1 Fully turbulent

Fig. 8.4 Planform of the

Kolbe wing [14]

the failure of such low-order panel methods to compute the flow past thin wings was
attributed to an inadequate doublet distribution in the chord-wise direction. Earlier
boundary-layer studies based on such improperly computed inviscid flow fields even
led authors come to the conclusion that the boundary-layer flow over swept wings is
predominantly two-dimensional. This of course is not true.

8.2.1 The Kolbe Wing

The Kolbe wing is a swept wing with a large sweep angle ϕ0 of the leading edge [12].
This angle, the aspect ratio Λ, the taper ratio λt , the free-stream Mach number M∞ ,
the angle of attack α, the Reynolds number Recm , the reference (mean) chord length
cm , and the state of the boundary layer regarding the results of a boundary-layer
computation are given in Table 8.1. The planform of the wing is shown in Fig. 8.4.
The chord section normal to the leading edge is a NACA 64-010 airfoil.6
Consider now Fig. 8.5. It shows the dimensionless span-wise circulation distri-
bution Γ /(cm u ∞ ), its derivative d Γ /d y, and the distribution of the kinematically
active vorticity content Ω ≡ Ωs , all computed with the panel methods 1 and 2.

6 The
large leading-edge sweep of the wing combined with a large angle of attack—see below—
makes it necessary to investigate whether leading-edge separation can be present.
8.2 Panel Method (Model 4) Solutions—Proper and Improper Results in Flow Domain 0 189

Fig. 8.5 Kolbe wing [14]: comparison of the results of panel method 1 and 2: circulation distribution
Γ , its derivative d Γ /d y, and the vorticity content Ω = Ωs /u ∞ as functions of the half-span
coordinate 2y/b

The Γ (y) results of method 1 and the lower-order method 2 do not differ much,
likewise the results regarding lift and induced drag (not shown). The point-wise inves-
tigation of the velocity components at the trailing edge with Eq. (4.20) yields values
of Ω (=Ωs /u ∞ ), which for the higher-order method 1 agree very well with d Γ /d y,
which means that the compatibility condition, Sect. 4.4, is fulfilled. However, for
method 2 with Ω ≈ 0 they are wrong throughout.
The trailing-edge flow shear angle ψ, given in Fig. 8.6 reflects this result. Panel
method 1 shows the expected rise of ψl (=ψe /2) toward the wing tip, whereas method
2 yields a nearly zero shear angle.7 In contrast to that, the vortex-line angle ε is the
same for both methods. The vortex line for the back-swept trailing edge is deflected
in wing-tip direction by ε ≈ 5◦ . At the root and at the tip ε is approximately zero.
Regarding the matter of properly or improperly computed inviscid velocity fields
and the boundary-layer computations based on them, we study Fig. 8.7. Results
of three-dimensional boundary-layer computations are shown, performed with the
external inviscid flow fields found with panel method 1 and 2. The boundary-layer
method was the integral method of Cousteix-Aupoix in the MBB version [17].
We look at the streamlines of the external inviscid flow and the skin-friction lines
of the fully turbulent boundary layer at both the suction (upper) and the pressure
(lower) side of the wing. Upstream of and at the trailing edge on both sides the
inviscid streamlines found with method 1 show directions according to the vortex-
line angle ε and the shear angle ψl in Fig. 8.6. The inviscid streamlines found with
panel method 2, however, approach the trailing edge nearly in chord direction. This
holds in particular for the pressure side.

7 For the definition of the trailing-edge flow shear angle ψ and the vortex-line angle ε see Fig. 4.11.
e
190 8 Large Aspect-Ratio Wing Flow

Fig. 8.6 Kolbe wing [14]: comparison of the results of panel methods 1 and 2: trailing-edge flow
shear angle ψl and vortex-line angle ε as functions of the half-span coordinate 2y/b

Note that only method 1 yields the points of inflection of the inviscid streamlines
near approximately 75 per cent chord length at almost the whole pressure side, Fig.
8.7b. The error in the inviscid flow field and therefore in the boundary-layer solution
found with method 2 extends over approximately 50 per cent chord length upstream
of the trailing edge of both the suction and the pressure side of the wing.

8.2.2 The Forward-Swept Wing

The application of the higher-order panel method 1 and of some lower-order panel
methods to a forward-swept wing, which was studied in [13], gives a similar result
as above. The leading edge sweep angle ϕ0 of the wing, the aspect ratio Λ, the taper
ratio λt , the free-stream Mach number M∞ (incompressible flow), and the angle of
attack α are given in Table 8.2. The planform of the wing is given in Fig. 8.8.
All methods agree rather well regarding the span-wise circulation distribution
Γ (y). However, only for method 1 the derivative d Γ /d y(y) and the vorticity content
Ωs (y) are compatible, Fig. 8.9.
The trailing-edge flow shear angle ψl (= ψ/2) in Fig. 8.10 again shows the expected
trend only for method 1.
The vortex-line angle is negative with ε ≈ −7 ◦ , i.e., the vortex lines are deflected
in wing-root direction. Actually it is the whole flow field at both sides of the wing
which is deflected toward the wing’s root.8 The results of the lower-order methods

8 Thisdeflection is the reason for the well observed unwelcome accumulation of boundary-layer
material at the wing root (and the fuselage) of forward swept wings. This can lead to adverse
separation phenomena and, with rear-mounted engines at the aft end of the fuselage, makes special
measures necessary regarding the position of the engines. The general deflection of course is also
a property of the trailing vortex layer which leaves the wing’s trailing edge.
8.2 Panel Method (Model 4) Solutions—Proper and Improper Results in Flow Domain 0 191

Fig. 8.7 Kolbe wing: comparison of the results of panel methods 1 and 2 and a three-dimensional
boundary-layer method [14]. Streamlines of the external inviscid flow and skin-friction lines of the
turbulent boundary layer: a suction side, b pressure side of the wing
192 8 Large Aspect-Ratio Wing Flow

Table 8.2 Geometrical and flow parameters of the forward-swept wing example [14]
ϕ0 [◦ ] Λ λt M∞ α [◦ ]
−35.1 4.5 0.35 0 4

Fig. 8.8 Planform of the forward swept wing [14]

Fig. 8.9 Forward swept wing: comparison of results of several panel methods [14]. Circulation
distribution Γ , its derivative d Γ /d y, and the vorticity content Ω as functions of the half-span
coordinate 2y/b

show much scatter, depending on where the kinematic flow condition is implemented,
either on the skeleton plane of the wing, or on the true wing surface.
8.3 Creation of Lift in an Euler Solution (Model 8) for a Lifting … 193

Fig. 8.10 Forward swept wing [14]: comparison of results of several panel methods. trailing-edge
flow shear angle ψl and vortex-line angle ε as functions of the half-span coordinate 2y/b

Fig. 8.11 Planform of the considered wing [18]

8.3 Creation of Lift in an Euler Solution (Model 8) for a

Lifting Large-Aspect Ratio Wing—Proof of Concept in
Flow Domain 0 and 1

The considered large, rather medium aspect-ratio wing is a trapezoidal wing with
small leading-edge sweep ϕ0 = 25◦ , Fig. 8.11. The results were presented first at the
Symposium on the International Vortex-Flow Experiment on Euler Code (Model 8 of
Table 1.3) Validation in 1986, FFA Bromma, Sweden [18]. All figures in this section
are from that presentation. The results prove the viability of Model 8 solutions for
lifting high aspect-ratio wings.
194 8 Large Aspect-Ratio Wing Flow

Table 8.3 Geometrical and flow parameters of the trapezoidal wing example [18]
ϕ0 [◦ ] Λ λt M∞ α [◦ ]
25 3.75 0.3 0.3 5

Table 8.4 Computed force and moment coefficients of the trapezoidal wing example [18]
α = 5◦ Euler solution thick wing, Linear theory thin wing,
M∞ = 0.3 M∞ = 0.01
CL 0.342 0.315
C Di 0.0103 0.0085
CM −0.155 −0.187

8.3.1 The Computation Case and Integral Results

The leading edge sweep angle ϕ0 of the wing, the aspect ratio Λ, the taper ratio
λt , the free-stream Mach number M∞ (weakly compressible flow), and the angle of
attack α are given in Table 8.3. The wing has 12 per cent thickness, the chord-wise
section is the NACA 64-012 airfoil. The computation was made with the Euler code
described in [19].
The computed lift (C L ), induced drag (C Di ) and pitching moment (C M ) coeffi-
cients are given in Table 8.4. They are compared with the results of a linear method
(Model 4). Because that method did not take into account the wing’s thickness and
moreover was made for M∞ = 0.01, the lift and induced drag results are about 10
per cent lower. This is the right order of magnitude. The results further give proof
that Euler methods (Model 8) can be viable design tools.

8.3.2 Details of the Computed Flow Field of Domain 0 and 1

In Fig.8.12 the computed inviscid streamlines on the surface of the wing and in the
near wake are given. The streamlines at the wing surface show well the expected
pattern, which was sketched in Fig. 4.17. The trailing-edge flow shear angle ψe
increases from nearly zero close to the symmetry plane of the wing to about 25◦ at
about 90 per cent half span. This behavior is reflected by the distribution of Ωs and
dΓ /dy in Fig. 8.19.
The vortex-line angle  is nearly zero, because the trailing edge is unswept. Down-
stream of the trailing edge the streamlines lie in the skeletal plane of the wake, hence
no shear is discernible there. A roll-up tendency of the trailing-edge vortex layer is
not discernible, either. At the wing’s tip two streamlines indicate the flow around the
edge from the lower to the upper side. Whether the Euler solution here approximates
8.3 Creation of Lift in an Euler Solution (Model 8) for a Lifting … 195

Fig. 8.12 Right-hand side of the wing and the near wake with the computed inviscid surface
streamlines (view from above) [18]

Fig. 8.13 Plot of the flow vectors over the rear part of the wing and in the near wake at 40 per cent
half-span [18]

a tip vortex, cannot be decided. At the trailing edge the said streamlines lie well away
from the tip region.
In Sect. 5.1 it was argued regarding the wing’s wake—and this is our major point—
that the discontinuity layer of potential theory is widened up in the Euler solution
due to the numerical diffusive transport. The Euler wake hence represents the trailing
vortex layer of the wing. This Euler wake is sketched in the right part of Fig. 5.1.
Regarding the Euler boundary layer and the Euler wake in the present solution we
look at Fig. 8.13, which at 40 per cent half-span shows a velocity-vector plot. On the
wing’s surface the Euler boundary layer is barely discernible. (A computation with
a coarser grid, however, had a pronounced Euler boundary layer.) The kinematically
inactive wake part of the Euler wake is also non-existent, as is demanded, Fig. 5.1.
The kinematically active part of the Euler wake at x/c = 1.05 behind the wing—
i.e. at five per cent chord behind the unswept trailing edge—extends over roughly
four cells in z-direction, Fig. 8.14. The details of the wake are unclear. Anyway, in
terms of the velocity profiles the demanded features of the Euler wake are present.
The Euler wake is more pronounced in terms of the total-pressure loss, as seen
in the following Fig. 8.15. Shown are the profiles 1 − pt / pt∞ as functions of z
196 8 Large Aspect-Ratio Wing Flow

Fig. 8.14 Cross-flow vector plot in the plane x/c = 1.05 behind the right half of the wing (c is the
mid-wing chord length) [18]

Fig. 8.15 Distribution of the total pressure loss 1 − pt / pt∞ (z) at four span stations downstream
of the trailing edge of the wing at x/c = 1.05 [18]

above and below the trace of the chord plane at the four stations 0.25, 0.5, 0.75,
and 0.9 half-span of the right-hand side of the wing at the mid-wing chord location
x/c = 1.05 behind the trailing edge. Unexplained is the apparent total-pressure gain
present mainly below the Euler wake at the three inner stations. At the outer station
(0.9 half-span) obviously grid-resolution problems are present, causing a significant
total-pressure loss outside the wake.
In the following three figures the dimensionless velocity components u(z), v(z),
and w(z) are given at the four span stations, again at x/c = 1.05. The components
u(z) and v(z) are not the components shown in Fig. 5.1 in the local wake coordinate
system. The vortex-line angle  is almost zero at all span stations, therefore for
convenience the velocity components are approximated by the components found in
the Cartesian x, y, z-reference coordinate system shown in the lower left part of Fig.
4.17.
In Fig. 8.16 the kinematically inactive part of the Euler wake u(z) indeed is nearly
uniform.
In Fig. 8.17 the shear of the flow between the upper and the lower side of the
Euler wake is well discernible. It increases from the wing’s root toward the wing tip
according to the increase of the trailing-edge flow shear angle ψe seen in Fig. 8.12.
The asymmetry of the kinematically active v-profiles at the inner stations is due to
the small negative vortex-line angle . Again grid-resolution problems are visible.
8.3 Creation of Lift in an Euler Solution (Model 8) for a Lifting … 197

Fig. 8.16 Distribution of the velocity component u(z) at four span stations downstream of the
trailing edge of the wing at x/c = 1.05 [18]

Fig. 8.17 Distribution of the velocity component v(z) at four span stations downstream of the
trailing edge of the wing at x/c = 1.05 [18]

Fig. 8.18 Distribution of the velocity component w(z) at four span stations downstream of the
trailing edge of the wing at x/c = 1.05 [18]

The w-components are stronger positive below the Euler wake than above, Fig.
8.18. At the location of the Euler wake they are nearly zero, as is to be expected.
198 8 Large Aspect-Ratio Wing Flow

Fig. 8.19 Spanwise circulation distribution Γ of the Euler solution and of linear theory, the span-
wise derivative dΓ /dy, and the local vorticity content Ωs [18]

8.3.3 The Circulation and the Kinematically Active Vorticity

Content

Figure 8.19 shows the distribution of the circulation Γ in spanwise direction found
with the Euler solution.9 It is compared with the result of the linear Model 4 method.
In Fig. 8.19 also the compatibility condition, Sect. 4.4, is evaluated. The spanwise
derivative dΓ /dy and the local kinematically active vorticity content Ωs agree quite
well with each other from approximately 30 per cent span up to approximately 95
per cent span. Hence in that span interval the compatibility condition, Eq. (4.20), is
very well fulfilled. In the wing’s mid-section that is not the case, like near the wing
tip, where the flow around it disturbs the picture.
In Fig. 8.20 finally the magnitude of the vorticity ω(z) is plotted at the four span-
wise stations. Actually it is the vorticity component ωx connected to the kinematically
active part of the Euler wake. The structure of its distribution is in accordance with
the hypothesis put forward here.

9 Inthe mid-section of the wing, the Euler solution shows an irregularity, which seems to be due to
the sharp apex of the wing.
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 199

Fig. 8.20 Distribution of the magnitude of the vorticity ω(z) (=ωx ) at four span stations downstream
of the trailing edge of the wing at x/c = 1.05 [18]

8.4 RANS/URANS Solution (Model 10) for the CRM Case:

Flow Domains 0, 1, and 2

8.4.1 Introduction

The trailing vortex layer and the wake-vortex system of a transport aircraft configu-
ration are investigated in the Flow Domains 0, 1 and 2. The section’s content is based
on the master thesis of S. Pfnür, [3], a student of the third author of this book. Most
of the figures are from his thesis.
The considered aircraft configuration is the NASA/Boeing Common Research
Model (CRM), [20], which serves as the test configuration of the AIAA CFD Drag
Prediction Workshop series, see, e.g., [21, 22].
The CRM shape is a generic wing/body/horizontal tail configuration with tran-
sonic supercritical wing design, Fig. 8.2. The airfoil sections and the twist of the
wing correspond to the 1-g wing shape at cruise condition.10
The CRM shape was designed for CFD validation purposes, hence no propulsion
units were added. The wing, moreover, has a constant trailing-edge thickness over the
whole span. The clean wing is attached without a wing-fuselage fairing. Its trailing
edge shows a kink, the Yehudi break, mentioned already on Sect. 8.1. The CRM
planform is given in Fig. 8.21.
Indicated is the wing-tip reference point (WRP) and the x ∗ , y ∗ , z ∗ - reference
coordinate system, with z ∗ being normal to the x ∗ –y ∗ plane. The WRP-location is
x ∗ = x/b = 0, y ∗ = y/(b/2) = 1, z ∗ = z/(b/2) = 0.
The geometrical parameters of the CRM computation case are given in Table 8.5.
Note that the reference area Ar e f is the Wimpress area, which Boeing introduced,
Sect. 3.16.
The flight parameters, Table 8.6, correspond to that of “Case 1b” of the workshop,
with the setting angle of the horizontal tail plane α H T P = 0◦ . Despite α H T P = 0◦ ,
due to the downwash induced by the lifting wing, a negative vertical force is present

10 The 1-g shape is the shape of the elastic wing due to the aerodynamic load at nominal level flight,

see, e.g., [7].

200 8 Large Aspect-Ratio Wing Flow

Fig. 8.21 Planform of the Common Research Model [3]. WRP is the wing-tip reference point
(x ∗ = 0, y ∗ = 1, z ∗ = 0)

Table 8.5 Geometrical parameters of the CRM computation case [3]

ϕ L E [◦ ] ϕ25 [◦ ] λt b [m] lμ [m] Λ Ar e f (m2 )
37.5 35 0.275 58.763 7.00532 9 383.69

Table 8.6 Flight parameters of the CRM computation case [3]

M∞ Relμ T∞ [K] Tw [K] α [◦ ] β [◦ ] Viscous flow
portions
0.85 5·106 310.93 310.93 2 0 Fully turbulent

on the horizontal tail plane, like a trim force. Accordingly the horizontal tail plane
first has a trailing vortex layer and then a trailing vortex pair.
The force and moment coefficients for the longitudinal motion of this case were
found to be C L = 0.43, C D = 0.0245, and C M = 6.8 × 10−5 , with the reference
point at 33.67 m from the fuselage nose.

8.4.2 Computation Method and Grid Properties

The computation method employed was the Reynolds-averaged Navier-Stokes

(RANS) method TAU (Model 10), the code for unstructured grids developed at
the DLR [23]. The code as unsteady RANS (URANS) code also permits an unsteady
computation approach.
Two turbulence models were employed, the Spalart-Allmaras S A/k − ω model
(original publication [24]) in the S A − neg version, and the Menter − SST model
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 201

Fig. 8.22 The inserted PS grid at x ∗ = 4 [3]. Indicated are the spanwise and vertical numbers of
elements

(original publication [25], SST stands for Shear-Stress Transport). Initial turbulence
model and grid studies were made with both turbulence models in the RANS mode,
the final investigations with the Menter − SST model in the URANS mode. This
was made because of a certain unsteady solution behavior.
Of course all results obtained in domains of strong ordinary separation must be
seen with some reservations. This more so because in some cases a strong dependence
on local discretization properties was observed.
The basic grid is an unstructured one, determined following the gridding guideline
given in [20]. After some adaptations the final unstructured grid had 30.17·106 nodes
and 132.3·106 elements.
Despite this final resolution it was necessary to insert a partly structured grid (PS
grid) in order to correctly capture the wake-flow properties. It extended from the mid
wing chord location downstream to x ∗ = 9. Again some adaptations were made. The
final medium-sized PS grid had 45.2·106 nodes and 114.2·106 elements. Figure 8.22
shows the element distribution of the PS grid at the cross-section x ∗ = 4, a location
approximately halfway along the extended near field.
We show a selection of results, highlighting separation and vortex phenomena.
We discuss these results in much detail in order to cover all aspects of separated and
vortical flow over the CRM wing. A detailed study of aspects of the skin-friction line
topology of this case can be found in [26].
With reference to the introduction, Sect. 8.1 and Fig. 8.1, we consider the flow
features in the Domains 0, 1 and 2, i.e. from over the wing to the extended near field.
202 8 Large Aspect-Ratio Wing Flow

Fig. 8.23 Schematic of a

horse-shoe vortex at the
wing/fuselage intersection
[10]

8.4.3 Flow Domain 0: The Flow over the Wing,

We first study the phenomena at the wing’s root, then at the wing’s tip and finally at
its trailing edge.
• Wing Root If a wing is attached to the fuselage without a wing-root fairing, the
flow separates ahead of the wing’s leading edge and a horse-shoe vortex is created,
Fig. 8.23.11
This vortex is the cause of a drag increment and further it can lead to buffet at
high angles of attack. Work in the 1930s by Th. von Kármán and co-workers led to
the introduction of smooth wing-root fairings (fillets) [28]. Such fairings effectively
eliminate these horse-shoe vortices. Wing-root fairings are the rule today for large
transport airplanes. Farther below we give an example.
At the CRM configuration no root fairing was introduced. The root area hence
permits to study separation and vortex phenomena.
Figure 8.24 gives a view toward the wing-root area from above.
In the left part we see the large pressure coefficient c p at the wing’s leading edge,
which leads to the root separation. Over the wing the strong flow acceleration in
chord direction reduces the static pressure (blue color), which is impressed also on
the side of the fuselage. The cross-section extent of the horse-shoe vortex is small.
The absolute value of the skin-friction coefficient |c f | in the right part of the
figure indicates separation at the fuselage side ahead of the wing’s root area. Along
the leading edge the strong flow acceleration in chord direction is reflected by high
skin friction. Toward the trailing edge we see a reduction of the absolute value of
skin friction.
Figure 8.25 shows the view toward the wing-root area from below. On the left
side the c p -distribution indicates a much weaker flow acceleration in chord direction
compared to that at the upper side of the wing. The horse-shoe vortex now has a
much larger spanwise extent than at the upper side of the wing. This in particular
holds toward and in the vicinity of the trailing edge.
The amount of the skin-friction coefficient |c f | shows a large extent of the sepa-
ration area below the wing’s root. At the fuselage generally the skin friction has low
values.

11 In [27] this situation is called flow obstacle, although there the swept obstacle is not a topic.
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 203

Fig. 8.24 Wing-root area from above (suction side of the wing) [3]. Left part: pattern of the
skin-friction lines and distribution of the surface-pressure coefficient c p . Right part: pattern of the
skin-friction lines and distribution of the absolute value of the skin-friction coefficient c f ≡ |c f |

Fig. 8.25 Wing root area from below (pressure side of the wing) [3]. Left part: pattern of the
skin-friction lines and distribution of the surface-pressure coefficient c p . Right part: pattern of the
skin-friction lines and distribution of the absolute value of the skin-friction coefficient c f ≡ |c f |

A closer look at what happens at the wing’s leading edge in the vicinity of the
intersection with the fuselage is given in Fig. 8.26. The flow arriving from the front
part of the fuselage is forced—due to the presence of the wing root—to separate. We
observe then a succession of singular points and associated singular lines.
The appearance of the singular lines happens first at the saddle point S1 , from
which to the left (lower wing side) and to the right (upper wing side) the primary
separation line originates. Next then lies the nodal point N1 . Originating from it we
see to the left (lower wing side) an attachment line (secondary attachment line), which
tapers off after a short distance (open-type ending). To the right the attachment line
is not well developed and is shortly squeezed between the primary and the secondary
separation lines.
204 8 Large Aspect-Ratio Wing Flow

Fig. 8.26 Detail wing root topology [3]. Upper part: pattern of the skin-friction lines and distribution
of the surface-pressure coefficient c p . Lower part: pattern of the skin-friction lines and distribution
of the absolute value of the skin-friction coefficient c f ≡ |c f |
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 205

Fig. 8.27 LTA leading edge with root fairing [26]. Left part: pattern of skin-friction lines and
distribution of the surface-pressure coefficient c p . Right part: pattern of the skin-friction lines and
distribution of the absolute value of the skin-friction coefficient |c f |

Next comes the saddle point S2 . From it a secondary separation line originates.
To the left it is well visible until further away below the wing, where it becomes
indistinguishable from the primary separation line, Fig. 8.25. To the right we have
a similar picture. The secondary separation line above the wing seems to join the
primary separation line, too, Fig. 8.24.
Finally we come to the nodal point N2 , which represents a nice star nodal point.
In the picture to the left from it the attachment line at the wing’s leading edge is well
discernible. Along it the skin friction clearly has a finite value, as is to be expected.
Its amount also is constant or nearly constant in spanwise direction up to the wing
tip.
Oddly enough, below N2 to the left an open-type attachment line appears to be
present. Probably only the skin-friction lines concentrate along the cutting curve of
the wing surface with the fuselage.
The pattern of singular points at the wing root discussed in [26] is much simpler.
Only the saddle point S1 is present and the nodal point N2 . We note that the simulations
there were made with a not—further specified—SA-turbulence model. Pfnür in his
grid refinement studies found a similar pattern for a coarse PS grid. That result was
obtained with the S A − neg turbulence model [3]. Anyway, the different results show
that here a problem is hidden. Experimental data, which could clear up the situation,
are not available. Even then one would have to be careful to draw conclusions, because
it is a matter of the range of similarity parameters, which also have an influence.
All these issues disappear, if a proper wing-root fairing is employed, Fig. 8.27.
The figure, taken from [26], shows for a large transport aircraft (LTA) with similar
geometrical and flow parameters a smooth transition from the flow past the fuselage
to that at the wing’s leading edge. The wing’s attachment line begins as open-type
attachment line!
Back to the present CRM case: in Fig. 8.28 the dimensionless axial vorticity
component
ωx b
ζ= (8.3)
2u ∞
206 8 Large Aspect-Ratio Wing Flow

Fig. 8.28 Dimensionless axial vorticity component ζ around the wing/fuselage intersection [3].
Left part: leading-edge region. Right part: upper trailing-edge region

is plotted at several downstream locations in the wing-root domain. The left part
shows in the leading-edge area at the suction (upper) side of the wing that the vortex
apparently has a rather small cross-section. The blue color indicates that it rotates in
clockwise direction. There is no evidence of a secondary vortex.
On the pressure (lower) side we have counter-clockwise rotation. The cross-
section is larger and the secondary vortex is clearly indicated.
In the trailing-edge region of the suction side we have no indication of the vortex,
except for the last stations, where separation is indicated, right part of Fig. 8.28. On
both the wing and the fuselage surface the boundary layer thickening in streamwise
direction is visible.
At the upper side of the wing near the trailing edge the pattern of the skin-friction
lines shows a larger separation event in form of a focus point F, Fig. 8.29.
In this focus point a vortex filament leaves the surface, Sect. 7.2.2. It is to be
expected that at the side of the fuselage a counter-rotating vortex filament leaves the
surface. That is well indicated by the skin-friction line pattern there. The experimental
investigations did not indicate either phenomenon.
• Wing Tip At the well rounded wing tip of the CRM configuration the pattern
of the skin-friction lines indicates a smooth flow around it from the lower side to the
upper side of the wing, Fig. 8.30.
The attachment line’s end is of open-type, Fig. 8.31. The line simply fans out.
At the left side of the figure we observe the expected pressure distribution: relative
pressure maximum along the attachment line, Sect. 7.3, fast expansion in chordwise
direction on the suction side, less strong expansion on the pressure side, all with only
small or even zero gradients along the leading edge. At the outer end of the leading
edge, however, we too see a strong expansion toward the outer side of the wing tip.
The skin-friction line pattern, right part of the figure, shows the relative minimum
of |c f | along the attachment line. It lies very close to the attachment line, because
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 207

Fig. 8.29 Pattern of the

skin-friction lines at the
upper side of the
trailing-edge area of the
wing with a focus point
separation [3]. Such a point
also is present at the side of
the fuselage

Fig. 8.30 Wing tip seen

from below: pattern of the
skin-friction lines and
distribution of the
surface-pressure coefficient
c p [3]

all gradients in x-direction are small. Toward the upper side of the wing and to the
tip we see maxima of |c f |, which are due to the flow acceleration there.
Figure 8.32 illustrates the flow around the wing tip to the upper side (suction side)
of the wing.
The pressure first decreases around the wing tip, accelerating the flow toward the
upper wing side. There a general pressure rise toward the trailing edge is present.
The result is that downstream of mid-chord an open-type separation line begins to
form along the upper side of the wing tip.
The round wing tip becomes sharp toward the wing’s trailing edge, which, how-
ever, has a finite thickness. The sharp-edged part locally causes a strong drop of the
surface pressure (blue color) ahead of and at the sharp tip portion, Fig. 8.33, upper
208 8 Large Aspect-Ratio Wing Flow

Fig. 8.31 End of the attachment line at the wing tip without singular point (open-type) [3]. Left
part: pattern of the skin-friction lines and distribution of the surface-pressure coefficient c p . Right
part: pattern of the skin-friction lines and distribution of the absolute value of the skin-friction
coefficient c f ≡ |c f |

Fig. 8.32 The wing tip seen

from above: pattern of the
skin-friction and distribution
of the surface-pressure
coefficient c p [3]

part. Obviously this leads to a large transport of kinematically active vorticity content
into the tip vortex. The flow field beneath the tip vortex becomes rearranged similar
to what happens at the lee side of a delta wing at angle of attack, Chap. 10.
The pressure field leads to a strong outward bending of the skin-friction lines near
the trailing edge on the upper side (there also the surface pressure is very low). The
bending is terminated by a pressure rise and an open-type secondary separation line
comes into being. All is reflected also in the skin-friction field in the lower part of
the figure.
The secondary vortex is formed just inboard of the wing tip. Topological con-
siderations demand the presence of an attachment line between the two separation
lines, which is clearly indicated. This secondary attachment line is of open type, too.
Moreover we see that a tertiary separation line is formed together with the topologi-
cally necessary—tertiary—attachment line. We conclude that below the primary tip
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 209

Fig. 8.33 Detail of the wing tip skin-friction line topology [3]. Upper part: pattern of the skin-
friction lines and distribution of the surface-pressure coefficient c p . Lower part: pattern of the
skin-friction lines and distribution of the absolute value of the skin-friction coefficient c f ≡ |c f |
210 8 Large Aspect-Ratio Wing Flow

Fig. 8.34 Levels of the dimensionless axial vorticity component ζ in the domain of the tip-vortex
system [3]

Fig. 8.35 Topological schematic of the open lee-side flow-field in the Poincaré surface just upstream
of the right wing’s trailing edge. The view is in positive x-direction—right wing from front too back

vortex both a secondary and a tertiary vortex are present, which means that we have
a tip-vortex system.
The levels of the dimensionless axial vorticity component ζ are reflecting the
tip-vortex system, Fig. 8.34. Note that ζ > 0 means that the—primary—tip vortex
turns clockwise, when looking at the right wing in positive x-direction. This is the
same direction as with the trailing vortex behind the right wing of the aircraft. ζ <
0 means counter-clockwise turning, which the secondary vortex does. The tertiary
vortex again turns clockwise. The net vorticity content is that of the whole tip-vortex
system.
The sketch of the open lee-side flow field in Fig. 8.35 shows how the arrangement
of the vortex phenomena in the Poincaré surface can be understood. The wing’s
contour is highly simplified. The fuselage is completely omitted.
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 211

Table 8.7 Figure 8.35: the singular points and their meaning
Singular point Type Kind of flow
F1 , F2 , F3 Focus Vortex center
   
S1 , S3 , S5 , S7 Half-saddle Attachment line
   
S2 , S4 , S6 , S8 Half-saddle Separation line

Fig. 8.36 Domain of the wing-tip vortex system: downstream development of the axial velocity
u/u ∞ (left side) and of the turbulence intensity T u (right side) [3]

In Table 8.7 we collect the singular points from Fig. 8.35 and their type and
meaning.
In order to apply the topological rule for the Poincaré surface, Rule 2 from
Sect. 7.4, we have to look at the whole wing, i.e., also at the left wing. Therefore,
taking into account also the singular points present at that wing side, we arrive at

     
1  1  14
N+ N − S+ S = (6 + 0) − 0 + = −1.
2 2 2
(8.4)
The fulfillment of Rule 2 tells us that the topology is a valid one. This, of course, is a
necessary, but not a sufficient condition. The surface skin-friction line pattern seen in
Fig. 8.33, however, makes the observation valid. Still remaining are the reservations
regarding the employed turbulence model.
Two general properties of the tip-vortex system are given in Fig. 8.36: the devel-
opment in downstream direction of the mean axial velocity u/u ∞ (left side) and of
the turbulence intensity T u (right side).
Regarding the axial velocity we first observe its expected reduction within the
boundary layer. In the evolving—primary—tip vortex the velocity defect still is
there, but an increase of u/u ∞ is visible in the vortex core. The free-stream velocity
with u/u ∞ = 1, however, is not reached within the vortex.
212 8 Large Aspect-Ratio Wing Flow

Due to the employed turbulence model, the turbulent kinetic energy k is a quantity
determined during the solution process. The—isotropic—turbulence intensity T u is
found with
√
2k
Tu = . (8.5)
u∞

In Fig. 8.36, right side, we observe the well developed turbulence intensity. In the
evolving primary tip vortex a distinct increase of T u is visible. The values of T u are
almost twice as high as in the boundary layer.

8.4.4 Excursion: The Wing-Tip Vortex System and

Non-linear Lift

So far we have discussed the phenomena present in the wing-tip—or wing side-
edge—flow regime. Three questions arise: (1) what does the tip-vortex system effec-
tuate regarding the wing’s performance as such, (2) are there similarities to the flow
past other wing configurations than the large aspect-ratio wing of the CRM con-
figuration? (3) how is the near-field/extended near field affected? We consider the
situation at the right wing only.
1. The tip-vortex system, as a whole turning in the same direction as the trailing
vortex, downstream of the trailing edge and the wing tip is merging with the
latter. In this process extra circulation is added to the trailing vortex, regarding
the amount see below. This, of course, leads to an increase of the induced drag.
When looking at the surface pressure in the tip-vortex regime, Fig. 8.32, we see
a distinct low-pressure area there (blue color). Such a phenomenon leads to what
is called “non-linear lift”, compared to the “linear” lift, found in potential theory
of lifting wings (Model (4) Of course, the respective surface area is very small
compared to the total wing’s surface. Nevertheless, an increment is added to the
wing’s lift (and pitching moment). However, this non-linear lift generally is not
a topic with large aspect-ratio wings, as the contribution is negligible.
The non-linear lift increment of course adds an increment to the root bending
moment, too, and locally also to the wing’s torsion (nose down in this example).
2. The above increments of induced drag, lift, pitching moment and structural forces
and moments obtain another importance when looking at wing configurations
other than the large aspect-ratio wing.
(a) Non-linear lift due to the low pressure area generated below the tip-vortex
system first of all is found on wings with small aspect ratio. L. Prandtl in 1921
gave experimental data for rectangular wings, which show a non-linear behavior
of the lift coefficient for Λ < 3, see in this regard the discussions in [6, 9]. The
strength of the non-linear lift also depends on the shape of the wing’s side edge,
its sweep, and whether it is sharp or rounded.
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 213

(b) The effect also is seen at aerodynamic trim and control surfaces as well as
at lift-enhancement surfaces like slats, Fowler flaps, etc. It is to surmise that at
such surfaces, too, side-edge (tip) vortex phenomena lead to first-order effects
regarding the forces and moments, when the surface has a low aspect ratio. An
early discussion can be found in the doctoral thesis of B. Göthert from the year
1940 [29].
(c) The really prominent example with non-linear lift is that of the delta wing at
angles of attack where lee-side vortices appear, Chap. 10. These vortices appear
at swept leading edges when a critical combination of the normal angle of attack
α N and the normal leading-edge Mach number M N is exceeded, Sect. 10.2.6.12
The swept leading edge then can be seen to act like a side edge.
The fluid-mechanical mechanisms present at delta wings basically are the same
as we find them at the wing side edges of large aspect-ratio wings, regardless of
whether sharp or round leading edges are present. Basically the same also is true
for the flow-field topology in terms of the pattern of the skin-friction lines over
the wing’s surface and the pattern found in the Poincaré surface, Fig. 8.35.
3. The tip-vortex system obviously affects the roll-up process of the trailing vortex
layer. In the case of the large aspect-ratio wing, the influence is rather weak. The
stronger the tip-vortex or side-edge system is, the more it will dominate the roll-up
process.
This is reflected by the length of the roll-up process. We have noted in Sect. 4.4
that in the case of large aspect-ratio swept or unswept wings the roll-up process
practically is completed a few half-span distances behind the wing. In the case
of low-aspect ratio wings that location can be already at a chord length or less
behind the wing.
Hence the extent and the flow properties of the near field and the extended near field
are affected by the strength of the tip vortex system. In particular the magnitude of
the downwash behind the wing depends to a degree on whether the trailing vortex
layer is rolled up or not. This then influences the effectiveness of the horizontal
tail surfaces [9].
We summarize: non-linear lift appears to be present at every finite-span lifting
wing, regardless of its aspect ratio. It contributes the more to the overall lift, the
smaller the aspect ratio of the wing is. At large aspect ratio wings its amount is
negligible. At aerodynamic trim, control and lift enhancement surfaces, which often
have a low aspect ratio, the flow around the side edges is similar to that around
a wing tip. Therefore sizeable non-linear effects can be expected. For delta wings
and strake configurations their non-linear lift effects are an important aerodynamic
property not only in view of their influence on the overall lift, but also in terms of
the maneuverability and agility of the aircraft.

12 Note that at the leading edge the vortex separation is beginning at the rear and then moves forward

with increasing angle of attack.

214 8 Large Aspect-Ratio Wing Flow

Fig. 8.37 Calculated skin-friction lines (black) and external (boundary-layer edge) streamlines
(red) at the upper side of the wing [3]. The angle β between the two lines in the projection on the
wing’s surface is color-coded (insert)

8.4.5 Flow Domain 0 Contd.: Trailing-Edge Flow and the

Compatibility Condition

Figure 8.37 shows flow patterns on the upper (suction) side of the wing. We recall
that if an external inviscid streamline locally is curved, the skin-friction line below
it is more strongly curved [26]. A point-of-inflection in the external streamline, of
course, accordingly alters the skin-friction line pattern.
Figure 8.37 also shows—indicated by the color strip—the presence of a shock
wave roughly at mid-chord and in spanwise direction between y ∗ ≈ 0.25 and y ∗ ≈
0.7. The wing’s shape was designed for the nominal 1-g cruise case with camber and
twist, the outboard airfoil sections being supercritical ones. For y ∗  0.25 and y ∗ 
0.7 the airfoil sections have a thickness to chord length ratio small enough that no
shock wave is present.13
What does the shock wave indicated in Fig. 8.37 effectuate? First of all the pre-
shock boundary-layer edge Mach number appears to be below Me  1.3 − 1.35. If

13 Note that along the attachment line the flow velocity does not change much. This is indicated,
too, by the pressure distribution along it, see the figure of the attachment line’s end, Fig. 8.31,
left side. Hence the consideration of the local chord properties—without regard to their spanwise
location—is permitted, though not in an exact quantitative sense.
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 215

this is given, the boundary layer does not separate due to the presence of the shock,
it only thickens, Sect. 6.1.2.
The shock wave locally obviously is not orthogonal to the upstream flow. Hence
both the external inviscid streamline and all streamlines in the boundary layer, includ-
ing the skin-friction line, are deflected. The figure shows a rather small deflection of
the external inviscid streamlines, but a large one of the skin-friction lines, indicated
by the black color.
Closer to the trailing edge a general increase is seen of the local angle β between
the lines. The shock wave appears to influence that patterns only weakly. The angle
β in general has a maximum along the trailing edge, which, however is diminished
toward the wing tip.
Nevertheless, the presence of the shock wave must be considered in view of the
compatibility condition at the trailing edge, Sect. 4.4. The shock wave in the inviscid
flow field leads to a velocity defect and an, if only slight, deflection of the external
streamlines. (The velocity defect also manifests itself in a shock-decambering effect,
Sect. 6.1.2.) Both effects appear to be small as we will see below.
Of special interest is the appearance of the overlaid patterns of the external inviscid
streamlines at the upper and the lower side of the wing, Fig. 8.38. At the trailing edge
the angle between them, the local trailing-edge flow shear angle ψe , Sect. 4.2.6, is
the decisive element of the compatibility condition at the trailing edge. If the external
inviscid velocities at the trailing edge are constant or nearly constant in spanwise
direction, which is the case here, it is only ψe , which governs the compatibility
condition.

Fig. 8.38 The inviscid external streamlines at the upper and the lower side of the wing [3]
216 8 Large Aspect-Ratio Wing Flow

The visual inspection of the streamline pattern at the trailing edge shows that the
angle ψe increases toward the wing tip, as demanded by the compatibility condition,
Sect. 4.4.
Important properties of the near wake, i.e., the trailing vortex layer, just behind
the wing’s trailing edge at x/c = 1.005 are plotted in Fig. 8.39. In this figure the
location y ∗ = y/(b/2) in a sense follows the trailing edge, i.e., it is not function of
x ∗ = const. The evaluation is made in the local wake coordinate system, Fig. 4.11,
where the trailing-edge flow shear angle ψe is defined as well as the vortex-line angle
ε. Along the trailing edge the Kutta direction is not taken into account.
The nomenclature is somewhat different from that used in Fig. 4.12. The lateral
edge velocity components there, veu and vel , are now ve2u and ve2l , the thicknesses δu
and δl are z e∗u and z e∗l , and the upper trailing-edge flow shear angle ψeu is ψu .14
The dimensionless circulations σ(ψ) and σ(Cl ), as well as the derivative
dσ(Cl )/dy ∗ are defined in Sect. 3.16, the vorticity content Ω 1 in Sect. 4.2.6.
The lateral velocity component ve2 (y ∗ ) = veu cos ψu = vel cos ψl in Fig. 8.39a at
the upper side of the wing is negative, i.e., it is directed toward the wing root. At
the lower side it is directed toward the wing tip. As was to be expected, too, both
components increase toward the wing tip, hence also the difference between them.
At y ∗ = 0.37 the kink in both curves indicates the location of the Yehudi break. Near
the wing root, both curves change sign.
The graphs in Fig. 8.39b indicate the spanwise distribution of the wake thicknesses
at the upper and the lower side. The wing’s chord length decreases from the root to
the tip. Hence the boundary-layer running length decreases, too. This is the reason
why the upper and lower wake thicknesses decrease toward the wing tip.15 The sharp
rise of the wake thickness at the wing tip’s upper side is due to the tip-vortex system.
The slightly thicker wake leaving the upper side reflects the presence of a larger
portion of an adverse pressure gradient. All this confirms what was found in an
earlier investigation of such flows [30].
The upper side trailing-edge flow shear angle ψu (y ∗ ) and the vortex-line angle
ε(y ∗ ) in Fig. 8.39c show the expected behavior. The (negative at the wing’s upper
side) shear angle ψu increases from the root to the tip, indicating the increase of the
strength of the wake’s kinematically active vorticity content in that direction. At the
left the influence of the fuselage is visible, and also that of the Yehudi break. The
vortex-line angle ε is positive because of the backward sweep of the trailing edge,
Sect. 4.2.6. That sweep from the root up to the Yehudi break is small, Fig. 8.21, but
then quite large. Hence from that break on, ε is nearly constant until the wing tip is
approached.

14 Note that the upper indices ’2’ of the velocity components do not indicate that they are squared
velocities. They simple indicate the lateral direction in the local wake coordinate system, t in Fig.
4.11.
15 This observation assumes a more or less two-dimensional behavior of the wing’s boundary layers.

The justification is twofold: (1) the thickness of the boundary layer along the attachment line is
more or less constant, and with this the initial conditions for the chordwise flow, (2) a larger three-
dimensionality is present only in the immediate vicinity of the attachment line. Hence for qualitative
considerations and also for crude estimations the assumption of two-dimensionality is permissible.
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 217

Fig. 8.39 Properties of the trailing vortex layer and the compatibility condition at the location
x/c = 1.005 immediately behind the trailing edge [3]: a lateral velocity component ve2 (y ∗ ) at the
upper and the lower side, b wake thickness z e∗ (y ∗ ) at the upper and the lower side, c trailing-edge
flow shear angle ψu (y ∗ ) at the upper side and vortex-line angle ε(y ∗ ), and d non-dimensional
circulation σ(y ∗ ), its derivative in y ∗ -direction and the local vorticity content Ω(y ∗ )

Figure 8.39d gives the vorticity content Ω1 (solid black line)—computed with the
flow properties at the trailing edge—and the dimensionless circulation distribution
dσ(Cl )/dy ∗ (blue dashed line)—found from the computed circulation distribution.
Both are connected via the compatibility condition Eq. (4.20), Sect. 4.4.
Because we consider the wing’s right-hand side, both entities are negative. In
order to help the reader in the understanding of the result, we discuss that now in
terms of the magnitudes |Ω1 | and |dσ(Cl )/dy ∗ |.
218 8 Large Aspect-Ratio Wing Flow

Fig. 8.40 Spanwise

circulation distribution
obtained with the
compatibility condition σ(ψ)
and with the
Kutta-Joukowsky theorem
σ(Cl ), and comparison with
the elliptical distribution σ ∗
[3]

Ω1 at the wing root is positive and changes sign at y ∗ ≈ 0.15. |Ω1 | then increases
and at the Yehudi break reaches a local maximum with |Ω1 | ≈ 0.08. A further weak
increase is present up to y ∗ ≈ 0.8. Beyond that a strong increase happens up to |Ω1 |
≈ 0.35.
The spanwise integration of Ω1 , beginning at the wing tip, results in the circulation
distribution, here given as dimensionless distribution σ(ψ) (blue dashed line). The
value rises from σ(ψ) = 0 at the wing tip to the maximum σ(ψ) ≈ 0.071 at y ∗ ≈
0.2. The root circulation at y ∗ = 0.13 is σ0 (ψ) = 0.0697.
Sofar we have dealt with the vorticity content Ω1 found from flow properties at
the wing’s trailing edge and derived from it the circulation distribution σ(ψ). Now
we check these results with the help of the Kutta-Joukowsky theorem. This is made
by relating with Eq. (3.31) locally the circulation Γ (y) to the computed pressure
distribution, respectively the local lift coefficient Cl (y).
The resulting dimensionless circulation σ(Cl ) (green dash-dotted line) is in rea-
sonable agreement with σ(ψ). The former has somewhat higher values at the outboard
and lower at the inboard side. The root circulation is almost equal. Also a reasonable
agreement shows the differentiated circulation dσ(Cl )/dy ∗ with Ω1 . At the Yehudi
break the difference is larger and even larger at the root.
In Fig. 8.40 the two obtained circulation distributions are compared with the
elliptic distribution σ ∗ , Sect. 3.16. σ(Cl ) and σ ∗ show a good agreement for y ∗ >
0.6, whereas σ(ψ) is smaller by up to 15 per cent in this region. For y ∗ < 0.6 the
elliptic distribution is smaller, while the other two agree rather well, except for the
root region, see above.
The root circulations are σ0∗ = 0.061, σ0 (ψ) = 0.0697 and σ0 (Cl ) = 0.068. The
two computed values are well above the elliptic one, and differ themselves only by
about 2.5 per cent.
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 219

Our conclusion is that the concept of kinematically active and inactive vorticity
content, which culminates in the formulation of the compatibility condition, is a
viable one. The differences in the spanwise circulation distributions are due to some
properties of the flow fields at the upper and the lower wing surface. These are the
(inviscid) flow deflection and the velocity defect due to the shock wave at the wing’s
upper surface as well as local separation effects at the trailing edge due to its finite
thickness.
Further we must keep in mind that the vorticity-content concept assumes Re →
∞, whereas here we have a Reynolds number of finite magnitude.

8.4.6 Flow Domain 1: The Trailing Vortex Layer in the Near

Field

The wake profiles in the near field behind the trailing edge—Domain 1, see Fig.
8.1—are studied, as well as the wake properties at x ∗ = 0.5.
• The Profiles of the Trailing Vortex Layer Close to the Trailing Edge. We
begin with an investigation of the wake profiles just behind the trailing edge of the
wing, Fig. 8.41. The three dimensionless velocity components v i are those indicated
in Fig. 4.12, except for v 3 .16 The coordinate system in vertical position is located with
z ∗ = 0 lying at the local trailing-edge position. In spanwise direction the locations
are with y ∗ = 0.13 close to the wing root, with y ∗ = 0.4 just outboard of the Yehudi
break, and with y ∗ = 0.99 very close to the wing tip. The x/c-locations are related
to the local chord length with x/c = 1.03 lying very close to the trailing edge and
the two others at x/c = 1.17 and x/c = 3.35.
The streamwise v 1 /u ∞ -profile represents the boundary-layer profile with kine-
matically inactive vorticity content leaving the wing’s trailing edge. The extent of
the profiles declines in z ∗ -direction, like that of the others, from the wing root to the
wing tip because the chord length is reducing in that direction, see the discussion of
Fig. 8.39. Despite the shock wave at the upper side of the wing, the velocities at the
upper and the lower side are nearly the same. The profiles in stream direction are
filling up quickly. At x/c(y) = 3.35 they are nearly vanished. At the same time the
wake is sinking. This is most pronounced at y ∗ = 0.13.
The lateral v 2 /u ∞ -profile is the wake profile with the kinematically active vorticity
content. The shear is well discernible, with the local shear angle ψ increasing toward
the wing tip. Only at y ∗ = 0.13ψ is negative due to the wing-body interference, see
also the spanwise circulation distribution in Fig. 8.40.
The downwash, represented by v 3 /u ∞ , is stronger at the upper side of the wing
than at the lower side. At y ∗ = 0.99 the downwash is positive at the lower side, i.e. it
is an upwash. This appears to be induced by the combination of the evolving trailing
vortex and the tip-vortex system.

16 The superscript index notation does not mean that the velocity components are contravariant ones

as they are defined for instance in [26].

220 8 Large Aspect-Ratio Wing Flow

Fig. 8.41 Wake profiles at three different spanwise and axial positions [3]. The dimensionless
velocity components are given in the local wake coordinate system, Fig. 4.11

The maximum of the dimensionless vorticity ωb/2u ∞ increases toward the wing
tip. Its profile also mirrors the decrease of the wake thickness in that direction, like
that of the v 1 /u ∞ -profile.
Quantitatively all compares well with the results of the Euler simulation in
Sect. 8.3. There of course the u(z)-profile, the equivalent to v 1 /u ∞ , is unity through-
out, Fig. 8.16. The reader is asked to study also the results in [30].
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 221

Fig. 8.42 Distribution of the

dimensionless axial vorticity
ζ at x ∗ = 0.5 [3]

• The Properties of the Trailing Vortex Layer at x ∗ = 0.5. Next we study near-
field properties at the location x ∗ = 0.5. For a visualization of this location see Fig.
8.50. We consider the right side of the flow field, y ∗ > 0. At x ∗ = 0.5 the roll-up
process of the trailing vortex layer already has resulted in a distinct trailing vortex,
Fig. 8.42.
In the trailing vortex layer the dimensionless axial vorticity ζ, eq. (8.3), increases
from ζ ≈ 2 near the wing root to ζ ≈ 6 near the wing tip. The Yehudi break has a
weak influence, visible in the vortex layer. The inset in Fig. 8.42 shows the trailing
vortex in more detail. In its core we have ζ ≈ 25.
Some wiggles present in the picture are due to grid properties. The merging of
the tip-vortex system into the trailing vortex is vaguely indicated.
The figure also shows, at and below the horizontal stabilizer, a domain with weak
negative vorticity. This is due to a downward force, which is present at the stabilizer,
like a trim force. A trailing vortex layer leaves the trailing edge and rolls up into a
pair of trailing vortices, counter-rotating to the wing’s trailing vortices.
The downward force is present, although the stabilizer nominally has zero angle
of attack. However, because trailing vortex layer and vortices of the wing induce
a downwash velocity component w, see Sect. 8.1, the stabilizer effectively has a
negative angle of attack.
Figure 8.43 at y ∗ = 0.15 visualizes the downwash velocity component w/u ∞
behind the wing and in particular at the horizontal stabilizer. Note that the stabilizer
itself induces an upwash velocity.
Of interest are also the velocity distributions and the turbulence intensity in the
trailing vortex layer and the vortex. Particularly interesting is the axial velocity u/u ∞ ,
Fig. 8.44. We observe a distinct deficit of the axial velocity both in the vortex layer
222 8 Large Aspect-Ratio Wing Flow

Fig. 8.43 Dimensionless vertical velocity w/u ∞ behind the wing at y ∗ = 0.15 [3]

Fig. 8.44 The dimensionless

velocity u/u ∞ at x ∗ = 0.5

and the vortex. The vortex layer shows minimum velocities of u/u ∞ ≈ 0.93, whereas
in the vortex core we have u/u ∞ ≈ 0.97.
At x ∗ = 0.5 the roll-up of the trailing vortex layer is not yet completed and
the trailing vortex is not yet fully developed. (The vortex is young with its age
τ ∗ = 0.0039.) In Fig. 8.45 we compare the vortex development at this location with
two analytical vortex models.
The computed circumferential velocity vθ,num (red symbols) is compared to the
analytically found velocities of models, which represent fully developed vortices:
Rankine vortex vθ,Rankine (green symbols), Eq. (3.16) and the Lamb-Oseen vortex
vθ,Lamb−Oseen (blue symbols), Eq. (3.17), both to be found in Sect. 3.10.
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 223

Fig. 8.45 Comparison of

numerical and analytical
circumferential velocities
vθ (r/(b/2)) at x ∗ = 0.5. The
vortex age is τ ∗ = 0.0039 [3]

Two numbers are needed for the latter, the viscous core radius rc and the root
circulation Γ0 . The first one was determined from the numerical solution, see Fig.
3.9, the second one is taken in non-dimensional form from Fig. 8.40.
Figure 8.45 indicates that at the location rc the numerically found value of the
circumferential velocity is much below the analytical ones. The latter also differ
much. In the vortex core, r < rc the linear velocity distributions are well obtained,
outside of it the agreement at least is reasonable.

8.4.7 Flow Domain 2: The Trailing Vortices Appear

We look at the roll-up process of the wake and at the wing’s circulation in Domain
2, see Fig. 8.1.
• The Roll-Up Process in Terms of the Dimensionless Axial Vorticity ζ. In
the extended near field we follow the development of the trailing vortices. First we
have a look at their interaction—in terms of the axial vorticity ζ—with the trailing
vortices of the horizontal tail plane. That surface is located between x ∗ = 0.15 and
0.30. We only describe the situation behind the right wing.
The trailing vortex grows in diameter and moves inwards. The trailing vortex
layer weakens, grows in thickness and is sinking faster than the trailing vortex. (The
author of [3] notes that the choice of the turbulence model has an influence on this
development.) At x ∗ = 5 we see that a part of the vortex layer interacts with that
of the horizontal tail plane, Fig. 8.46. This part finally will not be ingested into the
trailing vortex.
The roll-up process is not finished at x ∗ = 9, Fig. 8.47. The separated part of
the trailing vortex layer—in a vortex pairing process with the trailing vortex of the
horizontal tail plane, which acts as a trim surface—now is counterrotating to the
wing’s trailing vortex. The vortex age of the trailing vortex τ ∗ = 0.071 still is low. In
224 8 Large Aspect-Ratio Wing Flow

Fig. 8.46 Distribution of the

dimensionless axial vorticity
ζ at x ∗ = 5 [3]

Fig. 8.47 Distribution of the

dimensionless axial vorticity
ζ at x ∗ = 9 [3]

[31] it is reported that for a high-lift transport aircraft the roll-up process was finished
at τ ∗ = 0.25 − 0.3.
We again compare now the numerical and analytical circumferential velocities
vθ (r/(b/2)) as we did it at x ∗ = 0.5, Fig. 8.45. Now the agreement of the numerical
and the Lamb-Oseen data is very good, Fig. 8.48. The Rankine vortex of course
approximates well the circumferential velocity only away from r/(b/2) = 0.1 and
below r/(b/2) = 0.05. However, it must be expected that further downstream the
agreement will be less good due to the assumptions in the analytical models.
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 225

Fig. 8.48 Comparison of

numerical and analytical
circumferential velocities
vθ (r/(b/2)) at x ∗ = 9 [3]

The motion of the—still not fully developed—trailing vortex is represented by

the vortex trajectory, Fig. 8.49.
In Sect. 4.4 we have noted that the distance b0 between the fully developed trailing-
vortex pair increases with increasing aspect ratio. The ratio b0 /b, which also is called
the load factor s, Sect. 3.16, for the elliptical circulation distribution is
b0 π
s = s∗ = = = 0.7854, (8.6)
b 4
only for wings with very large aspect ratio s → 1.
In the y ∗ -z ∗ plane, Fig. 8.49, the vortex center is seen to move from the initial
position at the WRP, Fig. 8.21, toward the wing’s root, first upward and then strongly
downward.17 In the x ∗ − y ∗ plane the position of the vortex center appears to reach
its final spanwise position at x ∗ > 10 with y ∗ = s < 0.8.
We summarize: (1) the vortex is still young at x ∗ = 10 (the roll-up process is not
yet finished), (2) the solution becomes erroneous for x ∗ > 9 (see the graph in the
z ∗ -x ∗ plane and below), and (3) the circulation is the original one minus that of the
counterrotating trailing vortices of the horizontal tail plane, Fig. 8.50. If in [3] the
final value of the load factor is found to be s = 0.715 < s ∗ = 0.7854, these three
items must be taken into account as possible cause for this small value.
• The Wing’s Circulation in Flow Domain 2. The dimensionless circulation
G(x ∗ ) = Γ (x ∗ )/Γ0 in the extended near field is given in Fig. 8.50. The upper blue
line is the isolated wing’s circulation. It is nearly constant up to x ∗ = 9. Up to this
location the inserted PS grid, Sect. 8.4.2, is present. The evaluated data at x ∗ = 10
show that the original grid resolution is insufficient there, remember the remarks in
Sect. 5.3.

17 The WRP was chosen as initial position, which, however, is a makeshift, because the actual flow
situation at that location is highly complex, Fig. 8.33.
226 8 Large Aspect-Ratio Wing Flow

Fig. 8.49 Trajectory of the right-hand side trailing vortex [3]. Upper part left: y ∗ -z ∗ plane. Upper
part right: x ∗ -y ∗ plane. Lower part: x ∗ -z ∗ plane

The isolated wing’s circulation up to x ∗ ≈ 3 is slightly above G = 1. This prob-

ably is due to inaccuracies in the determination of the root circulation caused by
the wing/fuselage interference. The lower curve represents the net circulation. The
opposite circulation of the horizontal tail plane was determined once at x ∗ = 0.4
and then assumed to be constant. Even if the accuracy of the approach is not fully
satisfying, the result shows that the influence of the horizontal tail plane, which in
effect is a trim surface, is rather large. This points to the fact that the aircraft’s trim
needs attention in order to minimize the trim drag.
The development of the isolated trailing vortex as it grows in the roll-up pro-
cess in x-direction is given in Fig. 8.51. The development is shown in terms of the
dimensionless circulation G(x ∗ ) = Γ (x ∗ )/Γ0 . G increases in downstream direction
because the trailing vortex layer successively is absorbed by the trailing vortex.
8.4 RANS/URANS Solution (Model 10) for the CRM Case: Flow Domains 0, 1, and 2 227

Fig. 8.50 The dimensionless circulation G(x ∗ ) in the extended near field [3]. The upper blue curve
is G of the wing alone, the lower black one that with the opposite circulation of the horizontal tail
plane taken into account

Fig. 8.51 The development

of the dimensionless
circulation G(x ∗ ) of the
trailing vortex [3]

Figure 8.51 contains two curves: Gωx ) and G(vθ ), both as function of x ∗ . The first
one was found by applying Stokes theorem by integrating ωx over the cross-section
of the vortex, Sect. 3.3, the second one by computing the line integral around the
vortex with vθ .
228 8 Large Aspect-Ratio Wing Flow

Fig. 8.52 Development of the characteristics of the trailing vortex [3]. Left: the viscous core radius
rc (x ∗ ). Right: the maximum circumferential velocity component vθ (x ∗ )

The results in a sense are qualitative ones, because of the difficulty to define the
integration domains. The trailing vortex is not a circular one, it is moreover connected
to the trailing vortex layer. Both curves initially grow with a large gradient in x ∗ -
direction, which at x ∗  3 becomes smaller. The values at x ∗ = 9 clearly show that
there the roll-up process is not finished.
Interesting for us is the value of G(ωx ) close to the WRP. There it represents the
net circulation contained in the tip-vortex system. This circulation is G(x ∗ = 0) =
0.068. This means that the net circulation of the tip-vortex system is about seven per
cent of the circulation of the trailing vortex.
In closing this section we have a look at some of the characteristic properties of
the evolving trailing vortex. We have to note that the magnitudes of the obtained
data to a degree depend on the applied turbulence model. All quantities considered
change strongly in the region up to two wing spans downstream of the trailing edge
(x ∗  2)and then rather slowly.
In Fig. 8.52 the viscous core radius rv , Sect. 3.16, increases very fast throughout
(left part of the figure). The maximum circumferential velocity has a rather large
magnitude in the beginning with vθmax ≈ 0.27 and then decreases fast to vθmax ≈ 0.07
(right part of the picture). This strong reduction must be seen as the consequence of
the fast growth of the viscous core radius. (Beyond x ∗ = 9 all results have deficits
because of the insufficient grid resolution there (see above).)
The axial velocity defect (u ∞ − u min )/u ∞ (x ∗ ) in Fig. 8.53 initially has a value
of about 10 per cent, but then decreases fast to around one per cent (left part of
the figure). A similar drop is seen of the maximum dimensionless axial vorticity
component ζ = ωx b/2u ∞ (right part of the figure).
8.5 Concluding Remarks 229

Fig. 8.53 Development of the characteristics of the trailing vortex [3]. Left: the axial velocity defect
(u ∞ − u min )/u ∞ (x ∗ ). Right: the maximum dimensionless axial vorticity component ζmax (x ∗ )

8.5 Concluding Remarks

We have demonstrated by means of Unit Problems the capabilities of three kinds of

mathematical flow models regarding the simulation and the analysis of the vortical
and separated flow past large aspect-ratio lifting wings.
In Sect. 8.2 the concept of the local vorticity content was employed in order to
investigate the capabilities of several panel methods (Model 4) in the Flow Domain 0.
It turned out that even if the overall results are in reasonable agreement, the flow fields
near the trailing edges of the investigated wings were not computed correctly by most
of the methods. This concerned methods with lower-order singularity formulations.
Their solutions did not fulfill the trailing-edge compatibility condition, Eq. (4.20).
Only a higher-order formulation gave the proper results. The vortex-line angle  in
general was computed correctly.
The application of a discrete numerical solution of the Euler equations (Model 8)
to the flow past a large aspect-ratio wing in Sect. 8.3 yielded satisfactory results in
the Flow Domains 0 and 1. Usually the results of such simulations are considered
in terms of the forces and moments only. Here the properties of the Euler wake and
partly also of the Euler boundary layer were investigated. The near-wake properties
were found to be qualitatively correct. At the trailing edge the compatibility condition
is fulfilled.
In Sect. 8.4 the flow past the wing of the Common Research Model was inves-
tigated with a RANS/URANS method (Model 10). Considered were the flow and
wake properties in the Flow Domains 0, 1 and 2.
In Domain 0 also the flow at the wing root, along the leading edge and at the wing
tip was studied. Because the CRM has no wing root/fuselage fairing, a complex
separation pattern was found there, similar as at the wing tip. The result must be
considered with some reservation, because the choice of the turbulence model and
230 8 Large Aspect-Ratio Wing Flow

the discretization approach influence the outcome [3]. The pattern of the skin-friction
lines revealed that both the classical and the open-type separation, Sect. 7.1.4, were
present.
The end of the attachment line in the vicinity of the wing tip was found to be of
open-type. The wing-tip separation completely happens at the rear upper side of the
wing. Besides the primary vortex, a secondary and even a tertiary vortex was found.
Therefore we speak of the wing-tip vortex system.
Wing-tip separation always was found to be of open-type. The distribution of the
surface pressure coefficient reveals a distinct low pressure area near the trailing edge
of the wing tip. Such an area is the cause of the so-called non-linear lift. Hence this
kind of lift is present also on large aspect-ratio wings. However, it has such a small
magnitude that it does not play a role.
The trailing-edge compatibility condition was investigated, too. It was found that
the concept of kinematically active vorticity permits a good explanation of the lifting
wing’s flow situation at the trailing edge.
The roll-up process of the wing’s trailing vortex layer into the pair of trailing
vortices in Flow Domain 1 and 2 was visualized. It was found that the wing tip
vortex system substantially contributes to the wake’s circulation.
Also visualized was the interaction of the wing’s wake with the wake of the
horizontal tail plane, which effectively acts like a trim surface. It was further shown
that beyond the end of the inserted PS grid the original grid is sof coarse that the
circulation of the wake/trailing vortex system is not preserved downstream. This is
a matter, which always needs attention.

8.6 Problems

Problem 8.1 Check with the correlations given in Fig. 10.23 whether at the Kolbe
wing at α = 8.2◦ the possibility of leading edge separation and hence lee-side vortices
exists. How is the situation, if the wing would have a sharp leading edge?

Problem 8.2 Make the same check for the forward-swept wing. Assume M∞ =
0.25.

Problem 8.3 In Table 8.4 the lift found with the Euler solution for M∞ = 0.3 is
compared with that found with linear theory for M∞ = 0.01. Apply the Prandtl-
Glauert rule for M∞ = 0.3 to the result of linear theory. How do now the results
compare?

Problem 8.4 The lift coefficient of the CRM case was found to be C L = 0.445.
Assume an elliptical circulation distribution and compute the induced drag coefficient
C Di . How large in per cent is it in relation to the total drag coefficient C D = 0.025?

Problem 8.5 For the CRM case three root-circulation values were obtained, Sect. 8.4.
Assume elliptical circulation distributions for all of them and compute C L∗ and C D∗ i .
8.6 Problems 231

Compare with the data given on Sect. 8.4 and in Problem 8.4. Choose the Oswald
factor to be e = 0.8.

Problem 8.6 From Fig. 8.50 it can be concluded how much the overall circulation
is reduced due to the horizontal tail surface. How are the lift and the induced drag
affected?

Problem 8.7 Consider an aircraft flying at constant speed at the beginning of the
cruise segment, Fig. 1.3, of a long-distance flight. The wing span is b = 60 m,
disregarding the fuselage, the aspect ratio is Λ = 9. The mass of the aircraft is
250,000 kg, the flight speed is u ∞ = 800 km/h, the flight altitude H = 10 km.
Assume that Model 4 holds and that a spanwise elliptic circulation distribution
is present. (a) How large is the circulation Γ0 of the trailing vortices? (b) How large
is Γ0 , if the wing span would only be b = 30 m, and the aspect ratio Γ = 4.5? (c)
How large is the wing loading Ws for the two cases? (d) What do the results mean
regarding the strength of the trailing vortices? (e) How are the coefficients of lift and
induced drag affected?

Problem 8.8 An elliptic wing of aspect ratio Λ = 10 flies at M∞ = 0.8. What is the
lift curve slope? Use the lifting-line model (low speed) in Appendix A.4 and correct
the value.

References

1. Breitsamter, C.: Wake vortex characteristics of transport aircraft. Prog. Aerosp. Sci. 47(1),
89–134 (2011)
2. Obert, E.: Aerodynamic Design of Transport Aircraft. IOS Press, Delft (2009)
3. Pfnür, S.: Numerical Analysis of the Trailing-Edge Vortex Layer and the Wake-Vortex System of
a Generic Transport-Aircraft Configuration. Master’s Thesis, Institut für Luft- und Raumfahrt,
Technical University München, Germany (2015)
4. Schrauf, G., Laporte, F.: AWIATOR—wake-vortex characterization methodology. In: KATnet
Conference on Key Aerodynamic Technologies, Bremen, Germany (2005). Accessed 20–22
June 2005
5. Prandtl, L.: Tragflügeltheorie, I. und II. Mitteilung. Nachrichten der Kgl. Ges. Wiss. Göttingen,
Math.-Phys. Klasse, 451–477 (1918) and 107–137 (1919)
6. Anderson Jr., J.D.: Fundamentals of Aerodynamics, 5th edn. McGraw Hill, New York (2011)
7. Vos, R., Farokhi, S.: Introduction to Transonic Aerodynamics. Springer Science+Business
Media, Dordrecht (2015)
8. Küchemann, D.: The Aerodynamic Design of Aircraft. Pergamon Press, Oxford. Also AIAA
Education Series, p. 2012. AIAA, Reston, V (1978)
9. Schlichting, H., Truckenbrodt, E.: Aerodynamik des Flugzeuges, Vol. 1 and 2, Springer-
Verlag, Berlin/Göttingen/Heidelberg (1959), also: Aerodynamics of the Aeroplane, 2nd edition
(revised). McGraw Hill Higher Education, New York (1979)
10. Rudnik, R.: Transportflugzeuge. In: Rossow, C.-C., Wolf, K., Horst, P. (eds.) Handbuch der
Luftfahrzeugtechnik, pp. 83–113. Carl Hanser Verlag, München, Germany (2014)
11. Rizzi, A., Oppelstrup, J.: Aircraft Aerodynamik Design with Computational Software. Cam-
bridge University Press (2020)
232 8 Large Aspect-Ratio Wing Flow

12. Kolbe, D.C., Boltz, F.W.: The Forces and Pressure Distributions at Subsonic Speeds on a Plane
Wing Having 45◦ of Sweepback, an Aspect Ratio of 3, and a Taper Ratio of 0.5. NACA RM
A51G31 (1951)
13. Hirschel, E.H., Sacher, P.: A Comparative Theoretical Study of the Boundary-Layer Develop-
ment on Forward Swept Wings. In: R.K. Nangia (ed.), Proceedings of International Conference
on Forward Swept Wings, Bristol, 1982. University of Bristol, U.K. (1983)
14. Hirschel, E.H., Fornasier, L.: Flowfield and Vorticity Distribution Near Wing Trailing Edges.
AIAA-Paper 1984–0421 (1984)
15. Fornasier, L.: HISSS–A Higher-Order Subsonic/Supersonic Singularity Method for Calculating
Linearized Potential Flow. AIAA-Paper 1984–1646 (1984)
16. Sytsma, H.S., Hewitt, B.L., Rubbert, P.E.: A Comparison of Panel Methods for Subsonic Flow
Computation. AGARD-AG-241 (1979)
17. Hirschel, E.H.: Das Verfahren von Cousteix-Aupoix zur Berechnung von turbulenten, dreidi-
mensionalen Grenzschichten. MBB-UFE122-AERO-MT-484, Ottobrunn, Germany (1983)
18. Hirschel, E.H., Rizzi, A.: The Mechanism of Vorticity Creation in Euler Solutions for Lifting
Wings. In: A. Elsenaar, G. Eriksson (eds.), Proceedings Symposium on the International Vortex-
Flow Experiment on Euler Code Validation, FFA Bromma, Sweden, pp. 127–162 (1987).
Accessed 1–3 Oct 1986
19. Rizzi, A., Eriksson, L.-E.: Computation of flow around wings based on the Euler equations. J.
Fluid Mech. 148, 45–71 (1984)
20. Vassberg, J.C., DeHaan, M.A., Rivers, S.M., Wahls, R.A.: Development of a Common Research
Model for Applied CFD Validation Studies. AIAA-Paper 2008–6919 (2008)
21. Vassberg, J.C., Tinoco, E.N., Mani, M., Rider, B., Zickuhr, T., Levy, D.W., Brodersen, O.P.,
Eisfeld, B., Crippa, S., Wahls, R.A., Morrison, J.H., Mavriplis, D.J., Murayama, M.: Summary
of the fourth AIAA computational fluid dynamics drag prediction workshop. J. Aircr. 51(4),
1070–1089 (2014)
22. Levy, D.W., Laflin, K.R., Tinoco, E.N., Vassberg, J.C., Mani, M., Rider, B., Rumsey, C.L.,
Wahls, R.A., Morrison, J.H., Brodersen, O.P., Crippa, S., Mavriplis, D.J., Murayama, M.:
Summary of data from the fifth computational fluid dynamics drag prediction workshop. J.
Aircr. 51(4), 1194–1213 (2014)
23. Gerhold, T.: Overview of the Hybrid RANS Code TAU. In Kroll, N., Fassbender, J. (eds.)
MEGAFLOW—Numerical Flow Simulation for Aircraft Design. Notes on Numerical Fluid
Mechanics and Multidisciplinary Design, NNFM 89, pp. 81–92. Springer, Berlin (2005)
24. Spalart, P.R., Allmaras, S.R.: A One-Equation Turbulence Model for Aerodynamic Flows.
AIAA-Paper 1992–0439 (1992)
25. Menter, F.R.: Two-equation eddy-viscosity turbulence models for engineering applications.
AIAA J. 32(8), 1598–1605 (1994)
26. Hirschel, E.H., Cousteix, J., Kordulla, W.: Three-Dimensional Attached Viscous Flow.
Springer, Berlin (2014)
27. Délery, J.: Three-Dimensional Separated Flow Topology. ISTE, London and Wiley (2013)
28. Kármán, Th, von, : Aerodynamics-Selected Topics in the Light of their Historical Development.
Cornell University Press, Ithaca, New York (1954)
29. Göthert, B.: Systematische Untersuchungen an Flügeln mit Klappen und Hilfsklappen (System-
atic Investigations at Wings with Flaps and Servo-Flaps). Doctoral Thesis, Technical Univer-
sity Braunschweig, Germany (1940), also Jahrbuch 1940 der Luftfahrtforschung, pp. 278–307
(1940)
30. Wanie, K.M., Hirschel, E.H., Schmatz, M.A.: Analysis of Numerical Solutions for Three-
Dimensional Lifting Wing Flows. Z. f. Flugwissenschaften und Weltraumforschung (ZFW)
15, 107–118 (1991)
31. Breitsamter, C.: Nachlaufwirbelsysteme großer Transportflugzeuge - Experimentelle Charak-
terisierung und Beeinflussung (Wake-Vortex Systems of Large Transport Aircraft—
Experimental Characterization and Manipulation). Inaugural Thesis, Technische Universität
München, 2007, utzverlag, München, Germany (2007)
Chapter 9
Particular Flow Problems of Large
Aspect-Ratio Wings

The topics of this book are the basic principles and Unit Problems of separated
and vortical flow in aircraft wing aerodynamics. This chapter is devoted to short
considerations of application-oriented topics. Complete literature reviews are not
intended, we give compact accounts of selected topics and provide references for
further reading.
In the following Sect. 9.1 the shock-wave/boundary-layer interaction occurring at
the supercritical airfoil is considered. It follows a sketch of the flow and separation
phenomena at a high-lift system, Sect. 9.2. The wing in high-lift configuration is the
topic of Sect. 9.3. The effect of the nacelle-strake vortex is discussed in Sect. 9.4.
Section 9.5 treats the topic of wing-tip devices and Sect. 9.6 finally the problem of
aircraft-wake control.

9.1 Supercritical Airfoil—Shock-Wave/Boundary-Layer

Interaction

The subject of this section is transonic shock wave/boundary-layer interaction, see,

e.g., [1]. Why is this of importance? Breguet’s range formula in its simplest form
relates the flight range R with the parameters flying speed v∞ , aerodynamic quality
(lift to drag ratio) C L /C D , specific impulse Isp , as well as the structure parameters
empty mass m e , payload mass m P and fuel mass m F 1 :
 
CL mF
R = v∞ Isp ln 1 + . (9.1)
CD me + m P

This formula can also be written in the form

1 The relation of the specific impulse Isp to the specific fuel consumption b [kg/s N] of the propulsion
system is Isp = 1/(b g).
© Springer-Verlag GmbH Germany, part of Springer Nature 2021 233
E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_9
234 9 Particular Flow Problems of Large Aspect-Ratio Wings
 
CL mF
R = M ∞ a∞ Isp ln 1 + . (9.2)
CD me + m P

Of interest in our context is the combination of the terms

CL
M∞ ,
CD

which tells us that the flight Mach number should be as high as possible and at the
same time also the aerodynamic quality, the lift to drag ratio. In Sect. 2.4.2 it was
shown that a phenomenon exists, the drag divergence, which manifests itself as a
strong drag increase once a certain flight Mach number, the drag-divergence Mach
number Mdd , is reached. The thinner the airfoil, the higher is Mdd .
Wings of transonic transport aircraft are swept and have supercritical airfoils,
Sect. 2.4.2. The present section shortly discusses and illustrates flow and separation
phenomena up to transonic buffet, which are of relevance for the drag-divergence
Mach number of a supercritical airfoil, respectively wing. Even if in reality the
flow past a swept wing is three-dimensional, the two-dimensional case is a fair
approximation to the reality, of which an example is shown in Fig. 8.37.
Consider Fig. 9.1. It gives in more detail the flow field past the lifting transonic
airfoil, which was considered in Sect. 4.2.3, Fig. 4.6. The sonic line—broken line—
is indicated also in the boundary layer below the supersonic flow pocket at the
suction side of the airfoil. The terminating shock wave is slightly curved, as is to be
demanded, and at the airfoil’s surface in the inviscid picture is to end normal to it,
for both propositions see Sect. 4.2.3.

Fig. 9.1 Schematic of the flow field of a supercritical airfoil, after [2]. The thickness δ of the
boundary layer is not to scale
9.1 Supercritical Airfoil—Shock-Wave/Boundary-Layer Interaction 235

Fig. 9.2 Schematic of weak shock-wave/turbulent boundary-layer interaction [2]

The sonic line is shown to curve around the tip (1) of the shock wave and attaches
to it in (2). The interpretation is that the shock wave above (2) is oblique to the
local flow direction. Therefore the flow behind it is supersonic and then decelerates
isentropically to sonic and subsonic flow.
The shock wave interacts with the boundary layer in the interaction domain. Two
cases are to be distinguished, weak and strong shock-wave/boundary-layer interac-
tion, in our example with turbulent flow.
The interaction is considered to be weak, if the boundary layer does not separate
due to the shock wave, and strong, if it does. The critical pre-shock Mach number at
the shock foot is M f oot ≈ 1.3–1.35, [1], see also Sect. 2.2.
• Weak shock-wave/boundary-layer interaction. First we consider the weak interac-
tion case, Fig. 9.2. The external inviscid flow is supersonic. The flow in the bound-
ary layer of course also is supersonic down to a distance from the wall, which
depends on the overall boundary-layer flow properties. In the subsonic part of the
boundary layer information about the outer pressure field can travel upstream. A
fan of isentropic compression waves results above the sonic line and the pressure
rise due to the shock wave is smeared out. No separation occurs, only a thickening
of the boundary layer.
The shock wave just above the boundary layer and in its upper part is slightly
oblique to the flow. That leads to weak supersonic flow behind the shock wave
with isentropic recompression. The result is the “supersonic pocket” behind the
shock wave, marked in the figure with M > 1.
Downstream of the interaction domain the boundary layer is subsonic and the wall
shear stress is reduced. Overall the boundary layer is at risk to separate early at
the rear of the wing. That would change the flow-off separation there to a complex
trailing-edge separation pattern.
236 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.3 Schematic of strong shock-wave/turbulent boundary-layer interaction [2]

• Strong shock-wave/boundary-layer interaction. Once the pre-shock Mach number

is large enough, the shock wave will lead to separation of the boundary layer. A
separation bubble is formed, Fig. 9.3. The upstream displacement effect in the
subsonic part now has such a strength that the shock wave at the outer edge of
the boundary layer is split into two oblique shock waves, the lambda shock. At
the foot of the upstream shock wave an isentropic compression fan is present. The
supersonic flow pocket is much larger than that of the weak interaction case. From
the triple point shown in the figure a slip line ensues, in the form of a vortex sheet
like shown in Fig. 4.5.
As in the case of weak interaction, downstream of the interaction domain the
boundary layer thickness has increased and that to a higher degree. The wall shear
stress is reduced more severely, the boundary layer increasingly is more prone to
suffer separation at the rear of the airfoil, i.e. trailing-edge separation. Increasing
the pre-shock Mach number would move forward this separation until it directly
begins at the interaction domain.

Regarding the thickening of the boundary layer downstream of the interaction

domain we note that another interesting effect exists. All boundary-layer thicknesses
depend on the inverse of some power of the Reynolds number: δi ∝ Re−n , Appendix
A.5.4. In [3] it is shown that across a normal shock wave the unit Reynolds number
always decreases. Behind the interaction domain this effect alone would lead to an
increase of the boundary-layer thickness. Whether and how this effect in our case
plays a role is not known.
Summarizing we note that shock-wave/boundary-layer interaction leads to several
adverse effects. They are, on the one hand the increase of drag due to the increased
form drag and in addition the wave drag, on the other hand the reduction of lift due
to boundary-layer and shock-wave decambering, Sect. 6.1.
9.1 Supercritical Airfoil—Shock-Wave/Boundary-Layer Interaction 237

Fig. 9.4 Lift coefficient C L : schematic of separation and interaction effects as function of the flight
Mach number M∞ [2]

If trailing-edge separation occurs, respectively total separation behind the shock

wave, drag divergence happens, Sect. 2.4.2. For the laminar wing weak and strong
interactions are of particular importance, because the laminar boundary layer does
not accept as much adverse pressure gradient as the turbulent one.
Figure 9.4 schematically shows how the lift coefficient is affected in the different
Mach-number domains.2 In the subsonic domain the maximum lift coefficient is
reached when total separation, or wing stall happens. In the transonic domain first
the lift is reduced and once total separation behind the shock wave is present, the
strong transonic lift dip occurs, which goes along with unsteadiness of the flow field:
transonic buffet. Interaction with aeroelastic effects of the wing then leads to the
transonic buffeting. In aircraft operations transonic buffet and in particular transonic
buffeting limit the flight Mach number. The buffet onset boundary in the figure is
indicated as design point.
The supercritical airfoil is to reduce shock-wave/boundary-layer interaction
effects by minimizing the overspeed—the increment of speed over the freestream
speed—in particular at the suction side of the wing. Weak and strong interaction
effects in this way are pushed to higher flight Mach numbers. For the basics of inter-
action effects see, e.g., [1], for the implication for the transonic wing design, e.g., [2].
The reader interested in the respective early German developments will find ample
information in [5].

2 The figure goes back to F. Thomas [4].

238 9 Particular Flow Problems of Large Aspect-Ratio Wings

Both the experimental and the numerical simulation of shock-wave/boundary-

layer interaction and transonic buffet are highly demanding. The experimental inves-
tigation needs both Mach-number and Reynolds-number similarity. Wind tunnel
test-section effects must be overcome.
The numerical simulation faces the problem of turbulence modeling in unsteady
flow and in the presence of shock waves.3 Numerical studies in particular of the
buffet onset have shown that obviously simulations based on the RANS or URANS
equations (Model 10 of Table 1.3) are not suited well. In [6] it is demonstrated that
zonal detached-eddy simulation (Model 11), rather than RANS and URANS and
the standard detached-eddy simulation, permits to reproduce the self-sustained flow
oscillations on the airfoil.
In [7] results of delayed detached-eddy simulation (Model 11) of transonic buf-
fet and URANS simulation are presented. The former successfully simulates the
unsteady shock-wave/boundary-layer interaction. Alternate vortex shedding and a
spanwise undulation are present. Near the trailing edge the flow unsteadiness is
overestimated.
Results of experimental investigations in order to provide test cases for numerical
simulation approaches are reported in [8]. The authors observe that the experimental
results suggest that the transonic buffet essentially is a two-dimensional mean-flow
phenomenon, although three-dimensional effects are present.
URANS simulations of transonic buffet on different wing shapes are published in
[9]. The phenomenon is studied on two-dimensional, infinite swept and finite-span
wing configurations. For small wing sweep angles the mechanisms are like those for
two-dimensional flow. For moderate sweep angles a propagation of lateral pressure
disturbances was found, which is not present in two dimensions. At high-sweep
wings, when the wing becomes stalled, shock buffet has vanished. Wing tip effects
are shown to be large at low-aspect ratio wings.
Much effort was and is spent on ways and devices to influence and reduce
shock-wave/boundary-layer interaction effects on supercritical wings. Passive and
active means in particular for laminar wings were investigated for instance in the
EUROSHOCK projects of the European Union in the late 1990s [10, 11].
The bump, a small distensible local contour modification in the shock region
mainly to reduce the strength of the shock wave, was found to be the most effective
device when drag reduction is the main driver. Benefits are present also regarding
transonic buffet. The problems come in when the bump must be adaptive to changing
flight conditions. The potential for turbulent flow is to be explored. The laminar wing
still is an important topic. Increasing ecological pressure will lead to more efforts in
research and industry.
The adaptive bump in turbulent flow with the function of both wave-drag reduction
and buffet control is a challenging subject, too. In [12] URANS investigations are
reported on the effect of adaptive two-dimensional bumps and of “smart vortex
generators” in the form of three-dimensional bumps. On a supercritical, unswept wing
section the effect of bump flow conditions, crest height, and streamwise positioning

3 If laminar-turbulent transition would be involved, the situation would be much more challenging.
9.1 Supercritical Airfoil—Shock-Wave/Boundary-Layer Interaction 239

of two- and three-dimensional shock control bumps was analyzed regarding buffet
behavior and overall performance.
Two-dimensional shock-control bumps were found to improve buffet behavior
thanks to the shock strength reduction with positive effects on flow separation. Three-
dimensional bumps yield the same buffet-affecting mechanisms. The finite spanwise
extent makes them less dominant. Also shown was that the strength of the vortical
wake of three-dimensional bumps can be tuned by appropriate bump shaping. The
tuned strength then correlates positively with delayed buffet onset.
For both laminar and turbulent flow control smart structures and materials are
required. Moreover the overall problem of drag reduction needs to be addressed:
sub-boundary layer devices, mass-less air jets, vortex generators as well as reversed-
flow flaps and mini flaps for the control of the flow at the trailing edge, which always
is of finite thickness, as was shown in Sect. 6.3.
In view of practical applications installation penalties and actual benefits are to
be identified. This is of large importance in order to chose the right course in the
efforts, see, e.g., [13].

9.2 Flow Past a High-Lift System

High-lift systems in the form of lift-enhancement surfaces are indispensable devices

in particular on transport aircraft wings. They permit to achieve the necessary airfield
performance, i.e., the needed high-lift capabilities for landing but also for take-off.
The overall demands are low approach speed, low take-off drag, all to be achieved
with an in total small and simple system with low weight and low complexity as
formulated to the point in [14, 15]. A large variety of leading edge and trailing edge
devices is in use, see, e.g., [16, 17] and also [18].
A high-lift system usually has a slat in the front and a single-slotted—or even
double- or triple-slotted—Fowler flap at the rear. The deployed flaps increase the
camber of the wing and hence the lift, which goes along with a reduction of the
lift-to-drag ratio. Other nose devises than the slat are the Krueger flap and the hinged
leading-edge droop-nose device (DND), see, e.g., [14, 15].
We are interested in the flow and separation phenomena, which are present at a
high-lift configuration with slat. Consider Fig. 9.5, which shows the basic configura-
tion of a two-dimensional high-lift airfoil and the arising phenomena. The gap or slot
between the slat and the main wing, as well as that between the main wing and the
Fowler flap are the devices, which make the system a viable one. The flow through
them, driven by the pressure difference between the lower (pressure) and the upper
(suction) side, prevents boundary-layer separation at the upper side of both the main
wing and the Fowler flap.
At the upper side of the slat due to the strong flow acceleration a supersonic flow
pocket arises despite the typical low free-stream/flight Mach number at which the
high-lift system is activated. The supersonic flow pocket is terminated by a nearly
normal shock wave with the associated shock-wave/boundary-layer interaction.
240 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.5 High-lift configuration and schematic of the flow phenomena (after [14]). Laminar-
turbulent boundary-layer transition is not indicated everywhere, but always is present

At the lower side of the slat a geometry-induced (cove-) separation is present.

From the trailing edge of the slat a shear layer or wake with kinematically inactive
vorticity content arises via flow-off separation. The situation is the same at the end
of the main wing, again with cove separation.
The flow over the Fowler flap at the rear is even more complex, because of the
flow through the rear gap. The critical location is that at the rear of the upper side of
the flap, where the flow is prone to ordinary, i.e., adverse pressure-gradient induced
separation. In all cases confluence of the different boundary layers leads to complex
flow situations.4
The two-dimensional flow picture shown in Fig. 9.5 changes for a real wing,
Fig. 9.6. Three-dimensional effects due to the finite wing span, the sweep of the wing,
spanwise segmentation and nacelle interactions reduce the efficiency of a nominally
two-dimensional high-lift system.
We illustrate with results of numerical simulations some of the flow and separation
phenomena arising at a high-lift system [19]. The results were obtained in the frame
of a research project of the European Union [20]. The topic of this project was the
accuracy and reliability of hybrid RANS-LES (Model 11) methods, in particular the
grey-area mitigation.
The test case was a three-element airfoil at M∞ = 0.15 and Rec = 2.094 · 106 ,
with c = 0.6 m being the length of the airfoil with retracted slat and flap. The angle of
attack was α = 6◦ . Figure 9.7 illustrates the expected complex separation phenomena
and simulation challenges.
Experimental data was obtained, too. Experimental uncertainties regarding flow
three-dimensionality, side-wall effects, laminar-turbulent transition and angle-of-

4 Theshear layers leaving the trailing edges of all elements of the high-lift system are also noise
sources due to the finite thickness of the edges, Sect. 6.3.
9.2 Flow Past a High-Lift System 241

Fig. 9.6 The wing movables of the A350 XWB-900 [15]. DND: hinged leading-edge droop-nose
device

Fig. 9.7 Geometry and salient flow features of the three-element airfoil [19]

attack corrections precluded the use of the obtained data as strict validation data.
The data is used rather as demonstration data.
Five European partners applied their zonal or embedded hybrid RANS-LES
(Model 11) methods to the test case.5 A few figures are shown in order to illus-
trate the obtained results.
Mean flow-field data was obtained by time averaging and by spatial averaging
in the homogeneous spanwise direction. Figure 9.8 shows the streamlines past the
configuration resulting from the different simulation approaches.
The streamline topologies show the same overall global behavior. Each of the
three wing elements has its own attachment point of the external inviscid flow. (We
note that the boundary layer at an attachment point has a finite thickness [21].) Below

5 We do not present the details of the numerical approaches. The interested reader is asked to consult

the original publication [19].

242 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.8 Mean flow-field streamlines achieved with non-zonal (left) and embedded (right) numer-
ical simulations [19]

Fig. 9.9 Mean surface-pressure coefficient distributions achieved with non-zonal (left) and embed-
ded (right) numerical simulations [19]

the wing elements the boundary layers are attached, as is the case also at the upper
side of the slat and the main wing.
At the upper side of the Fowler flap all solutions indicate separation and show
recirculation areas of different extensions. It must be kept in mind that the flow
patterns represent time-averaged data. The reality sees vortex shedding from the
upper side of the flap.
Surface-pressure coefficient distributions found by the partners, again time- and
span-averaged, are given in Fig. 9.9. Included too are experimental results. The com-
parison with them is affected by the uncertainties mentioned above. Nevertheless, it
appears that each of the simulations yields acceptable results.
At the pressure sides of all wing elements the computed pressure distributions
agree well with each other and the experimental ones. At the upper sides generally
9.2 Flow Past a High-Lift System 243

Fig. 9.10 Schematic of sub boundary-layer vortex generators (SBLVGs) [22]

the non-zonal approaches appear to be closer to each other and the experimental
values. This is not the case with the results of the embedded simulations, although
the differences are not large altogether. The in reality to be expected periodic vortex
shedding from the upper side of the Fowler flap of course is not reflected in the data.
Interesting is the question, how that flap separation can be influenced and, in gen-
eral, how the performance of a high-lift system can be enhanced. Several passive and
active means in this regard were investigated in the research and technology project
AWIATOR (Aircraft Wing with Advanced Technology Operation) of the European
Union and in the research project IHK (Innovative High-Lift Configurations) of the
German Aerospace Research Program [22].
Purely passive means of flow control are sub-boundary-layer vortex generators
(SBLVGs). They have a height of 0.8–0.9 of the boundary-layer thickness δ, i.e., they
are located in the ‘outer layer’ or ‘turbulent layer’ of the turbulent boundary layer.
Their height hence is smaller than that of ordinary vortex generators. Their location
is just ahead of the largest thickness of the Fowler flap, Fig. 9.10. When the flap is
retracted, they are below the shroud in the cove.
The SBLVGs basically are slender triangles. Typical dimensions, spacings and
orientations are given in Fig. 9.10. Like ordinary vortex generators, but with less
eigen drag, they produce streamwise coherent vortices of small scale. In this way,
fluid from the outer layer with large momentum is transported toward the wall,
effectively reducing the separation disposition of the boundary layer.
SBLVGs, being devices with low complexity, weight and cost penalty, have been
tested in wind tunnels and full-scale also in flight with an Airbus A340. The flight
tests have shown an increase of lift due to the larger flap angle—35◦ instead of 32◦ —
without separation and without a reduction of the aerodynamic efficiency L/D at
take-off. If separation is shifted to a higher flap angle, also buffet at the horizontal
tail plane is shifted.
The devices as add-on devices can increase the performance of a given under-
performing flap system. On the other hand SBLVGs have the potential to create
smaller and lighter Fowler flap systems. They are studied, too, as means for flow
control of engine inlets, see, e.g., [23].
244 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.11 Schematic of an adaptive Mini-TED device at a Fowler flap [22]

Active means to enlarge the lift of a take-off configuration are adaptive miniature
trailing-edge devices (Mini-TEDs), Fig. 9.11. They were tested on an A340 with a
length of 2 per cent clean-wing chord, i.e., 90–150 mm length [22].
A Mini-TED located at the very trailing edge with 90◦ deflection actually is a
Gurney flap. Such a flap enlarges the camber of an airfoil or wing. In [22] it is
reported that adaptive Mini-TEDs were shown to improve the performance of the
high-lift configuration of a large aspect-ratio wing, and to reduce drag and buffet
onset in cruise condition.
Mini-TEDs can be used to adapt wing twist in cruise, and, depending on the
spanwise location, for roll and spanwise load control. Wind-tunnel tests have shown
that also lift and the lift-to-drag ratio can be improved for a canard-delta fighter
aircraft configuration at low speed [22].
Adaptive elements and devices like Mini-TEDs, adaptive drooped hinge flaps
(ADHF), pulsed blowing—also at the wing’s leading edge—have been proven to be
effective devices to improve the performance of high-lift systems and to be means for
load control and other interesting applications. However, compared to the passive
SBLVGs they lead to a higher system complexity, as well as mass, volume and
energy increments. Their application on aircraft depends on a number of requirements
besides their actual effects.

9.3 The Wing in High-Lift Condition

We treat, following—very abbreviated—the discussion by R. Rudnik in [24], the

major vortex phenomena occurring at a swept, large aspect-ratio wing together with
the underwing engine in high-lift configuration. The configuration is typical for a
modern transport aircraft designed for transonic flight, Fig. 9.12.
Separation and vortex phenomena occurring at the clean lifting large aspect-ratio
wing were treated in Chap. 8. The flow past the isolated high-lift system was the
9.3 The Wing in High-Lift Condition 245

Fig. 9.12 Transport aircraft: locations, which are prone to ordinary separation [24]

topic of the preceding Sect. 9.2. Now we discuss the phenomena present at a swept,
tapered wing’s root, its tip and in particular at the nacelle of the underwing mounted
engine, all in presence of the employed high-lift system.
At the high-lift configuration in Fig. 9.12 four locations are indicated, where sep-
aration and vortex phenomena can be and are indeed present:

• (a) the wing-root area,

• (b) the region around the engine,
• (c) the trailing edge of the Fowler-flap system,
• (d) the wing tip.

Also in the background of our topic are the stalling characteristics of swept,
tapered wings. The wing’s sweep leads to an extra thickening of the upper-side
boundary layer at the outer wing section and there to a higher separation tendency.
This tendency is due to the wing upper and lower flow-fields shear, and is increased
due to the wing’s taper. If we have a parallel isobar pattern, or one near to it, [25],
the surface pressure gradients, both favorable and adverse, become the larger, the
smaller the local chord length is. This mainly concerns the upper side of the wing,
where the boundary layer then becomes prone to separation. If the stall begins at the
outer wing, the roll control by means of the aileron is no longer possible.
The stall characteristics of swept wings and the aerodynamic demands regarding
take-off and landing have led to the presently employed high-lift systems, which
permit a safe and controllable low-speed flight. A typical high-lift system consists of
slats or Krueger flaps along the wing’s leading edge and single-slotted or multiple-
246 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.13 Schematic of the vortex phenomena present at a large aspect-ratio wing in high-lift
configuration [24]

slotted Fowler flap systems along the wing’s trailing edge. The principle of a high-lift
system is shown in Fig. 9.5.6
We look now at the effects present at the full-wing configuration, which have
a significant influence on its high-lift performance. High-lift performance means
the achievable maximum lift coefficient and the stalling characteristics, which both
are strongly influenced by the engine installation. Also playing a role are the flow
phenomena at the wing root and the wing tip, and in particular also those at the lateral
edges of elements of the high-lift system.
Both ordinary and flow-off separation lead to vortex sheets and vortices, which
at and downstream of their location of origin influence the flow pattern, see the
considerations in Sect. 2.2.
We now investigate the occurrence and effects of separation and vortices at the
wing shown in Fig. 9.13.
The high-lift system is assumed to consist of a leading-edge flap and a trailing-
edge flap, the former being a slat and the latter a Fowler flap. These high-lift elements
in span direction are sub-divided into several elements, as shown in Fig. 9.6.7 Because
of the small gaps between the elements in span direction, they can be considered as

6 At propeller-driven subsonic aircraft with no or only small wing sweep usually no leading-edge
devices are employed. Also playing a role there is the favorable interference between the propeller
wake and the wing flow.
7 In our discussion we assume that instead of the droop nose device (DND) inboard of the engine

also a slat is present, like outboard of it.

9.3 The Wing in High-Lift Condition 247

uninterrupted. This does not hold for the slat at the position of the engine. Because
present-day bypass engines are mounted close to the wing, see the following Sect. 9.4,
the slat has a cutout at the location of the pylon.
Regarding the trailing-edge flap system, present-day aircraft usually have no
cutout behind the engine. This became possible with the single-slot flap system,
which did replace the older two-slot flap system. The former does not extend as far
downward as the latter. The end-to-end Fowler flap system of course has a partition
at the Yehudi break. In Fig. 9.6 this is indicated as inboard flap and outboard flap.
An important aspect regarding the trailing-edge flap system is that no interference
with the engine jet takes place. This demands a sufficient vertical downward location
of the engine and a proper setting angle of the engine against the horizontal.
We discuss the phenomena present at the four locations, and begin with the wing-
root area, location (a) in Fig. 9.12. As was shown in Sect. 8.4.3, at the clean wing’s
root separation is present, leading to a horse-shoe vortex. This is only the case, if the
wing is attached to the fuselage without a fairing. With a properly designed fairing
the flow does not separate at the wing’s leading edge and a horse-shoe vortex is not
present, Fig. 8.27.
In the high-lift configuration the situation is different, Fig. 9.13. Two vortices
are indicated at the wing’s root, where the extended slat leads to an interruption
of the leading-edge contour: the clockwise turning slat-horn vortex and the weaker
counterrotating onglet vortex. The slat horn has the purpose to create a strong slat-end
vortex. Both vortices reduce the separation inclination of the boundary layer along
the wing root.
At larger angle of attack large scale separation will begin at the rear of the wing
as indicated in Fig. 9.12, location (a). That separation is considered to be rather
uncritical, compared to separation beginning at the outer wing. The induced roll
moment is small, moreover the associated pitch-down moment reduces the angle of
attack. This is a good-natured stall behavior, which, however, should not occur too
early. Then the maximum lift capacity would be limited.
The complex flow situation at location (b) is due to the interference of the nacelle
flow with the wing flow in the high-lift configuration. At the edges of the slat cutout
two counterrotating vortices evolve. At the left-hand wing shown in Fig. 9.13 the
vortex at the inboard slat edge turns counterclockwise, at the outboard slat edge
clockwise. The slat cutout leads to a reduction of the maximum lift of the wing. This
reduction is only partly compensated by a lift increment due to the flow past the
nacelle. That flow results in the so-called nacelle vortices, which are counterrotating,
too. They come into being because of boundary-layer separation at the sides of the
nacelle.
With close-coupled engines at high angles of attack only the inboard nacelle vortex
passes over the wing, the other below it. The inboard nacelle vortex induces an upwind
above the wing and hence enhances the separation tendency there. The separation
zone is beginning behind the slat cutout, grows laterally and moves downstream with
an inboard tendency, as indicated in Fig. 9.13.
In that figure also a nacelle-strake vortex is indicated. It runs over the slat cutout
and is an effective means to eliminate the lift loss due to the close coupled high-
248 9 Particular Flow Problems of Large Aspect-Ratio Wings

bypass ratio engine. The mechanism of the nacelle-strake vortex is discussed and
visualized in the following Sect. 9.4.
Location (c) in Fig. 9.12 denotes the possible separation problem at the Fowler
flap. If the high-lift system works properly, flow-off separation happens at the trailing
edge of the flap. As discussed in the preceding Sect. 9.2, at the highly loaded flap
ordinary separation can happen upstream of the flap’s trailing edge. In that case of
course the effect of the high-lift system is reduced. Means to reduce the flap separation
were sketched in that section, too.
Location (d) finally indicates the problem area at the wing tip. The tip-vortex
system at a lifting large aspect-ratio wing is discussed in Sect. 8.4.3. Wing-tip devices
are the topic of Sect. 9.5. The problem at location (d) is due to the fact that the slat
generally ends—due to structural constraints of the wing—at a location somewhat
ahead of the wing-tip section. Consequently the effect of the slat is missing there
and separation already occurs at a smaller angle of attack than at the main part of the
wing. At the outer edge of the deployed slat a vortex is present, which influences the
separated flow regime and furthermore interacts with the tip-vortex system.
We have given only an overall discussion of the separation phenomena occurring
at a wing in high-lift condition. At what angle of attack and at what location of a
wing separation happens is a consequence of the layout of the high-lift system and
the engine installation of the aircraft under consideration, as well as the desired stall
behavior of that aircraft [24].

9.4 The VHBR Engine and the Nacelle-Strake Vortex

The still increasing bypass ratio of jet engines leads to increasing nacelle diameters.8
The underwing arrangement of the engines is limited by the needed ground clearance.
In order to avoid a mass critical extension of the landing gear the jet engines are
becoming ever closer coupled to the wing, see, e.g., [26] and also Fig. 3.18.
Consider the left part of Fig. 9.14. Schematically it shows a very high bypass-ratio
engine (8  VHBR  12) closely mounted below the wing. The ground clearance
is indicated. The slat is lowered and because of the engine’s proximity to the wing,
above the nacelle the slat has a cutout. The right part of the figure shows that this
cutout leads to the so-called nacelle-wake separation, see, e.g., [27]. This separation
is due to complex interacting flow phenomena at the nacelle and the slat cutout. The
result is a local wing stall, which reduces the high-lift effect and leads to an oversizing
of the high-lift system. The currently employed remedy for VHBR engines is a rather
simple device, the nacelle strake, which induces the nacelle-strake vortex, see e.g.,
[28].
The nacelle strake is a geometrically small device, placed at the inboard side
of the nacelle, where below the wing and between the fuselage and the nacelle a
channel effect is present. With a proper location the strake induces at and above the

8 The higher the bypass ratio, the lower are the fuel consumption and the noise emission.
9.4 The VHBR Engine and the Nacelle-Strake Vortex 249

Fig. 9.14 Schematic of an engine wing arrangement with lowered slat (after [29]) Left part: VHBR
engine and the slat cutout. Right part: the resulting nacelle-wake separation at the main wing part

required angle of attack the nacelle-strake vortex with its location behind the slat
cutout and above the wing. There the vortex induces a suction-pressure field, like a
lee-side vortex—or more in general, a side-edge vortex—, which in turn suppresses
the nacelle-wake separation. With a correctly placed nacelle strake the induced vortex
leads to the recovery of a large part of the lift loss due to the engine mounting.
The interaction of the nacelle/pylon/nacelle-strake flow with that of the high-
lift system, in particular the slat, is a topic of high importance. It was treated, for
instance, in the frame of the EUROLIFT II project of the European Union. In [27] it
was demonstrated for a high angle of attack case that RANS (Model 10) simulations
today are well suited to deal with the existing flow problems. Suitable mathematical
models, accurate numerical solutions and comprehensive experimental data are the
requirements for advances in wing and nacelle stall simulations [30].
The flow situation at a close-coupled flow-through nacelle with the nacelle-strake
vortex is shown in Fig. 9.15. The sense of vortex rotation is made visible with the
amount of the vorticity component ωx : red indicates the clockwise direction, blue the
opposite, as indicated at the upper left of the figure. Several longitudinal vortices arise
at the slat cutout. They interact with the flow at the upper side of the wing, which at
and downstream of the leading edge is highly three-dimensional. The nacelle-strake
vortex at the inboard side of the nacelle is well discernible.
In EUROLIFT II project the flow past the KH3Y configuration—representing a
modern transport aircraft—was studied [27], see also [31]. The configuration com-
plexity was increased in three steps from Stage 1 to Stage 3, Fig. 9.16.
The Stage 1 configuration is the baseline high-lift configuration with an onglet and
a slat horn, but without engine. The slat deflection angle is δs = 26.5◦ and the Fowler-
flap angle δ F = 32◦ . In Stage 2 a through-flow nacelle together with slat cutout is
added. The nacelle represents a modern high by-pass engine. Stage 3 finally includes
the nacelle strake.
RANS simulations and wind-tunnel measurements were made for a low and a high
Reynolds-number case below and at maximum lift conditions. We first look at results
of the low Reynolds-number case. The free-stream Mach number is M∞ = 0.176
and the Reynolds number Rel = 1.33 · 106 .
250 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.15 Vortices at a close-coupled flow-through nacelle at pre-stall condition [29]

Fig. 9.16 Extension stages of the KH3Y configuration (right-hand wing) from a baseline high-lift
configuration (Stage 1) to the realistic full aircraft configuration (Stage 3) [27]

Figure 9.17 shows both the computed skin-friction lines and the distribution of
the wall shear-stress component c f x for the configurations Stage 2 and 3 at three
angles of attack. When c f x becomes negative—red color in the figure—, the flow is
considered to be separated.
At the lowest of the angles of attack no major separation phenomena are present. At
α = 15.92◦ separation develops at the inboard slat close to the pylon. With increasing
angle of attack this separation finally extends in spanwise direction and down to the
end of the main wing. When the nacelle strake is present, Stage 3, no slat separation
occurs at the three angles of attack and up to maximum lift. The beginning of lift
breakdown is then dominated by the separation at the 6th slat track. These RANS
results are in accordance with the wind-tunnel results.
We look now at some results for the high Reynolds-number case. The free-stream
Mach number is M∞ = 0.204, the Reynolds number Rel = 25 · 106 , and the angle
of attack α = 17.5◦ . The slat deflection angle and the Fowler-flap angle are the same
as for the low Reynolds-number case.
9.4 The VHBR Engine and the Nacelle-Strake Vortex 251

Fig. 9.17 Low Reynolds-number case (right-hand wing): numerically predicted wall shear-stress
distributions and skin-friction line patterns for Stage 2 (left) and Stage 3 (right) at three angles of
attack [27]:  0.00 (red)  c f x  0.011 (blue)

The nacelle-strake vortex effect, visualized with the skin-friction line patterns
and the c f x -distribution computed for Stage 2 and Stage 3, is shown in Fig. 9.18. We
look at the nacelle at the right-hand wing. The c f x -distribution is coded such that red
color indicates negative values, which means that the flow is separated. Blue color
indicates a high wall-shear stress.
The visualization of the associated vortex phenomena is presented in Fig. 9.19.
The color denotes the amount of the vorticity component ωx . Red indicates clockwise,
blue counter-clockwise rotating, see also Fig. 9.15.
The combined look at the two figures for Stage 2 shows at the inboard side of
the nacelle the convergence of the skin-friction lines and a strong counter-clockwise
rotating vortex. A vortex filament, clockwise rotating, emerges from the upper side
of the inboard slat. At the right-side end—looking at the figure—of the outboard slat
a counter-clockwise rotating vortex emerges.
On the main-wing element from the left to the right in the figure the skin-friction
lines indicate a separation line, bending away to the left, an attachment line, which
tapers off soon, and in the large red patch possibly also a vortex filament leaving the
surface. The red patch indicates the large separation region, where the high surface
252 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.18 The effect of the nacelle-strake vortex on the skin-friction line pattern and the c f x -
distribution for the high Reynolds-number case at the right-hand wing [27]:  0.00 (red)  c f x 
0.011 (blue). Left part: Stage 2 (no nacelle strake), right part: Stage 3 (with nacelle strake)

Fig. 9.19 The vortex visualizations for the high Reynolds-number case at the right-side wing [27]:
Left part: Stage 2 (no nacelle strake), right part: Stage 3 (with nacelle strake)

pressure leads to the reduction of the high-lift effect. Also seen on the upper side of
the pylon is a short separation line, leading to the small counter-clockwise rotating
vortex above the pylon.
For Stage 3 the strong counter-clockwise rotating strake vortex already changes
the skin-friction line pattern at the inboard side of the nacelle. The vortex filament
separation at the upper side of the inboard slat has disappeared. Only small changes
are present at the right-side end of the outboard slat. The pylon vortex still is indicated.
On the main wing, due to the presence of the nacelle-strake vortex, the skin-
friction line pattern has changed completely. It now indicates, together with the c f x
distribution, that the flow basically is attached. From the left to the right in the figure
we see a separation line, an attachment line, and a second separation line. The color
of the c f x -distribution in each case qualitatively confirms this: low values along
9.4 The VHBR Engine and the Nacelle-Strake Vortex 253

the separation line, high values along the attachment line, there also present—not
shown—a relative maximum of the wall pressure.9 Unfortunately no surface pressure
distributions are available, so that the effect of the nacelle-strake vortex in this regard
cannot be demonstrated.
With current VHBR engines the nacelle-strake vortex is sufficient to suppress
the nacelle-wake separation. This passive means is no more effective for ultra high
bypass-ratio engines (UHBR  12), which presently are beginning to come into
service. In order to achieve the needed ground clearance, such engines are even closer
coupled to the wing than the VHBR engines. The slat cutout then becomes too large
for the passive nacelle strake to be effective. Active means, like sealing elements,
continuous or pulsed slot blowing etc. are being studied or being introduced, see
for example the discussion and reporting in [29]. From the side of aircraft design of
course system complexity problems arise with such means, which are not present
with passive means.

9.5 Wing-Tip Devices

In the introduction to Chap. 8 we have discussed the matter of induced drag. The
larger the aspect ratio Λ of a wing is, the smaller is the induced drag Di . The induced
drag is part of the total drag of an aircraft. Its magnitude depends on the given flight
phase and in cruise flight at C L /C D |opt can be up to 50 per cent of the total drag [32].
The span and hence the aspect ratio of the wing of a transport aircraft at least is
restricted due to the ramp or apron size at an airport, for instance (length × span)
60 m × 60 m. This then is an incentive to use wing-tip devices in order to compensate
the restricted wing span. Another incentive would be to improve the aerodynamic
performance of a given aircraft by adding wing-tip devices.
Wing-tip devices in their earliest forms were the end-plates, located at the tips
of the wing or of the horizontal tailplane. The idea behind it was—as one can still
find in the literature—that the device hinders the relief of the high pressure at the
underside of the wing around the wing tip to the upper side with its low pressure.
In this picture the corresponding flow around the wing tips is the cause of the “tip
vortices”, which on their part are the cause of the induced drag.
In the background obviously Prandtl’s lifting-line model prevails. Even today
one can find in widely used text books that no clear difference is made between the
trailing vortex and the tip vortex. The trailing vortex evolves due to the wing upper and
lower flow-fields shear, Sect. 4.3.2. At the lifting large aspect-ratio wing it emerges
out of the trailing vortex layer behind the wing, being present in full only some wing
half-span distances behind the wing’s trailing edge, see for example Fig. 4.18, which
shows the situation as it appears in the frame of potential-flow (Model 4) theory.

9 Attachment and separation lines are singular lines, for the related flow properties see Sect. 7.3,
respectively [21].
254 9 Particular Flow Problems of Large Aspect-Ratio Wings

The tip vortex, or the tip-vortex system, as we have seen in Sect. 8.4.3, comes
into being due to ordinary separation at the upper side of the wing-tip area. The flow
there of course passes from the lower side of the wing around the wing tip to the
upper side. There, depending on the given pressure field the separation happens. A
wing-tip device indeed influences this process as well as the trailing vortex layer
behind the wing.
As mentioned above, the induced drag of a wing decreases as the wing’s aspect
ratio increases. Ground-operational demands restrict wing spans. A wing-tip device
in form of a winglet or a sharklet is a means to compensate a span restriction.10
Other applications are the adaptation/extension of a given wing in order to improve
the aerodynamic performance.
The shape of a winglet or sharklet hence is designed to reduce the induced drag
and at the same time changes the strength of the tip vortex, or tip-vortex system. In
Sect. 8.4 it was seen for the investigated case that the circulation of the tip vortex is
about seven per cent of the total wake circulation. Hence a reduction of the strength
of the tip-vortex system would have an effect, but whether positive or negative is not
clear over the large range of operating conditions.
A wing-tip device of course leads to an increase of the viscous drag of the wing,
i.e., the skin-friction drag and the form drag increases. If the extension leads to
a corner flow situation, interference effects and shock-wave drag do occur in the
transonic flight domain. The device further increases the wing’s mass. Hence the
sought effect of a wing-tip device is reduced by these effects, see below.
The purpose of this section is to give the reader an overview of some aircraft-
relevant issues associated with wing-tip devices. No review is intended of the
approaches and the vast literature regarding wing-tip devices. In-depth studies of the
fluid-mechanical situation at such devices, like that for the CRM wing in Sect. 8.4.3
regarding the wing tip-vortex system and its interaction with the trailing vortex layer,
are not available.
A wing-tip device must not only be seen in view of a reduction of the induced
drag. For the aircraft it has other effects, too. In order to obtain an overall picture,
A. Büscher in his doctoral thesis uses the concept of the “equivalent drag reduction”
relative to a wing with planar wing tip [34, 35].
The equivalent drag reduction coefficient ΔC Dequiv as an objective function for
design work has five terms:

ΔC Dequiv = ΔC Daer o + ω · ΔC Dwr bm + ΔC Dmass + ΔC Dtrim + ΔC Dmech . (9.3)

ΔC Dequiv being negative indicates a reduction of the equivalent drag.

10 The classical winglet is a vertical surface at the wing tip, whereas now wing-tip extensions evolve

smoothly out of the wing’s surface. Sharklet so-called are winglets solely at the aircraft of the A320
family of Airbus, see, e.g., [33].
9.5 Wing-Tip Devices 255

Basically it is assumed that for a given aircraft, respectively wing configuration,

a wing-tip device is added. For some of the terms the reference configuration of the
wing tip is the Küchemann wing tip. The five terms are summarily presented.

1. ΔC Daer o represents the effect of the wing-tip device as such, i.e., the net aerody-
namic drag reduction. It consists of three or four parts: due to the wing-tip device
the reduction of the induced drag (negative sign), due to the device’s surface,
trailing edge, and volume the increase of skin-friction drag, form drag, and at
transonic flight wave drag (all three with positive sign).
2. ω · ΔC Dwr bm : if a wing-tip device is added, it leads to an increase of the bending
moment along the wing and hence also of the wing-root bending moment (wrbm).
This leads to an adjustment of the wing structure and with that to an increase of the
wing’s mass. Hence the empty mass of the aircraft is increased. If the maximum
take-off weight is to be kept, despite the reduced fuel demand limitations of either
the disposable load or the operational range of the aircraft are the consequence.
The mass-increase effect summarily is taken into account by the term ω. This
term is discussed below.
3. ΔC Dmass is due to the mass of the wing-tip device and a possible extra wing mass
increase due to its attachment on the wing.
4. ΔC Dtrim is an increment of the trim drag due to the change of the pitching moment
of the aircraft. It consists of two parts. The first represents the change of the
pitching moment due to the wing-tip device, the second its change due to the
changed induced drag.
5. ΔC Dmech is due to the extra mass, if movables at the wing-tip device are used at
low speed to improve the high-lift performance of the device.

The factor ω (0  ω  1) is a weighting factor between the empty mass of the

aircraft and the drag, called ‘wing-root bending moment weighting factor’. Its value
depends on the chosen design objective of the aircraft layout: (a) direct operating
costs (DOC), or (b) maximum take-off weight (MTOW), or (c) flight range (RANGE).
Hence three scenarios can be considered in which two of the factors are fixed and
the third one is variable. The determining component of the aircraft is the wing. It is
the only aircraft component, which directly reacts on the design modifications with
drag and mass changes.
In the first case the range is the variable parameter. If the MTOW is not changed,
only a small drag reduction is necessary in order to reduce the fuel consumption so
much that the DOC do not rise. Hence ω = ω R AN G E is small. The resulting design
has a high aerodynamic quality.
If, however, due to an increase of the empty mass the MTOW can be increased,
the situation is different. Due to the direct coupling to the DOC at constant RANGE
the MTOW increase can only be restricted, if the drag reduction is much larger than
in the first case. Consequently ω = ωmtow is larger for this case.
The third scenario of a drag reduction demands constant MTOW and RANGE.
In order to reach this despite the rise of the empty mass a large reduction of fuel
consumption is necessary, which only can be achieved with a large drag reduction.
256 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.20 The long-range

aircraft reference
configuration [34]

Fig. 9.21 Schematic of the

long-range configuration and
the spanwise partition in
LIDCA [34]

The resulting reduction of the DOC is ensured with a high ω = ω D OC , which leads
to a more mass-optimized design.
The considered wing-tip devices in Büscher’s thesis are based on a high ω. Hence
the factor is referred to as ω D OC . Part of the parameter study was made with the
smaller factor ω M T O W . The change of the factor implies a reduction of the root-
bending moment of about 40 per cent.
We look now at some overall results. The study was made for a long-range research
configuration with Küchemann wing tip, but without nacelles at high-speed condi-
tions and also for a high-lift take-off configuration with nacelles and extended slats,
flaps and ailerons. The long-range research configuration is shown in Fig. 9.20. The
flight Mach number was M∞ = 0.85, the Reynolds number Re L = 54.2·106 , with L
= 11.5 m, and the lift coefficient C L = 0.5, equivalent to cruise flight.
The study was based on the lifting-line method LIDCA (Model 4), which includes
a database module for airfoil data. The method was validated with RANS (Model
10) computations. The LIDCA configuration is shown in Fig. 9.21.
The aspect ratio of the wing is Λ = 8, the semispan 39.5 m. Changes of the wing
geometry due to the wing-tip device were restricted to outside of 95 per cent of the
wing’s semispan. That location is indicated in the lower left part of Fig. 9.21. The
9.5 Wing-Tip Devices 257

Fig. 9.22 Visualization of

the most effective wing-tip
devices together with the
Küchemann wing tip and the
Large Winglet [34]

resulting semispan of the wing was restricted to 42.5 m, and the extension of anhedral
devices in downward direction to 2.75 m, measured from the attachment location.
The following results concern the four most effective wing-tip devices for the
long-range case, which were found with LIDCA: concept 1, 2, 4, and 5 [34]. RANS
computations were made in order to check the results. Figure 9.22 shows the concepts
1, 2 and 5, the last one being a sharklet concept. Concept 4 is the same as concept
1, but with a smaller sweep angle of the 25 per cent line—40◦ instead of 45◦ —and a
reduced wing-tip area. Both are anhedral devices with the angle ν = −20◦ . Concept
2 has the dihedral angle ν = 10◦ . The figure in addition illustrates the Küchemann
wing tip of the reference concept and the Large Winglet, which also was considered
in the study.
The results in terms of the equivalent drag reduction obtained for the four concepts
are given in Fig. 9.23. We note that the results for all concepts were found with the
high wing-root bending moment weighting factor ω D OC . The figure contains two
basic results: (1) concept 4 leads to the largest equivalent drag reduction; (2) the
RANS solutions, which were made for the validation of the LIDCA results, confirm
all LIDCA results and in particular also that for concept 4. The RANS solutions
throughout yield a higher drag reduction for all concepts, with a delta of about 0.3
per cent.
The decomposed LIDCA results in Fig. 9.24 yield the following outcome: (1) for
concept 1 and 2 each of the increments of ΔC Daer o , ΔC Dwr bm and ΔC Dmass have almost
the same magnitude. The slightly higher aerodynamic increment of concept 1 is due
to the higher absolute value of ν and the larger effective span width; (2) concept
4 has the smallest increments of all concepts. The somewhat smaller aerodynamic
increment is compensated by the smaller wing-root bending moment increment and
the smaller mass increment; (3) concept 5 has the highest aerodynamic increment,
but also the highest increments of the wing-root bending moment increment and the
device mass, hence it has the smallest equivalent drag reduction coefficient, which,
however, is near to that of concept 2.
These results, which in a highly condensed manner present the total results of [34],
show the merit of wing-tip devices as non-planar wing tip shapes. Their advantages
258 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.23 The equivalent drag reduction coefficient ΔC Dequiv for the four most effective wing-tip
concepts [34]

Fig. 9.24 Decomposition of ΔC Dequiv for the four concepts [34]

are the simple geometry and the rather unproblematic integration at a new or an
existing wing in the sense of a retrofit solution without a degradation of the wing’s
performance at other conditions.
It appears, however, that if the wing span is not restricted, a wing-tip device has
no advantage over a planar wing extension. If, however, the wing’s span is restricted,
or the performance of a given wing is to be improved, non-planar wing tip shapes
definitely have benefits for the aircraft. The wing-tip device can have a positive or a
negative dihedral, i.e., an anhedral, both show the desired effect. The devices today
usually have the positive dihedral, since that does not lead to restrictions regarding
the ground clearance.
Restrictions in span due to the ramp dimensions can be overcome in another way,
which, however, up to date was not a viable one. But now Boeing with its 777X
9.5 Wing-Tip Devices 259

series introduced the—upward—foldable wing extension.11 On ground the span of

the aircraft is 64.8 m, in flight 71.8 m. The folding mechanism with its actuator
of course has its own mass, which acts against the benefit of the extension. The
energy demand certainly is small, but the overall system complexity of the aircraft
is increased.

9.6 The Wake-Vortex Hazard: The Problem and Means to

Control It

As we have shown and discussed in Chap. 8, the trailing vortex layer of the lifting
wing rolls up into the pair of trailing vortices. In this process it absorbs the two wing-
tip vortex systems, the vortex sheets and vortices originating from the fuselage, the
tail unit, the engines, flap tracks etc. and last but not least the counterrotating trailing
vortex layer from the trim surface. The resulting flow field summarily is called the
aircraft vortex wake. This wake can pose the so-called wake-vortex hazard for other,
following aircraft [36, 37].
Potential flow theory (Model 4 of Table 1.3) can be used to achieve basic insight
regarding this hazard. The theory yields for an elliptic wing-circulation distribution
that the strength of the trailing vortex Γ0 is proportional to the wing’s lift, Sect. 3.16,
and hence to the mass of the aircraft. This result basically holds also outside of the
range of validity of potential flow theory.
Behind the aircraft we see the situation sketched in Fig. 9.25. The vortex wake
is marked by the two trailing vortices, a downwash between them and an upwash
at their outer sides. When another aircraft enters the aircraft wake or flies in it, it
meets the wake-vortex hazard. That manifests itself in increased air turbulence, which
leads to increased dynamic structural loads, as well as passenger discomfort. Further
happens a loss of lift, and a rolling moment may be induced, too. The hazard may
arise everywhere on the flight path, but in particular during take-off from the runway,
during approach to and landing on the runway.
The wake-vortex hazard became a larger topic when the Boeing 747 entered
service in the beginning of the 1970s. With a maximum take-off mass (MTOM) of
up to 448 t it was much heavier than the Boeing 707 with at most 142 t MTOM.
About 35 years later, when the Airbus A380 with about 570 t MTOM was to enter
service, the wake-vortex hazard again attracted much attention.
Starting already in the early 1970s comprehensive investigations were dedicated
to reduce the wake-vortex hazard in terms of diminishing the wake field impact on
a follower aircraft, in particular the induced rolling moment.
Strategies to minimize the wake vortex hazard can be divided into two categories
[38]. In the first category, the focus is on a Low Vorticity Vortex wing (LVV) design,
which reduces the wake vortex hazard by enhancing the dispersion of the vorticity
field. After the roll-up of the trailing vortex layer is completed, the resulting trailing

11 Airbus studies such and alternative foldable solutions.

260 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.25 Schematic of the aircraft’s wake with the trailing vortex pair and the related downwash
and upwash [37]

Fig. 9.26 Wake alleviation

concepts [39]: change from
the high to the low vorticity
vortex (LVV)

vortices should feature a larger core size and smaller swirl velocities at the core
radius, Fig. 9.26 [39]. Consequently, the induced rolling moment is diminished.
In the second category, the focus is on a Quickly Decaying Vortex (QDV). An
enhanced wake vortex decay may be achieved by promoting three-dimensional insta-
bilities by means of passive or active devices. Particularly, the growth rates of the
long wave “Crow” instability, [40], have to be increased significantly, Fig. 9.27, e.g.,
9.6 The Wake-Vortex Hazard: The Problem and Means to Control It 261

Fig. 9.27 Wake alleviation concepts [39]: the quickly decaying vortex (QDV). t ∗ : non-dimensional
time, T ∗ : non-dimensional time up to vortex linking

by the interaction of multiple vortex systems or active excitation. The Crow insta-
bility is a sinusoidal vortex-pair instability due to an initial disturbance, Sect. 3.11.3.
The vortex trajectories are deflected laterally and finally connect, forming a chain of
vortex rings, which typically happens in Domain 4 behind the aircraft, Fig. 8.1.
A variety of configurative measures has been investigated for the implementation
of these concepts [39]. Considering the LVV strategy, wing control surfaces, such as
spoilers, and modified variants, e.g., flap edge elements, are used to create areas of
highly turbulent flow aimed to expand the core size of the trailing vortices [41].
Also, a modified wing-load distribution may minimize the induced rolling moment
for a following aircraft. The alteration of the circulation distribution of the wake-
generating wing can be obtained using for instance differential flap or spoiler settings.
It has been shown that a wing with an outboard partially deflected flap and an inboard
fully deflected flap produces—at least in the extended near field—a smaller induced
rolling moment than a wing with a standard flap setting [39].
Different measures can also be attributed to the QDV approach. Because a multiple
vortex system shows instabilities, which can grow more rapidly, passive devices aim
to promote these kinds of instabilities through the deliberate production of distinct
vortices in addition to those coming from the wing tip and the flap edge.
The production of additional distinct vortices can also be achieved by differential
flap setting. The efficiency of these concepts depends on the persistence of such
additional vortices, which is determined by configurational details of the aircraft.
Particularly, active devices are considered as a possible powerful means to amplify
262 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.28 Examples of passive means for vortex control—spoilers and flap elements [39]: a four
engine large transport aircraft in approach configuration, b delta spoiler at the wing’s upper side at
the outboard nacelle station, c differential flap setting

wake vortex instabilities. The operational implementation, however, would run into
the problem of heightened system complexity.
Passive and active means to alleviate the wake-vortex hazard are now sketched in
some detail. Generally it must be observed that the system complexity connected to
alleviation means, is an obstacle for the implementation of such means.
• Passive Means. Viscous and convective mechanisms are used to spread the
vorticity field of the trailing vortex over a wider spatial area, reducing peak values
in axial vorticity and circumferential velocities, Fig. 9.26.
The radial transport of vorticity and thus the expansion of the vortex core is
supported by increasing the turbulence intensity of the rolling up vortex layer and/or
in the formation area of concentrated single vortices. Flap edge devices, spoiler
elements and wing fins are applied to produce turbulent wakes enhancing the turbulent
mixing in specific regions of the wake-vortex near field [42, 43].
As a first example, the effect of delta-shaped spoiler elements has been studied on
a four engine large transport aircraft model (scale 1:20) for the high-lift approach con-
figuration, compare Table 1.1. Wind and water tunnel experiments were performed,
Fig. 9.28a and b, [39, 44]. Regarding the full-scale case this configuration belongs
to the category of a heavy leader aircraft (>136,000 kg take-off weight).
9.6 The Wake-Vortex Hazard: The Problem and Means to Control It 263

The spoiler elements are used to create a highly turbulent flow in the region of
outboard flap-vortex shedding, and help the merging of the outboard flap and outboard
nacelle vortex. These vortices contribute significantly to the formation of the rolled-
up, final trailing vortex. The concentrated region of highly turbulent flow, created
by the delta-spoiler burst leading-edge vortices and the turbulent wake emanating
from the spoiler trailing edge, affects the merging area of the outboard flap and
outboard nacelle vortices. It is aimed to enlarge the radial vorticity distribution, also
influencing the region of the free circulation center.
During the roll-up process, with the merger of the outboard flap and outboard
nacelle vortices, the highly turbulent flow is fed into the core region of the main
trailing vortex. Relative to the reference case (no spoiler deflection), a marked dis-
persion of the vorticity field is achieved in the wake-vortex extended near flow field—
Fig. 8.1—resulting in a reduction of the axial peak vorticity.
A significant expansion of the viscous core by 50–90 per cent occurs accompanied
by a reduction in maximum induced velocities of about 50 per cent. The alleviation
in the maximum induced rolling moment acting on a follower aircraft of the light
weight category (<7,000 kg take-off weight) is about 30 per cent. This effect has
been also documented to some extent for the wake vortex far field. The use of the
delta-type spoilers affects only slightly the overall flight performance.
The second example of a passive measure concentrates on influencing the wake
by deploying the inboard and outboard trailing-edge flaps at different angles, called
differential flap setting (DFS), Fig. 9.28c. Increasing the number of dominant near
field vortices raises also the number of merging processes throughout the roll-up
process, thus enlarging the main vortex core and reducing peak cross-flow velocities.
Again, investigations have been conducted on a four engine large transport aircraft
model in high-lift approach configuration [39].
The differential setting of inboard and outboard trailing-edge flaps results in an
inboard or outboard wing loading. Typically, the angle of attack must be increased to
keep the lift coefficient at the level of the reference case. Differential flap setting is
associated with the shedding of additional vortices, namely at the outboard side edge
of the inboard flap and the inboard side edge of the outboard flap, respectively. These
vortices are counter-rotating to each other while their respective strengths depend on
the flap deployment angle.
The vortices influence the sequence of vortex merging in the extended near field,
so that the persistence of a multiple vortex system is supported over a larger dis-
tance downstream. An inboard wing loading leads to a main vortex of less strength
and enhances its merging with the wing-tip vortex. The reduction in the maximum
induced rolling moment with respect to a light-weight class follower aircraft in the
extended near field is up to 48 per cent. The outboard loading results in a vortex
topology similar to that of the reference case. Due to the interaction of the additional
vortices a reduction in the maximum induced rolling moment of about 44 per cent
is shown for the extended near field. To a lesser extent this effect is also measurable
in the far field.
• Active Means. In the mid 1970s, theoretical and experimental studies dealt
with the influence of sinusoidally moving flaps on the wake vortex behavior and
264 9 Particular Flow Problems of Large Aspect-Ratio Wings

Fig. 9.29 Example of active means for vortex control: winglet flaps performing harmonic oscilla-
tions [47]

predicted a reduction in the wake life span by a factor of three [45]. In the last years,
such means were carefully re-investigated.
An active system was proposed by J.D. Crouch et al. based on periodic oscil-
lations of control surfaces, for example, ailerons and flaperons [46]. The perturba-
tions influence the vortex wake to trigger inherent instabilities which, after sufficient
amplification, may cause an earlier breakup of the trailing vortices into vortex rings
resulting in a rapid decay (QDV concept). Oscillating devices may include inner and
outer ailerons [46], trailing-edge flaps [45], and winglet flaps [47], Fig. 9.29.
The presence of instability mechanisms propagating in streamwise direction along
the vortex wake can lead to a relevant distortion of the trailing vortex pair, accelerating
its decay and dispersion. Velocity fluctuations caused by turbulence or fed actively
into the vortex wake represent inhomogeneous forcing terms with respect to stability
theory.
The influence of turbulence on wake-vortex life spans was investigated, e.g., by
S.C. Crow and E.J. Bate [48]. Regarding linear stability theory, the velocity pertur-
bation at the core of the trailing vortex is then composed by the component due to
mutual induction and the component related to superimposed fluctuations. Introduc-
ing such fluctuations at the frequency of the inherent instability may significantly
reduce the time for vortex linking, Fig. 9.27.
9.6 The Wake-Vortex Hazard: The Problem and Means to Control It 265

Rapid wake decay will then start within a shorter trailing distance compared to the
non-perturbed case. The forcing frequency is adapted to the wave number associated
with the highest amplification rate of the considered instabilities. The wavelength
of the Crow-type instability obtained by linear stability analysis is approximately
λCr ow ≈ 8 b0 [40].
The quantity b0 is the lateral separation distance of the vortex pair, with b0 = 0.76–
0.78b for a typical transport aircraft, see also Sect. 4.4. Thus, the wavelength λCr ow
corresponds to a reduced frequency of kCr ow ≈ 0.08, with k = f (b/2)/U∞ and
f = u ∞ /λCr ow . The characteristic reduced frequency kCr ow is then used to define
the excitation frequency related to the oscillating devices for sub- or full-scale testing.
It has been demonstrated in model studies that by such means the time up to contact
of the trailing vortices could be halved with the subsequent exchange of circulation
via the symmetry plane and the beginning of wake-vortex decay. The potential of a
large winglet with two integrated trailing-edge flaps has been investigated in [47].
Harmonic oscillations of the two winglet flaps with 180◦ phase shift ensure the
temporal stability of the overall aerodynamic coefficients. Regarding the wake-vortex
extended near field, the results document an increase by a factor of 2–5 in the power
spectral densities of the velocity fluctuations at reduced frequencies corresponding
to the Crow-type instabilities. This increase in spectral energy significantly raises the
excitation level attributed to the inherent instabilities triggering an increase in their
amplification.

9.7 Problems

Problem 9.1 Breguet’s range equation: The lift-to-drag ratio times the Mach number
should be as high as possible. Discuss the limiting flow phenomena.

Problem 9.2 What are the configurational means to have M∞ L/D as high as pos-
sible, (a) for the wing, (b) for the whole aircraft?

Problem 9.3 What does a vortex generator produce? Would it work also in laminar
flow?

Problem 9.4 Is a sub-boundary-layer vortex generator (SBLVG) more effective in

a laminar boundary layer than in a turbulent one? Consider Fig. 2.1.

Problem 9.5 What is the effect of a wing-root/fuselage fairing?

Problem 9.6 Which relative saving in fuel mass Δ m F = (m F2 − m F1 )/m F1 in per-

cent is obtained by reducing the zero drag coefficient C D0 by 3 per cent as well as the
induced drag coefficient C Di by 2 per cent? The portion of the zero drag coefficient
is 80 per cent of the total drag coefficient C D . Steady level cruise flight is considered
for which range R, flight speed u ∞ , specific fuel consumption b and take-off weight
m T O W remain unchanged. The aircraft take-off weight is m T O W =310,000 kg and
266 9 Particular Flow Problems of Large Aspect-Ratio Wings

the fuel mass of the baseline case is m F1 =113,000 kg (Index 1: baseline case; index
2: case with reduced drag coefficient).

Problem 9.7 Sketch the spanwise circulation distribution Γ (y) for a wing with
vertically orientated winglets with respect to an optimum circulation distribution.

Problem 9.8 What is a reason for installing wing-tip devices the one pointing
upward and the other downward?

Problem 9.9 How large is the excitation frequency for an active system enhancing
wake vortex decay, e.g., performing oscillations of trailing-edge flaps, for (a) the full
scale case with u ∞ = 80 m/s and a wing span of b = 60 m, and (b) the wind tunnel
situation with a model of 1:25 scale and a test section velocity of u ∞ = 25 m/s?

References

1. Délery, J.: Transonic shock-wave boundary-layer interactions. In: Babinsky, H., Harvey, J.K.
(eds.) Shock-Wave Boundary-Layer Interactions, pp. 5–86. Cambridge University Press, Cam-
bridge (2011)
2. Vos, R., Farokhi, S.: Introduction to Transonic Aerodynamics. Springer Science+Business
Media, Dordrecht (2015)
3. Hirschel, E.H.: Basics of Aerothermodynamics, 2nd edn (revised). Springer, Cham (2015)
4. Thomas, F.: Entwurfsgerechte Tragflügelaerodynamik. In: Bericht uber die Sitzung des WGLR-
Fachausschusses fur Aerodynamik, - Entwurfsaerodynamik. Deutsche Luft- und Raumfahrt,
Mittellung 67-24 (1967)
5. Meier, H.U. (ed.): German Development of the Swept Wing–1935-1945. Library of Flight,
AIAA, Reston (2010)
6. Deck, S.: Numerical simulation of transonic buffet over a supercritical airfoil. AIAA J. 43(7),
1556–1566 (2005)
7. Grossi, F., Braza, M., Hoarau, Y.: Prediction of transonic buffet by Delayed Detached-Eddy
simulation. AIAA J. 52(10), 2300–2312 (2014)
8. Jacquin, L., Molton, P., Deck, S., Maury, B., Soulevant, D.: Experimental study of shock
oscillation over a transonic supercritical profile. AIAA J. 47(9), 1985–1994 (2009)
9. Iovnovich, M., Raveh, D.E.: Numerical study of shock buffet on three-dimensional wings.
AIAA J. 53(2), 449–463 (2015)
10. Stanewsky, E., Délery, J., Fulker, J., Geissler, W. (eds.): EUROSHOCK—Drag Reduction by
Passive Shock Control. Results of the Project EUROSHOCK, AER2-CT92-0049, supported
by the European Union, 1993–1995. Notes on Numerical Fluid Mechanics, Vol. 56. Vieweg,
Braunschweig Wiesbaden (1997)
11. Stanewsky, E., Délery, J., Fulker, J., Matteis, P. de (eds.): Drag Reduction by Shock and Bound-
ary Layer Control. Results of the Project EUROSHOCK II, supported by the European Union,
1996–1999. Notes on Numerical Fluid Mechanics and Multidisciplinary Design, NNFM80.
Springer, Berlin (2002)
12. Mayer, R., Lutz, Th, Krämer, E.: Numerical study on the ability of shock control bumps for
buffet control. AIAA J. 56(5), 1978–1987 (2018)
13. Thiede, P. (ed.): Aerodynamic drag reduction technologies. In: Proceedings of the
CEAS/DragNet European Drag Reduction Conference, 19–21 June 2000, Potsdam, Germany.
Notes on Numerical Fluid Mechanics, NNFM76. Springer, Berlin (2001)
References 267

14. Flaig, A., Hilbig, R.: High-lift design for large civil aircraft. AGARD-CP-515, 31.1–31.12
(1993)
15. Strüber, H.: The aerodynamic design of the A350 XWB-900 high lift system. Paper presented
at the 29th Congress of the International Council of the Aeronautical Sciences, St. Petersburg,
Russia, 7–12 Sept 2014 (2014)
16. Rudolph, P.K.C.: High-lift systems on commercial airliners. NASA Contractor Rep. 4746
(1996)
17. Wild, J.: Hochauftrieb. In: Rossow, C.-C., Wolf, K., Horst, P. (eds.) Handbuch der Luft-
fahrzeugtechnik, pp. 184–192. Carl Hanser Verlag, München (2014)
18. Reckzeh, D.: Multifunctional wing moveables: design of the A350 XWB and the way to future
concepts. Paper presented at the 29th Congress of the International Council of the Aeronautical
Sciences, St. Petersburg, Russia, 7–12 Sept 2014 (2014)
19. Probst, A., Probst, S., Schwamborn, D.: 3-Element Airfoil. In: C. Mockett, W. Haase, D.
Schwamborn (eds.): Go4Hybrid—Grey area mitigation for hybrid RANS-LES methods.
Results of the 7th Framework Research Project Go4Hybrid, funded by the European Union,
2013–2015. Notes on Numerical Fluid Mechanics and Multidisciplinary Design, NNFM134,
Springer, Cham, pp. 155–171 (2018)
20. Mockett, C., Haase, W., Schwamborn, D. (eds.): Go4Hybrid: grey area mitigation for hybrid
RANS-LES methods. Results of the 7th Framework Res. Project Go4Hybrid, funded by the
European Union, 2013–2015. Notes on Numerical Fluid Mechanics and Multidisciplinary
Design, NNFM134, Springer, Cham (2018)
21. Hirschel, E.H., Cousteix, J., Kordulla, W.: Three-Dimensional Attached Viscous Flow.
Springer, Berlin (2013)
22. Breitsamter, C.: Innovative High-Lift Concepts of the Integrated Research Programs AWIATOR
and IHK. LTH-AD 04 02 004 (2009)
23. Jiràsek, A.: Vortex Generator Modeling and its Application to Optimal Control of Airflow
in Inlet. Doctoral Thesis, KTH Royal Institute of Technology, Rep TRITA-AVE 2006:66,
Stockholm, Sweden (2006)
24. Rudnik, R.: Transportflugzeuge. In: Rossow, C.-C., Wolf, K., Horst, P. (eds.) Handbuch der
Luftfahrzeugtechnik, pp. 83–113. Carl Hanser Verlag, München (2014)
25. Küchemann, D.: The Aerodynamic Design of Aircraft. Pergamon Press, Oxford, : also AIAA
Education Series, p. 2012. Va, AIAA, Reston (1978)
26. Lengers, M.: Industrial assessment of overall aircraft driven local active flow control. Paper
presented at the 29th Congress of the Internatioal Council of the Aeronautical Sciences, St.
Petersburg, Russia, 7–12 Sept 2014 (2014)
27. von Geyr, H., Schade, N., van der Burg, J.W., Eliasson, P., Esquieu, S.: CFD Prediction of max-
imum lift effects on realistic high-lift commercial-aircraft configurations within the european
project EUROLIFT II. AIAA-Paper, 2007–4299 (2007)
28. Haines A.B.: Scale effects on aircraft and weapon aerodynamics. AGARD-AG-323 (1994)
29. Fricke, S., Ciobaca, V., Wild, J., Norman, D.: Numerical studies of active flow control applied at
the engine-wing junction. In: R. Radespiel, R. Niehuis, N. Kroll, K. Behrends (eds.): Advances
in Simulation of Wing and Nacelle Stall. Closing Symposium of DGLR Research Unit FOR
1066, 1–2 Dec 2014, Braunschweig, Germany. Notes on Numerical Fluid Mechanics and
Multidisciplinary Design, NNFM131, Springer, Cham, pp. 397–411 (2016)
30. Radespiel, R., Niehuis, R., Kroll, N., Behrends, K. (eds.): Advances in Simulation of Wing
and Nacelle Stall. Closing Symp. of DGLR Research Unit FOR 1066, Dec. 1–2, 2014,
Braunschweig, Germany. Notes on Numerical Fluid Mechanics and Multidisciplinary Design,
NNFM131, Springer-Verlag Cham Heidelberg (2016)
31. Wichmann, G.: Maximum lift effects on realistic high-lift transport aircraft configurations with
nacelle strakes. LTH-AD 04 01 013 (2011)
32. Nicolai, L.M.: Fundamentals of Aircraft Design. METS, Inc., San Jose (1975)
33. Heller, G., Kreuzer, P., Dirmeier, S.: Development and integration of a new high performance
wingtip device for transonic aircraft. ICAS 2002 Congress, pp. 122.1–122.7 (2002)
268 9 Particular Flow Problems of Large Aspect-Ratio Wings

34. Büscher, A.: Flügelendformen zur Leistungssteigerung eines Langstreckenflugzeuges (Wing-

Tip Devices for the Performance Enhancement of a Long-Range Transport Aircraft). Doc-
toral Thesis, Technical University Braunschweig, Germany. Shaker Verlag, Aachen, ZLR-
Forschungsbericht 2008-06 (2008)
35. Büscher, A., Radespiel, R., Streit, T.: Modeling and design of wing tip devices at various
flight conditions using a databased aerodynamic prediction tool. Aerospace Sci. Technol. 10,
668–678 (2006)
36. Rossow, V.J.: Lift-generated vortex wakes of subsonic transport aircraft. Prog. Aerosp. Sci.
35(6), 507–660 (1999)
37. Breitsamter, C.: Nachlaufwirbelsysteme großer Transportflugzeuge - Experimentelle Charak-
terisierung und Beeinflussung (Wake-Vortex Systems of Large Transport Aircraft—
Experimental Characterization and Manipulation). Inaugural Thesis, Technische Universität
München, 2007, utzverlag, München, Germany (2007)
38. Gerz, T., Holzäpfel, F., Darracq, D.: Commercial aircraft wake vortices. Prog. Aerosp. Sci.
38(3), 181–208 (2002)
39. Breitsamter, C.: Wake vortex characteristics of transport aircraft. Prog. Aerosp. Sci. 47(1),
89–134 (2011)
40. Crow, S.C.: Stability theory for a pair of trailing vortices. AIAA J. 8(12), 2172–2179 (1970)
41. Coustols, E., Stumpf, E., Jacquin, L., Moens, F., Vollmers, H., Gerz, T.: “Minimised Wake”:
A Collaborative Research Programme on Aircraft Wake Vortices. AIAA-Paper, 2003–0938
(2003)
42. Croom, D.R.: The development and use of spoilers as vortex attenuators. In: Proceedings of
the NASA Symposium on Wake Vortex Minimization. NASA SP-409, pp. 339–368 (1976)
43. Schell, I., Özger, E., Jacob, D.: Influence of different flap settings on the wake vortex structure
of a rectangular wing with flaps and means of alleviation. Aerosp. Sci. Technol. 4(2), 79–90
(2000)
44. Hünecke, K.: From formation to decay–extended-time wake vortex characteristics of transport-
type aircraft. AIAA Paper, 2002–3265 (2002)
45. Donaldson, C.duP., Bilanin, A.J.: Vortex wakes of conventional aircraft. AGARDograph No.
204 (1975)
46. Crouch, J.D., Miller, G., Spalart, P.R.: Active control system for breakup of airplane trailing
vortices. AIAA J. 39(12), 2374–2381 (2001)
47. Breitsamter, C., Allen, A.: Transport aircraft wake influenced by oscillating winglet flaps. J.
Aircraft 46(1), 175–188 (2009)
48. Crow, S.C., Bate, E.J.: Lifespan of trailing vortices in a turbulent atmosphere. J. Aircraft 13(7),
476–482 (1976)
Chapter 10
Small Aspect-Ratio Delta-Type Wing
Flow

Studied in this chapter are the lee-side vortex systems of small aspect-ratio
delta-type wings, which lead to the phenomenon of non-linear lift. Small aspect-
ratio delta-type wings encompass delta wings, usually in combination with a canard
and hybrid wings, i.e., trapezoidal wings in combination with strakes.
The lee-side vortex system appears depending on several geometrical and flow
parameters. The topic has many more facets than the topic of Chap. 8 “Large Aspect-
Ratio Wing Flow”.
Whereas large aspect-ratio wings are the wings of choice in the subsonic (unswept
wing) and the transonic (swept wing) flight regimes, the small aspect-ratio delta-type
wing (slender wing) is the wing of choice for the supersonic/hypersonic flight regime.
Note that the pure “thin wing”, as the Lockheed F-104 Starfighter had, in general
seems not to be an option for the latter flight regime.1
In the Introduction, Sect. 10.1, we distinguish between five different classes of
aircraft of small aspect-ratio delta-wing type, which we characterize. It follows a
short review of the history of the delta wing.
The remainder of the chapter is concerned with the phenomena and Unit Problems
concerning mainly the basic delta wing. In Sect. 10.2 an account is given of the
fundamentals of lee-side flow fields of such wings.
In Sect. 10.3 we discuss the results obtained with different flow-physical and
mathematical models applied to the flow field past the configuration of the Vortex-
Flow Experiment 1 (VFE-1). That wing is a wing with sharp leading edges. In the
following Sect. 10.4 we give proof that indeed discrete numerical solutions of the
Euler equations (Model 8) are viable solutions, if the wing has sharp leading edges.
For the VFE-1 configuration it is shown that along the leading edge the compatibility
condition is fulfilled and that the creation of vorticity and the generation of entropy
follow the same mechanism as for large aspect-ratio wings. For a rather long time
this was a matter of conjectures and speculations. With the work from the year 1996
the problem was settled.

1 Itis interesting to note, however, that Aerion and Lockheed work on the Mach 1.4 AS2 business
jet, which has a thin trapezoidal laminar wing.
© Springer-Verlag GmbH Germany, part of Springer Nature 2021 269
E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_10
270 10 Small Aspect-Ratio Delta-Type Wing Flow

The next section, Sect. 10.5, discusses results found for the VFE-2 configuration,
which has round leading edges. Both a subsonic and a transonic case are considered,
as well as vortex breakdown. The flow past the SAGITTA wing with a spanwise
varying leading-edge contour is treated in Sect. 10.6. Finally a hypersonic case is
presented, Sect. 10.7. The Blunt Delta Wing (BDW) at M∞ = 7.15 has a well devel-
oped lee-side vortex system, but almost no non-linear lift, because the lee side lies
in the so-called hypersonic shadow.

10.1 Introduction

10.1.1 Flight Vehicle Classes

The five distinctly different classes of aircraft of small aspect-ratio delta-type wings
are: (1) the supersonic transport aircraft, (2) the winged re-entry vehicle, (3) the hyper-
sonic airbreather, (4) the modern fighter aircraft, which itself is a flight vehicle with
many different configurational realizations, and (5) the corresponding unmanned
aircraft systems (UAS), also called drones.
In general we put emphasis on the classes (4) and (5), because here we have got
much material related to the separated and vortical flow past them.
1. The supersonic transport aircraft. Aircraft of this type first must cross the drag-
divergence domain in the transonic regime, Sect. 2.4.2, and then fly at small
to medium supersonic speeds. In the supersonic regime the drag coefficient
decreases, but the actual drag increases with the increasing dynamic pressure,
where the square of the flight speed is the driving factor.
All this calls for a slender configuration with, in addition, a sharp nose of the fuse-
lage and small leading-edge radii of wing and aerodynamic trim and stabilization
surfaces. In addition to these demands, the configuration must allow for sufficient
low-speed aerodynamic properties (take-off, climb, approach and landing). The
slenderness (half span/length ratio) of highly swept wings should not be smaller
than (b/2)/L ≈ 0.3 [1]. Otherwise low-speed properties become critical: too low
dC L /dα, dutch-roll behavior and lateral/directional stability problems at high α.
The first operational–and now decommissioned—aircraft of this type were the
Soviet Tupolev Tu-144 and the French-British Concorde. The overall 61.66 m
long Concorde, Fig. 10.1, in particular, had an ogive wing and no elevator unit.
The ogive or ogee wing, [1], is a double-delta type of wing with a smeared-out
transition from the first to the second delta and a rounded wing tip.2 The first delta
can and is also considered as being a strake.
The wing of a supersonic aircraft has to satisfy two contradictory demands: (1)
low wave drag, (2) sufficiently good low-speed performance. The first demand

2 To the double-delta types of wings we count the true double-delta wing, the ogive or ogee wing,
the strake-delta wing, and the strake-trapezoidal wing, also called “hybrid wing”.
10.1 Introduction 271

Fig. 10.1 The plan view of the Concorde configuration

calls for a wing so slender that the leading edge is a—nominally—subsonic one.
The second demand calls for a sufficiently large aspect ratio in order to achieve the
needed low-speed performance. This, of course goes together with a wave-drag
increment in the supersonic speed domain.
The double-delta wing in its inboard part is to meet the first demand, the outboard
part the second demand. Their combination moreover has the welcome effect of a
small movement of the neutral point during the transition from sub- to supersonic
flight. This allows for a trim schedule with minimized trim drag.
The wing shape of a given aircraft of course always reflects a compromise between
aerodynamic and many other demands. In the following sections we discuss the
flow past slender wings together with the implications of the wing shape for the
associated vortex flow.
Back to our example: the leading edge of the Concorde wing at supersonic flight
roughly can be considered as nominally subsonic up to about half root chord
and then, at the outboard delta, as nominally supersonic, and eventually, with
increasing sweep at the wing tip, as subsonic again.
Newer projects of supersonic transport aircraft, also business aircraft aim for
an increase of the aerodynamic performance and additionally also for a drastic
reduction of the sonic boom. The latter is mandatory to allow flight over populated
areas, which was not permitted for the Concorde.
2. The winged re-entry vehicle. A winged re-entry vehicle is a glider, which essen-
tially flies a braking mission [2]. For our purpose it suffices to distinguish three
major trajectory parts: (1) the upper part down to approximately 50 km, (2) the
intermediate part, and (3) approach and landing.
Part 1 of the trajectory is the major braking phase, where also the largest thermal
loads are present. In part 2 the glider must have sufficiently high cross-range and
down-range capabilities, mainly in order to cope with uncertainties of the prop-
erties of the atmosphere, see, e.g., [2]. In part 3 then a sufficiently low landing
speed is demanded.
We look at the configuration of the now decommissioned American Space Shuttle
Orbiter (32.8 m length), Fig. 10.2. It has a double-delta wing with large leading-
edge radii, and also a blunt fuselage nose with large radius.
272 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.2 The plan view of

the Space Shuttle Orbiter
configuration

The large radii are necessary in order to cope with the thermal loads, in particular
in Part 1 of the trajectory. The first delta (or strake) has the leading-edge angle
ϕ L E1 = 81◦ , and the second delta ϕ L E2 = 45◦ .3
Part 1 was flown with a high angle of attack, at the beginning with α ≈ 40◦ . The
aerodynamic performance in terms of the lift-to-drag ratio then was L/D ≈ 1. In
this way the required deceleration was achieved. Part 2 was flown with a smaller
angle of attack and L/D = 2 to 3, and Part 3 finally with L/D  4 at α ≈ 10◦ .
3. The hypersonic airbreather. G. Cayley’s design paradigm says that one ought to
assign functions like lift, propulsion, etc., unambiguously to corresponding sub-
systems, for instance the wing, the propulsion system, etc., see, e.g., [3]. These
subsystems and their functions are to be coupled only weakly and linearly. Con-
sidering Fig. 10.3 we see, however, that obviously with increasing flight Mach
number the coupling of the lift system and the external part of the propulsion
system grows, i.e., an increasing part of the lower vehicle surface caters to both
functions.4
The reason for that is the constraint on the dynamic pressure q∞ = 0.5ρ∞ u 2∞ ,
see, e.g., [2]. The dynamic pressure is a measure of the mechanical loads on the
structure of both the airframe and the propulsion system. In addition all aero-
dynamic forces and moments are proportional to it. The ranges of q∞ found in
the literature are large, 50 kPa being an often cited value. If both the ramjet and
the scramjet mode are to be employed, ranges of 25 kPa  q∞  95 kPa have
been proposed.
In any case, the dynamic-pressure constraint means that the faster one flies, the
higher has to be the flight altitude. Accordingly, because the air density decreases
with altitude, the larger has to be the inlet capture area and hence the propul-

3 Another re-entry vehicle configuration type is the lifting body. In the 1990s the Crew Return
Vehicle (CRV) was studied (X-38). The present Dream Chaser of the Sierra Nevada Corporation
Space Systems is of that type, too. Boeing, however, for its Phantom Express (XS-1) has chosen a
double-delta wing configuration.
∞ ≈ 3.5, that of ramjet
4 The flight Mach-number range of turbojet propulsion extends up to M

propulsion up to M∞ ≈ 6 to 7, and that of scramjet propulsion up to M∞ ≈ 10 to 12, see, e.g., [2].

10.1 Introduction 273

Fig. 10.3 Growth of the external part of the propulsion system (forebody, inlet) and the external
nozzle area (hatched) of airbreathing aircraft with cruise Mach number M∞ [4]

sion system package, although forebody pre-compression then effectively reduces

both, see, e.g., [2]. This is behind the configuration patterns sketched in Figs. 10.3
and 10.4.
We show as an example the plan view of the two-stage-to-orbit space transporta-
tion system SÄNGER, Fig. 10.4, which was the reference concept of a former
German Hypersonics Technology Programme [2, 5]. The lower stage—82.40 m
long—was to transport the upper stage to an altitude of about 32 km and to release
it at a flight Mach number M∞ ≈ 6.8.
The plan view reveals a double-delta like wing shape of the lower stage. The first
delta presents a subsonic leading edge. The second part then serves several pur-
poses like that of the supersonic transport aircraft discussed above. The planned
Lockheed Martin SR-72, a 30 m long M∞ = 6 drone, has a similar wing arrange-
ment.
How hypersonic scramjet airbreathers, apart from experimental vehicles, eventu-
ally will look like, is an open question. The pure waverider, like the “caret wing”,
or osculating cone shapes, is not an option. Integration of the propulsion system,
integration of aerodynamic trim and control surfaces (straight hinge lines), blunt
rounded leading edges in order to cope with the thermal loads, and off-design and
low-speed behavior are the demands, which must be met [2].

Fig. 10.4 The plan view of

the SÄNGER TSTO space
transportation system [5]
274 10 Small Aspect-Ratio Delta-Type Wing Flow

4. The modern fighter aircraft. While the shapes of subsonic and transonic transport
aircraft have converged to a certain degree to what are thought to be optimum
configurations, Sect. 8.1, this is not the case for modern fighter aircraft, in partic-
ular if they are to operate in the transonic and the (low) supersonic flight domain.
The many different requirement profiles of such aircraft have led to quite different
vehicle shapes, where also signature aspects (radar, infrared) have played a large
role.
As Fig. 10.5 shows, the main wing planforms are the swept wing (backward and
forward), the pure delta wing, the delta wing with canard, and the hybrid wing,
i.e., the trapezoidal wing with strake. In our short characterization of these wings
we follow the in-depth discussion given by E. Krämer [6].
The classical swept wing, also the trapezoidal wing, cannot meet the demands put
on the aircraft under consideration. We shortly discuss two major configuration
classes, which are in operation: (a) the delta wing with a canard, (b) the hybrid
wing, i.e., the combination of the trapezoidal wing with a strake.
(a) The delta wing with a non-swept trailing edge and a high leading-edge sweep
(ϕ L E = 50◦ to 60◦ ) has large benefits. Together with a canard it is a choice
for unstable configurations, for instance the Eurofighter Typhoon, the Dassault
Rafale, and the Saab JAS 39 Gripen.
Generally the canard is not used primarily for trim or control of the aircraft. That
is the task of the corresponding control surfaces at the wing’s trailing edge. The
main purpose of the canard is to shift forward the neutral point in order to achieve
the desired degree of longitudinal instability. In this way drag and flight charac-
teristics of the aircraft can be optimized.
Subsidiary tasks are the support of the trailing-edge control surfaces and the
shortening of the landing distance. The reader may note that the location of the
canard relative to the delta wing is quite different for the given examples: close
coupled in the case of the Gripen and the Rafale, weakly coupled in the case of
the Eurofighter. Of course also the vertical positions are different, which we do
not show.
(b) A hybrid wing is the trapezoidal wing combined with a strake (leading-edge
extension). The trapezoidal wing typically has a well rounded leading edge with
rather small sweep angle (ϕ L E = 25◦ to 45◦ ). The strake then is a slender, far
forward reaching inner wing with a high sweep angle and a small leading-edge
radius. The leading edge may be straight or slightly curved.
At high angle of attack the large sweep leads to a strong vortex system, which on
the one hand results in a non-linear lift increment and on the other hand stabilizes
the flow past the trapezoidal wing. Hence at high angle of attack uncontrolled
separation first happens at the outer part of the trapezoidal wing, but the separated
area is much smaller compared to that at the latter wing alone.
The hybrid wing combines favorably low and high speed flight properties. Many
of the current American and Russian fighter aircraft have a hybrid wing, see
Fig. 10.5. We do not discuss variable-sweep configurations.
10.1 Introduction 275

Fig. 10.5 Plan views of typical active fighter aircraft, the Suchoi Su-47 being only a demonstrator
[6]
276 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.6 The plan view of

the Raytheon UCAV
demonstrator

A special case is the F-16XL—a derivative of the F-16 Fighting Falcon—with a

cranked-arrow delta wing. It can be considered as a cross-breed of a double-delta
like wing and a hybrid wing.
The delta wing with canard and the hybrid wing both have their merits. In some
instances one has advantages over the other. It depends on the mission require-
ments and on the assessment of the single mission elements, which type finally is
preferred. Interesting papers in this regard are that of K.E. Modin and U. Clareus
on the evolution of the Gripen configuration [7] and of Clareus on its aerody-
namic highlights [8]. It is noted that it is a delicate task to make a fair comparative
evaluation of the merits of candidate layouts. And finally, the experience of the
design team, even the product philosophy of the manufacturing company may
play a role [6].
5. Unmanned aircraft systems (UAS).
Of interest here are unmanned systems with planforms like shown in Fig. 10.6.
Such systems are combat aerial vehicles, which have a very high agility, that is
they permit very high g-loads, not achievable with piloted aircraft. At the same
time they have very low RADAR and IR signatures. Examples of studies, demon-
strators and projects are the Raytheon UCAV Demonstrator, Dassault Neuron,
BAE Taranis, Boeing X-45, Northrop Grumman X-47.
High g-loads are achieved, for instance, with high angles of attack, which then,
depending on the planform of the flight vehicle and the leading-edge properties,
lead to complex lee-side vortex systems.

All the considered slender configurations have in common—at according angles

of attack—lee-side vortex systems of different kinds, interactions of different vortex
pairs like in the case of the delta wing with canard, and interactions of vortices with
other flow portions. The vortex behavior as such, for instance vortex breakdown, is
also of importance.
The aircraft designer must have knowledge and experience of the effects and the
behavior of lee-side vortex systems. During the shape definition of the aircraft the
designer must be able to provide and place the necessary configuration elements.
And finally he must be able to interpret in experimental and numerical simulations
the resulting flow fields and their influence on the aircraft.
Our aim is to provide and foster the understanding of the basic properties of the
corresponding vortex systems and their behavior. We further want to demonstrate the
10.1 Introduction 277

capabilities of numerical simulation methods for such flows and the interpretation
of the results from a fluid-mechanical point of view.
Generally we do not deal with configurational issues of aircraft. In the following
chapter, Chap. 11, we discuss selected flow problems of small aspect-ratio delta-type
wings.

10.1.2 Notes on the History

Small aspect-ratio delta-type wings were designed and actually flown before their
properties and flow phenomena were fully understood and scientifically confirmed.
This sometimes had grave consequences.5
The history of this type of wings probably began in the year 1935, when H. Winter
reported on experimental work with slender wings, slender not necessarily meaning
delta-type [11]. Adolf Busemann’s concept of the swept wing from 1935, [12], soon
led to experimental work on such wings: H. Ludwieg, 1939 [13].6
A first non-linear theory for rectangular wings was published by W. Bollay in
1939 [15]. K. Gersten 22 years later formulated a non-linear theory for wings with
arbitrary planform [16].
N.R. Payen in France in the late 1930s designed a tandem-wing configuration
aircraft [17]. The Pa 22 was to participate in an air race. The originally foreseen
engine was an early form of a ramjet. Later on a piston engine was installed. The
trapezoidal frontal wing, the canard, had nearly zero leading-edge sweep, whereas
the delta wing behind had a sweep angle of 67◦ , its span being only 85 per cent of
that of the frontal wing.
During the German occupation of France a flight-test program was initiated, mod-
ifications became necessary. In 1943 the aircraft more or less was destroyed during
an Allied air raid. The underlying philosophy of the design is not clear, possible
influences on later French aircraft designs cannot be traced.
In Germany during World War II, A. Lippisch developed an experimental vehicle
with a slender delta wing, the DM-1 glider, see, e.g., [18]. The well-rounded leading
edge of the wing had a sweep angle of 60◦ . Lippisch intended to obtain first insight
into the low-speed domain for the development of future supersonic aircraft.
The glider in 1945, after the war, was transported to the USA. There at that time
R.T. Jones had published his now classical NACA Report 863 [19]. A fundamental
result of Jones’ work was the significance of the velocity component normal to the
leading edge: reduction of the compressibility effects in the transonic domain and
the benefits of the “subsonic leading edge” in the supersonic flight domain.

5 These notes are based on the introduction to the doctoral thesis of the fourth author of this book
[9], but see also [10].
6 In [14] a very detailed account of the German work on that topic during the first half of the 1940s

can be found, also treated is the work on the area rule and what now is called the supercritical airfoil,
see also Sect. 2.4 of the present book.
278 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.7 DM-1:

manipulation of the leading
edge and the result [9]

In 1946 the original DM-1 glider was tested in the large subsonic wind tunnel
at NASA Langley by H.A. Wilson and J.C. Lovell [20]. An initially unexplained
result was that the configuration had a much smaller maximum lift compared to
that previously obtained in sub-scale tests. Lovell and Wilson rightly conjectured
that the smaller maximum lift was due to the absence of the strong laminar leading-
edge separation found in the sub-scale tests. They decided to obtain that effect by
adding a sharp elongated leading-edge nose, which then effectively rose the lift curve,
Fig. 10.7.
The merits of Wilson and Lovell are (1) that they recognized the importance
of Reynolds-number and nose radius for the formation of the leading-edge vortex
system and (2) that they devised and tested manipulation means. It seems that they
unknowingly suppressed the effect of the enlarged leading-edge sweep angle, origi-
nally 60◦ , theirs 64◦ .
Anyway, their work in the USA triggered developments like the XF-92A, the first
delta configuration aircraft, which flew supersonic, Fig. 10.8, and later the F-102
Delta Dagger and the F-106 Delta Dart. After the not very successful aircraft B-58
Hustler and the B-70 Valkyrie the interest in delta configurations waned in the USA,
however, did not expire completely.
Interesting is that Lippisch already in 1939 went to the Messerschmitt company
and developed, emanating from the DSF 194 experimental aircraft, the Messerschmitt
Me 163 interceptor, Fig. 10.9.
That rocket propelled aircraft became operational in 1943. With a leading-edge
sweep angle of 25◦ it would have had a subsonic leading edge up to a Mach number
M∞ = 1.1. Therefore it was not the blueprint for supersonic configurations as the
DM-1 was.
Nevertheless, by the end of World War II every German aircraft company had one
or more delta-wing projects—in any case with turbojet propulsion—on the drawing-
board or already in the wind tunnel, see, e.g., [21, 22] and in particular also [14].
10.1 Introduction 279

Fig. 10.8 From the

experimental configuration
DM-1 to the experimental
supersonic fighter
configuration XF-92a [9]

Fig. 10.9 Plan view of the

Messerschmitt Me 163 [21]

Figure 10.10 gives an impression of the range of such projects, although also swept-
wing configurations are included.
In Germany after WW II aircraft development was banned. Remarkable is the
observation that many German aircraft specialists went to the victorious powers and
continued to work in their field [24]. To be mentioned in particular is that German
design teams soon were active again. In the first half of the 1950s Kurt Tank with
most of his original Focke-Wulf team developed in Argentina the M∞ = 1.1 fighter
FMA IAe 33 Pulqui II. That aircraft had a swept wing with a large leading-edge
sweep. In the second half of the 1950s, now in India, followed the development
of the M∞ = 1.2 delta-wing fighter-bomber HAL HF-24 Marut. During the 1960s
Willy Messerschmitt with his design team moved from Spain to Egypt and developed
the M∞ = 2 delta-wing interceptor Helwan HA-300.
280 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.10 Examples of jet aircraft with swept wings and delta wings in Germany until 1945 [23].
P: prototyp, F: flight tests, B: production order, SE: in service, all others: project studies

While in the USA the interest in delta configurations waned, in Europe, in particu-
lar, the topic was picked up again.7 Names like Robert Legendre, Hermann Behrbohm
and Dietrich Küchenmann are connected with their successful developments.8 The
French Mirage family (Dassault), the Swedish aircraft family Draken, Viggen and
Gripen (Saab), the French/British Concorde, and the Eurofighter Typhoon are testi-
mony to that.
The roots of the delta-wing Eurofighter go back to the late 1960s, when the
German Airforce had defined their F-104 successor needs. At the end of the 1970s
Germany, the UK, and Italy—Spain joined later—were on their way to the launch
of the development of the Eurofighter in 1988. In [26] this way is sketched—from a
German point of view—with the different intermediate vehicle studies and designs,
including that of the MBB/Rockwell experimental aircraft X-31A, of which two
vehicles were built and flown.
It certainly was to the merit of the delta-wing application side that the interest
was aroused on the research side.
Experimental studies with simple wing shapes, delta and delta-like plan forms
were beginning in the 1950s. T. Örnberg in Sweden, [27], D.J. Marsden, R.W. Simp-
son and W.J. Rainbird in Great Britain, [28], H. Werlé in France, [29], and D. Hummel
and G. Redeker in Germany, [30–32]—to name a few of the researchers—provided

7 Regarding the development of the topic in the USA see, e.g., [25].
8 We note without further comment that after WW II Hermann Behrbohm came from the Messer-
schmitt company to Saab in Sweden, and Dietrich Küchemann from the Aerodynamische Versuch-
sanstalt (AVA) in Göttingen to the Royal Aircraft Establishment (RAE) in the UK. Robert Legendre
since 1950 was at ONERA in France.
10.1 Introduction 281

basic data regarding forces and moments, pressure distributions, and flow-field obser-
vations over sharp-edged slender wings.
Theoretical work by W. Bollay and K. Gersten was mentioned above [15, 16]. In
the following years vortex-line approaches, restricted to the leading-edge region,
were developed in order to overcome the shortcomings of Bollay’s ansatz. We
mention the work by R. Legendre, [33], C.E. Brown and W.H. Michael, [34],
D. Küchemann, [35], K.W. Mangler and J.H.B. Smith, [36].
Common in their work was the use of the theory of slender bodies with conical
flow. This led to the violation of the flow-off condition at the wing’s trailing edge.
Partly useful overall forces were obtained, but the unrealistic pressure distribution
at the trailing edge resulted in a wrong pitching moment. To be emphasized is E.C.
Polhamus’ concept of the leading-edge suction analogy from 1966, which permitted
to describe the non-linear lift behavior by means of linear potential theory [37].
Early on also the phenomenon of vortex breakdown, observed in the experiments,
was tackled by the theoreticians. M.G. Hall used an analogy to two-dimensional
boundary-layer flow, [38, 39], H. Ludwieg derived his criterion for inviscid vortex
flow from the stability theory, [40], whereas T.B. Benjamin started from a wave-
propagation theory [41]. Common to these theories is that they assume a cylindrical
structure of the vortex, respectively its core. That was in contrast to the experimental
findings.
The reader interested in the (early) history of the development of slender-
wing research is referred to the review articles in the first half of the 1980s by
E.C. Polhamus, [25, 42] and regarding the theoretical side to the articles of J.H.B.
Smith, [43], and H.W.M. Hoeijmakers [44].
At that time the second mathematization wave of fluid mechanics and aerody-
namics, [3], was gaining momentum. Growing computer power, due to increased
chip performance, new computer architectures (vector computer, massively parallel
computer), soaring algorithm and computer science, and not least, in academia and
industry a rising acceptance of the discrete numerical methods for the solution of the
governing equations of fluid flow were the driving factors, see, e.g., the introduction
to [45].
Analytical and semi-empirical methods for the calculation of non-linear effects
in the flow past slender wings gave way to non-linear panel methods (Model 4 of
Table 1.3) and then to discrete numerical solutions of the Euler equations (Model 8)
and finally of the Navier–Stokes and RANS equations (Model 9 and 10). Overviews,
in particular also of the Model 4 methods can be found in [46] and of the Model 10
methods developed in Europe in [47].
Model 8 to 10 solutions are discrete numerical one-domain approaches. Such
approaches have the charm that the whole flow field can be treated without any
preconditioning, apart from the discretization: the separation, either flow-off (sharp
LE) or ordinary (round LE) separation from the leading edges, the roll-up of the
separating vortex sheets (the feeding layers) into the primary vortices, the same for
the secondary vortices, and the flow-off separation from the trailing edge and the roll-
up process of the trailing vortex layer and its interaction with the lee-side vortices,
Fig. 10.11.
282 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.11 General features

of flow past a delta wing [48]

In the beginning, due to the rather weak computer power at that time, for slender
wings in the subsonic and transonic domain only Euler (Model 8) solutions could
be realized, the first one worldwide being that of L.-E. Eriksson and A. Rizzi in
1981 [48]. At that time in the USA, for instance, space-marching Euler solutions for
supersonic flight cases were the topic.
The significance of the work of Eriksson and Rizzi is that they demonstrated
that it is possible to “capture” the vortical structures in the flow field, Fig. 10.11,
instead to “fit” them with vortex sheets and vortices as singularities. Regarding the
methods, which fit the vortical structures, see, e.g., [49] and also [46]. The capturing
is analogous to shock capturing, see our Sect. 5.3. Eriksson and Rizzi thus initiated a
sea change in approach and paradigm to treat vortical flow fields past small aspect-
ratio delta-type wings.
Discrete numerical Euler (Model 8) solutions, i.e., inviscid solutions,

1. demand that sharp leading edges are present in order to fix the location of sepa-
ration (flow-off separation),
2. they cannot produce secondary and higher-order vortices, and
3. regarding vortex breakdown, they fail, if viscous effects play a major role.

Vortex sheets, the roll-up process and the resulting vortices of course require
kinematically active vorticity to be present. In Model 4 solutions this is hidden in
the singularity sheets and vortex cores. In Model 8 solutions this vorticity appears
by itself, together with the associated total-pressure loss, or, respectively, an entropy
rise, Chap. 5 of this book.
10.1 Introduction 283

This appearance of vorticity from the beginning gave rise to speculations. The
two first authors of the present book postulated that the situation at the delta wing’s
leading edges is the same as that at the trailing edge of a large aspect-ratio wing.
From this the concept of kinematically active and inactive vorticity content of a shear
layer originated—Chap. 4 of the present book—as well as the explanation of the
appearance of the kinematically active vorticity content in discrete (Model 8) Euler
solutions, Chap. 5. At the Symposium on the International Vortex-Flow Experiment
on Euler Code Validation in 1986, see below, they presented the large aspect-ratio
wing case, see Sect. 8.3 of the present book.
Nevertheless, the debate was still going on. In 1993 S.M. Hitzel in his doctoral
thesis, for instance, saw the origin of the vorticity in the cross-flow shocks in the lee-
side flow field [50]. These, of course, do not appear in purely subsonic flowfields.
It was a doctoral student of the first author of this book, R. Hentschel [51], who in
1996 finally gave the proof for the sharp-edged delta wing by applying the concept
of the kinematically active vorticity content, Chap. 4. His work is sketched as Unit
Problem in Sect. 10.4 of the present chapter.
The advent of the Euler solutions did raise the desire for dedicated experimental
data. Previously obtained material was not suited well. In 1982 at FFA the second
author of this book was in need of dedicated experimental data to check his Euler
simulation of delta-wing flow, see above.
This led to a chain of events, and eventually to what became the “International
Vortex-Flow Experiment on Euler Code Validation (VFE-1)” with European and
American partners, which was carried out from 1984 to 1986. The configuration was
a cropped sharp-edged delta wing with 65◦ leading-edge sweep. The symposium
with the presentation of the results took place in Stockholm, Sweden, October 1–3,
1986 [52]. Modeling and computing was mostly on Euler (Model 8) level, only one
contribution was on Navier–Stokes (Model 9) level. Reports on VFE-1 can be found
in [53, 54].
In the meantime the AGARD Symposium on Aerodynamics of Vortical Type
Flows in Three Dimensions was held in Rotterdam, 1983. The proceedings, AGARD-
CP-342, show that modeling and computing was on Euler (Model 8) level at most.
Then, in 1988, the Symposium Transsonicum III in Göttingen already saw several
RANS (Model 10) approaches, also to flow past slender configurations.
Work on discrete numerical solution methods for the Euler equations (Model 8) by
A. Eberle at the Military Aircraft Division of Messerschmitt-Bölkow-Blohm (MBB)
led to the EUFLEX code, see [55]. Applications to generic delta-wing configurations
were performed and led, for instance, to the vortex-flow field over a generic delta
wing/canard configuration as shown in Fig. 10.12 [56].
A major step forward was the probably first application to a full fighter-aircraft
configuration [57]. The computation domain had for that time a very high number of
520,000 cells. Figure 10.13 presents the computed surface-pressure field at M∞ =
0.85, for the angle of attack α = 7.5◦ and the sideslip angle β = 5◦ .
Computer and algorithm development advanced further and it became possible
to attack in full the problem of viscous flow on RANS basis (Model 10). That would
open the possibility to treat configurations with round leading edges and to tackle
284 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.12 Vortical flow

over a generic delta
wing/canard configuration
[56]

Fig. 10.13 Euler solution

for a fighter-aircraft
configuration at angle of
attack and sideslip:
surface-pressure field [57]

the problem of secondary and higher order lee-side vortices. Consequently in 2001
D. Hummel and G. Redeker proposed the second vortex-flow experiment (VFE-2)
[58]. The suggested configuration—a NASA configuration [59]—had interchange-
able sharp and round leading edges, and again a sweep angle of 65◦ . The “Second
International Vortex-Flow Experiment” took place from 2003 to 2008 in the frame
of a task group [60, 61].9

9 AGARD, the Advisory Group for Aerospace Research and Development of NATO merged in 1996

with the NATO Defence Research Group (DRG) to become the NATO Research and Technology
Organization (RTO), now the Science and Technology Organization (STO) of NATO.
10.1 Introduction 285

In 2001 RTO started a Task Group on Vortex Breakdown AVT-80 (AVT: Applied
Vehicle Technology Panel). AVT-113, since 2003, had the objective of “Understand-
ing and Modeling Vortical Flows to Improve the Technology Readiness Level for
Military Aircraft”. Other AVT work did follow up to STO-TR-AVT-183 “Reliable
Prediction of Separated Flow Onset and Progression for Air and Sea Vehicles”. The
configurations considered in AVT-113 were F-16XL and also VFE-2, AVT-161 con-
sidered X-31 and SACCON (stability and control configuration), and AVT-183 the
diamond delta wing.
Nowadays the discrete numerical methods of aerodynamics have reached a high
level of usability. Limitations are present regarding the modeling of laminar-turbulent
transition and aspects of ordinary turbulent separation. Applications range from Euler
solutions in pre-design work to full RANS/URANS solutions in design work and this
for actual complex aircraft configurations. That is mirrored in STO-TR-AVT-201
“Extended Assessment of Stability and Control Prediction Methods for NATO Air
Vehicles”.
In his recent Lanchester Lecture (2017) J.M. Luckring gave an all-embracing
account of the development of the field of vortex-flow aerodynamics [62]. The second
author of this book together with J.M. Luckring discusses in [63] the evolution and
use of discrete numerical methods—CFD—for separated flows relevant to military
aircraft.

10.2 Non-linear Lift Exemplified with the Plain Delta Wing

This section is devoted to a short presentation of elements of the problem identifica-

tion. We study the phenomenon of lee-side vortices by considering the flow past the
simplest possible configuration, the plain delta wing. This is a Unit Problem in the
sense given in Fig. 1.12.
First we look at the relevant wing’s geometrical and flow parameters, then at
the non-linear lift as an aerodynamic phenomenon. The basic influence of sharp
and round leading edges and the limitations due to vortex breakdown and vortex
overlapping are noted. Then correlations of the observed lee-side flow phenomena are
presented, which mainly are based on experimental findings. A short consideration
follows of the means to influence the lee-side flow field (a deeper discussion of this
topic is given in Chap. 11).

10.2.1 Geometrical and Flow Parameters

The geometrical parameters of the plain delta wing are listed, together with relevant
flow parameters.
(a) Regarding the wing’s geometry we have the following parameters.
– Plan form:
286 10 Small Aspect-Ratio Delta-Type Wing Flow

1. span b,
2. leading-edge sweep angle ϕ◦ ,
3. aspect ratio Λ = b2 /A, with A being the wing’s reference area,
4. taper ratio λ = ct /cr .
– Chord geometry (L is the wing’s length):
1. nose radius r◦ ,
2. thickness ratio t/L,
3. camber ratio f /L.
(b) Free-stream parameters and wing attitude:
1. Mach number M∞ = u ∞ /a∞ , Reynolds number Re L = ρ∞ u ∞ L/μ∞ , and free-
stream temperature T∞ , where a∞ , ρ∞ , and μ∞ are free-stream speed of sound,
density, and viscosity.
2. angle of attack α,
3. angle of side slip β.
(c) Characteristics of the vortex flow:
1. Type of separation, either flow-off at sharp edges, or ordinary in all other cases,
2. laminar, transitional, turbulent flow,
3. flow topology and structural changes of the flow field,
4. vortex breakdown behavior.
(d) Geometrical properties of the vortex flow:
1. Position of the primary vortex axis in terms of angles
– against the free-stream direction: αV , or against the wing’s surface plane: αV ,
– against the wing’s center line: ϕV ,
2. and the vortex-breakdown location xbd /L.

10.2.2 The Non-linear Lift as Aerodynamic Phenomenon

Non-linear lift arises, because flow turns around the side edges of a lifting wing, and
lee-side vortices appear, which produce suction pressure at the upper side of the wing,
see the discussion in Sect. 8.4.3. The non-linear lift effect schematically is shown in
Fig. 10.14. In the case of a large aspect-ratio wing the curve of the lift coefficient
C L (α) is straight, which means that the non-linear lift increment is negligibly small.
The potential-flow gradient is
 
dC L dC L dC L /dα|2D
= /|2D / 1 + , (10.1)
dα dα πΛ

With dC L /dα|2D = 2 π we obtain

10.2 Non-linear Lift Exemplified with the Plain Delta Wing 287

Fig. 10.14 Schematic of

non-linear lift behavior, after
[6]

dC L 2πΛ
= . (10.2)
dα Λ+2

In the case of the small aspect-ratio wing the gradient increases non-linearly with
increasing angle of attack. For the linear case—slender body theory of R.T. Jones
[64]—broken lower line in Fig. 10.14, it is

dC L π
= Λ. (10.3)
dα 2
This value for small aspect ratios (Λ < 1) does not depend on the Mach number.
Consider now Fig. 10.15, taken from [9]. Sketched in Poincaré surfaces are typical
lee-side vortex-flow topologies observed over slender delta wings. (See in this regard
also Fig. 7.18.)
Shown is only the primary vortex system. The topologies are correlated with the
lift coefficient C L as function of the angle of attack α:

• (0) At small angles of attack no lee-side vortices are observable, linear lift behavior.
• (1) Lee-side vortices begin to appear. Their formation starts outward at the trailing
edge of the wing. With increasing α they creep forward at the leading edge until
the wing’s apex is reached. The lift coefficient C L still shows a linear behavior.
• (2) The lee-side vortex system is fully developed. An open lee-side flow field
is present. The slope of the lift coefficient becomes non-linear. The axes of the
vortices cross the wing’s tip. With α increasing the vortex axes are moving inwards
and upwards: ϕV and αV are variable.
• (3) The lee-side flow field is of closed type. The inwards motion of the lee-side
vortices ceases: ϕV = const., αV still increases, i.e., the axes of the primary vortices
in a sense adapt to the free-stream direction.
288 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.15 Flat delta wing: correlation of lee-side vortex systems (Poincaré surfaces in the lower
part of the figure) and lift behavior with angle-of-attack domains [9]. For simplicity secondary and
possible higher-order vortex pairs have been omitted. Also not indicated is the presence of two
primary attachment lines at the windward side, see Sect. 7.5

• (4) Vortex breakdown coming from behind reaches the trailing edge of the wing
(lift break shown in C L α at α L B ) and moves toward the wing’s apex until the
maximum lift is reached at αmax .

Figure 10.16 vividly visualizes the lee-side vortex flow past a fighter model in
a water tunnel. The lower figure shows the situation at low angle of attack with
beginning lee-side vortices and weakly non-linear lift. Shown in the middle is the
lee-side vortex flow and the non-linear lift, both fully developed. The upper figure
shows the model in post-stall with vortex breakdown, respectively fully separated
flow.
Regarding the flow field at the windward side of the wing see the discussion of
Fig. 7.18.
10.2 Non-linear Lift Exemplified with the Plain Delta Wing 289

Fig. 10.16 Fighter model in

a water tunnel [65].
Schematic assignment of
lee-side vortex phenomena to
the lift polar

10.2.3 The Matter of Secondary and Higher-Order Lee-Side

Vortex Pairs

In Sect. 7.4.3 we already met a secondary lee-side vortex pair in the Poincaré surface
at a delta wing, Fig. 7.10. This phenomenon is indicated also in Fig. 10.11.
These vortices also influence the wing’s lee-side pressure distribution and hence
the non-linear lift. In the paper of D. Hummel from 1978 a sketch is given in this
regard [66]. Figure 10.17a shows the lee-side vortex pairs. The feeding layers of the
primary vortices originate via flow-off separation at the, in this case, sharp leading
edges. The feeding layers of the counter-rotating secondary vortices originate via
ordinary separation—induced by the pressure field due to the primary vortex pair—
below the primary vortices.
Figure 10.17b is a sketch of the induced suction10 peaks as seen at the time of
[66] in 1978. However, if the lee-side flow field is established as shown there, the
primary suction peak in general is much more pronounced and the secondary one
much smaller, as examples in the following sections show.
With Fig. 8.35 we show a sketch of the topological schematic in the Poincaré
surface at the large aspect-ratio wing tip. The figure, based on the skin-friction line
patterns in the foregoing figures, shows the existence of a tertiary vortex. Tertiary

10 Suction pressure denotes a negative surface pressure coefficient c p and suction peak accordingly
a relative minimum of c p .
290 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.17 Basic features of

lee-side vortex flow at delta
wings [66]. Closed lee-side
flow field with a the primary
and secondary lee-side
vortex pairs, b the induced
lee-side suction peaks

lee-side vortex pairs present at delta wings were reported early on, for instance by
T. Örnberg in 1954 [27].
Before we have a look at Örnberg’s figures, we need to consider the nomencla-
ture. Örnberg speaks about a rearward vortex and a forward vortex, Fig. 10.18 (left).
Alternatively he calls the primary vortex the “edge vortex” and the secondary one
the “surface vortex”, Fig. 10.18 (right).
The pair of tertiary vortices in view of the primary vortex pair is a pair of co-
rotating vortices. The VFE-2 community has chosen to speak about two primary
vortex pairs over the lee side of the delta wing, that is about a “doubled primary”
vortex system. Of course one could speak about a primary, a secondary and a ter-

Fig. 10.18 Schematic of the—suggested—lee-side flow (left) at the ϕ0 = 58◦ delta wing at α ≈
25◦ , (right) at the ϕ0 = 70◦ delta wing at α ≈ 15◦ [27]
10.2 Non-linear Lift Exemplified with the Plain Delta Wing 291

tiary vortex pair. However, we keep the VFE-2 nomenclature when discussing this
phenomenon in Sect. 10.5.2.

10.2.4 Basic Influences of Sharp and Round Leading Edges

It seems that the differences of sharp and round leading edges of small aspect-ratio
delta-type wings in view of the non-linear lift were discovered in 1946 by H.A.
Wilson and J.C. Lovell, while they worked on the DM-1 glider, Sect. 10.1.2.
Regarding the formation of the primary lee-side vortices in relation to the leading-
edge shape J.M. Luckring gave a systematic account [67], Fig. 10.19.
At the—clean—sharp-edged wing the primary vortex originates along the sharp
leading edge via flow-off separation (primary separation). The emerging vortex
sheet—the feeding layer—, which carries both kinematically active and inactive
vorticity content, then rolls up into the primary vortex. That in turn induces, via
ordinary separation, the secondary vortex, see above. In Fig. 10.19 it is indicated that
the vortex origin is fixed at the wing’s apex.11
The properties of the primary vortex then depend on the angle of attack and the
Mach number, via the normal—to the leading edge—angle of attack α N and the

Fig. 10.19 Formation of the primary lee-side vortex pair: basic influences of sharp and round
leading edges [67]

11 Observations in wind-tunnel experiments show that this is not necessarily so. See in this regard
also Fig. 7.18.
292 10 Small Aspect-Ratio Delta-Type Wing Flow

normal leading-edge Mach number M N , Sect. 10.2.6. Reynolds number, angle of

side-slip, the wing’s cross-section generally do not play large roles.
The situation is different at the round-edged wing. The onset and the progression
of the primary separation not only depends on the flow conditions—as is the case
with the sharp-edged wing—but also on the geometry of the leading edge including
a more pronounced effect of the Reynolds number.
In the VFE-2 literature the cross- flow bluntness parameter

rle (x)
pb (x) = (10.4)
b (x)

is used for qualitative considerations [67]. The terms in that parameter are the leading

edge radius rle and the local span width b (x) being defined in the wing’s cross-
section. The local span width is a function of the location in x-direction, while the
leading-edge radius rle may be constant, as is the case with the VFE-2 configuration.
In Sect. 6.4 we show how the parameter pb approach can be supported in a simple
way.
From the beginning the origin of the vortex is not at the wing’s apex, as indicated by
the distance xv in Fig. 10.19. The vortex sheet emerges via ordinary separation at some
location of the leading edge, generally on its upper side. Basically the emergence
happens in the manner, which was sketched in Fig. 10.15 for the sharp-edged delta
wing. The origin moves from the back of the wing forward with increasing angle of
attack. Of course again the Mach number, the Reynolds number, the leading-edge
radius and others are determining the event.
There remains the question what a sharp or a round leading edge is. In an experi-
ment or in a discrete numerical simulation, the leading edge can be sharp in the sense
of the word, or rounded as demanded. The reality of an aircraft is different, as we
show in Sect. 6.3. There was and is no “sharp” leading edge, except that of the thin
trapezoidal wing of the Starfighter F-104. The reasons are not enough robustness in
handling, and weather conditions (rain, hail).
Actually it must be asked, at what radius does a round leading edge act like an
aerodynamically sharp one, Sect. 6.4. Regarding Euler (Model 8) solutions, the effect
of trailing-edge rounding on the implicit Kutta condition at an airfoil was discussed
in [55] on p. 327. A general investigation of the problem is not known to the authors
of this book.
The gedankenexperiment made in Sect. 6.4 does not help directly. But note that in
any case a factor of two in the inverse thickness ratio δ, from for instance δ = 150 to
δ = 300—equivalent to pb = 0.0033 to pb = 0.00167—results in a factor of about
four in c pmax and a factor of about seven for dc p /dϕ|max . The locations of these
maxima are very close to ϕ = 90◦ at the upper side of the leading edge.
Regarding Model 9 and 10 solutions for round-edged delta wings, we are not
directly concerned with the question of round or sharp leading edges.
10.2 Non-linear Lift Exemplified with the Plain Delta Wing 293

10.2.5 Limits: Vortex Breakdown and Vortex Overlapping

The slenderness of the wing—sweep angle of the leading edge—as well as the
angle of attack have upper bounds, because two phenomena can occur and are to be
handled or even avoided: vortex breakdown—Sect. 3.13—and vortex overlapping—
vortex pairing—, Sect. 3.12.3, both usually considered in relation only to the primary
lee-side vortices.
Vortex breakdown, usually beginning at the rear of the wing, leads to a lift loss and
to a pitch-up. Hence it is a phenomenon of utmost importance for small aspect-ratio
delta-type wings.
Here we show with correlations from [9] the stability boundaries of vortex break-
down and vortex overlapping at flat delta wings with sharp leading edges. Global
data like those given in Fig. 10.20 can be found in many publications. However, they
go back to [68] and hence are relevant primarily for the low speed flight domain.
Note the leading-edge sweep-angle domains, which are ϕ0 ≈ 45 to 70◦ for pure
delta wings, ϕ0 ≈ 70 to 80◦ for strakes, and ϕ0 ≈ 30 to 45◦ for trapezoidal wings.
The angle of attack, at which vortex breakdown appears at the trailing edge of the
wing, is the critical angle αbd , Fig. 10.20. For α > αbd even small changes of the
free-stream properties and the geometry of the wing can lead to an erratic behavior
of the wing’s forces and moments. With increasing α the breakdown location moves
upstream toward the wing’s tip until a complete destruction of the lee-side vortex
system happens.

Fig. 10.20 Flat, sharp-edged delta wing: correlation of the influence of leading-edge sweep angle
ϕ0 and angle of attack α on the behavior of the lee-side vortex system [9]
294 10 Small Aspect-Ratio Delta-Type Wing Flow

We note that the breakdown boundary ranges from rather small angles of attack
at leading-edge sweep angles of about ϕ0 ≈ 50◦ to α ≈ 33◦ at ϕ0 ≈ 77◦ .
Above ϕ0 ≈ 77◦ the lee-side vortex system becomes asymmetric (vortex overlap-
ping), with a strong influence on forces and moments. This phenomenon is visualized
in Fig. 7.19. For higher sweep angles the critical angle of attack becomes smaller.
With leading-edge sweep and angle of attack outside of the “healthy” domain—
without a distance producing fuselage—maximum lift is reduced, roll instability
appears for increasing ϕ0 at always smaller angles of attack, even for the symmetric
flight attitude (β = 0◦ ), see also Sect. 11.1. The same phenomenon also is present at
slender fuselage noses and rocket configurations.

10.2.6 Correlations of Lee-Side Flow Fields

Of large interest in particular for the aircraft designer is the knowledge of fundamental
dependencies of the lee-side flow field on the wing’s geometry, flight parameters and
vehicle attitude. This then leads to a first appraisal of the required aerodynamic
properties of the aircraft.
Main influencing factors from the side of the wing geometry are the leading-edge
sweep angle, the leading-edge nose radius—down to the sharp leading edge—and
from the free-stream side the angle of attack and the flight Mach number. A first
comparatively coarse correlation was given by A. Stanbrook and L.C. Squire [69].
They recognized that the flow properties at flat, thin wings with high leading-edge
sweep depend on the flow relationships normal to the leading edge in terms of the
“normal angle of attack α N ” and the “normal leading-edge Mach number M N ”:

α N = arctan(tan α/ cos ϕ0 ), (10.5)

M N = M∞ cos ϕ0 (1 + sin2 α tan2 ϕ0 )0.5 . (10.6)

The authors of [69] proposed criteria for the existence of lee-side vortex systems,
now called the Stanbrook/Squire boundary, Fig. 10.21.
For sharp-edged delta wings the relation for that boundary approximately is (α N
in degree)
M N = 0.75 + 0.94 · 10−3 α1.63
N , (10.7)

and for round-edged wings, for α N  6◦ ,

M N = 0.2 + 0.375 · 10−3 (α N + 23.3)2 . (10.8)

That above M N ≈ 1 not only attached flow exists, but that, due to shock interfer-
ences, also vortex flow develops, was shown by W. Ganzer, H. Hoder and J. Szodruch,
[70], see also Sect. 10.7.
10.2 Non-linear Lift Exemplified with the Plain Delta Wing 295

Fig. 10.21 Formation of lee-side vortex systems, the Stanbrook/Squire boundaries [9]: a thin
wing/sharp leading edge, b thick wing/round leading edge

We do not intend to give a review of the literature about correlations. In [9] several
correlations of experimental data can be found (selection):

• Vortex position αV relative to the wing’s lee-side surface as function of the angle
of attack α and the lateral position ϕV .
• Vortex location αV as function of the angle of attack α.
• Correlation of the location of the vortex axis αV with the Reynolds number Re L
and the chord thickness.
• Vortex-axis angle ϕV /ϕ0 as function of M N for turbulent secondary separation.
• Influence of the angle of side-slip β on the lateral position ϕV .

In Figs. 10.22 and 10.23 summarizing correlations from [9] are presented of pos-
sible flow patterns at sharp- and round-edged delta wings, see also [55].
The limiting curves (hatched) are transition regions between the flow patterns.
Deviations up to ±10 percent are possible due to the spectrums of the experimental
data. In order to minimize Reynolds-number effects, only experimental data with
296 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.22 Possible flow patterns at wings with sharp leading edges [9]

turbulent secondary separation were included. The indicator M N at the abscissa

denotes the upper limit for the formation of leading-edge vortex systems. Again
a scatter of up to ±10 percent is possible due to configuration particularities. The
MILLER/WOOD legend refers to [71].
Comparing Figs. 10.22 and 10.23, the Mach-number influence with regard to
attached flow (0) and vortex development (1) is much stronger for round-edged,
compared to sharp-edged wings. Region (2), in which fully developed and stable
vortex systems are present, is large for the sharp-edged, and small for the round-
edged wings. In the domain of higher angles of attack (3), nearly identical flow
patterns are present for both sharp and rounded leading edges. Differences are due
to effects of vortex breakdown, viz. adverse pressure gradients at the trailing edge
and/or shock-wave interactions. Large differences are present in the regions (4)–(7).
10.2 Non-linear Lift Exemplified with the Plain Delta Wing 297

Fig. 10.23 Possible flow patterns at wings with round leading edges [9]

10.2.7 Flow-Physical Challenges

The flow-physical challenges when dealing with lee-side vortex systems of small
aspect-ratio delta-like wings are manifold. In research this regards analytical, exper-
imental and numerical investigations, and in design of aircraft with such wings in
addition, for instance, the treatment of elastic properties of the whole airframe, sur-
face features, and unsteady flight conditions.
Here we look at the challenges, which already some basic flow-field features pose.
We follow the considerations of J.M. Luckring and O.J. Boelens in [72], where they
look at the phenomena present at the simplest possible lee-side vortex field over a
flat blunt-edged swept wing, Fig. 10.24. That was made in view of the phenomena
present at a diamond wing with the leading-edge sweep angle ϕ0 = 53◦ . We aim for
298 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.24 The flow-physical challenges of lee-side vortex systems of small aspect-ratio delta-like
wings [72]

a broader look at the challenges, and also have added two more challenges, no. 6 and
7.

1. Incipient separation at the blunt leading edge is a phenomenon where the under-
standing still is insufficient. The flow toward the leading edge evolves at the wind-
ward side of the wing from the primary attachment line, see Figs. 7.21 and 7.10.
The properties of this flow—streamline pattern and flow momentum, depending
to a degree on the shape, in particular the radius of the leading edge—play a role
and whether the flow is laminar or turbulent. The most complicated case would
be, if laminar-turbulent transition plays a role.
Transition is not only governed by the properties of the flow field as such, but also
by the surface properties, in particular the surface roughness and also by the level
of free-stream turbulence and the thermal state of the surface, see, e.g., [73]. This
all must be regarded, and that in particular, if experimental data or flight-test data
are in the background of an investigation.
2. The properties of the evolving primary vortex depend on the separation location
and the flow properties present there. In any case the vortex is different from a
vortex emanating from a sharp leading edge, also whether that is sharp in the
sense of the word, or whether it is to be considered as aerodynamically sharp,
Sect. 6.4.
3. The secondary vortex is induced by the primary one. It evolves from a separation
line—the secondary separation line, in the figure the full line between (3) and
(5)—and again it is to ask, whether laminar or turbulent flow is present along
this line. Laminar-turbulent transition at some location on this line—embedded
transition—would very much complicate the situation. In any case the evolution
of the secondary vortex has a repercussion on the primary vortex.
10.2 Non-linear Lift Exemplified with the Plain Delta Wing 299

4. The attached flow field portion of course is very strongly affected by the presence
of the primary and also the secondary vortex. A role plays the separation location
at the leading edge. Again also possible laminar-turbulent transition comes in.
The overall flow field topology moreover plays a role: open or closed lee-side
flow field, Sect. 7.4.3.
5. The matter of a possible inner vortex pair was considered above in Sect. 10.2.3. It
can be seen as a tertiary or as a second primary vortex pair. A question is, whether
it is a phenomenon appearing only at low Reynolds numbers and on top of that is
connected to laminar-turbulent transition.
6. The issue of the wind-tunnel environment is not restricted to the question of the
free-stream turbulence. In principle it is the test-section environment, which mat-
ters. Ideally it should have no influence on the experimental outcome. If it does,
numerical investigations of the case—which anyway should be made in the frame
of a collaborative approach—must treat the model-in-test-section situation.
Aeroelastic deformation under aerodynamic load of both the model and its sup-
port must be known and reduced as much as possible. Otherwise a numerical
investigation must take it into account, too.
7. Flow unsteadiness has two main aspects regarding the simulation with discrete
numerical methods. One is the large-scale massive separation, marked by unsteady
vortex shedding, also the unsteady strong shock-wave/boundary-layer interaction,
the other is the inherent small-scale unsteadiness of turbulent flow. The second
author of this book and J.M. Luckring recently gave a comprehensive review in
particular regarding the first aspect—separated flow simulation in view of flight
vehicle aerodynamics [63].
Problems of the first aspect cannot be simulated successfully with Model 10
methods, RANS methods, even with unsteady RANS, i.e., URANS methods. For
such problems scale-resolving methods, Model 11, appear to be the methods of
choice, with regard to computational effort in combination with RANS/URANS
methods as hybrid approaches.
The second aspect regards the fact that turbulent vortices, as well as turbulent
feeding layers, do not have, also like turbulent boundary layers, smooth edges,
see the remark on Sect. 1.5. The rugged unsteady edge-flow properties are the
source of dynamic loads, vibration excitation, etc. Whether they influence the
large-scale flow phenomena over the wing, is not known. In any case, if they
have an influence, numerical simulation must employ scale-resolving, Model 11,
methods.

This listing of flow-physical challenges is not intended to scare off the reader. It
is meant to show that a number of challenges still exists. The discussion of several
Unit Problems in the following sections shows that these challenges have been iden-
tified, and partly have been overcome. The former vortex-flow experiments VFE-1
and VFE-2 and the follow-up AVT task groups had their successes not least because
of their combination of experimental and theoretical/numerical work. Such collab-
orative approaches are the key to further improvement of the understanding of the
300 10 Small Aspect-Ratio Delta-Type Wing Flow

flow physics of lee-side vortex systems and their description for flight-vehicle design
purposes.

10.2.8 Manipulation of Lee-Side Flow Fields, an Overview

In the shape-definition process of a flight vehicle means to influence the vortex system
in relevant flight domains can be of interest [9]. Here we just note a few of them, in
Chap. 11 several sections can be found in this regard:

• Wing-planform shaping and optimization.

• Fuselage forebody strakes.
• Spanwise blowing over the wing.
• Geometrical alignment of the wing design.

10.3 Vortical Flow Past the Sharp-Edged VFE-1 Delta

Wing—Different Models

The first concerted international, now classical approach to the problem of vortical
flow past sharp-edged delta wings is the Vortex-Flow Experiment I (VFE-1), the first
Unit Problem of lee-side vortices, which we discuss. We have cited reports on VFE-1
in Sect. 10.1.2: [52–54]. A concise look at this type of vortical flow is given in [58].
The VFE-1 configuration is a cropped delta wing with sharp leading edges, short
sharp side edges, and nearly flat upper and lower surfaces, the latter except for the
support sting, Fig. 10.25.
Force and surface-pressure measurements, as well as flow-field surveys, all for a
number of flow conditions were performed in wind tunnels worldwide. Summarized
are the experimental results in [54].
The general features of the flow field found at a sharp-edged delta wing at geo-
metrical conditions, which lead to a lee-side vortex field, Fig. 10.11, are

• flow-off separation at the leading edges,

• the resulting vortex layers, containing both kinematically active and inactive vor-
ticity, roll up under eigen-induction, Sect. 3.12.1,
• and thus form the primary lee-side vortex pair,
• which on the lee side of the wing induces, via ordinary separation, a pair of
secondary vortices (even tertiary vortices are possible).
• This lee-side vortex system is connected to a strong lee-side surface pressure reduc-
tion, the suction pressure, which generates the non-linear lift, see also Sect. 8.4.3.
10.3 Vortical Flow Past the Sharp-Edged VFE-1 Delta Wing—Different Models 301

Fig. 10.25 The VFE-1 configuration [58]

Table 10.1 Geometrical and flow parameters of the cropped delta wing example
ϕ0 (◦ ) L (m) b (m) M∞ T∞ (K) Tw Re L α (◦ )
65 0.6 0.476 0.4 300 Adiabatic 3.1 · 106 9

Selected results of numerical simulations of such flow fields with different flow-
physical and mathematical models are now presented. We show and discuss results
of discrete numerical solutions of the Euler equations (Model 8, Table 1.3) and of
the Navier–Stokes/RANS equations (Model 9 and 10), the latter with two different
turbulence models.
The geometrical and flow parameters are given in Table 10.1.
For this sub-sonic case we find the normal angle of attack to be α N = 20.5◦ and
the normal Mach number to be M N = 0.18. For these numbers the correlation in
Fig. 10.22 yields that, region (2), a stable and fully developed leading-edge vortex
system is to be expected. Vortex breakdown is expected to happen around α = 20◦ ,
Fig. 10.20, an angle much higher than the present one.
The results presented in Fig. 10.26, generated by W. Fritz at EADS, München,
Germany, show some of the manyfold aspects—and pitfalls—to be considered when
using discrete numerical methods.
302 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.26 VFE-1 configuration without support sting: surface-pressure coefficient c p (η) at x/c =
0.6 (above) and 0.8 (below), experimental data with support sting [58]. Influence of describing
equations (Euler, RANS), grid resolution and turbulence model

In the upper left and the lower left part of the figure we immediately note that the—
grid-independent—Euler solution yields the (primary) suction peak more outboard
and much higher than that experimentally obtained. At the very leading edge we also
see a much higher value. The secondary separation and hence the secondary suction
peak, of course, is not present in the Euler solution.
The RANS solutions give an acceptable agreement with the experimental data
for the 193×129×81 “fine” grid, for which grid-independence can be assumed. The
primary suction peak agrees well with that found in the experiment. At the location
of the secondary vortex the suction peak is much more pronounced than in the
experimental data.
Two different turbulence models have been employed: (a) the Baldwin–Lomax
model with the Degani–Schiff modification, and (b) the Wilcox k − ω model. The
computations were made fully turbulent, i.e., possible laminar flow portions were
neglected.
10.3 Vortical Flow Past the Sharp-Edged VFE-1 Delta Wing—Different Models 303

In the upper right and the lower right part of Fig. 10.26 the results of the simu-
lations with these two models are shown. The differences are small, but visible and
significant. At the lower side of the wing all results are nearly identical. The differ-
ences to the experimental data are due to the influence of the support sting, neglected
in the computations.
Interesting differences become evident when looking at the computed total-
pressure loss and the eddy viscosity at the location x/c = 0.8, Figs. 10.27 and 10.28.
With both turbulence models the computed total-pressure loss indicates well the
boundary layers at the upper and the lower side of the wing, and the primary and the
secondary vortex at the upper side. In Fig. 10.28 with the Wilcox model the contours
are somewhat fuller than in Fig. 10.27. There the secondary vortex appears to be
larger and even a tertiary vortex seems to be indicated. The feeding layers, at least
of the primary vortex, are indicated only weakly.
The picture is quite different when looking at the computed eddy viscosity, each
in the lower part of the figures. With the Baldwin–Lomax model the eddy viscosity
appears, with limitations, only in the boundary layers, whereas in the vortex area it
is barely indicated. The picture is quite different with the Wilcox model, Fig. 10.28.
There all appears to be well pronounced, inboard of the vortex area, however, exag-
gerated.

Fig. 10.27 Baldwin–Lomax model: total-pressure loss (upper part) and eddy viscosity (lower part)
at x/c = 0.8 [58]
304 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.28 Wilcox k − ω model: total-pressure loss (upper part) and eddy viscosity (lower part) at
x/c = 0.8 [58]

These results tell us two things regarding the flow in the non-linear lift domain of
sharp-edged, low aspect-ratio lifting wings, here delta-type wings:

• Turbulence modeling in RANS (Model 10) solutions is of large importance. A

sufficiently exact simulation of the crucial vortex phenomena may be possible in
certain cases, in the present case (Unit Problem) to a degree even with the Baldwin–
Lomax model. If complex vortex ensembles, typical for aircraft configurations,
are present, a careful investigation is necessary in order to choose an adequate
turbulence model.
If laminar-turbulent transition is present, in general and in particular embedded
along attachment lines, it must be determined, whether it has a non-negligible
influence on the important flow features or not. If yes, then the topic of transition
prediction becomes important, if not, this is a “lucky situation”.
• In the present case, the two RANS solutions—also the one with the Baldwin–
Lomax Model even with a very erroneously computed eddy viscosity—yield rather
good results compared to the experimental data. Regarding the overall forces and
the pitching moment, even the Euler (Model 8) solution gives results with dis-
crepancies, which are acceptable in predesign work, see the next section. In that
section we also discuss in detail the flow field past the VFE-1 wing.
The reason for this reasonably good Euler solution is that to first order the proper-
ties of the inviscid flow field are the deciding ones, not the properties of the viscous
flow, the boundary layers. In Sect. 4.3 we have shown for large aspect-ratio wings,
how the appearance of lift can be understood: the second break of symmetry. We
have seen that at angle of attack overall a shear—representing the kinematically
10.3 Vortical Flow Past the Sharp-Edged VFE-1 Delta Wing—Different Models 305

active vorticity—is present in the streamline pattern of the external inviscid flow,
the upper and lower wing flow-fields shear.
This shear, although it cannot always be shown in the way as it is possible for the
large aspect-ratio wing, is present at any lifting wing, also in the present case. It
appears at the trailing edge and also at the leading edges. In the following section
we show this for the VFE-1 configuration in detail and give proof, that the mech-
anism, which leads to the lee-side vortices, is the same that leads to the trailing
vortex layer of large aspect-ratio wings.

10.4 Creation of Lift in the Euler Solution (Model 8) for the

Sharp-Edged VFE-1 Delta Wing—Proof of Concept

With this second Unit Problem we give proof that at non-linearly lifting sharp-edged
delta wings the generation of lee-side vortices, hence the generation of vorticity and
the rise of entropy, follows the same mechanism as we have seen at the trailing edge
of lifting large aspect-ratio wings, Sect. 8.4. Moreover, this is a proof that discrete
Euler (Model 8) solutions of such cases in principle are viable solutions.
We consider the flow past the small aspect-ratio VFE-1 delta wing, which we
already have treated in the last section. Figure 10.29 gives the coordinate convention.
The results were obtained in the frame of the doctoral thesis of R. Hentschel—a
doctoral student of the first author of this book—on three-dimensional self-adaptive
grid generation [51, 74]. Except for one figure, all figures in this section are from
that thesis.

Fig. 10.29 Planform of the considered sharp-edged cropped delta wing [51]. The lengths are non-
dimensionalized with the length of the wing, L = 0.6 m
306 10 Small Aspect-Ratio Delta-Type Wing Flow

10.4.1 The Computation Case and Integral Results

The case goes back to the International Vortex-Flow Experiment on Euler Code
Validation (VFE-1). The wing is the AFWAL configuration of NLR [75], see also
Fig. 10.25. The geometrical and flow parameters for this transonic case are given in
Table 10.2. The results discussed in the previous section were obtained with different
parameters.
Basically it was one of the objectives of the doctoral work [51] to show how
the vorticity and the lee-side vortices arise in a discrete modeled Euler (Model 8)
solution. The creation of the vorticity was studied by employing the local vorticity-
content concept, Chap. 4. A Navier–Stokes/RANS solution (Model 9/10) was made
for comparison. The Baldwin–Lomax turbulence model was used, however only at
the upper (leeward) side of the wing, due to indications from the experiment. The
flow at the lower (windward) side was treated as laminar flow.
The combination of the two characteristic parameters, the normal angle of attack
α N = 22.64◦ and the normal leading-edge Mach number M N = 0.38, is such that
in this case stable and fully developed leading-edge vortices as well as embedded
cross-flow shocks can be expected, according to the correlations given in Sect. 10.2.6.
Vortex breakdown for the present leading-edge sweep and angle of attack is of no
concern, because it is expected to happen around α = 20◦ , Fig. 10.20.
That the results of the Euler and the Navier–Stokes/RANS simulations are reliable
is demonstrated in Table 10.3 with the comparison of experimental and computed
force and moment coefficients. The data are selected from the detailed data given in
[51]. The maximum deviations are around 10 per cent.
We further show computed and measured surface pressure coefficients in the
cross-section at the location x = 0.6, Fig. 10.30. The data at this location are more
or less representative for the whole wing.

Table 10.2 Geometrical and flow parameters of the cropped delta wing example [51]
ϕ0 (◦ ) L (m) b (m) M∞ T∞ (K) Re L α (◦ )
Tw
65 0.6 0.476 0.85 300 Adiabatic 9 · 106 10

Table 10.3 Measured and computed force and moment coefficients [51]
Source CL CD CM L/D
Experiment NLR 0.4573 0.0846 –0.271 5.405
Euler simulation 0.5022 0.0831 –0.3030 6.047
Navier–Stokes/RANS simulation 0.4765 0.0890 –0.2846 5.354
10.4 Creation of Lift in the Euler Solution (Model 8) ... 307

Fig. 10.30 Comparison of computed and measured distributions of the pressure coefficient c p (y/s)
at the cross-section at x = 0.6 [51]. Open and full circles are experimental data, the full lines are the
interesting computed ones. Left: comparison with the Euler (Model 8) solution, right: comparison
with the Navier–Stokes/RANS (Model 9/10) solution

At the lower side of the wing we see a very good agreement of the computed and
the measured pressure coefficients. At the upper side we find the pressure distribution
typical for Euler simulations of such flow fields: the pressure peak is too large and its
location is too much to the outboard. All this is due to the missing secondary vortex.
Well discernible is the cross-flow shock.
The agreement of the Navier–Stokes/RANS data with the experimental data is
much better. The suction peak nearly has the right position. The peak value also is
too high, but not as much as that of the Euler simulation. The shock wave has a better
resolution.
The data show at the first glance, why lift, drag and the moment found with the
Euler solution are in a rather reasonable agreement with the experimental data. At
the lower side the surface pressure fields show a very good agreement, at the upper
side the deviations in the Euler data compensate each other to a degree. If this is the
case, in conceptual design work Model 8 simulations can be used without too big
risks. This, however, only holds for a given configuration class, which can be very
narrow.
We ask now what is to be expected regarding the flow field found with the Euler
simulation, in particular regarding the appearance of the lee-side vortices.
308 10 Small Aspect-Ratio Delta-Type Wing Flow

10.4.2 Details of the Computed Flow Field

In the reality (Model 1) besides the primary lee-side vortex pair generally a secondary
vortex pair is present, each vortex pair with the associated feeding (vortex) layers.
(Even tertiary vortices can appear.) This feature cannot be expected to be found in an
Euler simulation (Model 8). Figure 10.31 visualizes this schematically with cross-
section views of two virtual flow realizations for an open lee-side flow field. At the
left side the situation present in reality is sketched, at the right side that to be expected
from the Euler simulation. Possible cross-flow shocks are not indicated.
A detailed picture of the flow field found with the present Euler simulation is
given in Fig. 10.32. It regards the cross section at x ≈ 0.86—just at the beginning of
the straight side edge—in the span interval 0.23  2y/b  0.42. The side edge of
the wing is located at 2y/b = 0.3964.
Shown in the upper left part of the figure is the final self-adapted grid. The grid
was adapted in the computational space with the entropy gradient |∇s|. This sensor
yields very good results in the recognition of boundary layers, vortex layers, contact
discontinuities and shock waves.
We see that, beginning at the sharp side edge, the trace of the vortex layer (feeding
layer of the lee-side vortex) is well captured (see also the iso-lines of the total-pressure
loss, lower right figure). The vortex layer first curves away from the edge and then
comes around toward the center of the wing. The core area of the lee-side vortex lies
at about 80 per cent half-span. The grid there is refined down to the wing’s surface.
Located in this domain is also the expected cross-flow shock.
The vortex layer is well discernable too in the Mach number plot (lower left) and
in the total-pressure loss plot (lower right). A spiraling of the vortex layer is not
observable. At 2y/b ≈ 0.251 the total-pressure loss plot indicates the location of the
secondary attachment line A2 , Fig. 10.33, right side. The—inviscid—surface flow at
2y/b  0.251 is in outward direction, showing a vortex-layer structure.

Fig. 10.31 Schematics, not to scale, of vortex systems (open lee-side flow fields) in Poincaré
surfaces of delta wings at higher angles of attack (view from behind onto the right-hand side of
the wing). Left: perceived reality (Model 1) as well as expected result of Navier–Stokes/RANS
simulations (Model 9/10), right: expected result of Euler simulations (Model 8)
10.4 Creation of Lift in the Euler Solution (Model 8) ... 309

Fig. 10.32 Cross section of the flow field at x ≈ 0.86 [51]. Upper left part: the self-adapted grid;
upper right part: iso-pressure coefficient lines (–1.6  c p  0.02); lower left part: iso-Mach number
lines (0.7  M  2.02); lower right part: iso-lines of total-pressure loss (0  1 − pt / pt∞  0.35)

The cross-flow shock is located at 2y/b ≈ 0.35. It lies below the lee-side vortex
and does not appear to extend far upwards. Outward its location no vortex-layer
structure is visible. Hentschel notes that details of the computed flow pattern can
depend much on the grid’s topology and fineness.
Surface streamlines of the computed inviscid flow field are shown in Fig. 10.33.
(Note that we show only one half each of the lower and the upper side.) The primary
attachment lines A1 at the lower side of the wing have a quasi-conical appearance.
They are located close to the leading edges. Between them the flow is only nearly
two-dimensional because the lower side of the wing is not fully flat.12 In the small
domains between the primary attachment lines and the leading edges the flow is in
outward direction with slightly curved streamlines. The primary separation lines S1
lie at the very leading edges, because these are sharp. We have flow-off separation
there.
We have indicated the location of the primary separation line S1 at the sharp
leading edges. The lee-side vortex leads to a secondary attachment line A2 . Between
the secondary attachment lines we see a more or less two-dimensional flow toward
the wing’s trailing edge. This points to an open lee-side flow field. The secondary
attachment line tapers off at about 90 per cent of the wing’s length. This indicates
that there the lee-side vortices lift off from the surface.

12 A more or less truly two-dimensional pattern inboard of the primary attachment lines is only to be

expected at the fully flat windward side of a configuration. Such a pattern is always desired in view
of the onset flow into inlets of propulsion systems and over aerodynamic trim and control surfaces.
310 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.33 Surface

streamlines of the Euler
simulation [51]. Left: lower
(windward) side, right: upper
(leeward) side (open lee-side
flow field). Primary
attachment line: A1 , primary
separation line (sharp
leading edge): S1 , secondary
attachment line: A2 . The
flow-off directions at the
leading edges and the sharp
side edges are denoted by ν

Between A2 and the leading edge with S1 the surface flow is directed outboard.
The cross-flow shock, Fig. 10.32, leads to a kink in the surface streamlines. At the
leading edge finally the flow-off from the upper surface evidently has a larger angle
than at the lower side: ν2 > ν1 . The resulting angle ν2 − ν1 between the flow at the
upper and the lower side is the leading-edge flow shear angle ψeL E . In Fig. 10.37 we
come back to the shear-angle distribution.
At this point we have to remark that it has not been possible to explain the flow-
field situation as we did it for the lifting large aspect-ratio wing, Sect. 4.3.2. There
we could show that a shear exists between the inviscid flow fields at the pressure
and the suction side of the wing, the upper and lower flow-fields shear. The break of
symmetry of the surface streamline pattern (the second break) then could be seen as
the origin of the wing’s lift.
Back to our wing: at the side edge of the wing the flow-off angles are the other
way around: ν4 < ν3 . This means that the side-edge flow shear angle ψeS E is negative,
Fig. 10.37. Hence at the wing’s straight side edges each a counterrotating secondary
vortex develops. (This is not to be confused with the secondary vortices arising below
the primary ones, if viscous flow is considered.) The phenomenon of counterrotat-
ing vortices at straight side edges was reported also in [76], where numerical and
experimental results are presented.
At the inner part of the trailing edge the shear angle ψe is nearly zero. This means
that there only little kinematically active vorticity content is leaving the wing’s trailing
edge. Regarding the flow field at and downstream of the trailing edge of delta wings
see also Sect. 3.12.3.
From the Navier–Stokes/RANS solution in [51] we discuss only the skin-friction
line pattern at the wing’s lower and upper surface, Fig. 10.34. At the lower side—left
10.4 Creation of Lift in the Euler Solution (Model 8) ... 311

part of the figure—the pattern is quite similar to that found with the Euler simulation.
The primary attachment line A1 again lies close to the leading edge.
The skin-friction lines toward the leading edge and in the domain between the
primary attachment lines, however, show a pattern slightly different to that given by
the Euler solution in Fig. 10.33. As long as no fundamental difference exists between
the Euler and the Navier–Stokes/RANS simulation, this must be expected. Even if the
boundary-layer flow is only weakly three-dimensional, the skin-friction line pattern
in principle is different from the streamline pattern of the appendant external inviscid
flow [73].
At the upper side of the wing again we have got an open lee-side flow field. The
secondary attachment line A2 has a location quite similar to that of A2 found with
the Euler simulation. However, it tapers off much earlier, which seems to indicate
that the lift-off of the lee-side vortex happens earlier than with the Euler simulation.
At this time, though, we must allow for the possibility that this result is due to the
applied turbulence model. Anyway the question arises whether turbulent attachment
happens along A2 or laminar-turbulent transition. (The influence of laminar or tur-
bulent flow on the strength of the lee-side vortex pairs of delta wings is discussed in
the Concluding Remarks, Sect. 10.8.)
If we accept the solution in this regard, we observe now the fundamental differ-
ence to the Euler result: the—primary—lee-side vortex pair induces a pair of small
secondary lee-side vortices. The feeding layer leaves the surface at the secondary
separation line S2 . Of course then between S2 and the primary separation line S1

Fig. 10.34 Skin-friction

lines of the
Navier–Stokes/RANS
simulation [51]. Left: lower
(windward) side, right: upper
(leeward) side.
Nomenclature see Fig. 10.33
312 10 Small Aspect-Ratio Delta-Type Wing Flow

at the sharp leading edge the—tertiary—attachment line A3 must be present. The

flow from the tertiary attachment line toward the primary separation line S1 at the
sharp leading edge is well discernible. The presence of a cross-flow shock as well
of counter-rotating side-edge vortices cannot be deduced from the skin-friction line
pattern.

10.4.3 The Circulation and the Kinematically Active Vorticity

Content in the Euler Simulation

We give a very compact summary of the results of [51] regarding the connection of
the circulation and the kinematically active vorticity content in the Euler simulation
(Model 8) of the flow past the cropped delta wing.
The circulation Γ in the lee-side vortices of the wing is found by the integration
of the vorticity in surfaces F normal to the longitudinal axis—the x-axis—of the
wing (theorem of Stokes, Sect. 3.3):
  
Γ = v ds = ω d F. (10.9)

The result is given with the upper curve in Fig. 10.35. The circulation Γ increases
along the x-axis—represented here by the (minus) i-coordinate—due to the vorticity
fed into the flow field by the two vortex layers, which originate at the sharp leading
edges of the wing.
The circulation increases until the straight side edge is reached (location 3). There
the secondary vortices due to the straight side edges, which counter-rotate to the
primary vortices, introduce negative circulation. Hence the total circulation decreases
until the constant value behind the wing is reached (left of location 2 at i = 20). That
value is only approximately constant, because of the coarse grid there, see in this
regard also Sect. 5.3.
Regarding the vorticity content we first consider locally the geometry of the vortex
layer, which leaves the sharp leading edge of the wing, Fig. 10.36. Sketched is an
element of that layer with the width ds. The layer has the thickness dn. The grey
surface lies approximately parallel to the wing’s surface, cutting the vortex layer at
right angles. Indicated on it is the decomposed wake structure in the idealized reality,
Fig. 4.12, right part. At the right in Fig. 10.36 the broken line, designated “inviscid
shape”, would be the uniform u(z)-profile of the idealized inviscid flow. Together
with the other profile it represents the ideal Euler wake shown in Fig. 5.1, right part.
Due to the properties of the present vortex layer, which are more complex than
those of the shear layers treated in Chap. 4, the vorticity content was found with the
integration of the vorticity 
Ω= ω dn (10.10)
10.4 Creation of Lift in the Euler Solution (Model 8) ... 313

Fig. 10.35 Circulation Γ , its derivative dΓ /di and the vorticity content Ω ds (circles) found with
the Euler solution [51]. The flow is from the right to the left. Note that the coordinate i decreases
in downstream direction

Fig. 10.36 Schematic of a sharp leading edge with emanating vortex layer [51]
314 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.37 Leading-edge flow shear angle ψ (= ψe ), vorticity content Ω/|U | and the absolute
value |U | as function of the location i [51]

in the following way.

In the case of the Euler wake ω results from the vortex-resembling velocity profile
lying in the grey surface element cutting the vortex layer in Fig. 10.36. The change
of the circulation between location i 2 and i 1 is then connected to the integral of the
vorticity in that surface element according to the compatibility condition Eq. (4.20):
 
dΓ
= ω dn ds. (10.11)
di

In Fig. 10.35 the result is shown. Note that the coordinate i decreases in down-
stream direction, hence the derivative dΓ /di initially goes negative. Negative is also
ds. The wiggly development of Γ , reflected in dΓ /di, is due to different refinement
levels of the grid. Nevertheless, the agreement between dΓ /di and Ω ds is satis-
factory. Close to the maximum of Γ the vorticity content decreases and the sign of
dΓ /di changes properly. The general development and the minimum of dΓ /di are
well reproduced by Ω.
The compatibility condition, Eq. (4.20), holds for sub-critical flow, i.e., shock-free
flow with the same absolute value of the velocity vector at the upper and the lower
side of the leading edge. In our case a cross-flow shock is present at the upper side
of the wing. Its influence on the velocity at the leading edge appears to be relatively
weak as Fig. 10.37 indicates.
The values 2 tan ψ/2 and Ω/|U | are in satisfactory agreement. Along the swept
leading edge the leading-edge shear angle results to ψe ≈ 30◦ to 40◦ . At the location
of the straight side edge the angle becomes negative, pointing to the presence of the
side-edge counterrotating vortex.
This short summary in particular shows how in the discrete modeled Euler solution
(Model 8) for the sharp-edged delta wing the creation of the lee-side vortex pair can
10.4 Creation of Lift in the Euler Solution (Model 8) ... 315

be understood. A kinematically active vorticity content is introduced into the flow by

the shear of the flow directions at the upper and the lower sides of the sharp leading
edges. The two resulting vortex layers curl up by self induction and are forming the
pair of lee-side vortices. These then are responsible for the non-linear lift increment.
With ordinary separation at regular surfaces the introduction of kinematically
active vorticity content into the separating shear layer cannot be demonstrated as
simply as in this case. In principle, however, the mechanism should be the same.

10.5 Vortical Flow Past the Round-Edged VFE-2 Delta

Wing

In the year 2001 D. Hummel and G. Redeker proposed a new vortex-flow experiment,
the VFE-2 [58]. The configuration was to be chosen such that the following flow
regimes were covered:
(i) Attached flow without vortex formation (0◦  α  4◦ ).
(ii) Separated vortical flow without vortex breakdown (4◦  α  20◦ ).
(iii) Separated vortical flow with vortex breakdown (20◦  α  40◦ ).
(iv) Separated deadwater-type flow (40◦  α  90◦ ).

10.5.1 The Wing and the Subsonic Computation Case

An appropriate configuration was a plain delta wing with a leading-edge sweep of

ϕ◦ = 65◦ . The wing should have a flat inner portion and interchangeable leading
edges. These demands were fulfilled by the NASA NTF delta wing configuration,
which is shown in Fig. 10.38. The aspect ratio of the wing is 1.85, three rounded and
one sharp leading edge were available.
The general aim of VFE-2 was to provide flow-field-data for comparisons
with numerical results [77]. Experimental topics were laminar-turbulent transition,
surface-pressure measurements in the onset-of-separation domains of configurations
with round leading edges, boundary-layer measurements, skin-friction line patterns
in view of the determination of secondary and tertiary separation lines, flow-field
measurements regarding the primary and secondary vortices (components of veloc-
ity and vorticity, turbulent energy and eddy viscosity), vortex-breakdown flow fields
including surface-pressure fluctuations caused by the spiral mode of breakdown.
Numerical topics were validation and improvement of the existing codes, code-
to-code comparisons on common structured and unstructured grids, investigations of
turbulence models in RANS (Model 10) simulations, assistance related to the setup
and evaluation of new wind-tunnel experiments, synergistic effects through test runs
on the simple VFE-2 configuration prior to full-scale aircraft investigations.
316 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.38 The NASA NTF delta-wing configuration [59]. Mean aerodynamic chord c = 2/3· c R ,
1 in. = 2.54 cm

The work on VFE-2 took place from 2003 to 2008 within the RTO Task Group
AVT-113. In the special issue “VFE-2” of Aerospace Science and Technology,13
edited by D. Hummel and R.M. Cummings, in eight papers reports on that work
were given. Summarizing articles are—besides others—those by J.M. Luckring and
D. Hummel [78], and W. Fritz and R.M. Cummings [79].
We discuss selected results of the investigations. The flow parameters are given
in Table 10.4. Note that experimental parameters can be slightly different.
The normal angle of attack and the normal leading-edge Mach number of this case
are α N = 29.22◦ and M N = 0.188. This case hence lies in the regime (2) “develop-
ment of the vortex system” in Fig. 10.23 and therefore is covered by the experimental
data underlying the correlations there. The critical angle of attack regarding vortex
breakdown is α = 20◦ , i.e., well above of that of the present case.

Table 10.4 Flow parameters of the subsonic VFE-2 case

M∞ Re∞,c c (m) T∞ (K) Twall α (◦ ) Viscous
flow
0.4 3 · 106 0.4358 297.4 Adiabatic 13 Fully
turbulent

13 Volume 24, Issue 1, pp. 1–294 (January−February 2013).

10.5 Vortical Flow Past the Round-Edged VFE-2 Delta Wing 317

10.5.2 Two Pairs of Primary Vortices

Experimental investigations at DLR Göttingen with the parameters given in Table 10.4
yielded for the medium-radius rounded leading edge an interesting picture of vortex
flow: co-rotating vortices, i.e., a “doubled primary” vortex system [80].
In the upper part of Fig. 10.39 the surface pressure is given, obtained with Pressure
Sensitive Paint (PSP) measurements, together with the velocity field, found with
Particle Image Velocimetry (PIV) measurements in cuts at x = const. locations [81].
In the lower part the result of a numerical simulation (RANS with the k-ω model)
of the flow field at EADS Munich is in a convincing agreement with the experimental
result [82]. The figure gives total-pressure loss contours at x = const. locations.
Regarding the effect of laminar-turbulent transition in the actual flow field see the
concluding remarks in Sect. 10.8.
Overall the experimental and the numerical simulation give the following picture:
at about x = 0.4 an inner vortex pair appears, at about x = 0.5 an outer vortex pair
appears, around x = 0.7 both vortex pairs appear to have the same size, further
downstream the outer vortex pair becomes stronger, as is to be expected, whereas
the inner pair decays.
A look at the computed skin-friction line pattern provides deeper insight, Fig. 10.40.
The RANS computation was made with DLR’s TAU code, fully turbulent, employ-
ing the Spalart–Allmaras turbulence model [83]. We note that the computation was
made with a slightly larger angle of attack than before: α = 13.3◦ .
The figure gives the top view at the left-hand side of the wing. The free-stream is
in direction from the left lower corner of the figure upward along the center line of
the wing.
In Sect. 10.2.4 we have noted that at a wing with round leading edge the formation
of the (primary) lee-side vortex begins at the rear of the wing and with increasing
angle of attack moves forward toward the wing’s apex. In our case the angle of attack
is such that the primary vortex originates at about 45 per cent of the wing’s length.
Hence at the apex of the wing and downstream of it up to x/c R ≈ 0.4 the pattern
of the skin-friction lines indicates that the flow from the lower side passes without
separation around the round leading edge to the upper side of the wing.14 In the
VFE-2 literature this is attributed to the rather large value of the cross-flow bluntness
 
parameter pb = r L E /b , with b being the local span, Sect. 6.4.
In our case at x/c R = 0.1, for instance, we have for the medium-radius leading
edge pb = 0.0107. For this value the inverse thickness ratio of the flattened ellip-
tical cylinder of our gedankenexperiment in Sect. 6.4 is δ = 46.7. Figure 6.4, right
part, shows for δ = 50 a peak value c p = −2,500 at ϕ = 90◦ . The magnitude of
the adverse pressure gradient, Fig. 6.5, right part, is dc p /dϕ = 1,471 at ϕ = 90.7◦
behind the ‘leading edge’.
At x/c R = 0.7 we have for the medium-radius leading edge pb = 0.00153. The
inverse thickness ratio then is δ = 326.48. Figure 6.4, right part, now shows for

14 This is similar to the flow pattern at the blunt delta wing, Sect. 10.7: the laminar flow goes around

the blunt leading edge and separation and vortex formation happens at the upper side of the wing.
318 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.39 Pressure (surface color), velocity (vectors), and vorticity (vector color) distributions
of the VFE-2 configuration with medium-radius rounded leading edges. Upper part: PSP and PIV
measurements [81]. Lower part: numerical solution [82]
10.5 Vortical Flow Past the Round-Edged VFE-2 Delta Wing 319

Fig. 10.40 Surface pressure and skin-friction lines at the left upper side of the VFE-2 delta wing
[83]. The dashed red lines indicate the locations of the vortex axes

δ = 300 a peak value c p = −90,000 at ϕ = 90◦ . The magnitude of the adverse

pressure gradient is dc p /dϕ = 278,940 at ϕ = 90.2◦ , behind, but very close to the
‘leading edge’.
If we project these results on our case, we see that with constant radius r L E the
leading edge at the front part of the wing indeed acts as the round one, which it is.

But at the rear of the wing the increasing b leads to an effective decrease of the
bluntness parameter. The leading edge becomes relatively sharper, fixing there the
primary separation location like the aerodynamically sharp leading edge does.
The flow around the effectively blunt leading edge is present up to x/c R ≈ 0.4,
however, at x/c R  0.2 apparently with a small bubble separation directly at the
leading edge, where a thin strip of blue color indicates low pressure.
In any case up to x/c R ≈ 0.4 the skin-friction lines first point away from the
leading edge and then curve around into a direction first parallel to the leading edge
and then toward it. At the same time from the center line of the wing skin-friction lines
converge toward those coming from the leading edge eventually forming a pattern
of converging lines, which typically is found upstream of open-type separation, see
for example Fig. 7.21 in [73]. All that can be considered as the domain of incipient
separation.
320 10 Small Aspect-Ratio Delta-Type Wing Flow

At x/c  0.45 on the wing’s surface the outer broken red line depicts the location
of the center of the outer primary vortex. Below it we see in blue the induced low
pressure domain. The outer primary vortex originates immediately at the leading
edge.
When following a path from the leading edge of the wing toward its center line,
we find a succession of singular lines, however, being not in all cases unambiguous:

• the primary separation line (barely visible), S1 , leading to the outer primary vortex,
• the primary attachment line, A1 ,
• the secondary separation line, S2 ,
• the secondary attachment line, A2 ,
• the tertiary separation line, leading to the inner primary vortex, S3 ,
• the tertiary attachment line, A3 ,
• the fourth separation line, S4 .

Figure 10.41 schematically shows at x/c R ≈ 0.7, with some reservations, the flow-
field in the Poincaré surface. All indicators point to an open lee-side flow field. At the

lower side of the wing A0 denotes the (left-hand) attachment line, and at the upper
side F1 the outer primary vortex, F2 its secondary vortex, and F3 the second (inner)
primary vortex. The attachment lines A show up as quarter-saddle points S  at the
lower side of the wing and as half-saddle points S  at its upper side.
We have not sketched a possible secondary vortex due to the inner primary vortex,
because we see no clear indication for that. Neither the skin-friction line pattern nor
the surface pressure distribution point to a secondary inner vortex.
To check the topological structure shown in Fig. 10.41, we apply Rule 2’ from
Sect. 7.4. Of course we also have to take into account the singular points at the other
side of the wing. Rule 2’ then reads

Fig. 10.41 Left side of the wing: topological schematic of the open lee-side flow-field in the
Poincaré surface at about x/c R ≈ 0.7. The view is in positive x-direction. One prime denotes a
half-saddle point, two primes denote a quarter-saddle point
10.5 Vortical Flow Past the Round-Edged VFE-2 Delta Wing 321

Fig. 10.42 Surface pressure and streamlines (particle traces) at the left upper side of the VFE-2
delta wing [83]

     
1  1   1   13 2
N+ N − S+ S + S = (6 + 0) − 0 + + = −1,
2 2 4 2 4
(10.12)
telling us that the topological structure is a viable one.
The formation of the inner primary vortex in this case to a degree seems to be
coupled to the formation of the outer primary vortex, Fig. 10.42. Particle traces from
within and from above the boundary layer show that, see also Sect. 10.8.
The convergence of the skin-friction lines, noted above, is seen here to lead to a
thickening of the boundary layer ahead of the separation area. This of course indicates
a rise of the kinematically active and inactive vorticity content. The particle traces,
as to be expected, mainly enter the inner primary vortex, a few of them also the outer
one.
Downstream of this area, due to the decreasing bluntness parameter, the outer
primary vortex strongly is fed with kinematically active vorticity, which is not the
case for the inner primary vortex. Hence downstream the latter decays due to the
always present viscous effects.
An experimental investigation of a larger VFE-2 configuration at a lower Reynolds
and Mach number than that discussed here was performed by A. Furman and the
third author of the present book [84]. We shortly discuss an aspect of that case. The
flow parameters are given in Table 10.5.
322 10 Small Aspect-Ratio Delta-Type Wing Flow

Table 10.5 Flow parameters of the selected low sub-sonic case VFE-2 case of [84]
M∞ Re∞,c c (m) T∞ (K) Twall α (◦ ) Viscous
flow
0.14 2 · 106 0.98 292.5 Adiabatic 13 lam./turb.
transition

Whereas in the case discussed above possible bubble-type laminar-turbulent tran-

sition was present at x/c R ≈ 0.3, [78], in the present case at the upper side of the
wing laminar flow was observed from the apex downstream to x/c R ≈ 0.3, Fig. 10.43.
Again an inner and an outer primary vortex is observed (right-hand side of the wing).
The inner vortex originates from ordinary laminar separation (blue color). Further
downstream, after laminar-turbulent transition of the boundary layer has happened,
the inner vortex is fed by turbulent flow (red color).
The outer primary vortex is sketched as initially originating via laminar, but then
mainly via turbulent separation. Basically we see the same skin-friction line pattern
and the same open lee-side flow field as shown in the Poincaré surface in Fig. 10.41.
The outer primary separation line S1 lies well at the upper side of the wing.

Fig. 10.43 Surface oil-flow pattern (left) and sketch of the flow topology (right) of the VFE-2
configuration with medium-radius rounded leading edges [84]
10.5 Vortical Flow Past the Round-Edged VFE-2 Delta Wing 323

10.5.3 Vortex Breakdown

In Sect. 10.2.5 we have shown that limits exist of wing slenderness and angle of
attack because of the possibly occurring phenomena of vortex breakdown and vortex
overlapping. Vortex breakdown was an important topic of the VFE-2 work. Following
the summary given in [78], as well as the discussions in other VFE-2 papers, we give
a short overview.
Important first of all is the observation that at high angles of attack the flow past the
wing may become unsteady. This, of course, has consequences for both experimental
and numerical investigations. Regarding the latter, we note that unsteady Reynolds-
averaged Navier–Stokes (URANS) solutions (Model 10) should be employed, pos-
sibly also scale-resolving methods like Large-Eddy Simulation (LES) as well as its
derivatives Detached-Eddy Simulation (DES) and Delayed Detached-Eddy Simula-
tion (DDES), all belonging to the class of Model 11 methods.
– Subsonic Case, M∞ = 0.4 For the VFE-2 configuration with the leading-edge
sweep angle ϕ0 = 65◦ the correlation in Fig. 10.20 yields the critical angle of attack
αbd ≈ 20◦ .15
Vortex breakdown begins in a region of strong streamwise deceleration, located
commonly at the wing’s trailing edge. With increasing angle of attack its onset moves
upstream. At the VFE-2 wing vortex breakdown of spiral type, Sect. 3.13, of the outer
primary vortex was observed. In the subsonic domain (M∞ = 0.4) it began around
α = 20◦ . For the sharp and the medium-rounded leading edge it had progressed over
the whole wing already at α = 23◦ .
In a numerical study—RANS with the k-ω turbulence model—by W. Fritz it was
found that weak vortex breakdown can appear at a lower angle of attack than αbd
≈ 20◦ [82]. The computation was made for the medium rounded leading edge at
a higher Reynolds number than that given in Table 10.4, viz. Re∞,c = 6 · 106 , the
angle of attack was α = 18◦ .
The outer and the inner primary vortex were found to be located close to the apex
(again with coupled origins), the secondary vortex is not visualized, Fig. 10.44. The
inner primary vortex appears to be weak with little influence on the pressure field.
The core of the outer primary vortex is compact up to x/c R ≈ 0.8. Downstream of
this location a spiral-type vortex breakdown is present. Behind the wing’s trailing
edge the vortex appears to be re-configured, Sect. 3.13.
The pink-colored bubble represents an iso-surface of zero axial (x-direction)
velocity. Inside the bubble the axial velocity is negative, which is an indicator of
vortex breakdown, being in this case a weak one. Ahead and behind the pink bubble
the axial velocity is positive. At the wing’s thick rounded trailing edge, Fig. 10.38,
the pink color indicates ordinary separation, the kink at the outside probably being
due to the secondary vortex, which locally suppresses separation. The inner primary
vortex appears to be healthy throughout.

15 Note that the VFE-2 wing in this case has a blunt trailing edge, Fig. 10.38, whereas the correlation

is based on wings with sharp trailing edges. The result hence must be seen with a certain reservation.
324 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.44 Computed flow field at α = 18◦ . Right-hand side of the wing: surface pressure, left-hand
side: stream traces with 3-D volume ribbons [82]

Why was vortex breakdown found at a lower angle of attack than expected? The
author attributes it to uncertainties regarding the location and the influence of laminar-
turbulent transition and to deficits in the turbulence modeling, in his simulation being
the k-ω model. The comparison with experimental data, not shown here, reveals
differences in the flow-field data at the front of the wing, whereas the agreement is
good at the rear of the wing.
In any case, Fig. 10.44 tells us why vortex breakdown is of concern for the aero-
dynamic design of flight vehicles with delta wings. Vortex breakdown usually begins
at the rear of the wing, in our case at around 70 per cent of the wing’s length. The
suction effect of the lee-side vortex system—low surface pressure—and with that
the non-linear lift, is reduced. Hence a loss of lift occurs. Critical is that this only
happens at the rear of the wing. Therefore the lift loss goes together with a pitch-up
of the wing, which can be rather strong.
If the breakdown occurs asymmetrically, a sideslip and a rolling moment results.
If with increasing angle of attack the breakdown has moved up to the wing’s apex,
the non-linear lift has gone completely. This all may go together with effects of
unsteadiness.
– Transonic Case, M∞ = 0.85 In the summarizing paper [78] it is discussed how
in a sense premature vortex breakdown in the transonic regime (M∞ = 0.85) can be
understood. The onset of breakdown was found to happen suddenly at some place
in the middle of the wing. It possibly was triggered by an interaction of the vortex
10.5 Vortical Flow Past the Round-Edged VFE-2 Delta Wing 325

Table 10.6 Flow parameters of the transonic sharp-edged VFE-2 case of [85]
M∞ Re∞,c c (m) T∞ (K) Twall α (◦ ) Viscous
flow
0.85 6 · 106 0.436 281.4 Adiabatic 23 Turbulent

with the terminating shock wave of the local supersonic flow regime formed by the
sting mount.
A detailed account of shock-wave effects on the vortex breakdown at the VFE-2
wing with sharp leading edges was given by S. Crippa, a doctoral student of the second
author of this book [85], see also [86]. Crippa performed Detached Eddy Simulations
(DES, Model 11), and we discuss a few of his findings. The flow parameters are given
in Table 10.6.
For this case we find the normal angle of attack and the normal leading-edge
Mach number to be α N = 45.13◦ and M N = 0.47. This case lies in the regime (3)
“vortices fixed in span direction, …, embedded shock waves possible” in Fig. 10.22
and hence is covered by the experimental data underlying the correlations there. The
critical angle of attack regarding vortex breakdown, now with the valid correlation
for sharp-edged delta wings, is as quoted already above α = 20◦ .
Figure 10.45 gives the computed vortex-breakdown location as function of the
time for the last computation cycle. The breakdown location is defined as the foremost
chordwise location, where fully-reversed flow in the outer primary vortex is present.
An inner primary vortex was not detected at that angle of attack.
The figure demonstrates the high degree of unsteadiness of vortex breakdown.
The vortex breakdown location rather suddenly moves within ≈7 ms downstream

Fig. 10.45 Vortex-breakdown location as function of the time [85]. Time steps denoted by red dots
are those of which details of the flow field are presented in Figs. 10.46 and 10.47
326 10 Small Aspect-Ratio Delta-Type Wing Flow

from x/c R ≈ 0.55 to x/c R ≈ 0.73. The subsequent upstream motion then is rather
gradual.
The reason for the different gradients of the upstream and the downstream motion
can be found by an inspection of the flow field in the symmetry plane of the wing.
Note that the tip of the support sting lies at x/c R = 0.63, Fig. 10.38. We do the
inspection in the following Figs. 10.46 and 10.47 for the red-dot locations shown in
Fig. 10.45.
Of the two figures the first one, Fig. 10.46, shows the flow-field properties at the
time step t = 0.1492 s for a better inspection in an enlarged mode. We see a single
shock wave at x/c R ≈ 0.51, which moves upstream (indicated by the black arrow in
the lower left frame). The associated surface-pressure coefficient (c p ) distribution is
shown in the upper left frame. (This frame lies at and near the center line of the wing,
see the upper right frame.) The flow comes from the left, the nose of the support sting
is located at the lower right corner, the foremost part of the reversed-flow isosurface
(grey indicates zero axial velocity) is at the upper right corner. The low c p ahead of the
shock wave is in green, yellow and red. The bundling of the iso-c p -lines indicates the
location of the shock wave, behind it the high pressure is in light blue. The pressure
rises across the local wing span toward the center line, though to different degrees, the

Fig. 10.46 Sub-frames showing instant flow-field properties at the time step t = 0.1492 s [85].
Upper left frame: view from above at the suction side of the wing ahead of the support sting with the
surface-pressure distribution, flow comes from the left. Upper right frame: frontal-isometric view
of the right-hand half-span suction side with the surface pressure distribution and the reversed-flow
isosurface (grey). Lower left frame: surface-pressure coefficient c p (x/c R ) at the intersection of the
symmetry plane and the wing surface, location corresponds to the lower edge of the figure above
it. Lower right frame: normal view along the symmetry plane with the local Mach-number range,
blue = subsonic, green = sonic, red = supersonic
10.5 Vortical Flow Past the Round-Edged VFE-2 Delta Wing 327

Fig. 10.47 Sub-frames showing instant flow-field properties at different time steps [85]. Upper left
part: t = 0.1547 s, upper right part: t = 0.1560 s, lower left part: t = 0.1586 s, lower right part:
t = 0.1650 s. For the contents of the sub-frames see the legend of Fig. 10.46

suction effect of the lee-side vortex is gone. Unsteady lift loss and upward pitching
moment increments are the consequence.
The lower right frame shows that in the symmetry plane shortly downstream of
the wing’s apex a supersonic flow pocket is present, terminated by a curved shock
wave, which impinges nearly orthogonal on the wing’s surface.16 At the support
sting the flow again is accelerated to supersonic speed. That flow pocket then is
terminated close to the trailing edge of the wing. The vortex breakdown at x/c R ≈
0.57 is associated with the first shock wave, as indicated in the upper left frame.
The presence of two shock waves needs a closer inspection. The first shock wave,
whose footprint crosses the whole span of the wing (upper left frame), is the primary
one.17
The second shock wave appears to be solely due to the support sting, which
is a half-ogive followed by a half-cylindrical part. The half-ogive induces by its
displacing effect a flow expansion right from its blunt nose—like a blunt body does,
in this case in transonic flow. The flow speed becomes supersonic (red color) and
then is terminated by the nearly normal shock wave.
That occurs at the tip of the half-ogive shaped forebody of the sting, at the inter-
section with the cylindrical rear part of the sting. There a discontinuity of the surface

16 For inviscid two-dimensional flow over a regular surface theory demands orthogonal impinge-

ment, Sect. 4.2.3.

17 How and whether it interacts with the to be conjectured cross-flow shocks is not clear. In [87] it

is shown for a similar case that the embedded cross-flow shock wave lies very close to the wing’s
surface between the primary and the secondary vortex, see also Fig. 10.32.
328 10 Small Aspect-Ratio Delta-Type Wing Flow

curvature is present, barely detectable in the side view (lower right part) of Fig. 10.38.
That discontinuity induces the terminating recompression shock wave, which in the
respective frames in Fig. 10.47 is also seen to occur—approximately—at the same
location. The shock wave hence is the trace of the stand-up collar-like shock wave
surface, standing around the tip of the half-ogive shaped forebody of the support
sting. However, whether and how it possibly interacts with the vortex-breakdown
evolution is not clear.
At the next time step, t = 0.1547 s, upper left part in Fig. 10.47, in the symmetry
plane the rear supersonic pocket forms a kind of a bridge to the flat wing portion
ahead of the tip of the support sting. Accordingly a double-shock system lies ahead
of the sting tip, indicated by the two horizontal bars in the lower left frame of the
upper left part of the figure.
The first, now weaker shock has moved upstream to x/c R ≈ 0.48, the—also
weak—second shock is located at x/c R ≈ 0.6 close to the sting’s tip. The vortex
breakdown has jumped back—conjectured to be due to this shock arrangement, see
above—from the position behind the first shock to a position behind the second
shock, upper right frame of the upper left part of the figure. Possibly the weakening
of the first shock wave and the forward motion of the second one induces this kind
of abrupt pass-over of the breakdown to the rear position.
The frame in the upper right part of Fig. 10.47, 0.0013 s later, shows that the two
supersonic regions in the symmetry plane have merged. The downstream motion
toward the sting tip of the first shock wave slowly continues. At the time step t =
0.1586 s the first shock wave has reached its furthermost downstream location at
x/c R ≈ 0.61. That also holds for the vortex breakdown position, which lies at its
furthermost downstream position at x/c R ≈ 0.73, Fig. 10.45. The shock wave has
ceased much of its strength and does no more push back the vortex breakdown
location.
After that the first shock wave begins to move upstream again, which too moves
upstream the vortex breakdown position. At t = 0.1650 s, lower right part of
Fig. 10.47, the strength of the shock has increased, the vortex breakdown has moved
forward to x/c R ≈ 0.64. Shortly after that time step the next cycle is beginning when
the shock has reached its most forward position, which is coupled to a decrease of
the shock strength.
We have seen hence that the flow field in conjunction with the vortex break-
down is highly unsteady and for the transonic case is coupled to the presence of an
embedded highly volatile shock-wave system. A conclusion drawn in [85] is that
a time-dependent scale-resolving solution permits to recognize disturbances in the
region between the shock wave(s), the sting tip and the primary vortex. These distur-
bances travel upstream toward the first shock wave. The frequencies of the upstream
moving disturbances and the spiral motion of the destructed vortex core after the
breakdown were found to be very similar.
Also reported in [85] is a consideration of axial vortex-core flow properties.
Figure 10.48 shows a result for M∞ = 0.8, Re∞,c = 2 · 106 , and α = 26◦ . We present
the result in order to illustrate the breakdown situation in terms of the core-flow
velocity.
10.5 Vortical Flow Past the Round-Edged VFE-2 Delta Wing 329

Fig. 10.48 Axial core velocity u/u ∞ as function of x/c R , sharp-edged case [85]. Blue line: result
of the numerical simulation, black dots: experimental data (PIV)

Computed and measured data is compared. The flow field was considered to be
steady. The Reynolds number in the experiment was larger with Re∞,c = 3 · 106 ,
but this was considered to be acceptable, because the sharp-edged wing case was
treated. There was a certain problem, because in the experiment vortex breakdown
was found to occur between x/c R = 0.6 and 0.7, whereas in the computation it was
further upstream at x/c R = 0.4. Hence the differences in the figure.
Nevertheless, the comparison reveals the typical behavior of the axial core velocity
during vortex breakdown. From the wing’s apex the axial vortex-core velocity is
building up fast to an overspeed of u/u ∞ ≈ 1.5 at x/c R = 0.05. The velocity rises
further and ahead of the computationally found breakdown location at x/c R = 0.4
we see the high axial velocity u/u ∞ = 1.88. The experimental value is somewhat
higher with u/u ∞ = 1.962 at x/c R = 0.5, with smaller values downstream of that
location.
The vortex breakdown at x/c R = 0.4 is marked by a steep drop of the axial velocity
to u/u ∞ = −0.6. Behind that location the velocity stays negative, but rises slowly
to a small positive value at the end of the wing. A re-configuration of the vortex over
the wing does not happen.

10.6 Partly Developed Swept Leading-Edge Vortices,

the SAGITTA Case

Selected results from the thesis of a doctoral student of the third author of this
book, A. Hövelmann, are presented as Unit Problem. Hövelmann investigated both
experimentally and numerically the vortex-flow phenomena over the moderately
swept, low aspect-ratio AVT-183 configuration with round leading edges, and over
the SAGITTA configuration with varying leading-edge contour [88]. We discuss
330 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.49 The SAGITTA configuration [88]. Left part: plan view. Right part: planform parameters

some of the results found for the latter configuration. All figures in this section are
from [88].
The SAGITTA (Latin: arrow) diamond wing is relatively thick and with basically
round leading edges. Signature requirements, however, led to a sharp leading-edge
contour up to 20 per cent in spanwise direction of the configuration, Fig. 10.49, left
part. The leading-edge sweep angle ϕ L E = ϕ0 = 55◦ is rather small. SAGITTA’s
geometrical planform parameters are given in Fig. 10.49, right part.
Note that SAGITTA is a slightly cropped diamond wing. The overall length is
cr = 1.2 m and the small wing-tip chord ct = 0.025 cr . The moment-reference point
is denoted by xmr p . The aspect ratio is Λ = 2.001, the taper ratio λ = 0.025, and the
mean aerodynamic chord lμ = 0.801 m. The original wing over the full span has a
symmetric NACA 64A012 airfoil section.
We present selected results of numerical simulations of the flow past the original
SAGITTA configuration (Geo 1), the configuration with completely sharp leading
edge (Geo 5), and the configuration with completely round leading edge (Geo 6).
The URANS (Model 10) computations were made assuming fully turbulent flow, the
Spalart–Allmaras one-equation turbulence model was applied. The flow parameters
are given in Table 10.7.
The wing of the SAGITTA configuration consists of four segments, Fig. 10.50.18
Segment I of the Geo 1 configuration has a sharp leading edge, the other three
segments have rounded ones with r L E /cr = 0.99 per cent, which amounts to δ = 52.
This means that even at the rear an aerodynamically sharp leading edge is not yet
present, Sect. 6.4. Geo 5 has a sharp leading edge throughout, Geo 6 a round leading
edge throughout.

Table 10.7 Flow parameters of the sub-sonic SAGITTA case [88]

M∞ Re∞,lμ lμ (m) T∞ (K) Twall α (◦ ) Viscous
flow
0.13 2.3 · 106 0.801 288.15 Adiabatic s. Table 10.8 Turbulent

18 Inthe following figures segment I is denoted with IB (inboard), segments II and III are denoted
with MB (midboard).
10.6 Partly Developed Swept Leading-Edge Vortices, the SAGITTA Case 331

Fig. 10.50 Segments and leading-edge contours of SAGITTA’s right wing [88]. Left part: Geo 1
configuration with partly sharp leading edge. Middle part: Geo 5 configuration with fully sharp
leading edge. Right part: Geo 6 configuration with fully round leading edge

In order to get a first impression of which lee-side vortex-flow patterns are to be

expected for either a sharp or a round leading edge, we look at the correlations in
Figs. 10.22 and 10.23. The results together with the normal angle of attack α N and the
normal leading-edge Mach number M N are listed in Table 10.8. A look at Fig. 10.20
tells us that—at sharp-edged delta wings—the angle of attack above which vortex
breakdown happens, is very low with αbd ≈ 9◦ .

Table 10.8 Possible lee-side vortex-flow patterns at SAGITTA with fully sharp and with fully
round leading edge (LE) for M∞ = 0.13, ϕ0 = 55◦ , and three angles of attack
α (◦ ) α N (◦ ) MN LE Possible
vortex-flow
pattern
8 13.76 0.076 Sharp Development of
vortex system
Round No vortex system
16 26.36 0.080 Sharp Stable and fully
developed vortex
system
Round Development of
vortex system
24 37.82 0.086 Sharp Stable and fully
developed vortex
system
Round Development of
vortex system
332 10 Small Aspect-Ratio Delta-Type Wing Flow

From the correlation data in Table 10.8 we can expect that for the wing with
round leading edge not much of non-linear lift will evolve, in particular also because
vortex breakdown is beginning already at a low angle of attack. Although for the
sharp-edged wing the situation looks better, also for it the early beginning of vortex
breakdown will limit the extent of non-linear lift.
First we discuss selected computational results of the flow past the Geo 1 con-
figuration.19 Considered is the upper-side flow field of the right-hand side of the
wing for three angles of attack, Fig. 10.51. (It appears that the flow field, like those
of the other cases, is an open-type lee-side flow field.) In the figures time-averaged
surface-pressure coefficients c p are shown as well as streamlines, which are started
close to the leading edge. In the figures, also in the corresponding ones for Geo 5
and Geo 6, the main flow direction is always from the lower right to the upper left.
At segment I of Geo 1 already at α = 8◦ a leading-edge vortex is present, upper
part of Fig. 10.51 (IB = inboard). At the intersection of segment I and II the vortex
detaches and moves inward. (At segment II the flow still is attached.) The segment
I vortex grows with increasing angle of attack, the suction level rises, too, the time-
averaged axial vorticity level becomes high, but decreases downstream. At the other,
round-edged segments, the flow is attached, at segment IV, however, a first indication
of separation is visible. It is possibly due to the small cropped part of the wing.
At α = 16◦ the situation has changed, middle part of Fig. 10.51.20 The flow past
segment IV now is separated, irregular recirculation is indicated, and the small suc-
tion peak present there at lower angle of attack has disappeared. The flow past the
segments II and III is still attached.
Finally at α = 24◦ , lower part of Fig. 10.51, a very complex flow pattern is present.
The midboard (MB) lee-side vortex, which came into being at segment III by ordinary
separation, dominates the flow field downstream. Underneath it a suction peak is not
present, vortex breakdown now clearly is present. Secondary separation exists. At
segment II incipient separation is indicated. The lee-side vortex coming from segment
I, the inboard (IB) vortex, has formed secondary separation already at a lower angle
of attack.
The skin-friction line pattern and the surface-pressure coefficient c p , for this angle
of attack, both time-averaged, are shown in Fig. 10.52. At segment I (IB) the primary
separation line PSL appears to lie above the sharp leading edge, not exactly at its
edge. In direction toward the center line of the wing follow the secondary attachment
line SAL, the secondary separation line SSL and the primary attachment line PAL,
the latter strongly curved into the direction of the wing’s center line. (The reader
should notice that we usually denote with ‘primary’ the attachment line or lines at
the windward side of a configuration, see Fig. 7.21. Here PAL is the attachment line
of the primary vortex emanating from PSL.)

19 The reader should note that the Geo 1 configuration resembles the by H.A. Wilson and J.C. Lovell

manipulated leading edge contour of the DM-1 glider, Fig. 10.7.

20 In [88] the lee-side flow development is visualized in Δα = 4◦ steps. Due to space restrictions

we discuss only in 8◦ steps.

10.6 Partly Developed Swept Leading-Edge Vortices, the SAGITTA Case 333

Fig. 10.51 Upper-side flow field at the right-hand side of the SAGITTA Geo 1 wing: time-averaged
surface-pressure coefficient c p and field streamlines [88]. Upper part: α = 8◦ , middle part: α = 16◦ ,
lower part: α = 24◦
334 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.52 Top view at the right-hand side of the SAGITTA Geo 1 wing at α = 24◦ showing the
time-averaged skin-friction lines and the time-averaged surface-pressure coefficient c p [88]. LEV:
leading-edge vortex, PSL: primary separation line, SSL: secondary separation line, PAL primary
attachment line, SAL: secondary attachment line. The freestream direction is upward from below

Overall we see an open lee-side flow pattern. Whether the separation and attach-
ment lines are of open-type beginning, cannot be assured. An open-type ending is
evident. More detailed than in Fig. 10.51 we see that the suction peak underneath the
IB leading-edge vortex is well developed, but very soon ceases to exist due to the
vortex breakdown.
At the following MB segments we find a well discernible flow past the round
leading edge and the primary separation line PSL located away from the leading
edge at the wing’s upper side. Downstream it comes closer to the leading edge.
There a secondary separation line SSL and a secondary attachment line SAL appear.
Beginning and ending of the singular lines appears to be of open type. The suction
peak underneath the MB leading-edge vortex is seen to be well developed, and also
very soon to cease to exist due to the vortex breakdown.
We contrast now these results first with those found for the sharp-edged Geo 5
wing, and then with those found for the round-edged Geo 6 wing, both sketched
also in Fig. 10.50. We begin with Geo 5. At α = 8◦ a lee-side vortex exists along the
whole leading edge, upper part of Fig. 10.53.
The correlation data in Table 10.8 suggest the beginning of the development of the
vortex system for this angle of attack. Similar, but stronger than for the respective
case of Geo 1, at segment IV an indication of separation or vortex breakdown is
present. Influence of the small cropped part of the wing? Anyway, the amount of
suction pressure along the leading edge is small, hence the non-linear lift at this
angle of attack is small, too.
Figure 10.53, middle part, shows that for α = 16◦ the lee-side vortex is fully
developed with a larger amount of non-linear lift than at the smaller angle of attack.
The non-linear lift, however, is degraded by the breakdown of the vortex, which
10.6 Partly Developed Swept Leading-Edge Vortices, the SAGITTA Case 335

Fig. 10.53 Upper-side flow field at the right-hand side of the fully sharp-edged SAGITTA Geo 5
wing: time-averaged surface-pressure coefficient c p and field streamlines [88]. Upper part: α = 8◦ ,
middle part: α = 16◦ , lower part: α = 24◦
336 10 Small Aspect-Ratio Delta-Type Wing Flow

already has reached segment III. At α = 24◦ finally the vortex has reached a large
strength, indicated by the high suction pressure underneath it, lower part of Fig. 10.53.
But vortex breakdown has reached segment II. Hence the adverse effects of vortex
breakdown, loss of non-linear lift and potentially a pitch-up moment are present.
We look now at the results found for the round-edged SAGITTA Geo 6 wing. At
α = 8◦ attached flow is present along the whole leading edge, except for a small area
at the cropped wing tip, upper part of Fig. 10.54. The correlation data in Table 10.8
suggest that no vortex system exists for this angle of attack. At the segments III and
IV a small amount of suction pressure is present.
For α = 16◦ attached lee-side flow is present up to segment IV. Sizeable suc-
tion pressure is found. At the aft part of the wing, segment IV, a vortex is present,
middle part of Fig. 10.54. That vortex obviously underwent breakdown almost from
its beginning. Flow reversal is indicated, the suction pressure is gone at almost the
whole leading edge of segment IV.
At α = 24◦ along the first half of the leading edge the flow still is attached, with
a rather strong suction pressure visible, lower part of Fig. 10.54. Vortex breakdown
has moved upstream to the beginning of segment II. The suction pressure mostly has
disappeared beyond segment II.
When comparing the results found for the three SAGITTA configurations Geo 1,
Geo 5 and Geo 6, we assert that mostly rather weak lee-side vortex phenomena are
present, even secondary ones, that suction pressure exists to a small extent, and that
vortex breakdown happens already at small angles of attack. The results as a whole
suggest that non-linear lift is small for all three configurations. Hence its influence
on the aerodynamic properties of SAGITTA is small, too.
In [88] computed aerodynamic coefficients and derivatives of the longitudinal
motion are compared for the three configurations. We show the result for the lift and
the pitching moment coefficient and their derivatives. In the figures also results for
Geo 1 found with the Athena Vortex Lattice method AVL are plotted.
The lift coefficient C L (α) and its derivative dC L (α)/dα are given in Fig. 10.55.
In the whole angle-of-attack interval the lift results, left part of the figure, including
the AVL results, indeed do not differ much from each other. SAGITTA’s lift almost
fully is due to its plan form.
That non-linear effects are present, even if only small, can be seen when looking at
the lift-coefficient derivative in the right part of Fig. 10.55. The linear AVL method
yields a lift slope, which evenly falls with increasing angle of attack. This is in
contrast to the non-linear results.
The derivative of the sharp-edge Geo 5 configuration increases to a maximum at
α ≈ 11◦ . The reason for this is the existence of the full-span leading-edge vortex
with its increasing suction level. Beyond that angle of attack vortex breakdown leads
to a general reduction of the derivative.
For the round-edged Geo 6 configuration we see a weak maximum of dC L (α)/dα
at α ≈ 8◦ , followed by a weak continuous decrease. The Geo 1 configuration with the
sharp-edged segment I and rounded leading edges in segments II, III, and IV shows
the strongest non-linear effect with a maximum of dC L (α)/dα at around α ≈ 18◦ ,
10.6 Partly Developed Swept Leading-Edge Vortices, the SAGITTA Case 337

Fig. 10.54 Upper-side flow field at the right-hand side of the fully round-edged SAGITTA Geo 6
wing: time-averaged surface-pressure coefficient c p and field streamlines [88]. Upper part: α = 8◦ ,
middle part: α = 16◦ , lower part: α = 24◦
338 10 Small Aspect-Ratio Delta-Type Wing Flow

followed by a very steep decrease. That appears mainly to be due to the breakdown
of the segment I vortex at the rear of the wing.
The pitching-moment coefficient Cm (α) in the left part of Fig. 10.56 is negative—
due to the choice of the reference point xmr p , see the right part of Fig. 10.49—hence
SAGITTA in pitch is stable—depending on the center of gravity location versus the
chosen reference point. The leading-edge contours play a minor role only above α
≈ 10◦ . Due to the symmetric airfoil of the configuration the pitching moment is zero
at α = 0◦ . The result found with the linear AVL method agrees with the others up to
α ≈ 13◦ .
Strong deviations are found of the pitching-moment derivative dCm (α)/dα seen in
the right part of Fig. 10.54. They are due to the diverse effects present, i.e., separation,
vortex development with the associated suction pressure, and vortex breakdown with
the associated pitch up.
We highlight a few of the consequences of the effects appearing at the three
configurations. The fully sharp-edged Geo 5 configuration shows a nearly constant
derivative up to α ≈ 14◦ , where a small relative maximum occurs. With increasing
α then it drops sharply to dCm (α)/dα = −0.375 at α ≈ 24◦ . The fully round-edged

Fig. 10.55 Longitudinal motion of configurations Geo 1, Geo 5 and Geo 6 [88]. Left part: lift
coefficient C L (α). Right part: lift-coefficient derivative dC L (α)/dα

Fig. 10.56 Longitudinal motion of configurations Geo 1, Geo 5 and Geo 6 [88]. Left part: pitching
moment coefficient Cm (α). Right part: pitching-moment derivative dCm (α)/dα
10.6 Partly Developed Swept Leading-Edge Vortices, the SAGITTA Case 339

Geo 6 configuration almost everywhere exhibits a steady drop of the derivative down
to dCm (α)/dα = −0.43 at α = 25◦ . Geo 1 finally up to α = 15◦ has a steady drop
of the derivative like Geo 6 has. Then a steep increase leads to a relative maximum at
α = 21◦ , followed by a small decrease down to dCm (α)/dα = −0.28 at α = 24◦ .
Hence due to the intricate flow phenomena along the leading edge with varying
contours, Geo 1 shows an appreciably varying longitudinal stability behavior.

10.7 Laminar Hypersonic Flow Past a Round-Edged Delta

Wing

The wing of this Unit Problem is the Blunt Delta Wing (BDW), which was a study
configuration in the European HERMES project, see, e.g., [2]. The BDW configu-
ration is a very strongly simplified re-entry vehicle configuration flying at moderate
angle of attack.
The flow past the BDW was investigated experimentally and numerically by many
authors. Here we discuss a few results from the thesis of a doctoral student of the
first author of this book, S. Riedelbauch [89], see also [90]. Riedelbauch studied the
BDW’s surface-radiation cooling, here we discuss the flow field and the possible
non-linear lift.21 All figures in this section except two are from [89].
The configuration is a simple slender delta wing with a blunt nose and rounded
leading edges, Fig. 10.57. The lower side has a dihedral (γ = 15◦ , lower part of
Fig. 10.57) and therefore is only approximately flat.
Navier–Stokes (Model 9) computations with perfect-gas assumption were per-
formed with the parameters given in Table 10.9. Although in reality laminar-turbulent
transition would occur with the flight parameters listed there, the flow was assumed
to be laminar throughout. We note in passing that at the wing’s surface radiation
cooling was modeled with the surface-emission coefficient being  = 0.85.
The normal angle of attack and the normal leading-edge Mach number of this
hypersonic case are α N = 38.08◦ and M N = 3. This case lies far out to the right in
Fig. 10.23 and hence is not covered by the experimental data underlying the correla-
tions there. Anyway, the leading edge is supersonic, and lee-side cross-flow shocks
are to be expected.

Table 10.9 Computation parameters of the blunt delta wing (“flight” case) [89]
M∞ H (km) T∞ (K) Re∞ u L (m) ϕ 0 (◦ ) α (◦ ) ε Boundary
(m−1 ) layer
7.15 30 226.506 2.69 · 14 70 15 0.85 Laminar
106

21 Aspects of the radiation cooling of the BDW are discussed in [91], the topic of flow properties in
this regard along singular lines is treated in [73], see also our Sect. 7.3.
340 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.57 Configuration of

the BDW and the coordinate
convention [89]: a side view,
b view from above, c
cross-section B - B

The cross-flow bluntness parameter ranges from pb = 0.143 at x/L = 0.1 to

pb = 0.074 at x/L = 1. Hence we are well away from an aerodynamically sharp
leading edge.
Experimental investigations were performed at FFA Sweden with a BDW model
having a modified nose [92]: sharp in comparison to the blunt nose shown in
Fig. 10.57. The parameters partly were different and are given in Table 10.10:
The computation for the blunt-nosed BDW with the experimental data yielded
the distribution of the surface pressure coefficient c p (y) at the location x/L = 0.5
shown in Fig. 10.58. The lift predominantly is created at the windward side of the
configuration. At the lee side is to a degree present the hypersonic shadow effect [91].
That means that not much of a negative c p is existing. The classical suction peak of
delta wings is only very weakly indicated, hence in this case one really cannot speak
of non-linear lift being present.

Table 10.10 Experimental parameters of the sharp-nosed blunt delta wing [92]
M∞ T∞ (K) Re∞u L (m) Tw (K) ϕ 0 (◦ ) α (◦ ) Boundary
(m−1 ) layer
7.15 74 39 · 106 0.15 288 70 15 Laminar
10.7 Laminar Hypersonic Flow Past a Round-Edged Delta Wing 341

Fig. 10.58 Distribution of

the computed surface
pressure coefficient c p (y) at
x/L = 0.5 [89]. Note that
the windward side is above
and the lee side below

Fig. 10.59 Detail of the

distribution of the computed
[89] surface pressure
coefficient c p (y) at
x/L = 0.5 of the lee side and
experimental [92] data ()

A detail of the lee-side pressure distribution is shown in Fig. 10.59. Indicated are
a few measured data, the triangles [92]. The deviation from the computed data is less
than 10 per cent. The absolute minimum of the pressure is c pmin = −0.02492 at y/c
= 0.78, close to the vacuum pressure, see Appendix A.1, of c pvac = −0.02794.
342 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.60 Computed

skin-friction line pattern
(blunt-nosed wing, above
[89]) and oil-flow picture
(sharp-nosed wing, below
[92]). The free-stream comes
from the left

It usually is taken as granted that discrete numerical solutions of the Navier–Stokes

equations (Model 9) do not suffer from modeling problems, like RANS solutions. The
discretization, of course, must be correct, as well as the whole solution procedure.
We present, besides the comparison of the surface pressure data above, a visual
comparison between experimentally and numerically found skin-friction line pat-
terns, Fig. 10.60.
In the lower part of that figure the upper side of the BDW wind-tunnel model from
[92] is shown. The sharp nose, compared to the blunt one of the actual BDW seen
in the upper part, clearly is discernible. Well visible is the influence of the model
support’s sting at the lower right.
Apart from the nose and the aft region the agreement between the computed and
the measured skin-friction line patterns is acceptable. The oil-flow picture by far does
not contain the fine details, which we see in the computed picture. In the oil-flow
picture recognizable are only the primary (S1 ) and the secondary (S2 ) separation
lines.
We now have a look at the topology of the skin-friction field computed for the
“flight” case, Table 10.9. We wish to identify attachment and separation lines.
We see at the lower (windward) side of the configuration in Fig. 10.61a the classical
skin-friction line pattern present at the windward side of a delta wing. Because the
lower side of the BDW is not fully flat, the flow exhibits a slight two-dimensionality
between the two primary attachment lines A1 . The latter are marked by strongly
divergent skin-friction lines. The forward stagnation point, which is a nodal point,
10.7 Laminar Hypersonic Flow Past a Round-Edged Delta Wing 343

Fig. 10.61 Selected

computed skin-friction lines
at the surface of the BDW
[89]: a look at the lower side,
b look at the upper side of
the configuration. The
free-stream comes from the
left

Sect. 7.2, lies also at the lower side, at about 3 per cent of the body length. The primary
attachment lines are almost from the beginning parallel to the leading edges, i.e., they
do not show a conical pattern.
At the upper (leeward) side of the wing the situation is quite different, Fig. 10.61b.
The flow streams from the lower side around the blunt leading edge well onto the
upper side, where it then separates. In the lower part of Fig. 10.61b we see along
the vertical line—from the leading edge toward the symmetry line—a succession
of separation and attachment lines: the primary separation line S1 , the secondary
attachment line A2 , a secondary separation line S2 , and the tertiary attachment line
A3 in the middle of the lower side, the latter indicating a closed lee-side flow field.
All is mirrored on the upper part of the picture.
Again a conical pattern is not discernible, except for a small portion near the nose.
The secondary separation lines are almost parallel to the single tertiary attachment
line along the upper symmetry line of the wing.
Both the primary and the secondary separation lines are lines of “open-type sep-
aration”, i.e., the separation line does not begin in a singular point on the surface,
344 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.62 Selected

computed skin-friction lines
at the upper side of the BDW
near the nose (detail of
Fig. 10.61b) [89]

Fig. 10.63 Structure of the closed lee-side flow field in the Poincaré surface [89]

Sect. 7.1.4. Figure 10.62 shows this, and also that all attachment lines are of open-
type, too.
With these surface patterns we can qualitatively sketch a structure of the leeward-
side flow in the Poincaré surface, Fig. 10.63. The computed cross-flow shocks are
indicated. They lie above the lee-side vortex system, quite in contrast to the findings
in Sect. 10.4.
The two primary attachment lines A1 show up as quarter-saddle points, whereas
half-saddle points indicate the two primary separation lines S1 , the two secondary
attachment lines A2 , the two secondary separation lines S2 , and the tertiary attachment
line A3 . Above the wing the two focus points F1 of the primary lee-side vortices, the
two focus points F2 of the secondary lee-side vortices, and the free saddle point S0
above the lee-side vortices (closed lee-side flow field) are indicated.
We apply the topological rule 2’, Eq. (7.25). Because the focus points are counted
as nodal points, we find
10.7 Laminar Hypersonic Flow Past a Round-Edged Delta Wing 345
   
1 1 1
4+ 0 − 1 + 7 + 2 = −1, (10.13)
2 2 4

and conclude that the topology is a valid one.

10.8 Concluding Remarks

After the introduction and the short historical notes we presented some basics of
the phenomenon of non-linear lift appearing at delta wings, Sect. 10.2. Correlations
of empirical data were given showing when lee-side vortex systems are expected to
be present, Sect. 10.2.6, and when vortex breakdown and overlapping can happen,
Sect. 10.2.5. These correlations proved to give reasonable first insight regarding the
appearance and behavior of the vortex systems in the Unit Problems under consid-
eration.
The first Unit Problem then illustrated the significance of different flow-physical
and mathematical models for the computational simulation of lee-side vortex flow,
Sect. 10.3. We have shown with simulations for the VFE-1 configuration with low-
order turbulence models and even with the Euler equations (Model 8), that they can
give reasonable results for pre-design work, provided that the delta wing has sharp
leading edges.22
The second Unit Problem, Sect. 10.4, was devoted to the proof that at sharp-edged
delta wings the creation of vorticity including the entropy rise and eventually the
appearance of the—primary—lee-side vortices follow the same mechanism, which
is present at large aspect-ratio wings. The proof made use of the concept of kinemat-
ically active vorticity content and the compatibility condition, both Chap. 4, which
in Chap. 8 were applied to the wing of the CRM configuration, Sect. 8.4.
We have shown and discussed only a small number of the results achieved with
the round-edged vortex-flow experiment VFE-2, which is the third Unit Problem
treated. Important is that the relative bluntness of the leading edge, i.e. the parameter

pb = r L E /b plays a major role: with constant leading-edge radius it diminishes
from the front of the wing toward the rear. This means that the leading edge becomes
relatively sharper in downstream direction. Accordingly the formation of the primary
lee-side vortex is affected.
In the present case the wing has—in terms of the nomenclature of the VFE-2
community—a second primary lee-side vortex system, located ahead of the first one
and probably coupled to that, Sect. 10.5.2. This phenomenon was discussed and after
that the phenomenon of vortex breakdown. Both a subsonic and a transonic case

22 Regarding the numerical simulation of vortical flow fields we assume that always sufficient
investigations were made in order to ensure grid independency. This holds for all simulations with
Model 8, 9, and 10 methods. Regarding experimental simulations, similar requirements must be
fulfilled with respect to the experimental set-up, the free-stream properties, the test section as
test environment, the model properties—including the sharpness of the leading edges—and the
measuring devices.
346 10 Small Aspect-Ratio Delta-Type Wing Flow

was considered. For the latter in particular the high degree of flow unsteadiness was
demonstrated.
The following Unit Problem was the flow past the SAGITTA configuration at
the low subsonic Mach number M∞ = 0.13. The configuration has a sharp leading
edge directly behind its apex and downstream a round one. Moreover with ϕ0 = 55◦
it has a smaller leading-edge sweep than the VFE-1 and the VFE-2 configuration,
which both have ϕ0 = 65◦ . This difference appears not to be much, but the empirical
correlations already show—also due to the small Mach number—that the lee-side
vortex system is developed only weakly and that vortex breakdown is to happen
already at a small angle of attack. Accordingly non-linear lift effects due to the
lee-side vortex system are rather weak.
The last Unit Problem, the laminar flow at hypersonic speed past a blunt-edged
wing at high angle of attack, revealed a well established lee-side vortex system. The
skin-friction line pattern shows well how primary and secondary singular lines govern
the vortex system of the closed lee-side flow field. However, due to the hypersonic
shadow effect, a non-linear lift is almost non-existent.
In closing these remarks we list some of the general observations made during the
VFE-1 and in particular the VFE-2 investigations, mainly following [78, 79]. The
observations regard the influence of the angle of attack, Reynolds and Mach number,
vortex breakdown, and laminar-turbulent transition and turbulence modeling.

• Angle of attack At wings with sharp leading edges the lee-side vortex system above
a threshold angle generally appears to be present along the whole leading edge.
Experimental observations, however, challenge that to a certain degree.
If the wing has a round leading edge, the primary lee-side vortex first appears at
the rear of the wing and with increasing angle of attack moves forward. This effect
appears to be connected to the bluntness parameter pb . The gedankenexperiment in
Sect. 10.2.4 indicates that for a given bluntness the actual adverse pressure gradient
also depends on sin2 α, i.e., increasing angle of attack increases the separation
disposition at the very leading edge.
• Reynolds number In experiments it frequently was observed that in low Reynolds-
number delta wing cases with presumably laminar flow throughout, the lee-side
vortices were stronger than with turbulent flow in high Reynolds-number cases,
and hence also the non-linear lift. This holds for wings with round and with sharp
leading edges, for the latter only regarding the secondary vortices.
This observation can be commented on with a consideration of the—kinematically
active—vorticity content of the involved boundary layers at separation.
Figure 10.64 shows schematically the flow situation observed at a delta wing with
round leading edges. Both the pairs of the primary and the secondary vortices are
sketched. The possible cross-flow shocks are not indicated. On the left of the figure
the situation arising with laminar flow is indicated, on the right that with turbulent
flow.
Because the laminar boundary layer cannot negotiate as much adverse pressure
gradient as the turbulent one, it will separate earlier than the turbulent one. This
10.8 Concluding Remarks 347

Fig. 10.64 Schematic of ordinary separation in the cross-section of a delta wing with round leading
edges and a closed-type lee-side flow field [55]. Pressure coefficient distribution (without cross-flow
shock effects) and separation locations of a laminar flow, b turbulent flow. c Location of primary
and secondary separation and vortices

holds for both the primary and the secondary separation and also for these locally
highly three-dimensional boundary-layer flows.
We assume now that the structure of separation, in particular the shear between the
upper and the lower inviscid flow at the leading edges—and also at the location
of secondary separation—is not completely different in the two cases (a) and (b).
If this can be accepted, the kinematically active vorticity content at separation is
larger in the laminar case than in the turbulent one. This is due to the fact that the
separation in the laminar case happens at a considerably higher velocity level than
in the turbulent case.
This tells us two things. First, the matter of separation prediction with numeri-
cal methods (Model 10), in particular at wings with round swept leading edges,
hinges on the suitability of the employed turbulence model. At wings with round
leading edges and with regard to the secondary vortices, discrete modeled Euler
solutions (Model 8) are not suited to describe these flows. Second, if embedded
laminar-turbulent transition happens at the lee side of the wing, the properties of
secondary separation and vortices hardly can be simulated properly. The reason
is that today only empirical or semi-empirical transition criteria and models are
348 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.65 Onset of separation of the outer primary vortex as function of the angle of attack (AOA ≡
α) [93]. Upper part: Rec = 2 · 106 , lower left part: Rec = 6 · 106 , lower right part: Rec = 60 · 106

available, but no non-empirical ones [73].

Increasing the Reynolds number moves a vortex system downstream. This obser-
vation can be explained with the concept of the strength of a boundary layer,
Sect. 2.1. For two-dimensional flow it was demonstrated by means of empirical
separation criteria that for a given adverse pressure gradient an increase of the
Reynolds number reduces the separation tendency of both laminar and turbulent
flow. This result, with due reservations, can be generalized to hold for the present
flow problems.
Indeed, S. Crippa and A. Rizzi, [93], in this sense illustrate the Reynolds-number
effect at the VFE-2 configuration at M∞ = 0.4, Fig. 10.65. We show the results
for three Reynolds numbers. Of concern was how to determine the onset. In the
numerical simulation two approaches were employed: (1) convergence of skin-
friction lines at the separation location (blue color), and (2) tracing of the vortex
core location (red color). The results from the experiment (black color) are also
specified. All approaches more or less agree well with each other.
In all cases the Reynolds-number effect is well illustrated. At small angle of
attack (AOA ≡ α) the agreement of experimental and numerical data is question-
able, with increasing angle it becomes better. Anyway, at small α for the highest
Reynolds number the separation onset is at x/cr = 0.6. The location moves for-
10.8 Concluding Remarks 349

Fig. 10.66 Onset of separation of the outer and the inner primary vortex as function of the angle
of attack (AOA ≡ α) [93]. Left part: Rec = 6 · 106 , right part: Rec = 60 · 106

ward with increasing α. For the lower Reynolds numbers it is found more forward
at x/cr = 0.35 and 0.25. At the largest angles of attack the separation location
seems to be no more affected by the magnitude of the Reynolds number.
An interesting result is shown in Fig. 10.66: the location of the onset of separation
of the inner primary vortex obviously is coupled to that of the outer primary vor-
tex. Because of uncertainties in the data, the results are plotted only for the higher
Reynolds numbers.
The, to a high degree, constant distance between the onset locations of the outer
and the inner primary vortex points to the fact that the two vortex pairs are closely
coupled. This is the same with the secondary vortex of the outer primary vortex.
Hence, as initially mentioned, one better should speak of the tertiary vortex pair
instead of the inner primary vortex pair.
• Mach number From the two Mach number cases considered of the VFE-2 case,
M∞ = 0.4 (subsonic) and M∞ = 0.8 (transonic), it was seen that basically the
flow-field topology remains the same. In the transonic case, the formation of the
outer primary vortex begins at a smaller angle of attack. Its axis is shifted inboard,
the inner primary vortex appears to be either very weak or not present at all.
• Vortex breakdown Vortex breakdown at large angles of attack basically is a highly
unsteady phenomenon. This means that computational simulation must be made
with unsteady approaches, e.g., URANS or scale-resolving methods (Model 10 or
11 methods). Vortex breakdown first occurs in the vicinity of the trailing edge and
with increasing angle of attack moves upstream.
For the VFE-2 configuration with sharp and medium-radius leading edges vortex
breakdown of the outer primary vortex was seen to be present along the whole
leading edge at α = 23◦ . At the SAGITTA configuration with a smaller leading-
edge sweep vortex breakdown begins to appear already around α = 9◦ .
The correct prediction of the onset of breakdown by means of numerical simu-
lations is difficult. In all such simulations vortex breakdown already occurred at
smaller angles of attack than found experimentally. The reason for that probably
is a too low prediction of the axial velocity of the vortex. Shortcomings of the
350 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.67 Schematic of the effect of laminar-turbulent transition on the separation behavior at
delta wings with round leading edges [94]

employed turbulence models and/or the URANS approach as such are the causes
for that. Scale-resolving methods appear to be the way out.
• Laminar-turbulent transition and turbulence modeling The laminar-turbulent tran-
sition affects ordinary separation behavior. As explained in Sect. 2.1, a turbulent
boundary layer can negotiate a larger adverse pressure gradient than a laminar
one. The other way around this means that with a given adverse pressure field
the turbulent boundary layer will separate at a more downstream location than the
laminar one.
The laminar-turbulent transition behavior at a given pressure field, Mach num-
ber, free-stream and body-surface conditions depends on the magnitude of the
Reynolds number, see, e.g., [73].

Figure 10.67 from [94] visualizes the effect of increasing Reynolds number on the
transition location and the separation behavior at a delta wing with round leading
edge. At the lowest Reynolds number (left sketch) laminar primary (open-type)
separation happens close to the leading edge. Once transition has occurred, the
separation location of the now turbulent boundary layer moves to the upper side
of the wing.
The sketch in the middle shows that with increasing Reynolds number transition
happens more upstream and most of the turbulent primary separation line now is
located at the upper side of the wing. The further increase of the Reynolds number
leads to transition upstream of the separation location and hence the beginning of
10.8 Concluding Remarks 351

the fully turbulent primary open-type separation effectively is shifted downstream

with the separation line. Of course, as before, it is located at the upper side of the
wing.
If transition occurs in the attached boundary layer ahead of the location where the
vortex emerges, simulation approaches have a problem, if the separation phenom-
ena are influenced. That problem is increased, if embedded transition occurs, for
instance along attachment lines, or even in the evolving or the fully evolved vor-
tices. Transition criteria are not available for such flows, the popular en approaches,
[73], are not viable.
Comparisons of numerical RANS and URANS simulations with different turbu-
lence models have shown that important differences in the obtained flow phenom-
ena can occur. In this regard best-practice approaches would be helpful.

10.9 Problems

Problem 10.1 With flight vehicles we observe the following general wing shapes:
(a) more or less rectangular wings, (b) swept wings, (c) slender delta-like wings.
(1) In what Mach-number domains are these shapes found? What is behind this?
(2) Are there other wing shapes? (3) What is the difference between an hypersonic
airbreather and a winged re-entry vehicle like the former Space Shuttle Orbiter?
Problem 10.2 Two vortex phenomena limit the “healthy” flow regime above a delta
wing at angle of attack. Give a short description of these phenomena and their influ-
ence on the aerodynamic properties of the wing.

Problem 10.3 Consider the DM-1 glider in Fig. 10.7. Assume an angle of attack α =
30◦ and a flight Mach number M∞ = 0.3. What do the correlations in Sect. 10.2.6
say regarding the occurrence of lee-side vortices (a) for the configuration with the
original round leading edge and ϕ0 = 60◦ and (b) for the manipulated configuration
with sharp leading edge and ϕ0 = 64◦ ?
Problem 10.4 Consider Fig. 10.41. Would the lee-side flow field as a closed one be
viable? Apply Rule 2’.

Problem 10.5 Sketch qualitatively the spanwise distributions of both the lift
C L (y)l(y) and the lift coefficient C L (y) for a delta wing of about 65◦ leading-edge
sweep at about 20◦ angle of attack. Which separation scenarios occur? What is the
impact on the lift characteristics?

Problem 10.6 Consider a planar delta wing of ϕ0 = 65◦ leading-edge sweep with
nearly sharp leading edges and an aspect ratio of Λ = 1.9. Which pitching moment
characteristics as function of angle of attack may be attributed to this wing?

Problem 10.7 Consider a double-delta wing featuring sharp leading edges at angle
of attack of about α ≈ 20◦ . The leading-edge sweep of the forward portion of the
352 10 Small Aspect-Ratio Delta-Type Wing Flow

Fig. 10.68 Delta wing at angle of attack α

Fig. 10.69 Double delta wing

wing (strake) is ϕ01 = 75◦ , the leading-edge sweep of the rearward portion of the
wing is ϕ02 = 55◦ .
(1) Provide a topological consistent sketch of a leading-edge vortex system with
respect to primary and secondary vortex structures for a cross flow plane intersecting
the wing in the rearward portion. (2) Which trends may be expected with respect
to pitch-up and roll-reversal tendencies? Which flow-physics effects are associated
with these tendencies?
Problem 10.8 A flat delta wing with dimensions according to sketch below,
Fig. 10.68, flies at an angle of attack of α. Where is the center of pressure located?

Problem 10.9 A slender wing has the dimensions in Fig. 10.69. What is the slope
of the lift curve, dC L /dα?

Problem 10.10 Simplify the case of a wing-and-horizontal stabilizer to a two-

dimensional problem where the wing and stabilizer are a flat plate, Fig. 10.70. The
angle of attack is zero. Using the lumped vortex method, what is the vortex strength
of the wing, with dimensions according to the sketch.
References 353

Fig. 10.70 Two-dimensional wing-stabilizer configuration

References

1. Küchemann, D.: The Aerodynamic Design of Aircraft. Pergamon Press, Oxford (1978), also
AIAA Education Series. AIAA, Reston (2012)
2. Hirschel, E.H., Weiland, C.: Selected Aerothermodynamic Design Problems of Hypersonic
Flight Vehicles. Springer, Berlin, and Progress in Aeronautics and Astronautics, vol. 229.
AIAA, Reston (2009)
3. Hirschel, E.H.: Towards the virtual product in aircraft design? In: Periaux, J., Champion, M.,
Gagnepain, J.-J., Pironneau, O., Stoufflet, B., Thomas, P. (eds.) Fluid Dynamics and Aero-
nautics New Challenges. CIMNE Handbooks on Theory and Engineering Applications of
Computational Methods, Barcelona, Spain, pp. 453–464 (2003)
4. Edwards, C.L.W., Small, W.J., Weidner, J.P.: Studies of scramjet/air-frame integration tech-
niques for hypersonic aircraft. AIAA-Paper 75-58 (1975)
5. Kuczera, H., Sacher, P.: Reusable Space Transportation Systems. Springer, Berlin, and Praxis
Publishing, Chichester (2011)
6. Krämer, E.: Kampfflugzeuge. In: Rossow, C.-C., Wolf, K., Horst, P. (eds.) Handbuch der Luft-
fahrzeugtechnik, pp. 113–150. Carl Hanser Verlag, München (2014)
7. Modin, K.E., Clareus, U.: Aerodynamic design evolution of the SAAB JAS 39 Gripen aircraft.
AIAA-Paper 91-3195 (1995)
8. Clareus, U.: Aerodynamic highlights of a fourth generation delta canard fighter aircraft. In:
Proceedings of the 6th Annual International Symposium, 14–19 August 2001, Zhukovsky,
Russia (1995)
9. Staudacher, W.: Die Beeinflussung von Vorderkantenwirbelsystemen schlanker Tragflügel (The
manipulation of leading-edge vortex systems of slender wings). Doctoral thesis, University
Stuttgart, Germany (1992)
10. Hitzel, S.M., Schmidt, W.: Slender wings with leading-edge vortex separation: a challenge for
panel methods and Euler solvers. J. Aircr. 21(10), 751–759 (1984)
11. Winter, H.: Strömungsvorgänge an Platten und profilierten Körpern bei kleinen Spannweiten.
Forschung im Ingenieurwesen 6, 67–71 (1935)
12. Busemann, A.: Aerodynamischer Auftrieb bei Überschallgeschwindigkeit. Proc. Volta-
Kongress, Rom 1935, 328–360 and Luftfahrtforschung 12, 210–220 (1935)
13. Ludwieg, H.: Pfeilflügel bei hohen Geschwindigkeiten. Aerodynamische Versuchsanstalt Göt-
tingen, Bericht 39/H/18 (1939)
14. Meier, H.U. (ed.): German Development of the Swept Wing–1935-1945. Library of Flight.
AIAA, Reston (2010)
15. Bollay, W.: A non-linear wing theory and its application to rectangular wings of small aspect
ratio. ZAMM 19, 21–35 (1939)
16. Gersten, K.: Nichtlineare Tragflächentheorie, insbesondere für Flügel mit kleinem Seitenver-
hältnis. Ingenieur-Archiv 30(6), 431–452 (1961)
17. Pelletier, A.J.: Paper darts to deltas—The designs of Roland Payen. Air Enthu. Mag. 68, 33–44
(1997)
18. Wagner, W.: The History of German Aviation: The First Jet Aircraft. Schiffer Military/Aviation
History, Atglen, PA (1998)
354 10 Small Aspect-Ratio Delta-Type Wing Flow

19. Jones, R.T.: Wing planforms for high-speed flight. NACA 863 (1945)
20. Wilson, H.A., Lovell, J.C.: Full-scale investigation of the maximum lift flow characteristics of
an airplane having approximately triangular planform. NACA RM L6K20 (1946)
21. Nowarra, H.J.: Die deutsche Luftrüstung 1933-1945 (four parts). Bernard & Graefe Verlag,
Bonn (1989)
22. Masters, D.: German Jet Genesis. Jane’s Publishing Company, London (1982)
23. Heinzerling, W.: Die Geschichte des Pfeilflügels. Teil 2: Die projektmäßige Anwendung des
Pfeilflügels. DGLR-Jahrbuch 1981, 071-1–071-33 (1981)
24. von Gersdorff, K.: Transfer of German aeronautical knowledge after 1945. In: Hirschel, E.H.,
Prem, H., Madelung, G. (eds.) Aeronautical Research in Germany—From Lilienthal Until
Today, pp. 325–344. Springer, Berlin (2004)
25. Polhamus, E.C.: Vortex lift research: early contributions and some current challenges. In:
Campbell, J.F., Osborne, R.F., Foughner, J.T. (eds) Vortex Flow Aerodynamics, vol. 1, NASA
CP-2416, pp. 1–30 (1986)
26. Friedrich, O., Herbst, W., Obermeier, E.: Evolution of the New European Fighter—A German
Industrial Perspective. MBB Military Aircraft Division, München (1989)
27. Örnberg, T.: A note on the flow around delta wings. KTH-AERO-TN 38 (1954)
28. Marsden, D.J., Simpson, R.W., Rainbird, W.J.: An investigation into the flow over delta wings
at low speeds with leading-edge separation. Cranfield College of Aeronautics, Rep. No. 114
(1958)
29. Werlé, H.: Sur l’Éclatement de Tourbillions d’Apex d’une Aile Delta aux Faible Vitesses. La
Recherche Aeronautique No. 74 (1960)
30. Hummel, D.: Experimentelle Untersuchung der Strömung auf der Saugseite eines schlanken
Deltaflügels. ZFW 13(7), 247–252 (1965)
31. Hummel, D.: Zur Umströmung scharfkantiger Deltaflügel bei großen Anstellwinkeln. ZFW
15(10), 376–385 (1967)
32. Hummel, D., Redeker, G.: über den Einfluß des Aufplatzens der Wirbel auf die Aerody-
namischen Beiwerte von Deltaflügeln mit kleinem Seitenverhältnis beim Schiebeflug. WGLR
Jahrbuch 232–240 (1967)
33. Legendre, R.: Ecoulement en Voisinage de la Pointe Avant d’une Aile à Forte Flèche aux
Incidences Moyennes. La Recherche Aeronautique No. 30 (1952)
34. Brown, C.E., Michael, W.H.: Effect of leading-edge separation on the lift of a delta wing. J.
Aeronaut. Sci. 21, 690–694 (1954)
35. Küchemann, D.: A non-linear lifting-surface theory for wings of small aspect ratio with edge
separation. RAE Aero Report No. 2540 (1955)
36. Mangler, K.W., Smith, J.H.B.: A theory of the flow past a slender delta wing with leading-edge
separation. RAE Aero Report No. 2593 (1957)
37. Polhamus, E.C.: A concept of the vortex lift of sharp edge delta wings, based on a leading-edge
suction analogy. NASA TN D-3767 (1966)
38. Hall, M.G.: A theory for the core of a leading-edge vortex. J. Fluid Mech. 11(2), 209–228
(1961)
39. Hall, M.G.: A numerical method for solving the equations of a vortex core. RAE TR 65106
(1965)
40. Ludwieg, H.: Zur Erklärung der Instabilität der über angestellten Deltaflügeln auftretenden
freien Wirbelkerne. ZFW 10, 242–249 (1962)
41. Benjamin, T.B.: Theory of the vortex breakdown phenomenon. J. Fluid Mech. 14(4), 593–629
(1962)
42. Polhamus, E.C.: Applying slender wing benefits to military aircraft. J. Aircr. 21(8), 545–559
(1984)
43. Smith, J.H.B.: Theoretical modelling of the three-dimensional vortex flows in aerodynamics.
AGARD CP-342, Paper 17 (1983)
44. Hoeijmakers, H.W.M.: Computational vortex flow aerodynamics. AGARD CP-342, Paper 18
(1983)
References 355

45. Hirschel, E.H., Krause, E. (eds.): 100 Volumes of ‘Notes on Numerical Fluid Mechanics and
Multidisciplinary Design’. NNFM100. Springer, Berlin (2009)
46. Rom, J.: High Angle of Attack Aerodynamics. Springer, New York (1992)
47. Vos, J.B., Rizzi, A., Darracq, D., Hirschel, E.H.: Navier-Stokes solvers in European aircraft
design. Prog. Aerosp. Sci. 38, 601–697 (1992)
48. Eriksson, L.-E., Rizzi, A.: Computation of vortex flow around wings using the Euler equations.
In: Viviand, H. (ed.) Proceedings of the Fourth GAMM-Conference on Numerical Methods in
Fluid Mechanics, Paris, 7–9 October 1981. Notes on Numerical Fluid Mechanics, vol. 5, pp.
87–105. Vieweg-Verlag, Braunschweig (1982)
49. Hoeijmakers, H.W.M.: Computational aerodynamics of ordered vortex flow. Doctoral thesis,
Technical University Delft, The Netherlands (1989)
50. Hitzel, S.M.: Zur numerischen Simulation dreidimensionaler, kompressibler Flügelvorderkan-
tenwirbelsysteme in einem Euler-Verfahren (About the numerical simulation of leading-edge
vortex systems with an Euler Method). Doctoral thesis, University Stuttgart, Germany (1993)
51. Hentschel, R.: Entwicklung und Anwendung eines dreidimensionalen selbstadaptiven Ver-
fahrens zur Simulation viskoser Strömungen auf der Basis strukturierter Gitter (Development
and application of a three-dimensional self-adaptive method for the simulation of viscous flows
on the basis of structured grids). Doctoral thesis, University Stuttgart, Germany (1996)
52. Elsenaar, A., Eriksson, G. (eds.): International vortex-flow experiment on Euler code validation.
In: Proceedings of the Symposium Stockholm, 1–3 October 1986. FFA, Bromma, Sweden
(1987)
53. Drougge, G.: The international vortex flow experiment for computer code validation. In: ICAS-
Proceedings 1988, vol. 1, pp. XXXV–XLI (1988)
54. Elsenaar, A., Hjelmberg, L., Bütefisch, K.-A., Bannink, W.J.: The international vortex flow
experiment. Vortex Flow Aerodynamics, AGARD-CP-437, 9-1–9-23 (1988)
55. Eberle, A., Rizzi, A., Hirschel, E.H.: Numerical Solutions of the Euler Equations for Steady
Flow Problems. Notes on Numerical Fluid Mechanics, vol. 34. Vieweg, Braunschweig (1992)
56. Schwarz, W.: IEPG-TA15 computational methods in aerodynamics, MBB contribution to the
theoretical program of Phase 2. MBB-FE122-AERO-MT-833 (1986)
57. Eberle, A., Misegades, K.: Euler solution for a complete fighter aircraft at sub- and supersonic
speed. AGARD CP-412, 17-1–17-12 (1986)
58. Hummel, D., Redeker, G.: A new vortex flow experiment for computer code validation. In:
Proceedings of the RTO AVT Symposium on Vortex Flow and High Angle of Attack Aerody-
namics. Loen, Norway, 7–11 May 2001. RTO-MP-069, Paper Nr. 8 (2002)
59. Chu, J., Luckring, J.M.: Experimental surface pressure data obtained on 65◦ delta wing across
Reynolds number and Mach number ranges. NASA TM 4645 (1996)
60. Hummel, D.: The second international vortex flow experiment (VFE-2): objectives and first
results. J. Aerosp. Eng. 220(6), 559–568 (2006)
61. Hummel, D.: The second international vortex flow experiment. In: Hirschel, E.H., Krause, E.
(eds.) 100 Volumes of ‘Notes on Numerical Fluid Mechanics and Multidisciplinary Design’,
NNFM100, pp. 231–240. Springer, Berlin (2009)
62. Luckring, J.M.: The discovery and prediction of vortex flow aerodynamics. Aeronaut. J.
123(1264), 729–804 (2019)
63. Rizzi, A., Luckring, J.M.: Evolution and use of CFD for separated flow simulations relevant
to military aircraft. In: Proceedings of the AVT-307 Research Symposium—Separated Flow:
Prediction, Measurement and Assessment for Air and Sea Vehicles. Trondheim, Norway, 7–9
October 2019. STO-MP-AVT-307, Paper No. 3 (2019)
64. Jones, R.T.: Properties of low-aspect-ratio pointed wings at speeds below and above the speed
of sound. NACA Rep. 835 (1946)
65. Hirschel, E.H.: Neue strömungsphysikalische Aspekte in der Aerodynamik. Physik in unserer
Zeit, Jahrg. 15(5), 145–155 (1984)
66. Hummel, D.: On the vortex formation over a slender wing at large angles of incidence. AGARD
CP-247, 15-1–15-17 (1978)
356 10 Small Aspect-Ratio Delta-Type Wing Flow

67. Luckring, J.M.: Initial experiments and analysis of blunt-edge vortex flows for VFE-2 config-
urations at NASA Langley, USA. Aerosp. Sci. Technol. 24, 10–21 (2013)
68. Wentz Jr., W.H., Kohlmann, D.L.: Vortex breakdown on slender sharp-edged wings. J. Aircr.
8(3), 156–161 (1971)
69. Stanbrook, A., Squire, L.C.: Possible types of flow at swept leading edges. Aeronaut. Q. 15(1),
77–82 (1964)
70. Ganzer, W., Hoder, H., Szodruch, J.: On the aerodynamics of hypersonic cruise vehicles in
off-design conditions. In: ICAS 1978, Proceedings, vol. 1, pp. 152–161 (1978)
71. Wood, R.M., Miller, D.S.: Fundamental aerodynamic characteristics of delta wings with
leading-edge vortex flows. J. Aircr. 22(6), 479–485 (1985)
72. Luckring, J.M., Boelens, O.J.: A unit-problem investigation of blunt leading-edge separation
motivated by AVT-161 SACCON research. RTO-MP-AVT-189, 27-1–27-27 (2011)
73. Hirschel, E.H., Cousteix, J., Kordulla, W.: Three-Dimensional Attached Viscous Flow.
Springer, Berlin (2014)
74. Hentschel, R.: The creation of lift by sharp-edged delta wings. An analysis of a self-adaptive
numerical simulation using the concept of vorticity content. Aerosp. Sci. Technol. 2, 79–90
(1998)
75. Elsenaar, A., Hoeijmakers, H.W.M.: An experimental study of the flow over a sharp-edged
delta wing at subsonic and transonic speeds. Vortex Flow Aerodynamics, AGARD-CP-494,
15-1–15-19 (1991)
76. Burggraf, U., Ehlers, T.: Aerodynamische Untersuchungen an Doppeldeltaflügelkonfiguratio-
nen mit Seitenkante. Strömungen mit Ablösung, DGLR-Bericht 94-04, pp. 185–190 (1994)
77. Hummel, D.: The international vortex flow experiment 2 (VFE-2): background, objectives and
organization. Aerosp. Sci. Technol. 24, 1–9 (2013)
78. Luckring, J.M., Hummel, D.: What was learned from the new VFE-2 experiments. Aerosp.
Sci. Technol. 24, 77–88 (2013)
79. Fritz, W., Cummings, R.M.: What was learned from the numerical simulations for the VFE-2.
AIAA-Paper 2008-399 (2008)
80. Konrath, R., Klein, Ch., Schröder, A., Kompenhans, J.: Combined application of pressure
sensitive paint and particle image velocimetry to the flow above a delta wing. In: Proceedings
of the 12th International Symposium on Flow Visualization, DLR Göttingen, Germany, 10–14
September 2006, pp. 1–14. Optimage Ltd., Edinburgh, UK (2006)
81. Konrath, R.: Personal communication (2019)
82. Fritz, W.: Numerical simulation of the peculiar subsonic flow field about the VFE-2 wing with
rounded leading edge. Aerosp. Sci. Technol. 24, 45–55 (2013)
83. Schütte, A., Lüdecke, H.: Numerical investigations on the VFE-2 65-degree rounded leading
edge delta wing using the unstructured DLR TAU-Code. Aerosp. Sci. Technol. 24, 56–65
(2013)
84. Furman, A., Breitsamter, C.: Turbulent and unsteady flow characteristics of delta wing vortex
systems. Aerosp. Sci. Technol. 24, 32–44 (2013)
85. Crippa, S.: Advances in vortical flow prediction methods for design of delta-winged aircraft.
Doctoral thesis, Kungliga Tekniska Högskolan (KTH), Rep. TRITA-AVE 2008:30, Stockholm,
Sweden (2008)
86. Schiavetta, L.A., Boelens, O.J., Crippa, S., Cummings, R.M., Fritz, W., Badcock, K.J.: Shock
effects on delta wing vortex breakdown. AIAA-Paper 2008-0395 (2006)
87. Donohoe, S.R., Bannink, W.J.: Surface reflective visualizations of shock-wave/vortex interac-
tions above a delta wing. AIAA J. 35(10), 1568–1573 (1997)
88. Hövelmann, A.: Analysis and control of partly-developed leading-edge vortices. Doctoral the-
sis, Technical University München, Germany, Verlag Dr. Hut, München (2017)
89. Riedelbauch, S.: Aerothermodynamische Eigenschaften von Hyperschallströmungen über
strahlungsadiabate Oberflächen (Aerothermodynamic properties of hypersonic flows past
radiation-cooled surfaces). Doctoral thesis, Technische Universität München, Germany, 1991.
Also DLR-FB 91-42 (1991)
References 357

90. Riedelbauch, S., Hirschel, E.H.: Aerothermodynamic properties of hypersonic flow over
radiation-adiabatic surfaces. J. Aircr. 30(6), 840–846 (1993)
91. Hirschel, E.H.: Basics of Aerothermodynamics, 2nd Revised edn. Springer, Cham (2015)
92. Linde, M.: Experimental test on a planar delta wing at high Mach number and high angle of
attack. FFA TN 1988-59, Bromma, Sweden (1988)
93. Crippa, S., Rizzi, A.: Numerical investigation of Reynolds number effects on a blunt leading-
edge delta wing. AIAA-Paper 2006-3001 (2006)
94. Hummel, D.: Effects of boundary layer formation on the vortical flow above slender delta wings.
In: Proceedings of the RTO AVT Specialists’ Meeting on Enhancement of NATO Military
Flight Vehicle Performance by Management of Interacting Boundary Layer Transition and
Separation. Prague, Czech Rep., 4–7 October 2004. RTO-MP-AVT-111, 30-1–30-32 (2004)
Chapter 11
Selected Flow Problems of Small
Aspect-Ratio Delta-Type Wings

The main topic of this book are the basics of separated and vortical flow in aircraft
wing aerodynamics. However, like in Chap. 9, we present also here some application-
oriented topics. The implications of in particular vortex breakdown are important
not only for the aerodynamic coefficients, but also for their derivatives. They are
determining the vehicle’s stability and control characteristics, i.e., the flight properties
and the handling qualities, and last, but not least, its flight envelope. These issues are
shortly discussed in Sect. 11.1.
Consequently we discuss some means to influence lee-side vortex systems by
geometrical shaping of a wing, and also by other active and passive means. We
concentrate on configurational design aspects when trying to draw benefits from the
lee-side vortex induced lift with respect to performance and controllability of the
complete configuration. Simultaneously introduced adverse effects—for example
regarding the longitudinal stability—have to be analyzed and reduced/eliminated in
the process of aircraft (pre-) design.
Generation of leading-edge vortex systems at delta wings in many cases will
go together with their manipulation. Geometrical measures may become visible as
so-called trigger devices at various locations—fuselage, wing, tail unit—of the con-
figuration under consideration. We give overviews in Sects. 11.2, 11.3 and 11.4.
Besides the geometrical measures, concentrated and distributed spanwise blow-
ing over the wing upper surface allows to generate vortex systems via “artificial”
cross-flow variations and also to stabilize a geometrically introduced vortex system,
Sect. 11.5. After that in Sect. 11.6 a step by step modification of a wing geometry is
shown which leads to a design with the desired aerodynamic properties.
In the review article [1] about control of leading-edge vortices, also rather exotic
approaches are treated.

© Springer-Verlag GmbH Germany, part of Springer Nature 2021 359

E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_11
360 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Fig. 11.1 Lift characteristic

of a delta wing with lee-side
vortices

11.1 Lift and Stability Problems Connected to Lee-Side

Vortex Systems of Delta-Type Wings

With Figs. 11.1 and 11.2 we recapitulate main insight from the preceding Chap. 10.
Figure 11.1 shows the basic phenomenon of lee-side vortices and the non-linear lift
due to them. The latter arises already at rather small angles of attack.
Figure 11.2 gives in detail, but idealized, the flow, respectively the lee-side primary
vortex phenomena occurring at sharp and round-edged thin planar wings.
Along the abscissa the leading-edge sweep angle ϕ0 ranges from about 50◦ to 85◦ .
Along the ordinate the angle of attack α ranges from about 2◦ to 40◦ . Occurring in
these intervals are the interesting vortex phenomena, shown here for the right-hand
side of the wing.
Indicated are five domains of interest. In domain 0 no lee-side vortex is present,
the lee-side flow is attached. Domain 1 indicates the evolution of the lee-side vortex,
which with increasing α is moving from the back of the wing—the trailing edge—
toward its tip. (Note that indicated in the figure is only the evolution of the primary
vortex.) The higher the sweep angle ϕ0 is, the smaller is the angle of attack, at
which the lee-side vortex is beginning to appear. This, by the way, is true for all
of the phenomena indicated in Fig. 11.2, except for the vortex breakdown and the
maximum attainable angle of attack.
Domain 2 sees the fully developed lee-side vortex moving inward with increasing
angle of attack. In Domain 3 the span-wise location of the vortex is fixed. Domain 4,
in reddish color, at its lower bound indicates the beginning of vortex breakdown—
αvor tex bur sting —, and at its upper bound the maximum angle of attack αmax .1 Beyond
this angle total stall is present, maximum lift is surpassed.

1 Vortex breakdown is also denoted as vortex bursting.

11.1 Lift and Stability Problems Connected to Lee-Side Vortex Systems … 361

Fig. 11.2 Flow phenomena at sharp- and round-edged thin planar delta wings as function of the
leading-edge sweep angle ϕ0 and the angle of attack α. The broken lines constitute the upper bounds
of the sharp leading-edge domains 1 and 2

Important is that both the lower and the upper bound of domain 4 indicate that
both, the beginning of vortex breakdown and its reaching the wing tip, start at the
lowest angle of attack for the wing with smallest leading-edge sweep ϕ0 ≈ 50◦ .
With increasing sweep angle the critical angles of attack move from α = 5◦ to
approximately 37◦ , respectively from 24◦ to 40◦ .
Regarding the vortex overlapping over the wing, Sect. 10.2.5, we note that the
influence of the fuselage of a full configuration reduces the likelihood of its occur-
rence. The effect, of course, then can occur at the aircraft’s forebody. The general
problem is that a straightforward transfer of data and behavior of the isolated wing
to the full aircraft in general is not possible.
Together with the non-linear lift comes an unwelcome effect, the pitch-up ten-
dency of the wing, respectively the aircraft. Figure 11.3 in the upper part shows the
coefficient of the pitching moment Cm (α), which builds up like the non-linear lift
component, reaching its maximum just ahead of C L max . A proper sizing and position-
ing of the horizontal stabilizer can correct this tendency.
Two other unwelcome effects are indicated in the lower part of that figure, the
behavior of the coefficients of the rolling moment Clβ —concerning the dihedral or
lateral stability—and the yawing moment Cn β —concerning the directional stability.
The divergence of both the rolling moment and the yawing moment happens
at the divergence angle of attack αdiv , which is smaller than αmax of the lift. The
consequence of the two divergencies may be the loss of the controllability of the
aircraft. This comes in terms of aileron reversal, reduced rudder efficiency, and spin
362 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Fig. 11.3 Wing-body-tail configuration: typical idealized curve characteristics of the lift coefficient
C L , the pitching-moment coefficient Cm , the rolling-moment coefficient Cl , and the yawing-moment
coefficient Cn as function of the angle of attack α. β is the sideslip angle. Lower right part: divergence
behavior

danger. Furthermore asymmetries in breakdown between the two wing sides may
cause additional moments, which are to be actively controlled.
In addition strong unsteadiness of the flow is present, once vortex breakdown
happens. The resulting dynamic structural loads also are of great concern. Hence the
aerodynamical useful maximum lift cannot be exploited due to these issues.
This is demonstrated in Fig. 11.4 for thin wings with sharp leading edge in the
low-speed domain.
The maximum lift coefficient C Lmax is obtained for a leading-edge sweep angel
ϕ0 ≈ 66◦ , upper part of the figure. In the lower part it is shown that for the sweep-
angle interval ϕ0  50◦ the divergence angle αdiv is distinctly smaller than αmax of
the lift. Hence the lift potential of the wing cannot be exploited due to the lateral
stability and control problem.
Regarding the pitch-up problem, Fig. 11.5 shows an empirical limiting curve for
tolerable pitch-up. The curve is based on [3]. The correlation of J.A. Shortal and
B. Maggin, [4], lies closely to the left of it. Below the yellow-shaded curve the pitch-
up of the delta-wing configuration is controllable, above not. Note that wings with
11.1 Lift and Stability Problems Connected to Lee-Side Vortex Systems … 363

Fig. 11.4 Thin delta wings with sharp leading edge in the low-speed domain: maximum lift coef-
ficient C Lmax , maximum angle of attack αmax , and divergence angle of attack αdiv as function of
the leading-edge sweep angel ϕ0 (based on [2])

Fig. 11.5 Delta-wing configuration: empirical limiting curve for tolerable pitch-up (based on [3])
364 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

the aspect ratio Λ below the broken line—Λ = 4 tan ϕ0 —are cropped delta wings,
like the VFE-1 configuration, Fig. 10.25.
Since a long time configurational and other measures are studied and also
employed successfully in order to extend the useful angle-of-attack domain of air-
craft. Flyability and controllability in the stall and also the post-stall domain are of
major concern. Flight in the post-stall domain of course demands thrust-vectoring of
the propulsion system, or even control thrusters, like those employed on the former
Space-Shuttle Orbiter.
In the following sections we discuss shortly some topics in this regard. Always
only the longitudinal motion, and stability and control of the flight vehicle or the
isolated wing is being considered.

11.2 Wing-Planform Shaping and Optimization

Two topics are presented in a compact manner, first the effects of the geometry of a
delta wing at low flight Mach number, and then that of small aspect-ratio wings of
high-speed flight vehicles.

11.2.1 Effects of the Wing Geometry

To generate and/or manipulate leading-edge vortex systems, the geometry of the wing
provides the driving parameters. Besides the wing planform (aspect ratio/leading-
edge sweep), the wing section characteristics (leading-edge radius/camber/thickness)
are contributors. But note that the “father” of the design is the delta-wing (planform).
In this context adjustment of the aircraft forebody cross section can play an important
role in view of the directional/longitudinal stability at high angle of attack (“shark
nose” type forebody, nose strakes).
Regarding the wing planform and its modifications, we note that the driving
parameter for the development and the stability of leading-edge vortex systems is
the leading edge sweep angle ϕ0 . For the pure delta-wing the sweep is coupled to
the aspect ratio via Λ = 4/tan ϕ0 . We note:

– (a) To develop a “full” leading-edge vortex system over a (sharp-edged) delta-wing

the sweep angle should exceed ϕ0 = 53◦ , hence the wing aspect ratio Λ should
be less than 3.
– (b) When the vortex breakdown is reaching—from the aft of the wing—the trailing
edge, lift development is affected by a break. The respective angle is α L B , Fig. 11.6,
low-speed data. We note that a break in this form does not appear in all cases.
Therefore a break is not indicated in the respective figures in the foregoing section.
– (c) Vortex breakdown then will move forward with increasing angle of attack α.
Maximum lift is reached for an angle of attack approximately half of the leading-
11.2 Wing-Planform Shaping and Optimization 365

Fig. 11.6 Flat sharp-edged delta wings [2]: maximum angle of attack αmax and angle of attack of
lift break α L B as function of the leading-edge sweep angle ϕ◦ . Low-speed wind-tunnel data of a
wing-body configuration

Fig. 11.7 Flat sharp-edged delta wings [2]: location of vortex breakdown x/li at αmax as function
of the leading-edge sweep angle ϕ◦ . Low-speed wind-tunnel data

edge sweep angle ϕ0 . This behavior is indicated in Fig. 11.6, too. Figure 11.7, again
with low-speed data, shows the location of vortex breakdown as function of ϕ0 .
The subsequent increase of the separation area over the wing raises the danger of
longitudinal and lateral instabilities, both static and dynamic ones.
366 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

11.2.2 Wings of High-Speed Vehicles

After having presented these results, relevant for low flight speed, say M∞  0.4, we
now turn to higher Mach-number flight vehicles, which generally have wings with
small aspect ratio. Two representative flight vehicles were examined in Sect. 10.1,
the Concorde, Fig. 10.1 and the Space Shuttle Orbiter, Fig. 10.2.
Lee-side vortex phenomena occurring over the wings of high-speed vehicles are
correlated with the normal angle of attack α N and the normal leading-edge Mach
number M N , Sect. 10.2.6.
The two flight vehicles feature the typical sharp—Concorde—and round—Space
Shuttle Orbiter—leading edges. We note that the Orbiter at hypersonic flight is outside
of the scope of our considerations, but see for such configurations the Blunt Delta
Wing case in Sect. 10.7. General aspects of the hypersonic flight regime can be found
in [5].
To make use of additional nonlinear lift a basic wing planform can be modified by
highly swept leading-edge extensions, positioned inboard at the wing-body juncture.
Such “strakes” are found at the Concorde and at the fighter aircraft F-16 and F-18,
Fig. 10.5.
At the Concorde the ogive wing can be considered in the following way: a highly
swept delta-type wing is modified—inboard and in the tip region—by continuously
increasing the leading-edge sweep towards the body and the wing tip location. Thus
strake-type planform parts are produced at these locations, which smoothly merge
into the shape of the main wing, yielding in this way the ogive wing shape.
Comparing the wings of the Concorde and the F-16 and F-18, two major geomet-
rical differences are visible:

1. The basic wing of Concorde is a highly swept delta type with low aspect ratio. In
contrast to that the fighters have trapezoidal wings with low leading-edge sweep
and medium aspect ratios Λ > 3.
2. The inboard strakes differ in area and shape. Additional surface due to the strake
is less than one per cent for the Concorde, but around 10 per cent for the fighters,
and transition of the leading-edge sweep into the basic wing is abrupt for the
fighters.

The reason for these differences in wing shapes is found in their design points. The
Concorde is a supersonic transport aircraft—without horizontal stabilizer—with a
cruise Mach number M∞ = 2. The sweep of its wing is adapted to this Mach number
and presents a subsonic leading edge and hence reduced wave drag. At subsonic Mach
numbers the configuration benefits from the development of the stable leading-edge
vortices and their high nonlinear lift contribution at take-off and landing. In turn the
needed angle of attack is reduced there. Note that the ground-effect also increases
with decreasing aspect ratio.
For the fighters the strake alone is the origin of their high non-linear lift for sub-
sonic/transonic high angle-of-attack maneuvering. The effect exemplarily is shown
in Fig. 11.8 [6]. Compared is the maximum lift in terms of C L max and the buffet onset
11.2 Wing-Planform Shaping and Optimization 367

Fig. 11.8 Comparisons of the aerodynamic ranges of maximum lift C L max (left) and buffet-onset
lift C L bo (right) of basic wing and basic wing with strake as function of the flight Mach number
M∞ [6]. Low-speed wind-tunnel data

lift in terms of C L bo as functions of the flight Mach number M∞ for a trapezoidal

wing with and without strake.
The geometrical data of the wings are given in the left part of Fig. 11.8: the basic
wing has the leading-edge sweep ϕ0 = 32◦ , the strake ϕ0 = 75◦ . The simple strake
with an additional area of 10 per cent of the basic wing was in the subsonic region as
effective as an increase of the basic-wing area by 50 per cent for the same maximum

Fig. 11.9 Maximum lift

C L max of the basic wing with
three different strake types as
function of the flight Mach
number M∞ [6].
Wind-tunnel data
368 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Fig. 11.10 Lift-to-drag ratio

L/D of the basic wing with
three different strake types as
function of the lift coefficient
C L at M∞ = 0.5 [6]

lift. Regarding buffet-onset its efficiency increased by more than 50 per cent. Note
that in the transonic regime around M∞ = 0.8 the values of C L bo drop, and that by a
greater amount for the strake wing.
The increase of the lift due to the strake is always combined with a pitch-up
tendency of the wing-body configuration. A tailless configuration is not possible in
this case, but a low positioned horizontal tail allows for a correction, see the F-16
and the F-18 in Fig. 10.5 as examples.
The influence of three strake types on the maximum lift coefficient C L max as
function of M∞ is shown in Fig. 11.9. Strake 1 has a straight leading edge, strake 5
(F16) is of Concorde wing type and strake 9 (F-18) a “gothic” one. All strakes are
effective below M∞ = 0.9, strake 9 being the most effective one.
Regarding the lift-to-drag ratio L/D—the aerodynamic efficiency—of the basic
wing with the three strakes, at M∞ = 0.5 strake 5 is the best with L/D = 11.8 at
C L = 0.3, Fig. 11.10.

11.3 Wing Sections and Leading-Edge Flaps

Besides wing-planform optimization the shaping of wing sections is an approach

to improve and optimize the performance of a small aspect-ratio delta-type wings.
On the other hand leading-edge flaps also allow to improve the performance of such
wings. This section gives short discussions of both topics.
11.3 Wing Sections and Leading-Edge Flaps 369

11.3.1 Wing Sections

The chord profile of the wing is usually defined by its nose radius r◦ , the thickness
ratio t/L and the camber ratio f /L, L being the reference length of the wing, see
Sect. 10.2.1. With respect to lee-side vortex systems the driving parameter from the
wing-section side is the physical dimension of the leading-edge radius.
The larger this radius, the larger—in terms of the angle of attack—is the domain of
attached flow around it, which is fully developed and stable, if the flow is subsonic,
Sect. 10.2.4. Note again the importance of the normal angle of attack α N and the
normal leading-edge Mach number M N . These are the relevant criteria when we
look at the flow past wings with highly swept leading edges, Sect. 10.2.6. The type
of separation may be ordinary, locally separation bubbles may be present.
If at a delta-type wing a lee-side vortex system is present, its development and
existence domain, as well as the vortex breakdown disposition can be taken from
the correlations given in Sect. 10.2.6—Fig. 10.22 for wings with sharp leading edges
and Fig. 10.23 with round leading edges—and in Sect. 10.2.5.
In Fig. 10.23 the correlations correspond to a leading-edge Reynolds number Re N
= r N v N /ν =  2·104 . r N , respectively r◦ , is the leading-edge radius, v N is the free-
stream velocity component normal to the leading edge, ν the kinematic viscosity.
If the magnitude of the Reynolds number is above that value, fully developed flow
around the leading edge is present.
The lower boundary of the leading-edge Reynolds number is Re N ≈ 7·103 . At
this number “full” leading-edge suction is beginning to appear.
In design aerodynamics of delta-type wings the concept of the leading-edge suc-
tion analogy of E.C. Polhamus [7, 8], was and still is used. That concept is based
on an analogy between the lee-side suction pressure due to the lee-side vortices at
a sharp-edged wing—see for instance Fig. 10.17—and the suction pressure at the
leading edge of a blunt-edged wing.2 The concept allows to obtain the so called
vortex lift as well as the induced drag due to it. Here it is used in order to quantify
the Reynolds-number effect.
We give only an overall picture, for details see [7], and also [2]. The lift coefficient
results from the suction analogy to

CL = CL p + CLv , (11.1)

where C L p is the—nota bene linear—potential-theory (Model 4) lift and C L v the lift

due to the lee-side vortex system. These two parts are written as

2 The suction-pressure concept goes back to the application of circulation theory (Model 4). Consider

an airfoil at angle of attack. The only force is the lift force L, normal to the freestream direction.
Its component in x-direction of the airfoil is Px = −L sinα = S, i.e., the suction force, which is
forward directed [9]. It is attributed to the low pressure—the suction pressure—due to the flow
around the leading edge of the airfoil, which is a high-speed flow. The higher the angle of attack,
the larger is S. In our context it regards the flow around the swept leading edges of delta wings and
the resulting non-linear lift due to the lee-side vortex system.
370 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

C L p = K p sin α cos2 α, (11.2)

with K p = dC L p /dα and

C L v = K v sin2 α cos α, (11.3)

with

K 2p1
K v = (K p − ) (11.4)
πΛ cos ϕ0

as derivative of the normal-force coefficient of the by 90◦ rotated, fictitious leading-

edge suction force, i.e., K v = dC L /dα2 .
Remembering Eq. (8.2), the induced drag can be written in a general way as

C Di = K i C L2 , (11.5)

with K i being the induced-drag factor. This factor varies between two limits. The
first limit is K 1 = 1/π Λ, which is considered to represent full leading-edge suction.
Zero leading-edge suction is given in the limit K 2 ≈ 1/(dC L /dα).
K 2 is an approximation, because the “exact” induced drag coefficient in this case is
C Di = C L tan α. This simple consideration directly shows that K 2 and hence C Di are
reduced by the development of the non-linear lift due to the lee-side vortex system,
as for a demanded constant lift a lower angle of attack would be necessary.
Coming back to the influence of the above leading-edge Reynolds number Re N on
the development of the vortex lift, Fig. 11.11 shows the percentage of leading-edge
suction S ∗ attained at a cropped delta-wing/body configuration with rounded leading
edge as function of the leading-edge Reynolds number Re N in the low-speed regime,
see [10] and also [11].
The experimental data were found for dC L /dα values always corresponding to
the lift condition C L = C L (L/D)opt . For the Reynolds number Re N  2·104 , the
value underlying Fig. 10.23, it is demonstrated that—for this case—the maximum
attainable lee-side suction pressure is about 90 per cent.
Compared to the situation with sharp leading edges, the development of a stable
lee-side vortex system is shifted to higher angles of attack and in consequence the
amount of useable nonlinear lift is reduced. Consequently, the induced drag at a given
angle of attack reduces with increasing leading-edge radius, which, for instance, may
be helpful in view of the specific excess power (SEP) goals of the design process.3
In general, one has to be careful using wind-tunnel data, because Reynolds-number
effects can play a large role.

3 Thespecific excess power is the power surplus needed/available for the acceleration of the flight
vehicle or for its climbing performance.
11.3 Wing Sections and Leading-Edge Flaps 371

Fig. 11.11 Effect of the

Reynolds number Re N on
the development of the
leading-edge suction force
[10]. The full line graph is
the mean value of measured
data S ∗ (hatched area)

11.3.2 Leading-Edge Flaps

The effect of the wing section, in particular the leading-edge or nose radius, was
the topic of the preceding sub-section. Now the matter of—movable—leading-edge
flaps at small aspect-ratio delta-type wings is considered. Such flaps allow to some
extent to manipulate the development, position and effects of lee-side vortex systems
in the relevant angle-of-attack regime [10, 11].
The manipulation via the extension of leading-edge flaps is done in view of per-
formance, stability around all axes and handling qualities of the considered aircraft.
Magnitude (and direction) of the flap deflection should correspond to the intention
of finding/producing the most effective angle of attack the leading-edge region is
“feeling” for a given flight condition.
In the background of the following two figures is the lift condition C L =
C L (L/D)opt . Figure 11.12 shows for sharp-edged delta wings with different chord-
wise airfoil shapes the comparison of the development of leading-edge suction S ∗ as
function of the leading-edge sweep angle ϕ0 .
The lowest development of S ∗ is found for the symmetric chordwise shape, warp-
ing with the shown shape increases S ∗ , and an even stronger increase is found for
the nose-flap shape. Up to a leading-edge sweep angle ϕ0 ≈ 40◦ the leading-edge
suction is nearly constant, for higher sweep angles it drops markedly.
We are treating mainly small aspect-ratio delta-type wings. This means high
leading-edge sweep angles ϕ0 . Figure 11.13 shows the development of the leading-
edge suction for a rounded leading-edge delta wing with two different chord (camber)
shapes. The wing has the leading-edge sweep ϕ0 = 63◦ . Because it is a delta wing,
the aspect ratio is Λ = 2 and the factor for induced drag for 100 per cent leading-edge
suction hence is K 1 = 1/(π Λ) = 0.16, that for zero suction is K 2 ≈ 0.34. Therefore
a loss of 10 per cent of S ∗ will increase K 1 by (K 2 − K 1 )/K 1 × 0.1 ≈ 11 per cent.
372 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Fig. 11.12 Effect of nose flap and wing warp for sharp-edged delta wings as function of the
leading-edge sweep angle ϕ0 [10]

Fig. 11.13 Effect of conical camber of a delta wing as function of the free-stream Mach number
M∞ [10]

Note that the wing in Fig. 11.13 has a five per cent thick chord section with rounded
leading edge, the wings in Fig. 11.12 have sharp leading edges. Our example is found
in Fig. 11.12 to the right for ϕ0 = 63◦ . The figure gives at that angle for the symmetric
chord section S ∗ ≈ 40 per cent, for the warped one S ∗ ≈ 50 per cent, and for that
with the nose flap S ∗ ≈ 62 per cent.
The corresponding factors of lift-dependent drag, the actual K 1 -factors, then are
0.268, 0.25, 0.23 (K 1 = 0.16 for 100 per cent S ∗ , see before). We conclude that the
11.3 Wing Sections and Leading-Edge Flaps 373

leading-edge flap has produced a reduction of the lift-dependent drag—for the case
of a sharp leading edge—of (0.268 − 0.23)/0.268 = 14.2 per cent.
Deeper insights in our theme can be found in the papers by R.K. Nangia et al.
[12], J.M. Brandon et al. [13], the latter showing flight results of the F-106B aircraft,
and also by N.T. Frink et al. [14].

11.4 Fuselage Forebody Strakes

Fuselage forebody strakes are designed to (re-)stabilize the—usually military—delta-

type aircraft at high, up to post-stall angles of attack in lateral/directional motion (clβ ,
cn β ).
Deflecting strakes and/or asymmetrical activated strakes allow to generate vortices
on the forebody. Correct positioning of such devices on the forebody—x- and z-
location—is mandatory and in many cases a hard task for the aircraft designer.
When making use of forebody-strakes, one has to be aware of two imminent
dangers:
1. Forebody strakes contribute to static directional stability at high angle of attack,
but their effect regarding dynamic stability can be adverse, because they can lead
to auto-rotation of the aircraft around its vertical axis (spin).
2. When at sideslip condition the windward forebody-strake vortex (with the strake’s
higher effective angle of attack and dynamic pressure) is overlapping the leeward
one, the nose-strake effect becomes destabilizing for the aircraft in its lateral
motion.
Complete knowledge of the flow field around the aircraft therefore is mandatory
for success. We show as example a few figures from a study of S.M. Hitzel and
R. Osterhuber [15]. Investigated was a supersonic combat configuration in order to
extend its maneuverability to very high angles of attack.
At the configuration leading-edge vortex flow with mutual interaction and break-
down effects dominate the aircraft’s behavior. URANS simulations around the con-
figuration with all its geometrical details exhibited that asymmetric vortex breakdown
and interaction features at side slip were the cause of the limitations of maneuver-
ability.
The study finally led to the application of fuselage forebody strakes and leading-
edge root extensions (LERX), see below, which considerably enhanced the lateral
stability and the attainable angle of attack. All that was confirmed by wind-tunnel
and flight tests.
Figure 11.14 from the publication [15] shows very detailed the computed flow
field over the basic—Standar d—delta-canard configuration at high angle of attack
and sideslip. Note that also the deployed slats were modeled and hence the resulting
slat vortices were taken into account.
The large figure shows the side of the aircraft with the advancing wing, the small
figure at the low left shows the aircraft with receding wing (sideslip toward to the
374 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Fig. 11.14 Computed surface pressure, vorticity cross-sections and vortex-core paths at the
Standar d configuration at subsonic speed and high angle of attack [15]

port side of the aircraft). This is the same also in the next figure. In both figures it
can be seen that the canard vortex (canard wake) wake passes high over the aircraft.
The interaction with the lee-side flow field appears to be small.
The realized changes on the Standar d configuration in order to improve its
behavior were rather small. The narrow rectangular fuselage fence on both sides
below the cockpit was replaced by a delta-shaped fuselage strake. Further a leading-
edge root extension, connecting the intake-ramp side edge with the wing’s leading
edge, was implemented. The combination of the fuselage strake and the LERX led
to the desired improvements. The configuration is called the E F E M configuration,
which stands for EF2000 Enhanced Maneuverability.
A visualization of the resulting flow field is shown in Fig. 11.15. The changes
of the flow field compared to that of the Standar d configuration are obvious. The
vortex from the intake ramp now appears to be stronger. The vortex axis—yellow
color—is not absorbed by the strake vortex, in contrast to the standard case, where
it merges with that of the fence vortex.
The E F E M fuselage strake vortex—axis in red color—is much stronger than the
Standar d vortex and it follows the wing surface closer. The wing’s lee-side vortex
system as whole is stabilized and over the wing the non-linear lift is increased. All
this is stronger at the advancing wing than at the receding one and leads to the stable
rolling moment of the E F E M fuselage.
This is demonstrated in the next two figures. Figure 11.16 shows for the subsonic
flight domain, how the rolling-moment coefficient cl (α) for the two configurations
behaves.
11.4 Fuselage Forebody Strakes 375

Fig. 11.15 Computed surface pressure, vorticity cross-sections and vortex-core paths at the E F E M
configuration at subsonic speed and high angle of attack [15]

Fig. 11.16 Wind-tunnel

results of the rolling moment
cl as function of the angle of
attack α and the sideslip
angles β = 0◦ and ∓ 5◦ of
both the Standar d and the
E F E M aircraft in the
subsonic domain [15]

For the sideslip angle β = 0◦ the configurations in a large angle-of-attack range

behave well, but then become uncontrollable. For the Standar d configuration at β =
∓5◦ the rolling moment cl abruptly drops to zero at a lower angle of attack. In contrast
to that the E F E M configuration exhibits an acceptable roll-moment evolution to a
much larger angle of attack.
For the transonic flight domain a similar pattern emerges, Fig. 11.17. At β = 0◦ for
both configurations the favorable angle-of-attack range is larger than in the subsonic
domain. For β = ∓5◦ the Standar d configuration shows a markedly asymmetric
behavior, whereas the E F E M configuration apparently behaves better than in the
subsonic case.
With this short presentation of selected results from [15] we have shown that
simple forebody strake devices can be employed in order to influence and enhance
376 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Fig. 11.17 Wind-tunnel

results of the rolling moment
cl as function of the angle of
attack α and the sideslip
angles β = 0◦ and ∓5◦ of
both the Standar d and the
E F E M aircraft in the
transonic domain [15]

the lateral stability of small aspect-ratio delta-type aircraft. For more information
and insight regarding forebody strakes and their function the reader is referred to
[16, 17] and also the paper of the third author of this book [18].

11.5 Spanwise Blowing

The basic ideas regarding the technique of spanwise (concentrated) blowing were
published in the late 1960s. They can be attributed—to the knowledge of the
authors—to J.J. Cornish and C.J. Dixon, both with Lockheed Corporation, USA [19,
20], and to Ph. Poisson-Quinton from ONERA, France. Therefore this technique can
be regarded as the result of a “transatlantic alliance”.
The topic “spanwise blowing” was booming in the 1970s. The effects and the
efficiency of this technique have been well established from the experimental side,
spanning from the improvement of performance—increase of lift and reduction of
lift-dependent drag—to stability issues around all axes, depending on the location of
the blowing jet (wing, body, tail). Also control augmentation of rudders and ailerons
have been a topic.
The fourth author of this book was the leader of the German (MBB, DFVLR) and
French (ONERA) working group “Wings with Controlled Separation”, which from
1969 to 1978 experimentally investigated wing flows with stable leading-edge vortex
systems, and from 1975 to 1982 the effects of concentrated spanwise blowing.
In [21] an overview is given on the major results of the work:

– The merits and and shortcomings of the technique of concentrated spanwise blow-
ing are discussed regarding the aerodynamic performance as well as stability and
control aspects of aircraft.
– The limits of the aerodynamic efficiency are established on an empirical/theoretical
basis and compared to the experimental results. Upper and lower application
boundaries are identified.
11.5 Spanwise Blowing 377

Fig. 11.18 Low-speed modular pilot model [21]

– A view is given regarding the practical (non-) application of this technique and the
reasons why the technique did not find application in operational fighter aircraft.
Discussed are competitive approaches.
The following short overview is on results concerning only the longitudinal motion
of the aircraft. We concentrate on the vortex development and manipulation and on
the major effects of spanwise blowing. The blowing coefficient cμ is defined in the
following way:
ṁ v j
cμ = , (11.6)
q∞ Ar e f

where ṁ is the mass flux of the jet, v j the jet velocity—their product being the jet
thrust—q∞ the dynamic pressure of the free-stream, and Ar e f the wing’s reference
area.
The effect of spanwise blowing is demonstrated with a few results, which were
found with the modular low-speed pilot model with a hybrid wing, Fig. 11.18.
The slat and Fowler flap have downward deflection angles 0◦  δ  90◦ , the
aileron downward and upward angles −90◦  δ  90◦ .
Table 11.1 gives the main geometrical parameters of the trapezoidal wing and the
strake of the modular pilot model.
The main parameters of the investigation are collected in Table 11.2, with c being
the mean chord length.
378 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Table 11.1 Geometrical parameters of the modular pilot model wing [21]. λt is the taper ratio of
the wing
ϕ0 (◦ ) Λ λt ϕ0strake (◦ )
32 3.2 0.3 75

Table 11.2 Low-speed modular pilot model: investigation parameters [21]

M∞ Re∞,c c (m) Angle of attack Side-slip angle Blowing coefficient

0.2 ≈2.1·106 0.42 −5◦ α 90◦ −15◦ β 15◦ 0  cμ  0.4

Spanwise blowing in this case refers to a concentrated transversal jet. That is

directed over the suction side of the trapezoidal wing or the strake and blows roughly
parallel to its plane and leading edge.
Two applications of spanwise blowing are listed now. They can be employed
separately or simultaneously depending on the kind of flow over the wing or the
strake.

1. If the wing has the appropriate geometry—a delta wing or a strake planform and
the adequate angle of attack—and accordingly a stable lee-side vortex system,
the latter can be manipulated by spanwise blowing. Regarding the pilot model,
this applies for the strake.
2. If the wing does not have the appropriate geometry, which leads to a lee-side vortex
system, spanwise blowing can be employed to generate, stabilize and control a
lee-side vortex system. Regarding the pilot model, this applies for the trapezoidal
wing.

11.5.1 Blowing at a Pilot Model Without Strake and

Forebody Fins

The wing of the pilot-model, Fig. 11.18, without strake can be considered as a mod-
erate aspect-ratio/small leading-edge sweep wing. Experiments with this wing as
reference configuration were made in order to study the effect of span-wise blow-
ing on the aerodynamic properties [22–28]. The optimum nozzle position on the
fuselage side over the wing (indicated at the left wing in the figure)—including the
jet-direction (sweep angle and elevation-angle over the wing surface plus the nozzle
location above the wing)—was found in an experimental approach.
Figure 11.19 shows the results for the longitudinal motion at M∞ = 0.2 for the
near optimum jet/nozzle position, here being at 40 per cent root chord (x D /li = 0.4)
at the wing/body intersection. The nozzle diameter is d = 15 mm, ϕ D = 15◦ is the
nozzle angle against the 40 per cent chord line.
11.5 Spanwise Blowing 379

Fig. 11.19 Effect of spanwise blowing (cμ = 0, 0.05, 0.1, 0.2, 0.4) on lift, pitching moment and
drag coefficients [21]. Basic trapezoidal wing, strake and forebody fins off

Analyzing the effects of spanwise blowing on the lift, pitching moment and drag
characteristics we find:
– Maximum lift and angle of attack increase with increasing cμ , the same tendency
is found for the nonlinearity of the lift and the lift curve slope dC L /dα. These
effects clearly indicate that a lee-side vortex system is induced by the spanwise
blowing.
– For low angle of attack a quasi-camber effect is induced for the lift, but without an
influence on the pitching moment. Increasing cμ extends the longitudinal stability,
and hence the constant location of the neutral point up to higher angles of attack
and higher lift.
– The lift-dependent—induced—drag is reduced by spanwise blowing. This effect
must be solely attributed to the nonlinear lift development, i.e., not to any regain
of leading-edge suction.

11.5.2 Blowing at the Pilot Model with Strake and Without

Forebody Fins

The wing of the pilot model with strake can be considered as a high leading-edge
sweep (inboard)/medium aspect-ratio hybrid wing. Also with this wing on the pilot
model experimental investigations were made.
The optimum position of the jet nozzle was found at 10 per cent of total root
chord of the wing (indicated at the right-hand strake in Fig. 11.18). The nozzle was
380 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Fig. 11.20 Effects of wing planform and type of blowing on the lift increment ΔC L /cμ |opt as
function of the blowing coefficient cμ [21]

housed again in the body and blowing parallel to the strake’s leading edge (sweep
angle ϕ L E strake = 75◦ ) and also parallel to the wing’s surface.
The effects and efficiencies of blowing on the wing and on the strake are shown
in the next two figures, Figs. 11.20 and 11.21. The results are given in comparison to
the results for (a) the basic wing, clean and with strake, Fig. 11.20, and (b) the wing
clean and with strake in high-lift configuration—(leading-edge) slats 25◦ down and
trailing-edge flaps (single slotted Fowler flaps) 30◦ down, Fig. 11.21.
Figure of merit is the ratio of “induced lift increment due to jet” to “blowing
coefficient” ΔC L /cμ , hence the capability to transform jet momentum into useful
aerodynamic lift.
Figure 11.20 gives results of blowing for the clean configuration: (a) upper curve:
wing only, (b) middle curve: wing and strake, blowing over the wing only, (c) lower
curve: wing and strake, blowing over the strake and the wing.
Evidently spanwise blowing always is more efficient on the clean basic wing
(upper curve) than on the strake-wing combination (middle curve). This was to be
expected. Splitting-up the total blowing momentum on the strake and the wing in
equal amounts, say combining the two optimum positions on the strake and the basic
wing is unfavorable, lower curve. In each case the absolute lift increment ΔC L goes
up with increasing blowing. For all cases shown, the efficiency of spanwise blowing
ΔC L /cμ is reduced with increasing cμ .
An analogous comparison, now for a high-lift configuration, is given in Fig. 11.21.
The upper curve gives the blowing efficiency for the basic wing with slat and Fowler
11.5 Spanwise Blowing 381

Fig. 11.21 High-lift configurations: blowing efficiency ΔC L /cμ |opt as function of the blowing
coefficient cμ [21]. Fowler flap δ f = 30◦ , slat δ N = −25◦ , nozzle diameter d = 15 mm

flap and spanwise blowing, the lower curve that for the same basic wing, but with a
strake added and blowing over that strake.
The two curves clearly demonstrate that spanwise blowing over the wing is most
efficient for the basic wing in high-lift condition, which means for the configuration
that is “farthest” away from spanwise concentrated vortex development. Blowing is
less efficient for the wing with strake. Its absolute lift coefficient ΔC L is always only
a little above 50 per cent of that of the wing without strake.

11.5.3 Summarizing Remarks

Spanwise blowing is a very interesting technique, seen from a fluid mechanical and
aerodynamical point of view.4 Why did it not find application in aircraft, in particular
fighter aircraft? The reasons for that are manifold. We do not reproduce the discussion
given in [21] regarding the aerodynamic, configurational and mission characteristics
opposing spanwise blowing, we only point out some aspects of the issue.
Figure 11.22 gives a comparison of the effects of wing geometry, high flap and
slat settings—see legend of Fig. 11.21—and spanwise blowing with respect to the
achieved maximum lift increment ΔC L max . The reference of the comparison is the
pilot model with the trapezoidal wing in clean configuration. The winner in any case
is the combination of trapezoidal wing and strake, not necessarily with blowing.
Spanwise blowing is bound to low flight-speed applications. This becomes evident
when examining Eq (11.6) in more detail. Remembering that the dynamic pressure
can be written as

4 Wenote that detailed investigations of the ensuing flow changes with discrete numerical methods
(Model 10 and 11 of Table 1.3) would be desirable, at least from an academic point of view.
382 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Fig. 11.22 In reference to the clean trapezoidal wing: effects of strake, high-lift system (slat and
Fowler flap) and spanwise blowing in terms of ΔC L max [21]. Left side: trapezoidal wing alone.
Right side: trapezoidal wing in combination with strake. The indicated angles of attack α are those
at maximum lift

1 1
q∞ = ∞ u 2∞ = γ p∞ M∞
2
, (11.7)
2 2
we find that the increment due to blowing decreases with increasing flight Mach
number.
Keeping in mind that pressure and mass-flux of the blowing devices are limited
by the aircraft’s engine characteristics, it is understandable that spanwise blowing
anyway is limited to low flight Mach numbers.
Another aspect is that when spanwise blowing was investigated, already the com-
bination low/medium aspect-ratio trapezoidal wing and strake was in full develop-
ment. It then found its application for instance with the F-16 and the F-18, Fig. 10.5.
Spanwise blowing, however, should be seen in the wider context of “jets in cross
flow”. The AGARD Symposium on “Computational and Experimental Assessment
of Jets in Cross Flow”, held from 19th to 22nd April 1993 in Winchester, United
Kingdom, had shown the breadth of jet-in-cross-flow applications [29]. Topics like
thrust vectoring, jet impingement, hypersonic cross flow, vehicle control, internal
flow, and turbine cooling were presented and discussed.
We would like to single out the hypersonic cross-flow jet topic. On the re-entry
trajectory of the Space Shuttle Orbiter reaction control devices were active in order to
11.5 Spanwise Blowing 383

control roll, pitch and yaw partly down to angles of attack α ≈ 20 to 25◦ at M∞  5,
when the aerodynamic stabilization, trim and control surfaces had left the hypersonic
shadow of the fuselage and the wing, see, e.g., [5].

11.6 Design Example

This section is devoted to a short presentation of a study regarding the pitch-up

problem of slender wings, i.e., delta-type wings, as it was discussed very detailed in
[2]. The study was based on an experimental investigation of a fighter configuration
with a double-delta wing. The approach was to carefully and comprehensively deter-
mine the connection of the observed pitch-up phenomena with the occurring lee-side
vortex structures. These basically multi-vortex systems were observed by means of
oil-flow pictures, thread tufts and vapor-screen photographs. Six-component force
measurements were made in order to determine the corresponding aerodynamic force
and moment coefficients.

11.6.1 Basic Configuration

The basic configuration of the supersonic light-weight fighter design (M∞ = 1.4)
was a (tailless) wing-body-vertical stabilizer configuration, similar to that of the F-
16XL. The width of the fuselage was approximately 15 per cent of the total span of
the model.
Starting point for the layout of the wing geometry was a plane, 3.4 per cent
thick double-delta wing with an inboard leading-edge sweep angle of ϕ0,i = 70◦ ,
an outboard sweep angle of ϕ0,a = 50◦ with a leading-edge kink at 69 per cent half
span, see the inserted wing plan view in Fig. 11.23. The sweep angle of the trailing
edge was slightly negative with ϕT E = −6.50◦ and constant along the wing’s span.
Design problems were found in particular for the subsonic flight regime—say M∞
< 0.6—mainly concerning an intolerable pitch-up tendency.
We show in Fig. 11.23 for the basic wing configuration the pitching moment in four
different angle-of-attack regimes together with the observed multi-vortex systems in
the Poincaré surfaces at x/L = 0.9.5
Four characteristic regimes of the development of the pitching moment Cm (α)
can be distinguished, with the assigned structures of the velocity-field topology to a
degree being arguable:

5 Note that at the lower (windward) side the primary attachment line always is assumed to be located

at the middle of the wing, see the discussion in Sects. 7.4.3 and 7.5. Note further that focus points
are nodal points and therefore denoted with N , instead of F, as for instance in Sect. 7.4.3.
384 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Fig. 11.23 The basic wing configuration at M∞ = 0.2 [2]. The pitching-moment behavior Cm (α)
and the flow-topological structures in the Poincaré surface at x/L = 0.9 (horizontal broken line in
the inserted wing plan view) at four angle-of-attack positions of the wing

– (1) A linear development regime with a constant negative dCm /dα.

Longitudinal stability is given for the angle-of-attack domain 0◦  α  10◦ . Above
the wing we have an open lee-side flow field, Sect. 7.4.3. Regarding the topology
of the velocity field we observe two primary vortices in the Poincaré surface, one
due to the inner delta wing with the focus point N1 , the other due to the outer delta
wing with the focus point N2 . Below these vortices we see a secondary vortex
each.
11.6 Design Example 385

– (2) Pitch up develops in the angle-of-attack domain 10◦  α  20◦ .

Above the wing still an open lee-side flow field exists. The vortex of the outer
delta wing moves faster up than that of the inner delta wing. Both are now coupled
via the free saddle point S. The outer vortex moves toward the inner and begins to
break up near the wing’s trailing edge (not indicated in the figure). Both vortices
then begin to rotate around the free saddle point, which in the experiment was
observed as wiggly line of the vortex centers (vapor-screen pictures and thread
tufts).
The summary effect of the vortex development moves the non-linear lift effect
forward, which leads to the pitch-up bend of the Cm curve.
– (3) The maximum instability present at α = 20◦ is preserved up to the
stall regime 20◦  α  30◦ .
Regarding the topology of the velocity field we observe first of all that now a
closed lee-side flow field exists. This means that the whole upper wing surface
now is dominated by the separated flow. The vortex of the outer wing part has lost
the contact to the leading edge. It almost completely has undergone breakdown.
Obviously a balance is present between the increase of nonlinear lift and the
forward movement of the center of pressure. The result is that pitching moment
still grows, but that the gradient dCm /dα is nearly constant.
– (4) Above α ≈ 28◦ the gradient dCm /dα begins to decrease and finally
a high degree of stability is recovered, which leads to pitch-down
beyond αmax .
With the approach to the maximum angle of attack αmax only one lee-side vortex
system—that from the inner delta part—is present. The secondary vortex below it
is barely observable in the oil-flow and the vapor-screen picture. Vortex breakdown
has moved far forward. The Poincaré surface actually does no more correspond
with the location x/L = 0.9, but with a much more forward lying location.
What happens, if the angle-of-attack is increased beyond αmax ? The flow pattern
approaches that past a vertically located flat plate. The resulting normal force slips
back into the center of the surface. The configuration becomes highly stable. Above
αmax  40◦ the danger of “deep stall” arises, if the aircraft—due to insufficient
control-surface function—restabilizes and a balanced flight attitude with Cm = 0
is established.

11.6.2 Modifications of the Wing Geometry

To solve, respectively reduce the inherent pitch-up problems, the effects of some
geometrical measures were investigated with wind-tunnel tests in the low-speed
domain. The most promising variants were then checked with a high-speed model
in the domain 0.5  M∞  2.0. The geometric measures are:
– (a) Change of the leading-edge sweep in the apex region, the main wing, the
leading-edge extension and the outboard wing.
386 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Fig. 11.24 Effect of the modifications of the wing planform on the pitch-up behavior [2]. Upper
part: wing planforms with (1) basis, (2) and (3) apex modification, (4) apex and leading-edge
modifications. Lower part: the effects of the modifications

– (b) Change of the outboard wing span.

– (c) Placing of a trigger-device, a “fence”, on the wing kink.
– (d) Variation of the anhedral of the outboard wing (downward) with νa = 0◦ /15◦
/30◦ /45◦ .

These modifications are shown in Fig. 11.24 (upper part) together with their effects
on the pitch-up behavior (lower part). To quantify the effects, the C L -Cm graph was
employed. In that way the location of the neutral point x N
xcog − x N dCm
= (11.8)
lμ dC L

can be used to show the effect. Here xcog is the location of the center of grav-
ity, and lμ the mean aerodynamic chord length. In order to compare the effect of
11.6 Design Example 387

the modifications, all configurations were put to the same measure of instability
(dCm /dC L )|C L →0 = 0.03 or 3 per cent.
The effects on the pitch-up tendency are the following:
– The modification of the inner delta wing—configuration (2)—yields a reduction
of the measure of the pitch-up tendency to 6 per cent, compared to 9 per cent
of the original configuration (1). The effect goes together with a weak unstable
break. That is the excursion from the nearly straight-line graph, marked with a fine
hatching.
– If the leading-edge sweep of the inner delta-wing modification of 55◦ is reduced to
48◦ —configuration (3)—the measure of the pitch-up tendency is further reduced
to 4.6 per cent.
– Configuration (4) is configuration (2), but with a small leading-edge extension
(LEX). That is located just ahead of the outer delta wing with a final sweep of
90◦ at aft the kink-point with the outward wing. It has the lowest measure of the
pitch-up tendency of the configurations with 3.8 per cent.
The result is that any of these modifications is helpful in reducing the basic (strong)
pitch-up tendency of 9 per cent. A trigger-device—a fence at the kink-point between
the outboard and the inboard wing—had a detrimental effect. It led to a too high
distortion of the inboard vortex system (not shown here).

11.6.3 The Final Configuration

The final configuration was found after introducing two further variations to con-
figuration (4), now of the outboard wing: 1. modifications of the anhedral angle νa ,
and/or 2. the leading-edge sweep ϕa
Figure 11.25 presents the effects of the two variations.
Changing the anhedral angle is very effective, left part of the figure. With νa = 30◦
the pitch-up tendency has almost disappeared, and fully with νa = 45◦ . The anhedral
angles lead to a stable vortex outboard of the kink, introduced by the spanwise
velocity components of the inboard leading-edge vortex system.
Also helpful for the longitudinal stability is a reduction of outboard leading-edge
sweep, right side of Fig. 11.25. This, however, is excluded due to supersonic design
aspects.
It should be noted that the anhedral reduces the effective angle of attack with its
increase by a quasi twist effect of the outboard wing. Reduction of the outboard wing
span/aspect ratio also reduces the proneness of pitch-up, but is not tolerable due to
performance aspects.
Finally—combining and extrapolating the effects from above—the wing, respec-
tively the aircraft configuration with the desired longitudinal stability properties was
found with Λ = 1.6, inner and middle leading-edge sweep 50◦ /67◦ plus leading-edge
extension with 50◦ sweep, the anhedral position at 69 per cent span and the anhedral
angle νa = 30◦ .
388 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

Fig. 11.25 Effect of anhedral-angle νa and leading-edge sweep ϕ0a modifications of the outer
wing on the pitch-up behavior of wing No. 4 from Fig. 11.24 [2]. Upper part: wing planforms and
anhedral angle modifications. Lower part: the effects of the modifications

Figure 11.26 shows the resulting pitching-moment behavior again in four different
angle-of-attack regimes in combination with the observed flow-topological structures
in the Poincaré surfaces at x/L = 0.9.
We look again at four characteristic trends of the development of the pitching
moment Cm (α) together with the structures of the velocity-field topology. The new
structures are noted now with a dashed line.

– (1’) A linear development regime with a constant negative dCm /dα.

Above the wing again an open lee-side flow field is present. From the kink of the
lowered outboard wing with the anhedral angle νa = 30◦ via flow-off separation
a vortex with the focus point N3 develops. Compared to the original secondary
11.6 Design Example 389

Fig. 11.26 The final wing: flow-topological structures in the Poincaré surface and pitching-moment
behavior Cm (α) at four angle-of-attack positions [2]

vortex—its focus point denoted N in (1) in Fig. 11.23—this vortex is stable and
with increasing angle of attack keeps its location above the outboard wing. In
the experiments it was observed that to achieve this, obviously a minimum size
is necessary of the anhedral angle of about νa = 15◦ . The overall topological
structures (1) and (1’) are the same. The numbers of free nodal points and saddle
points are the same.
– (2’) The linear development regime of the pitching moment persists.
The flow topology basically is the same as that in the regime (1’). The tendency
seen in (2) of the vortices N1 and N2 to rotate around each other has disappeared.
390 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

– (3’) Still linear development up to α ≈ 28◦ , but a weak pitch-up.

Regarding the topology of the velocity field we observe that still an open lee-side
flow field exists. The outer vortex has moved up, the inner toward the wing’s center.
The gradient dCm /dα is nearly constant, but shows a 0.2 per cent pitch-up.
– (4’) Above α ≈ 28◦ the gradient dCm /dα begins to drop strongly.
The overall flow topology with a closed lee-side flow field is almost the same as
observed in (4). The secondary vortex is not observable. Vortex breakdown again
begins around α L B = 27◦ .

Regarding the performance of this final configuration we note that the wing has
a more or less linear characteristic of the pitching moment with only a very small
pitch-up tendency. The maximum lift is slightly reduced with C L max = 1.3 at α =
32◦ , compared to C L max = 1.37 at α = 34◦ of the original wing.
The modification of the aerodynamic load distribution led at the same total load
to a reduction of the wing-root bending moment of about 12 per cent. Overall an
improvement of the flight performance is given. The potential of flight with artificially
controlled longitudinal stability is enlarged. Acceleration and steady and unsteady
turning performances are enhanced as well as the flight characteristics. For the full
discussion the reader is referred to [2].
The knowledge and the interpretation of the topologies of the velocity fields in
Poincaré surfaces revealed the vortical flow properties as the driving influences for
the configuration development. Figure 11.27 shows the photo of the high speed model
of a possible aircraft configuration.

Fig. 11.27 A possible aircraft configuration with the final wing design [2]
11.7 Problems 391

11.7 Problems

Problem 11.1 Consider a delta wing at angle of attack α < α L B with lee-side
vortices. When α L B is reached, vortex breakdown begins to happen at the rear of the
wing. Assume that it occurs symmetrically. Which aerodynamic properties of the
wing are being affected and how? What is the flow-physical effect?

Problem 11.2 Consider the divergence of the rolling moment Clβ and the yawing
moment Cn β . What are the basic dependencies on (a) the general flight conditions,
(b) the flow structure over the wing?

Problem 11.3 Give the reasons for the planform geometry found for the Concorde
(take-off → cruise) with respect to (a) flight performance and (b) stability and han-
dling qualities.

Problem 11.4 Compare Figs. 11.12 and 11.13 for the relative effects of (a) leading-
edge radius, (b) camber, and (c) nose flap for the configuration with the aspect ratio
Λ = 2, the flight Mach number M∞ = 0.3, and the leading-edge sweep angle ϕ0 =
63◦ .

Problem 11.5 What are the—positive—effects of concentrated spanwise blowing?

Which aerodynamic properties of an aircraft/wing are affected? What are the reasons
for its “non-application”?

Problem 11.6 Explain the positive effects of anhedral of the outboard wing
(Sect. 11.6) via consideration of the flow structure (topological aspects). Discuss
Figs. 11.23 and 11.26 .

References

1. Gursul, I., Wang, Z., Vardaki, E.: Review of flow control mechanisms of leading-edge vortices.
Prog. Aerosp. Sci. 43(3), 246–270 (2007)
2. Staudacher, W.: Die Beeinflussung von Vorderkantenwirbelsystemen schlanker Tragflügel (The
Manipulation of Leading-Edge Vortex Systems of Slender Wings). Doctoral Thesis, University
Stuttgart, Germany (1992)
3. Gottmann, Th., Groß, U., Staudacher, W.: Flügel kleiner Streckung, Teil 1: Grundsatzunter-
suchungen, Band 1: Analysebericht. MBB/LKE127/S/R/1563 (1985)
4. Shortal, J.A., Maggin, B.: Effect of Sweepback and Aspect Ratio on Longitudinal Stability
Characteristics of Wings at Low Speeds. NACA TN-1093 (1985)
5. Hirschel, E.H., Weiland, C.: Selected Aerothermodynamic Design Problems of Hypersonic
Flight Vehicles. Progress in Aeronautics and Astronautics, AIAA, Reston, VA, vol. 229.
Springer, Berlin, Heidelberg (2009)
6. Staudacher, W.: Zum Einfluss von Flugelgrundrissmodifikationen auf die aerodynamischen
Leistungen von Kampflugzeugen. Jahrestagung DGLR/OGFT, Innsbruck, Austria, DGLR Nr.
73–71, 24–28 (1973)
392 11 Selected Flow Problems of Small Aspect-Ratio Delta-Type Wings

7. Polhamus, E.C.: A Concept of the Vortex Lift of Sharp Edge Delta Wings, Based on a Leading-
Edge Suction Analogy. NASA TN D-3767 (1966)
8. Polhamus, E.C.: Application of the Leading-Edge Suction Analogy of Vortex Lift to the Drag
Due to Lift of Sharp-Edged Delta Wings. NASA TN D-4739 (1968)
9. Schlichting, H., Truckenbrodt, E.: Aerodynamik des Flugzeuges, vol. 1 and 2, Springer,
Berlin/Gättingen/Heidelberg, 1959, also: Aerodynamics of the Aeroplane, 2nd edn. (revised).
McGraw Hill Higher Education, New York (1979)
10. Staudacher, W.: Abschätzung des Reynolds-Zahl-Einflusses auf den induzierten Widerstand
schlanker Flügel. MBB-UFE122-Aero-Mt-399 (1980)
11. Henderson, W.P.: Effects of Wing Leading Egde Radius and Reynoldy Number on Longitudinal
Aerodynamic Characteristics of Highly Swept Wing-Body Configurations at Subsonic Speeds.
NASA TN D-8361 (1976)
12. Nangia, R.K., Miller, A.S.: Vortex flow dilemmas and control of wing planforms for high
speeds. In: Proceedings RTO AVT Symposium on Vortex Flow and High Angle of Attack
Aerodynamics. Loen, Norway, May 7 to 11, 2001. RTO-MP-069, Paper Nr. 9 (2002)
13. Brandon, J.M., Hallissy, J.B., Brown, P.W., Lamar, J.E.: In-flight flow visualization results of
the F-106B with a vortex flap. In: Proceedings RTO AVT Symposium on Vortex Flow and High
Angle of Attack Aerodynamics. Loen, Norway, May 7 to 11, 2001. RTO-MP-043, Paper Nr.
43 (2002)
14. Frink, N.T., Huffman, J.K., Johnson Jr., T.D.: Vortex Flap Flow Reattachment Line and Subsonic
Longitudinal Aerodynamic Data on 50◦ to 74◦ Delta Wings on Common Fuslage. NASA TM
84618 (1983)
15. Hitzel, S.M., Osterhuber, R.: Enhanced maneuverability of a delta-canard combat aircraft by
vortex flow control. J. Aircraft 55(3), 1–13 (2017)
16. Erickson, G.E., Gilbert, W.P.: Experimental Investigation of Forebody and Wing Leading-Edge
Vortex Interactions at High Angles of Attack. AGARD-CP-342, 11-1–11-28 (1983)
17. Malcolm, G.N., Skow, A.M.: Enhanced Controllability Through Vortex Manipulation on
Fighter Aircraft at High Angles of Attack. AIAA-Paper 86–277 (1986)
18. Breitsamter, C.: Strake effects on the turbulent fin flowfield of a high-performance fighter
aircraft. In: W. Nitsche, H.-J. Heinemann, R. Hilbig (eds.), New Results in Numerical and
Experimental Fluid Mechanics II. Contributions to the 11th AG STAB/DGLR Symposium
Berlin, Germany 1998. Notes on Numerical Fluid Mechanics, vol. 72, pp. 69–76. Vieweg-
Verlag, Braunchweig, Wiesbaden (1999)
19. Cornish, J.J.: High Lift Applications of Spanwise Blowing. 7th ICAS Congress, ICAS Paper
70-09 (1970)
20. Dixon, C.J.: Lift and Control Augmentation by Spanwise Blowing Over Trailing Edge Flaps
and Control Surfaces. AIAA-Paper 72–781 (1972)
21. Staudacher, W.: Effects, Limits and Limitations of Spanwise Blowing. AGARD-CP-534, 26-
1–26-10 (1993)
22. Staudacher, W.: Flügel mit kontrollierter Abläsung. DGLR Nr. 77–028, 24–28 (1977)
23. Staudacher, W., Egle, S., Boddener, W., Wulf, R.: Grundsätzliche Untersuchungen über span-
nweitiges Blasen und stabilisierten Wirbelauftrieb. MBB, Ottobrunn, I.B. 77-125/UFE 1320
(1977)
24. Staudacher, W.: Interference Effects of Concentrated Blowing and Vortices on a Typical Fighter
Configuration. AGARD-CP-285, 19-1–19-13 (1980)
25. Staudacher, W.: Influence of Jet-Location on the Efficiency of Spanwise Blowing. 12th ICAS
Congress, ICAS Paper 13-02 (1980)
26. Staudacher, W.: Verbesserung der aerodynamischen Leistungen durch konzentriertes Aus-
blasen. MBB, Ottobrunn, FE 122 /S/R 1499 (1980)
27. Gottmann, Th., Hünecke, K., Staudacher, W.: Einfluss des Spannweitigen Blasens auf die
Aerodynamischen Leistungen. Handbuch der Luftfahrttechnik (LTH) (1986)
28. Huffman, J.K., Hahne, D.E., Johnson Jr., T.D.: Aerodynamic effect of distributed spanwise
blowing on a fighter configuration. J. Aircraft 24(10), 673–679 (1987)
29. N.N.: Computational and experimental assessment of jets in cross flow. In: Proceedings of the
AGARD Symmposium, Winchester, UK, April 19–22, 1993. AGARD-CP-534 (1993)
Chapter 12
Solutions of the Problems

Useful relations are given in Appendix A and and constants and atmospheric data in
Appendix B. Perfect gas is assumed throughout with γ = 1.4.

12.1 Problems of Chapter 2

Problem 2.1
The speed of sound is a∞ = 299.5 m/s, the flight speed u ∞ = 149.8 m/s at M∞ = 0.5
and u ∞ = 239,6 m/s at M∞ = 0.8. The unit Reynolds numbers are Re∞ u
= 4.247·106
for M∞ = 0.5 and Re∞ = 6.796·10 for M∞ = 0.8. For the viscosity we use the
u 6

power-law approximation with ω = 0.65.

At the location x/c = 0.5 we obtain for M∞ = 0.5 the wall shear stress τw = 0.94
N m−2 and the displacement thickness δ1 = 0.0013 m, and for M∞ = 0.8 τw = 1.89
N m−2 and δ1 = 0.0011 m.
At the location x/c = 1 for M∞ = 0.5 the results are τw = 0.67 N m−2 and δ1 =
0.0019 m. and at M∞ = 0.8 τw = 1.33 N m−2 and δ1 = 0.0016 m.
Problem 2.2
For turbulent flow we find at the location x/c = 0.5 for M∞ = 0.5 τw = 10.2 N m−2
and δ1 = 0.0046 m, and for M∞ = 0.8 τw = 23.8 N m−2 and δ1 = 0.0042 m.
At the location x/c = 1 for M∞ = 0.5 the results are τw = 8.91 N m−2 and δ1 =
0.0079 m. and at M∞ = 0.8 τw = 20.8 N m−2 and δ1 = 0.0074 m.
Problem 2.3
For laminar flow we find at the location x/c = 0.5 for M∞ = 0.5 the wall shear stress
τw = 0.877 N m−2 and the displacement thickness δ1 = 0.0018 m, and for M∞ =
0.8 τw = 1.75 N m−2 and δ1 = 0.0015 m.
At the location x/c = 1 for M∞ = 0.5 the results are τw = 0.62 N m−2 and δ1 =
0.0026 m. and at M∞ = 0.8 τw = 1.24 N m−2 and δ1 = 0.0022 m.
© Springer-Verlag GmbH Germany, part of Springer Nature 2021 393
E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3_12
394 12 Solutions of the Problems

Problem 2.4
For turbulent flow we find at the location x/c = 0.5 for M∞ = 0.5 τw = 8.03 N m−2
and δ1 = 0.0052 m, and for M∞ = 0.8 τw = 18.1 N m−2 and δ1 = 0.0048 m.
At the location x/c = 1 for M∞ = 0.5 the results are τw = 6.99 N m−2 and δ1 =
0.009 m. and at M∞ = 0.8 τw = 15.7 N m−2 and δ1 = 0.0084 m.
Problem 2.5
The results at once show that the skin friction decreases with increasing x, whereas the
displacement thickness increases. This of course can directly be seen when looking
at the equations. The Mach number does not have directly an effect. The effect is
due to the fact that the Reynolds number increases with increasing Mach number.
Important is that with increasing wall temperature a strong decrease of the skin
friction happens, but only for the turbulent case. This is a typical thermal-surface
effect, which is of special importance with high supersonic and hypersonic flow
[1]. Regarding the total drag of, for instance, an airfoil, the skin-friction drag can
be reduced with an increase of the wall temperature. But because the displacement
thickness increases, the viscous-effects induced pressure drag, i.e., the form drag,
increases too. Hence one has to look at the overall effect of a temperature increase.
Problem 2.6
At H = 10 km altitude we have the speed of sound a∞ = 299.5 m/s and the density
ρ∞ = 0.4135 kg/m3 . The flight velocities are u ∞ = 59.9 m/s, 119,8 m/s, 239,6 m/s
and the drag at these velocities D = 4.45 N, 17.8 N, 71.2 N. The drag increases
quadratically with the flight speed.
Problem 2.7
The Prandtl–Glauert rule reads
Clic
Clc =  .
1 − M∞
2

The incompressible lift coefficient results to Clic = 0.1908, and the compressible
coefficients are Clc = 0.194, 0.208, 0.31.
Problem 2.8
L = 143,9 N, 617.3 N, 3,680.3 N.
Problem 2.9
To arrive at actual forces, the force coefficients have to be multiplied with the dynamic
pressure q∞ and the reference area Ar e f . The forces hence depend on the square of the
flight velocity and the density at the flight altitude. The problems reflect the situation
in the compressible flight domain below the critical Mach number. There the drag
coefficient of an airfoil or wing is more or less independent of the Mach number and
the lift coefficient changes with the Mach number according to the Prandtl–Glauert
rule.
12.2 Problems of Chapter 3 395

12.2 Problems of Chapter 3

Problem 3.1
(a) Model 4, circulation theory. (b) The circulation Γ0 is constant on all four legs.
(c) Helmholtz’s theorems. (d) The two-dimensional airfoil is a wing of infinite span.
Hence the trailing vortices are located at y = ±∞. (e) The situation is the same as
at the airfoil.
Problem 3.2
Steady level flight: lift = weight. For the flight altitude of H = 10 km, the atmosphere
data (Table B.2) are: p∞ = 26,500 Pa, ρ∞ = 0.4135 kg/m 3 . Hence, at a Mach number
of M∞ = 0.82 the flight velocity is u ∞ = 245.6 m/s (with γ = 1.4).
Referring to force equilibrium, the lift coefficient can be calculated:
L = m g. C L = 2mg/(ρ∞ u ∞ b2 ) = 0.45.
Using Eq. (3.40), and taking into account the spanwise load factor for an elliptical
circulation distribution of s = π/4, the induced downward velocity w0 is (it can be
also calculated directly inserting the expression for C L ):
w0 = 1.59 m/s.
The vertical distance Δz of the wake vortex downward movement for a flight path
length of Δx = 50 km then is (Δx = u ∞ Δt, Δz = w0 Δt):
Δz = w0 /u ∞ Δx = 324 m.
Problem 3.3
Applying Eq. (3.38), the root circulation Γ0 is obtained to Γ0 = 475.34 m 2 /s.
The expressions for the tangential (circumferential) velocity vθ related to the
Lamb-Oseen vortex model and the Burnham-Hallock vortex model are given in Eqs.
(3.17) and (3.21).
At the core radius r = rc we find then for the first one vθ,max = 30,1 m/s (β =
1.256), and for the second one vθ,max = 20,0 m/s.
The difference reflected by the ratio v L O /v B H = 2 (1 - e−β ) = 1.43 is attributed to
the approach of characterizing the vortex core structure: the Lamb-Oseen model as
an analytical solution accounting for a temporal vortex core growth due to viscosity,
whereas the Burnham-Hallock model relaxes the peak velocity at the viscous core
by quadratic terms.
Problem 3.4
Applying the Kutta–Joukowsky relation with the lift L is the same for the spanwise
circulation distribution Γ (y) and a distribution with spanwise constant Γ0 we obtain
via Eq. (3.34) with L = ρ∞ u ∞ Γ0 b0 :
 +b/2  +b/2  
1 1 2y 2
L = ρ∞ u ∞ Γ (y)dy = ρ∞ u ∞ Γ0 1 − ( ) dy
b −b/2 b −b/2 b

and
396 12 Solutions of the Problems

Fig. 12.1 Biconvex circular

arc airfoil 10 per cent thick

 +b/2    +b/2  
1 2y 2 4 y3 1 2
b0 = 1 − ( ) dy = y − 2 =b 1− = b,
b −b/2 b b 3 −b/2 3 3

and finally
b0 2
s= = .
b 3
The lateral distance b0 of the rolled-up wake vortex system or the center of the free
circulation centroid, respectively, is located at 66 per cent of the wing span b. Con-
sequently, the distance between the trailing vortex trajectories is smaller compared,
for instance, to that related to an elliptical circulation distribution (b0 = 0.785 b, Eq.
(3.37)). The parabolic circulation distribution also reflects a more inboard loading.
The reduced lateral distance may result in an earlier vortex trajectory contact caused
by the Crow instability when progressing downstream and, therefore, an earlier wake
vortex decay (Fig. 12.1).
Problem 3.5
As indicated in the sketch, the radius for a circle intersecting the chord line in x =
±0.5 and the z-axis in z = 0.05 is

R 2 = 0.52 + (R − 0.05)2 ; R = 2.525.

The equations for the circular contour and its slope are

x 2 + (2.525 − 0.05 + z)2 = 2.5252 , z = −2.475 + 6.38 − x 2 ,
dz x
= −√ .
dx 6.38 − x 2

The vertical velocity in the first collocation point w1 consists of contributions

from all three source distributions

w1 = w11 + w12 + w13 .

The boundary conditions

12.2 Problems of Chapter 3 397

dz
wi = u ∞ ( )i
dx
gives the source intensities

w1 = σ1 /2; σ1 /u ∞ = 2(dz/d x)1 = 2 ∗ 0.1331 = 0.2662.

w2 = 0; σ2 /u ∞ = 0.
w3 = σ3 /2; σ3 /u ∞ = −0.2662.

From the known source intensities the longitudinal disturbance velocities can be
calculated:

u 11 = 0; collocation point is in the middle of source distribution.

u 12 = 0; second source distribution has zero intensity.
σ3
u 13 = 2π ln| 1/6−2/3
1/6−1
| = 0.0216 u ∞ .

In small disturbance theory c p = −2 u/u ∞ , hence

c p = −0.0432.

Problem 3.6
The lift force may be written alternatively as

L = ρu ∞ Γ
1 1
L = ρu 2∞ C L c = ρu 2∞ 2παc.
2 2
Thus
Γ = u ∞ παc.

Problem 3.7
The vertical velocities in the collocation points are:

1 1
w1 = u ∞ α − Γ1 + Γ2 ,
2πc/2 2π(c/2 + c)
1 1
w2 = u ∞ α − Γ 1 − Γ2 ,
2π(3c/2 + c) 2πc/2

w should be zero.
Substitute u ∞ απc with A:
398 12 Solutions of the Problems

1
A = Γ1 − Γ2 ,
1 + 2
1
A= Γ1 + Γ2 .
3 + 2

The solution is

3 + 2
Γ1 = A ,
2(1 + )
1 + 2
Γ2 = A .
2(1 + )

And because the lift force is L = ρu ∞ Γ , then

1 3 + 2
L 1 /(1/2ρ∞ u 2∞ 2παc) = .
2 1+
1 1 + 2
L 2 /(1/2ρ∞ u 2∞ 2παc) = .
2 1+

The lift on a single airfoil would be

L ∞ = 1/2ρu 2∞ 2πα, and one may write

1 3 + 2
L 1 /L ∞ = ,
2 1+
1 1 + 2
L 2 /L ∞ = .
2 1+

Problem 3.8
The speed of sound is a∞ = 299.5 m/s, the flight speed u ∞ = 239.6 m/s at M∞ = 0.8.
The results are: lift coefficient C L = 0.413, induced drag C Di = 0.00861, total drag
C D = 0.0258, lift-to-drag ratio L/D = 16.

12.3 Problems of Chapter 4

Problem 4.1
With the relations from Sect. A.5.4 we find for the laminar case:
x = 2.5 m, δ/x = 0.0015, x = 5 m, δ/x = 0.0011,
and for the turbulent case:
x = 2.5 m, δ/x = 0.015, x = 5 m, δ/x = 0.013.
√
Even if these data do not scale with the 1/ Re assumption in Sect. 4.1, the result
is acceptable.
12.3 Problems of Chapter 4 399

Problem 4.2
The vorticity-content integral over the hatched area is
 2π  re  2π  re  
1 d(r vθ )
ωz r dθ dr = r dθ dr = 2πre vθe .
0 0 0 0 r dr

Problem 4.3
The terms suction side and pressure side are colloquial. Suction side means a negative
pressure coefficient c p is present, pressure side a positive one, compare Fig. 2.3. There
the better (?) terms upper side and lower side are used. The lift is due to the pressure
difference between the lower and the upper side.
Problem 4.4
A vortex line is defined by v × ω = 0. We make the proof with the vorticity-content
vector Ω. We obtain with v = [u, 0, 0] and Ω = [Ωs , 0, 0]

⎡ ⎤
es et e z
⎣ u 0 0 ⎦ = es (0 + 0) + et (0 + 0) + e z (0 + 0) = 0.
Ωs 0 0

Problem 4.5
Circulation theory adds to the translative free stream a potential (Rankine) vortex
with such a circulation Γ that the combined flow field obeys the Kutta condition,
i.e., that the flow at the trailing edge does not turn around to the upper side of the
airfoil, Fig. 4.14, but smoothly leaves the trailing edge, Fig. 4.15,
Problem 4.6
For y = 0 kinematically active vorticity content leaves the wing’s trailing edge in
the trailing vortex layer, thus reducing Γ . After the rolling-up of the trailing vortex
layer the vorticity is concentrated in the trailing vortices and Γ0 is present again. The
condition is the compatibility condition, presented in Sect. 4.4.
Problem 4.7
The trailing vortex layer must be sketched as an rectangle. It carries only an kine-
matically inactive vorticity content, coming from the boundary layers at the upper
and the lower side of the wing.
Problem 4.8
The finite extent shows the strength of the trailing vortex layer, which, however, at
that location carries only kinematically inactive vorticity.
400 12 Solutions of the Problems

12.4 Problems of Chapter 5

Problem 5.1
The answer lies with the integral Eq. 4.6 in Sect. 4.1. Only the velocities at the upper
and the lower bound are counting, not the shape of u(z).
Problem 5.2
Model 8 approaches permit, after a suitable discretization of the computation domain,
a fast computation of important aerodynamic properties of the wing or flight vehicle.
Compared to Model 4 approaches (today usually panel methods), compressibility
effects are included. However, viscous effects, including ordinary separation can
only be described with Model 9 and 10 methods. Today these are mainly RANS or
URANS methods. Massive separation can only be treated with scale-resolving meth-
ods (Model 11). Generally turbulence modeling is a major topic, laminar-transition
can be a big problem.
Problem 5.3
The computation process basically has the following steps: (1) parametrization of
the CAD body surface, (2) discretization of the body surface, (3) discretization of the
computation domain. Step 2 and 3 in particular can be very cost-driving, depending
on the chosen Model approach.
Model 4 approaches in the form of panel methods demand only step 1 and 2,
all other approaches also step 3. Shock waves appearing in the flow field in Model
8 to 11 approaches are “captured” today. This makes necessary a sufficiently fine
discretization of the computation domain in the vicinity of a shock wave. That also
holds for the body surface, if a shock wave originates there. Boundary layers and
strong interaction domains in Model 8 to 11 approaches in particular need a very
fine discretization on the body surface and in the respective flow domain, which is
not the rule for trailing vortex layers and vortices.
All this basically holds for rigid bodies. If flow fields past non-rigid bodies, includ-
ing movables (slats, flaps, others), are to be computed, the discretization must follow
the characteristic time scales. In the end all this calls for self-adapting discretization
methods, which still are not fully adopted.
The computation costs rise with the Model approach level. Lowest are the costs
for panel methods. Statistical turbulence models in RANS/URANS methods can be
nasty cost drivers, in particular also all scale-resolving approaches.
In aircraft design work all this is to be considered in comparison to wind-tunnel
measurements, too. There design, manufacturing and finally the instrumentation of
the wind-tunnel model are very time and cost consuming. This also is the reason
why the computational approaches are progressing, which of course mainly is due
to the still increasing computer power, storage capacities, and not least the progress
in solution-algorithm development.
Problem 5.4
Case 1 for the two-dimensional, and case 2 for the three-dimensional wake. The
function u(z) contributes to the entropy rise, too.
12.4 Problems of Chapter 5 401

Problem 5.5
Only a proper grid resolution yields the necessary resolution of the characteristic flow
phenomena. This holds for Model 4 and 8 to 11 approaches, where the flow field as
such, boundary layers, shock waves, interaction domains, trailing vortex layers and
vortices need to be described to the demanded degree of accuracy.

12.5 Problems of Chapter 6

Problem 6.1
Boundary-layer decambering at a lifting airfoil is due to the different boundary-layer
properties at the suction side and the pressure side at the trailing edge of the airfoil.
At the suction side due to the larger loading of the boundary layer with an adverse
pressure gradient the displacement thickness is larger than at the pressure side. This
effect is present in both the subcritical and the supercritical case.
In the supercritical case an additional effect is present. Consider the case with a
supersonic flow pocket at the suction side only. The terminating shock wave impinges
vertically on the airfoil’s surface. Across the shock wave a total-pressure loss occurs.
Hence at the trailing edge the flow momentum at the suction side—with the shock
wave—is smaller than at the pressure side—without a shock wave. This leads at
the trailing edge of the airfoil to an upward deflection of the flow—shock-wave
decambering—which together with the boundary-layer decambering leads to a reduc-
tion of the airfoil’s lift.
Problem 6.2
From Table B.2 we find T∞ = 281.651 K, ρ∞ = 1.112 kg/m3 , μ∞ = 1.758·10−5 N s
m−2 . The speed of sound is a∞ = 336.438 m/s, and the flight speed u ∞ = 33.64 m/s.
With the thickness of the trailing edge h = 0.01 m, we obtain the Re = 21,280.5.
Hence the Strouhal number is Sr ≈ 0.2 and the shedding frequency f ≈ 673 Hz. The
shedding frequency hence is not at the upper bound of the human hearing domain,
which is f ≈ 16,000–20,000 Hz. The lower bound is f ≈ 16–21 Hz.
Problem 6.3
The higher the angle of attack, the higher is the actual pressure gradient, because the
relevant freestream component increases with the angle of attack.
Problem 6.4
p − p∞
cp = .
q∞

Vacuum means p = 0, hence we get with the squared speed of sound a 2 = γ R T∞

− p∞ −2 p∞ −2 γ ρ∞ R T∞ −2
cp = = = = .
q∞ ρ∞ u ∞
2 γ ρ∞ u ∞
2 γ M∞
2
402 12 Solutions of the Problems

Fig. 12.2 Equivorticity lines of a wing with (a) no-slip boundary conditions and (b) with perfect-
slip boundary conditions

Problem 6.5
The leading edge of this wing is aerodynamically sharp and sheds a vortex sheet in
Fig. 12.2b. However in Fig. 12.2a there is an interaction between the vortex and the
boundary layer, which doesn’t occur in Fig. 12.2b.
Problem 6.6
At higher Reynolds number the primary vortex induces a higher velocity near the
surface of the cylinder that interacts and retards the motion in the boundary layer
causing separation and the creation of the pair of secondary vortices.
Problem 6.7
See the discussion in Sect. 3.4.2 concerning Fig. 3.4.

12.6 Problems of Chapter 7

Problem 7.1
Such lines are found in Figs. 8.27, 8.31 and 8.32 on page 205 ff., with a detailed
presentation in Fig. 8.33. In Fig. 8.27 the open-type beginning of the attachment
line on a wing’s leading edge is shown, in Fig. 8.31 the open-type ending of such
an attachment line, and in Fig. 8.33 the open-type beginning and also ending of
separation and attachment lines.
Problem 7.2

For the flow field at the left Rule 2 holds with four half saddle points S and one focus
point N : 1 − 0.5 · 4 = −1. For the flow field at the right the originator of the figure
proposes to consider the end of the separation bubble shifted to infinity downstream.
Then again Rule 2 holds. Make a sketch for yourself.
Problem 7.3
We find an half-saddle each at the lower and the upper symmetry line. Rule 2 is
fulfilled.
Problem 7.4
The resulting eigenvalues are
12.6 Problems of Chapter 7 403

   2
1 ∂τwx ∂τwz 1 ∂τwx ∂τwz ∂τwx ∂τwz
λ1,3 = + ± − +4 ,
2 ∂x ∂z 2 ∂x ∂z ∂z ∂x

1 ∂p
λ2 = .
2 ∂y

Compare with Eqs. (7.7) and (7.8).

Problem 7.5
The prerequisite for plane-of-symmetry flow is that the flow occurs along a geodesic.
The geodesic is defined as a curve, whose tangent vectors always remain parallel to
it, if they are conveyed along it, see, e.g., [2]. It defines the shortest path between
two points on a surface.
Problem 7.6
On the convex side of the attachment line the approaching streamlines change from
concave to convex, hence inflection points are present.
Problem 7.7
Off-surface flow portraits are the patterns of the streamlines, which leave the surface,
which always happens only in singular points.
Problem 7.8
Write down the summary and compare with Sect. 7.3.
Problem 7.9
In two-dimensional flow the wall shear-stress changes sign. In three-dimensional
flow this does not happen. Instead one observes a local convergence of skin-friction
lines, the occurrence of a |τw |-minimum line and the bulging of the boundary-layer
thickness and the displacement thickness.

12.7 Problems of Chapter 8

Problem 8.1
The normal angle of attack results to α N = 11.93◦ and the normal leading-edge Mach
number to M N = 0.179. The correlations show that at the round leading edge the flow
is attached, but if the leading edge would be sharp, a lee-side vortex system could
develop.
Problem 8.2
The normal angle of attack results to α N = 4.88◦ and the normal leading-edge Mach
number to M N = 0.205. The correlations show that at the round leading edge the flow
is attached. For the sharp leading edge the correlation lies on the boundary between
“attached flow” and “development of the vortex system”.
404 12 Solutions of the Problems

Problem 8.3
The application of the rule yields for the linear theory a C L ic = 0.315015, and for M∞
= 0.3 the value C L c = 0.33. This compares well with the result of the Euler simulation
with C L c = 0.342.
Problem 8.4
The aspect ratio of the wing is  = 9. In the ideal case with Eq. (8.1) the induced
drag coefficient results to C Di = 0.007 or 28 per cent of the total drag.
With an Oswald factor of e = 0.8, the result is C Di = 0.0087 or 35 per cent of the
total drag.
Problem 8.5
In Sect. 3.16 we find
Γ
σ= .
u ∞ b/2

With the Mach number M∞ and the free-stream temperature T∞ the free-stream
speed is found: u ∞ = 300.47 m/s. For the three root circulations we obtain Γ0 =
538.50 m2 /s, 615.30 m2 /s, 600.29 m2 /s.
With the Sutherland equation, Appendix A.5.1, the free-stream viscosity is found
to be μ∞ = 1.899 · 10−5 N s m−2 , hence ρ∞ = 0.0451 kg/m3 .
With Eq. (3.46) the lift coefficients for the three root-result circulations result to
C L = 0.431, 0.493, 0.481, and finally C Di = 0.0082, 0.0107, 0.010.
The agreements with the lift coefficient on page 200 and the induced-drag coef-
ficient from Problem 8.4 are reasonable.
Problem 8.6
The dimensionless circulation G(x ∗ ) = Γ (x ∗ )/Γ0 for the case without the horizontal
tail surface is G(x ∗ ) ≈ 1, and with it G(x ∗ ) ≈ 0.85. The latter amounts to a loss of
about 15 per cent for Γ0 .
The lift is affected according to Eq. (3.46) and the induced drag to Eq. (3.48).
Problem 8.7
(a) With Eq. (3.34) we find with ρ∞ = 0.4135 kg/m3 and b0 = (π/4)b the root
circulation

L m2
Γ0 = = 566.2 .
ρ∞ u ∞ b0 s

(b) For the smaller wing the root circulation results to Γ0 = 1,132.4 m2 /s.
(c) The wing loading for large transport aircraft is Ws = 3,000 to 8,000 N/m2 .
Here it is in the first case Ws = 6,130 N/m2 and in the second case Ws = 12,262 N/m2 ,
which is well above the average values.
(d) If the same mass of an aircraft is transported with a smaller wing, the strength
of the trailing vortex increases.
12.7 Problems of Chapter 8 405

(e) The lift coefficient is two times, and the induced drag coefficient four times
higher for the smaller wing. Compute the values for yourself.
Problem 8.8
The lifting-line theory in Appendix A.4, gives the incompressible flow lift coefficient
for an angle of attack of α: C L = 2πα/(1 + 2/), where  = b2 /A is the aspect
ratio, b the span and A the wing area. According to the Prandtl–Glauert rule for
three dimensional flow, the y and z coordinates of the real configuration are to be
multiplied with β to obtain the analogue configuration for incompressible flow,

β = sqr t (1 − M∞
2
) = 0.6,

b02 β 2 b0.8
2
0 = = caverage = β0.8 = 0.6 · 10 = 6.
S0 b0.8

This scaling also applies to the wing incidence, then α0 = βα0.8 :

C L0 = (dC L /dα)0 · α0 = 2π/(1 + 2/6) · βα0.8 = 2.828α0.8

In the transformation back to the real M∞ = 0.8 flow, all incompressible flow
pressures should be divided by β 2 , and consequently we find

C L0.8 = C L0 /β 2 = 2.828 α0.8 /0.36 = 7.856 α0.8

and the lift curve slope becomes

(dC L /dα)0.8 = 7.856.

12.8 Problems of Chapter 9

Problem 9.1
The limiting flow phenomena are the drag divergence in the transonic flight regime
and the associated lift drop. The drag divergence is due to boundary-layer/shock-
wave interaction, which in the end leads to the transonic dip of the lift and to the
transonic buffet phenomenon, due to the high unsteadiness of the separated wing
flow.
Problem 9.2
The configurational means for the wing are the wing sweep and the supercritical
airfoil.
The wing’s sweep usually is backward, but can also be forward. If the wing is
swept, the free-stream “is seeing” a wing with smaller thickness to chord-length
ratio, than it is seeing at the unswept wing. Hence the overspeed is lesser and the
406 12 Solutions of the Problems

occurrence of the supersonic flow pocket with its terminating shock wave—usually
considered to be present, but not necessarily, only at the suction side of the wing—is
shifted to a higher flight Mach number.
The supercritical airfoil, or chord section, has the same effect due to its shape,
which mainly is a flattened upper surface, and in addition a highly cambered aft
section.
For the whole aircraft the means is the area rule, which concerns the overall
axial cross-section distribution of fuselage, wing, engine nacelles, empennage. The
distribution in axial direction should be as smooth as possible in order to minimize
interference drag.
While the swept wing easily is discernible, the supercritical airfoil and the area
ruling in a sense are “hidden” to the observer.
Problem 9.3
The vortex generator basically is transporting fluid with high momentum from the
upper boundary-layer regime and the adjacent inviscid external flow into the wall-
near part of the boundary layer. The effect then is an increase of the flow momentum
of the wall-near boundary-layer flow, and hence a reduction of separation proneness.
The vortex generator in principle also works in laminar flow. But if an adverse
pressure gradient is present, which is the cause of separation proneness, the flow
anyway will undergo laminar-turbulent transition.
Problem 9.4
In general not. The reason is that the laminar boundary layer has a less full flow
profile than the turbulent one. Hence less flow momentum can be transported into
the wall-near part of the boundary layer.
Problem 9.5
Without a wing-root/fuselage fairing the flow coming from the forward part of the
fuselage meets the wing’s root, which is a flow obstacle, at a swept wing a swept
one. Ahead of that obstacle three-dimensional boundary-layer separation happens.
The result is a horse-shoe like vortex around the wing-root/fuselage location. This
vortex is the cause of a drag increment and further it can lead to buffet at high angles
of attack.
A proper wing-root/fuselage fairing leads to a smooth flow passage over the wing’s
root and hence avoids separation and hence the horse-shoe vortex with its adverse
side effects.
Problem 9.6
The Breguet range equation is used to calculate the relative saving in fuel mass. The
formula is given by Eq. (9.1). The mass m T O W related to the take-off weight refers
to the sum of the masses due to empty weight m e , fuel m F and payload m P : m T O W
= m e + m F + m P . The range R is kept constant and the flight speed u ∞ , the specific
fuel consumption b, the lift coefficient C L and the take-off weight m T O W remain the
same:
12.8 Problems of Chapter 9 407

Fig. 12.3 The resulting

spanwise circulation
distribution Γ (y)

 
CL u∞ mT OW
R= ln = const.
CD b g mT OW − m F

The drag coefficient is composed as C D1 = C D01 + C Di1 = 0.8 C D1 + 0.2 C D1 , the

index 1 denoting the baseline case.
The case with reduced drag level is (index 2) C D2 = C D0 + C Di2 = 0.8o × 0.97
C D1 + 0.2 × 0.98 C D1 = 0.972 C D1 .
For constant range R it follows:
   
CL u∞ mT OW CL u∞ mT OW
ln = ln .
C D1 bg m T O W − m F1 C D2 bg m T O W − m F2
   
mT OW 1 mT OW
ln = ln .
m T O W − m F1 0.972 m T O W − m F2

 0.972  
m T O W − m F1 m T O W − m F2
= .
mT OW mT OW
 0.972 
m F1
m F2 = m T O W 1− 1− = 110, 483 kg.
mT OW

The relative fuel saving results in Δm F = −2.23 per cent.

Problem 9.7
According to Prandtl’s lifting line model a spanwise circulation distribution Γ (y)
providing a minimum induced drag is of elliptical shape (blue colored area) derived
from a modeling of horseshoe vortices for a planar wing surface neglecting wake vor-
tex roll-up. Taking this elliptical circulation distribution as an optimum with respect
to induced drag, vertically orientated winglets have to be loaded in an equivalent way
as it would be reflected by the virtual increase in span (orange and red areas related
to the winglet case; blue and orange colored area are of same magnitude for same
lift level). “Unloaded” winglets do not provide substantial effects on induced drag
reduction (Fig. 12.3).
408 12 Solutions of the Problems

Problem 9.8
Referring to Fig. 9.22 and Eq. (9.3), an increase in effective span may be realized by
wing-tip devices independent of upward or downward orientation. From a combina-
tion of both the elastic wing deformation due to highly loaded wing-tip devices can
be adjusted in such a way that deformations are limited or adjusted in a desired way
for further increase of aerodynamic efficiency or reduction of structural dynamic
loads.
Problem 9.9
According to Chap. 9.6 the characteristic reduced frequency attributed to the Crow
type instability is kCr ow ≈ 0.08. Here, the reduced frequency is defined as k =
f (b/2)/u ∞ . Thus, the full scale case results in an oscillation frequency of f =
kCr ow u ∞ /(b/2) = 0.21 Hz. The oscillation frequency for the wind tunnel situation
then f = kCr ow u ∞ /(b/2) = 1.67 Hz.

12.9 Problems of Chapter 10

Problem 10.1
(a) More or less rectangular wings are found in the subsonic flight domain, i.e.,
below the critical Mach number.
(b) Swept wings are typical for transport aircraft operating in the transonic flight
domain. The wing’s sweep, together with a supercritical chord section, permits to fly
at Mach numbers above the critical one, shifting the drag-divergence Mach number
up to high values. Usually backswept wings are employed. Forward swept wings in
principle are feasible, but for several reasons did not find broad use, like the oblique
wing, which is swept forward at one side of the aircraft and backward at the other
side.
(c) The slender, delta-like wing is typical for aircraft operating in the supersonic
flight domain, like the former Concorde. The main reason is to have a subsonic
leading edge in order to reduce wave-drag increments. Fighter configurations either
have delta wings or hybrid wings with a strake-trapezoidal wing combination.
Both—still hypothetical—hypersonic airbreathers and winged re-entry vehicles—
like the former Space Shuttle Orbiter—have highly swept leading edges. The re-entry
vehicle on the initial flight trajectory flies a breaking mission at very high angle of
attack, and on the lower flight trajectory aims for reasonable down and cross range.
The airbreather must have low overall drag on the whole flight trajectory, like any
other aircraft.
Problem 10.2
These phenomena are vortex breakdown and vortex overlapping. Vortex breakdown—
also called vortex bursting—happens, when a vortex interacts with an axially adverse
pressure field. Due to their breakdown, which as a rule is beginning at the rear of the
wing, the lee-side vortices loose the capability to induce the suction force beneath
them. Consequently a loss of lift occurs and at the same time a pitch-up of the wing
12.9 Problems of Chapter 10 409

Fig. 12.4 Sketch of the

principle spanwise
distributions of C L (y)l(y)
and C L (y)

happens. If vortex breakdown occurs asymmetrically, side forces and moments are
induced. The higher the leading-edge sweep angle of the wing, the higher is the angle
of attack, at which vortex breakdown happens.
Vortex overlapping occurs at wings with very high leading-edge sweep angle and
at high angles of attack. This phenomenon is also observed at fuselages. It leads to
side forces at nominally zero sideslip angle and to difficulties in aircraft control.
Problem 10.3
For the original round-edged configuration at α = 30◦ and M∞ = 0.3 the normal angle
of attack is α N = 49.1◦ and the normal Mach number M N = 0.2. For the manipulated
sharp-edged configuration this data are α N = 52.8◦ and M N = 0.19.
The correlations hence for both cases show vortices fixed in span direction, but
elevating. The fact that for the original case obviously no lee-side vortices were
present, must be traced back to the combination of leading-edge radius and Reynolds
number, which were sub-critical.
Problem 10.4
Yes, in the Poincaré surface the left and right attachment points A3 , being half-saddle
points, move into the center and above the wing, there combined forming a full saddle
point S. The half-saddle point S4 at the center of the upper surface is retained there,
but the flow directions in it change (Fig. 12.4).
Problem 10.5
The spanwise lift and lift-coefficient distributions, C L (y)l(y) and C L (y), in principle
can be sketched as follows:
The dominating separation scenario is due to leading-edge separation resulting
in a fully developed leading-edge vortex at α ≈ 20◦ , Fig. 11.2, providing an inboard
loading of C L (y)l(y). Further, the wing tip area is characterized by irregular sepa-
rated flow as the corresponding sections exhibit stall, which is reflected by the local
outboard maximum in the C L (y) trend. Consequently, the wing tip section does not
contribute effectively to lift production. Therefore often cropped delta-wing plan-
forms are used for practical applications.
410 12 Solutions of the Problems

Fig. 12.5 Sketch of the

topological representation of
the closed lee-side flow field
in a Poincaré surface

Problem 10.6
According to Fig. 11.3 a pitch-up tendency is present at high angle of attack.
Figure 11.5 provides a correlation of wing leading-edge sweep and aspect ratio based
on a variety of configurations indicating “controllable” and “unsafe” pitch-up ten-
dencies. The combination of ϕ0 = 65◦ leading-edge sweep and aspect ratio of 
= 1.9 probably results in an uncontrollable pitch-up tendency. Further measures,
e.g., leading-edge flap setting, have to be implemented to account for unacceptable
longitudinal stability margins.
Problem 10.7
(1) The topological representation of the flow field in a cross-flow plane—a
Poincaré surface—can be sketched as shown in Fig. 12.5, compare Fig. 11.23.
Following the topological rule, Eq. (7.25), outlined in Sect. 7.4.3, the sketch can
be proven for consistency. Two interacting primary vortices and one secondary vortex
are shown resulting in three nodes and one saddle point per wing side (2 x (3N −
1S)) and one saddle point for the symmetry plane. Further, there are two separation
and one attachment half saddle points S  per wing side and two attachment quarter
saddle points S  at the lower side. Thus, it matches the rule of Eq. (7.25):
 1   1   1  
( N+ N )−( S+ S + S ) = −1,
2 2 4
with
1 1
6 − (3 + 7 + 2) = −1.
2 4
(2) Due to the increased suction on the forward (strake) wing part a pitch-up
tendency for angles of attack of 20◦ and higher most probably will be present, see
Figs. 11.3 and 11.24. Vortex breakdown will take place in the rearward area of the
wing, thus decreasing the suction level there, while upstream the fully developed
leading-edge vortex for the strake section still produces high lift, which then results
in a pitch-up tendency with respect to typical locations of the center of gravity.
Following the trends given in Figs. 11.4 and 11.23, for some sideslip angle a roll
reversal for angles of attack beyond 20◦ is also expected, because the windward
leading-edge vortex system is prone to vortex breakdown. This is caused by the
12.9 Problems of Chapter 10 411

change in the effective wing leading-edge sweep as it decreases at the windward side
and increases at the leeward side.
Problem 10.8
From Appendix A.4—small aspect-ratio wing theory—one finds:

l(x) = 4π · αq∞ · y(x) · dy(x)/d x,

bx
y= ,
2c
b
dy/d x = ,
2c
bx b b
l(x)/(παq∞ ) = 4 = ( )2 x,
2
 c c 2c c
b
L/(παq∞ ) = l(x)/(παq∞ ) · d x = ( )2 · c2 /2 = b2 /2,
c
0 c
b
M/(παq∞ ) = x · l(x)/(παq∞ ) · d x = ( )2 · c3 /3 = b2 c/3,
0 c
2
xcp = M/L = c.
3
Problem 10.9
In Appendix A.4—small aspect-ratio wing theory—is an expression l(x) = 4 π α q
y(x) dy(x)/d x. Thus the lift reads
 1  1
L= l(x)d x = 4παq∞ y(x) · dy(x)/d x · d x,
0 0
 1
1 2
L = 4παq∞ dy = 2παq∞ · 0.1252 .
x=0 2

The lift curve slope is

dC L L
=d /dα,
dα q∞ A
A = 0.375,
dC L /dα = 0.262.

Alternatively the lift contributions from the front part and the rear part may be added
together.
Problem 10.10
In the two collocation points, one on the wing and one on the stabilizer, sum the
vertical velocity components due to the free stream and the two vortices. On the
wing the sum should be zero, and on the stabilizer −u ∞ δ.
412 12 Solutions of the Problems

1 1
w1 = u ∞ · 0 − Γ1 + Γ2 = 0,
2π (c/2) 2π (4c + c/4 + c/8)
1 1
w2 = u ∞ · 0 − Γ 1 − Γ2 = −u ∞ δ,
2π (4c + 3c/4 + 3c/8) 2π (c/4)

1 4
−Γ1 + Γ2 = 0,
πc 35π c
4 2
−Γ1 − Γ2 = −u ∞ δ.
41π c πc

Γ1 πc
= 0.0568.
u∞δ

12.10 Problems of Chapter 11

Problem 11.1
Lift decreases, drag rises, and, very importantly, a pitch-up happens. This behavior
is due to the loss of suction force, which disappears below the vortex pair where the
breakdown happens. Because that happens at the rear of the wing, the upward directed
suction force there vanishes, and the consequence in particular is the pitch-up.
Problem 11.2
(a) High angle of attack and asymmetries of the configuration and/or the freestream
are the basic causes. Lateral/directional divergence problems introduced by side-slip
effects (Cn β , Clβ ) lead to non-controllability of the aircraft. Hence these problems
must be reduced by proper configurational development and/or restrictions in the
flight envelope. (b) The flow structure due to sideslip basically becomes asymmetric
due to a reduction of the effective leading-edge sweep on the windward side, and an
increase on the leeward side. See regarding both items also the correlations in the
Sects. 10.2.5 and 10.2.6.
Problem 11.3
Low aspect ratio and high leading-edge sweep are the drivers in supersonic—here
M∞ = 2—wing design, which results in a subsonic leading edge. A softly varying
area distribution—supersonic area ruling—gives the chance to minimize the drag.
This helps to increase the lift-to-drag ratio at supersonic cruise by reduction of the
(zero-lift) wave drag.
To keep the induced drag at a low level, the supersonic trim drag has to be reduced,
too. This demands to keep the backward shift of the neutral point at supersonic flight
at low level. With that high (negative) trim-flap deflections can be avoided. Conse-
quently the lift-to-drag ratio is increased, thus either lowering the fuel consumption
or extending the flight range of the aircraft.
12.10 Problems of Chapter 11 413

Stability and handling qualities of such aircraft—here the Concorde—usually are

more critical at low subsonic flight speeds, say at take-off and landing and in the
high angle-of-attack flight regime. Aspects of stability divergence—pitch, yaw and
roll motion—are of importance. Controllability also may be a critical design driver,
see Sect. 11.2.
Problem 11.4
First it is to remark that these curves only show results for the special case C L = C L opt
= C L (L/D)opt . This means that C L is not constant for these curves, but is varying
along them, as happens with C D0 and C Di .
In Fig. 11.12 our case ( = 2, ϕ0 = 63◦ ) is included for three wing-chord shapes:
symmetric, warped, full-span nose-flap. The figure is valid for sharp leading-edge
wings with increasing sweep angle ϕ0 .
For our case, ϕ0 = 63◦ , we find the following leading-edge suction ratios S ∗ (in
percent of attained to ideal—100 per cent— leading-edge suction):
• symmetric profile: S ∗ = 40 per cent,
• warped profile: S ∗ = 48 per cent,
• full-span nose flap: S ∗ = 64 per cent.
Turning to Fig. 11.13, the data for the point  = 2, ϕ0 = 63◦ now are given for the
wing with rounded leading edge and a relative thickness of 5 per cent for the cases
“symmetric profile” and “conical camber” of the wing.
For our flight Mach number M∞ = 0.3 we obtain:
• symmetric profile: S ∗ = 60 per cent,
• conical camber: S ∗ = 95 per cent.
These results clearly demonstrate the superiority of wings with rounded leading
edge, when we look for high suction rates at subsonic Mach numbers in order to end
up with a high lift-to-drag ratio.
Problem 11.5
The positive Effects of concentrated spanwise blowing over the upper side of the
wing are, see Figs. 11.19 to 11.21:
1. Increase of the maximum lift by stabilizing an existing leading-edge vortex system
(wings with high leading-edge sweep) or by generation—and stabilization—of
concentrated vortex systems over wings with low/medium aspect ratio. In both
cases the additional lift increases non-linearly with increasing angle of attack.
2. This is also reflected in the development of lift-dependent drag, which is reduced
for constant lift due to the lower angle of attack necessary for it.
3. Most efficient in terms of Δ C L /cμ is blowing over wings that do not develop lee-
side vortex systems. But the efficiency of Δ C L /cμ is decreasing with increasing
blowing coefficient cμ .
The reasons for its “non-application” can be found in the definition of the blowing
coefficient
414 12 Solutions of the Problems

ṁ v j
cμ = .
q∞ Ar e f

The efficiency of blowing decreases with increasing flight speed: cμ ∝ 1/u 2∞ .

Consequently the induced extra lift will decrease with increasing speed.
Hence spanwise blowing is effective—and applicable—for low flight speeds only.
But there it leads to engine thrust reduction, because of the needed mass flux of the
jet ṁ. Therefore this technique did not find its way into application.
Problem 11.6
Topological flow structures and the development of the pitching moment with increas-
ing angle of attack are presented in the Figs. 11.23 and 11.26 for the planar and for
the cranked wing.
Comparing the figures we find:
(a) The planar wing develops a strong pitch-up tendency beginning already at
medium angle of attack α ≈ 10◦ , the cranked wing on the other hand shows a quasi-
linear stability curve Cm α (0  α < 35◦ ) and finally a—safe—pitch-down tendency.
(b) These tendencies correspond to the following flow structures:

(A) Low to medium angle of attack: inserts (1) and (2) in the figures.

Here both wings show quite similar flow characteristics (1):

• Three nodes are present due to vortices induced by
(a) the highly swept leading edge of the inboard wing ⇒ leading-edge separation.
(b) the leading-edge kink (high sweep of the inboard wing, reduced sweep of the
outboard wing), plus the crank of the outboard wing,
(c) the outboard wing side edge.
• Open lee-side flow field at both wings.
• An internal saddle point due to interaction of the three vortex systems, which exist
over the wings.

An increase of the angle of attack—(2)—induces a pitch-up tendency for the

planar wing only. The flow structure is altered by the beginning of a local unsteadiness
and the increasing strength of the vortex system of the outboard wing.
No change of the flow structure is found for the cranked wing, (2) in Fig. 11.26.
Both wings still show an open lee-side flow field.
(B) Medium to high angle of attack: inserts (3) and (4) in the figures.

The main change in the flow structure of the planar wing—(2) ⇒ (3)—is the
appearance of a central, free-flow saddle point—representative for a closed lee-side
flow field—in combination with an upward movement of the leading-edge vortex of
the inboard wing.
In contrast to that the cranked wing still presents an open lee-side flow field (3)
and a stronger induced effect of the crank vortex, driven by the primary leading-edge
vortex of the highly swept inboard wing.
12.10 Problems of Chapter 11 415

A further increase of the angle of attack—(3) ⇒ (4)—has tremendous conse-

quences for the planar wing. The location of vortex breakdown of the primary leading-
edge vortex is now far forward over the wing, hence the lift force is concentrated
there, with a strong pitch-up as a result.
In contrast to that the cranked wing, (4) in Fig. 11.26, is producing two stable
vortex systems: see the nodal point N1 from the leading edge vortex of the inboard
wing and the nodal point N2 —induced and stabilized by N1 —originating over the
crank. Now we also have a closed lee-side flow field.
From this comparison it is clear that the positive strong interaction of N1 and
N2 is the reason for the now “healthy” development of the pitching moment of the
cranked-wing configuration.

References

1. Hirschel, E.H.: Basics of Aerothermodynamics. 2nd, revised edition. Springer, Cham Heidel-
berg, New York (2015)
2. Hirschel, E.H., Cousteix, J., Kordulla, W.: Three-Dimensional Attached Viscous Flow. Springer,
Berlin, Heidelberg
Appendix A
Useful Relations

Useful relations are given for quantitative considerations, both exact ones and approx-
imative ones. Details can be found, for instance, in [1–4] and in relevant monographs.

A.1 Pressure Relations

The pressure coefficient c p is defined as

p − p∞
cp = , (A.1)
q∞

where q∞ = 0.5ρ∞ u 2∞ is the dynamic pressure of the free-stream. For air as perfect
gas the pressure coefficient c p reads with M∞ = u ∞ /a∞
 
2 p
cp = −1 , (A.2)
γ M∞2 p∞
√
a = γ RT∞ is the speed of sound, γ = c p /cv = 1.4 (perfect gas) the ratio of specific
heats, and R = 287, 06 m2 s−2 K−1 , the specific gas 
constant (air).
The expansion limit—the maximum speed vm = 2 c p Tt , with Tt being the total
temperature—is reached with p → 0. With this we obtain the vacuum pressure
coefficient
2
c pvac = − . (A.3)
γ M∞2

The pressure coefficient can be expressed in terms of the ratio local speed u to
free-stream value u ∞ , and the free-stream Mach number M∞ :

© Springer-Verlag GmbH Germany, part of Springer Nature 2021 417

E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3
418 Appendix A: Useful Relations
⎧ ⎫
  γ−1
γ

2 ⎨ γ−1 2 u 2 ⎬
cp = 2 ⎩
1 + M ∞ [1 − ] − 1 . (A.4)
γ M∞ 2 u∞ ⎭

In subsonic compressible flow we get for the stagnation point (isentropic com-
pression) with u = 0
  γ
2 γ − 1 2 γ−1
c pstag = 1+ M∞ −1 . (A.5)
γ M∞2 2

In the case of supersonic flow of course the total-pressure loss across the shock
must be taken into account.
For incompressible flow the pressure coefficient is found with the help of
Bernoulli’s equation, see, e.g., [1], and reads

u2
cp = 1 − . (A.6)
u 2∞

We note that at a stagnation point c p is always larger for compressible flow than
that for incompressible flow with c pstag = 1.

A.2 Vortex-Induced Velocity

A.2.1 Introduction

With Fig. 3.1 we introduced several vortex models for finite-span wing flow. The
following useful relations, derived from these models, provide further insight into
how well these models work and can be used in actual applications [4]. We begin in
two dimensions with a description of the vortex bound to the airfoil in Fig 3.4, and
after that proceed to the three-dimensional case.

A.2.2 The Two-Dimensional Case

• Vortex Model for Airfoil Flow

A distribution of vorticity γ(x) on an airfoil is a solution to Laplace’s equation and
forms the basis for the analysis of a very thin cambered airfoil (i.e. a plate): Glauert’s
thin-airfoil theory.
The solution satisfies the boundary conditions, if the combination of the velocity
induced by the vortices cancels the component of the freestream normal to the plate:
Appendix A: Useful Relations 419

dz
wi (x) = u ∞ (α − ),
dx
where small-angle approximations have been introduced.
The basic approximation of thin-airfoil theory is that the velocity induced at some
point x due to the vorticity at x  may be approximated by the velocity induced at the
same x position on the x-axis due to a vortex on the x-axis, i.e., the vortex can be
shifted down on the x-axis.
The velocity induced by this bit of vorticity is computed from the basic vortex
singularity, the formula known as the Biot-Savart law, Sect. 3.9. For the element of
vorticity at x  , it reads in two dimensions:

1 γ(x  )
d w(x) = d x .
2π (x − x  )

So, the total induced velocity at the point x is given by:

1
1 γ(x  )
w(x) = d x .
2π 0 (x − x  )

Combining this expression with the flow-tangency boundary condition, we have

the basic integral equation to be solved for the unknown vorticity distribution:
1
1 γ(x  ) dz

d x = α − .
2π u ∞ 0 (x − x ) dx

The approach to solving this equation is to change variables:

cos θ = 1 − 2x,

with θ varying from 0 to π, and to write γ as a Fourier series.

If we consider only symmetric airfoils, i.e., dz/d x = 0—no camber—we obtain
the solution

γ(θ) = 2u ∞ α cot (θ/2).

For an arbitrary airfoil, γ is expanded in a Fourier series that then leads to a

solution.
• Lumped-Vortex Method Glauert’s thin-airfoil theory (see above) describes the
vortex distribution γ over a cambered airfoil.
The center of gravity for this distribution lies in the point x = c/4. Replacing the
continuous distribution γ(θ) with the resultant point vortex Γ = u ∞ πcα at x = c/4
, lift force per unit span and moment remain the same (Fig. A.1). (Clearly, the Kutta
condition says that there shall be no vortex on the trailing edge.)
With only the one variable Γ (instead of a whole distribution) it is only possible
to satisfy the boundary condition in just one point. Let this point lie a distance kc
420 Appendix A: Useful Relations

Fig. A.1 Substitution of the

distributed vorticity with a
single vortex [4]

downstream from the leading edge. The contribution from the discrete vortex has to
balance the vertical component of the free stream velocity:
Γ
w(x = kc) = − + u∞α = 0 ⇒ k = 3/4.
2π(kc − c/4)

If the airfoil chord were to be divided into several parts, N panels, and on each
panel i is put a vortex Γi , it can be shown, that the boundary condition is satisfied at
the 3/4 local chord on each panel.

A.3 The Three-Dimensional Case—the Elementary

Horse-Shoe Vortex

From Biot-Savart’s law, the induced vertical velocity w—with the notation from the
left part of the figure below—is (Fig. A.2)
−Γ
w(x, y, 0) = (cos β1 − cos β2 ).
4πd
The complete horse-shoe vortex as it would be used in the vortex-lattice method
with coordinates for the collocation point (x, y) and the vortex definition points
(xa , ya ) and (xb , yb ), is shown in the right part of the figure.
The vertical velocity for the slightly simpler case with xa = xb is
Appendix A: Useful Relations 421

Fig. A.2 Influence of an elementary horse-shoe vortex [4]

 
−Γ (x − xa )2 + (y − ya )2
w(x, y, 0) = 1+ +
4π (y − ya ) x − xa
 
Γ (x − xa )2 + (y − yb )2
+ 1+ . (A.7)
4π (y − yb ) x − xa

A.4 Wing-Lift Predictions

A.4.1 Prandtl’s Theory for the Large Aspect-Ratio Wing

• Concept
Rather than representing the wing with just one horse-shoe shaped vortex, the wing is
represented by several of them [4]. In this way the bound circulation in the wing can
vary from the root to the tip. The strength of the trailing-vortex filaments is related
to the circulation on the wing by Γwake = ΔΓwing . A wake vortex is shed from the
wing whenever the wing circulation changes, as illustrated in Fig. A.3.
In the limit, as the number of horse-shoe vortices goes to infinity, the trailing
vortex layer is a sheet of vorticity. The trailing vortex strength per unit length in the
y-direction is the derivative of the total circulation on the wing at that station.
From this model we can derive the basic relations for finite-span wings. The
vorticity strength in the trailing vortex layer is given by: γ = dΓ /dy, and since the
wing circulation changes most quickly near the wing tip, the strength of the trailing
vortex layer is largest there. The roll-up of the trailing vortex layer, which can be
seen as beginning there, leads to the pair of trailing vortices. The two wing-tip vortex
systems, Sect. 8.4.4, are drawn into the trailing vortex layer and hence into the trailing
vortices.
• Lift Prediction
The derivation of the method is based on the concept that lift is caused by a change in
the flow direction at the wing—the downwash at the location of the wing—induced
by vortices representing the wing surface. The local effective angle of attack, αe f f ,
is obtained through a reduction of the geometric angle α, with the induced angle
αinduc (y):
422 Appendix A: Useful Relations

Fig. A.3 Horse-shoe vortex arrangement to simulate a lifting wing [4]

αe f f (y) = α − αinduc (y).

The reduction is due to the wake downstream of the wing, also a distribution of
vorticity. Both wing and wake deflect the air stream. As the name implies the wing-lift
distribution is replaced with a lifting line. Therefore the method can only predict how
the force of the lifting line varies in the spanwise direction. The variation is described
through Fourier coefficients. The coefficients are determined by the geometry of the
wing in a number of positions along the span.
An especially simple result is obtained  for a flat elliptical wing, for which the
vortex distribution becomes Γ (y) = Γ0 1 − (2y/b)2 , where b is the wing span.
The method gives for the special case of a plane elliptical wing the constant
vertical velocity
w(x = c/4, y) = −Γ0 /(2b).

The effective angle of attack at a span station y is

αe f f (y) = α − w(x = c/4, y)/u ∞ = α − Γ0 /(2bu ∞ ).

Defining the load at a span station y as l(y), and the local lift coefficient cl (y), one
can write
l(y) = c L (y) q c(y) dy = ρ u ∞ Γ (y)dy,

and
Appendix A: Useful Relations 423

Fig. A.4 Lift-curve slope

dC L /dα as function of the
aspect ratio Λ of wings [4]

Γo = cl (y) u ∞ co /2. (A.8)

At each station the relation between flow deflection and the lift is taken from two
dimensional flow theory

cl (y) = 2π (α − Γo /(2 b u ∞ )). (A.9)

Elimination of Γ between Eqs (A.8) and (A.9), and integration over the span gives
with α in radian
dC L 2πΛ
= , (A.10)
dα Λ+2

where Λ is the wing aspect ratio. For Λ → ∞ we obtain the two-dimensional case
dC L
= 2π. (A.11)
dα
A graph is given in Fig. A.4.
As the wing vortex is placed at the 1/4 chord points of all sections, the resulting
lift force will act in the 1/4 chord point of the reference chord.
For plane wings of elliptical planform the induced-drag coefficient C Di can be
shown to be

C L2
C Di = , (A.12)
πΛ
see Sect. 3.16.
424 Appendix A: Useful Relations

Fig. A.5 Sketch of the flow

pattern as a delta wing at
incidence penetrates a
vertical plane normal to the
flight direction [5]

A.4.2 R.T. Jones’ Theory for Small Aspect-Ratio Wings

R.T. Jones devised his theory by reasoning on physical grounds in the following way
[5]. Fix a vertical reference plane in space. A slender delta wing at angle of attack α
runs with a speed of u ∞ straight through the plane, and a streamline pattern, like the
one in Fig. A.5, is established. It is also the pattern of a flat plate mowing downward
with a velocity u ∞ α.
Observe, that the width of the wing and the scale of the flow pattern will vary, as
the wing moves through the plane. This increase in the width needs a force from the
cut of the wing corresponding to the downwash velocity u ∞ α times the local change
of the mass involved m  . The reacting force on the wing cross section is noted l(x):

l(x) = u ∞ α dm  /dt = u 2∞ α dm  /d x, (A.13)

because u ∞ = d x/dt. From two dimensional flow theory it is known that (a com-
plicated proof is given in [4])
m  = π y12 ρ∞ ,

where y1 is the y-coordinate of the leading edge at a distance x from the apex, and
m  is the mass in a cross section with y1 as radius. Hence we obtain

l(x) = u 2∞ α 2πρ∞ y1 (x) dy1 (x)/d x = 4π α (ρ∞ u 2∞ /2) y1 (x) dy1 (x)/d x.

Consider a delta wing with the chord c, the aspect ratio Λ and lift L, then y1 (x)
becomes Λ x/4.
Or if the lift coefficient is based on the wing area:
dC L π
= Λ. (A.14)
dα 2
Figure A.4 plots dC L /dα according to the theories for small and for large aspect
ratios, respectively. The vortex lattice method, Fig. 3.1, is one suitable approach to
fill in data for the most important interval of moderate aspect ratios. Then it will be
possible to judge, within how wide errors, the analytical methods (lifting-line and
Appendix A: Useful Relations 425

small aspect-ratio) are valid. Note that even for an aspect ratio of Λ = 15 there is
quite a way to the two dimensional value of dC L /dα = 2π.

A.5 Estimation of Boundary-Layer Properties

For a quick estimations of boundary-layer properties we give approximate boundary-

layer relations in generalized form. In order to be self-consistent, we take them over
partly from [3]. The relations are based on work of G. Simeonides [6], and hold
for two-dimensional and not too strongly three-dimensional attached viscous flow,
as long as the flow is of boundary-layer type. The body surface is assumed to be
hydraulically smooth.
The relations are valid for both laminar and turbulent flow in a Reynolds number
range up to 107 . It is assumed that the temperature does not exceed 1,500 K. Although
derived originally for hypersonic flow problems, the relations can be used also for all
lower-speed problems. Of course it is necessary, as with all approximate relations,
to check the range of applicability and to establish the error range. This holds in
particular, if the relations are of empirical or semi-empirical character.
In the following relations we use for laminar flow the exponent n = 0.5 and for
turbulent flow n = 0.2.

A.5.1 Viscosity and Thermal Conductivity

Expressions for the viscosity of air are the Sutherland equation and simple power-law
approximations. The Sutherland equation reads

T 1.5
μ Suth = 1.458 · 10−6 [N s/m 2 ]. (A.15)
T + 110.4

A simple power-law approximation is μ = cμ T ωμ . For the temperature range T

 200 K the approximation—with the constant cμ1 computed at T = 97 K—it can
be written

μ1 = cμ1 T ωμ1 = 0.702 · 10−7 T, (A.16)

and for T  200 K—with the constant cμ2 computed at T = 407.4 K—

μ2 = cμ2 T ωμ2 = 0.04644 · 10−5 T 0.65 . (A.17)

Regarding the thermal conductivity of air we note that for temperatures up to

1,500–2,000 K, an approximate relation due to C.F. Hansen—similar to Sutherland’s
equation for the viscosity of air—can be used [7]
426 Appendix A: Useful Relations

T 1.5
k H an = 1.993 · 10−3 [N /(s K )]. (A.18)
T + 112.0

A simple power-law approximation can also be formulated for the thermal conduc-
tivity: k = ck T ωk . For the temperature range T  200 K the approximation reads—
with the constant ck1 computed at T = 100 K—

k1 = ck1 T ωk1 = 9.572 · 10−5 T, (A.19)

and for T  200 K—with the constant ck2 computed at T = 300 K—

k2 = ck2 T ωk2 = 34.957 · 10−5 T 0.75 . (A.20)

Often the thermal conductivity is found via the Prandtl number:

μc p
Pr = , (A.21)
k

with c p [m 2 /(s 2 K )] being the (mass) specific heat at constant pressure.

A.5.2 Reference Temperature and Recovery Temperature

In order to take into account Mach-number and wall-temperature effects in boundary-

layer flow, the reference-temperature concept is used [8, 9].
The reference temperature T ∗ is empirically composed of the boundary layer edge
temperature Te , the wall temperature Tw , and the recovery temperature Tr :

T ∗ = 0.28Te + 0.5Tw + 0.22Tr [K ]. (A.22)

The recovery or adiabatic wall temperature Tr is defined by1

 
v2 γ−1 2
Tr = Te + r ∗ e = Te 1 + r ∗ Me , (A.23)
2c p 2

with r ∗ being the recovery factor, which is a function of the Prandtl number Pr at
the reference temperature T ∗ , Me is the Mach number at the edge of the boundary
layer.
The Prandtl number depends rather weakly on the temperature, see, e.g., [1, 3].
∗
√to assume r = r = const. For laminar
Usually it is sufficient √ flow the recovery factor
can be taken as r = Pr , and for turbulent flow r = 3 Pr . With the Prandtl number
at low temperatures, Pr ≈ 0.74, we get rlam ≈ 0.86 and rtur b ≈ 0.90.

1 The total temperature Tt is found with r ∗ = 1.

Appendix A: Useful Relations 427

The characteristic Reynolds number for a boundary-layer like attached viscous

flow at the location x is postulated to read
ρ∗ u e x
Rex∗ = . (A.24)
μ∗

Density ρ∗ and viscosity μ∗ are reference data, characteristic for the boundary layer.
They are determined with the local pressure p and the reference temperature T ∗ ,
with ve being the external inviscid flow velocity.
Introducing the boundary layer edge data as reference flow data into Eq. (A.24)
yields
ρe u e x ρ∗ μe ρ∗ μe
Rex∗ = = Ree,x , (A.25)
μe ρe μ∗ ρe μ∗

with Ree,x = ρe u e x/μe . This relation can be simplified. If we apply it to boundary

layer like flows, we can write, because p ≈ pe ≈ pw = const.
ρ∗ Te
= ∗. (A.26)
ρe T

If, for simplicity, we further introduce the power-law expression for the viscosity,
we obtain
∗
μ∗ (T ∗ )ω
= . (A.27)
μe (Te )ωe

Only if T ∗ and Te are both in the same temperature interval, ω ∗ and ωe are equal,
we get:
 ∗ ω
μ∗ T
= . (A.28)
μe Te

Introducing Eqs. (A.26) and (A.28) into Eq. (A.25) reduces the latter to
 1+ω
Te
Rex∗ = Ree,x . (A.29)
T∗

A.5.3 Skin Friction and Heat Transfer

For the skin friction over a flat plate we get in generalized form, with C = 0.332 for
laminar flow and C = 0.0296 for turbulent flow [6]
428 Appendix A: Useful Relations
 1−n  n
−n T∞ μ∗
τw = Cμ∞ u ∞ x (Re∞ )
u 1−n
[N /m 2 ], (A.30)
T∗ μ∞

respectively
 n(1+ω)−1
T∗
τw = Cμ∞ u ∞ x −n (Re∞ ) ,
u 1−n
(A.31)
T∞

with Re∞u
= ρ∞ u ∞ /μ∞ [m −1 ] being the unit Reynolds number.
The skin-friction coefficient c f then is
 1−n  n
τw T∞ μ∗
= c f = 2 C x −n u −n
(Re∞ ) . (A.32)
0.5ρ∞ u 2∞ T∗ μ∞

The heat flux in the gas at the wall reads, again with C = 0.332 for laminar flow
and C = 0.0296 for turbulent flow,

 1−n  n
T∞ μ∗  1−n
qgw = C x −n k∞ Pr 1/3 (Tr − Tw ) u
Re∞ [kg/s 3 ], (A.33)
T∗ μ∞

respectively
 n(1+ω)−1
−n T∗  1−n
qgw = C x k∞ Pr 1/3
(Tr − Tw ) u
Re∞ . (A.34)
T∞

A.5.4 Boundary-Layer Thicknesses

The boundary-layer thicknesses of the Blasius and the 1/7-power law—

incompressible—boundary layers can be written in generalized form:

x 1−n
δi = Ci [m], (A.35)
(Re∞u )n

with the unit Reynolds number being Re∞ u

= ρ∞ u ∞ /μ∞ , and i = 0 for the boundary-
layer thickness δ (where we leave away the lower index 0), i = 1 for the displacement
thickness δ1 , i = 2 for the momentum thickness δ2 .
The constants Ci for the different thicknesses and the exponent n, both for laminar
and turbulent flow are given in Table A.1.
u u∗
Redefining the original Re∞ with the relation Eq. (A.25) as Re∞ and introducing
the reference temperature, Eq. (A.35) becomes
Appendix A: Useful Relations 429

Table A.1 Constants Ci in Eq. (A.35). The exponent in that equation is n = 0.5 for laminar flow
and n = 0.2 for turbulent flow. Also given is the shape factor H12 = δ1 /δ2
δ δ1 δ2 H12
Laminar BL: Ci = 5 1.721 0.664 2.591
Turbulent BL: Ci = 0.37 0.046 0.036 1.278

 −n
x 1−n ρ∗ μ∞
δi = Ci , (A.36)
(Re∞u )n ρ∞ μ∗

respectively
 n(1+ω)
x 1−n T∗
δi = Ci . (A.37)
(Re∞u )n T∞

Alternate formulations for high Mach-number flow can be found, for instance, in
[1].
The characteristic thicknesses Δ of laminar and turbulent boundary layers can
be used to explain quite a number of phenomena in laminar and turbulent attached
viscous flow, see [1, 3]. They also must be taken into account in grid generation for
numerical methods.
The characteristic thicknesses govern the wall shear stress and the heat flux in
the gas at the wall of attached viscous flow of boundary layer type. In the laminar
flow domain, the characteristic thickness Δlam is approximately the boundary layer
thickness δlam , the 99 per-cent thickness.
In the turbulent domain Δtur b is the thickness of the viscous sub-layer δvs , not the
thickness δtur b . δvs is much smaller than the boundary layer thickness δtur b . However,
at and in the vicinity of singular points and singular lines the characteristic thickness
can not be approximated by δlam or δvs .
The explicit relation for the thickness of the viscous sub-layer usually is not given
in the boundary-layer literature. Exceptions are for instance the book of E.R.G. Eckert
and R.M. Drake [9] and the report of G. Simeonides [10]. The relation of the viscous
sub-layer, and hence of the characteristic thickness Δtur b , reads for the flat plate:
 0.8
x 0.2 ρ∞ μ∗
δvs = Δtur b = 33.78 , (A.38)
(Re∞
u )0.8 ρ∗ μ∞

respectively:
 0.8(1+ω)
x 0.2 T∗
δvs = Δtur b = 33.78 . (A.39)
(Re∞
u )0.8 T∞
430 Appendix A: Useful Relations

Note the different dependencies of δvs and δtur b on x and Re∞

u
: δvs ∝ x n (Re∞ )
u 1−n
u −n
compared to δtur b ∝ x (Re∞ ) (Eq. (A.35)), where n = 0.2. The authors of [9]
1−n

give n = 0.1 for the δvs -relation.

Regarding the treatment at junctions, laminar-turbulent transition locations etc.
with the concept of the virtual origin see [3].

References
1. Hirschel, E.H.: Basics of Aerothermodynamics. 2nd, revised edition. Springer, Cham, Heidel-
berg, New York (2015)
2. Hirschel, E.H., Weiland C.: Selected Aerothermodynamic Design Problems of Hypersonic
Flight Vehicles. Progress in Astronautics and Aeronautics, AIAA, Reston, Va, vol. 229.
Springer, Heidelberg (2009)
3. Hirschel, E.H., Cousteix, J., Kordulla, W.: Three-Dimensional Attached Viscous Flow. Springer,
Berlin, Heidelberg (2014)
4. Rizzi, A., Oppelstrup, J.: Aircraft Aerodynamic Design with Computational Software. Cam-
bridge University Press (2020)
5. Jones, R.T.: Properties of Low-Aspect-Ratio Pointed Wings at Speeds Below and Above the
Speed of Sound. NACA Rep. 835 (1946)
6. Simeonides, G.: Generalized reference-enthalpy formulation and simulation of viscous effects
in hypersonic flow. Shock Waves 8 3, 161–172 (1998)
7. Hansen, C.F.: Approximations for the Thermodynamic and Transport Properties of High-
Temperature Air. NACA TR R-50 (1959)
8. Rubesin, M.W., Johnson, H.A.: A critical review of skin friction and heat transfer solutions of
the laminar boundary layer of a flat plate. Trans. ASME 71, 385–388 (1949)
9. Eckert, E.R.G., Drake, R.M.: Heat and Mass Transfer. MacGraw-Hill, New York (1950)
10. Simeonides, G.: On the Scaling of Wall Temperature Viscous Effects. ESA/ESTEC EWP -
1880 (1996)
Appendix B
Constants, Atmosphere Data, Units, and
Conversions

In this book, units are in general the SI units (Système International d’Unités), see
[1, 2], where also the constants can be found. In the following sections we give
first constants and air properties, Sect. B.1, and then a selection of atmosphere data,
Sect. B.2. The basic units, the derived units, and conversions to US units are given
in Sect. B.3.1.1

B.1 Constants and Air Properties

See Table B.1.

Molar universal gas constant: R0 = 8.314472 · 103 kg m2 s−2 kmol−1 K−1 =
= 4.97201·104 lbm ft2 s−2 (lbm -mol)−1 ◦ R−1

Standard gravitational
acceleration of earth at
sea level: g0 = 9.80665 m s−2 = 32.174 ft s−2

Table B.1 Molecular weights and gas constants of air constituents for the low temperature domain
[3, 4]. ∗ is the U.S. standard atmosphere value [5], + the value from [4]
Gas Molecular weight M [kg kmol−1 ] Specific gas constant R [m2 s−2 K−1 ]
Air 28.9644∗ (28.97+ ) 287.06
N2 28.02 296.73
O2 32.00 259.83

1 Details can be found, for instance, at http://physics.nist.gov/cuu/Reference/contents/html

© Springer-Verlag GmbH Germany, part of Springer Nature 2021 431
E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3
432 Appendix B: Constants, Atmosphere Data, Units, and Conversions

Table B.2 Properties of the 15 ◦ C U.S. standard atmosphere as function of the altitude [5]
Altitude H Temperature Pressure p Density ρ Dynamic Thermal
[km] T [K] [Pa] [kg m−3 ] viscosity μ conductivity k
[N s m−2 ] [W m−1 K−1 ]
0.0 288.150 1.013 · 105 1.225 · 100 1.789 · 10−5 2.536 · 10−2
1.0 281.651 8.988 · 104 1.112 · 100 1.758 · 10−5 2.485 · 10−2
2.0 275.154 7.950 · 104 1.007 · 100 1.726 · 10−5 2.433 · 10−2
3.0 268.659 7.012 · 104 9.092 · 10−1 1.694 · 10−5 2.381 · 10−2
4.0 262.166 6.166 · 104 8.193 · 10−1 1.661 · 10−5 2.329 · 10−2
5.0 255.676 5.405 · 104 7.364 · 10−1 1.628 · 10−5 2.276 · 10−2
6.0 249.187 4.722 · 104 6.601 · 10−1 1.595 · 10−5 2.224 · 10−2
7.0 242.700 4.110 · 104 5.900 · 10−1 1.561 · 10−5 2.170 · 10−2
8.0 236.215 3.565 · 104 5.258 · 10−1 1.527 · 10−5 2.117 · 10−2
9.0 229.733 3.080 · 104 4.671 · 10−1 1.493 · 10−5 2.063 · 10−2
10.0 223.252 2.650 · 104 4.135 · 10−1 1.458 · 10−5 2.009 · 10−2
12.0 216.650 1.940 · 104 3.119 · 10−1 1.421 · 10−5 1.953 · 10−2
14.0 216.650 1.417 · 104 2.279 · 10−1 1.421 · 10−5 1.953 · 10−2

B.2 Atmosphere Data

See Table B.2.

B.3 Units and Conversions

Basic and derived SI units are listed of the major flow, transport, and thermal entities.
In the left column name and symbol are given and in the right column the unit
(dimension), with → the symbol used in Appendix C, and in the line below its
conversion.

B.3.1 SI Basic Units

length, L [m], → [L]

1.0 m = 100.0 cm = 3.28084 ft
1,000.0 m = 1.0 km

mass, m [kg], → [M]

1.0 kg = 2.20462 lbm
Appendix B: Constants, Atmosphere Data, Units, and Conversions 433

time, t [s] (= [sec]), → [t]

temperature, T [K], → [T]

1.0 K = 1.8 ◦ R
⇒ TKelvin = (5/9) (TFahrenheit + 459.67)
⇒ TKelvin = TCelsius + 273.15

amount of substance, mole [kmol], → [mole]

1.0 kmol = 2.20462 lbm -mol

B.3.2 SI Derived Units

area, A [m2 ], → [L2 ]

1.0 m2 = 10.76391 ft2

volume, V [m3 ], → [L3 ]

1.0 m3 = 35.31467 ft3

speed, velocity, v [m s−1 ], → [L t −1 ]

1.0 m s−1 = 3.28084 ft s−1

force, F [N] = [kg m s−2 ], → [M L t−2 ]

1.0 N = 0.224809 lbf

pressure, p [Pa] = [N m−2 ], → [M L−1 t −2 ]

1.0 Pa = 10−5 bar = 9.86923·10−6 atm =
= 0.020885 lbf ft−2

density, ρ [kg m−3 ], → [M L−3 ]

1.0 kg m−3 = 0.062428 lbm / ft −3 =
= 1.94032·10−3 lbf s2 ft−4

(dynamic) viscosity, μ [Pa s] = [N s m−2 ], → [M L−1 t −1 ]

1.0 Pa s = 0.020885 lbf s ft −2

kinematic viscosity, ν [m2 s−1 ], → [L2 t −1 ]

1.0 m2 s−1 = 10.76391 ft2 s−1

shear stress, τ [Pa] = [N m−2 ], → [M L−1 t −2 ]

1.0 Pa = 0.020885 lbf ft −2
434 Appendix B: Constants, Atmosphere Data, Units, and Conversions

energy, enthalpy, work, [J] = [N m], → [M L2 t −2 ]

quantity of heat 1.0 J = 9.47813·10−4 BTU =
= 23.73036 lbm ft2 s−2 = 0.737562 lbf s−2

(mass specific) internal [J kg−1 ] = [m2 s−2 ], → [L2 t −2 ]

energy e, enthalpy h 1.0 m2 s−2 = 10.76391 ft2 s−2

(mass) specific heat, cv , c p [J kg−1 K−1 ] = [m2 s−2 K−1 ], → [L2 t −2 T−1 ]
specific gas constant, R 1.0 m2 s−2 K−1 = 5.97995 ft2 s−2 ◦ R−1

power, work per unit time [W] = [J s−1 ] = [N m s−1 ], → [M L2 t −3 ]

1.0 W = 9.47813·10−4 BTU s−1 =
= 23.73036 lbm ft2 s−3

thermal conductivity, k [W m−1 K−1 ] = [N s−1 K−1 ], → [M L t−3 T−1 ]

1.0 W m−1 K−1 =
= 1.60496·10−4 BTU s−1 ft −1 ◦ R−1 =
= 4.018342 lbm ft s−3 ◦ R−1

heat flux, q [W m−2 ] = [J m−2 s−1 ], → [M t−3 ]

1.0 W m−2 = 0.88055·10−4 BTU s−1 ft −2 =
= 2.204623 lbm s−3

References
1. Taylor, B.N. (ed.): The International System of Units (SI). US Dept. of Commerce, National
Institute of Standards and Technology, NIST Special Publication 330 (2001), US Government
Printing Office, Washington, D.C. (2001)
2. Taylor, B.N.: Guide for the Use of the International System of Units (SI). US Dept. of Com-
merce, National Institute of Standards and Technology, NIST Special Publication 816 (1995),
US Government Printing Office, Washington, D.C. (1995)
3. Hirschfelder, J.O., Curtiss, C.F., Bird, R.B.: Molecular Theory of Gases and Liquids. Wiley,
New York (corrected printing) (1964)
4. Bird, R.B., Stewart, W.E., Lightfoot, E.N.: Transport Phenomena, 2nd edn. Wiley, New York
(2002)
5. N.N.: U.S. Standard Atmosphere. Government Printing Office, Washington, D.C. (1976)
Appendix C
Symbols

Only the important symbols are listed. If a symbol appears only locally or infrequent,
it is not included. Dimensions are given in terms of the SI basic units: length [L],
time [t], mass [M], temperature [T], and amount of substance [mole], Appendix B.
For actual dimensions and their conversions see Appendix B.3.

C.1 Latin Letters

A, Ar e f reference area, [L2 ]

a speed of sound, [Lt−1 ]
b wing span [L]
b0 lateral distance of trailing vortex centers, [L]
CD drag coefficient [−]
C D∗ i drag coefficient at elliptical circulation distribution, [−]
CL lift coefficient [−]
Cl local lift coefficient, [−]
Cl rolling-moment coefficient, [−]
Cm pitching-moment coefficient, [−]
Cn yawing-moment coefficient, [−]
Cz local normal force coefficient, [−]
cf skin-friction coefficient, [−]
cp (mass) specific heat at constant pressure, [L2 t −2 T−1 ]
cp pressure coefficient, [−]
c pstag stagnation pressure coefficient, [−]
c pvac vacuum pressure coefficient, [−]
D diameter [L]
D drag [MLt−2 ]
Di induced drag [MLt −2 ]
Di∗ induced drag at elliptic circulation distribution, [MLt−2 ]
D0 zero-lift drag, [MLt −2 ]
© Springer-Verlag GmbH Germany, part of Springer Nature 2021 435
E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3
436 Appendix C: Symbols

e Oswald efficiency factor, [−]

e span-efficiency factor, [−]
F force [M Lt −2 ]
G normalized circulation, [−]
g gravitational acceleration, [Lt −2 ]
H altitude [L]
H shape factor, [−]
k thermal conductivity, [MLt−3 T−1 ]
L length [L]
L lift [MLt−2 ]
L∗ lift at elliptical circulation distribution, [MLt−2 ]
L/D lift-to-drag ratio [−]
M Mach number, [−]
Me boundary-layer edge Mach number, [−]
M∞ flight Mach number [−]
m mass, [M]
n exponent in boundary-layer relations, [−]
Pr Prandtl number, [−]
p pressure [ML−1 t −2 ]
pb cross-flow bluntness parameter, [−]
p∞ free-stream pressure [ML−1 t −2 ]
q heat flux, [Mt −3 ]
q∞ free-stream dynamic pressure, [ML−1 t −2 ]
R gas constant, [L2 t −2 T−1 ]
R0 universal gas constant, [ML2 t −2 mole−1 T−1 ]
R radius [L]
RN nose radius [L]
Re Reynolds number, [−]
Reu unit Reynolds number, [L−1 ]
r recovery factor, [−]
Sr Strouhal number, [−]
s entropy, [L 2 /t 2 T ]
s spanwise load factor, [−]
T temperature [T]
Te boundary-layer edge temperature [T]
Tr recovery temperature, [T]
Tt total temperature, [T]
Tw wall temperature, [T]
T∞ free-stream temperature [T]
T∗ reference temperature, [T]
t time [t]
t0 time scale, [t]
t, n, z external inviscid streamline-oriented coordinates
V magnitude of velocity vector [Lt−1 ]
V velocity vector, [−]
Appendix C: Symbols 437

u, v, w Cartesian velocity components [Lt−1 ]

u e , ve boundary-layer edge velocity, [Lt−1 ]
u ∞ , v∞ free-stream velocity, flight speed, [Lt−1 ]
vn cross-flow velocity component, [Lt −1 ]
vt stream-wise velocity component, [Lt−1 ]
Ws wing loading, [M t−2 L−1 ]
w0 downward velocity, [Lt−1 ]
xbd vortex-breakdown location, [L]
x, y, z Cartesian coordinates, [L]
x, y, z body axis coordinates, [L]
x ∗, y∗, z∗ CRM reference coordinates, [L]

x i (i  = 1, 2, 3) Cartesian reference coordinates

C.2 Greek Letters

α angle of attack [◦ ]
αV , αV , ϕV vortex-axis angles, [◦ ]
β sideslip angle [◦ ]
Γ circulation, [L2 t −1 ]
Γ0 root circulation, [L2 t −1 ]
γ flight-path angle [◦ ]
γ ratio of specific heats, [−]
Δc characteristic boundary layer thickness, [L]
δ flow boundary layer thickness [L]
δlam laminar boundary-layer thickness, [L]
δtur b turbulent boundary-layer thickness, [L]
δvs viscous sub-layer thickness, [L]
δ1 displacement thickness, [L]
δ2 momentum-flow displacement thickness, [L]
ε vortex-line angle, [◦ ]
ε surface emissivity coefficient, [−]
η f lap flap setting [◦ ]
ηslat slat setting [◦ ]
Λ aspect ratio, [−]
λ taper ratio, [−]
μ viscosity, [ML−1 t −1 ]
μe boundary-layer edge viscosity [ML−1 t −1 ]
ρ density [ML−3 ]
ρe boundary-layer edge density [ML−3 ]
ρ∞ free-stream density [ML−3 ]
τw wall shear stress, skin friction, [ML−1 t −2 ]
σ dimensionless circulation, [−]
ϕ0 sweep angle of leading edge [◦ ]
438 Appendix C: Symbols

ψe trailing-edge flow shear angle, [◦ ]

Ω vorticity-content vector, [−]
ω exponent in the power-law equations of viscosity
and heat conductivity, [−]
ω vorticity vector, [−]

C.3 Indices

C.3.1 Upper Indices

u
unit
∗
with elliptical circulation distribution
∗
reference-temperature value

C.3.2 Lower Indices

D drag
e boundary-layer edge, external (inviscid flow)
ic incompressible
inv inviscid
k thermal conductivity
L lift
L length
LE leading edge
lam laminar
ref reference
TE trailing edge
t total
tr transition
tur b turbulent
vac vacuum
vs viscous sub-layer
w wall
μ viscosity
0 leading edge
∞ infinity
Appendix C: Symbols 439

C.4 Other Symbols

O( ) order of magnitude
v, v vector
< > average
Appendix D
Abbreviations, Acronyms

AGARD Advisory Group for Aerospace Research

& Development of NATO
AIAA American Institute of Aeronautics and Astronautics
ARA Aircraft Research Association
AVA Aerodynamische Versuchsanstalt Göttingen
AVT Applied Vehicle Technology Panel
BDW Blunt Delta Wing
BL boundary layer
CFD computational fluid dynamics
CRM Common Research Model
DDES delayed detached-eddy simulation
DES detached-eddy simulation
DLR German Aerospace Center
DNS direct numerical simulation
DRG Defence Research Group of NATO
FFA National Aerospace Research Center Sweden
FOI Swedish Defence Research Agency
HISSS Higher-Order Subsonic-Supersonic Singularity (method)
HTP horizontal tail plane
ISW infinite swept wing
LE leading edge
LES large-eddy simulation
LTA large transport airplane
MBB Messerschmitt-Bölkow-Blohm GmbH
NACA National Advisory Committee for Aeronautics
NASA National Aeronautics and Space Administration
NATO North Atlantic Treaty Organization
NLR National Aerospace Laboratory of the Netherlands
NS Navier–Stokes
ONERA National Aerospace Research Center France
RAE Royal Aircraft Establishment
© Springer-Verlag GmbH Germany, part of Springer Nature 2021 441
E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3
442 Appendix D: Abbreviations, Acronyms

RANS Reynolds-averaged Navier–Stokes

RTO Research and Technology Organization of NATO
STO Science and Technology Organization of NATO
TE trailing edge
URANS unsteady Reynolds-averaged Navier–Stokes
Permissions

Figures reproduced with permission by

Aerospace Science and Technology: Figs. 10.19, 10.39 below, 10.40, 10.42, 10.43,
10.44.
Airbus: Fig. 9.6.
American Institute of Aeronautics and Astronautics: Figs. 3.4, 3.11, 3.12, 9.29, 10.3,
11.14 – 11.17.
Bernard & Graefe: Fig. 10.9.
DGLR: Figs. 10.10, 11.8 – 11.10.
DLR: Figs. 6.2, 7.2, 9.16 – 9.19.
Hanser Verlag, München: Figs. 1.1 left, 1.3, 1.5 left and right, 8.23, 9.12, 9.13, 10.5,
10.14.
Jefferys, D.: Fig. 1.1, right.
J. Comp. Physics: Fig. 3.13, 3.15.
MIT Press: Fig. 3.6.
ONERA: Fig. 7.18.
Progr. Aerospace Science: Figs. 3.16, 3.19, 8.1, 9.26 – 9.28.
Shaker Verlag, Aachen: Figs. 9.20 – 9.24.
Springer Verlag: Figs. 1.8, 2.3, 2.4, 3.2, 3.20, 3.21, 4.18, 7.5, 9.1 – 9.4, 9.7 – 9.9,
9.14, 9.15, 10.4.
uhdwallpapers: Fig. 1.7.
utzverlag, München: Figs. 1.2, 1.4, 3.9, 3.10, 3.22, 9.25.
Verlag Dr. Hut, München: Figs. 10.49 – 10.56.

The permissions to reprint all other figures have been provided directly from the
authors, see Acknowledgements at the beginning of the book.

© Springer-Verlag GmbH Germany, part of Springer Nature 2021 443

E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3
Subject Index

A Boeing 787, 186

Aerion AS2, 269 Boundary-layer thickness, 167
Air 1/7-power, 428
material properties, 431 Blasius, 428
transport properties, 425 characteristic, 429
Airbus A320, 254 extremum, 164
Airbus A350, 186 Boundary-layer velocity profile
Airbus A380, 259 laminar, 31
Approach turbulent, 31
one-domain, 29, 281 Breguet range formula, 233
three-domain, 29 Buffet, 79, 202, 366
two-domain, 29 onset boundary, 237
Area rule, 37, 41 transonic, 131, 234, 237f.
Aspect ratio Buffeting, 80f.
large, 2, 12, 35f., 85, 93, 114, 119, 127, transonic, 237
168, 174, 181f., 187, 193, 212 Bypass ratio, 248
small, 2, 12, 14, 40, 121, 186, 212
Attachment
indicator, 165 C
open-type, 152, 162, 205, 208, 344 CAST 7 airfoil, 100
point, 148f., 162, 167 Circulation
Attachment line, 15, 150f., 153, 162ff., 251 bound, 84
embedded, 149 root, 84
inviscid, 149f., 163 Circulation theory, 16
primary, 149, 162, 167, 342 Common Research Model, 133, 187, 199ff.
secondary, 149, 167, 343, 344 Concorde, 186, 270, 280, 366
tertiary, 149, 167, 343, 344 Crocco’s theorem, 58, 101
AVT-183 configuration, 329 Cross-flow
bluntness parameter, 136, 292, 317f., 340
shock, 344
B Crow instability, 68, 184, 260
B-58, 278
B-70, 278
Barotropic fluid, 53 D
Blunt Delta Wing, 339ff., 366 Decambering, 113, 159
Boeing 707, 259 boundary-layer, 22, 39, 127ff., 236
Boeing 747, 80, 259 shock-wave, 22, 39, 59, 129f., 215, 236
© Springer-Verlag GmbH Germany, part of Springer Nature 2021 445
E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3
446 Subject Index

Detachment Flow field

line, 150f., 163, 164 closed lee-side, 74, 168, 174, 287, 299, 343,
point, 150f., 166 344, 385, 390
Displacement open lee-side, 168, 174, 210, 287, 290, 299,
effect, 34 308, 309, 320, 384, 385, 388, 390
thickness, 34, 39, 131, 165, 403 Fowler flap, 246
DM-1, 277, 291, 332
Drag
equivalent reduction, 254ff. G
form, 39, 107, 120, 185, 236, 255 Galilean invariance, 29, 47
induced, 16ff., 35ff., 51, 86f., 93f., 107f., Gas constant
111f., 121, 185, 189, 212, 253ff., specific, 431
369–371, 379 universal, 431
interference, 38 Geodesic, 163
pressure, 37 Gravitational acceleration at sea level, 431
skin-friction, 37, 107, 120, 185, 255 Gripen, 68, 274, 280
transonic, 39 Ground clearance, 248
trim, 226, 255 Gurney flap, 244
viscous, 38, 107, 254
wave, 38, 236, 255, 270, 366 H
zero-lift, 39, 87, 134 HA-300, 279
Drag divergence, 37–39, 132f., 234, 237, 270 Hansen equation, 425
Drag Prediction Workshop, 133, 199 Heat flux, 153
Draken, 280 in the gas at the wall, 164, 428f.
Droop-noose device, 239 HF-24, 279
DSF 194, 278 Horse-shoe vortex, 183, 202, 247
Hypersonic flow/flight, 56f., 79, 133, 164, 186,
269, 271f., 339, 382
E
Hypersonic shadow effect, 340, 383
EFEM, 374
Entropy, 121ff.
Equivalence theorem, 37 I
Equivalent inviscid source distribution, 35 Interaction
Euler boundary layer, 102, 121, 195 global, 18, 35f., 93
Euler wake, 120, 195 shock-wave/boundary-layer, 39, 57, 131,
Eurofighter Typhoon, 7, 186, 274, 280 235ff., 299
EUROLIFT II, 249 strong, 18, 29, 35f., 39, 93, 105, 128
vortex, 73f.
weak, 34f., 38, 148
F
F-102, 278
F-104, 40, 186, 280 K
F-106, 278 Kármán vortex street, 14, 135, 150
F-106B, 373 Kolbe wing, 188ff.
F-16, 276, 366f. Krueger flap, 245
F-16XL, 276 Küchemann wing tip, 255f.
F-18, 366f. Kutta
F-35A, 186 direction, 22, 111, 127, 131, 140
F/A-18, 81 panel, 140f.
Feeding layer, 124, 308 panel free-floating, 140
Flatback airfoil, 135 Kutta condition, 16, 18, 23, 37, 52f., 109ff.,
Flight-path angle, 5 121, 127, 131, 150, 419
Flow downwash, 16ff., 35, 51, 111f., 185, 199, explicit, 17ff., 140
213, 219f., 259, 421f. implicit, 17ff., 37, 140, 292
Subject Index 447

Kutta-Joukowsky theorem, 82 O
Onglet vortex, 247
Onset flow, 176
L Orr-Sommerfeld equation, 161
Lagrangian description, 47 Oswald efficiency factor, 87
Lambda shock, 236
Laminar wing, 238
Leading edge P
aerodynamically sharp, 136ff., 150, 319, Panel method, 20, 187ff.
330 Paradox of d’Alembert, 16, 18, 108
extension, 387 Phase-plane analysis, 154
root extension, 373 Phase portrait, 148, 154f., 170
subsonic, 40, 271, 273, 277, 366 Poincaré surface, 162ff., 172, 210, 213, 287,
suction analogy, 281, 369ff. 289, 308, 320, 344, 383, 384ff.
supersonic, 271 Points-of-inflection line, 164
Leading-edge flow shear angle, 310, 314 Prandtl-Glauert rule, 42, 230
Lift Prandtl number, 426
enhancement surface, 213, 239 Pressure pmax -line, 163
non-linear, 52, 212ff., 230, 274f., 281, Pulqui II, 279
285ff., 324, 334f., 345ff., 360, 366, 370
Lift divergence, 39
R
Lifting-line model, 14, 16, 20, 46, 51, 113, 185,
Rafale, 274
253
Re-entry flight/vehicle, 57, 270, 271, 339, 382
Lift-to-drag ratio, 87, 185, 239, 243, 244, 272,
Reference area, 82
368
Airbus Gross, 82
Locality principle, 36, 111
Boeing Wimpress, 82, 199
Local wake coordinate system, 105, 111
Rolling-moment divergence, 361
Local wing stall, 248

S
M SAGITTA, 329ff.
Mach number SÄNGER, 273
critical, 11, 39 Separation, 36, 156, 166
drag-divergence, 39f., 132, 234 control, 78
pre-shock, 35, 39, 101, 130f., 214, 236f. cove, 240
subcritical, 12, 38 definition, 15
supercritical, 22 flow-off, 11ff., 16, 23, 30, 35, 109ff., 115,
Me 163, 278 127, 138, 140, 149, 159, 182, 230, 235,
MiG 21, 133 248, 281, 289, 300, 347, 388
Mini-TED, 244 incipient, 298
Mirage, 280 indicator, 165
Molecular weight, 431 open-type, 30, 152, 162f., 207, 319, 343
ordinary, 11, 12, 14, 30, 35, 149, 183, 254,
281, 289, 315, 350
N point, 15, 162
Nacelle-strake vortex, 79, 248ff. squeeze-off, 11
Non-parallel effects, 161 Separation line, 15, 148ff., 152f., 158, 160,
Normal angle of attack, 123, 291, 294, 306, 162ff., 208, 251, 342
316, 325, 331, 339, 366, 369 primary, 30, 203, 309f., 332, 343f.
Normal leading-edge Mach number, 124, 213, secondary, 203, 208, 298, 311, 334, 343f.
292, 294, 306, 316, 325, 331, 339, 366, tertiary, 208, 315
369 Separatrix, 151
No-slip wall condition, 9, 153 Sharklet, 254, 257
448 Subject Index

Shock Surface-curvature effect, 33, 161

capturing, 124 Sutherland equation, 425
fitting, 124 Symmetry break
Shock wake, 100 first, 17, 20, 56, 109, 111
Singular line, 15, 148, 153, 162ff., 203, 320, second, 17, 20, 56, 111f., 115, 304
339, 429
Singular point, 10, 11, 15, 148f., 153ff., 156f.,
159, 161, 162, 165, 170, 203, 205, 320, T
343, 429 Temperature
center, 155 adiabatic, 426
focus, 155, 159, 165 external inviscid flow, 426
half-nodal, 161 recovery, 426f.
half-saddle, 161, 168 reference, 427f.
nodal, 155, 162, 165 total, 426
quarter-saddle, 167f. wall, 164
saddle, 151, 155f., 162f. Thermal state surface extrema, 164, 298
star node, 156 Thin wing, 40, 132, 186, 269
Skin friction, 153, 427 Topology
τw -min line, 164f. skin-friction field, 148ff.
line, 149ff., 158, 162f., 165ff., 175, 191, Total pressure loss, 59, 99f., 101, 121, 122,
203ff., 214, 250ff., 289, 310ff., 332, 130f., 195, 282, 303, 308, 317, 418
342ff., 348 Trailing edge
Slat, 245 blunt, 11, 21, 113, 132ff., 141, 150, 323
Slat cutout, 249 cusp, 21, 132
Slat-horn vortex, 247 finite angle, 21, 132
Trailing-edge flow shear angle, 105, 111, 189,
Slip flow, 9
190, 194, 215, 310
Solenoidality, 48
Transonic lift dip, 237
Space Shuttle Orbiter, 186, 271, 366
Tu-144, 270
Span-efficiency factor, 87
Spanwise load factor, 84, 113
Specific excess power, 370 U
Specific impulse, 233 Unmanned Aerial System, 8, 276
SR-72, 273 Upstream effect, 35, 36
Stagnation point, 150f., 162, 167, 418
Stall
deep, 385 V
Standard atmosphere, 432 Vacuum pressure coefficient, 417
Streamline, 9f., 34, 53, 57, 107, 110f., 148, VHBR engine, 248
150f., 152, 156, 158, 159, 160f., 162, Viggen, 280
189f. Viscosity, 425
external inviscid, 97, 104 Viscous sub-layer, 429f.
limiting, 150 Vortex
surface, 163 bound/lifting, 16, 51
Strouhal number, 135 control, 78
Structural stability, 170ff. layer, 13f.
flow field, 36 lee-side, 6, 14, 36, 41, 69, 74, 77, 93, 119,
Subsonic flow/flight, 36, 187 124, 161ff., 168, 172ff., 208, 213, 249,
Suction 269ff., 359ff.
peak, 12, 289, 302, 332, 340 secondary, 36, 124, 174, 183, 206, 208f.,
pressure, 286, 289, 300, 334, 336, 369, 370 210, 230, 281, 282, 284, 289, 290, 298,
Supersonic flow/flight, 187 300f., 310, 320, 323, 346, 384, 385, 388
Supersonic flow pocket, 38, 39, 57, 98f., 129, shedding, 12, 23
141, 234f., 239, 327 starting, 16, 46, 51ff., 99
Subject Index 449

tertiary, 124, 183, 210, 230, 289, 290, 300f., kinematically active, 2, 14, 17, 22, 35f.,
349 48f., 56, 70, 96ff., 112ff., 120f., 124, 187,
trailing, 4f., 11, 14, 16, 20, 35, 46, 51f., 188, 198, 208, 216ff., 283, 291, 312f.,
63f., 66, 68f., 73, 75, 93, 113, 114, 124, 321, 345f.
181, 183, 183f., 184, 200, 210, 219, 221f., kinematically inactive, 3, 14, 17, 22, 35f.,
223, 225, 228, 230, 253, 259f., 263, 264 48f., 56, 96ff., 219f., 240, 283, 291, 321
Vortex age, 86, 222
Vortex breakdown, 14, 75f., 80, 281, 285, 288,
293, 296, 301, 306, 315, 323ff., 331ff., W
345, 349, 359f., 364, 373 Wake flow, 12, 35f., 37, 65, 68, 70, 99, 112
bubble type, 75 Wake-vortex hazard, 67, 259
spiral type, 75, 323 Wind turbine, 135
Vortex core Wing
jet-type flow, 75 forward swept, 40, 106, 162, 186, 187,
wake-type flow, 75 190f., 274
Vortex filament, 49, 62f., 73, 159, 206, 251 hybrid, 6
Vortex-Flow Experiment infinite swept, 163f.
first, 193, 283, 300, 306 ogive, 270, 366
second, 284 ONERA M6, 55
Vortex generator, 78, 243 pitch-up, 79, 293, 324, 336, 338, 361, 362,
sub-boundary-layer, 79, 243 368, 383, 385
Vortex layer root bending moment, 255, 390
trailing, 1, 11, 12, 16, 35, 51, 60, 64, 69f., root-fuselage fairing, 79, 199, 202
75, 84, 93, 105, 111ff., 119, 123f., 181, tip device, 253ff.
185, 187, 194, 199f., 213, 216, 221f., Wing flow-fields shear, 17, 105, 110, 112, 135,
223f., 253f., 259f., 281f., 305 182, 245, 253, 305, 310
Vortex line, 46, 49, 52, 61, 73f., 107, 114, 131,
Winglet, 254
152, 189, 190, 281
Wing stall, 237
Vortex-line angle, 106, 111, 131, 182, 187,
Wing-tip
189, 190, 194, 216, 229
vortex, 4, 7f., 14, 52, 75, 168, 182, 195,
Vortex model
208f.
Batchelor, 66
vortex system, 52, 113, 183, 187, 210f.,
Burger, 65
212f., 216, 230, 248, 254ff.
Burnham-Hallock, 66
Wing-tip reference point, 199
Lamb-Oseen, 65, 222
point, 49
potential, 49, 62, 64
Rankine, 49, 64f., 96, 222 X
ring, 51 X-15, 133
Vortex overlapping, 285, 293, 323, 345, 361 X-31A, 280
Vortex pairing, 73, 223 XF-92A, 278
Vortex re-configuration, 76, 323, 329
Vortex re-connection, 78
Vortex stretching, 59, 73 Y
Vortex tilting, 59 Yawing-moment divergence, 361
Vorticity content, 30, 93f., 95, 190, 210, 229 Yehudi break, 133, 182, 199, 247
Author Index

A Boltz, F.W., 232

Abbott, I.H., 144 Borst, H.V., 144
Abell, C.J., 178 Brandon, J.M., 373, 392
Ackeret, J., 144 Braza, M., 266
Allen, A., 268 Breitsamter, C., 26, 89–91, 231, 267, 268, 356,
Allmaras, S.R., 232 392
Anderson, Jr., J.D., 2, 16, 25, 37, 42, 43, 45, Brodersen, O.P., 232
89, 117, 144, 231 Brown, C.E., 281, 354
Andronov, A.A., 178 Brown, P.W., 392
Atkinson, S.A., 75, 90 Burggraf, U., 356
Burg, J.W. van der, 267
Büscher, A., 254, 268
B Busemann, A., 277, 353
Babinsky, H., 42, 266 Bush, R.H., 27
Badcock, K.J., 356 Bütefisch, K.A., 144, 355
Baker, G.R., 27
Bakker, P.G., 178
Bangga, G.S.T.A., 144 C
Bannink, W.J., 355, 356 Calvo, J.B., 42
Baron, J.R., 125 Campbell, J.F., 90, 354
Batchelor, G., 66 Carter, E.C., 178
Bate, E.J., 264, 268 Cayley, G., 272
Batemann, T.E.B., 117 Cebeci, T., 2, 25, 42
Becker, J., 91 Chambers, J.R., 90
Behrbohm, H., 280 Champion, M., 353
Behrends, K., 27, 267 Charbonnier, D., 90
Benjamin, T.B., 281, 354 Chu, J., 355
Berg, D.E., 144 Chyczewski, T.S., 27
Betz, A., 37, 55, 89 Ciobaca, V., 267
Bier, N., 27 Clareus, U., 276, 353
Bilanin, A.J., 268 Cornish, J.J., 376, 392
Bippes, H., 178 Cousteix, J., 2, 25, 42, 90, 117, 144, 177, 232,
Bird, R.B., 42, 434 267, 268, 356, 392, 415, 430
Bloor, D., 16, 26, 52, 89 Coustols, E., 268
Boddener, W., 392 Crippa, S., 232, 325, 348, 356, 357
Boelens, O.J., 27, 297, 356 Croom, D.R., 268
Bollay, W., 277, 353 Crouch, J.D., 264, 268
© Springer-Verlag GmbH Germany, part of Springer Nature 2021 451
E. H. Hirschel et al., Separated and Vortical Flow in Aircraft Wing Aerodynamics,
https://doi.org/10.1007/978-3-662-61328-3
452 Author Index

Crow, S.C., 69, 90, 264, 268 Fricke, S., 267

Cummings, R.M., 316, 356 Friedrich, O., 354
Curtiss, C.F., 434 Frink, N.T., 373, 392
Fritz, W., 301, 316, 323, 352
Fulker, J., 266
D Furman, A., 321, 356
D’Alembert, J.-B. le Rond, 16, 18, 108
Dallmann, U., 43, 148, 177, 178
Darracq, D., 268, 355 G
Davey, A., 165, 178 Gagnepain, J.-J., 353
Deck, S., 266 Ganzer, W., 294, 356
DeHaan, M.A., 232 Gebing, H., 43
Deister, F., 145 Gehri, A., 90
Délery, J., 1, 25, 42, 148, 152, 177, 232, 266 Geissler, W., 266
Dirmeier, S., 267 Gerhold, T., 232
Dixon, C.J., 376, 392 Gersdorff, K. von, 354
Donaldson, C.duP., 268 Gersten, K., 2, 25, 42, 277, 353
Donohoe, S.R., 356 Gerz, T., 268
Dovgal, A.V., 25 Geyr, H. von, 267
Drake, R.M., 429, 430 Gilbert, W.P., 392
Drela, M., 2, 25, 45, 59, 89 Goldstein, S., 10, 26
Drougge, G., 355 Gordon, I.I., 178
Duraisamy, H., 27
Görtler, H., 26, 178
Durbin, P.A., 27
Görtz, S., 77, 90
Göthert, B., 37, 43, 213, 232
Gottmann, Th., 391, 392
E
Grasjo, I., 90
Eberle, A., 26, 117, 125, 145, 283, 355, 392
Grossi, F., 266
Eckert, E.R.G., 429, 430
Groß, U., 391
Eckert, M., 26
Guillaume, M., 90
Edwards, C.L.W., 353
Gursul, I., 391
Egle, S., 392
Ehlers, T., 356
Eichelbrenner, E.A., 15, 26
Eisfeld, B., 27, 42, 232 H
Eliasson, P., 267 Haase, W., 27, 117, 267
Elle, B.J., 75, 90 Hahne, D.E., 392
Elsenaar, A., 11, 26, 117, 125, 232, 355, 356 Haines A.B., 267
Erickson, G.E., 392 Hallissy, J.B., 392
Eriksson, G., 117, 125, 232, 355 Hall, M.G., 281, 354
Eriksson, L.-E., 232, 282, 355 Hansen, C.F., 425, 430
Esquieu, S., 267 Harvey, J.K., 42, 266
Heinemann, H.-J., 392
Heinzerling, W., 354
F Heller, G., 267
Farokhi, S., 27, 43, 144, 186, 231, 266 Helmholtz, H. von, 16, 60, 76
Fassbender, J., 232 Henderson, W.P., 392
Fellows, K.A., 178 Henne, P.A., 144
Fischer, J., 100, 117 Hentschel, R., 283, 305, 355
Flaig, A., 267 Herberg, H., 43
Flügge, S., 89 Herbst, W., 354
Fornasier, L., 117, 178, 187, 232 Herr, M., 144
Foughner, J.T., 354 Hess, J.L., 27
Frenzl, O., 37 Hewitt, B.L., 232
Author Index 453

Hilbig, R., 267, 392 Kordulla, W., 2, 25, 42, 90, 117, 144, 177, 232,
Hilgenstock, A., 178 267, 356, 392, 415, 430
Hirschel, E.H., 2, , 25–27, 42, 43, 89, 90, 117, Körner, H., 26
125, 144, 145, 177–179, 232, 266, 267, Kozlov, V.V., 25
353–357, 391, 392, 415, 430 Krämer, E., 26, 144, 266, 274, 353
Hirschfelder, J.O., 434 Krause, E., 26, 76, 90, 355
Hitzel, S.M., 26, 283, 353, 355, 373, 392 Kreuzer, P., 267
Hjelmberg, L., 355 Kroll, N., 27, 232, 267
Hoarau, Y., 266 Küchemann, D., 1, 2, 25, 47, 132, 144, 186,
Hoder, H., 294, 356 231, 267, 280, 353, 354, 392
Hoeijmakers, H.W.M., 71, 90, 125, 281, Kuczera, H., 353
354–356 Kullberg, E., 90
Hoerner, S.F., 144 Kutta, M.W., 16, 26, 45, 109
Holzäpfel, F., 268
Hornung, H., 178
Horst, P., 2, 25, 26, 45, 89, 231, 267, 353 L
Hövelmann, A., 329, 356 Laflin, K.R., 232
Hünecke, K., 268 Lamar, J.E., 392
Huffman, J.K., 392 Lanchester, F.W., 16, 26, 46
Hummel, D., 74, 90, 280, 284, 289, 315, 316, Laporte, F., 231
354–357 Larsson, R., 90
Hünecke, K., 392 Lawaczeck, O., 144
Hunt, J.C.R., 166, 178 Legendre, R., 280, 354
Le Moigne, Y., 73, 90
Lengers, M., 267
I Leontovich, E.A., 178
Iovnovich, M., 266 Leschziner, M., 27
Levy, D.W., 232
Lightfoot, E.N., 42, 434
J Lighthill, M.J., 1, 25, 27, 53, 89, 103, 114, 117,
Jacob, D., 268 128, 144, 151, 165, 178
Jacquin, L., 266, 268 Linde, M., 357
Jefferys, D., 26 Lippisch, A., 37, 277
Jiràsek, A., 79, 90, 267 Lorenz-Meyer, W., 144
Johnson, H.A., 430 Lovell, J.C., 278, 291, 332, 354
Johnson, Jr., T.D., 392 Luber, W., 91
Jones, R.T., 277, 287, 354, 355, 424, 430 Lucchi, C.W., 144
Joukowski, N., 16, 26, 45 Luckring, J.M., 27, 285, 291, 297, 299, 316,
355, 356
Lüdecke, H., 356
K Ludwieg, H., 37, 277, 281, 354
Kaplan, W., 178 Ludwig, T., 90
Kármán, Th. von, 26, 202, 232 Lugt, H.J., 1, 25, 45, 54, 61, 89, 152, 178
Kawalki, K.H., 37 Lutz, Th., 144, 266
Keune, F., 37
Keye, S., 27
Klein, Ch., 356 M
Klein, F., 55, 89 Mabey, D.G., 90
Klopfer, G.H., 144 Madelung, G., 43, 89, 354
Knopp, T., 42 Maggin, B., 362, 391
Kohlmann, D.L., 356 Ma, H.-Y., 1, 25, 45, 89, 144
Kolbe, D.C., 232 Maier, A.G., 178
Kompenhans, J., 356 Malcolm, G.N., 392
Konrath, R., 356 Mandanis, G., 90
454 Author Index

Mangler, K.W., 117, 281, 354 P

Mani, M., 232 Pandolfi, M., 27
Marsden, D.J., 280, 354 Payen, N.R., 277
Maruhn, K., 117 Peake, D.J., 1, 25, 148, 152, 175, 177, 178
Masters, D., 354 Peckham, D.H., 75, 90
Matteis, P. de, 266 Pelletier, A.J., 353
Maury, B., 266 Perez, E.S., 125
Mavriplis, D.J., 232 Periaux, J., 353
Mayer, R., 266 Perry, A.E., 178
Meier, H.U., 43, 266, 353 Peterka, J.A., 178
Menter, F.R., 232 Pfnür, S., 199, 205, 231
Merazzi, S., 90 Pironneau, O., 353
Messerschmitt, W., 279 Platzer, M., 26, 178
Michael, W.H., 281, 354 Poisson-Quinton, Ph., 376
Miller, A.S., 392 Polhamus, E.C., 281, 354, 369, 392
Miller, D.S., 356 Powell, K.G., 125
Miller, G., 268 Prandtl, L., 9, 10, 15, 16, 25, 26, 46, 113, 117,
Misegades, K., 355 185, 212, 231
Mockett, C., 27, 267 Prem, H., 43, 89, 354
Probst, A., 267
Modin, K.E., 276, 353
Probst, S., 267
Moens, F., 268
Puffert, W., 117
Moffat, H.K., 177, 178
Pulliam, T.H., 27
Molton, P., 266
Morrison, J.H., 232
Mosinskis, G.J., 42 R
Müller, R., 117 Radespiel, R., 27, 267
Murayama, M., 232 Rainbird, W.J., 280, 354
Murman, E.M., 125 Raveh, D.E., 266
Reckzeh, D., 267
Redeker, G., 280, 284, 315, 354, 355
N Rider, B., 232
Nangia, R.K., 232, 373, 392 Riedelbauch, S., 178, 339, 356, 357
Newsome, R.W., 145 Rivers, S.M., 232
Nicolai, L.M., 267 Rizzi, A., 2, 26, 27, 55, 89–91, 117, 125, 145,
Niehuis, R., 27, 267 186, 231, 232, 282, 348, 355, 357, 392,
Nitsche, W., 392 415, 430
Nixon, D., 144 Rohlmann, D., 27
Norman, D., 267 Rom, J., 1, 25, 125, 355
Nowarra, H.J., 354 Rosenhead, L., 27, 89, 178
Rossow, C.-C., 2, 25, 26, 45, 89, 231, 267, 353
Rossow, V.J., 268
Rubbert, P.E., 232
O Rubesin, M.W., 430
Oberkampf, W.L, 27 Rudnik, R., 186, 231, 244, 267
Obermeier, E., 354 Rudolph, P.K.C., 267
Obert, E., 144, 186, 231 Rumsey, C.L., 27, 232
Oppelstrup, J., 2, 26, 91, 145, 186, 231, 357,
415, 430
Örnberg, T., 280, 290, 354 S
Osborne, R.F., 354 Sacher, P., 232, 353
Osterhuber, R., 373, 392 Sachs, G., 26
Oswatitsch, K., 10, 26, 37, 156, 159, 178 Sacks, A.H., 64, 70, 90, 117
Özger, E., 268 Saffmann, P.G., 27, 45, 55, 89
Author Index 455

Samuel, W.W.E., 43 T
Schade, N., 267 Tank, K., 279
Schell, I., 268 Taylor, B.N., 434
Schiavetta, L.A., 356 Thiede, P., 266
Schlichting, H., 2, 25, 42, 43, 45, 89, 117, 144, Thomas, F., 237, 266
186, 231, 392 Thomas, P., 353
Schmatz, M.A., 232 Thomson, W., 50
Schmid, A., 91 Thwaites, B., 33, 42, 139
Schmidt, W., 353 Tinoco, E.N., 232
Schneider, W., 26, 178 Tobak, M., 1, 25, 148, 152, 175, 177, 178
Schrauf, G., 231 Tomac, M., 27
Trucano, T.G., 27
Schröder, A., 356
Truckenbrodt, E., 2, 25, 43, 45, 89, 117, 144,
Schulte-Werning, B., 171, 178
186, 231, 392
Schütte, A., 356
Tsinober, A., 177, 178
Schwamborn, D., 27, 117, 178, 267 Turk, M., 178
Schwarz, W., 355
Sears, W.R., 150, 178
Sedin, Y.C.J., 68, 90 V
Sensburg, O., 90 Van Dyke, M., 36, 43, 178
Serrin, J.B., 89 Vardaki, E., 391
Shapiro, A.H., 42 Vassberg, J.C., 232
Shortal, J.A., 362, 391 Viviand, H., 178, 355
Simeonides, G., 425, 430 Vollmers, H., 178, 268
Simpson, R.W., 280, 354 Von Doenhoff, A.E., 144
Skow, A.M., 392 Vos, J.B., 90, 355
Small, W.J., 353 Vos, R., 27, 43, 144, 186, 231, 266
Smith, A.M.O., 42
Smith, B.R., 27
Smith, J.H.B., 117, 281, 354 W
Soulevant, D., 266 Wagner, W., 353
Spalart, P.R., 27, 232, 268 Wahls, R.A., 232
Spreiter, J.R., 64, 70, 90, 117 Wang, K.C., 152, 178
Squire, L.C., 294, 356 Wang, Z., 391
Stanbrook, A., 294, 356 Wanie, K.M., 232
Stanewsky, E., 117, 266 Weidner, J.P., 353
Weiland, C., 27, 145, 178, 179, 353, 391, 430
Staudacher, W., 353, 391, 392
Weis-Fogh, T., 56, 89
Stephani, P., 90
Wentz, W.H.Jr., 356
Stewart, W.E., 42, 434
Werlé, H., 75, 90, 178, 280, 354
Stivers, Jr., L.S., 144 White, F.M., 2, 25
Stouflet, B., 353 Wichmann, G., 267
Stratford, B.S., 34, 42, 139 Widnall, S.E., 64, 89
Streeter, V.L., 42 Wild, J., 267
Streit, T., 268 Wilson, H.A., 278, 291, 332, 354
Strüber, H., 267 Winkel, M.E.M. de, 178
Stumpf, E., 268 Winter, H., 277, 353
Su, W.-H., 43 Wolf, K., 2, 25, 26, 45, 89, 231, 267, 353
Sytsma, H.S., 232 Wood, R.M., 356
Szodruch, J., 294, 356 Woo, H., 178
456 Author Index

Wu, J.-Z., 1, 25, 45, 89, 144 Zayas, J.R., 144

Wulf, R., 392 Zhang, H.-Q., 43
Zhou, M.-D., 1, 25, 45, 89, 144
Z Zickuhr, T., 232
Zahm, A.F., 117 Zierep, J., 117, 144