Passive Active Ve Reactive. Flow Controlpdf

Flow Control
PASSIVE, ACTIVE, AND REACTIVE FLOW MANAGEMENT

The ability to actively or passively manipulate a ow eld to bring
about a desired change is of immense technological importance. The
potential benets of improving ow control systems range from saving
billions of dollars in fuel costs for land, air, and sea vehicles to achieving
more economically competitive and environmentally sound industrial
processes involving uid ows.
This book provides a thorough, up-to-date treatment of the basics
of owcontrol and control practices that can be used to produce desired
effects. Among topics covered are transition delay, separation preven-
tion, drag reduction, lift augmentation, turbulence suppression, noise
abatement, and heat and mass transfer enhancement. The nal chapter
explores the frontiers of ow control strategies, especially as applied to
turbulent ows.
Intended for engineering and physics students, researchers, and
practitioners, Flow Control brings together in a single source a wealth
of information on current practices and state-of-the-art developments
in this very active eld.
Mohamed Gad-el-Hak received his Ph.D. in uid mechanics from the
Johns Hopkins University in 1973. He has since taught and conducted
research at the University of Southern California, University of Virginia,
Institut National Polytechnique de Grenoble and Universit e de Poitiers,
has lectured extensively at seminars in the United States and overseas,
and is currently Professor of Aerospace and Mechanical Engineering at
the University of Notre Dame.
Professor Gad-el-Hak is world renowned for advancing several
novel diagnostic tools for turbulent ows, including the laser-induced
uorescence (LIF) technique for ow visualization; for discovering the
efcient mechanismvia which a turbulent region rapidly grows by desta-
bilizing a surrounding laminar ow; for introducing the concept of
targeted control to achieve drag reduction, lift enhancement and mix-
ing augmentation in boundary layer ows; and for developing a novel
viscous pump suited for microelectromechanical systems (MEMS) ap-
plications.
The author is a fellow of the American Society of Mechanical
Engineers, a life member of the American Physical Society, an asso-
ciate fellow of the American Institute of Aeronautics and Astronautics
and a research fellow of the American Biographical Institute.
In 1998, Professor Gad-el-Hak was named the Fourteenth ASME
Freeman Scholar. In 1999, he was awarded the prestigious Alexander
von Humboldt PrizeGermanys highest research award for senior
U.S. scientists and scholars in all disciplinesas well as the Japanese
Government Research Award for Foreign Scholars.
Flow Control
PASSIVE, ACTIVE, AND REACTIVE
FLOW MANAGEMENT
MOHAMED GAD-EL-HAK
University of Notre Dame
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, So Paulo
Cambridge University Press
The Edinburgh Building, Cambridge CB2 2RU, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521770064
Gad-el-Hak 2000
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without
the written permission of Cambridge University Press.
First published 2000
This digitally printed first paperback version 2006
A catalogue record for this publication is available from the British Library
Library of Congress Cataloguing in Publication data
Gad-el-Hak, M.
Flow control : passive, active, and reactive flow management / Mohamed Gad-el-Hak.
p. cm.
Includes bibliographical references and index.
ISBN 0-521-77006-8 (hb)
1. Fluid dynamics. 2. Hydraulic control. I. Title.
QC151.G32 2000
629.8042 dc21
99-052941
ISBN-13 978-0-521-77006-4 hardback
ISBN-10 0-521-77006-8 hardback
ISBN-13 978-0-521-03671-9 paperback
ISBN-10 0-521-03671-2 paperback
To Dilek, Kamal, and Yasemin, sine qua non
Contents
Preface page xiii
Nomenclature xv
1 Introduction 1
Prologue 1
1.1 What Is Flow Control? 1
1.2 Five Eras of Flow Control 3
1.3 Has the Field Crested? 4
2 Governing Equations 10
Prologue 10
2.1 The Fundamental Equations 10
2.1.1 Conservation of Mass 11
2.1.2 Newtons Second Law 12
2.1.3 Conservation of Energy 13
2.2 Closing the Equations 14
2.3 Compressibility 17
2.4 Equations of Motion at the Wall 20
2.5 Turbulent Flows 21
2.6 The K arm an Integral Equation 22
3 Unifying Principles 25
Prologue 25
3.1 Control Goals and Their Interrelation 25
3.2 Classication Schemes 27
3.3 Free-Shear and Wall-Bounded Flows 30
3.4 Regimes of Reynolds and Mach Numbers 31
3.5 Convective and Absolute Instabilities 33
4 Coherent Structures 34
Prologue 35
4.1 The Changing Paradigms of Turbulence 35
4.1.1 Visualization 36
4.1.2 First Principles 38
4.1.3 Statistical View of Turbulence 38
vii
viii CONTENTS
4.1.4 Reemergence of Coherent Structures 39
4.1.5 The Latest Tools 40
4.2 What Is a Coherent Structure? 42
4.3 Free-Shear Flows 42
4.4 Wall-Bounded Flows 44
4.4.1 Overview 46
4.4.2 Open Issues 53
5 Reynolds Number Effects 58
Prologue 57
5.1 The Last Conundrum 59
5.2 The Issues 61
5.2.1 Field versus Laboratory Flows 61
5.2.2 Reynolds Number 62
5.3 Flow Regimes 63
5.3.1 Viscous Region 64
5.3.2 Constant-Reynolds-Stress Region 66
5.3.3 Outer Layer 68
5.4 Comparison with Other Shear Flows 69
5.5 Mean Flow 73
5.5.1 The Rise and Fall of the Log-Law 74
5.5.2 Alternatives to the Logarithmic
Prole 77
5.5.3 The Illusory Asymptotic State 79
5.5.4 Is Self-Preservation Ever Achieved? 80
5.6 Higher-Order Statistics 81
5.6.1 Root-Mean-Square Velocity
Fluctuations 82
5.6.2 Reynolds Stress 85
5.7 Coherent Structures 90
5.7.1 Bursting Period 90
5.7.2 High Reynolds Number 92
5.7.3 Small Structures in Outer Layer 94
5.7.4 Inner Structures 94
5.8 Flow Control 97
5.8.1 Introductory Remarks 97
5.8.2 Riblets 97
5.8.3 Recovery Response 98
5.8.4 Control of High-Reynolds-Number Flows 101
5.9 Summary of Reynolds Number Effects 101
6 Transition Control 104
Prologue 104
6.1 Free-Shear versus Wall-Bounded Flows 104
6.2 Routes to Transition 105
6.3 Linear Stability Theory 108
6.3.1 Critical Re and Transition 109
CONTENTS ix
6.4 Stability Modiers 110
6.4.1 Wall Motion 111
6.4.2 Suction 112
6.4.3 Shaping 114
6.4.4 Wall Heating and Cooling 116
6.5 Wave Cancellation 118
7 Compliant Coatings 120
Prologue 120
7.1 Introduction 120
7.2 Compliant Coatings prior to 1985 122
7.3 Free-Surface Waves 125
7.4 System Instabilities 127
7.5 The Kramer Controversy 129
7.6 Transitional Flows 131
7.6.1 Linear Stability Theory 131
7.6.2 Coating Optimization 133
7.6.3 Practical Examples 139
7.6.4 The Dolphins Secret 140
7.7 Turbulent Wall-Bounded Flows 141
7.8 The Future 145
7.9 Parting Remarks 148
8 Separation Control 150
Prologue 150
8.1.1 The Phenomenon of Separation 150
8.1.2 Separation Control 151
8.1.3 Method of Control 152
8.1.4 Outline 153
8.2 Steady and Unsteady Separation 153
8.2.1 Steady Separation 153
8.2.2 Unsteady Separation 154
8.2.3 Vortex-Induced Separation 158
8.3 Equations of Motion at the Wall 161
8.3.1 Equation for Turbulent Flows 161
8.4 Velocity Prole Modiers 163
8.4.1 Shaping 163
8.4.2 Transpiration 165
8.4.3 Wall Heat Transfer 168
8.5 Moving-Wall and Time-Dependent
Separations 168
8.5.1 Moving Walls 168
8.5.2 Time-Dependent Separation 170
8.6 Three-Dimensional Separation 172
8.7 Turbulators 172
8.8 Momentum Addition to Near-Wall Flow 174
x CONTENTS
8.8.2 Passive Vortex Generators 176
8.8.3 Passive Turbulence Amplication 177
8.8.4 Active Wall-Region Momentum Addition 178
8.9 Separation Provocation 183
8.9.1 Delta Wings 183
8.10.1 Recapitulation 185
8.10.2 Computational Fluid Dynamics 186
8.10.3 Applications and Comparisons 187
8.10.4 Recommendations 188
9 Low-Reynolds-Number Aerodynamics 189
Prologue 189
9.2 Low-Reynolds-Number Airfoils 191
9.3 Conditions for Bubble Formation 192
9.4 Bubbles Breakdown 194
9.5 Control Goals and Their Interrelation 195
9.6 Separation Control 197
9.7 Transition Control via Shaping 198
9.8 Turbulators 202
10 Drag Reduction 205
Prologue 205
10.1 Background 206
10.2 Drag Reduction 209
10.3 Skin-Friction Reduction 210
10.4 Reduction of C
f
in Turbulent Flows 212
10.4.1 Reduction of Near-Wall Momentum 212
10.4.2 Introduction of Foreign Substance 214
10.4.3 Methods Involving Geometry 218
10.4.4 Recent Innovations 222
10.5 Relaminarization 223
10.6 Synergism 225
11 Mixing Enhancement 229
Prologue 229
11.2 Mixing 231
11.2.1 Kinematical Description 233
11.2.2 Enhancement Strategies 235
11.3 Early Transition 235
11.3.1 Wall-Bounded Flows 236
11.3.2 Free-Shear Flows 237
CONTENTS xi
11.4 Turbulence Enhancement 238
11.4.1 Turbulent Wall-Bounded Flows 239
11.4.2 Turbulent Free-Shear Flows 239
11.5 Heat Transfer 240
11.6 Chaotic Mixing 244
12 Noise Reduction 250
Prologue 250
12.2 Fundamentals of Acoustics 252
12.3 Sources of Sound 255
12.4 Noise Control 259
12.4.1 Relation to Other Flow-Control Goals 259
12.4.2 Passive Noise Control 261
12.4.3 Compliant Coatings 263
12.4.4 Jet Noise 264
12.4.5 Turbomachinery Blades 266
12.4.6 Antisound 267
13 Microelectromechanical Systems 269
Prologue 269
13.2 Flow Physics 272
13.2.1 Fluid Mechanics Issues 272
13.2.2 Fluid Modeling 273
13.2.3 Continuum Model 277
13.2.4 Compressibility 280
13.2.5 Boundary Conditions 282
13.2.6 Molecular-Based Models 289
13.2.7 Liquid Flows 299
13.2.8 Surface Phenomena 303
13.3 Typical Fluid Applications 308
13.3.2 Turbulence Measurements 308
13.3.3 Micropumps 310
13.3.4 Microturbines 312
13.3.5 Microbearings 313
14 Frontiers of Flow Control 318
Prologue 318
14.1.1 The Taming of the Shrew 318
14.1.2 Control of Turbulence 319
14.1.3 Outline 320
14.2 Suction 320
14.3 Coherent Structures 322
xii CONTENTS
14.4 Scales 324
14.4.1 Sensor Requirements 327
14.5 Reactive Control 329
14.5.2 Targeted Control 330
14.5.3 Reactive Feedback Control 334
14.5.4 Required Characteristics 336
14.6 Magnetohydrodynamic Control 338
14.6.2 The von K arm an Equation 339
14.6.3 Vorticity Flux at the Wall 340
14.6.4 EMHD Tiles for Reactive Control 341
14.7 Chaos Control 343
14.7.1 Nonlinear Dynamical Systems Theory 343
14.7.2 Chaos Control 345
14.8 Microfabrication 349
14.9 Soft Computing 352
14.9.1 Neural Networks for Flow Control 355
Epilogue 359
Bibliography 363
Index 417
Preface
Words are food for thought; utter the right ones and you will have created a delicacy
to be savored. And so will begin every chapter in this book with the rich words
spoken through the ages by the wisest of all men.
The ability to actively or passively manipulate a oweld to effect a desired change is
of immense technological importance, and this undoubtedly accounts for the subjects
being hotly pursued at present by more scientists and engineers than any other topic in
uid mechanics. The art of owcontrol is as old as prehistoric man, whose sheer per-
severance resulted in the invention of streamlined spears, sickle-shaped boomerangs,
and n-stabilized arrows. The German engineer Ludwig Prandtl pioneered the sci-
ence of ow control at the beginning of the twentieth century.
The potential benets of realizing efcient ow-control systems range fromsaving
billions of dollars in fuel cost for land, air, and sea vehicles to achieving economically
and environmentally more competitive industrial processes involving uid ows. The
purpose of this book is to provide an up-to-date view of the fundamentals of some
basic ows and control practices that can be employed to produce needed effects.
Understanding of some basic mechanisms in free and wall-bounded turbulence has
increased substantially in the last fewyears. This understanding suggests that taming
of turbulencethe quintessential challenge in the eld of ow controlis possible
so as to eliminate some of its deleterious effects while enhancing its useful traits.
The present monograph covers current ideas and methods for effecting such con-
trol to achieve transition delay, separation prevention, drag reduction, lift augmenta-
tion, turbulence suppression, noise abatement, and heat and mass transfer (including
combustion) enhancement in both free-shear and wall-bounded ows. Futuristic
ow-control strategies will be explored as well. Sophisticated reactive control sys-
tems consisting of a colossal number of microsensors and microactuators communi-
cating with each other via massively parallel computers will in the future improve, by
leaps and bounds, the uids engineers ability to control the all too common turbulent
ows.
Speaking of turbulence, some readers may rightly sense that this particular topic
is getting more than its fair share of coverage. The argument is often made that
laminar ows are easier to control and that every effort should be made to control
the ow at this stage before it is too late. I could not agree more. However, there are
no known strategies that could delay transition (provided that is what one wants to
xiii
xiv PREFACE
accomplish) beyond a Reynolds number, based on freestream velocity and distance
from leading edge, of Re
x
= O[10
7
]. The boundary layer ow around the fuselage
of a large commercial airplane reaches that Reynolds number after only 1 m, which
leaves more than 50 m to go. So turbulence for better or worse is there to stay and
must be dealt with whatever the difculties.
The present book is intended for engineers, physicists, educators, and graduate
students at universities and research laboratories in government and industry who
have an interest in the physics and control of laminar, transitional, and turbulent
ows. The different chapters here are written in a pedagogic style and are designed
to attract newcomers to the eld. A knowledge of basic uid dynamics is assumed,
and the book should help the reader in navigating through the colossal literature
available on ow-control fundamentals and practices.
The book is organized into fourteen chapters as follows: The subject is intro-
duced and the important governing equations are derived in the rst two chapters.
Chapter 3 provides the unifying principles for the broad eld of ow control. Coher-
ent structures and Reynolds number effects are reviewed in the next two chapters.
Chapter 6 introduces aspects of transition control, and compliant coatings are dis-
cussed in Chapter 7. The all-important separation control and low-Reynolds-number
aerodynamics are covered in Chapters 8 and 9. The technologically very important
drag reduction is reviewed in Chapter 10. Mixing enhancement and noise reduc-
tion are briey discussed in Chapters 11 and 12. Microelectromechanical systems
(MEMS) is an emerging eld that offers unique possibilities for fabricating inexpen-
sive, communicative, dense arrays of microsensors and microactuators for use in
reactive ow control. Both ow physics and typical applications of MEMS are cov-
ered in Chapter 13. The nal chapter explores the frontiers of owcontrol, including
futuristic ideas for achieving a leap in the performance of control systems.
Hans Wolfgang Liepmann, my academic grandfather, wrote in a recent retrospec-
tive of the early days of research on boundary-layer transition (Applied Mechanics
Reviews, vol. 50, no. 2, pp. R1R4, 1997): The importance of asking the right
question is as obvious in principle as it is nebulous in research. Always present is the
temptation to look for the key under the street light and not in the dark corner where
it was lost. Indeed, in my opinion, much of the extensive turbulence research today
follows this line and fails to address the real problems. How true those words of
wisdom are to the eld of ow control and especially to the difcult task of taming
turbulence. The author of the present book aspires for the right questions to be asked
and for the answers to be sought even in the darkest corners.
Mohamed Gad-el-Hak
Notre Dame, Indiana
1 January 2000
Nomenclature
Few of the symbols below are used to indicate more than one variable or parameter.
This is done to preserve long-held conventions. For example, c symbolizes both
chord length and wave speed. The proper usage is clearly indicated when the symbol
appears in the text.
Non-Greek Symbols
A control surface
|A| wave amplitude
a channel half-width; also pipe radius; also rotor radius
a
o
speed of sound
B second constant in the log-law
Br Brinkman number (Ec Pr)
C Clausers outer-layer shape parameter
C
D
drag coefcient (2D,
U
2
c; per unit span)

C
f
local skin-friction coefcient (2
w
,
o
U
2
o
)
C
L
lift coefcient (2L,
U
2
c; per unit span)

C
q
suction coefcient (|v
w
|,U
o
)
c airfoils chord; also complex wave speed
c
i
imaginary part of c (determines temporal growth of disturbance)
c
propagation velocity for longitudinal waves

c
p
specic heat at constant pressure
c
r
disturbance phase velocity (real part of c)
c
t
propagation velocity for transverse waves
c
v
specic heat at constant volume
c
+
x
streamwise phase velocity in wall units
D drag; also damping coefcient
De Dean number ((Re,2)(
a,R))
D
K.E.
dissipation of turbulence kinetic energy
d depth
dE energy ux
E elastic modulus
Ec Eckert number (v
2
o
,c
p
LT)
e internal energy per unit mass
F external forcing term
xv
xvi NOMENCLATURE
F exural rigidity of a compliant surface; also probability distribution
F external force vector
F
i j
force eld between two molecules i and j
F
(T) dimensionless function that characterizes viscosity variation with

temperature
f frequency; also normalized velocity distribution
f
(o)
Maxwellian velocity distribution
G rst elastic constant (shear modulus of rigidity)
Gr Grashof number (g LT L
3
,
2
)
g acceleration of gravity
g
i
component of body force per unit mass
H shape factor (ratio of displacement thickness to momentum thickness)
h riblet height; also channel depth; also convective heat transfer coefcient
i imaginary number (i =
1); also used as a sufx in tensor notations

J ( f. f
) collision integral
Kn Knudsen number (L,L)
k spring constant; also Boltzmann constant
k
longitudinal wave number

L lift; also characteristic ow dimension
L mean free path of gas molecules
L
s
slip length; also characteristic dimension of solid
L
o
s
limiting value of slip length
probe length; also mixing length
viscous length scale (=,u
+
probe length in wall units (=,
)
M( ) dispersion relation
Ma Mach number (v
o
,a
o
)
m mass; also molecular mass
N number of molecules
Nu Nusselt number (dimensionless temperature gradient at the surface)
n modal amplication; also exponent in velocity power-law; also number
density
n outward unit normal vector
P bursting period
P sensoractuator power consumption
P mean pressure
P
K.E.
production of turbulence kinetic energy
Pe P eclet number (v
o
L,
T
= Pr Re)
Pr Prandtl number (,
T
)
p thermodynamic pressure
p
i
inlet pressure
p
o
outlet pressure
p
pressure uctuations
q
2
twice turbulence kinetic energy
q
k
heat ux vector (k = 1. 2. 3)
q
x
tangential heat ux
q
y
normal heat ux
R radius of curvature
NOMENCLATURE xvii
R gas constant
Ra Rayleigh number (=Gr Pr)
Re Reynolds number (v
o
L,)
Re
a
Reynolds number based on centerline velocity and pipe radius
Re
c
Reynolds number based on chord and freestream velocity
Re
crit
critical Reynolds number
Re
x
Reynolds number based on distance from leading edge and freestream
velocity
Re
ratio of boundary-layer thickness (or pipe radius) to viscous length scale

r
c
cut-off distance
S slip factor

2
v
v
Kn
St Strouhal number (f L,v

o
)
s riblet spanwise spacing; also entropy
T temperature
T period of forcing signal; also time constant for temperature change
T
gas
gas temperature at wall
T
longitudinal tension of thin plate

T
o
reference temperature
T
w
wall temperature
T
ambient temperature
t time
t
viscous time-scale (given by the inverse of mean wall vorticity)

U upper wall speed in a Couette cell
U time-averaged streamwise velocity; also bulk velocity
U(y) unidirectional laminar ow
U
c
centerline velocity
U
i
mean velocity component
U
o
(x. t) velocity outside boundary layer
U
freestream velocity
u velocity vector
u
uid
uid velocity at wall
u
k
instantaneous velocity component (u. v. w)
u
k
velocity uctuations
u
wall
wall velocity
u
viscous velocity scale; also called friction velocity (=
,t
w
,)
u
another symbol for friction velocity

u
i
u
k
(kinematic) Reynolds stress
uv tangential Reynolds stress
V control volume
V time-averaged normal velocity
V(r) potential energy
V
i j
(r) potential between two molecules i and j
v
m
mean molecular speed
v
w
injectionsuction velocity; also upwarddownward wall velocity
v
o
characteristic velocity
W(
y
) universal wake function

x position vector
x
k
Cartesian coordinates (k = 1. 2. 3)
xviii NOMENCLATURE
x
i
spatial coordinate
y
+
distance normal to surface in wall units
y
+
p
location of peak Reynolds stress (expressed in wall units)
Greek Symbols
angle of attack; also complex wave number; also isothermal
compressibility coefcient
rate of change of angle of attack (=
d
dt
)
i
imaginary part of (determines spatial growth of disturbance)
r
wave number (real part of )
coefcient of thermal expansion (also called bulk expansion coefcient)
I circulation
specic heat ratio (isentropic exponent)
(x) local circulation in a vortex sheet per unit length
shear rate

c
critical value of shear rate
Lp
pressure drop normalized using a viscous pressure scale (Lp4a

2
,
2
)
LT characteristic temperature difference
Lu|
w
tangential velocity slip
LU
+
maximum deviation of mean velocity prole from log-law (in wall units)
Lx center-to-center distance between discrete vortices
boundary-layer thickness; also distance between molecules
ki
Kronecker delta (unit second-order tensor)
momentum thickness
displacement thickness
c characteristic energy scale
c
wf
energy scale for walluid interaction
normalized rotor eccentricity
ratio of surface wave speed to transverse wave speed
compliant coating displacement in y-direction; also Kolmogorov length
scale
O second elastic constant
von K arm an constant
f
thermal conductivity of uid
s
thermal conductivity of solid
A Pohlhausen parameter (ratio of pressure forces to viscous forces)
second coefcient of viscosity
+
z
spanwise mean-streak-spacing (in wall units)
j rst coefcient of viscosity
kinematic viscosity (=j,)
T
thermal diffusivity (=
f
,c
p
)
compliant coating displacement in x-direction
molecular velocity vector
r
relative speed between two molecules
H prole parameter
uid density
NOMENCLATURE xix
s
solid density
ki
stress tensor (surface force per unit area)
molecular diameter (molecular length scale)
T
thermal accommodation coefcient
v
tangential-momentum-accommodation coefcient
wf
length scale for wall-uid interaction
shear stress; also molecular time scale; also tangential momentum ux
w
wall shear stress
characteristic roughness height
dissipation function
dissipation function for incompressible ows

(y) perturbation amplitude-function in the OrrSommerfeld equation
stream function
O mean vorticity
O
z
mean spanwise vorticity
radian frequency (=2 f ); also rotational speed
i
instantaneous vorticity component
Subscripts
0 conditions at y = 0
i imaginary part; also incident variable
longitudinal direction
o conditions outside boundary layer; also reference quantity
r real part; also reected variable
rms root mean square
s solid variable
t transverse direction
w conditions at wall
far-eld or ambient conditions
Superscripts
turbulence uctuations; also derivative w.r.t. (y,)
+ variable expressed in wall units
dimensionless variable; also postcollision value

dimensionless variable
Abbreviations
AFOSR (U.S.) Air Force Ofce of Scientic Research
AGARD (NATO) Advisory Group for Aerospace Research and Development
AIAA American Institute of Aeronautics and Astronautics
AICHE American Institute of Chemical Engineers
AIP American Institute of Physics
xx NOMENCLATURE
ARC (British) Aeronautical Research Council
ARO (U.S.) Army Research Ofce
ASME American Society of Mechanical Engineers
CFD computational uid dynamics
DNS direct numerical simulations
DOD (U.S.) Department of Defense
DOE (U.S.) Department of Energy
FISI ow-induced surface instabilities
ICAS International Council of the Aeronautical Sciences
IEEE Institute of Electrical and Electronics Engineers
IOP (British) Institute of Physics
LEBU large-eddy breakup device
LES large-eddy simulations
LFC laminar ow control
MEMS microelectromechanical systems
MRS MooreRottSears (criterion)
NACA (U.S.) National Advisory Committee for Aeronautics
NASA (U.S.) National Aeronautics and Space Administration
NATO North Atlantic Treaty Organization
NLF natural laminar ow
NSF (U.S.) National Science Foundation
OGY OttGrebogiYorke (strategy for controlling chaos)
OLD outer-layer device
ONR (U.S.) Ofce of Naval Research
SAE (U.S.) Society of Automotive Engineers
TSI TollmienSchlichting instabilities
TWF traveling-wave utter
CHAPTER ONE
Introduction
Thinking is one of the greatest joys of humankind.
(Galileo Galilei, 15641642)
The farther backward you can look, the farther forward you are likely to
see.
(Sir Winston Leonard Spencer Churchill, 18741965)
PROLOGUE
The subject of ow control is broadly introduced in this rst chapter, leaving much
of the details to the subsequent chapters of the book. The ability to manipulate a
oweld actively or passively to effect a desired change is of immense technological
importance, and this undoubtedly accounts for the subjects being more hotly pur-
sued by scientists and engineers than any other topic in uid mechanics. The potential
benets of realizing efcient ow-control systems range from saving billions of dol-
lars in annual fuel costs for land, air, and sea vehicles to achieving economically and
environmentally more competitive industrial processes involving uid ows. In this
monograph both the classical tools and the more modern strategies of owcontrol are
covered. Methods of control to achieve transition delay, separation postponement,
lift enhancement, drag reduction, turbulence augmentation, and noise suppression
are considered. The treatment is tutorial at times, which makes the material accessible
to the graduate student in the eld of uid mechanics. Emphasis is placed on external
boundary-layer ows, although applicability of some of the methods discussed for
internal ows as well as free-shear ows will be mentioned. An attempt is made to
present a unied view of the means by which different methods of control achieve a
variety of end results. Performance penalties associated with a particular method
such as cost, complexity, and trade-off will be elaborated upon throughout the
book.
1.1 What Is Flow Control?
Before we get started it might be a good idea to explain what is meant by ow con-
trol. The topic, as discussed in this monograph, is not related to ow-rate control
1
2 INTRODUCTION
via manual or automatic valves. Rather it is the attempt to alter the character or
disposition of a oweld favorably that is of concern to us here. Many denitions
for the topic exist, and some differentiate between (active) ow control and (passive)
ow management. My favorite denition was offered by Flatt
1
in 1961 as applied
to wall-bounded ows but could easily be extended to free-shear ows: Bound-
ary layer control includes any mechanism or process through which the boundary
layer of a uid ow is caused to behave differently than it normally would were
the ow developing naturally along a smooth straight surface. Prandtl (1904) pi-
oneered the modern use of ow control in his epoch-making presentation to the
Third International Congress of Mathematicians held at Heidelberg, Germany. In
just eight pages (as required for acceptance by the congress) of a paper titled

Uber
Fl ussigkeitsbewegung bei sehr kleiner Reibung (On Fluid Motion with Very Small
Friction), Prandtl introduced the boundary-layer theory, explained the mechanics of
steady separation, opened the way for understanding the motion of real uids, and
described several experiments in which the boundary layer was controlled.
Prandtl (1904) used active control of the boundary layer to show the great inu-
ence such a control exerted on the ow pattern. He used suction to delay boundary-
layer separation from the surface of a cylinder. Notwithstanding Prandtls success,
aircraft designers in the three decades following his convincing demonstration were
accepting lift and drag of airfoils as predestined characteristics with which no man
could or should tamper (Lachmann 1961). This predicament changed mostly owing
to the German research in boundary-layer control pursued vigorously shortly before
and during the Second World War. In the two decades following the war, extensive
research on laminar ow control, in which the boundary layer formed along the ex-
ternal surfaces of an aircraft is kept in the low-drag laminar state, was conducted in
Europe and the United States, culminating in the successful ight-test program of the
X-21 in which suction was used to delay transition on a swept wing up to a chord
Reynolds number of 4.7 10
7
. The oil crisis of the early 1970s brought renewed in-
terest in novel methods of ow control to reduce skin-friction drag even in turbulent
boundary layers. In the 1990s, the need to reduce the emissions of greenhouse gases
and to construct supermaneuverable ghter planes, faster and quieter underwater
vehicles, and hypersonic transport aircraft (e.g., the U.S. National Aerospace Plane)
provided new challenges for researchers in the eld of ow control.
The major goal of this monograph is to introduce the subject broadly, to present
a unied viewof the different control methods to achieve a variety of end results, and
to provide an up-to-date view of the fundamentals of some basic ows and control
practices. The number of journal articles and proceedings available on ow control
is daunting. For example, a recent bibliography on the rather narrow sub-subject of
skin-friction reduction by polymers and other additives cites over 4900 references
(Nadolink and Haigh 1995). Therefore, citations in the present book will be rather
selectiveindeed very limited. To avoid having all 441 pages of this book as a gigan-
tic list of references, numerous good articles will not be cited. At times, the treatment
of the subject matter herein is pedagogical and is designed to attract newcomers to
the eld and to help the reader in navigating through the colossal literature available
on ow-control fundamentals and practices. Emphasis is placed on the technolog-
ically important external boundary-layer ows, although applicability of some of
1
All references are listed alphabetically in the bibliography at the end of the book.
1.2 FIVE ERAS OF FLOW CONTROL 3
the methods discussed for internal ows as well as free-shear ows will be covered
as appropriate. The same vorticity considerations brilliantly employed by Lighthill
(1963a) to place the boundary layer correctly in the ow as a whole are used to
explain many of the ow-control techniques discussed in this book. The history of
the subject is briey and broadly recalled in the following section.
1.2 Five Eras of Flow Control
Flow control involves passive or active devices to effect a benecial change in wall-
bounded or free-shear ows. Whether the task is to delay or advance transition,
to suppress or enhance turbulence, or to prevent or provoke separation, useful end
results include drag reduction, lift enhancement, mixing augmentation, and ow-
induced noise suppression. Broadly, there are perhaps ve distinct eras in the devel-
opment of the art and science of this challenging albeit very useful eld of research
and technology: the empirical era (prior to 1900), the scientic era (19001940), the
World War II era (19401970), the energy crisis era (19701990), and the 1990s and
beyond.
The art of ow control probably has its roots in prehistoric times when stream-
lined spears, sickle-shaped boomerangs, and n-stabilized arrows evolved empirically
(Williams 1987) by the sheer perseverance of archaic Homo sapiens who knew noth-
ing about air resistance or aerodynamic principles. Three aerodynamically correct
wooden spears were recently excavated in an open-pit coal mine near Hannover,
Germany (Dennell 1997). Archaeologists dated the carving of those complete spears
to about 400,000 years ago (Thieme 1997), which strongly suggests early Stone Age
ancestors possessing resourcefulness and skills once thought to be characteristics that
came only with fully modern Homo sapiens.
Modern man also artfully applied ow-control methods to achieve certain techno-
logical goals. Relatively soon after the dawn of civilization and the establishment of an
agricultural way of life 8000 years ago, complex systems of irrigation were built along
inhabited river valleys to control the waterow, thus freeing man from the vagaries
of the weather. Some resourceful albeit mischievous citizens of the Roman Empire
discovered that adding the right kind of diffuser to the calibrated convergent nozzle
ordinarily installed at home outlets of the public water main signicantly increased
the charge of potable water over that granted by the emperor. For centuries, farmers
knew the value of windbreaks to keep topsoil in place and to protect fragile crops.
The science of ow control originated with Prandtl (1904), who introduced the
boundary-layer theory, explained the physics of the separation phenomena, and de-
scribed several experiments in which a boundary layer was controlled. Thus, the
scientic method to control a oweld, or the second era of ow control, was born.
Slowly but surely, the choice of ow-control devices was no longer a trial and error
feat, but physical reasoning and even rst principles were more often than not used
for rational design of such artifacts.
Stimulated by the Second World War and the subsequent Cold War, that trend
accelerated signicantly during the third era (19401970). Military needs of the
superpowers dictated the development of fast, highly maneuverable, efcient aircraft,
missiles, ships, submarines, and torpedoes, and ow control played a major role
in achieving these goals. Natural laminar ow (in which shaping is used to delay
4 INTRODUCTION
transition), laminar owcontrol (in which nonpassive strategies are employed to keep
the owlaminar), and polymer drag reduction were notable achievements during this
era. Partial summaries of ow-control research during this period are presented in
the books edited by Lachmann (1961) and Wells (1969).
The energy crises exemplied by the 1973 Arab oil embargo brought about a
noticeable shift of interest from the military sector to the civilian one. During the
period 19701990, government agencies and private corporations around the world,
but particularly in the industrialized countries, invested valuable resources in the
search for methods to conserve energy, and hence drag reduction for civilian air, sea,
and land vehicles; for pipelines; and for other industrial devices was emphasized. The
availability of fast, inexpensive computers made it possible to simulate numerically
complex ow situations that had not been approachable analytically. Some control
strategies, for example, transition-delaying compliant coatings (Gad-el-Hak 1996a),
were rationally optimized using computational uid dynamics. Large-eddy breakup
devices (LEBUs) and riblets are examples of control methods developed during this
period to reduce skin-friction drag in turbulent boundary layers. Good sources of
information on these and other devices introduced during the fourth era are the books
edited by Hough (1980); Bushnell and Hefner (1990); and Barnwell and Hussaini
(1992). Numerous meetings devoted to ow control, particularly drag reduction,
were held during this period. Plentiful fuel supplies during the 1990s and the typical
short memory of the long gas lines during 1973 have, unfortunately, somewhat dulled
the urgency and enthusiasm for energy conservation research as well as practice.
Witnessat least in the United Statesthe awakening of the long-hibernated gas-
guzzler automobile and the recent run on house-size sport utility vehicles, a.k.a. land
barges.
For the 1990s and beyond, more complex reactive control devices geared specif-
ically toward manipulating the omnipresent coherent structures in transitional and
turbulent shear ows (Cantwell 1981; Robinson 1991) are being pursued by several
researchers. Theoretical advances in chaos control and the development of micro-
electromechanical systems (MEMS) and neural networks should help such efforts.
Articles specically addressing reactive control strategies include those by Wilkinson
(1990); Moin and Bewley (1994); and Gad-el-Hak (1994, 1996b). The MEMS are
discussed in Chapter 13, and reactive ow control is emphasized in the last chapter
of this book.
All ve eras of ow control are seen from the perspective of the history of the
universe timeline shown in Figure 1.1.
1.3 Has the Field Crested?
There are those who consider uid mechanics to be a mature subject that led to very
useful technological breakthroughs in the past but that the pace of improvements is
fast reaching the point at which returns on investment in research are not sufciently
impressive. As we approach the twenty-rst century, the skeptics claim, little new
scientic or engineering breakthroughs are to be expected from the aging eld of
study. It may be worth remembering that much the same was said about physics
toward the end of the nineteenth century. Self-satised that almost all experimental
observations of the time could be tted into either Newtons theory of mechanics
1.3 HAS THE FIELD CRESTED? 5
Figure 1.1: History of the universe timeline. All dates are approximate, and the
timescale is highly nonlinear.
or Maxwells theory of electromagnetics, the majority of physicists believed that the
work of their successors would be merely to make measurements to the next decimal
place. That was just before the theory of relativity and quantum mechanics were
discovered!
6 INTRODUCTION
Technology has its share of amusing anecdotes as well. In the 1860s, Abraham
Lincolns commissioner of patents recommended that the commission be closed in
a few years because the rate of discovery had become so great that everything that
needed to be discovered would have been discovered by then. The patent commission
would simply have no future business. Opportunity is dead! All possible inventions
have been invented. All great discoveries have been made. In 1899before the
airplane, laser, and computer were inventedthe commissioner of the U.S. Ofce of
Patents, Charles H. Duell, urged President McKinley to abolish this ofce because
everything that can be invented has been invented. Foolish, fallacious statements
like those are frequently attributed to various myopic patent ofcials of the past and
are perpetuated even by the most respected writers and speakers of our time. We cite
here three of the most recent perpetuators who all of course wanted only to show
how ignorant and unimaginative the hapless patent ofcer must have been: Daniel
E. Koshland, Jr., the editor-in-chief of the periodical Science, in an editorial on the
future of its subject matter (Koshland 1995); the cyberseer and megaentrepreneur
Bill Gates in the hardcoverbut not the paperbackedition of the instant best-seller
The Road Ahead (Gates, Myhrvold, and Rinearson 1995); and the president of the
National Academy of Sciences, Bruce Alberts, in a fund-raising letter dated April
1997, widely distributed to friends of science in the United States. The denitive
history of the preceding and related apocryphal anecdotes was documented by Eber
Jeffery, who in 1940 conducted an exhaustive investigation of their authenticity and
origin. Jeffery traced the then widely circulated tales to a testimony delivered before
the United States Congress by Henry L. Ellsworth, the commissioner of patents in
1843, who told the lawmakers that the rapid pace of innovation taxes our credulity
and seems to presage the arrival of that period when human improvement must end.
According to Jeffery (1940), this statement was a mere rhetorical ourish intended
to emphasize the remarkable strides forward in inventions then current and to be
expected. Indeed, Commissioner Ellsworth asked the Congress to provide him with
extra funds to cope with the ood of inventions he anticipated. Jeffery concluded that
no document could be found to establish the identity of the mysterious commissioner,
examiner, or clerk who thought that all invention was a thing of the past. This was
not true then and is certainly not true now, for both science and technology have
indeed an endless frontier (Bush 1945; Petroski 1997).
Our own subeld of ow control provides a good rebuttal to the uid mechanics
critics. How much resources are spent annually to overcome the drag of air, water,
and land vehicles and the uid resistance in pipelines, ducts, and countless other
man-made systems? How many airplane crashes could have been averted if lifting
surfaces more resistant to stall had been designed? Can ghter aircraft achieve
supermaneuverability? How much pollution is generated as a result of incomplete,
inefcient combustion in furnaces, incinerators, internal combustion engines, gas
turbines, and numerous other reactors? As mandated by the 1997 Kyoto Protocol, by
the year 2012 the industrialized countries must reduce their emissions of greenhouse
gases 68% below those produced in 1990. How can they achieve this target? How
much does it cost to y across continents at hypersonic or even supersonic speeds?
Can we economically utilize ultrafast submarines, instead of snail-paced ships, for
hauling people and commerce across the seas? For transport through urban areas
whose already congested roads have no more room for growth, can we envision a
personal helicopter (skycar) that, aided by a sophisticated network of computers, is
as easy to navigate and safer to drive than the automobile? And how about medical
applications such as articial organs, microdosage delivery systems, and diagnostic
tools? All these challenges are areas that future advances in ow control in particular
and uid mechanics in general, unforeseen to us today, could help overcome. The
book edited by Lumley, et al. (1996) provides an excellent outline of the near-future
prospects of 27 different uid dynamics research areas, whereas the one edited by
Gad-el-Hak, Pollard, and Bonnet (1998) focuses exclusively on ow control.
True, uid mechanics will be a very different discipline during the third millen-
nium. The entire eld, together with much of the rest of the physical sciences, may
be set for dramatic changes. In no small part, rapidly advancing computer technol-
ogy will be responsible for those changes. As argued in the editorial by Gad-el-Hak
and Sen (1996) and the paper by Gad-el-Hak (1998a), gigantic computers com-
bined with appropriate software may be available during the twenty-rst century
to routinely integrate the full, instantaneous NavierStokes equations. The black
box would prompt its operator for the geometry and ow conditions and would
then present a rst-principles numerical solution to the specic engineering problem,
high-Reynolds-number turbulent ows included.
In a turbulent ow, the ratio of the large eddies (at which the energy maintaining
the ow is input) to the Kolmogorov microscale (the ow smallest length scale) is
proportional to Re
3/4
(Tennekes and Lumley 1972). Each excited eddy requires at
least one grid point to describe it. Therefore, to adequately resolve, via direct nu-
merical simulations (DNS), a three-dimensional ow, the required number of modes
would be proportional to (Re
3/4
)
3
. To describe the motion of small eddies as they
are swept around by large ones, the time step must not be larger than the ratio of
the Kolmogorov length scale to the characteristic rms velocity. The large eddies, on
the other hand, evolve on a time scale proportional to their size divided by their rms
velocity. Thus, the number of time steps required is again proportional to Re
3/4
. Fi-
nally, the computational work requirement is the number of modes the number of
time steps, which scales with Re
3
, that is, an order of magnitude increase in computer
power is needed as the Reynolds number is doubled (Karniadakis and Orszag 1993).
Because the computational resource required varies as the cube of the Reynolds num-
ber, it may not be possible to simulate very high-Reynolds-number turbulent ows
any time soon (Karniadakis 1999).
Despite the bleak assessment above, one wonders whether gigantic computers
combined with appropriate software will be available during the twenty-rst century
to routinely solve, using DNS, practical turbulent ow problems? The black box
would prompt its operator for the geometry and ow conditions and would then
generate a numerical solution to the specic engineering problem. Nobody, except
the software developers, needs to know the details of what is going on inside the
black box, not even which equations are being solved. This situation is not unlike
using a present-day word processor or hand calculator. A generation of users of the
NavierStokes computers would quickly lose the aptitude, and the desire, to perform
simple analysis based on physical considerations, much the same as the inability of
some of todays users of hand calculators to manually carry out long divisions. The
need for rational approximations, so prevalent today in uid mechanics teaching and
practice, would gradually wither.
During the late 1990s, the supercomputer power approached the teraop (i.e.,
10
12
oating-point operations per second). This is about right to compute a ow
8 INTRODUCTION
with a characteristic Reynolds number of 10
8
, sufcient to simulate the ow around
an airfoil via DNS, around a wing via large-eddy simulation, or around an entire
commercial aircraft via Reynolds-averaged calculation. It is anticipated that the 10-
teraop computer will be available by mid-2000, and the 100-teraop machine by
2004, bettering the Moores law by triplinginstead of doublingthe chip speed
every 18 months. An exaop (10
18
ops) computer is needed to carry out direct
numerical simulation of the complete airplane (Karniadakis and Orszag 1993), and
thus we are still a long way from that.
Silicon-based computer powers have manifested spectacular recent advances
something like a factor of 10,000 improvement in speed and capacity during the
past 20 years.
2
If one is to extrapolate those recent advances to the next 50 years or
so, using direct numerical simulations to solve the turbulence problem for realistic
geometries and eld Reynolds numbers may begin to approach feasibility. Unfortu-
nately, however, silicon microchips are rapidly approaching their physical limits with
little room for further growth. The smallest feature in a microchip is dened by its
smallest linewidth, which in turn is related to the wavelength of light employed in
the basic lithographic process that is used to create the chip. The linewidths of chips
manufactured in 1999 is 180 nm, and most experts doubt that chips using current
complementary metal oxide semiconductor (CMOS) technology can shrink below
50-nm linewidthsa size that could be reached around 2012 (Lerner 1999). There
is even a naturalin contrast to a technologicalbarrier to extreme miniaturization
(Schulz 1999). A recent report by a team of researchers from Lucent Technologies
(Muller et al. 1999) has uncovered a troubling natural brick wall. The thin layer of
silicon dioxide insulation that separates the various conducting or semiconducting
components on a chip is rendered useless if its thickness falls below 0.7 nm or the
equivalent of four silicon atoms across. At the current rate of progress, such a funda-
mental limit on the thinnest usable silicon dioxide gate dielectric will also be reached
around 2012.
Fortunately, this kind of linear thinking may be misleading. Revolutionary com-
puting machines that bear little resemblance to todays chip-based computers may
be developed in the future. A recent article in Science (Glanz 1995) discusses ve
such futuristic computing concepts: quantumdots, quantumcomputers, holographic
association, optical computers, and deoxyribonucleic acid (DNA) computers. Very
recently, the prospect for molecular computing was unveiled (Collier et al. 1999;
Service 1999; Markoff 1999). We elaborate below on the latter two computers as
examples.
The DNAcomputer is a novel concept introduced and actually demonstrated only
late in 1994 by Adleman, who in turn was inspired by Feynmans (1961) original
vision of building even smaller submicroscopic computers. The idea for the massively
parallel DNA computing system is to use the four basic chemical units of DNA
(adenine, thymine, guanine, and cytosine, designated A, T, G, and C, respectively)
as computing symbols and to utilize the genetic material for information storage
and computations. Computer theorists argue that a problem could be set up by
synthesizing DNA molecules with a particular sequence that represents numerical
2
Although loosely related, this is consistent with the law named after the cofounder of Intel Corporation,
Gordon Moore, who in 1965 predicted that the transistor density on a semiconductor chip would double
and its price would halve roughly every 18 months. Incidentally, Moores law has been bettered in 1997.
information and by letting the molecules react in a test tube, producing newmolecules
whose sequence is the answer to the problem. Thus, the same genetic machinery
that generates living organisms could be used to solve previously unapproachable
mathematical puzzles. Crude estimates indicate that a mere 500 g of DNA molecules
(a human body contains about 300 g of deoxyribonucleic acid) suspended in 1000 l
of uid would have a memory equivalent to all the electronic computers ever made!
In such primordial reacting soup, 4 months of manipulating the DNA molecules
would yield an answer to a problem that would have required more operations than
all those ever performed on all the conventional computers ever builtat least up to
the mid-1990s.
Just as this book was going to the press, a research team from HewlettPackard
company and the University of California at Los Angeles (Collier et al. 1999) an-
nounced that they have fashioned simple computing components no thicker than a
single rotaxane
3
molecule, thus plunging deeply into a Lilliputian world that promises
computers 10
11
times as fast as todays most powerful personal computers. This infant
eld of research is termed moletronics or molecular electronics and uses chemical pro-
cesses rather than light to form molecular-thin onoff switches. Future developments
in moletronics will open new windows onto a once speculative but now increasingly
probable vista of molecular-scale computers, sensors, and actuators.
To close this section on a positive note, neither uid mechanics nor ow control
appears to be approaching a pinnacle. The best is yet to come. Future computers will
allow solutions to the most daunting problems in both elds, and rational designs
based on rst principles will be routine for the most complex uid systems. So-
phisticated reactive control systems consisting of a colossal number of microsensors
and microactuators communicating with each other via massively parallel comput-
ers will in the future improve, by leaps and bounds, the uids engineers ability to
reduce drag, increase lift, suppress ow-induced noise, enhance mixing, and achieve
any other desired ow-control goal.
3
Rotaxane is a synthetic organic compound created by chemists at the University of California at Los
Angeles.
CHAPTER TWO
Governing Equations
No knowledge can be certain if it is not based upon mathematics.
(Leonardo da Vinci, 14521519)
You are not educated until you know the Second Law of Thermodynamics.
(Charles Percy (Baron) Snow, 19051980)
PROLOGUE
There is no doubt that rational design (i.e., based on rst principles) of ow-control
devices is always preferable to a trial and error approach. Rational design of course is
not always possible owing to the extreme complexity of the equations involved, but
one tries either analytically or, more commonly to date, numerically. The search for
useful compliant coatings, discussed in Chapter 7, is a case in point. The window of
opportunity for a successful coating is so narrow that the probability of nding the
right one by experimenting is near nil. Fortunately, the analytical and numerical tools
to guide the initial choice for a transition-delaying compliant surface are currently
available. On the other hand, the oweld associated with a typical, deceivingly
simple vortex generator for airplane wings is so complex that its design is still done
to date more or less empirically.
The proper rst principles for owcontrol are those for uid mechanics itself. The
principles of conservation of mass, momentum, and energy govern all uid motions.
Additionally, all processes are constrained by the second law of thermodynamics. In
general, a set of partial, nonlinear differential equations expresses those principles,
and, together with appropriate boundary and initial conditions, constitute a well-
posed problem. In the following we provide a simple derivation of the general form
of the three conservation relations. Readers desirous of more details could consult
any advanced textbook in uid dynamics (e.g., Batchelor 1967; Hinze 1975; Landau
and Lifshitz 1987; Tritton 1988; Sherman 1990; Kundu 1990; Currie 1993; Panton
1996).
2.1 The Fundamental Equations
Each of the fundamental laws of uid mechanics, conservation of mass, momentum,
and energy, will be expressed rst as an integral relation and then as a differential
10
2.1 THE FUNDAMENTAL EQUATIONS 11
Figure 2.1: Arbitrary control volume.
equation. The uid motions under consider-
ation are nonrelativistic (i.e., their charac-
teristic velocities are far below the speed of
light). Thus, mass and energy are not inter-
changeable, and each is separately conserved.
The only additional assumption we will
make initially is that the uid is a conti-
nuum, implying that the derivatives of all the
dependent variables exist in some reasonable
sense. In other words, local properties such
as density and velocity are dened as aver-
ages over elements large compared with the microscopic structure of the uid but
small enough in comparison with the scale of the macroscopic phenomena to per-
mit the use of differential calculus to describe them. The continuum approxima-
tion is almost always met. The resulting equations therefore cover a very broad
range of situations, the exception being ows with spatial scales that are not much
larger than the mean distance between the uid molecules, as for example in the
case of rareed gas dynamics; shock waves that are thin relative to the mean free
path; or ows in micro and nano-devices (see Chapter 13). The present general ap-
proach will become particularly useful when considering high-speed ows, variable-
viscosity ows (e.g., in the case of surface heating and cooling), non-Newtonian uids
(e.g., concentrated solutions of drag-reducing polymers), and other unconventional
situations.
Consider the arbitrary, simply connected control volume V xed in an inertial,
Eulerian frame and depicted in Figure 2.1. Fluid can pass in or out through the control
surface A, which is also xed in space and time. The outward unit normal vector eld
is n. To derive an integral form of any of the three conservation relations, we simply
state that the rate of change of any physical property contained within the control
volume must be equal to the net rate of inux of that property through the control
surface plus the net rate of gain from surface as well as volume sources. The three
properties we consider in turn in the following subsections are mass, momentum,
and energy.
2.1.1 Conservation of Mass
For a chemically nonreacting uid, that mass cannot be destroyed or created
translates into the following statement: Rate of change of mass inside the control
volume = (mass convected out through the control surface). Thus, the integral
form of the continuity equation reads in Cartesian tensor notations
t
___
V
dV =
__
A
u
k
dA
k
(2.1)
where is the uid density and u
k
is an instantaneous velocity component (u, v, w).
The independent variables are time t and the three spatial coordinates x
1
, x
2
, and x
3
or (x, y, z). The innitesimal volume dV = dxdy dz. Finally, the Einsteins summation
convention applies to all repeated indices.
12 GOVERNING EQUATIONS
Applying the Gauss divergence theorem of vector calculus converts the surface
integral in Eq. (2.1) into a volume one, and the equation reads
___
V
t
dV =
___
V
x
k
(u
k
) dV (2.2)
Note that because time and space are independent, the time derivative and volume
integral are commuted in the left-hand side of the preceding equation.
The last step to derive the differential form of the continuity equation involves
simply moving the right-hand side of Eq. (2.2) to the left-hand side and arguing that
because the control volume is arbitrary, the resulting equation can be fullled only if
the integrand is zero at each point in space. Thus, at every point in spacetime
t
+

x
k
(u
k
) = 0 (2.3)
2.1.2 Newtons Second Law
If the physical property under consideration is momentum, then according to
Newtons second lawsurface and body forces constitute, respectively, the surface and
volume sources of momentum. Conservation of momentum thus reads: Time rate of
increase of momentum inside the control volume =(rate of convective outux of
momentum across the control surface) + surfaces forces + body forces. The integral
form of the momentum equation can therefore be written as
1
t
___
V
u
i
dV =
__
A
(u
k
n
k
) u
i
dA
+
__
A
ki
n
k
dA+
___
V
g
i
dV (2.4)
where
ki
is the second-order stress tensor (surface force per unit area), and g
i
is the
body force per unit mass. Body forces apply to the entire mass of a uid element and
are due to external elds such as gravity and electromagnetic potential. A discussion
of the latter kind of body force will be included in Chapter 14. Equation (2.4) is of
course a vector equation valid for i = 1, 2, and 3.
Once again all the surface integrals in Eq. (2.4) are converted to volume integrals
using the divergence theorem, and thus the momentum integral equation reads
___
V
t
(u
i
) dV =
___
V
x
k
(u
k
u
i
) dV
+
___
V
ki
x
k
dV +
___
V
g
i
dV (2.5)
The differential formof the momentumequation follows immediately fromequat-
ing the integrand in Eq. (2.5) to zero, thus
t
(u
i
) +

x
k
(u
k
u
i
) =

ki
x
k
+g
i
(2.6)
1
If the frame of reference is noninertial (i.e., if it accelerates linearly, rotates relative to an inertial frame,
or both), additional acceleration terms are needed in the momentum balance.
2.1 THE FUNDAMENTAL EQUATIONS 13
This can readily be simplied by invoking the continuity equation (2.3) to read nally
_
u
i
t
+u
k
u
i
x
k
_
=

ki
x
k
+g
i
(2.7)
Each term in the Cauchys equation of motion (2.7) has units of force per unit vol-
ume. An equation for the angular momentum can be derived in a way similar to
the derivation of the linear momentum equation above. Also, taking the curl of
Eq. (2.7) leads to the equation for vorticity.
2.1.3 Conservation of Energy
Here the conserved physical property under consideration is the intrinsic energy
of a thermodynamic system. For convenience, we choose the intrinsic energy per unit
mass to be
2
the sum of internal energy e and kinetic energy
1
2
u
i
u
i
. For nonrelativistic
motions, energy cannot be created or destroyed. The rst law of thermodynamics
states: Time rate of change of intrinsic energy inside the control volume =(rate of
convective outux of energy across the control surface) (rate of heat transfer due to
conduction and radiation out of the control surface) + rate of work done by surface
and body forces. The integral formof the energy equation can therefore be written as
t
___
V
_
e +
1
2
u
2
i
_
dV =
__
A
u
k
_
e +
1
2
u
2
i
_
dA
k
__
A
q
k
dA
k
+
__
A
u
i
ki
dA
k
+
___
V
u
i
g
i
dV
(2.8)
where q
k
is the sum of heat ux vectors due to conduction and radiation, and a
summation over i is implied in u
2
i
. All the surface integrals in Eq. (2.8) are converted
to volume integrals using the divergence theorem, and thus the integral form of the
conservation of energy equation now reads
___
V
t
_
_
e +
1
2
u
2
i
__
dV
=
___
V
x
k
_
u
k
_
e +
1
2
u
2
i
__
dV
___
V
q
k
x
k
dV +
___
V
(u
i
ki
)
x
k
dV +
___
V
u
i
g
i
dV (2.9)
The differential form of the energy equation follows from equating the integrand
in Eq. (2.9) to zero and invoking the continuity equation (2.3), thus
_
e +
1
2
u
2
i
_
t
+u
k
_
e +
1
2
u
2
i
_
x
k
_
=
q
k
x
k
+
(u
i
ki
)
x
k
+u
i
g
i
(2.10)
2
We can include the potential energy as part of the intrinsic energy. In that case, however, the work done
by the body forces should not be included in the energy balance.
An equation for the kinetic energy per unit volume
1
2
u
i
u
i
can be derived simply
by taking the dot product of u
i
with the momentum equation (2.7). The resulting
equation reads
_
1
2
u
2
i
_
t
+u
k
_
1
2
u
2
i
_
x
k
_
= u
i
ki
x
k
+u
i
g
i
(2.11)
and states that the rate of increase of kinetic energy following a uid material ele-
ment = rate of work done by the net surface forces + rate of work done by body
forces. The thermal energy equation is readily obtained by subtracting Eq. (2.11)
from Eq. (2.10) to yield
_
e
t
+u
k
e
x
k
_
=
q
k
x
k
+
ki
u
i
x
k
(2.12)
The second term in the right-hand side is the deformation work done (per unit time)
by the surface forces and contains both reversible and irreversible contributions (as
dictated by the second lawof thermodynamics) to internal energy. Body forces do not
contribute directly to internal energy, and Eq. (2.12) states that the rate of increase
of internal energy of a uid material element is proportional to the sum of the net
rate of heat inow and the net rate of work done by surface forces.
2.2 Closing the Equations
Equations (2.3), (2.7), and (2.12) constitute ve differential equations for the 17 un-
knowns , u
i
,
ki
, e, and q
k
. Absent any body couples, the stress tensor is symmetric,
having only six independent components, which reduces the number of unknowns
to 14. Obviously, the continuum ow equations do not form a determinate set. To
close the conservation equations, the relation between the stress tensor and defor-
mation rate, the relation between the heat ux vector and the temperature eld, and
the appropriate equations of state relating the different thermodynamic properties
are needed. The stressrate of strain relation and the heat uxtemperature relation
are approximately linear if the ow is not too far from thermodynamic equilibrium.
This is a phenomenological result but can be rigorously derived from the Boltzmann
equation for a dilute gas if it is assumed that the ow is near equilibrium (see
Chapter 13).
For a Newtonian, isotropic, Fourier,
3
ideal gas, for example, these relations read
ki
= p
ki
+
_
u
i
x
k
+
u
k
x
i
_
+
_
u
j
x
j
_
ki
(2.13)
q
i
=
f
T
x
i
+Heat ux due to radiation (2.14)
de = c
v
dT and p = RT (2.15)
3
Newtonian implies a linear relation between the stress tensor and the symmetric part of the deformation
tensor (rate of strain tensor). The isotropy assumption reduces the 81 constants of proportionality in
that linear relation to two constants. Fourier uid is that for which the conduction part of the heat ux
vector is linearly related to the temperature gradient, and again isotropy implies that the constant of
proportionality in this relation is a single scalar.
2.2 CLOSING THE EQUATIONS 15
where p is the thermodynamic pressure, and are the rst and second coefcients
of viscosity, respectively,
ki
is the unit second-order tensor (Kronecker delta),
f
is the uid thermal conductivity, T is the temperature eld, c
v
is the specic heat
at constant volume, and R is the gas constant given by the Boltzmann constant
divided by the mass of an individual molecule (R=k/m). The Stokes hypothesis
relates the rst and second coefcients of viscosity, +
2
3
=0, although the validity
of this assumption has occasionally been questioned (Gad-el-Hak, 1995). Given the
preceding constitutive relations, and if radiative heat transfer is neglected,
4
Eqs. (2.3),
(2.7), and (2.12), respectively, read
t
+

x
k
(u
k
) = 0 (2.16)
_
u
i
t
+u
k
u
i
x
k
_
=
p
x
i
+g
i
+

x
k
_
_
u
i
x
k
+
u
k
x
i
_
+
ki

u
j
x
j
_
(2.17)
c
v
_
T
t
+u
k
T
x
k
_
=

x
k
_
f
T
x
k
_
p
u
k
x
k
+ (2.18)
The three components of the vector equation (2.17) are the NavierStokes equations
expressing the conservation of momentum for a Newtonian uid. In the thermal
energy equation (2.18), is the always positive
5
dissipation function expressing the
irreversible conversion of mechanical energy to internal energy as a result of the
deformation of a uid element. The second term on the right-hand side of Eq. (2.18)
is the reversible work done (per unit time) by the pressure as the volume of a uid
material element changes. For a Newtonian, isotropic uid, the viscous dissipation
rate is given by
=
1
2
_
u
i
x
k
+
u
k
x
i
_
2
+
_
u
j
x
j
_
2
(2.19)
There are now six unknowns, , u
i
, p, and T, and the ve coupled equations (2.16),
(2.17), and (2.18) plus the equation of state relating pressure, density, and temper-
ature. These six equations together with sufcient numbers of initial and boundary
conditions constitute a well-posed, albeit formidable, problem. The system of equa-
tions (2.16)(2.18) is an excellent model for the laminar or turbulent ow of most
uids such as air and water under most circumstances, including high-speed gas ows
for which the shock waves are thick relative to the mean free path of the molecules.
6
Considerable simplication is achieved if the owis assumed to be incompressible,
which is usually a reasonable assumption provided that the characteristic ow speed
is less than 0.3 of the speed of sound.
7
The incompressibility assumption is readily
satised for almost all liquid ows and many gas ows. In such cases, the density
is assumed to be either a constant or a given function of temperature (or species
4
A reasonable assumption when dealing with low to moderate temperatures because the radiative heat
ux is proportional to T
4
.
5
As required by the second law of thermodynamics.
6
This condition is met if the shock Mach number is less than 2.
7
Although, as will be demonstrated in the following section, there are circumstances when even a low-
Mach-number ow should be treated as compressible.
concentration).
8
The governing equations for such ow are
u
k
x
k
= 0 (2.20)
_
u
i
t
+u
k
u
i
x
k
_
=
p
x
i
+

x
k
_
_
u
i
x
k
+
u
k
x
i
__
+g
i
(2.21)
c
p
_
T
t
+u
k
T
x
k
_
=

x
k
_
f
T
x
k
_
+
incomp
(2.22)
where
incomp
is the incompressible limit of Eq. (2.19). These are ve equations for the
ve dependent variables u
i
, p, and T. Note that the left-hand side of Eq. (2.22) has the
specic heat at constant pressure c
p
and not c
v
. It is the convection of enthalpyand
not internal energythat is balanced by heat conduction and viscous dissipation.
This is the correct incompressible-ow limit of a compressible uid, as discussed
in detail in Section 10.9 of Panton (1996), which is a subtle point, perhaps, but
one that is frequently missed in textbooks. The system of equations (2.20)(2.22)
is coupled if either the viscosity or density depends on temperature; otherwise, the
energy equation is uncoupled from the continuity and momentum equations and can
therefore be solved after the velocity and pressure elds are determined.
In nondimensional form, the incompressible ow equations read
u
k
x
k
= 0 (2.23)
_
u
i
t
+u
k
u
i
x
k
_
=
p
x
i
+
Gr
Re
2
T
i 3
+

x
k
_
F
(T)
Re
_
u
i
x
k
+
u
k
x
i
__
(2.24)
_
T
t
+u
k
T
x
k
_
=

x
k
_
1
Pe
T
x
k
_
+
Ec
Re
F
(T)
incomp
(2.25)
where F
(T) is a dimensionless function that characterizes the viscosity variation

with temperature, and Re, Gr, Pe, and Ec are, respectively, the Reynolds, Grashof,
P eclet, and Eckert numbers. These dimensionless parameters determine the relative
importance of the different terms in the equations.
For both the compressible and the incompressible equations of motion, the trans-
port terms are neglected away fromsolid walls in the limit of innite Reynolds number
(i.e., zero Knudsen number). The uid is then approximated as inviscid and noncon-
ducting, and the corresponding equations read as follows (for the compressible case):
t
+

x
k
(u
k
) = 0 (2.26)
_
u
i
t
+ u
k
u
i
x
k
_
=
p
x
i
+g
i
(2.27)
c
v
_
T
t
+ u
k
T
x
k
_
= p
u
k
x
k
(2.28)
8
Within the so-called Boussinesq approximation, density variations have negligible effect on inertia but
are retained in the buoyancy terms. The incompressible continuity equation is therefore used.
2.3 COMPRESSIBILITY 17
The Euler equation (2.27) can be integrated along a streamline, and the resulting
Bernoullis equation provides a direct relation between the velocity and pressure.
All the eld equations recited thus far are valid at all points in space and time and
can be used for either laminar or turbulent ows. Three forms of the equations of
motion are particularly useful to ow control: the x, y, and z momentum equations
written at the wall, the ensemble average of the instantaneous equations for turbulent
ows, and the integral form of the differential equations. Those forms will be given
in the three sections immediately following the section on compressibility.
2.3 Compressibility
The issue of whether to consider the continuum ow compressible or incompress-
ible seems to be rather straightforward but is in fact full of potential pitfalls. If the
local Mach number is less than 0.3, then the ow of a compressible uid like air
canaccording to the conventional wisdombe treated as incompressible. But the
well-known Ma <0.3 criterion is only a necessary but not a sufcient condition to
allow a treatment of the ow as approximately incompressible. In other words, there
are situations in which the Mach number can be exceedingly small, whereas the ow
is compressible. As is well documented in heat transfer textbooks, strong wall heat-
ing or cooling may cause the density to change sufciently and the incompressible
approximation to break down, even at low speeds. Less known is the situation en-
countered in some microdevices in which the pressure may strongly change owing
to viscous effects even though the speeds may not be high enough for the Mach
number to go above the traditional threshold of 0.3. Corresponding to the pressure
changes would be strong density changes that must be taken into account when writ-
ing the continuum equations of motion. In this section, we systematically explain all
situations in which compressibility effects must be considered.
Let us rewrite the full continuity equation (2.3) as follows
D
Dt
+
u
k
x
k
=0 (2.29)
where
D
Dt
is the substantial derivative (

t
+ u
k

x
k
) expressing changes following a
uid element. The proper criterion for the incompressible approximation to hold
is that (
1
D
Dt
) is vanishingly small. In other words, if density changes following a uid
particle are small, the ow is approximately incompressible. Density may change
arbitrarily from one particle to another without violating the incompressible ow as-
sumption. This is the case, for example, in the stratied atmosphere and ocean where
the variable-density/temperature/salinity ow is often treated as incompressible.
From the state principle of thermodynamics, we can express the density changes
of a simple system in terms of changes in pressure and temperature,
=( p, T) (2.30)
From the chain rule of calculus,
1
D
Dt
=
Dp
Dt

DT
Dt
(2.31)
where and are, respectively, the isothermal compressibility coefcient and the
bulk expansion coefcienttwo thermodynamic variables that characterize the uid
susceptibility to change of volume, which are dened by the following relations:
( p, T)
1
T
(2.32)
( p, T)
1
p
(2.33)
For ideal gases, =1/p, and =1/T. Note, however, that in the following arguments
it will not be necessary to invoke the ideal gas assumption.
The ow must be treated as compressible if pressure or temperature changes, or
both are sufciently strong. Equation (2.31) must of course be properly nondimen-
sionalized before deciding whether a termis large or small. Here in, we followclosely
the procedure detailed in Panton (1996).
Consider rst the case of adiabatic walls. Density is normalized with a reference
value
o
, velocities with a reference speed v
o
, spatial coordinates and time with,
respectively, L and L/v
o
, and the isothermal compressibility coefcient and bulk
expansion coefcient with reference values
o
and
o
, respectively. The pressure
is nondimensionalized with the inertial pressure scale
o
v
2
o
. This scale is twice the
dynamic pressure (i.e., the pressure change as an inviscid uid moving at the reference
speed is brought to rest).
Temperature changes for the case of adiabatic walls result from the irreversible
conversion of mechanical energy into internal energy via viscous dissipation. Tem-
perature is therefore nondimensionalized as follows:
T
=
T T
o
_
o
v
2
o
o
_ =
T T
o
Pr
_
v
2
o
c
po
_ (2.34)
where T
o
is a reference temperature;
o
,
o
, and c
p
o
are, respectively, reference viscos-
ity, thermal conductivity and specic heat at constant pressure; and Pr is the reference
Prandtl number (
o
c
p
o
)/
o
.
The scaling used above for pressure is based on the Bernoullis equation and
therefore neglects viscous effects. This particular scaling guarantees that the pressure
term in the momentum equation will be of the same order as the inertia term. The
temperature scaling assumes that the conduction, convection, and dissipation terms
in the energy equation have the same order of magnitude. The resulting dimensionless
form of Eq. (2.31) reads
1
Dt
=
o
Ma
2
_
Dp
Dt

PrB
A
DT
Dt
_
(2.35)
where the superscript

indicates a nondimensional quantity, Ma is the reference
Mach number (Ma v
o
/a
o
, where a
o
is the reference speed of sound), and A and
B are dimensionless constants dened by A
o

o
c
p
o
T
o
, and B
o
T
o
. If the
scaling is properly chosen, the terms having the

superscript in the right-hand side
should be of order one, and the relative importance of such terms in the equations of
motion is determined by the magnitude of the dimensionless parameter(s) appearing
to their left (e.g., Ma, Pr, etc.). Therefore, as Ma
2
0, temperature changes due to
2.3 COMPRESSIBILITY 19
viscous dissipation are neglected (unless Pr is very large, as for example in the case
of highly viscous polymers and oils). Within the same order of approximation, all
thermodynamic properties of the uid are assumed constant.
Pressure changes are also neglected in the limit of zero Mach number. Hence,
for Ma <0.3 (i.e., Ma
2
<0.09), density changes following a uid particle can be
neglected and the owcan then be approximated as incompressible.
9
However, there
is a caveat in this argument. Pressure changes due to inertia can indeed be neglected at
small Mach numbers, and this is consistent with the way we nondimensionalized the
pressure termabove. If, on the other hand, pressure changes are mostly due to viscous
effects, as is the case, for example, in a long microduct or a microgas bearing, pressure
changes may be signicant even at low speeds (low Ma). In that case the term
Dp
Dt
in
Eq. (2.35) is no longer of order one and may be large regardless of the value of Ma.
Density then may change signicantly, and the ow must be treated as compressible.
Had pressure been nondimensionalized using the viscous scale (
o
v
o
L
) instead of the
inertial one (
o
v
2
o
), the revised Eq. (2.35) would have had Re
1
appearing explicitly
in the rst termin the right-hand side, accentuating the importance of this termwhen
viscous forces dominate.
A similar result can be gleaned when the Mach number is interpreted as follows:
Ma
2
=
v
2
o
a
2
o
= v
2
o
s
=

o
v
2
o
p
=

o
(2.36)
where s is the entropy. Again, the preceding equation assumes that pressure changes
are inviscid, and therefore a small Mach number means negligible pressure and
density changes. In a ow dominated by viscous effectssuch as that inside a
microductdensity changes may be signicant even in the limit of zero Mach
number.
Experiments in gaseous microducts conrm the arguments above. For both low-
and high-Mach-number ows, pressure gradients in long microchannels are non-
constant, consistent with the compressible ow equations. Such experiments were
conducted by, among others, Prudhomme, Chapman, and Bowen (1986); Pfahler
et al. (1991); van den Berg, Seldam, and Gulik (1993); Liu et al. (1993, 1995); Pong
et al. (1994); Harley, Bau, and Zemel (1995); Piekos and Breuer (1996); Arkilic
(1997); and Arkilic, Schmidt, and Breuer (1995, 1997a,b). We will return to this
issue in Chapter 13.
Identical arguments can be made in the case of isothermal walls. Here strong
temperature changes may be the result of wall heating or cooling even if viscous
dissipation is negligible. The proper temperature scale in this case is given in terms
of the wall temperature T
w
and the reference temperature T
o
as follows:
T =
T T
o
T
w
T
o
(2.37)
where

T is the new dimensionless temperature. The nondimensional form of
9
With an error of about 10% at Ma = 0.3, 4% at Ma = 0.2, 1% at Ma = 0.1, and so on.
Eq. (2.31) now reads
1
Dt
=
o
Ma
2
Dp
Dt
B
_
T
w
T
o
T
o
_
D

T
Dt
(2.38)
Here we notice that the temperature term is different from that in Eq. (2.35). The
Ma is no longer appears in this term, and strong temperature changes such as large
(T
w
T
o
)/T
o
may cause strong density changes regardless of the value of the Mach
number. Additionally, the thermodynamic properties of the uid are not constant
but depend on temperature, and as a result the continuity, momentum, and energy
equations all couple. The pressure term in Eq. (2.38), on the other hand, is exactly
as it was in the adiabatic case, and the same arguments made before apply: the ow
should be considered compressible if Ma > 0.3 or if pressure changes due to viscous
forces are sufciently large.
There are three additional scenarios in which signicant pressure and density
changes may take place without inertial, viscous, or thermal effects. First is the case
of quasi-static compression and expansion of a gas in, for example, a pistoncylinder
arrangement. The resulting compressibility effects are, however, compressibility of
the uid and not of the ow. Two other situations in which compressibility effects
must also be considered are problems with length scales comparable to the scale
height of the atmosphere and rapidly varying ows, as in sound propagation (see
Lighthill 1963b).
2.4 Equations of Motion at the Wall
In wall-bounded ows, the streamwise (x) and spanwise (z) momentum equations
written at the wall give very useful expressions of the wall uxes of, respectively, the
spanwise and streamwise vorticity. For an incompressible uid over a nonmoving
wall of small curvature, and if body forces are neglected, the streamwise, normal,
and spanwise momentum equations, written at y = 0, follow from Eq. (2.21) and
read
v
w
u
y
y=0
+
p
x
y=0
y=0
u
y
y=0
=
2
u
y
2
y=0
(2.39)
0 +
p
y
y=0
0 =
2
v
y
2
y=0
(2.40)
v
w
w
y
y=0
+
p
z
y=0
y=0
w
y
y=0
=
2
w
y
2
y=0
(2.41)
These equations are instantaneous and are valid for both laminar and turbulent ows.
Here, v
w
is the (positive) injection or (negative) suction velocity through the wall.
Upward or downward motion of the wall acts analogously to injection or suction.
The right-hand side of each of the three equations represents the curvature of the cor-
responding velocity prole, which in the case of the streamwisespanwise equation
is the same as the ux of spanwisestreamwise vorticity from the wall. Negative cur-
vature implies that the velocity prole is fuller and that the wall is a source (positive
2.5 TURBULENT FLOWS 21
ux) of vorticity. Whether the wall is a source or a sink of spanwisestreamwise
vorticity depends on whether uid is sucked or injected from the wall, whether
the streamwisespanwise pressure gradient is favorable or adverse, and whether the
wall viscosity is lower or higher than that above the surface. For a canonical tur-
bulent ow, the instantaneous velocity proles in all three directions change shape
continuously as a result of the random changes in the pressure eld.
2.5 Turbulent Flows
All the equations thus far are valid for nonturbulent as well as turbulent ows. How-
ever, in the latter case the dependent variables are in general random functions of
space and time. No straightforward method exists for obtaining stochastic solutions
of these nonlinear partial differential equations, and this is the primary reason why
turbulence remains as the last great unsolved problem of classical physics. Dimen-
sional analysis can be used to obtain crude results for a few cases, but rst-principles
analytical solutions are not possible even for the simplest conceivable turbulent
ow.
The contemporary attempts to use dynamical systems theory to study turbulent
ows have not yet reached fruition, especially at Reynolds numbers far above transi-
tion (Aubry et al. 1988), although advances in this theory have helped somewhat with
reducing and displaying the massive bulk of data resulting fromnumerical and exper-
imental simulations (Sen 1989). The recent book by Holmes, Lumley, and Berkooz
(1996) provides a useful, readable introduction to the emerging eld. It details a
strategy via which knowledge of coherent structures, nite-dimensional dynamical
systems theory, and the KarhunenLo` eve or proper orthogonal decomposition could
be combined to create low-dimensional models of turbulence that resolve only the
organized motion and describe their dynamical interactions. The utility of the dynam-
ical systems approach as an additional arsenal to tackle the turbulence conundrum
has been demonstrated only for turbulence near transition or near a wall, and thus
the ow would be relatively simple and a relatively small number of degrees of free-
dom would be excited. Holmes et al. (1996) summarize the (partial) successes that
have been achieved thus far using relatively small sets of ordinary differential equa-
tions and suggest a broad strategy for modeling turbulent ows as well as other
spatiotemporally complex systems.
The brute-force numerical integration of the instantaneous equations using the
supercomputer is prohibitively expensiveif not impossibleat practical Reynolds
numbers (Moin and Mahesh 1998). For the present at least, a statistical approach, in
which a temporal, spatial, or ensemble average is dened and the equations of motion
are written for the various moments of the uctuations about this mean, is the only
route available to get meaningful engineering results. Unfortunately, the nonlinearity
of the NavierStokes equations guarantees that the process of averaging to obtain
moments results in an open system of equations in which the number of unknowns
is always greater than the number of equations, and more or less heuristic modeling
is used to close the equations. This is known as the closure problem and again makes
obtaining rst-principle solutions to the (averaged) equations of motion impossible.
To illustrate the closure problem, consider the (instantaneous) continuity and
momentum equations for a Newtonian, incompressible, constant density, constant
viscosity turbulent ow. In this uncoupled version of the four equations (2.20) and
(2.21), for the four random unknowns u
i
and p, no general stochastic solution is
known to exist. But would it be feasible to obtain solutions for the nonstochastic
mean-ow quantities? As was rst demonstrated by Osborne Reynolds (1895) over
a century ago, all the eld variables are decomposed into a mean and a uctuation. Let
u
i
=U
i
+u
i
and p=P + p
, where U
i
and P are ensemble averages for the velocity
and pressure, respectively, and u
i
and p
are the velocity and pressure uctuations

about the respective averages. Note that temporal or spatial averages could be used in
place of ensemble average if the oweld is stationary or homogeneous, respectively.
In the former case, the time derivative of any statistical quantity vanishes. In the latter,
averaged functions are independent of position. When the decomposed pressure and
velocity are substituted into Eqs. (2.20) and (2.21), the equations governing the mean
velocity and mean pressure for an incompressible, constant-viscosity turbulent ow
become
U
k
x
k
= 0 (2.42)
_
U
i
t
+U
k
U
i
x
k
_
=
P
x
i
+

x
k
_
U
i
x
k
u
i
u
k
_
+ g
i
(2.43)
where, for clarity, the primes have been dropped from the uctuating velocity com-
ponents u
i
and u
k
.
This is now a system of 4 equations for the 10 unknowns U
i
, P, and u
i
u
k
.
10
The
momentum equation (2.43) is written in a form that facilitates the physical inter-
pretation of the turbulence stress tensor (Reynolds stresses), u
i
u
k
, as additional
stresses on a uid element to be considered along with the conventional viscous
stresses and pressure. An equation for the components of this tensor may be derived,
but it will contain third-order moments such as u
i
u
j
u
k
, and so on. The equations are
(heuristically) closed by expressing the second- or third-order quantities in terms of
the rst- or second-moments, respectively. For comprehensive reviews of these rst-
and second-order closure schemes, see Lumley (1983, 1987), Speziale (1991), and
Wilcox (1993).
2.6 The K arm an Integral Equation
The next useful equation results from integrating in the normal direction (from the
wall to the edge of the shear layer) and combining the continuity and streamwise
momentum equations, the latter of which is written in terms of a generic shear stress
(
xy
/y). The usual boundary layer approximations are invoked in the momentum
equation: (1) neglecting the normal stress terms relative to the shear stress terms, and
(2) assuming the pressure to be impressed upon the boundary layer from without.
The nal result is known as the K arm an integral equation. For a two-dimensional
(or axisymmetric) wall-bounded ow of a compressible uid and with body force
10
The second-order tensor u
i
u
k
is obviously a symmetric one with only six independent components.
2.6 THE K
ARM
AN INTEGRAL EQUATION 23
neglected, this equation is derived in Kays and Crawford (1993) and reads
C
f
2
=
1
U
2
o
t
(U
o

) +

x

w
v
w
o
U
o
+
__
2 +

_
1
U
o
U
o
x
+
1
o
x
+
1
R
R
x
_
(2.44)
Similar integral relations can be derived from the kinetic energy and thermal en-
ergy differential equations. In the momentum integral equation, the skin-friction
coefcient, the displacement thickness, and the momentum thickness are given by,
respectively,
C
f

w
1
2
o
U
2
o
(2.45)

_

0
_
1
u
o
U
o
_
dy (2.46)

_

0
u
o
U
o
_
1
u
U
o
_
dy (2.47)
Also, R is the transverse radius of curvature of the body of revolution (the corre-
sponding term in Eq. (2.44) drops for a two-dimensional ow),
o
and U
o
are the
density and velocity outside the boundary layer, respectively,
w
and v
w
are the density
and normal velocity of uid injected or sucked through the surface (or the upward
or downward wall velocity), and
w
is the shear stress at the wall.
Because the streamwise momentumequation leading to the K arm an integral equa-
tion was written in terms of a generic shear stress, the integral momentum equation
(2.44) is, in fact, valid for laminar and turbulent ows as well as for Newtonian
and non-Newtonian uids; the only assumption being made is that the ow is two-
dimensional (or axisymmetric) in the mean. In case of a turbulent ow, the mean
streamwise velocity, U(t, x, y), is used in the denition of
and
. For a two-
dimensional ow of a Newtonian uid,
w
=
U
y
y=0
+
V
x
y=0
(2.48)
The second termon the right-hand side of this equation vanishes if there is no injection
or suction (or upward or downward wall motion), or if the transpiration is spatially
uniform.
There is a seeming controversy that periodically ares up in the literature: Is
the K arm an integral equation accurate for a turbulent ow to within the rst-order
boundary-layer approximation? In the case of a two-dimensional, incompressible
turbulent ow, the streamwise momentum equation contains terms related to the
turbulence normal stresses, namely,
U
t
+U
U
x
+ V
U
y
=
P
o
x
+
x
_
v
2
u
2
_
+
xy
y
(2.49)
In the last term of this equation,
xy
is the sum of the viscous shear stress (
U
y
)
and the Reynolds shear stress ( uv). The second term on the right-hand side has
contributions from the fact that the pressure within the boundary layer is not quite
equal to that at the edge (the difference being v
2
) and from the turbulence nor-
mal stress term ( u
2
). The viscous normal stress terms corresponding to those two
Reynolds stress terms are, respectively,
V
y
and
U
x
. These clearly have smaller
orders of magnitude and are therefore dropped from Eq. (2.49). The question is
whether or not to retain the Reynolds normal stress terms. If the turbulence normal
stresses are retained, Eq. (2.49) cannot be integrated to yield the K arm an integral
equation. Given that the turbulence rms uctuations within a boundary layer are
typically one order of magnitude less than the mean velocity and that the term in
question is the difference between two small numbers, careful order-of-magnitude
analysis shows that indeed the second term on the right-hand side of Eq. (2.49)
can be neglected within the boundary-layer approximation (see also Hinze 1975).
The K arm an integral equation (2.44) for a turbulent wall-bounded ow is therefore
rst-order accurate.
CHAPTER THREE
Unifying Principles
Mechanics is the paradise of the mathematical sciences because by means
of it one comes to the fruits of mathematics.
(Leonardo da Vinci, 14521519)
What experience and history teach is thisthat people and governments
never have learned anything from history, or acted on principles deduced
from it.
(Georg Wilhelm Friedrich Hegel, 17701831)
PROLOGUE
A particular control strategy is chosen based on the kind of ow and the control goal
to be achieved. Flow-control goals are strongly, often adversely, interrelated, and
there lies the challenge of making the tough compromises. There are several different
ways for classifying control strategies to achieve a desired effect. Presence or lack of
walls, Reynolds and Mach numbers, and the character of the ow instabilities are all
important considerations for the type of control to be applied. All these seemingly
disparate issues are what places the eld of owcontrol in a unied framework. They
will be discussed in turn in this chapter.
3.1 Control Goals and Their Interrelation
What does the engineer want to achieve when attempting to manipulate a particular
oweld? Typically he or she aims at reducing the drag; at enhancing the lift; at
augmenting the mixing of mass, momentum, or energy; at suppressing the ow-
induced noise; or at a combination thereof. To achieve any of these useful end results
for either free-shear or wall-bounded ows, transition fromlaminar to turbulent ow
may have to be delayed or advanced, ow separation may have to be prevented or
provoked, and nally turbulence levels may have to be suppressed or enhanced. All
these engineering goals and the corresponding ow changes intended to effect them
are schematically depicted in Figure 3.1. None of these are particularly difcult if
taken in isolation, but the challenge is to achieve a goal using a simple device that
is inexpensive to build as well as to operate, and, most important, has minimum
25
26 UNIFYING PRINCIPLES
Figure 3.1: Engineering goals and corresponding ow changes.
side effects. For this latter hurdle, the interrelation among control goals must be
elaborated, and this is what is attempted below.
Section 3.3 will contrast free-shear and wall-bounded ows. For the purposes of
the present argument, we focus on the latterparticularly the technologically very
important boundary layers. An external wall-bounded ow, such as that developing
on the exterior surfaces of an aircraft or a submarine, can be manipulated to achieve
transition delay, separation postponement, lift increase, skin-friction and pressure-
drag
1
reduction, turbulence augmentation, heat-transfer enhancement, or noise sup-
pression. These objectives are not necessarily mutually exclusive. The schematic in
Figure 3.2 is a partial representation of the interrelation between one control goal
and another. To focus the discussion further, think of the ow developing on a lift-
ing surface such as an aircraft wing. If the boundary layer becomes turbulent, its
resistance to separation is enhanced, and more lift could be obtained at increased
incidence. On the other hand, the skin-friction drag for a laminar boundary layer
can be as much as an order of magnitude less than that for a turbulent one. If tran-
sition is delayed, lower skin friction as well as lower ow-induced noise is achieved.
However, the laminar boundary layer can only support a very small adverse pressure
gradient without separation, and subsequent loss of lift and increase in form drag
occur. Once the laminar boundary layer separates, a free-shear layer forms and, for
moderate Reynolds numbers, transition to turbulence takes place. Increased entrain-
ment of high-speed uid because of the turbulent mixing may result in reattachment
1
Pressure drag includes contributions from ow separation, displacement effects, induced drag, wave
drag, and, for time-dependent motion of a body through a uid, virtual mass.
3.2 CLASSIFICATION SCHEMES 27
Figure 3.2: Interrelation between ow-control goals.
of the separated region and formation of a laminar separation bubble.
2
At higher
incidence, the bubble breaks down, either separating completely or forming a longer
bubble. In either case, the form drag increases and the lift-curves slope decreases.
The ultimate goal of all this is to improve the airfoils performance by increasing the
lift-to-drag ratio. However, induced drag is caused by the lift generated on a lifting
surface with a nite span. Moreover, more lift is generated at higher incidence, but
form drag also increases at these angles.
The preceding discussion points to potential conicts as one tries to achieve a
particular control goal only to adversely affect another goal. An ideal method of
control that is simple, inexpensive to build and operate, and does not have any
trade-offs does not exist, and the skilled engineer has to make continual compromises
to achieve a particular design goal.
3.2 Classication Schemes
There are different classication schemes for ow-control methods. One is to con-
sider whether the technique is applied at the wall or away fromit. Surface parameters
2
See Chapter 9 for more details about the formation and control of separation bubbles.
that can inuence the ow include roughness, shape, curvature, rigid-wall motion,
compliance, temperature, and porosity. Heating and cooling of the surface can in-
uence the ow via the resulting viscosity and density gradients. Mass transfer can
take place through a porous wall or a wall with slots. Suction and injection of pri-
mary uid can have signicant effects on the oweld, inuencing particularly the
shape of the velocity prole near the wall and thus the boundary-layer susceptibil-
ity to transition and separation. Different additives, such as polymers, surfactants,
microbubbles, droplets, particles, dust, or bers, can also be injected through the
surface in water- or air-wall-bounded ows. Control devices located away from the
surface can also be benecial. Large-eddy breakup devices (also called outer-layer
devices or OLDs); acoustic waves bombarding a shear layer from outside; additives
introduced in the middle of a shear layer; manipulation of freestream turbulence
levels; and spectra, gust, and magneto- and electrohydrodynamic body forces are
examples of ow-control strategies applied away from the wall.
A second scheme for classifying ow-control methods considers energy expendi-
ture and the control loop involved. As shown in the schematic in Figure 3.3, a control
device can be passive, requiring no auxiliary power and no control loop, or active,
requiring energy expenditure. As for the action of passive devices, some prefer to use
the term ow management rather than ow control (Fiedler and Fernholz 1990), re-
serving the latter terminology for dynamic processes. Active control requires a control
Figure 3.3: Classication of ow-control strategies.
3.2 CLASSIFICATION SCHEMES 29
Figure 3.4: Different control loops for active ow control. (a) Predetermined, open-loop
control; (b) Reactive, feedforward, open-loop control; (c) Reactive, feedback, closed-loop
control.
loop and is further divided into predetermined or reactive categories. Predetermined
control includes the application of steady or unsteady energy input without regard
to the particular state of the ow. The control loop in this case is open, as shown
in Figure 3.4a, and no sensors are required. Because no sensed information is being
fed forward, this open control loop is not a feedforward one. This subtle point is
often misunderstood in the literature and results in predetermined control being con-
fused with reactive, feedforward control. Reactive
3
control is a special class of active
control in which the control input is continuously adjusted based on measurements
of some kind. The control loop in this case can either be an open feedforward one
(Figure 3.4b) or a closed feedback loop (Figure 3.4c). Classical control theory deals,
for the most part, with reactive control.
The distinction between feedforward and feedback is particularly important when
dealing with the control of ow structures that convect over stationary sensors and
3
Also termed, by some, interactive ow control. Reactive could be confused with chemically reacting
ows but is more proper linguistically to describe this particular control strategy.
actuators. In feedforward control, the measured variable and the controlled variable
differ. For example, the pressure or velocity can be sensed at an upstream location,
and the resulting signal is used together with an appropriate control law to trigger an
actuator, which in turn inuences the velocity at a downstream position. Feedback
control, on the other hand, necessitates that the controlled variable be measured, fed
back, and compared with a reference input. Reactive feedback control is further clas-
sied into four categories: adaptive, physical-model-based, dynamical systems-based,
and optimal control (Moin and Bewley 1994). We will return to these categories in
Chapter 14.
Yet another classication scheme is to consider whether the control technique
directly modies the shape of the instantaneousmean velocity prole or selectively
inuences the small dissipative eddies. An inspection of the NavierStokes equations
written at the surface, Eqs. (2.39)(2.41), indicates that the spanwise and streamwise
4
vorticity uxes at the wall can be changed, either instantaneously or in the mean, via
wall motion or compliance, suction or injection, streamwise or spanwise pressure
gradient (respectively), or normal viscosity gradient. These vorticity uxes determine
the fullness of the corresponding velocity proles. For example, suction (or down-
ward wall motion), favorable pressure gradient, or lower wall viscosity results in
vorticity ux away from the wall, making the surface a source of spanwise and
streamwise vorticity. The corresponding fuller velocity proles have negative cur-
vature at the wall and are more resistant to transition and to separation but are
associated with higher skin-friction drag. Conversely, an inectional velocity prole
can be produced by injection (or upward wall motion), adverse pressure gradient,
or higher wall viscosity. Such prole is more susceptible to transition and to sep-
aration and is associated with lower, even negative, skin friction. Note that many
techniques are available to effect a wall-viscosity gradient; for example, surface heat-
ing or cooling, lm boiling, cavitation, sublimation, chemical reaction, wall injection
of lower or higher viscosity uid, and the presence of shear thinning or thickening
additive.
Flow-control devices can alternatively target certain scales of motion rather than
globally change the velocity prole. Polymers, riblets, and LEBUs, for example,
appear to selectively damp only the small dissipative eddies in turbulent wall-bounded
ows. These eddies are responsible for the (instantaneous) inectional prole and
the secondary instability in the buffer zone,
5
and their suppression leads to increased
scales, a delay in the reduction of the (mean) velocity-prole slope, and consequent
thickening of the wall region. In the buffer zone, the scales of the dissipative and
energy-containing eddies are roughly the same and, hence, the energy-containing
eddies will also be suppressed, resulting in reduced Reynolds stress production, mo-
mentum transport, and skin friction.
3.3 Free-Shear and Wall-Bounded Flows
Free-shear ows, such as jets, wakes, and mixing layers, are characterized by inec-
tional mean-velocity proles and are therefore susceptible to inviscid instabilities.
4
Streamwise vorticity exists only if the velocity eld is three-dimensional, instantaneously or in the mean.
5
See Chapter 5 for a description of the different regions of a turbulent wall-bounded ow.
3.4 REGIMES OF REYNOLDS AND MACH NUMBERS 31
Viscosity is only a damping inuence in this case, and the prime instability mech-
anism is vortical induction. Control goals for such ows include transition delay
or advancement, mixing enhancement, and noise suppression. External and internal
wall-bounded ows, such as boundary layers and channel ows, can also have inec-
tional velocity proles, but, in the absence of adverse pressure gradient and similar
effects, are characterized by noninectional proles and viscous instabilities, which
are then to be considered. These kind of viscosity-dominated, wall-bounded ows are
intrinsically stable and therefore are generally more difcult to control. Free-shear
ows and separated boundary layers, on the other hand, are intrinsically unstable
and lend themselves more readily to manipulation.
Free-shear ows originate from some kind of surface upstream, be it a nozzle,
a moving body, or a splitter plate, and ow-control devices can therefore be placed
on the corresponding walls, albeit far from the fully developed regions. Examples of
such control include changing the geometry of a jet exit from circular to elliptical
(Gutmark and Ho 1986), using periodic suction or injection in the lee side of a blunt
body to affect its wake (Williams and Amato 1989), and vibrating the splitter plate of
a mixing layer (Fiedler, Glezer, and Wygnanski 1988). These and other techniques are
extensively reviewed by Fiedler and Fernholz (1990), who offer a comprehensive list
of appropriate references, and more recently by Gutmark, Schadow, and Yu (1995);
Viswanath (1995); Fiedler (1998); and Gutmark and Grenstein (1999).
3.4 Regimes of Reynolds and Mach Numbers
The Reynolds number is the ratio of inertial to viscous forces and, absent centrifu-
gal, gravitational, electromagnetic and other unusual effects, Re determines whether
the ow is laminar or turbulent. It is dened as Re v
o
L/, where v
o
and L are,
respectively, suitable velocity and length scales, and is the kinematic viscosity. For
low-Reynolds-number ows, instabilities are suppressed by viscous effects and the
ow is laminar, as can be found in systems with large uid viscosity, small length
scale, or small velocity. The large-scale motion of highly viscous volcanic molten
rock and air and water ow in capillaries and microdevices are examples of laminar
ows. Turbulent ows seem to be the rule rather than the exception and occur in
or around most important uid systems such as airborne and waterborne vessels,
gas and oil pipelines, material processing plants, and the human cardiovascular and
pulmonary systems.
Because of the nature of their instabilities, free-shear ows undergo transition
at extremely low Reynolds numbers as compared with wall-bounded ows. Many
techniques are available to delay laminar-to-turbulence transition for both kinds of
ows, but these techniques are ineffective in the case of indenitely high Reynolds
numbers. Therefore, for Reynolds numbers beyond a reasonable limit, one should
not attempt to prevent transition but rather deal with the ensuing turbulence. Of
course early transition to turbulence can be advantageous in some circumstances
such as to achieve separation delay, enhanced mixing, or augmented heat trans-
fer. The task of advancing transition is generally simpler than trying to delay it.
Chapter 6 specically addresses the control of laminar-to-turbulence transition. For
now, we briey discuss transition control for various regimes of Reynolds and Mach
numbers.
Three Reynolds number regimes can be identied for the purpose of reducing skin
friction in wall-bounded ows. First, if the owis laminar, typically at Reynolds num-
bers based on distance fromleading edge <10
6
, then methods of reducing the laminar
shear stress are sought. These are usually velocity-prole modiers such as adverse
pressure gradient, injection, cooling (in water), and heating (in air) that reduce the
fullness of the prole at the increased risk of premature transition and separation.
Secondly, in the range of Reynolds numbers from 110
6
to 410
7
, active and pas-
sive methods to delay transition as much as possible are sought. These techniques can
result in substantial savings and are broadly classied into two categories: stability
modiers and wave cancellation. The skin-friction coefcient in the laminar at plate
can be as much as an order of magnitude less than that in the turbulent case. Note,
however, that all the stability modiers such as favorable pressure gradient, suction,
or heating (in liquids) result in an increase in the skin friction over the unmodied
Blasius layer. The object is, of course, to keep this penalty below the potential saving,
that is, the net drag will be above that of the at-plate laminar boundary layer but
presumably well below the viscous drag in the at-plate turbulent ow. Thirdly, for
Re > 4 10
7
, transition to turbulence cannot be delayed with any known practical
method without incurring a penalty that exceeds the saving.
6
The task is then to
reduce the skin-friction coefcient in a turbulent boundary layer. Relaminarization
(Narasimha and Sreenivasan 1979) is an option, although achieving a net saving here
is problematic at present.
The Mach number is the ratio of a characteristic ow velocity to local speed of
sound, Ma v
o
/a
o
. It determines whether the ow is incompressible (Ma < 0.3) or
compressible (Ma > 0.3). The latter regime is further divided into subsonic (Ma < 1),
transonic (0.8 < Ma < 1.2), supersonic (Ma > 1), and hypersonic (Ma > 5). Each
of these owregimes lends itself to different optimummethods of control to achieve a
given goal. Take laminar-to-turbulence transition control as an illustration (Bushnell
1994). During transition, the eld of initial disturbances is internalized via a pro-
cess termed receptivity, and the disturbances are subsequently amplied by various
linear and nonlinear mechanisms. If it is assumed that bypass mechanisms, such as
roughness or high levels of freestream turbulence, are identied and circumvented,
delaying transition then is reduced to controlling the variety of possible linear modes:
TollmienSchlichting modes, Mack modes, crossow instabilities, and G ortler insta-
bilities. TollmienSchlichting instabilities dominate the transition, process for two-
dimensional boundary layers having Ma <4 and are damped by increasing the Mach
number, by wall cooling (in gases), and by the presence of favorable pressure gra-
dient. Contrast this with the Mack modes, which dominate for two-dimensional
hypersonic ows. Mack instabilities are also damped by increasing the Mach num-
ber and by the presence of favorable pressure gradient but are destabilized by wall
cooling. Crossow and G ortler instabilities are caused by, respectively, the develop-
ment of inectional crossow velocity prole and the presence of concave streamline
curvature. Both of these instabilities are potentially harmful across the speed range
but are largely unaffected by Mach number and wall cooling. The crossow modes
are enhanced by favorable pressure gradient, whereas the G ortler instabilities are
insensitive. Suction suppresses, to different degrees, all the linear modes discussed
here.
6
Passive compliant coatings may be an exception. See Chapter 7.
3.5 CONVECTIVE AND ABSOLUTE INSTABILITIES 33
3.5 Convective and Absolute Instabilities
In addition to grouping the different kinds of hydrodynamic instabilities as inviscid
or viscous, one could also classify them as convective or absolute based on the linear
response of the system to an initial localized impulse (Huerre and Monkewitz 1990).
A ow is convectively unstable if, at any xed location, this response eventually
decays in time (in other words, if all growing disturbances convect downstream from
their source). Convective instabilities occur when there is no mechanismfor upstream
disturbance propagation, as for example in the case of rigid-wall boundary layers. If
the disturbance is removed, then perturbation propagates downstream, and the ow
relaxes to an undisturbed state. Suppression of convective instabilities is particularly
effective when applied near the point where the perturbations originate.
If any of the growing disturbances has zero group velocity, the ow is absolutely
unstable. This means that the local systemresponse to an initial impulse grows in time.
Absolute instabilities occur when a mechanismexists for upstreamdisturbance prop-
agation, as for example in the separated ow over a backward-facing step in which
the ow recirculation provides such a mechanism. In this case, some of the growing
disturbances can travel back upstreamand continually disrupt the oweven after the
initial disturbance is neutralized. Therefore, absolute instabilities are generally more
dangerous and more difcult to control; nothing short of complete suppression will
work. In some owsfor example, two-dimensional blunt-body wakescertain re-
gions are absolutely unstable whereas others are convectively unstable. The upstream
addition of acoustic or electric feedback can change a convectively unstable ow to
an absolutely unstable one, and self-excited ow oscillations can thus be generated.
In any case, identifying the character of ow instability facilitates its effective control
(i.e., suppressing or amplifying the perturbation as needed).
CHAPTER FOUR
Coherent Structures
It was six men of Indostan
To learning much inclined,
Who went to see the Elephant
(Though all of them were blind),
That each by observation
Might satisfy his mind.
The rst approached the Elephant,
And happening to fall
Against his broad and sturdy side,
At once began to bawl: God bless me! but the Elephant
Is very like a wall!
The second, feeling of the tusk,
Cried: Ho! what have we here
So very round and smooth and sharp?
To me tis mighty clear
This wonder of a Elephant
Is very like a spear!
The third approached the animal,
And happening to take
The squirming trunk within his hands,
Thus boldly up and spake: I see, quoth he, the Elephant
Is very like a snake.
The fourth reached out his eager hand,
And felt about the knee.
What most this wondrous beast is like
Is mighty plain, quoth he;
Tis clear enough the Elephant
Is very like a tree!
The fth, who chanced to touch the ear,
Said: Een the blindest man
Can tell what this resembles most: Deny the fact who can,
This marvel of an Elephant
Is very like a fan!
34
Cambridge Books Online Cambridge University Press, 2009
4.1 THE CHANGING PARADIGMS OF TURBULENCE 35
The sixth no sooner had begun
About the beast to grope,
Than, seizing on the swinging tail
That fell within his scope,
I see, quoth he, the Elephant
Is very like a rope!
And so these men of Indostan
Disputed loud and long,
Each in his own opinion
Exceeding stiff and strong,
Though each was partly in the right,
And all were in the wrong!
MORAL
So, oft in theologic wars,
The disputants, I ween,
Rail on in utter ignorance
Of what each other mean,
And prate about an Elephant
Not one of them has seen!
(John Godfrey Saxes, 18161887, version of the fable
The Blind Men and the Elephant, which occurs in the Ud ana,
a canonical Hindu scripture)
PROLOGUE
The relatively recent realization that organized structures play an important role in
all turbulent shear ows leads quite naturally to the concept of turbulence control
via direct interference with these deterministic events. Active, predetermined, open-
loop control can be employed, but perhaps reactive control can be used much more
effectively where specic coherent structures are sensed and then targeted for modu-
lation to achieve a useful end result such as drag reduction, lift enhancement, and so
forth. Reactive control strategies require sensors, actuators, and appropriate control
algorithms, as will be discussed in Chapter 14, but for the present chapter we provide
a gentle introduction to the fascinating world of coherent structures in transitional
and turbulent ows. First, we briey describe the different historical perspectives on
turbulence. Secondly, we dene what is meant by organized motions. This is followed
by a summary of what is known about coherent structures in free-shear ows. Lastly,
the important topic of coherent motions in wall-bounded ows is covered. This last
topic will be further explored from the perspective of Reynolds number effects in the
following chapter.
4.1 The Changing Paradigms of Turbulence
Turbulence is the last great unsolved problem of classical physics. Or so it goes for
a quote variously attributed to one of the great modern physicists Albert Einstein,
36 COHERENT STRUCTURES
Richard Feynman, Werner Heisenberg, or Arnold Sommerfeld. But in fact the closest
sentiments to this quote that could be traced were expressed by the classical physicist
Horace Lamb (1895), who actually wrote, starting with the second edition of his
celebrated book Hydrodynamics under the heading of Turbulent Motion: It remains
to call attention to the chief outstanding difculty of our subject. Ahumorous fable,
also attributed to several of the great ones, goes as follows. As he lay dying, the
modern physicist asked God two questions: Why relativity (or quantum mechanics,
depending on who is departing), and why turbulence? I really think, said the famed
physicist, He may have an answer to the rst question. No one knows how to
obtain stochastic solutions to the well-posed set of partial differential equations that
govern turbulent ows. Averaging those nonlinear equations to obtain statistical
quantities always leads to more unknowns than equations, and ad hoc modeling is
then necessary to close the problem. So, except for a rare few limiting cases, rst-
principle analytical solutions to the turbulence conundrum are not possible. In the
words of John Lumley (1996a), turbulence is a difcult problem that is unlikely
to suddenly succumb to our efforts. We should not await sudden breakthroughs
and miraculous solutions but rather keep at it slowly building one small brick at a
time.
Our struggle to conquer turbulence has been long and arduous with lots of sweat,
few victories, and much frustration. Not surprisingly, the way turbulence is being
viewed as a complex physical phenomenon has changed over the years. Indeed, key
ideas in the eld continue to rise and fall (Liepmann 1979) and perhaps even to rise
again! We now know that turbulence is random uctuations superimposed on mean
ow. A turbulent ow contains motions with numerous time and length scales that
are random in the sense that there is zero probability of any ow variable having a
particular value, and there is zero energy in any one particular frequency or wave
number. In other words, the probability density function and spectrum of any ow
variable are continuous and nite (Hunt, Carruthers, and Fung 1991). But we also
know that the seemingly random mess is at least in part deterministic: a combination
of coherent and incoherent motions.
Historically, there are perhaps ve doctrines in approaching the ve-century-old
conundrum: visualization, rst principles, the statistical approach, coherent struc-
tures, and modern tools. Each of these dogmas is described in turn in the following
ve subsections.
4.1.1 Visualization
Perhaps more than any other tool available to attack the problemof uid mechan-
ics in general and turbulence in particular, ow visualization is singly responsible for
many of the most exciting discoveries in the eld. Because visualization is relatively
simple and quick and is capable of giving both global and local behavior, rendering
the uid motion accessible to visual perception can yield invaluable qualitative as
well as quantitative information about a complex ow. However, one has to be ex-
tremely vigilant to avoid the many possible pitfalls when interpreting visual images
of uid ows, particularly time-dependent ows (Gad-el-Hak 1992).
Leonardo da Vinci pioneered the visualization genre close to 500 years ago. Much
of Leonardos notebooks of engineering and scientic observations were translated
into English in a magnicent two-volume book by MacCurdy (1938). Succulent
descriptions of the smooth and eddying motions of water alone occupy 121 pages.
Figure 4.1: Leonardo da Vincis sketch of water exiting from a square hole into a pool; circa 1500.
In them, one can easily discern the Renaissance geniuss foreshadowing of some of
the turbulence physics to be discovered centuries after his time. Particularly relevant
to the present subject, the words eddies and eddying motions percolate throughout
Leonardos treatise on liquid ows.
Figure 4.1 represents perhaps the worlds rst use of visualization as a scientic
tool to study a turbulent ow. Around 1500, Leonardo sketched a free water jet
issuing from a square hole into a pool. He wrote, Observe the motion of the surface
of the water, which resembles that of hair, which has two motions, of which one is
caused by the weight of the hair, the other by the direction of the curls; thus the water
has eddying motions, one part of which is due to the principal current, the other to the
random and reverse motion.
1
Reecting on this passage, Lumley (1992) speculates
that Leonardo da Vinci may have pregured the now famous Reynolds turbulence
decomposition nearly 400 years prior to Osborne Reynolds own ow visualization
and analysis!
In describing the swirling water motion behind a bluff body, da Vinci provided
the earliest reference to the importance of vortices in uid motion: So moving water
strives to maintain the course pursuant to the power which occasions it and, if it
nds an obstacle in its path, completes the span of the course it has commenced
by a circular and revolving movement. Leonardo accurately sketched the pair of
quasi-stationary, counterrotating vortices in the midst of the random wake.
Finally, da Vincis words . . . The small eddies are almost numberless, and large
things are rotated only by large eddies and not by small ones, and small things are
1
This particular translation from the Italian original was made by Ugo Piomelli, University of Maryland.
It differs in several key points from that of MacCurdy (Watch the movement of the surface of water,
how like it is to that of hair, which has two movements, one following the undulation of the surface, the
other the lines of the curves: thus water forms whirling eddies, part following the impetus of the chief
current, part the rising and falling movement.), but on the basis of careful analysis Piomellis phrasing
appears to be more precise.
turned by both small eddies and large presage Richardsons cascade,
2
coherent struc-
tures, and large-eddy simulations, at least (John L. Lumley, private communication).
4.1.2 First Principles
At the time of Leonardo da Vinci, neither calculus nor the laws of mechanics
were available, of course.
3
Little more than a century and half after the incompara-
ble Newtons Principia Mathematica was published in 1687, the rst principles of
viscous uid ows were rmed in the formof the NavierStokes equations with major
contributions by Navier in 1823, Cauchy in 1828, Poisson in 1829, Saint-Venant in
1843, and Stokes in 1845. These equations of uid motion were derived in Chapter 2.
With very few exceptions, the NavierStokes equations provide an excellent model
for both laminar and turbulent ows. But in the latter case no analytical solutions
are possible for several reasons: (1) the governing partial differential equations are
nonlinear; (2) the dependent variables are random functions of space and time; and
(3) the usual simplications and symmetries do not apply to the instantaneous, three-
dimensional owquantities (although they may apply to statistical quantities). Thus,
even though the turbulence problem was well dened mathematically, no one could
integrate the equations of motion or for that matter do much with themwhen dealing
with stochastic phenomena.
For a century and half only laminar ows and their stability were tractable an-
alytically. Recent advances in computer power made it possible to return to rst
principles in the case of a turbulent owand numerically integrate the NavierStokes
equations for simple geometries and rather lowReynolds numbers. Notwithstanding
their rather severe restrictions and limitations, direct numerical simulations (DNS)
are therefore the only route available for a direct, brute-force onslaught on the tur-
bulence problem. On the other hand, numerical solutions overwhelm the senses with
data while providing little physical understanding.
4.1.3 Statistical View of Turbulence
As is clear fromthe previous subsection, up until recently not much could be done
with rst principles for the turbulence problem. But do we really need the kind of
detailed information that DNS provides for all the random ow variables? Would a
statistical description yielding such quantities as mean or root-mean-square values,
correlation functions, spectra, and probability distributions sufce? The answer is
an emphatic yes, but there is a price to pay for the lowered expectations.
On the basis of a painstaking ow visualization study, Osborne Reynolds (1883)
made the observation that a pipe ow is either direct or sinuous (laminar or tur-
bulent), depending on the value of a dimensionless parameter that we now call the
Reynolds number. Eleven years later, on 24 May 1894, Reynolds read to an audience
of the British Royal Society a follow-up paper in which he attempted to provide a
physical explanation for his earlier observation. That second treatise, published in
1895, is considered by many to have pioneered the modern scientic approach to
2
Richardsons (1920, 1922) poetic description (widely recited but often misquoted) of the turbulence
eddies within a cumulus cloud reads: Big whirls have little whirls that feed on their velocity, and little
whirls have lesser whirls and so on to viscosity.
3
Refer to Figure 1.1.
the turbulence problem and even to have reshaped the direction of uid mechanics
research in general for the next century. Reynolds (1895) asserted that a turbulent
oweld could be decomposed into mean and uctuating parts. He thus was able
to write expressions for the time-averaged momentum, now known as the Reynolds
equations, in which convective stress terms appear as new unknowns. This was the
dawn of the statistical doctrine of turbulence research. Reynolds (1895) also derived
the transport equation for the turbulence kinetic energy and demonstrated that the
apparent stresses due to turbulence interacting with the mean velocity gradient lead
to a transfer of kinetic energy from the mean motion to the turbulent.
As was illustrated in Section 2.5, in the statistical approach a temporal, spatial, or
ensemble average is dened, and the equations of motion are written for the various
moments of the uctuations about this mean. Unfortunately, the nonlinearity of the
NavierStokes equations guarantees that the process of averaging to obtain moments
results in an open system of equations in which the number of unknowns is always
greater than the number of equations and more or less heuristic modeling is used
to close the equations. This is known as the closure problem, which again makes
obtaining rational solutions to the (averaged) equations of motion impossible.
Attempts to close the Reynolds equations are at the heart of the turbulence mod-
eling community. From the simple mixing length ideas of Prandtl, Taylor, and von
K arm an to the more involved Reynolds-stress or second-order modeling and beyond,
a whole new industry sprang out of the Reynolds decomposition (see, for example,
Lumley 1983, 1987; Speziale 1991; Wilcox 1993).
4.1.4 Reemergence of Coherent Structures
The recognition of coherent structures during the last few decades brought us
back a full circle to the time of Leonardo. Not only was visualization once again the
method of choice for the major discoveries, but it was also the reafrmation of the
importance of eddying motions and the copresence of large, organized motions and
small, random ones. In the view of Hussain (1986), the search for coherent events is
the embodiment of mans desire to nd order in apparent disorder.
The recent history of coherent structures is amply chronicled in the article by
Liu (1988). He asserts that the kernel idea germinated as a result of a discussion
that took place during the Fifth International Congress of Applied Mechanics held in
1938. There, both Tollmien and Prandtl, responding to a comment by von K arm an
regarding the difculties of reconciling a scalar mixing length with turbulence mea-
surements made in a channel, suggested that the measured turbulence uctuations
include both random and nonrandom elements. Dreyden (1948) pointed out that
the boundary layer measurements conducted at a later date at the National Bureau
of Standards supported the ideas of Tollmien and Prandtl. Dreyden also lamented
that there is no known procedure, experimental or theoretical, that can be used to
separate the random processes from the nonrandom ones.
Liepmann (1952), citing measurements in free-shear turbulent ows, in ow be-
tween rotating cylinders, and in the far-wake of a cylinder, emphasized the importance
of the presence of a secondary, large-scale structure superimposed upon the primary
turbulent shear ow. Townsend (1956) thoroughly exploited this concept in the rst
edition of his famed monograph on the structure of turbulent shear ows. He rec-
ognized the quasi-deterministic nature of large eddies and inferred their shapes from
the long-time-averaged spatial-correlation tensor measured in a Eulerian frame.
Townsends approach suffers from several shortcomings, including the lack of a
unique relationship between the correlation tensor and the unsteady ow that pro-
duces it.
In a later article, Liepmann (1962) once again underscored the splitting of seem-
ingly random uctuations into large-scale, deterministic structures and ne-grained
turbulence. Liepmann asserted the importance of large-scale structures in many tech-
nological problems in aerodynamic sound, combustion, and so forth. Liepmann
(1979) was perhaps the rst to suggest that the existence of deterministic eddies in
turbulent ows can be exploited for control purposes via direct interference with
these large structures.
The modern view of coherent structures resulted from ow visualization stud-
ies of low-Reynolds-number boundary layers conducted rst at the University of
Maryland and then at Stanford University during the late 1950s and 1960s. The new
doctrine did not pick up steam, however, until the milestone discovery of Brown and
Roshko (1971, 1974) of organized motion in a mixing layer at Reynolds numbers far
exceeding transitional ones. The large spanwise vortices, prominent in visual images
of the shear layer, were totally missed in the correlation studies of Townsend and
others.
4.1.5 The Latest Tools
Finally, dynamical systems and wavelets are the modern (i.e., the 1990s and be-
yond, at least for now) tools to tackle the last conundrum. A turbulent ow is a com-
plex, nonlinear dynamical system that has an innite number of degrees of freedom.
The issue here is the possibility of representing such a system with a reasonable
number of degrees of freedom. A Fourier decomposition would not do because it
requires a very large number of components.
Mechanical systems with three or more degrees of freedom are capable of chaotic
behavior exemplied by a strange attractor in phase space. Such systems are complex,
aperiodic, and random and display extreme sensitivity to initial conditions, but they
are nevertheless still deterministic. The book by Holmes et al. (1996) provides an
excellent introduction to the dynamical system approach to tackle the turbulence
problem.
Under rather severe restrictions, chaos theory allows the representation of tur-
bulence as a low-dimensional dynamical system. The ow is modeled as coherent
structures plus a parameterized turbulent background. The proper orthogonal de-
composition, or KarhunenLo` eve decomposition, has been of great use because it is
capable of representing the ow with a minimum number of modes. Such a repre-
sentation is useful on two fronts:
1. It provides an inexpensive
4
surrogate to the turbulent ow and in the process
sheds light on its basic physics.
2. Chaos control concepts can be utilized to achieve effective manipulation with
minimum energy expenditure once the ow is successfully represented as a
dynamical system with a reasonably small number of degrees of freedom. We
will return to the latter front in Chapter 14.
4
As compared with DNS.
The dynamical system approach thus far has been successful for ows near
transition or near a solid wall, in which cases the associated low Reynolds num-
ber implies that a relatively small number of degrees of freedom are excited and
that a large fraction of the energy is in the ordered, deterministic component of the
ow. More and more degrees of freedom are excited as the ow moves farther from
transition or away from a wall. In these cases, the structure of the strange attractor
becomes so complex as to negate the dynamical system approach advantages over
the classical statistical description.
The second modern tool is wavelets, which was introduced a little more than a
decade ago. Wavelet transform is used in many elds, including signal processing,
data compression, image coding, and numerical analysis (Wickerhauser 1994). The
technique is based on group theory and square integrable representation in terms of
basis functions, called wavelets, that are localized in both physical and wave-number
spaces. Farges (1992) review article provides a good introduction to the eld and
particularly its application to turbulence. The more recent paper by Vasilyev, Yuen,
and Paolucci (1997) offers a gentle introduction to the use of wavelets for numerically
solving complex, multiscale partial differential equations.
Wavelets allow the unfolding of a oweld into both space and scale and pos-
sibly even direction. Wavelets are localized analyzing functions that are dilated or
contracted before convolving with the signal under consideration to achieve scale
decomposition. Wavelet analysis can be viewed as a multilevel or multiresolution
representation of a function, each level of resolution consisting of basis functions
that have the same scale but are located at different positions. In contrast to Fourier
transform, wavelet transform is a local one, and the behavior of the signal at innity
plays no role in the analysis. Continuous wavelet transforms offer redundant unfold-
ing in terms of space and scale and are thus suited for tracking coherent motions and
their contributions to the energy spectrum. Discrete wavelet transforms, on the other
hand, allow orthonormal projection on a minimal number of independent modes
and may thus be used to model the ow dynamics.
Basis functions such as Mexican-hat wavelets or Daubechies scaling functions
can be used to decompose a velocity eld into eddies. Because coherent structures
are always of limited spatial extent, wavelet decomposition seems to be a better
representation of them than, say, Fourier representation. Fourier modes in the form
of (space-lling) trigonometric functions stretch off to innity and are thus suited
for decomposing a velocity eld into waves of different, independent wavelengths.
On the other hand, a limited-extent eddy is ideally represented by modes that act in
groups as the wave number increases. Wavelets offer complete representation at least
for homogeneous turbulence.
5
For inhomogeneous ows, a complete representation
using wavelets is not as readily achieved, and some difculties remain to be resolved.
6
Nevertheless, the technique offers some potential advantages from the point of view
of controlling turbulent ows. Specically, wavelet transforms may be used, for
example, as an efcient, unbiased strategy for real-time identication of coherent
structures from an instantaneous velocity signal. This step is of course at the heart
of effective reactive control.
5
As do a variety of conventional techniques with varying degrees of computational efciency and accuracy.
6
Those difculties in fact are being resolved as we go to the press (Samuel Paolucci, private commu-
nication).
4.2 What Is a Coherent Structure?
The statistical view that turbulence is essentially a stochastic phenomenon having a
randomly uctuating velocity eld superimposed on a well-dened mean has been
changed in the last few decades by the realization that the transport properties of all
turbulent shear ows are dominated by quasi-periodic, large-scale vortex motions
(Laufer 1975; Townsend 1976; Cantwell 1981; Fiedler 1988). Despite the extensive
research work in this area, no generally accepted denition of what is meant by co-
herent motion has emerged. In physics, coherence stands for a well-dened phase
relationship. We provide here two rather different views, the rst is general and the
second is more restrictive. According to Robinson (1991), a coherent motion is
dened as a three-dimensional region of the ow over which at least one fundamen-
tal ow variable (velocity component, density, temperature, etc.) exhibits signicant
correlation with itself or with another variable over a range of space and/or time that
is signicantly larger than the smallest local scales of the ow. The rather restrictive
denition is given by Hussain (1986): A coherent structure is a connected turbulent
uid mass with instantaneously phase-correlated vorticity over its spatial extent. In
other words, underlying the random, three-dimensional vorticity that characterizes
turbulence, there is a component of large-scale vorticity that is instantaneously co-
herent over the spatial extent of an organized structure. The apparent randomness of
the oweld is, for the most part, due to the randomsize and strength of the different
type of organized structures comprising that eld. Several other denitions are cat-
alogued by Delville et al. (1998). The same authors also provide a cook-book-style
approach to coherent structure identication using a variety of classical and modern
strategies.
The challenge is to identify a coherent structure well hidden in a sea of random
background when such a structure is present either in a visual impression of the ow
or in an instantaneous velocity, temperature, or pressure signal. This is of course not
a trivial task, and the ancient Hindu fable
7
about the six blind men, each trying from
his own limited perspective to identify how an elephant looks, immediately comes to
mind. Complicating the issue is that coherent structures change fromone type of ow
to another and even in the same type of ow as initial and boundary conditions vary.
The largest eddies are of the same scale as that of the ow and consequently cannot
be universal. Identifying a coherent structure based on certain dynamical properties
is more likely to succeed but is quite involved. On the other hand, a kinematic
detector based on its creators perception of the dynamic behavior of the organized
motion is simpler to employ but runs the risk of detecting the presence of nonexistent
objects (Lumley 1981). Beneting fromhindsight, the fewow-visualization pictures
depicted in the next two sections may help in exploring the nature of the whole
beast.
4.3 Free-Shear Flows
As indicated above, there is no universal coherent structure. Organized motions in
wakes are different from those in boundary layers. In a jet issuing from a nozzle with
7
Recited at the beginning of this chapter with all six blind men. This ancient story is often misquoted in
the recent literature with three, four, or ve of the lads.
4.3 FREE-SHEAR FLOWS 43
a thin, laminar boundary layer on its inner surface, coherent structures are easily
observed, whereas a jet issuing from a nozzle with a thick, turbulent boundary layer
has organized structures that are nearly undetectable. The proportion of organized
turbulence decreases, in general, with the level of disturbances in the incoming ow.
In general again, coherent structures in free-shear ows are easier to detect and
characterize than those in wall-bounded ows. According to Liu (1988), for free-
shear turbulent ows it is not necessary to conjecture that the local ne-grained
turbulence rearranges itself to give bursts of white noise in order to maintain the
hydrodynamically unstable waves as in the case of wall-bounded ows. The exis-
tence of large-scale, coherent motions in mixing layers, jets, and wakes is instead a
manifestation of the dynamic instability associated with the local inectional mean-
velocity proles. As a result, free-shear ows have pronounced organized motions
and wavelike structures (Fiedler 1988, 1998).
In a mixing layer, the dominant structures are rollerlike vortices as large as the
shear layer itself (Brown and Roshko 1971, 1974; Roshko 1976). The growth of
the mixing layer results from the amalgamation of neighboring large eddies rather
than fromtheir individual growth (Winant and Browand 1974). The large transverse
eddies in a mixing layer are strung together by a spaghetti-like net of smaller-scale,
streamwise, counterrotating vortices. If the mixing layer develops from undisturbed
conditions (i.e., thin, laminar boundary layers on the splitter plate), the rollerlike
vortices are energetic and only relatively slowly become three-dimensional and less
organized. If, on the other hand, the splitter plate has thick, turbulent boundary
layers, the proportion of organized motion is considerably less dominant.
The spark-shadowgraph photograph in Figure 4.2 shows the mixing of a fast-
moving stream of helium (top) and a slower stream of nitrogen (bottom), both mov-
ing from left to right. The two streams have the same mean momentum per unit
cross-sectional area,
1
U
2
1
=
2
U
2
2
, and originate from a splitter plate with laminar
boundary layers. The rollerlike vortices convect at nearly constant speed equal to
the average
1
2
(U
1
+U
2
). Increasing the Reynolds number produces more small-scale
structures without signicantly altering the large eddies.
The spanwise vortices remain as a permanent dominant feature even at high
Reynolds numbers. Brown and Roshkos (1971, 1974) experiment is important be-
cause it was the rst to show a signicant proportion of organized turbulence even
at Reynolds numbers far from transitional. Moreover, a much closer tie between
Figure 4.2: High-Reynolds-number mixing layer. The helium stream on top moves at a velocity of 10 m/s,
and the nitrogen stream on the bottom moves at a speed of 3.78 m/s. The whole test section is pressurized
to P = 8 atm, giving a Reynolds number based on downstream distance of the order of 10
6
[after Brown
and Roshko 1974].
Figure 4.3: Transitional subsonic air jet [after Bradshaw, Ferris, and Johnson 1964].
stability and turbulence has been established: The essentially two-dimensional vor-
tices in the fully turbulent ow are clearly related to the general instability modes of
a simple vortex sheet.
The organized structures in jets are not as dominant as those in mixing layers
having similar levels of disturbances in the incoming ow. The schlieren photograph
in Figure 4.3 depicts a subsonic air jet issuing at a speed of 12 m/s from a 5-cm-
diameter nozzle on the left, giving a Reynolds number of Re 410
4
. The laminar
jet undergoes transition shortly after exiting and eventually becomes turbulent. The
initial instabilities in the form of vortex rings appear to originate at the interface of
the jet and the surrounding potential ow and quickly become three-dimensional as
longitudinal striations develop.
In contrast to mixing layers, the spanwise vortices shed in the wake of bluff
bodies do not pair. The basic mechanism for wake growth is entrainment. The two
photographs in Figure 4.4 contrast the low-Reynolds-number turbulent wake (Re =
4300) behind a at plate at a 45
angle of attack and the wake behind the tanker

Argo Merchant, which was grounded on the Nantucket shoals in 1976. The labo-
ratory ow is visualized using aluminum akes suspended in water. Leaking crude
oil indicates that the Argo Merchant happened to be inclined at about 45
to the
current. Although the Reynolds number for the tank ship at Re 10
8
is more than
four orders of magnitude higher, the two wakes are remarkably similar.
4.4 Wall-Bounded Flows
We nowturn our attention to the more enigmatic boundary layers and channel ows.
In a wall-bounded ow, a multiplicity of coherent structures have been identied
4.4 WALL-BOUNDED FLOWS 45
Figure 4.4: Turbulent wakes behind bluff bodies. (a) Inclined at plate in water tunnel; from Cantwell
(1981). (b) Grounded tanker [NASAphotograph courtesy of O.M. Grifn, Naval Research Laboratory].
mostly through ow visualization experiments, although some important early dis-
coveries have been made using correlation measurements (e.g., Townsend 1961,
1970; Bakewell and Lumley 1967). Although the literature on this topic is vast, no
research-community-wide consensus has been reachedparticularly on the issues of
the origin of, and interaction between, the different structures, regeneration mech-
anisms, and Reynolds number effects. What follow are somewhat biased remarks
addressing these issues. At times diverse viewpoints will be presented, but for the
most part particular scenarios, which in my opinion are most likely to be true, will
be emphasized. The interested reader is referred to the book edited by Panton (1997a),
which emphasizes the self-sustaining mechanisms of wall turbulence, and the large
number of review articles available (e.g., Kovasznay 1970; Laufer 1975; Willmarth
1975a,b; Saffman 1978; Cantwell 1981; Fiedler 1986, 1988; Blackwelder 1988,
1998; Robinson 1991; Delville et al. 1998). The paper by Robinson (1991) in partic-
ular summarizes many of the different, sometimes contradictory, conceptual models
offered thus far by different research groups. Those models are aimed ultimately at
explaining how the turbulence maintains itself and range from the speculative to the
rigorous, but not one, unfortunately, is self-contained and complete. Furthermore,
the structure research dwells largely on the kinematics of organized motion, and little
attention is given to the dynamics of the regeneration process.
4.4.1 Overview
With few exceptions, most of the available structural information on wall-
bounded ows come from rather low-Reynolds-number experiments and numerical
simulations. Organized structures appear to be similar in all wall-bounded ows
only in the inner layer. The outer region of a boundary layer is by necessity different
from the core region of a pipe or channel ow. An overall view, whose source of
information is predominately low-Reynolds-number experiments, is presented here.
As will become clear throughout the discussion here and in the following chapter, the
picture that emerges at high Reynolds numbers is quite different, and structural in-
formation gleaned from low-Reynolds-number physical and numerical experiments
may not be very relevant to the more practically important high-Reynolds-number
ows.
In boundary layers, the turbulence production process is dominated by three kinds
of quasi-periodic eddies: the large outer structures, the intermediate Falco eddies, and
the near-wall eddies. Examples of these coherent structures visualized in laboratory-
scale boundary layers are depicted in Figures 4.54.7. Laser sheet illumination is
used in all three photographs. The large eddies forming on a at plate towed in
a water channel are seen in the side view in Figure 4.5. The ow (relative to the
plate) is from left to right. The articially tripped boundary layer has a momentum-
thickness Reynolds number at the observation station of Re
/ = 725 and
is marked with uorescent dye. The smoke-lled boundary layer shown in the top
view in Figure 4.6 depicts the characteristic pockets believed to be induced by the
motion of Falco eddies over the wall. In this case, the experiments are conducted in
a wind tunnel at a Reynolds number of Re
= 742, and the boundary layer is again

articially tripped. Finally, the top view in Figure 4.7 depicts the low-speed streaks
in the near-wall region of the same turbulent boundary layer previously shown in
side view in Figure 4.5. Flow direction is again from left to right, and the thin sheet
of laser used for illumination is parallel to and almost touching the wall.
Figure 4.8, from Gad-el-Hak, Blackwelder, and Riley (1981), shows a top view
of an articially generated turbulent spot evolving in a laminar boundary layer. The
displacement thickness Reynolds number at the spots initiation point is 625, which
is well above the critical Re for linear instability. Laser-induced uorescence is used
to visualize different cuts through the growing turbulent structure. In this gure, the
laser sheet is parallel to and very near the at wall. The dynamics within the spot
appear to be controlled by many individual eddies similar to those within a turbulent
boundary layer. The spot grows in the spanwise direction by an efcient mechanism,
Figure 4.5: Side view of a low-Reynolds-number turbulent boundary layer, Re
= 725. Flat
plate towed in a water tank. Large eddies are visualized using a sheet of laser and uorescent
dye [after Gad-el-Hak, Blackwelder, and Riley 1984].
Figure 4.6: Top viewof a low-Reynolds-number turbulent boundary layer, Re
=
742. Wind tunnel experiment. Pockets, believed to be the ngerprints of typical
eddies, are visualized using dense smoke illuminated with a sheet of laser [after
Falco 1980].
Books Online Cambridge University Press, 2009
Figure 4.7: Top view of a low-Reynolds-number turbulent boundary layer Re
= 725. Flat plate towed in

a water tank. Low-speed streaks are visualized using a sheet of laser and uorescent dye [after Gad-el-Hak,
Blackwelder, and Riley 1984].
Figure 4.8: Top view of an articially generated turbulent spot evolving in a laminar boundary layer. The
displacement thickness Reynolds number at the spots initiation point is Re
= 625 [after Gad-el-Hak

et al. 1981].
which Gad-el-Hak et al. (1981) have termed the growth by destabilization process
(see also Riley and Gad-el-Hak 1985). Near the edges of the spot, the dye lines are
sharp, which is indicative of the initial breakdown into chaotic motion. Toward the
middle, on the other hand, the dye becomes more diffused because the turbulence
there is older and more mixing has taken place.
In a turbulent boundary layer, the large, three-dimensional bulges (Figure 4.5)
scale with the layer thickness, , and extend across the entire boundary layer
(Kovasznay et al. 1970; Blackwelder and Kovasznay 1972). These eddies control the
dynamics of the boundary layer in the outer region such as entrainment, turbulence
production, and so forth. The large eddies are characterized by a sharp interface and
a highly contorted surface that exhibits a signicant amount of folding (Paiziz and
Schwarz 1974) and has a fractal dimension of close to 2.4 (Sreenivasan, Ramshankar,
and Meneveau 1989). They appear randomly (quasi-periodically) in space and time
and seem to be, at least for moderate Reynolds numbers, the residue of the transi-
tional Emmons spots (Zilberman, Wygnanski, and Kaplan 1977; Gad-el-Hak et al.
1981; Riley and Gad-el-Hak 1985). Note, however, that at higher Reynolds numbers
(Re
>5000) the very existence of the large eddy as an isolated coherent structure
has been questioned by Head and Bandyopadhyay (1981).
The Falco eddies are also highly coherent and three-dimensional. Falco (1974,
1977) named them typical eddies because they appear in wakes, jets, Emmons spots,
grid-generated turbulence, and boundary layers in zero, favorable, and adverse pres-
sure gradients. They have an intermediate scale of about 100 wall units. (Viscous
forces dominate over inertia in the wall region. The characteristic scales there are
obtained from the magnitude of the mean vorticity in the region and its viscous dif-
fusion away from the wall. Thus, the viscous time-scale, t
, is given by the inverse of

the mean wall vorticity
t
U
y
1
(4.1)
and the viscous length scale,
, is determined by the characteristic distance by which

the (spanwise) vorticity is diffused from the wall, and is thus given by
U
y
w
(4.2)
where is the kinematic viscosity. The wall-velocity scale (so-called friction velocity,
u
) follows directly from the time and length scales

u
U
y
w
=
(4.3)
where
w
is the shear stress at the wall, and is the uid density. A wall unit implies
scaling with the viscous scales, and the usual ()
+
notation is used; for example,
y
+
= y/
= y u
/.)
The Falco eddies appear to be an important link between the large structures
and the near-wall events. In the plan view shown in Figure 4.6, smoke lls the
near-wall region of a boundary layer, and the roughly circular regions devoid of
marked uid are called pockets. These undulations are very similar to the so-called
folds observed by Perry, Lim, and Teh (1981). Falco (1980) asserted that the pock-
ets are the footprints of some outer structures that induce uid toward the wall.
Robinson, Kline, and Spalart (1989) analyzed the database generated from the di-
rect numerical simulations of Spalart (1986, 1988). They concurred that the pock-
ets are the signature of local wallward motions, evidenced by spanwise divergence
of streamlines, above regions of high wall pressure. Low-pressure regions, on the
other hand, occur along lines of converging streamlines associated with outward
motion. These motions are, respectively, the sweep and ejection events depicted in
Figure 4.9.
The third kind of eddies exist in the wall region (0 y
+
100), where the
Reynolds stress is produced in an intermittent fashion. Half of the total production
of turbulence kinetic energy (uv U/y) takes place near the wall in the rst 5% of
the boundary layer at typical laboratory Reynolds numbers (smaller fraction of the
Figure 4.9: Proposed sequence of the bursting process. The arrows indicate the sequen-
tial events, and the ? indicates relationships with less supporting evidence [adapted
from Blackwelder 1998].
boundary layer thickness at higher Reynolds numbers), and the dominant sequence
of intense organized motions there are collectively termed the bursting phenomenon.
This dynamically signicant process, identied during the 1960s by researchers at
Stanford University (Kline and Runstadler 1959; Runstadler, Kline, and Reynolds
1963; Kline et al. 1967; Kim, Kline, and Reynolds 1971; Offen and Kline 1974,
1975), was reviewed by Willmarth (1975a) and Blackwelder (1978) and most recently
by Blackwelder (1998).
To focus the discussion on the bursting process and its possible relationships to
other organized motions within the boundary layer, we refer to the schematic in
Figure 4.9, adapted from Blackwelder (1998). Qualitatively, the process, according
to at least one school of thought, begins with elongated, counterrotating, streamwise
vortices having diameters of approximately 40 wall units or 40/u
. The estimate
for the diameter of the vortex is obtained from the conditionally averaged spanwise
velocity proles reported by Blackwelder and Eckelmann (1979). There is a distinc-
tion, however, between vorticity distribution and a vortex (Saffman and Baker 1979;
Robinson et al. 1989; Robinson 1991), and the visualization results of Smith and
Schwartz (1983) may indicate a much smaller diameter. In any case, with reference
to Figure 4.10, the counterrotating vortices exist in a strong shear and induce low-
and high-speed regions between them. Those low-speed streaks were rst visual-
ized by Francis Hama at the University of Maryland (see Corrsin 1957), although
Hamas contribution is frequently overlooked in favor of the subsequent and more
thorough studies conducted at Stanford University and cited above. The vortices
and the accompanying eddy structures occur randomly in space and time. How-
ever, their appearance is sufciently regular that an average spanwise wavelength
Figure 4.10: Model of near-wall structure in turbulent wall-bounded ows [after
Blackwelder 1978].
of approximately 80 to 100/u
has been identied by Kline et al. (1967) and

others.
It might be instructive at this point to emphasize that the distribution of streak
spacing is very broad. The standard of deviation is 3040 percent of the more
commonly quoted mean spacing between low-speed streaks of 100 wall units. Both
the mean and standard deviation are roughly independent of Reynolds number in
the rather limited range of reported measurements (Re
=3006500, see Smith and

Metzler 1983; Kim, Moin, and Moser 1987). Butler and Farrell (1993) have shown
that the mean streak spacing of 100/u
is consistent with the notion that this is

an optimal conguration for extracting the most energy over an appropriate eddy
turnover time. In their work, the streak spacing remains 100 wall units at Reynolds
numbers, based on friction velocity and channel half-width, of a
+
= 180360.
Kim et al. (1971) observed that the low-speed regions (Figure 4.7) grow down-
stream, lift up, and develop (instantaneous) inectional u(y) proles.
8
At approx-
imately the same time, the interface between the low- and high-speed uid begins
to oscillate, apparently signaling the onset of a secondary instability. The low-speed
region lifts up away from the wall as the oscillation amplitude increases, and then
the ow rapidly breaks up into a completely chaotic motion. The streak oscilla-
tions commence at y
+
10, and the abrupt breakup takes place in the buffer layer,
although the ejected uid reaches all the way to the logarithmic region. Because
the breakup process occurs on a very short timescale, Kline et al. (1967) called it a
burst.
Virtually all of the net production of turbulence kinetic energy in the near-wall
region occurs during these bursts. Corino and Brodkey (1969) showed that the low-
speed regions are quite narrow (i.e., 20/u
) and may also have signicant shear in

the spanwise direction. They also indicated that the ejection phase of the bursting
process is followed by a large-scale motion of upstream uid that emanates from the
outer region and cleanses (sweeps) the wall region of the previously ejected uid. The
sweep phase is, of course, required by the continuity equation and appears to scale
with the outer-ow variables. The sweep event seems to stabilize the bursting site, in
effect preparing it for a new cycle.
Considerably more has been learned about the bursting process during the last
two decades. For example, Falco (1980, 1983, 1991) has shown that when a typical
eddy, which may be formed in part by ejected wall-layer uid, moves over the wall, it
induces a high uv sweep (positive u and negative v). The wall region is continuously
bombarded by pockets of high-speed uid originating in the logarithmic and possibly
the outer layers of the ow. These pockets appear to scale, at least in the limited
Reynolds number range in which they have been observed, Re
= O[1000], with
wall variables and tend to promote and enhance the inectional velocity proles by
increasing the instantaneous shear, leading to a more rapidly growing instability. The
relation between the pockets and the sweep events is not clear, but it seems that the
former forms the highly irregular interface between the latter and the wall-region
uid. More recently, Klewicki, Murray, and Falco (1994) conducted a four-wire hot-
wire probe measurement in a low-Reynolds-number canonical boundary layer to
8
According to Swearingen and Blackwelder (1984), inectional u(z) proles are just as likely to be found
in the near-wall region and can also be the cause of the subsequent bursting events. See Figure 4.9.
clarify the roles of velocityspanwise vorticity eld interactions regarding the near-
wall turbulent stress production and transport.
Other signicant experiments were conducted by Tiederman and his students
(Donohue, Tiederman, and Reischman 1972; Reischman and Tiederman 1975;
Oldaker and Tiederman 1977; Tiederman, Luchik, and Bogard 1985), Smith and
his colleagues (Smith and Metzler 1982, 1983; Smith and Schwartz 1983), and the
present author and his collaborators. The rst group conducted extensive studies of
the near-wall region, particularly the viscous sublayer, of channels with Newtonian
as well as drag-reducing non-Newtonian uids. Smiths group, using a unique two-
camera, high-speed video system, was the rst to indicate a symbiotic relationship
between the occurrence of low-speed streaks and the formation of vortex loops in the
near-wall region. Gad-el-Hak and Hussain (1986) and Gad-el-Hak and Blackwelder
(1987a) have introduced methods by which the bursting events and large-eddy struc-
tures are articially generated in a boundary layer. Their experiments greatly facilitate
the study of the uniquely controlled simulated coherent structures via phase-locked
measurements.
4.4.2 Open Issues
There are at least four unresolved issues regarding coherent structures in wall-
bounded ows, not all of which are necessarily independent: How does a particu-
lar structure originate? How do different structures, especially the ones having dis-
parate scales, interact? How does the turbulence continue to regenerate itself? Does
the Reynolds number affect the different structures in any profound way? The pri-
mary difculty in trying to answer any of these queries stems from the existence of
two scales in the ow that become rather disparate at large Reynolds numbers. The
closely related issues of origin, innerouter interaction, and regeneration will be ad-
dressed in the following three subsubsections. Reynolds number effects are deferred
to Chapter 5.
Origin of Different Structures
Faced with the myriad of coherent structures existing in the boundary layer, a
legitimate question is, Where do they all come from and which one is dynamically
signicant? Sreenivasan (1988) offers a glimpse of the difculties associated with
trying to answer this question. The structural description of a turbulent boundary
layer may not be that complicated, however, and some of the observed structures may
simply be a manifestation of the different aspects of a more basic coherent structure.
For example, some researchers argue that the observed near-wall streamwise vortices
and large eddies are, respectively, the legs and heads of the omnipresent hairpin
vortices (Head and Bandyopadhyay 1981). Nevertheless, that still leaves us with a
minimum number of building blocks that must be dealt with.
If the large eddies are assumed to be dynamically signicant, then how are they
recreated? It is easy to argue that the conventional laminar-to-turbulence transition
cannot be responsible because the same large eddies appear even in heavily tripped
boundary layers in which the usual transition routes are bypassed. Wall events cannot
be responsible for creating large eddies because of their extremely small relative scale
at high Reynolds number. Furthermore, no hierarchical amalgamation of scales has
been observed to justify such proposition.
If, alternatively, wall events are assumed to dominate, then where do the stream-
wise vortices or the low-speed streaks come from, and what mechanism sustains
the bursting cycle? Mechanisms that assume local instability cannot be valid at large
Reynolds numbers for which the wall layer is, say, 0.1% of the boundary layer
thickness,
9
and it is difcult to conceive that 99.9% of the boundary layer has no
active role in the generation and maintenance of turbulence. On the other hand,
assuming the bursting events are triggered by the large eddies brings us back to the
original question of where the latter come from.
The preceding difculties explain the lack of a self-consistent model of the tur-
bulent boundary layer despite the enormous effort expended to establish such a
model. None of the existing models are complete in the sense that none accounts
for each aspect of the ow in relation to every other aspect. The question marks
in Figure 4.9 are just the tip of the iceberg. Developing a complete, self-consistent
model is more than an academic exercise, for a proper conceptual model of the
ow gives researchers the necessary tools to compute high-Reynolds-number practi-
cal ows using the Reynolds-averaged NavierStokes equations and to devise novel
ow-control strategies as well as to extend known laboratory-scale control devices
to eld conditions.
InnerOuter Interaction
There is no doubt that signicant interactions between the inner and outer lay-
ers take place. On energy grounds alone, it is known that in the outer layer the
dissipation is larger than the turbulence kinetic energy production (Townsend 1976).
It is therefore necessary for energy to be transported from the inner layer to the
outer layer simply to sustain the latter. How that is accomplished and whether co-
herent structures are the only vehicle to transport energy is not clear, but two distinct
schools of thought have emerged. In the rst, the large-scale structures dominate and
provide the strong buffeting necessary to maintain the low-Reynolds-number turbu-
lence in the viscous region (y
+
30). In the second view, rare, intense wall events
are assigned the active role and, through outward turbulence diffusion, provide the
necessary energy supply to maintain the outer region. As mentioned in the previous
subsubsection, both views have some loose ends.
On the basis of a large number of spacetime, two-point correlation measure-
ments of u and v, Kovasznay, Kibens, and Blackwelder (1970) suggested that the
outer region of a turbulent boundary layer is dominated by large eddies. The inter-
face between the turbulent ow and the irrotational uid outside the boundary layer
is highly corrugated with a root-mean-square slope in the xy plane of roughly 0.5.
The three-dimensional bulges are elongated in the streamwise direction with an aspect
ratio of approximately 2:1 and have a characteristic dimension in the wall-normal
direction of between 0.5 and . They appear quasi-periodically and are roughly sim-
ilar to each other. Kovasznay et al. (1970) allowed that the large eddies are passive in
the sense that the wall events and not these eddies are responsible for producing the
Reynolds stress. Kovasznay (1970) advanced the hypothesis that wall bursting starts
a chain reaction of some sort at all intermediate scales culminating into a sequence
of amalgamations that eventually lead to the large structures. As mentioned earlier,
9
This is the near-wall region extending fromthe surface to y
+
= 30 as a percentage of the boundary-layer
thickness when the Reynolds number is Re
10
5
.
Figure 4.11: Sketch of large-eddy structures as a
collection of smaller-scale hairpin vortices. From
Head and Bandyopadhyay (1981). (a) Typical
low-Reynolds-number boundary layer. (b) Typical
high-Reynolds-number ow.
such hierarchical amalgamation of scales
has not been directly observed in the lab-
oratory.
Head and Bandyopadhyay (1981),
on the other hand, suggested that the
very existence of the large eddies at
high Reynolds numbers is in doubt.
Their combined ow visualization and
hot-wire probe experiments are unique
in that an unusually large range of
Reynolds number was investigated,
Re
=50017,500, allowing them to unambiguously clarify Reynolds number ef-

fects on the structure of the boundary layer. Head and Bandyopadhyay maintained
that a large structure seen in typical ow-visualization experiments is nothing but the
slow overturning of a random collection of smaller-scale hairpin vorticesjust a few
or even a single isolated vortex loop at low Reynolds numbers (say, Re
<1000) but
a large number of them at high Reynolds numbers (say, Re
>5000). This is sketched

for typical low- and high-Reynolds-number boundary layers in Figures 4.11a and
4.11b, respectively. A brisker rate of rotation of the isolated (fat) vortex loop is
observed at the lowest Reynolds number consistent with prior observations of large
eddies in low-speed experiments. The hairpins are inclined at around 45
to the plane
of the ow over a major part of the layer thickness. In Head and Bandyopadhyays
(1981) view, the entire turbulent boundary layer consists very largely of vortex loops
that become increasingly elongated as the Reynolds number increases. The so-called
large eddies, on the other hand, do not appear to exhibit any particular coherent
motion beyond a relatively slow overturning or toppling due to shear.
Corroborative evidence for the hairpin angle of inclination of 45
comes from
the simultaneous, multiple-point hot-wire measurements of Alving et al. (1990) in a
canonical turbulent boundary layer and a boundary layer recovering from the effects
of strong convex curvature. Their cross-correlation results are consistent with the
observation of large-scale structures spanning the entire shear layer and inclined at
angles in the range of 35
45
near the outer edge of the boundary layer but at

continuously decreasing angles as the wall is approached.
The sketch in Figure 4.4.2b for a typical large eddy at high Reynolds number
is consistent with the statistical ndings of Brown and Thomas (1977), who have
shown by using conditional averaging techniques that a typical large structure in a
turbulent boundary layer has an upstream rotationalirrotational interface inclined
at 18
to the ow direction. Head and Bandyopadhyay (1981) have observed such

individual structures only at higher Re
(> 5000). It is possible to arrive precisely

at this slope by modeling the large structure to be composed of hairpin vortices
formed at regular intervals (Bandyopadhyay 1980). Such large structures composed
of many hairpin vortices have not been observed in the low-Reynolds-number DNS
simulations.
Regeneration Mechanisms
Robinson (1991) summarized many of the conceptual models advanced by dif-
ferent researchers to explain how a wall-bounded turbulent ow maintains itself.
Books Online Cambridge University Press, 2009
Among those reviewed are the models advocated by Willmarth and Tu (1967); Black
(1968); Offen and Kline (1975); Hinze (1975); Praturi and Brodkey (1978); Thomas
and Bull (1983); Acarlar and Smith (1987a,b); and Robinson (1990). Some of these
conceptual models emphasize a particular aspect of the owdynamics, as for example
the bursting cycle, whereas others are more ambitious and attempt to include both
the inner and outer structures and their interaction. Several viewpoints on the regen-
eration mechanisms of wall turbulence are represented in the books edited by Kline
and Afgan (1990) and by Panton (1997a).
Robinson (1991) also lists signicant contributions that utilize structural infor-
mation to predict statistical quantities or invoke a simplied form of the governing
equations to model the dynamics of the near-wall turbulence-production process.
Among the predictive models discussed are those by Landahl (1967, 1980, 1990);
Townsend (1976); Perry and Chong (1982); Perry et al. (1986, 1990); Walker and
Herzog (1988); Aubry et al. (1988); Hanratty (1990); and Berkooz, Holmes, and
Lumley (1991).
Other potentially useful predictive models not discussed by Robinson (1991) in-
clude those based on stability considerations (Malkus 1956, 1979), based on the
turbulence energy equation (Bradshaw, Ferriss, and Atwell 1967), based on the
uv velocity-quadrant statistical description of the organized motions (Nagano and
Tagawa 1990), and based on a single hairpinlike vortex in a unit domain of turbu-
lence production (Bandyopadhyay and Balasubramanian 1993, 1995, 1996). These
models account explicitly for Reynolds number effects and might, therefore, be useful
for practical Reynolds numbers.
Inevitably, in almost all the conceptual models the omnipresent hairpin vortex
(or horseshoe at lowRe) plays a key role. Such a vortex has been proposed earlier
by Theodorson (1952) on intuitive grounds as the primary structure responsible for
turbulence production and dissipation in the boundary layer. His tornadolike vortices
formastride near-wall, low-speed regions of uid and growoutward with their heads
inclined at 45
to the ow direction.
Black (1966, 1968) conducted a more rigorous analytical work to show the fun-
damental role of hairpin vortices in the dynamics of wall-bounded ows. His basic
premise is that the primary role of the random turbulent motion is not to transfer
mean momentumdirectly but rather to excite strong, three-dimensional instability of
the sublayer, which is a powerhouse of vorticity. In Blacks model, trains of discrete
horseshoes are generated by repetitive, localized, nonlinear instabilities within the vis-
cous sublayer. The vortical structures are shed and outwardly migrate from the near-
wall region in a characteristic, quasi-frozen spatial array. The horseshoes inviscidly
induce an outow of low-speed uid from within the vortex loops, creating motions
that would be seen by a stationary probe as sharp, intermittent spikes of Reynolds
stress. Because of the continuous creation of new vortex loops that replace older
elements, the lifetime of the vortical array is much longer than that for its individual
members. According to Black (1968), such organized structures are responsible for
the efcient mass and momentum transfer within a turbulent boundary layer.
Sreenivasan (1988) offers a similar model to that of Black (1968). The essential
structures of the boundary layer, including the hairpin vortices, result from the insta-
bility of a caricature ow in which all the mean-ow vorticity has been concentrated
in a single fat sheet.
As a parting remark in this subsection, it might be instructive to recall that
hairpin vortices play an important role also in the laminar-to-turbulence transition
of boundary-layer ows. Essentially, these hairpins are the result of the nonlinear
tertiary instability of the three-dimensional peak and valley pattern, which itself
is the secondary instability of the primary TollmienSchlichting waves (Klebanoff,
Tidstrom, and Sargent 1962).
CHAPTER FIVE
Reynolds Number Effects
What is reasonable is real; that which is real is reasonable.
(Georg Wilhelm Friedrich Hegel, 17701831)
Science is what you know, philosophy is what you dont know.
(Bertrand Arthur William Russell, 18721970)
PROLOGUE
This chapter deals with Reynolds number effects in turbulent shear ows with par-
ticular emphasis on the canonical zero-pressure-gradient boundary layer and two-
dimensional channel-ow problems. The Reynolds numbers encountered in many
practical situations are typically several orders of magnitude higher than those stud-
ied computationally or even experimentally. High-Reynolds-number research facil-
ities are expensive to build and operate, and the few that exist are heavily sched-
uled with mostly developmental work. For wind tunnels, additional complications
due to compressibility effects are introduced at high speeds. Likewise, full compu-
tational simulation of high-Reynolds-number ows is beyond the reach of current
capabilities. Understanding turbulence and modeling will therefore continue to play
vital roles in the computation of high-Reynolds-number practical ows using the
Reynolds-averaged NavierStokes equations. Because the existing knowledge base,
accumulated mostly through physical as well as numerical experiments, is skewed
toward the low Reynolds numbers, the key question in such high-Reynolds-number
modeling as well as in devising novel owcontrol strategies is, What are the Reynolds
number effects on the mean and statistical turbulence quantities and on the organized
motions? Understanding the Reynolds number effects is important for ow control
on two counts:
1. A passive or active control device developed in a low-Reynolds-number
facility may perform quite differently at high Re.
2. For reactive control, coherent structures are targeted. Those organized mo-
tions change scale and even character as the Reynolds number varies.
58
5.1 THE LAST CONUNDRUM 59
5.1 The Last Conundrum
Once more, turbulence is the last great unsolved problem of classical physics. Two
of the greatest achievements of turbulence research are the Kolmogorov universal
equilibrium theory and the universal logarithmic law of the wall. In fact, there is a
direct analogy between the two high-Reynolds-number asymptotes, one being con-
cerned with a cascade of energy and an inertial subrange in the frequency domain
and the other with a hierarchy of eddies and an inertial sublayer in the physical
space. The overall ow dynamics in the energy spectrum subrange and the wall-
bounded owsublayer is independent of viscosity. Dimensional reasoning, similarity,
and asymptotic analysis are the tools of choice to derive analytical expressions for
certain statistical ow quantities without actually solving the intractable governing
equations.
One of the fundamental tenets of boundary-layer research is the idea that any sta-
tistical turbulence quantity (mean, rms, Reynolds stress, etc.) measured at different
facilities and at different Reynolds numbers will collapse to a single universal pro-
le when nondimensionalized using the proper length and velocity scales (different
scales are used near the wall and away from it). This is termed self-similarity or self-
preservation and allows convenient extrapolation from the low-Reynolds-number
laboratory experiments to the much higher Reynolds number situations encountered
in typical eld applications. The universal logarithmic prole mentioned above de-
scribes the mean streamwise velocity in the overlap region between the inner and
outer layers of any wall-bounded ow and is the best-known result of the stated
classical idea.
The log-law has been derived independently by Ludwig Prandtl and G.I. Taylor
using mixing length arguments, by Theodore von K arm an using dimensional rea-
soning, and by Clark B. Millikan using asymptotic analysis. Those names belong
of course to the revered giants of our eld. Questioning the fundamental tenet or
its derivatives is, therefore, tantamount to heresy. But the questions and doubts
linger as evidenced from the work of Simpson (1970), Malkus (1979), Barenblatt
(1979), Long and Chen (1981), Wei and Willmarth (1989), George, Castillo, and
Knecht (1992), Bradshaw (1994), Sreenivasan (see the doctoral thesis by Kailasnath
1993), and Smits (see the doctoral thesis by Smith 1994), among others, who at
different times challenged various aspects of this law. And those are only the ones
who, with varying degrees of difculty, could get their work published. There is
strong suspicion, among the iconoclasts at least, that Reynolds number effects persist
indenitely for both mean velocity and, more pronouncedly, higher-order statistics,
and hence that true self-preservation is never achieved in a growing boundary layer.
In fairness to the high priests, their logarithmic law was always intended to be a
very high-Reynolds-number asymptote. These issues are discussed in greater detail
in the survey by Gad-el-Hak and Bandyopadhyay (1994). In this chapter, we give the
highlights.
If in fact the log-law is fallible, the implications are far-reaching (Gad-el-Hak
1997). Resolution of the full equations, via direct numerical simulations, at all but
the most modest values of Reynolds number is beyond the reach of current or near-
future computer capabilities. Modeling will, therefore, continue to play a vital role
in the computations of practical ows using the Reynolds-averaged NavierStokes
equations. Flow modelers, in attempting to provide concrete information for the
60 REYNOLDS NUMBER EFFECTS
designers of, say, ships, submarines, and aircraft, heavily rely on similarity principles
in order to model the turbulence quantities and circumvent the well-known closure
problem. Because practically all turbulence models are calibrated to reproduce the
law of the wall in simple ows, failure of this universal relation virtually guarantees
that Reynolds-averaged turbulence models will fail too. Finally, developers of ow-
control devices to reduce drag, enhance lift, and so forth, often have to extrapolate the
widely available low-speed (or more precisely low-Reynolds-number) results to high-
speed ows of practical interest for which no data are available. Such extrapolation
is not possible if the difcult-to-quantify Reynolds number effects persist indenitely.
For both scientists and engineers the message is essentially back to the drawing board!
In a community of conformists, the heretics never have it easy, of course. The
peer reviewsystem, although essential for weeding out the charlatans, the misguided,
and the fools, is somewhat biased against unorthodox ideas. Nevertheless, the latest
two theses to question the infallibility of the log-law are contained in the articles
by Barenblatt, Chorin, and Prostokishin (1997) and George and Castillo (1997).
The two teams tackle the same problem quite differently and independently. Both
papers offer concrete alternatives to the Reynolds-number-independent law of the
wall. Barenblatt et al. (1997) use scaling laws that invoke a zero-viscosity asymptote,
whereas George and Castillo introduce newtools they termthe asymptotic invariance
and near asymptotic principles that result in a new law of the wall with explicit
Reynolds number dependence. George and Castillos new law is deduced from
rst principles, ts existing mean-velocity data better, and is extendible to higher-
order statistics.
When the log-law and its consequences are challenged, the usual immediate re-
action is to doubt the credentials of the blasphemer. Something is wrong with his or
her model, experiment, numerical scheme, or with an endless list of potential pitfalls.
These are all genuine concerns that turn out to be valid most of the time, but para-
noiacs have enemies too! There is also the persistent, albeit misguided, argument that
the log-law ts the data well enough for engineering applications. If it isnt broke,
dont x it! This pragmatism is of course simultaneously the curse and the blessing of
science conducted by engineers. Moreover, although the errors involved in attempting
to t the log-law to existing mean-velocity data are quite tolerable considering our
inability to accurately measure the friction velocity (the velocity scale necessary to
collapse the plots), the corresponding errors for higher-order statistics are egregious.
The entire enterprise is not unlike the sixteenth-century debate over the Ptolemaic
view of the heavens and the Copernican model seeking to replace it. The former the-
ory served navigators well for over 1400 years. The Copernican theory made only
small corrections but radically changed humankinds view of their universe. And it
was a masterpiece in terms of its economy of postulates and assumptions, which is a
necessary condition for theoretical elegance.
In judging the alternatives to the log-law, one should of course ask the usual
barrage of questions that only a vibrant collection of skeptical neurons could muster.
Is the theory simple, elegant, and self-consistent? Does the model provide a better t to
the data? Does it explain previous contradictions? Is the theory based on a minimum
number of assumptions? Is it extendable to more complex situations? Are the results
asymptotically correct? Is the logic sound? Is the mathematics free of errors? A tting
end to this editorializing is to quote my (elder) academic sibling John Lumley: . . . A
theory that does all that in an effortless way is often called elegant. Tomorrow, it may
5.2 THE ISSUES 61
be wrong. Even so, it deserves to be regarded as one of the better things of which
man is capable (Lumley 1992).
5.2 The Issues
5.2.1 Field versus Laboratory Flows
It is difcult to overstate the technological importance of the turbulent wall-
bounded ow. Vast amounts of energy are spent in overcoming the turbulence
skin-friction drag in pipelines and on air, water, and land vehicles. For blunt
bodies (e.g., trucks and trains), the pressure drag resulting from boundary layer
separation can be several orders of magnitude higher than the skin friction, and
even more energy is wasted. Heat transfer and mixing processes crucially depend
on the turbulence transport for their efcient attainment. The Reynolds numbers
encountered in many practical situations are typically several orders of magnitude
higher than those studied computationally or even experimentally (Figure 5.1).
Yet, our knowledge of high-Reynolds-number ows is very limited, and a complete
understanding is yet to emerge. The existing knowledge base, accumulated mostly
through physical as well as numerical experiments, is clearly skewed toward low
Reynolds numbers. For many practical applications the key question is then, What
are the Reynolds number effects on the mean and statistical turbulence quantities
and on the organized motions of turbulence? One always hopes that the ow
characteristics become invariant at sufciently high Reynolds number. That merely
shifts the question to What is high enough?
Consider the simplest possible turbulent wall-bounded ow, that over a smooth
at plate at zero incidence to a uniform, incompressible ow or its close cousin, the
two-dimensional channel ow. Leaving aside for a moment the fact that such ideal-
ized ow does not exist in practice, where three-dimensional, roughness, pressure-
gradient, curvature, wall-compliance, heat-transfer, compressibility, stratication and
other effects may be present individually or collectively, the canonical problemitself is
not well understood. Most disturbing froma practical point of vieware the unknown
effects of Reynolds number on the mean ow, the higher-order statistical quantities,
and the ow structure. The primary objective of this chapter is to review the state of
the art of Reynolds number effects in shear-ow turbulence with particular emphasis
on the canonical boundary layer and channel-ow problems.
Figure 5.1: Ranges of momentum-thickness Reynolds number for different
facilities and for eld conditions.
5.2.2 Reynolds Number
Reynolds number effects are intimately related to the concept of dynamic similar-
ity. In a given owgeometry, if Land v
o
are the length and velocity scales, respectively,
the nondimensional equation of motion for an effectively incompressible, Newtonian
uid with negligible body forces is given by
u
t
+(u )u = p +
1
Re
2
u (5.1)
where Re = v
o
L,, p is pressure, and is the constant kinematic viscosity. This
seemingly supercial nondimensionalization reveals two important properties. The
rst is the concept of dynamic similarity. No matter how L, v
o
and are varied, as
long as Re is the same in two geometrically similar ows, they have the same solu-
tion. Small-scale model testing of large-scale real-life ows is based on this property.
Secondly, for a given geometry and boundary conditions, the effect of changing L,
v
o
, or , or any combination of them, can be described uniquely by the change of Re
alone. Although, the importance of Re was recognized earlier by Stokes, it has come
to be termed Reynolds number in recognition of Osborne Reynolds (1883) telling
demonstration of its effect on the onset of turbulence. Even today, the laminar-to-
turbulence transition is one of the most dramatic Reynolds number effects, and its
rational computation continues to be a research challenge. Equation (5.1) shows that
Re represents the relative importance of viscous and inviscid forces. Because three
forces (inertial, pressure, and viscous) are in equilibrium, the balance can be described
by the ratio of any two, although it has become customary to characterize the ow
by the ratio of inertial to viscous forces.
In this chapter, only turbulent ows are considered because they are widely preva-
lent. The report by Bushnell et al. (1993) treats Reynolds number similarity and
scaling effects in laminar and transitional ows. The understanding of the effects of
Reynolds number relies on our understanding of viscous forces. For a wall-bounded
ow, this is true no matter how high the Reynolds number is. Experience shows that,
aside from rareed gas ows, there is no practical Reynolds number for which the
no-slip boundary condition, which owes its origin to viscous effects, switches off.
Because the net viscous force on an element of incompressible uid is determined
by the local gradients of vorticity, the understanding of the vorticity distribution is
the key to determining Reynolds number effects. Vorticity can be produced only at a
solid boundary and cannot be created or destroyed in the interior of a homogeneous
uid under normal conditions.
The qualitative effects of Reynolds number on the scales of turbulence are demon-
strated by the velocity-uctuations spectra for lowand high Re. The large scale is only
weakly dependent on Reynolds number (Townsend 1976). However, as Reynolds
number increases, the small scales become physically smaller (larger wave numbers),
and the diversity of intermediate scales between the large and small increases. In terms
of organized motions in a turbulent boundary layer, the effect of Reynolds number
on the omnipresent vortex loops has been demonstrated in the ow-visualization
experiments of Head and Bandyopadhyay (1981). With increasing Reynolds num-
ber, the aspect ratio of the constituent horseshoe vortices increases while the vortices
become skinnier. Thus, the vortex loops typically observed at low Reynolds num-
ber are gradually supplanted by horseshoe vortices and then hairpin vortices as Re
increases.
5.3 FLOW REGIMES 63
Figure 5.2: Distribution of viscous and turbulence shear stresses in wall-
bounded ows [after Sreenivasan 1989].
5.3 Flow Regimes
An inspection of the distribution of viscous and turbulence shear stresses in a typical
wall-bounded ow demonstrates the presence of three distinct regions. Figure 5.2,
adapted by Sreenivasan (1989) from the smooth-pipe-ow data of Nikuradse (1932)
and Laufer (1954), shows such distribution in wall units (friction velocity, u
=
(
w
,)
1,2
, used as velocity scale, and the ratio of kinematic viscosity to friction veloc-
ity used as length scale; see Section 4.4.1). Pipe (or channel) ow data are preferable
to at-plate boundary layer experiments because the Reynolds stress, a rather dif-
cult quantity to measure accurately, can be computed exactly for fully developed
channel ows from the relatively simpler measurements of mean-velocity prole and
pressure gradient. The semilog plot in the gure enhances the importance of the thin
near-wall region relative to the rest of the shear layer.
The broken line in Figure 5.2 is the time-averaged viscous stress distribution com-
puted by differentiating the mean-velocity prole. Note that this laminar owconcept
of shear may not be relevant to the time-dependent turbulent owbecause turbulence
models based on the mean velocity gradients have not been widely successful (e.g.,
Bradshaw et al. 1967). Nevertheless, it is clear from the gure that the mean viscous
stress, j(U,y), is important only near the wall. This wall layer is followed by a
region of approximately constant Reynolds stress. Finally, an outer layer
1
is charac-
terized by a diminishing turbulence shear stress, reaching zero at the centerline of the
1
Called core region in internal ows.
pipe. Unlike the second and third regimes, the extent of the rst region does not de-
pend on Reynolds number. Both the viscous region and the constant-Reynolds-stress
region are similar in all wall-bounded ows. In contrast, the outer layer is different
in internal ows and boundary layers. Proles of the mean velocity and other turbu-
lence statistics can be constructed from scaling considerations of the three distinct
regimes, as will be demonstrated in the following three subsections.
Note that the Reynolds number used as a parameter in Figure 5.2 is dened as
Re
= u
a,; that is the channel half-width (or boundary-layer thickness) expressed

in wall units: a
+
(or
+
). Although numerically Re
and
+
are the same, their dif-
ference in signicance and usage should be claried. The variable
+
denotes the
ratio of the outer- to inner-layer thickness and represents the degree of shrinking
of the latter with respect to the former, which changes little with Reynolds number.
The variable emphasizes the disparity of the two scales and the diversity of the in-
termediate and interacting scales at higher Reynolds numbers. As shown by Head
and Bandyopadhyay (1981),
+
also indicates the reduction of the hairpin vortex
diameter and the increase in its aspect ratio as the Reynolds number increases. The
value of
+
in a typical laboratory experiment is O[1000], whereas it approaches
100,000 in the boundary layer developing over the space shuttle (Bandyopadhyay
1991). This variable is pertinent to the understanding of the mechanism of drag re-
duction by outer-layer devices (Anders 1990a). On the other hand, Re
is a Reynolds
number that is also called a stability parameter by Black (1968). In Blacks work and
later in Sreenivasans (1988), Re
indicates a Reynolds number associated with the

quasi-periodic instability and breakup process that is hypothesized to be responsible
for the regeneration of turbulence in a wall-bounded ow. Note that, for a smooth
wall, Re
increases monotonically with Re
and never reaches an asymptote.

5.3.1 Viscous Region
Viscosity appears to be important only up to y
+
=30. The viscous region
can be subdivided into two subregions: the viscous sublayer and the buffer layer
(Figure 5.3). Very close to the wall, 0 y
+
5, the turbulence shear stress is nearly
zero, which implies that the only relevant quantities there are the kinematic viscosity
and friction velocity u
.
2
In this viscous sublayer, several turbulence statistics can be
asymptotically estimated fromconsiderations of the no-slip condition and continuity
and dynamical equations. Following Monin and Yaglom (1971) and using experi-
mental data, Sreenivasan (1989) gives the following Taylors series expressions, in
wall units, for the mean streamwise velocity, for the root-mean-square value of the
three uctuating velocity components, and for the Reynolds stress, respectively:
U
+
= y
+
1 10
4
y
+
4
+1.6 10
6
y
+
5
+ (5.2)
u
+
= 0.3y
+
+c
1
y
+
2
+ (5.3)
v
+
= 0.008y
+
2
+c
2
y
+
3
+ (5.4)
w
+
= 0.07y
+
+c
3
y
+
2
+ (5.5)
uv
+
= 4 10
4
y
+
3
8 10
6
y
+
4
+ . (5.6)
2
For hydraulically rough walls (i.e., where the average roughness height is greater than the viscous
sublayer thickness), the relevant scaling parameters are the characteristic roughness height and friction
velocity.
5.3 FLOW REGIMES 65
Figure 5.3: Schematic of the different regions within a wall-bounded ow
at typical laboratory and eld Reynolds numbers.
where the overbar and prime denote, respectively, mean and root-mean-square
values.
For y
+
- 5, the leading term of each of the preceding expansions sufces. Note,
however, that the experimentally determined leading coefcients in Eqs. (5.3)(5.5)
are lower than those computed from direct numerical simulations. Mansour, Kim,
and Moin (1988) analyzed the channel-ow database generated by Kim et al. (1987)
and reported the following leading-term coefcients for u
+
, v
+
, and w
+
: 0.36,
0.0086, and 0.19, respectively. The last coefcient, in particular, is almost three
times that of the corresponding leading term determined experimentally. The reason
may be due to the rapid drop in the spanwise velocity uctuations as the wall is
approached, and thus a very small hot-wire probe or LDV focus would be needed to
realize the true value.
With three terms, Eq. (5.2) for the mean velocity is valid up to y
+
= 20. Note
that the constants in the preceding equations are not necessarily universal. As will be
discussed later, clearly discernible Reynolds number effects will be demonstrated for
all higher-order statistical quantities even in the near-wall region.
The buffer layer is where both the viscous stress and the turbulence shear stress
are important and the peak production and dissipation of turbulence kinetic energy
occur (at about y
+
=12, seemingly independent of the global Reynolds number).
In the buffer layer, the characteristic local Reynolds number of yu
, =30 is ex-
ceedingly low, and turbulence cannot be maintained unless buffeted constantly by
strong disturbances, presumably from the outer layer. This region merges with the
constant-Reynolds-stress layer.
5.3.2 Constant-Reynolds-Stress Region
This region, loosely interpreted
3
to include all points within the 3 dB points
of the peak Reynolds stress, extends from y
+
=30 to y,a =0.2, where a is the pipe
radius. Here, the distance fromthe wall y is much larger than the viscous length scale,
,u
, but much smaller than the pipe radius (or the boundary layer thickness, , for
an external ow). Note that the upper extent of this region is a constant fraction of
the boundary layer thickness but varies with Reynolds number when expressed in
wall units (see Figure 5.3).
In this region, viscous stresses are negligible, and the momentum ux is accom-
plished nearly entirely by turbulence. The only relevant length scale is y itself, and the
square root of the nearly constant Reynolds stress, (uv
max
)
1,2
, is the appropriate
velocity scale. Thus, the mean-velocity gradient may be expressed as
U
y

(uv
max
)
1
2
y
. (5.7)
The well-known logarithmic velocity prole follows directly from integrating
Eq. (5.7) and using the velocity at the edge of the viscous sublayer as a boundary
condition
U
+
=
1
ln y
+
+ B (5.8)
where is the von K arm an constant. Both and Bare presumably universal constants
and are determined empirically for at-plate boundary layers to be approximately
0.41 and 5.0, respectively. Slightly different values are used for the two constants in
the case of pipe or channel ows. In that case, Eq. (5.8) holds almost up to the center-
line of the channel. As the Reynolds number increases, the extent of the logarithmic
region (in wall units) increases, and the maximum Reynolds stress approaches the
value of the viscous stress at the wall
(uv
max
)
1
2
u
2
1 (5.9)
Several other methods can be used to derive the logarithmic velocity prole. Amix-
ing length, based on momentum transport, that simply varies linearly with distance
fromthe wall, = y, again yields Eq. (5.8). Millikans (1939) asymptotic analysis re-
covers the log-relation by assuming the existence of a region of overlap where both the
inner and outer laws are simultaneously valid (see also the rarely cited albeit relevant
3
Strictly speaking, the Reynolds stress is not really constant anywhere in a pipe or channel ow.
5.3 FLOW REGIMES 67
article by Izakson 1937). All models invariably rely on the presence of the constant-
stress layer experimentally observed at high Reynolds number. Despite copious evi-
dence for the existence of a logarithmic region in the mean-velocity prole, the whole
log-lawscenario has been periodically questioned (see, for example, Barenblatt 1979,
1993; Malkus 1979; Long and Chen 1981; George et al. 1992, 1994; Barenblatt and
Prostokishin 1993). We will return to this point later in Section 5.5.1.
The arguments used by Millikan (1939) to derive the logarithmic relation for
the boundary layer are analogous to those employed to establish the universal equi-
librium theory of turbulence, which was called the theory of local similarity by its
originator Kolmogorov (1941a,b,c; 1962).
4
For the boundary layer, an inertial sub-
layer exists at sufciently large Reynolds numbers, and the overall ow dynamics
is independent of viscosity, which merely provides a momentum sink of prescribed
strength at the wall. Similarly, an inertial subrange exists in the turbulence energy
spectrum when the Reynolds number is large enough. There, the wave number is
larger than that for the large eddies but smaller than the dissipative wave numbers.
The viscosity again provides the dissipative sink for kinetic energy at the small-scale
end of the turbulence spectrum. At high Re, the spectral shape in the inertial sub-
range is nearly completely determined by the energy ux across the wave number
domain. The amount of this energy ux, and therefore the eventual rate of dissipa-
tion, is determined by the dynamics of the large, energy-containing eddies. Changing
the viscosity (i.e., Re) has no inuence on the rate of energy dissipationit simply
changes the scale at which the conversion of kinetic energy to heat takes place.
Scaling arguments similar to those leading to Eqs. (5.7) and (5.8) can be used in
the constant-Reynolds-stress region to show that
5
u
= constant = 2.0 (5.10)

v
= constant = 1.0 (5.11)

w
= constant = 1.4 (5.12)

P
K.E.
= D
K.E.
=
u
3
y
(5.13)
where P
K.E.
and D
K.E.
are the production and dissipation of turbulence kinetic energy,
respectively. Additionally, a portion of the power spectrum for each of the three
velocity components exhibits a 1 power-law in this same region governed by a
constant turbulence shear stress transmitted across its different uid layers.
The total stress is approximately constant throughout the viscous layer and the
constant-Reynolds-stress region. This is the so-called inner layer (see Figure 5.4), and
for a smooth wall the mean-velocity prole there is given by the unique similarity
law of the wall rst formulated by Prandtl (1925a)
U
+
= f (y
+
) (5.14)
4
The book by Frisch (1995) is a fascinating reading of the legacy of Andrei Nikolaevich Kolmogorov and
his contribution to turbulence theory.
5
The values of the constants in Eqs. (5.10)(5.13) are determined empirically from mostly low-Reynolds-
number experiments. Again, Section 5.6 will reveal that these constants depend, in fact, on the Reynolds
number.
Figure 5.4: Normalized turbulence kinetic energy production rate as a function of normal distance from
the wall. Data from a typical laboratory at-plate boundary layer [after Kline et al. 1967].
where f is a universal function presumably independent of Reynolds number and
streamwise location. The inner law is the same for both internal and external ows.
5.3.3 Outer Layer
Beyond the constant-stress region, an outer layer is characterized by a diminishing
turbulence shear stress. Note that some researchers include the constant-Reynolds-
stress region as part of the outer region. This is perhaps an accurate inclusion because
the part of the boundary layer where the logarithmic law is valid is, strictly speaking,
the region of overlap between the inner and outer laws (see Figure 5.4).
In internal ows, intermittency of turbulence and interaction with potential
freestream are absent. There is, however, an interaction of turbulence from the op-
posite wall in the case of a two-dimensional channel, and this is even more complex
in the case of a circular pipe. Furthermore, fully developed conditions for pipes and
channels are dened as all time-averaged owquantities (except static pressure) being
independent of x. Therefore, the core region of a pipe or channel ow differs from
the outer layer of a growing boundary layer.
The appropriate length scale in the core region is the pipe radius a (or the
boundary-layer thickness for an external ow). The mean-velocity prole is char-
acterized by the velocity defect (U
o
U), where U
o
is the velocity at the edge of the
shear layer (centerline velocity U
c
for a fully-developed pipe ow). The velocity-defect
(or, more appropriately, momentum-defect) law, formulated by von K arm an (1930),
is given by a second universal function
(U
o
U)
u
= g
(5.15)
This equation is valid even in the logarithmic region and appears to be well conrmed
experimentally.
5.4 COMPARISON WITH OTHER SHEAR FLOWS 69
For a turbulent boundary layer, Coles (1956) combined the defect law and the
inner law to give the following empirical velocity prole valid throughout the entire
wall-bounded shear layer:
U
+
= f (y
+
) +
H
(5.16)
where is the von K arm an constant and H is a prole parameter that depends
strongly on Re for small Reynolds numbers. Coles idea is that a typical boundary-
layer ow can be viewed as a wakelike structure constrained by a wall. Intermittency
and entrainment give rise to the wakelike behavior of the outer part of the ow,
which is sensitive to pressure gradient and freestreamturbulence. The wall constraint
is closely related to the magnitude of the surface shear stress and is sensitive to the
wall roughness and other surface conditions.
For equilibrium ows (Clauser 1954), the prole parameter H is independent of
streamwise location. The universal wake function W(y,) is the same for all two-
dimensional boundary-layer ows. Its form is similar to that describing a wake ow
or, more precisely, the mean-velocity prole at a point of separation or reattachment.
For example, the wake function can be adequately represented by
W
= 2 sin
2
2

y
(5.17)
Note, however, that this simple expression obtained empirically by Coles (1969)
does not yield zero velocity gradient at the edge of the boundary layer, as it should.
Lewkowicz (1982) proposed an alternative quartic polynomial that removes this
deciency.
At the same Reynolds number, deviation of the actual mean-velocity distribution
fromthe logarithmic prole in the core region of a pipe or channel owis smaller than
that in the outer region of a boundary layer. In fact, as mentioned in Section 5.3.2,
the logarithmic velocity prole, Eq. (5.8) with slightly modied constants and B,
holds approximately up to the centerline of the pipe.
5.4 Comparison with Other Shear Flows
Before proceeding to investigate the specic effects of Reynolds number on the mean
and turbulence quantities of wall-bounded ows, it is instructive to give a coarse
comparison between such ows on the one hand and free-shear ows on the other.
As will be illustrated, the presence of the wall is of paramount importance to the issue
at hand. No matter howlarge the Reynolds number is, viscosity must be important in
a progressively shrinking region close to the wall, and Reynolds number dependence
persists indenitely.
Wakes, jets, and mixing layers are profoundly different from channel and pipe
ows and boundary layers. The absence of the wall in free-shear ows implies that
at sufciently high Reynolds numbers, the ow is nearly inviscid and by implication
Reynolds-number-independent (Dimotakis 1991, 1993). For wall-bounded ows, on
the other hand, there is always a small, progressively shrinking region near the surface
where viscosity must be important, no matter how large the Reynolds number is.
In boundary layers and channel ows, the overall behavior and gross structure
of turbulence are always affected by viscosity near the wall, whereas the direct
effect of viscosity gradually diminishes away from the surface. This implies that the
velocity and length scales must be different near the wall and away from it. The
disparity of scales for wall-bounded ows increases with Reynolds number, and true
self-preservation may never be achieved unless the inner and outer scales are forced
to be proportional at all Reynolds numbers. This latter scenario can be realized,
for example, in the very special case of ow between two planes converging at a
prescribed angle.
In wall-bounded ows, very large levels of turbulence uctuations are maintained
close to the wall despite the strong viscous as well as turbulence diffusion. As indicated
in Figure 5.4, at a typical laboratory Reynolds number of say Re
= O[10
3
], more
than about a third of the total turbulence kinetic energy production (and dissipation)
occurs in the 2% of the boundary-layer thickness adjacent to the wall. The fraction
of this thickness decreases as the Reynolds number increases. The near-wall region is
directly affected by viscosity,
6
and its importance to the maintenance of turbulence
is clearly disproportional to its minute size.
The thinness of the viscous sublayer presents a great challenge to both physical
and numerical experiments. Because this region is closest to the wall and is where
drag acts, it is extremely important at all Reynolds numbers. Yet in contemporary
direct numerical simulations, the viscous sublayer of ve wall units is resolved only
up to 1.4 wall units. In measurements, probe resolutions are even worse; other than in
low-Reynolds-number or oil-channel experiments, a probe length of
+
-7 is indeed
rare (Bandyopadhyay 1991).
In free-shear layers, on the other hand, energy production peaks near the inection
points of the mean-velocity prole. Production and dissipation are spread over the
entire ow width, as shown in Figure 5.5 depicting the turbulence energy balance
for a typical two-dimensional wake. Above a reasonably modest Reynolds number,
O[10
3
], all turbulence quantities become invariant to additional changes in Reynolds
number.
Despite these differences between boundary layers and free-shear ows, there
are also some similarities. The outer region of a boundary layer is characterized by
an intermittent rotationalirrotational ow, much the same as that observed in all
free-shear ows. Moreover, the outer ow is more or less inviscid at sufciently high
Reynolds numbers, which again is similar to jets, wakes, and mixing layers. The
interaction between the outer, or wake, region of a turbulent boundary layer and
the potential ow in the freestream is also similar to that in wakes and other free-
shear ows. The preceding discussion has addressed observational similarities and
differences between the wall-bounded turbulent ows and the free-shear layers. In
what follows, they are compared based on dynamic issues, such as the applicability
of an inectional inviscid breakdown mechanism, and it is shown that the subject is
still wide open.
Free-Shear Flows
Inviscid stability theory has been successfully used to predict the observed coher-
ent structures in turbulent free-shear ows, but it is not clear that similar arguments
can be made when a wall is present. In other words, it is not obvious that the same
6
Through the action of viscous stresses for a smooth wall, or through the action of pressure drag resulting
from the separated ow around discrete elements of sufcient size for rough walls.
5.4 COMPARISON WITH OTHER SHEAR FLOWS 71
Figure 5.5: Turbulence energy balance in a typical plane wake [after Townsend 1976].
inviscid, mean-ow breakdown mechanism responsible for generating the large ed-
dies in, say, a mixing layer is operable in the case of a boundary layer. Consider
the mean streamwise velocity and mean spanwise vorticity distributions for a plane
mixing layer, jet, and wake and for a two-dimensional boundary layer.
Atwo-dimensional mixing layer may be modeled as a single vortex sheet placed at
the location of peak vorticity (at the point of inection of the mean-velocity prole).
The local circulation per unit length of the vortex sheet, (x), is equal to the integral
of the mean spanwise vorticity, O
z
, across the shear layer
(x) =
O
z
dy =
dU
dy
dy = U
1
U
2
(5.18)
where U
1
and U
2
are, respectively, the velocities of the high- and low-speed sides of
the mixing layer. The vortex sheet is inviscidly unstable to two-dimensional pertur-
bations, and the resulting KelvinHelmholtz instability eventually evolves into the
omnipresent two-dimensional vortices observed even in high-Reynolds-number mix-
ing layers (Chapter 4). The resulting vortex blobs correspond to the saturation state
of this instability. The total circulation in the blobs is conserved, I = (x)Lx, where
Lx is the center-to-center distance between adjacent discrete vortices. Secondary in-
stabilities of the roll-up structures result in smaller longitudinal vortices and other
three-dimensional, hairpinlike eddies.
Similar reasoning leads to the in-phase counterrotating vortices for plane jets and
the staggered K arm an vortex street for two-dimensional wakes. Both jets and wakes
can be modeled as two vortex sheets with opposite signs of vorticity. Again, each
sheet is situated at the location of spanwise vorticity extrema, and total circulation
is conserved.
Wall-Bounded Flows
For a turbulent wall-bounded ow it is not obvious that the observed large-eddy
structures (Chapter 4) can be derived using similar inviscid arguments. If the bound-
ary layer is modeled by a vortex sheet in which the entire mean-ow vorticity has
been concentrated, the presence of the wall imposes a boundary condition that neces-
sitates the use of an image vortex sheet of opposite vorticity sign. Such considerations
led Sreenivasan (1988) to propose that the large eddies are the result of an instability
of a caricature of the real boundary layer. Two- and three-dimensional instabilities,
both inviscid and viscous, of this caricature ow lead to a plausible explanation for
many observed features, including the double-roller structures, hairpin eddies, and
low-speed streaks (see Chapter 4).
It turns out that a somewhat similar argument was advanced two decades earlier
by Black (1968). He treated the mean-ow breakdown as an intermittent, three-
dimensional, inviscid process in which a mechanism analogous to the starting vortex
of an impulsively started airfoil is in play. Black thereby successfully predicted the
formation of an array of hairpin vortices caused by a passing instability wave. Note
that, although Theodorsen (1955) predicted the formation of hairpin vortices, the
aspect of an array of them is absent in his work. In Blacks work u
, appears as
an important stability parameter. Mention should also be made of the waveguide
theory developed by Landahl (1967, 1972, 1977, 1980, 1990) to explain the cause
and effect relationships for the variety of coherent structures observed in turbulent
boundary layers.
In Sreenivasans (1988) model, a fat vortex sheet and its image are located on either
side of the wall at a distance corresponding to the position of the peak Reynolds stress.
Because of the absence of inection points in the interior of the canonical turbulent
ow, Sreenivasan (1988) chose the alternative location of peak uv based upon
experience with transitional boundary layers.
7
From all available boundary-layer as
well as channel-ow data, this location appears to scale with the geometric mean
of the inner and outer scales.
8
Sreenivasan (1988) termed that position the critical
layer,
9
although any evidence for the existence of such a two-dimensional layer, where
small perturbations are presumed to grow rapidly, is lacking. Using linear stability
theory, Sreenivasan successfully showed that the primary instability of the vortex
7
Note that in the past, both Clauser (1956) and Corrsin (1957) have also attempted to treat the turbulence
regenerative process as primarily similar to the breakdown mechanism of a critical laminar layer.
8
This powerful result is contrary to the universal law of the wall and will be analytically demonstrated
in Section 5.6.2.
9
In a transitional wall-bounded ow, the location of peak Reynolds stress coincides with the critical-layer
position. Because this location shows similar trends with Reynolds number to that in a turbulent ow,
Sreenivasan (1988) chose to place his proposed vortex sheet at this position.
5.5 MEAN FLOW 73
sheet and its image yields two-dimensional roll-up structures, which in turn excite
low-speed streaks and bursting. Subsequent instability of the roll-up structures leads
to hairpin eddies and double-roller structures. One problem with this picture is that,
unlike the case of free-shear ows, the predicted two-dimensional structures have
never been observed in an actual turbulent wall-bounded ow. Sreenivasan (1988)
himself allows that his simplistic model is unnished and has a number of weaknesses
but offers it as a target for useful criticism.
Experience with turbulence modeling, however, suggests that the turbulence in
a wall-bounded ow is not derived directly from the mean ow. In the earliest tur-
bulence models, shear stress is derived from the mean-velocity prole. Such models
have not been widely successful. Townsend (1976) and Bradshaw et al. (1967) have
argued that instead there is a much closer connection between the shear stress and the
turbulence structure. Townsends work was limited to the near-wall region, whereas
Bradshawet al. have extended the argument to the entire shear layer. Direct measure-
ments of typical eddies have supported their assertion (Falco 1974, Newman 1974).
There is a rst-order direct connection between the mean ow and turbulence in a
a free-shear ow but not in a wall-bounded ow. In a mixing layer, for example,
the experimentally observed two-dimensional rollers are the direct result of an in-
viscid instability of the mean-velocity prole. Their characteristic dimension is equal
to the layer thickness, and they contain almost all of the mean-ow vorticity. In a
wall-bounded ow, on the other hand, the three-dimensional hairpin vortices are the
result of a secondary or a tertiary instability, and their diameters are typically much
smaller than the boundary layer thickness. The hairpins contain only a portion of
the mean-ow vorticitythat is, they are farther removed from the mean ow.
The arguments above indicate that the existence of an inviscid breakdown mech-
anism responsible for the self-sustenance of the turbulence has not been rmly es-
tablished. In other words, it is not clear that the observed coherent structures in a
boundary layer or channel ow are the result of an instability of the mean ow or its
caricature. Until this issue is resolved, progress in the understanding of wall-bounded
ows will continue to lag behind that of free-shear ows. Despite the importance of
this dynamical issue, research on the organized nature of turbulent boundary layers
has remained largely conned to the kinematics, and high-payoff turbulence control
strategies are yet to be developed.
5.5 Mean Flow
Before investigating the issue of Reynolds number effects on coherent structures,
available data for the mean velocity and higher-order statistics of wall-bounded
ows are reviewed in the present and following sections. The mean-ow velocity in
the streamwise direction is a relatively easy quantity to measure, and almost every
paper on wall-bounded ows has such measurements (see, for example, Preston and
Sweeting 1944; Laufer 1951; Comte-Bellot 1965; Eckelmann 1974; Purtell,
Klebanoff, and Buckley 1981; Andreopoulos et al. 1984; Wei and Willmarth 1989).
Requirements for probe resolution are modest and, except very near the wall,
10
10
But here lies the caveat. It is the slope of the velocity prole at the wall that is needed to compute the
velocity scale used to collapse the data.
most published data are reliable to better than 1%. This is obviously not the case
for measurements of higher-order statistics, and this point will be revisited in the
following section.
For a turbulent wall-bounded ow, the region directly affected by viscosity, the
viscous sublayer plus the buffer layer, occupies the progressively smaller proportion
of the boundary-layer thickness as the Reynolds number increases (see Figure 5.3).
The rest of the ow is dominated by inertia, and the effect of viscosity enters only as
an inner boundary condition set by the viscous region. It is not surprising, therefore,
that the Reynolds number has a considerable effect on the velocity prole. As Re
increases, the mean-velocity prole becomes fuller, and the shape factor H (ratio of
displacement thickness to momentumthickness) decreases accordingly. For example,
at Re
= 2000, H = 1.41, and at Re
= 10,000, H = 1.33. The effect is even more

pronounced at Reynolds numbers lower than 2000. In a laminar at-plate ow,
in contrast, viscosity is important across the entire layer, and the shape factor is
independent of Reynolds number.
5.5.1 The Rise and Fall of the Log-Law
Available data appear to indicate that the wall-layer variables universally describe
the streamwise mean velocity in the inner layer of smooth at-plates, pipes, and
channels at all Reynolds numbers. The boundary-layer data of Purtell et al. (1981)
and Andreopoulos et al. (1984) and the channel ow data of Wei and Willmarth
(1989) are typical and cover a wide range of Reynolds numbers. In Figure 5.6
we depict the former set of low-Reynolds-number measurements as published in
Physics of Fluids. The graph indicates the presence of a logarithmic region for Re
as
Figure 5.6: Mean-velocity proles for different Reynolds numbers. Plot from the article by Purtell
et al. (1981) as it appears in Physics of Fluids.
5.5 MEAN FLOW 75
lowas 500. This is rather surprising when one considers that at this Reynolds number
a constant-stress region is virtually nonexistent and the maximum Reynolds stress is
substantially less than u
2
. However, in deriving the log-law, Eq. (5.8), the presumably

constant velocity scale has a weak, square-root dependence on the Reynolds stress.
The extent of the log-region, expressed in wall units, increases with Reynolds
number but is a constant fraction of the boundary layer thickness (see Figure 5.3).
Purtell et al.s (1981) measurements in low-Reynolds-number boundary layers
appear to indicate that the law of the wall does not vary with Reynolds number, thus
implying a truly constant value of the von K arm ans constant . Close to a decade
earlier, Huffman and Bradshaw (1972) analyzed the data from several other low-
Reynolds-number experiments and arrived at the same conclusion. At the other
Figure 5.7: Mean-velocity proles for different Reynolds numbers.
Data of Purtell et al. (1981) as replotted by George and Castillo
(1997). Note in particular how the curves in the bottom gure
do not collapse in the viscous sublayer contrary to the original
plots that were published in Physics of Fluids. In fairness to Purtell,
Klebanoff, and Buckley (1981), this particular practice of plotting
the analytical result U
+
=y
+
in the viscous sublayer region instead
of actual data is all too common in the open literature.
extreme, Grigson (1992) used model and eld data to show the constancy of for
Reynolds numbers up to Re
x
= 4 10
9
. Both the low- and high-Reynolds-number
results refute an earlier claimby Simpson (1970, 1976) that the von K arm an constant
varies with Reynolds number.
The single straight line in Figure 5.6 does not support the assertion by Landweber
(1953), Preston (1958), and Granville (1977) that the logarithmic region disappears
altogether at lowReynolds numbers. An important question is, What is the minimum
Re
at which a log-region is rst established? Coles (1962) analysis of the wake

component indicates that it is zero at Re
600. The data of Bandyopadhyay and

Ahmed (1993) indicate that Clausers outer-layer shape parameter C reaches zero at
Re
= 425. These can be regarded as two indications of a minimumvalue supporting

the experimental ndings of Purtell et al. (1981).
At lowReynolds numbers, the large scales of the turbulence uctuations dominate
its dynamics. The logarithmic region appears to be an inherent characteristic of the
turbulent boundary layer and to be associated with the large eddies. Because of the
persistence of the log-region to Reynolds numbers just above transition, Purtell et al.
(1981) suggest that the large-scale structures in the turbulent boundary layer are
related to, if not simply composed of, the hairpin eddies produced during the nal
stages of laminar-to-turbulence transition.
There is, however, a much more serious problem with Figure 5.6 and similar
mean-velocity plots abundant throughout the open literature. We quote here from
the recent work of George and Castillo (1997), who advocate an alternative to the
log-law.
11
The near wall region can not be addressed without rst deciding what values of u
to use for
a given data set. Unfortunately, direct measurements of
w
have, for the most part, not been
very satisfactory for a variety of reasons, not the least of which being that the results did not
conrmthe Millikan/Clauser theory. Therefore it has been common practice (since Clauser
1954) to choose values for u
which are consistent with the universal log law results for
the matched layer, by which is usually meant the constants chosen fromthe more abundant
pipe and channel ow data. Coles (1969) notes that the results obtained in this way are
seldom consistent with the momentum integral equation. Nonetheless, in spite of the fact
the measurement errors tend to cancel when the integrals are computed from the velocity
proles and therefore should be more accurate than the proles themselves, the friction
results calculated from them have been usually discarded in favor of the Clauser method.
Typical of such results is Figure 5 [reproduced in this book as Figure 5.7] which shows
the data of Purtell et al. (1981) in inner variables using the wall stress determined from
the Clauser method (bottommost) and from the velocity gradient near the wall (topmost).
Note the excellent collapse in the log region of the former, but also recognize that
the data should collapse best in this region because u
has been chosen to insure that it

doesin fact the choice of the tangency point for the log is obvious from the gure. More
interesting, however, is what has happened to the measurements for small values of y
+
- 8
or so where the data do not collapse at all.
This problem with the inner scaling has been noted before (e.g., Kline et al. 1967), and
it seems to have been customary to opt for the method which collapses the log layer,
and rationalize the problems this presents by assuming the measurements near the wall to
be in error. While certainly all measurement sets are subject to inaccuracies near the wall,
the choice represents more an expression of faith in the log law than a consequence of
11
George and Castillos nomenclature is slightly different from the one used in this book.
5.5 MEAN FLOW 77
careful error analysis. In fact, curiously, measurements in the linear regions of pipes and
channels seemto have been routinely made over the past 50 or so years with little difculty
as long as the ratio of wire diameter to distance from the wall was greater than about 100
(v. Monin and Yaglom, 1971, vol. I, gure 25). Perry and Abell (1975) measure to within
0.4 mm with 4 jm hot-wires in a pipe ow before noticing effects of wall proximity. By
the same criterion, the data for Purtell et al. should be expected to be valid to values of y
+
from 2 to 5 for 8 of the 11 data sets. Thus, either hot-wires have behaved very differently
in boundary layers than in pipes, or the experimenters have been unwilling to accept the
answers the wires were providing.
The values of u
from the velocity gradient were made using a linear t for the data in
the region 3 - y
+
- 6, and points closer to the wall were ignored since they were clearly
contaminated by near wall effects on the wires. (It will be seen below that there is evidence
that the shear stress has been underestimated by about 20% by this method, but with
relative errors which are nearly Reynolds number independent.) The data show clearly
that the linear region both exists and the data in it collapse nearly perfectly. Moreover the
extent of the linear region appears to be greater than for pipes and channels, but decreases
with increasing Reynolds number. As expected from the new theory, the data outside
the linear layer does not perfectly collapse in inner variables, but shows the Reynolds
number dependence which must be present if the inner and outer scale velocities are
different.
Similar proles for the near wall region were published by Blackwelder and Haritonidis
(1983) who documented the difference between the shear stress inferred from the velocity
gradient near the wall and froma t to the log region. These systematic differences were
typically 15% and are tabulated in Table I of their paper. Their plot of the velocity prole
(normalized using the shear stress determined from the linear region) is strikingly similar
to that of the Purtell data plotted in Figure 5 [Figure 5.7 herein]. Of special interest is
the fact that the very near wall data show clearly the onset of the near wall measurement
errors well inside the linear region, except for the highest Reynolds numbers.
In the absence of direct force measurements, the fact that the velocity prole near the
wall collapses in both these data sets using the wall shear stress inferred from the gradient
suggests strongly that both the mean velocity and the shear stress have been reasonably
estimated, at least at the lower Reynolds numbers. A similar inference cannot be made
using a value of the stress which collapses a log region, since (as pointed out above) it
is based on the assumptions that such a region exists and that the constants describing it
are Reynolds number independent. Clearly the second of these is incorrect if the proles
normalized using the shear stress inferred from the velocity gradient are correct, and the
rst may be wrong as well.
The rst important inference which can be seen from these gures is that the data
collapse (if a cross-over can be called collapse) in the log region only if the shear
stress is calculated from a method which forces it to by assuming such a layer exists, and
then only by compromising the collapse in the linear region. On the other hand, when the
shear stress inferred from the mean velocity gradient is used, the proles collapse well very
close to the wall, but not in the log layer. The lack of collapse in the linear layer when
the Clauser method is used is particularly serious since there are no adjustable constants
or Reynolds number dependencies here: the measurements must yield u
+
= y
+
if they are
to be believed.
A second important inference from the gures is that when the data are normalized
by a proper shear stress, the point of departure from the linear region moves closer to
the wall with increasing Reynolds number toward what might be a limiting value. This
is clear evidence that if there is an intermediate layer outside the linear region, then it
is only asymptotically independent of Reynolds number. The apparent Reynolds num-
ber dependence of the extent of the near-linear region is consistent with the new theory
presented earlier since only this innermost region of the ow is Reynolds number inde-
pendent in inner variables. Once the mesolayer is entered, neither u
nor U
completely
collapse the data (except in the innite Reynolds number limit). Many (Gad-el-Hak and
Bandyopadhyay 1994) have noted similar trends for the turbulence moment data near the
wall, especially u
2
and uv.
5.5.2 Alternatives to the Logarithmic Prole
Despite copious evidence for the existence of the log-law, the whole scenario lead-
ing to it has been periodically questioned. From a purely practical point of view, a
portion of the streamwise mean-velocity prole could equally well t either a log-
arithmic relation or a power-law. If the mean-velocity prole for a pipe ow (or
boundary layer) is to be tted with a power-law of the form
U
U
c
= b
y
a
1
n
(5.19)
the value of the exponent (1,n) will decrease as the velocity prole becomes fuller. In
other words, n increases as Re increases and H decreases. For example, the smooth-
pipe-ow data of Nikuradse (1932) indicate that n changes from 6 to 10 as the
Reynolds number varies in the range of Re
a
=2.53 10
3
to 1.85 10
6
(Schlichting
1979).
Sreenivasan (1989) argues that although the power-law used by engineers to
describe the mean-velocity prole has been discredited by scientists since Millikan
(1939) derived the logarithmic law from asymptotic arguments, the basis for the
power-law is a priori as sound as that for the log-lawparticularly at low Reynolds
numbers (see also Barenblatt 1979, 1993; Barenblatt and Prostokishin 1993). The
behavior of the exponent (1,n) as Re
a
is of particular interest. If it tends to
zero, the log-lawis recovered. If, on the other hand, the limiting value of the exponent
is a nonzero constant, the log-law does not strictly hold. This implies that an inertial
sublayer is lacking and, therefore, that viscous effects persist even at innitely large
Reynolds number. Such a scenario is consistent with the suggestion by Long and
Chen (1981) that the inner ow is the outcome of an interplay between wall effects
and outer effects. According to them, it is strange that the matched layer between
one characterized by inertia and another characterized by viscosity depends only
on inertia. Long and Chen suggest that a mesolayer intrudes between the inner
and outer regions, preventing the overlap assumed in the derivation of the classical
logarithmic velocity prole. Unfortunately, existing experimental or numerical mean-
velocity data cannot be readily used to explore this important issue because the
difference between a logarithmic relation and a power-law with a large but nite n
is imperceptible (see Kailasnath 1993 for a comprehensive review of available data).
George et al. (1992, 1993, 1994) and most recently George and Castillo (1997)
have provided the most serious challenge to the validity of the log-law for external
wall-bounded ows. As discussed in Section 5.5.1, George and his colleagues assert
that boundary layer data taken at different Reynolds numbers collapse in the log-
region only if the shear stress is calculated from a method (i.e., the Clausers method)
that forces it to by assuming such a layer exists. Such supercial collapse compro-
mises the collapse in the viscous sublayer where no adjustable constants or Reynolds
number dependence should exist. As an alternative, George et al. (1993) used
5.5 MEAN FLOW 79
measured shear stress to normalize the data of Purtell et al. (1981) and showed
that the proles collapse well very close to the wall but not in the log-region, where
clearly discernible Reynolds number dependence is depicted. To remedy the situ-
ation, George et al. (1993) proposed matching a newvelocity-defect lawwith explicit
Reynolds number dependence and the traditional law of the wall. The result in the
matched region is a power-law velocity prole of the form
U
U
o
= C
o
n
+ B
o
(5.20)
U
u
= C
i
n
+ B
i
(5.21)
where the coefcients B
i
, B
o
, C
i
, C
o
and the exponent n are Reynolds number de-
pendent, but all are asymptotically constant. In a subsequent article, George et al.
(1994) have shown that the additive coefcients B
i
and B
o
are identically zero. Using
the two equations above, the friction coefcient is readily expressed as the Reynolds
number to the power 2n,(1 +n) .
George et al. (1994) asserted that their approach removes many of the unsatisfying
features of the classical Millikan (1939) theory. Furthermore, they argued that a clear
distinction should be made between internal and external wall-bounded ows. For
fully developed pipe and channel ows, the streamwise homogeneity ensures that the
pressure gradient and wall-shear stress are not independent, and thus u
is the correct
scaling velocity for the entire shear layer, including the core region. This results in a
log-law, although the owhas no constant-stress layer. The growing, inhomogeneous
boundary layer, in contrast, is governed by a power-law even though it does, at least
for zero pressure gradient and high Re, have a constant-stress layer. The matched
region of a boundary layer retains a dependence on streamwise distance and hence
never becomes Reynolds number independent. The same arguments presented here
for the mean ow could be extended to higher-order statistics.
The difference in the inner region between at-plate boundary layers and fully
developed channel ows explored above might have serious implications on the prin-
ciple and use of the Preston tube, a device widely employed for measuring the local
mean friction-coefcient. Such a gauge is commonly calibrated in a pipe ow and
relies on the universality of the inner layer to compute the skin friction in a at-plate
boundary layer. Such extrapolation is thus questionable, and direct measurements of
wall-shear stress are clearly preferred.
George and Castillo (1993) have also extended the new scaling law described
earlier for the at-plate ow to boundary layers with pressure gradient. Inclusion of
roughness or compressibility effects could proceed along similar attempts made in
the past for the classical theory (Hama 1954, Coles 1962).
The fresh look at the turbulent boundary layer by George and his colleagues is
intriguing and deserves further scrutiny. Independent conrmation of their claims
is needed, and carefully controlled boundary layer experiments over a wide range
of Reynolds numbers would be most useful. Low-Reynolds-number experiments in
which the linear region is resolved and the wall-shear stress is measured directly would
be particularly valuable. If validated, their new theory indicates that the boundary
layer is governed by a different scaling law than commonly believed. Explicit, albeit
weak, Reynolds number dependence is shown for the mean velocity prole all the way
down to the edge of the viscous sublayer. The matched region retains a dependence
on streamwise distance, and hence Reynolds number effects will always persist for
all turbulence quantities.
5.5.3 The Illusory Asymptotic State
Although inner scaling appears
12
to collapse wall-layer mean-ow data onto a
single curve regardless of the Reynolds number, the situation is not that simple in
the outer layer. As discussed in Section 5.3.3, Coles (1956) proposed to represent the
entire mean-velocity prole in any two-dimensional turbulent boundary layer by a
linear superposition of two universal functions, the lawof the wall and the lawof the
wake. In fact, Coles suggested a simple extension of his empirical law to represent
even yawed and three-dimensional ows. Recall Eq. (5.16)
U
+
= f (y
+
) +
H
(5.22)
The rst term on the right-hand side is valid for any smooth wall-bounded ow, and
available evidence appears to indicate that the function f is independent of Reynolds
number, pressure gradient, and freestream turbulence. This term supposedly repre-
sents all mixing processes in the wall layer governed primarily by viscosity. The wall
constraint is felt mainly in the viscous sublayer, the buffer layer, and the logarithmic
portion of the velocity prole. For rough walls, particularly when the roughness is suf-
ciently pronounced, the viscous length scale is simply replaced by the characteristic
roughness height. For both smooth and rough walls, the appropriate velocity scale
is derived from the magnitude of the surface shear stress.
The second term, representing turbulence mixing processes dominated by inertia,
is the product of the universal wake function W(y,) and the ratio of the prole pa-
rameter H to the von K arm an constant . The parameter H depends on the pressure
gradient, the freestream turbulence, and whether the ow is internal or external but
is not directly affected by wall conditions such as roughness, and so forth. For a
at-plate boundary layer, the prole parameter increases with Reynolds number but
presumably asymptotically approaches a constant value at high enough Re. When
Coles (1962) plotted the change of the maximum deviation of the mean velocity
from the logarithmic law, LU
+
, with Reynolds number, it reached a constant value
for Re
6000. Coles (1962) termed the owat this high Reynolds number equilib-
rium, which led to the widespread perception that the ow becomes independent of
Reynolds number beyond this value. Because the maximum deviation, the so-called
strength of the wake component, occurs close to the edge of the boundary layer and
W(y,) has been normalized such that W(1) = 2, the strength of the wake component
is approximately related to the prole parameter by
LU
+
2
H
(5.23)
It is clear that the strength of the wake depends upon the somewhat arbitrary way in
which the logarithmic portion of the velocity prole is tted, that is, on the particular
values of and B chosen.
12
At least until the recent work of George and his colleagues cited in the previous subsection.
5.5 MEAN FLOW 81
Unfortunately, when Coles (1962) extended the plot to larger values of Reynolds
numbers, it became clear that the presumed asymptotic state is merely an illusion.
The strength of the wake component started decreasing again at about Re
>15,000,
although very slowly compared with the rise rate for Re
- 6000. After excluding all

data containing certain anomalies, Coles (1962) was puzzled by the persistent change
in behavior and the variation between data sets for Re
>15,000. The drop cannot

be explained from experimental uncertainties such as those caused by probe calibra-
tion problems, improper tripping devices, three-dimensional effects, high levels of
freestream turbulence, and pressure-gradient effects, and Coles left the issue open.
The rapid rise and the subsequent gradual fall of LU
+
with Re
appear to be
genuine and have been conrmed in several other experiments, as summarized by
Mabey (1979). The maximumLU
+
, reached at about Re
= 6000, is 2.7 for subsonic

ows (Smith and Walker 1959) but is higher by 30% or more for supersonic ows
(Lee, Yanta, and Leonas 1969; Hopkins, Keener, and Polek 1972; Mabey, Meier, and
Sawyer 1976).
5.5.4 Is Self-Preservation Ever Achieved?
At approximately Re
>15,000, the gradual departure of LU

+
from the appar-
ent low-Re
asymptote suggests that some new effects are gradually appearing in

the turbulence production process. A new, lower asymptote appears to have been
reached when Re
= 50,000, but the boundary layer might also continue to change

indenitely, as the inner and outer scales are forever disparting. The paucity of high-
Reynolds number reliable data makes it difcult to form a denitive conclusion.
Without denitive experiments at even higher Reynolds numbers, one can never
be sure of the universality of the defect law. Very fewexisting facilities can deliver the
required ultra-high-Reynolds-number ows while maintaining relatively low Mach
number (in the so-called low-speed, high-Reynolds number tunnels) in order to avoid
the added complication of compressibility effects. The worlds largest wind tunnel,
the 24 m37 m Full-Scale Aerodynamics Facility at NASA Ames, is capable of
generating a boundary layer with a momentum-thickness Reynolds number as high
as 3.7 10
5
(Saddoughi and Veeravalli 1994).
The largest available water tunnels and towing tanks can deliver momentum
thickness Reynolds numbers of approximately 3 10
4
and 9 10
4
, respectively
(Figure 5.1). Cryogenic tunnels, for example the National Transonic Facility at NASA
Langley, typically use nitrogen and run as high as Re
= 610
4
, but their Mach num-
ber is near one, and they are rather expensive as well as heavily scheduled (Bushnell
and Greene 1991). Tunnels using liquid helium I are an attractive, low-cost alter-
native to the much larger nitrogen tunnels (Donnelly 1991). Helium facilities can
match the Reynolds numbers of the transonic wind tunnels but with essentially zero
Mach number and much smaller sizes (e.g., 1 cm 1 cm test section). Instrumenting
the smaller facilities with high-resolution velocity or pressure probes is at present
problematic, although the rapidly developing microfabrication technology has the
potential for producing inexpensive megahertz-frequency and micron-size sensors
(see, for example, L ofdahl, Stemme, and Johansson 1989, 1991, 1992; Ho and Tai
1996, 1998; Gad-el-Hak 1999; L ofdahl and Gad-el-Hak 1999).
A commonly accessible large-scale, high-Reynolds-number facility is the atmo-
spheric boundary layer. The ow is virtually incompressible, and the momentum-
thickness Reynolds number in the atmosphere can be as much as four orders of
magnitude higher than that in typical laboratory experiments. Unfortunately, such a
natural laboratory has several drawbacks. Firstly, the wall in this case is almost
always rough, and direct comparison with the canonical boundary layer is difcult.
Secondly, the atmospheric experiments are not well controlled, and thus the ow
conditions are neither precisely repeatable nor documentable to the needed detail
(see however the thesis by Kailasnath 1993, who was able to carry out useful com-
parison between low-Reynolds-number laboratory data and high-Reynolds-number
atmospheric data).
The so-called super-pipe facility is currently operational at Princeton University
(Zagarola 1996; Zagarola et al. 1996). The pipe has a diameter of 12.7 cm and a
length-to-diameter ratio of 200. This high-pressure-air (200 atm) pipe owis capable
of providing very high Reynolds numbers of up to Re
a
= 2.3 10
7
at a reasonably
large scale and low Mach number and, it is hoped, will help in answering many of
the questions raised in this chapter.
For the present at least, it is simply not known if the mean velocity in a wall-
bounded ow ever achieves true self-preservation. As will be seen in the next section
for higher-order statistics, the evidence for lack of self-preservation is stronger.
5.6 Higher-Order Statistics
Compared with the mean ow, higher-order statistical quantities are more difcult to
measure, and the issue of Reynolds number effects is even murkier. For quantities such
as root-mean-square, Reynolds stress, skewness, and spectrum, the issues of spatial
as well as temporal probe resolutions, three-dimensional effects, and boundary-layer
tripping devices become much more critical. In contrast to mean ow, reliable data
for higher-order statistics are scarce.
A measurement probe essentially integrates the signal over its active sensing area
or volume. This means that velocity or pressure uctuations having scales smaller
than the sensor size are attenuated by the averaging process, and the measured root-
mean-square of the uctuations, for example, is smaller than the true value. Several
studies have shown the importance of probe size in the detection of small-scale struc-
tures in the near-wall region (for example, Willmarth and Bogar 1977; Schewe 1983;
Johansson and Alfredsson 1983; Blackwelder and Haritonidis 1983; Luchik and
Tiederman 1986; Karlsson and Johansson 1986; Ligrani and Bradshaw 1987; Wei
and Willmarth 1989; L ofdahl et al. 1992). As a rule of thumb, probe length much
larger than the viscous sublayer thickness is not acceptable for accurately measuring
turbulence levels and spectra anywhere across the boundary layer, and even smaller
sensing elements are required to resolve dynamical events within the sublayer itself.
In the case of a hot wire, for example, the probe diameter determines its spatial reso-
lution for measuring the time-averaged velocity, whereas the much larger wire length
limits the accuracy of measuring the velocity uctuations.
The issue of sufcient probe resolution is particularly acute when studying
Reynolds number effects. A probe that provides accurate measurements at low Re
may give erroneous results when the Reynolds number is increased and the scales to
be resolved become smaller relative to the probe size. The probe resolution should
be expressed in wall units, and as mentioned above
+
should not be much larger
than 5, where is the probe length.
5.6 HIGHER-ORDER STATISTICS 83
The classic idea of inner scaling is that any turbulence quantity measured at dif-
ferent Reynolds numbers and in different facilities will collapse, at least in the inner
layer, to a single universal prole when nondimensionalized using inner-layer vari-
ables. In contrast to mean-velocity proles, higher-order statistics do not in general
scale with wall-layer variables even deep inside the inner layer. In the following two
subsections, we review Reynolds number effects on the root-mean-square values of
the velocity uctuations and the Reynolds stress. The cited data plus those for the
spectra, skewness and atness factors, and rms and spectrum of the wall-pressure
uctuations were collected from the original publications and presented in more
details in the review article by Gad-el-Hak and Bandyopadhyay (1994). The main
conclusion of this survey is that Reynolds number effects on the turbulence quantities
persist deep into the viscous region.
5.6.1 Root-Mean-Square Velocity Fluctuations
The intensity of turbulence uctuations is dened by its root-mean-square value.
The streamwise velocity uctuations are more readily measured using, for example, a
single hot-wire probe or a two-beam laser Doppler velocimeter. Measuring the other
two velocity components, in contrast, requires two hot wires either in an X-array or
a V-array or four intersecting laser beams. Especially very close to the wall, few reli-
able measurements of the normal velocity components are reported in the literature,
and even fewer are available for the spanwise component. A notable exception is
the oil-channel data reported by Kreplin and Eckelmann (1979), who measured all
three velocity components inside the viscous sublayer. The at-plate data of Purtell
et al. (1981) taken in the range Re
=4655100 showed clear Reynolds number ef-

fects penetrating into the boundary layer much deeper in terms of the turbulence
intensity than they do for the mean velocity. Approximate similarity in wall units
is maintained only out to y
+
15 compared with mean velocity, which appears to
be similar throughout the entire inner layer (but see the discussion in Section 5.5.1).
Close inspection of Purtell et al.s data at the lowest Reynolds number of Re
=465
reveals that even in the viscous region itself some weak dependence on Reynolds
number is observed in the rms value of the streamwise velocity uctuations. Purtell
et al. attributed the systematic decrease in u
across most of the boundary layer to the

suppression of all but the largest turbulence eddies as the Reynolds number is reduced.
Reynolds number can of course be changed either by varying the freestream ve-
locity at a xed streamwise location or by holding the tunnel speed constant and con-
ducting the measurements at increasing downstream locations. To check the state of
development of the ow, Purtell et al. (1981) measured, for a xed freestreamvelocity,
the mean-velocity proles at four downstream distances from the tripping device. At
the station closest to the distributed roughness used to trip the boundary layer, they
report underdevelopment in the outer region of the mean ow but an undistorted
logarithmic region that produces friction velocity values in agreement with directly
measured u
-values computed from near-wall measurements of U,y. The same

trends were observed at higher Reynolds numbers by Klebanoff and Diehl (1952).
When normalized with the freestream velocity, u
also shows distortion in the

outer region for measurements not sufciently far downstreamof the tripping device.
However, the rms data plotted in inner variables at a constant freestreamvelocity and
four downstream stations exhibit such a strong Reynolds number dependence in the
outer layer that this distortion is obscured. Close inspection of the data again shows
a small but systematic Reynolds number effect even below y
+
= 10. At a given y
+
,
u
+
rms
increases with x, that is, with Re
.
One might argue that the strong Reynolds number effects indicated above will
eventually subside at sufciently high Reynolds number. This is not the case, at least
up to Re
of 15,406, as shown by Andreopoulos et al.s (1984) at-plate data. Their

rms values of the longitudinal velocity uctuations show strong Reynolds number
dependence all the way to the edge of the viscous sublayer. In the buffer layer, u
/u
decreases as the Reynolds number increases from3624 to 12,436, thereafter reaching

what seems to be a constant value. An opposite trend, which continues for all four
Reynolds numbers they tested, is observed in the logarithmic and wake regions.
Andreopoulos et al. (1984) indicated that the behavior of their u
-data in the buffer

region is in general agreement with the channel-owresults of Laufer (1951), Comte-
Bellot (1963, 1965), and Zaric (1972).
Andreopoulos et al. (1984) were able to measure the normal velocity compo-
nent only at the lowest Reynolds number, Re
=3620. A weak second peak in v
/u
appeared at y
+
=4 and seems to be unique to this particular experiment, but the
authors do not comment on this. Reliable v
measurements could not be obtained

by Andreopoulos et al. for higher Reynolds numbers owing to limitations of the
applicable velocity range of their triple-wire sensor.
Relatively more data are available for two-dimensional channel ows. Laufer
(1951) conducted his experiments at three Reynolds numbers, based on channel
half-width and centerline velocity, of Re
a
=12,300, 30,800, and 61,600. Comte-
Bellot (1965) covered the higher range of Re
a
=57,000 and 230,000. Kreplin
and Eckelmanns (1979) experiments were conducted at Re
a
=3850. Johansson
and Alfredsson (1982) provided data for Re
a
=6900, 17,300, and 24,450. Probe
resolution problems appear to exist in both Laufers and Comte-Bellots high-
Reynolds number data. In the former experiment, the hot-wire length increased from
3 wall units at the lowest Reynolds number investigated to 13 wall units at the highest
Reynolds number, a fourfold loss in spatial resolution. Laufer (1951) observed (erro-
neously) a corresponding decrease in the peak value of u
/u
with increasing Reynolds

number. Comte-Bellots (1965) probe length increased from 13 to 36 viscous lengths
as Re
a
increased from 57,000 to 230,000. Correspondingly, she also observed (erro-
neously) a decrease in the peak value of the nondimensional streamwise turbulence
intensity from 2.85 to 2.5. High-quality turbulence data obtained using sufciently
small probes and facilities void of signicant trip-memory effects are clearly lacking.
In any case, the available data indicate that the turbulence intensity proles do not
collapse even deep into the inner layer.
Similar trends are observed in the rms values of the velocity uctuations normal to
the wall. The peak v
is lower than that for the streamwise uctuations and occurs far-
ther away from the wall. Validity of inner scaling deep into the viscous region cannot
be ascertained from available results because of the scarcity of data for y
+
<30.
In view of the poor quality of most of the data, Wei and Willmarth (1989) syste-
matically investigated the Reynolds number effects using a unique high-resolution,
two-component laser-Doppler velocimeter. To reduce the amount of ambient light
in the vicinity of the measuring volume and thus to improve the signal-to-noise-ratio
of the LDV system, four laser beams were entered and exited into the test section via
two narrow slits located at the two side walls of a two-dimensional water channel.
Both slots were covered with an extremely thin (17 m) window of heat-shrinking
Mylar lm, which virtually eliminated optical refraction by the window. The laser
beams intersected at a single point away from the wall, and the effective probe
length ranged from 0.66 to 6.43 wall units as the Reynolds number was varied in the
range of Re
a
=300040,000. Beam refraction in the path between the laser source
and the measuring station was minimized using a specially designed optical head.
Wei and Willmarth (1989) ascribed the slight disagreement between the different
data sets available in the literature and theirs to a decreased spatial resolution of the
hot wires used by Laufer (1951), Comte-Bellot (1965), and Johansson and Alfredsson
(1982) at high Reynolds numbers. But, even in the newer data set, the uctuating
turbulence quantities do not scale with wall variables even at as close as 10 viscous
lengths from the wall. In fact, inner scaling does not seem to apply to v
across the
entire portion of the viscous region where measurements are available.
Additional support for Wei and Willmarths (1989) basic conclusion that turbu-
lence quantities in the near-wall region do not scale on wall variables comes from
the boundary layer experiments of Purtell et al. (1981) and Andreopoulous et al.
(1984) referenced earlier in this section as well as from the physical and numerical
channel-ow experiments conducted by Antonia et al. (1992) and the at-plate ex-
periments of Murlis, Tsai, and Bradshaw (1982), Wark and Nagib (1991), Naguib
(1992), and Naguib and Wark (1992).
13
Contrary to the prediction of the univer-
sal theory of the wall, the peak value of u
+
rms
, though occurring at the same y
+
,
increases with Reynolds number. It is, of course, conceivable that wall-layer scaling
might apply over the entire inner layer provided that the Reynolds number is high
enough, but at the moment at least such ultra-high-Reynolds-number experiments
cannot be conducted with sufcient probe resolution.
Coles (1978) summarized the results of fty different experiments conducted in
circular pipes, rectangular channels, and zero-pressure-gradient boundary layers. He
did remark that not all experiments are equally reliable. Nevertheless, Figure 3 of
Coles article indicates that, when the ratio of outer to inner length-scales,
+
, in-
creases from 100 to 10,000, the value of the rms streamwise velocity uctuations
measured at y
+
= 50 and normalized with the corresponding peak value measured
at y
+
15 systematically increases from about 0.6 to 0.9. This result is consistent
with a nonnegligible Reynolds number effect on the turbulence just outside the vis-
cous region.
5.6.2 Reynolds Stress
Reynolds Number Effects
Turbulence shear stress, or Reynolds stress uv, is the most important dynami-
cal quantity affecting the mean motion. The major portion of the momentum trans-
ported in a two-dimensional turbulent wall-bounded ow is accomplished by uv.
Therefore, modeling the behavior of Reynolds stress is one of the primary objec-
tives of various prediction schemes. Simultaneous measurements of the streamwise
and normal velocity uctuations are required to compute the Reynolds stress at any
particular point in the oweld. Provided this is done with high delity in the low-
to-moderate Reynolds-number laboratory experiments, extrapolation to the higher-
13
The channel-ow experiments of Walker and Tiederman (1990) and Harder and Tiederman (1991)
should also be mentioned. Their results do not agree with those of Wei and Willmarth (1989), but
Gad-el-Hak and Bandyopadhyay (1994) concluded that the latter set of data is more reliable.
Reynolds-number eld conditions is only possible if the Reynolds number effects
are well understood. In this subsection, we review those effects in boundary layers,
channels, and pipes.
In boundary layers, Andreopoulos et al. (1984) measured the normalized cross
correlation uv for a single Reynolds number of Re
= 3664. Again, they were

unable to measure the Reynolds stress reliably at higher Reynolds numbers owing
to limitations of the applicable velocity range of their triple-wire probe. The directly
measured turbulence shear stress was on the average 10%smaller than the theoretical
distribution deduced from the momentum balance and mean-ow data.
More data are available for two-dimensional channel ows. The (kinematic)
Reynolds stress is directly computed by averaging the product of the measured
u- and v-velocity uctuations. Eckelmann (1974), using an oil channel, covered the
low Reynolds numbers of 2800 and 4100. Alfredsson and Johansson (1984) con-
ducted their experiment at Re
a
= 7500, whereas Kastrinakis and Eckelmann (1983)
conducted theirs at Re
a
= 12,600. Comte-Bellot (1965) covered the higher range of
Reynolds number of 57,000 and 230,000. Wei and Willmarth (1989) covered the
Reynolds number range of 2970 to 39,582. Their high-resolution LDV allows mea-
surements very close to the wall where the Reynolds stress is decreasing. When plotted
against y/a, the maxima of the nondimensional turbulence shear stress proles in-
crease in magnitude and are closer to the wall as the Reynolds number increases.
But when plotted versus y
+
, the maximum value of the normalized turbulence shear
stress is not the same for each prole, indicating the lack of inner scaling in the
Reynolds number range investigated. When expressed in wall units, the location of
peak Reynolds stress moves away from the wall as the Reynolds number increases.
One advantage of investigating fully developed pipe or channel ows is the ability
to compare direct measurements with the computed Reynolds stress using the mean-
velocity prole and pressure gradient, two quantities that are easier to measure.
This has the advantage of enabling the accuracy of the directly measured Reynolds
stress to be checked, especially near the wall where probe resolution problems are
particularly acute. In a fully developed channel or pipe ow, the average normal and
spanwise velocities vanish, there are no mean longitudinal velocity or Reynolds stress
variations in the streamwise and spanwise directions, and the pressure gradient is a
constant. The longitudinal momentum equation could then be integrated to give an
exact relation between the Reynolds stress and mean-velocity distribution
uv =
dU
dy
+u
2
1
y
a
(5.24)
where U(y) is the streamwise mean-velocity distribution, and a is the channel half-
width or pipe radius. The friction velocity, or the slope of the velocity prole at the
wall, is related to the constant pressure gradient through
u
2
dU
dy
w
=
dp
dx
(5.25)
where p is the static pressure and and are the uid density and kinematic viscosity,
respectively. In wall units the momentum balance equation (5.24) reads
uv
+
=
dU
+
dy
+
+
1
y
+
a
+
(5.26)
Wei and Willmarth (1989) used Eq. ( 5.26) to compute the Reynolds stress proles,
and again the nondimensional proles at different Reynolds numbers do not collapse
in the outer and logarithmic regions and even well into the viscous region. Except
very close to the wall, the agreement between the directly measured Reynolds stress
and that computed from the measured mean velocity and pressure gradient is very
good and attests to the accuracy of the instantaneous velocity traces reconstructed,
ltered, and smoothed from the Doppler burst detector and processor signals. Wei
and Willmarth speculated that the divergence between the directly measured and
computed Reynolds stresses is due to insufcient spatial and temporal resolution in
the direct LDV measurement very close to the wall.
Wei and Willmarth (1989) also computed the turbulence kinetic energy produc-
tion using both the directly measured Reynolds stress and the momentum balance
equation. Except very close to the wall, the two methods of computing uv(dU/dy)
agree within 10%. Neither method leads to a prole collapse even in the viscous
region. The maximum value of kinetic energy production obtained from the momen-
tum balance increases with Reynolds number. Interestingly, although the position
of peak Reynolds stress, expressed in wall units, moves away from the wall as the
Reynolds number increases, the peak turbulence production seems to be xed at
y
+
1215. This point will be revisited later in this subsection.
The measurements of mean velocity and pressure drop in the smooth-pipe-owex-
periments of Nikuradse (1932) and Laufer (1954) were used to compute the Reynolds
stress proles shown previously in Figure 5.2. The Reynolds number in that gure,
Re
= u
a/, is the ratio of the pipe radius to the viscous length scale and varies over
the wide range of 14055,400. An approximately constant-turbulent-shear-stress re-
gion is clear at the highest Reynolds number. As in the channel ows, the peak value
of normalized Reynolds stress increases, and its location, relative to the viscous length
scale, moves away from the wall as the Reynolds number increases.
Peak Location
Sreenivasan (1989) analyzed several different wall-bounded ow experiments.
The distance from the wall, expressed in inner variables, where the streamwise tur-
bulence intensity peaks appears to be independent of Reynolds number (Figure 16
of his article). In contrast, the location where the largest normal uctuations occur
is a strong function of Reynolds number (Figure 17 of his article)
y
+
max
= [Re
]
0.75
(5.27)
where Re
is the pipe radius or boundary-layer thickness in wall units. Available data

on the spanwise intensity are scarce, and no conclusion can be reached on the scaling
of its peak position. For the total turbulence kinetic energy, however, the position of
its peak does scale on wall variables much the same as u
. This is because the near-

wall value of the total uctuation energy is essentially overwhelmed by the streamwise
component.
Similar to the normal uctuations, the peak Reynolds stress occurs at increasingly
higher values of y
+
as the Reynolds number increases, as shown in Figures 5.8a
and 5.8b, compiled from directly measured and computed turbulence shear-stress
data, respectively.
14
The lowermost two data points in Figure 5.8b correspond to
14
These same plots also appeared earlier in the article by Long and Chen (1981).
Figure 5.8: Location of peak Reynolds stress as a function of Reynolds number. Data compiled by
Sreenivasan (1988) from various wall-bounded ow experiments. Solid lines are least-square t.
(a) Directly measured Reynolds stress in different boundary layers. (b) Computed from measured
mean velocity in different channel ows; the lowermost two data points correspond to the critical-
layer position in typical transitional ows.
the location of the peak Reynolds stress, or the critical layer position, in typical
transitional boundary layer and channel ows. Aleast-square t of all the data points
in both gures leads to the same equation
y
+
p
= 2 [Re
]
0.5
(5.28)
where y
p
is the location of peak Reynolds stress. Note that although probe resolution
has a signicant effect on the magnitude of turbulence intensity, Reynolds stress, or
other higher-order statistics, a relatively long probe should have less effect on the
accuracy of determining the peak location of these quantities. It is therefore not
surprising that Sreenivasan (1989) could use a variety of data sources, including
some with insufcient probe resolution, to arrive at the correlations in Eqs. (5.27)
and (5.28).
Equation (5.28) indicates that the location of the peak turbulence stress scales
on the geometric mean of the inner and outer scales. Recall that, because this is the
position in Sreenivasans (1988) model discussed in Section 5.4, where all the mean
vorticity of a turbulent boundary layer has been assumed to be concentrated into
a single sheet, the correlation in Eq. (5.28) and its applicability to the critical-layer
position in transitional ows give some credence to his hypothesis.
Note that if a velocity prole is assumed for the case of fully developed channel
ow, exact expressions for the location of the peak Reynolds stress and turbulence
kinetic energy production can be derived from Eq. (5.24). For example, if the mean
prole is approximately logarithmic in the vicinity of y
p
, the peak Reynolds stress
occurs at
y
+
p
=
Re
0.5
= 1.56 [Re
]
0.5
(5.29)
whereas the peak production occurs at a xed y
+
. Furthermore, the value of the
Reynolds stress at its peak can also be computed from the streamwise momentum
equation (5.24)
[uv]
max
u
2
= 1 2 [ Re
]
0.5
(5.30)
Thus, both the maximum value of the Reynolds stress and the location of its peak
depend directly upon the outer lengthscale a (or ), and the peak location scales with
the geometric mean of the inner and outer scales.
At high Reynolds numbers the peak Reynolds stress occurs substantially outside
the viscous region. Note however that, owing to the shrinking of the inner layer
as the Reynolds number increases, this peak location moves closer to the wall as
a fraction of the boundary-layer thickness. Interestingly, although the most signif-
icant Reynolds-stress-producing activity does not occur at a universal value of y
+
,
the production of turbulence kinetic energy, uv(dU/dy), does always peak at y
+

1215. This implies that, at high Reynolds numbers where the two positions dispart,
the scales producing the Reynolds stress are quite different from those responsible
for the turbulence kinetic energy production. It is this observation that led Townsend
(1961) to hypothesize the existence of an active motion and an inactive motion within
the inner layer. The former is due to the vorticity eld of the inner layer proper and is
responsible for turbulence kinetic energy production. The statistical properties of the
active motion are presumably universal functions of the distance from the wall. The
inactive, larger-scale motion is partly due to the irrotational eld sloshing associated
with the pressure uctuations in the outer layer and partly the large-scale vorticity
eld of the outer-layer turbulence, which the inner layer sees as an unsteady external
stream (see also the substantiative measurements of Bradshaw 1967). The inactive
motion does not scale with inner variables and is characterized by intense velocity
uctuations. The effect of increasing the Reynolds number can then be thought of as
the increasing signicance of the inactive motion (see also Naguib 1992).
The primary conclusion of this and the previous subsubsections is that Reynolds
number does have an effect on the turbulence shear stress even in the inner layer.
Inner scaling fails to collapse the Reynolds stress proles. The peak value of uv
increases with Reynolds number, and its position moves outward when expressed in
wall units.
Asymptotic Theory
The results discussed in this section thus far indicate that inner scaling fails to
collapse the proles for the Reynolds stress and for the root-mean-square values of the
velocity uctuations. Considerable Reynolds number effects are exhibited even for y
+
values less than 100. Panton (1990a) pointed out that a turbulent wall-bounded ow
is fundamentally a two-layer structure, a classical single perturbation situation. At
nite Reynolds numbers, neither the inner representation nor the outer representation
is a uniformly valid approximation to the true answer in the matching region. As Re
varies, the overlap layer changes size, and the proportions of inner and outer effects
are altered.
A uniformly valid answer for the present singular perturbation problem can be
obtained by forming an additive composite expansion from the inner and outer ex-
pansions. Matching essentially replaces the two lost boundary conditions at y = 0
and y = , and the additive composite expansion is simply the sum of the inner and
outer ones minus the common part (see, for example, Van Dyke 1964). Systematic
changes with Reynolds number considered anomalies when turbulence quantities are
expressed as inner expansions can then be regarded as proper rst-order trends that
are expected when viewed in the proper light. Such treatments were demonstrated
for the mean ow (Panton 1990b), for the rms turbulent uctuations (Panton 1991),
and for the Reynolds stress (Panton 1990a). Root-mean-square values of the veloc-
ity uctuations or Reynolds stress expressed as additive composite expansions are
equivalent in accuracy to the mean velocity expressed as the law of the wall plus the
law of the wake.
For the mean ow, Panton (1990a) suggested the use of an inner velocity scale that
is different fromthe friction velocity (see Eq. (5.7) of this volume), the two being equal
only in the limit of innite Reynolds number. Within the framework of an asymptotic
theory (Yajnik 1970; Afzal 1976; Afzal and Bush 1985), the lowest-order equation
for the mean ow shows weak Reynolds number dependence, whereas that for the
Reynolds stress indicates a much stronger effect. According to Panton (1990a), the
logarithmic nature of the inner, outer, and composite expansions for the mean ow
dictates minimal Reynolds number effects. On the other hand, the Reynolds stress
behaves algebraically, and the innerouter effects are mixed in different proportions
and occur at different locations, resulting in strong Reynolds number dependency
even in the rst-order theory. Moreover, the additive composite expansion for the
Reynolds stress does not evince a constant-stress region; only the inner expansion
does so.
5.7 Coherent Structures
The next question to be discussed in this chapter relates to Reynolds number effects
on the coherent structures in wall-bounded ows. Reference is made to Section 4.4
of the previous chapter. There are several facets to that issue, and the present sec-
tion is divided into four subsections. To be addressed below are proper scaling for
the period between bursts, the possibility of profound structural changes after the
5.7 COHERENT STRUCTURES 91
well-known Reynolds number limit of Re
= 6000, the ratio of outer to inner scales,

and Reynolds number effects on inner structures.
5.7.1 Bursting Period
Because of the problems of threshold setting and probe resolution, bursting fre-
quency and its scaling have become the source of continuing controversy. Cantwell
(1981), based on a review of available literature, has concluded that this frequency
scales on outer variables, thus establishing a strong link between the inner and outer
regions of a wall-bounded ow. On the other hand, Blackwelder and Haritonidis
(1983) have shown that the frequency of occurrence of these events scales with the
viscous parameters, which is consistent with the usual boundary-layer scaling ar-
guments. Their results, obtained with a hot-wire probe whose length varied in the
range of
+
= 4.520 as the Reynolds number increased in the range of Re
= 1000
10,000, are shown in Figure 5.9. In Figure 5.9a, outer variables are used to normalize
the bursting frequency (or its inverse, the period between bursts). The nondimen-
sional frequency increases with Reynolds number, thus clearly indicating that outer
scaling is not applicable. On the other hand, the same data plotted in Figure 5.9b
using the viscous time scale to normalize the frequency indicate the validity of in-
ner scaling. Thus, the properly nondimensionalized bursting period is independent
of the Reynolds number in agreement with the observations of Kline et al. (1967),
Corino and Brodkey (1969), Donohue et al. (1972), Achia and Thompson (1977),
and Blackwelder and Eckelmann (1978). Blackwelder and Haritonidis (1983) have
suggested that past erroneous results are caused by insufcient spatial resolution of
the sensors used to detect the bursts.
On the basis of measurements in the atmospheric boundary layer, where the
Reynolds number is several orders of magnitude higher than in typical laboratory
experiments, Narasimha and Kailas (1986, 1987, 1990) still maintain that bursting
events scale on outer variables. To do otherwise, the insufcient time resolution of the
atmospheric data would simply not have allowed the detection of any dynamically
signicant events. Narasimha and Kailas cite other laboratory experiments to support
their position (e.g., Rao, Narasimha, and Badri Narayanan 1971; Ueda and Hinze
1975; Willmarth 1975a; Shah and Antonia 1989; Rajagopalan and Antonia 1984).
Adding to the present confusion, Bandyopadhyay (1982) has shown that the
bursting period is not a universal function and that both inner and outer variables
are involved in its scaling with Re
. He reviewed existing data and concluded that a

universal value of the bursting frequency scaled with either inner or outer variables
in various boundary layers ranging from relaminarized to separated does not exist.
Because a turbulent boundary layer is characterized by three integral variables (C
f
,
H, and Re
), verication of universality with Re
alone is clearly inadequate, and

the apparent confusion stems in part from the lack of experiments over a sufciently
wide range of shape factors H. Johansson and Alfredsson (1982) have also suggested
that the bursting period scale with intermediate scaling proportional to the geomet-
ric mean of the inner and outer scales. It should be noted, however, that, within
the framework of an asymptotic theory, mixed variables have little or no physical
signicance.
The arguments by both Blackwelder and Haritonidis (1983) and Narasimha and
Kailas (1987) are compelling, and the issue of scaling of the bursting events must,
for the moment at least, stay open. The laboratory experiments of the former group
Figure 5.9: Mean bursting frequency versus Reynolds number [after Blackwelder and
Haritonidis 1983]. (a) Scaling using outer-ow variables. (b) Scaling using inner-ow vari-
ables; the inverted triangles represent three additional data points froman untripped boundary
layer.
were well controlled, but the range of Reynolds numbers and range of shape factors
investigated are not large enough. The latter group experiments were conducted with
sufcient probe resolution, but the atmosphere can neither be controlled nor fully
documented. Moreover, the effects of roughness on the scaling are simply not known.
Controlled high-Reynolds-number experiments using smooth walls and probes with
sufcient resolution should settle the question.
5.7.2 High Reynolds Number
Does the boundary layer structure change when Re
> 6 10
3
? The Reynolds
number variations of U
+
attributed to Coles (1962), discussed in Section 5.5.3, at
rst suggest that an asymptotic state is reached approximately when Re
> 6 10
3
.
But, the higher-Reynolds-number data beyond that limit show that U
+
drops, al-
though very slowly compared with the rise rate for Re
<6 10
3
. The gradual depar-
ture of U
+
from the apparent low asymptote suggests that some new effects are ap-
pearing in the turbulence production process at approximately Re
>(6 to 15) 10
3
.
Experiments conducted in several different facilities are briey described below, and
they showthat profound changes in the coherent structures of different wall-bounded
ows may indeed take place at very high Reynolds numbers.
Relevant to the issue of structural changes when Re
>6 10
3
is the recent as-
sertion by Kailasnath (1993) that the skin friction, the pressure uctuations, and
the mean-velocity proles all show a distinct change of behavior at about the same
Reynolds number. For example, a power-lawt to existing skin-friction data for both
boundary layers and pipe ows indicates a break point, at Re
5000, that sepa-

rates two ranges of Reynolds numbers. This and other evidence prompted Kailasnath
(1993) to propose a transitional behavior for wall-bounded ows from the low- to
the high-Reynolds-number regimes and to suggest further that the turbulence regen-
eration mechanism is different in the two regimes.
The ow visualization results of Head and Bandyopadhyay (1981) briey men-
tioned in Section 5.2.2 indicate that the hairpin structures exhibit strong dependence
on Reynolds number for Re
< 7000, and hence the hairpins are atypical. At higher

Reynolds numbers, on the other hand, the hairpin vortex is found unambiguously.
In Head and Bandyopadhyays view, Falcos (1977) typical eddies are merely the
longitudinal cross sections of the tips of the hairpins. Perry and Chong (1982) and
Perry et al. (1986) concur with this view. Their model of the turbulent boundary
layer emphasizes a hierarchy of hairpin eddies as the quintessential structure of the
outer region. In wave number space, the analogous idea of a hierarchy of interacting
scales and energy transfer fromlarge eddies to smaller ones is, of course, not newand
has been proposed as early as 1920 by Richardson and formalized by Kolmogorov
(1941a).
Turbulent boundary layers ranging from relaminarized to separated cover the
entire range of possible shape factor H. The statistical properties of the turbulent
irrotational uid interface as well as the bursting period in such diverse layers can be
described by H (Fiedler and Head 1966, Bandyopadhyay 1982). As can be expected,
the location of the maximum deviation of the mean velocity from the logarithmic
law also correlates with the mean location of the intermittent layer. Changes in the
properties of the intermittent layer can take place when H drops below 1.3, which
is approximately when Re
> 10 10
3
. This is supported by the ow visualization
results of Head and Bandyopadhyay (1981) at Re
= 17.5 10
3
, which shows that
the outer part of the boundary layer is noticeably sparser: fewer of the hairpin vortices
reach , although more of them are produced per unit (dimensional) wall area.
In high-speed, high-Reynolds-number turbulent boundary layers, the mean loca-
tion of the intermittent layer and its standard deviation change signicantly according
to the results of Owen et al. (1975) at Ma
= 7.0 and Re
= 8510
3
. The intermit-
tency prole for the supersonic boundary layer is clearly fuller. Furthermore, at these
high Reynolds numbers, the boundary layer structures do not exhibit much over-
turning motion, which is typical of lower Reynolds numbers (see the recent book by
Smits and Dussauge, 1996, for useful information on supersonic shear layers). In the
statistical measurements of conventional boundary-layer properties at high Reynolds
numbers, these changes may not always seem dramatic, but their critical importance
may lie in the efciency of outer-layer or other control devices for drag reduction.
Morrison, Bullock, and Kronaner (1971) compared the sublayer spectra of the
longitudinal velocity uctuations for low- and high-Reynolds-number pipe ows.
Over the sufciently wide range of Reynolds numbers Re
a
=10,000100,000, the
shape of the two-dimensional spectra expressed in wall-layer variables is not uni-
versal. This result contradicts the earlier low-Reynolds-number one- and two-
dimensional spectral observations made by Bakewell and Lumley (1967) and
Morrison (1969). As the Reynolds number is increased, more energy appears in
the low-frequency, low-wave-number region. The additional energy results from dis-
turbances that convect at twice the characteristic velocity of the sublayer of 8u
.
The high Reynolds numbers appear to have the effect of randomizing the phase
velocity whereby the disturbances are no longer phase-correlated in the sublayer.
This additional evidence also suggests much change in the turbulence production
mechanism at very high Reynolds numbers. In fact, Morrison et al. (1971) have
strongly suggested that the low-speed streaks are unique to low-Reynolds-number
wall-bounded ows. Streaks would no longer appear at very high Reynolds numbers
at which a phase-correlated, wavelike turbulence might not exist within the viscous
sublayer.
Using a rake of X-wires and conditional averaging techniques, Antonia et al.
(1990) have examined the effects of Reynolds number on the topology of the large
structures in the range 1360 Re
9630. The instantaneous longitudinal sectional

streamlines in a moving frame of reference contain many rotational structures O[/2]
at the lowest Reynolds number. Very signicant Re
effects can be observed in the

instantaneous frames (see their Figure 5). As Re
is gradually increased to 9630, the

large rotational structures become much smaller and no longer dominate the outer
layer. When the large structures are selectively sampled and averaged, their foci are
found to be more circular at lower Reynolds numbers. As Re
is increased from
1360 to 9630, the location of the foci moves closer to the wall from 0.83 to 0.78.
This is consistent with the effect of Reynolds number on the mean location of the
intermittent layer for similar values of the shape factor H (Fiedler and Head 1966).
5.7.3 Small Structures in Outer Layer
In this subsection, a relation is developed relating the ratio of outer to inner
scales to Reynolds number changes. The boundary-layer thickness in wall units,
+
,
is related to the Reynolds number Re
via the skin-friction coefcient and the ratio

of boundary-layer thickness to momentum thickness
+

u
=
u
Re
(5.31)
C
f
2
1
2
Re
(5.32)
Figure 5.10: Reynolds number dependence of ratio of outer to inner scale. The
straight line is computed from the modied pipe-resistance formula and power-
law mean-velocity prole. Data compiled by Bandyopadhyay (1991) from different
experiments.
For a smooth at plate, an approximate empirical relation can be obtained by using
the modied pipe-resistance formula
C
f
= 0.0296(Re
x
)
1
5
(5.33)
and the 1/7th-power-law velocity prole. The ratio of the outer scale to inner scale
is thus given by
+
= 1.168(Re
)
0.875
(5.34)
Figure 5.10, taken from Bandyopadhyay (1991), shows that Eq. (5.34) describes the
data over the entire Reynolds number range for which measurements are available.
5.7.4 Inner Structures
In this subsection, Reynolds number effects on the inner structures are discussed. It
will be argued that vortex stretching is enhanced at higher Reynolds numbers and that
the low-speed streaks, commonly observed in low-Reynolds number experiments,
may become less important at higher speeds.
Wei and Willmarth (1989) have argued that in turbulent channel-ow literature,
where an inner-layer scaling has been claimed to hold, errors have crept into mea-
surements that are ascribable to large invasive probes and that sometimes small
but systematic variations with Reynolds number have been overlooked because
such a scaling was assumed a priori. Their measurements, discussed previously in
Section 5.6, show that u
rms
/u
turbulence intensity scales on inner variables only up

to y
+
10, which is well inside the inner region. The v
rms
/u
turbulence intensity
and the Reynolds shear stress uv/u
2
distributions do not scale on inner variables

anywhere in the channel. Interestingly, the maximumnormalized Reynolds stress and
normal turbulence intensity increase with Reynolds number. This they attributed to
the enhancement of the vortex-stretching mechanismin the inner region with increas-
ing Reynolds number.
Because the Reynolds number Re
a
or Re
is a dimensionless grouping of outer

variables, failure of the turbulence quantities in the inner region to scale only on
inner variables is an indication that the dynamics of the inner region structure are
affected by outer as well as inner variables. Wei and Willmarth (1989) argue that these
Reynolds-number dependencies are caused by changes in the coherent structure of
the turbulence close to the wall and that the turbulent owstructure at high Reynolds
number near solid boundaries (i.e., the hairpin vortex structure and interactions) will
differ signicantly from lower-Reynolds-number inner structures.
Morrison et al. (1971) have studied the organized motion in the sublayer region
of a pipe ow. As discussed in Section 5.7.2, they have measured the two-dimensional
frequencywave-number spectra of the longitudinal velocity uctuations at 10.6
10
3
Re
a
96.5 10
3
. An appeal of this data lies in its long-time averaged statistical
nature and the absence of any troublesome subjective threshold setting as used in
VITA or VISA techniques. For Re
a
<30 10
3
, the streamwise phase velocity c
+
x
in
the sublayer is independent of wave number and remains a constant throughout
the sublayer. Because this constant velocity coordinates the phase of the periodic
motions at different wall-normal locations, Morrison et al. concluded that the
sublayer turbulence is wavelike, and, in fact, at low Reynolds numbers it is likely
that the sublayer consists of relatively periodic waves.
The critical-layer height is estimated to be 9/u
because at that location the av-

erage uid velocity equals c
+
x
= 8. At Re
a
< Re
crit
= 30 10
3
, the characteristic
spanwise wavelength
+
z
of 135 agrees with Kline et al.s (1967) streak-spacing es-
timate of 130. However, for Re
a
> Re
crit
, the frequency/wave-number spectra for
the laterally spaced points lose their universal shape, and the relative amount of
low-frequency, low-wave-number energy increases with Reynolds numbers. This ad-
ditional energy, which becomes signicant at higher Reynolds numbers, results from
disturbances that convect at velocities much greater (the average being 16u
) than the
characteristic sublayer velocity of 8u
. This led Morrison et al. (1971) to conclude

that, at higher Reynolds numbers, the character of the sublayer will be substantially
altered inasmuch as an increasing amount of low-frequency, low-wave-number en-
ergy will be introduced. The disturbances responsible for this additional energy have
propagation velocities much larger than that which characterizes the sublayer at low
Reynolds numbers. The streaky structure that has been assumed to be character-
istic of the sublayer will become less important as the Reynolds number is increased,
and it is probable that the streaks may not be apparent at all at sufciently large
values.
By trial and error, Walsh (1990) has optimized the dimensions of drag-reducing
V-groove riblets (h
+
= s
+
= 12, where h and s are the riblet height and spanwise
spacing, respectively) at low Reynolds numbers (Re
< 6 10
3
). The value of s
+
does not scale with the spanwise mean-streak-spacing
+
z
100. The ndings of
Walsh, Sellers, and McGinley (1989) and Walsh (1990) that the riblet performance
does not change at transonic speeds and high Reynolds numbers (Ma
= 0.7 and
20 10
3
Re
50 10
3
) do not, therefore, invalidate Morrison et al.s (1971)
conclusion that the low-speed streaks will gradually become unimportant at high
Reynolds numbers.
5.8 FLOW CONTROL 97
Grass (1971) investigated the nature of inner and outer interaction in smooth
as well as rough wall-bounded ows. He maintains that the essential features of
this interaction do not change despite the presence of three-dimensional roughness
elements that protrude as much as 80 wall units into the inner layerwell outside
the viscous region. Because low-speed streaks are not observed on walls with three-
dimensional roughness, Grass results minimize the importance of the streaks in the
maintenance of turbulence.
A related issue is the importance of the intense but rare bursting events at high
Reynolds numbers. A partial answer is given by Kailasnath (1993), who used a sta-
tistical approach to obtain useful information on the structure of the instantaneous
momentum ux, thus sidestepping analysis conditioned on specic episodes and fo-
cusing instead on the contribution to the momentum ux associated with various
magnitudes of velocity uctuations. Kailasnaths nonepisodic approach reveals that
the contribution to the ux is dominated by medium amplitude velocity uctuations
in the range of 1.5u
, which are not rare events. This implies a diminishing impor-

tance of the rare, intense events taking place in a progressively shrinking near-wall
region as the Reynolds number increases.
5.8 Flow Control
5.8.1 Introductory Remarks
The ability to actively or passively manipulate a oweld to effect a desired change
is of immense technological importance. The term boundary-layer control includes
any mechanism or process through which the boundary layer of a uid ow is made
to behave differently than it normally would were the ow developing naturally
along a smooth, at surface. A boundary layer could be manipulated to achieve tran-
sition delay, separation postponement, lift enhancement, drag reduction, turbulence
augmentation, or noise suppression. These objectives are not necessarily mutually
exclusive. For example, by maintaining as much of a boundary layer in the laminar
state as possible, the skin-friction drag and the ow-generated noise are reduced.
However, a turbulent boundary layer is in general more resistant to separation than
a laminar one. By preventing separation, lift is enhanced, and the form drag is re-
duced. An ideal method of control that is simple, inexpensive to build and operate,
and does not have any trade-off does not exist, and the skilled engineer has to make
continuous compromises to achieve a particular goal.
Of all the various types of shear-ow control now extant, control of ow sep-
aration is probably the oldest and most economically important. The tremendous
increases in the capability of computational uid dynamics, which have occurred as
a direct result of increases in computer storage capacity and speed, are transforming
ow separation control from an empirical art to a predictive science. Control tech-
niques such as mitigation of imposed pressure gradients, blowing, and suction are all
readily parameterized via viscous CFD. Current inaccuracies in turbulence modeling
can severely degrade CFD predictions once separation has occurred; however, the
essence of ow separation control is the calculation of attached ows and estima-
tion of separation location (and indeed whether or not separation will occur), which
are tasks CFD can in fact perform reasonably well within the uncertainties of the
transition location estimation.
The knowledge of the Reynolds number effects is useful to ow control. This
is because experimental investigations at low Reynolds numbers (i.e., lower speeds,
smaller length scales, or both) are less expensive. Most ow-control devices are,
therefore, developed and tested at rather lowReynolds numbers, say Re
= O[1000].
Extrapolation to eld conditions is not always straightforward though, and it often
comes to grief. Although the riblets results seem to extrapolate favorably to eld
conditions, the verdict on large-eddy breakup devices is disappointing. These points
will be discussed in the following three subsections.
5.8.2 Riblets
Properly optimized, longitudinally grooved surfaces, called riblets, can lead to
a modest skin-friction drag reduction, in the range of 510%, in turbulent bound-
ary layers. The subject dates back to the mid-1960s but has attracted much attention
during the 1970s and 1980s. Walsh (1990) and Pollard (1998) provide up-to-date re-
views. The exact mechanismthrough which riblets achieve net drag reduction despite
the substantial increase in wetted surface area is still controversial. For the present
purpose, however, the issue of Reynolds number effects on riblets performance is
more pertinent. Flight tests and transonic tunnel experiments all indicate that the
inner variables are the proper scaling for the dimensions of the riblets, as discussed
in Section 5.7.4.
Chois (1989) spectrum measurements show that the energy of skin-friction
uctuations in the riblet groove drops by a decade compared with that of the smooth
surface over more than a decade of the ow-frequency range. Typically, over time
expressed in wall units of
+
= 170, the skin friction in the groove can remain below
average and at a quiescent state as if the uid in the groove is partially relaminarized.
The dye ow visualization of Gallagher and Thomas (1984) also shows that the
tracer remained quiescent and viscous-pool-like between the ribs, and it leaves the
groove only when a burst passes overhead. Like Gallagher and Thomas, Narasimha
and Liepmann (1988) have also suggested that the riblets create pools of slowviscous
ow in the valleys and thereby modify the interaction of the wall ow with the outer
ow. Black (1968) has analytically described the dynamics of the mean-velocity
prole of a canonical turbulent boundary layer in terms of a periodic competition
between the wall and outer layers whereby the thickness of the sublayer changes with
phase. The maximum thickness of the sublayer that the outer layer will allow can be
obtained from the extrapolation of the log-law to the sublayer prole. Consider the
equality
U
+
= y
+
= 2.41 ln(y
+
) +5.4 (5.35)
This is satised by y
+
= 11, which is nearly the same as the optimized riblet
height.
In a recent paper, Choi, Moin, and Kim (1993) used direct numerical simulations
to study the turbulent ow over a riblet-mounted surface. Quadrant analysis indi-
cates that drag-reducing riblets mitigate the positive Reynolds-shear-stress-producing
events. Choi et al. suggest that riblets with sufciently small spacing reduce viscous
drag by restricting the location of the streamwise vortices above the wetted surface
such that only a limited area of the riblets is exposed to the downwash of high-
speed uid induced by these vortices (see also the corroborating numerical results of
Kravchenko, Choi, and Moin 1993).
5.8 FLOW CONTROL 99
Once it is accepted that the riblet performance is unrelated to streaks, it comes
as no surprise that sand-grain roughness also has a drag-reducing behavior exactly
like riblets. Tani (1987, 1988) has reanalyzed Nikuradses (1933) experimental data
on sand-grain roughness and has shown that the performance of the roughness also
changes from drag reducing to drag increasing with increasing h
+
, where h is the
characteristic roughness height. The skin friction remains lower than that of the
smooth wall for h
+
<6. Compared with the optimized riblets, the drag reduction is
lower in magnitude but is still of the same order. Tani has also suggested that the
mechanism of drag reduction is likely to originate in the nearly quiescent regions of
the ow within the interstices of the roughness elements, as observable deep within
riblets.
Grass (1971) has shown in a channel ow that the inrush and outrush phases of
the production cycle are also present when the wall has a three-dimensional rough-
ness. Note, however, that walls with three-dimensional roughness elements do not
have smooth, wavelike, low-speed streaks, and although the outer-layer structure is
similar to that in a smooth wall, the near-wall stress ux has a different behavior
(Bandyopadhyay and Watson 1988).
5.8.3 Recovery Response
There are at least two modes of interaction between the inner and outer regions of
a boundary layer. In the rst, the outer structures obtain at least part of their energy
by convection and turbulence transport from the inner region of the upstream part
of the boundary layer. This view is supported by the near-constancy of the ratio
between turbulence stress and twice the turbulence kinetic energy, uv/q
2
, across
a major portion of the boundary layer. This is true even in the wake region (i.e.,
y/ 0.2), where the Reynolds stress and the rms velocity uctuations are rapidly
decreasing. According to Townsend (1976), the turbulent uid in that region has
been sheared sufciently long to attain its equilibrium structure. The second mode of
innerouter interaction involves the pressure effects of the inactive motion. Compared
with the convective mode, the pressure mode is much less extended in the streamwise
direction. In other words, the rst mode points toward long memory, whereas the
second is associated with short memory. This may be relevant to the performance of
different control devices, as discussed in the following two subsubsections.
Disturbances in Outer Layer
The long memory associated with the outer structure dependence on upstream
conditions contrasts with the short memory of the inner region. This was demon-
strated by Clauser (1956), who has shown that in a turbulent boundary layer
at a given Reynolds number disturbances survive much longer in the outer layer
(y/ > 0.2) than in the inner layer. He demonstrated this by placing a circular rod
in the outer and inner regions of a fully developed wall layer. For the rod placed
at y/ = 0.16, the decay of the maximum deviation of the distorted mean-velocity
prole from the equilibrium value was reduced to half of its initial value at a down-
streamdistance of 2. In contrast, the outer-layer rod at y/ = 0.6 caused a distortion
in the velocity prole that lasted four times longer, 8, and that did not completely
disappear even at 16 downstream of the rod (see Figure 13 of Clausers article).
Note that Clauser (1956) compared the response of the inner and outer layers at a
low Reynolds number and did not consider any Reynolds number effects. Inciden-
tally, some consider Clausers demonstration as the predecessor of the modern-day
drag-reducing experiments employing modications to the outer layer (Bushnell and
Hefner 1990).
In viscous drag reduction techniques in which a device drag penalty is involved, as
with outer-layer devices (OLD), a recovery length O[100] is desirable to achieve a net
gain. To date, drag reduction has been achieved only at lowReynolds numbers, Re
<
6 10
3
(Anders 1990a). However, when Anders examined his outer-layer devices
at higher Reynolds numbers, to his surprise, the drag reduction, performance was
reduced, and the device was no longer a viable candidate for viscous drag reduction.
Anders experiments in the range of Reynolds numbers of 2500 Re
18, 000
were conducted by towing a slender, axisymmetric body in a water channel. Two
outer-layer devices were used; the rst consisted of two NACA-0009 airfoil-section
rings placed in tandem 1.5 m downstream of the nose of the 3.7-m-long body. The
second device used consisted of two Clark Y low-Reynolds-number airfoil-section
rings, again placed in tandem. Both devices were optimized to yield lowest skin
friction downstream. Although both devices used by Anders (1990a) consistently led
to lower skin friction at all Reynolds numbers tested, net drag will, of course, be
increased by the device drag penalty. This penalty depends on, among other factors,
the thickness and angle of attack of the device; whether the boundary layer on the
device itself is laminar, transitional, or turbulent; and the presence and extent of
any separation bubble that might form on the outer-layer ribbon or airfoil. Anders
(1990a) reported a very modest net drag reduction for his airfoil devices of around 2%
at the lowest Reynolds number but a net drag increase of 15% at higher Reynolds
numbers.
Bandyopadhyay (1986a) used a large-area drag balance to investigate systemat-
ically the Reynolds number effects on both single- and tandem-ribbon devices. His
Reynolds number range of 1300 Re
3600 is lower than that of Anders (1990a),

but the loss-of-performance trends are the same. Note that the drag penalty for the
thin-ribbon devices used by Bandyopadhyay should be far smaller than that for the
airfoil devices used by Anders. We conclude that for both low Re
(<610
3
) and high
Re
(>6 10
3
), the effectiveness of OLD diminishes with the increase of Reynolds
number.
The continued drop in the skin-friction reduction with Reynolds number comes
as a surprise because the mean ow analysis of Coles (1962) indicates an asymp-
totic state of the outer layer to have been reached above Re
> 6 10
3
. The slow
drop in U
+
does not start until Re
>15, 000. Anders (1990a) attributed the ir-

reproducibility of the low-Reynolds-number behavior at higher values to a signif-
icant change in the turbulence structure at higher Re
, as discussed by Head and

Bandyopadhyay (1981). The structural changes as the Reynolds number increases
provide a simple explanation for the performance deterioration of outer-layer de-
vices. These devices presumably work by selectively suppressing the normal velocity
uctuations and thus decorrelating the streamwise and normal velocities. As dis-
cussed earlier, at high Reynolds numbers, fewer hairpin vortices reach the edge of
the boundary layer because of increased interactions among these vortices. The over-
turning motion of the large eddies observed at low Reynolds numbers is less at
higher Reynolds numbers, which reduces the u-turbulence suppression role for the
OLD.
5.8 FLOW CONTROL 101
Disturbances Close to Wall
It is clear from the preceding subsubsection that knowledge of Reynolds number
effects on the mean turbulent ow alone does not allow one to address all prac-
tical problems. This can be further demonstrated in the posttransition unexpected
result described herein. Klebanoff and Diehl (1952) have made measurements on
articially thickened boundary layers at zero pressure gradient. The rst 60 cm of
their splitter plate was covered with No. 16 oor sanding paper. The measurements
were carried out over a length of 320 cm at freestream velocities of 11, 17, and
33 m/s, giving three ranges of Reynolds numbers and producing a maximum Re
of 14,850. The Reynolds number at the end of the sand roughness (at x = 60 cm)
was Re
= 2640, 4050, and 7990 at the three above-mentioned speeds, respectively.

The recovery response of the turbulent boundary layer to the same wall disturbance
(meaning the same sand roughness) at the three different reference Reynolds num-
bers (that is freestream speeds) is characterized by plotting U
+
versus Re
. One
normally expects the recovery from wall disturbances to be the quickest (in x) at
the highest Reynolds numbers. Therefore, it comes as a surprise that, to the con-
trary, the return to the apparent equilibrium state (given by the behavior of U
+
in the canonical boundary layer) is clearly slowed down as the reference Reynolds
number is increased and not decreased! Even at the last station, where x = 320 cm
and x/ = 100, the recovery was not yet complete. The recovery length for an
incoming Re
of about 8 10
3
for the near-wall disturbance case is similar to the
outer-layer disturbance case observed in typical OLD at a similar Reynolds number.
This puzzling behavior leads to the question, Why are the near-wall transition-trip
disturbances surviving even beyond an x/ of 100 at such high Reynolds numbers
as 15 10
3
in a way similar to the behavior of outer-layer devices at much lower
Reynolds numbers? This question is clearly important to model testing in wind tun-
nels and code validation data, where roughness is used to trip and thicken the bound-
ary layer to simulate high Reynolds numbers or ight conditions (Bushnell et al.
1993).
5.8.4 Control of High-Reynolds-Number Flows
The preceding discussion indicates that posttransition memory is longer at higher
Reynolds numbers for certain trips. Wall-layer control may, therefore, have a long-
lasting effect, say O[100], if applied during transition. On the other hand, as per
Clauser (1956), if the wall control is applied in the fully developed turbulent region
of the ow, the effect does not last long. The relevance of Reynolds number effects
to ow control is particularly telling in the case of full numerical simulation because
it is currently limited to Reynolds numbers that are not that far from transitional
values.
It is instructive to recall typical Reynolds numbers encountered in the labora-
tory and in the eld. Other than a handful of large-scale facilities, boundary layers
generated in wind tunnels and water tunnels typically have Reynolds numbers of
Re
= O[1000]. A commercial aircraft traveling at a speed of 300 m/s at an alti-

tude of 10 km would have a unit Reynolds number of 10
7
/m. Owing to the much
smaller kinematic viscosity of water, a nuclear submarine moving at a modest speed
of 10 m/s (20 knots) would have the same unit Reynolds number of 10
7
/m. This
unit Reynolds number translates to a momentum thickness Reynolds number near
the end of either vehicle of roughly Re
= 300,000. The Reynolds number on the

space shuttle is as high as Re
= 430,000, an aircraft carrier can have a maximum of

Re
= 1.510
6
, and in the atmospheric boundary layer the Reynolds number is typ-
ically Re
= 10
6
10
7
. These ranges of Reynolds numbers together with the scopes of
operation of typical wind and water tunnels; direct numerical simulations; and three
large-scale facilities, the National Transonic Tunnel, NASALangley towing tank,
and the super-pipe are schematically shown in Figure 5.1.
It is clear fromthe discussion thus far in this section and fromthe strong Reynolds
number effects on the mean ow, higher-order statistics and coherent structures
demonstrated in Sections 5.55.7 that control devices developed and tested in the
laboratory cannot in general be readily extrapolated to eld conditions. Detailed
knowledge of high-Reynolds-number consequences is required prior to attempting
to control practical wall-bounded ows.
5.9 Summary of Reynolds Number Effects
The classical similarity theory of wall-bounded ows that asserts a universal descrip-
tion for the near-wall ow is found to be increasingly decient as the questions be-
come more detailed. We summarize below the Reynolds number effects on the mean
motion, higher-order statistics and coherent structures. A few items below were not
discussed directly in this chapter but can be found in more detail in the survey by
Gad-el-Hak and Bandyopadhyay (1994).
The widely accepted asymptotic state of the wake component is present

only in the range of 6 10
3
< Re
< 1.5 10
4
. At higher values, it drops,
although at a much slower rate than that in the range of Re
< 6 10
3
.
The Clausers shape parameter C is strongly Reynolds-number-dependent at

Re
< 10
3
and weakly above that.
Alternatives to the logarithmic mean-velocity prole have been periodically

proposed. Such heretical ideas deserve further scrutiny. Independent conr-
mation via well-controlled experiments that cover a wide range of Reynolds
numbers, resolve the linear region, and directly measure the wall-shear stress
is needed.
The freestream turbulence effect is dependent on Reynolds number.
Turbulence measurements with probe lengths greater than the viscous sub-
layer thickness (5 wall units) appear to be unreliable, particularly near the
wall.
Unlike the mean ow, the statistical turbulence quantities do not scale accu-
rately with the wall-layer variables over the entire inner layer. Such scaling
applies over only a very small portion of the inner layer adjacent to the wall.
At low Reynolds numbers, the peak u-turbulence intensity increases slightly

with Reynolds number in both channels and at plates.
The distance from the wall where the streamwise turbulence intensity peaks
appears to scale with inner variables.
In contrast, the corresponding distances, expressed in wall units, for both

the normal uctuations and the Reynolds stress move away from the wall
as the Reynolds number increases. At high Re, the peak normal turbulence
5.9 SUMMARY OF REYNOLDS NUMBER EFFECTS 103
intensity and the peak Reynolds stress occur substantially outside the viscous
region.
The wall-pressure rms increases slightly with Reynolds number.
Systematic changes in the mean and higher-order statistics as the Reynolds

number varies could be considered as proper rst-order trends within the
framework of an asymptotic theory. At nite Reynolds numbers, the additive
composite expansion formed from the inner and outer expansions of any
turbulence quantity provides the only uniformly valid approximation in the
matched region.
In at plates, trip memory can survive the statistical turbulence quantities at

even Re
>6 10
3
, where the mean ow is said to have reached an asymp-
totic state.
The Reynolds number dependence of the posttransition relaxation length of

both the mean and turbulence quantities is not well understood.
In pipe ows, the wave nature of the viscous sublayer, which is observable at
low Reynolds numbers, gives way to a poorly understood random process at
high Reynolds numbers.
Although the variously dened (small) length scales differ greatly from each
other at low Reynolds numbers, they all asymptotically approach the mixing
length at much higher Reynolds numbers (Re
> 10
4
).
The outer-layer structure changes continuously with Reynolds numbers, and

very little is known about the structure of very high-Reynolds-number tur-
bulent boundary layers.
The aspect ratioof the hairpin vortices increases with Reynolds number as
they also become skinnier. In a large structure, the number of constituent
hairpin vortices per unit wall area increases with Reynolds number.
Changes in the wall-bounded owphysics could be described as due to chang-

ing the scale ratio,
+
or a
+
, and not the Reynolds number per se. In a given
boundary layer,
+
changes downstream at a rate slightly lower than Re
.
The inuence of the wall changes from nonlocal to local as this scale ratio
increases.
There is a dire need for high-resolution, reliable measurements of mean and

statistical turbulence moments at high Reynolds numbers in smooth, at-plate
turbulent boundary layers.
Reynolds number effects in canonical ows cannot always be extrapolated

to noncanonical cases in a simple, straightforward manner.
CHAPTER SIX
Transition Control
I have no special talents. I am only passionately curious.
(Albert Einstein, 18791955)
That is the essence of science: ask an impertinent question, and you are on
the way to a pertinent answer.
(Jacob Bronowski, 19081974)
PROLOGUE
Delaying laminar-to-turbulence transition of a boundary layer has many obvious
advantages. Depending on the Reynolds number, the skin-friction drag in the laminar
state can be as much as an order of magnitude less than that in the turbulent condition
(Figure 6.1). For an aircraft or an underwater body, the reduced drag means longer
range, reduced fuel cost and volume, or increased speed. Flow-induced noise results
fromthe pressure uctuations in the turbulent boundary layer and, hence, is virtually
nonexistent in the laminar case. Reducing the boundary layer noise is crucial to the
proper operation of an underwater sonar. On the other hand, turbulence is an efcient
mixer, and rates of mass, momentum, and heat transfer are much lower in the laminar
state; thus, early transition may be sought in some applications as, for example, when
enhanced heat transfer rates are desired in heat exchangers or when rapid mixing
is needed in combustors. This chapter focuses on transition delay, particularly for
wall-bounded ows. Transition advancement will be discussed in Chapter 11.
6.1 Free-Shear versus Wall-Bounded Flows
Free-shear ows, such as jets, wakes, and mixing layers, are characterized by in-
ectional mean-velocity proles and are therefore susceptible to inviscid instabilities.
Viscosity is only a damping inuence in this case, and the prime instability mechanism
is vortical induction. Free-shear ows and separated boundary layers are, therefore,
intrinsically unstable, and their critical Reynolds numbers are typically at very low
values, Re
crit
= O[10]. Fiedler (1998) provides a comprehensive reviewof the control
of free-shear ows, including transition delay and advancement. However, technolog-
ically important free-shear ows will be basically turbulent, and ones efforts should
104
6.2 ROUTES TO TRANSITION 105
Figure 6.1: Skin-friction coefcient for a smooth at plate at zero
incidence to uniform, incompressible, isothermal ow.
be directed toward controlling the resulting random ow to improve mixing, inhibit
shear-layer growth, reduce ow-induced noise, and so forth. We will return to these
topics in Chapters 11 and 12.
External and internal wall-bounded ows, such as boundary layers and chan-
nel ows, can also have inectional velocity proles, but, in the absence of adverse
pressure-gradient and similar effects, are characterized by noninectional proles,
and viscous instabilities are then to be considered. Such viscosity-dominated wall-
bounded ows are intrinsically stable and therefore are generally more difcult to
control. Their critical Reynolds numbers are typically Re
|
crit
= O[10
3
], but transi-
tion can be delayed by pushing this Re to one or two orders of magnitude higher
values using a variety of active and passive control strategies. This is a fruitful ap-
proach for devices having moderate Re values such as aircraft wings and torpedoes.
But, for the fuselage of a large airplane, the Reynolds number becomes so high down-
streamthat, except for limited regions of the forebody, delaying transition becomes an
exercise in futility, and efforts to control the inevitable turbulent ow are then worth
pursuing. The present chapter emphasizes transition delay strategies for boundary
layers. Methods to advance transition as well as enhance turbulence are discussed in
Chapter 11.
6.2 Routes to Transition
The routes to transition are many, but some are more understood than others. Readers
who work in the eld are surely familiar with the instructive, albeit a bit messy, Map
of the Roads to Wall Turbulence drawn and redrawn during numerous presentations
by Mark Morkovin of the Illinois Institute of Technology. (See Figure 22.11 of the
book by Panton 1996 for a tamed reproduction of at least one version of this map.)
On a semi-innite, smooth at plate placed in a clean, uniform, incompressible,
isothermal ow with as little external disturbances as possible, the laminar ow
106 TRANSITION CONTROL
(somewhat downstream of the leading edge) is in the form of a Blasius prole (Fig-
ure 6.2b) and exists at low enough Reynolds number, typically Re
x
<6 10
4
. This
laminar shear layer is, however, unstable to small perturbations that invariably exist
in any ow. Squires (1933) theorem shows that two-dimensional traveling waves
(TollmienSchlichting [TS] waves) are the most dangerous for incompressible-ow
Figure 6.2: Normal proles in a boundary layer for streamwise velocity (left col-
umn), its derivative with respect to y (central column), and its second derivative
(right column). The proles are less full (more inectional) from top to bottom.
(a) Full prole; (b) Blasius prole; (c) Inectional prole; (d) Prole at the point
of steady, two-dimensional separation; (e) Prole with ow reversal.
6.2 ROUTES TO TRANSITION 107
instability and become unstable when the Reynolds number exceeds a critical value.
However, as soon as the TS waves are amplied, gain a certain amplitude, and non-
linear effects take hold, three-dimensional disturbances can no longer be excluded
(Itoh 1987). Following the linear step, the originally two-dimensional waves inher-
ently acquire a nearly periodic spanwise modulation and three-dimensional structures
evolve as a result of a secondary instability. Hairpin vortices develop, perhaps due
to a tertiary instability, and nally breakdown to turbulence occurs (Klebanoff et al.
1962). Not surprisingly, these nonlinear processes are far less understood than the
linear step. The linear amplication step is, however, the slowest of the successive
multiple steps in the transition process and, hence, factors that affect the linear am-
plication determine the magnitude of the transition Reynolds number.
Reshotko (1976, 1985, 1987) asserted that transition is a consequence of the
nonlinear response of the laminar shear layer (a very complicated oscillator) to ran-
dom forcing disturbances that result from freestream turbulence, radiated sound,
surface roughness, surface vibrations, or combination of these environmental fac-
tors. If the initial disturbances are small, transition Reynolds number depends upon
the nature and spectrum of these disturbances, their signature and excitation of the
normal modes in the boundary layer (receptivity; see, for example, Morkovin 1969;
Goldstein and Hultgren 1989), and the linear amplication of the growing normal
modes. Once wave interaction and nonlinear processes set in, transition is quickly
completed. If the initial disturbance levels are large enough, the relatively slow lin-
ear amplication step mentioned above is bypassed (Morkovin 1984, 1988), and
transition can occur at much lower Reynolds numbers. In fact, a sufciently violent
disturbance, u
rms
/U
10%, can cause transition of a laminar boundary layer to

advance to the position upstream at which perturbations of all wave numbers decay
(Klebanoff et al. 1955).
Other routes to transition include the TaylorG ortler vortices forming on a con-
cave surface as a set of counterrotating, streamwise standing eddies (Stuart 1963;
DiPrima and Swinney 1985), the crossow instabilities occurring on a swept wing
or a rotating disk (Reed and Saric 1987, 1989), and any other situation in which an
instability will occur as a result of disturbing the equilibrium of body forces, inertia,
and surface forces (Drazin and Reid 1981). Once again, these instabilities will gen-
erally occur at Reynolds numbers lower than the critical Reynolds number for the
growth of TollmienSchlichting waves.
To delay transition to as far a downstreamposition as possible, the following steps
may be taken. First, because factors that affect the linear amplication of Tollmien
Schlichting waves determine the magnitude of the transition Reynolds number, these
waves may be either inhibited or canceled. In the former method of control, the
growth of the linear disturbance is minimized by using any, or a combination of,
the so-called stability modiers that alter the shape of the velocity prole. These in-
clude increased length of favorable pressure gradient, wall transpiration, wall motion,
surface heating and cooling, wall curvature, and body forces. Wave cancellation of the
growing perturbation is accomplished through exploiting, but not altering, the stabil-
ity characteristics of the ow. Secondly, the forcing disturbances in the environment in
which the laminar shear layer develops may be reduced. This is accomplished by using
smooth surfaces, reducing the freestream turbulence and the radiated sound, mini-
mizing body vibration, and ensuring a particulate-free incoming ow or, in case of
a contaminated environment such as the ocean, using a particle-defense mechanism.
Practically achieved surface smoothness and levels of radiated noise place an upper
limit for the unit Reynolds number required for a successful laminar ow-control sys-
tem. For aircraft, this typically translates into a requirement for high-altitude opera-
tion (above 10 km). Thirdly, one may provide a owwhere other kinds of instabilities
such as TaylorG ortler vortices or crossowinstabilities, will not occur or at least will
not growat a rapid rate. This is done by avoiding as much as possible concave surfaces
or concave streamlines, minimizing the sweep on lifting surfaces, and so forth.
In the next section, the elements of linear stability theory are briey recalled,
again for the benets of newcomers to the eld. This is followed by a discussion of
the rst of the three steps outlined above, stability modiers and wave cancellation
methods. Most of the arguments presented here are for incompressible ows. A
brief comparison between compressible and incompressible shear ows is given in
Section 3.4.
6.3 Linear Stability Theory
The linear stability of a laminar ow is governed by the OrrSommerfeld equation
and appropriate boundary conditions. For an isothermal, incompressible ow the
OrrSommerfeld equation reads
(U c)(
2
) U
i
Re
2
2
+
4
) (6.1)
This fourth-order, linear, ordinary differential equation is derived from the Navier
Stokes equation (2.21) of Chapter 2 by assuming a two-dimensional small distur-
bance superimposed upon a steady, unidirectional mean ow, U(y). The streamfunc-
tion for the perturbation is given by
(x, y, t) = (y)e
i (xct)
(6.2)
In Eqs. (6.1) and (6.2), c is the wave speed, is the amplitude function of the uctu-
ations, the superscript

denotes derivative with respect to y, is the wave number, i
is the imaginary number (i =
1), and Re is the Reynolds number based upon the

freestreamvelocity and the boundary layer thickness. The OrrSommerfeld equation
has been nondimensionalized using the boundary layer thickness as a length scale and
the freestream velocity as a velocity scale. Though linear, this equation is notoriously
difcult to solve analytically.
The order of the OrrSommerfeld equation increases as forces additional to inertia
and viscous ones are incorporated into the momentum balance. For example, in case
of a rotating-disk ow, Coriolis and streamline-curvature terms are needed, leading
to a sixth-order stability equation.
In Eq. (6.1), either the temporal or the spatial growth of instability waves is con-
sidered as an eigenvalue problem. In the former case, a disturbance oscillates in space
but either grows or decays exponentially with time. Acomplex eigenvalue c =c
r
+i c
i
is determined for each pair of values of the real parameter and the Reynolds num-
ber. The real part of c is the phase velocity of the prescribed disturbance, and the sign
of the imaginary part determines whether the wave is temporary amplied (c
i
> 0),
temporary damped (c
i
< 0), or neutrally stable (c
i
= 0).
6.3 LINEAR STABILITY THEORY 109
The more realistic spatial stability problem involves disturbances that oscillate in
time but either grow or decay exponentially with downstream distance. In this case,
a complex eigenvalue =
r
+ i
i
is determined for each pair of values of the real
radian frequency = 2 f = c and the Reynolds number. The real part of is the
wave number, and the sign of
i
determines whether the wave is spatially amplied
(
i
< 0) or spatially damped (
i
> 0).
In either the temporal or spatial instability studies, the major difculty in nu-
merically integrating the OrrSommerfeld equation lies in its being highly stiff and
unstable (Lin 1945), which makes the use of conventional numerical schemes vir-
tually impossible. Explicit numerical methods with a step size commensurate with
the global behavior of the solution cannot be used to integrate Eq. (6.1) because of
the numerical instabilities that characterize this ordinary differential equation. The
papers by Orszag (1971) and Scott and Watts (1977) give examples of special codes
to handle stiff eigenvalue problems.
In deriving Eq. (6.1), the uid viscosity in Eq. (2.21) is assumed constant. If
viscosity varies in space as a result of, for example, surface heating or cooling, ad-
ditional terms containing the rst and second derivatives of viscosity with respect to
y result, and one obtains the so-called modied OrrSommerfeld equation (Wazzan,
Okamura, and Smith 1968)
(U c)(
2
) U
i
Re
[(
2
2
+
4
)
+ 2
) +
+
2
)] (6.3)
where (y) is the uid viscosity nondimensionalized using the freestream value, and
Re is dened using the freestream viscosity. This is the governing equation for study-
ing the stability of, for example, wall-bounded ows with surface heating or cooling.
6.3.1 Critical Re and Transition
The Reynolds number below which (linear) perturbations of all wave numbers
decay is termed the critical Reynolds number, Re
crit
, or the limit of stability. For a
given velocity prole U(y), the critical Reynolds number and the rate of growth of
perturbations depend strongly on the shape of the velocity prole. A prole with an
inectional point (
2
U/y
2
= 0) above the wall provides a necessary and sufcient
condition for inviscid instability (Rayleigh 1880; Tollmien 1935). Such proles must
have a positive curvature at y = 0 because
2
U/y
2
is negative at a large distance
from the wall (Figure 6.2). Even when viscous effects are included (right-hand side
of the OrrSommerfeld equation = 0), a velocity prole becomes more stable as
its second derivative near the wall becomes negative, [
2
U/y
2
]
o
< 0. The prole
is then said to be more full (Figure 6.2a), having a smaller ratio of displacement
thickness to momentum thickness than, for example, an inectional velocity prole
(Figure 6.2ce). In the former case, the critical Reynolds number is increased, the
range of amplied frequencies is diminished, and the amplication rate of unstable
waves is reduced.
In a spatially growing boundary layer, transition to turbulence involves the inte-
grated effect of the growth of unstable waves as they are convected downstream by
the mean ow. If | A| represents the amplitude of a wave, then its local spatial growth
is given by
d| A|
dx
=
i
| A| (6.4)
The net growth of a wave with a given real frequency, from the location x
o
, where it
rst becomes unstable, to the location x
oo
, where transition to turbulence occurs, is
given by
ln
| A|
| A
o
|
x
oo
x
o
i
dx = n (6.5)
Empirically, a modal amplication n of between 8 and 11 is found at transition to
turbulence (e.g., Smith and Gamberoni 1956; Van Ingen 1956; Jaffe, Okamura, and
Smith 1970).
Despite the apparent success of the amplitude-ratio method to predict transition,
Reshotko (1976) maintained that the method is defective in principle and perhaps
also in practice. Transition location depends strongly on the freestream turbulence
levels and other environmental factors. Moreover, nonlinear effects that are physi-
cally signicant in the transition process cannot be accounted for by procedures based
on linear stability theory. In general, the e
n
-method works fairly well for situations
in which the freestream turbulence level is below 1% (e.g., cruising aircraft), but the
strategy fails for noisy environments (e.g., turbine blades). Transition processes dom-
inated by TS instabilities can accurately be predicted using the e
n
-method, whereas
transition caused by crossow instabilities and the like cannot be forecast. Notwith-
standing these drawbacks, the e
n
-rule provides a practical scheme for estimating the
effectiveness of a transition-delay method of control (Bushnell and McGinley 1989;
Bushnell 1989).
6.4 Stability Modiers
Stability modiers are those methods of laminar ow control that alter the shape of
the velocity prole to minimize the linear growth of unstable waves. For a boundary
layer ow along a nonmoving wall
1
of small curvature, the three momentum equa-
tions at the surface were listed in Chapter 2. Herein, we allow both streamwise and
normal motions of the wall, and the instantaneous streamwise momentum equation
at y = 0 reads
w
u
t
y=0
+
w
u
w
u
x
y=0
+
w
v
w
u
y
y=0
+
p
x
y=0
y=0
u
y
y=0
=
2
u
y
2
y=0
(6.6)
Equation (6.6) is valid for a uid with variable properties and . For a two-
dimensional laminar boundary layer, vorticity is only in the spanwise direction and
is given by
z
= u/y. The right-hand side of Eq. (6.6), therefore, represents the
1
Wall was not allowed to move in the streamwise or spanwise directions. Normal wall motions are
analogous to wall transpiration, v
w
, and were, therefore, included in Eqs. (2.39)(2.41).
6.4 STABILITY MODIFIERS 111
ux of vorticity at the surface, as demonstrated by Lighthill (1963a). Any of the
terms on the left-hand side of Eq. (6.6) can affect the sign of the second derivative of
the velocity prole (or the direction of the vorticity ux) at the wall and, hence, both
the ow stability and its resistance to separation. Stability modiers do just that and
include wall motion such that the rst, second, or third term on the left-hand side of
Eq. (6.6) is negative, wall suction (v
w
< 0), favorable pressure gradient (p/x < 0),
or lower wall viscosity (/y > 0). Any one or a combination of these methods
will cause the curvature of the velocity prole at the wall to become more negative
and, hence, increase the lower critical Reynolds number and reduce the spatial or the
temporal amplication rates of unstable waves. Curvature effects and body forces
were not included in Eq. (6.6) but can also be very effective in advancing or delaying
transition. A discussion of Lorentz forces
2
is deferred to Chapter 14.
For historical reasons, boundary layers, which are stabilized by extending the
region of favorable pressure gradient, are known as natural laminar ow (NLF),
whereas other methods to modify the stability of a shear ow are termed laminar
ow control (LFC). It is clear from Eq. (6.6) that the effects of all those strategies are
additive. The term hybrid laminar ow control normally refers to the combination
of NLF and one of the LFC techniques. The recent review article by Joslin (1998)
discusses NLF and LFC specically as applied to aircraft.
The most common method to lower the viscosity near the surface is to heat the wall
for liquids (d/dT < 0) or to cool it for gases (d/dT > 0). Other strategies to effect
a near-wall viscosity gradient include surface-lm boiling, cavitation, sublimation,
chemical reaction, wall injection of lower or higher viscosity uid, and the presence
of shear thinning or thickening additive. The following four subsections describe in
turn the four stability modiers to delay the transition of a boundary layer: wall
motion, suction, shaping, and wall heating or cooling.
6.4.1 Wall Motion
Wall motion can be generated by either actively driving the surface or by using a
exible coating whose modulus of rigidity is low enough so that surface waves are
generated under the inuence of the stress eld in the uid. In the former case, the
wall motion is precisely controlled and can be made to affect the shape of the velocity
prole in a desired manner. For example, a transverse or longitudinal wave can be
forced to propagate on the surface of the solid and thus modulate the instantaneous
values of the rst three terms on the left-hand side of Eq. (6.6). This method of control
is, however, quite impractical and is used mainly to provide controlled experiments to
determine what type of wave motion is required to achieve a given result. The more
practical passive exible walls can be broadly classied into two categories: truly
compliant coatings that have extremely small damping and modulus of elasticity
and can therefore respond with little phase lag to the boundary layer ow, and the
more readily available resonant walls in which vibration modes in the solid, acting
as an active vibration damper, are excited by the ow-disturbance forcing function
(Bushnell, Hefner, and Ash 1977). The owstabilization in this case may be a result of
altering the phase relation between the instantaneous streamwise and normal velocity
components (i.e., the Reynolds stress uv) in the viscous region rather than changing
the curvature of the mean velocity prole at the wall.
2
Body forces that result when a conducting uid moves in the presence of a magnetic eld.
Theoretical studies of boundary-layer stability in the presence of a exible wall
started in the early 1960s, stimulated in part by the pioneering experimental work
of Kramer (1960a). In both the classical work (e.g., Benjamin 1960a, Landahl 1962;
Kaplan 1964) and the more recent research (e.g., Carpenter and Garrad 1985; Willis
1986; Yeo and Dowling 1987), steady-state stability theory has been used with an as-
sumed velocity prole that is not allowed to change as the wall moves. It is clear, how-
ever, that the wall displacement will induce a traveling pressure signal that, in turn,
will modulate the mean velocity prole. Either a quasi-steady stability analysis of the
modulated ow or true time-dependent calculations would be preferable to conven-
tional linear-stability theory, but neither has been attempted yet owing to the obvious
complexities of the problem. Notwithstanding the shortcoming of present stability
calculations, Willis (1986) obtained a very impressive agreement between linear the-
ory and a carefully conducted experiment. Eigenvalue calculations were performed to
predict the amplication factors for a range of modal frequencies. The exible coating
was a silicon-rubber and silicon-oil mix covered by a thin latex rubber skin stretched
across the surface. The experiments were conducted in a water towing tank using a at
plate, and controlled, harmonic, two-dimensional disturbances were introduced up-
streamof the compliant surface. A reduction of the wave growth by an order of mag-
nitude is feasible when this rather simple coating is used, almost eliminating transition
due to TollmienSchlichting type of instability. The exible wall itself is, however,
susceptible to other kinds of instability, and care must be taken to ensure that these
surface waves will not grow to an amplitude that will promote transition through a
roughness-like effect (see the reviewarticle by Riley, Gad-el-Hak, and Metcalfe 1988).
Although the original Kramers (1960a) experiments were discredited up until a
few years ago, more recent theoretical and experimental evidence conrms Kramers
results (Carpenter and Garrad 1985). Passive exible coatings with density the order
of the uid density appear to be capable of considerable transition postponement. All
of Chapter 7 is devoted to this simple control method that does not require energy
expenditure, slots, ducts, or internal equipment of any kind.
6.4.2 Suction
A second method for postponing transition is the application of wall suction. As
seen from Eq. (6.6), small amounts of uid withdrawn from the near-wall region of
the boundary layer change the curvature of the velocity prole at the wall and can
dramatically alter the stability characteristics of the boundary layer. Additionally,
suction inhibits the growth of the boundary layer, and thus the critical Reynolds
number based on thickness may never be reached.
Although laminar ow can be maintained to extremely high Reynolds numbers
provided that enough uid is sucked away, the goal is to accomplish transition delay
with the minimum suction ow rate. Not only will this reduce the power necessary
to drive the suction pump but also the momentum loss due to suction; hence, the
skin friction, is minimized. This latter point can easily be seen from the momentum
integral equation. If one rewrites the K arm an integral equation (2.44) for a steady,
incompressible ow ( =
o
= constant) over a at plate

dU
o
/dx = 0; dR/dx = 0
with uniform suction through the wall (v

w
negative), the equation reads
C
f
2
=
d
dx
+
|v
w
|
U
o
(6.7)
The second term on the right-hand side is the suction coefcient, C
q
, and although
withdrawing the uid through the wall leads to a decrease in the rate of growth of the
momentum thickness, C
f
increases directly with C
q
. Fluid withdrawn through the
wall has to come from outside the boundary layer where the streamwise momentum
per unit mass is at the relatively high level of U
. The second term is proportional

to the rate of momentum loss due to withdrawing a mass per unit time and area of
|v
w
|. Note that this term does not exist for pipe ows because of the mass-ow
constraint. Hence, this momentum penalty is not paid for channel ows with wall
transpiration, which is an important distinction between internal and external ows.
Although Prandtl (1904) used suction to prevent owseparation fromthe surface
of a cylinder near the beginning of this century, the rst experimental demonstration
that boundary-layer transition can be delayed by withdrawing near-wall uid did
not take place until about four decades later. Holstein (1940); Ackeret, Ras, and
Pfenninger (1941); Ras and Ackeret (1941); and Pfenninger (1946) used carefully
shaped, single- and multiple-suction slits to demonstrate the decrease in drag associ-
ated with delaying transition. Braslow et al. (1951) used continuous suction through
a porous wall to maintain laminar ow on an airfoil to a chord Reynolds number of
2.0 10
7
. Raspet (1952) conducted unique, noise-free experiments that conrmed
the large decrease in drag when suction is applied through the wings of a sailplane.
In the early 1960s, test ights of two X21 aircraft (modied U.S. Air Force B66)
indicated the feasibility of maintaining a laminar ow on a swept wing to chord
Reynolds numbers as high as 4.710
7
(Whites, Sudderth, and Wheldon 1966). The
wing surfaces contained many thin and closely spaced spanwise suction slots, and
the total airplane drag was reduced by 20% as compared with the no suction case.
Although discrete suction slots were used rst (because of the unavailability of
suitable porous surfaces in the early 1940s), the theoretical treatment of the problem
is considerably simplied by assuming continuous suction through a porous wall
where the characteristic pore size is much smaller than a boundary layer thickness.
In fact, the case of a uniform suction from a at plate at zero incidence is an exact
solution of the incompressible NavierStokes equation (2.21). If one assumes suction
weak enough that the potential ow outside the boundary layer is unaffected by the
loss of mass at the wall (sink effects), the asymptotic velocity prole in the viscous
region is exponential and has a negative curvature at the wall
U(y) = U
1 exp
|v
w
| y
(6.8)
The displacement thickness has the constant value
= /|v
w
|, where is the kine-
matic viscosity and |v
w
| is the absolute value of the normal velocity at the wall. In
this case, Eq. (6.7) reads
C
f
= 2C
q
(6.9)
Bussmann and M unz (1942) computed the critical Reynolds number for the preceding
asymptotic velocity prole to be Re
|
crit
U
/ = 70,000. From the value of
given above, the owis stable to all small disturbances if C

q
|v
w
|/U
>1.4 10
5
.
The amplication rate of unstable disturbances for the asymptotic prole is an order
of magnitude less than that for the Blasius boundary layer (Pretsch 1942). This
treatment ignores the development distance fromthe leading edge needed to reach the
asymptotic state. When this is included in the computation, a higher C
q
(1.18 10
4
)
is required to ensure stability (Iglisch 1944; Ulrich 1944). Wuest (1961) presented a
summary of transpiration boundary layer computations up to the early 1960s.
The more complicated analysis for the stability of a boundary layer with suc-
tion through discrete spanwise strips was only carried out satisfactorily relatively
recently. Reed and Nayfeh (1986) conducted a numerical-perturbation analysis of
a linearized, triple-deck, closed-form basic state of a at plate boundary layer with
suction through a nite number of spanwise porous strips. Their results were com-
pared with interacting boundary-layer calculations (Ragab and Nayfeh 1980) as well
as with the carefully conducted experiments of Reynolds and Saric (1986). Suction
applied through discrete strips can be as effective as suction applied continuously
over a much longer streamwise length. Reed and Nayfeh (1986) suggested a scheme
for optimizing the strip conguration. Their results showed that suction should be
concentrated nearer the leading edge (branch I of the neutral stability curve) when
disturbances are still small in amplitude.
Suction may be applied through porous surfaces, perforated plates, or carefully
machined slots. It is of course structurally impossible to make the whole surface of
an aircraft wing or the like out of porous material, and often strips of sintered bronze
or steel are used. A relatively inexpensive woven stainless steel, Dynapore, is now
available and provides some structural support (Reynolds and Saric 1986). Superior
surface smoothness and rigidity are obtained by drilling microholes in titaniumusing
the recently developed electron-beam technology. The lower requirement for a pres-
sure drop in the case of a perforated plate translates directly into pumping-power
saving. However, outowproblems may result fromregions of the wing having strong
pressure gradients (Saric and Reed 1986). Outow in the aft region of a suction strip
can cause large destabilizing effects and local three-dimensionality.
Although structurally a surface with multiple slits is more rigid than a porous
surface, slots are more expensive to fabricate accurately. Moreover, the higher mass-
ow rates associated with them may result in high-Reynolds-number instabilities
such as separation and backow, which adversely affect the stability of the basic
ow. The rule of thumb is that the Reynolds number, based on slot width (or hole
diameter in the case of a perforated plate) and the local suction velocity, should be
kept below 10 to avoid adverse effects on the boundary-layer stability. Nevertheless,
Saric and Reed (1986) claimed a hole Reynolds number an order of magnitude higher
than that without destabilization of the basic ow.
Delaying transition using suction is a mature technology, in which most of the
remaining problems are in the maintainability and reliability of suction surfaces and
the optimization of suction rate and distribution. To protect the delicate suction
surfaces on the wing of an aircraft from insect impacts and ice formation at low
altitudes, special leading edge systems are used (Wagner and Fischer 1984; Wagner
et al. 1984, 1988, 1990). Suction is less suited for underwater vehicles because of the
abundance of suspended ocean particulate that can clog the suction surface as well
as destabilize the boundary layer.
6.4.3 Shaping
The third method of control to delay laminar-to-turbulence transition is perhaps
the simplest and involves the use of suitably shaped bodies to manipulate the pres-
sure distribution. In Eq. (6.6), the pressure gradient term can affect the sign of the
curvature of the velocity prole at the wall and, hence, change the stability character-
istics of the boundary layer. According to the calculations of Schlichting and Ulrich
(1940), the critical Reynolds number based on displacement thickness and freestream
velocity changes from about 100 to 10,000 as a suitably nondimensionalized pres-
sure gradient varies from = 6 (adverse) to = +6 (favorable). The Pohlhausen
parameter, , is the ratio of pressure forces to viscous forces and is given by

dp
dx

2
U
o
=

2
dU
o
dx
(6.10)
where is the boundary layer thickness, U
o
is the velocity outside the boundary layer,
is the kinematic viscosity, is the dynamic viscosity, and dp/dx is the pressure
gradient.
For the case of a favorable pressure gradient, no unstable waves exist at innite
Reynolds number. In contrast, the upper branch of the neutral stability curve in the
case of an adverse pressure distribution tends to a nonzero asymptote so that a nite
region of wavelengths at which disturbances are always amplied remains even as
Re .
Streamlining a body to prevent separation and reduce form drag is quite an old
art going back to prehistoric times, as indicated in Chapter 1, but the stabilization
of a boundary layer by pushing the longitudinal location of the pressure minimum
as far back as possible dates to the 1930s and led to the successful development
of the NACA 6-Series NLF airfoils. Newer, low-Reynolds-number lifting surfaces
used in sailplanes, low-speed drones, and executive business jets have their maxi-
mum thickness point far aft of the leading edge. The recent success of the Voyagers
9-day, unrefueled ight around the world was due in part to a wing design employing
natural laminar ow to approximately 50% chord. Application of NLF technology
to underwater vehicles is feasible but somewhat more limited (Granville 1979).
For a lifting surface, the favorable pressure gradient extends to the longitudinal
location of the pressure minimum. Beyond this point, the adverse pressure gradient
becomes steeper and steeper as the peak suction is moved farther aft. For an airfoil, the
desired shift in the point of minimumpressure can only be attained in a certain narrow
range of angles of incidence. Depending on the shape, angle of attack, Reynolds
number, surface roughness, and other factors, the boundary layer either becomes
turbulent shortly after the point of minimum pressure or separates rst and then
undergoes transition. One of the design goals of NLF is to maintain attached ow
in the adverse pressure gradient region, and some method of separation control
(Chapter 8) may have to be used there.
Factors that limit the utility of NLF include crossow instabilities and leading
edge contamination on swept wings, insect and other particulate debris, high unit
Reynolds numbers at lower cruise altitudes, and performance degradation at higher
angles of attack due to the necessarily small leading edge radius of NLF airfoils.
Reductions of surface waviness and smoothness of modern production wings, spe-
cial leading edge systems to prevent insect impacts and ice formation, higher cruise
altitudes of newer airplanes, and higher Mach numbers all favor the application of
NLF (Runyan and Steers 1980). To paraphrase an older statement by Holmes (1988),
an NLF airfoil is no longer as nicky as Morris the cat. It is true that a boundary
layer that is kept laminar to extremely high Reynolds numbers is very sensitive to
environmental factors such as roughness, freestreamturbulence, radiated sound, and
so forth. However, the ow is durable and reliable within certain conservative design
corridors that must be maintained by the skillful designer and eventual operator of
the vehicle. Current research concentrates on understanding the achievability and
maintainability of natural laminar ow, expanding the practical applications of NLF
technology, and extending the design methodology to supersonic aviation (Bushnell
and Malik 1988; Bushnell 1989; Joslin 1998; Becker, Durst, and Lienhart 1999).
6.4.4 Wall Heating and Cooling
The last of the stability modiers to be considered in this chapter is the addition
or removal of heat from a surface, which causes the viscosity to vary with distance
from the wall. In general, viscosity increases with temperature for gases, whereas
the opposite is true for liquids. Thus, if heat is removed from the surface of a body
moving in air, the fth term on the left-hand side of Eq. (6.6) is negative. In that case,
the velocity gradient near the wall increases and the velocity prole becomes fuller
and more stable. The term containing the viscosity derivative will also be negative if
the surface of a body moving in water is heated. With heating in water or cooling in
air, the critical Reynolds number is increased, the range of amplied frequencies is
diminished, and the amplication rate of unstable waves is reduced. Substantial delay
of transition is feasible with a surface that is only a few degrees hotter (in water) or
colder (in air) than the freestream. Note that for gases in particular, surface cooling
affects both viscosity and density. The increased near-wall density has additional
benecial effects via the stable buoyancy (for certain orientations) and enhanced
near-wall momentum it induces.
The rst indirect evidence of this phenomenon was the observation that the drag
of a at plate placed in a wind tunnel increases by a large amount when the plate
is heated (Linke 1942). Both Frick and McCullough (1942) and Liepmann and Fila
(1947) showed that the transition location of a at-plate boundary layer in air at
low subsonic speeds is moved forward as a result of surface heating. The stability
calculations of Lees (1947) conrmed these experiments and, moreover, showed that
cooling has the expected opposite effects. The critical Reynolds number based on
distance from the leading edge increases from 10
5
to 10
7
when the wall of a at plate
placed in an air stream is cooled to 70% of the absolute ambient temperature. Even
a modest cooling of the wall to 0.95T
results in doubling of the critical Reynolds

number (Kachanov, Koslov, and Levchenko 1974).
With cooling, the range of amplied frequencies is diminished and the growth rate
of TS waves is reduced, resulting in a substantial increase in transition Reynolds
number. These same trends were dramatically conrmed in subsonic and supersonic
ights
3
by Dougherty and Fisher (1980), who studied the transition on an airborne
cone over the Mach number range of 0.552.0. They reported a transition Reynolds
number that varied approximately as T
7
w
, where T
w
is the wall temperature. For
aircraft, this method of transition delay is feasible only for a vehicle that uses a
cryofuel such as liquid hydrogen or liquid methane. In that case, a sizable heat sink is
readily available, the idea being that the fuel is used to cool the major aerodynamic
surfaces of the aircraft as it ows from the fuel tanks to the engines. Reshotko
3
In hypersonic ows, a different mode of stability, Macks second mode, dominates the transition process
and cooling and in fact promotes earlier transition to turbulence.
(1979) examined the prospects for the method and concluded that, particularly for
a hydrogen-fueled aircraft, substantial drag reductions are feasible. His engineering
calculations indicated that the weight of the fuel saved is well in excess of the weight
of the required cooling system.
The preceding effects are more pronounced in water ows owing to the larger
Prandtl number (good thermal coupling) and the stronger dependence of viscosity
on temperature.
4
In a typical low-speed situation, a surface heating of 1
C in water
has approximately the same effect on the curvature of the velocity prole at the wall
as a surface cooling of 20
C in air or a suction coefcient (in air or water) of 0.0003

(Liepmann, Brown, and Nosenchuck 1982). Wazzan et al. (1968, 1970) used the
modied fourth-order OrrSommerfeld equation (6.3) combined with the e
9
-method
of Smith and Gamberoni (1956) and conrmed that wall heating can produce large
increases in the transition Reynolds number of water-boundary layers. They predicted
a transition Reynolds number, based on freestream velocity and distance from the
leading edge of a at plate, as high as 210
8
for wall temperatures that are only
40
C above the ambient water temperature.

Lowell and Reshotko (1974) rened Wazzan et al.s (1968, 1970) calculations
by introducing a coupled sixth-order system of vorticity and energy disturbance
equations. The predicted critical Reynolds numbers for wall overheats of up to 2.8
C
were conrmed by the experiments of Strazisar, Reshotko, and Prahl (1977), who
measured the growth rates of small disturbances generated by a vibrating ribbon in
a heated at plate in water. These experiments did not yield data on transition or on
stability at higher overheats.
The transition predictions of Wazzan et al. (1968, 1970) at higher overheats were
partially conrmed by the very carefully conducted experiments of Barker and Gile
(1981), who used the entrance region of an electrically heated pipe. The displace-
ment thickness was much smaller than the pipe radius and, thus, the boundary-layer
development was approximately the same as that of a zero-pressure gradient at
plate. Barker and Gile reported a transition Reynolds number of 4.710
7
for a
wall overheat of 8
C. No further increase in (Re|

transition
) was observed as the wall
was heated further, which is in contradiction to the computations of Wazzan et al.
(1968, 1970). Barker and Gile (1981) investigated possible causes of this discrepancy,
including buoyancy effects, wall roughness, effects of geometry, ow asymmetries,
and suspended particulate matter. Their analysis and numerous related work (e.g.,
Kosecoff, Ko, and Merkle 1976; Chen, Goland, and Reshotko 1979; Hendricks and
Ladd 1983; Lauchle and Gurney 1984) indicate that increased concentration and
size of suspended particulate diminish the stabilizing effect of surface heating until
at some point surface heating no longer stabilizes the boundary layer but is in fact a
destabilizing inuence.
On a heated body of revolution in a high-speed water tunnel, Lauchle and Gurney
(1984) observed an increase in transition Reynolds number from4.510
6
to 3.610
7
for an average overheat of 25
C. Clearly, surface heating in water can be an extremely

effective method of transition delay, which accounts for drag reduction for small,
high-speed underwater vehicles in which the rejected heat from their propulsion
4
For water at room temperature, Pr 7, and absolute viscosity is decreased by approximately 2% for
each 1
C rise in temperature. For room temperature air, Pr 0.7, and absolute viscosity is decreased
by approximately 0.2% for each 1
C drop in temperature.
system can be used to increase the surface temperature along the body length. The
detrimental effects of freestream particulate alluded to earlier are, however, a ma-
jor obstacle at present for a practical implementation of this method of control.
Suspended particulate having wide-band concentration spectra are abundant in the
oceans, and particle-defense mechanisms must be sought before using any of the
transition-delay methods in a contaminated environment.
In addition to surface heating, several other techniques are available to lower the
near-wall viscosity (/y > 0) in a liquid boundary layer and thus favorably affect
the stability of the ow. These include, as mentioned earlier, lm boiling, cavitation,
sublimation, chemical reaction, or wall injection of a gas or lower-viscosity liquid.
Finally, a shear-thinning additive can be introduced into the boundary layer. Because
the shear increases as the wall is approached, the effective viscosity of the non-
Newtonian uid decreases there, and (/y) becomes positive.
6.5 Wave Cancellation
An alternative approach to the four stability modiers used for increasing the tran-
sition Reynolds number of a laminar boundary layer is wave cancellation. If the
frequency, orientation, and phase angle of the dominant element of the spectrum of
growing linear disturbances in the boundary layer are detected, a control system and
appropriately located disturbance generators may then be used to effect a desired
cancellation or suppression of the detected disturbances. In this case, the stability
characteristics of the boundary layer are exploited but not altered (Reshotko 1985).
Wave cancellation is feasible only when the disturbances are still relatively small,
their growth is governed by a linear equation, and the principle of superposition is
still valid.
The rst reported use of wave cancellation is that due to Schilz (1965, 1966).
He used a vibrating ribbon to excite a TS wave on a test plate that had a exible
surface. A unique wall-motion device ush mounted into the plate moved the exible
wall in a transverse, wavelike manner with a variety of frequencies and phase speeds.
A signicant amount of cancellation resulted when the exible wall motion had the
opposite phase but the same frequency and phase speed as the TS wave. Both Milling
(1981) and Thomas (1983) used two vibrating wires, one downstreamof the other, to
generate and later cancel a single frequency TS wave. Thomas (1983) observed that
interaction between the primary disturbance and background excitations prevented
complete cancellation of the primary wave. To further study the consequences of
wave interactions, Thomas applied the same method of control to eliminate two
interacting waves of different frequency. Although the primary waves were behaving
linearly, a nonlinear interaction gave rise to a low-amplitude difference frequency
that could only be partially reduced and ultimately led to transition. Thomas (1983)
concluded that it is not possible to return the ow completely to its undisturbed base
state because of wave interactions and that it is perhaps more appropriate to describe
this control method as wave superposition rather than wave cancellation.
The same principle of wave superposition can be applied using wall heating
and cooling (Liepmann et al. 1982; Liepmann and Nosenchuck 1982; Ladd and
Hendricks 1988), plate vibration (Gedney 1983), compliant wall (McMurray,
Metcalfe, and Riley 1983), or periodic suction and blowing (Biringen 1984).
6.5 WAVE CANCELLATION 119
Figure 6.3: Cancellation of articial instability waves using adaptive heat
[after Ladd and Hendricks 1988].
Liepmann and Nosenchuck (1982) used ush-mounted hot-lm probes to sense nat-
ural TS waves in a at-plate boundary layer in a water tunnel. A feed-forward
control loop was then used to synthesize and introduce disturbances of equal am-
plitude but of opposite phase via ush-mounted wall heaters. Ladd and Hendricks
(1988) performed their experiment in a water tunnel on a 9:1 neness-ratio ellipsoid.
Strip heaters were again used to create and actively attenuate TS waves. Ladd and
Hendricks applied digital ltering techniques to synthesize the attenuation signal.
The lter was able to adapt the attenuation signal actively to changes in amplitude
and frequency of the articially introduced instability wave with no loss in attenua-
tion downstream. A sample of Ladd and Hendricks results is depicted in Figure 6.3.
Time records of a ush-mounted hot-lm probe are shown for the natural ow, for
the articially introduced TS wave using a heating ring, and for the successfully
canceled wave using a second downstream heating ring.
The transition delay achieved by active wave cancellation is modest (typically a
factor of two or less increase in the transition Reynolds number based on distance
from the leading edge). Reshotko (1985) maintained that to achieve signicant delay
in transition using this technique would require an extensive array of disturbance de-
tectors and generators as well as prohibitively complicated control system that could
cancel both the primary and residual disturbance spectra. Signicant delay in transi-
tion is more readily achieved via the stability modiers summarized in Section 6.4.
CHAPTER SEVEN
Compliant Coatings
There is no greater impediment to progress in the sciences than the desire
to see it take place too quickly.
(George Christoph Lichtenberg, 17421799)
There is a river in the ocean: in the severest droughts it never fails, and in
the mightiest oods it never overows; its banks and its bottom are of
cold water, while its current is of warm; the Gulf of Mexico is its fountain,
and its mouth is the Arctic Seas. It is the Gulf stream. There is in the
world no other such majestic ow of waters.
(Matthew Fontaine Maury, 18061873)
PROLOGUE
Boundary layer manipulation via reactive control strategies is now in vogue. The
payoffs are handsome, but the difculties involved are daunting. This topic is de-
ferred to the last chapter of the book. There are, however, much simpler alternatives
to such sophisticated ow alteration devices, and the present chapter discusses one
such alternative: passive compliant walls. We particularly reviewthe important devel-
opments in the eld of compliant coatings that took place during the past decade or
so. During this period, progress in theoretical and computational methods somewhat
outpaced that in experimental efforts. There is no doubt that compliant coatings can
be rationally designed to delay transition and to suppress noise on marine vehicles
as well as other practical hydrodynamic devices. Transition Reynolds numbers that
exceed by an order of magnitude those on rigid-surface boundary layers can read-
ily be achieved. Recent theoretical work indicates that transition to turbulence can
be delayed indenitely, at least in principle, provided that optimized multiple-panel
compliant walls are used. There is renewed experimental evidence of favorable in-
teractions of compliant coatings even for air ows and even for turbulent boundary
layers, but more research is needed to conrm these latest results.
7.1 Introduction
A compliant wall, as opposed to a rigid one, offers the potential for favorable inter-
ference with a wall-bounded ow. Laminar-to-turbulence transition may be delayed
120
7.1 INTRODUCTION 121
or advanced, boundary layer separation may be prevented or triggered, ow-induced
noise may be modulated, and skin-friction drag in both laminar and turbulent ows
may be altered. The challenge is of course to nd a coating with the right physical
properties to achieve a desired goal.
Passive compliant coatings were around long before reactive ow control was
even contemplated. For better or worse, hydrodynamically speaking, the epidermis
of most nekton is pliable at their typical swimming speeds. For close to half a cen-
tury the science and technology of compliant coatings has fascinated, frustrated,
and occasionally gratied Homo sapiens searching for methods to delay laminar-to-
turbulence transition, to reduce skin-friction drag in turbulent wall-bounded ows,
to quell vibrations, and to suppress ow-induced noise. Compliant coatings offer a
rather simple method to delay laminar-to-turbulence transition as well as to interact
favorably with a turbulent wall-bounded ow. In its simplest form, the technique is
passive, relatively easy to apply to an existing vehicle or device, and perhaps not too
expensive. Unlike other drag-reducing techniques such as suction, injection, poly-
mer or particle additives, passive compliant coatings do not require slots, ducts, or
internal equipment of any kind. Aside from reducing drag, other reasons for the
perennial interest in studying compliant coatings are their many other useful ap-
plications, for example, as sound-absorbent materials in noisy ow-carrying ducts
in aero-engines and as exible surfaces to coat naval vessels for the purposes of
shielding their sonar arrays from the sound generated by the boundary-layer pres-
sure uctuations and of reducing the efciency of their vibrating metal hulls as sound
radiators.
The original interest in the eld was spurred by the experiments of Kramer
(1957), who demonstrated a compliant coating design based on dolphins epider-
mis and claimed substantial transition delay and drag reduction in hydrodynamic
ows. Those experiments were conducted in the seemingly less-than-ideal environ-
ment of Long Beach Harbor, California. Subsequent laboratory attempts to sub-
stantiate Kramers results failed, and the initial interest in the idea zzled. A similar
bout of excitement and frustration that dealt mostly with the reduction of skin-
friction drag in turbulent ows for aeronautical applications followed. Those results
were summarized in the comprehensive review by Bushnell et al. (1977). During the
early 1980s, interest in the subject was rejuvenated mostly as the result of modest
investment in resources by the Ofce of Naval Research in the United States and
the Procurement Executive of the Ministry of Defence in Great Britain.
1
Signi-
cant advances were made during this period in numerical and analytical methods to
solve the coupled uid-structure problem. New experimental tools were developed
to measure the minute yet important surface deformation caused by the unsteady
uid forces. Coherent structures in turbulent wall-bounded ows were routinely
identied, and their modulation by the surface compliance could readily be quanti-
ed.
Careful analyses by Carpenter and Garrad (1985) and Willis (1986) as well
as the well-controlled experiments reported by Daniel, Gaster, and Willis (1987)
1
Through all the ups and downs in the West, compliant coating research continued at more or less steady
pace in the former Soviet Union, but open-literature publications resulting from this work, either in
English or in Russian, are rather scarce. For some valuable references, see, for example, the books by
Aleyev (1977), and Choi (1991), and the article by Carpenter and Garrad (1985).
122 COMPLIANT COATINGS
and Gaster (1988)
2
have, for the rst time, provided direct conrmation of the
transition-delaying potential of compliant coatings, convincingly made a case for
the validity of Kramers original claims, and offered a plausible explanation for the
failure of the subsequent laboratory experiments. There is little doubt now that com-
pliant coatings can be rationally designed to delay transition and to suppress noise
on marine vehicles and other practical hydrodynamic devices. Transition Reynolds
numbers that exceed by an order of magnitude those on rigid-surface boundary
layers can readily be achieved. Although the number of active researchers in the
eld continues to dwindle, new and promising results are being produced. Recent
theoretical work by Davies and Carpenter (1997a) and Carpenter (1998) indicates
that transition to turbulence can be delayed indenitely, at least in principle, provided
that optimized multiple-panel compliant walls are used and that the freestream is a
low-disturbance environment. There is also recent evidence of favorable interactions
of compliant coatings even for air ows (Lee, Fisher, and Schwarz 1995) and even
for turbulent boundary layers (Lee, Fisher, and Schwarz 1993a; Choi et al. 1997).
This chapter emphasizes the signicant compliant coating research that took place
during the last 1015 years and suggests avenues for future research. The reader is
referred to prior reviews for more classical work on the subject such as those by
Bushnell et al. (1977), Gad-el-Hak (1986a, 1987a), Riley et al. (1988), Carpenter
(1990), and Metcalfe (1994). Following these introductory remarks, a somewhat
sketchy history of the subject, particularly prior to 1985, is recalled. This will help
place more recent developments in proper perspective.
7.2 Compliant Coatings prior to 1985
Before embarking on describing the recent accomplishments in the eld of compliant
coatings, we rst elaborate on its history before 1985. This seemingly arbitrary date
is chosen because it demarks the time after which tools for rationally designing a
compliant coating to delay transition became more readily available. The victories
and defeats of the subject matter will become clear through the discussion that fol-
lows. The idea of using compliant coatings for drag reduction motivated much of
the earlier work in this area and was rst introduced by Kramer (1957) based on his
earlier observation, while crossing the Atlantic ocean in 1946, of dolphins swimming
in water. He advanced the concept that the stability and transition characteristics of
a boundary layer may be inuenced by coupling it hydroelastically to a compliant
coating. In his pioneering paper and several subsequent publications, Kramer (1960a,
b,c; 1961; 1962; 1965; 1969) reported substantial drag reduction for towed under-
water bodies covered with compliant coating modeled after the dolphin skin. He
hypothesized that by tuning the elastic wall damping to a frequency near that of the
most unstable TollmienSchlichting wave, it would be possible to dissipate partially
the instability waves, thus delaying the transition to turbulence. Kramers tests were
performed by towing a test model behind a motor boat in Long Beach Harbor. Unfor-
tunately, many attempts by other investigators to repeat Kramers experiments under
more controlled conditions failed to yield similar conclusions (e.g., Puryears 1962
2
Originally reported in the thesis by Willis (1986) and later contrasted to theory in the paper by Lucey
and Carpenter (1995).
7.2 COMPLIANT COATINGS PRIOR TO 1985 123
experiment in a towing tank). This so-called Kramer controversy will be revisited in
Section 7.5.
Theoretical work by Benjamin (1960a,b), Betchov (1960), Landahl (1962), and
Kaplan (1964) indicated that drag reduction by delaying transition is possible.
However, the theoretically predicted successful coatings had specic characteristics
that would be extremely difcult to match in practice. It is important to stress that
almost all this early work addressed the delay of transition and ignored the potential
for reducing turbulence skin friction with compliant coatings.
During the mid-1960s, Benjamin (1966) explored the possibility that a compliant
coating may affect the skin-friction drag in a fully developed turbulent boundary
layer without necessarily delaying transition. Dinkelacker (1966) conducted careful
tests of a compliant surface in a water-pipe ow. He systematically attempted to
determine the repeatability of rigid-tube data, the inuence of small steps in the tube
wall, and the possible occurrence of organ-pipe acoustic modes. Dinkelackers results
seemed to indicate a modest reduction in drag by using a compliant wall.
Blick and his coworkers at the University of Oklahoma experimentally demon-
strated signicant reductions in turbulence skin friction for compliant surfaces in
air (Fisher and Blick 1966; Looney and Blick 1966; Smith and Blick 1966; Blick and
Walters 1968; Chu and Blick 1969). Subsequent tests by Lissaman and Harris (1969),
who attempted to substantiate Blicks conclusions, yielded only extremely modest
gains. In another study, McMichael, Klebanoff, and Meese (1980) demonstrated that
the apparent reduction in turbulence skin friction in the University of Oklahomas
experiments could be a consequence of experimental deciencies coupled with the im-
proper interpretation of data. McMichael et al. (1980) concluded that drag reduction
via compliant coating in gaseous ows would not be as successful as in liquids.
During the 1970s various compliant materials were tested in water at the Naval
Ocean Systems Center, the Naval Research Laboratory, the Naval Undersea Systems
Center, and the Advanced Technology Center of the LTV Corporation, all in the
United States. In no case was a statistically signicant reduction in drag measured.
Fischer and Ash (1974) presented a general review of concepts for reducing skin fric-
tion, including the use of compliant coatings. Bushnell et al. (1977), in summarizing
the work conducted at the NASA Langley Research Center and the general status of
compliant surface drag reduction, stated that, although it was possible to increase
the transition Reynolds number by perhaps a factor of two, there was no denitive
reduction of drag for turbulent ows in air. They also stated that drag reduction in
turbulent ows in water is potentially feasible and can be accomplished using sur-
faces that can be practically built. It is of particular interest to note that much of
the research on compliant coatings has been based on materials that attempt to repli-
cate dolphin skin. Yet, in the Russian book Nekton (Aleyev 1977) it is indicated that
the wrinkling (Figure 7.1) of the dolphin skin has no hydrodynamic-drag advan-
tage. Other characteristics of the dolphins skin may, however, be benecial. This
subject will be revisited in Section 7.6.4.
Bushnell et al. (1977) put forward the possibility of a feedback mechanism in
turbulent wall-bounded ows through which the quasi-periodic, coherent structures
termed bursts regenerate (Chapter 4). Older bursts grow, migrate away fromthe wall,
and interact to produce a pressure eld that contains pulses of sufcient duration and
amplitude to induce new bursts in the near-wall region. This model is supported by
the measurements of Burton (1974), who reported a strong correlation between the
Figure 7.1: Wavelike folds in compliant skin appearing during rapid swimming of a delphinidae (top
photograph) and a Homo sapiens (bottom photograph) [after Aleyev 1977].
occurrence of a burst and the imposition on the wall ow of a large, moving adverse
pressure gradient followed by a favorable pressure gradient. Bushnell et al. (1977)
hypothesized that a successful compliant coating would modulate the preburst ow
in the turbulent boundary layer by providing a pressure eld that would tend to
block the feedback mechanism and, thus, inhibit burst formation. This would result
in a reduction in the number of bursts occurring per unit time and also in the skin-
friction drag. Orszag (1979) assumed this conceptual model and performed numerical
calculations of wall boundary layer instability to explore the effects of complaint
surfaces. His results, although preliminary, indicated that turbulence drag reduction
may be possible for certain classes of materials. He concluded that compliant walls
that support only short wavelengths may have an appreciable effect in inhibiting
further bursts in a turbulent boundary layer.
During the U.S. Navy-sponsored research program conducted over the period
19801985, the subject of boundary layer interaction with compliant coatings has
been reexamined to answer the question: Can compliant coatings delay transition, or
signicantly reduce turbulence skin friction on bodies at high Reynolds numbers, or
both? Several signicant developments have been achieved by the many investigators
participating in this research program. Although unrefutable experimental evidence
of compliant coating drag reduction was still lacking by 1985, our understanding of
boundary-layer ow over a compliant surface has increased dramatically over this
period. That understanding proved crucial to the subsequent successes in the eld,
which is a subject that will be emphasized throughout this chapter.
7.3 FREE-SURFACE WAVES 125
7.3 Free-Surface Waves
Before addressing the complex issue of stability of the coupled uid-structure system,
it is instructive to study the solid as a wave-bearing medium when no uid (another
wave-bearing medium) is present. Some understanding of how a compliant surface
will respond to the owabove it can be obtained by examining the free-surface waves
of the coating. As pointed out by Rayleigh (1887), the surface waves can be modeled
as a linear combination of waves having displacements perpendicular and parallel to
the propagation direction. These are called transverse and longitudinal displacement
waves, respectively. If it is assumed that the coating is a single-layer, elastic solid
of thickness d attached to a rigid half-space at its lower boundary and bounded by
vacuum on its upper surface, both wave systems satisfy the wave equation (see, for
example, Landau and Liftshitz 1987)
2
t
2
c
2

2
x
2
+

2
y
2
= 0 (7.1)
where is a component of the displacement vector, and x and y are coordinates par-
allel and normal to the undisturbed surface, respectively. The propagation velocities
are c = c
t
=

G/
s
for the transverse waves and c = c
(+2G)/
s
for the
longitudinal waves, where Gand are elastic constants, and
s
is the density of the
solid.
The free-surface wave dispersion relationship is obtained by assuming that the
wave solutions have exponential dependence on the distance measured normal to the
surface y and have harmonic dependence on the distance measured along the surface
x, and on time t
= (A sinh y + B cosh y)e
i (k
xt)
(7.2)
= (C sinh y + D cosh y)e
i (k
xt)
(7.3)
where and are, respectively, the displacement in the x- and y-direction, and k
is
the longitudinal wave number. For real the waves decay exponentially with depth,
whereas for imaginary they oscillate. Substituting these solutions into the wave
equation (7.1), and applying the boundary conditions, no normal or tangential stress
at the surface y = 0, and no displacement at the bottom y = d, respectively,
(c
2
2c
2
t
)
x
+c
2
y
= 0 at y = 0 (7.4)
y
+

x
= 0 at y = 0 (7.5)
= = 0 at y = d (7.6)
Gad-el-Hak, Blackwelder, and Riley (1984) obtained the following dispersion rela-
tionship:
M( ) 4
k
2
(2
2
)

t
k
2
[4 +(2
2
)
2
] cosh
t
d cosh
d
+
2
t

2
k
4
+(2
2
)
2
sinh
t
d sinh
d = 0 (7.7)
Figure 7.2: Solutions of the dispersion equation for free-surface waves on a
single-layer, homogenous coating [after Gad-el-Hak, Blackwelder, and Riley
1984].
where
t
= k
1
2
,
= k
1
2
R
2
, Ris the ratio of transverse to longitudinal
wave speed, and = c
p
/c
t
is the ratio of the surface wave speed to transverse wave
speed.
Plots of the dispersion curves resulting from Eq. (7.7) are given in Figure 7.2 for
the case R= 0, which is close to typical experimental conditions. Only one solution
has a surface wave speed below c
t
( <1). This solution approaches its asymptote
= 0.956 for large k
d, which is the value of Rayleighs innite half-layer solution.

Also note that as increases for a given k
d, the possibility exists for a richer range

of interactions between the coating and a ow. Typical values for k
d in experiments,
with the wavelength 2/k
taken as the boundary layer thickness, are in the range of

15. Figure 7.2 indicates that interaction between a uid owand a compliant surface
may not be expected for ow speeds much below c
t
and that the best opportunity
for interactions will be for ow speeds well above c
t
. Unfortunately, hydroelastic
instability waves appear for freestreamspeeds somewhat above c
t
, thus causing large
surface deformations and limiting the opportunities for favorable interactions. The
waveform can be obtained by determining the constants A, B, C, and Din Eqs. (7.2)
and (7.3) using the four boundary conditions (7.4)(7.6).
Duncan and Hsu (1984) calculated the dispersion relations for a two-layer coating
by nding the zeroes of the determinant of the boundary condition coefcients. Their
results for a thin, stiff coating placed on a much thicker, soft coating are shown in
Figure 7.3. The upper coating has about 100 times the modulus of rigidity of the lower
surface. In this gure, the wave number is normalized by the total coating thickness,
whereas the phase speed of the free-surface waves is normalized by the transverse
wave speed of the lower layer. This normalization allows an easy comparison with
the single-layer case when the upper portion of the coating is replaced by a layer of
stiffer material. The gross shapes of the curves in Figure 7.3 are similar to those in
Figure 7.2, although a wrinkle is clearly seen in the two-layer case. The dashed line in
7.4 SYSTEM INSTABILITIES 127
Figure 7.3: Solutions of the dispersion equation for free-surface waves on a two-layer coating
[after Duncan and Hsu 1984].
Figure 7.3 connects the maxima of the wrinkle in each modal curve and is a solution
of the equation for long bending waves in a free plate. Duncan and Hsu (1984) have
hypothesized that the upper coating acts like a plate that is inuenced by the lower
layer. They have attributed the cutoff of the response of the bending waves in the up-
per layer at small wave numbers to the lack of slow-moving waves in the lower layer.
7.4 System Instabilities
From a fundamental viewpoint, a rich variety of uidstructure interactions exists
when a uid ows over a surface that can comply with the ow. Not surprisingly,
instability modes proliferate when two wave-bearing media are coupled. Some waves
are ow-based, some are wall-based, and some are a result of the coalescence of both
kind of waves. What is most appealing about compliant coatings is their potential to
inhibit, or to foster, the dynamic instabilities that characterize both transitional and
turbulent boundary layer ows and in turn to modify the mass, heat, and momentum
uxes and change the drag and the acoustic properties. Although it is relatively easy
to suppress a particular instability mode, the challenge is of course to prevent other
modes from growing if the aim is, say, to delay laminar-to-turbulence transition.
From a practical point of view, it is obvious that an in-depth understanding of the
coupled system instabilities is a prerequisite to the design of a coating that meets a
given objective.
There are at least three classication schemes for the uidstructure waves, each
with its own advantages and disadvantages. The original scheme was introduced
by Benjamin (1963) and divides the waves into three classes according to their re-
sponse to irreversible energy transfer to and from the compliant wall. Both class A
and class B disturbances are essentially oscillations involving conservative energy
exchanges between the uid and solid, but their stability is determined by the net
effect of irreversible processes such as dissipation in the coating or energy transfer
to the solid by nonconservative hydrodynamic forces. Class A oscillations are TS
waves in the boundary layer modied by the wall compliance (in other words by the
motion of the solid in response to the pressure and shear-stress uctuations in the
ow). TollmienSchlichting waves are stabilized by the irreversible energy transfer
fromthe uid to the coating but destabilized by dissipation in the wall. Class B waves
reside in the wall and result from a resonance effect much the same as wind-induced
waves over a body of water. Their behavior is the reverse of that for class A waves,
for they are stabilized by wall damping but destabilized by the nonconservative hy-
drodynamic forces. Essentially class B waves are amplied when the ow supplies
sufcient energy to counterbalance the coating internal dissipation. Finally, class C
waves are akin to the inviscid KelvinHelmholtz instability and occur when conser-
vative hydrodynamic forces cause a unidirectional transfer of energy to the solid.
The pressure distribution in an inviscid ow over a wavy wall is in exact antiphase
with the elevation. In that case, class C waves can grow on the solid surface only if
the pressure amplitude is so large as to outweigh the coating stiffness. Irreversible
processes in both the uid and solid have negligible effect on class C instabilities.
If one considers the total disturbance energy of the coupled uidsolid system, a
decrease in that energy leads to an increase in the amplitude of class A instabilities,
class B is associated with an energy increase, and virtually no change in total energy
accompanies class C waves. In other words, any nonconservative ow of activation
energy from or to the system must be accompanied by disturbance growth of class A
or B waves, whereas the irreversible energy transfer for class C instability is nearly
zero.
The second classication scheme was introduced by Carpenter and Garrad (1985,
1986). It simply divides the waves into uid-based (TollmienSchlichting instabilities
[TSI]) and solid-based (ow-induced surface instabilities [FISI]). The latter are closely
analogous to the instabilities studied in hydro- and aeroelasticity and include the
traveling-wave utter that moves at speeds close to the solid free-wave-speed (class B)
and the essentially static, and more dangerous, divergence waves (class C).
3
The
main drawback of this classication scheme is that under certain circumstances the
uid-based TS waves and the solid-based utter can coalesce to form a powerful
new instability termed transitional mode by Sen and Arora (1988). According to the
energy criterion advanced by Landahl (1962), this latest instability is a second kind
of class Cwaves. In a physical experiment, however, it is rather difcult to distinguish
between the static-divergence waves and the transitional ones.
The third scheme to classify the instability waves considers whether they are
convective or absolute (Huerre and Monkewitz 1990). As discussed in Section 3.5,
an instability mode is considered to be absolute if its group velocity is zero. On the
other hand, the unstable development of a disturbance is said to be convective when
none of its constituent modes possesses zero group velocity. Both classes A and B
are convective, whereas class C divergence and transitional modes are absolute.
As Carpenter (1990) points out, the occurrence of absolute instabilities would lead
3
Static-divergence waves were erroneously interpreted in the past by, for example, Gad-el-Hak,
Blackwelder, and Riley (1984), Duncan et al. (1985), and Yeo (1992) as being class A. The confu-
sion occurs when divergence is treated as a convective instability when in fact it is an absolute one
(Carpenter 1990).
7.5 THE KRAMER CONTROVERSY 129
Figure 7.4: Summary of all three classication schemes.
to profound changes in the laminar-to-turbulence transition process. It is therefore
pointless to consider reducing their growth rate or postponing their appearance
to a higher Reynolds number; nothing short of complete suppression would work.
Figure 7.4 combines and summarizes all three classication schemes.
7.5 The Kramer Controversy
It may be worth recalling in more detail the pioneering work of Max O. Kramer and
the controversy surrounding it. The entire eld of compliant coatings became the
Rodney Dangereld of uid mechanics research, getting no respect from a skeptical
community largely because Kramers original experiments lost credibility. However,
as will be seen below, the most recent evidence resurrects the good name of this
ingenious GermanAmerican and with it renewed condence in this waning and
waxing eld.
As already mentioned, Kramer (1957; 1960a,b,c; 1961; 1962; 1965; 1969) con-
ducted his original experiments by towing a model behind an outboard motor boat
in Long Beach Harbor, California. His early tests showed a drag reduction of more
than 50% when a dolphinlike skin was used. A typical successful coating used by
Kramer consisted of a exible inner skin, an outer diaphragm, and stubs, all made
of soft, natural rubber. The cavity between the outer diaphragm and the inner skin
was usually lled with a highly viscous damping uid, such as silicone oil, which in
Kramers view damped out the TollmienSchlichting waves.
Subsequent experiments to conrmKramers ndings were conducted in a towing
tank, a lake, or a water tunnel (Puryear 1962; Nisewanger 1964; Ritter and Messum
1964; Ritter and Porteous 1964). No signicant drag reduction was observed in any
of these investigations. Since then, many researchers have assumed that Kramers
results were in error and that his observed drag reduction could have come about
as a result of favorable changes to the form drag or the accidental excretion of the
silicone oil used as the damping uid during the tests. Surface discontinuities could
have favorably altered the pressure drag, and the released oil could have acted as a
drag-reducing polymer when released into the boundary layer and the ambient uid.
Carpenter and Garrad (1985) stated that it is probably no exaggeration to sug-
gest that the credibility of Kramers coatings is now rather low. Acceptance of his
results was not granted by the scientic community because the rigorous standards
of scientic investigation were not met, and the gradual improvements by Kramer to
meet these standards were not adequate (Johnson 1980). It did not help his cause any
that Kramers explanations of his own empirical results, though intuitively appealing,
were proven physically incorrect. For example, we now know that damping in the
solid destabilizes TSI.
Almost 30 years after Kramers original investigation, Carpenter and Garrad
(1985) presented a very careful analysis of his experiments (e.g., Kramer 1957) and
the subsequent tests (Puryear 1962; Nisewanger 1964; Ritter and Messum 1964;
Ritter and Porteous 1964) that attempted to provide independent evidence of the
drag-reducing capabilities of Kramers coatings. On the basis of their own rigorous
analysis of the hydrodynamic stability of ows over Kramer-type compliant surfaces,
Carpenter and Garrad argued that Kramers coatings were only marginally capable
of delaying transition. Any unfavorable factor, such as an adverse pressure gradient,
a step in which the compliant surface is joined to a rigid surface or an unusually high
freestream turbulence level, could badly affect the performance of the coating. Also,
a particular coating was designed for a restricted range of Reynolds number and was
therefore unlikely to delay transition outside that range.
Carpenter and Garrad (1985) contend that one or more of the above adverse fac-
tors may have existed in the experiments conducted by Puryear (1962), Nisewanger
(1964), Ritter and Messum (1964), and Ritter and Porteous (1964). Puryears (1962)
experiments were conducted using a prolate spheroid in a towing tank. He did not use
Kramers coating with the best performance, and serious problems were encountered
in making a smooth joint between the rigid and compliant surfaces. Nisewangers
(1964) tests were conducted by releasing a lighter-than-water body of revolution
from the bottom of a lake. His Kramer-like coating contained a uid with a viscosity
below that of the optimum uid as determined from Kramers results. Ritter and
Messum (1964), and Ritter and Porteous (1964) conducted their experiments in a
water tunnel using either a at plate or a cylindrical model with an elliptic nose. The
conventional ume had a relatively high freestreamturbulence level, which may have
rendered the facility unsuitable for transition experiments.
In addition to these adverse effects, some evidence existed in the tests conducted
to conrm Kramers results for the onset of a hydroelastic instability in the coating.
Such large-amplitude waves would certainly lead to drag increase, and their presence
may indicate that the boundary layer was already turbulent. On the basis of these
experiments, Carpenter and Garrad (1985) concluded that the results presented in
these tests should not be taken as conclusive evidence that the Kramer coatings are
incapable of delaying transition and that the case against Kramers coating may not
be so strong as popularly supposed.
Carpenters (1988) optimization procedure described in Section 7.6.2 results in
a compliant coating capable of delaying transition by a factor of 46 in Reynolds
number. It is therefore quite conceivable to design a Kramer-type coating that may
7.6 TRANSITIONAL FLOWS 131
lead to a drag reduction of the order reported by the original inventor himself. The
preceding analysis of Kramers tests illustrates the importance of carefully selecting
the ow facility to conduct compliant coating experiments. The background turbu-
lence in the facility should be particularly monitored if transition delay is sought.
This is precisely what was done in the successful experiments conducted by Gaster
(1988) to conrm the theoretical prediction of Carpenter and Garrad (1985), both
of which are described in more detail in the following section.
7.6 Transitional Flows
7.6.1 Linear Stability Theory
The hydrodynamic and the hydroelastic stability theories have reached an im-
pressive level of maturity during the last two decades. The linear theories can be
handled, for the most part, analytically, whereas the nonlinear stability theories are
more computer intensive. Perhaps no one has contributed more to the recent applica-
tion of the stability theory to compliant coatings than Peter W. Carpenter, originally
with the University of Exeter and presently with the University of Warwick. His list
of relevant publications includes 65 papers and is growing; obviously only a selected
few will be cited in the present short chapter.
Within the framework of the linear stability theory, two-dimensional small dis-
turbances are assumed to be superimposed upon a steady, unidirectional mean ow.
As discussed in Chapter 6, the nonlinear, partial NavierStokes equations are then
reduced to the well-known OrrSommerfeld equation, which is a fourth-order, linear,
ordinary differential equation. The order of this equation increases when additional
complexities are included in the problem. For rotating-disk ows, for example, Corio-
lis and streamline-curvature terms are incorporated, leading to a sixth-order stability
equation. The major difculty in integrating the OrrSommerfeld equation is that it
is highly stiff and unstable, which makes it virtually impossible to apply conventional
numerical schemes. Explicit codes with step size that is commensurate with the global
behavior of the solution lead to numerical instabilities, and alternative routines have
been developed to handle this stiff eigenvalue problem.
An added difculty when the walls are compliant is the interfacial conditions
that require continuity of velocity and stress. Those boundary conditions can also be
linearized, but special care should still be exercised in handling them. Appropriate
equations must be used for the compliant walls to be able to fully couple the uid
and solid dynamics. Many types of compliant surfaces exist, and thus there are
numerous models for the solid. Those models can be either surface-based or volume-
based (Figure 7.5). The former model reduces the spatial dimensions by one and
is therefore less computationally demanding. An example is the thin platespring
model used by Garrad and Carpenter (1982), Carpenter and Garrad (1985, 1986),
Domaradzki and Metcalfe (1987), Metcalfe et al. (1991), and Davies and Carpenter
(1997a,b), among others, to simulate Kramer-type coatings. This model is relatively
simple yet contains characteristics representative of a broad range of surfaces. If a
coordinate system is chosen with the x-axis lying along the undisturbed free surface
and the y-axis normal to this surface, then the equation for the y-component
4
of the
4
The x-component of momentum is usually neglected in such a model.
Figure 7.5: Volume-based and surface-based models of compliant coatings [adapted from Carpenter
1990].
momentum of the compliant coating reads
t
2
=
T
x
2
D
t

F
m
x
4

k
m
+ F (7.8)
where (x, t) is the y-displacement of the surface from its equilibrium state at time
t and position x, T
is the longitudinal tension and F the exural rigidity of the

thin plate, m is the mass per unit area, D is the damping coefcient, k is the spring
constant, and F is an external forcing term.
The volume-based models are based on the Navier equation and include single and
multilayer coatings (Duncan, Waxman, and Tulin 1985; Fraser and Carpenter 1985;
Buckingham et al. 1985; Yeo 1988) as well as isotropic and anisotropic materials
(Yeo 1990, 1992; Duncan 1988). The equations describing the stability of the coupled
system form a numerical eigenvalue problem for the complex wave number of the
disturbance. Duncan (1987) offers a useful comparison between the results obtained
from a surface-based model and a corresponding volume-based one.
Compliant walls do suppress the TollmienSchlichting waves due to the irre-
versible energy transfer from the uid to the solid, but solid-based instabilities
proliferate if the coating becomes too soft. For the class A TS waves, the wall
compliance reduces the rate of production (via Reynolds stress) of the disturbance
kinetic energy. Simultaneously, the viscous dissipation is increased, and thus the
balance between the energy production and removal mechanisms is altered in favor
of wave suppression.
Experimental validation of the stability calculations is rather difcult and requires
well-controlled tests in a quiet water or wind tunnel. Several careful experiments to
test the ow stability to two-dimensional as well as three-dimensional controlled
disturbances have been reported in the past few years (Daniel et al. 1987; Gaster
1988; Lee et al. 1995, 1997). Figures 7.6 and 7.7 show the remarkable agreement
between theory and towing tank experiments for both rigid-wall and compliant-wall
cases. The top part of each gure is the predicted amplication factors as function
of ow speed for a range of modal frequencies, and the bottom part is the measured
growth-decay cycle of articially induced TS waves. A simple compliant model
predicts a dramatic decrease in the instability of the ow, and this prediction agrees
well with the experimental observations when a thick, soft coating is covered with a
thin, stiff layer.
The articles by Lee et al. (1995, 1997) report the results of wind tunnel ex-
periments and actually demonstrate the stabilizing potential of compliant coatings
in aerodynamic owsa remarkable achievement that had been deemed imprac-
tical in the past (Bushnell et al. 1977; Carpenter 1990). Excellent agreements are
reported between the results of the stability theory and the hydrodynamic experi-
ments (Willis 1986; Daniel et al. 1987; Gaster 1988; Riley et al. 1988; Carpenter
1990; Lucey and Carpenter 1995). The article by Lucy and Carpenter, in particular,
applies the linear stability theory to predict the experimentally observed evolution
of both TollmienSchlichting waves and traveling-wave utter in water ows. For
the wind tunnel experiments, Carpenter (1998) has conducted the corresponding
calculations, but his preliminary results thus far are negative: the density of an ef-
fective coating must be comparable to the uid density; otherwise, no transition
delaying benets are observed. This theoretical result leaves open the question of
explaining the positive experimental ndings of Lee et al. (1995). Here, we show
one example of the suppression of TS waves in an air boundary layer developing
on top of a silicone elastomersilicone oil compliant surface. Figure 7.8 presents the
wind tunnel results of Lee et al. (1995). The coating in Figure 7.8a was made by
mixing 91% by weight of 100 mm
2
/s silicone oil with 9% of silicone elastomer. In
Figures 7.8b,c, and d, the corresponding mix was 90% and 10%, yielding about
35% higher modulus of rigidity. As compared with the rigid wall, the single-layer,
isotropic, viscoelastic compliant coating signicantly suppresses the rms amplitude
of the articially generated TollmienSchlichting waves across the entire boundary
layer for a range of displacement-thickness Reynolds numbers. Reductions in the
maximum rms amplitude of as much as 40% are observed for the softer coating
(Figure 7.8a), which may lead to delayed transition.
7.6.2 Coating Optimization
If a compliant coating is to be designed for use on an actual vehicle, a relevant
question may be, What are the optimumwall properties to give the greatest transition
delay? The large number of available parameters makes it imperative that a rational
(i.e., one derived from rst principles) selection process be conducted. For obvious
Figure 7.6: Growth curves for the rigid-wall case. (a) Theoretical prediction; (b) Experimen-
tal results [after Willis 1986].
reasons, the trial-and-error empirical approach used in the past (if it is soft, let us
try it!) should not even be contemplated. This should be particularly true now that
rational optimization procedures are becoming readily available, as described below.
A wall that is too compliant (i.e., too soft) can substantially delay transition via TSI
by shrinking its unstable region in the frequency-Reynolds number plane, but rapid
breakdown can occur through the amplication of wall-based instabilities (Lucey
and Carpenter 1995). Both kinds of FISI are potentially harmful. The divergence
Figure 7.7: Growth curves for the compliant-wall case. A two-layer coating with an elastic
modulus of the lower layer of E = 5000 N m
2
. (a) Theoretical prediction; (b) Experimental
results [after Willis 1986].
instabilities are absolute and nearly static and yield to wholesale deformations of the
surface, which are likely to trigger premature transition due to a roughnesslike effect
(Figure 7.9). Flutter instabilities, though convective, are also dangerous. As shown
in the stability diagram in Figure 7.10, their narrow band of unstable frequencies
extends indenitely as Reynolds number increases downstream. Thus, once these
instabilities are encountered at some downstreamlocation, sustained growth follows.
Figure 7.8: Comparison of distribution of rms-amplitude of the TSI of rigid surface
and compliant surface across boundary layer. Wind tunnel experiments of Lee et al.
(1995). , rigid surface; , compliant surface. (a) Re
= 1274; (b) Re
= 1105;
(c) Re
= 1225; (d) Re
= 1350. Inset shows locations of ribbon () and probe ()

relative to neutral-stability curve.
This is unlike the broadband TollmienSchlichting instabilities that grow and then
decay as the different waves travel downstreamand pass through the lower and upper
branches of their neutral-stability curve (Figure 7.10).
A workable strategy for coating optimization suggested by Carpenter (1988) is to
choose a restricted set of wall properties such that the coating is marginally stable with
respect to FISI (both utter and divergence). The remaining disposable wall parame-
ters can then be varied to obtain the greatest possible transition (via TSI) delay. For
the plate-spring, surface-based model, for example, there are two disposable parame-
ters: the wall damping and the critical wave number for divergence. The down-stream
location of the transition region is estimated froman e
n
-criterion, where n is typically
chosen in the range of 710. The lower exponent represents the approximate limit of
validity of the linear stability theory for a low-disturbance environment and provides
a rather conservative calculation.
Although wall dissipation destabilizes TollmienSchlichting waves, a viscoelastic
coating with moderate level of damping leads to greater delay in transition as com-
pared with purely elastic surfaces. Apparently the stabilizing effects of wall damping
Figure 7.9: Static-divergence waves on a single-layer viscoelastic coating subjected to the pressure uc-
tuations of a turbulent boundary layer. Freestream velocity is 6.3 times the transverse-wave speed in the
solid [after Gad-el-Hak, Blackwelder, and Riley 1984].
on traveling-wave utter allow a softer wall to be used and thus more than offset the
adverse effects of coating dissipation on TSI.
Coating optimization with respect to TSI growth rate is performed at a rather nar-
rowrange of Reynolds numbers. On a growing boundary layer, the Reynolds number
increases monotonically, and a compliant coating will not be optimumover the whole
Figure 7.10: Typical neutral-stability curves for TollmienSchlichting
waves and for traveling-wave utter. The solid is a double-layer, Gaster-
type compliant coating [after Lucey and Carpenter 1995].
Figure 7.11: Static-divergence waves are clearly seen under the turbulent wedge generated by a single
roughness element embedded into an otherwise laminar boundary layer [after Gad-el-Hak, Blackwelder,
and Riley 1984].
length of a vehicle. Carpenter (1993) suggested that a multiple-panel wall placed in
series, with each panel optimized for a particular range of Reynolds numbers, is
likely to produce larger transition delays than a single-panel wall. His calculations
for a two-panel, plate-spring-type compliant wall indicate an additional performance
improvement of more than 30% over an optimized single-panel wall.
It seems reasonable that a large number of panels, say 10, in series would lead
to superior performance, but of course the calculations involved become prohibitive
very quickly. An additional benet from using multipanels is that shorter panels are
more resistant to both static-divergence waves and traveling-wave utter (Carpenter
1993; Lucey and Carpenter 1993; Dixon et al. 1994), thus allowing softer panels to
be used that further suppress TSI and improve the coating performance.
In the at plate and similar boundary layers in low-disturbance environments,
the quasi-two-dimensional TollmienSchlichting waves dominate the laminar-to-
turbulence transition. Various receptivity processes are responsible for generating
two-dimensional instabilities, which are probably initially three-dimensional and
randomly distributed. Low-disturbance environments could be realized, for example,
in free ight and marine vehicles. Very recently, Davies and Carpenter (1997a) and
Carpenter (1998) have shown that in such environments complete suppression of the
TollmienSchlichting waves is possible provided that optimized multiple-panel com-
pliant walls are used with each panel tailored to suit its local surrounding. Assured by
the experimental observations that static-divergence waves are only observed when
the wall-bounded ow is turbulent (see Figure 7.11, reproduced from Gad-el-Hak,
Blackwelder, and Riley 1984),
5
Carpenters (1998) new assumptions are somewhat
5
In other words, one less instability mode to worry about when the ow is laminar and the objective is
to keep it so.
less conservative than those used in earlier theories (Carpenter 1993; Dixon et al.
1994). The new nding raises the possibility of maintaining laminar ow in situa-
tions where the TS instabilities are the primary cause of transition,
6
to indenitely
high Reynolds numbersa very profound prospect indeed.
Work on nonlinear stability theory has recently been in the forefront and conrms
that transition-delaying coatings, optimized using the linear theory, maintain their
benecial effects into the latter stages of transition to turbulence (Metcalfe et al. 1991;
Joslin and Morris 1992; Thomas 1992a,b). Lee et al. (1997) experimentally studied
the effects of a compliant surface on the growth rates of both the subharmonic and
three-dimensional uid-based instabilities of a laminar boundary layer in air. Their
results suggest that a delay of the excitement of the secondary instability can be
achieved by suppressing the growth of the primary waves using surface compliance.
7.6.3 Practical Examples
Most of the theoretical as well as experimental compliant coating research has
been concerned with canonical boundary layers. Nevertheless, an attempt is made
herein to estimate the potential benet of applying the technique for eld applications
in which strong three-dimensional and pressure-gradient effects and, for aeronautical
applications, compressibility effects may be present. The typical Reynolds numbers,
based on vehicle speed and overall length, for a hydrofoil, a torpedo, and a nuclear
submarine are, respectively, of the order of 10 million, 50 million, and 1 billion.
Applying e
n
-type calculations (with the exponent chosen conservatively to be n =
7) to an optimum two-panel, plate-spring-type compliant wall, Carpenter (1993)
computed a transition Reynolds number of 13.6210
6
as compared with 2.2510
6
for a rigid wall.
7
This means that the laminar region that would normally extend
over 23%, 5%, and 0.2% of the respective vehicle lengths would, with the use of an
optimum coating, extend over a larger length of 100%, 27%, and 1%. When one
computes the corresponding overall drag coefcients using standard methods for a
mixed laminar-turbulent boundary layer over a at plate, the potential reduction in
skin-friction drag using the optimum two-panel compliant wall can be as much as
83%, 19%, and 0% for the three respective vehicles. Obviously the large submarine
does not benet, as far as drag reduction is concerned, from the use of transition-
delaying compliant coating, but the smaller vehicles do. However, extending the
laminar region on a submarine even by 1 m can be signicant for sonar applications
requiring longer quiet regions of the boundary layer.
For aeronautical applications, a cruising commercial jet aircraft has a fuselage
Reynolds number of the order of 0.5 billion and a wing Reynolds number of the
order of 50 million. Again, increasing the transition Reynolds number by a factor of
5 or so is signicant for the wing but not for the fuselage. Skin-friction reduction
of the order of 20% is achievable for the wings (whose skin-friction drag accounts
6
Excludes, for example, ows where the walls, streamlines, or both are concave (G ortler instabilities),
where the boundary layer is three-dimensional and the crossowvelocity prole is inectional (crossow
instabilities), or where the wall is rough or the freestream turbulence levels are unusually high (bypass
transition).
7
Contrast this sixfold increase to the 30%higher transition Reynolds number reported in the experiments
of Gaster (1988), who did not attempt to optimize his single-panel, two-layer, silicone rubberlatex
rubber coating. Theoretical calculations by Dixon et al. (1994) indicate that an optimum Gaster-type
coating would provide a 500% higher transition Reynolds number.
for about 50% of the skin friction of the entire aircraft and 25% of the total drag).
8
Finding a compliant coating that would reduce the turbulent skin-friction drag
would of course be very benecial for both the typical fuselage and long submarine.
The estimates above were made for a simple plate-spring model. Using more than
two panels can provide further transition delay. More complex compliant surfaces,
particularly anisotropic ones designed specically to suppress the Reynolds stress
uctuations, can conceivably offer more spectacular savings. Such custom-designed
coatings can also favorably interact with fully turbulent ows. Even for laminar
ows, the calculations involved when complex, wall-based models are used, though
straightforward in principle, are quite demanding in practice.
7.6.4 The Dolphins Secret
The ability to swim or to y with minimum skin friction and pressure drag is of
extreme importance to the Darwinian survival of certain nektonic and avian species.
Homo sapiens interested in building the fastest submarine or the most fuel-efcient
aircraft have much to learn about alternative drag-reducing approaches fromhumble
earthlings. As Kramer remarked close to half a century ago, a school of porpoises,
including the young and the old, the weak and the strong, showing off their seemingly
effortless glide along a fast ocean liner is a sight to behold.
Cetaceans appear to possess unusually low overall drag coefcients. This is the
basic factor for the so-called Grays (1936) paradox in which a steady-state energy
balance based on the anticipated muscle power of various nektons, including the dol-
phin, failed to explain their unusually fast swimming speeds. Gray clocked bottlenose
dolphins, Tursiops trucatus, swimming at speeds exceeding 10 m/s for a period of 7 s.
If one assumes that the power output of cetaceans is equal to that of other mammals
(35 W/kg of body weight), then such speeds are reached under turbulent ow con-
ditions only if dolphins can expend several times more power than their muscles can
generate. Lang (1963) concluded on the basis of energy considerations that dolphins
can not exceed a speed of 6 m/s for periods greater than 2 hours.
Transition delay is of course an obvious, albeit arduous, technique for achieving
about an order of magnitude lower skin-friction drag, but does the dolphin possess
an exotic means by which such a difcult ow-control goal can be accomplished?
Obviously the dolphin is not sharing its secrets with other fellowmammals. Kramers
(1957, 1961) invention of a compliant coating tried to mimic the dolphins epidermis
and claimed drag reduction of as much as 60%. His explanation for the dolphins
secret is that its skin, like his successful compliant coating, is capable of substantially
delaying laminar-to-turbulence transition. Kramers work was discredited for a while
but now seems to be back in vogue, as remarked in Section 7.5. The calculations
presented in Section 7.6.3 indicate that it is quite possible to design a Kramer-type
coating that delays transition by a factor of 46 in Reynolds number and that drag
reduction of the order reported by Kramer is also quite possible. Does the dolphin
or other similar fast swimmers possess such a coating?
In a recent article, Bushnell and Moore (1991) quoted the relevant energetic and
controlled swimming studies but concluded by supporting the explanation offered
8
Note that the dominant mechanism leading to transition on a swept wing is crossow instability and
not TS waves. However, recent evidence (Cooper and Carpenter 1995, 1997a,b,c) indicates benecial
effects of compliant surfaces even for the former kind of instability.
7.7 TURBULENT WALL-BOUNDED FLOWS 141
by Au and Weihs (1980) that dolphins, which must periodically breath air, achieve
high-speed swimming by simply porpoising, that is, momentarily leaping out of the
water, thereby reducing their drag force by a factor of 800 (density ratio of air and
water). This more than pays for the additional interfacial or wave drag and accounts
for the abnormally low apparent drag coefcients inferred from the assumption of
fully submerged travel.
I, however, do not concur with the preceding nal solution to the Grays paradox.
Dolphins have been clocked at sustained and burst speeds of close to 10 and 20 m/s,
respectively. Delphinus delphis has a typical length of 2 m. This leads to sustained
and burst Reynolds numbers based on the overall length of the order of 20 million
and 40 million, respectively. Carpenter (1993) reported the results of optimizing
a rather simple plate-spring coating. By using a single panel, as compared with a
rigid surface, a 4.6-fold increase in transition Reynolds number is estimated, which
leads to a drag reduction of 36% at the typical dolphins sustained speed and 20%
at burst speed. Using a mere two-panel coating, the transition Reynolds number
becomes 6.1 times the value for a rigid surface, and the potential drag reductions
for the sustained and burst speeds are now 52% and 30%, respectively.
These lower levels of skin friction are compatible with the available muscle power
for a dolphin of the size used above. Admittedly, the preceding estimates were made
for a at-plate boundary layer and may not hold when pressure-gradient and other
shape effects are taken into account. Additionally, the dolphin also has pressure drag
on top of the (much larger) skin friction. On the other hand, cetaceans have had
millions of years of evolutionary adaptations to hone their coatings for maximum
speed and efciency, and it is quite conceivable that their epidermis is much more
complex, and hydrodynamically benecial, than the simple ones computed in the
examples above. Moreover, each portion of the skin could have been optimized for
the appropriate range of local Reynolds numbers. Therefore, the dolphins appar-
ent success is not incompatible with having optimum compliant coatings to delay
laminar-to-turbulence transition substantially and therefore to attain inordinately
low coefcients of drag.
Other fascinating questions related to the amazing swimming abilities of the dol-
phin include the possibility that its excreted mucin is a drag-reducing additive. Is
there a hydrodynamic advantage to the warm-blooded cetaceans because their epi-
dermal temperature is higher than the ambient one (in which case the near-wall water
viscosity is lowered and the turbulent boundary layer may be relaminarized)? Does
the dolphins particular body shape during coasting (with no attendant overall body
deformation) or actual swimming (accompanied by appropriate body oscillations)
offer additional drag-reducing advantages? Also, what are the potential benets to
the porpoise when it uses ship-generated bow waves for body surng? These sub-
jects, though related to the preceding discussion, are outside the scope of the present
discussion and are therefore left for another circumstance.
7.7 Turbulent Wall-Bounded Flows
Unlike the laminar and transitional ows investigated in Section 7.6, compliant coat-
ing effects on turbulent boundary layers are rather difcult to study theoretically. In
fact any turbulent ow is largely unapproachable analytically. For a turbulent ow,
the dependent variables are random functions of space and time, and no straight-
forward method exists for analytically obtaining stochastic solutions to the govern-
ing nonlinear, partial differential equations. The statistical approach to solving the
NavierStokes equations always leads to more unknowns than equations (the clo-
sure problem), and solutions based on rst principles are again not possible. Direct
numerical simulations (DNS) of the canonical turbulent boundary layer have thus
far been carried out up to a very modest momentum-thickness Reynolds number of
1410 (Spalart 1988).
How would one go about rationally choosing a coating to achieve a particular
control goal for a turbulent boundary layer? Analytical optimization procedures such
as those used to delay transition (Section 7.6.2) would not work for fully turbulent
ows. To analyze the full problem, direct numerical simulations of the turbulent
boundary layer should be coupled to a nite-element model of the compliant coat-
ing, a task that is extremely time-consuming and expensive and taxes the fastest su-
percomputer around. Modeling the turbulence by an eddy-viscosity or even a more
sophisticated closure scheme is less computationally demanding, but there is no guar-
antee that turbulence models developed primarily for rigid surfaces would work for
a compliant surface. In fact, it is not difcult to argue that closure models based on
mean quantities completely miss the all-important spectral contents of a uidsolid
interaction and will therefore never work.
A turbulent boundary layer is characterized by a hierarchy of coherent struc-
tures. Near the wall, the dynamics are dominated by the quasi-periodic bursting
events (Robinson 1991). A crude, albeit resourceful, attempt to model a turbulent
boundary-layer interaction with a single-layer, isotropic, viscoelastic coating has been
advanced by Duncan (1986). He approximates the turbulent ow over the coating
by a potential ow with a superimposed pressure pulse, convecting downstream,
that mimics the pressure footprint of a single bursting event. To relate the prob-
lem to a real turbulent ow, the pressure pulse characteristics are taken from actual
boundary layer measurements and the potential ow is modied to incorporate the
reduced magnitudes and phase shifts found experimentally in boundary layer ows
over moving wavy walls. At low ow speeds (relative to the transverse-wave speed in
the solid), the coating response to the pressure pulse is stable and primarily localized
under it. At intermediate speeds, the response is still stable but includes a discernible
wave pattern tagging along behind the pressure pulse. At the highest speed stud-
ied, large-amplitude, unstable waves develop on the compliant surface, much the
same as the FISI observed experimentally. Duncan and Sirkis (1992) have recently
extended this model to anisotropic compliant coatings. They reported that certain
anisotropic surfaces provide more effective control over the amplitude and angular
extent of the generated stable response pattern. Larger amplitudes are generated as
compared with isotropic surfaces, thus providing for greater potential for modifying
the turbulence.
Whenever the ow speed in a turbulent boundary layer becomes sufciently large
compared with the transverse free-wave speed in the solid, ow-induced surface
instabilities proliferate. The pressure uctuations within the ow are an order of
magnitude larger than the normal and tangential viscous stresses and drive the coating
response. In laminar wall-bounded ows it is difcult to observe the hydroelastic
waves in their unstable state. As soon as utter or divergence waves grow, rapid
breakdown to turbulence takes place in the boundary layer, and the ow is no longer
laminar.
7.7 TURBULENT WALL-BOUNDED FLOWS 143
Most of the experimental studies concerning compliant coating effects on tur-
bulent boundary layers focused on documenting the unstable ow-induced surface
instabilities. When divergence waves or utter are unstable, the effects, though
adverse, are pronounced and are somewhat easier to document. Only recently few
hardy souls have attempted to investigate the wall-bounded ows when these FISI
are stable or neutrally stable. Obviously, the latter kind of studies have to await the
development of rened techniques to measure the minuscule surface deformation
and the associated coherent structure modulation when the FISI are neutrally stable.
Both Gad-el-Hak (1986a,b) and Hess, Peattie, and Schwarz (1993) introduced
nonintrusive methods for the point measurement of the instantaneous vertical surface
displacement of a compliant coating, whereas Lee, Fisher, and Schwarz (1993b) of-
fered an optical holographic interferometer, in connection with an interactive fringe-
processing system, to capture whole-eld random topographic features. The latter
technique is more expensive to set up but offers higher spatial resolution, of the order
of 1 m, and yields simultaneous surface displacement information on a large section
of the compliant coating. Both the local and global methods were initially employed
to document the unstable surface response to the pressure uctuations in turbulent
boundary layers. The holographic interferometer was recently used to record the
surface topography in the presence of stable ow-induced deformations (Lee et al.
1993a).
The onset speed and wave characteristics of the solid-based class B and C in-
stabilities were systematically documented in a series of towing-tank experiments
(Gad-el-Hak 1986b). Divergence waves were observed on a single-layer viscoelas-
tic coating made from a PVC plastisol. The utter appeared on an elastic coating
made from common household gelatin, but, in the absence of damping, its threshold
speed was consistently lower than that for divergence (Figure 7.12). The damping
in the PVC coating stabilized the traveling-wave utter, and hence only divergence
was observed there.
9
For the elastic coating, utter appeared rst and dominated the
observed surface deformation. For both kind of waves, the threshold speed decreases
with coating thickness; in other words, thin surfaces (relative to the displacement
thickness of the boundary layer) are less susceptible to hydroelastic instabilities than
thick ones.
Typical proles of unstable class B and C waves were also recorded in the
same hydrodynamic experiments using a laser displacement gauge (Figure 7.13.
The vertical displacement at a point associated with the slow-moving, asymmetric,
large-amplitude divergence waves contrasts with the faster, more-or-less symmetric,
smaller-amplitude utter. Both waves cause roughness-like effect, but the static di-
vergence is the more dangerous instability. The phase speed of the static-divergence
waves is of the order of 1% of the freestream speed, and their wavelength is about
510 times the coating thickness. The corresponding quantities for the utter are
40% and 1.53, respectively.
Hess (1990) and Lee et al. (1993a) also investigated compliant coating effects
on turbulent boundary layers. Both experiments were conducted in the same water
tunnel, but the second article focused on the stable interaction between the uid and a
9
Parenthetically, this and similar earlier observations led Gad-el-Hak (1986a) and others to the wrong
conclusion, as stated in footnote 3 in Section 7.4, regarding the classication of the divergence waves.
To reiterate, the class B utter is stabilized by damping, whereas the class C divergence is largely
unaffected.
Figure 7.12: Onset speed dependence on thickness.
, traveling-wave
utter on an elastic coating; , static-divergence waves on a coating
with damping [after Gad-el-Hak 1986b].
Figure 7.13: Typical unstable displacements of an elastic coating () and vis-
coelastic coating () [after Gad-el-Hak 1986b].
7.8 THE FUTURE 145
Figure 7.14: Variations of mean dimensionless spanwise spacing of the low-
speed streaks in a turbulent boundary layer. Filled

, rigid surface; lled ,
compliant surface, friction velocity u
obtained from rigid-surface mean-

velocity measurements; , compliant surface, u
obtained from compliant-

surface mean-velocity measurements. These data are from the water tunnel
experiments of Lee et al. (1993a). Other symbols in the gure are from
classical rigid-surface measurements.
single-layer, homogeneous, viscoelastic coating made of a mixture of silicone rubber
and silicone oil. Lee et al.s coating was chosen based on the criterion established by
Duncan (1986). In the presence of a stable wave pattern on the compliant surface,
the ow visualization experiments indicated low-speed streaks with increased
spanwise spacing (by as much as 80%; see Figure 7.14) and elongated spatial
coherence compared with those obtained on a rigid surface. More signicantly, for
the particular compliant coating investigated, an intermittent relaminarization-like
phenomenon was observed at low Reynolds numbers. Lee et al. (1993a) also
reported a slight thickening of the buffer region and viscous sublayer and an upward
vertical shift in the compliant law-of-the-wall. The streamwise turbulence intensity,
the local skin-friction coefcient and the Reynolds stress across the boundary layer
were all reduced, indicating a possible interruption of the feedback loop, which
allows the turbulence to be self-sustaining. Thus, potentially favorable interaction
between a compliant coating and a turbulent boundary layer has been demonstrated
for the rst time. The more recent hydrodynamic experiments by Choi et al. (1997)
provide additional evidence for favorable interaction and indicate a total drag
reduction for a long, slender body of revolution of the order of 7%.
7.8 The Future
The diminishing pool of researchers remaining active in the eld of compliant coatings
includes teams from the University of Warwick, University of Nottingham, Johns
Hopkins University, University of Houston, University of Maryland, and the Institute
of Thermophysics in Novosibirsk. Alarger pool was involved during the early 1980s,
but the realities of research funding combined with the checkered past of the eld
led to the present decline.
Few suggestions for future research are given in here. The optimization proce-
dures discussed in Section 7.6.2 have not been validated experimentally. Gaster-type
experiments should be repeated using optimized coatings, including multi-panel ones.
The recent claims by Davies and Carpenter (1997a) and Carpenter (1998) regarding
the possibility of maintaining laminar owto indenitely high Reynolds numbers are
very profound. Experiments, particularly eld ones in low-disturbance environments
specically designed to test those claims would be extremely useful.
The results of the transitional-boundary-layer, wind-tunnel experiments reported
by Lee et al. (1995, 1997) are intriguing and y in the face of the conventional wis-
dom. They indicate that compliant coatings are capable of delaying transition even
for air ows. Past calculations using a plate-spring model and considering the ex-
tremely large density of typical walls compared with the density of air indicated
that very imsy coatings would be required to achieve transition delay and that the
situation gets worse as the air speed increases. This led Carpenter (1990) among
others to conclude that the use of wall compliance is impractical for aeronautical
applications. But the plate-spring results do not apply in any straightforward way
to the homogeneous, single-layer walls studied by Lee et al. (1995). Validating the
recent favorable results using both independent experiments and numerical simula-
tions would open the door for aerodynamic applications, which is something that
was seriously considered but later abandoned by NASA and the aerospace industry.
The optimization procedures developed by Carpenter (1988) for transitional hydro-
dynamic ows should be extended to air ows. Experiments should be conducted
using the resulting optimized coatings.
More complex coatings could potentially yield superior performance as com-
pared with the relatively simple walls studied thus far. Multipanels, multilayers,
anisotropic coatings, and combinations thereof should be investigated. In any such
research program, experiments have to be guided with theoretical results. As already
mentioned, trying to pick a compliant coating by trial and error is a very inefcient
use of limited resources and will perhaps never work.
Favorably modulating a fully turbulent ow, in contrast to merely delaying tran-
sition, is also of great practical importance. The experimental results reported by
Lee et al. (1993a) are very encouraging, but the coating used was chosen based on
a rather simplistic model of the turbulence pressure uctuations. To custom design
compliant coatings to achieve particular control goals for turbulent wall-bounded
ows, direct numerical simulations of the coupled uid-structure system have to be
performed. Turbulence modeling via classical closure schemes, although sufcient
for some simple ows over rigid surfaces, will perhaps not yield reliable results for
compliant walls. Direct numerical simulations, on the other hand, require extensive
computer resources and are quite expensive to carry out. The bottom line is that
relatively large investment in resources is required for this task, but the enormous
potential payoffs could easily justify the expenditure.
Most of the research thus far has considered incompressible, zero-pressure-
gradient, at-plate boundary layers. Effects of compressibility, pressure gradient, and
three-dimensionality on the performance of compliant coatings are largely unknown.
7.8 THE FUTURE 147
Such studies will yield invaluable information for eld application of the control
technique for both air and water ows. Most practical aerodynamic ows are in
the moderate-to-high Mach number regime, and compressibility effects must there-
fore be investigated before compliant coatings are used on actual aircraft. Related
to the pressure-gradient effects is the question of separated ows: Does compliant
coating affect separation favorably or adversely? Other stability modiers, such as
favorable pressure gradient, suction, or heating or cooling, do delay transition as
well as prevent separation. It is not known whether compliant coatings also have
this dual benet, and it may be benecial to research the possibility. Finally, real
ows are three-dimensional and involve complex geometries. A model problem for
three-dimensional ows is the rotating disk. Few experiments were conducted using
a rotating disk with a compliant face (Hansen and Hunston 1974). More recently,
Cooper and Carpenter (1995; 1997a,b,c) analyzed the crossow (type I; inviscid)
as well as the viscous (type II) uid-based instabilities that develop in the same
three-dimensional ow. The preliminary results are encouraging and indicate that
compliant coatings can suppress the more dangerous type I instabilities.
Active compliant coatings, or more properly internally driven exible coatings,
though bringing us back to the complexity of reactive control systems, is an emerging
area deserving of further research. Energy expenditure is required to drive the wall,
but the potential for signicant net drag reduction is higher than that for passive
coatings. The feasibility of the concept for stabilizing laminar boundary layers has
been shown through numerical experiments (Metcalfe et al. 1986). Internally driven
exible coatings could also be used to suppress the Reynolds stress and reduce the
skin-friction drag in turbulent wall-bounded ows, but any realistic eld application
of the technique has to await further development of reasonably priced and rugged
microfabricated sensors and actuators (Gad-el-Hak 1994, 1996b). This topic will be
revisited in Chapter 14.
Using the subject of compliant coatings to make a point, I would like to end this
section with a personal commentary. There is a growing impatience among our fellow
citizens with the glacial pace of transferring knowledge from the laboratories to the
factories and thus getting back the invested research dollars of yesterday in the form
of stronger industrial competitiveness tomorrow. In this era of instant gratication,
what matters is not what basic research has done for you over the long haul, but what
research has done lately. In his closing remarks on the occasion of the presentation of
the 1990 American Physical Society Fluid Dynamics Prize, Lumley (1992) lamented
that the United States is a curiously unsympathetic environment for a theoretician,
or any scientist interested in fundamental work. At a later pulpit, Lumley (1996a)
observed that
public support of science in the United States has always been characterized by a certain
inconstancy, a ckleness, a tendency to unsuitable liaisons and the transfer of affection
from one pretty face to another at the drop of a hat. This is nowhere more evident than
in the support of research in turbulence; the great need for a solution, coupled with the
evident slow rate of progress, means that new ideas sometimes attract more attention than
they are subsequently found to deserve. The turbulence community itself is somewhat
prone to this kind of behavior.
Through all its ups and downs, compliant coating research provides a good case
study of what my academic sibling John Lumley is talking about. Although in the
general scheme of things this basic research is but a drop in the ocean, its triumphs
and debacles are not untypical. The usual 5-year cycle for academic research is just
enough to get off the ground. One should always remember the words of George
Lichtenberg recited at the beginning of this chapter.
When the compliant coating research program was reignited in the early 1980s,
few veterans from the 1960s and 1970s were around to share their valuable experi-
ences, and the newcomers have had to climb the learning curve from its bottom.
Nevertheless, we now know how to carry out stability calculations, solve fully
coupled uidsolid problems, numerically simulate turbulent ows, conduct well-
controlled experiments for both transitional and turbulent ows, reliably measure
surface deformation, identify as well as quantify coherent structures, optimize a
coating for a particular task using rst principles, and so forth. In other words, the
compliant coating research community has now most of the tools it needs for signif-
icant further progress. Unfortunately, this community has recently been forced into
an early retirement.
As amply illustrated in this chapter, compliant coating research, despite its check-
ered history, offers the potential for substantial transition delay and favorable inter-
actions with turbulent boundary layers. It requires modest commitment of resources,
but the payoff is extraordinary. It might be worth recalling that a mere 10% reduc-
tion in the total drag of an aircraft translates into a saving of $1 billion in annual
fuel cost (at 1999 prices) for the commercial eet of aircraft in the United States
alone. Contrast this benet to the annual expenditure of less than $2 million for the
5-year compliant coating research program that was sponsored by the U.S. Ofce
of Naval Research in 1980. Private capital will not and cannot step in place of the
government to support leading-edge research with long-term promise but without
short-term return. Do we have the will, desire, patience, and resources to continue
the journey toward real-life applications? In so many similar circumstances in the
past, the answer was no. Scarce resources were spent for 5 years only to be hastily
diverted to newer areaslong before the fruits of our labor have even had a chance
of being reaped. But even in these difcult times of trying to reduce the federal bud-
get decit, one hopes for a different road to prosperityand I do not mean for the
researchers involvedthis time around.
And while we are at it, cutting funds for basic research in general may be popular
but is certainly unwise. Todays research generates the knowledge from which the
future is built, and a responsible government must strike a balance between near-term
goals and long-term economic growth and prosperity. Reducing spending on current
outlays is one thing, but investing less in the countrys future is an entirely different
matter. Fundamental knowledge provides the foundation for a nations productivity
and economic growth, sustains its high standard of living, improves its quality of
health and environment, and ensures its security. The overall detrimental impact
of reducing government funding for basic research probably will not be felt for a
generation, but it will be felt.
7.9 Parting Remarks
Passive compliant coatings present a much simpler alternative to reactive ow-
control strategies aimed at favorably interfering with wall-bounded ows. The last
1015 years have witnessed renewed interest in compliant coatings as a means to
7.9 PARTING REMARKS 149
achieve benecial ow-control goals. Signicant advances were made in numerical
and analytical techniques to solve the coupled uid structure problem. Novel exper-
imental tools were developed to measure the stable as well as the unstable surface
deformations caused by the pressure uctuations in the boundary layer. In turbulent
wall-bounded ows, coherent structures were routinely identied, and their modu-
lation by wall compliance could be quantied.
Most signicant results in the eld thus far were obtained when a strong cooper-
ation existed between theory and experiment. Recent theoretical work indicates that
complete suppression of the TollmienSchlichting waves may be possible, provided
that optimized multiple-panel compliant walls are used. The new nding raises the
possibility of maintaining laminar owto indenitely high Reynolds numbers, which
is a very profound prospect indeed. Recent experiments indicate favorable compli-
ant coating interactions even for aerodynamic ows and even for turbulent boundary
layers. More research is needed, however, to conrm these latest results.
The coupled systeminstabilities are nowwell understood, and compliant coatings
can therefore be rationally designed to achieve substantial, perhaps even indenite,
transition delay in hydrodynamic ows. That recent shift from random to rational
search for the right kind of coating is not unlike the great paradigm change in syn-
thetic chemistry that took place near the beginning of the twentieth century. Increased
understanding of the molecular geometry of organic compounds changed the scene
from a hapless alchemist muddling around hoping to chance the right combination
of ingredients, heat, pressure, and catalysts to produce something useful, to a pro-
fessional chemist guring out what he or she wants and working backward from the
shape of a desired molecule for, say, a synthetic hormone. In fact, the present anal-
ogy is apt: the uid dynamist working with the NavierStokes equations can tell the
chemist the exact properties of the compliant coating to be synthesized to achieve a
given goal. If the needed molecular structure is too complicated, futuristic nanoscale
machines can assemble the required molecules directly, element by element.
CHAPTER EIGHT
Separation Control
Logical consequences are the scarecrows of fools and the beacons of wise
men.
(Thomas Henry Huxley, 18251895)
Aristotle discovered all the half-truths which were necessary to the
creation of science.
(Alfred North Whitehead, 18611947)
PROLOGUE
Under certain conditions, wall-bounded ows separate. To improve the performance
of natural or man-made ow systems, it may be benecial to delay or advance this
detachment process. This chapter reviews the status and outlook of separation control
for steady and unsteady ows. Passive and active techniques to prevent or to provoke
ow detachment are considered, and suggestions are made for further research.
8.1 Introduction
8.1.1 The Phenomenon of Separation
Fluid particles in a boundary layer are slowed down by wall friction. If the ow
is sufciently retarded, for example, owing to the presence of an adverse pressure
gradient, the momentum of those particles will be reduced by both the wall shear
and the pressure gradient. In terms of energy principles, the kinetic energy gained at
the expense of potential energy in the favorable-pressure-gradient region is depleted
by viscous effects within the boundary layer. In the adverse-pressure-gradient region,
the remaining kinetic energy is converted to potential energy but is too small to
surmount the pressure hill, and the motion of near-wall uid particles is eventually
arrested. At some point (or line), the viscous layer departs or breaks away from the
bounding surface. The surface streamline nearest to the wall leaves the body at this
point, and the boundary layer is said to separate (Maskell 1955). At separation,
the rotational ow region next to the wall abruptly thickens, the normal velocity
component increases, and the boundary-layer approximations are no longer valid.
150
Because of the large energy losses associated with boundary-layer separation, the
performance of many practical devices is often limited by the separation location. For
example, if separation is postponed, the pressure drag of a bluff body is decreased,
the circulation and hence the lift of an airfoil at high angle of attack is enhanced,
and the pressure recovery of a diffuser is improved. On the other hand, the high-
lift capabilities of delta wings are achieved by provoking separation and forming
leading-edge vortices.
Great strides have been made in the past fewdecades in establishing a rmanalyti-
cal foundation for steady, two-dimensional separation. On the other hand, theoretical
or numerical analysis of three-dimensional or unsteady separation is considerably less
developed, and progress to date in these areas has depended crucially on experimental
work. From a practical point of view, the local separation over, say, a lifting surface
has a very strong effect on the global aerodynamic properties, and a thorough, fun-
damental understanding of the phenomenon is obviously needed. Separation on such
a surface occurs just prior to or at maximum loading, thus greatly inuencing the
devices optimum performance.
The breakthrough in unsteady separation research was achieved by Moore, Rott,
and Sears during the 1950s. Before their work, it was believed that steady and un-
steady separations have the same characteristics; namely, the point of vanishing wall
shear, the termination of the boundary layer, and the beginning of the wake or bub-
ble of separated uid. Rott (1956), in analyzing the unsteady ow in the vicinity of
a stagnation point, noted that the point of vanishing wall shear does not coincide
with the point of boundary-layer detachment. In the same year, Sears made the as-
sumption that the unsteady separation point is characterized by the simultaneous
vanishing of the shear and the velocity at a point within the boundary layer as seen
by an observer in a coordinate convected with the separation velocity. Two years
later, Moore (1958), while investigating a steady ow over a moving wall, arrived at
the analogous model to that for unsteady separation. He stated that for a slowly
moving wall separation occurs when, at some point in the boundary layer, the prole
velocity and shear simultaneously vanish.
8.1.2 Separation Control
Of all the various types of shear ow control now extant, control of ow separa-
tion, historically referred to as boundary-layer control or BLC, is probably the oldest
and most economically important. Separation control is of immense importance to
the performance of air, land, and sea vehicles; turbomachines; diffusers; and a variety
of other technologically important systems involving uid ow. Generally it is desired
to postpone separation so that form drag is reduced, stall is delayed, lift is enhanced,
and pressure recovery is improved. However, in some instances it may be benecial to
provoke separation. For example, to improve the subsonic high-lift performance of
an airfoil optimized for supersonic ight, a ap may be used to initiate leading-edge
separation followed by reattachment.
Flowseparation causes signicant deviations frominviscid pressure distributions.
Such deviations can be either favorable, as in the vortex lift associated with lee
surface separation on swept leading edges and the use of spoilers to obviate ground-
effect-lift during landing or, more commonly, detrimental, resulting, for example, in
high form drag and reduced diffuser efciency. Typical applications and benets of
ow separation control include effective low-Reynolds-number airfoils for remotely
152 SEPARATION CONTROL
piloted vehicles (RPVs), propellers, windmills, helicopters, and so forth; efcient
inlets and diffusers; improved axial ow compressors; increased C
L
max
for greater
payload, reduced engine power and noise at takeoff, shorter runways and reduced
approach speed; supermaneuverability or birdlike ight; efcient and effective stall
or spin control; reduced drag on missiles, automobiles, ships, and helicopters; and
a myriad of applications in industrial aero- and hydrodynamics. As examples of
estimated benets, a 5% improvement in landing C
L
max
would allow a 25% payload
increase (Butter 1984); mitigation of military stall and spin accidents, which from
1977 to 1986 involved over 150 aircraft; application of ow separation control to
tractor-trailer truck drag reduction, which could save in excess of 50 million barrels of
oil per year (Muirhead and Saltzman 1979); and diffuser suction in supersonic closed-
circuit wind tunnels, which would result in an estimated 3050% power reduction
on a device that can draw in excess of 20 MW (Bushnell and Trimpi 1986).
8.1.3 Method of Control
Flowseparation control is currently employed, for example, via vortex generators
on the wings of most Boeing aircraft; via blown aps on older generation supersonic
ghters or leading-edge extensions and strakes on newer generations; and via passive
bleed in the inlets of supersonic engines on, for example, the SR71 and Concorde.
Future possibilities for aeronautical applications of ow separation control include
providing structurally efcient alternatives to aps or slats; cruise application on con-
ventional takeoff and landing aircraft (CTOL), including BLC on thick spanloader
wings (Smith and Thelander 1973); and cruise application on high-speed civil trans-
ports (HSCT) for favorable interference-wave drag reduction, increased leading-edge
thrust, and enhanced fuselage and upper surface lift (Bushnell 1990). In fact, much
of the remaining gains to be made in aerodynamics appear to involve various types of
ow control, including separated ow control (e.g., Hazen 1984; Chambers 1986).
Typical, in some cases serious, problems associated with ow-separation con-
trol include parasitic device drag or energy consumption; system weight, volume,
complexity, reliability or cost; performance sensitivity to body attitude or orienta-
tion; and, especially in the case of the automobile, styling (Poisson-Quinton 1950;
Silhanek 1969; Decken 1971; Sovran, Morel, and Mason 1978). The status of ow-
separation control is still typied by two comments made by Darby (1954): The
Germans were experimenting with BLC in the wind tunnel in the 1920sNACA be-
gan wind-tunnel tests in the late 1920s. It was not until 1949 that American agencies
became interestedthe skepticism seems to be breaking down at long last, but any
branch of technology which has gone so long without bearing fruit certainly requires
a close scrutiny. There are an almost bewildering number of systems, at least for
separation control, which work.
In general, the eld of ow-separation control is far richer than the conventional
view, which usually considers only suction, injection, and vortex generators. Many
decades of research have proven that separation control, in most of its guises, will
work. The task of researchers and designers nowis to improve reliability, where neces-
sary, and increase net gains through innovation. This richness of existing approaches
for ow-separation control is mirrored in the extensive literature partially available
for low-speed ows in the publications by Lachmann (1961), Colin and Williams
(1971), Chang (1976), Adkins (1977), Gad-el-Hak (1989, 1990), Lin, Howard,
Bushnell, and Selby (1990), and Gad-el-Hak and Bushnell (1991a,b) and references
8.2 STEADY AND UNSTEADY SEPARATION 153
therein as well as herein. For high-speed ows, excellent reviews for separation
control in shock-boundary-layer interactions are provided by Delery (1985) and
Viswanath (1988).
Given an imposed pressure eld, the kernel problem in separation postponement
is to add momentumto the very near-wall region of the owby either transferring mo-
mentumfromowregions farther fromthe wall, which are still momentumrich, or by
direct addition of power drawn fromthe propulsive system. Probably the most popu-
lar ow-separation control technique has been to add momentumto the near-wall re-
gion either actively (e.g., tangential blowing or wall jets) or passively (e.g., boundary-
layer tripping, turbulence enhancement, or vortex generators of various scales).
8.1.4 Outline
The objective of this chapter is to summarize the status and outlook for ow
separation control with an emphasis on more recent developments such as opti-
mized vortex generators; various methods of turbulence control, including dynamic
inputs; and wall-heat transfer. Direct comparisons between the various approaches
on common test problems will be made. The recent advances in computational uid
dynamics are allowing the transformation of much of BLC from an empirical art to
a predictive science, and this subject will be briey discussed. Owing to space limita-
tions, the presentation does not include large-scale-provoked separation or control
of vortices (e.g., Williams and Amato 1989; Bushnell and Donaldson 1990; Panton
1990c; Williams and Papazian 1991) nor the related problem of asymmetric nose
body vortex control on missiles and similar vehicles at high angles of attack (Vakili
1990). However, control by vortices is included in this chapter. Separation control
is addressed for both nominally two-dimensional and three-dimensional ows, al-
though the knowledge base concerning the latter is noticeably decient. Typically,
three-dimensional separation occurs sooner but is less catastrophic than the quasi-
two-dimensional case (Driver 1989).
In this chapter, we review methods to control separation for steady and unsteady
boundary layers. Both passive and active techniques to prevent or to provoke ow
detachment are considered. For historical reasons, most of the separation control
methods discussed here were developed for aircraft wings. The results, however,
pertain to fundamental properties of uid ow and could readily be extended to a
variety of systems such as diffusers, steam turbine blades, wind turbine rotors, pump
impellers, and off-shore structure components. To set the appropriate mechanistic
framework for discussing the variety of control tools available, differences between
steady and unsteady separations are rst considered.
8.2 Steady and Unsteady Separation
8.2.1 Steady Separation
Separation of a steady, two-dimensional boundary layer was explained rst by
Prandtl (1904) in his milestone presentation
Uber Fl ussigkeitsbewegung bei sehr

kleiner Reibung in which he introduced the boundary-layer theory. Fluid particles
near the surface are retarded by the friction of the wall and by any adverse pressure
gradient present in the freestream. If the near-wall uid has insufcient momentum
for it to continue its motion, it will be brought to rest at the separation point (line).
Further downstream, the adverse pressure forces will cause reverse ow. Because the
velocity at the wall is always zero, the gradient [u/y]
y=0
must be positive upstream
of separation, zero at the point of separation, and negative in the reverse-owregion.
For an axially symmetric ow, the line of separation becomes a circle, and the point
of vanishing shear still coincides with the point of separation.
In Prandtls view, the separation point is entirely determined by external condi-
tions. Boundary-layer separation is accompanied by a thickening of the rotational
ow region and ejection of vorticity. Downstream of the separation point the shear
layer either passes over the region of recirculating uid and reattaches to the body
surface or forms a wake and never reattaches to the body. The characteristic dimen-
sion of the recirculating region is quite large in the latter case and is of the order of
the body height.
Analytically, the solution of the steady, two-dimensional, laminar boundary-
layer equations with a prescribed external-pressure (or external-velocity) distribu-
tion breaks down at the point of separation, and this is commonly known as the
Goldsteins singularity, in honor of Sydney Goldstein, who, in 1948, rst noted the
singular behavior of the solution near a point of zero skin friction. (For a review
of other singularities occurring in the equations for three-dimensional or unsteady
boundary-layer ows, or both, see the article by Williams 1985.) The singularity
of the boundary-layer equations at separation is obviously not a physical property
of the ow and can be overcome by prescribing either the displacement thickness or
the wall-shear distribution instead of the external pressure. This kind of analysis is
termed inverse calculation.
8.2.2 Unsteady Separation
For two-dimensional ow over moving walls, two-dimensional unsteady ows,
and three-dimensional steady and unsteady ows the point (line) of vanishing wall
shear does not necessarily coincide with separation, and this greatly complicates the
problem. This was rst observed by Rott (1956) while analyzing the unsteady
ow in the vicinity of a stagnation point. He observed that, although the wall shear
vanished with an accompanying reverse ow, there was no singularity or breakdown
of the boundary-layer assumptions. In seeking a generalized model for separation,
Sears (1956) postulated that the unsteady separation point is characterized by the
simultaneous vanishing of the shear and the velocity at a point within the boundary
later as seen by an observer moving with the velocity of the separation point.
Moore (1958), while investigating a steady owover a moving wall, arrived at the
same model for unsteady separation. On the basis of an intuitive relationship between
steady ow over a moving wall and unsteady ow over a xed wall, Moore was able
to sketch the expected velocity proles for both cases, as seen in Figures 8.1 and 8.2.
He considered the possibility that a Goldstein-type singularity occurs at the location
where the velocity prole simultaneously has zero velocity and shear at a point above
the moving wall. Equivalently, for unsteady separation on a xed wall, the separation
point is the location at which both the shear and velocity vanish in a singular fashion
in a frame of reference moving with the separation point. The main drawback of
this MooreRottSears (MRS) model in the xed-wall case is that the speed of the
separation point is not known a priori, making it difcult to locate this point and
forcing researchers to rely on more qualitative measures for unsteady separation.
Sears and Telionis (1972a, 1975) were the rst to prove that the boundary-layer
Figure 8.1: Streamlines and velocity proles when a boundary
layer over a xed or moving wall separates. (a) Fixed wall;
(b) Wall moving downstream; (c) Wall moving upstream.
Figure 8.2: Moores (1958) postulated velocity
proles for unsteady separation on a xed wall.
() Moving separation point; () xed
separation point.
separation singularity accompanies detachment of unsteady ows. They suggested
that the existence of such singularity could serve as a criterion for unsteady sep-
aration when numerically integrating the
boundary-layer equations.
Because of the preceding difculty,
early attempts to verify this important
MRS model of unsteady separation consid-
ered the more tractable problem of steady
separation over moving walls. Both Vidal
(1959) and Ludwig (1964) experimentally
investigated a shrouded rotating cylinder
in steady ow. As expected, separation
was delayed (moved downstream) when
the wall moved in the freestream direction
and was advanced when the wall moved
opposite the main ow. The measured
velocity proles corresponded to those
hypothesized by Moore (1958). Telionis and Werle (1973) presented an analytical
verication of the separation model for the case of a downstream moving wall.
Until the work of Williams and Johnson (1974a,b), the relationship between un-
steady boundary-layer separation over xed walls and steady separation over moving
walls has been an intuitive one. By considering the rather general class of unsteady,
two-dimensional boundary-layer problems that could be treated by the method of
semisimilar solutions, Williams and Johnson (1974a) were able to transform an un-
steady problem in three independent variables into an equivalent problem in two
independent variables. This transformation then made it possible to use conven-
tional numerical techniques for solving steady, nonsimilar boundary-layer problems
and allowed the authors to investigate some time-dependent ows with separation
occurring in a coordinate system for which unsteady separation is most easily identi-
ed and analyzed. As a practical example, Williams and Johnson (1974a) veried the
MRS model for unsteady separation for a particular time-dependent retarded ow,
the steady ow equivalent of which is the classical linearly retarded ow studied by
Howarth (1938).
In a subsequent article, Williams and Johnson (1974b) established a rigorous ana-
lytical link between unsteady separation over a xed wall and steady separation over
a moving wall for the special case in which the external velocity distribution in the
xed coordinate system is a function only of a linear combination of the streamwise
coordinate and time and in which the wall moves downstream with the (constant)
speed of the unsteady separation point. Once more, Williams and Johnson were able
to transform a given unsteady ow into a steady ow over a wall moving with the
speed of the separation point and then to relate this back to the unsteady ow.
Archetypes of Unsteady Separation
There are two distinctly different kinds of unsteady separation. On a body oscil-
lating in pitch, the pressure gradient may vary both in magnitude and form. On the
other hand, if the freestream velocity changes periodically with time, the impressed
(mean) pressure gradient over a xed surface varies in magnitude but not in form
during the oscillation. Detailed velocity measurements in the rst kind are scarce
because of the obvious difculties involved in instrumenting a moving body. Accord-
ingly, the exact nature of unsteady separation for this case remains unclear despite
the abundance of qualitative ow visualization data (Gad-el-Hak 1987b).
Consider rst the case of an airfoil oscillating in pitch. Sudden changes in lift,
drag, and pitching moment occur near the onset of separation. The dynamic stall
is characterized by two distinctly different ow phenomena (Ericsson 1971, 1988;
Ericsson and Reding 1971, 1987). The rst is quasi-steady and is due to time-lag
and boundary-layer improvement effects. The second phenomenon is transient and
concerns the effects of the forward movement of the separation point. As the airfoil
leading edge moves upward, the boundary layer between the stagnation and separa-
tion points experiences a moving wallwall jet effect very similar to that observed on
a rotating cylinder (Ericsson and Reding 1984, 1986, 1987; Ericsson 1988). Thus,
the boundary layer has a fuller velocity prole as compared with the steady case
and is therefore more resistant to separation. On the downstroke the effect is the
opposite, promoting separation. At stall, a vortex is shed from the leading edge and
is convected downstream over the chord. For high reduced frequency, this spilled
vortex remains over the airfoil a signicant portion of the cycle, and extra lift is
produced.
Figure 8.3: Envelopes of velocity proles for periodic os-
cillations. The dashed lines and the arrows indicate the ex-
treme positions of the tangents to the proles at the wall.
Schematic representation of Despard and Millers (1971)
data.
The relatively simpler second kind of unsteady separation was investigated by
Despard and Miller (1971), who performed a series of experiments to measure the
instantaneous velocity proles in oscillating boundary layers subjected to adverse
pressure gradients. An array of 10 hot-wire probes was used to measure the normal
proles of streamwise velocity at several points along the rear portion of an airfoil-
like body placed in a low-speed wind tunnel. The freestream velocity was varied
periodically by a set of rotating shutter valves located downstream of the test sec-
tion. The resulting envelopes of the velocity proles throughout the entire cycle of
oscillation are shown schematically in Figure 8.3 for four different stations along the
body. Within half a boundary layer thickness of the beginning of the region of wake
formation, the point S in the gure exhibits reverse ow or zero velocity gradient
at the wall for the entire cycle of proles. This led Despard and Miller to propose a
practical denition of unsteady separation as the farthest upstream point at which
there is zero velocity or reverse ow at some point in the velocity prole throughout
the entire oscillation cycle. This approach avoids the difculty of determining the
speed of the separation point as required by the MRS model. Point Z in Figure 8.3
is the farthest upstream point at which the wall shear stress passes through zero dur-
ing a complete cycle of oscillation. Therefore, the point of vanishing shear oscillates
between S and Z, but the outer ow remains attached until S. Thus, Despard and
Millers results provided the rst experimental evidence that a thin layer of reversed
owcan be embedded at the bottomof an otherwise attached boundary layer, as pos-
tulated by Moore (1958). Despard and Miller further concluded that the presence
of an oscillating velocity component and a time-dependent adverse pressure gradient
cause the separation point to be displaced upstream of its steady-ow position. The
distance between the two points diminishes as the frequency of oscillation increases
but is rather insensitive to variations in the amplitude of the outer ow oscillations.
Shortly after this important experiment, Sears and Telionis (1975) provided a
review of the numerical evidence for the validity of the MRS model. Numerical
integrations of unsteady laminar and turbulent boundary layers showed no evidence
of the Goldstein singularity, even in a region of partially reversed ow and through
the point of zero skin friction. The singularity did appear, however, at the point of
unsteady separation, as dened by the MRS criterion.
Some controversy still exists over the precise denition of unsteady separation.
Telionis (1979) reiterated that separation means the location on the solid boundary
where the ow stops creeping over the skin of the body and breaks away from the
wall, thus generating a turbulent wake. Sears and Telionis (1972a,b) argued that
abrupt changes of boundary-layer properties in the rst-order boundary-layer equa-
tions may signal the approach of the point of separation or of the Goldstein singular-
ity. Despard and Miller (1971), on the other hand, dened separation as the farthest
upstream station at which the shear uctuates between zero and some negative value
throughout an entire cycle of freestream oscillation. Tsahalis and Telionis (1974)
numerically studied the unsteady separation in an oscillating ow and found that
Despard and Millers criterion could be veried under certain but not all situations.
Turbulent Boundary Layers
The carefully executed, well-documented experiments of Simpson and his col-
leagues (Shiloh, Shivaprasad, and Simpson 1981; Simpson, Chew, and Shivaprasad
1981a,b, 1983; Simpson and Shivaprasad 1983) are very useful in understanding
some of the basics of steady and unsteady separation in turbulent ows. Simpson
et al. investigated the structure of separating, nominally two-dimensional, turbulent
boundary layers in a steady freestream (Shiloh et al. 1981; Simpson et al. 1981a,b)
as well as in a sinusoidally oscillating freestream (Simpson et al. 1983; Simpson and
Shivaprasad 1983). Their experiments illustrate both the similarities and differences
between steady and unsteady separations. Upstream of where intermittent backow
begins, the owbehaves in a quasi-steady manner. Downstream, there are non-quasi-
steady effects on the ensemble-averaged ow structure. Also, the hysteresis effects in
the ensemble-averaged velocity proles are more pronounced for higher-reduced-
frequency freestream oscillations. Simpson (1989) provides a comprehensive review
of the topic of turbulent ow separation.
8.2.3 Vortex-Induced Separation
Separation can be triggered experimentally with concentrated vortices moving
close to a wall. Harvey and Perry (1971) placed a half-span rectangular wing in a
wind tunnel and observed the evolution of the resulting single tip vortex as it passes
over a downstream moving oor (Figure 8.4). Total-head surveys in planes across
the ow were conducted to observe separation and the formation of a secondary
vortex of opposite sense to the main one. The secondary vortex causes a rebounding
of the trailing vortex, as sketched in Figure 8.5. Similar effects are present when a
ring vortex approaches a wall normal to its propagation direction (Magarvey and
McLatchy 1964; Schneider 1980). In the latter case, a downstream-moving, unsteady
separation is produced.
Ho (1986) attributed the sparseness of a positive experimental proof (or disproof)
of the MRS criterion for unsteady separation to the difculty of measuring the sepa-
rated velocity near the wall. The temporal variation of the resulting three-dimensional
velocity eld necessitates the use of a large number of small, fast-response probes and
a colossal data acquisition and storage system. This problem is particularly acute for
an upstream moving separation in which ow reversal occurs, leading to additional
measuring difculties. No ow reversal occurs in a downstream moving separation,
and hot-wire probes can then be used to survey the oweld in the vicinity of the
wall.
Figure 8.4: Surveys of total head when a tip vortex passes over a down-
stream moving wall [after Harvey and Perry 1971].
Detailed measurements of the downstream-moving separation caused by periodic
ring vortices impinging normally onto a at plate were conducted by Didden and Ho
(1985). They forced an axisymmetric jet to produce a train of primary ring vortices
in the air-jet shear layer sketched at the top of Figure 8.6. The phase-locked ow
visualization pictures obtained by Didden and Ho using smoke clearly show that the
wall-jet boundary layer evolves into a secondary vortex that is counterrotating with
respect to the primary vortex and periodically separates. The unsteady separation
is induced by the primary vortex and moves downstream in the radial mean-ow
direction, thus allowing detailed measurements with hot-wire probes. The phase-
averaged measurements of normal and radial velocity components provide data for
locating the onset of separation in space and time.
Figure 8.5: Suggested interpretation of the
total-head surveys. (a) Asection downstream
of the initial separation; (b) Subsequent de-
velopment of the secondary vortex [after
Harvey and Perry 1971].
Like all other cases discussed thus far, the unsteady separation phenomenon
observed in the impinging jet ow involves a strong viscousinviscid interaction. The
detailed data obtained by Didden and Ho (1985) reveal the following sequence of
events. The primary ring vortex in the inviscid
region produces a fast-moving stream while it
approaches the wall. The resulting accelerat-
ing ow causes a negative wall pressure in the
low-pressure region downstream from the jet
axis. Further downstream, pressure recovery
results in an unsteady adverse pressure gradi-
ent, which retards the owin the viscous layer.
Hence, a strong shear layer is generated at
the viscousinviscid interface. The instability
of this shear layer leads to the rollup of the
vortex sheet and the formation of a secondary
vortex. The ejection of this secondary vortex,
believed to be associated with the onset of un-
steady separation, leads to an abrupt increase
in the momentumthickness and the formation
of a bulge that convects downstream at the
same speed as the zero-shear-stress point of
the radial velocity prole. In other words, the
unsteady separation originates from a local
Figure 8.6: Forced axisymmetric jet impinging normally onto a
at plate. Phase-averaged pressure traces at the instant of sepa-
ration (t/T = 0.15) and after a short time interval (t/T =0.3)
[after Didden and Ho 1985].
shear layer, which is initiated in the viscous region by the unsteady adverse pressure
gradient produced by the primary vortex.
At the instant of apparent separation, a sharp, high-amplitude suction peak occurs
in the surface pressure signature of the secondary vortex, as shown with the broken
line in Figure 8.6 representing the (normalized) phase-averaged surface pressure trace
at t/T = 0.15, where t is time and T is the period of the forcing signal. After the
passing of both the primary and secondary vortices, the boundary layer reattaches,
and a high-level pressure is found in the wake of separation, as shown with the solid
line in the same gure a short time after separation (t/T = 0.3).
Given the appearance of a sharp, negative peak pressure at about the same time as
the secondary vortex and because the relatively more gentle negative pressure induced
by the primary vortex as it passed very nearly parallel to the wall was not expected
to change, Didden and Ho (1985) conjectured that the observed high-amplitude
pressure uctuations are caused by the unsteady separation rather than by the primary
vortex. They reiterated that, although viscosity does not play an important role in
the stability of the spatially developing local shear layer, viscous effects are essential
for producing the shear layer before the onset of the unsteady separation.
Because a local shear layer has been observed in many other separated ows, Ho
(1986) contends that such a shear layer is a generic owmodule to all separated ows.
This interesting idea of associating unsteady separation with a spatially developing
local shear layer allows researchers in the subject area to borrow from the extensive
knowledge available in free-shear-layer studies (Ho and Huang 1982; Ho and Huerre
1984). Many examples of this association are provided in the review article by Gad-
el-Hak (1987b).
The qualitative criteria of separation, namely, the thickening of the boundary
layer, the ejection of vorticity, and the vanishing of the wall shear, do not appear at the
8.3 EQUATIONS OF MOTION AT THE WALL 161
same time or convect at the same speed in the unsteady case. The MRS criterion for un-
steady separation is a quantitative one. However, as mentioned earlier, the convection
speed of the separation point is not known a priori, and this greatly increases the dif-
culty of the problem. By choosing an appropriate frame of reference based on a well-
documented oweld, Didden and Ho (1985) were able to provide positive proof of
the MRS criterion for a downstream-moving separation. Moreover, at the separation
position and in a frame moving at the local velocity of the point of zero shear stress
(or the convection speed of the boundary layers bulge), the phase-averaged stream-
lines were found to form a closed recirculating region in qualitative agreement with
the calculations reported by Walker (1978) for a line vortex near a wall.
8.3 Equations of Motion at the Wall
To examine the local and instantaneous effects of many of the separation control tools
to be described in this chapter, we need to bring to the forefront three equations from
Chapter 2. Recall the NavierStokes equations written at the wall from Section 2.4.
In wall-bounded ows, the streamwise (x) and spanwise (z) momentum equations
written at the wall give very useful expressions of the wall uxes of, respectively, the
spanwise and streamwise vorticity. For an incompressible uid over a nonmoving
wall of small curvature and with body forces neglected, the streamwise, normal, and
spanwise momentum equations, written at y = 0, read
v
w
u
y
y=0
+
p
x
y=0
y=0
u
y
y=0
=
2
u
y
2
y=0
(8.1)
0 +
p
y
y=0
0 =
2
v
y
2
y=0
(8.2)
v
w
w
y
y=0
+
p
z
y=0
y=0
w
y
y=0
=
2
w
y
2
y=0
(8.3)
The preceding equations are instantaneous and are valid for both laminar and
turbulent ows. Here, v
w
is the (positive) injection or (negative) suction velocity
through the wall. Upward or downward motion of the wall acts analogously to
injection or suction. The right-hand side of each of the three equations represents the
curvature of the corresponding velocity prole, which in the case of the streamwise
spanwise equation is the same as the ux of spanwisestreamwise vorticity from
the wall. Negative curvature implies that the velocity prole is fuller and that the
wall is a source (positive ux) of vorticity. Whether the wall is a source or a sink of
spanwisestreamwise vorticity depends on whether uid is sucked or injected from
the wall, whether the streamwisespanwise pressure gradient is favorable or adverse,
and whether the wall viscosity is lower or higher than that above the surface.
8.3.1 Equation for Turbulent Flows
For a turbulent ow, Eqs. (8.1)(8.3) are still valid instantaneously. For a canon-
ical turbulent ow, the instantaneous velocity proles in all three directions change
shape continuously as a result of the random changes in the pressure eld. But what
about the mean quantities? Recall fromSection 2.5 the equation for the mean momen-
tumfor a Newtonian, incompressible turbulent ow. Let u
i
= U
i
+u
i
and p = P+p
,
where U
i
and P are ensemble averages for the velocity and pressure, respectively, and
u
i
and p
are the velocity and pressure uctuations about the respective averages. The
continuity and momentumequations governing the mean velocity and mean pressure
for an incompressible turbulent ow become
U
k
x
k
= 0 (8.4)
U
i
t
+U
k
U
i
x
k
=
P
x
i
+

x
k
U
i
x
k
u
i
u
k
+ g
i
(8.5)
where, for clarity, the primes have been dropped fromthe uctuating velocity compo-
nents u
i
and u
k
. At the wall, the equation for the streamwise component of momentum
reads
v
w
U
y
y=0
+
P
x
y=0
d
dT
T
y
y=0
U
y
y=0
+
uv
y
y=0
=
2
U
y
2
y=0
(8.6)
where T is the mean temperature eld, and viscosity is assumed to change as a result
of surface heat transfer. The right-hand side of Eq. (8.6) is the ux of mean spanwise
vorticity,
z
= U/y, at the surface. In the absence of wall transpiration, mean
pressure gradient, and surface heating or cooling, the rst three terms on the left-
hand side of Equation (8.6) vanish. The fourth term on the left-hand side is the slope
of the normal prole of uv at y = 0. This term could be asymptotically estimated as
the wall is approached. Consider a Taylors series expansion in powers of y in the
neighborhood of the point y = 0. As a result of the no-slip condition, the streamwise
velocity uctuations u varies linearly with y. To conserve mass, the normal velocity
uctuations must vary as y
2
. It follows then that very near the wall (within the
viscous sublayer), the tangential Reynolds stress uv varies at least as y
3
and that
uv/y varies as y
2
. At the wall itself, y = 0 and [ uv/y]
y=0
= 0, although close
to the wall the slope of the tangential Reynolds stress prole is quite large.
Therefore, it can be deduced from Eq. (8.6) that the mean streamwise velocity
prole for the canonical turbulent boundary layer (two-dimensional, incompressible,
isothermal, zero-pressure-gradient, over an impervious, rigid surface) will have a zero
curvature at the wall. Notwithstanding this common characteristic with the Blasius
boundary layer (Figure 6.2b of Chapter 6), the turbulent boundary layer is quite
different fromthe laminar one. As pointed out by Lighthill (1963a), turbulent mixing
concentrates most of the mean vorticity much closer to the wall as compared with
the laminar case. The mean vorticity at the wall, [U/y]
y=0
, is typically an order of
magnitude larger than that in the laminar case. This explains the higher skin-friction
drag associated with a turbulent ow. Turbulent mixing also causes the mean vorticity
to migrate away fromthe wall, and about 5%of the total is found much farther from
the surface. The ux of mean spanwise vorticity is zero at the wall itself but very large
close to it, reaching a maximum at about the same location where the root-mean-
square vorticity uctuations peak (near the edge of the viscous sublayer). This trait
is responsible for the turbulent boundary layers resistance to separation.
In the next six sections, available and contemplated ow-control methods to delay
or to provoke separation will be discussed. The equations developed in this section
for laminar and turbulent boundary layers will help in presenting a unied view of
the different control techniques.
8.4 VELOCITY PROFILE MODIFIERS 163
8.4 Velocity Prole Modiers
As mentioned in Section 8.2.1, Prandtl (1904) was the rst to explain the mechanics of
separation. He provided a precise criterion for its onset for the case of a steady, two-
dimensional boundary layer developing over a xed wall. If such a ow is retarded,
the near-wall uid may have insufcient momentum to continue its motion and will
be brought to rest at the point of separation. Fluid particles behind this point move
in a direction opposite to the external stream, and the original boundary-layer uid
passes over a region of recirculating ow. Because the velocity at the wall is always
zero, the gradient [u/y]
y=0
will be positive upstream of separation, zero at the
point of separation, and negative in the reverse-ow region. The velocity prole at
separation must then have a positive curvature at the wall. However, [
2
u/y
2
]
y=0
is negative at a large distance from the wall, which means the velocity prole at
separation must have a point of inection somewhere above the wall, as shown in
Figure 6.2d. Because [
2
u/y
2
]
y=0
> 0 is a necessary condition for a steady, two-
dimensional boundary layer to separate, the opposite (i.e., a negative curvature of
the velocity prole at the wall) must be a sufcient condition for the boundary-layer
ow to remain attached.
The preceding arguments naturally lead to several possible methods of control
to delay (or advance) steady, two-dimensional separation that rely on modifying the
shape of the velocity prole near the wall. Namely, the object is to keep [
2
u/y
2
]
y=0
as negative as possible, or in other words to make the velocity prole as full as
possible (Figure 6.2a). In this case, the magnitude of the spanwise vorticity decreases
monotonically away from the wall, and the surface vorticity ux is in the positive
y direction. Not surprisingly, then, methods of control to postpone separation that
rely on changing the velocity prole are similar to those used to delay laminar-to-
turbulence transition (Chapter 6).
Inspection of the streamwise momentum equation (8.1) indicates that separation
control methods may include the use of wall suction (v
w
<0), favorable pressure
gradient (p/x < 0), or lower wall viscosity (/y > 0). Obviously, any one or a
combination of these methods may be used in a particular situation. For example,
beyond the point of minimum pressure on a streamlined body the pressure gradient
is adverse and the boundary layer will separate if the pressure rise is sufciently steep;
however, enough suction may be applied there to overcome the retarding effects of
the adverse pressure gradient and to prevent separation. Each of these velocity prole
modiers is covered in more detail in the following three subsections.
8.4.1 Shaping
For any two-dimensional, subsonic owover a closed-surface body, adverse pres-
sure gradient always occurs somewhere in the aft region. Streamlining can greatly
reduce the steepness of the pressure rise, leading to the prevention or postponement
of separation. Numerous biological species are endowed with body shapes that avoid
separation and allow for minimum uid resistance to their motion in air or water.
Archaic Homo sapiens discovered, through scores of trials and errors, the value of
streamlining spears, sickle-shaped boomerangs, and n-stabilized arrows (Williams
1987). In supersonic ows, pressure always rises across a shock wave, and a boundary
layer may separate as a result of the wave interaction with the viscous ow (Young
1953; Lange 1954).
Laminar boundary layers can only support very small adverse pressure gradients
without separation. In fact, if the ambient incompressible uid decelerates in the
streamwise direction faster than U
o
x
0.09
, the ow separates (Schlichting 1979).
On the other hand, a turbulent boundary layer, being an excellent momentum con-
ductor, is capable of overcoming much larger adverse pressure gradients without
separation. In this case, separation is avoided for external ow deceleration up to
U
o
x
0.23
(Schlichting 1979). The efcient momentum transport that characterizes
turbulent ows provides the mechanismfor mixing the slower uid near the wall with
the faster uid particles farther out. The forward movement of the boundary-layer
uid against pressure and viscous forces is facilitated, and separation is thus post-
poned. According to the experimental results of Schubauer and Spangenberg (1960),
a larger total pressure increase without separation is possible in the case of a turbu-
lent ow by having larger adverse pressure gradient in the beginning and continuing
at a progressively reduced rate of increase.
To expand the attached ow operational envelope to off-design conditions using
the concept of pressure gradient mitigation generally requires some form of variable
geometry such as vanes, slats, or aps. These can be combined with other separation
control techniques such as active or passive blowing, a rotating cylinder at the ap
knee, or use of an injection-stabilized trapped vortex. For low-speed ows, bodies
can now be designed for incipiently separated ow over large surface areas using
the so-called Stradford closure to be discussed in the following subsubsection (Smith
1977; Smith, Stokes, and Lee 1981). Although it minimizes skin-friction drag, this
approach exacerbates the attached-ow, viscous-induced form or pressure drag, and
the minimum drag body is actually a less extreme design. Also, such bodies tend to
generate large separated ow regions off-design, and therefore some form of standby
ow separation control would probably be required to ensure reasonable off-design
performance.
Skin Friction
The skin friction downstream of the separation line is negative. However, the
increase in pressure drag that results from ow detachment is far greater than the
saving in skin-friction drag. The articles by Stradford (1959a,b) provide useful dis-
cussion on the prediction of turbulent-boundary-layer separation and the concept of
ow with continuously zero skin-friction throughout its region of pressure rise. By
specifying that the turbulent boundary layer be just at the condition of separation,
without actually separating, at all positions in the pressure rise region, Stradford
(1959b) experimentally veried that such ows achieve a specied pressure rise in
the shortest possible distance and with the least possible dissipation of energy. A lift-
ing surface that could utilize the Stradford distribution immediately after transition
from laminar to turbulent ow would be expected to have very low skin-friction
as well as pressure drag. Liebeck (1978) successfully followed this strategy using a
highly polished wing to achieve the best lift-to-drag ratio (over 200) of any airfoil
tested in the low-Reynolds-number range of 5 10
5
2 10
6
. He argued that the
entire pressure-recovery region of an airfoils upper surface would be operating at its
maximumcapacity if the adverse pressure distribution were uniformly critically close
to separation. By assuming an incipient-separation, turbulent-ow prole, Liebeck
calculated the pressure eld required and then used an inverse calculation procedure
to derive the airfoil shape from the given critical-velocity distribution.
When one attempts to reduce skin-friction drag by driving the boundary layer
toward separation, a major concern is the ow behavior at off-design conditions,
as discussed earlier. A slight increase in angle of attack, for example, can lead to
separation and consequent large drag increase as well as loss of lift. High performance
airfoils with lift-to-drag ratio of over 100 utilize carefully controlled adverse pressure
gradient to retard the near-wall uid, but their performance deteriorates rapidly
outside a narrow envelope (Carmichael 1974).
Separation Bubbles
Laminar-to-turbulence transition on the upper surface of a lifting surface typically
occurs at the rst onset of adverse pressure gradient if the Reynolds number exceeds
10
6
. The separation-resistant turbulent boundary layer that evolves in the pressure
recovery region results in higher maximumlift and a relatively larger angle of stall. At
lower Reynolds numbers, and depending on the severity of the initial adverse gradient
(hence on the airfoil shape), laminar separation may take place before transition. For
sufciently low Re, the separated owwill not reattach to the surface. However, in the
intermediate Reynolds number range of typically 10
4
10
6
, transition to turbulence
takes place in the free-shear layer owing to its increased susceptibility (Gad-el-Hak
1990). Subsequent turbulent entertainment of high-speed uid causes the ow to
return to the surface, thus forming what is known as a laminar separation bubble,
a topic to be discussed in more detail in the next chapter. Regardless of whether the
ow subsequently reattaches, the laminar separation leads to higher form drag and
lower maximum lift. Delicate contouring of the airfoil near the minimum pressure
point to lessen the severity of the adverse pressure gradient may be used to accomplish
separation-free transition (Pfenninger and Vemuru 1990).
8.4.2 Transpiration
The second method to avert separation by changing the curvature of the velocity
prole at the wall involves withdrawing the near-wall uid through slots or porous
surfaces. Prandtl (1904) applied suction through a spanwise slit on one side of a
circular cylinder. His ow visualization photographs convincingly showed that the
boundary layer adhered to the suction side of the cylinder over a considerably larger
portion of its surface. By removing the decelerated uid particles in the near-wall
region, the velocity gradient at the wall is increased, the curvature of the velocity
prole near the surface becomes more negative, and separation is avoided. In the
following, an approximate method to compute the amount of suction needed to
prevent laminar separation is briey recalled (Prandtl 1935).
For a laminar boundary layer, the ratio of pressure forces to viscous forces is
proportional to the Pohlhausen parameter, , as given by
dp
o
dx

2
U
o
=
dU
o
dx
(8.7)
where is the boundary layer thickness, U
o
is the velocity outside the boundary layer,
is the kinematic viscosity, is the dynamic viscosity, and dp
o
/dx is the streamwise
pressure gradient. At separation, [u/y]
y=0
= 0, and Equation (8.1) reads
dp
o
dx
=
2
u
y
2
y=0
(8.8)
The Pohlhausen parameter at the point of separation of a laminar boundary layer
is = 12, and from Eqs. (8.7) and (8.8) the expressions for the curvature of the
velocity prole at the wall and the boundary-layer thickness become, respectively,
2
u
y
2
y=0
=
12 U
o
2
(8.9)
=
12
dU
o
dx
(8.10)
The velocity distribution dU
o
/dx is determined from the potential ow solution. As
an example, suppose we wish to compute the suction coefcient C
q
|v
w
| /U
o
,
which is just sufcient to prevent laminar separation from the surface of a cylinder.
By assuming that the velocity proles in the vicinity of separation are identical with
that at the point of separation and computing (dU
o
/dx) at the downstreamstagnation
point, Prandtl (1935) used the momentumintegral equation and the preceding results
to make a simple estimate of the required suction
C
q
=4.36 Re
0.5
(8.11)
where Re is the Reynolds number based on the cylinder diameter and the freestream
velocity.
Several researchers have used similar approximate methods to calculate the lam-
inar boundary layer on a body of arbitrary shape with arbitrary suction distribution
(see, e.g., Schlichting and Pechau 1959; Chang 1970). A particularly simple calcula-
tion was made by Truckenbrodt (1956). He reduced the problem to solving a rst-
order ordinary differential equation. As an example, for a symmetrical Zhukovskii
airfoil with uniform suction, Truckenbrodt predicted a suction coefcient just suf-
cient to prevent separation of
C
q
=1.12 Re
0.5
(8.12)
where Re is the Reynolds number based on the airfoil chord and the freestream
velocity.
For turbulent boundary-layers, semi-empirical methods of calculation are in-
evitably used owing to the well-known closure problem. Suction coefcients in the
range of C
q
= 0.0020.004 are sufcient to prevent separation on a typical air-
foil (Schlichting 1959; Schlichting and Pechau 1959). Optimally, the suction should
be concentrated on the low-pressure side of the airfoil just a short distance behind
the nose where, at large angles of attack, the largest local adverse pressure gradient
occurs.
Passive Suction
For the high-speed, shock-boundary-layer interaction case, a passive porous sur-
face can be used to mitigate the local pressure gradients and obviate separation as
well as reduce wave drag and shock losses (Bahi, Ross, and Nagamatsu 1983; Savu
and Trifu 1984; Nagamatsu et al. 1985, 1987; Raghunathan 1985; Stanewsky and
Krogmann 1985; Barnwell et al. 1985; Kovalnogov, Fomin, and Shapovalov 1987)
in both transonic and supersonic ows (Bauer and Hernandez 1988). The basic de-
vice, sketched in Figure 8.7, is an empty subsurface plenum covered by a porous
surface and located underneath the shock-boundary-layer interaction region. Such a
Figure 8.7: Schematic of porous airfoil for passive suction.
passive porous surface allows mass to self-bleed from downstream of the shock to
upstream, resulting in a more gradual viscousinviscid interaction, a series of weaker
shock waves, and reduced pressure gradients. Flow separation is then delayed and
wave drag is minimized. Suction implemented via passive bleed is employed as an
integral part of the technologically important case of high-speed inlet design (Delery
1985; Viswanath 1988).
Suction Optimization
Flow separation control by suction is the other conventional technique which,
along with blowing, is within the capacity of contemporary CFD for design and op-
timization via tailored and perhaps distributed ow proles and is well reviewed in
readily available literature. Suction can be instituted via active or passive systems
with the boundary-layer diverter constituting the reductio ad absurdum. The param-
eter space for separation control by suction (or injection) includes distribution of
mass transfer location vis- ` a-vis adverse pressure gradient regions, spatial distribution
(discrete, continuous), exit orientation, tailoring of orice velocity proles, active or
passive source or sink, and distribution of suction and injection both chordwise and
spanwise.
In summary, studies of suction control at high speeds indicate that suction holes
should probably be inclined in the upstream direction (Purohit 1987), that upstream
control is less effective than direct control of the separated ow region, and that
optimumdistributed suction requires several independent plenumchambers to avoid
local backows and consequent loss of suction control. An additional possibility for
high-speed ows is to utilize the embedded near-wall spanwise vortex structure as-
sociated with swept shocks to remove the inner portion of the incident boundary
layer (S.M. Bogdonoff, private communication). This may require provision of an
additional control shock upstream of the interaction. The increased interest in, and
favorable ight experience with, hybrid laminar ow control (where suction is em-
ployed in the leading-edge region ahead of wing box) has revived interest in suction
control for the subsonic case. Preliminary studies indicate that the LFCsuction system
for cruise application is not incompatible with leading-edge region suction separa-
tion control or high-lift requirements for both CTOL and SST (supersonic transport)
applications. Flight experiments of leading-edge suction systems for high lift (e.g.,
Hunter and Johnson 1954) and LFC (Hefner and Sabo 1987) do not indicate any
stoppers. A combined approach using the same suction surface and system for both
LFC at cruise and leading-edge-region high lift for takeoff and landing has consider-
able promise. For LFC, there is the added benet of eliminating the joints and other
factors associated with conventional leading-edge, variable-geometry devices.
8.4.3 Wall Heat Transfer
The third term in Eq. (8.1) points to yet another method to delay boundary-layer
separation. By transferring heat from the wall to the uid in liquids or from the uid
to the wall in gases, this term adds a negative contribution to the curvature of the
velocity prole at the wall and, hence, causes the separation point to move farther aft.
If the surface of a body in a compressible gas is cooled, the near-wall uid will have
larger density and smaller viscosity than that in the case with no heat transfer. The
smaller viscosity results in a fuller velocity prole and higher speeds near the wall.
Combined with the larger density, this yields a higher momentum for the near-wall
uid particles and, hence, the boundary layer becomes more resistant to separation.
Although this method of control has been successfully applied to delay transition in
both water and air ows as discussed in Chapter 6, its use to prevent separation has
been demonstrated only for high-speed gaseous ows. These effects are conrmed
via the analytical results of Libby (1954), Illingworth (1954), and Morduchow and
Grape (1955). Experimental verication is provided by the work of Gadd, Cope,
and Attridge (1958); Bernard and Siestrunck (1959); and Lankford (1960, 1961).
Excellent summaries of the problem of heat transfer effects on the separation of a
compressible boundary layer are available in Gadd (1960) and Chang (1970, 1976).
Active separation control by wall cooling in air is straightforward. The tech-
nique might be particularly appropriate for high-altitude, long-endurance vehicles
(HALE) having thick, low-Reynolds-number wings and cryogenic fuel to provide
the requisite heat sink (e.g., Baullinger and Page 1989). Unfortunately, this method
is mainly restricted to cryogenically-fueled aircraft. This makes it particularly ap-
propriate for hypersonic applications such as shock-boundary-layer interactions on
hydrogen-fueled vehicles. Cooling in air works according to both experiment and
theory for low (Macha, Norton, and Young 1972; Lin and Ash 1986) as well as high
(Spaid 1972; Ogorodnikov, Grin, and Zakharov 1972) speeds.
In liquids, surface heating lowers the near-wall viscosity but the density remains
essentially unchanged. Using simple asymptotic analysis of the coupled energy and
momentum equations, Aroesty and Berger (1975) compared the effectiveness of wall
heating with suction as a means of delaying separation for a prescribed adverse
pressure gradient in a water boundary layer. They concluded that surface heating
can be used in water to delay separation somewhat. However, it seems that this
analytical result has not been conrmed experimentally.
In addition to surface heating and cooling, several other methods are available to
establish a viscosity gradient in a wall-bounded ow and thus to affect the location
of separation. These include lm boiling, cavitation, sublimation, chemical reaction,
wall injection of a secondary uid having lower or higher viscosity, and the intro-
duction into the boundary layer of shear-thinning or shear-thickening additive.
8.5 Moving-Wall and Time-Dependent Separations
8.5.1 Moving Walls
The moving-wall effects can be exploited to postpone separation. For a surface
moving downstream, the relative motion between the wall and the freestream is
minimized, and thus the growth of the boundary layer is inhibited. Furthermore, the
8.5 MOVING-WALL AND TIME-DEPENDENT SEPARATIONS 169
surface motion injects additional momentuminto the near-wall ow. Prandtl (1925b)
demonstrated the effects of rotating a cylinder placed in a uniform stream at right
angles to its axis. Separation is completely eliminated on the side of the cylinder
where the wall and the freestream move in the same direction. On the other side
of the cylinder, separation is developed only incompletely. In fact, for high enough
values of circulation, the entire oweld can be approximated by the potential ow
theory. The asymmetry causes a force on the cylinder at a right angle to the mean
ow direction. This important phenomenon, known as the Magnus effect (Magnus
1852; Swanson 1961), is exploited in several sports balls (Mehta 1985a) and even in
an experimental device used for propelling ships known as the Flettners (1924) rotor.
From a practical point of view, wall motion for body shapes other than circular
cylinders or spheres is prohibitively complicated, although it is feasible to replace a
small portion of the surface of, say, an airfoil by a rotating cylinder, thus energizing the
boundary layer and avoiding separation (Alvarez-Calderon 1964). Rotating cylinders
have been successfully employed to delay separation at the leading and trailing edges
of airfoils and control surfaces (Johnson et al. 1975; Mokhtarian and Modi 1988;
Mokhtarian et al. 1988a,b; Modi et al. 1989), at ap junctures (Alvarez-Calderon
1964; Lee 1974; Tennant et al. 1975; Modi et al. 1980), and in diffusers (Tennant
1973). Critical parameters include the rotational speed and the gap between the
cylinder and xed surface.
Flight tests were conducted on a YOV-10A STOL-type aircraft having aps with
rotating cylinders at their leading edges (Cichy, Harris, and Mackay 1972; Weiberg
et al. 1973; Cook, Mickey, and Quigley 1974). With the aps in lowered position,
the cylinders were rotated at high speed, and lift coefcients as high as 4.3 were
recorded at a modest ying speed of 30 m/s along approaches up to 8
. Modi and his

colleagues (Modi et al. 1980, 1981, 1989; Mokhtarian and Modi 1988; Mokhtarian,
Modi, and Yokomizo 1988a,b) carried out a comprehensive wind tunnel test program
involving a family of airfoils each having one or more rotating cylinders located at
the leading edge, the trailing edge, or the upper surface, as sketched in Figure 8.8.
Under optimum conditions, the lift coefcient increased by as much as 200%, and
the stall angle was delayed to 48
.
More recently, Modi, Fernando, and Yokomizo (1990) extended the concept of
moving surface boundary-layer control to reduce the drag of land vehicles. On a
scale model of a typical tractor-trailer truck conguration having a splined rotating
Figure 8.8: Various rotating cylinder congurations used to increase
lift and delay stall of an airfoil.
Figure 8.9: Tractor-trailer truck with rotating cylinders.
cylinder at the top leading edge of the trailer (Figure 8.9), drag was lowered by as
much as 27% when the cylinder surface velocity was 3.7 times the freestream speed.
Modi et al. (1990) maintained that this separation control concept is essentially
semi-passive, requiring a negligible amount of power for its implementation.
Creating a wall-slip layer biases the mean ow such that a larger pressure gradi-
ent can be tolerated before separation occurs. There are essentially two techniques
for establishing a slip layer on the wall to mitigate separation. The rst of these is
actually to translate the wall itself (e.g., moving belts or embedded rotating cylin-
ders, as discussed above). The other approach is the establishment of stabilized cavity
vortex ows. Either small-scale (Migay 1960a,b; 1961; 1962a,b,c; Stull and Velkoff
1975; Howard and Goodman 1985) or large-scale (Ringleb 1961; Adkins 1975;
Adkins, Mathaus, and Yost 1980; Burd 1981; Chow, Chen, and Huang 1985; Krall
and Haight 1972; Haight et al. 1974) vortex stabilization is necessary; otherwise, the
trapped vortex will generally periodically shed downstream, causing disrupted op-
eration and higher drag and losses. Stabilization techniques include injection (Krall
and Haight 1972; Haight, Reed, and Morland 1974), suction (Adkins 1975; Adkins
et al. 1980; Burd 1981; Chowet al. 1985), and viscous forces via low-cavity Reynolds
number (e.g., Howard and Goodman 1985). The vortex ap is the latest version of
such a device. These slip-layer separation control strategies work for moderately sep-
arated ows, but in extreme cases reverse ow can still occur away from the surface
(Zhuk and Ryzhov 1980).
8.5.2 Time-Dependent Separation
For time-dependent ows, the separation point is no longer stationary but rather
moves along the surface of the body. The MooreRottSears (MRS) criterion states
that unsteady separation occurs when the shear and velocity vanish simultaneously
and in a singular fashion at a point within the boundary layer as seen by an observer
moving with the separation point (Rott 1956; Sears 1956; Moore 1958). Steady
separation is clearly included in this model as a special case. The main drawback of the
MRS criterion is that the speed of the separation point is not known a priori, making
it difcult to locate this point and forcing researchers to rely on more qualitative
measures for unsteady separation.
In analogy to the moving-wall case, unsteady separation is (temporally) post-
poned when the separation point moves upstream, as is the case on the suction side
of an airfoil undergoing a pitching motion from small to large angles of attack (Fig-
ure 8.10). Conversely, when an airfoil is pitched from large to small attack angle, the
separation point on the suction side moves downstream, and separation is advanced
much the same as the case of a wall moving upstream.
An airfoil oscillating sinusoidally through high angles of attack can produce very
high lift coefcients and maintain ow attachment well beyond static stall attack
8.5 MOVING-WALL AND TIME-DEPENDENT SEPARATIONS 171
Figure 8.10: A NACA 0012 rectangular wing during a pitching-up motion between 0
and 30
. Aspect
ratio = 4; Re
c
= 1.25 10
4
; reduced frequency f c/U
= 0.2 [after Gad-el-Hak 1992].

angles (McCroskey 1977, 1982). During the upstroke, the separation point moves
upstream, and reverse ow exists in an attached and mathematically well-behaved
boundary layer. The global aerodynamic properties of a pitching airfoil are strongly
inuenced by the local unsteady separation. Sudden changes of lift, drag, and pitching
moment occur near the onset of separation and the spillage of a leading edge vortex.
These effects are particularly signicant at high frequencies and large amplitudes.
Moment stall is observed when the reverse-owregion extends over most of the airfoil
and a large-scale vortex is formed near the leading edge. A discontinuous increase in
circulation is associated with the spilled vortex. During this phase of the cycle, lift
continues to increase. Lift stall follows moment stall and occurs when the separation
vortex reaches the latter half of the airfoil and a double-peaked pressure distribution
results on the suction side. In other words, the suction on the upper surface of the
airfoil continues to increase at the initial stages of separation, and a sudden decrease
in suction does not occur until the leading edge is in the wake of separation.
Similar phenomena are observed on three-dimensional lifting surfaces undergo-
ing pitching motion (Gad-el-Hak 1986c, 1988a,b; Gad-el-Hak and Ho 1986a). For
highly swept wings, both steady as well as unsteady ows are vortex dominated.
The latter ow is characterized by the existence of unsteady large- and small-scale
vortices that go through a growth-decay cycle with hysteresis during each period
(Gad-el-Hak and Ho 1985, 1986b; Kandil and Chuang 1990a,b; Atta and Rockwell
1990; Huyer, Robinson, and Luttges 1990).
According to Ericsson (1967, 1988), the forces on an airfoil oscillating in pitch
will deviate from the static forces realized at the instantaneous angle of attack due to
the superposition of two effects. The rst is the frequency-induced normal velocity
distribution over the airfoil. This so-called q effect can be visualized as a frequency-
induced camber. The second is the effect of attack angle rate of change, the so-called
effect. This can be visualized as a frequency-induced change of the mean velocity
vector or plunging. During the upstroke, a pitching airfoil will appear as having a
positive camber and as plunging. According to the unsteady Bernoullis equation, the
local pressure gradient is less adverse in the dynamic case. Thus, the boundary layer
at a particular during the upstroke has a more favorable upstream time history as
compared with the static case. The opposite effects take place during the downstroke.
Insects, most of which mate and eat while airborne, exploit unsteady separation
effects to achieve remarkable aerodynamic characteristics. The dragony, in existence
for approximately 250 million years, presumably survived innumerable life and death
aerodynamic struggles (Luttges et al. 1984; Luttges 1989). The enviably large lift
coefcients generated by the chalcid wasp during hovering suggest the existence of
an efcient, unsteady lift-generation mechanism (Weis-Fogh 1973; Lighthill 1973;
Maxworthy 1979, 1981; Ellington et al. 1996; Brookes 1997).
8.6 Three-Dimensional Separation
Three-dimensional boundary layers are more common in practical ow situations
than two-dimensional ones. Bodies of revolution at nonzero angle of attack, ow
near wing tips, turbine blades, pump impellers, and low-aspect-ratio wings are ex-
amples of owelds in which three-dimensional effects dominate. As mentioned ear-
lier, the point of boundary-layer separation from a three-dimensional body does not
necessarily coincide with the point of vanishing wall shear. Instead, the shear stress
at the wall is equal to zero only at a limited number of points along the separation
line. The number and type of these critical or singular points must satisfy certain
topological laws (Lighthill 1963a; Tobak and Peake 1982).
The projection of the limiting streamlines as the distance fromthe wall goes to zero
coincides with the skin-friction lines on the surface of the body. Oil-streak techniques
and the like are usually used to obtain separation and attachment patterns for steady,
three-dimensional ows (Maltby 1962). A necessary condition for the occurrence of
owseparation is the convergence of skin-friction lines onto a particular line. Because
of the three-dimensionality of the ow, the near-wall uid may move in a direction in
which the pressure gradient is more favorable and not against the adverse pressure in
the direction of the main ow, as is the case for two-dimensional ows. Consequently,
three-dimensional boundary layers are in general more capable of overcoming an
adverse pressure gradient without separation.
Three-dimensional relief of the streamwise adverse pressure gradient may be ex-
ploited to delay separation. Properly designed corrugated trailing edges can provide
sufcient easement to postpone the separation at higher angles of attack. In nature,
three-dimensional serrated geometry is to be found in the trailing edges of the ns and
wings of many aquatic animals and birds (Norman and Fraser 1937; Lighthill 1975).
For man-made lifting surfaces, the same concept was tested in the low-Reynolds-
number regime by Vijgen et al. (1989). They reported a modest 5% increase in the
maximum lift-to-drag ratio when triangular serrations were added to the trailing
edge of a natural-laminar-ow airfoil (Gad-el-Hak 1990).
8.7 Turbulators
A turbulent boundary layer is more resistant to separation than a laminar one, and,
mostly for that reason, transition advancement may be desired in some situations.
8.7 TURBULATORS 173
In low-Reynolds-number terminology, the transition-promoting devices are called
turbulators. For a zero-pressure-gradient boundary layer, transition typically occurs
at a Reynolds number based on distance from leading edge of the order of 10
6
. The
critical Reynolds number Re
crit
below which perturbations of all wave numbers de-
cay is about 6 10
4
. To advance the transition Reynolds number, one may attempt
to lower the critical Re, increase the growth rate of TollmienSchlichting waves, or
introduce large disturbances that can cause bypass transition. The rst two routes
involve altering the shape of the velocity prolemaking it less fullby using wall
motion, injection, adverse pressure gradient, surface heating in gases or cooling in
liquids, or other strategies to increase the near-wall viscosity. The third route, expos-
ing the boundary layer to large disturbances, is much simpler to implement though
more difcult to analyze (Smith and Kaups 1968; Cebeci and Chang 1978; Nayfeh,
Rageb, and Al-Maaitah 1986; Cebeci and Egan 1989).
Morkovin (1984) broadly classied the large disturbances that can cause bypass
transition into steady or unsteady ones originating in the freestream or at the body
surface. The most common example is single, multiple, or distributed roughness
elements placed on the wall. The mechanical roughness elements, in the form of ser-
rations, strips, bumps or ridges, are typically placed near the airfoils leading edge.
If the roughness characteristic length is large enough, the disturbance introduced is
nonlinear and bypass transition takes place. For a discrete three-dimensional rough-
ness element with a height-to-width ratio of one, Tani (1969) reports a transition
Reynolds number of Re
300 for a roughness Reynolds number of Re
10
3
.
Here, Re
is based on the velocity outside the boundary layer and the displacement
thickness (Re
U
o

/), and Re
is based on the height of the roughness element

and the velocity in the undisturbed boundary layer at the height of the element
(Re
U() /). Note that the transition Reynolds number, Re
, indicated above
is below the critical Re
|
crit
= 420 predicted from the linear stability theory. For a
roughness Reynolds number of about 600, transition occurs at Re
10
3
. For a
smooth surface, transition typically takes place at Re
2.6 10
3
. An important
consideration when designing a turbulator is to produce turbulence and suppress
laminar separation without causing the boundary layer to become unnecessarily
thick. A thick turbulent wall-bounded owsuffers more drag and is more susceptible
to separation than a thin one. Consistent with this observation, available data indi-
cate that a rough airfoil has higher lift-to-drag ratio than a smooth one for Re
c
<10
5
but that this trait is reversed at higher Reynolds numbers.
For low-Reynolds-number airfoils, performance may be improved by reducing
the size of the laminar separation bubble through the use of transition ramps (Eppler
and Somers 1985), boundary layer trips (Davidson 1985; Van Ingen and Boermans
1986), or even pneumatic turbulators (Pfenninger and Vemuru 1990). Donovan and
Selig (1989) provide extensive data using both methods for 40 airfoils in the Reynolds
number range of Re
c
= 6 10
4
3 10
5
. A long region of roughly constant adverse
pressure gradient on the upper surface of a lifting surface (termed a bubble ramp)
achieves a lower drag than the more conventional laminar-type velocity distribu-
tion in which initially the pressure remains approximately constant and then quickly
recovers. Trips were also used in Donovan and Seligs experiments to shorten the sepa-
ration bubble. Asimple two-dimensional trip performed as well or better than zig-zag
tape, hemisphere bumps, and normal blowing. Donovan and Selig concluded that an
airfoil that performs poorly at low Reynolds numbers can be improved through the
use of turbulators. Trips were less effective, however, at improving airfoils, which
normally had low drag. We will return to low-Reynolds-number situations in the
following chapter.
Other large disturbances that can lead to early transition include high turbulence
levels in the freestream, external acoustic excitations, particulate contamination, and
surface vibration. These are often termed environmental tripping. Transition can also
be effected by detecting naturally occurring TS waves and articially introducing in-
phase waves. Transition can be considerably advanced, on demand, using this wave
superposition principle.
Early transition can also be achieved by exploiting other routes to turbulence such
as TaylorG ortler or crossow vortices (Taylor 1923; G ortler 1955; Gregory, Stuart,
and Walker 1955; Reed and Saric 1987, 1989). For example, a very mild negative
curvature of 0.003/
results in the generation of strong streamwise vortices. In this

case, transition Reynolds number is lowered from Re
2600 for the at-plate case

to Re
700 for the concave surface (Tani 1969). For high-Mach-number ows,
the general decay in spatial amplication rate of TS waves makes conventional
tripping more difcult as the Mach number increases (Reshotko 1976). For these
ows, trips that generate oblique vorticity waves of wavelength appropriate to cause
resonant growth of three-dimensional modes may be most effective to advance the
transition location. For example, on a sharp cone at Ma = 3.5, Corke and Cavalieri
(1997) and Corke, Cavalieri, and Matlis (1999) used an array of plasma (corona-
discharge) electrodes to excite oblique vorticity-mode pairs and successfully advanced
the transition of the supersonic boundary layer. A similar strategy to generate pairs
of oblique waves was employed by Corke and Mangano (1989) in an incompressible
Blasius boundary layer. There, Corke and Mangano used a spanwise array of heaters
placed at the position of the critical layer to generate time-periodic, spanwise-phase-
varying velocity perturbations.
The last issue to be considered in this section is augmentation of the turbulence
for a shear ow that has already undergone transition. Notwithstanding that the
turbulence levels and the Reynolds stresses are highest immediately following tran-
sition (Harvey, Bushnell, and Beckwith 1969), the newly developed turbulent ow
is in general less capable of resisting separation than a corresponding ow at higher
speeds (Lissaman 1983). Turbulence augmentation in the low-Reynolds-number case
is then a useful control goal to energize the ow and to enhance its ability to resist
separation at higher angles of attack. Roughness will enhance the turbulence, but
its associated drag must be carefully considered. Other devices to enhance the tur-
bulence mixing include vane-type vortex generators, which draw energy from the
external ow, or Wheeler-type or Kuethe-type generators, which are fully submerged
within the boundary layer and presumably have less associated drag penalty (Rao
and Kariya 1988). These and other devices will be detailed in the next section.
8.8 Momentum Addition to Near-Wall Flow
Near-wall momentum addition is the usual approach of choice for control
of residual ow separation remaining after mitigation of the causative pressure
eld or for off-design conditions. Common to all these control methods is the
8.8 MOMENTUM ADDITION TO NEAR-WALL FLOW 175
Figure 8.11: Passive blowing through lead-
ing-edge slats and trailing-edge aps.
supply of additional energy to the near-wall
uid particles that are being retarded in the
boundary layer. The additional longitudinal
momentum is provided either from an exter-
nal source or through local redirection into
the wall region. Passive techniques do not
require auxiliary power but do have an as-
sociated drag penalty. These include inten-
tional tripping of transition from laminar to
turbulent ow upstream of what would be
a laminar separation point (Mangalam et al.
1986; Harvey 1986); using boundary-layer fences to prevent separation at the tips
of swept-back wings; placing an array of vortex generators on the body to raise the
turbulence level and enhance the momentum and energy in the neighborhood of the
wall (Mehta 1985b; Rao and Kariya 1988); employing a rippled trailing edge (Werle,
Paterson, and Presz 1987), streamwise corrugations (Mabey 1988), or stepped af-
terbodies to form a system of captive vortices in the base of a blunt body (Kenteld
1985a,b; Kidd, Wikoff, and Cottrell 1990); and using a screen to divert the ow and
increase the velocity gradient at the wall.
Active methods to postpone separation require energy expenditure. Obviously,
the energy gained by the effective control of separation must exceed that required by
the device. In addition to suction or heat transfer reviewed in Sections 8.4.2 and 8.4.3,
uid may be injected parallel to the wall to augment the shear-layer momentum or
normal to the wall to enhance the mixing rate (Horstmann and Quast 1981). Either
a blower is used or the pressure difference that exists on the aerodynamic body itself
is utilized to discharge the uid into the retarded region of the boundary layer. The
latter method is found in nature in the thumb pinion of a pheasant, the split-tail of
a falcon, or the layered wing feathers of some birds.
In man-made devices, passive blowing through leading-edge slots and trailing-
edge aps is commonly used on aircraft wings (Smith 1975; Lin et al. 1992), as
sketched in Figure 8.11. Although in this case direct energy expenditure is not re-
quired, the blowing intensity is limited by the pressure differentials obtainable on
the body itself. Nevertheless, the effect of passive blowing on lift and drag can be
dramatic. This is shown convincingly in Figure 8.12 for the NACA 23012 airfoil
section with no ap, with a single trailing-edge ap, and with a double-slotted ap.
Compared with the clean (no ap) case, when a single trailing-edge ap is used the
maximum lift is increased by about 175%, whereas the section drag at C
L
max
is in-
creased by more than 180%. The corresponding numbers when a double-slotted
ap is used are, respectively, 230% and 500%. The induced drag, which must be
considered in addition to the section drag, is proportional to C
2
L
max
, and the total
drag, therefore, rises sharply at low aircraft speeds. Very recently, Thomas et al.
(1997, 1999) emphasized the importance of unsteady effects caused by the conuent
boundary layer
1
on the performance of multielement airfoils.
Many of the control methods discussed in this section can be used for external
as well as internal ows. For example, Viets (1980) used an asymmetrical rotating
1
This termdenotes the (typically) turbulent owgenerated by the interaction of the wake of a leading-edge
slat with the boundary layer on the primary lifting surface.
Figure 8.12: Effects of trailing-edge aps on lift and section drag. NACA23012 airfoil section
[adapted from data by Abbott and von Doenhoff 1959].
camembedded in the wall to produce large eddies in a turbulent boundary layer with
zero- and adverse-pressure gradients. By using this device in a wide-angle diffuser,
Viets, Ball, and Bougine (1981a) were able to postpone the natural separation and
dramatically improve the diffusers performance.
8.8.2 Passive Vortex Generators
Passive momentumaddition is most commonly carried out via one of two general
approaches: either macro overturning of the mean ow using embedded streamwise
vortices generated by xed lifting surfaces or Reynolds stress amplication, which
leads to increased cross-stream momentum transfer. Conventional passive vortex
generators (VGs) date from the 1940s (Taylor 1948a) and are simple and effective
and therefore generally the rst tried as a x to an existing ow separation problem.
Passive VGs have been applied, for example, to compressor blades (Staniforth 1958),
diffusers (Henry, Wood, and Wilbur 1956; Feir 1965; Brown, Nawrocki, and Paley
1968), airfoils (Pearcey 1961; Nickerson 1986; Bragg and Gregorek 1987), and the
afterbody of aircraft fuselages (Calarese, Crisler, and Gustafson 1985; Wortmann
1987). What these embedded vortices do is cause overturning of the near-wall ow
via macro motions. Fluid particles with high streamwise momentum are swept along
helical paths toward the surface to mix with and, to a certain extent, to replace the
retarded near-wall ow. The vortex inuence upon the turbulence can actually be
debilitating owing to streamline-curvature-induced stabilization.
Passive vortex generators are essentially small-aspect-ratio airfoils mounted nor-
mal to the surface. Their individual parameterization includes planform shape, sec-
tion prole and camber, yaw angle, aspect ratio, and height with respect to the
boundary-layer thickness. The spatial relationship of the devices is also critical (e.g.,
corotating versus contrarotating biplane and wing, use of downstream reenforcers,
and spacing). Counterrotating vortices force large regions of vorticity to rise above the
surface and hence are often not as efcient as corotating devices. However, corotating
vortices with too close a spacing undergo mutual vorticity cancellation; therefore,
spacing is critical in this case.
Nominal guidelines for conventional VGs are given in the articles by Taylor
(1948b), Henry et al. (1956), and Pearcey (1961): aspect ratio of 0.51; rectangular or
triangular planformwith either simple at plate or low-Reynolds-number airfoil cross
section, yaw angle less than 15
, and individual height of the order of the boundary-

layer thickness. For corotating devices, the spanwise spacing should be greater than
three times the device height to avoid interactive cancellation or annihilation of the
vortical structures. The recommended nominal spanwise spacing for contrarotating
vortex production devices is on the order of 45 device heights. These conventional
vortex-generating devices produce sizable parasitic drag. When employed to han-
dle off-design problems, the VGs cause a reduction in cruise performance unless the
devices are retracted when not needed. The articles by Schubauer and Spangenberg
(1960); Gadetskii, Serebriiskii, and Fomin (1972); Liandrat, Aupoix, and Cousteix
(1986); Cutler and Bradshaw (1986, 1989); Inger and Siebersma (1988); Mehta
(1988); and Briedenthal and Russell (1988) provide generic studies of vortex genera-
tor physics and operation. Application to supersonic ows is summarized by Gartling
(1970).
Several approaches are now available to optimize the performance of passive
vortex generators. The rst of these is the use of downstream reenforcers, which are
vortex generators of the same sense located in the path of the upstream-generated vor-
tex to maintain the strength of the overturning motion (Kuethe 1973; Wheeler 1984;
Lin and Howard 1989; Rao and Kariya 1988). A second optimization approach is
surprisingly recent and consists of simply reducing the device height from the order
of to O[/5] or less (Rao and Kariya 1988; Lin and Howard 1989; Lin, Howard,
Bushnell, and Selby 1990a). This size reduction signicantly diminishes the parasitic
drag and is enabled by the extreme fullness of the mean velocity prole in a turbulent
boundary layer. In other words, with sizable longitudinal momentum levels readily
available quite close to the surface, it is not necessary to have O[] devices. Such
sub- devices must be placed closer to the nominal separation location and therefore
may be less suitable than larger devices for situations in which the separation region
is not relatively localized.
Alternative approaches for separation control by streamwise vortices are the
V-shaped cutouts of the NACAs ush inlet, leading-edge serrations (Harris and
Bartlett 1972; Soderman 1972; Barker 1986), and use of large-scale (oweld versus
boundary-layer scale) vortical motions to control separation on highly swept wings.
Such motions can be generated by either auxiliary lifting devices (e.g., canards) or
by simple abrupt planform variations such as wing leading-edge extensions (LEXs).
Such large-scale vortical motions alter both the near-wall momentum and the basic
pressure eld, generating increased lift. Problems with such an approach include vor-
tex bursting, which is usually caused primarily by adverse pressure gradients. Control
of bursting is actually control of vortices and is a subsidiary problem of separation
control by vortices (Vakili 1990).
8.8.3 Passive Turbulence Amplication
The fundamental ow-separation-control approach associated with turbulence
amplication is augmentation of the cross-stream momentum via Reynolds stresses
inasmuch as the overall pressure rise for incipient separation is directly propor-
tional to the square root of the skin-friction coefcient of the undisturbed bound-
ary layer (Hayakawa and Squire 1982). The zeroth-order consideration for most
separation control is to ensure a turbulent rather than laminar boundary-layer state.
For Reynolds numbers less than O[10
7
], this may necessitate use of various boundary-
layer tripping devices such as roughness and waviness, passive or active mass transfer,
acoustic elds, body vibration, or even elevated freestream disturbance levels.
Once a turbulent boundary-layer ow is established, large-scale dynamic trans-
verse vortical entities can be generated to augment the innate turbulent Reynolds
stress eld beyond the usual amplication concomitant with the adverse pressure
gradients usually associated with separation. Applicable devices include transverse
cylinders or ow control rails mounted just above, or in the outer part of, the bound-
ary layer (Sajben, Chen, and Kroutil 1976). Airfoils with chord of O[] mounted
parallel to the body in the outer part of the boundary layer and set at a nonzero angle
of attack also generate large transverse dynamic vortical motions (Corke, Guezennec,
and Nagib 1980), as can embedded cavity (Helmholtz) resonator surfaces. Dynamic
three-dimensional horse-shoe or hairpin eddies can be generated in the tip ow of
three-dimensional stubs mounted normal to the surface. Except for the ow-control
rail, which works quite well, these passive devices have received only limited attention
and development for ow-separation control.
Howard and Goodman (1985, 1987) recently investigated the effectiveness of two
passive techniques to reduce owseparation: transverse rectangular grooves and lon-
gitudinal V-grooves placed in the aft shoulder region of a bluff body. Both types of
grooves were benecial in reducing the form drag on a body at zero and moderate
angles of yaw. Basically, the grooves work by redirecting the outerowmomentumto
the near-wall region through the three-dimensionalization of a two-dimensional ow.
This is an example of locally mitigating the adverse pressure gradient through par-
tial boat-tailing. Lin, Howard, and Selby (1990b) extended the longitudinal groove
approach to the low-speed, two-dimensional, rearward-facing ramp. Their closely
packed grooves resulted in a reduction of the reattachment distance by up to 66%.
Selby and Miandoab (1990) reported that the base pressure of a blunt trailing-edge
airfoil with surface grooves increased with increasing groove depth and angle. They
speculated that minimally attached ow in the grooves is the mechanism by which
uid of higher momentum is redirected to the base ow region to effect an increase
in the pressure.
Additional turbulence amplication ploys include destabilizing longitudinal sur-
face curvature to increase wall shear: concave for boundary layers and convex for wall
jets (Coanda effect), and for high speeds, use of upstream shock-wave-interaction,
which can amplify turbulence signicantly (Anyiwo and Bushnell, 1982; Zang,
Hussaini, and Bushnell 1984) while also delaying separation (Schoeld 1985; Gold
Feld and Zatoloka 1979). The use, generation, or both of freestream turbulence
and disturbance elds with intermediate scale motions also amplify turbulence in
the boundary layer and delay separation (Hoffmann 1981; Sasaki and Kiya 1985;
Hoffmann, Kassir, and Larwood 1988; Isomoto and Honami 1989).
8.8.4 Active Wall-Region Momentum Addition
Active momentum addition techniques include streamwise vortex generation via
discrete blowing or injection, turbulence or Reynolds stress amplication through use
of dynamically activated or driven devices, and direct tangential injection of high-
velocity uid. The two passive techniques discussed in Sections 8.8.2 and 8.8.3, vortex
generation and turbulence amplication, can also be employed via active systems. Of
particular interest is the use of discrete jet injection for streamwise vortex generation.
This concept arose in the 1950s (Wallis and Stuart 1958; Pearcey 1961; Kukainis
1969; Wimpenny 1970) and is the subject of current research (Zhang and Sheng
1987; Reynolds et al. 1988; Johnston and Nishi 1989, 1990; Compton and Johnston
1991; McManus et al. 1994, 1996; McManus and Magill 1996, 1997; Johari and
McManus 1997). What is particularly appealing is the ability to deploy or retract
the vortex generators as required, which eliminates parasitic drag in the retracted
nonblowing condition. The associated uid supply lines operate at high pressure and
are thus relatively small. They might even be utilizable as structural reenforcement
elements. The status of research in this area is such that, although it is clear that
discrete jets will generate vortices and delay separation, the approach has not yet been
optimized. Research is required to determine optimal injection orientation, spacing,
individual hole geometry and size, velocity, pressure, and nally location vis- ` a-vis the
adverse pressure gradient regions. Papell (1984) addresses vortex generation within
the injection jet itself. Another use of injection for vortex generation is spanwise
injection along the leading edges of swept wings for upper surface separation control
(e.g., Bradley and Wray 1974). Employing spanwise arrays of small, skewed, pitched
jets from holes in the surface, Johnston and Nishi (1990) have shown that the jets
can produce longitudinal vortices strong enough to reduce substantially and nearly
eliminate a large stalled region of a turbulent separated ow.
Arelatively recent development in owseparation control is active turbulence am-
plication in the bounding shear layer for ows already separated. The fundamental
concept is excitation and, through phasing, enhanced interaction of large transverse
eddy structures near and downstreamof the separation point, thereby amplifying the
mixing in the shear layer bounding the separated ow region. This increases entrain-
ment and generally reduces the extent of separation. For initially laminar ows, the
zeroth-order inuence of dynamic forcing is to trip transition and enhance the eddy
dynamics of the low-Reynolds-number shear-layer (e.g., Collins 1979, 1981; Mullin,
Greated, and Grant 1980; Sigurdson and Roshko 1985; DeMeis 1986; Durbin and
McKinzie 1987; Huang, Maestrello, and Bryant 1987; Neuburger and Wygnanski
1988; Zaman and McKinzie 1989; Bar-Sever 1989).
Various dynamic devices have been tried for the initially turbulent ow case,
including acoustic drivers (Bhattacharjee, Scheelke, and Troutt 1985), oscillating
embedded plates, spoilers, and aps (Reisenthel, Nageb, and Koga 1985; Roos and
Kegelman 1986; Miau, Chen, and Chow 1988; Chen and Shi Ying 1989; Katz,
Nishri, and Wygnanski 1989a,b); both chordwise and spanwise dynamic blowing
(Oyler and Palmer 1972; Ely and Berrier 1975; Vakili, Wu, and Bhat 1988; Vakili
1990); rotating cams (Viets et al. 1979; 1981a,b,c; 1984); and even dynamic motions
of the entire body. One of the few large-scale dynamic experiments thus far is also
one of the earliest (Oyler and Palmer 1972). The same incremental increase in lift
was achieved with only 50% of the blowing mass ow required for the steady-state
case. What is obvious from the initially turbulent, dynamic-input separation control
research thus far is that, once again, the method works. What is not so obvious is
how well it would work in engineering applications and the nature and operating
range of the optimal dynamic devices.
Wall Jets
Direct tangential injectionwall jetwas and still is the preferred and straight-
forward ow separation control technique that has been applied to military ghters
and STOL transports. High-pressure air can be used, enabling relatively small interior
lines as opposed to suction control which, although generally more energy efcient
than blowing at both low and high speeds, usually requires larger interior ducting
(Gratzer 1971). In some applications, a lighter gas is introduced to reduce the rate
at which heat is exchanged between the wall and the external stream and, thus, to
provide thermal protection at high supersonic speeds.
High-pressure bleed air was readily available from the early jet engines but less
so for modern high-bypass-ratio turbofans. Computational uid dynamics (CFD)
can now be used to design the system and optimize the injection velocity prole
for a given mass ow for optimal separation delay (see, for example, Saripalli and
Simpson 1980). Performance can be further enhanced by convex longitudinal curva-
ture (Coanda effect). Tangential blowing is also used to stabilize a trapped vortex, par-
ticularly in the knee region of wing aps. The literature for the steady blowing case is
both extensive and readily available, and most information dates fromthe 1950s. Sep-
aration control by blowing at high speeds is covered in the reviews by Delery (1985)
and Viswanath (1988). Of possible interest for separation control via direct tangen-
tial injection is the application of turbulence control techniques to reduce the mixing
between the injected and incident ows and thereby preserve the high near-wall mo-
mentum for a larger extent downstream (McInville, Hassan, and Goodman 1985).
Tangential jet blowing over the upper surface of a rounded trailing-edge airfoil
sets an effective Kutta condition by xing the location of separation. This circulation
control concept was initially described by Cheeseman and Seed (1967), and a sub-
stantial database has been gathered since then for the purpose of performance eval-
uation (Kind 1967; Wood and Nielsen 1985; Novak, Cornelius, and Roads 1987).
More recently, McLachlan (1989) conducted an experimental study of the ow past
a two-dimensional circulation control airfoil under steady leading- and trailing-edge
blowing. In the range of chord Reynolds numbers of 1.210
5
3.910
5
, McLachlan
observed a dramatic increase in the lift coefcient when trailing-edge blowing was
used to control the location of the rear separation points. He reported a gain in the lift
coefcient of the order of 80 times the injected momentumcoefcient. When leading-
edge blowing was employed simultaneously, a slight decrease in lift was observed.
Quite recently, McManus et al. (1994, 1996), McManus and Magill (1996, 1997),
and Johari and McManus (1997) conducted a series of laboratory experiments
in which pulsed jets were employed to improve the leading-edge ap effectiveness
on high-performance ghters. Their separation control technique exploits the re-
markable vorticity-generation capabilities of impulsively started jets to enhance
boundary-layer momentum transport and thus suppress stall for both compressible
and incompressible ows. In addition to the customary streamwise vortices, large-
scale turbulent vortex rings are also generated in the ow as a result of pulsing the
wall jets. The net effect is to achieve fairly good stall suppression with substantially
lower jet-mass-ow rate.
Although not directly related to separation control, jets in the form of thin sheets
exiting in the spanwise direction from the tips of a straight wing can be used to
enhance the lift through an effective enlargement of the wing span (Wu et al. 1983,
1984; Tavella et al. 1988; Lee et al. 1989; Vakili 1990). This application was rst
reported in 1956 by Ayers and Wild and suggests the possibility of using such an
arrangement in place of conventional ailerons or aps to alter the aerodynamic forces
acting on an aircraft (Tavella, Wood, Lee, and Roberts 1986). Either a single long
slot (Lee et al. 1986; Tavella, Lee, and Wood 1986) or several short ones (Wu et al.
1983, 1984) are used along the entire chord at the wing tips. The enlargement of the
aspect ratio associated with the lateral displacement of the tip vortices via blowing
leads also to a reduction of the induced drag as well as to a benecial effect on stall.
Additional Active Control Methods
Other active methods for controlling boundary-layer separation and reattach-
ment include acoustic excitations (Collins and Zelenevitz 1975; Ahuja, Whipkey, and
Jones 1983; Ahuja and Burrin 1984; Zaman, Bar-Sever, and Mangalam1987; Huang
et al. 1987), periodic forcing of the velocity eld via an oscillating ap or wire (Koga,
Reisenthel, and Nagib 1984; Sigurdson and Roshko 1985; Reisenthel et al. 1985;
Roos and Kegelman 1986; Katz et al. 1989a,b; Bar-Sever 1989), and oscillatory
surface heating (Maestrello, Badavi, and Noonan 1988).
As early as 1948, Schubaur and Skramstad observed that sound at particular fre-
quencies and intensities could enhance the momentum exchange within a boundary
layer and could, therefore, advance the transition location. Collins and Zelenevitz
(1975) and Collins (1979, 1981) introduced the external acoustic excitation tech-
nique to enhance the lift of an airfoil. In this case, sound is radiated onto the wall from
a source outside the boundary layer. Using the same technique at chord Reynolds
numbers up to 110
6
, Ahuja et al. (1983) and Ahuja and Burrin (1984) successfully
demonstrated that sound at a preferential frequency and sufcient amplitude can
postpone the separation of a turbulent boundary layer developing on an airfoil in
both pre- and poststall regimes. The optimum frequency was found to be 4U
/c
(i.e., Strouhal number St = 4), where U
is the freestreamvelocity and c is the airfoil

chord. Goldstein (1984) speculated that the delay in separation in Ahuja et al.s
(1983) experiment resulted from enhanced entrainment promoted by instability
waves that were triggered on the separated shear layer by the acoustic excitation.
Zaman et al. (1987) conducted further study of the benecial interaction between
external acoustic excitation and the separated ow around an airfoil at high angles
of attack. They found that the most effective separation control is achieved at fre-
quencies at which the acoustic standing waves in their wind tunnel induce transverse
velocity uctuations in the vicinity of the lifting surface. The loudspeakers used by
Zaman et al. (1987) as well as by Ahuja et al. (1983) essentially excited the reso-
nant modes in their respective wind tunnels, and one of these modes forced the shear
layers to separate from their respective airfoils. Extremely high level of excitation
was required, however, to maintain these wind-tunnel resonance modes, making the
external acoustic excitation technique impractical for eld applications. Zaman et al.
(1987) speculated that a more effective separation control can be obtained by direct
introduction of velocity disturbances.
Huang et al. (1987) introduced the internal acoustic excitation technique in which
sound is emanated from a hole or a slot on the surface of a lifting surface. The loud-
speaker in this case is essentially used as a piston to produce localized vorticity
perturbations at the leading edge of the airfoil. More recently, Hsiao et al. (1990) re-
ported improved aerodynamic performance of a two-dimensional airfoil using sound
emitted from three narrow wall slots located near the leading edge. The sound pres-
sure levels used in their experiment was substantially lower than that used externally
by other researchers. Additionally, a correct Strouhal number scaling was achieved.
Hsiao, Liu, and Shyu (1990) concluded that the enhancement of momentum trans-
port resulting from the sound excitation produces a suction peak at the leading edge,
an increase of lift, and a narrower wake as long as the excitation frequency is locked
in to the most unstable frequency of the separated shear layer.
To disturb the velocity eld directly, Koga et al. (1984) used a computer-controlled
spoilerlike ap in a at-plate turbulent boundary layer with and without modeled
upstream separation. They were able to manipulate the separated ow region and its
reattachment length characteristics by varying the frequency, amplitude, and wave-
formof the oscillating ap. Reynolds and Carr (1985) offered a plausible explanation,
from the viewpoint of a vorticity framework, for the experimental observations of
Koga et al. It seems that the large-scale vortical structures produced by forcing play
a major role in enhancing mixing and entrainment, thus leading to reattachment.
The active ap controls the size of the separated region by providing an additional
mechanism for removing vorticity from this zone; namely, large-scale vortex convec-
tion. More recent experiments by Nelson, Kogan, and Eaton (1987, 1990) seem to
conrm that the dominant mechanism of vorticity transport behind an oscillating
spoiler is convective.
Periodic forcing of the velocity eld has been shown to reduce reattachment
length in both laminar and turbulent ows on a number of other basic geometrical
congurations (Sigurdson and Roshko 1985; Roos and Kegelman 1986; Katz et al.
1989a,b; Bar-Sever 1989). At a chord Reynolds number in the range of 1 10
5
3 10
5
, Bar-Sever (1989) used an oscillating wire to introduce transverse velocity
uctuations into a separated shear layer on an airfoil at high incidence. The effec-
tiveness of this separation control technique is depicted in Figure 8.13, showing the
variation of unforced and forced lift coefcients as a function of angle of attack.
For each angle, the forced case represents the best lift achieved at any combination
of forcing frequency and amplitude. At = 20
, the controlled forcing moved the

separation from the leading edge to about 0.8c. A wide band of forcing reduced fre-
quencies (0.72.7) was found to be effective, although diminished inuence occurred
at lower frequencies.
Figure 8.13: Lift coefcient variation with angle of attack for unforced and best
forced case; Re
c
= 1.5 10
5
[after Bar-Sever 1989].
8.9 SEPARATION PROVOCATION 183
Figure 8.14: Freestreamap. (a) During super-
sonic ight; (b) During takeoff and landing.
8.9 Separation Provocation
Although most of the control methods re-
viewed thus far are designed to prevent sep-
aration, under certain circumstances the de-
signer may wish to provoke separation. Con-
trolled separation is associated with the for-
mation of jets, throttling action in household
faucets, and noisy as well as musical acoustic
effects generated by ow. During takeoff and
landing of a supersonic aircraft, a freestream
ap may be used to provoke leading-edge
separation followed by reattachment at the
leading edge of the ap, thus forming a
thicker pseudobody with the desired aerody-
namic shape for subsonic speeds, as shown
in Figure 8.14 adapted from Hurley (1961). The airfoil has then a relatively high lift
coefcient, although its lift-to-drag ratio is quite low.
The detached bow shock forming upstream of a blunt body in supersonic ight
may be changed into a weaker, attached oblique shock by placing a spike in front of
the body. The pressure rise and the presence of a solid surface on which a boundary
layer forms causes the ow to separate downstream of the spike tip. A properly
designed spike may result in lower drag, higher lift, and corresponding change in
pitching moment (Wood 1961).
Periodic separation may also be provoked by changes in the wall geometry.
Francis et al. (1979) initiated separation on an airfoil by periodically inserting and
removing a spoiler at the wall. Viets et al. (1984) studied separation inducement and
control by use of a cam-shaped rotor mounted on an airfoil. The cam, either driven or
free-wheeling, periodically extended out into the ow, causing the boundary layer to
separate. Large-scale coherent spanwise structures were periodically generated and
were responsible for the owdetachment, as explained by Reynolds and Carr (1985).
8.9.1 Delta Wings
On a sharp-leading-edge delta wing, the separation position is xed, and a strong
shear layer is formed along the entire edge (Lee and Ho 1989, 1990). The shear layer
is wrapped up in a spiral fashion, which results in a large-bound vortex on each side
of the wing. The two vortices appear on the suction surface of the wing in the form
of an expanding helix when viewed from the apex. The low pressure associated with
the vortices produces additional lift on the wing, often called nonlinear or vortex lift,
which is particularly important at large angles of attack. The experiments of Gad-
el-Hak and Blackwelder (1985) have indicated that small discrete vortices are shed
parallel to the leading edge at a repeatable frequency determined by the angle of attack
and Reynolds number (Figure 8.15). Repeated vortex pairings result in the formation
of progressively larger vortices. This process can be modulated by weak, periodic
suction or injection through a leading edge slot. In particular, when the perturbation
frequency is a subharmonic of the natural shedding frequency, the evolution of the
bounded shear layer is dramatically altered (Gad-el-Hak and Blackwelder 1987a;
Blackwelder, Gad-el-Hak, and Srnsky 1987).
(a)
(b)
Figure 8.15: Sharp-leading-edge, 60
-sweep delta wing in steady ight. Reynolds number based on root

chord Re
c
= 1.3 10
4
; = 10
. (a) Top view; (b) End view [after Gad-el-Hak and Blackwelder 1985].
Wood and Roberts (1986, 1988) examined the feasibility of vortex control by
tangential mass injection at the leading edge of a 60
-sweep delta wing. Their initial

experimental results indicated that modest continuous blowing is capable of extend-
ing the regime of stable, controlled vortical ow over the upper surface of the wing
by approximately 30
angle of attack. Increases in maximum normal force of 30%

were achieved, and signicant rolling moments were produced at attack angles of
3560
. On a delta wing with rounded leading edges, the blowing seems to control
the location of the crossow separation points and hence the trajectories of the en-
suing vortices. Wood and Roberts (1988) proposed that this blowing scheme may be
a practical solution for changing the normal force without changing attitude for the
production of both steady and transient control moments at extremely high angles
of attack and for increasing the lift-to-drag ratio of very slender bodies at modest
attack angles. In a later experiment, Wood, Roberts, and Celik (1990) have shown
that the effects of asymmetric leading-edge blowing are uncoupled at prestall angles
of attack. In this case, the overall forces and moments for symmetric blowing can
simply be deduced by superposition of asymmetric blowing situations. On the other
hand, the response of the vortical oweld is strongly coupled for asymmetric blow-
ing at poststall conditions. Wood et al.s results imply that tangential leading-edge
blowing may result in substantial rolling moments at conditions under which other
control devices cease to be effective.
Controllable leading-edge aps provide an active method of inuencing the large
vortices on a delta wing (Rao 1979; Marchman, Manor, and Plentovich 1980). These
devices appear to be capable of improving L/D ratio of a given wing primarily
through drag reduction. Unlike the control by blowing discussed above, the use of
aps results in only a modest improvement in the angle-of-attack envelope.
8.10 Parting Remarks
8.10.1 Recapitulation
The performance of many practical devices is often controlled by the separation lo-
cation. For a steady, two-dimensional boundary layer, the streamwise velocity and its
normal gradient vanish at the point (or line) of separation. For three-dimensional or
unsteady ows, the point of vanishing shear does not necessarily coincide with sepa-
ration. Under some circumstances, a thin layer of reverse owcan be embedded at the
bottom of an otherwise attached and mathematically well-behaved boundary layer.
Advancing or delaying boundary layer detachment is of immense importance to
the performance of biological and man-made systems involving uid ow. Postpon-
ing separation may lead to pressure-drag reduction, lift enhancement, stall delay,
and pressure-recovery improvement. Provoking ow detachment, on the other hand,
may improve the subsonic performance during takeoff and landing of a thin airfoil
optimized for supersonic cruising.
Methods of separation control that rely on modifying the shape of the velocity
prole near the wall include shaping, transpiration, and establishing a normal viscos-
ity gradient through surface heating or cooling, lm boiling, cavitation, sublimation,
chemical reaction, wall injection of a secondary uid having lower or higher viscosity,
and the introduction into the boundary layer of shear-thinning or shear-thickening
additive. To delay separation, any one or a combination of the preceding techniques
is congured to make the velocity prole as full as possible. The wall in this case
is a source of spanwise vorticity. To advance separation, the reverse technique is
employed to make the wall a sink of vorticity.
For a moving wall, separation is delayed when the motion of the surface is in
the same direction as that of the freestream. This phenomenon is exploited in sports
balls, Flettners rotor, and some experimental STOL-type aircraft. In analogy with
the moving wall case, unsteady separation is postponed when the separation point
moves upstream and is advanced when the point of detachment moves downstream.
Airfoils oscillating through high angles of attack can produce very high lift and
maintain ow attachment well beyond their static stall angles. Insects exploit these
unsteady separation effects to achieve remarkable aerodynamic characteristics.
In three-dimensional ows, the near-wall uid may move in a direction in which
the pressure gradient is more favorable and not against the adverse pressure in the
direction of the main ow, as is the case for two-dimensional ows. Consequently,
three-dimensional boundary layers are in general more capable of overcoming an
adverse pressure gradient without extensive or catastrophic separation. A turbulent
boundary layer is more resistant to separation than a laminar one, and mostly for that
reason transition advancement may be desired in some situations. In low-Reynolds-
number terminology, the transition-promoting devices are called turbulators and are
in the form of single, multiple, or distributed roughness elements placed on the wall.
The mechanical roughness elements, in the formof serrations, strips, bumps or ridges,
are typically placed near the airfoils leading edge.
Many other passive and active methods to postpone separation for low- and
high-speed ows are available. Common to all these control methods is an attempt
to supply additional energy to the near-wall uid particles that are being retarded in
the boundary layer. Passive techniques include intentional tripping of transition from
laminar-to-turbulent ow upstream of what would be a laminar separation point,
using boundary-layer fences to prevent separation at the tips of swept-back wings,
placing an array of vortex generators on the body to enhance the momentum and
energy in the neighborhood of the wall, utilizing a rippled trailing edge, employ-
ing streamwise corrugations, and using a screen to divert the ow and increase the
velocity gradient at the wall. Active methods to postpone separation include uid
injection parallel to the wall to augment the shear-layer momentum or normal to the
wall to enhance the mixing rate, the use of acoustic excitations, periodic forcing of
the velocity eld via an oscillating ap or wire, and oscillatory surface heating.
8.10.2 Computational Fluid Dynamics
The tremendous increases in CFD capability that have occurred as a direct result
of increases in computer storage capacity and speed are transforming ow separa-
tion control from an empirical art to a predictive science. Control techniques such
as mitigation of imposed pressure gradients, blowing, and suction are all readily
parameterized via viscous CFD. Current inaccuracies in turbulence modeling can
severely degrade CFDpredictions once separation has occurred; however, the essence
of separation control is the calculation of attached ows, estimation of separation
location, and indeed whether or not separation will occur. These tasks can in fact be
performed reasonably well via CFDwithin the uncertainties of the transition location
estimation (e.g., Smith 1975). This latter uncertainty has been signicantly reduced
for low-disturbance freestreams and smooth surfaces using CFD along with the
e
n
-method discussed in Chapter 6 (Bushnell and McGinley 1989). Therefore it is now
possible, at least to rst order, to design bodies over which the ow will not separate
(Cooke and Brebner 1961; Pinebrook and Dalton 1983; Yang et al. 1984; Dillner,
May, and McMasters 1984; Waggoner and Allison 1987; Szodruch and Shneider
1989), and this is generally done, for aircraft, for the cruise condition. This is not
currently done, for example, for helicopters and automobiles.
The cab-top fairings on tractor-trailer trucks (Kirsch 1974) and fairings at after-
body junctures (Howard et al. 1981) are useful experimental attempts at reducing
the causative adverse pressure elds. An exploratory study documented by White-
head and Bertram (1971) indicated that continuous-curvature surface geometry
can, presumably via minimization of pressure gradients, profoundly reduce three-
dimensional separation. Current CFD capability allows the design of intersection
regions that are properly lleted to obviate formation of the usual organized horse-
shoe vortices found in three-dimensional separated ows (e.g., Lakshmanan et al.
1988).
8.10.3 Applications and Comparisons
It should be evident from the discussion in this chapter thus far that there are a
large number of strategies that can control or mitigate ow separation. The choice of
which technique to employ is a function of the particular ow situation and purpose
for the control. Considerations include system aspects such as volume, weight,
complexity, cost, reliability, overall energy and drag budgets, and any requirements
for dynamic response and styling. Aside from design for separation minimization,
signicant applications thus far have been relatively limited though important.
Examples of successful separation control systems include active injection on several
1950s and 1960s era ghters, STOL transports, the ubiquitous vortex generators,
passive bleed for supersonic inlets, leading edge extensions for the 1970s and 1980s
ghters, cab top fairing for tractor-trailer trucks, and various types of blown aps
on transport airplanes.
Particularly intriguing in the future are possible applications at cruise, especially at
high speeds, as well as standby techniques for off-design situations, thereby allowing
extremely tight designs. In addition, increased knowledge of, and control innovations
for, vortical ows should allowsolutions to the problems of high-angle-of-attack ma-
neuvering and stall and spin prevention or recovery. Alternate techniques should also
be sought for either supplementation or replacement of conventional high-lift ap
systems. Candidate approaches include use of the LFC leading-edge suction system
along with air jet vortex generators and perhaps either rotating wings or airport ski
jumps. Also, tremendous energy saving potential exists in further separation control
or form drag reduction for vehicles that have notoriously high drag: automobiles,
helicopters and tractor-trailer trucks.
Various comparisons have been made between candidate separation control tech-
niques on common test beds. Vakili et al. (1985) reported that vortex generators
perform better than a ow-control rail for a diffusing S-duct. For a swept wing,
experiments indicated that slot blowing is more effective than either vane or air-jet
vortex generators (Kukainis 1969). For CTOL wings, improvements in the design of
mechanical high-lift systems have tended to keep pace with the trend toward higher
wing loading, and thus benets from BLC techniques have not appeared sufciently
attractive (Gratzer 1971). Butter (1984) and Dillner et al. (1984) provided excellent
analyses and discussions of current CTOL high-lift design practice, problems, and
performance. As a suggestion for further work, a technique that has not yet been re-
searched sufciently is the three-dimensionalization of a nominally two-dimensional
problem which, from the work of Ball (1971), McLean and Herring (1974), and Lin
and Howard (1989) tends to reduce the extent of separation.
8.10.4 Recommendations
Flow-separation control will be of increasing importance in the future as declin-
ing petroleum reserves, concern over the greenhouse effect, economic development
of the Third World, continued population growth, and increasing economic com-
petition force stringent energy conservation and efciency improvements. Much of
the remaining gains in aerodynamics involve some form of viscous ow control,
including ow-separation control. Particularly intriguing is the possibility of replac-
ing or at least augmenting conventional high-lift devices (i.e., aps and slats). In the
limit, this may require wing rotation because some increase in wing angle of attack
would probably be required.
With the exception of vortex generators, the rapid development of CFD has
transformed much of conventional ow separation control from art to science, in-
cluding the capability of mitigating the causative pressure elds and optimizing sev-
eral control approaches. However, the effects of combinations of separation control
approaches and the potential for synergistic benets require considerable further
research.
The use of ow-separation control at cruise for high-speed, supersonic civil trans-
ports may allow signicant increases in lift-to-drag ratio via increased leading-edge
thrust, upper surface and fuselage lift contribution, and effective favorable interfer-
ence wave drag reduction fromshock-boundary-layer interaction separation control.
The art and science of owseparation control is far richer than the conventional view
of blowing or suction and vortex generators. The recent research regarding miniatur-
ized and jet-injection vortex generators allows reduction of parasitic cruise drag for
such devices and is particularly intriguing for several applications, including replace-
ment or supplementation of conventional high-lift systems. Microelectromechanical
systems (MEMS) offer the potential for extremely small actuators and thus even more
effective and less expensive means for ow control (Ho and Tai 1996, 1998). We will
return to this topic in Chapter 14.
Suggested future research directions include further work on microdevices de-
signed specically for separation control; optimization of air-jet vortex generators,
including pulsed injection; three-dimensionalization of nominally two-dimensional
surfaces; use of weak upstreamshocks for turbulence amplication and swept shocks
for spanwise removal of the near-wall low-momentum region; downstream vortex
reenforcers and miniaturized near-wall vortex generators; and techniques to force
momentum toward the wall (e.g., downwash from embedded lifting surfaces).
CHAPTER NINE
Low-Reynolds-Number Aerodynamics
Oh, how much is today hidden by science! Oh, how much it is expected
to hide!
(Friedrich Wilhelm Nietzsche, 18441900)
My philosophy of life is work. Bringing out the secrets of nature and
applying them for the happiness of man. I know of no better service to
render during the short time we are in this world.
(Thomas Alva Edison, 18471931)
PROLOGUE
Among the goals of external ow control are separation postponement, lift enhance-
ment, transition delay or advancement, and drag reduction. These objectives are not
necessarily mutually exclusive. For low-Reynolds-number lifting surfaces, where the
formation of a laminar separation bubble may have a dominant effect on the ow-
eld, the interrelation between the preceding goals is particularly salient, present-
ing an additional degree of complexity when ow control is attempted to achieve,
say, maximum lift-to-drag ratio. This chapter discusses the aerodynamics of low-
Reynolds-number lifting surfacesparticularly the formation and control of separa-
tion bubbles.
9.1 Introduction
Insects, birds and bats have perfected the art of ight through millions of years of
evolution. Mans dream of ying dates back to the early Greek myth of Daedalus
and his son Icarus, but the rst successful heavier-than-air ight took place less than
a century ago. Today, the Reynolds numbers for natural and man-made iers span
the amazing range from 10
2
to 10
9
, insects being at the low end of this spectrum and
huge airships occupying the high end (Carmichael 1981).
The function of the airfoil section on those iers is to produce lift. Inevitably,
viscous effects, compressibility effects, and the nite span of the lifting surface all
ensure that drag is also produced. A thrust must be generated by some sort of a
power plant to overcome this streamwise resistance to the motion. The lift-to-drag
189
190 LOW-REYNOLDS-NUMBER AERODYNAMICS
Figure 9.1: Airfoil performance as function of chord Reynolds number [after
McMasters and Henderson 1980].
ratio is a measure of the effectiveness of the airfoil. In general, this ratio is very low
at low Reynolds numbers and improves with increases in this parameter. As shown
in Figure 9.1, reproduced from McMasters and Henderson (1980), the maximum
[C
L
/C
D
] improves dramatically in the range of chord Reynolds numbers of Re
c
=
10
4
10
6
. Below 10
4
, typical of insects and small model airplanes, the boundary
layer around the lifting surface is laminar. Stalling in this case is caused by an abrupt
separation of the laminar ownear the leading edge as the angle of attack is increased
to modest values.
1
The maximum lift is limited, and the drag increases signicantly
when the lifting surface stalls.
For Re
c
> 10
6
, typical of large aircraft, boundary-layer transition to turbulence
usually takes place ahead of the theoretical laminar separation point. A turbulent
boundary layer can negotiate quite severe adverse pressure gradients without separa-
tion, and this kind of lifting surface often experiences a trailing-edge stall at relatively
high angles of attack. The stall is preceded by a movement of the separation point
forward from the trailing edge with increasing incidence (McCullough and Gault
1951).
In the range of Reynolds numbers of 10
4
10
6
, termed low Reynolds number
for the purpose of this chapter (Lissaman 1983), many complicated phenomena
take place within the boundary layer. Separation, transition and reattachment can
all occur within a short distance and dramatically affect the performance of the
lifting surface. The laminar separation bubble that commonly forms in this range
of Reynolds numbers plays an important role in determining the boundary layer
behavior and the stalling characteristics of the airfoil (Tani 1964). As indicated in
Figure 9.1, the maximum lift-to-drag ratio for a smooth airfoil increases by two
orders of magnitude in this Reynolds number regime. Remotely piloted aircraft and
turbine blades are examples of lifting surfaces having this range of Reynolds numbers.
1
The arguments and the gure herein are for steady-state ows. Insects and other very low-Reynolds-
number iers exploit unsteady effects to achieve remarkable performances, as was discussed in Chapter 8.
9.2 LOW-REYNOLDS-NUMBER AIRFOILS 191
The skilled designer has available a variety of passive and active techniques to
effect a benecial change in the complex oweld that characterizes this intermediate
range of Reynolds numbers. Roughness and shaping are among the simplest passive
methods to ensure ow attachment beyond a critical angle of attack and, thus, an
improved performance. Wall transpiration and heat transfer are examples of active
control methods to improve the lift-to-drag ratio. Although these broad ow-control
strategies are similar to those discussed elsewhere in this book, the emphasis here is on
the low-Reynolds-number regime dominated by the formation of laminar separation
bubbles.
This chapter discusses the aerodynamics of low-Reynolds-number lifting surfaces,
particularly the formation and control of separation bubbles. Flow-control goals in
those cases are strongly interrelated leading to certain difculties in choosing the
best control strategy for a particular end result. These potential conicts will be
elaborated.
9.2 Low-Reynolds-Number Airfoils
In the range of Reynolds numbers of Re
c
=10
4
10
6
, a substantial improvement in
the lift-to-drag ratio of an airfoil takes place. According to Carmichael (1981), this
is the Reynolds number regime in which we nd man and nature together in ight:
large soaring birds; large radio-controlled model aircraft; foot-launched ultralight,
man-carrying hang-gliders; human-powered aircraft; and the more recently devel-
oped remotely-piloted-vehicles (RPVs) used for military and scientic sampling,
monitoring, and surveillance. Three review articles on low-Reynolds-number aero-
dynamics by Tani (1964), Lissaman (1983), and Mueller (1985a) are particularly
recommended. There is also a wealth of information available in the proceedings of
the following conferences dedicated to the subject matter: (1) Conference on Low
Reynolds Number Airfoil Aerodynamics, University of Notre Dame, Notre Dame,
Indiana, 1719 June 1985 (Mueller 1985b); (2) International Conference on Aero-
dynamics at Low Reynolds Numbers 10
4
< Re <10
6
, Royal Aeronautical Society,
London, Great Britain, 1518 October 1986; (3) Conference on LowReynolds Num-
ber Aerodynamics, University of Notre Dame, Notre Dame, Indiana, 57 June 1989
(Mueller 1989); and (4) Eighth International Conference on Remotely Piloted Vehi-
cles, Bristol, Great Britain, 24 April 1990.
In this range of Reynolds numbers, very complex ow phenomena take place
within a short distance on the upper surface of an airfoil at incidence. Unless arti-
cially tripped, the boundary layer remains laminar at the onset of pressure recovery,
and the airfoils performance is then entirely dictated by the laminar ows poor re-
sistance to separation. The separated ow forms a free-shear layer, which is highly
unstable, and transition to turbulence is readily realized. Subsequent reattachment of
the separated region may take place because of the increased entrainment associated
with the turbulent ow. Provided that the high-speed uid entrained into the wall re-
gion supplies sufcient energy to maintain the circulating motion against dissipation,
a separation bubble forms.
The precise conditions for the occurrence of separation, transition, and reattach-
mentin other words for the formation of a laminar separation bubbledepend on
the Reynolds number, the pressure distribution, the surface curvature, the surface
Figure 9.2: Sketch of a laminar separation bubble.
roughness, and the freestream turbulence as well as other environmental factors. If
the Reynolds number is sufciently high, transition takes place near the minimum
pressure point ahead of the location at which separation would have occurred if the
boundary layer had remained laminar. For moderate Reynolds numbers, separation
takes place before transition. The laminar boundary layer can only support very small
adverse pressure gradient without separation. As stated in Chapter 8, if the ambient
incompressible uid decelerates in the streamwise direction faster than U
o
x
0.09
,
the owseparates. The separated owwill not reattach to the surface, and no bubble
will be formed if the Reynolds number is sufciently low. However, for the intermedi-
ate Reynolds number range (typically 10
4
10
6
), the separated owproceeds along the
direction of the tangent to the surface at the separation point (von Doenhoff 1938a)
and transition to turbulence takes place in the free-shear layer owing to its increased
transition susceptibility. Subsequent turbulent entrainment of high-speed uid causes
the ow to return to the surface, thus forming what is known as a laminar separa-
tion bubble, as sketched in Figure 9.2. Downstream of the point of reattachment, the
newly formed turbulent boundary layer is capable of negotiating quite severe adverse
pressure gradients without separation. The ability of a turbulent boundary layer to
resist separation improves as the Reynolds number increases (Lissaman 1983).
9.3 Conditions for Bubble Formation
It is clear fromthe preceding arguments that bubble formation is conned to a certain
range of Reynolds numbers and that this range changes from one airfoil to another
as well as from one environment to another. A rough rule according to Carmichael
(1981) is that the Reynolds number based on freestream velocity and the distance
fromseparation to reattachment is approximately 510
4
. In general, then, an airfoil
with chord Reynolds number less than 510
4
will experience laminar separation
9.3 CONDITIONS FOR BUBBLE FORMATION 193
Figure 9.3: Qualitative pressure distributions for two airfoils at in-
cidence. () Potential ow solution; () experiment. (a)
Short bubble on upper surface and subsequent rear separation of
turbulent boundary layer; (b) Long bubble.
with no subsequent reattachment. For chord Reynolds numbers slightly higher than
5 10
4
, a long bubble is expected. Shorter bubbles are formed at higher Reynolds
numbers. Tani (1964) asserted that a Reynolds number typical of local conditions
in the boundary layer is more appropriate to characterize a separation bubble than
the chord Reynolds number. Typically, the Reynolds number, based on the boundary-
layers displacement thickness
2
and the velocity just outside the rotational owregion
at the point of separation, is more than 500 for a short bubble
3
and less than 500
for a long one. The corresponding bubbles streamwise extent, normalized with the
displacement thickness at the point of separation, is 10
2
and 10
4
, respectively (Tani
1964).
The short separation bubble generally has a length of the order of a few percent
of the chord. It merely represents a transition-forcing (tripping) mechanism to allow
reattachment of an otherwise separated shear layer. Such a bubble does not greatly
affect the peak suction, as determined from the potential ow solution around the
airfoil. Except for the appearance of a minute bump in the lift curve (C
L
versus
curve, where C
L
is the lift coefcient and is the angle of attack), the presence
of a short bubble has no signicant effect on the pressure distribution around the
lifting surface, as depicted in Figure 9.3a. On the other hand, a long bubble may be as
much as 0.2c0.3c, where c is the airfoil chord, and signicantly changes the pressure
distribution by effectively altering the shape over which the outer potential ow is
developed. In this case, the sharp suction-peak near the leading edge is generally not
realized, and a suction plateau of a reduced level extends over the region occupied
by the bubble (Figure 9.3b). A long bubble tends to increase in length as incidence
2
The displacement thickness
for a laminar boundary layer near the leading edge of an airfoil is extremely
small and cannot accurately be measured. Instead the momentum thickness
is computed from the

pressure distribution using Thwaites (1949) formula, and the ratio (
) at the separation point is

assumed to be 3.7.
3
The bubble gets shorter at higher Re
; and at Re
6000, bubble formation is precluded (bubbles

streamwise extent approaches zero) by transition to turbulence in the boundary layer.
is increased, leading to a corresponding decrease in the slope of the lift curve as well
as an increase in the pressure drag.
9.4 Bubbles Breakdown
In general, the lift-to-drag ratio is higher for an airfoil having a shorter bubble. As the
result of many factors, a short bubble forming at low incidence may move forward
and contract in streamwise extent as the angle of attack is increased (Tani 1964).
Within the bubble, a small region of constant pressure exists followed by pressure
recovery. At higher incidence, the bubble bursts and no longer reattaches, thus a
leading-edge stall ensues (Jones 1934). This process is often irreversible, meaning that
reducing the angle of attack will not immediately unburst the bubble. As sketched in
Figure 9.4, strong lift hysteresis effects are thus observed as the attack angle or the
Reynolds number is recycled (Schmitz 1967; Mueller 1985c; Bastedo and Mueller
1985; Brendel and Mueller 1988). Similar hysteresis effects are also seen in drag.
For thin airfoils of small nose-radius, pressure recovery commences very near the
leading edge, and the adverse pressure gradients are severe at high angles of attack.
Separation bubble may occur on these airfoils even at chord Reynolds numbers ex-
ceeding 10
6
. At large incidence, the short bubble breaks down into a long one. With
Figure 9.4: Illustration of lift hysteresis as the angle of attack
(top plot) or the Reynolds number (bottom plot) is recycled.
9.5 CONTROL GOALS AND THEIR INTERRELATION 195
increasing angles of attack, the reattachment point moves progressively backward
until it reaches the trailing edge, at which stage its maximum thickness is typically
3% of the chord. A further increase in incidence leads to completely detached ow
and the so-called thinairfoil stall. A comprehensive review of the different kinds of
stall on thin airfoils is given by Crabtree (1957).
Aerodynamics data for both thin and thick airfoils in the low-Reynolds-number
regime are accessible in several recent articles and conference proceedings (e.g., Burns
1981; Mueller and Burns 1982; Render 1984; Render, Stollery, and Williams 1985;
Leibeck and Camacho 1985; Bastedo and Mueller 1986; Stollery and Dyer 1989;
Mueller 1985b, 1989). Available experimental data on bubbles formation and burst-
ing indicates that transition to turbulence in the separated shear-layer and subsequent
reattachment will occur if the Reynolds number based on displacement thickness at
the point of laminar separation exceeds a critical value that is not necessarily univer-
sal (TaniOwenKlanfer criterion). A lower limit for this Reynolds number seems to
be Re
350. Bursting occurs if the pressure recovered in the reattachment process

in terms of the dynamic pressure at separation (pressure recovery coefcient) exceeds
a certain critical value (Crabtree criterion). Again, this critical value changes from
one airfoil to another, but an upper limit of 0.35 appears to be valid for many shapes.
Crabtree (1954) assumes that bubbles breakdown occurs because a maximumpossi-
ble value of pressure exists that can be recovered in the turbulent entrainment process
that causes the ow reattachment. This implies the existence of a maximum possible
value of the shear stress setup in the turbulent entrainment region so as to counteract
the pressure gradient. At breakdown, caused by either an increase in incidence or a
decrease in Reynolds number, the TaniOwenKlanfer criterion is satised, but the
Crabtree criterion is about to be violated.
The question of primary concern to us in this chapter is how to control the
ow around a low-Reynolds-number airfoil to achieve an improved performance.
The interrelation between the different control goals is particularly salient when a
separation bubble exists, and this issue will be tackled in the next section.
9.5 Control Goals and Their Interrelation
According to Tani (1964), all three kinds of stall, trailing-edge stall, leading-edge stall,
and thin-airfoil stall, may occur for a given airfoil at different Reynolds numbers or
for different airfoils at a given Reynolds number. Aparticular lifting surface produces
higher lift at higher incidence limited by the angle at which the airfoil stalls. At that
point, drag increases dramatically, and the lifting surface performance deteriorates
rapidly. Flow control is aimed at improving this performance. Among the practical
considerations that must be taken into account for both active and passive control
devices are their cost of construction and operation, complexity, and potential trade-
offs or penalties associated with their use. It is this latter point in particular that
presents an additional degree of complexity for controlling low-Reynolds-number
lifting surfaces. Achieving a benecial effect for one control goal may very well
adversely affect another goal, and design compromises must often be made.
Among the desired goals of external ow modication are separation or
reattachment control, transition delay or advancement, and lift enhancement and
drag reduction. These objectives are not necessarily mutually exclusive, and for
Figure 9.5: Interrelation between ow-control goals.
low-Reynolds-number lifting surfaces the interrelations between these goals are
particularly conspicuous, presenting an additional degree of complexity. As men-
tioned before, in the range of Reynolds numbers of 10
4
10
6
, a laminar separation
bubble may form and may have a dominant effect on the oweld and the airfoils
performance.
Recall the discussion in Section 3.1. The schematic in Figure 9.5 is a partial rep-
resentation of the interrelation between one control goal and another for a lifting
surface. If the boundary layer becomes turbulent, its resistance to separation is en-
hanced, and more lift could be obtained at increased incidence. On the other hand,
the skin-friction drag for a laminar boundary layer can be as much as an order of
magnitude less than that for a turbulent one. If transition is delayed, lower skin fric-
tion as well as lower ow-induced noise is achieved. However, the laminar boundary
layer can only support very small adverse pressure gradient without separation, and
subsequent loss of lift and increase in form drag occur. Once the laminar boundary
layer separates, a free-shear layer forms, and for moderate Reynolds numbers tran-
sition to turbulence takes place. Increased entrainment of high-speed uid due to
the turbulent mixing may result in reattachment of the separated region and forma-
tion of a laminar separation bubble. At higher incidence, the bubble breaks down,
either separating completely or forming a longer bubble. In either case, the form
drag increases, and the lift-curves slope decreases. The ultimate goal of all this is
9.6 SEPARATION CONTROL 197
to improve the airfoils performance by increasing the lift-to-drag ratio. However,
induced drag is caused by the lift generated on a lifting surface with a nite span.
Moreover, more lift is generated at higher incidence, but form drag also increases at
these angles.
All of the above points to potential conicts as one tries to achieve a particular
control goal only to affect another goal adversely. An ideal method of control that
is simple, inexpensive to build and operate, and does not have any trade-off does
not exist, and the skilled engineer has to make continuous compromises to achieve
a particular design goal. Keeping this in mind, we now proceed to review selected
control methods available to the designer of low-Reynolds-number lifting surfaces
such as turbine blades and remotely piloted aircraft.
9.6 Separation Control
Fluid particles in a boundary layer are slowed down by wall friction. If the external
potential ow is sufciently retarded owing, for example, to the presence of an ad-
verse pressure gradient, the momentum of those particles will be consumed by both
the wall shear and the pressure gradient. At some point (or line), the viscous layer
departs or breaks away from the bounding surface. The surface streamline nearest to
the wall leaves the body at this point, and the boundary layer is said to separate. At
separation, the rotational ow region next to the wall abruptly thickens, the normal
velocity component increases, and the boundary-layer approximations are no longer
valid. Because of the large energy losses associated with boundary-layer separation,
the performance of a lifting surface is often controlled by the separation location.
If separation is postponed, the pressure drag is decreased, and the circulation, and
hence the lift at high angles of attack, is enhanced.
As discussed in the previous chapter, separation control methods include the
modulation of pressure gradient via shaping, wall suction, surface cooling in gases,
and surface heating in liquids. The rst of these strategies is the simplest and is most
suited for low-Reynolds-number situations. Streamlining can greatly reduce the
steepness of the pressure rise. In contrast to turbulent ows, laminar boundary layers
can only support very small adverse pressure gradients without separation. Transition
on the upper surface of a lifting surface typically occurs at the rst onset of adverse
pressure gradient if the Reynolds number exceeds 10
6
. The separation-resistant
turbulent boundary layer that evolves in the pressure recovery region results in higher
maximum lift at relatively large angle of stall. Depending on the severity of the initial
adverse gradient, and hence on the airfoil shape, laminar separation may take place
prior to transition. Regardless of whether or not the ow subsequently reattaches,
the laminar separation leads to higher form drag and lower maximum lift. Delicate
contouring of the airfoil near the minimum pressure point to lessen the severity of
the adverse pressure gradient may be used to accomplish separation-free transition.
As an example of the effects of the lifting surfaces shape on its performance,
consider the lift curves for the three airfoil sections NACA 63
3
018, NACA 63009,
and NACA 64A006. These airfoils have maximum thicknesses of 0.18c, 0.09c, and
0.06c, respectively, where c is the chord. The respective leading edge radii are 0.021c,
0.006c, and 0.003c. Figure 9.6, adapted fromthe measurements by McCullough and
Gault (1951), depicts C
L
versus curves for the three sections at a chord Reynolds
Figure 9.6: Lift curves for three airfoils at
Re
c
= 5.8 10
6
. (a) NACA 63
3
018; (b) NACA
63009; (c) NACA64A006 [after McCullough and
Gault 1951].
number of 5.8 10
6
. For the thick sec-
tion, NACA 63
3
018, transition takes
place near the minimum pressure point
(Fitzgerald and Mueller 1990). Stalling
in this case is of the trailing-edge type
and is preceded by a gradual movement
of the separation point of the turbulent
boundary layer forward from the trail-
ing edge as increases. A laminar sepa-
ration bubble is formed on the other two
sections at small incidence. However, the
NACA63009 section experiences a sud-
den leading-edge stall when the bubble
bursts with no subsequent reattachment,
whereas the NACA 64A006 section ex-
periences a more gradual thin-airfoil stall
(McCullough 1955; Crabtree 1957). In
the latter case, the short bubble breaks
down into a longer bubble at an angle of
attack of 5
, causing a slight discontinu-

ity in the lift curve. Subsequent increase
in leads to a movement of the reattach-
ment point toward the trailing edge. The
maximum lift in this case is about 40% lower than that for the thick airfoil. The stall
angle is also lower.
A second example is provided for an airfoil specically designed for the low-
Reynolds-number regime. The carefully contoured Eppler 61 has a maximum thick-
ness of 0.056c and is highly cambered. Mueller and Burns (1982) reported lift, drag,
and smoke visualization data for this airfoil section in the range of Reynolds num-
bers of 3 10
4
2 10
5
. A sample of their lift and drag curves at three different
speeds is depicted in Figure 9.7. At a negative angle of attack of about = 3
, the
ow around the cambered airfoil separates at the leading edge on the lower surface
without further reattachment. Zero lift is measured at this angle and is correlated
with the appearance of smooth smokelines above and below the airfoil to form an
uncambered, symmetrical shape, as shown in the top photograph in Figure 9.8. In
this gure, the chord Reynolds number is Re
c
= 8.710
4
, and three angles of attack
are depicted: = 3
, = 0
, and = 8
. The results explain the deviation of the

zero-lift angle as well as the shape of the lift curve fromthose predicted by thin airfoil
theory. Strong Reynolds number effects are evident in both the lift and drag curves.
At increasing angles of attack, the performance of the Eppler 61 airfoil is similar
to that of the other thin airfoil depicted in Figure 9.6c, although the maximum lift
coefcient is higher in the former case.
9.7 Transition Control via Shaping
Streamlining a body to prevent separation and reduce form drag is quite an old
art, but the stabilization of a boundary layer by pushing the longitudinal location
of the pressure minimum as far back as possible dates to the 1930s and led to the
9.7 TRANSITION CONTROL VIA SHAPING 199
Figure 9.7: Reynolds number effects on lift (top curves) and drag (bottom
curves) of the Eppler 61 airfoil [after Mueller and Burns 1982].
Figure 9.8: Smoke-wire ow visualization of the low-Reynolds-number Eppler 61 airfoil.
Re
c
= 8.7 10
4
. (a) = 3
; (b) = 0
; (c) = 8
[after Mueller and Burns 1982].

9.7 TRANSITION CONTROL VIA SHAPING 201
successful development of the NACA 6-Series NLF airfoils. Newer, low-Reynolds-
number lifting surfaces used in sailplanes, low-speed drones, and executive business
jets have their maximumthickness point far aft of the leading edge. The recent success
of the Voyagers 9-day, unrefueled ight around the world was due in part to a wing
design employing natural laminar ow to approximately 0.5c.
The favorable pressure gradient extends to the longitudinal location of the pres-
sure minimum. Beyond this point, the adverse pressure gradient becomes steeper and
steeper as the peak suction is moved further aft. For an airfoil, the desired shift in the
point of minimum pressure can only be attained in a certain narrow range of angles
of incidence. On the basis of the shape, angle of attack, Reynolds number, surface
roughness, and other factors, the boundary layer either becomes turbulent shortly
after the point of minimum pressure or separates rst and then undergoes transition.
One of the design goals of NLF is to maintain attached ow in the adverse pressure
gradient region, and some method of separation control may have to be used there.
Subtle changes in the magnitude and extent of favorable pressure gradient, lead-
ing edge radius, and other shape variables can have pronounced effects on the
airfoils performance. As an example, consider the lift and drag characteristics of
a conventional section (NACA 23015) and a laminar-ow section (NACA 66
2
215). Both airfoils are cambered, and both have maximum thickness of 0.15c.
However, the maximum thickness point is located at 0.3c for the conventional
section and at 0.45c for the laminar-ow one. The leading-edge radius is 0.025c
and 0.014c for the two respective airfoils. The point of minimum pressure, as
computed for the basic symmetric section at zero incidence, is located at 0.15c
for the conventional airfoil, whereas this point is pushed back to 0.6c for the
Figure 9.9: Lift curves for (a) conventional sec-
tion (NACA 23015); and (b) laminar-ow sec-
tion (NACA 66
2
215). Chord Reynolds number
in both cases is Re
c
= 9 10
6
[after Abbott and
von Doenhoff 1959].
laminar-ow airfoil. As seen from the lift
curves depicted in Figure 9.9 for a chord
c
= 9 10
6
, the
laminar-ow section has slightly higher lift
at small angles of attack than the conven-
tional section, but stalling occurs at lower
incidence and the maximum lift is smaller
for the laminar-ow airfoil. The lift-drag
polars for the same two airfoils are shown
in Figure 9.10. The sudden increase in drag
for the laminar-ow section occurs at an
angle of attack of about = 1.5
. This
is caused by a forward movement of the
minimum pressure point and correspond-
ing accentuation of the adverse pressure
gradient as the ow on the upper surface
accelerates sharply to round the airfoils
sharp nose. The increased adverse pressure
gradient leads to early transition and cor-
responding drag increase. The maximum
lift-to-drag ratios for the conventional and
laminar-ow airfoils are 125 and 102, re-
spectively. However, the laminar-ow sec-
tion is designed to cruise in the low-drag
Figure 9.10: Lift-drag polars for the two airfoils depicted in
Figure 9.9 [after Abbott and von Doenhoff 1959].
region ( <1.5
). At the design lift coefcients of 0.3 and 0.2 for the conventional
and laminar-ow sections, the lift-to-drag ratios are 47.6 and 59.5, respectively.
Factors that limit the utility of NLF include crossow instabilities and leading-
edge contamination on swept wings, insect, and other particulate debris; high unit
Reynolds numbers of conventional aircraft at lower cruise altitudes (not a limiting
factor for low-Reynolds-number aircraft); and performance degradation at higher
angles of attack due to the necessarily small leading-edge radius of NLF airfoils.
Reductions of surface waviness and smoothness of modern production wings, special
leading-edge systems to prevent insect impacts and ice accretion (Korkan, Cross, and
Cornell 1986; Cebeci 1989; Zaman and Potapczuk 1989), higher cruise altitudes of
newer airplanes, and higher Mach numbers all favor the application of NLF (Runyan
and Steers 1980).
9.8 Turbulators
A turbulent boundary layer is more resistant to separation than a laminar one, and
mostly for that reason transition advancement may be desired in some situations.
In low-Reynolds-number terminology, the transition promoting devices are called
turbulators. For a zero-pressure-gradient boundary layer, transition typically occurs
at a Reynolds number based on distance from leading edge of the order of 10
6
. The
critical Reynolds number Re
crit
below which perturbations of all wave numbers de-
cay is about 6 10
4
. To advance the transition Reynolds number, one may attempt
to lower the critical Re, increase the growth rate of TollmienSchlichting waves, or
introduce large disturbances that can cause bypass transition. The rst two routes
involve altering the shape of the velocity prolemaking it less fullusing wall
motion, injection, adverse pressure gradient, surface heating in gases or cooling in
liquids, or other strategies to increase the near-wall viscosity. The third route, expos-
ing the boundary layer to large disturbances, is much simpler to implement though
more difcult to analyze (Smith and Kaups 1968; Cebeci and Chang 1978; Nayfeh
et al. 1986; Cebeci and Egan 1989).
Morkovin (1984) broadly classied the large disturbances that can cause by-
pass transition into steady or unsteady ones originating in the freestream or at the
body surface. The most common example is single, multiple, or distributed rough-
ness elements placed on the wall. The mechanical roughness elements, in the form
of serrations, strips, bumps, or ridges, are typically placed near the airfoils leading
edge. If the roughness characteristic length is large enough, the disturbance intro-
duced is nonlinear, and bypass transition takes place. For a discrete three-dimensional
roughness element of height-to-width ratio of one, Tani (1969) reported a transition
300 for a roughness Reynolds number of Re
10
3
.
Here, Re
is based on the velocity outside the boundary layer and the displacement
thickness (Re
U
o

/), and Re
is based on the height of the roughness element

and the velocity in the undisturbed boundary layer at the height of the element
(Re
U() /). Note that the transition Reynolds number, Re
, indicated above
|
crit
=420 predicted from the linear stability theory. For a
roughness Reynolds number of about 600, transition occurs at Re
10
3
. For a
smooth surface, transition typically takes place at Re
2.6 10
3
. An important
consideration when designing a turbulator is to produce turbulence and suppress
laminar separation without causing the boundary layer to become unnecessarily
thick. A thick, turbulent, wall-bounded ow suffers more drag and is more suscep-
tible to separation than a thin one. Consistent with this observation, available data
(Figure 9.1) indicate that a rough airfoil has higher lift-to-drag ratio than a smooth
one for Re
c
<10
5
but that this trait is reversed at higher Reynolds numbers.
For low-Reynolds-number airfoils, performance may be improved by reducing the
size of the laminar separation bubble through the use of transition ramps (Eppler and
Somers 1985), boundary layer trips (Davidson 1985; Van Ingen and Boermans 1986),
or even pneumatic turbulators (Pfenninger and Vemuru 1990). Donovan and Selig
(1989) provided extensive data using both methods for 40 airfoils in the Reynolds
number range of Re
c
= 6 10
4
3 10
5
. A long region of roughly constant adverse
pressure gradient on the upper surface of a lifting surface (termed a bubble ramp)
achieves a lower drag than the more conventional laminar-type velocity distribu-
tion in which the pressure initially remains approximately constant and then quickly
recovers. Trips were also used in Donovan and Seligs experiments to shorten the sepa-
ration bubble. Asimple two-dimensional trip performed as well or better than zig-zag
tape, hemisphere bumps, and normal blowing. Donovan and Selig concluded that
an airfoil that performs poorly at low Reynolds numbers can be improved through
the use of turbulators. Trips were less effective, however, at improving airfoils that
normally had low drag. Other large disturbances that can lead to early transition
include high turbulence levels in the freestream, external acoustic excitations, partic-
ulate contamination, and surface vibration. These are often termed environmental
tripping. Transition can also be effected by detecting naturally occurring TS waves
and articially introducing in-phase waves. Transition can be considerably advanced,
on demand, using this wave superposition principle.
9.9 Parting Remarks
Available and contemplated ow control methods particularly suited for low-
Reynolds-number lifting surfaces have been surveyed. The ow around these air-
foils is dominantly affected by the formation of a separation bubble. The laminar
separation makes the interrelation between transition, separation, lift, and drag con-
trols particularly salient, presenting an additional degree of complexity. Several of
these control techniques are suited for high- as well as low-speed airfoils.
Low-Reynolds-number airfoils span the range of 10
4
10
6
. In this regime, laminar
separation, transition, and reattachment may all occur within a short distance on the
upper surface of an airfoil at incidence. The precise conditions for the occurrence of
a laminar separation bubble depend on the local Reynolds number, pressure distri-
bution, surface curvature, surface roughness, and freestream turbulence as well as
other environmental factors.
To control the ow around a low-Reynolds-number airfoil to achieve improved
performance, one must carefully consider potential conicts in trying to achieve a
particular control goal while inadvertently causing an adverse effect on another goal.
A laminar boundary layer is less able to resist separation but is characterized by a
very low skin-friction drag. However, if the ow separates, lift decreases and form
drag increases substantially. In the low-Reynolds-number regime, laminar separation
is often followed by transition to turbulence in the separated free-shear layer and
subsequent reattachment to forma closed bubble. Bubble bursting at higher incidence
leads to loss of lift and increased drag.
CHAPTER TEN
Drag Reduction
There does not exist a category of science to which one can give the name
applied science. There are science and the applications of science, bound
together as the fruit of the tree which bears it.
(Louis Pasteur, 18221895)
Science becomes dangerous only when it imagines that it has reached its
goal.
(George Bernard Shaw, 18561950)
PROLOGUE
Drag is the force by which a uid resists the relative motion of a solid. An equal
and opposite reaction force acts on the body surface as a result of the uid deforma-
tion, and drag is the component of this force parallel to the direction of the relative
velocity vector. The uid can be external or internal to the solid boundaries, and
the solid surface can be rigid or compliant. Billions of gallons of fossil fuel are used
annually to overcome the drag encountered by vehicles moving in air or water and
the uid resistance in gas, water, or oil pipelines. Flow control aims at minimizing
this drag force, and the subject is explicitly or implicitly interwoven in every chapter
of this book. Delaying laminar-to-turbulence transition (Chapter 6) is usually sought
in order to benet from the much lower skin friction associated with laminar bound-
ary layers. Preventing separation (Chapter 8) means reducing the pressure drag, and
when separation is provoked, as for example in delta wings, it is desirable to keep the
associated drag penalty to a minimum. Compliant coatings (Chapter 7) are sought
to reduce drag as well as to achieve other benecial effects. Low-Reynolds-number
vehicles (Chapter 9) notoriously suffer from low lift-to-drag ratio, and ow control
attempts to remedy that. When higher rates of mass, momentum, or heat transfer are
desired (Chapter 11), it is imperative that the concomitant drag penalty be kept to a
minimum. The futuristic control systems envisioned in Chapter 14 deal for the most
part with drag reduction. Reducing drag saves enormous resources spent to overcome
it and is clearly advantageous to the environment, the economy, and the overall in-
dustrial competitiveness of any country. This chapter is a central, albeit brief, treatise
on drag reduction that unites the disparate pieces of the puzzle discussed in more
detail elsewhere in the book and ties a few loose ends not covered in other chapters.
205
206 DRAG REDUCTION
10.1 Background
Whenever there is relative motion between a solid body and the uid in which it is
immersed, the body experiences a net force due to the action of the uid. As depicted
in Figure 10.1, this surface force can be decomposed into (1) a drag component
parallel to the relative velocity vector, (2) a lift component perpendicular to the
direction of motion, and (3) a side force resulting from any body asymmetries about
the liftdrag (LD) plane. For bodies that are symmetric about this plane, the side
force, yawing moment, and rolling moment all vanish, leaving only lift, drag, and
pitching moment acting on the body. If the body has a second plane of symmetry
perpendicular to the LDplane, the principal chord line of the body is the intersection
of those two symmetry planes and is parallel to the freestream. In this case, both lift
and pitching moment also vanish, and the body experiences only drag.
The total force acting on the body is equal and opposite to that on the uid and
is obtained by integrating the shear stress and pressure along the body surface
F =

A
w
dA

A
pdA. (10.1)
Figure 10.1: Forces and moments on a body immersed in a uniform ow.
10.1 BACKGROUND 207
This surface force is related to the uid deformation and, in the case of a Newtonian
uid, the relation between the surface force per unit area (i.e., stress tensor) and rate
of strain tensor is linear (see Section 2.2). In the limit of zero relative velocity between
the body and uid, the rst term on the right-hand side of Eq. (10.1) vanishes, and
the second term is the integral of the hydrostatic pressure around the body. This
is the buoyancy force that Archimedes had shown over 2200 years ago to be equal
and opposite to the weight of the displaced uid.
The component of F parallel to the direction of motion is termed drag, and a
body will move at a constant speed when the thrust generated by its propulsion
system is equal to the drag force. The total drag consists of skin friction, equal to the
streamwise component of the integral of all tangential stresses over the body surface,
and pressure drag, equal to the streamwise component of the integral of all normal
stresses. Flow separation is the major source of pressure dragat least for blunt
bodieswith additional contributions due to displacement effects of the boundary
layer, wave resistance in a supersonic ow or at an airwater interface, drag induced
by lift on a nite body, and virtual mass effects for time-dependent motions. The
latter pressure drag exists even in inviscid uids and occurs because, when the
body velocity changes, the total kinetic energy of the uid changes also. As depicted
in Figure 10.2, the term form drag is used to denote the portion of the pressure drag
that remains after subtracting out the drag due to lift. The sum of form drag and
skin-friction drag sometimes is termed prole drag, which indicates the total drag
that would exist had the aspect ratio been innite (i.e., zero-induced drag).
At very low Reynolds numbers, say <1, viscous forces dominate, and pressure
drag and skin-friction drag are about the same order of magnitude even when sepa-
ration is absent. In this creeping ow regime, the drag is proportional to the relative
velocity, the drag coefcient is inversely proportional to the Reynolds number, and
the body shape has some but no major effect on the total drag.
At high Reynolds numbers, say greater than 1000, inertial effects dominate except
for a thin layer near the body surface for which viscous forces and inertia are equally
important. Drag for blunt bodies is large and is dominated by pressure drag caused
by global separation, which usually occurs near the position of minimum pressure
or maximum body thickness. The pressure drag is typically several orders of magni-
tude higher than the skin-friction drag. If Re is sufciently high, the boundary layer
is turbulent and is better able to resist separation longer, resulting in a narrower
wake and lower total drag as compared with the laminar boundary layer case. The
dimples on a golf ball, though increasing the skin-friction drag, lead to improved
performance because the increased surface roughness trips the boundary layer and
thus signicantly lowers the dominant pressure drag.
For streamlined bodies at high Re, separation can usually be avoided all the way
to the vicinity of the trailing edge. In that case, most of the drag is due to skin
friction and, for that reason, higher total drag is observed when the boundary layer
is turbulent. Still, the total drag is much smaller than that for blunt body having the
same surface area. Astreamlined wing at sufciently high angle of attack is effectively
a blunt body, and the resulting leading-edge separation is termed stall and leads to
substantial increase in drag and loss of lift.
For all high-Reynolds-number cases, the total drag D can be estimated from a
wake survey using the momentum integral equation for steady, incompressible ow
208 DRAG REDUCTION
Figure 10.2: Fluid forces on immersed bodies.
with negligible body forces
D=

A
U (U
U) dA (10.2)
The only restriction here is that the upstream and downstream planes at which
the freestream velocity U
and wake prole U(y, z) are measured are far enough

from the body such that the pressure in both planes is equal to the ambient pres-
sure. At low Re, this condition is satised at distances that are too far from the
body, at which location the velocity decit (U
U) would be too small to measure

accurately.
The object of this chapter is to review available methods to reduce total drag. For
a vehicle, reduced drag means longer range, reduced fuel cost and volume, higher
payload, or increased speed. For pipes and channels, for which close to 100% of
the drag is due to skin friction, drag reduction can result in improved throughput,
reduced pumping power, or reduced duct size and cost.
10.2 DRAG REDUCTION 209
10.2 Drag Reduction
Nature provides numerous instances in which drag reduction is essential for the
survival of many species of air and marine animals (Bushnell and Moore 1991).
Here, we cite instead two examples of the importance of minimizing the drag of
man-made vehicles. At present, the annual fuel cost for all commercial airlines in
the United States is about $10 billionan expenditure that, thanks to declining oil
prices adjusted for ination, has not changed much in over a decade (Hefner 1988;
Bushnell 1998). At subsonic cruising speeds, approximately half of the total drag of
conventional takeoff and landing aircraft is due to skin friction. Hence, a reduction in
skin-friction drag of 20%translates into an annual fuel saving of $1 billion. Not only
is this a substantial sum, but many also believe that a return of the 1973 energy crisis
is inevitable (Phillips 1979; Kannberg 1988). Although the world may have another
100-year supply of natural petroleum, it is estimated that the supply of the United
States will be virtually exhausted early in the next century (Nagel, Alford, and Dugan
1975). Fuel conservation is certainly an important tool to ward off future shortages.
Astated goal of the United States National Aeronautics and Space Administration
(NASA) is to improve the lift-to-drag ratio of the commercial eet of aircraft by a
factor of two during the next two decades. This will maintain the competitive edge
that the U.S. aerospace industry nowenjoys over European and Asian manufacturers.
Doubling L/D would have to be done with a combination of clever drag reduction
and high-lift devices. There is no shortage of ideas, but application to the real world
is another matter altogether (Bushnell 1998; Perrier 1998). For example, to be ac-
ceptable to an airline, a drag-reducing device must not add too much weight to the
aircraft because this causes a proportional drag increase, must not be too expensive
because this increases the capital cost, and must not interfere with maintenance by
either compromising safety or increasing operational cost.
The second example is from the military sector. The amount of propulsive power
available for an underwater vehicle is limited by the volume allocated to its power
plant and the efciency of the various propulsive components. For these vehicles,
about 90% of the total drag is due to skin friction. Accordingly, a reduction in skin-
friction drag of 20% translates into an increase in speed of 6.8%. Although modest,
this extra speed may be vital for the survival of a submarine being chased by another
underwater vehicle.
Attempts to reduce drag go back to antiquity, as mentioned in Chapter 1. Stream-
lining and other control methods summarized in Chapter 8 can eliminate most of the
pressure drag due to ow separation. Some form drag remains, however, even when
the ow continues to be attached to the trailing edge. Owing to the displacement
effects of the boundary layer, the pressure distribution around the body differs from
the symmetric distribution predicted by potential ow theory. This remnant drag can
be reduced by keeping the boundary layer as thin as possible.
For a blunt body, passive and active methods to reduce wake momentum decit
are available. For missiles, for example, the base drag componentwith no jet ow
at the basecan be as much as 50%of the total drag. In general, compared with two-
dimensional bodies, the base drag penalty is lower for axisymmetric bodies because
the vortex shedding process is much less intense in the latter case. Base drag is also
generally lower at supersonic speeds inasmuch as compressibility effects tend to sup-
press vortex shedding. At moderate Reynolds numbers, Strykowski and Sreenivasan
210 DRAG REDUCTION
(1990) have demonstrated that placing a tiny cylinder in just the right place along-
side and well outside the wake of a much larger circular cylinder can signicantly
suppress vortex shedding from the mother cylinder and thus reduce its drag.
Boundary-layer tripping for advancing transition, trailing-edge splitter plate for
disrupting the vortex formation process, base cavities, ventilated cavities, locked-
vortex afterbodies, multistep afterbodies, and afterbodies employing nonaxisym-
metric boat-tailing concepts are among the passive techniques to modify the oweld
around a bluff body. Viswanath (1995) provided a comprehensive review of these
passive methods for subsonic and supersonic ows around two-dimensional and
axisymmetric bodies. Among the active techniques to increase base pressure and thus
to reduce pressure drag are transpiration and vibration. Continuous or pulsating
base bleed is used to modify the ow in the separated region. The latter, with zero
net mass addition, has been shown to be very effective when pulsating at twice the
K arm an shedding frequency (Williams and Amato 1989).
Wave resistance and induced drag can also be reduced by geometric design. By
sweeping the wings of a subsonic aircraft, drag divergence is delayed to higher Mach
numbers, thus allowing the aircraft to y at higher speeds without experiencing a
sudden increase in drag. Additionally, the so-called area rule or coke-bottle effect
typically leads to a factor of two reduction in wave drag at Mach number of one
(Whitcomb 1956). For surface ships, for which typically half the drag is due to
wave resistance, bow and stern bulbs can reduce the energy dissipated into waves at
the airwater interface. An even simpler solution to reduce wave drag is to operate
the ship well below the hull design speed, as is the case for supertankers.
The induced drag of an aircrafts wing, about 25% of the airplanes total drag
at subsonic cruising speeds, is inversely proportional to its aspect ratio and, hence,
a lifting surface is typically designed with as large an aspect ratio as permissible by
structural considerations and desired degree of maneuverability. End plates or other
vortex diffusers can also be used to reduce the induced drag further. In December
1986, the Voyager aircraft completed the rst nonstop, nonrefueled ight around
the world. The primary design goal was to use fuel sparingly to sustain the 9-day
ight. Several innovations contributed to this success, including the use of lightweight
graphite honeycomb composites; unique wing, propeller, and body design; and a
high-specic-energy, air-cooled power plant. But, additionally, the unusually large
wingspan used (33.8 m, including the winglets) contributed to a signicant reduction
in induced drag and proportional fuel saving. The Daedalus became in 1988 the rst
man-powered, heavier-than-air vehicle to sustain ight for a distance of 110 km.
Here, too, a wing with a large aspect ratio and ultra-light-weight material contributed
to the record-breaking ight.
Most of the current research efforts are directed toward reducing the skin-friction
drag, and this topic will occupy the remainder of this chapter. According to Bushnell
(1983), the leverage in this area of research is quite considerable and justies the
study of unusual or high-risk approaches on an exploratory basis.
10.3 Skin-Friction Reduction
As stated, skin friction accounts for about 50% and 90% of the total drag of, respec-
tively, subsonic aircraft and underwater vehicles. Let us start by estimating typical
10.3 SKIN-FRICTION REDUCTION 211
Reynolds numbers for the vehicles whose drag is to be minimized. A commercial air-
craft traveling at a speed of 300 m/s would have a unit Reynolds number of 210
7
/m
at sea level and 1 10
7
/m at an altitude of 10 km. Owing to the much smaller kine-
matic viscosity of water, an underwater vehicle moving at a modest speed of 10 m/s
(20 knots) would have the same unit Reynolds number of 1 10
7
/m.
Three ow regimes can be identied for the purpose of reducing skin friction.
First, if the ow is laminar, typically at Reynolds numbers based on distance from
leading edge of <10
6
, then methods of reducing the laminar shear stress are sought.
Recall from Section 2.4 the streamwise momentum equation written at the wall
v
w
u
y
y=0
+
p
x
y=0
y=0
u
y
y=0
=
2
u
y
2
y=0
(10.3)
The right-hand side of this equation is the wall ux of spanwise vorticity. When this
term is positive, the vorticity ux is negative, and the wall is then a sink of spanwise
vorticity. The streamwise velocity prole in this case is inectional and has lower
slope at the wall, meaning lower skin friction. Therefore, any or a combination of
the following techniques can be used to lower the laminar skin friction: upward wall
motion, injection of uid normal to the wall, adverse pressure gradient, wall heating
in air, or wall cooling in water. Note that any of these methods will promote ow
instability and separation. These tendencies have to be carefully considered when
deciding how far to go with the attempt to lower C
f
.
Two other techniques can be used to lower laminar skin friction. Narasimha
and Ojha (1967) considered the higher-order effects of moderate longitudinal sur-
face curvature. Their similarity solutions show a denite decrease in skin friction
when the surface has convex curvature in all cases, including zero pressure-gradient.
Narasimha and Ojha (1967) attributed the decrease in C
f
to the tendency for the
velocity in the potential ow region to decrease away from the surface. The second
technique is used in rareed gas ows. Appropriate surface preparation can be used
to lower the tangential momentum accommodation coefcient and, thus, introduce
a measurable slip velocity at the wall (Steinheil et al. 1977; Gampert, Homann, and
Rieke 1980). We will return to the issue of slip velocity and accommodation coef-
cient in Chapter 13.
Secondly, in the range of Reynolds numbers from 1 10
6
to 4 10
7
, active and
passive methods to delay transition as much as possible are sought. These tech-
niques were reviewed in Chapter 6 and can result in substantial savings. As shown in
Figure 6.1, the skin-friction coefcient in the laminar at-plate can be as much as an
order of magnitude less than that in the turbulent case. Note, however, that all the
stability modiers discussed in Section 6.4 result in an increase in the skin friction
over the unmodied Blasius layer. The object is, of course, to keep the penalty below
the saving (i.e., the net drag will be above that of the at-plate laminar boundary
layer but well below the viscous drag in the at-plate turbulent ow).
Thirdly, for Re >4 10
7
, transition to turbulence cannot be delayed with any
known practical method
1
without incurring a penalty that exceeds the savings. The
task is then to reduce the skin-friction coefcient in a turbulent boundary layer. This
1
The passive compliant coating that could indenitely suppress TollmienSchlichting instabilities as dis-
cussed in Section 7.6.2 may be an exception to this statement, but as of this writing the very promising
theoretical nding of Section 7.6.2 is yet to be validated experimentally.
212 DRAG REDUCTION
topic will be covered in the following section. Relaminarization will be covered in
Section 10.5, although achieving a net saving here is problematic at present. Futur-
istic control systems to reduce skin friction in turbulent wall-bounded ows will be
discussed in Chapter 14.
For a vehicle with a unit Reynolds number of 110
7
/m, the rst regime exists
in the rst few centimeters and, hence, is of no great consequence. Transition delay
is feasible on the wing and other appendages of an aircraft but not on its much
longer fuselage. On the fuselage, where half of the skin-friction drag takes place,
some alteration of the turbulence structure is sought. Short underwater bodies, such
as torpedoes, are ideal targets for applying transition control methods. On the much
longer submarine, these techniques are feasible on the rst few meters of its surface.
Beyond that, turbulence drag reduction or relaminarization is sought.
10.4 Reduction of C
f
in Turbulent Flows
At very large Reynolds numbers, transition can no longer be postponed, and the
boundary layer becomes turbulent and, thus, an excellent momentum conductor.
Such a ow is less prone to separation but is characterized by large skin friction. Sev-
eral techniques are available to reduce the turbulent skin-friction coefcient, but only
very few are in actual use. The basic reason is that the majority of these methods are
relatively new and are thus still in the research and development stage. The various
drag-reduction methods discussed here will be grouped into four categories: tech-
niques that reduce the near-wall momentum, methods involving the introduction
of foreign substances, techniques involving the geometry, and recent innovations.
Futuristic strategies are deferred to the last chapter of this book. Several excellent re-
views of the various turbulent skin-friction reduction methods are available: Bushnell
(1983), Bandyopadhyay (1986b), Wilkinson et al. (1988), Pollard (1997, 1998), and
Blackwelder (1998). Bushnell and McGinley (1989) and Gad-el-Hak (1989) provided
more general articles addressing turbulence control in wall ows to achieve a variety
of desired goals. The former article in particular summarizes the known sensitivi-
ties of wall-bounded ows to various inputs having a rst-order inuence upon the
ow structures. The books edited by Bushnell and Hefner (1990), Gad-el-Hak et al.
(1998), and Meng (1998) offer many useful articles on drag reduction.
10.4.1 Reduction of Near-Wall Momentum
Methods of skin-friction drag reduction in turbulent boundary layers that rely
on reducing the near-wall momentum are reviewed in this subsection. Recall the
denition of local skin-friction coefcient
C
f

w
1
2
U
2
o
=
2
U
2
o
U
y
y=0
= 2
U
o
2
, (10.4)
where U is the time-averaged velocity component in the streamwise direction, U
o
is the velocity outside the boundary layer, and u
is the friction velocity. Thus, one

brute-force way of achieving drag reduction is to lower the mean-velocity gradient at
the wall or make the curvature of the velocity prole there as positive as possible, as
was done in the laminar case. Equation 10.3 is valid instantaneously and is therefore
applicable to the turbulent case. The inuence of wall transpiration, shaping, or
10.4 REDUCTION OF C
f
IN TURBULENT FLOWS 213
heat transfer on the mean velocity prole is complicated by the additional effects
of these modulations on the Reynolds stress term. However, these inuences are
qualitatively in the same direction as in the simpler laminar case. Thus, lower skin
friction is achieved by driving the turbulent boundary layer toward separation. This
is accomplished by injecting uid normal to the wall, shaping to produce adverse
pressure gradient, surface heating in air, surface cooling in water, or any other strategy
designed to increase the uid viscosity at the wall. These methods of control in general
result in an increase in turbulence intensity (Wooldridge and Muzzy 1966).
Although in the reverse ow region downstream of the separation line the skin
friction is negative, the increase in pressure drag is far more than the saving in skin-
friction drag. The goal of these methods of control is to avoid actual separation, that
is, lower C
f
but not any lower than zero (the criterion for steady, two-dimensional
separation). The papers by Stratford (1959a,b) provide useful discussion on the pre-
diction of turbulent-boundary-layer separation and the concept of ow with contin-
uously zero skin-friction throughout its region of pressure rise. By specifying that
the turbulent boundary layer be just at the threshold of separation, without actually
separating, at all positions in the pressure rise region, Stratford (1959b) experimen-
tally veried that such a owachieves a specied pressure rise in the shortest possible
distance and with the least possible dissipation of energy. An airfoil that could utilize
the Stratfords distribution immediately after transition from laminar to turbulent
ow would be expected to have a very low drag (Liebeck 1978).
Two additional methods to reduce the near-wall momentum are tangential in-
jection and ion wind. In the former, a low-momentum uid is tangentially injected
from a wall slot. For some distance downstream, the wall senses the lower injection
velocity and not the actual freestream. Bushnell (1983) termed this situation a wall
wake as opposed to a wall jet (used in high-lift devices to keep the ow attached).
Ion wind, generated by an electrode in the wall and discharging either to space or to
a ground on the wall further downstream, causes an increase in the normal velocity
component near the wall. Because there is no net mass transfer through the surface,
the near-wall longitudinal momentum is again reduced, and the boundary layer is
driven toward separation (Malik, Weinstein, and Hussaini 1983). The source and
amount of power to produce a measurable drag reduction are among the outstand-
ing issues to be considered with this technique. Among the possibilities is the use of
the natural buildup of static electricity on an aircraft that at present is dissipated at
discharge points to avoid large-scale arcing.
When attempting to reduce drag by driving the boundary layer toward separation,
a major concern is the owbehavior at off-design conditions. Aslight increase in angle
of attack, for example, can lead to separation and consequent large drag increase as
well as loss of lift. High-performance airfoils with lift-to-drag ratio of over 100
utilize carefully controlled adverse pressure gradient to retard the near-wall uid,
but their performance deteriorates rapidly outside a narrow envelope (Carmichael
1974).
In the case of tangential or normal injection, the source of uid to be injected
is important. Large ram-drag penalty is associated with using freestream uid, and
reaching a break-even point is highly problematic. A lower loss source is uid with-
drawn from somewhere else to control transition or separationfor example, from
the wings and empennage of an aircraft. This point will be discussed further in
Section 10.6.
214 DRAG REDUCTION
Figure 10.3: Computed skin-friction reduction due to tangential injection
[after Howard et al. 1975].
Howard, Hefner, and Srokowski (1975) conducted nite-difference calculations
to show the effect of single- and multislot injection on skin-friction drag of a fuselage
shape typical of long-haul subsonic transport. A sample of their results is shown
in Figure 10.3 for up to 10 tangential slots. The abscissa in the gure indicates the
streamwise location of each slot, , normalized with the total length of the fuselage,
T
. The benecial effect of injection is most pronounced immediately downstream of
the slot exit and diminishes with distance from the slot. Performance deterioration
is clearly a result of the mixing between the low-momentum slot ow and the high-
momentum boundary-layer ow. Note, however, that the asymptotic level of skin
friction with multislots is lower than that in the unperturbed ow. Without taking
into account the system penalties for collecting, ducting, and injecting the secondary
ow, Howard et al. (1975) computed a reduction in total skin-friction drag of the
order of 50%. They maintained that further optimization may provide even greater
skin-friction reductions to help compensate for system losses.
As was the case when surface heating or cooling was considered for transition
delay or separation postponement, this technique should be used for skin-friction
reduction only when a heat source or sink is readily available. Rejected heat from
the vehicles power plant is a free source of energy, and cryofuel is an excellent heat
sink that does not require energy expenditure.
10.4.2 Introduction of Foreign Substance
Turbulent skin-friction drag can be reduced by the addition of several foreign sub-
stances. Examples include long-chain molecules, surface-active agents and microbub-
bles in liquid ows, and small solid particles and bers in either gases or liquids. In
general, the addition of these substances leads to a suppression of the Reynolds stress
production in the buffer zone that links the linear and the logarithmic portions of
10.4 REDUCTION OF C
f
the mean velocity prole. Thus, the turbulent mixing is inhibited, and a consequent
reduction in the viscous shear stress at the wall is achieved.
Drag reduction by solutions of macromolecules (molecular weight >10
5
) is per-
haps the more mature technology with well over 5000 research papers in existence
(Virk 1975; White and Hemmings 1976; Hoyt 1979; Hendricks et al. 1989; Nadolink
and Haigh 1995; Gyr and Bewersdorff 1995). Toms (1948) observed that the addi-
tion of a few parts per million of polymethyl methacrylate to a turbulent pipe ow of
monochlorobenzene reduced the pressure drop substantially belowthat of the solvent
alone at the same ow rate. Successful long-chain molecules for water ow include
carboxymethyl cellulose, guar gum, copolymer of polyacrylamide and polyacrylic
acid, and polyethylene oxide. Skin-friction drag reduction of up to 80% is possible
in both external and internal ows with 100 ppm or less of the largest molecules.
The viscosity of these dilute solutions of polymers is typically 10% higher than
that of the solvents, although most polymer solutions are detectably shear thinning
at higher concentrations (Berman 1978). The polymer molecules at rest are in the
form of spherical random coils in which each monomer is joined to the preceding
one at a random angle. The onset of drag reduction is associated with the expansion
of these coils outside the viscous sublayer. Hinch (1977) asserted that this unfolding
explains the dramatic effects of few parts per million (by weight) on the drag. The
effective hydrodynamic volume fraction shows large concentration once the polymer
molecules are stretched. Therefore, it is the volume fraction of spheres just enclosing
the separate polymers rather than the weight that is the appropriate measure of the
added substance.
When the macromolecules are coiled, the viscosity of the dilute solution changes
by a fewpercent, whereas if they are fully stretched the apparent viscosity increases by
several orders of magnitude. Onset of drag reduction occurs when D
M
u
/ = 0.015,
where D
M
is the effective molecular diameter. This diameter is about two orders of
magnitude less than the shortest dynamically signicant length in a typical turbulent
ow (Virk et al. 1967; Lumley 1969). Numerous experiments have indicated that the
drag reduction is associated with a thickening of the buffer layer and a corresponding
shift of the logarithmic part to accommodate the larger buffer zone (Reischman and
Tiederman 1975). The effect of the additive may then be represented as a virtual slip
at the wall (Virk et al. 1967).
The percentage drag reduction increases as the polymer concentration is increased
and the buffer layer is thickened. This process tends to an asymptote when the buffer
layer reaches the center of the pipe or the edge of the boundary layer. As the Reynolds
number is increased at a given concentration, a second limit in drag reduction is
reached when the macromolecules become fully stretched. Moreover, degradation or
breaking of the molecules by the ow occurs at sufciently high strain rates, with a
consequent loss of the ability to reduce the drag (Berman and George 1974). Because
of this latter point, care has to be exercised when mixing the macromolecules into
the solvent as well as when delivering the solution into the ow.
By injecting the polymer in different regions of the boundary layer, Wells and
Spangler (1967), Wu and Tulin (1972), and McComb and Rabie (1979) have shown
that the additive must be in the wall region for drag reduction to occur. Furthermore,
the channel-ow experiments of Tiederman et al. (1985) have clearly demonstrated
that drag reduction is measured downstream of the location where the additives in-
jected into the viscous sublayer begin to mix in signicant quantities with the buffer
216 DRAG REDUCTION
region (i.e., 10 < u
y/ < 100. At streamwise locations where drag reduction does

occur, the dimensionless spanwise streak spacing increases and the average bursting
rate decreases, although the decrease in bursting frequency is larger than the corre-
sponding increase in low-speed streak spacing. Tiederman et al. (1985) concluded
that the polymers have a direct effect on the ow processes in the buffer region and
that the linear sublayer appears to have a rather passive role in the interaction of the
inner and outer portions of a turbulent wall layer.
Many mechanisms have been proposed to explain the experimental observations.
Notable among these is the work of Lumley (1973, 1977), Landahl (1973, 1977),
and the more recent quantitative theory by Ryskin (1987). According to Lumleys
model, the macromolecules are expanded outside the viscous sublayer owing to the
uctuating strain rate. The resulting increase in effective viscosity damps only the
small dissipative eddies. The suppression of small eddies in the buffer layer leads to
increased scales, a delay in the reduction of the velocity prole slope, and consequent
thickening of the wall region. In the viscous sublayer, the scales remain unchanged
because the effective viscosity of the dilute solution in steady shear is only slightly
affected. In the buffer zone, the scales of the dissipative and energy-containing eddies
are roughly the same and, hence, the energy-containing eddies will also be suppressed,
resulting in reduced momentum transport and reduced drag.
Lumleys more conservative general hypothesis is not incompatible with Landahls
specic, rather speculative mechanism (Lumley and Kubo 1984). The polymer sup-
presses the small eddies responsible for the inectional prole and the secondary
instability in the buffer zone. Kim et al. (1971) have shown that the greatest tur-
bulent kinetic energy production for a Newtonian boundary layer occurs near the
beginning of the buffer zone at y 12/u
. A sharp maximum in the turbulent

intensity,
u
2
, occurs at the same location. For a dilute solution of drag-reducing
polymer, the peak of
u
2
moves away from the wall and becomes much broader
as compared with the solvent alone. Production of Reynolds stress in this important
buffer zone is thus diminished.
In a recent article, Sreenivasan and White (1999) challenged the prevailing theories
for polymer drag reduction. They maintained that the connection between the uc-
tuating strain rates and the large extensional viscosity is only circumstantial and that
polymer coils can only be partially stretched in a random eld of strain rates. If true,
the elongational viscosity cannot be measurably higher than the solvent viscosity, and
a different mechanismmust therefore be responsible for the observed drag reduction.
Sreenivasan and White (1999) argued that the elastic theory advanced by de Gennes
(1990) is compatible with at least two experimental observations: the dependence
of drag reduction onset on polymer concentration and the maximum drag-reduction
asymptote. Basically, the elastic energy stored by the partially stretched polymers has
an effect on the ow only when exceeding the turbulent energy. When the two ener-
gies become comparable, the elastic energy interferes with the usual turbulent cascade
mechanismby not allowing it to proceed all the way to the Kolmogorov scales. Thus,
the nonlinear action that generates small scales of turbulence will be terminated at
some scale larger than the Kolmogorov scale, leading to increased buffer layer thick-
ness and reduced drag. Sreenivasan and White (1999) carefully computed the order
of magnitude of the expected ow behavior and successfully compared the results to
existing experiments. Their results, albeit tentative, provide the framework needed
for further targeted studyparticularly at high Reynolds numbers.
10.4 REDUCTION OF C
f
The use of drag-reducing polymers in pipe lines is cost-effective in several applica-
tions. When it was discovered that the winter cooling of crude oil in the trans-Alaskan
pipeline had been underestimated, polymer additives offered an inexpensive alterna-
tive to increasing the number or power of the pumping stations. Polymer has also
been used in the the IraqTurkey pipeline. Long-chain molecules are presently used
in waste-disposal plants, re-ghting equipment, and high-speed water-jet cutting.
For surface ships or underwater vehicles practical considerations include the cost of
the polymer and how, with what, and when to mix it as well as how to eject it into
the boundary layer (i.e., using pumps or pressurized tanks), through slots or porous
surfaces, and so forth. A very important additional consideration is the portion of
the payload that has to be displaced to make room for the additive. Oil compa-
nies appear to have concluded that the use of polymer for supertankers is just at
the break-even point economically. That assessment will undoubtedly change as fuel
prices rise and perhaps as manufacturing costs of polymers drop. For submarines and
torpedoes, where space is a prime consideration, it seems that the volume occupied
by the additive is a more important obstacle to its routine use to date.
Other substances that can be added to a clear uid to reduce skin-friction drag
include surfactants (Savins 1967; Patterson, Zakin, and Rodriguez 1969; Zakin
et al. 1971; Aslanov et al. 1980), microbubbles (McCormick and Bhattacharya 1973;
Bogdevich and Malyuga 1976; Bogdevich et al. 1977; Madavan, Deutsch, and Merkle
1984, 1985; Merkle and Deutsch 1985, 1989), large length-to-diameter particles such
as bers, (Hoyt 1972; Lee et al. 1974; McComb and Chan 1979, 1981), and spher-
ical particles (Soo and Trezek 1966; Pfeffer and Rosetti 1971; Radin 1974; Radin,
Zakin, and Patterson 1975; Povkh, Bolonov, and Eidelman 1979). Lumley (1977,
1978a) argued that the same mechanism discussed in conjunction with the polymer
case governs the interaction of most of these substances with the turbulent boundary
layer. The additives affect only the dissipative scales of the turbulence, thus increas-
ing the scales of dissipation in the buffer layer. The energy-containing eddies are also
suppressed and, consequently, the momentum transport is reduced.
A surface-active substance, such as soap, is that which, when dissolved in water
or in an aqueous solution, reduces its surface tension. The majority of surfactants
appear to reduce the turbulent skin-friction drag only in the presence of electrolytes.
These additives are stable against mechanical destruction, which affords a certain ad-
vantage over polymer molecules that break down when the strain rate is sufciently
high. Although drag-reducing polymer ows undergo transition at a lower Reynolds
number than Newtonian ows (Lumley and Kubo 1984), some recent experiments
indicate that surfactants are very effective in delaying laminar-to-turbulence transi-
tion (A. J. Smits, private communication; Sabadell 1988).
The idea of placing a thin layer of gas between the wall and its water-boundary
layer dates back to the last century. Substantial drag reductions are potentially pos-
sible because of the lower density of the gas. Unfortunately, the various instabilities
associated with gasliquid interface result in drag increase. Extremely small gas bub-
bles (microbubbles) injected through a porous wall or produced by electrolysis do
not suffer from the stability problems of a gas lm and result in skin-friction reduc-
tion as high as 80%. Maximum reduction is obtained when the gas volume fraction
approaches the bubble-packing limit. Skin friction is reduced because of the substan-
tially lower density and also because of the usual increase in bulk viscosity due to
particles (Batchelor and Green 1972), which damps the small-scale motions in the
218 DRAG REDUCTION
buffer layer. Legner (1984) presented a simple phenomenological model to predict
microbubble drag reduction in turbulent boundary layers. Application of the mi-
crobubbles technique for surface ships is quite feasible. Compressed atmospheric air is
injected through a microporous skin over portions of the hull. This has the additional
advantage of reducing fouling drag and fouling maintenance costs because water is
kept away from the hull surface. The situation is different for underwater vehicles.
There, a source of air is not readily available, and using electrolysis for bubble produc-
tion will not yield net energy saving. Moreover, the presence of bubbles may adversely
affect the ow-induced noise, which is an important consideration for naval vehicles.
Spherical or large length-to-diameter particles can be used in both air and water
ows. For spherical particles, drag reduction of up to 50% is feasible for certain pa-
rameter values, although drag increase is also possible. Heavy but small particles may
reduce the drag by inducing a stable density stratication near the wall, thus driving
the turbulence toward relaminarization. Lighter particles that are large enough to
interact with the smallest turbulence scales reduce the drag through the same mech-
anism as in the polymer case (i.e., suppressing the dissipative eddies and increasing
the scale of dissipation).
McComb and Chan (1981) reported up to 80% drag reduction using the natu-
rally occurring macrober chrysotile asbestos dispersed at a nominal concentration
of 300 ppm by weight in a 0.5% aqueous solution of Aerosol OT. The individual
brils of chrysotile asbestos have a mean diameter of 40 nm and a mean length-to-
diameter ratio in an undegraded suspension in the range of 10
3
10
4
. Like polymers,
macrobers readily break down (McComb and Chan 1979) and have to be care-
fully dispersed into the solvent and delivered to the boundary layer. In principle,
bers can also be used in gases, although their drag-reducing potential has yet to be
demonstrated.
10.4.3 Methods Involving Geometry
Under this classication are some of the most recently researched techniques to
reduce the turbulent skin-friction drag. These include large-eddy breakup devices
(LEBUs), riblets, compliant surfaces, wavy walls, and other surface modications.
The book edited by Bushnell and Hefner (1990) offers comprehensive reviews of these
and other drag-reduction strategies. With few exceptions, there is little theoretical
basis for howthese geometrical modications affect the skin friction, and most of the
present knowledge comes from experiments. Needless to say, the lack of analytical
framework makes optimization for a given ow condition as well as extension to
other ow regimes very tedious tasks.
The LEBUs are designed to sever, alter, or break up the large vortices that form
the convoluted outer edge of a turbulent boundary layer. A typical arrangement con-
sists of one or more splitter plates placed in tandem in the outer part of a turbulent
boundary layer, as sketched in Figure 10.4. It is of course very easy to reduce substan-
tially the skin friction in a at-plate boundary layer by placing an obstacle above the
surface. What is difcult is to ensure that the devices own skin-friction and pressure
drag do not exceed the saving. The original screen fence device of Yajnik and Acharya
(1978) and the various sized honeycombs used by Hefner, Weinstein, and Bushnell
(1980) did not yield a net drag reduction. In low-Reynolds-number experiments, very
thin elements placed parallel to a at plate have a device total drag that is nearly equal
to laminar skin friction. A net drag reduction of the order of 20% is feasible with
10.4 REDUCTION OF C
f
Figure 10.4: Sketch of a tandem arrangement of an LEBU.
two elements placed in tandem with a spacing of O[10] (Corke et al. 1980, 1981).
These ribbons have typically a thickness and a chord of the order of 0.01 and ,
respectively, and are placed at a distance from the wall of 0.8. Several experiments
report a more modest drag reduction (Hefner, Anders, and Bushnell 1983) or even a
drag increase, but it is believed that a slight angle of attack of the thin element can
result in a laminar separation bubble and a consequent increase in the device pres-
sure drag. Flat ribbons at small, positive angles of attack produce larger skin-friction
reductions. This is consistent with the analytical result that a device producing posi-
tive lift away from the wall is more effective (Gebert 1988). In any case, a net drag
reduction should be achievable, at least for devices having chord Reynolds numbers
<10
6
, if extra care is taken to polish and install the LEBU.
Net drag reduction has been documented even in the presence of favorable or
adverse pressure gradient in the main ow (Bertelrud, Truong, and Avellan 1982;
Plesniak and Nagib 1985). Additionally, Anders and Watson (1985) have demon-
strated that an LEBU having an airfoil shape is nearly as effective in reducing the net
drag as a at ribbon. A wing is several orders of magnitude stiffer than a thin ribbon,
which affords a certain advantage for extending the technique to eld conditions.
At ight Reynolds and Mach numbers, the possibility of paying for transitional or
turbulent skin friction as well as wave drag on the device itself is real (Anders et al.
1984, 1988), and achieving net drag reduction in this circumstance is problematic at
present (Anders 1990b). For an airplane, an LEBU is likely to take the form of a ring
around the fuselage.
Two analytical attempts to explain the mechanisms involved with the LEBUs are
noted in here. Basically, the LEBU acts as an airfoil on a gusty atmosphere, and a
vortex unwinding mechanism is activated. Dowlings (1985) inviscid model indicates
that the vorticity shed from the LEBUs trailing edge as a result of an incident line
vortex convected past the device tends to cancel the effect of the incoming vortex and
to reduce the velocity uctuations near the wall. Atassi and Gebert (1987) and Gebert
(1988) modeled the incoming turbulent rotational ow as two- or three-dimensional
harmonic disturbances. They use the rapid distortion approximation and unsteady
aerodynamic theory to compute the uctuating velocity downstreamof thin-plate and
airfoil-shaped devices. The uctuating normal velocity component is most effectively
suppressed for a range of frequencies that scales with the freestream velocity and
the device chord. This important result determines the optimum size of an LEBU
by selecting this frequency range to correspond to that of the large-scale eddies in a
given turbulent boundary layer.
220 DRAG REDUCTION
Figure 10.5: Fin shapes for longitudinally ribbed
surface.
The second geometrical modication is
the riblets, which are wall grooves aligned
with the freestream. Small longitudinal
striations in the surface interacting favor-
ably with the near-wall structures in a tur-
bulent boundary layer can produce a mod-
est drag reduction in spite of the increase
in surface area. The early work employed
rectangular ns with height and spacing of
O[100/u
]. The turbulent bursting rate

was reduced by about 20%, and a modest
4% net drag reduction was observed (Liu,
Kline, and Johnston 1966). In a later re-
nement of this technique, Walsh and his
colleagues at NASALangley examined
the drag characteristics of longitudinally
ribbed surfaces having a wide variety of n
shapes (see Figure10.5) that included rect-
angular grooves, V-grooves, razor blade
grooves, semicircular grooves, and alter-
nating transverse curvature (Walsh and
Weinstein 1978; Walsh 1980, 1982, 1983, 1990; Walsh and Lindemann 1984). A net
drag reduction of 8%is obtained using V-groove geometry with sharp peak and either
sharp or rounded valley. Optimum height and spacing of the symmetric grooves are
about 15/u
. Although these dimensions would be extremely small for the typical

Reynolds numbers encountered on an airplane or a submarine (peak-to-valley height
35 m), such riblets need not be machined on the surface. Thin, low-specic-gravity
plastic lms with the correct geometry on one side and an adhesive on the other are
presently available commercially, and existing vehicles could readily be retrotted.
In fact, these tapes were successfully tested at Ma = 0.7 on a T33 airplane and on a
Lear jet. The performance of the riblets in ight was similar to that observed in the
laboratory (Anders et al. 1988). In water, riblets were employed on the rowing shell
during the 1984 Summer Olympics by the United States rowing team. Similar riblets
were also used on the submerged hull of the winner of the 1987 Americas Cup yacht
race, the Stars and Stripes, with apparent success.
Riblets are thought to restrain the movement of the near-wall longitudinal vortices
and therefore maintain their coherence. The stabilized rolls or coherent structures
present a barrier to the usual cascade to small scales and hence impede the rate of
energy loss with a resulting drag reduction despite the increased surface area. Riblets
are effective in the presence of moderate, adverse, and favorable pressure gradients.
The loss of drag-reducing effectiveness as ow conditions vary off-design is gradual.
Moreover, the percentage drag reduction slowly decreases to zero as the yaw angle
between the ow and the grooves goes up to 30
. Surprisingly, drag does not increase

at yaw angles >30
. Remaining practical problems include cost, weight penalty, par-

ticulate clogging, ultraviolet radiation, and lm porosity and resistance to hydraulic
uids and fuel. Recent reviews of riblets were presented by Pollard (1997, 1998).
Curiously, fast sharks have a surface covering of dermal denticles with ow-
aligned keels having the optimal riblet spacing as determined by Walsh (Bechert,
Hoppe, and Reif 1985). As the shark grows, new keels are added onto the sides of
10.4 REDUCTION OF C
f
the denticles without changing the keel-to-keel spacing. Aclue to howriblets result in
a net drag reduction despite the increase in wetted area is provided by the experiment
of Hooshmand et al. (1983), who showed that, for an optimum V-grooved surface,
C
f
is reduced by 40% in the valleys and increased by 10% at the peaks. It appears
that the riblets severely retard the ow in the valleys, which places a slip boundary
condition upon the turbulence production process. Lumley and Kubo (1984) argued
that the streamwise vortices in the near-wall region are forced to negotiate the sharp
peaks of the riblets, causing increased losses. To stay in equilibrium, the eddies must
increase their energy gain, and one way to do that is to grow larger (energy gain
(size)
2
; loss (size)). As was the case with polymers, the larger scales result in the
secondary instability and the sharp change in prole slope occurring farther from
the wall. The mean velocity is, thus, higher for the same friction velocity, and the
coefcient of skin friction is reduced.
The third geometrical modication to reduce the skin friction is xed or moving
wavy walls. Kendall (1970) observed that the integrated skin friction in a turbulent
boundary layer over rigid, sinusoidal waves can be as much as 25% below that over
an equivalent at surface. Kendalls waves were relatively shallow, having a height-
to-wavelength ratio of h/ = 0.028, with = O[]. Sigal (1971) reported a reduction
in integrated skin friction of about 12% when using 10% larger amplitude waves
(h/ = 0.031) than that used by Kendall. The transverse waves provide alternating
regions of longitudinal concave and convex curvatures along with alternating adverse
and favorable pressure gradients. The coincidence of convex curvature and favorable
pressure gradient promotes relaminarization as the wave crests are approached, and
the near-wall momentum is reduced in regions having adverse pressure gradient. Ac-
cordingly, the integrated skin-friction drag is reduced. Unfortunately, viscous forces
cause a downstream phase shift between the pressure distribution and the wave rel-
ative to the 180
out-of-phase relationship predicted from potential ow theory for

a sinusoidal wave. This phase shift causes an additional pressure drag not present in
the at-surface case. The net result is that the total drag for the wavy surface increases
despite the decrease in skin friction. Cary, Weinstein, and Bushnell (1980) argued that
there may be two approaches to overcome this problem. First, because pressure drag
varies as (h/)
2
, further reduction of h/ may yield a net drag reduction. Secondly,
certain nonsinusoidal waves may have more favorable pressure and curvature effects
on the viscous drag and a less skewed pressure distribution relative to the waves and
therefore less pressure drag. Cary et al.s numerical results indicate the soundness of
the rst approach. Net drag reduction of about 13% is attained for the optimum
h/ of 0.005. This result seems to be insensitive to the ratio of wave height to the
viscous scaleat least for hu
/ < 33. Experiments to test the second approach

were only partially successful (Lin et al. 1983). Skewed waves with gradual, straight
downstream-facing slope and steeper, sinusoidal upstream-facing surface have lower
pressure drag than a symmetric sine wave. However, the asymmetric surface is not
as effective in reducing the viscous drag with the result that the net drag reduction is
only 12%, which is hardly worth the effort (Wilkinson et al. 1987, 1988).
Moving wavy walls are divided into two categories: driven, noninteractive walls
and compliant surfaces. In the former technique, a at or wavy wall is translated
in the streamwise direction. The rectilinear wall motion essentially acts as a slip
boundary condition and reduces the skin friction. Obviously the boundary layer could
be completely eliminated when U
wall
= U
. Amoving wavy wall can produce a thrust

at high enough translational speed but causes an additional pressure drag at lower
222 DRAG REDUCTION
speed. In any case, this method of control is in general impractical and is used mainly
to provide controlled experiments to determine what type of wave motion is required
to achieve a given result, which is much the same as the laminar ow-control case.
Wall motion can also be generated by using a exible coating with sufciently
low modulus of rigidity (Chapter 7). The idea is very appealing if a particular kind of
uidsolid interaction yields a net drag reduction. Covering a vehicle with a compli-
ant coating is relatively simple; does not require modication of existing design; does
not require slots, ducts, or internal equipment of any kind; and no energy expenditure
is needed. The search for such a surface has been very elusive, however, despite re-
ports of substantial drag reduction by some researchers. Irreproducibility seems to be
an outstanding characteristic of the body of compliant coating experimental evidence
(Bushnell et al. 1977; McMichael et al. 1980; Gad-el-Hak 1986a,b, 1987a, 1996a).
The window of opportunity for a favorable coating response is rather narrow. For
ow speeds below the transverse wave speed in the solid, no signicant interaction
between the uid and the compliant surface takes place. Above the transverse wave
speed, ample opportunity for interaction exists; however, hydroelastic instabilities
appear, and the resulting large-amplitude surface waves increase the drag because
of the roughness-like effects (Gad-el-Hak, Blackwelder, and Riley 1984; Gad-el-Hak
and Ho 1986b; Gad-el-Hak 1996a). A second constraint concerns the density ratio
of the uid and the solid. The extremely lowdensity of air makes nding a reasonable
coating material highly unlikely, although in liquids the situation is different. Flexible
surfaces that interact favorably with TollmienSchlichting waves in water boundary
layers are readily available (Gad-el-Hak 1996a). For turbulent boundary layers, the
task is to nd a coating that responds to the ow, particularly to its preburst wall-
pressure signature, with a surface motion that can alter, modify, or interrupt the se-
quence of events leading to a burst. The analytical work of Purshouse (1977) indicates
that an anisotropic coating would be more likely to achieve a net drag reduction in a
turbulent boundary layer. More discussion on these issues was presented in Chapter 7.
Other geometric modications on the surface with a drag-reducing potential in-
clude xed waves aligned with the ow (essentially large-scale riblets), micro air-
bearings, compound or three-dimensional riblets, sieves, furry (wheateld-type) sur-
faces, and sword-sh conguration. These techniques have been reviewed by Bushnell
(1983) and more recently by Wilkinson et al. (1988). The last of these methods is
the most straightforward. As a boundary layer thickens, the Reynolds number in-
creases and the coefcient of skin friction decreases, with the result that the local
skin friction is higher in the forward part of a vehicle and quite low in the aft end.
Consider, for example, a 50-m-long aircraft with a unit Reynolds number of 10
7
/m
and ignore pressure gradient and other effects. Near the nose, the skin-friction coef-
cient is about 0.003 (Figure 6.1). Near the tail, Re 510
8
and C
f
0.0018 (i.e.,
40% lower). By substantially reducing the wetted area of the forebodythe sword-
sh congurationdrag reduction may be attainable. This technique is particularly
useful when combined with a convex aftbody.
10.4.4 Recent Innovations
Though against long-held believes that rough surfaces have more drag than their
smooth counterparts, the riblet experience has apparently inspired Sirovich and
Karlsson (1997) to search for other kind of roughness that may lead to skin-friction
reduction. Their nding is one of those rare instances in which a better understanding
10.5 RELAMINARIZATION 223
of the physics of a turbulent ow led to a successful ow-control strategy rather than
the more common reverse owof ideas (Carpenter 1997). Sirovich and his colleagues
believeas do many researchersthat the streamwise vortices are the primary struc-
tures in the near-wall region. But, additionally, weaker oblique plane waves exist in
this region propagating downstream and interacting with the streamwise vortices.
Bursting events are the results of such interaction. Computer simulations have indi-
cated that articially randomizing the phases of the vortices and oblique waves leads
to less bursting and consequent drag reduction. Sirovich and Karlsson (1997) tested
this hypothesis in a fully developed turbulent channel ow at modest Reynolds num-
bers. Indeed surfaces covered with special randomized patterns of small V-shaped
protuberances produced drag lower by up to 12% as compared to smooth surface.
Regular patterns of the same roughness led to a rise in drag by 20% and correspond-
ing increase in mixing.
There are two other drag-reducing strategies worth noting before closing
this section. Jung, Mangiavacchi, and Akhavan (1992) and Akhavan, Jung, and
Mangiavacchi (1992) reported a drag reduction of up to 40% in a direct numerical
simulation of a turbulent channel ow when one of the duct walls undergoes high-
frequency, spanwise oscillations. The maximum drag reduction is achieved when the
oscillation period is P
+
Pu
2
/ = 100. The rapid oscillations signicantly damped

the normal and tangential Reynolds stresses, whereas the peak location of each stress
moved away from the wall and the buffer layer thickened. In a at-plate boundary
layer experiment, Laadhari, Skandaji, and Morel (1994) conrmed the turbulence
suppression observed in the DNS data, but they did not measure the skin friction. In
a more recent wind-tunnel experiment, Choi, DeBisschup, and Graham (1998) and
Choi and Clayton (1998) did measure the skin friction using the Preston tube and
near-wall slope of the mean velocity prole. They reported a 45% drag reduction
at optimum oscillations. Choi and Graham (1998) extended the at-plate work to
show that circular-wall-oscillations in a turbulent pipe ow can lead to skin-friction
reduction of up to 25%. The lower skin friction in both physical and numerical ex-
periments has of course to be carefully balanced against the energy used to generate
the wall oscillations.
The second strategy has recently been proposed by Schoppa and Hussain (1998).
Using direct numerical simulations of turbulent channel ow, they indicated drag
reduction of 20% when imposed counterrotating vortices are used and 50% when
colliding, z-directed wall jets are employed. The former method can be realized exper-
imentally using large-scale, passive vortex generators, whereas the lattermuch like
the wall oscillations aboveis a predetermined active control method (see Figure 3.3)
that does not require sensors or control logic but does require energy expenditure
that has to be accounted for when computing the net saving. Schoppa and Hussains
strategy has yet to be validated using physical experiments.
10.5 Relaminarization
Several articles dene and explain the reversion of a turbulent ow to the laminar
state, and this phenomenon is variously also known as retransition, inverse or reverse
transition, or relaminarization (Preston 1958; Patel and Head 1968; Bradshaw1969).
Narasimha and Sreenivasan (1979) provided a comprehensive review and analysis
224 DRAG REDUCTION
of the phenomenon. Instances of relaminarization may be found in a stably stratied
atmosphere, spatially or temporally accelerated shear ows, coiled pipes, swirling
ames, tip vortices behind nite lifting surfaces, far wakes of nite drag-producing
bodies, air ow as it progresses from the trachea to the segmental bronchi in human
lungs, and shear ows subjected to magnetic elds. Various authors use one or more
of the following syndromes to diagnose the reversion of a turbulent ow: reduction
in heat transfer or skin-friction coefcient, cessation of bursting events near the wall
or entrainment of potential ow at the outer region of a boundary layer, reduction
in high-frequency velocity uctuations, breakdown of the log-law in the wall layer,
spreading of intermittency to wall region, and overlapping of energy-containing and
-dissipating eddy scales. In the quasi-laminar state, the velocity uctuations are not
necessarily zero, but rather their contribution to the dynamics of the mean ow be-
comes inconsequential. In other words, the turbulent uctuations inherited from the
previous history of the ow are no longer inuencing the transport of mass, momen-
tum, or energy. Narasimha and Sreenivasan (1979) adopted a pragmatic denition
of reversion. They maintained that a ow has relaminarized if its subsequent devel-
opment can be understood without recourse to any model for turbulent shear ow.
The most obvious mechanism for the occurrence of relaminarization is dissipa-
tion. When the Reynolds number goes down in a turbulent ow (e.g., by enlarging a
duct or by branching a channel ow), the viscous dissipation may exceed the produc-
tion of turbulent energy, and the owmay revert to a quasi-laminar state. This type of
reversion tends to be rather slow. Turbulent energy can also be destroyed or absorbed
by the work done against external forces such as buoyancy (e.g., in stably stratied
uids), centrifugal force (e.g., in convex boundary layers), or Coriolis force (e.g., in
rotating channel ows). The governing parameter in these absorptive-type reversions
is the Richardson number (Ri ), and relaminarization proceeds rapidly once the criti-
cal value of Ri is exceeded. In both dissipative- and absorptive-type of reversion, the
turbulence energy is decreased, but more signicantly the velocity components that
generate the crucial Reynolds stresses are decorrelated. This explains the strong
effects on turbulence of, say, a very mild positive curvature (So and Mellor 1973;
Smits and Wood 1985). In this case, the amplitude as well as the phase of the differ-
ent components of the uctuating motion are affected by the additional strain rate
imposed upon the owparticularly away from the wall.
The third mechanism to effect relaminarization is observed in highly accelerated
ows. Narasimha and Sreenivasan (1973) argued that, in the outer layer of such
ows, the pressure forces dominate over the nearly frozen Reynolds stress eld. The
uv term remains at about the same level as the zero-pressure-gradient case over
much of the boundary layer. Near the wall, however, dissipation dominates, and the
ow is stabilized by the acceleration. The turbulence-reenergizing bursts diminish in
frequency and may stop altogether, and a thin, new laminar boundary layer grows
from the wall within the old boundary layer. According to Morkovin (1988), the
turbulence in the outer layer is starved, and the inner laminar layer is buffeted
by the decaying, wakelike turbulence of the outer region. The skin friction and the
heat transfer rate in the relaminarized inner layer is less than those expected in a
corresponding turbulent ow.
Narasimha and Sreenivasan (1979) asserted that several different reverting ows
can be considered in the light of the three archetypes summarized above. It is
not uncommon, however, to effect relaminarization through a combination of
10.6 SYNERGISM 225
mechanisms rather than a single one. In internal ows, relaminarization can be ac-
complished by gentle wall injection (to avoid separation) or by wall heating (for
gases). In both these cases, the ow accelerates, thus activating the third mechanism
above. Moreover, when a gas is heated, its kinematic viscosity is increased, resulting
in a lower Reynolds number that may fall below the critical value, thus activating
the rst mechanism.
In external ows, Eq. 10.3 indicates that wall suction, surface heating in liquids,
or surface cooling in gases may have the same effect on the near-wall region of a
turbulent boundary layer as favorable pressure gradient. It appears that only the rst
of these stimuli has been experimentally demonstrated. Dutton (1960) observed that
a suction coefcient C
q
|v
w
|/U
o
0.01 is sufcient to relaminarize a boundary
layer at Re = O[10
6
]. The initially turbulent owapproaches the laminar asymptotic
state (Section 6.4) appropriate to the particular value of C
q
used. Although the mean-
velocity gradient near the wall is increased, the Reynolds shear stress is reduced at a
faster rate, leading to a reduction in the turbulent energy production.
The question of immediate concern to this chapter is whether or not suction, or in
fact any of the other reversion stimuli, is a viable drag reduction method for turbulent
boundary layers. The suction rate necessary for relaminarization is too high to be
applied protably over an entire surface, although it may be feasible to apply massive
suction through a spanwise slot (perhaps enough to ingest the entire mass ow in the
boundary layer) followed by gentle suction (or any of the other stability modiers
discussed in Section 6.4) to prevent transition of the newly formed laminar sublayer.
This issue will be discussed again in the next section.
Convex (or positive) curvature affects the boundary layer locally in much the same
way as an LEBU. With a relatively large radius of curvature, of the order of 10,
positive Reynolds stress is produced in the outer ow, and the wall shear is reduced.
Downstream of the end of the curvature region, the ow relaxes very slowly to its
unperturbed state. Wilkinson et al. (1988) suggested that a convex surface may be
better suited for drag reduction than an LEBU deviceparticularly in supersonic
ows in which wave drag penalty on the LEBU blade is rather large. Obviously,
a vehicle surface cannot be all convex, and short regions of concave curvature on
the way to the convex portions are unavoidable. Bushnell (1983) speculated that
the limited extent of the concave regions may not allow formation of any lasting
alterations to the turbulent structures. Net reductions of drag of the order of 20%
appear to be obtainable using surface curvature. Bandyopadhyay (1990) provided
a useful review of viscous drag reduction using convex curvature concepts. Little
research is available on the feasibility of employing other relaminarization techniques
to achieve drag reduction.
10.6 Synergism
Ideally, one searches for combinations of drag reduction methods for which the total
favorable effect is greater than the sum. For example, Lee et al. (1974) have shown
that the use of polymers and bers together can produce much larger reduction in
skin friction than either additive alone. A slightly different strategy, but perhaps no
less benecial, is to employ a second control method that reduces the drag penalty
of a given technique while adding some saving of its own. An example of this is the
226 DRAG REDUCTION
use of uid that is being withdrawn from a certain portion of a vehicle for injection
into another location, thus avoiding the ram drag penalty associated with blowing
freestream uid. For an aircraft, tangential slot injection or normal distributed injec-
tion may be used to reduce the skin friction on portions of the fuselage, as discussed
earlier. Possible sources of low-loss air include LFC suction from the wings and em-
pennage, relaminarization slots on the fuselage, and active or passive bleed air for
separation control.
Several possible combinations of drag reduction techniques were briey men-
tioned elsewhere in this chapter. The sword-sh conguration (small wetted area)
may be used in the forward part of a vehicle, where the local skin friction is high,
followed by the rest of the body, where the boundary layer is thicker and the skin-
friction coefcient is low. Anose spike followed by a short region of concave curvature
and then a longer convex portion may be used in supersonic ows. Massive suction
through a spanwise slot may be used to ingest the entire mass ow in a turbulent
boundary layer followed by any of the stability modiers discussed in Section 6.4 to
maintain the growing laminar boundary layer downstream.
The suction rate necessary for establishing an asymptotic turbulent boundary
layer independent of streamwise coordinate (d
/dx = 0) is much lower than the rate

required for relaminarization but still not lowenough to yield net drag reduction. For
Re = O[10
6
], Favre et al. (1966), Rotta (1970), and Verollet et al. (1972), among oth-
ers, reported an asymptotic suction coefcient of C
q
0.003. FromEq. 2.44 rewritten
for a steady, incompressible, zero-pressure-gradient boundary layer on a at plate,
the corresponding skin-friction coefcient is C
f
= 2C
q
= 0.006, indicating higher
skin friction than if no suction were applied. The problem is that uid withdrawn
through the wall has to come from outside the boundary layer where the streamwise
momentum per unit mass is at the relatively high level of U
, as was discussed in
Section 6.4 in connection with using suction for transition delay. To achieve a net
drag reduction with suction, the process must be further optimized. The results of
El ena (1975, 1984) and more recently of Antonia et al. (1988) indicate that suction
causes an appreciable stabilization of the low-speed streaks in the near-wall region.
The maximum turbulence level at y
+
12 drops from 15 to 12% as C
q
varies from
0 to 0.003. More dramatically, the tangential Reynolds stress near the wall drops by
a factor of two for the same variation of C
q
. The dissipation length scale near the
wall increases by 40%, and the integral length scale by 25% with the suction.
Gad-el-Hak and Blackwelder (1987c, 1989) suggested that one possible means
of optimizing the suction rate is to be able to identify where a low-speed streak is
located and to apply a small amount of suction under it. Assuming that the pro-
duction of turbulent energy is due to the instability of an inectional U(y) velocity
prole, one needs to remove only enough uid so that the inectional nature of the
prole is alleviated. An alternative technique that could reduce the Reynolds stress is
to inject uid selectively under the high-speed regions. The immediate effect would
be to decrease the viscous shear at the wall, resulting in less drag. In addition, the
velocity proles in the spanwise direction, U(z), would have a smaller shear, U/z,
because the injection would create a more uniform ow. Because Swearingen and
Blackwelder (1984) and Blackwelder and Swearingen (1990) have found that inec-
tional U(z) proles occur as often as inection points are observed in U(y) proles, in-
jection under the high-speed regions would decrease this shear and hence the resulting
instability. Because the inectional proles are all inviscidly unstable with growth
10.6 SYNERGISM 227
rates proportional to the shear, the resulting instabilities would be weakened by the
suction or injection process. Gad-el-Hak and Blackwelder (1987c, 1989) proposed to
combine suction with nonplanar surface modications. Minute longitudinal rough-
ness elements, if properly spaced in the spanwise direction, greatly reduce the spatial
randomness of the low-speed streaks (Johansen and Smith 1986). By withdrawing
the streaks forming near the peaks of the roughness elements, less suction should be
required to achieve an asymptotic boundary layer. Experiments by Wilkinson and
Lazos (1987) and Wilkinson (1988) combined suction or blowing with thin-element
riblets. Although no net drag reduction has yet been attained in these experiments,
the results indicate some advantage in combining suction with riblets, as proposed
by the patent of Blackwelder and Gad-el-Hak (1990).
An LEBU inuencing the outer structures of a turbulent boundary layer may be
combined with drag reduction methods that mainly inuence the near-wall region.
Walsh and Lindemann (1984) showed that the performance of riblets in the presence
of an LEBU is approximately additive and not quite a synergetic effect. Wilkinson
et al. (1988) proposed using an LEBU device to reduce the free mixing between
tangentially injected uid and the boundary layer ow, thus extending the low skin-
friction region. Bushnell (1983) suggested that the suction mass-ow rate required
to achieve relaminarization of a turbulent boundary layer may be reduced by the use
of an LEBU ahead of the suction region.
The LEBU model of Gebert (1988) indicates that the uctuating normal veloc-
ity component is most effectively suppressed for a particular incoming harmonic
disturbance. The modest drag reduction currently achieved using an LEBU device
in a natural boundary layer may, therefore, be greatly enhanced by cyclically gen-
erating the large outer structures using the method developed by Gad-el-Hak and
Blackwelder (1987a). In this case, the large eddies arrive at a given location at a
precisely controlled period that matches the optimum frequency for the LEBU de-
vice. Alternatively, Goodmans (1985) technique to generate closely spaced turbulent
spots in the spanwise and streamwise directions may be used for tripping a laminar
boundary layer. An LEBU device placed downstream will encounter more uniformly
paced large eddies, and its performance may be enhanced. Of course, in both schemes
the penalty associated with articially generating the large-eddy structures must be
considered.
In a recent article, Fontaine et al. (1999) explored the drag reduction achieved by
combining carbon dioxide microbubbles injection with homogenous dilute solutions
of either drag-reducing polymer (polyethylene oxide) or surfactant (Aerosol OT). At
low gas-injection rates, their results indicated that the combination of microbubbles
injection with polymer additives can yield drag reduction levels exceeding those of the
individual systems, but no synergismwas observed. Additionally, reducing the bubble
size through the addition of the surfactant did not favorably affect the characteristics
of microbubbles drag reduction. Fontaine et al. (1999) speculated that, though the
bubble size was reduced by as much as 23% in the presence of the surface-tension-
reducing Aerosol OT, the diminished bubble diameter was still large compared with
the turbulent scales in their water boundary layer.
Most of the combined drag reduction techniques discussed or proposed in this
section are speculative owing to the scarcity of research in the area of synergism.
Unguided by any theoretical framework and faced with the complexity of each of the
methods used individually, one is tempted to stay away from the much more difcult
228 DRAG REDUCTION
situation in which two or more techniques are combined. However, the potential
for achieving a total effect larger than the sum should stimulate more research for
combinations of methods. In fact, in some situations one may not have much of
a choice. Certain techniques such as suction may not yield net drag reduction and
must, therefore, be combined with something else to achieve the desired goal. Other
methods such as riblets yield such modest drag reductions (less than 10%) that it
may not be worth the effort to implement the individual technique.
Drag reduction is a subject that is explicitly or implicitly interwoven in every chapter
of this book. The practical benets of achieving even a modest reduction in uid resis-
tance to moving bodies are enormous. There is no shortage of ideas on howto reduce
the drag. Many challenges remain, however, when attempting to apply laboratory
curiosities to real-life situations. For airplanes and submarines, streamlining has led
to signicant reduction in the pressure drag component attributed to separation. That
is not the case for most land vehicles which, despite substantial improvement during
the last few decades, are still more or less blunt bodies. Natural laminar ow (NLF)
and laminar ow control (LFC) make it possible to have aircraft wings with substan-
tial laminar regions and thus to benet from the lower skin friction associated with
this ow regime. Several techniques like polymer injection, LEBUs, and riblets can
reduce the skin friction in turbulent boundary layers, but there are certain obstacles
for the widespread use of any of these methods. Polymers, in spite of their very im-
pressive percentage drag reduction and their successful application to internal ows,
are not sufciently inexpensive and require large volume for their storage onboard
ships and submarines. Riblets lead to very modest drag reduction, and LEBUs are not
effective at high Reynolds numbers. Newer strategies such as the random roughness
of Sirovich and Karlsson (1997) or the large-scale vortex generators of Schoppa and
Hussain (1998) still require more validationparticularly at eld Reynolds numbers.
The grand problem remains to nd a practical, cost-effective method to reduce the
turbulent skin-friction drag encountered on the long fuselage of commercial aircraft
or of water vessels. We will return to this topic in Chapter 14.
CHAPTER ELEVEN
Mixing Enhancement
Now I think hydrodynamics is to be the root of all physical science, and is
at present second to none in the beauty of its mathematics.
(William Thomson (Lord Kelvin), 18241907)
I pass with relief from the tossing sea of Cause and Theory to the rm
ground of Result and Fact.
(Sir Winston Leonard Spencer Churchill, 18741965,
in The Malakand Field Force)
PROLOGUE
This chapter deals with the enhancement of rates of mass, momentum, and heat
transfer. This is desired to improve the performance of mixers of nonreacting species;
chemical reactors, including combustors; lifting surfaces; heat exchangers, including
those in HVAC systems; and numerous other man-made devices. In nature, the ef-
cient turbulence mixing is responsible for the nearly uniform distribution of oxygen,
carbon dioxide, and internal energy in the Earths atmosphere and oceans. Without
that more-or-less uniform distribution, seasonal and latitudinal temperature changes
would be even more extreme. Oxygen would be more concentrated in the equatorial
rain forests, and carbon dioxide would be more concentrated in industrial and ur-
ban centers. Under such circumstances, city dwellers at least would not survive long
on this third planet from the sun. In plants and animals, the efcient transport of
mass and energy through limited spaces is also vital for their survival. The subject of
mixing enhancement is very broad, is important to entire industries, and is covered
extensively in many journals and books. The coverage here by necessity will be rather
brief, focusing mainly on enhancing convective mass, momentum, and heat transport
in both laminar and turbulent ows.
Flowcontrol can be used to augment mixing in both free-shear and wall-bounded
ows by increasing the effective area through which transport takes place, by setting
off resonant ow instabilities, by advancing laminar-to-turbulence transition, and by
enhancing the turbulence once the shear ow is already turbulent. In cases in which
early transition is not possible (or desired), as for example when the Reynolds num-
ber is not high enough because of small density, small length-scale, small velocity,
or large viscosity, chaotic advection can be effected in laminar ows to enhance the
229
230 MIXING ENHANCEMENT
convective rates of mass and heat transfer. The effective generation of secondary
ows, recirculation zones, or ow unsteadiness is an additional tool to enhance
mixing in both laminar and turbulent ows. Successful examples of this strategy
include the use of vortex generators on airplane wings and coiled tubes in heat ex-
changers. There is a drag penalty associated with increasing the surface area as well
as with all the other methods used to enhance mixing, and attempts must be made
to minimize this penalty.
11.1 Introduction
According to Ottino (1990), the Reynolds number in problems in which mixing is
important varies by an amazing 40 orders of magnitude. Over periods measured in
millions of years, mixing in the Earth mantle may take place at Reynolds numbers
as low as 10
20
, and for mixing in the interior of the stars the Reynolds number may
reach 10
20
; the Reynolds number of man-made devices is somewhere between these
two extremes. The ow regimes may be creeping, laminar or turbulent, diffusion
dominated or advection dominated, subsonic or hypersonic, and so forth. The uids
involved may be a single uid or many, miscible or immiscible, Newtonian or non-
Newtonian, one-phase or multiphase, and so on.
In accordance with the user group, the mass transport phenomenon is variously
termed mixing, stirring, blending, fusing, mingling, coalescing, or agitating. In this
chapter, mixing is used generically to denote the transport of mass, momentum,
or energy. There is no monolithic solution to the mixing problem. It is obvious,
for example, that exploiting turbulence to enhance mixing is possible only in some
problems but impractical in many others.
At times, we may desire to enhance the rate of mass or heat transfer without
unduly increasing the rate of momentum transferthe latter translating directly into
a skin-friction drag penalty. The decoupling between momentum transfer on the
one hand and heat and mass transfer on the other is not always easy to accomplish
of course, because the tools used to increase the rate of convective transport of
one quantity typically increase the transport rate of all others. At other times, we
may purposefully enhance the rate of momentum transfer, for example, to prevent
boundary-layer separation. In this case, the concomitant increase in skin-friction
drag is quite acceptable as compared with the alternative: a much higher increase in
pressure drag had the ow separated.
Given unlimited time and space, desired transport of chemical species, momen-
tum, or energy can eventually be achieved solely by molecular diffusion processes.
In the combustion chamber of a jet engine, however, time and space are both rare
commodities. Mixing the fuel with its rapidly moving oxidant must take place in the
shortest distance possible and must also be nearly complete to ensure higher ther-
modynamic efciency, less pollution, and so forth. Diffusion-dominated mixing
although good enough for the laminar ame of a candleis clearly too slow for the
jet engine and many other applications.
This chapter focuses on mixing enhancement in advection-dominated (i.e., high-
Reynolds-number) ows. Note, however, that, regardless of the global Reynolds
number (or more appropriately for mass and heat transfer the P eclet number) regime,
ultimately molecular diffusion dominates at sufciently small-scale, high-gradient
11.2 MIXING 231
regions. Mixing of species by advection is related to the stretching and folding of
material surfaces in three-dimensional owelds or material lines in two-dimensional
ones. This is the mechanism by which distant uid particles eventually come in con-
tact with one another, that is, become mixed. The enhanced stretching and folding
processes can result in greater surface areas and higher gradients available for the
molecular diffusion mechanism to effectively nish the job by smearing and ulti-
mately eliminating any remaining discontinuities or gradients in velocity, temper-
ature, species concentration, and so forth. Provided that the oweld is already
known and that the advection dominates, nearly complete description of the mixing
processes can be obtained purely kinematically down to the scales at which molecular
diffusion becomes appreciable. The book by Ottino (1989a) emphasizes this point
of view and provides a comprehensive treatment of the kinematics of mixing.
For some applications, the efcient mixing or transport mechanism of turbulence
may be feasible as well as desirable. For example, a turbulent boundary layer in
general is more resistant to separation than a laminar one, turbulence is used to
homogenize uid mixtures and to accelerate chemical reactions, and turbulent heat
exchangers are much more effective than laminar ones.
For other applications, the Reynolds number may be too low, and advancing
laminar-to-turbulence transition may be impractical. Mixing the highly viscous in-
gredients needed to make chocolate is such an application. In this case, it does not
make much economic sense to increase the owspeed or scale sufciently to promote
turbulence. Instead, newly advanced techniques to effect chaotic mixing in laminar
ows may be used to improve the efciency of the mass transfer processes involved.
Increasing the surface area using external and internal ns and inducing appropriate
secondary ows are additional strategies to improve the rates of mass, momentum,
and heat transfer in both laminar and turbulent ows.
11.2 Mixing
Stirring or mixing is basically a process that involves a reduction of length scales (i.e.,
thinning of material volumes) accomplished by stretching and folding of material sur-
faces. This archetypal mixing process is not unlike that found in the interior of stars
or that used in making puff pastries. The primary transport mechanism associated
with stretching and folding is macroscopic uid motion termed advection. Micro-
scopic uid motions, on the other hand, are responsible for the molecular diffusion
processes. The rate of transport of mass, momentum, or heat by molecular diffusion
depends on (1) the surface area, (2) the appropriate diffusion coefcient (e.g., binary
mass diffusion coefcient when blending two chemical species, viscosity coefcient
when transporting momentum or vorticity, and thermal conductivity when conduct-
ing heat), and (3) the gradient of the transported quantity. For simple uids like air
and water under most circumstances, the transport coefcients are scalars, and the
respective diffusion relations are linear but can be nonlinear as well as quite involved
for complex uids and ows. For example, the heat conducted per unit time per
unit area in an isotropic, Fourier uid is related linearly to the temperature gradient
q =
f
T (11.1)
where q is the heat ux due to molecular diffusion, T is the temperature eld, and
the thermal conductivity
f
is a scalar that in general depends weakly on temperature
and pressure.
1
The diffusion coefcients are physical (thermodynamic) properties of the uid(s)
involved and, short of modest changes with pressure and temperature, not much can
be done about them. Increasing the surface area available for the diffusion process
results in proportional increase in the rate of transport. Similarly, increasing the gra-
dient of the transported quantity proportionally enhances the diffusion process. That
is where mixing by advection comes into play. If done right, the stretching and folding
of material surfaces can dramatically increase both the area and the gradient available
for the molecular diffusion to become pronounced. Secondary ow, turbulence, and
chaotic advection can all lead to mixing enhancement. As always in ow control, the
challenge is to accomplish the task with a strategy that is simple, is inexpensive to
construct and operate, and has the least adverse effect on other ow considerations.
We start by recalling the important nondimensional parameters for the problem.
The Reynolds number is a characteristic ratio of advective momentumux to diffusive
momentum ux. Thus,
Re
v
o
L
(11.2)
where v
o
and L are respectively the convective velocity and length scales of the ow,
and is the uid kinematic viscosity. The heat transfer P eclet number is dened as
the ratio of advective heat transfer to conduction
Pe RePr =
v
o
L
T
(11.3)
where Pr is the Prandtl number (/
T
), and
T
is the uid thermal diffusivity
(
f
/c
p
). Similarly, the mass transfer P eclet number is dened as the ratio of ad-
vective mass transfer to mass diffusion
Pe ReSc =
v
o
L
D
AB
(11.4)
where Sc is the Schmidt number (/D
AB
), and D
AB
is the binary mass diffusion
coefcient between chemical species Aand B. The focus in this chapter is on mixing
by advection. Therefore, the Reynolds number and both P eclet numbers are assumed
to be large.
Finally, the Nusselt number is equal to the dimensionless temperature gradient
at the wall and provides a measure of the convective heat transfer occurring at the
surface
Nu
hL
f
=
T
=0
(11.5)
where h is the convection heat transfer coefcient,
f
is the uid thermal conduc-
tivity, and the superscript

denotes dimensionless quantity. Similarly, the Sherwood
1
Note that once
f
varies with T, the heat ux-temperature relation (11.1) becomes nonlinear.
11.2 MIXING 233
number is equal to the dimensionless concentration gradient at the wall and provides
a measure of the convective mass transfer occurring at the surface
Sh
h
m
L
D
AB
=
C
A
y
=0
(11.6)
where h
m
is the convection mass transfer coefcient, and C
A
is the concentration (or
mass fraction) of species A.
11.2.1 Kinematical Description
To analyze advection-dominated mixing, we may simply follow the uid surfaces
as they stretch and fold by the rate-of-strain eld. In principle, this kinematical
description of the process requires only knowledge of the velocity eld and of the
initial distribution of the particular transported quantity. At some scale of the motion
at which molecular diffusion becomes important, the pure kinematical description is
inadequate, and more involved analysis is needed. As a background, we rst elaborate
on the difference between a passive scalar and a passively transported tracer. Although
the literature on these two seemingly simple concepts is sufciently muddled, we will
argue below that a passive scalar is not necessarily a passively transported tracer. As
maintained by Hassan Aref (private communication), there are different degrees of
passivity.
A passive scalar, or passive contaminant, is one that does not affect the dynamics
of the ow in which it resides. In this case, the conservation equation for that scalar
is decoupled from the continuity and momentum equations describing the ow dy-
namics. For example, temperature is considered a passive contaminant if the ow
is incompressible and its density and viscosity are both independent of temperature.
In this case, the energy equation is decoupled from the NavierStokes equations.
Provided that the thermal conductivity is constant, the decoupled energy equation
is linear and can usually readily be integrated after the velocity eld is computed
from the continuity and momentum equations. The exception is turbulent owelds,
whichshort of direct numerical simulationsare known only probabilistically, and
nonintegrable laminar ows, that is, those admitting chaotic solutions to the energy
equation despite a well-behaved velocity eld (see Section 11.6).
A tracer that follows uid particlesa passively transported traceris one for
which diffusion is negligible and no source terms exist in its conservation equation.
For example, temperature T is passively transported or simply advected by the ow
and described by
DT
Dt
=
T
t
+u T = 0 (11.7)
if both conduction (molecular diffusion of heat) and dissipation are neglected in
the energy equation. The former requires an innite P eclet number, and the latter
requires the ratio of Eckert number to Reynolds number to be vanishingly small (see
Eq. (2.25)).
It is clear then that a passive scalar is not necessarily a passively transported tracer,
and vice versa. For example, the low-speed ow of a low-Prandtl-number uid (such
as mercury) can be described by a decoupled energy equation provided that the
temperature differences involved are not too high. In other words, the temperature
in this case can be treated as a passive contaminant. Heat conduction is important,
however, and the energy diffuses appreciably as it is advected by the ow and cannot
therefore be considered a passively transported tracer. For a reverse example, consider
the high-speed ow of a gas away from solid surfaces. For such a compressible,
inviscid, nonconducting ow, the energy equation is coupled to the momentum and
continuity equations, and the temperature eld does affect the owdynamics, that is,
T can no longer be considered a passive scalar. But Eq. (11.7) still holds, albeit with
an unknown velocity eld, and T is a passively transported tracer constant along any
particular particle path.
Similar arguments can be made if species concentration replaces temperature. In
this case, the mass transfer P eclet number determines the importance of advection
relative to molecular diffusion, and chemical reactions, if any, provide the source
terms in the species conservation equation. Only a nonreacting, nondiffusing species
can be considered a passively transported tracer.
Within the preceding approximation, the passively transported tracer T remains
constant along any particular particle path, and the Lagrangian equivalent to the
Eulerian equation (11.7) reads
dX
dt
= u (x, t) (11.8)
where X is the instantaneous position of a uid particle
2
characterized by its initial
position X
o
at a reference time say t = 0, and u is the Eulerian velocity eld, which is
presumably a known function of position and time determined from integrating the
continuity and momentumequations. Equation (11.8) is a dynamical systemdescrib-
ing the particle paths. For time-dependent ows, the particle paths, streamlines, and
streaklines may not coincide, and the dynamical system (11.8) is nonautonomous.
To compute the spatial distribution of a passively transported tracer at the present
time t, Eq. (11.8) is integrated backwards, in the direction of negative time, to deter-
mine the initial positions of all uid particles whose present positions are of interest.
The value of T at any present uid particle position is then simply equated to the
(presumably known) tracer value at the initial particle position at t = 0. In accor-
dance with the velocity eld, Eq. (11.8) can be linear or nonlinear and can be either
integrable or nonintegrable. Chaotic (i.e., nonintegrable) solutions can be expected
(but not guaranteed) if the number of degrees of freedom of the dynamical system
is at least three and if Eq. (11.8) is nonlinear. Two-dimensional, unsteady ows and
three-dimensional steady ows can thus exhibit chaotic behavior of particle trajec-
tories even if the velocity eld itself is nonchaotic.
As a result of chaotic particle motions, enhanced mixing and transport take place
because a group of adjacent particles in chaotic motion will tend to deviate exponen-
tially from one another, and this leads to substantial mixing with the surrounding
uid in a much shorter time than would be expected for regular motion. The resulting
tremendous stretching of nite uid regions paves the way for an enhanced activity
by diffusion due to the greater surface area and gradient available for the molecular
processes. We will return to the issue of chaotic advection in Section 11.6.
2
Note that x and X are the same quantity: position in physical space. The former is the spatial position
an independent variablein the Eulerian frame. The latter symbol is the material point position elda
dependent variablein the Lagrangian frame.
11.3 EARLY TRANSITION 235
11.2.2 Enhancement Strategies
To close this section on mixing, the following list categorizes the variety of strate-
gies available to enhance the transport of mass, momentum, and heat:
Increase the surface area of the solid using external and internal ns.
Set off resonant ow instabilities.
Advance laminar-to-turbulence transition.
Enhance turbulence Reynolds stresses.
Induce appropriate secondary ows, recirculation zones, or ow unsteadiness

in both laminar and turbulent regimes.
Effect chaotic advection in laminar ows.

The rst itemon this list deals with the solid surface area available for the transport of
mass, momentum, and energy. All other items exploit the augmented stretching and
folding of material surfaces to increase the uid surface areas and gradients available
for the molecular diffusion mechanism to nish the mixing task effectively.
11.3 Early Transition
Turbulence transports mass, momentum, and energy with equal zeal. In fact, the
ability to disperse and mix is the most prominent syndrome of turbulent ows. Thus,
as compared with laminar ows, turbulent boundary layers are thicker, are more
resistant to separation, have higher skin friction, and are characterized by higher
coefcients of convective mass and heat transfer. Mixing in free-shear ows is also
dramatically enhanced by early transition to turbulence, an occurrence that is readily
achieved at relatively low Reynolds numbers.
An important characteristic of turbulent ows is the random motion ability to
transport or mix mass, momentum, and energy. The respective rates of transport are
typically several orders of magnitude greater than the corresponding uxes owing to
molecular diffusion. The turbulence Reynolds number, dened using a characteristic
turbulence velocity u (such as the rms of the streamwise velocity uctuations) and a
length scale (such as the characteristic size of large eddies),
Re
u
(11.9)
gives approximately the ratio of rate of momentum transfer due to turbulence to
that due to molecular motion (i.e., due to viscous transport). Looking at it from a
different perspective, in a problem with an imposed length scale, Re
1
is the ratio
of a turbulence time scale to a molecular diffusion time scale that would prevail
in the absence of turbulence. Turbulence is almost always a high-Reynolds-number
phenomenon, and hence the turbulence diffusion is much faster than the molecular
one. For example, the warmth of a hot-water radiator is felt throughout a typical-
sized room in minutes rather than hours or even days. This enhanced transport,
as compared with molecular diffusion, is caused by the weak natural-convection
turbulent currents that are driven by the temperature and density gradients in the
vicinity of the radiator.
The Reynolds number above should not be confused with that in Eq. (11.2).
Consider, for example, a high-Reynolds-number, boundary-layer-type ow. In this
case, the Reynolds number in Eq. (11.2) is related to the ratio of the convective length
scale along the ow to the much smaller diffusive length scale across the ow. The
turbulence Reynolds number in Eq. (11.9), on the other hand, reects the ratio of
the cross-streamlength scale in a turbulent boundary layer to the cross-streamlength
scale in a laminar owhaving the same v
o
L/. The tangential Reynolds stress (uv)
very efciently drives the cross-streamgrowth in the turbulent owcase, whereas the
generally smaller viscous stress (U/y) is responsible for the more modest laminar
boundary layer growth. Having a larger Re in Eq. (11.2) than that in Eq. (11.9)
merely indicates that the streamwise length scale is larger than the cross-stream one
even in the turbulent ow case (i.e., the boundary layer is still relatively thin).
Similar arguments can be made for heat and mass transfer. The turbulence P eclet
number Pe dened using the thermal diffusivity
T
or the binary mass diffusivity
D
AB
,
Pe
u
T
(11.10)
Pe
u
D
AB
(11.11)
compares respectively the heat transfer rate or the mass transfer rate due to turbulence
to that due to molecular diffusion.
11.3.1 Wall-Bounded Flows
Advancing transition for wall-bounded ows is discussed rst, and the topic of
free-shear ows is deferred to the following subsection. As a means for preventing
ow separation, particularly for low-Reynolds-number lifting surfaces, early transi-
tion in wall-bounded ows was considered in Chapters 8 and 9. Methods to advance
boundary-layer transition are of course the mirror image of the transition-delaying
strategies discussed in Chapter 6.
Recall that in low-Reynolds-number terminology the transition-promoting de-
vices are called turbulators. For a zero-pressure-gradient boundary layer, transition
typically occurs at a Reynolds number based on distance from leading edge of the
order of 10
6
. The critical Reynolds number Re
crit
below which perturbations of all
wave numbers decay is about 6 10
4
. To advance the transition Reynolds number,
one may attempt to lower the critical Re, increase the growth rate of Tollmien
Schlichting waves, or introduce large disturbances that can cause bypass transition.
The rst two routes involve altering the shape of the velocity prolemaking it
less fullusing wall motion, injection, adverse pressure gradient, surface heating in
gases or cooling in liquids, or other strategies to increase the near-wall viscosity. The
third route, exposing the boundary layer to large disturbances, is much simpler to
implement though more difcult to analyze.
Morkovin (1984) broadly classied the large disturbances that can cause bypass
transition into steady or unsteady ones originating in the freestream or at the body
surface. The most common example is single, multiple, or distributed roughness ele-
ments placed on the wall. For lifting surfaces, for example, the mechanical roughness
elements, in the form of serrations, strips, bumps, or ridges, are typically placed near
11.3 EARLY TRANSITION 237
the airfoils leading edge. If the roughness characteristic length is large enough, the
disturbance introduced is nonlinear, and bypass transition takes place. For a discrete
three-dimensional roughness element of height-to-width ratio of one placed on a at
plate, Tani (1969) reported a transition Reynolds number of Re
300 for a rough-

ness Reynolds number of Re
10
3
. Here, Re
is based on the velocity outside the

boundary layer and the displacement thickness, and Re
is based on the height of

the roughness element and the velocity in the undisturbed boundary layer at the
height of the element. Note that the transition Reynolds number indicated above
|
crit
= 420 predicted from the linear stability theory for a
zero-pressure-gradient at plate. For a roughness Reynolds number of about 600,
transition occurs at Re
10
3
. For a smooth surface, in contrast, transition typically
takes place at Re
2.6 10
3
.
Other large disturbances that can lead to early transition include high turbulence
levels in the freestream, external acoustic excitations, particulate contamination, and
surface vibration. These are often termed environmental tripping. Transition can also
be effected by detecting naturally occurring TS waves and articially introducing
in-phase waves. Transition can be considerably advanced, on demand, using this
wave superposition principle.
Early transition can also be achieved by exploiting other routes to turbulence such
as TaylorG ortler or crossow vortices. For example, a very mild negative curvature
of (0.003/
) results in the generation of strong streamwise vortices. In this case, the

transition Reynolds number is lowered from Re
2600 for the at-plate case to

Re
700 for the concave surface. For high-Mach-number ows, the general decay
in spatial amplication rate of TS waves makes conventional tripping more difcult
as the Mach number increases. For these ows, trips that generate oblique vorticity
waves of appropriate wavelength to cause resonant growth of three-dimensional
modes may be most effective to advance the transition location. For example, on a
sharp cone at Ma = 3.5, Corke and Cavalieri (1997) and Corke et el. (1999) used an
array of plasma (corona-discharge) electrodes to excite oblique vorticity-mode pairs
and successfully advanced the transition of the supersonic boundary layer.
11.3.2 Free-Shear Flows
As discussed in Chapter 3, free-shear ows, such as jets, wakes, and mixing layers,
are characterized by inectional mean-velocity proles and are therefore susceptible
to inviscid instabilities. Viscosity is only a damping inuence in this case, and the
prime instability mechanism is vortical induction. Control goals for such ows in-
clude transition delay or advancement, mixing enhancement, and noise suppression.
Free-shear ows and separated boundary layers are intrinsically unstable and lend
themselves more readily to manipulation.
Free-shear ows originate from some kind of surface upstream, be it a nozzle, a
moving body, or a splitter plate, and ow-control devices can therefore be placed on
the corresponding walls, albeit far fromthe fully developed regions. Examples of such
control include changing the geometry of a jet exit from circular to elliptic (Gutmark
and Ho 1986), using periodic suction or injection in the lee side of a blunt body to
affect its wake (Williams and Amato 1989), and vibrating the splitter plate of a mixing
layer (Fiedler et al. 1988). These and other techniques are extensively reviewed by
Fiedler and Fernholz (1990), who offer a comprehensive list of relevant references,
and more recently by Gutmark et al. (1995), Viswanath (1995), and Fiedler (1998).
Because of the nature of their instabilities, free-shear ows undergo transition at
extremely lowReynolds numbers as compared with wall-bounded ows. The critical
Reynolds number for jets, wakes, and mixing layers is typically Re
crit
= O[10]. Many
techniques are available to delay laminar-to-turbulence transition for both kinds of
ows, but none would do that to indenitely high Reynolds numbers. Therefore,
for Reynolds numbers beyond a reasonable limit, one should not attempt to prevent
transition but rather deal with the ensuing turbulence. Of course early transition
to turbulence can be advantageous in some circumstances, for example to achieve
separation delay, enhanced mixing or augmented heat transfer. The task of advancing
transition is generally simpler than trying to delay it and is more relevant to the
present chapter. Unless the uid is very viscous or the speed and length scales are
extremely small, jets, wakes, and mixing layers will readily undergo laminar-to-
turbulence transition close to their source.
In addition to grouping the different kinds of hydrodynamic instabilities as invis-
cid or viscous, one could also classify them as convective or absolute based on the
linear response of the system to an initial localized impulse (Huerre and Monkewitz
1990). A ow is convectively unstable if, at any xed location, this response eventu-
ally decays in time; in other words, if all growing disturbances convect downstream
from their source. Convective instabilities occur when there is no mechanism for
upstream disturbance propagation, as, for example, in the case of rigid-wall bound-
ary layers. If the disturbance is removed, then perturbation propagates downstream
and the ow relaxes to an undisturbed state. Suppression or enhancement of con-
vective instabilities is particularly effective when applied near the point where the
perturbations originate.
If any of the growing disturbances has zero group velocity, the ow is abso-
lutely unstable. This means that the local system response to an initial impulse grows
in time. Absolute instabilities occur when a mechanism exists for upstream distur-
bance propagation, as, for example, in the separated ow over a backward-facing
step where the ow recirculation provides such mechanism. In this case, some of
the growing disturbances can travel back upstream and continually disrupt the ow
even after the initial disturbance is neutralized. In some ows (e.g., two-dimensional
blunt-body wakes), certain regions are absolutely unstable whereas others are con-
vectively unstable. The upstreamaddition of acoustic or electric feedback can change
a convectively unstable ow to an absolutely unstable one, and self-excited ow os-
cillations can thus be generated (e.g., Reisenthel, Xiong, and Nagib 1991). In any
case, identifying the character of ow instability facilitates its effective control (i.e.,
suppressing or amplifying the perturbation as needed).
11.4 Turbulence Enhancement
As emphasized earlier, an important characteristic of turbulent ows is the random-
motion ability to transport or mix mass, momentum, and energy. The respective rates
of transport are typically several orders of magnitude greater than the corresponding
uxes due to molecular diffusion. But can one improve on that impressive feat?
The mass, energy, and momentum transport in a turbulent ow are accomplished
via strong correlations between the corresponding uctuating quantities (e.g., c
A
v,
v, uv, where c
A
is the concentration uctuations of species A, is the temperature
uctuations, and u and v are respectively the uctuating streamwise and cross-stream
11.4 TURBULENCE ENHANCEMENT 239
velocity components). It is obvious, then, that to enhance turbulence transport one
has to strengthen the aforementioned correlations. This can be achieved by effecting
appropriate secondary ows, recirculation zones, or owunsteadiness. We offer here
a few examples of passive and active control methods to accomplish this, rst for
boundary layers and then for free-shear ows.
11.4.1 Turbulent Wall-Bounded Flows
To enhance turbulence in wall-bounded ows, one may reverse the stimuli dis-
cussed in Section 10.5 in connection with relaminarization by using injection instead
of suction in external ows, decelerating ows instead of accelerating ones, negative
curvature (concave) instead of positive curvature (convex), unstable stratication
instead of a stable one, and so on. The reverse stimuli will lead to an increase in
turbulence production and enhanced mass, momentum, and heat transfer. Rough-
ness will also enhance the turbulence, but its associated drag penalty must be care-
fully considered. Other devices to enhance the turbulence mixing include vane-type
vortex generators, which draw energy from the external ow, or Wheeler-type or
Kuethe-type generators, which are fully submerged within the boundary layer and
presumably have less associated drag penalty (Rao and Kariya 1988). These and other
passive and active strategies to energize the near-wall turbulence and augment the
cross-stream momentum transport via Reynolds stresses were extensively discussed
in Section 8.8 in connection with separation delay.
The turbulence levels in the near-wall region can be modestly augmented by
freestream excitation. In a unique experiment, Bandyopadhyay (1988) used a pitch-
ing airfoil located outside a turbulent boundary layer to resonate the viscosity-
dominated unsteady ow in a row of minute transverse square cavities at the wall.
In a narrow band of pitching frequencies, the near-wall turbulence levels increased
by as much as 6%, whereas the rest of the boundary layer seemed transparent to the
external disturbances. This result is particularly signicant because it demonstrates
that near-wall events may be controlled from the freestream and, therefore, that a
dynamic coupling between disparate scales may exist.
11.4.2 Turbulent Free-Shear Flows
Turbulent free-shear ows, like their wall-bounded counterparts, are dominated
by large-scale coherent structures that, in turn, are lled with incoherent motions of
smaller scales. The ratio of the largest to the smallest scales varies as Re
(3/4)
, and
the large eddies control the most important dynamics of the ow. To strengthen the
correlations described above, most turbulence-enhancement strategies for free-shear
ows center around controlling the processes that produce and sustain the large
eddies. The goal is to achieve an order one effect using an order perturbation. The
control is best carried out in the early stages of development because the growth of
shear layers is rather sensitive to their initial conditions.
As examples of targeting large-scale structures, we rst focus on ow in turbu-
lent jets. The dominant structures in axisymmetric jets are toroidal rings in the near
eld and helical structures in the far eld. Superposition of a plane axial acoustic
perturbation having a frequency that falls within the sensitive range of the basic ow
locks in the initial formation of vortex rings to the excitation frequency (Zaman and
Hussain 1980). This causes the subsequent growth and spreading of the jet by vortex
pairing to be augmented. Other strategies to enhance mixing in jets include (1) us-
ing an elliptical orice, thus producing elliptical vortex structures that exhibit large
majorminor axis oscillations caused by a curvature-dependent self-induction and
interaction between azimuthal and axial vortices (Ho and Gutmark 1987; Hussain
and Husain 1989; Husain and Hussain 1991, 1993; Gutmark and Grinstein 1999);
(2) using suction at the jet lip to produce local regions of backow and large-
amplitude, self-excited oscillations associated with the resulting absolute instability
(Strykowski and Niccum 1991); and (3) forcing the jet with multiple frequencies and
modes to make the near-eld ring vortices slightly eccentric (Reynolds 1996).
In this latter strategy, the rings act to tilt one another, and thus they can be made
to move in different directions. By adding helical perturbations to axial ones, the
vortex rings leave the nozzle lip eccentrically, and for an axial-to-helical frequency
ratio of 2, the jet can be split into two distinct branches that emerge at an angle
to the jet axisthe bifurcating jet of Juvet and Reynolds (1989). If the ratio of
frequencies is incommensurate but in the range of 1.73.5, the vortex rings are sent
in all directions and the jet explodes into a shower of vortex ringsthe blooming jet
of Lee and Reynolds (1985). As compared with the spreading angle of about 4
for an
unperturbed axisymmetric turbulent jet, the spreading angle of a blooming jet can be
as much as 75
, which is achieved at Re
D
=5000 and an axialorbital frequency ratio
of 2.3. Note, however, that as the vortex rings wander away fromthe primary jet axis,
the blooming jet does not remain centered. This prompted some (e.g., Husain and
Hussain 1983) to termthe enhanced mixing as apparent and not real. With increased
intensity of the required acoustic perturbations, these remarkable phenomena have
been extended to jet Reynolds numbers as high as 10
5
.
The ability of reactants to come together is a critical factor in determining the
rate at which the product of a chemical reaction is formed. This is particularly true
in reactions in which the reactive and diffusive time scales are longer than the ad-
vective time scale. In this case, the overall reaction yield is very sensitive to uid
mixing. For fast chemical reactions, mixing is relatively slow, desired reactions are
slowed, and undesired ones are often enhanced. Product selectivity is reduced, and
excessive production of waste results. A notorious example is the highly undesirable
formation of nitric oxide (NO) and nitrogen dioxide (NO
2
) when oxygen combines
with atmospheric nitrogen rather than with the hydrocarbon fuel in internal com-
bustion engines. Both NO
x
compounds are considered toxic, and NO is related to
the formation of photochemical smog. Improvements of combustion processes not
only decrease fuel consumption but also reduce pollutant generation (Pope 1996).
For example, simply imparting swirl to the incoming air in jet-engine combustors
improves fuel economy and pollution emission. The same technique is now used
in high-efciency home oil burners. Active control strategies to enhance mixing in
combustion applications are reviewed by Haile et al. (1998).
Lastly, methods to enhance mixing in compressible free-shear ows are reviewed
by Gutmark et al. (1995). There the challenge stems from the inherently low growth
rates of supersonic and hypersonic shear layers.
11.5 Heat Transfer
The size, performance, and efciency of many engineering systems are often limited
by the ability to transfer heat to or from the system. Consider the supercomputer, for
which the speed of computing is limited by the distance through which the electrons
11.5 HEAT TRANSFER 241
must travel at nearly the speed of light. Electronic chips and other components must,
therefore, be miniaturized and packaged tightly. Tremendous amounts of heat must
be removed from a rather modest surface area. The central processing unit (CPU) of
a present-generation supercomputer is not much larger than, say, a refrigerator. To
limit the temperature of the CRAY2 to a tolerable range, its entire CPU is immersed
in a dielectric liquid coolant (trade name FC75). This is obviously an expensive, but
perhaps unavoidable, approach to achieving the desired goal.
If a boundary-layer-type ow is considered and radiation is neglected, the surface
heat loss per unit time due to the macroscopic motion of a uid is given by Newtons
law of cooling
q
s
=
f
A
T
y
y=0
+
= hA(T
s
T
o
) (11.12)
where
f
is the thermal conductivity of the uid, Ais the surface area, h is the convec-
tion heat-transfer coefcient that depends on the nature of the boundary layer and
uid motion as well as a host of uid thermodynamic and transport properties, T
s
is
the surface temperature, and T
o
is the uid temperature at the edge of the shear layer.
It is clear, then, that to maximize heat removal by convection, the surface area has
to be maximized using ns or inserts, for example; the convection coefcient has to
be maximized via secondary-ow generation, chaotic advection, early transition, or
turbulence enhancement; or the temperature difference between the solid surface and
the outer ow has to be maximized or a combination of these techniques needs to
be used. Identical arguments can be made regarding convective mass transfer. There
species concentration replaces temperature, and mass diffusion coefcient replaces
thermal conductivity.
Convective heat transfer enhancement methods were reviewed by Bergles and
Morton (1965), Bergles (1978, 1981, 1985), Webb et al. (1981), Bergles and Webb
(1985), Nakayama (1986),

Zukauskas (1986), Webb (1987, 1993), and Chang and
Sen (1995). The Journal of Enhanced Heat Transfer is devoted to the topic and
provides up-to-date references. The simplest scheme involves the introduction of
distributed roughness on the heat-transfer surfaces. This would destabilize the wall
region and intensify the mixing process, leading to an increase in the convection-
heat-transfer coefcient. Patera (1986) and Patera and Miki c (1986) introduced the
concept of resonant heat transfer enhancement based on excitation of shear-layer
instabilities in internal separated ows. In a two-dimensionally periodically grooved
channel, TollmienSchlichting-like waves forced by the KelvinHelmholtz shear layer
instabilities at the grooves edge become unstable and take the form of self-sustained
oscillations for Reynolds numbers greater than a critical value. For lower Re, oscil-
latory perturbations result in subcritical resonant excitation. For both laminar and
turbulent ows, the resulting large-scale motions lead to signicant lateral mixing
and correspondingly dramatic enhancement of convection heat transfer coefcient.
Other devices to enhance convective heat transfer in pipes include coil-spring wire
inserts, twisted tape inserts, and longitudinal and helical ribs. These artifacts intro-
duce additional no-slip surfaces as well as increase the momentum ux in a direction
transverse to the ow, leading to signicant pressure-drop penalty.
The secondary motion in the transverse plane of a coiled tube provides higher
heat transfer rates compared with straight tubes. The same geometry can of course
Figure 11.1: Secondary ow in a curved segment of a pipe.
be used to enhance the rate of mass transfer in chemical reactors, for example. The
Dean number is dened as
De
Re
2
a
R
(11.13)
where Re is the Reynolds number based on the bulk velocity and pipe diameter, a is
the radius of the cross section, and R is the radius of curvature. This dimensionless
parameter measures the curvature effects compared with viscous effects. At a critical
De, a secondary ow emerges in a curved pipe as a second stable laminar ow. Fluid
particles near the ow axis have higher velocity and hence are acted upon by a larger
centrifugal force than the slower particles in the vicinity of the wall. This results in
a pair of streamwise vortices with ow directed outwards in the center and inwards
(toward the center of curvature) near the wall, as indicated in Figure 11.1.
A material line initially perpendicular to the secondary upwelling is stretched
the most, albeit linearly, as it is advected downstream by the primary ow.
3
The
secondary ow causes faster or slower primary ow and corresponding higher or
lower shear stress near the outside or inside of the bend. The coiling leads to an
increase in the convection coefcient as well as the pressure drop. At higher Dean
numbers, other instabilities develop, and the ow eventually becomes turbulent.
There is no accompanying increase in surface area when a tube is coiled, and if the
Reynolds number (or more precisely the Dean number) is not too high, the ow
remains laminar.
4
The increase in Nusselt number or Sherwood number in this case
is typically more than the increase in pressure drop, making coiled tubes more useful
for practical heat exchangers and chemical reactors (Shah and Joshi 1987).
Forming even more twists in a coiled tube may generate tertiary ow that en-
hances the mixing even further. In a recent article, Knight (1999a) speculated that
the unusually complex plumbing patterns of human arteries, respiratory ducts, and
urinary tracts may have evolved for precisely this advantage. He then raised the in-
triguing possibility of improving on present heart bypassesin which a section of
blood vessel in the form of a planar arch is spliced into a coronary artery to divert
blood around a blockageusing instead twisted, nonplanar bypasses. The resulting
helical bypass, indicated schematically in Figure 11.2, may lead to improved mixing,
smoothing out the undesired shear-stress extremes at the outside and inside of the
3
Contrast this linear stretching to the exponential one associated with chaotic advection (Section 11.6).
4
In fact, the centrifugal forces accompanying tight coiling may be sufciently strong to relaminarize what
would otherwise be a turbulent pipe ow. The stabilizing inuence in this case is analogous to that in
stably stratied ows.
11.5 HEAT TRANSFER 243
Figure 11.2: Conventional planar bypass and proposed twisted bypass graft.
bend and thus discouraging the accumulation of fat, proteins, and cholesterol at the
junction to the large artery. The spiral bypass better resembles natural blood vessels
and may lower the risk of two common modes of graft failure: (1) plaque accumula-
tion at the dead spot in the blood stream where the graft meets the primary artery,
and (2) excessive division of muscle cells in the arterial wall, which is an abnormality
thought to be triggered or at least aggravated by the presence of high-shear-stress
regions.
In microdevices (see Chapter 13), both radiative and convective heat loss or gain
are enhanced by the huge surface-to-volume ratio. For a device with a characteristic
length of 1 m, this ratio is a million times that of a device having a length of 1 m.
Consider a device having a characteristic length L
s
. Use of the lumped capacitance
method to compute the rate of convective heat transfer, for example, is justied
if the Biot number (hL
s
/
s
, where h is the convection heat transfer coefcient
and
s
is the thermal conductivity of the solid) is less than 0.1. Small L
s
implies
small Biot number and a nearly uniform temperature within the solid. Within this
approximation, the rate at which heat is lost by the solid to the surrounding uid is
given by
s
L
3
s
c
s
dT
s
dt
= hL
2
s
(T
s
T
) (11.14)
where
s
and c
s
are respectively the density and specic heat of the solid, T
s
is its
(uniform) temperature, and T
is the ambient uid temperature. Solving Eq. (11.14)

for constant convection coefcient is trivial, and the temperature of a hot surface
drops exponentially with time from an initial temperature T
i
,
T
s
(t) T
T
i
T
= exp
t
T
(11.15)
where the time constant T is given by
T =

s
L
3
s
c
s
hL
2
s
(11.16)
For small devices, the time it takes the solid to cool down is proportionally small.
Clearly, the millionfold increase in surface-to-volume ratio implies a proportional
increase in the rate at which heat escapes. Identical scaling arguments can be made
regarding mass transfer.
11.6 Chaotic Mixing
Stirring by chaotic advection can be an attractive alternative for situations in which
turbulence is either not possible or not desired. Mixing of highly viscous uids or
designers not willing to pay the turbulence drag penalty are two such circumstances.
The idea here is to effect chaotic particle paths even for well-behaved velocity elds. A
group of adjacent particles in chaotic motion will tend to deviate exponentially from
each other, and this leads to substantial mixing with the surrounding uid in a much
shorter time than would be expected for regular motion. The resulting transport can
be more effective than that in ordinary laminar owseven those with superimposed
secondary motions. The concept of chaotic advection has been demonstrated theo-
retically, experimentally, and numerically for mass and energy transport. Chaotic
advection, or Lagrangian chaos, is possible for time-dependent, two-dimensional (or
three-dimensional) owelds as well as for steady three-dimensional ows.
Consider the diffusion equation for a passive scalar eld (x, t) having a constant
diffusivity D. In the Eulerian description, the partial differential equation that governs
the diffusion process is linear because the velocity eld is obtained independently from
the continuity and momentum equations. The diffusion equation with no source
terms reads
t
+u = D
2
(11.17)
where the advecting velocity eld u(x, t) is a prescribed function of the spatial coor-
dinates x and time t. In a turbulent ow, u is a random vector eld and therefore is
also expected to be random. But the surprising resultif one considers the linearity of
the governing equationis that the scalar eld can exhibit chaotic behavior even
for simple laminar ows. This was demonstrated rather elegantly by Aref (1984),
who coined the expression chaotic advection.
When the problem of a passively transported tracer is analyzed from a
Lagrangian viewpoint, the governing equations may be nonlinear and nonintegrable.
The resulting Lagrangian chaos is a spatial not a temporal one. Therefore, Eq. (11.17)
11.6 CHAOTIC MIXING 245
is not the proper dynamical system to represent the problem, and its linearity or lack
thereof is irrelevant to the chaotic behavior of . The velocity at a point may be
constant or periodic, whereas the passively transported scalar may be aperiodic (i.e.,
chaotic) in space. Consider the trajectories of individual advected particles
dX
dt
= u(x, y, z, t) (11.18)
dY
dt
= v(x, y, z, t) (11.19)
dZ
dt
= w(x, y, z, t) (11.20)
The initial conditions are X|
t=0
= X
o
. These equations describe a nite-dimensional
dynamical system and are the Lagrangian equivalent of Eq. (11.17) for the case
of zero diffusivity. Chaotic particle paths lead to chaotic spatial distribution of the
transported tracer.
The preceding equations are deterministicmeaning the forcing terms on the
right-hand side are not randomfor nonturbulent velocity elds. For steady, two-
dimensional ow, the streamlines and particle paths coincide.
5
The equations for
particle paths can in this case be written in terms of the streamfunction, (x, y),
X=

y
,

Y =
x
(11.21)
In the language of dynamical systems theory, the preceding are just Hamiltons canon-
ical equations for one degree of freedomand hence are integrable when autonomous.
Therefore, a regular velocity eld leads to regular passive scalar distribution. On the
other hand, if the two-dimensional oweld is time-dependent, however innocuous,
the nonautonomous Hamiltonian dynamical system can be nonintegrable, produc-
ing stochastic particle paths. Note that for plane ow, the phase-space ow of a
Hamiltonian dynamical systemcorresponds to the conguration-space motion of ad-
vected particles. For three-dimensional ows, steady and unsteady laminar velocity
elds can result in chaotic particle paths and therefore highly complex distribution of
the passively transported tracer. In the steady case, a regular three-dimensional veloc-
ity eld can produce chaotic particle paths and identically irregular streamlinesan
amazing counterintuitive effect. These results provide a remarkable contrast between
the Eulerian and Lagrangian representation of the same ow: regular Eulerian ow-
elds can, under the right circumstances, lead to highly irregular advection patterns.
Unlike the fractal dimension test, which can only be used in dissipative dynamical
systems, the Lyapunov exponent provides a quantitative criterion for the presence
of chaos in both conservative (Hamiltonian) and dissipative deterministic systems. A
positive exponent implies chaotic dynamics and measures the sensitivity of the system
to changes in its initial conditions. If the initial distance between two trajectories in
phase space is d
o
, at a small but later time the distance is on the average
d(t) = d
o
e
t
(11.22)
5
As they also do in steady, three-dimensional ows.
where is the Lyapunov exponent. In the mixing problem, the Lyapunov exponent
can be used to measure the rate at which an innitesimal uid volume is stretched.
Each dimension of the ow has an associated exponent, and for a conservative,
incompressible dynamical system, the sum of all exponents must be zero. For a
nonchaotic system, stretching occurs at a linear rate, and thus all Lyapunov exponents
must be zero. For a system that exhibits chaotic particle paths, the largest Lyapunov
exponent is positive, denoting exponential separation of neighboring particles. The
exponent must therefore be negative in at least one other dimension, implying a
diminution of length scale in that direction.
If the two-dimensional or three-dimensional laminar oweld is cleverly chosen,
the dynamical system (Eqs. (11.18)(11.20)) becomes nonintegrable with resulting
stochastic response in the Lagrangian advection characteristics of a passively trans-
ported tracer. In this case, a group of adjacent particles will tend to deviate exponen-
tially from one another, and this leads to substantial mixing with the surrounding
uid in a much shorter time than would be expected for regular motion. The re-
sulting tremendous stretching of nite uid regions paves the way for an enhanced
activity by diffusion owing to the availability of greater surface area and gradient for
the molecular processes. Despite the exponential divergence of neighboring states (in
phase space), the stretching (i.e., divergence of neighboring trajectories) and folding
(i.e., connement to bounded space) mechanism is necessary to keep chaotic trajec-
tories within a nite volume of phase space. The stretching and folding operation
corresponds to what is called in dynamical systems theory a horseshoe map. Chaotic
mixing can result from the breakup of homoclinic or hetroclinic loops into tangles.
Those entanglements result from repeated intersections of the stable and unstable
manifolds, giving rise to chaotic behavior. This causes a set of particles to be well
mixed spatially as it is advected by even an innocuous oweld.
The utility of chaotic advection has been demonstrated for mass as well as energy
transport. Chaotic mixing can be effected even at extremely low Reynolds or P eclet
numbers where turbulence is not a viable option, making the technique particularly
useful for a range of applications in chemical engineering, low-speed ows, heat
transfer, and materials processing. Because Lagrangian chaos can be achieved even
at extremely low speeds, the enhanced scalar transport is not necessarily associated
with enhanced momentum transport. This is a tremendous practical advantage. In
contrast, scalar transport enhancement by turbulence is often at the expense of signif-
icant pressure drop, skin-friction increase, or both, and therefore power expenditure.
Herein we recall just a few examples from the recent deluge of demonstrations.
The original analytical work in the eld of chaotic advection is, as indicated earlier,
attributable to Aref (1984), who also cited several earlier contributions by other
researchers. Aref idealized an incompressible, inviscid uid being stirred in a tank
using point vortex agitator(s). Together with their images in the bounding contour, the
agitators provide the source of unsteady potential ow. The motion is assumed wholly
two-dimensional. Aref (1984) demonstrated by way of a numerical integration of
the time-dependent version of Eq. (11.21) that chaotic particle motions are possible
for certain unsteady movements of the stirring vortex. Two protocols were used
to generate the successful unsteadiness: either a stirrer that jumps back and forth
between two xed positions or two stirrers at xed positions that are run alternately
for a given time interval. In either case, efcient stirring was achieved when chaotic
particle paths were generated using the successful protocol.
Ottino (1989a,b; 1990) detailed the body of analytical and experimental work
conducted by his research group. Chaotic mixing of miscible and immiscible uids
has been achieved in time-dependent creeping ows in two simple two-dimensional
geometries. The rst is the journal-bearing arrangement in which two eccentric cylin-
ders counterrotate alternately for a xed time period. The second geometry is a cavity
ow in which the upper and lower walls are forced to move alternatelyeach paral-
lel to itself but in opposite directionsusing two independently driven timing belts.
In both cases good mixing is achieved after just few periods for certain optimum
movement protocols.
Again, for the creeping ow between two counterrotating eccentric cylinders,
Ghosh et al. (1992) have computed the enhancement in the cross-stream heat (or
mass) transport associated with either a steady recirculation region or chaotic advec-
tion. As compared with pure conduction (or molecular diffusion), the steady-state
enhancement is proportional to the square root of a characteristic width of the re-
circulation region, but the effect diminishes as the P eclet number is lowered from its
asymptotic innite value. If time-periodic forcing is used to perturb the steady recir-
culation region, 100%time-averaged enhancement over pure diffusion is achieved at
the optimum forcing frequency, leading to homoclinic entanglement of the recircula-
tion regions bounding separatrices and hence to chaotic advection transport across
those separatrices.
To avoid the practical difculties of using time-periodic perturbations at a bound-
ary, Acharya, Sen, and Chang (1992) investigated analytically, numerically, and ex-
perimentally the possibility of effecting chaotic advection in helical pipes by spatial
periodicity in the downstream direction. Using as a base ow a conventional, xed-
axis coiled tube, they perturbed the customary secondary ow by periodic changes
in the coiling. One such perturbation is shown in Figure 11.3, where each successive
loop of the coil lies on a different, mutually perpendicular plane. Chaotic particle
paths are observed when the switching length exceeds the critical value of one-half of
the loop circumference. As compared with a conventional coiled tube using realistic
uids (i.e., nite Re, Pr, and Pe), the alternating-axis coiled heat exchanger lead-
ing to chaotic advection enhances the heat transfer by 68% with only a 1.52.5%
pressure-drop penalty. Heat transfer enhancement by chaotic advection is reviewed
by Chang and Sen (1995).
The same alternating-axis coiled tube shown in Figure 11.3 was investigated in
an analytical study directed toward enhancing the rate of mass transfer. Sawyers
et al. (1996) documented the benets of chaotic advection on the overall yield in a
slow, bimolecular chemical reaction. The two reactants are initially separated, and
the ow is assumed laminar and steady with high mass P eclet number. In a straight
tube, there is no ow in the transverse direction, and the interface separating the
two reactants does not change as the ow progresses downstream. Mixing in this
case is achieved solely by the very slow molecular diffusion process. Regular mixing
produced by the secondary transverse ow in a helical coil linearly stretches the
interface separating two reactants, leading to some improvement in mixing. Chaotic
mixing, by contrast, leads to an exponentially growing interface length that can be
related to a positive Lyapunov exponent. The area-averaged product mass fraction
increases proportionally in this case, leading to a much higher yield of the reaction.
Numerical solutions show that the enhancement by chaotic mixing also exists for
fast reactions.
Figure 11.3: Conventional coiled tube and alternating-axis tube.
In a different kind of three-dimensional geometry, Sawyers, Sen, and Chang
(1998) computationally investigated the heat transfer enhancement in the lami-
nar ow inside corrugated channels. They compared the regular mixing in two-
dimensional sinusoidal corrugations (with primary ow perpendicular to the corru-
gations) with the chaotic mixing resulting when the corrugations are sinusoidal in
two orthogonal directions. The conventional geometry and the egg-carton one are
Figure 11.4: Channels with (a) two-dimensional corrugations; (b) egg-
carton corrugations.
schematically depicted in Figure 11.4. As compared with at-plate heat exchangers,
the convection-heat-transfer coefcient is higher for the two-dimensional corruga-
tions because of the presence of recirculation zones. The three-dimensional corruga-
tions cause the boundary of those zones to be broken, thus allowing uid particles to
cross between the recirculation regions and the main ow. This allows heat transfer
across what was a barrier to advection in the two-dimensional corrugated chan-
nel. The resulting enhanced mixing reduces the uid temperature nonuniformities,
thereby steepening the temperature gradient near the boundaries and increasing the
heat transfer between the uid and the channel walls.
Finally, for readers desiring either a gentle primer or more detailed treatment, the
following books offer a broad range of sophistication and readability on the topics
of nonlinear dynamical systems theory, chaos and chaotic advection: Gleick (1987),
Ottino (1989a), Baker and Gollub (1990), Kim and Stringer (1992), Moon (1992),
Ott (1993), Aref and Naschie (1995), and Nayfeh and Balachandran (1995).
CHAPTER TWELVE
Noise Reduction
The root of the matter is that the greatest stimulus of scientic discovery
are its practical applications.
(Lewis Fry Richardson, 18811953)
However far modern science and technics have fallen short of their
inherent possibilities, they have taught mankind at least one lesson:
Nothing is impossible.
(Lewis Mumford, 18951990)
PROLOGUE
This chapter is concerned with noise suppression. Noise in the context of this book
is undesired soundparticularly that generated by a uid ow. Although tangential
to the primary topics of this monograph, noise control is an important and broad
topic and deserves, and is addressed in, its own books. Hence, the treatment here is
rather cursory and consists simply of introducing the topic and relating noise control
to other ow control topics discussed elsewhere in this book. Particularly for cold
subsonic ows, small-scale turbulence uctuations and unsteady ow oscillations,
either in free-shear modes or interacting with solid surfaces, provide the primary
sources for the ow-induced sound energy. For hot supersonic ows, the interaction
of the turbulence large eddies with the ow is the dominant noise source. In either
case, therefore, controlling the owmodulates the sound eld favorably or adversely.
The goal is of course to reduce noise pollution either for human comfort, for military
advantage, or for the prevention of violent structural vibrations. Both passive and
active control strategies can be employed to suppress ow-induced noise. This goal
can be realized by reducing the vibrations of solid surfaces, eliminating or suppressing
turbulence, or cleverly manipulating certain owinstabilities and coherent structures.
12.1 Introduction
Sound ordinarily refers to audible pressure uctuations in the ambient air. However,
sound can propagate as well in liquids and solids, and its frequencies can be lower
250
(infrasonic waves below 20 Hz) or higher (ultrasonic waves above 20 kHz) than
those that can stimulate the human ear and brain to the sensation of hearing. Sound
can be a pleasant one like that emanating from a musical instrument in the hands
of a talented Homo sapiens. It can also be an unpleasant noise such as that radiated
from a jackhammer or an angry companion.
Sound can be generated by the vibrations of solid surfaces such as the strings
of violins and similar musical instruments or the diaphragms of loudspeakers. It
can also be generated by ow oscillations either directly or, more effectively, as a
result of interactions with solid surfaces. Unsteady addition or removal of heat can
also generate sound such as in singing ames. Noise is undesired sound, and the kind
generated as a result of uid owis of primary interest in this book. In 1952, Lighthill
focused on the sources of sound in the absence of vibrating surfaces and termed the
genre sound generated aerodynamically. The eld is now known as aeroacoustics
or, in water applications, hydroacoustics.
Audible sound levels vary over an enormous range, and therefore a relative log-
arithmic scale is typically used to express the sounds power, level of pressure uc-
tuations or intensity (energy ux or product of pressure and velocity perturbations).
Sound measured in decibels (dB) is computed from the relation: 10 log (X/Y), where
Xis either the sound power, the mean-square pressure uctuations, or the mean sound
intensity, and Y is a corresponding reference value typically related to the threshold
of hearing. More subjectively, sound can also be measured in phons, where a loud-
ness level of N phons is judged by the average ear to be as loud as a pure tone of
frequency 1 kHz at a sound-pressure level of NdB. There is also the sone scale, which
is a linear measure of loudness that is normalized so that 1 sone is a sound whose
loudness level is 40 phons. A sound of 10 sones is 10 times as loud as a sound of
1 sone, and the audible sounds lie in a range of 0100 sones. Finally, there is the
perceived noise decibel (PNdB), the A-weighted sound level (dBA), B-weighted, C-
weighted, . . . .
The envelope for man-made machines is continuously being pushed toward both
the miniature and the giant. As these machines become larger and more powerful,
their radiated, often unpleasant sound levels typically increase. Noise pollution is an
undesired byproduct of the present technological age, and efforts to minimize it must
always be considered in any eventual product. To place machine noise in perspective,
a human whisper involves acoustic power of roughly 10
10
watts (or 20 dB on the
relative logarithmic scale), a shout 10
5
w (70 dB), a jackhammer 1 w (120 dB),
a jet engine at takeoff 10
5
w (170 dB), and a sizable rocket at launch 10
7
w
(190 dB). Therefore, a single modern jet plane generates at takeoff more noise than the
combined shouting power of the entire worlds population. For humans, unprotected
exposure to sound levels above 100 dB for more than 15 minutes can cause hearing
damage, and permanent hearing loss results when exposed to sound levels above
110 dB. The threshold of pain is between 130 and 140 dB.
For some machines, like supersonic transport, noise control becomes one of the
limiting technologies. Out of concern for both air and noise pollution, the develop-
ment of commercial supersonic transport (SST) in the United States was summarily
halted in 1971 when the Congress voted to stop all further Federal funding for the
faster-than-sound aircraft. Though the Europeans and the Soviets proceeded, with
mixed success, to develop their own versions of SST, the United States has not yet
252 NOISE REDUCTION
recovered from that 30-year-old debacle. Noise control for jet engines remains a
formidable barrier to the successful development of supersonic as well as hypersonic
commercial aircraft.
Ultrasound is now used routinely in medicine to image tissues transparent to
X-rays and in nuclear power plants to test for cracks and structural integrity. More
relevant to this book, acoustic energy can be targeted toward certain ow regions
to control transition (Chapter 6) or separation (Chapters 8 and 9). But the focus
in this chapter is on attempting to reduce undesired ow-induced noise, not on the
use of sound to control the ow or to perform other useful tasks. As we attempt to
achieve other ow-control goals, however, the sound eld may be affected favorably
or adversely, and any unacceptable trade-off has to be scrutinized. Noise suppression
may of course be the primary control goal, and the effect of that on drag, lift, mixing,
and so forth, also has to be considered.
The main objective of this chapter is to introduce the topic of ow-induced noise
and to describe a variety of passive and active strategies to reduce noise pollution
either for human comfort, for military advantage, or for preventing violent structural
vibrations. Understanding the sources of noise is an essential step toward devising
ways to suppress the noise, and the chapter will start with a brief discussion of the
fundamentals of acoustics and sources of sound.
The following sections merely scratch the surface of the important eld of noise
control. Readers who desire more information can consult the numerous journal
articles available in the eld appearing in, for example, the Journal of Sound and
Vibration, Journal of the Acoustical Society of America, Archives of Acoustics Quar-
terly, and Journal of Vibration and Acoustics. A more gentle indoctrination in the
topic can be had, however, by reading the books by, among others, Dowling and
Ffowcs Williams (1983), Blake (1986), Bolton (1988), Wilson (1989), Foreman
(1990), Tokhi and Leitch (1992), Atassi (1993), and Fung (1994). The proceedings
and the two periodicals Noise/News and Noise Control Engineering Journal pub-
lished by the Institute of Noise Control Engineering of the United States of America
provide a wealth of information for the practically oriented. More fundamentally,
the Annual Review of Fluid Mechanics has published 11 articles related to aero-
and hydroacoustics in its rst 30 volumes. Finally, the pioneering articles on sound
generated aerodynamically authored by Sir James Lighthill can all conveniently be
found in volume III of his collected work (Hussaini 1997).
12.2 Fundamentals of Acoustics
Sound is pressure perturbations that are transmitted in a compressible mediumas suc-
cessive compressive and rarefaction waves. Sound waves are longitudinal, mechanical
waves; in other words, the material particles transmitting such a wave oscillate in
the direction of propagation of the wave itself, and the waves carry energy and thus
require a source of some sort. The elastic nature of the mediumcauses the vanishingly
weak pressure pulses to propagate while maintaining quasi-thermodynamic equilib-
rium adiabatically (i.e., isentropically) and essentially with no viscous losses. Sound
can, of course, propagate, reect, refract, scatter and dissipate in gases, liquids and
solids, but propagation through uids is the primary concern here.
12.2 FUNDAMENTALS OF ACOUSTICS 253
Acoustic disturbances are nearly always small, which allows linearization
of the governing equations.
1
For example, the root-mean-square of the pressure
perturbations at the threshold of hearing is 2 10
10
atmospheres, and at the
threshold of pain is 2 10
3
atm. Consider a still uid having uniform pressure p
o
and density
o
. Sound perturbs this eld such that
p(x, t) = p
o
+ p
(x, t) (12.1)
(x, t) =
o
+
(x, t) (12.2)
u
k
(x, t) = 0 +v
k
(x, t) (12.3)
where v
k
(x, t) is the minute uid particle velocity due to the passage of sound waves.
It can easily be shown (e.g., Moran and Shapiro 1995) that the rate of propagation
of an innitesimal pressure pulse, the so-called speed of sound, in an otherwise still
substance is given by
a
2
o
=

o
o

o
=
(12.4)
where
o
is the ratio of specic heats, and
o
is the isothermal compressibility coef-
cient (dened in Chapter 2), both of which are computed at the uniform ambient
conditions. The derivative p/ is computed at constant entropy. Because the en-
tropy is constant, the pressure and all other thermodynamic properties can be con-
sidered as a function of a single variable; for example, p = p() only. In other words,
sound is a disturbance in which the pressure can be determined from a knowledge of
the density alone. The sound speed is itself a thermodynamic property whose value
depends on the state of the medium through which sound propagates. For an ideal
gas,
o
= 1/p
o
and therefore a
o
=
o
RT
o
, and the sound speed depends only on
the gas type and its temperature. Owing to the much higher pressures required for
liquids to change their volumes, the speed of sound in liquids is generally higher than
in gases and higher yet than in solids. For a bubbly liquid, however, the speed of
sound can be much lower than in either the gas or liquid alone.
We now restrict the discussion to uids only. To develop the equation govern-
ing the propagation of sound waves, we start with the exact conservation relations
for a continuum uid (Chapter 2). In a Eulerian frame, the continuity equation in
differential form reads
t
+

x
k
(u
k
) = 0 (12.5)
and Newtons second law reads
t
(u
k
) +

x
i
(u
i
u
k
) =

ik
x
i
+g
k
(12.6)
1
Each of the simplifying features implemented here can of course be violated under certain unusual
circumstances. For example, viscous effects become important when sound propagates over very long
distances, and sound levels of 194 dB generate a not-too-small pressure uctuations of 1 atm. There
is also the issue of sound generated aerodynamically being strong enough to back-react with the uid
motions producing it. Nonlinear acoustics and other unusual effects are beyond the scope of this chapter.
254 NOISE REDUCTION
where is the uid density, u
k
is an instantaneous velocity component,
ik
is the
stress tensor (surface force per unit area), and g
k
is the body force per unit mass.
The stress tensor can be split into two parts, hydrostatic pressure and normal and
tangential viscous stresses
ik
= p
i k
+
i k
(12.7)
For the weak acoustic perturbations, we linearize Eqs. (12.5) and (12.6) and
neglect viscous and gravity forces.
2
The resulting mass and momentum conservation
equations read, respectively,
t
+
o
v
k
x
k
= 0 (12.8)
o
v
k
t
+
p
x
k
= 0 (12.9)
If the velocity perturbation is eliminated by taking /x
k
of Eq. (12.9) and the result
is subtracted from /t of Eq. (12.8),
t
2

2
p
x
k
x
k
= 0 (12.10)
and Eq. (12.4) is used to eliminate the density and to yield nally the wave equation
for pressure perturbations
2
p
t
2
a
2
o
2
p
x
k
x
k
= 0 (12.11)
This is a hyperbolic partial differential equation that describes the propagationat
the constant speed of sound a
o
of pressure perturbations in an otherwise still uid.
Identical equations can be written for all the other small quantities describing the
acoustic eld; for example, density perturbations (
=p
/a
2
o
), velocity v, and velocity
potential ( p
=
o

t
).
3
12.3 Sources of Sound
Sound is small pressure perturbations organized into wavessound waves. If we
insist that sound must not anticipate its cause or in other words that information
about the current source activity should not be contained in past waves that anticipate
the present,
4
the homogenous equation (12.11) in the absence of boundaries has
2
The relevant Reynolds number is 2a
o
/, and is of the order of 10
8
in air at the most audible frequency
but is much higher in water. Hence, viscous forces can be safely neglected for sound propagation over
distances much less than 2a
o
/ wavelengths. The relevant Froude number square is a
2
o
/g, and
therefore gravity is neglected for sound waves with wavelength 12 km (in air) or 200 km (in water),
which is a condition that is readily satised for most waves.
3
For one-dimensional sound waves, the particle velocity is given by v = p
/
o
a
o
, and v satises the
one-dimensional version of the wave equation.
4
This important principle is called the causality condition and in simple elds is indistinguishable from
the radiation condition, which states that in open space, only outward-traveling waves can have any
real existence.
12.3 SOURCES OF SOUND 255
only one solution: p
(x, t) = 0 everywhere. In other words, complete silence is the

only homogeneous sound eld in unbounded space. Obviously, sound must derive
its energy from a source of some sort. Mathematically, a nonzero source term on
the right-hand side of Eq. (12.11) makes the differential equation nonhomogenous
and can force nontrivial solutions. But what are some of those sources? A sound
source is a forcing term in the wave equation and occurs when the criteria leading
to the homogenous equation are violated in some way. Having a source term in
either the continuity or momentum equation, breaking down the state equation p =
p(), having a body in relative motion with a uid, or having velocity uctuations
sufciently strong to break down the linearization criterion all violate the essence of
the homogenous wave equation: an expression of the conservation laws in a weakly
disturbed homogenous material in which p = p() only.
Vibrating solid surfaces, machinery containing rotating blades, unsteady free-
shear or wall-bounded ows, and unsteady expansion or contraction of uid due, for
example, to unsteady heating or cooling are all possible sources of sound. Determin-
ing precisely the source(s) of a particular sound is not, however, an easy problem
least of all because a particular source provides a unique sound eld, but the converse
is certainly not true (Dowling and Ffowcs Williams 1983). Analyzing a sound eld
with the most sophisticated instruments cannot tell us the precise nature of its source;
this problem has no unique solution!
What are the mechanisms that cause the kinetic energy associated with the motion
of a uid or a solid to be converted to acoustic energy? Having a lot of pressure uc-
tuations in a low-speed, free-shear turbulent ow, for example, does not necessarily
result in a strong sound eld. In fact, a minuscule portion of the kinetic energy associ-
ated with the turbulence uctuations is converted to acoustic energy and radiated as
sound. Lighthill (1962) termed the pressure uctuations in an incompressible turbu-
lent owpseudosound. These uctuations are induced directly by uctuating vortical
motions in the owproper. Though we can hear pseudosound via a suitably placed
microphone, stethoscope, or even our own ears, these pressure uctuations lack the
essential quality of propagation at or around the characteristic sound speed. The
turbulent eddies are convected with the owitself, and within themthe pressure uc-
tuations are balanced, for the most part, by uctuations in uid acceleration. Without
the benet of a surface that may vibrate in reaction to the unsteady pressure forces, a
very small percentage of the pressure uctuations radiates as pressure waves, satises
the wave equation, and is responsible for what is termed aerodynamic soundreal
sound. The radiated sound falls off as the inverse rst power of the distance, and
thus in the far eld it dominates over the pseudosound, which falls off at least as
the inverse square of the distance. But how does one estimate the true sound as a
small percentage of the pseudosound? This and other questions posed in this section
were not answered satisfactorily until the pioneering work of Lighthill published
in 1952 in which he introduced a rst-principle theory for sound generated aero-
dynamically.
Having an unsteady source of mass () that forces the amount of matter in a xed
region in space to uctuate can generate sound effectively. This is the loudly efcient
mechanism by which puffs of steam or compressed air generate sound in a uidic
siren. Less effectively, a bells surface, when struck, generates sound by adding puffs
of momentum to the adjacent air. One can think of the siren as being replaced by
a hypothetical distribution of point mass singularities, or monopole sound sources.
The vibrating surface introduces no new uid and is equivalent to a distribution of
256 NOISE REDUCTION
dipoles, forcing the rates of momentum (i.e., mass ux u
k
) in a xed region in space
to uctuate. A point dipole involves a neighboring pair of equal point monopoles
with opposite signs. Though the instantaneous sum of the two sources is zero, the
net-sound eld of a dipole is nonzero only because the sources at innitesimally dif-
ferent positions radiate sound to the point (x, t) at slightly different retarded time. A
dipole has both a magnitude and direction, and the destructive interference or can-
cellation involved makes it a less efcient radiator of sound than the omnidirectional
monopole. The nal sound source in this hierarchy is a distribution of quadrupoles
that forces the rates of momentum ux (u
i
u
k
) across xed surfaces to uctuate.
A quadrupole consists of two positive monopoles and two negative ones and thus
it introduces neither new mass nor new momentum. No external force acts on the
turbulent uid because internal actions and reactions balance. In this case, the total
dipole strength is zero, and the effects of different hypothetical dipoles nearly, but
not quite, cancel out at large distances. A quadrupole is the least efcient source of
the triad because it involves a double tendency for source elements to cancel and
is the pathetically ineffective mechanism by which sound is generated in subsonic,
free-shear turbulent ows.
To answer the question of how sound is generated aerodynamically, Lighthill
(1952) considered the oweld to occupy a small part of a very large volume of
uid of which the remainder is at rest (Figure 12.1). The portion that contains the
intense velocity uctuations provides the source region outside of which the wave
equation (12.11) is valid. To illustrate Lighthills acoustic analogy, we seek the differ-
ence between the exact statements of the natural lawsvalid in the owregionand
their acoustical approximationsvalid in the far eld or the still uid region. Then
the equations governing the uctuations of density in the real uid are compared
with those that would be appropriate to a uniform acoustic medium at rest. The dif-
ference between the two sets of equations is considered as a hypothetical distribution
of sound sources.
Figure 12.1: Sound emitted at y is heard by the observer at x at a time
|x y|
ao
after
emission.
12.3 SOURCES OF SOUND 257
We start with the exact expressions for the conservation of mass and momentum,
the latter without body forces. Taking /x
k
of Eq. (12.6) and subtracting the result
from /t of Eq. (12.5), we get
t
2
=

2
t
2
=

2
x
i
x
k
(u
i
u
k
ik
) (12.12)
Subtracting
a
2
o
x
k
x
k
=

2
x
i
x
k
a
2
o
i k
(12.13)
from both sides of Eq. (12.12), we reach the desired inhomogenous equation for
sound
t
2
a
2
o
x
k
x
k
=

2
T
ik
x
i
x
k
(12.14)
where T
ik
=u
i
u
k
i k
+ p
i k
a
2
o
i k
, the Lighthill stress tensor, and we have used
Eq. (12.7).
In effect, we have rearranged the governing equations to isolate a linear wave
operator, the remaining terms being treated as a source that drives the wave motion.
The source eld for the acoustic waves is a quadrupole distribution whose instan-
taneous strength per unit volume is the tensor T
ik
. The double divergence structure
of the right-hand side of Eq. (12.14) indicates a double tendency for destructive in-
terference and reveals the quadrupole nature of the source term. Contributions to
the instantaneous Lighthill stress tensor come from normal and tangential Reynolds
stresses, normal and tangential viscous stresses, and the difference between the ac-
tual hydrostatic pressure in the active ow region and that which would prevail in a
uniform acoustic medium at rest. In the sound eld proper, p
a
2
o

, the weak ow
is inviscid and the Reynolds stress is second order in the small-velocity uctuations.
In other words, T
ik
0 outside the active ow region. Inside though, strong heat-
ing or cooling, pressure forces on a rigid body moving relative to the ow, viscous
stresses, and turbulent uctuations can all be sources of aerodynamic sound. The
density uctuations in the real ow are exactly those that would occur in a uniform
acoustic medium subject to an external stress system given by the difference between
the effective stresses in the real ow and the stresses in the uniform acoustic medium
at rest.
Consider an observer located in the far eld at x, the solution of Eq. (12.14) reads
(x, t) =

2
x
i
x
k
V
T
ik
y, t
|xy|
a
o
4a
2
o
|x y|
dV (12.15)
The integration is carried out for all sources located in the ow region at y. Every
quadrupole element at y generates a eld that travels out at the speed of sound,
reaching the observer at x at a time |x y|/a
o
later (Figure 12.1). The effect of each
source element falls off inversely as the distance traveled |x y|.
The integration above can in principle be carried out if the Lighthill stress tensor
T
ik
is known. For incompressible turbulent ows, this tensor is dominated by the
258 NOISE REDUCTION
instantaneous Reynolds stress u
i
u
k
, a random quantity that in general is unknown
analytically and is very difcult to measure. Therefore, heuristic models, based on
a combination of experimental data, computations, and theoretical concepts, or
simple dimensional analysis is used to estimate the sound eld. For example, Lighthill
(1952, 1954) estimated the mean-square value of the radiated sound eld froma cold
subsonic jet to be
2

2
o
Ma
8
D
2
|x|
2
, Ma 1 (12.16)
where Ma is the Mach number based on the jet-exit velocity; the ambient sound speed,
D, is its diameter; and |x| is the distance between the jet exit and the observer. The
small Mach number makes it possible to assume a compact source (D Ma ,
where is the characteristic wavelength of the generated sound) and greatly simplies
the dimensional analysis. Basically, for a compact source, the variation of retarded
time as y ranges over all the source volume in the integral (12.15) is ignored, in effect
making the source distribution equivalent to a point quadrupole.
Mercifully, less than 0.01%of the turbulence kinetic energy is converted to acous-
tic energy, and this percentage diminishes as the Mach number drops further from
unity.
5
At low Mach numbers, jets as well as all compact sources are acoustically
very inefcient, and
2
Ma
8
. The presence of a rigid surface, of much higher den-
sity than the surrounding uid,
6
in the ow proper is equivalent to a distribution of
dipoles and increases the sound-generation efciency so that
2
Ma
6
. For exible
bodies capable of dilating, the resulting monopole singularities lead to an even more
efcient sound generation, and
2
Ma
4
. And nally, for compressible turbulent
jets, the source region can no longer be considered compact, the structure of the
sound eld is dominated by retarded time variations over the source region, and the
number of eddies that can be heard at any one time increases linearly with the Mach
number. In this case, the mean-square value of the radiated sound eld is estimated
to be a much higher
2
Ma
3
. These dimensional analysis results are well conrmed
with experiments.
The Lighthill acoustic analogy provides reasonable predictions for subsonic free-
shear ows and for noise mechanisms associated with the nearly steady aerodynamic
loading of turbomachinery blades. The alternative stochastic wave model more re-
cently developed by Tam (1995) allows direct calculations of sound radiation with-
out recourse to the acoustic analogy. The semiempirical stochastic model theory has
been successful in predicting jet noise when Mach waves are radiated (as a result
of the supersonic convection speeds of large-eddy structures) or when broadband
shock-associated noise results from the passage of large-scale structures through the
shock-cell pattern in imperfectly expanded supersonic jets. We will return to su-
personic jet noise mechanisms in the next section where, armed with knowledge of
aerodynamic sound sources, we discuss methods to suppress ow-induced noise for
both low-speed and high-speed ows.
5
The acoustic efciency is the ratio of acoustic power radiated to jet power delivered. For subsonic jets,
10
4
Ma
5
.
6
Nekton, having about the same density as their surrounding uid, swim in water outstandingly more
quietly than the heavier-than-air avian species can y in air.
12.4 NOISE CONTROL 259
12.4 Noise Control
Noise suppression involves the reduction of noise sources efciency and ability to
convert kinetic energy to sound power, the interruption of sound transmission, the
accelerated dissipation of acoustic energy into heat, or the active cancellation of
sound waves using out-of-phase waves. These strategies can be implemented by a
variety of passive, active, and reactive devices ranging in complexity from simple ear
plugs to sophisticated antisound systems.
Unhealthy noise pollution is easier to quantify than what is considered a mere
annoyance. Unprotected exposure to sound levels above 110 dB can result in per-
manent hearing loss. Sound that merely irritates, on the other hand, is obviously a
subjective matter at any level. For most people, save teenagers, sound levels above
90 dB are uncomfortable and distracting. Noise containing discernible pure tones
is, however, more likely to invoke a negative response than broadband noise of the
same level. The screech tones radiating from imperfectly expanded supersonic jets
can be maddening.
Noise control, a relatively young eld of research, is required to maintain an ac-
ceptable or even pleasant environment, to avoid detection of certain aquatic vessels,
to facilitate the proper operation of sonars on underwater vehicles, and nally to pre-
vent violent structural vibrations. Heckl (1988) broadly classied noise suppression
methods into four categories:
Commonsense application: speed reduction, load reduction, increase of mass,

general purpose enclosures, mufers, screens, earmuffs.
Add-on measures: multiple walls and oors, including anking transmission,

tuned absorbers and vibration coatings, elastic foundations, impedance mis-
matches in the propagation paths, and shift of resonances away from exciting
frequencies.
Changes at the source: reduction of time derivatives, for example, in hammers,

punch presses, exhausts, pumps, and valves; interruption of feedback loops or
other instabilities (wheel squeal caused by stick-slip, self-excitation in burners,
whistles); reduction of roughness and free clearances (rolling noise, piston
slap).
Active methods of control: sound reection and absorption in waveguides

(ducts, beams), helmets with active noise control, interruption of feedback
loops or other instabilities such as in burners, compensation of periodic exci-
tation.
In the next six subsections, we will very briey discuss several aspects of the art
and science of noise suppression. Once again, we will merely scratch the surface, and
the interested reader is encouraged to consult the books and journals listed at the
end of Section 12.1.
12.4.1 Relation to Other Flow-Control Goals
We have established in Section 12.3 that turbulence is a major source of ow-
induced noiseat least for cold subsonic ows. It is logical, therefore, that tur-
bulence suppression, particularly at the noise-inducing scales, reduces the levels of
radiated sound. Delaying laminar-to-turbulence transition, suppressing turbulence
particularly its tangential Reynolds stressesor, in the extreme, relaminarization can
260 NOISE REDUCTION
each help to reduce noise levels. Controlling the mixing processes or targeting par-
ticular coherent structures can also modulate the sound eld favorably or adversely.
Strategies to achieve any of these goals are thoroughly discussed throughout this
book. If turbulence is to be enhanced, for example, to prevent ow separation, the
adverse effect of that control on noise levels has to be scrutinized. Conversely, if noise
is to be controlled by suppressing turbulence, the trade-off in terms of drag or lift
penalty has to be considered carefully. Occasionally, suppressing noise simultaneously
leads to additional desired ow traits. For example, delaying laminar-to-turbulence
transition leads to lower noise levels as well as lower skin-friction darg. As a second
example, using antisound to reduce combustion-generated noise can also eliminate
undesired ow instabilities. Lastly, for subsonic turbulent ows, weakening the ne-
scale structures is benecial to noise reduction, whereas breaking up the large eddies
in supersonic ows leads to lower noise intensities.
As seen in Section 12.3, turbulence generates sound equivalently to a quadrupole
source distribution. Organized structures in a turbulent ow play an important role
in the source process and, therefore, controlling these eddies may be the key to noise
suppression (Ffowcs Williams 1977). At the lowMach number typically encountered
in underwater applications (O[0.01]), unbounded turbulent ows have an extremely
lowacoustic efciency according to Lighthills (1952) theory. The ne-scale eddies are
responsible for most of the radiated sound. The large differences between the noise
levels predicted by the quadrupole theory and those actually generated by a turbulent
boundary layer are often explained via acoustic scattering processes (Crighton 1984).
Small rigid bodies like ribs, rivet heads, edges and angles, gas bubbles, or sharp edges
(Blake and Gershfeld 1989) can destroy the cancellation between opposing elements
of the quadrupole and may lead to the more efcient radiation of a dipole or even a
monopole. Thus, to reduce noise, these efcient sound generators should be avoided
as much as practical.
A navys ability to conceal its submarines or to have reliable underwater commu-
nications is a crucial military advantage. In contrast to air, water transmits sound
waves far better than it does electromagnetic waves like visible light or radio waves.
Therefore, sound is used extensively for underwater communications. Sonar in water
plays a similar role to radar in air. For water vessels, sonars are used for navigation
and detection. There are at least two military advantages to suppressing the noise
sources of aquatic vessels. First, noise radiated from a submarine, say, as a result of
its hull vibrations, its power plant, its propulsion system, or any other sound source
makes the vehicle less quiet and hence more susceptible to detection by enemy sonars.
Second, if a sonars transmission and receiving elements are buried beneath a turbu-
lent boundary layer, the hydrophones receive the pseudosound resulting from the
pressure uctuations in the turbulent ow along with any legitimate incoming sig-
nal. The signal-to-noise ratio is lowered in this case, and the sonars ability to detect
enemy vessels and incoming torpedoes as well as to receive friendly communications
is hampered as a consequence.
Mostly because of the latter problem, sonars are typically placed at the nose
of submarines where the boundary layer is laminar and the pseudosound minimal.
Search rates of these listening devices can be drastically improved by placing addi-
tional hydrophones farther downstream. However, the turbulence-induced noise is
sufcient to interfere with the proper operations of these devices, and noise must
therefore be suppressed before the search rate can be improved. A second example
of the undesired side effects of turbulence comes from the aeronautics eld. On
commercial aircraft, the turbulent boundary layer outside the fuselage contributes
signicantly to cabin noise. Techniques to reduce ow-induced noise are therefore
sought. Methods to delay transition (Chapter 6) or to relaminarize an already tur-
bulent boundary layer (Chapter 10) should be useful to suppress the turbulence-
induced noise in both of the preceding examples. It is feasible to delay transition on
both vehicles, where the unit Re is about 10
7
/m, for the rst few meters. Beyond
that, relaminarization or at least partial suppression of turbulence is the only option.
As indicated in Section 10.5, achieving relaminarization over the entire length of
an aircraft or a submarine may require large energy expenditure and is, therefore,
not very practical. However, it may not be necessary to relaminarize an entire ve-
hicle. Instead, it may be more feasible to target selected portions or even selected
coherent structures of the boundary layer where, for example, a hydrophone is to be
placed.
12.4.2 Passive Noise Control
Passive noise control devices include shields, rigid and compliant walls, mufers,
silencers, resonators, and absorbent materials. These can be placed around a noisy
machine or used along an exhaust pipe. They work by reducing the acoustic energy
transmitted to the environment either by reecting it back or by dissipating part of
the energy into heat. Radiated sound is also reduced when vibrations are suppressed
using, for example, a properly designed foundation to isolate a machine from a
concrete oor or an energy-absorbent viscoelastic liner to wrap around an entire
machine or vehicle.
When a solid wall interrupts sound propagation in air, part of the acoustic energy
is reected, part is absorbed, and part is transmitted. The sound pressure incident
on the wall causes the solid to vibrate, and the fraction of incident sound energy
transmitted depends on the relative impedance (product of density and sound speed)
of the wall and uid. A wall constructed of dense material tends to be most effective
in reducing noise transmission; the major part of the incident acoustic energy is ei-
ther reected or dissipated into heat owing to impedance mismatch, absorption, and
damping. Generally, passive sound suppression devices work best to shield from,
and to discourage, the transmission of high-frequency (short-wavelength) sound
waves.
To illustrate this last statement, consider a planar sound wave incident perpendic-
ular to a rigid plane wall of mass per unit area m. The uid on either side of the wall
has the same density
o
and sound speed a
o
, and the incident sound radian frequency
is . The wall material has higher density and sound speed than the uid. The aim is
to compute the portion of the incident acoustic energy that would be reected and
the portion that would be transmitted through to the other side of the wall. The
boundary conditions are continuity of the uid and solid particle velocities on either
side of the wall and a pressure difference across the wall that provides the force nec-
essary to accelerate a unit area of the wall material. The particle velocity condition
on the incident side of the wall immediately implies that both the transmitted and
reected waves have the same frequency as the incident wave but not the same wave-
length. The velocity and pressure boundary conditions can easily be used together
with the conventional relations for one-dimensional waves between pressure, particle
262 NOISE REDUCTION
velocity, and acoustic energy to give the following expression for the energy trans-
mission coefcient ETC (see, for example, Dowling and Ffowcs Williams 1983):
Energy transmission coefcient
Energy transmitted through the surface
Energy of the incident beam
ETC =
4
2
o
a
2
o
4
2
o
a
2
o
+
2
m
2
(12.17)
It is clear that very little high-frequency sound is transmitted through solid walls;
most of it is reected. In contrast, low-frequency waves travel through walls with
very little attenuation. The proper dimensionless parameter is
m
o
a
o
1 =sound waves are mostly reected
m
o
a
o
1 =sound waves are mostly transmitted
Low-frequency sound gets through all but the most massive walls.
Sound-absorbent coatings are made either of porous material or perforated ma-
terial and are often used to dissipate acoustic energy. In the case of porous material
like foam or expanded polystyrene, the passage of sound waves causes the air in-
side the pores to vibrate. The extremely narrow channels cause a large resistance
to the uid motion and the dissipation of sound energy into heat. Additional en-
ergy losses result from the heat exchange between the heated, compressed air or the
cooled, rareed air inside the pores and the solid skeleton. Perforated materials typ-
ically consist of a porous surface glued to a honeycomb structure and are designed
to create beer-bottle-like geometries (i.e., numerous Helmholtz resonators). There,
resonant-frequency pressure perturbations at the neck produce large velocities in the
resonator. Again, large velocity in a narrow passage results in considerable viscous
dissipation and reduction of acoustic energy. The depth of the perforations and the
volume of the backing cavities can be optimized to resonate in the middle of the
frequency range, where sound attenuation is desired.
Mufers are typically used to quiet exhaust pipes in internal combustion engines.
A mufer usually consists of inlet and outlet pipes of equal diameters and an expan-
sion chamber between them. The area ratio and the length of the expansion chamber
determine the mufer transmission loss, and the chamber length is tuned so that
maximum attenuation occurs at the dominant frequencies of the incoming sound.
Silencers are similar devices used in ducts or uid intakes in industrial applications.
The sound energy in mufers and silencers is conserved. Any reduction in the trans-
mitted wave energy must be coupled to a corresponding increase in the reected
energy. Both passive devices typically employ a combination of sound-absorbent
materials designed to convert the high-frequency component of the noise into me-
chanical vibration and heat and destructive interference between the transmitted and
reected waves in the expansion chamber for the lower range of frequencies. The
sound-absorbent material results in a net reduction in acoustic energy. Mufers and
silencers can also utilize Helmholtz resonators attached to the side of the duct and
optimized to suppress noise in a narrow range of frequencies.
12.4.3 Compliant Coatings
Fluidstructure interaction is particularly important in underwater noise radiation
owing to the relatively high density and bulk modulus of water. In contrast to a rigid
wall, a compliant wall is sufciently soft to deform elastically as a result of the
pressure and viscous forces of a moving uid. From a fundamental viewpoint, a rich
variety of uidstructure interactions exist when a uid ows over a surface that
can comply with the ow. The transverse wave speed in the solid (square root of the
ratio of the wall modulus of rigidity and its density) relative to a characteristic ow
speed determines the degree of compliance of the wall; a value below one indicates
a compliant wall. Not surprisingly, instability modes proliferate when two wave-
bearing media are coupled. Some waves are ow-based, some are wall-based, and
some are a result of the coalescence of both kind of waves. A compliant coating can
in principle be designed to achieve desired ow traits such as transition delay, noise
suppression, and lower skin-friction drag in turbulent ow. The topic is thoroughly
discussed in Chapter 7.
A compliant coating that causes transition delay or reduction in turbulence skin
friction will also lead to attenuation of sound radiated by the boundary layer. This
is because the wall-pressure uctuations are dependent on turbulence levels, which
in turn are related to the wall shear stress. The reverse is not necessarily true (i.e., a
coating may attenuate the ow noise without affecting the hydrodynamic drag). In
fact, the technology exists today for manufacturing energy-absorbent compliant lin-
ers for sound absorption, vibration reduction, and noise shielding, whereas the search
for drag-reducing coatingparticularly for turbulent owshas thus far eluded re-
searchers for about 50 years (Riley et al. 1988; Gad-el-Hak 1996a). As mentioned
earlier, a submarine hull vibrates as a result of the surface forces in the surround-
ing ow or as a result of vibrations in other parts of the vessel such as its power
plant. The hull vibrations are an effective source of radiated sound and can readily
be detected by sonars far away. A properly designed compliant coating can act as
an effective attenuator of hull vibrations and therefore reduce the likelihood of det-
ection.
Flow noise is inuenced by surface exibility through two distinct mechanisms:
either the surface acts as a sounding board excited by the turbulent pressure eld
or the surface compliance induces a change in the turbulence structure and, hence,
modies the pressure uctuations (Ffowcs Williams 1965; Purshouse 1976; Dowling
1983, 1986). As mentioned earlier, ow-induced noise is currently considered the
limiting performance factor for sonar systems placed on surface ships, submarines,
and towed arrays. Major advances in the reduction of self-noise have been achieved
by exploiting the rst mechanism above, and further reductions may be possible if
the nature of the turbulent boundary layer and its wall-pressure uctuations can be
altered. Von Winkle (1961) and Barger and von Winkle (1961) reported pressure
uctuation measurements on a streamlined body of revolution free-falling in a wa-
ter tank. Flush-mounted hydrophones fabricated from lead-zirconate titanite were
used to measure the instantaneous pressure. Their experiment indicated a dramatic
reduction in the pressure coefcient when the body was covered with a Kramer-
type compliant coating (see Chapter 7). This result may, however, have been due to
transition delay caused by the exible surface and not the changes in the turbulence
structure.
264 NOISE REDUCTION
12.4.4 Jet Noise
The quest to power commercial aircraft with ever more powerful jet engines
has, in large measure, inuenced noise suppression research for the last half century.
Inspecting Eq. (12.16) reveals the strong dependence of the noise levels from cold
subsonic jets on the Mach number and a weaker dependence on the jet diameter. This
leads to what is by far the most powerful strategy to reduce jet noise: by reducing
the jet velocity with a reciprocal increase in diameter to maintain a given thrust, the
resulting noise reduction is proportional to the sixth power of velocity. The acoustic
efciency is the ratio of acoustic power radiated to jet power delivered and varies as
Ma
5
. By means of the downward trend in Mach number, jet engines of ever larger
diameter have permitted an increase in engine power to be delivered at community-
acceptable noise levels. Even at a prescribed thrust, there are of course weight and
drag penalties when larger engines are utilized, and these have to be weighed carefully
against the benet (or necessity) of lower noise levels.
Turbojet engines with bypass reconcile the need to achieve high Carnot efciency
while maintaining a good Froude efciency, so to speak. Jet engines were designed
as early as the 1950s with a 40% bypass ratio. Part of the energy of the gas coming
out of the combustion chamber was used to impel a surrounding additional 40% of
the air mass, which bypassed the combustion process altogether. For a given thrust
and engine efciency, the sound energy was reduced at this bypass ratio by a factor
of 5 (7 dB). For even more gain, higher bypass ratios are sought in modern designs.
Though they work in practice, there are some questions about the exact reasons
why high-bypass-ratio engines have less noise than conventional engines. There is
a reduction in the global Mach number of the jet and a concomitant reduction in
radiated noise. Additionally, bypass air reduces the shear magnitude between the
high-speed primary air and the freestream air. This in turn reduces the turbulence
energy production and thus radiated sound. Because of the difference in tempera-
tures of the hot primary air, the cooler bypass air, and the cold freestream, there
is, however, a strong shear in Mach number (but not velocity) between the primary
and bypass air and a strong shear in velocity (but not Mach number) between the
bypass and freestream air. It appears that the Mach number is more critical to the
sound generation, but the detailed measurements have not been done yet to pinpoint
the sources of the sound. Therefore, at present the improvements in sound levels are
semiempirical without knowledge of the precise causes.
There are two other techniques to reduce noise from subsonic jets, often utilized
simultaneously. Neither strategy is suited for high-bypass-ratio engines. The use of
either a multinozzle noise suppressor (e.g., Pratt & Whitney engines powering some
Boeing 707s) or deeply corrugated jet exit (e.g., RollsRoyce engines powering some
Boeing 707s) benets from both the so-called shielding principle and a reduction
of the quadrupole intensity. The shielding principle (Lighthill 1962) states that if
an observer is listening to a jet noise and is placed near the peak of its directional
distribution, then the presence of another jet behind the rst, shielded fromhis or her
line of direct observation, does not signicantly increase the noise heard. The reason
is the scattering of the far jet noise by the near jet ow, which causes a directional
redistribution of the noise from the far jet and shifts its direction of peak intensity.
In effect, the far jet sound is heard at a directional average, not at a peak value.
In addition to the shielding effect, both the multinozzle jet and corrugated-exit jet
result in a shorter extent of the mixing layer, bringing the adjustment region where
turbulence intensities begin to decrease closer to the orice, and a reduced relative
shear between the jet and the ambient air, which suppresses the turbulence intensities
in the mixing region. These advantages effectively reduce the quadrupole intensity
and thus the radiating sound. The total noise reduction in both the Pratt & Whitney
and RollsRoyce engines is around 10 dB, whereas the weight, thrust, and drag
penalties resulting from the special noise-suppression designs are each around
0.7%. These penalties reduce the aircraft height above a typical community-noise
measuring point, adversely affecting the radiated noise by less than 1 dB.
Both ne-scale and large-scale coherent structures in turbulent ows are capable
of generating noise. However, the relative importance of the noise they produce
depends on the jet Mach number and uid temperature. Noise from cold subsonic
jets is caused primarily by the small-scale structures. Unless the jet is very hot,
7
the
turbulence convection Mach number, relative to the ambient sound speed, is below
one, and the large eddies are ineffective noise generators. In contrast to subsonic jets,
large-scale structures are the primary source of noise in hot supersonic jets. In this
case, the large turbulence structures convect downstreamat supersonic Mach number
relative to the ambient sound speed. The supersonic eddies are capable of producing
intense Mach wave radiation, which easily predominates over the noise from ne-
scale turbulence. The resulting broadband sound is called turbulence mixing noise,
and it radiates in the downstream direction. Tams (1995) stochastic instability wave
model treats the large coherent structures as being statistically equivalent to the
instability waves of the jet. These waves in turn are considered as a wavy wall
having the same wavelength and wave speed as the instability waveabove which the
supersonic ow convects and generates intense Mach (sound) waves that propagate
in a preferred downstream direction. The Mach angle can be computed readily from
the phase velocity c
p
of the most amplied instability wave and the ambient sound
speed a
o
: = sec
1
Ma
c
= sec
1
c
p
a
o
, where Ma
c
is the convective Math number.
Most supersonic jets are imperfectly expanded, and this leads to the formation
of a pattern of oblique shocks or expansion fans and two additional mechanisms
for noise generation. For an underexpanded jet, an expansion fan is initiated at the
nozzle lip to allow the gradual reduction of static pressure to ambient conditions.
For an overexpanded jet, an oblique shock is formed at the lip to allow the abrupt
increase in static pressure to match that outside the jet. Once formed, the expansion
fan or shock wave is convected downstream until impinged on the mixing layer on
the other side. The subsonic ow outside the jet cannot support either a shock or an
expansion fan, and hence both have to reect back into the jet plume. This process is
repeated many times, forming a quasi-periodic pattern of shocks or expansions fans.
The shock-cell pattern may be regarded as disturbances trapped inside the jet by the
mixing layer at the peripheral. The jet owbehaves as a waveguide to the disturbances
forming the shock pattern, which in turn leads to the two shock-associated sources
of noise described below.
Both broadband shock noise and discrete screech tones result from the passage
of large-scale structures through the shock-cell pattern. In contrast to the turbulence
mixing noise, the two shock-associated sounds radiate primarily in the upstream
7
The jet Mach number based on the local sound speed in the hot uid is lower than the acoustically more
relevant Mach number based on the sound speed in the cold ambient uid.
266 NOISE REDUCTION
direction. The broadband shock-associated noise is produced as a result of the con-
structive scattering of the large eddies by the shock cells in the jet plume. Because
the quasi-periodic shock pattern can be conceived as made up of a superposition of
waveguide modes of different wavelengths, each mode scatters sound in a preferred
(upstream) direction. The screech noise is due to an acoustic feedback phenomenon.
Here acoustic disturbances impinging on the nozzle lipwhere the mixing layer is
thin and receptive to external excitationexcite the intrinsic instability modes of
the jet. The resulting waves convect downstream, gain strength, and interact with
the aforementioned shock-cell pattern. The unsteady interaction generates acoustic
waves that propagate upstream outside the jet and upon reaching the lip region close
the feedback loop. The resulting sound is intense at a particular fundamental fre-
quency, the screech tone frequency, and can sometimes be heard up to the fourth or
fth harmonic.
Noise radiated from supersonic jet engines cannot be reduced by simply increas-
ing their size and thus reducing their Mach number, as done in subsonic jets. Extra-
wide supersonic engines would generate unacceptable shock-wave drag as well as
supersonic-boomemissions. Systems to suppress noise in supersonic jets are generally
more complex than their subsonic counterparts, making noise control a formidable
barrier to the successful development of supersonic as well as hypersonic commercial
aircraft. Improved mixing in supersonic jets breaks up their large-scale structures and
has invariably the benecial side effect of substantial noise reduction. Enhanced mix-
ing devices include ejectors, taps, acoustically treated mixers and ejectors, and uncon-
ventional nozzle geometries. Rectangular and triangular jets radiate less noise than
circular jets at high subsonic and supersonic speeds (Ahuja, Hayden, and Entrekin
1993; Rice and Raman 1993). Jet engine exhausts are scalloped to reduce the scale
of their coherent structures and hence reduce the radiated noise. An active strategy to
accomplish a similar feat involves forcing the jet acoustically to cause the natural train
of unequally spaced vortex rings to become more uniform. The formation of larger
scales is thus delayed, leading to a reduction of the jet spreading rate as well as radi-
ated noise (Hussain and Hasan 1985). To reduce the shock-associated broadband and
screech noise, researchers have attempted to produce shock-free supersonic plumes
with nonaxisymmetric distributions of exit-plane momentum thickness or to create
jets that are excited by oscillating control devices or by feedback of time-dependent
disturbances within the initial jet column.
12.4.5 Turbomachinery Blades
The noise radiated from fans, propellers, turbofans, propfans, turbine blades,
and the like is dominated by a series of pure tones whose fundamental is the basic
rotational frequency multiplied by the number of blades. Any uctuations generated
by the turbulent ow itself are merely the high-frequency noise background to the
pure tones and their harmonics. The turbomachine thrust and torque result from the
blades periodic forces acting on the uid. These forces are acoustically equivalent to
a distribution of dipoles. The turbomachine disk diameter is typically comparable to
the generated acoustic wavelength, and thus the sound source cannot be treated as
compact, thus complicating the use of the acoustic analogy to compute the radiated
noise.
Upstream nonuniformities are caused by atmospheric gusts, inlet distortions,
blade tip vortices, rotor-stator wake interactions, interactions between blade rows,
and so forth. In the blade frame of reference, these nonuniformities appear as con-
vected vortical waves and induce unsteady forces on the blade surfaces, causing
forced vibrations and radiation of additional broadband as well as harmonic noise
components. If the frequency of this aerodynamic excitation matches the turboma-
chine natural frequency, the forced vibration can be destructive. The high-frequency,
broadband noise due to vortex shedding and lift uctuations on high-speed blades
is even more problematic to compute as compared with the pure tones. In hydrody-
namic applications, the length scale associated with blade-row interactions is small
compared with the wavelength of the radiated sound, and the source can therefore be
treated as acoustically compact, making the use of the Lighthills analogy convenient.
As in jets, here also the objective is to reduce the noise levels without compro-
mising the machine performance measured, for example, in terms of thrust per unit
of airow (specic thrust) or thrust per unit of engine weight. Making the turbo-
machine larger while reducing the rotational frequency of its blades does lead to a
reduction of radiated noise but adds weight, space, and drag penalties. Noise related
to the uctuating forces on the blades can be suppressed by reducing those uctua-
tions or by attenuating the radiated sound through the engine intake. McCune and
Kerrebrock (1973) maintained that the perceived noise can be reduced by (a) reduc-
ing the radiated power, (b) shifting the radiated power to a frequency at which the ear
is less sensitive or where attenuation is more rapid, and (c) changing the directivity
pattern. They provided useful analysis of the performance penalties associated with
noise reduction in turbomachines.
12.4.6 Antisound
In addition to passive devices, noise control can also be accomplished actively
or reactively. Both of these strategies require energy expenditure. An example of
predetermined active control is the forced oscillations applied to a jet exit to equalize
the spacing between successive vortex ring instabilities (see Section 12.4.4). A second
example of active control is the beaming of a continuous broadband noise in an ofce
space to mask other intrusive sounds such as conversation in an adjacent cubicle. The
result of masking is an increase in the noise level, but the subjective effect may be
a reduction in annoyance. Reactive sound control is often termed antisound and is
designed to reduce the noise levels, but the reactive systemis quite a bit more involved
than a mere predetermined active control system.
Reactive noise suppression systems are based on generating sound by auxiliary
source with such an amplitude and phase that, in the region of interest, the sound
wave interference from the original and auxiliary source results in considerable re-
duction of the noise levels. What makes these systems possible is, of course, the
linearity of the governing equations and the consequent validity of the principle of
superposition. Though the idea can perhaps be traced to the 1950s, signicant ad-
vances in antisound were not made until the 1980s by, especially, Ffowcs Williams
(1984). In principle, it is possible to cancel an existing sound eld by introducing a
newsound eld of equal amplitude and opposite phase. The elements of an antisound
system consist of sensors to measure the existing eld, acoustic sources to produce
the antisound, and microprocessor-based controllers that utilize inputs from the sen-
sors to determine the appropriate output signals. These systems work best to cancel
low-frequency sound. For higher-frequency sound, additional acoustic modes can
propagate as well as disperse, complicating the application of the antisound concept.
268 NOISE REDUCTION
Fortunately, passive devices work well at high frequencies, and a combined system
may be able to suppress most of the noise.
The earliest application of antisound involved the reduction of fan noise in ducts.
We cite here a few more recent examples. Rotating stall instability in axial compres-
sors has been successfully controlled by active methods. Delay of surge in gas turbine
engines has been demonstrated, allowing a signicant increase in power beyond the
surge boundary for uncontrolled engines. The uttering of a exible wing in a wind
tunnel has been successfully stabilized by activating a wall-mounted loudspeaker.
Other existing or potential applications include reduction of cabinet noise in aircraft
and automobiles, combustion chambers in jet engines, electric power transformers,
diesel engines, electric motors, pumps, compressors, and other turbomachinery. The
pilots of the Voyager aircraft, which completed the rst nonstop, nonrefueled ight
around the world in December 1986, used an active-noise-control communications
headset system.
The older antisound systems with xed gain in the feedback loop could not achieve
high noise reduction. However, with the recent availability of adaptive lter systems,
much more impressive suppression of noise is feasible (Tichy, Warnaka, and Poole
1984). These systems are capable of adjusting the feedback loop for the magnitude
and phase relationship of the spectral components and can quickly compensate for
the sound path changes. These reactive-control devices seem to work best for low-
frequency sound, for which passive silencers are relatively ineffective, and when
the source is localized and accessible. Local control is obviously easier than global.
Moreover, effective control is achieved when the systemresponse within the frequency
band of interest is dominated by relatively few modes.
Combustion chambers often act as acoustic resonators, and attempts to increase
their density of energy release can lead to undesired combustion instabilities that can
be violent enough to quench the ame or even to cause structural damage. When
low-energy antisound systems were applied to combustion chambers to reduce the
noise generated by the combustion instabilities, both the sound eld and those buzz
instabilities were suppressed, dramatically improving the chamber performance. It
is conceivable that properly designed antisound systems can repeat this success in
other uid-ow situations in which instabilities contribute to the radiated sound as
well as limit performance. Review papers on active control of noise are available by
Warnaka (1982); Ericksson et al. (1988); Van Laere and Sas (1988); and Warner,
Waters, and Bernhard (1988). Antisound techniques seem to have gone from the
laboratory to application and commercialization in a remarkably short time.
CHAPTER THIRTEEN
Microelectromechanical Systems
If nature were not beautiful, it would not be worth studying it. And life
would not be worth living.
(Jules Henry Poincar e, 18541912)
Everything should be made as simple as possible, but not simpler.
(Albert Einstein, 18791955)
PROLOGUE
This chapter provides background material on an essential element of the targeted
ow-control strategy to be detailed in Chapter 14. Manufacturing processes that
can create extremely small machines have been developed in recent years. Micro-
electromechanical systems (MEMS) refer to devices that have characteristic length of
less than 1 mm but more than 1 m, that combine electrical and mechanical compo-
nents, and that are fabricated using integrated circuit batch-processing techniques.
Electrostatic, magnetic, electromagnetic, pneumatic, and thermal actuators, motors,
valves, gears, cantilevers, diaphragms, and tweezers of less than 100-m size have
been fabricated. These have been used as sensors for pressure, temperature, mass
ow, velocity, and sound; as actuators for linear and angular motions; and as simple
components for complex systems such as microheat engines and microheat pumps.
Many of these microsensors and microactuators are potentially very useful for fu-
turistic distributed control systems that are particularly suited for turbulent ows, as
will be discussed in the next chapter. In the present chapter, we review the status of
our understanding of uid-ow phenomena particular to microdevices. In terms of
applications, this and the following chapter emphasize the use of MEMS as sensors
and actuators for ow diagnosis and control.
13.1 Introduction
How many times when you are working on something frustratingly tiny, like your wifes
wrist watch, have you said to yourself, If I could only train an ant to do this! What I
would like to suggest is the possibility of training an ant to train a mite to do this. What
are the possibilities of small but movable machines? They may or may not be useful, but
they surely would be fun to make.
269
270 MICROELECTROMECHANICAL SYSTEMS
Figure 13.1: The scale of things in meters. Lower scale continues in the upper bar from left to right.
(From the talk Theres Plenty of Room at the Bottom, delivered by Richard
P. Feynman at the annual meeting of the American Physical Society, Pasadena,
California, 29 December 1959.)
Toolmaking has always differentiated our species from all others on Earth. Aero-
dynamically correct wooden spears were carved by archaic Homo sapiens close to
400,000 years ago. Man builds things consistent with his size, typically in the range
of two orders of magnitude larger or smaller than himself, as indicated in Figure 13.1.
Though the extremes of length scale are outside the range of this gure, human be-
ings, at slightly more than 10
0
m, amazingly t right in the middle of the smallest
subatomic particle, which is approximately 10
26
m, and the extent of the observ-
able universe, which is 1.42 10
26
m (15 billion light years). An egocentric uni-
verse indeed! But humans have always striven to explore, build, and control the
extremes of length and time scales. In the voyages to Lilliput and Brobdingnag of
Gullivers Travels, Jonathan Swift (1726) speculated on the remarkable possibilities
that diminution or magnication of physical dimensions provide. The Great Pyra-
mid of Khufu was originally 147 mhigh when completed around 2600 B.C., whereas
the Empire State Building constructed in 1931 is presentlyafter the addition of a
television antenna mast in 1950449 m high. At the other end of the spectrum of
man-made artifacts, a dime is slightly less than 2 cm in diameter. Watchmakers have
practiced the art of miniaturization since the thirteenth century. The invention of
the microscope in the seventeenth century opened the way for direct observation of
microbes and plant and animal cells. Smaller things were man-made in the latter half
of this century. The transistorinvented in 1947in todays integrated circuits has a
size
1
of 0.18 m (180 nanometers) in production and approaches 10 nm in research
laboratories using electron beams. But what about the miniaturization of mechani-
cal partsmachinesenvisioned by Feynman (1961) in his legendary speech quoted
above?
1
The smallest feature in a microchip is dened by its smallest linewidth, which in turn is related to the
wavelength of light employed in the basic lithographic process that is used to create the chip.
Manufacturing processes that can create extremely small machines have been de-
veloped in recent years (Angell, Terry, and Barth 1983; Gabriel et al. 1988, 1992;
OConnor 1992; Gravesen, Branebjerg, and Jensen 1993; Bryzek, Peterson, and
McCulley 1994; Gabriel 1995; Hogan 1996; Ho and Tai 1996, 1998; Tien 1997;
Busch-Vishniac 1998; Amato 1998). Electrostatic, magnetic, electromagnetic, pneu-
matic and thermal actuators, motors, valves, gears, cantilevers, diaphragms, and
tweezers of less than 100-m size have been fabricated. These have been used as
sensors for pressure, temperature, mass ow, velocity, and sound; as actuators for
linear and angular motions; and as simple components for complex systems such as
microheat engines and microheat pumps (Lipkin 1993; Garcia and Sniegowski 1993,
1995; Sniegowski and Garcia 1996; Epstein and Senturia 1997; Epstein et al. 1997).
The technology is progressing at a rate that far exceeds that of our understanding of
the unconventional physics involved in the operation as well as the manufacturing of
those minute devices (Knight 1999b). The present chapter focuses on one aspect
of such physics: uid-ow phenomena associated with microscale devices. In terms
of applications, this and the following chapter will emphasize the use of MEMS as
sensors and actuators for ow diagnosis and control.
The term microelectromechanical systems (MEMS) refers to devices that have
characteristic length of less than 1 mm but more than 1 m, that combine
electrical and mechanical components, and that are fabricated using integrated circuit
batch-processing technologies. Current manufacturing techniques for MEMS include
surface silicon micromachining; bulk silicon micromachining; lithography, electrode-
position, and plastic molding (or, in its original German, lithographie galvanofor-
mung abformung, LIGA); and electrodischarge machining (EDM). As indicated in
Figure 13.1, MEMS are more than four orders of magnitude larger than the diameter
of the hydrogen atombut about four orders of magnitude smaller than the traditional
man-made artifacts. Nanodevices (some say NEMS) further push the envelope of
electromechanical miniaturization.
Despite Feynmans demurring regarding the usefulness of small machines, MEMS
are nding increased applications in a variety of industrial and medical elds with
a potential worldwide market in the billions of dollars. Accelerometers for automo-
bile airbags, keyless entry systems, dense arrays of micromirrors for high-denition
optical displays, scanning electron microscope tips to image single atoms, microheat
exchangers for cooling of electronic circuits, reactors for separating biological cells,
blood analyzers, and pressure sensors for catheter tips are but a fewin current use. Mi-
croducts are used in infrared detectors, diode lasers, miniature gas chromatographs,
and high-frequency uidic control systems. Micropumps are used for ink-jet print-
ing, environmental testing, and electronic cooling. Potential medical applications
for small pumps include controlled delivery and monitoring of minute amounts of
medication, manufacturing of nanoliters of chemicals, and development of an arti-
cial pancreas. Several new journals are dedicated to the science and technology of
MEMS, for example the IEEE/ASME Journal of Microelectromechanical Systems,
Journal of Micromechanics and Microengineering, and Microscale Thermophysical
Engineering.
Not all MEMS devices involve uid ows, but this chapter will focus on the ones
that do. Microducts, micropumps, microturbines, and microvalves are examples of
small devices involving the owof liquids and gases. Microelectromechanical systems
can also be related to uid ows in an indirect way. The availability of inexpensive,
batch-processing-produced microsensors and microactuators provides opportunities
for targeting small-scale coherent structures in macroscopic turbulent shear ows.
Flow control using MEMS promises a quantum leap in control system performance.
This chapter will cover both the direct and indirect aspects of microdevices and uid
ows. Section 2 addresses the question of modeling uid ows in microdevices, and
Section 3 gives a brief overview of typical applications of MEMS in the eld of uid
mechanics. The papers by Gad-el-Hak (1999) and L ofdahl and Gad-el-Hak (1999)
provide more detail on MEMS applications in turbulence and ow control. In this
book, Chapter 14 details reactive owcontrol strategies particularly suited for taming
turbulence.
13.2 Flow Physics
13.2.1 Fluid Mechanics Issues
The rapid progress in fabricating and utilizing microelectromechanical systems
during the last decade has not been matched by corresponding advances in our under-
standing of the unconventional physics involved in the operation and manufacture
of small devices. Providing such understanding is crucial to designing, optimizing,
fabricating, and operating improved MEMS devices.
Fluid ows in small devices differ from those in macroscopic machines. The oper-
ation of MEMS-based ducts, nozzles, valves, bearings, turbomachines, and so forth
cannot always be predicted fromconventional owmodels such as the NavierStokes
equations with a no-slip boundary condition at a uidsolid interface, as routinely
and successfully applied for larger ow devices. Many questions have been raised
when the results of experiments with microdevices could not be explained via tra-
ditional ow modeling. The pressure gradient in a long microduct was observed to
be nonconstant, and the measured owrate was higher than that predicted from the
conventional continuum ow model. Load capacities of microbearings were dimin-
ished, and electric currents needed to move micromotors were extraordinarily high.
The dynamic response of micromachined accelerometers operating at atmospheric
conditions was observed to be overdamped.
In the early stages of development of this exciting new eld, the objective was to
build MEMS devices as productively as possible. Microsensors were reading some-
thing, but not many researchers seemed to know exactly what. Microactuators were
moving, but conventional modeling could not precisely predict their motion. After
a decade of unprecedented progress in MEMS technology, perhaps the time is now
ripe to take stock, slow down a bit, and answer the many questions that have arisen.
The ultimate aim of this long-term exercise is to achieve rational-design capability
for useful microdevices and to be able to characterize the operations of microsensors
and microactuators denitively and with as little empiricism as possible.
In dealing with uid ow through microdevices, one is faced with the question
of which model to use, which boundary condition to apply, and how to proceed
to obtain solutions to the problem at hand. Obviously, surface effects dominate
in small devices. The surface-to-volume ratio for a machine with a characteristic
length of 1 m is 1 m
1
, whereas that for a MEMS device having a size of 1 m is
10
6
m
1
. The millionfold increase in surface area relative to the mass of the minute
device substantially affects the transport of mass, momentum, and energy through
13.2 FLOW PHYSICS 273
the surface. The small length scale of microdevices may invalidate the continuum
approximation altogether. Slip ow, thermal creep, rarefaction, viscous dissipation,
compressibility, intermolecular forces, and other unconventional effects may have to
be taken into account, preferably using only rst principles such as conservation of
mass, Newtons second law, conservation of energy, and so forth.
In this section, we discuss continuum as well as molecular-based ow models
and the choices to be made. Computing typical Reynolds, Mach, and Knudsen num-
bers for the ow through a particular device is a good start to characterize the ow.
For gases, microuid mechanics has been studied by incorporating slip boundary
conditions, thermal creep, viscous dissipation, and compressibility effects into the
continuum equations of motion. Molecular-based models have also been attempted
for certain ranges of the operating parameters. Use is made of the well-developed
kinetic theory of gases, embodied in the Boltzmann equation, and direct simulation
methods such as Monte Carlo. Microuid mechanics of liquids is more complicated.
The molecules are much more closely packed at normal pressures and temperatures,
and the attractive or cohesive potential between the liquid molecules as well as be-
tween the liquid and solid ones plays a dominant role if the characteristic length of
the ow is sufciently small. In cases in which the traditional continuum model fails
to provide accurate predictions or postdictions, expensive molecular dynamics sim-
ulations seem to be the only rst-principle approach available to characterize liquid
ows in microdevices rationally. Such simulations are not yet feasible for realistic
ow extent or number of molecules. As a consequence, the microuid mechanics of
liquids is much less developed than that for gases.
13.2.2 Fluid Modeling
There are basically two ways of modeling a oweld: either as the uid really is,
a collection of molecules, or as a continuum in which the matter is assumed contin-
uous and indenitely divisible. The former modeling is subdivided into deterministic
methods and probabilistic ones, whereas in the latter approach the velocity, density,
pressure, and so forth are dened at every point in space and time and conservation
of mass, energy, and momentum leads to a set of nonlinear partial differential equa-
tions (Euler, NavierStokes, Burnett, etc.). Fluid-modeling classication is depicted
schematically in Figure 13.2.
The continuum model, embodied in the NavierStokes equations, is applicable
to numerous ow situations. The model ignores the molecular nature of gases and
liquids and regards the uid as a continuous medium describable in terms of the
spatial and temporal variations of density, velocity, pressure, temperature, and other
macroscopic owquantities. For dilute gas ows near equilibrium, the NavierStokes
equations are derivable from the molecularly based Boltzmann equation but can also
be derived independently for both liquids and gases. In the case of direct derivation,
some empiricism is necessary to close the resulting indeterminate set of equations.
The continuum model is easier to handle mathematically (and is also more familiar
to most uid dynamists) than the alternative molecular models. Continuum models
should therefore be used as long as they are applicable. Thus, careful considerations
of the validity of the NavierStokes equations and the like are in order.
Basically, the continuummodel leads to fairly accurate predictions as long as local
properties such as density and velocity can be dened as averages over elements large
compared with the microscopic structure of the uid but small enough in compar-
ison with the scale of the macroscopic phenomena to permit the use of differential
Figure 13.2: Molecular and continuum ow models.
calculus to describe them. Additionally, the ow must not be too far from thermo-
dynamic equilibrium. The former condition is almost always satised, but it is the
latter that usually restricts the validity of the continuum equations. As will be seen
in Section 13.2.3, the continuum ow equations do not form a determinate set. The
shear stress and heat ux must be expressed in terms of lower-order macroscopic
quantities such as velocity and temperature, and the simplest (i.e., linear) relations
are valid only when the ow is near thermodynamic equilibrium. Worse yet, the
traditional no-slip boundary condition at a solid-uid interface breaks down even
before the linear stressstrain relation becomes invalid.
To be more specic, we temporarily restrict the discussion to gases for which
the concept of mean free path is well dened. Liquids are more problematic, and we
defer their discussion to Section 13.2.7. For gases, the mean free path L is the average
distance traveled by molecules between collisions. For an ideal gas modeled as rigid
spheres, the mean free path is related to temperature T and pressure p as follows:
L =
1
2n
2
=
kT
2p
2
(13.1)
where n is the number density (number of molecules per unit volume), is the
molecular diameter, and k is the Boltzmann constant.
The continuum model is valid when L is much smaller than a characteristic ow
dimension L. As this condition is violated, the owis no longer near equilibrium, and
the linear relation between stress and rate of strain and the no-slip velocity condition
are no longer valid. Similarly, the linear relation between heat ux and temperature
gradient and the no-jump temperature condition at a solid-uid interface are no
longer accurate when L is not much smaller than L.
The length scale L can be some overall dimension of the ow, but a more precise
choice is the scale of the gradient of a macroscopic quantity, as, for example, the
density , as dened by
L =

(13.2)
The ratio between the mean free path and the characteristic length is known as the
Knudsen number
Kn =
L
L
(13.3)
and generally the traditional continuum approach is valid, albeit with modied
boundary conditions, as long as Kn < 0.1.
There are two more important dimensionless parameters in uid mechanics, and
the Knudsen number can be expressed in terms of these two. The Reynolds number
is the ratio of inertial forces to viscous ones
Re =
v
o
L
(13.4)
where v
o
is a characteristic velocity, and is the kinematic viscosity of the uid. The
Mach number is the ratio of ow velocity to the speed of sound
Ma =
v
o
a
o
(13.5)
The Mach number is a dynamic measure of uid compressibility and may be consid-
ered as the ratio of inertial forces to elastic ones. From the kinetic theory of gases,
the mean free path is related to the viscosity as follows:
=

=
1
2
Lv
m
(13.6)
where is the dynamic viscosity, and v
m
is the mean molecular speed, which is
somewhat higher than the sound speed a
o
,
v
m
=
a
o
(13.7)
where is the specic heat ratio (i.e., the isentropic exponent). Combining Eqs.
(13.3)(13.7), we reach the required relation
Kn =
_
2
Ma
Re
(13.8)
In boundary layers, the relevant length scale is the shear-layer thickness , and
for laminar ows
L

1
Re
(13.9)
Kn
Ma
Re
Ma
Re
(13.10)
where Re
is the Reynolds number based on the freestreamvelocity v

o
and the bound-
ary layer thickness , and Re is based on v
o
and the streamwise length scale L.
Rareed gas ows are in general encountered in ows in small geometries such
as MEMS devices and in low-pressure applications such as high-altitude ying and
high-vacuumgadgets. The local value of the Knudsen number in a particular owde-
termines the degree of rarefaction and the degree of validity of the continuum model.
The different Knudsen number regimes are determined empirically and are therefore
only approximate for a particular owgeometry. The pioneering experiments in rare-
ed gas dynamics were conducted by Knudsen in 1909. In the limit of zero Knudsen
number, the transport terms in the continuum momentum and energy equations are
negligible, and the NavierStokes equations then reduce to the inviscid Euler equa-
tions. Both heat conduction and viscous diffusion and dissipation are negligible, and
the owis then approximately isentropic (i.e., adiabatic and reversible) fromthe con-
tinuum viewpoint, whereas the equivalent molecular viewpoint is that the velocity
distribution function is everywhere of the local equilibrium or Maxwellian form. As
Kn increases, rarefaction effects become more important, and eventually the contin-
uum approach breaks down altogether. The different Knudsen number regimes are
depicted in Figure 13.3 and can be summarized as follows:
Euler equations (neglect molecular diffusion): Kn 0 (Re )
NavierStokes equations with no-slip boundary conditions: Kn 10
3
NavierStokes equations with slip boundary conditions: 10
3
Kn 10
1
Transition regime: 10
1
Kn 10
Free-molecule ow: Kn > 10
We will return to these regimes in the following subsections.
As an example, consider air at standard temperature (T = 288 K) and pressure
( p = 1.01 10
5
N/m
2
). A cube 1 m to the side contains 2.54 10
7
molecules
separated by an average distance of 0.0034 m. The gas is considered dilute if the
ratio of this distance to the molecular diameter exceeds 7, and in the present exam-
ple this ratio is 9, which barely satises the dilute gas assumption. The mean free
path computed from Eq. (13.1) is L = 0.065m. A microdevice with characteristic
length of 1 m would have Kn = 0.065, which is in the slip-ow regime. At lower
Figure 13.3: Knudsen number regimes.
pressures, the Knudsen number increases. For example, if the pressure is 0.1 atm and
the temperature remains the same, Kn = 0.65 for the same 1-m device, and the
ow is then in the transition regime. There would still be over 2 million molecules
in the same one-micron cube, and the average distance between them would be
0.0074 m. The same device at 100 km altitude would have Kn = 3 10
4
, which is
well into the free-molecule ow regime. The Knudsen number for the ow of a light
gas like heliumis about three times larger than that for air owat otherwise the same
conditions.
Consider a long microchannel for which the entrance pressure is atmospheric
and the exit conditions are near vacuum. As air goes down the duct, the pressure
and density decrease, whereas the velocity, Mach number, and Knudsen number
increase. The pressure drops to overcome viscous forces in the channel. If isothermal
conditions prevail,
2
density also drops, and conservation of mass requires the ow
to accelerate down the constant-area tube. The uid acceleration in turn affects the
pressure gradient, resulting in a nonlinear pressure drop along the channel. The
Mach number increases down the tube, limited only by the choked-ow condition
(Ma = 1). Additionally, the normal component of velocity is no longer zero. With
lower density, the mean free path increases, and Kn correspondingly increases. All
ow regimes depicted in Figure 13.3 may occur in the same tube: continuum with
no-slip boundary conditions, slip-ow regime, transition regime, and free-molecule
ow. The air ow may also change from incompressible to compressible as it moves
down the microduct. A similar scenario may take place if the entrance pressure is,
say, 5 atm, while the exit is atmospheric. This deceivingly simple duct ow may in
fact manifest every single complexity discussed in this section.
In the following six subsections, we discuss in turn the NavierStokes equations,
compressibility effects, boundary conditions, molecular-based models, liquid ows,
and surface phenomena.
13.2.3 Continuum Model
We recall in this subsection the traditional conservation relations in uid mechan-
ics. These equations were derived in Chapter 2. Here, we reemphasize the precise
assumptions needed to obtain a particular form of those equations. A continuum
uid implies that the derivatives of all the dependent variables exist in some reason-
able sense. In other words, local properties such as density and velocity are dened
as averages over elements large compared with the microscopic structure of the uid
but small enough in comparison with the scale of the macroscopic phenomena to
permit the use of differential calculus to describe them. As mentioned earlier, such
conditions are almost always met. For such uids, and on the assumption that the
laws of nonrelativistic mechanics hold, the conservation of mass, momentum, and
energy can be expressed at every point in space and time as a set of partial differential
equations as follows:
t
+

x
k
(u
k
) = 0 (13.11)
2
More likely the ow will be somewhere in between isothermal and adiabatic, Fanno ow. In that case
both density and temperature decrease downstream, the former not as fast as in the isothermal case.
None of that changes the qualitative arguments made in the example.
_
u
i
t
+u
k
u
i
x
k
_
=

ki
x
k
+g
i
(13.12)
_
e
t
+u
k
e
x
k
_
=
q
k
x
k
+
ki
u
i
x
k
(13.13)
where is the uid density, u
k
is an instantaneous velocity component (u, v, w),
ki
is the second-order stress tensor (surface force per unit area), g
i
is the body force
per unit mass, e is the internal energy, and q
k
is the sum of heat ux vectors due to
conduction and radiation. The independent variables are time t and the three spatial
coordinates x
1
, x
2
, and x
3
or (x, y, z).
Equations (13.11), (13.12), and (13.13) constitute 5 differential equations for
the 17 unknowns , u
i
,
ki
, e, and q
k
. Absent any body couples, the stress tensor
is symmetric, having only 6 independent components, which reduces the number of
unknowns to 14. Obviously, the continuumowequations do not forma determinate
set. To close the conservation equations, the relation between the stress tensor and
deformation rate, the relation between the heat ux vector, and the temperature eld
and appropriate equations of state relating the different thermodynamic properties
are needed. The stressrate of strain relation and the heat uxtemperature gradient
relation are approximately linear if the ow is not too far from thermodynamic
equilibrium. This is a phenomenological result but can be rigorously derived fromthe
Boltzmann equation for a dilute gas if it is assumed that the ow is near equilibrium
(see Section 13.2.6). For a Newtonian, isotropic, Fourier, ideal gas, for example,
these relations read
ki
= p
ki
+
_
u
i
x
k
+
u
k
x
i
_
+
_
u
j
x
j
_
ki
(13.14)
q
i
=
T
x
i
+Heat ux due to radiation (13.15)
de = c
v
dT and p = RT (13.16)
where p is the thermodynamic pressure, and are the rst and second coefcients
of viscosity, respectively,
ki
is the unit second-order tensor (Kronecker delta), is
the thermal conductivity, T is the temperature eld, c
v
is the specic heat at constant
volume, and Ris the gas constant, which is given by the Boltzmann constant divided
by the mass of an individual molecule (k = mR). The Stokes hypothesis relates the
rst and second coefcients of viscosity, thus +
2
3
= 0, although the validity of
this assumption for other than dilute, monatomic gases has occasionally been ques-
tioned (Gad-el-Hak 1995). With the preceding constitutive relations and radiative
heat transfer neglected, Eqs. (13.11), (13.12), and (13.13) respectively read
t
+

x
k
(u
k
) = 0 (13.17)
_
u
i
t
+u
k
u
i
x
k
_
=
p
x
i
+g
i
+

x
k
_
_
u
i
x
k
+
u
k
x
i
_
+
ki
u
j
x
j
_
(13.18)
c
v
_
T
t
+u
k
T
x
k
_
=

x
k
_
T
x
k
_
p
u
k
x
k
+ (13.19)
The three components of the vector equation (13.18) are the NavierStokes equations
expressing the conservation of momentum for a Newtonian uid. In the thermal
energy equation (13.19), is the always positive dissipation function expressing
the irreversible conversion of mechanical energy to internal energy as a result of the
deformation of a uid element. The second termon the right-hand side of Eq. (13.19)
is the reversible work done (per unit time) by the pressure as the volume of a uid
material element changes. For a Newtonian, isotropic uid, the viscous dissipation
rate is given by
=
1
2
_
u
i
x
k
+
u
k
x
i
_
2
+
_
u
j
x
j
_
2
(13.20)
There are nowsix unknowns, , u
i
, p, and T, and the ve coupled equations (13.17),
(13.18), and (13.19) plus the equation of state relating pressure, density, and temper-
ature. These six equations together with a sufcient number of initial and boundary
conditions constitute a well-posed, albeit formidable, problem. The system of equa-
tions (13.17)(13.19) is an excellent model for the laminar or turbulent ow of most
uids such as air and water under many circumstances, including high-speed gas ows
for which the shock waves are thick relative to the mean free path of the molecules.
Considerable simplication is achieved if the ow is assumed incompressible,
which is usually a reasonable assumption provided that the characteristic ow speed
is less than 0.3 of the speed of sound. The incompressibility assumption is readily
satised for almost all liquid ows and many gas ows. In such cases, the density is
assumed either a constant or a given function of temperature (or species concentra-
tion). The governing equations for such ows are
u
k
x
k
= 0 (13.21)
_
u
i
t
+u
k
u
i
x
k
_
=
p
x
i
+

x
k
_
_
u
i
x
k
+
u
k
x
i
__
+g
i
(13.22)
c
p
_
T
t
+u
k
T
x
k
_
=

x
k
_
T
x
k
_
+
incomp
(13.23)
where
incomp
is the incompressible limit of Eq. (13.20). These are now ve equations
for the ve dependent variables u
i
, p, and T. As noted in Section 2.2, the left-hand
side of Eq. (13.23) has the specic heat at constant pressure c
p
and not c
v
.
For both the compressible and the incompressible equations of motion, the trans-
port terms are neglected away fromsolid walls in the limit of innite Reynolds number
(Kn 0). The uid is then approximated as inviscid and nonconducting, and the
corresponding equations read (for the compressible case)
t
+

x
k
(u
k
) = 0 (13.24)
_
u
i
t
+u
k
u
i
x
k
_
=
p
x
i
+g
i
(13.25)
c
v
_
T
t
+u
k
T
x
k
_
= p
u
k
x
k
(13.26)
The Euler equation (13.25) can be integrated along a streamline, and the resulting
Bernoullis equation provides a direct relation between the velocity and pressure.
13.2.4 Compressibility
The issue of whether to consider the continuumowcompressible or incompress-
ible seems to be rather straightforward but is in fact full of potential pitfalls. This issue
was covered in Chapter 2, but portions of that material are repeated here because
of their importance to MEMS ows and to provide continuity between the different
subsections in this section. If the local Mach number is less than 0.3, then the ow of
a compressible uid like air canaccording to the conventional wisdombe treated
as incompressible. But the well-known Ma < 0.3 criterion is only a necessary but
not a sufcient one to allow treatment of the ow as approximately incompressible.
In other words, there are situations in which the Mach number can be exceedingly
small while the owis compressible. As is well documented in heat transfer textbooks,
strong wall heating or cooling may cause the density to change sufciently and the
incompressible approximation to break down, even at low speeds. Less known is
the situation encountered in some microdevices in which the pressure may strongly
change because of viscous effects even though the speeds may not be high enough
for the Mach number to go above the traditional threshold of 0.3. Corresponding
to the pressure changes are strong density changes that must be taken into account
when writing the continuum equations of motion. In this section, we systematically
explain all situations relevant to MEMS for which compressibility effects must be
considered.
Let us rewrite the full continuity equation (13.11) as follows:
D
Dt
+
u
k
x
k
= 0 (13.27)
where
D
Dt
is the substantial derivative (

t
+u
k

x
k
) expressing changes following a uid
element. The proper criterion for the incompressible approximation to hold is that
(
1
D
Dt
) is vanishingly small. In other words, if density changes following a uid particle
are small, the ow is approximately incompressible. Density may change arbitrarily
from one particle to another without violating the incompressible ow assumption.
This is the case, for example, in the stratied atmosphere and ocean, where the
variable-densitytemperaturesalinity ow is often treated as incompressible.
From the state principle of thermodynamics, we can express the density changes
of a simple system in terms of changes in pressure and temperature by
= ( p, T). (13.28)
From the chain rule of calculus,
1
D
Dt
=
Dp
Dt

DT
Dt
(13.29)
where and are, respectively, the isothermal compressibility coefcient and the
bulk expansion coefcient, both previously dened in Section 2.3. The ow must be
treated as compressible if pressure and temperature changes are sufciently strong.
Equation (13.29) must of course be properly nondimensionalized before it is decided
whether a term is large or small. Here, we follow closely the procedure detailed in
Panton (1996).
Consider rst the case of adiabatic walls. Density is normalized with a reference
value
o
, velocities with a reference speed v
o
, spatial coordinates and time with,
respectively, L and L/v
o
, and the isothermal compressibility coefcient and bulk
expansion coefcient with reference values
o
and
o
, respectively. The pressure
is nondimensionalized with the inertial pressure scale
o
v
2
o
. This scale is twice the
dynamic pressure, that is, the pressure change as an inviscid uid moving at the
reference speed is brought to rest.
Temperature changes for the case of adiabatic walls result from the irreversible
conversion of mechanical energy into internal energy via viscous dissipation. Tem-
perature is therefore nondimensionalized as follows:
T
=
T T
o
_
o
v
2
o
o
_
=
T T
o
Pr
_
v
2
o
c
po
_
(13.30)
where T
o
is a reference temperature,
o
,
o
, and c
p
o
are, respectively, reference vis-
cosity, thermal conductivity, and specic heat at constant pressure, and Pr is the
reference Prandtl number, (
o
c
p
o
)/
o
.
In the present formulation, the scaling used for pressure is based on the Bernoullis
equation and therefore neglects viscous effects. This particular scaling guarantees
that the pressure term in the momentum equation will be of the same order as the
inertia term. The temperature scaling assumes that the conduction, convection, and
dissipation terms in the energy equation have the same order of magnitude. The
resulting dimensionless form of Eq. (13.29) reads
1
Dt
=
o
Ma
2
_
Dp
Dt

PrB
A
DT
Dt
_
(13.31)

indicates a nondimensional quantity, Ma is the reference
Mach number, and Aand B are dimensionless constants dened by A
o
o
c
p
o
T
o
,
and B
o
T
o
. If the scaling is properly chosen, the terms having the

superscript
in the right-hand side should be of order one, and the relative importance of such
terms in the equations of motion is determined by the magnitude of the dimension-
less parameter(s) appearing to their left (e.g., Ma, Pr, etc.). Therefore, as Ma
2
0,
temperature changes due to viscous dissipation are neglected (unless Pr is very large
as, for example, in the case of highly viscous polymers and oils). Within the same
order of approximation, all thermodynamic properties of the uid are assumed
constant.
Pressure changes are also neglected in the limit of zero Mach number. Hence,
for Ma <0.3 (i.e., Ma
2
<0.09), density changes following a uid particle can be
neglected, and the ow can then be approximated as incompressible. However, there
is a caveat in this argument. Pressure changes due to inertia can indeed be neglected
at small Mach numbers, and this is consistent with the way we nondimensionalized
the pressure term above. If, on the other hand, pressure changes are mostly due to
viscous effects, as is the case for example in a long duct or a gas bearing, pressure
changes may be signicant even at low speeds (low Ma). In that case the term
Dp
Dt
in
Eq. (13.31) is no longer of order one and may be large regardless of the value of Ma.
Density then may change signicantly, and the ow must be treated as compressible.
Had pressure been nondimensionalized using the viscous scale (
o
v
o
L
) instead of the
inertial one (
o
v
2
o
), the revised equation (13.31) would have Re
1
appearing explicitly
in the rst term in the right-hand side, thus accentuating the importance of this term
when viscous forces dominate.
A similar result can be gleaned when the Mach number is interpreted as follows:
Ma
2
=
v
2
o
a
2
o
= v
2
o
s
=

o
v
2
o
p
=

o
(13.32)
where s is the entropy. Again, the preceding equation assumes that pressure changes
are inviscid, and therefore small Mach number means negligible pressure and density
changes. In a ow dominated by viscous effectssuch as that inside a microduct
density changes may be signicant even in the limit of zero Mach number.
Identical arguments can be made in the case of isothermal walls. Here strong
temperature changes may be the result of wall heating or cooling, even if viscous
dissipation is negligible. The proper temperature scale in this case is given in terms
of the wall temperature T
w
and the reference temperature T
o
as follows
T =
T T
o
T
w
T
o
(13.33)
where

T is the new dimensionless temperature. The nondimensional form of
Eq. (13.29) now reads
1
Dt
=
o
Ma
2
Dp
Dt
B
_
T
w
T
o
T
o
_
D

T
Dt
(13.34)
Here we notice that the temperature term is different from that in Eq. (13.31), for
Ma is no longer appearing in this term, and strong temperature changes, that is, large
(T
w
T
o
)/T
o
, may cause strong density changes regardless of the value of the Mach
number. Additionally, the thermodynamic properties of the uid are not constant
but depend on temperature, and as a result, the continuity, momentum, and energy
equations are all coupled. The pressure term in Eq. (13.34), on the other hand, is
exactly as it was in the adiabatic case, and the same arguments made before apply:
the ow should be considered compressible if Ma > 0.3 or if pressure changes due
to viscous forces are sufciently large.
Experiments in gaseous microducts conrm the preceding arguments. For both
low- and high-Mach-number ows, pressure gradients in long microchannels are
nonconstant, consistent with the compressible owequations. Such experiments were
conducted by, among others, Prudhomme et al. (1986); Pfahler et al. (1991); van den
Berg et al. (1993); Liu et al. (1993, 1995); Pong et al. (1994); Harley et al. (1995);
Piekos and Breuer (1996); Arkilic (1997); and Arkilic, Schmidt, and Breuer (1995,
1997a,b). Sample results will be presented in the following subsection.
13.2.5 Boundary Conditions
The equations of motion described in Section 13.2.3 require a certain number
of initial and boundary conditions for proper mathematical formulation of ow
problems. In this subsection, we describe the boundary conditions at a uidsolid
interface. Boundary conditions in the inviscid ow theory pertain only to the veloc-
ity component normal to a solid surface. The highest spatial derivative of velocity in
the inviscid equations of motion is rst-order, and only one velocity boundary con-
dition at the surface is admissible. The normal velocity component at a uidsolid
interface is specied, and no statement can be made regarding the tangential velocity
component. The normal velocity condition simply states that a uid-particle path
cannot go through an impermeable wall. Real uids are of course viscous, and the
corresponding momentum equation has second-order derivatives of velocity, thus
requiring an additional boundary condition on the velocity component tangential to
a solid surface.
Traditionally, the no-slip condition at a uidsolid interface is enforced in the
momentum equation, and an analogous no-temperature-jump condition is applied
in the energy equation. The notion underlying the no-slip and no-jump condition
is that, within the uid, there cannot be any nite discontinuities of velocity or
temperature. Those would involve innite velocity or temperature gradients and thus
produce innite viscous stress or heat ux that would destroy the discontinuity in
innitesimal time. The interaction between a uid particle and a wall is similar to that
between neighboring uid particles, and therefore no discontinuities are allowed at
the uidsolid interface either. In other words, the uid velocity must be zero relative
to the surface, and the uid temperature must equal to that of the surface. But strictly
speaking these two boundary conditions are valid only if the uid owadjacent to the
surface is in thermodynamic equilibrium. This requires an innitely high frequency
of collisions between the uid and the solid surface. In practice, the no-slip and
no-jump condition leads to fairly accurate predictions as long as Kn < 0.001 (for
gases). Beyond that, the collision frequency is simply not high enough to ensure
equilibrium, and a certain degree of tangential-velocity slip and temperature jump
must be allowed. This is a case frequently encountered in MEMS ows, and we
develop the appropriate relations in this subsection.
For both liquids and gases, the linear Navier boundary condition empirically
relates the tangential velocity slip at the wall u|
w
to the local shear
u|
w
= u
uid
u
wall
= L
s
u
y
w
(13.35)
where L
s
is the constant slip length, and
u
y
|
w
is the strain rate computed at the wall.
In most practical situations, the slip length is so small that the no-slip condition holds.
In MEMS applications, however, that may not be the case. Once again we defer the
discussion of liquids to Section 13.2.7 and focus for now on gases.
On the assumption that isothermal conditions prevail, the preceding slip relation
has been rigorously derived by Maxwell (1879) from considerations of the kinetic
theory of dilute, monatomic gases. Gas molecules, modeled as rigid spheres, contin-
uously strike and reect from a solid surface just as they continuously collide with
each other. For an idealized perfectly smooth wall,
3
the incident angle exactly equals
the reected angle, and the molecules conserve their tangential momentum and thus
exert no shear on the wall. This is termed specular reection and results in perfect
slip at the wall. For an extremely rough wall, on the other hand, the molecules
3
At the molecular scale.
reect at some random angle uncorrelated with their entry angle. This perfectly dif-
fuse reection results in zero tangential momentum for the reected uid molecules
to be balanced by a nite slip velocity in order to account for the shear stress trans-
mitted to the wall. A force balance near the wall leads to the following expression
for the slip velocity:
u
gas
u
wall
= L
u
y
w
(13.36)
where L is the mean free path. The right-hand side can be considered as the rst
term in an innite Taylor series, which is sufcient if the mean free path is relatively
small enough. Equation (13.36) states that signicant slip occurs only if the mean
velocity of the molecules varies appreciably over a distance of one mean free path.
This is the case, for example, in vacuum applications, ow in microdevices, or both.
The number of collisions between the uid molecules and the solid in those cases
is not large enough for even an approximate ow equilibrium to be established.
Furthermore, additional (nonlinear) terms in the Taylor series would be needed as L
increases and the ow is further removed from the equilibrium state.
For real walls some molecules reect diffusively and some reect specularly. In
other words, a portion of the momentum of the incident molecules is lost to the
wall, and a (typically smaller) portion is retained by the reected molecules. The
tangential-momentum-accommodation coefcient
v
is dened as the fraction of
molecules reected diffusively. This coefcient depends on the uid, the solid, and
the surface nish and has been determined experimentally to be between 0.20.8
(Thomas and Lord 1974; Seidl and Steinheil 1974; Porodnov et al. 1974; Arkilic
et al. 1997b; Arkilic 1997), the lower limit being for exceptionally smooth surfaces,
whereas the upper limit is typical of most practical surfaces. The nal expression
derived by Maxwell for an isothermal wall reads
u
gas
u
wall
=
2
v
v
L
u
y
w
(13.37)
For
v
= 0, the slip velocity is unbounded, whereas for
v
= 1, Eq. (13.37) reverts
to Eq. (13.36).
Similar arguments were made for the temperature-jump boundary condition by
von Smoluchowski (1898). For an ideal gas ow in the presence of wall-normal
and tangential temperature gradients, the complete slip-ow and temperature-jump
boundary conditions read
u
gas
u
wall
=
2
v
v
1
_
2RT
gas
w
+
3
4
Pr( 1)
RT
gas
(q
x
)
w
=
2
v
v
L
_
u
y
_
w
+
3
4
T
gas
_
T
x
_
w
(13.38)
T
gas
T
wall
=
2
T
T
_
2( 1)
( +1)
_
1
R
_
2RT
gas
(q
y
)
w
=
2
T
T
_
2
( +1)
_
L
Pr
_
T
y
_
w
(13.39)
where x and y are the streamwise and normal coordinates, and are respectively
the uid density and viscosity, R is the gas constant, T
gas
is the temperature of the
gas adjacent to the wall, T
wall
is the wall temperature,
w
is the shear stress at the
wall, Pr is the Prandtl number, is the specic heat ratio, and (q
x
)
w
and (q
y
)
w
are,
respectively, the tangential and normal heat ux at the wall.
The tangential-momentum-accommodation coefcient
v
and the thermal-
accommodation coefcient
T
are given by, respectively,
v
=

i
w
(13.40)
T
=
dE
i
dE
r
dE
i
dE
w
(13.41)
where the subscripts i , r, and w stand for, respectively, incident, reected, and
solid wall conditions; is a tangential momentum ux; and dE is an energy
ux.
The second term in the right-hand side of Eq. (13.38) is the thermal creep, which
generates slip velocity in the uid opposite to the direction of the tangential heat ux
(i.e., ow in the direction of increasing temperature). At sufciently high Knudsen
numbers, streamwise temperature gradient in a conduit leads to a measurable pres-
sure gradient along the tube. This may be the case in vacuumapplications and MEMS
devices. Thermal creep is the basis for the so-called Knudsen pumpa device with
no moving partsin which rareed gas is hauled from one cold chamber to a hot
one.
4
Clearly, such a pump performs best at high Knudsen numbers and is typically
designed to operate in the free-molecule ow regime.
In dimensionless form, Eqs. (13.38) and (13.39), respectively, read
u
gas
u
wall
=
2
v
v
Kn
_
u
_
w
+
3
2
( 1)
Kn
2
Re
Ec
_
T
_
w
(13.42)
T
gas
T
wall
=
2
T
T
_
2
( +1)
_
Kn
Pr
_
T
_
w
(13.43)

indicates dimensionless quantity, Kn is the Knudsen number,
Re is the Reynolds number, and Ec is the Eckert number dened by
Ec =
v
2
o
c
p
T
= ( 1)
T
o
T
Ma
2
(13.44)
where v
o
is a reference velocity, T = (T
gas
T
o
), and T
o
is a reference tempera-
ture. Note that very low values of
v
and
T
lead to substantial velocity slip and
temperature jump even for ows with small Knudsen number.
The rst term in the right-hand side of Eq. (13.42) is rst-order in Knudsen
number, whereas the thermal creep term is second-order, meaning that the creep
phenomenon is potentially signicant at large values of the Knudsen number.
Equation (13.43) is rst-order in Kn. Using Eqs. (13.8) and (13.44), the thermal
4
The terminology Knudsen pump has been used by, for example, Vargo and Muntz (1996), but according
to Loeb (1961), the original experiments demonstrating such pumps were carried out by Osborne
Reynolds.
creep term in Eq. (13.42) can be rewritten in terms of T and Reynolds number.
Thus,
u
gas
u
wall
=
2
v
v
Kn
_
u
_
w
+
3
4
T
T
o
1
Re
_
T
_
w
(13.45)
It is clear that large temperature changes along the surface or low Reynolds numbers
lead to signicant thermal creep.
The continuum NavierStokes equations with no-slip or no-temperature jump
boundary conditions are valid as long as the Knudsen number does not exceed 0.001.
First-order slip or temperature-jump boundary conditions should be applied to the
NavierStokes equations in the range of 0.001 <Kn <0.1. The transition regime
spans the range of 0.1 <Kn <10, and second-order or higher slip or temperature-
jump boundary conditions are applicable there. Note, however, that the Navier
Stokes equations are rst-order accurate in Kn, as will be shown in Section 13.2.6,
and are themselves not valid in the transition regime. Either higher-order contin-
uum equations (e.g., Burnett equations) should be used there, or molecular modeling
should be invoked, abandoning the continuum approach altogether.
For isothermal walls, Beskok (1994) derived a higher-order slip-velocity condition
as follows:
u
gas
u
wall
=
2
v
v
_
L
_
u
y
_
w
+
L
2
2!
_
2
u
y
2
_
w
+
L
3
3!
_
3
u
y
3
_
w
+
_
(13.46)
Attempts to implement the slip condition of Eq. (13.46) in numerical simulations are
rather difcult. Second-order and higher derivatives of velocity cannot be computed
accurately near the wall. On the basis of asymptotic analysis, Beskok (1996) and
Beskok and Karniadakis (1994, 1999) proposed the following alternative higher-
order boundary condition for the tangential velocity, including the thermal creep
term:
u
gas
u
wall
=
2
v
v
Kn
1 bKn
_
u
_
w
+
3
2
( 1)
Kn
2
Re
Ec
_
T
_
w
(13.47)
where b is a high-order slip coefcient determined from the presumably known no-
slip solution, thus avoiding the computational difculties mentioned above. If this
high-order slip coefcient is chosen as b = u
w
/2u
w
, where the prime denotes deriva-
tive with respect to y and the velocity is computed from the no-slip NavierStokes
equations, Eq. (13.47) becomes second-order accurate in Knudsen number. Beskoks
procedure can be extended to third- and higher-orders for both the slip-velocity and
thermal creep terms.
Similar arguments can be applied to the temperature-jump boundary condition,
and the resulting Taylor series reads in dimensionless form (Beskok 1996)
T
gas
T
wall
=
2
T
T
_
2
( +1)
_
1
Pr
_
Kn
_
T
_
w
+
Kn
2
2!
_
2
T
y
2
_
w
+
_
(13.48)
Again, the difculties associated with computing second- and higher-order derivatives
Figure 13.4: Variation of mass owrate as a function of ( p
2
i
p
2
o
).
Original data acquired by S.A. Tison and plotted by Beskok et al.
(1996).
of temperature are alleviated by using an identical procedure to that utilized for the
tangential velocity boundary condition.
Several experiments in low-pressure macroducts or in microducts conrm the
necessity of applying the slip boundary condition at sufciently large Knudsen num-
bers. Among them are those conducted by Knudsen (1909), Pfahler at al. (1991),
Tison (1993), Liu et al. (1993, 1995), Pong et al. (1994), Arkilic et al. (1995), Harley
et al. (1995), and Shih et al. (1995, 1996). The experiments are complemented by the
numerical simulations carried out by Beskok (1994, 1996); Beskok and Karniadakis
(1994, 1999); and Beskok, Karniadakis, and Trimmer (1996). Here we present se-
lected examples of the experimental and numerical results.
Tison (1993) conducted pipe owexperiments at very lowpressures. His pipe had
a diameter of 2 mm and a length-to-diameter ratio of 200. Both inlet and outlet pres-
sures were varied to yield Knudsen numbers in the range of Kn =0200. Figure 13.4
shows the variation of mass ow rate as a function of ( p
2
i
p
2
o
), where p
i
is the inlet
pressure and p
o
is the outlet pressure.
5
The pressure drop in this rareed pipe ow
is nonlinear, which is characteristic of low-Reynolds-number, compressible ows.
Three distinct ow regimes are identied: (1) the slip-ow regime, 0<Kn <0.6;
(2) transition regime, 0.6<Kn <17, where the mass ow rate is almost constant
as the pressure changes; and (3) free-molecule ow, Kn >17. Note that the de-
marcation between these three regimes is slightly different from that mentioned in
Section 13.2.2. As stated, the different Knudsen number regimes are determined
empirically and are therefore only approximate for a particular ow geometry.
Shih et al. (1995) conducted their experiments in a microchannel using helium
as a uid. The inlet pressure varied, but the duct exit was atmospheric. Microsen-
sors where fabricated in situ along their MEMS channel to measure the pressure.
5
The original data in this gure were acquired by S.A. Tison and plotted by Beskok et al. (1996).
Figure 13.5: Mass ow rate versus inlet pressure in a microchannel [after Shih
et al. 1995].
Figure 13.5 shows their measured mass ow rate versus the inlet pressure. The data
are compared to the no-slip solution and the slip solution using three different val-
ues of the tangential-momentum-accommodation coefcient, 0.8, 0.9, and 1.0. The
agreement is reasonable with the case
v
= 1.0, indicating perhaps that the channel
used by Shih et al. was quite rough on the molecular scale. In a second experiment
(Shih et al. 1996), nitrous oxide was used as the uid. The square of the pressure distri-
bution along the channel is plotted in Figure 13.6 for ve different inlet pressures. The
experimental data (symbols) compare well with the theoretical predictions (solid
lines). Again, the nonlinear pressure drop shown indicates that the gas ow is com-
pressible.
Arkilic (1997) provided an elegant analysis of the compressible, rareed ow
in a microchannel. The results of his theory are compared with the experiments
of Pong et al. (1994) in Figure 13.7. The dotted line is the incompressible ow
Figure 13.6: Pressure distribution of nitrous oxide in a microduct. Solid
lines are theoretical predictions [after Shih et al. 1996].
Figure 13.7: Pressure distribution in a long microchannel. The symbols are
experimental data while the solid lines are different theoretical predictions [after
Arkilic 1997].
solution, where the pressure is predicted to drop linearly with streamwise distance.
The dashed line is the compressible ow solution that neglects rarefaction effects
(assumes Kn = 0). Finally, the solid line is the theoretical result that takes into account
both compressibility and rarefaction via slip-ow boundary condition computed at
the exit Knudsen number of Kn = 0.06. That theory compares most favorably with
the experimental data. In the compressible ow through the constant-area duct,
density decreases, and thus velocity increases in the streamwise direction. As a result,
the pressure distribution is nonlinear with negative curvature. A moderate Knudsen
number (i.e., moderate slip) actually diminishes, albeit rather weakly, this curvature.
Thus, compressibility and rarefaction effects lead to opposing trends, as pointed out
by Beskok et al. (1996).
13.2.6 Molecular-Based Models
In the continuum models discussed in Section 13.2.3, the macroscopic uid prop-
erties are the dependent variables, whereas the independent variables are the three
spatial coordinates and time. The molecular models recognize the uid as a myriad of
discrete particles: molecules, atoms, ions, and electrons. The goal here is to determine
the position, velocity, and state of all particles at all times. The molecular approach
is either deterministic or probabilistic (refer to Figure 13.2). Provided that there is a
sufcient number of microscopic particles within the smallest signicant volume of
a ow, the macroscopic properties at any location in the ow can then be computed
from the discrete-particle information by a suitable averaging or weighted averaging
process. This subsection discusses molecular-based models and their relation to the
continuum models previously considered.
The most fundamental of the molecular models is a deterministic one. The motion
of the molecules is governed by the laws of classical mechanics, although, at the
expense of greatly complicating the problem, the laws of quantummechanics can also
be considered in special circumstances. The modern molecular dynamics computer
simulations (MD) have been pioneered by Alder and Wainwright (1957, 1958, 1970)
and reviewed by Ciccotti and Hoover (1986); Allen and Tildesley (1987); Haile
(1993); and Koplik and Banavar (1995). The simulation begins with a set of N
molecules in a region of space, each assigned a random velocity corresponding to
a Boltzmann distribution at the temperature of interest. The interaction between
the particles is prescribed typically in the form of a two-body potential energy, and
the time evolution of the molecular positions is determined by integrating Newtons
equations of motion. Because MD is based on the most basic set of equations, it is
valid in principle for any ow extent and any range of parameters. The method is
straightforward in principle, but there are two hurdles: (1) the choice of a proper and
convenient potential for particular uid and solid combinations and (2) the colossal
computer resources required to simulate a reasonable oweld extent.
For purists, the former difculty is a sticky one. There is no totally rational
methodology by which a convenient potential can be chosen. Part of the art of MD is
to pick an appropriate potential and validate the simulation results with experiments
or other analytical or computational results. A commonly used potential between
two molecules is the generalized LennardJones 612 potential to be used in Sec-
tion 13.2.7 and further discussed in Section 13.2.8.
The second difculty, and by far the most serious limitation of molecular dynamics
simulations, is the number of molecules Nthat can realistically be modeled on a digital
computer. Because the computation of an element of trajectory for any particular
molecule requires consideration of all other molecules as potential collision partners,
the amount of computation required by the MDmethod is proportional to N
2
. Some
saving in computer time can be achieved by cutting off the weak tail of the potential
(see Figure 13.12) at, say, r
c
= 2.5 and shifting the potential by a linear term in r so
that the force goes smoothly to zero at the cutoff. As a result, only nearby molecules
are treated as potential collision partners, and the computation time for Nmolecules
no longer scales with N
2
.
The state of the art of molecular dynamics simulations in the 1990s is such that
with a few hours of CPU time, general-purpose supercomputers can handle around
10,000 molecules. At enormous expense, the fastest parallel machine available can
simulate around 1 million particles. Because of the extreme diminution of molecu-
lar scales, these capabilities translate into regions of liquid ow of about 0.01 m
(100

A) in linear size over time intervals of around 0.001 s, which is just enough
for continuum behavior to set in for simple molecules. To simulate 1 s of real time
for complex molecular interactions (e.g., including vibration modes, reorientation
of polymer molecules, collision of colloidal particles, etc.) requires unrealistic CPU
time measured in thousands of years.
The MD simulations are highly inefcient for dilute gases, for their molecular
interactions are infrequent. These simulations are more suited for dense gases and
liquids. Clearly, molecular dynamics simulations are reserved for situations in which
the continuumapproach or the statistical methods are inadequate to compute impor-
tant ow quantities from rst principles. Slip boundary conditions for liquid ows
in extremely small devices is such a case, as will be discussed in Section 13.2.7.
An alternative to the deterministic molecular dynamics is the statistical approach
in which the goal is to compute the probability of nding a molecule at a particular
position and state. If the appropriate conservation equation can be solved for the
probability distribution, important statistical properties such as the mean number,
momentum, or energy of the molecules within an element of volume can be computed
froma simple weighted averaging. In a practical problem, it is such average quantities
that concern us rather than the detail for every single molecule. Clearly, however, the
accuracy of computing average quantities, via the statistical approach, improves as
the number of molecules in the sampled volume increases. The kinetic theory of dilute
gases is well advanced, but that for dense gases and liquids is much less so owing to
the extreme complexity of having to include multiple collisions and intermolecular
forces in the theoretical formulation. The statistical approach is well covered in books
such as those by Kennard (1938); Hirschfelder, Curtiss, and Bird (1954); Schaaf and
Chambr e (1961); Vincenti and Kruger (1965); Kogan (1969); Chapman and Cowling
(1970); Cercignani (1988); and Bird (1994) as well as in review articles such as those
by Kogan (1973); Muntz (1989); and Oran, Oh, and Cybyk (1998).
In the statistical approach, the fraction of molecules in a given location and state
is the sole dependent variable. The independent variables for monatomic molecules
are time, the three spatial coordinates, and the three components of molecular ve-
locity. These describe a six-dimensional phase space.
6
For diatomic or polyatomic
molecules, the dimension of phase space is increased by the number of internal de-
grees of freedom. Orientation adds an extra dimension for molecules that are not
spherically symmetric. Finally, for mixtures of gases, separate probability distribution
functions are required for each species. Clearly, the complexity of the approach in-
creases dramatically as the dimension of phase space increases. The simplest problems
are, for example, those for steady, one-dimensional ow of a simple monatomic gas.
To simplify the problem we restrict the discussion here to monatomic gases hav-
ing no internal degrees of freedom. Furthermore, the uid is restricted to dilute gases
and molecular chaos is assumed. The former restriction requires the average dis-
tance between molecules to be an order of magnitude larger than their diameter .
That will almost guarantee that all collisions between molecules are binary colli-
sions and thus enable one to avoid the complexity of modeling multiple encounters.
7
The molecular chaos restriction improves the accuracy of computing the macro-
scopic quantities from the microscopic information. In essence, the volume over
which averages are computed has to have a sufcient number of molecules to re-
duce statistical errors. It can be shown that computing macroscopic ow properties
by averaging over a number of molecules will result in statistical uctuations with
a standard deviation of approximately 0.1%, if one million molecules are used,
and around 3% if one thousand molecules are used. The molecular chaos limit
requires the length scale L for the averaging process to be at least 100 times the
average distance between molecules (i.e., typical averaging over at least one million
molecules).
Figure 13.8, adapted from Bird (1994), shows the limits of validity of the di-
lute gas approximation (/ >7), the continuum approach (Kn <0.1, as discussed
6
The evolution equation of the probability distribution is considered; hence, time is the seventh indepen-
dent variable.
7
Dissociation and ionization phenomena involve triple collisions and therefore require separate treatment.
Figure 13.8: Effective limits of different ow models [after Bird 1994].
previously in Section 13.2.2), and the neglect of statistical uctuations (L/ >100).
Using a molecular diameter of =4 10
10
m as an example, the three limits are
conveniently expressed as functions of the normalized gas density /
o
or number
density n/n
o
, where the reference densities
o
and n
o
are computed at standard condi-
tions. All three limits are straight lines in the loglog plot of Lversus /
o
, as depicted
in Figure 13.8. Note the shaded triangular wedge inside which both the Boltzmann
and NavierStokes equations are valid. Additionally, the lines describing the three
limits very nearly intersect at a single point. As a consequence, the continuum break-
down limit always lies between the dilute gas limit and the limit for molecular chaos.
As density or characteristic dimension is reduced in a dilute gas, the NavierStokes
model breaks down before the level of statistical uctuations becomes signicant. In
a dense gas, on the other hand, signicant uctuations may be present even when the
NavierStokes model is still valid.
The starting point in statistical mechanics is the Liouville equation, which ex-
presses the conservation of the N-particle distribution function in 6N-dimensional
phase space,
8
where Nis the number of particles under consideration. If only external
8
Three positions and three velocities for each molecule of a monatomic gas with no internal degrees of
freedom.
forces that do not depend on the velocity of the molecules,
9
the Liouville equation
for a system of N mass points reads
F
t
+
N
k=1
F
x
k
+
N
k=1
F
k
k
= 0 (13.49)
where F is the probability of nding a molecule at a particular point in phase space, t
is time,
k
is the three-dimensional velocity vector for the kth molecule, x
k
is the three-
dimensional position vector for the kth molecule, and F is the external force vector.
Note that the dot product in the preceding equation is carried out over each of the
three components of the vectors , x, and F and that the summation is over all mol-
ecules. Obviously such an equation is not tractable for a realistic number of particles.
A hierarchy of reduced distribution functions may be obtained by repeated inte-
gration of the Eq. (13.49). The nal equation in the hierarchy is for the single particle
distribution which also involves the two-particle distribution function. If molecular
chaos is assumed, that nal equation becomes a closed one (i.e., one equation in one
unknown) and is known as the Boltzmann equation, the fundamental relation of
the kinetic theory of gases. That nal equation in the hierarchy is the only one that
carries any hope of obtaining analytical solutions.
Asimpler direct derivation of the Boltzmann equation is provided by Bird (1994).
For monatomic gas molecules in binary collisions, the integro-differential Boltzmann
equation reads
(nf )
t
+
j
(nf )
x
j
+ F
j
(nf )
j
= J (f , f
), j = 1, 2, 3 (13.50)
where nf is the product of the number density and the normalized velocity distribu-
tion function (dn/n = f d), x
j
and
j
are respectively the coordinates and speeds
of a molecule,
10
F
j
is a known external force, and J (f , f
) is the nonlinear collision

integral that describes the net effect of populating and depopulating collisions on the
distribution function. The collision integral is the source of difculty in obtaining
analytical solutions to the Boltzmann equation and is given by
J ( f , f
) =
_

_
4
0
n
2
(f
1
f f
1
)
r
d(d)
1
(13.51)
indicates postcollision values, f and f

1
represent two different
molecules,
r
is the relative speed between two molecules, is the molecular cross-
section, is the solid angle, and d = d
1
d
2
d
3
.
Once a solution for f is obtained, macroscopic quantities such as density, velocity,
temperature, and so forth, can be computed from the appropriate weighted integral
of the distribution function. For example,
= mn = m
_
(nf ) d (13.52)
9
This excludes Lorentz forces, for example.
10
Constituting, together with time, the seven independent variables of the single-dependent-variable
equation.
u
i
=
_

i
f d (13.53)
3
2
kT =
_
1
2
m
i
i
f d (13.54)
If the Boltzmann equation is nondimensionalized with a characteristic length L
and characteristic speed [2(k/m)T]
1/2
, where k is the Boltzmann constant, m is the
molecular mass, and T is temperature, the inverse Knudsen number appears explicitly
in the right-hand side of the equation as follows:
t
+

f
x
j
+

F
j
j
=
1
Kn
J (

f ,

f
), j = 1, 2, 3 (13.55)
where the superscript represents a dimensionless variable, and

f is nondimension-
alized using a reference number density n
o
.
The ve conservation equations for the transport of mass, momentum, and en-
ergy can be derived by multiplying the Boltzmann equation above by, respectively,
the molecular mass, momentum, and energy and then integrating over all possible
molecular velocities. Subject to the restrictions of dilute gas and molecular chaos
stated earlier, the Boltzmann equation is valid for all ranges of Knudsen number
from 0 to . Analytical solutions to this equation for arbitrary geometries are dif-
cult mostly because of the nonlinearity of the collision integral. Simple models of
this integral have been proposed to facilitate analytical solutions; see, for example,
Bhatnagar, Gross, and Krook (1954).
There are two important asymptotes to Eq. (13.55). First, as Kn , molec-
ular collisions become unimportant. This is the free-molecule ow regime depicted
in Figure 13.3 for Kn > 10, where the only important collision is that between a
gas molecule and the solid surface of an obstacle or a conduit. Analytical solutions
are then possible for simple geometries, and numerical simulations for complicated
geometries are straightforward once the surface-reection characteristics are accu-
rately modeled. Secondly, as Kn 0, collisions become important, and the ow
approaches the continuum regime of conventional uid dynamics. The second law
species a tendency for thermodynamic systems to revert to the equilibrium state,
smoothing out any discontinuities in macroscopic ow quantities. The number of
molecular collisions in the limit Kn 0 is so large that the ow approaches the
equilibrium state in a time short compared with the macroscopic time scale. For
example, for air at standard conditions (T = 288 K; p = 1 atm), each molecule
experiences, on the average, 10 collisions per nanosecond and travels 1 micron in
the same time period. Such a molecule has already forgotten its previous state after
1 ns. In a particular oweld, if the macroscopic quantities vary little over a distance
of 1 m or over a time interval of 1 ns, the ow of air at standard temperature and
pressure is near equilibrium.
At Kn = 0, the velocity distribution function is everywhere of the local equilib-
rium or Maxwellian form
f
(0)
=
n
n
o
3/2
exp[(
u)
2
] (13.56)
where

and u are, respectively, the dimensionless speeds of a molecule and of the
ow. In this Knudsen number limit, the velocity distribution of each element of the
uid instantaneously adjusts to the equilibrium thermodynamic state appropriate to
the local macroscopic properties as this molecule moves through the oweld. From
the continuum viewpoint, the ow is isentropic, and heat conduction and viscous
diffusion and dissipation vanish from the continuum conservation relations.
The ChapmanEnskog theory attempts to solve the Boltzmann equation by con-
sidering a small perturbation of

f from the equilibrium Maxwellian form. For small
Knudsen numbers, the distribution function can be expanded in terms of Kn in the
form of a power series
f =

f
(0)
+Kn

f
(1)
+Kn
2

f
(2)
+ (13.57)
By substituting the preceding series in the Boltzmann equation (13.55) and equating
terms of equal order, the following recurrent set of integral equations results:
J
_

f
(0)
,

f
(0)
_
= 0,

J
_

f
(0)
,

f
(1)
_
=

f
(0)
t
+

f
(0)
x
j
+

F
j

f
(0)
j
, . . . (13.58)
The rst integral is nonlinear, and its solution is the local Maxwellian distribution,
Eq. (13.56). The distribution functions

f
(1)
,

f
(2)
, and so forth each satisfy an in-
homogeneous linear equation whose solution leads to the transport terms needed
to close the continuum equations appropriate to the particular level of approxima-
tion. The continuum stress tensor and heat ux vector can be written in terms of
the distribution function, which in turn can be specied in terms of the macroscopic
velocity and temperature and their derivatives (Kogan 1973). The zeroth-order equa-
tion yields the Euler equations, the rst-order equation results in the linear transport
terms of the NavierStokes equations, the second-order equation gives the nonlinear
transport terms of the Burnett equations, and so on. Keep in mind, however, that
the Boltzmann equation as developed in this subsection is for a monatomic gas. This
excludes the all-important air, which is composed largely of diatomic nitrogen and
oxygen.
As discussed in Sections 13.2.2, 13.2.3, and 13.2.5, the NavierStokes equations
can and should be used up to a Knudsen number of 0.1. Beyond that, the transition
owregime commences (0.1 <Kn <10). In this owregime, the molecular mean free
path for a gas becomes signicant relative to a characteristic distance for important
ow-property changes to take place. The Burnett equations can be used to obtain
analytical and numerical solutions for at least a portion of the transition regime for
a monatomic gas, although their complexity has precluded realizing many results for
realistic geometries (Agarwal, Yun, and Balakrishnan 1999). There is also some de-
gree of uncertainty about the proper boundary conditions to use with the continuum
Burnett equations, and experimental validations of the results have been very scarce.
Additionally, as the gas ow further departs from equilibrium, the bulk viscosity
(= +
2
3
, where is the second coefcient of viscosity) is no longer zero, and the
Stokes hypothesis no longer holds (see Gad-el-Hak 1995 for an interesting summary
of the issue of bulk viscosity).
In the transition regime, the molecularly based Boltzmann equation cannot easily
be solved either unless the nonlinear collision integral is simplied. Thus, clearly the
transition regime is in dire need of alternative solution methods. The MDsimulations,
as mentioned earlier, are not suited for dilute gases. The best approach for the tran-
sition regime right now is the direct simulation Monte Carlo (DSMC) method de-
veloped by Bird (1963, 1965, 1976, 1978, 1994) and briey described below. Some
recent reviews of DSMC include those by Muntz (1989), Cheng (1993), Cheng and
Emmanuel (1995), and Oran et al. (1998). The mechanics as well as the history of
the DSMC approach and its ancestors are well described in the book by Bird (1994).
Unlike molecular dynamics simulations, DSMC is a statistical computational ap-
proach to solving rareed gas problems. Both approaches treat the gas as discrete par-
ticles. Subject to the dilute gas and molecular chaos assumptions, the direct simulation
Monte Carlo method is valid for all ranges of Knudsen number, although the method
becomes quite expensive for Kn <0.1. Fortunately, this is the continuum regime in
which the NavierStokes equations can be used analytically or computationally. The
DSMC method is therefore ideal for the transition regime (0.1 <Kn <10), where
the Boltzmann equation is difcult to solve. The Monte Carlo method is, like its
name sake, a random number strategy based directly on the physics of the individual
molecular interactions. The idea is to track a large number of randomly selected, sta-
tistically representative particles and to use their motions and interactions to modify
their positions and states. The primary approximation of the DSMC method is to
uncouple the molecular motions and the intermolecular collisions over small time
intervals. A signicant advantage of this approximation is that the amount of com-
putation required is proportional to N, in contrast to N
2
, for molecular dynamics
simulations. In essence, particle motions are modeled deterministically, whereas col-
lisions are treated probabilistically, each simulated molecule representing a large
number of actual molecules. Typical computer runs of DSMC in the 1990s involve
tens of millions of intermolecular collisions and uidsolid interactions.
The DSMC computation is started from some initial condition and followed in
small time steps that can be related to physical time. Colliding pairs of molecules in a
small geometric cell in physical space are randomly selected after each computational
time step. Complex physics such as radiation, chemical reactions, and species concen-
trations can be included in the simulations without the necessity of nonequilibrium
thermodynamic assumptions that commonly afict nonequilibrium continuum-ow
calculations. Because DSMC is more computationally intensive than classical con-
tinuum simulations, it should therefore be used only when the continuum approach
is not feasible.
The DSMC technique is explicit and time marching and therefore always pro-
duces unsteady ow simulations. For macroscopically steady ows, Monte Carlo
simulation proceeds until a steady ow is established within a desired accuracy at
sufciently large time. The macroscopic ow quantities are then the time average of
all values calculated after reaching the steady state. For macroscopically unsteady
ows, ensemble averaging of many independent Monte Carlo simulations is carried
out to obtain the nal results within a prescribed statistical accuracy.
13.2.7 Liquid Flows
From the continuum point of view, liquids and gases are both uids obeying
the same equations of motion. For incompressible ows, for example, the Reynolds
number is the primary dimensionless parameter that determines the nature of the
oweld. It is true that water, for example, has density and viscosity that are, respec-
tively, three and two orders of magnitude higher than those for air, but if the Reynolds
number and geometry are matched, liquid and gas ows should be identical.
11
For
MEMS applications, however, we anticipate the possibility of nonequilibrium ow
conditions and the consequent invalidity of the NavierStokes equations and the
no-slip boundary conditions. Such circumstances can best be researched using the
molecular approach. This was discussed for gases in Section 13.2.6, and the corre-
sponding arguments for liquids will be given in this subsection. The literature on
non-Newtonian uids in general and polymers in particular is vast (for example,
the bibliographic survey by Nadolink and Haigh, 1995, cites over 4900 references
on polymer drag reduction alone) and provides a rich source of information on the
molecular approach for liquid ows.
Solids, liquids, and gases are distinguished merely by the degree of proximity
and the intensity of motions of their constituent molecules. In solids, the molecules
are packed closely and conned, each hemmed in by its neighbors (Chapman and
Cowling 1970). Only rarely would one solid molecule slip from its neighbors to
join a new set. As the solid is heated, molecular motion becomes more violent, and
a slight thermal expansion takes place. At a certain temperature that depends on
ambient pressure, sufciently intense motion of the molecules enables them to pass
freely fromone set of neighbors to another. The molecules are no longer conned but
are nevertheless still closely packed, and the substance is now considered a liquid.
Further heating of the matter eventually releases the molecules altogether, allowing
them to break the bonds of their mutual attractions. Unlike solids and liquids, the
resulting gas expands to ll any volume available to it.
Unlike solids, both liquids and gases cannot resist nite shear force without con-
tinuous deformation, that is, the denition of a uid medium. In contrast to the
reversible, elastic, static deformation of a solid, the continuous deformation of a
uid resulting from the application of a shear stress results in an irreversible work
that eventually becomes random thermal motion of the molecules, that is, viscous
dissipation. In a 1-m cube, there are around 25 million molecules of air at standard
temperature and pressure. The same cube would contain around 34 billion molecules
of water. Thus, liquid ows are continuum even in extremely small devices through
which gas ows would not. The average distance between molecules in the gas ex-
ample is one order of magnitude higher than the diameter of its molecules, whereas
that for the liquid phase approaches the molecular diameter. As a result, liquids are
almost incompressible. Their isothermal compressibility coefcient and bulk ex-
pansion coefcient are much smaller compared with those for gases. For water, for
example, a hundredfold increase in pressure leads to a less than 0.5% decrease in
volume. Sound speeds through liquids are also high relative to those for gases, and
as a result most liquid ows are incompressible.
12
The exception is propagation of
ultra-high-frequency sound waves and cavitation phenomena.
The mechanismby which liquids transport mass, momentum, and energy must be
very different from that for gases. In dilute gases, intermolecular forces play no role,
and the molecules spend most of their time in free ight between brief collisions during
which the molecules direction and speed abruptly change. The random molecular
motions are responsible for gaseous transport processes. In liquids, on the other
11
Barring phenomena unique to liquids such as cavitation, free surface ows, and so forth.
12
Note that we distinguish between a uid and a owbeing compressible or incompressible. For example,
the ow of the highly compressible air can be either compressible or incompressible.
hand, the molecules are closely packed though not xed in one position. In essence,
the liquid molecules are always in a collision state. Applying a shear force must create
a velocity gradient so that the molecules move relative to one another ad innitum
as long as the stress is applied. For liquids, momentum transport due to the random
molecular motion is negligible compared with that due to the intermolecular forces.
The straining between liquid molecules causes some to separate from their original
neighbors, bringing them into the force eld of new molecules. Across the plane of
the shear stress, the sum of all intermolecular forces must, on the average, balance
the imposed shear. Liquids at rest transmit only normal force, but when a velocity
gradient occurs, the net intermolecular force will have a tangential component.
The incompressible NavierStokes equations describe liquid ows under most
circumstances. Liquids, however, do not have as well-advanced a molecular-based
theory as that for dilute gases. The concept of mean free path is not very useful for
liquids, and the conditions under which a liquid ow fails to be in quasi-equilibrium
state are not well dened. There is no Knudsen number for liquid ows to guide
us through the maze. We do not know, from rst principles, the conditions under
which the no-slip boundary condition becomes inaccurate or the point at which the
(stress)(rate of strain) relation or the (heat ux)(temperature gradient) relation fails
to be linear. Certain empirical observations indicate that those simple relations that
we take for granted occasionally fail to model liquid ows accurately. For example,
it has been shown in rheological studies (Loose and Hess 1989) that non-Newtonian
behavior commences when the strain rate approximately exceeds twice the molecular
frequency-scale
=
u
y
2
1
(13.59)
where the molecular time scale is given by
=
_
m
2
_
1
2
(13.60)
where m is the molecular mass, and and are, respectively, the characteristic
length and energy scales for the molecules. For ordinary liquids such as water, this
time scale is extremely small, and the threshold shear rate for the onset of non-
Newtonian behavior is therefore extraordinarily high. For high-molecular-weight
polymers, on the other hand, m and are both many orders of magnitude higher
than their respective values for water, and the linear stress-strain relation breaks
down at realistic values of the shear rate.
The moving contact line, when a liquid spreads on a solid substrate, is an example
for which slip ow must be allowed to avoid singular or unrealistic behavior in the
NavierStokes solutions (Dussan and Davis 1974; Dussan 1976, 1979; Thompson
and Robbins 1989). Other examples for which slip ow must be admitted include
corner ows (Moffatt 1964; Koplik and Banavar 1995) and extrusion of polymer
melts from capillary tubes (Pearson and Petrie 1968; Richardson 1973; Den 1990).
Existing experimental results of liquid ow in microdevices are contradictory.
This is not surprising given the difculty of such experiments and the lack of a
guiding rational theory. Pfahler et al. (1990, 1991), Pfahler (1992), and Bau (1994)
summarize the relevant literature. For small-length-scale ows, a phenomenological
approach for analyzing the data is to dene an apparent viscosity
a
calculated so
that if it were used in the traditional no-slip NavierStokes equations instead of the
uid viscosity , the results would be in agreement with experimental observations.
Israelachvili (1986) and Gee et al. (1990) found that
a
= for thin-lm ows as
long as the lm thickness exceeds 10 molecular layers (5 nm). For thinner lms,
a
depends on the number of molecular layers and can be as much as 10
5
times
larger than . Chan and Horns (1985) results are somewhat different; the apparent
viscosity deviates from the uid viscosity for lms thinner than 50 nm.
In polar-liquid ows through capillaries, Migun and Prokhorenko (1987) re-
ported that
a
increases for tubes smaller than 1 m in diameter. In contrast, Debye
and Cleland (1959) reported
a
smaller than for parafn ow in porous glass with
average pore size several times larger than the molecular length scale. Experimenting
with microchannels ranging in depths from 0.5 to 50 m, Pfahler et al. (1991) found
that
a
is consistently smaller than for both liquid (isopropyl alcohol, silicone oil)
and gas (nitrogen, helium) ows in microchannels. For liquids, the apparent viscos-
ity decreases with decreasing channel depth. Other researchers using small capillaries
report that
a
is about the same as (Anderson and Quinn 1972; Tuckermann and
Pease 1981, 1982; Tuckermann 1984; Guvenc 1985; Nakagawa, Shoji, and Esashi
1990).
The preceding contradictory results point to the need for replacing phenomeno-
logical models by rst-principles ones. The lack of a molecular-based theory of
liquidsdespite extensive research by the rheology and polymer communities
leaves molecular dynamics simulations as the nearest weapon to the rst-principles
arsenal. Molecular dynamics simulations offer a unique approach to checking the
validity of the traditional continuum assumptions. However, as was pointed out in
Section 13.2.6, such simulations are limited to exceedingly minute ow extent.
Thompson and Troian (1997) provided molecular dynamics simulations to quan-
tify the slip-ow boundary condition dependence on shear rate. Recall the linear
Navier boundary condition introduced in Section 13.2.5,
u
w
= u
uid
u
wall
= L
s
u
y
w
(13.61)
where L
s
is the constant slip length, and
u
y
|
w
is the strain rate computed at the
wall. The goal of Thompson and Troians simulations was to determine the degree of
slip at a solidliquid interface as the interfacial parameters and the shear rate change.
In their simulations, a simple liquid underwent planar shear in a Couette cell as shown
in Figure 13.9. The typical cell measured 12.517.22h in units of molecular length
scale , where the channel depth h varied in the range of 16.7124.57, and the
corresponding number of molecules simulated ranged from1152 to 1728. The liquid
was treated as an isothermal ensemble of spherical molecules. A shifted Lennard
Jones 612 potential was used to model intermolecular interactions with energy and
length scales and and cut-off distance r
c
= 2.2:
V(r) = 4
__
r
_
12
_
r
_
6
_
r
c
_
12
+
_
r
c
_
6
_
(13.62)
The truncated potential was set to zero for r > r
c
.
Figure 13.9: Velocity proles in a Couette ow geometry at differ-
ent interfacial parameters. All three proles are for U =
1
, and
h = 24.57. The dashed line is the no-slip Couette-ow solution
[after Thompson and Troian 1997].
The uidsolid interaction was also modeled with a truncated LennardJones po-
tential with energy and length scales
wf
and
wf
, and cut-off distance r
c
. The equilib-
riumstate of the uid was a well-dened liquid phase characterized by number density
n = 0.81
3
and temperature T = 1.1/k, where k is the Boltzmann constant.
The steady-state velocity proles resulting from Thompson and Troians (1997)
MD simulations are depicted in Figure 13.9 for different values of the interfacial
parameters
wf
,
wf
, and n
w
. These parameters, shown in units of the corresponding
uid parameters , , and n, characterize, respectively, the strength of the liquid
solid coupling, the thermal roughness of the interface, and the commensurability
of wall and liquid densities. The macroscopic velocity proles recover the expected
ow behavior from continuum hydrodynamics with boundary conditions involving
varying degrees of slip. Note that when slip exists, the shear rate no longer equals
U/h. The degree of slip increases (i.e., the amount of momentum transfer at the
walluid interface decreases) as the relative wall density n
w
increases or the strength
of the walluid coupling
wf
decreases; in other words, when the relative surface
energy corrugation of the wall decreases. Conversely, the corrugation is maximized
when the wall and uid densities are commensurate and the strength of the wall
uid coupling is large. In this case, the liquid feels the corrugations in the surface
energy of the solid owing to the atomic close-packing. Consequently, there is efcient
momentum transfer and the no-slip condition applies, or in extreme cases, a stick
boundary condition takes hold.
Figure 13.10: Variation of slip length and viscosity as functions of shear rate
[after Thompson and Troian 1997].
Variations of the slip length L
s
and viscosity as functions of shear rate
are shown in parts (a) and (b) of Figure 13.10, for ve different sets of interfa-
cial parameters. For Couette ow, the slip length is computed from its denition,
L
s
= u|
w
/ = (U/ h)/2. The slip length, viscosity, and shear rate are normal-
ized in the gure using the respective molecular scales for length , viscosity
3
,
and inverse time
1
. The viscosity of the uid is constant over the entire range
of shear rates (Figure 13.10b), indicating Newtonian behavior. As indicated earlier,
non-Newtonian behavior is expected for 2
1
, which is well above the shear
rates used in Thompson and Troians simulations.
At low shear rates, the slip-length behavior is consistent with the Navier model
(i.e., is independent of the shear rate). Its limiting value L
o
s
ranges from0 to 17 for
the range of interfacial parameters chosen (Figure 13.10a). In general, the amount
of slip increases with decreasing surface energy corrugation. Most interestingly, at
high shear rates the Navier condition breaks down as the slip length increases rapidly
with . The critical shear-rate value for the slip length to diverge,
c
, decreases as
the surface energy corrugation decreases. Surprisingly, the boundary condition is
nonlinear even though the liquid is still Newtonian. In dilute gases, as discussed in
Section 13.2.6, the linear slip condition and the NavierStokes equations, with their
Figure 13.11: Universal relation of slip length as a function of shear
rate [after Thompson and Troian 1997].
linear stressstrain relation, are both valid to the same order of approximation in
Knudsen number. In other words, deviation from linearity is expected to take place
at the same value of Kn = 0.1. In liquids, in contrast, the slip length appears to
become nonlinear and to diverge at a critical value of shear rate well below the shear
rate at which the linear stressstrain relation fails. Moreover, the boundary condition
deviation from linearity is not gradual but is rather catastrophic. The critical value
of shear rate
c
signals the point at which the solid can no longer impart momentum
to the liquid. This means that the same liquid molecules sheared against different
substrates will experience varying amounts of slip and vice versa.
On the basis of the preceding results, Thompson and Troian (1997) suggested a
universal boundary condition at a solidliquid interface. Scaling the slip length L
s
by
its asymptotic limiting value L
o
s
and the shear rate by its critical value collapses
the data in the single curve shown in Figure 13.11. The data points are well described
by the relation
L
s
= L
o
s
_
1

c
_
1
2
(13.63)
The nonlinear behavior close to a critical shear rate suggests that the boundary
condition can signicantly affect ow behavior at macroscopic distances from the
wall. Experiments with polymers conrm this observation (Atwood and Schwalter
1989). The rapid change in the slip length suggests that for ows in the vicinity of

c
, small changes in surface properties can lead to large uctuations in the apparent
boundary condition. Thompson and Troian (1997) concluded that the Navier slip
condition is but the low-shear-rate limit of a more generalized universal relationship
that is nonlinear and divergent. Their relation provides a mechanismfor relieving the
stress singularity in spreading contact lines and corner ows, for it naturally allows
for varying degrees of slip on approach to regions of higher rate of strain.
To place the preceding results in physical terms, consider water
13
at a tem-
perature of T = 288 K. The energy scale in the LennardJones potential is then
= 3.62 10
21
J. For water, m = 2.99 10
26
kg, = 2.89 10
10
m, and at
standard temperature n = 3.35 10
28
molecules/m
3
. The molecular time scale can
thus be computed, = [m
2
/]
1/2
= 8.31 10
13
s. For the third case depicted
in Figure 13.11 (the open squares),
c
= 0.1, and the critical shear rate at which
the slip condition diverges is thus
c
= 1.2 10
11
s
1
. Such an enormous rate of
strain
14
may be found in extremely small devices having extremely high speeds. On
the other hand, the conditions to achieve a measurable slip of 17 (the solid circles
in Figure 13.10) are not difcult to encounter in microdevices: density of the solid
four times that of liquid and energy scale for walluid interaction that is one-fth
of the energy scale for liquid.
The limiting value of slip length is independent of the shear rate and can be
computed for water as L
o
s
= 17 = 4.91 10
9
m. Consider a water microbearing
having a shaft diameter of 100 m, rotation rate of 20,000 rpm, and a minimum gap
of h = 1 m. In this case, U = 0.1 m/s, and the no-slip shear rate is U/h = 10
5
s
1
.
When slip occurs at the limiting value just computed, the shear rate and the wall-slip
velocity are computed as follows:
=
U
h +2L
o
s
= 9.90 10
4
s
1
(13.64)
u|
w
= L
s
= 4.87 10
4
m/s (13.65)
As a result of the Navier slip, the shear rate is reduced by 1% from its no-slip
value, and the slip velocity at the wall is about 0.5% of U, which is small but not
insignicant.
13.2.8 Surface Phenomena
As mentioned in Section 13.2.1, the surface-to-volume ratio for a machine with a
characteristic length of 1 m is 1 m
1
, whereas that for a MEMS device having a size
of 1 m is 10
6
m
1
. The millionfold increase in surface area relative to the mass of
the minute device substantially affects the transport of mass, momentum, and energy
through the surface. Obviously, surface effects dominate in small devices. The surface
boundary conditions in MEMS ows have already been discussed in Sections 13.2.5
and 13.2.7. In microdevices, it has been shown that it is possible to have measurable
slip velocity and temperature jump at a soliduid interface. In this subsection, we
illustrate other ramications of the large surface-to-volume ratio unique to MEMS
and provide a molecular viewpoint for surface forces.
In microdevices, both radiative and convective heat loss or gain are enhanced by
the huge surface-to-volume ratio. Consider a device having a characteristic length L
s
.
Use of the lumped capacitance method to compute the rate of convective heat transfer,
for example, is justied if the Biot number (hL
s
/
s
, where h is the convective heat
13
Water molecules are complex ones forming directional, short-range covalent bonds and thus require
a more complex potential than the LennardJones to describe the intermolecular interactions. For
the purpose of the qualitative example described here, however, we use the computational results of
Thompson and Troian (1997), who employed the LJ potential.
14
Note however that
c
for high-molecular-weight polymers would be many orders of magnitude smaller
than the value developed here for water.
transfer coefcient of the uid and
s
is the thermal conductivity of the solid) is
less than 0.1. Small L
s
implies small Biot number and a nearly uniform temperature
within the solid. Within this approximation, the rate at which heat is lost to the
surrounding uid is given by
s
L
3
s
c
s
dT
s
dt
= hL
2
s
(T
s
T
) (13.66)
where
s
and c
s
are, respectively, the density and specic heat of the solid, T
s
is its (uni-
form) temperature, and T
is the ambient uid temperature. Solution of Eq. (13.66)

is trivial, and the temperature of a hot surface drops exponentially with time from
an initial temperature T
i
,
T
s
(t) T
T
i
T
= exp
_
t
T
_
(13.67)
where the time constant T is given by
T =

s
L
3
s
c
s
hL
2
s
(13.68)
For small devices, the time it takes the solid to cool down is proportionally small.
Clearly, the millionfold increase in surface-to-volume ratio implies a proportional
increase in the rate at which heat escapes. Identical scaling arguments can be made
regarding mass transfer.
Another effect of the diminished scale is the increased importance of surface forces
and the waning importance of body forces. On the basis of biological studies, Went
(1968) concluded that the demarcation length scale is around 1 mm. Below that,
surface forces dominate over gravitational forces. A 10-mm piece of paper will fall
down when gently placed on a smooth, vertical wall, whereas a 0.1-mm piece will
stick. Try it! Stiction is a major problem in MEMS applications. Certain structures
such as long, thin polysilicon beams and large, thin comb drives have a propensity to
stick to their substrates and thus fail to perform as designed (Mastrangelo and Hsu
1992; Tang, Nguyen, and Howe 1989).
Conventional dry friction between two solids in relative motion is proportional
to the normal force, which is usually a component of the moving device weight.
The friction is independent of the contact-surface area because the van der Waals
cohesive forces are negligible relative to the weight of the macroscopic device. In
MEMS applications, the cohesive intermolecular forces between two surfaces are
signicant, and the stiction is independent of the device mass but is proportional to
its surface area. The rst micromotor did not movedespite large electric current
through ituntil the contact area between the 100-micron rotor and the substrate
was reduced signicantly by placing dimples on the rotors surface (Fan, Tai, and
Muller 1988, 1989; Tai and Muller 1989).
One last example of surface effects that to my knowledge has not been investigated
for microows is the adsorbed layer in gaseous wall-bounded ows. It is well known
(see, for example, Brunauer 1944; Lighthill 1963b) that when a gas ows in a duct,
the gas molecules are attracted to the solid surface by the van der Waals and other
forces of cohesion. The potential energy of the gas molecules drops upon reaching the
surface. The adsorbed layer partakes of the thermal vibrations of the solid, and the gas
molecules can only escape when their energy exceeds the potential energy minimum.
In equilibrium, at least part of the solid would be covered by a monomolecular
layer of adsorbed gas molecules. Molecular species with signicant partial pressure
relative to their vapor pressuremay locally formlayers two or more molecules thick.
Consider, for example, the ow of a mixture of dry air and water vapor at standard
temperature and pressure. The energy of adsorption of water is much larger than that
for nitrogen and oxygen, making it more difcult for water molecules to escape the
potential energy trap. It follows that the lifetime of water molecules in the adsorbed
layer signicantly exceeds that for the air molecules (by 60,000 folds, in fact) and, as
a result, the thin surface layer would be mostly water. For example, if the proportion
of water vapor in the ambient air is 1:1000 (i.e., very lowhumidity level), the ratio of
water to air in the adsorbed layer would be 60:1. Microscopic roughness of the solid
surface causes partial condensation of the water along portions having sufciently
strong concave curvature. Thus, surfaces exposed to nondry air ows are mainly
liquid water surfaces. In most applications, this thin adsorbed layer has little effect
on the owdynamics despite the density and viscosity of liquid water being far greater
than those for air. In MEMS applications, however, the layer thickness may not be an
insignicant portion of the characteristic ow dimension, and the water layer may
have a measurable effect on the gas ow. A hybrid approach of molecular dynamics
and continuum ow simulations or MDMonte Carlo simulations may be used to
investigate this issue.
It should be noted that quite recently Majumdar and Mezic (1998, 1999) have
studied the stability and rupture into droplets of thin liquid lms on solid surfaces.
They pointed out that the free energy of a liquid lm consists of a surface tension
component as well as highly nonlinear volumetric intermolecular forces resulting
from van der Waals, electrostatic, hydration, and elastic strain interactions. For
water lms on hydrophilic surfaces such as silica and mica, Majumdar and Mezic
(1998) estimated the equilibrium lm thickness to be about 0.5 nm (2 monolayers)
for a wide range of ambient-air relative humidities. The equilibrium thickness grows
very sharply, however, as the relative humidity approaches 100%.
Majumdar and Mezics (1998, 1999) results open many questions. What are
the stability characteristics of their water lm in the presence of air ow above it?
Would this water lm affect the accommodation coefcient for microduct air ow?
In a modern Winchester-type hard disk, the drive mechanism has a readwrite head
that oats 50 nm above the surface of the spinning platter. The head and platter
together with the air layer in between form a slider bearing. Would the computer
performance be affected adversely by the high relative humidity on a particular day
when the adsorbed water lm is no longer thin? If a microduct hauls liquid water,
would the water lm adsorbed by the solid walls inuence the effective viscosity
of the water ow? Electrostatic forces can extend to almost 1 micron (the Debye
length), and that length is known to be highly pH-dependent. Would the water owbe
inuenced by the surface and liquid chemistry? Would this explain the contradictory
experimental results of liquid ows in microducts discussed in Section 13.2.7?
The few examples above illustrate the importance of surface effects in small de-
vices. From the continuum viewpoint, forces at a soliduid interface are the limit
of pressure and viscous forces acting on a parallel elementary area displaced into the
uid when the displacement distance is allowed to tend to zero. From the molecular
point of view, all macroscopic surface forces are ultimately traced to intermolecular
forces, which subject is extensively covered in the book by Israelachvilli (1991) and
references therein. Here we provide a very brief introduction to the molecular view-
point. The four forces in nature are (1) the strong and (2) weak forces describing
the interactions between neutrons, protons, electrons, and so forth; (3) the electro-
magnetic forces between atoms and molecules; and (4) gravitational forces between
masses. The range of action of the rst two forces is around 10
5
nm, and hence
neither concerns us overly in MEMS applications. The electromagnetic forces are ef-
fective over a much larger though still small distance on the order of the interatomic
separations (0.10.2 nm). Effects over longer rangeseveral orders of magnitude
longercan and do rise from the short-range intermolecular forces. For example,
the rise of the liquid column in capillaries and the action of detergent molecules in
removing oily dirt from fabric are the result of intermolecular interactions. Grav-
itational forces decay with the distance to second power, whereas intermolecular
forces decay much quicker, typically with the seventh power. Cohesive forces are
therefore negligible once the distance between molecules exceeds few molecular di-
ameters, whereas massive bodies like stars and planets are still strongly interacting,
via gravity, over astronomical distances.
Electromagnetic forces are the source of all intermolecular interactions and the
cohesive forces holding atoms and molecules together in solids and liquids. They can
be classied into (1) purely electrostatic forces arising from the Coulomb force be-
tween charges, interactions between charges, permanent dipoles, quadrupoles, and
so forth; (2) polarization forces arising from the dipole moments induced in atoms
and molecules by the electric eld of nearby charges and permanent dipoles; and
(3) quantum mechanical forces that give rise to covalent or chemical bonding and
to repulsive steric or exchange interactions that balance the attractive forces at very
short distances. The HellmanFeynman theorem of quantum mechanics states that
once the spatial distribution of the electron clouds has been determined by solv-
ing the appropriate Schr odinger equation, intermolecular forces may be calculated
on the basis of classical electrostatics, in effect reducing all intermolecular forces
to Coulombic forces. Note, however, that intermolecular forces exist even when
the molecules are totally neutral. Solutions of the Schr odinger equation for general
atoms and molecules are not easy of course, and alternative modeling is sought to
represent intermolecular forces. The van der Waals attractive forces are usually rep-
resented with a potential that varies as the inverse-sixth power of distance, whereas
the repulsive forces are represented with either a power or an exponential potential.
A commonly used potential between two molecules is the generalized Lennard
Jones (LJ 612) pair potential given by
V
i j
(r) = 4
_
c
i j
_
r
_
12
d
i j
_
r
_
6
_
(13.69)
where V
i j
is the potential energy between two particles i and j , r is the distance
between the two molecules, and are respectively characteristic energy and length
scales, and c
i j
and d
i j
are parameters to be chosen for the particular uid and solid
combinations under consideration. The rst term on the right-hand side is the strong
repulsive force that is felt when two molecules are at extremely close range compara-
ble to the molecular length scale. That short-range repulsion prevents overlap of the
molecules in physical space. The second term is the weaker van der Waals attractive
force that commences when the molecules are sufciently close (several times ).
That negative part of the potential represents the attractive polarization interaction
of neutral, spherically symmetric particles. The power of 6 associated with this term
is derivable from quantum mechanics considerations, whereas the power of the re-
pulsive part of the potential is found empirically. The LennardJones potential is zero
at very large distances, has a weak negative peak at r slightly larger than , is zero
at r = , and is innite as r 0.
The force eld resulting from this potential is given by
F
i j
(r) =
V
i j
r
=
48
_
c
i j
_
r
_
13
d
i j
2
_
r
_
7
_
(13.70)
Atypical LJ 612 potential and force eld are shown in Figure 13.12, for c = d = 1.
The minimum potential V
min
= corresponds to the equilibrium position (zero
force) and occurs at r = 1.12. The attractive van der Waals contribution to the
minimum potential is 2, whereas the repulsive energy contribution is +. Thus,
the inverse 12th-power repulsive force term decreases the strength of the binding
energy at equilibrium by 50%.
The LJ potential is commonly used in molecular dynamics simulations to model
intermolecular interactions between dense gas or liquid molecules and between uid
and solid molecules. As mentioned in Section 13.2.7, such potential is not accurate
for complex substances such as water whose molecules form directional covalent
bonds. As a result, MD simulations for water are much more involved.
Figure 13.12: Typical LennardJones 612 potential and the intermolecular force eld resulting from
it. Only a small portion of the potential function is shown for clarity.
13.3 Typical Fluid Applications
The physics of uid ows in microdevices was covered in Section 13.2. In this
section, we provide several examples of useful applications of MEMS devices in uid
mechanics. The list is by no means exhaustive but includes the use of MEMS-based
sensors and actuators for ow diagnosis and control, a recently developed viscous
micropumpmicroturbine, and analysis of a journal microbearing. The articles by
Gad-el-Hak (1999) and L ofdahl and Gad-el-Hak (1999) offer more detail on some
of the topics covered in this section. From the perspective of the present monograph,
using microsensors and microactuators for owcontrol is a very important topic and
will be deferred to the following chapter.
13.3.2 Turbulence Measurements
Microelectromechanical systems offer great opportunities for better ow diagno-
sis and controlparticularly for turbulent ows. The batch processing fabrication
of microdevices makes it possible to produce large numbers of identical transducers
within extremely tight tolerance. Microsensors and microactuators are small and in-
expensive, combine electronic and mechanical parts, have low energy consumption,
and can be distributed over a wide area. In this subsection we discuss the advantages
of using MEMS-based sensors for turbulence measurements, and in the following
subsection the issue of ow control will be addressed.
Turbulence remains largely an enigmaanalytically unapproachable yet prac-
tically very important. For a turbulent ow, the dependent variables are random
functions of space and time, and no straightforward method exists for analytically
obtaining stochastic solutions to the governing nonlinear, partial differential equa-
tions. The statistical approach to solving the NavierStokes equations sets a more
modest aim of solving for the average ow quantities rather than the instantaneous
ones. But as a result of the nonlinearity of the governing equations, this approach
always leads to more unknowns than equations (the closure problem), and solutions
based on rst principles are again not possible. Turbulence, therefore, is a conun-
drum that appears to yield its secrets only to physical and numerical experiments,
provided that the wide band of relevant scales is fully resolveda far-from-trivial
task, particularly at high Reynolds numbers.
Aturbulent oweld is composed of a hierarchy of eddies having a broad range of
time and length scales. The largest eddies have a spatial extension of approximately
the same size as the width of the oweld, whereas the smallest eddies are of the size
at which viscous effects become dominant and energy is transferred from kinetic into
internal. The ratio of the smallest length scalethe Kolmogorov microscale to
the largest scale is related to the turbulence Reynolds number as follows:
_
u
3
4
= Re
3
4
(13.71)
Similar expressions can be written for time and velocity scales (see, for example,
Tennekes and Lumley 1972). Not only does a sensor have to be sufciently small
to resolve the smallest eddies, but multisensors distributed over a large volume are
needed to detect any ow structures at the largest scale. Clearly the problem worsens
as the Reynolds number increases.
13.3 TYPICAL FLUID APPLICATIONS 309
In wall-bounded ows, the shear-layer thickness provides a measure of the largest
eddies in the ow. The smallest scale is the viscous wall unit. Viscous forces dominate
over inertia in the near-wall region. The characteristic scales there are obtained from
the magnitude of the mean vorticity in the region and its viscous diffusion away from
the wall. Thus, the viscous time scale, t
, is given by the inverse of the mean wall

vorticity
t
=
_
U
y
w
_
1
(13.72)
where U is the mean streamwise velocity. The viscous length scale,
, is determined
by the characteristic distance by which the (spanwise) vorticity is diffused from the
wall and is thus given by

U
y
w
(13.73)
where is the kinematic viscosity. The wall velocity scale (so-called friction velocity
u

u
U
y
w
=
_
(13.74)
where
w
is the mean shear stress at the wall, and is the uid density. A wall unit
implies scaling with the viscous scales, and the usual ()
+
notation is used; for example,
y
+
= y/
= yu
/. In the wall region, the characteristic length for the large eddies
is y itself, whereas the Kolmogorov scale is related to the distance from the wall y as
follows:
+

u
(y
+
)
1
4
(13.75)
where is the von K arm an constant ( 0.41). As y
+
changes in the range of 15 (the
extent of the viscous sublayer), changes from 0.8 to 1.2 wall units.
It is clear from the above that the spatial and temporal resolutions for any probe
to be used to resolve high-Reynolds-number turbulent ows are extremely tight. For
example, both the Kolmogorov scale and the viscous length scale change from few
microns at the typical eld Reynolds numberbased on the momentum thickness
of 10
6
to a couple of hundred microns at the typical laboratory Reynolds number of
10
3
, and MEMS sensors for pressure, velocity, temperature and shear stress are at
least one order of magnitude smaller than conventional sensors (Ho and Tai 1996;
L ofdahl, K alvesten, and Stemme 1996). The small size of MEMS sensors improves
both the spatial and temporal resolutions of the measurements, typically by a few
microns and few microseconds, respectively. For example, a microhot-wire (called
hot point) has very small thermal inertia, and the diaphragm of a micropressure-
transducer has correspondingly fast dynamic response. Moreover, the microsensors
extreme miniaturization and low energy consumption make them ideal for moni-
toring the ow state without appreciably affecting it. Lastly, literally hundreds of
microsensors can be fabricated on the same silicon chip at a reasonable cost, making
them well suited for distributed measurements. The UCLACaltech team (see, for
example, Ho and Tai 1996, 1998 and references therein) has been very effective in
developing many MEMS-based sensors and actuators for turbulence diagnosis and
control.
13.3.3 Micropumps
There have been several studies of microfabricated pumps. Some of them use
nonmechanical effects. The Knudsen pump mentioned in Section 13.2.5 uses the
thermal-creep effect to move rareed gases from one chamber to another. Ion-drag
is used in electrohydrodynamic pumps (Bart et al. 1990; Richter et al. 1991; Fuhr
et al. 1992); these rely on the electrical properties of the uid and are thus not suitable
for many applications. Valveless pumping by ultrasound has also been proposed
(Moroney, White, and Howe 1991) but produces very little pressure difference.
Mechanical pumps based on conventional centrifugal or axial turbomachinery
will not work at micromachine scales at which the Reynolds numbers are typically
small (on the order of 1 or less). Centrifugal forces are negligible and, furthermore,
the Kutta condition through which lift is normally generated is invalid when inertial
forces are vanishingly small. In general there are three ways in which mechanical
micropumps can work:
Positive-displacement pumps. These are mechanical pumps with a membrane

or diaphragm actuated in a reciprocating mode and with unidirectional inlet
and outlet valves. They work on the same physical principle as their larger
cousins. Micropumps with piezoelectric actuators have been fabricated (Van
Lintel, Van de Pol, and Bouwstra 1988; Esashi, Shoji, and Nakano 1989;
Smits 1990). Other actuators, such as thermopneumatic, electrostatic, elec-
tromagnetic, or bimetallic, can be used (Pister, Fearing, and Howe 1990;
D oring et al. 1992; Gabriel et al. 1992). These exceedingly minute positive-
displacement pumps require even smaller valves, seals, and mechanismsa
not-too-trivial micromanufacturing challenge. In addition there are long-term
problems associated with wear or clogging and consequent leaking around
valves. The pumping capacity of these pumps is also limited by the small
displacement and frequency involved. Gear pumps are a different kind of
positive-displacement device.
Continuous, parallel-axis rotary pumps. A screw-type, three-dimensional de-

vice for lowReynolds numbers was proposed by Taylor (1972) for propulsion
purposes and shown in his seminal lm. The pump has an axis of rotation
parallel to the ow direction, implying that the powering motor must be sub-
merged in the ow, the ow be turned through an angle, or that complicated
gearing would be needed.
Continuous, transverse-axis rotary pumps. This is the class of machines that

was recently developed by Sen, Wajerski, and Gad-el-Hak (1996). They have
shown that a rotating body, asymmetrically placed within a duct, will produce
a net ow due to viscous action. The axis of rotation can be perpendicular to
the ow direction, and the cylinder can thus be easily powered from outside
a duct. A related viscous-ow pump was designed by Odell and Kovasznay
(1971) for a water channel with density stratication. However, their de-
sign operates at a much higher Reynolds number and is too complicated for
microfabrication.
Figure 13.13: Schematic of micropump developed by
Sen et al. (1996).
As evidenced from the third item
above, it is possible to generate axial
uid motion in open channels through
the rotation of a cylinder in a viscous
uid medium. Odell and Kovasznay
(1971) studied a pump based on this
principle at high Reynolds numbers.
Sen et al. (1996) carried out an exper-
imental study of a different version of
such a pump. The novel viscous pump,
shown schematically in Figure 13.13, consists simply of a transverse-axis cylindri-
cal rotor eccentrically placed in a channel so that the differential viscous resistance
between the small and large gaps causes a net ow along the duct. The Reynolds
numbers involved in Sen et al.s work were low (0.01 < Re 2a
2
/ < 10, where
is the radian velocity of the rotor, and a is its radius) and typical of microscale devices
but achieved using a macroscale rotor and a very viscous uid. The bulk velocities
obtained were as high as 10% of the surface speed of the rotating cylinder. Sen et al.
(1996) have also tried cylinders with square and rectangular cross sections, but the
circular cylinder delivered the best pumping performance.
A nite-element solution for low-Reynolds-number, uniform ow past a rotat-
ing cylinder near an impermeable plane boundary has already been obtained by
Liang and Liou (1995). However detailed two-dimensional NavierStokes simula-
tions of the pump described above have been carried out by Sharatchandra, Sen, and
Gad-el-Hak (1997), who extended the operating range of Re beyond 100. The effects
of varying the channel height H and the rotor eccentricity have been studied. It
was demonstrated that an optimum plate spacing exists and that the induced ow
increases monotonically with eccentricity, the maximum ow rate being achieved
with the rotor in contact with a channel wall. Both the experimental results of Sen
et al. (1996) and the two-dimensional numerical simulations of Sharatchandra et al.
(1997) have veried that, at Re < 10, the pump characteristics are linear and there-
fore kinematically reversible. Sharatchandra et al. (1997, 1998a) also investigated
the effects of slip ow on the pump performance as well as the thermal aspects of
the viscous device. Wall slip does reduce the traction at the rotor surface and thus
lowers the performance of the pump somewhat. However, the slip effects appear to
be signicant only for Knudsen numbers greater than 0.1, which is encouraging from
the point of view of microscale applications.
In an actual implementation of the micropump, several practical obstacles need
to be considered. Among these are the larger stiction and seal design associated
with rotational motion of microscale devices. Both the rotor and the channel have
a nite, in fact rather small, width. DeCourtye, Sen, and Gad-el Hak (1998) nu-
merically investigated the viscous micropump performance as the width of the
channel W became exceedingly small. The bulk ow generated by the pump de-
creased as a result of the additional resistance to the ow caused by the side walls.
However, effective pumping was still observed with extremely narrow channels.
Finally, Sharatchandra et al. (1998b) used a genetic algorithm to determine the op-
timum wall shape to maximize the micropump performance. Their genetic algo-
rithm uncovered shapes that were nonintuitive but yielded vastly superior pump
performance.
Though most of the micropump discussion above is of ow in the steady state,
it should be possible to give the eccentric cylinder a nite number of turns or even
a portion of a turn to displace a prescribed minute volume of uid. Numerical com-
putations will easily show the order of magnitude of the volume discharged and the
errors induced by acceleration at the beginning of the rotation and deceleration at
the end. Such system can be used for microdosage delivery in medical applications.
13.3.4 Microturbines
DeCourtye et al. (1998) have described the possible utilization of the inverse
micropump device (Section 13.3.3) as a turbine. The most interesting application of
such a microturbine would be as a microsensor for measuring exceedingly small ow
rates on the order of nanoliters per second (i.e., microow metering for medical and
other applications).
The viscous pump described in Section 13.3.3 operates best at low Reynolds
numbers and should therefore be kinematically reversible in the creeping-owregime.
A microturbine based on the same principle should, therefore, lead to a net torque in
the presence of a prescribed bulk velocity. The results of three-dimensional numerical
simulations of the envisioned microturbine are summarized in this subsection. The
Reynolds number for the turbine problem is dened in terms of the bulk velocity,
because the rotor surface speed is unknown in this case,
Re =
U(2a)
(13.76)
where U is the prescribed bulk velocity in the channel, a is the rotor radius, and is
the kinematic viscosity of the uid.
Figure 13.14 shows the dimensionless rotor speed as a function of the bulk velocity
for two dimensionless channel widths W= and W= 0.6. In these simulations, the
dimensionless channel depth is H = 2.5, and the rotor eccentricity is /
max
= 0.9.
The relation is linear, as was the case for the pump problem. The slope of the lines is
0.37 for the two-dimensional turbine and 0.33 for the narrowchannel with W= 0.6.
This means that the induced rotor speed is, respectively, 0.37 and 0.33 of the bulk
Figure 13.14: Turbine rotation as a function of the bulk velocity
in the channel [after DeCourtye et al. 1998].
velocity in the channel.
15
For the pump, the corresponding numbers were 11.11
for the two-dimensional case and 100 for the three-dimensional case. Although it
appears that the side walls have bigger inuence on the pump performance, it should
be noted that in the turbine case a vastly higher pressure drop is required in the three-
dimensional duct to yield the same bulk velocity as that in the two-dimensional duct
(dimensionless pressure drop of p
p4a
2
/
2
= 29 versus p
= 1.5).
The turbine characteristics are dened by the relation between the shaft speed
and the applied load. A turbine load results in a moment on the shaft, which at
steady state balances the torque due to viscous stresses. At a xed bulk velocity,
the rotor speed is determined for different loads on the turbine. Again, the turbine
characteristics are linear in the Stokes (creeping) ow regime, but the side walls have
weaker, though still adverse, effect on the device performance as compared with the
pump case. For a given bulk velocity, the rotor speed drops linearly as the external
load on the turbine increases. At large enough loads, the rotor will not spin, and
maximum rotation is achieved when the turbine is subjected to zero load.
At present it is difcult to measure ow rates on the order of 10
12
m
3
/s
(1 nl/s). One possible way is to collect the efuent directly over time. This is use-
ful for calibration but is not practical for on-line ow measurement. Another is to
use heat transfer from a wire or lm to determine the local ow rate, as in a thermal
anemometer. Heat transfer from slowly moving uids is mainly by conduction, and
thus temperature gradients can be large. This is undesirable for biological and other
uids easily damaged by heat. The viscous mechanism that has been proposed and
veried for pumping may be turned around and used for measuring. As demonstrated
in this subsection, a freely rotating cylinder eccentrically placed in a duct will rotate
at a rate proportional to the ow rate owing to a turbine effect. In fact, other geome-
tries such as a freely rotating sphere in a cylindrical tube should also behave similarly.
The calibration constant, which depends on systemparameters such as geometry and
bearing friction, should be determined computationally to ascertain the practical vi-
ability of such a microow meter. Geometries that are simplest to fabricate should
be explored and studied in detail.
13.3.5 Microbearings
Many of the micromachines use rotating shafts and other moving parts that carry
a load and need uid bearings for support, and most of them operate with air or
water as the lubricating uid. The uid mechanics of these bearings are very different
compared with that of their larger cousins. Their study falls in the area of microuid
mechanics, an emerging discipline that has been greatly stimulated by its applications
to micromachines and is the subject of this chapter.
Macroscale journal bearings develop their load-bearing capacity from large pres-
sure differences that are a consequence of the presence of a viscous uid, an ec-
centricity between the shaft and its housing, a large surface speed of the shaft,
and a small clearance to diameter ratio. Several closed-form solutions of the no-
slip ow in a macrobearing have been developed. Wannier (1950) used modied
Cartesian coordinates to nd an exact solution to the biharmonic equation governing
15
The rotor speed can never, of course, exceed the uid velocity even if there is no load on the turbine.
Without load, the integral of the viscous shear stress over the entire surface area of the rotor is exactly
zero, and the turbine achieves its highest, albeit nite, rpm.
two-dimensional journal bearings in the no-slip, creeping ow regime. Kamal (1966)
and Ashino and Yoshida (1975) worked in bipolar coordinates; they assumed a gen-
eral form for the stream function with several constants that were determined using
the boundary conditions. Though all these methods work if there is no slip, they can-
not be readily adapted to slip ow. The basic reason is that the ow pattern changes
if there is slip at the walls, and the assumed form of the solution is no longer valid.
Microbearings are different in the following aspects: (1) being so small, it is dif-
cult to manufacture them with a clearance that is much smaller than the diameter of
the shaft; (2) because of the small shaft size, their surface speed, at normal rotational
speeds, is also small;
16
and (3) air bearings in particular may be small enough for
noncontinuum effects to become important. For these reasons the hydrodynamics of
lubrication are very different at microscales. The lubrication approximation that is
normally used is no longer directly applicable, and other effects come into play. From
an analytical point of view there are three consequences of the preceding character-
istics: uid inertia is negligible, slip ow may be important for air and other gases,
and relative shaft clearance need not be small.
In a recent study, Maureau et al. (1997) analyzed microbearings represented as
an eccentric cylinder rotating in a stationary housing. The ow Reynolds number
was assumed small, the clearance between shaft and housing was not small relative
to the overall bearing dimensions, and there was slip at the walls due to nonequi-
librium effects. The two-dimensional governing equations were written in terms of
the streamfunction in bipolar coordinates. Following the method of Jeffery (1920),
Maureau et al. (1997) succeeded in obtaining an exact innite-series solution of the
NavierStokes equations for the specied geometry and ow conditions. In contrast
to macrobearings and owing to the large clearance, ow in a microbearing is char-
acterized by the possibility of a recirculation zone that strongly affects the velocity
and pressure elds. For high values of the eccentricity and low slip factors, the ow
develops a recirculation region, as shown in the streamlines plot in Figure 13.15.
From the innite-series solution the frictional torque and the load-bearing ca-
pacity can be determined. The results show that both are similarly affected by the
eccentricity and the slip factor: they increase with the former and decrease with the
latter. For a given load, there is a corresponding eccentricity that generates a force
sufcient to separate shaft from housing (i.e., sufcient to prevent solid-to-solid con-
tact). As the load changes, the rotational center of the shaft shifts a distance necessary
for the forces to balance. It is interesting to note that for a weight that is vertically
downwards, the equilibrium displacement of the center of the shaft is in the horizon-
tal direction. This can lead to complicated rotor dynamics governed by mechanical
inertia, viscous damping, and pressure forces. A study of these dynamics may be of
interest. Real microbearings have nite shaft lengths, and end walls and other three-
dimensional effects inuence the bearing characteristics. Numerical simulations of
the three-dimensional problem can readily be carried out and may also be of interest
to the designers of microbearings. Other potential research includes determination of
a criterion for onset of cavitation in liquid bearings. From the results of these studies,
information related to load, rotational speed, and geometry can be generated that
would be useful for the designer.
16
The microturbomachines being developed presently at MIT operate at shaft rotational speeds on the
order of 1 million rpmand are therefore operating at different owregimes fromthose considered here.
Figure 13.15: Effect of slip factor and eccentricity on the microbearing stream-
lines. From top to bottom, eccentricity changes as = 0.2, 0.5, 0.8. From left
to right, slip factor changes as S (
2v
v
)Kn = 0, 0.1, 0.5 [after Maureau et al.
1997].
Finally, Piekos et al. (1997) have used full NavierStokes computations to
study the stability of ultra-high-speed gas microbearings. They concluded that it
is possibledespite signicant design constraintsto attain stability for specic
bearings to be used with the MIT microturbomachines (Epstein and Senturia 1997;
Epstein et al. 1997), which incidentally operate at much higher Reynolds numbers
(and rpm) than the micropumps, microturbines, and microbearings considered thus
far in this and the previous two subsections. According to Piekos et al. (1997), high-
speed bearings are more robust than low-speed ones owing to their reduced running
eccentricities and the large loads required to maintain them.
The 40-year-old vision of Richard Feynman of building minute machines is now
a reality. Microelectromechanical systems have witnessed explosive growth during
the last decade and are nding increased applications in a variety of industrial and
medical elds. The physics of uid ows in microdevices and some representative
applications have been explored in this chapter. Although we nowknowconsiderably
more than we did just few years ago, much physics remains to be explored so that
rational tools can be developed for the design, fabrication, and operation of MEMS
devices.
The traditional NavierStokes model of uid ows with no-slip boundary condi-
tions works only for a certain range of the governing parameters. This model basically
demands two conditions:
1. That the uid is a continuum, which is almost always satised because there
are usually more than 1 million molecules in the smallest volume in which
appreciable macroscopic changes take place. This is the molecular chaos
restriction.
2. The ow is not too far from thermodynamic equilibrium, which is satised
if there are a sufcient number of molecular encounters during a time period
small compared with the smallest time scale for ow changes. During this
time period the average molecule will have moved a distance small compared
with the smallest-ow length scale.
For gases, the Knudsen number determines the degree of rarefaction and the
applicability of traditional ow models. As Kn 0, the time and length scales of
molecular encounters are vanishingly small compared with those for the ow, and
the velocity distribution of each element of the uid instantaneously adjusts to the
equilibrium thermodynamic state appropriate to the local macroscopic properties as
this molecule moves through the oweld. From the continuum viewpoint, the ow
is isentropic and heat conduction and viscous diffusion and dissipation vanish from
the continuum conservation relations, leading to the Euler equations of motion. At
small but nite Kn, the NavierStokes equations describe near-equilibriumcontinuum
ows.
Slip ow must be taken into account for Kn > 0.001. The slip boundary condi-
tion is at rst linear in Knudsen number then nonlinear effects take over beyond a
Knudsen number of 0.1. At the same transition regime (i.e., 0.1 <Kn <10), the lin-
ear (stress)(rate of strain) and (heat ux)(temperature gradient) relationsneeded
to close the NavierStokes equationsalso break down, and alternative continuum
equations (e.g., Burnett) or molecular-based models must be invoked. In the tran-
sition regime, provided that the dilute gas and molecular chaos assumptions hold,
solutions to the difcult Boltzmann equation are sought, but physical simulations
such as Monte Carlo methods are more readily executed in this range of Knudsen
number. In the free-molecule ow regime (i.e., Kn >10), the nonlinear collision in-
tegral is negligible and the Boltzmann equation is drastically simplied. Analytical
solutions are possible in this case for simple geometries, and numerical integration
of the Boltzmann equation is straightforward for arbitrary geometries provided that
the surface-reection characteristics are accurately modeled.
Gaseous ows are often compressible in microdevices even at lowMach numbers.
Viscous effects can cause sufcient pressure drop and density changes for the ow to
be treated as compressible. In a long, constant-area microduct, all Knudsen number
regimes may be encountered, and the degree of rarefaction increases along the tube.
The pressure drop is nonlinear, and the Mach number increases downstream, limited
only by choked-ow condition.
Similar deviation and breakdown of the traditional NavierStokes equations oc-
cur for liquids as well, but there the situation is more murky. Existing experiments
are contradictory. There is no kinetic theory of liquids, and rst-principles prediction
methods are scarce. Molecular dynamics simulations can be used, but they are lim-
ited to extremely small ow extents. Nevertheless, measurable slip is predicted from
MD simulations at realistic shear rates in microdevices.
Microelectromechanical systems are nding increased applications in the diag-
nosis and control of turbulent ows. The use of microsensors and microactuators
promises a quantum leap in the performance of reactive-ow control systems and is
now in the realm of the possible for future practical devices (Chapter 14). Simple,
viscous-based micropumps can be utilized for microdosage delivery, and microtur-
bines can be used for measuring ow rates in the nanoliter per second range. Both of
these can be of value in several medical applications. Much nontraditional physics
is still to be learned, and many exciting applications of microdevices are yet to be
discovered. The future is bright for this emerging eld of science and technology.
Richard Feynman was right about the possibility of building mite-size machines but
was somewhat cautious in forecasting that such machines, although they would be
fun to make, might or might not be useful.
CHAPTER FOURTEEN
Frontiers of Flow Control
There is always an easy solution to every human problemneat, plausible
and wrong.
(Henry Louis Mencken, 18801956)
As for the future, your task is not to foresee, but to enable it.
(Antoine-Marie-Roger de Saint-Exup ery, 19001944, in
Citadelle (The Wisdom of the Sands))
PROLOGUE
In contrast to the early chapters, in the present chapter I will emphasize the frontiers
of the eld of ow control, pondering mostly the control of turbulent ows. I will
review the important advances in the eld that have taken place during the past few
years and are anticipated to dominate progress in the future, essentially covering the
fth era outlined in Section 1.2. By comparison with laminar ow control or sepa-
ration prevention, the control of turbulent ow remains a very challenging problem.
Flow instabilities quickly magnify near critical ow regimes, and therefore delay-
ing transition or separation is a relatively easier task. In contrast, classical control
strategies are often ineffective for fully turbulent ows. Newer ideas for turbulent
ow control to achieve skin-friction drag reduction, for example, focus on the direct
onslaught on coherent structures. Spurred by the recent developments in chaos con-
trol, microfabrication, and soft computing tools, reactive control of turbulent ows
through which sensors detect oncoming coherent structures and actuators attempt to
modulate these quasi-periodic events favorably is nowin the realmof the possible for
future practical devices. In this last chapter, I will provide estimates for the number,
size, frequency, and energy consumption of the sensor and actuator arrays needed to
control the turbulent boundary layer on a full-scale aircraft or submarine.
14.1 Introduction
14.1.1 The Taming of the Shrew
When one considers the extreme complexity of the turbulence problem in general
and the unattainability of rst-principles analytical solutions in particular, it is not
318
surprising that controlling a turbulent ow remains a challenging task mired in
empiricism and unfullled promises and aspirations. Brute force suppression, or
taming, of turbulence via active, energy-consuming control strategies is always
possible, but the penalty for doing so often exceeds any potential benets. The
artice is to achieve a desired effect with minimum energy expenditure. This is of
course easier said than done. Indeed, suppressing turbulence is as arduous as the
taming of the shrew. The former task will be emphasized throughout this chapter,
but for now we reect on a short verse from the latter.
From William Shakespeares The Taming of the Shrew:
Curtis (Petruchios servant, in charge of his country house): Is she so hot a shrew as shes
reported?
Grumio (Petruchios personal lackey): She was, good Curtis, before this frost. But thou
knowst winter tames man, woman, and beast; for it hath tamed my old master, and my
new mistress, and myself, fellow Curtis.
14.1.2 Control of Turbulence
Numerous methods of ow control have already been successfully implemented
in practical engineering devices. Delaying laminar-to-turbulence transition to reason-
able Reynolds numbers and preventing separation can readily be accomplished using
a myriad of passive and predetermined active control strategies. Such classical tech-
niques have been reviewed in the preceding chapters of this book and by, among oth-
ers, Bushnell (1983, 1994); Wilkinson et al. (1988); Bushnell and McGinley (1989);
Gad-el-Hak (1989); Bushnell and Hefner (1990); Fiedler and Fernholz (1990);
Gad-el-Hak and Bushnell (1991a,b); Barnwell and Hussaini (1992); Viswanath
(1995); and Joslin, Erlebacher, and Hussaini (1996). Yet, very few of the classi-
cal strategies are effective in controlling free-shear or wall-bounded turbulent ows.
Serious limitations exist for some familiar control techniques when applied to certain
turbulent ow situations. For example, in attempting to reduce the skin-friction drag
of a body having a turbulent boundary layer by using global suction, the penalty
associated with the control device often exceeds the savings derived from its use.
What is needed is a way to reduce this penalty to achieve a more efcient control.
Flow control is most effective when applied near the transition or separation
pointsin other words, near the critical ow regimes where ow instabilities mag-
nify quickly. Therefore, delaying or advancing laminar-to-turbulence transition and
preventing or provoking separation are relatively easier tasks to accomplish. To re-
duce the skin-friction drag in a nonseparating turbulent boundary layer, where the
mean ow is quite stable, is a more challenging problem. Yet, even a modest reduc-
tion in the uid resistance to the motion of, for example, the worldwide commercial
airplane eet is translated into fuel savings estimated to be in the billions of dollars.
Newer ideas for turbulent ow control focus on the direct onslaught on coherent
structures. Spurred by the recent developments in chaos control, microfabrication,
and soft computing tools, reactive control of turbulent ows is now in the realm of
the possible for future practical devices.
The primary objective of this chapter is to advance possible scenarios by which
viable control strategies of turbulent ows can be realized. As will be argued in
the following presentation, future systems for control of turbulent ows in general
and turbulent boundary layers in particular could greatly benet from the merging
of the science of chaos control, the technology of microfabrication, and the newest
320 FRONTIERS OF FLOW CONTROL
computational tools collectively termed soft computing. Control of chaotic, nonlinear
dynamical systems has been demonstrated theoretically as well as experimentally
even for multi-degree-of-freedom systems. Microfabrication is an emerging tech-
nology that has the potential for producing inexpensive, programmable sensor or
actuator chips that have dimensions of the order of a few microns (see Chapter 13).
Soft computing tools include neural networks, fuzzy logic, and genetic algorithms
and are now more advanced as well as more widely used as compared with just
few years ago. These tools could be very useful in constructing effective adaptive
controllers.
Such futuristic systems are envisaged as consisting of a large number of intelligent,
interactive, microfabricated wall sensors and actuators arranged in a checkerboard
pattern and targeted toward specic organized structures that occur quasi-randomly
within a turbulent ow. Sensors detect oncoming coherent structures, and adaptive
controllers process the sensors information and provide control signals to the actu-
ators, which in turn attempt to modulate the quasi-periodic events favorably. Finite
numbers of wall sensors perceive only partial information about the entire oweld
above. However, a low-dimensional dynamical model of the near-wall region used
in a Kalman lter can make the most of the partial information from the sensors.
Conceptually, all of that is not too difcult, but in practice the complexity of such a
control system is daunting, and much research and development work still remain.
14.1.3 Outline
In this chapter, I will review the important developments in the eld of ow con-
trol that took place during the past fewyears and suggest avenues for future research.
The emphasis will be on reactive ow control for future vehicles and other indus-
trial devices. This chapter is organized into ten sections. A particular example of
a classical control system, suction, is described in the following section. This will
serve as a prelude to introducing the selective suction concept. In Section 14.3, the
different hierarchy of coherent structures that dominate a turbulent boundary layer
and that constitute the primary target for direct onslaught are briey recalled. The
characteristic lengths and sensor requirements of turbulent ows are discussed in
Section 14.4. Reactive ow control and the selective suction concept are described
in Section 14.5. The number, size, frequency, and energy consumption of the sensor
and actuator units required to tame the turbulence on a full-scale air or water vehicle
are estimated in that same section. Section 14.6 introduces the topic of magnetohy-
drodynamics and suggests a reactive owcontrol scheme using electromagnetic body
forces. Sections 14.7, 14.8, and 14.9 consider, in turn, the emerging areas of chaos
control, microfabrication, and soft computingparticularly as they relate to reactive
control strategies. Finally, brief concluding remarks are given in Section 14.10.
14.2 Suction
To set the stage for introducing the concept of targeted or selected control in
Section 14.5, I will rst discuss in this section global control as applied to wall-
bounded ows. Aviscous uid that is initially irrotational will acquire vorticity when
an obstacle is passed through the uid. This vorticity controls the nature and struc-
ture of the boundary layer ow in the vicinity of the obstacle. For an incompressible,
wall-bounded ow, the ux of spanwise or streamwise vorticity at the wall, and hence
14.2 SUCTION 321
whether the surface is a sink or a source of vorticity, is affected by the wall motion
(e.g., in the case of a compliant coating), transpiration (suction or injection), stream-
wise or spanwise pressure gradient, wall curvature, normal viscosity gradient near
the wall (caused by, for example, heating or cooling of the wall or introduction of a
shear-thinning or shear thickening additive into the boundary layer), and body forces
such as electromagnetic ones in a conducting uid (Section 14.6). These alterations
separately or collectively control the shape of the instantaneous as well as the mean
velocity proles, which in turn determines the skin friction at the wall, the boundary
layer ability to resist transition and separation, and the intensity of turbulence and
its structure.
For illustration purposes, I will focus on global wall suction as a generic con-
trol tool. The arguments presented here and in subsequent sections are equally valid
for other global control techniques such as geometry modication (body shaping),
surface heating or cooling, electromagnetic control, and so forth. Transpiration pro-
vides a good example of a single control technique that is used to achieve a variety
of goals. Suction leads to a fuller velocity prole (vorticity ux away from the wall)
and can, therefore, be employed to delay laminar-to-turbulence transition, post-
pone separation, achieve an asymptotic turbulent boundary layer (i.e., one having
constant momentum thickness), or relaminarize an already turbulent ow. Unfortu-
nately, global suction cannot be used to reduce the skin-friction drag in a turbulent
boundary layer. The amount of suction required to inhibit boundary-layer growth is
too large to effect a net drag reduction. This is a good illustration of a situation in
which the penalty associated with a control device may exceed the savings derived
from its use.
Small amounts of uid withdrawn from the near-wall region of a boundary layer
change the curvature of the velocity prole at the wall and can dramatically alter the
stability characteristics of the ow. Concurrently, suction inhibits the growth of the
boundary layer, and thus the critical Reynolds number based on thickness may never
be reached. Although laminar ow can be maintained to extremely high Reynolds
numbers provided that enough uid is sucked away, the goal is to accomplish transi-
tion delay with the minimum suction ow rate. This will reduce not only the power
necessary to drive the suction pump but also the momentum loss due to the addi-
tional freestream uid entrained into the boundary layer as a result of withdrawing
uid from the wall. That momentum loss is, of course, manifested as an increase in
the skin-friction drag.
The case of uniformsuction froma at plate at zero incidence is an exact solution
of the NavierStokes equation. The asymptotic velocity prole in the viscous region
is exponential and has a negative curvature at the wall. The displacement thickness
has the constant value
= /|v
w
|, where is the kinematic viscosity and |v
w
| is the
absolute value of the normal velocity at the wall. In this case, the familiar von K arm an
integral equation reads: C
f
= 2C
q
. Bussmann and M unz (1942) computed the critical
Reynolds number for the asymptotic suction prole to be Re
/ = 70,000.
From the value of
given above, the ow is stable to all small disturbances if

C
q
|v
w
|/U
>1.4 10
5
. The amplication rate of unstable disturbances for the
asymptotic prole is an order of magnitude less than that for the Blasius bound-
ary layer (Pretsch 1942). This treatment ignores the development distance from the
leading edge needed to reach the asymptotic state. When this is included into the
computation, a higher C
q
= 1.1810
4
is required to ensure stability (Iglisch 1944;
Ulrich 1944).
In a turbulent wall-bounded ow, the results of El ena (1975, 1984) and Antonia
et al. (1988) indicate that suction causes an appreciable stabilization of the low-speed
streaks in the near-wall region. The maximumturbulence level at y
+
13 drops from
15 to 12% as C
q
varies from 0 to 0.003. More dramatically, the tangential Reynolds
stress near the wall drops by a factor of 2 for the same variation of C
q
. The dissipation
length scale near the wall increases by 40%and the integral length scale by 25%with
the suction.
The suction rate necessary for establishing an asymptotic turbulent boundary
layer independent of streamwise coordinate (d
/dx = 0) is much lower than the

rate required for relaminarization (C
q
0.01) but still not low enough to yield
net drag reduction. For Reynolds number based on distance from leading edge
Re
x
= O[10
6
], Favre et al. (1966), Rotta (1970), and Verollet et al. (1972), among
others, reported an asymptotic suction coefcient of C
q
0.003. For a zero-pressure-
gradient boundary layer on a at plate, the corresponding skin-friction coefcient is
C
f
= 2C
q
= 0.006, indicating higher skin friction than if no suction were applied.
To achieve a net skin-friction reduction with suction, the process must be further
optimized. One way to accomplish that is to target the suction toward particular
organized structures within the boundary layer and not to use it globally as in clas-
sical control schemes. This point will be revisited in Section 14.5, but the coherent
structures to be targeted and the length scales to be expected are rst detailed in
Sections 14.3 and 14.4.
14.3 Coherent Structures
The discussion in Section 14.2 indicates that achieving a particular control goal is
always possible. The challenge is reaching that goal with a penalty that can be toler-
ated. Suction, for example, would lead to a net drag reduction if only we could reduce
the suction coefcient necessary for establishing an asymptotic turbulent boundary
layer to below one-half of the unperturbed skin-friction coefcient. A more efcient
way of using suction, or any other global control method, is to target particular
coherent structures within the turbulent boundary layer. Before discussing this selec-
tive control idea, I will briey describe in this and the following section the different
hierarchy of organized structures in a wall-bounded ow and the expected scales of
motion. Coherent structures were more fully discussed in Chapter 4.
The classical view that turbulence is essentially a stochastic phenomenon having
a randomly uctuating velocity eld superimposed on a well-dened mean has been
changed in the last few decades by the realization that the transport properties of all
turbulent shear ows are dominated by quasi-periodic, large-scale vortex motions
(Laufer 1975; Cantwell 1981; Fiedler 1988; Robinson 1991). Despite the extensive
research work in this area, no generally accepted denition of what is meant by
coherent motion has emerged. In physics, coherence stands for well-dened phase
relationship. For the present purpose we adopt the rather restrictive denition given
by Hussain (1986): A coherent structure is a connected turbulent uid mass with
instantaneously phase-correlated vorticity over its spatial extent. In other words,
underlying the random, three-dimensional vorticity that characterizes turbulence,
there is a component of large-scale vorticity that is instantaneously coherent over the
spatial extent of an organized structure. The apparent randomness of the oweld
is, for the most part, due to the random size and strength of the different types of
organized structures comprising that eld.
In a wall-bounded ow, a multiplicity of coherent structures have been identi-
ed mostly through ow visualization experiments, although some important early
discoveries have been made using correlation measurements (e.g., Townsend 1961,
1970; Bakewell and Lumley 1967). Although the literature on this topic is vast, no
research-community-wide consensus has been reachedparticularly on the issues of
the origin of, and interaction between, the different structures, regeneration mech-
anisms, and Reynolds number effects. What follow are somewhat biased remarks
addressing those issues gathered mostly through low-Reynolds number experiments.
The interested reader is referred to the book edited by Panton (1997a) and the large
number of review articles available (e.g., Kovasznay 1970; Laufer 1975; Willmarth
1975a,b; Saffman 1978; Cantwell 1981; Fiedler 1986, 1988; Blackwelder 1988,
1998; Robinson 1991; Delville et al. 1998). The article by Robinson (1991) in partic-
ular summarizes many of the different, sometimes contradictory, conceptual models
offered thus far by different research groups. Those models are aimed ultimately at
explaining how the turbulence maintains itself and range from the speculative to the
rigorous, but not one of them, unfortunately, is self-contained and complete. Further-
more, the structure research dwells largely on the kinematics of organized motion,
and little attention is given to the dynamics of the regeneration process.
In a boundary layer, the turbulence production process is dominated by three
kinds of quasi-periodicor, depending on ones viewpoint, quasi-randomeddies:
the large outer structures, the intermediate Falco eddies, and the near-wall events. The
large, three-dimensional structures scale with the boundary layer thickness, , and
extend across the entire layer (Kovasznay et al. 1970; Blackwelder and Kovasznay
1972). These eddies control the dynamics of the boundary layer in the outer region,
such as entrainment, turbulence production, and so forth. They appear randomly in
space and time and seem to be, at least for moderate Reynolds numbers, the residue
of the transitional Emmons spots (Zilberman et al. 1977; Gad-el-Hak et al. 1981;
Riley and Gad-el-Hak 1985). The Falco eddies are also highly coherent and three-
dimensional. Falco (1974, 1977) named them typical eddies because they appear in
wakes, jets, Emmons spots, grid-generated turbulence, and boundary layers in zero,
favorable, and adverse pressure gradients. They have an intermediate scale of about
100/u
(100 wall units; u
is the friction velocity and /u
is the viscous length

scale). The Falco eddies appear to be an important link between the large structures
and the near-wall events.
The third kind of eddies exists in the near-wall region (0 < y < 100/u
) where
the Reynolds stress is produced in a very intermittent fashion. Half of the total
production of turbulence kinetic energy (uvU/y) takes place near the wall in
the rst 5% of the boundary layer at typical laboratory Reynolds numbers (smaller
fraction at higher Reynolds numbers), and the dominant sequence of eddy motions
there is collectively termed the bursting phenomenon. This dynamically signicant
process was reviewed by Willmarth (1975a); Blackwelder (1978); Robinson (1991);
and more recently Panton (1997a). Qualitatively, the process, according to at least
one school of thought, begins with elongated, counterrotating, streamwise vortices
having diameters of approximately 40/u
. The vortices exist in a strong shear and

induce low- and high-speed regions between them, as was sketched in Figure 4.10
of Chapter 4. The vortices and the accompanying eddy structures occur randomly
in space and time. However, their appearance is regular enough that an average
spanwise wavelength of approximately 80 to 100/u
has been identied by Kline

et al. (1967) and others. Kline et al. also observed that the low-speed regions grow
downstream and develop inectional U(y) proles. At approximately the same time,
the interface between the low- and high-speed uid begins to oscillateapparently
signaling the onset of a secondary instability. The low-speed region lifts up away
from the wall as the oscillation amplitude increases, and then the ow rapidly breaks
down into a completely chaotic motion. Because this latter process occurs on a very
short time scale, Kline et al. called it a burst. Corino and Brodkey (1969) showed
that the low-speed regions are quite narrow (i.e., z = 20/u
) and may also have

signicant shear in the spanwise direction. Virtually all of the net production of
turbulence kinetic energy in the near-wall region occurs during these bursts.
Considerably more has been learned about the bursting process during the last
two decades. For example, Falco (1980, 1983, 1991) has shown that when a typical
eddy, which may be formed in part by ejected wall-layer uid, moves over the wall, it
induces a high uv sweep (positive u and negative v). The wall region is continuously
bombarded by pockets of high-speed uid originating in the logarithmic and possi-
bly the outer layers of the ow. These pockets tend to promote, enhance, or both,
the inectional velocity proles by increasing the instantaneous shear, leading to a
more rapidly growing instability. Blackwelder and Haritonidis (1983) have shown
convincingly that the frequency of occurrence of these events scales with the viscous
parameters consistent with the usual boundary layer scaling arguments. An excellent
review of the dynamics of turbulent boundary layers has been provided by Sreeni-
vasan (1989). More information about coherent structures in high-Reynolds number
boundary layers is given by Gad-el-Hak and Bandyopadhyay (1994). The book edited
by Panton (1997a) emphasizes the self-sustaining mechanisms of wall turbulence.
14.4 Scales
In this section, we develop the equations needed to compute the characteristic lengths
and sensor requirements of turbulent ows in general and wall-bounded ows in
particular. Turbulence is a high-Reynolds-number phenomenon characterized by the
existence of numerous length and time scales. The spatial extension of the length
scales is bounded from above by the dimensions of the oweld and from below
by the diffusive and dissipative action of the molecular viscosity. If we limit our
interest to shear ows, which are basically characterized by two large length scales,
one in the streamwise direction (the convective or longitudinal length scale) and
the other perpendicular to the ow direction (the diffusive or lateral length scale),
we obtain a more well-dened problem. Moreover, at sufciently high Reynolds
numbers, the boundary layer approximation applies, and it is assumed that there is a
wide separation between the lateral and the longitudinal length scales. This leads to
some attractive simplications in the equations of motion such as that the elliptical
NavierStokes equations are transferred to the parabolic boundary-layer equations
(e.g., Hinze 1975). Thus, in this approximation, the lateral scale is approximately
equal to the extension of the ow perpendicular to the ow direction (the boundary
layer thickness), and the largest eddies typically have this spatial extension. These
eddies are most energetic and play a crucial role in both the transport of momentum
14.4 SCALES 325
and contaminants. A constant energy supply is needed to maintain the turbulence,
and this energy is extracted from the mean ow into the largest, most energetic
eddies. The lateral length scale is also the relevant scale for analyzing this energy
transfer. However, there is an energy destruction in the ow due to the action of
the viscous forces (the dissipation), and for the analysis of this process other smaller
length scales are needed.
As the eddy size decreases, viscosity becomes a more signicant parameter because
one property of viscosity is its effectiveness in smoothing out velocity gradients. The
viscous and the nonlinear terms in the momentum equation counteract each other
in the generation of small-scale uctuations. Although the inertial terms try to pro-
duce smaller and smaller eddies, the viscous terms check this process and prevent the
generation of innitely small scales by dissipating the small-scale energy into heat.
In the early 1940s, the universal equilibrium theory was developed by Kolmogorov
(1941a,b). One cornerstone of this theory is that the small-scale motions are sta-
tistically independent of the relatively slower large-scale turbulence. An implication
of this is that the turbulence at the small scales depends only on two parameters,
namely, the rate at which energy is supplied by the large-scale motion and the kine-
matic viscosity. In addition, it is assumed in the equilibrium theory that the rate of
energy supply to the turbulence should be equal to the rate of dissipation. Hence,
in the analysis of turbulence at small scales, the dissipation rate per unit mass is
a relevant parameter together with the kinematic viscosity . Kolmogorov (1941a)
used simple dimensional arguments to derive a length, a time, and a velocity scale
relevant for the small-scale motion given respectively by
=
_
_
1
4
(14.1)
=
_
_1
2
(14.2)
= ()
1
4
(14.3)
These scales are accordingly called the Kolmogorov microscales, or sometimes the
inner scales of the ow. Because they are obtained through a physical argument, these
scales are the smallest scales that can exist in a turbulent ow and they are relevant
for both free-shear and wall-bounded ows.
In boundary layers, the shear-layer thickness provides a measure of the largest
eddies in the ow. The smallest scale in wall-bounded ows is the viscous wall unit,
which will be shown below to be of the same order as the Kolmogorov length scale.
Viscous forces dominate over inertia in the near-wall region, and the characteristic
scales there are obtained from the magnitude of the mean vorticity in the region and
its viscous diffusion away from the wall. Thus, the viscous time-scale, t
, is given by
the inverse of the mean wall vorticity
t
=
_
U
y
w
_
1
(14.4)
where U is the mean streamwise velocity. The viscous length-scale,
, is determined
by the characteristic distance by which the (spanwise) vorticity is diffused from the
wall and is thus given by
U
y
w
(14.5)
where is the kinematic viscosity. The wall velocity scale (so-called friction velocity,
u

u
U
y
w
=
_
(14.6)
where
w
is the mean shear stress at the wall, and is the uid density. A wall
unit implies scaling with the viscous scales, and the usual ()
+
notation is used; for
example, y
+
= y/
= y u
/. In the wall region, the characteristic length for the

large eddies is y itself, whereas the Kolmogorov scale is related to the distance from
the wall y as follows:
+

u
(y
+
)
1
4
(14.7)
where is the von K arm an constant (0.41). As y
+
changes in the range of 15
(the extent of the viscous sublayer), changes from 0.8 to 1.2 wall units.
We now have access to scales for the largest and smallest eddies of a turbulent
ow. To continue our analysis of the cascade energy process, it is necessary to nd a
connection between these diverse scales. One way of obtaining such a relation is to use
the fact that at equilibrium the amount of energy dissipating at high wave numbers
must equal the amount of energy drained from the mean ow into the energetic
large-scale eddies at low wave numbers. In the inertial region of the turbulence
kinetic energy spectrum, the ow is almost independent of viscosity, and because the
same amount of energy dissipated at the high wave numbers must pass this inviscid
region, an inviscid relation for the total dissipation may be obtained by the following
argument. The amount of kinetic energy per unit mass of an eddy with a wave number
in the inertial sublayer is proportional to the square of a characteristic velocity for
such an eddy, u
2
. The rate of transfer of energy is assumed to be proportional to the
reciprocal of one eddy turnover time, u/, where is a characteristic length of the
inertial sublayer. Hence, the rate of energy that is supplied to the small-scale eddies
via this particular wave number is of order u
3
/, and this amount of energy must be
equal to the energy dissipated at the highest wavenumber, expressed as

u
3
(14.8)
Note that this is an inviscid estimate of the dissipation because it is based on large-
scale dynamics and does not either involve or contain viscosity. More comprehensive
discussion of this issue can be found in Taylor (1935) and Tennekes and Lumley
(1972). Froman experimental perspective, this is a very important expression because
it offers one way of estimating the Kolmogorov microscales fromquantities measured
in a much lower wave number range.
14.4 SCALES 327
Because the Kolmogorov length and time scales are the smallest scales occurring
in turbulent motion, a central question will be how small these scales can be without
violating the continuum hypothesis. By looking at the governing equations, it can be
concluded that high dissipation rates are usually associated with large velocities, and
this situation is more likely to occur in gases than in liquids, and thus it would be
sufcient to showthat for gas ows the smallest turbulence scales are normally much
larger than the molecular scales of motion. The relevant molecular length scale is the
mean free path, L, and the ratio between this length and the Kolmogorov length-scale,
, is the microstructure Knudsen number and can be expressed as (see Corrsin 1959):
Kn =
L
Ma
1
4
Re
(14.9)
where the turbulence Reynolds number Re and the turbulence Mach number Ma
are used as independent variables. It is obvious that a turbulent ow will interfere
with the molecular motion only at high Mach number and low Reynolds number,
and this is a very unusual situation occurring only in certain gaseous nebulae.
1
Thus, under normal condition the turbulence Knudsen number falls in the group
of continuum ows. However, measurements using extremely thin hot wires, small
MEMS sensors, or ows within narrow MEMS channels can generate values in the
slip-ow regime and even beyond, and this implies that, for instance, the no-slip
condition may be questioned, as was discussed in Section 13.2.5.
14.4.1 Sensor Requirements
It is the ultimate goal of all measurements in turbulent ows to resolve the largest
and smallest eddies that occur in the ow. At the lower wave numbers, the largest
and most energetic eddies occur, and normally there are no problems associated with
resolving these eddies. Basically, this is a question of having access to computers with
sufciently large memory for storing the amount of data that it may be necessary to
acquire from a large number of distributed probes, each collecting data for a time
period long enough to reduce the statistical error to a prescribed level. However,
at the other end of the spectrum, both the spatial and the temporal resolutions are
crucial, and this puts severe limitations on the sensors to be used. It is possible to
obtain a relation between the small and large scales of the ow by substituting the
inviscid estimate of the total dissipation rate, Eq. (14.8), into the expressions for the
Kolmogorov microscales, Eqs. (14.1)(14.3). Thus,
_
u
3
4
= Re
3
4
(14.10)
u
_
u
1
2
= Re
1
2
(14.11)
u

_
u
1
4
= Re
1
4
(14.12)
1
Note that in microduct ows and the like, the Re is usually too small for turbulence to even exist.
Thus the issue of turbulence Knudsen number is mute in those circumstances even if rarefaction effects
become strong.
where Re is the Reynolds number based on the speed of the energy containing eddies
u and their characteristic length . Because turbulence is a high-Reynolds-number
phenomenon, these relations show that the small length, time, and velocity scales
are much less than those of the larger eddies and that the separation in scales widens
considerably as the Reynolds number increases. Moreover, this also implies that the
assumptions made on the statistical independence and the dynamical equilibrium
state of the small structures will be most relevant at high Reynolds numbers. Another
interesting conclusion that can be drawn from the preceding relations is that if
two turbulent owelds have the same spatial extension (i.e., same large-scale) but
different Reynolds numbers, there would be an obvious difference in the small-scale
structure in the two ows. The low-Reynolds-number ow would have a relatively
coarse small-scale structure, whereas the high-Re ow would have much ner small
eddies.
To resolve the smallest eddies spatially, sensors that are of about the same size as
the Kolmogorov length scale for the particular ow under consideration are needed.
This implies that, as the Reynolds number increases, smaller sensors are required. For
instance, in the self-preserving region of a plane-cylinder wake at a modest Reynolds
number, on the basis of the cylinder diameter of 1840, the value of varies in the range
of 0.50.8 mm(Aronson and L ofdahl 1994). For this case, conventional hot wires can
be used for turbulence measurements. However, an increase in the Reynolds number
by a factor of 10 will yield Kolmogorov scales in the micrometer range and call
for either extremely small conventional hot wires or MEMS-based sensors. Another
example illustrating the Reynolds number effect on the requirement of small sensors
is a simple two-dimensional, at-plate boundary layer. At a momentum thickness
= 4000, the Kolmogorov length scale is typically of the

order of 50 m, and to resolve these scales it is necessary to have access to sensors
that have a characteristic active measuring length of the same spatial extension.
Severe errors will be introduced in the measurements by using sensors that are
too large, for such a sensor will integrate the uctuations due to the small eddies over
its spatial extension, and the energy content of these eddies will be interpreted by the
sensor as an average cooling. When measuring uctuating quantities, this implies
that these eddies are counted as part of the mean ow, and their energy is lost. The
result will be a lower value of the turbulence parameter, and this will wrongly be
interpreted as a measured attenuation of the turbulence (see for example Ligrani and
Bradshaw 1987). However, because turbulence measurements deal with statistical
values of uctuating quantities, it may be possible to loosen the spatial constraint of
having a sensor of the same size as to allow a sensor dimensions that are slightly
larger than the Kolmogorov scalesay on the order of .
For boundary layers, the wall unit has been used to estimate the smallest
necessary size of a sensor for accurately resolving the smallest eddies. For instance
Keith, Hurdis, and Abraham (1992) stated that 10 wall units or less is a relevant
sensor dimension for resolving small-scale pressure uctuations. Measurements of
uctuating velocity gradients, essential for estimating the total dissipation rate in
turbulent ows, are another challenging task. Gad-el-Hak and Bandyopadhyay
(1994) argued that turbulence measurements with probe lengths greater than the
viscous sublayer thickness (about 5 wall units) are unreliableparticularly near
the surface. Many studies have been conducted on the spacing between sensors
14.5 REACTIVE CONTROL 329
necessary to optimize the formed velocity gradients (see Aronson, Johansson, and
L ofdahl 1997 and references therein). A general conclusion from both experiments
and direct numerical simulations is that a sensor spacing of 35 Kolmogorov
lengths is recommended. When designing arrays for correlation measurements or for
targeted control, the spacing between the coherent structures will be the determining
factor. For example, when targeting the low-speed streaks in a turbulent boundary
layer, several sensors must be situated along a lateral distance of 100 wall units,
the average spanwise spacing between streaks. All this requires quite small sensors,
and many attempts have been made to meet these conditions with conventional
sensor designs. However, although conventional sensors like hot wires have been
fabricated in the micrometer-size range (for their diameter but not their length), they
are usually handmade, difcult to handle, and are too fragile, and here the MEMS
technology has really opened a door for new applications.
It is clear from the preceding discussion that the spatial and temporal resolu-
tions for any probe to be used to resolve high-Reynolds-number turbulent ows are
extremely tight. For example, the Kolmogorov scale and the viscous length scale
change from a few microns at the typical eld Reynolds numberbased on the mo-
mentum thicknessof 10
6
, to a couple of hundred microns at the typical laboratory
Reynolds number of 10
3
. Microelectromechanical systems sensors for pressure, ve-
locity, temperature and shear stress are at least one order of magnitude smaller than
conventional sensors (Ho and Tai 1996, 1998; L ofdahl et al. 1996; L ofdahl and
Gad-el-Hak 1999). Their small size improves both the spatial and temporal reso-
lutions of the measurements, typically a few microns and a few microseconds, re-
spectively. For example, a micro-hot-wire (called hot point) has very small thermal
inertia, and the diaphragm of a micropressure transducer has correspondingly fast
dynamic response. Moreover, the microsensors extreme miniaturization and lowen-
ergy consumption make themideal for monitoring the owstate without appreciably
affecting it. Lastly, literally hundreds of microsensors can be fabricated on the same
silicon chip at a reasonable cost, making them well suited for distributed measure-
ments and control. The UCLACaltech team (see, for example, Ho and Tai 1996,
1998 and references therein) has been very effective in developing many MEMS-based
sensors and actuators for turbulence diagnosis and control.
14.5 Reactive Control
Targeted control implies sensing and reacting to a particular quasi-periodic struc-
ture in the boundary layer. The wall seems to be the logical place for such reactive
control because of the relative ease of placing something in there, the sensitivity of the
owin general to surface perturbations, and the proximity and therefore accessibility
to the dynamically all important near-wall coherent events. According to Wilkinson
(1990), there are very few actual experiments that use embedded wall sensors to ini-
tiate a surface actuator response (Alshamani, Livesey, and Edwards 1982; Wilkinson
and Balasubramanian 1985; Nosenchuck and Lynch 1985; Breuer, Haritonidis, and
Landahl 1989). This 10-year-old assessment is fast changing, however, with the
introduction of microfabrication technology that has the potential for producing
small, inexpensive, programmable sensor and actuator chips. Witness the more recent
reactive control attempts by Kwong and Dowling (1993); Reynolds (1993); Jacobs
et al. (1993); Jacobson and Reynolds (1993a,b; 1994, 1995, 1998); Fan, Hofmann,
and Herbert (1993); James, Jacobs, and Glezer (1994); and Keefe (1996). Fan et al.
and Jacobson and Reynolds even considered the use of self-learning neural networks
for increased computational speeds and efciency. Recent reviews of reactive ow
control include those by Gad-el-Hak (1994, 1996b); Lumley (1996b); McMichael
(1996); Mehregany, DeAnna, and Reshotko (1996); and Ho and Tai (1996).
Numerous methods of ow control have already been successfully implemented
in practical engineering devices. Yet, limitations exist for some familiar control tech-
niques when applied to specic situations. For example, in attempting to reduce the
drag or enhance the lift of a body having a turbulent boundary layer using global
suction, global heating or cooling, or global application of electromagnetic body
forces, the actuators energy expenditure often exceeds the saving derived from the
predetermined active control strategy. What is needed is a way to reduce this penalty
to achieve a more efcient control. Reactive control geared specically toward ma-
nipulating the coherent structures in turbulent shear ows, though considerably more
complicated than passive control or even predetermined active control, has the poten-
tial to do just that. As will be argued in this and the following sections, future systems
for control of turbulent ows in general and turbulent boundary layers in particular
could greatly benet from the merging of the science of chaos control, the technol-
ogy of microfabrication, and the newest computational tools collectively termed soft
computing. Such systems are envisaged as consisting of a large number of intelligent,
communicative wall sensors and actuators arranged in a checkerboard pattern and
targeted toward controlling certain quasi-periodic, dynamically signicant coherent
structures present in the near-wall region.
14.5.2 Targeted Control
As discussed in Chapter 10, successful techniques to reduce the skin friction in
a turbulent ow, such as polymers, particles, or riblets, appear to act indirectly
through local interaction with discrete turbulent structures, particularly small-scale
eddies, within the ow. Common characteristics of all these methods are increased
losses in the near-wall region, thickening of the buffer layer, and lowered production
of Reynolds shear stress (Bandyopadhyay 1986b). Methods that act directly on the
mean ow, such as suction or lowering of near-wall viscosity, also lead to inhibition of
Reynolds stress. However, skin friction is increased when any of these velocity-prole
modiers is applied globally.
Could these seemingly inefcient techniques (e.g., global suction) be used more
sparingly and be optimized to reduce their associated penalty? It appears that the
more successful drag-reducing methods (e.g., polymers) act selectively on particular
scales of motion and are thought to be associated with stabilization of the secondary
instabilities. It is also clear that energy is wasted when suction or heating or cooling
is used to suppress the turbulence throughout the boundary layer when the main in-
terest is to affect a near-wall phenomenon. One ponders, what would become of wall
turbulence if specic coherent structures were to be targeted by the operator through a
reactive control scheme for modication? The myriad of organized structures present
in all shear ows are instantaneously identiable, quasi-periodic motions (Cantwell
1981; Robinson 1991). Bursting events in wall-bounded ows, for example, are both
intermittent and random in space as well as time. The random aspects of these events
reduce the effectiveness of a predetermined active control strategy. If such structures
are nonintrusively detected and altered, on the other hand, net performance gain
might be achieved. It seems clear, however, that temporal phasing as well as spatial se-
lectivity would be required to achieve proper control targeted toward randomevents.
A nonreactive version of the preceding idea is the selective suction technique,
which combines suction to achieve an asymptotic turbulent boundary layer and lon-
gitudinal riblets to x the location of low-speed streaks. Although far fromindicating
net drag reduction, the available results are encouraging, and further optimization
is needed. When implemented via an array of reactive control loops, the selective
suction method is potentially capable of skin-friction reduction that approaches
60%.
The genesis of the selective suction concept can be found in the articles by
Gad-el-Hak and Blackwelder (1987c, 1989) and the patent by Blackwelder and
Gad-el-Hak (1990). These researchers have suggested that one possible means of
optimizing the suction rate is to be able to identify where a low-speed streak is located
and to apply a small amount of suction under it. Assuming that the production of
turbulence kinetic energy is due to the instability of an inectional U(y) velocity
prole, one needs to remove only enough uid so that the inectional nature of the
prole is alleviated. An alternative technique that could reduce the Reynolds stress
is to inject uid selectively under the high-speed regions. The immediate effect of
normal injection would be to decrease the viscous shear at the wall, resulting in less
drag. In addition, the velocity proles in the spanwise direction, U(z), would have a
smaller shear, U/z, because the suction or injection would create a more uniform
ow. Because Swearingen and Blackwelder (1984) and Blackwelder and Swearingen
(1990) have found that inectional U(z) proles occur as often as inection points
are observed in U(y) proles, suction under the low-speed streaks, injection under the
high-speed regions, or both, would decrease this shear and hence the resulting insta-
bility. The combination of selective suction and injection is sketched in Figure 14.1.
In Figure 14.1a, the vortices are idealized by a periodic distribution in the spanwise
direction. The instantaneous velocity proles without transpiration at constant y and
z locations are shown by the dashed lines in Figures 14.1b and 14.1c, respectively.
Clearly, the U (y
o
, z) prole is inectional, having two inection points per wave-
length. At z
1
and z
3
, an inectional U(y) prole is also evident. The same proles
with suction at z
1
and z
3
and injection at z
2
are shown by the solid lines. In all cases,
the shear associated with the inection points would have been reduced. Because
the inectional proles are all inviscidly unstable with growth rates proportional
to the shear, the resulting instabilities would be weakened by the suction or injection
process.
The feasibility of the selective suction as a drag-reducing concept has been demon-
strated by Gad-el-Hak and Blackwelder (1989) and is indicated in Figure 14.2. Low-
speed streaks were articially generated in a laminar boundary layer using three span-
wise suction holes as per the method proposed by Gad-el-Hak and Hussain (1986),
and a hot-lm probe was used to record the near-wall signature of the streaks. An
open, feedforward control loop with a phase lag was used to activate a predeter-
mined suction from a longitudinal slot located in between the spanwise holes and
the downstream hot-lm probe. An equivalent suction coefcient of C
q
= 0.0006
Figure 14.1: Effects of suction or injection on velocity proles. Broken lines: reference
proles. Solid lines: proles with transpiration applied. (a) Streamwise vortices in the
yz plane, suction or injection applied at z
1
, z
2
, and z
3
. (b) Resulting spanwise velocity
distribution at y = y
o
. (c) Velocity proles normal to the surface.
was sufcient to eliminate the articial events and prevent bursting. This rate is ve
times smaller than the asymptotic suction coefcient for a corresponding turbulent
boundary layer. If this result is sustained in a naturally developing turbulent boundary
layer, a skin-friction reduction of close to 60% would be attained. Gad-el-Hak and
Blackwelder (1989) proposed to combine suction with nonplanar surface modica-
tions. Minute longitudinal roughness elements, if properly spaced in the spanwise
direction, greatly reduce the spatial randomness of the low-speed streaks (Johansen
and Smith 1986). By withdrawing the streaks forming near the peaks of the rough-
ness elements, less suction should be required to achieve an asymptotic boundary
layer. Experiments by Wilkinson and Lazos (1987) and Wilkinson (1988) combine
suction or blowing with thin-element riblets. Although no net drag reduction has
yet been attained in these experiments, their results indicate some advantage of com-
bining suction with riblets, as proposed by Gad-el-Hak and Blackwelder (1987c,
1989).
The recent numerical experiments of Choi, Moin, and Kim (1994) also validated
the concept of targeting suction or injection to specic near-wall events in a turbu-
lent channel ow. On the basis of complete interior ow information and using the
rather simple, heuristic control lawproposed earlier by Gad-el-Hak and Blackwelder
(1987c), Choi et al.s (1994) direct numerical simulations indicated a 20% net drag
Figure 14.2: Effects of suction froma streamwise slot on ve articially induced burst-
like events in a laminar boundary layer (from Gad-el-Hak and Blackwelder 1989). (a)
C
q
= 0.0. (b) C
q
= 0.0006.
reduction accompanied by signicant suppression of the near-wall structures and the
Reynolds stress throughout the entire wall-bounded ow. When only wall informa-
tion was used, a drag reduction of 6% was observed, which is a rather disappointing
result if one considers that sensing and actuation took place at every grid point along
the computational wall. In a practical implementation of this technique, even fewer
wall sensors would perhaps be available measuring only a small subset of the acces-
sible information and thus requiring even more sophisticated control algorithms to
achieve the same degree of success. Low-dimensional models of the near-wall ow
(Section 14.7) and soft computing tools (Section 14.9) can help in constructing more
effective control algorithms.
Time sequences of the numerical oweld of Choi et al. (1994) indicated the
presence of two distinct drag-reducing mechanisms when selective suction or injection
is used:
1. First, deterring the sweep motion, without modifying the primary streamwise
vortices above the wall, and consequently moving the high-shear regions
from the surface to the interior of the channel, thus directly reducing the skin
friction.
2. Changing the evolution of the wall vorticity layer by stabilizing and prevent-
ing lifting of the near-wall spanwise vorticity, thus suppressing a potential
source of new streamwise vortices above the surface and interrupting a very
important regeneration mechanism of turbulence.
Three modern developments have relevance to the issue at hand. Firstly, the re-
cently demonstrated ability to convert a chaotic systemto a periodic one may provide
optimal nonlinear control strategies for further reduction in the amount of suction
(or the energy expenditure of any other active wall-modulation technique) needed
to attain a given degree of ow stabilization. This is important because, as can
be seen from Eq. (2.44) of Chapter 2, net drag reduction achieved in a turbulent
boundary layer increases as the suction coefcient decreases. Secondly, to remove
the randomly occurring low-speed streaks selectively, for example, would ultimately
require reactive control. In that case, an event is targeted, sensed, and subsequently
modulated. Microfabrication technology provides opportunities for practical imple-
mentation of the required large array of inexpensive, programmable sensoractuator
chips. Thirdly, newly introduced soft computing tools include neural networks, fuzzy
logic, and genetic algorithms and are nowmore advanced as well as more widely used
as compared to just fewyears ago. These tools could be very useful in constructing ef-
fective adaptive controllers. All three novel developments will be discussed in turn in
Sections 14.714.9.
14.5.3 Reactive Feedback Control
As was schematically depicted in Figure 3.3 of Chapter 3, a control device can be
passive, requiring no auxiliary power, or active, requiring energy expenditure. Ac-
tive control is further divided into predetermined or reactive. Predetermined control
includes the application of steady or unsteady energy input without regard to the
particular state of the ow. The control loop in this case is open, as was shown in
Figure 3.4a of Chapter 3, and no sensors are required. Because no sensed information
is being fed forward, this open control loop is not a feedforward one. Reactive control
is a special class of active control in which the control input is continuously adjusted
based on measurements of some kind. The control loop in this case can either be an
open, feedforward one (Figure 3.4b) or a closed, feedback loop (Figure 3.4c).
The distinction between feedforward and feedback is particularly important when
dealing with the control of ow structures that convect over stationary sensors and
actuators. In feedforward control, the measured variable and the controlled variable
differ. For example, the pressure or velocity can be sensed at an upstream location,
and the resulting signal is used together with an appropriate control law to trigger
an actuator that in turn inuences the velocity at a downstream position. Feedback
control, on the other hand, necessitates that the controlled variable be measured, fed
back, and compared with a reference input.
Moin and Bewley (1994) categorized reactive feedback control strategies by ex-
amining the extent to which they are based on the governing ow equations. Four
categories are discerned: adaptive, physicalmodel-based, dynamicalsystems-based,
and optimal control (Figure 3.3 of Chapter 3). Note that, except for adaptive control,
the other three categories of reactive feedback control can also be used in the feed-
forward mode or the combined feedforwardfeedback mode. Also, in a convective
environment such as that for a boundary layer, a controller will perhaps combine
feedforward and feedback information and may include elements from each of the
four classications. Each of the four categories is briey described below.
Adaptive schemes attempt to develop models and controllers via some learning
algorithm without regard to the details of the ow physics. System identication is
performed independently of the ow dynamics or the NavierStokes equations that
govern these dynamics. An adaptive controller tries to optimize a specied perfor-
mance index by providing a control signal to an actuator. To update its parameters,
the controller thus requires feedback information relating to the effects of its con-
trol. The most recent innovation in adaptive ow-control schemes involves the use
of neural networks that relate the sensor outputs to the actuator inputs through
functions with variable coefcients and nonlinear, sigmoid saturation functions. The
coefcients are updated using the so-called back-propagation algorithm, and com-
plex control laws can be represented with a sufcient number of terms. Hand tuning
is required, however, to achieve good convergence properties. The nonlinear adaptive
technique has been used with different degrees of success by Fan et al. (1993) and
Jacobson and Reynolds (1993b, 1995, 1998) to control, respectively, the transition
process and the bursting events in turbulent boundary layers. We will return to this
subject in Section 14.9.1.
Heuristic physical arguments can instead be used to establish effective control
laws. That approach obviously will work only in situations in which the dominant
physics are well understood. An example of this strategy is the active cancellation
scheme used by Gad-el-Hak and Blackwelder (1989) in a physical experiment and
by Choi et al. (1994) in a numerical experiment to reduce the drag by mitigating
the effect of near-wall vortices. As mentioned earlier, the idea is to oppose the near-
wall motion of the uid, caused by the streamwise vortices, with an opposing wall
control, thus lifting the high-shear region away from the surface and interrupting the
turbulence regeneration mechanism.
Nonlinear dynamical systems theory allows turbulence to be decomposed into a
small number of representative modes whose dynamics are examined to determine
the best control law. The task is to stabilize the attractors of a low-dimensional
approximation of a turbulent chaotic system. The best-known strategy is the
OttGrebogiYorke (OGY) method, which, when applied to simpler, small-number-
of-degrees-of-freedom systems, achieves stabilization with minute expenditure of
energy. This and other chaos control strategies, especially as applied to the more
complex turbulent ows, will be revisited in Section 14.7.2.
Finally, optimal control theory applied directly to the NavierStokes equations
can, in principle, be used to minimize a cost function in the space of the control. This
strategy provides perhaps the most rigorous theoretical framework for ow control.
As compared with other reactive control strategies, optimal control applied to the full
NavierStokes equations is also the most computer-time intensive. In this method,
feedback control laws are derived systematically for the most efcient distribution
of control effort to achieve a desired goal. Abergel and Temam (1990) developed
such an optimal control theory for suppressing turbulence in a numerically simu-
lated, two-dimensional NavierStokes ow, but their method requires impractical
full ow-eld information. Choi, Temam, Moin, and Kim (1993) developed a more
practical, wall-information-only, suboptimal control strategy that they applied to the
one-dimensional stochastic Burgers equation. Later application of the suboptimal
control theory to a numerically simulated turbulent channel ow has been reported
by Moin and Bewley (1994) and Bewley et al. (1997, 1998). The recent book edited
by Sritharan (1998) provides eight articles that focus on the mathematical aspects of
optimal control of viscous ows.
14.5.4 Required Characteristics
The randomness of the bursting events necessitates temporal phasing as well as
spatial selectivity to effect selective control. Practical applications of methods tar-
geted at controlling a particular turbulent structure to achieve a prescribed goal
would therefore require implementing a large number of surface sensors and ac-
tuators together with appropriate control algorithms. This strategy for controlling
wall-bounded turbulent ows has been advocated by, among others and in chrono-
logical order, Gad-el-Hak and Blackwelder (1987c, 1989); Lumley (1991, 1996);
Choi, Moin, and Kim (1992); Reynolds (1993); Jacobson and Reynolds (1993b,
1995); Moin and Bewley (1994); Gad-el-Hak (1994, 1996b, 1998b); McMichael
(1996); Mehregany et al. (1996); Blackwelder (1998); Delville et al. (1998); and
Perrier (1998).
It is instructive to estimate some representative characteristics of the required
array of sensors and actuators. Consider a typical commercial aircraft cruising at
a speed of U
= 300 m/s and at an altitude of 10 km. The density and kinematic

viscosity of air and the unit Reynolds number in this case are, respectively, =
0.4 kg/m
3
, = 3 10
5
m
2
/s, and Re = 10
7
/m. Assume further that the portion
of fuselage to be controlled has turbulent boundary layer characteristics that are
identical to those for a zero-pressure-gradient at plate at a distance of 1 m from
the leading edge. In this case, the skin-friction coefcient
2
and the friction velocity
are, respectively, C
f
= 0.003 and u
= 11.62 m/s. At this location, one viscous wall

unit is only /u
= 2.6 m. In order for the surface array of sensors and actuators

to be hydraulically smooth, it should not protrude beyond the viscous sublayer, or
5/u
= 13 m.
Wall-speed streaks are the most visible, reliable, and detectable indicators of the
preburst turbulence production process. The detection criterion is simply lowvelocity
near the wall, and the actuator response should be to accelerate (or to remove) the
low-speed region before it breaks down. Local wall motion, tangential injection,
suction, heating or electromagnetic body force, all triggered on sensed wall-pressure
or wall-shear stress, could be used to cause local acceleration of near-wall uid.
The recent numerical experiments of Berkooz, Fisher, and Psiaki (1993) indicate
that effective control of bursting pairs of rolls may be achieved by using the equivalent
of two wall-mounted shear sensors. If the goal is to stabilize or to eliminate all
low-speed streaks in the boundary layer, a reasonable estimate for the spanwise
and streamwise distances between individual elements of a checkerboard array is,
respectively, 100 and 1000 wall units,
3
or 260 m and 2600 m, for our particular
example. A reasonable size for each element is probably one-tenth of the spanwise
separation, or 26 m. A(1 m1 m) portion of the surface would have to be covered
with about n = 1.5 million elements. This is a colossal number, but the density of
2
Note that the skin friction decreases as the distance fromthe leading increases. It is also strongly affected
by such things as the externally imposed pressure gradient. Therefore, the estimates provided here are
for illustration only.
3
These are equal to, respectively, the average spanwise wavelength between two adjacent streaks and
the average streamwise extent for a typical low-speed region. One can argue that these estimates
are too conservative: once a region is relaminarized, it would perhaps stay as such for quite a while
as the ow convects downstream. The next row of sensors and actuators may therefore be relegated to a
downstream location well beyond 1000 wall units. Relatively simple physical or numerical experiments
could settle this issue.
sensors and actuators could be considerably reduced if we moderated our goal of
targeting every single bursting event (and also if less conservative assumptions were
used).
It is well known that not every low-speed streak leads to a burst. On the average, a
particular sensor would detect an incipient bursting event every wall-unit interval of
P
+
=P u
2
/ =250, or P =56 s. The corresponding dimensionless and dimensional

frequencies are f
+
= 0.004 and f = 18 kHz, respectively. At different distances
from the leading edge and in the presence of nonzero-pressure gradient, the sensors
actuators array would have different characteristics, but the corresponding numbers
would still be in the same ballpark, as estimated here.
As a second example, consider an underwater vehicle moving at a speed of U
=
10 m/s. Despite the relatively low speed, the unit Reynolds number is still the same
as estimated above for the air case, Re = 10
7
/m, owing to the much lower kinematic
viscosity of water. At 1 m from the leading edge of an imaginary at plate towed
in water at the same speed, the friction velocity is only u
= 0.39 m/s, but the wall

unit is still the same as in the aircraft example, /u
= 2.6 m. The density of the

required sensorsactuators array is the same as computed for the aircraft example,
n = 1.5 10
6
elements/m
2
. The anticipated average frequency of sensing a bursting
event is, however, much lower at f = 600 Hz .
Similar calculations have been recently made by Gad-el-Hak (1993, 1994, 1998b);
Reynolds (1993); and Wadsworth et al. (1993). Their results agree closely with
the estimates made here for typical eld requirements. In either the airplane or the
submarine case, the actuators response need not be too large. As will be shown
in Section 14.8, wall displacement on the order of 10 wall units (26 m in both
examples), suction coefcient of about 0.0006, or surface cooling or heating on the
order of 40
C/2
C(in the rst and second example, respectively) should be sufcient

to stabilize the turbulent ow.
As computed in the two examples above, both the required size for a sensor
actuator element and the average frequency at which an element would be activated
are within the presently known capabilities of microfabrication technology. The num-
ber of elements needed per unit area is, however, alarmingly large. The unit cost of
manufacturing a programmable sensoractuator element would have to come down
dramatically, perhaps matching the unit cost of a conventional transistor,
4
before the
idea advocated here would become practical.
A consideration in addition to the size, amplitude, and frequency response is
the energy consumed by each sensoractuator element. Total energy consumption
by the entire control system obviously has to be low enough to achieve net sav-
ings. Consider the following calculations for the aircraft example: One meter from
the leading edge, the skin-friction drag to be reduced is approximately 54 N/m
2
.
Engine power needed to overcome this retarding force per unit area is 16 kW/m
2
,
or 10
4
W/sensor. If a 60% drag-reduction is achieved,
5
this energy consumption
is reduced to 4320 W/sensor. This number will increase by the amount of energy
consumption of a sensoractuator unit, but one hopes not back to the uncontrolled
4
The transistor was invented in 1947. In the mid-1960s, a single transistor sold for around $70. In 1997,
Intels Pentium II processor (microchip) contained 7.5 10
6
transistors and cost around $500, which is
less than $0.00007 per transistor!
5
A not-too-farfetched goal according to the selective suction results discussed earlier.
levels. The voltage across a sensor is typically in the range of V =0.11 V, and its
resistance in the range of R = 0.11 M. This means a power consumption by a
typical sensor in the range of P =V
2
/R = 0.110 W, which is well below the an-
ticipated power savings due to reduced drag. For a single actuator in the form of a
spring-loaded diaphragm with a spring constant of k = 100 N/m oscillating up and
down at the bursting frequency of f = 18 kHz with an amplitude of y = 26 m,
the power consumption is P = (1/2) ky
2
f = 600 W/actuator. If suction is used
instead, C
q
= 0.0006, and if a pressure difference of p = 10
4
N/m
2
is assumed
across the suction holes or slots, the corresponding power consumption for a single
actuator is P = C
q
U
p/n = 1200 W/actuator. It is clear, then, that when the

power penalty for the sensor and actuator is added to the lower-level drag, a net
savings is still achievable. The corresponding actuator power penalties for the sub-
marine example are even smaller (P = 20 W/actuator for the wall motion actuator,
and P = 40 W/actuator for the suction actuator), and larger savings are therefore
possible.
14.6 Magnetohydrodynamic Control
Magnetohydrodynamics (MHD) is the science underlying the interaction of an
electrically conducting uid with a magnetic eld. Several decades ago several re-
searchers from around the world were vigorously attempting to exploit the electric
current and thus electric power that results when a uid conductor moves in the
presence of a magnetic eld. Thus, the uid internal energy (or enthalpy) is directly
converted into electrical energy, eliminating the traditional intermediate mechanical
step. The usual turbine and generator in conventional power plants are therefore
combined in a single unitin essence an electromagnetic turbine with no moving
parts. The inverse device is an electromagnetic pump in which, in the presence of
a magnetic eld, a conducting uid is caused to move when electric current passes
through it. An excellent primer for the topic of MHD can be found in the book by
Shercliff (1965).
Examples of conducting uids (in descending order of electrical conductivities)
include liquid metals, plasmas, molten glass, and seawater. The electrical conductivity
of mercury is 10
6
mhos/m, and that for seawater is 4 mhos/m. Possible useful
applications include MHD power generation, propulsion, and liquid metals used
as coolants for magnetically conned fusion reactors. Though ctitious, the Soviet
submarine in Tom Clancys best-seller The Hunt for Red October was endowed with
an extremely quiet MHD propulsion system. With the collapse of Communism, the
interest in this eld has shifted somewhat to more peaceful uses such as ow control
or, in other words, the ability to manipulate a oweld to achieve a benecial goal
such as drag reduction, lift enhancement, and mixing augmentation. Here the so-
called Lorentz body forces that result when a conducting uid moves in the presence
of a magnetic eld are exploited to effect desired changes in the oweld (e.g., to
suppress turbulence).
Of particular interest to this chapter is the development of efcient reactive control
strategies that employ the Lorentz forces to enhance the performance of sea vessels.
14.6 MAGNETOHYDRODYNAMIC CONTROL 339
Methods to delay transition, prevent separation, and reduce skin-friction drag in tur-
bulent boundary layers are sought. Control strategies targeted toward certain coher-
ent structures in a turbulent oware particularly sought. We start here by developing
the von K arm an momentum integral equation and the instantaneous equations of
motion at the wall in the presence of Lorentz force. This should prove useful for
making quick estimates of system behavior under different operating conditions. Re-
sults are usually not as accurate as solving the differential conservation equations
themselves but are more accurate than dimensional analysis. This is followed by an
outline of a suggested reactive control strategy that exploits electromagnetic body
forces.
14.6.2 The von K arm an Equation
The Lorentz force (body force per unit volume) is given by the cross product of
the current density vector j and the magnetic ux density vector B
F = j B (14.13)
where the current is the sum of the applied and induced contributions
j = (E +u B) (14.14)
where is the electrical conductivity of the uid, E is the applied electric eld vector,
and u is the uid velocity vector. Thus, the Lorentz force is in general given by
F = (E B) +(u B) B (14.15)
Owing to the rather lowelectrical conductivity of seawater, the induced Lorentz force
is typically very small in such applications, and the only way to effect a signicant
body force is to cross the magnetic eld with an applied electric eld. For such an
application, therefore, the term electromagnetic forcing (EMHD) is commonly used
in place of MHD.
Assume a two-dimensional boundary-layer ow of a conducting uid. Further
assume that the magnetic and electric elds are generated using alternating electrodes
and magnets parallel to the streamwise direction (see Figure 14.3). Basically, both
elds are in the plane (y, z). The resulting Lorentz force has components in the
streamwise (x), normal (y), and spanwise (z) directions given by, respectively,
F
x
= (E
y
B
z
E
z
B
y
) u
B
2
z
+ B
2
y
(14.16)
F
y
= 0 vB
2
z
(14.17)
F
z
= 0 vB
z
B
y
(14.18)
where u and v are, respectively, the streamwise and normal velocity components.
Each of the rst terms on the the right-hand sides of Eqs. (14.16)(14.18) is due to
the applied electric eld. The second termis the induced Lorentz force and is negligible
for low-conductivity uids such as seawater. Therefore, for such an application and
particular arrangement of electrodes or magnets, the electromagnetic body force is
predominately in the streamwise direction, F
x
(E
y
B
z
E
z
B
y
).
We now write the continuity and streamwise momentum equations. For an in-
compressible ow and with the gravitational body force neglected
u
x
+
v
y
= 0 (14.19)
u
t
+u
u
x
+v
u
y
=
p
x
+

y
+ (E
y
B
z
E
z
B
y
) u
B
2
z
+ B
2
y
(14.20)
where is the viscous shear stress. The pressure term can be written in terms of
the velocity outside the boundary layer U
o
, which in general is a function of x and
t. Note that the total drag should be the sum of the usual viscous and form drags
plus (or minus) the Lorentz force. In Eq. (14.20) the induced Lorentz force is always
negative (for positive u). The applied force can be positive or negative, depending
on the direction of the applied electric eld E relative to the magnetic eld B. In any
case, when the Lorentz force is positive, it constitutes a thrust (negative drag).
At a given x location, the streamwise momentum equation is readily integrated
in y from the wall to the edge of the boundary layer. If we invoke the continuity
equation and use the usual denitions for the skin friction coefcient (C
f
) and the
displacement and momentum thicknesses (
and
), the resulting integral equation

reads
C
f

2
w
U
2
o
=
2
U
2
o
(U
o
)
t
+2
x
2
v
w
U
o
+2
2 +

1
U
o
U
o
x
+
2
U
2
o

0
[E
y
B
z
E
z
B
y
] dy
2
U
2
o

0
u
B
2
z
+ B
2
y
dy (14.21)
where v
w
is the injection (or suction) velocity normal to the wall. If is not constant,
it has to be included in the two integrals on the right-hand side of Eq. (14.21).
It is clear that the following effects contribute to an increase in the skin friction:
temporal acceleration of freestream, growth of momentum thickness, wall suction,
favorable pressure gradient, and positive streamwise Lorentz force. When a Lorentz
force points opposite to the ow direction, the skin friction decreases, and the ve-
locity prole tends to become more inectional. But once again remember to add
(or subtract) the drag (or thrust) due to the applied as well as induced Lorentz
forces.
Note that Eq. (14.21) is valid for steady or unsteady ows, for Newtonian or
non-Newtonian uids, and for laminar or turbulent ows (mean quantities are used
in the latter case). The two integrals in Eq. (14.21) can be computed once the electric
and magnetic elds are known and an approximate velocity prole is assumed. For
seawater, the last integral in Eq. (14.21) is neglected.
14.6.3 Vorticity Flux at the Wall
Another very useful equation is the differential momentumequation written at the
wall (y = 0). Here we assume a Newtonian uid but allow viscosity to vary spatially
and the ow to be three-dimensional. For a nonmoving wall, the instantaneous
14.6 MAGNETOHYDRODYNAMIC CONTROL 341
streamwise, normal, and spanwise momentum equations read respectively
v
w
u
y
y=0
+
p
x
y=0
y=0
u
y
y=0
[E
y
B
z
E
z
B
y
]
y=0
=
2
u
y
2
y=0
(14.22)
0 +
p
y
y=0
0 F
y
y=0
=
2
v
y
2
y=0
(14.23)
v
w
w
y
y=0
+
p
z
y=0
y=0
w
y
y=0
F
z
y=0
=
2
w
y
2
y=0
(14.24)
where F
y
|
y=0
and F
z
|
y=0
are, respectively, the wall values of the Lorentz force in
the normal and spanwise directions. In Eq. (14.22), the streamwise Lorentz force
at the wall F
x
|
y=0
is due only to the applied electric eld (the induced force is zero
at the wall because u vanishes there for nonmoving walls). The expression for F
x
in Eq. (14.22) is for an array of alternating electrodes and magnets parallel to the
streamwise direction,
6
but the streamwise Lorentz force can readily be computed for
any other conguration of magnets and electrodes. If the electric conductivity varies
spatially, the value of at the wall should be used in Eq. (14.22).
The right-hand side of Eq. (14.22) is the (negative of) wall ux of spanwise vortic-
ity, whereas that of Eq. (14.24) is the (negative of) wall ux of streamwise vorticity.
Much like transpiration, pressure gradient, or viscosity variations, the wall value of
the Lorentz force determines the sign and intensity of wall vorticity ux. Stream-
wise force contributes to spanwise vorticity ux, and spanwise force contributes to
streamwise vorticity ux. Take for example a positive streamwise Lorentz force. This
is a negative term on the left-hand side of Eq. (14.22) that makes the curvature of
the streamwise velocity prole at the wall more negative (i.e., instantaneously fuller
velocity prole). The wall is then a source of spanwise vorticity, and the ow is more
resistant to transition and separation. In a turbulent ow, a positive streamwise
Lorentz force leads to suppression of normal and tangential Reynolds stresses.
14.6.4 EMHD Tiles for Reactive Control
In this subsection, I will outline a reactive control strategy that exploits the Lorentz
forces to modulate the ow of an electrically conducting uid such as seawater. The
idea here is to target the low-speed streaks in the near-wall region of a turbulent
boundary layer. The electric eld is applied only when and where it is needed, and
hence the power consumed by the reactive control system is kept far below that con-
sumed by a predetermined active control system. According to recent numerical and
experimental results, brute force application of steady or time-dependent Lorentz
forceto reduce drag, for exampledoes not achieve the break-even point. The
reason for this is the high energy expenditure by predetermined control systems (see,
for example, the two meeting proceedings edited by Gerbeth 1997 and Meng 1998).
6
Note that for such an array, the applied part of both F
y
|
y=0
and F
z
|
y=0
is identically zero, but the induced
part is nonzero if v
w
= 0. In seawater applications, the induced Lorentz force is negligible regardless of
any suction, injection, or normal wall motion.
Figure 14.3: EMHD tiles for reactive control of turbulent boundary layers.
The top view in Figure 14.3 depicts the distribution of EMHD tiles to achieve
targeted control in a turbulent wall-bounded ow. Each tile consists of two stream-
wise strips constituting the north and south poles of a permanent magnet and another
two strips constituting positive and negative electrodes.
7
The tiles are staggered in
a checkerboard conguration and are separated by 100 wall units in the spanwise
direction (z) and 1000 wall units in the streamwise direction (x). The length and
width of each tile is about 50 wall units.
8
A shear stress sensor having a spatial resolution of 25 wall units is placed just
upstream and along the centerline of each tile. Each sensor and corresponding actu-
ator are connected via a closed-feedback control loop. The tile is activated to give
a positive or negative streamwise Lorentz force when a low- or high-speed region
is detected. The control law used can be based on simple physical arguments or a
more complex self-learning neural network. Other possible, albeit more sophisti-
cated, control laws can be based on nonlinear dynamical systems theory or optimal
or suboptimal control theory.
For a simpler physical or numerical experiment that does not require sensors and
complex closed-loop control, a single low-speed streak is articially generated in a
laminar boundary layer using two suction holes, as depicted in the (xz) view in
Figure 14.4. The two holes are separated in the spanwise direction by 100 wall units,
and each is about 0.5 wall unit in diameter. Such a method was successfully used by
Gad-el-Hak and Hussain (1986) to generate articial low-speed streaks as well as sim-
ulated bursting events in a laminar environment. In this same experiment, high-speed
regions and counterrotating streamwise vortices are also generated, and, if desired,
one can readily target any of these simulated events for elimination. The necessary
actuation would of course depend on the kind of coherent structure to be targeted.
A single EMHD tile is placed about 50 wall units directly downstream of the
suction holes and is activated with a suitable time delay after a suction pulse is
7
The cathode and anode are interchangeable, depending on the desired direction of the Lorentz force.
8
The length and width of each magnet pole (or electrode) are thus 50 and 12.5 wall units, respectively.
14.7 CHAOS CONTROL 343
Figure 14.4: Control of a single articial low-speed streak.
appliedin essence using a simple, predetermined, open-loop control. A single elec-
tric pulse
9
triggers a positive Lorentz force to eliminate the articially generated
low-speed streak. The experiment would of course be repeated several times, and the
actuator location, phase lag, strength, and duration could be optimized to achieve
the desired goal.
The streamwise Lorentz force necessary to eliminate the resulting articial low-
speed streak could be computed from the measurements, and thus the energy expen-
diture necessary to eliminate all the low-speed streaks in a real turbulent boundary
layer could readily be estimated.
14.7 Chaos Control
14.7.1 Nonlinear Dynamical Systems Theory
In the theory of dynamical systems, the so-called buttery effect denotes sensitive
dependence of nonlinear differential equations on initial conditions with phasespace
solutions initially very close together separating exponentially. The solution of non-
linear dynamical systems of three or more degrees of freedom may be in the form of
a strange attractor whose intrinsic structure contains a well-dened mechanism to
produce a chaotic behavior without requiring random forcing. Chaotic behavior is
complex, aperiodic and, though deterministic, appears to be random.
A question arises naturally. Just as small disturbances can radically grow within
a deterministic system to yield rich, unpredictable behavior, can minute adjustments
to a system parameter be used to reverse the process and control (i.e., regularize)
the behavior of a chaotic system? Recently, that question was answered in the afr-
mative theoretically as well as experimentallyat least for system orbits that reside
on low-dimensional strange attractors (see the review by Lindner and Ditto 1995).
Before describing such strategies for controlling chaotic systems, we rst summarize
the recent attempts to construct a low-dimensional dynamical systems representa-
tion of turbulent boundary layers. Such construction is a necessary rst step to be
able to use chaos-control strategies for turbulent ows. Additionally, as argued by
9
Note that in this simulated environment the electric pulse would have the same polarity if only low-speed
streaks were to be obliterated.
Lumley (1996b), a low-dimensional dynamical model of the near-wall region used
in a Kalman lter (Banks 1986; Petersen and Savkin 1999) can make the most of
the partial information assembled from a nite number of wall sensors. Such a lter
minimizes in a least-squares sense the errors caused by incomplete information and
thus globally optimizes the performance of the control system.
Boundary layer turbulence is described by a set of nonlinear partial differential
equations and is characterized by an innite number of degrees of freedom. This
makes it rather difcult to model the turbulence using a dynamical systems approx-
imation. The notion that a complex, innite-dimensional ow can be decomposed
into several low-dimensional subunits is, however, a natural consequence of the real-
ization that quasi-periodic coherent structures dominate the dynamics of seemingly
randomturbulent shear ows. This implies that low-dimensional, localized dynamics
can exist in formally innite-dimensional extended systems such as open turbulent
ows. Reducing the ow physics to nite-dimensional dynamical systems enables a
study of its behavior through an examination of the xed points and the topology of
their stable and unstable manifolds. From the dynamical systems theory viewpoint,
the meandering of low-speed streaks is interpreted as hovering of the ow state near
an unstable xed point in the low-dimensional state space. An intermittent event
that produces high wall stressa burstis interpreted as a jump along a heteroclinic
cycle to a different unstable xed point that occurs when the state has wandered
too far from the rst unstable xed point. Delaying this jump by holding the system
near the rst xed point should lead to lower momentum transport in the wall re-
gion and, therefore, to lower skin-friction drag. Reactive control means sensing the
current local state and through appropriate manipulation keeping the state close to
a given unstable xed point, thereby preventing further production of turbulence.
Reducing the bursting frequency by say 50% may lead to a comparable reduction
in skin-friction drag. For a jet, relaminarization may lead to a quiet ow and very
signicant noise reduction.
In one signicant attempt the proper orthogonal, or KarhunenLo` eve, decom-
position method has been used to extract a low-dimensional dynamical system
from experimental data of the wall region (Aubry et al. 1988; Aubry 1990). Aubry
et al. (1988) expanded the instantaneous velocity eld of a turbulent boundary layer
using experimentally determined eigenfunctions which are in the form of stream-
wise rolls. They expanded the NavierStokes equations using these optimally cho-
sen, divergence-free, orthogonal functions applied a Galerkin projection and then
truncated the innite-dimensional representation to obtain a 10-dimensional set of
ordinary differential equations. These equations represent the dynamical behavior of
the rolls and are shown to exhibit a chaotic regime as well as an intermittency due to
a burst-like phenomenon. However, Aubry et al.s (1988) 10-mode dynamical system
displays a regular intermittency in contrast both to that in actual turbulence as well as
to the chaotic intermittency encountered by Pomeau and Manneville (1980) in which
event durations are distributed stochastically. Nevertheless, the major conclusion of
Aubry et al.s (1988) study is that the bursts appear to be produced autonomously
by the wall region even without turbulence but are triggered by turbulent pressure
signals from the outer layer. More recently, Berkooz et al. (1991) generalized the
class of wall-layer models developed by Aubry et al. (1988) to permit uncoupled
evolution of streamwise and cross-stream disturbances. Berkooz et al.s results sug-
gest that the intermittent events observed in Aubry et al.s representation do not arise
solely because of the effective closure assumption incorporated but are rather rooted
deeper in the dynamical phenomena of the wall region. The book by Holmes et al.
(1996) details the Cornell research groups attempts at describing turbulence as a
low-dimensional dynamical system.
In addition to the reductionist viewpoint exemplied by the work of Aubry et al.
(1988) and Berkooz et al. (1991), attempts have been made to determine directly the
dimension of the attractors underlying specic turbulent ows. Again, the central
issue here is whether or not turbulent solutions to the innite-dimensional Navier
Stokes equations can be asymptotically described by a nite number of degrees of
freedom. Grappin and L eorat (1991) computed the Lyapunov exponents and the
attractor dimensions of two- and three-dimensional periodic turbulent ows without
shear. They found that the number of degrees of freedom contained in the large
scales establishes an upper bound for the dimension of the attractor. Deane and
Sirovich (1991) and Sirovich and Deane (1991) numerically determined the number
of dimensions needed to specify chaotic RayleighB enard convection over a moderate
range of Rayleigh numbers, Ra. They suggested that the intrinsic attractor dimension
is O[Ra
2/3
].
The corresponding dimension in wall-bounded ows appears to be dauntingly
high. Keefe, Moin, and Kim (1992) determined the dimension of the attractor un-
derlying turbulent Poiseuille ows with spatially periodic boundary conditions. Us-
ing a coarse-grained numerical simulation, they computed a lower bound on the
Lyapunov dimension of the attractor to be approximately 352 at a pressure-gradient
Reynolds number of 3200. Keefe et al. (1992) argued that the attractor dimension in
fully-resolved turbulence is unlikely to be much larger than 780. This suggests that
periodic turbulent shear ows are deterministic chaos and that a strange attractor
does underlie solutions to the NavierStokes equations. Temporal unpredictability
in the turbulent Poiseuille ow is thus due to the exponential spreading property of
such attractors. Although nite, the computed dimension invalidates the notion that
the global turbulence can be attributed to the interaction of a fewdegrees of freedom.
Moreover, in a physical channel or boundary layer, the ow is not periodic and is
open. The attractor dimension in such a case is not known but is believed to be even
higher than the estimate provided by Keefe et al. for the periodic (quasi-closed) ow.
In contrast to closed, absolutely unstable ows such as TaylorCouette systems in
which the number of degrees of freedom can be small, local measurements in open,
convectively unstable ows, such as boundary layers, do not express the global dy-
namics, and the attractor dimension in that case may inevitably be too large to be
determined experimentally. According to the estimate provided by Keefe et al. (1992),
the colossal data required (about 10
D
, where Dis the attractor dimension) for mea-
suring the dimension simply exceed current computer capabilities. Turbulence near
transition or near a wall is an exception to that bleak picture. In those special cases,
a relatively small number of modes are excited, and the resulting simple turbulence
can therefore be described by a dynamical system of a reasonable number of degrees
of freedom.
14.7.2 Chaos Control
There is another question of greater relevance here. Given a dynamical system
in the chaotic regime, is it possible to stabilize its behavior through some kind of
active control? Although other alternatives have been devised (e.g., Fowler 1989;
H ubler and L uscher 1989; Huberman 1990; Huberman and Lumer 1990), the recent
method proposed by workers at the University of Maryland (Ott, Grebogi, and Yorke
1990a,b; Shinbrot et al. 1990, 1992a,b, 1998; Shinbrot, Ditto, Grebogi, Ott, Spano,
and Yorke 1992; Romeiras et al. 1992) promises to be a signicant breakthrough.
Comprehensive reviews and bibliographies of the emerging eld of chaos control can
be found in the articles by Shinbrot et al. (1993); Shinbrot (1993, 1995, 1998); and
Lindner and Ditto (1995).
Ott et al. (1990a) demonstrated, through numerical experiments with the Henon
map, that it is possible to stabilize a chaotic motion about any prechosen unstable
orbit through the use of relatively small perturbations. The procedure consists of
applying minute time-dependent perturbations to one of the system parameters to
control the chaotic system around one of its many unstable periodic orbits. In this
context, targeting refers to the process whereby an arbitrary initial condition on
a chaotic attractor is steered toward a prescribed point (target) on this attractor.
The goal is to reach the target as quickly as possible using a sequence of small
perturbations (Kostelich et al. 1993a).
The success of the OttGrebogiYorke (OGY) strategy for controlling chaos
hinges on the fact that beneath the apparent unpredictability of a chaotic system
lies an intricate but highly ordered structure. Left to its own recourse, such a system
continually shifts from one periodic pattern to another, creating the appearance of
randomness. An appropriately controlled system, on the other hand, is locked into
one particular type of repeating motion. With such reactive control the dynamical
system becomes one with a stable behavior. The OGY method can be simply illus-
trated by the schematic in Figure 14.5. The state of the system is represented as
the intersection of a stable manifold and an unstable one. The control is applied
intermittently whenever the system departs from the stable manifold by a prescribed
tolerance; otherwise, the control is shut off. The control attempts to put the sys-
tem back onto the stable manifold so that the state converges toward the desired
trajectory. Unmodeled dynamics cause noise in the system and a tendency for the
state to wander off in the unstable direction. The intermittent control prevents that,
and the desired trajectory is achieved. This efcient control is not unlike trying to
balance a ball in the center of a horse saddle (Moin and Bewley 1994). There is one
Figure 14.5: The OGY method for controlling
chaos.
stable direction (front or back) and one
unstable direction (left or right). The
restless horse is the unmodeled dynamics
intermittently causing the ball to move in
the wrong direction. The OGY control
needs only to be applied, in the most di-
rect manner possible, whenever the ball
wanders off in the left or right direction.
The OGY method has been success-
fully applied in a relatively simple ex-
periment by Ditto, Rauseo, and Spano
(1990) and Ditto and Pecora (1993) at
the Naval Surface Warfare Center in
which reverse chaos was obtained in
a parametrically driven, gravitationally
buckled, amorphous magnetoelastic rib-
bon. Garnkel et al. (1992) applied the
same control strategy to stabilize drug-induced cardiac arrhythmias in sections of
a rabbit ventricle. Other extensions, improvements, and applications of the OGY
strategy include higher-dimensional targeting (Auerbach et al. 1992; Kostelich et al.
1993b); controlling chaotic scattering in Hamiltonian (i.e., nondissipative, area
conservative) systems (Lai, Deng, and Grebogi 1993; Lai, T el, and Grebogi 1993);
synchronizing identical chaotic systems that govern communication, neural, or bio-
logical processes (Lai and Grebogi 1993); using chaos to transmit information
(Hayes, Grebogi, and Ott 1994; Hayes, Grebogi, Ott, and Mark 1994); controlling
transient chaos (Lai et al. 1994); and taming spatiotemporal chaos using a sparse
array of controllers (Chen, Wolf, and Chang 1993; Qin, Wolf, and Chang 1994;
Auerbach 1994).
In a more complex system, such as a turbulent boundary layer, there exist numer-
ous interdependent modes and many stable as well as unstable manifolds (directions).
The ow can then be modeled as coherent structures plus a parameterized turbulent
background. The proper orthogonal decomposition (POD) is used to model the co-
herent part because POD guarantees the minimum number of degrees of freedom
for a given model accuracy. Factors that make turbulence control a challenging task
are the potentially quite large perturbations caused by the unmodeled dynamics of
the ow, the nonstationary nature of the desired dynamics, and the complexity of
the saddle shape describing the dynamics of the different modes. Nevertheless, the
OGY-control strategy has several advantages that are of special interest in the con-
trol of turbulence: (1) the mathematical model for the dynamical system need not be
known, (2) only small changes in the control parameter are required, and (3) noise
can be tolerated (with appropriate penalty).
Recently, Keefe (1993a,b) made a useful comparison between two nonlinear con-
trol strategies as applied to uid problems: OttGrebogiYorkes feedback method
described above and the model-based control strategy originated by H ubler (see, for
example, H ubler and L uscher 1989; L uscher and H ubler 1989), the H-method. Both
novel control methods are essentially generalizations of the classical perturbation
cancellation technique: apply a prescribed forcing to subtract the undesired dynam-
ics and impose the desired one. The OGY strategy exploits the sensitivity of chaotic
systems to stabilize existing periodic orbits and steady states. Some feedback is needed
to steer the trajectories toward the chosen xed point, but the required control signal
is minuscule. In contrast, H ublers scheme does not explicitly make use of the system
sensitivity. It produces general control response (periodic or aperiodic) and needs
little or no feedback, but its control inputs are generally large. The OGY strategy
exploits the nonlinearity of a dynamical system; indeed, the presence of a strange
attractor and the extreme sensitivity of the dynamical system to initial conditions are
essential to the success of the method. In contrast, the H-method works equally for
both linear and nonlinear systems.
Keefe (1993a) rst numerically examined the two schemes as applied to fully de-
veloped and transitional solutions of the GinzburgLandau equation, an evolution
equation that governs the initially weakly nonlinear stages of transition in several
ows and that possesses both transitional and fully chaotic solutions. The Ginzburg
Landau equation has solutions that display either absolute or convective instabilities
and is thus a reasonable model for both closed and open ows. Keefes main conclu-
sion is that control of nonlinear systems is best obtained by making the maximumuse
possible of the underlying natural dynamics. If the goal dynamics is an unstable non-
linear solution of the equation and the owis nearby at the instant control is applied,
both methods perform reliably and at low-energy cost in reaching and maintaining
this goal. Predictably, the performance of both control strategies degrades owing
to noise and the spatially discrete nature of realistic forcing. Subsequently, Keefe
(1993b) extended the numerical experiment in an attempt to reduce the drag in a
channel ow with spatially periodic boundary conditions. The OGY method reduces
the skin friction to 6080% of the uncontrolled value at a mass-ux Reynolds num-
ber of 4408. The H-method fails to achieve any drag reduction when starting from
a fully turbulent initial condition but shows potential for suppressing or retarding
laminar-to-turbulence transition. Keefe (1993a) suggested that the H-strategy might
be more appropriate for boundary layer control, whereas the OGY-method might
best be used for channel ows.
It is also relevant to note here the work of Bau and his colleagues at the University
of Pennsylvania (Singer, Wang, and Bau 1991; Wang, Singer, and Bau 1992) who
devised a feedback control to stabilize (relaminarize) the naturally occurring chaotic
oscillations of a toroidal thermal convection loop heated frombelowand cooled from
above. On the basis of a simple mathematical model for the thermosyphon, Bau and
his colleagues constructed a reactive control system that was used to alter the ow
characteristics inside the convection loop signicantly. Their linear control strategy,
perhaps a special version of the OGYchaos control method, consists simply of sensing
the deviation of uid temperatures fromdesired values at a number of locations inside
the thermosyphon loop and then altering the wall heating either to suppress or to
enhance such deviations. Wang et al. (1992) also suggested extending their theoretical
and experimental method to more complex situations such as those involving B enard
convection (Tang and Bau 1993a,b). Hu and Bau (1994) used a similar feedback
control strategy to demonstrate that the critical Reynolds number for the loss of
stability of planar Poiseuille ow can be signicantly increased or decreased.
Other attempts to use low-dimensional dynamical systems representation for ow
control include the work of Berkooz et al. (1993); Corke, Glauser, and Berkooz
(1994); and Coller, Holmes, and Lumley (1994a,b). Berkooz et al. (1993) applied
techniques of modern control theory to estimate the phase-space location of dynam-
ical models of the wall-layer coherent structures and used these estimates to control
the model dynamics. Because discrete wall sensors provide incomplete knowledge
of phase-space location, Berkooz et al. maintained that a nonlinear observer, which
incorporates past information and the equations of motion into the estimation proce-
dure, is required. Using an extended Kalman lter, they achieved effective control of
a bursting pair of rolls with the equivalent of two wall-mounted shear sensors. Corke
et al. (1994) used a low-dimensional dynamical system based on the proper orthogo-
nal decomposition to guide control experiments for an axisymmetric jet. By sensing
the downstream velocity and actuating an array of miniature speakers located at the
lip of the jet, their feedback control succeeded in converting the near-eld instabili-
ties from spatial-convective to temporal-global. Coller et al. (1994a,b) developed a
feedback control strategy for strongly nonlinear dynamical systems, such as turbulent
ows, subject to small randomperturbations that kick the systemintermittently from
one saddle point to another along heteroclinic cycles. In essence, their approach is to
use local, weakly nonlinear feedback control to keep a solution near a saddle point as
long as possible but then to let the natural, global nonlinear dynamics run their course
when bursting (in a low-dimensional model) does occur. Though conceptually related
to the OGY strategy, Coller et al.s method does not actually stabilize the state but
merely holds the system near the desired point longer than it would otherwise stay.
14.8 MICROFABRICATION 349
Shinbrot and Ottino (1993a,b) offered yet another strategy presumably most
suited to controlling coherent structures in area-preserving turbulent ows. Their
geometric method exploits the premise that the dynamical mechanisms that produce
the organized structures can be remarkably simple. By repeated stretching and fold-
ing of horseshoes that are present in chaotic systems, Shinbrot and Ottino have
demonstrated numerically as well as experimentally the ability to create, destroy, and
manipulate coherent structures in chaotic uid systems. The key idea to create such
structures is to place folds of horseshoes near low-order periodic points intentionally.
In a dissipative dynamical system, volumes contract in state space, and the colocation
of a fold with a periodic point leads to an isolated region that contracts asymptoti-
cally to a point. Provided that the folding is done properly, it counteracts stretching.
Shinbrot and Ottino (1993a) applied the technique to three prototypical problems:
a one-dimensional chaotic map, a two-dimensional one, and a chaotically advected
uid. Shinbrot (1995, 1998) and Shinbrot et al. (1998) provided recent reviews of
the stretching and folding as well as other chaos control strategies.
14.8 Microfabrication
This topic was covered in more detail in the previous chapter. Manufacturing pro-
cesses that can create extremely small machines have been developed in recent years
(Angell et al. 1983; Gabriel et al. 1988, 1992; OConnor 1992; Gravesen et al.
1993; Bryzek et al. 1994; Gabriel 1995; Hogan 1996; Ho and Tai 1996, 1998;
Tien 1997; Busch-Vishniac 1998; Amato 1998). In this emerging microfabrication
technology, under intensive development only since 1990, electronic and mechanical
components are combined on a single silicon chip using photolithographic micro-
machining techniques. Motors, electrostatic actuators, pneumatic actuators, valves,
gears, and tweezers of typical size O[10 m] have been fabricated. These have been
used as sensors for pressure, temperature, velocity, mass ow, or sound, and as ac-
tuators for linear and angular motions. Current usage for microelectromechanical
systems (MEMS) includes accelerometers for airbags and guidance systems, pressure
sensors for engine air intake and blood analysis, rate gyroscopes for antilock brakes,
microrelays and microswitches for semiconductor automatic test equipment, and
microgrippers for surgical procedures (OConnor 1992; Hogan 1996; Paula 1996;
Ouellette 1996; Ashley 1996; Robinson, Helvajian, and Jansen 1996a,b). There is
considerable work under way to include other applications, one example being the
microsteam engine described by Lipkin (1993). A second example is the 3 cm
1.5 cm digital light processor that contains 0.52 million individually addressable
micromirrors, each measuring 16 mm. Texas Instruments, Inc., is currently
producing such a device with a resolution of 2000 1000 pixels, for high-denition
televisions and other display equipments. The company maintains that when mass
produced, such device costs on the order of $100 (i.e., less than $0.0001 per
actuator).
The new Journal of Microelectromechanical Systems and Journal of Microme-
chanics and Microengineering are dedicated to this technology, and the older Sensors
and Actuators is increasingly allotting more of its pages to MEMS. Entire sessions in
scientic meetings have been increasingly assigned to MEMS applications in uid
mechanics (see, for example, the presentations by McMichael, Tai, Mehregany,
Mastrangelo, and Yun, all made at the AIAA Third Shear Flow Control Conference,
Orlando, Florida, 69 July 1993, and the volumes edited by Bandyopadhyay, Breuer,
and Blechinger 1994 and Breuer, Bandyopadhyay, and Gad-el-Hak 1996). Recent
reviews of the use, or potential use, of MEMS in ow control include those by
Gad-el-Hak (1994, 1996b); Lumley (1996); McMichael (1996); Mehregany et al.
(1996); and Ho and Tai (1996, 1998).
Microelectromechanical systems would be ideal for the reactive ow-control con-
cept advocated in this chapter. Methods of ow control targeted toward specic
coherent structures involve nonintrusive detection and subsequent modulation of
events that occur randomly in space and time. To achieve proper targeted control
of these quasi-periodic vortical events, temporal phasing as well as spatial selectiv-
ity are required. Practical implementation of such an idea necessitates the use of a
large number of intelligent, communicative wall sensors and actuators arranged in
a checkerboard pattern. Section 14.5 provides estimates for the number, character-
istics, and energy consumption of such elements required to modulate the turbulent
boundary layer that develops along a typical commercial aircraft or nuclear sub-
marine. An upper-bound number to achieve total turbulence suppression is about
one million sensors or actuators per square meter of the surface, although as argued
earlier the actual number needed to achieve effective control could perhaps be 12
orders of magnitude below that.
The sensors would be expected to measure the amplitude, location, and phase
or frequency of the signals impressed upon the wall by incipient bursting events.
Instantaneous wall-pressure or wall-shear stress can be sensed, for example. The
normal or in-plane motion of a minute membrane is proportional to the respective
point force of primary interest. For measuring wall pressure, microphone-like devices
respond to the motion of a vibrating surface membrane or an internal elastomer.
Several types are available, including variable capacitance (condenser or electret),
ultrasonic, optical (e.g., optical-ber and diode-laser), and piezoelectric devices (see,
for example, L ofdahl et al. 1993, 1994; L ofdahl and Gad-el-Hak 1999). Apotentially
useful technique for our purposes has been tried at MIT (Warkentin et al. 1987;
Young et al. 1988; Haritonidis et al. 1990a,b). An array of extremely small (0.2
mm in diameter) laser-powered microphones (termed picophones) was machined in
silicon using integrated-circuit fabrication techniques and used for eld measurement
of the instantaneous surface pressure in a turbulent boundary layer. The wall-shear
stress, though smaller and therefore more difcult to measure than pressure, provides
a more reliable signature of the near-wall events.
Actuators are expected to produce a desired change in the targeted coherent
structures. The local acceleration action needed to stabilize an incipient bursting
event can be in the form of adaptive wall, transpiration, wall-heat transfer, or elec-
tromagnetic body force. Traveling surface waves can be used to modify a locally
convecting pressure gradient such that the wall motion follows that of the coherent
event causing the pressure change. Surface motion in the formof a Gaussian hill with
height y
+
= O[10] should be sufcient to suppress typical incipient bursts (Lumley
1991b; Carlson and Lumley 1996). Such time-dependent alteration in wall geome-
try can be generated by driving a exible skin using an array of piezoelectric devices
(which dilate or contract depending on the polarity of current passing through them),
electromagnetic actuators, magnetoelastic ribbons (made of nonlinear materials that
change their stiffness in the presence of varying magnetic elds), or Terfenol-d rods
(a novel metal composite developed at Grumman Corporation that changes its length
14.8 MICROFABRICATION 351
when subjected to a magnetic eld). Other exotic materials that can be used for ac-
tuation should also be noted. For example, electrorheological uids (Halsey and
Martin 1993) instantly solidify when exposed to an electric eld and may thus be
useful for the present application. Recently constructed microactuators specically
designed for ow control include those by Wiltse and Glezer (1993); James et al.
(1994); Jacobson and Reynolds (1995); Vargo and Muntz (1996); and Keefe (1996).
Suction or injection at many discrete points can be achieved simply by connecting
a large number of minute streamwise slots arranged in a checkerboard pattern to
a low-pressure or high-pressure reservoir located underneath the working surface.
The transpiration through each individual slot is turned on and off using a corre-
sponding number of independently controlled microvalves. Alternatively, positive-
displacement or rotary micropumps (see, for example, Sen et al. 1996; Sharatchandra
et al. 1997) can be used for blowing or sucking uid through small holes or slits. On
the basis of the results of Gad-el-Hak and Blackwelder (1989), equivalent suction
coefcients of about 0.0006 should be sufcient to stabilize the near-wall region. If
it is assumed that the skin-friction coefcient in the uncontrolled boundary layer is
C
f
= 0.003 and further assumed that the suction used is sufcient to establish an
asymptotic boundary layer (d
/dx = 0, where
is the momentum thickness), the

skin friction in the reactively controlled case is then C
f
= 0+2C
q
= 0.0012, or 40%
of the original value. The net benet will, of course, be reduced by the energy ex-
penditure of the suction pump (or micropumps) as well as the array of microsensors
and microvalves.
Finally, if the bursting events are to be eliminated by lowering the near-wall viscos-
ity, direct electric-resistance heating can be used in liquid ows, and thermoelectric
devices based on the Peltier effect can be used for cooling in the case of gaseous
boundary layers. The absolute viscosity of water at 20

C decreases by approxi-
mately 2% for each 1

C rise in temperature, whereas for room-temperature air,
decreases by about 0.2% for each 1

C drop in temperature. The streamwise mo-
mentumequation written at the wall can be used to showthat a suction coefcient of
0.0006 has approximately the same effect on the wall curvature of the instantaneous
velocity prole as a surface heating of 2

C in water or a surface cooling of 40

C in
air (Liepmann and Nosenchuck 1982; Liepmann et al. 1982).
Sensors and actuators of the types discussed in this section can be combined on
individual electronic chips using microfabrication technology. The chips can be in-
terconnected in a communications network that is controlled by a massively parallel
computer or a self-learning neural network (Section 14.9), perhaps each sensor or ac-
tuator unit communicating only with its immediate neighbors. In other words, it may
not be necessary for one sensor or actuator to exchange signals with another far-away
unit. Factors to be considered in an eventual eld application of chips produced using
microfabrication processes include sensitivity of sensors, sufciency and frequency
response of actuators action, fabrication of large arrays at affordable prices, surviv-
ability in the hostile eld environment, and energy required to power the sensors or
actuators. As argued by Gad-el-Hak (1994, 1996b) and in Chapter 13, sensor and
actuator chips currently produced are small enough for typical eld application, and
they can be programmed to provide a sufciently large or fast action in response to a
certain sensor output (see also Jacobson and Reynolds 1995). Present prototypes are,
however, still quite expensive as well as delicate. But so was the transistor when rst
introduced! It is hoped that the unit price of future sensor and actuator elements will
follow the same dramatic trends witnessed in case of the simple transistor and even
the much more complex integrated circuit. The price anticipated by Texas Instru-
ments for an array of one million mirrors hints that the technology is well in its way
to mass produce phenomenally inexpensive microsensors and microactuators. Addi-
tionally, current automotive applications are a rigorous proving ground for MEMS:
under-the-hood sensors can already withstand harsh conditions such as intense heat,
shock, continual vibration, corrosive gases, and electromagnetic elds.
14.9 Soft Computing
The term soft computing was coined by Lot Zadeh of the University of California,
Berkeley, to describe several ingenious modes of computations that exploit tolerance
for imprecision and uncertainty in complex systems to achieve tractability, robust-
ness, and low cost (Yager and Zadeh 1992; Bouchon-Meunier, Yager, and Zadeh
1995a,b; Jang, Sun, and Mizutani 1997). The principle of complexity provides the
impetus for soft computing: as the complexity of a systemincreases, the ability to pre-
dict its response diminishes until a threshold is reached beyond which precision and
relevance become almost mutually exclusive (Noor and Jorgensen 1996). In other
words, precision and certainty carry a cost. By employing modes of reasoning
probabilistic reasoningthat are approximate rather than exact, soft computing can
help in searching for globally optimal design or achieving effectual control while
taking into account system uncertainties and risks.
Soft computing refers to a domain of computational intelligence that loosely lies
in between purely numerical (hard) computing and purely symbolic computations.
Alternatively, one can think about symbolic computations as a form of articial
intelligence lying in between biological intelligence and computational intelligence
(soft computing). The schematic in Figure 14.6 illustrates the general idea. Articial
intelligence relies on symbolic information-processing techniques and uses logic as
representation and inference mechanisms. It attempts to approach the high level of
human cognition. In contrast, soft computing is based on modeling low-level cogni-
tive processes and strongly emphasizes modeling of uncertainty as well as learning.
Computational intelligence mimics the ability of the human brain to employ modes
of reasoning that are approximate. Soft computing provides a machinery for the
numeric representation of the types of constructs developed in the symbolic articial
intelligence. The boundaries between these paradigms are of course fuzzy.
The principal constituents of soft computing are neurocomputing, fuzzy logic,
and genetic algorithms, as depicted in Figure 14.6. These elements, together with
probabilistic reasoning, can be combined in hybrid arrangements, resulting in better
systems in terms of parallelism, fault tolerance, adaptivity, and uncertainty manage-
ment. To my knowledge, only neurocomputing has been employed for uid-ow
control, but the other tools of soft computing may be just as useful for construct-
ing powerful controllers and have in fact been used as such in other elds such as
large-scale subway controllers and video cameras. A brief description of these three
constituents follows.
Neurocomputing is inspired by the neurons of the human brain and how they
work. Neural networks are information-processing devices that can learn by adapting
synaptic weights to changes in the surrounding environment; can handle imprecise,
14.9 SOFT COMPUTING 353
Figure 14.6: Tools for soft computing.
fuzzy, noisy, and probabilistic information; and can generalize from known tasks
(examples) to unknown ones. Actual engineering-oriented hardware are termed ar-
ticial neural networks (ANN), whereas algorithms are called computational neural
networks (CNN). The nonlinear, highly parallel networks can perform any of the
following tasks: classication, pattern matching, optimization, control, and noise re-
moval. As modeling and optimization tools, neural networks are particularly useful
when good analytic models are either unknown or extremely complex.
An articial neural network consists of a large number of highly interconnected
processing elementsessentially equations known as transfer functionsthat are
analogous to neurons and are tied together with weighted connections that are
analogous to synapses. Aprocessing unit takes weighted signals fromother units, pos-
sibly combines them, and gives a numeric result. The behavior of neural networks
how they map input datais inuenced primarily by the transfer functions of the
processing elements, how the transfer functions are interconnected, and the weights
of those interconnections. Learning typically occurs by examplethrough exposure
to a set of inputoutput data for which the training algorithm adjusts the connection
weights (synapses). These connection weights store the knowledge necessary to solve
specic problems. As an example, it is now possible to use neural networks to sense
(smell) odors in many different applications (Ouellette 1999). The electronic noses
(e-noses) are on the verge of nding commercial applications in medical diagnostics,
environmental monitoring, and the processing and quality control of foods. Neural
networks, as used in uid-ow control, will be covered in the following subsection.
Fuzzy logic was introduced by Lot Zadeh in 1965 as a mathematical tool to
deal with uncertainty and imprecision. The book by Yager and Zadeh (1992) is an
excellent primer to the eld. For computing and reasoning, general concepts (such
as size) are implemented into a computer algorithm by using mostly words (such as
small, medium, or large). Fuzzy logic, therefore, provides a unique methodology for
computing with words. Its ratiocination is based on three mathematical concepts:
fuzzy sets, membership function, and possibility. As dictated by a membership func-
tion, fuzzy sets allow a gradual transition from belonging to not belonging to a set.
The concept of possibility provides a mechanism for interpreting factual statements
involving fuzzy sets. Three processes are involved in solving a practical problemusing
fuzzy logic: fuzzication, analysis, and defuzzication. Given a complex, unsolvable
problem in real space, these three steps entail enlarging the space and searching for
a solution in the new superset and then specializing this solution to the original real
constraints.
Genetic algorithms are search algorithms based loosely on the mechanics of nat-
ural selection and natural genetics. They combine survival of the ttest among string
structures with structured yet randomized information exchange and are used for
search, optimization, and machine learning. For control, genetic algorithms aim at
achieving minimum cost function and maximum performance measure while satis-
fying the problem constraints. The books by Goldberg (1989); Davis (1991); and
Holland (1992) provide a gentle introduction to the eld.
In the Darwinian principle of natural selection, the ttest members of a species are
favored to produce offspring. Even biologists cannot help but be awed by the com-
plexity of life observed to evolve in the relatively short time suggested by the fossil
records. A living being is an amalgam of characteristics determined by the (typically
tens of thousands) genes in its chromosomes. Each gene may have several forms or al-
ternatives called alleles that produce differences in the set of characteristics associated
with that gene. The chromosomes are therefore the organic devices through which
the structure of a creature is encoded, and this living being is created partly through
the process of decoding those chromosomes. Genes transmits hereditary characters
and form specic parts of a self-perpetuated deoxyribonucleic acid (DNA) in a cell
nucleus. Natural selection is the link between the chromosomes and the performance
of their decoded structures. Simply put, the process of natural selection causes those
chromosomes that encode successful structures to reproduce more often than those
that do not.
In an attempt to solve difcult problems, John H. Holland of the University of
Michigan introduced in the early 1970s the man-made version of the procedure of
natural evolution. The candidate solutions to a problem are ranked by the genetic al-
gorithmaccording to howwell they satisfy a certain criterion, and the ttest members
are the most favored to combine among themselves to form the next generation of
the members of the species. Fitter members presumably produce even tter offspring
and therefore better solutions to the problem at hand. Solutions are represented by
binary strings, and each trial solution is coded as a vector called a chromosome.
The elements of a chromosome are described as genes, and their varying values at
specic positions are called alleles. Good solutions are selected for reproduction
based on a tness function using genetic recombination operators such as crossover
14.9 SOFT COMPUTING 355
and mutation. The main advantage of genetic algorithms is their global parallelism
in which the search efforts to many regions of the search area are simultaneously
allocated.
Genetic algorithms have been used for the control of different dynamical systems
such as the optimization of robot trajectories. But to my knowledge and at the time
of writing this chapter (mid-1999), the control of uid ow is yet to benet from
this powerful soft computing tool. In particular, when a nite number of sensors are
used to gather information about the state of the ow, a genetic algorithm, perhaps
combined with a neural network, can adapt and learn to use current information to
eliminate the uncertainty created by insufcient a priori information.
14.9.1 Neural Networks for Flow Control
Biologically inspired neural networks are nding increased applications in many
elds of science and technology. Modeling of complex dynamical systems, adaptive
noise canceling in telephones and modems, bomb sniffers, mortgage-risk evaluators,
sonar classiers, and word recognizers are but a few of the existing usages of neural
nets. The book by Nelson and Illingworth (1991) provides a lucid introduction to
the eld, and the review article by Antsaklis (1993) focuses on the use of neural nets
for the control of complex dynamical systems. For ow-control applications, neural
networks offer the possibility of adaptive controllers that are simpler and potentially
less sensitive to parameter variations as compared with conventional controllers.
Moreover, if a colossal number of sensors and actuators is to be used, the mas-
sively parallel computational power of neural nets will surely be needed for real-time
control.
The basic elements of a neural network are schematically shown in Figure 14.7.
Several inputs are connected to the nodes (neurons or processing elements) that form
the input layer. There are one or more hidden layers followed by an output layer.
Note that the number of connections is higher than the total number of nodes. Both
numbers are chosen based on the particular application and can be arbitrarily large
for complex tasks. Simply put, the multitaskalbeit simplejob of each processing
element is to evaluate each of the input signals to that particular element, calculate
the weighted sumof the combined inputs, compare that total to some threshold level,
Figure 14.7: Elements of a neural network.
and nally determine what the output should be. The various weights are the adaptive
coefcients, which vary dynamically as the network learns to perform its assigned
task; some inputs are more important than others. The threshold, or transfer, function
is generally nonlinear; the most common one is the continuous sigmoid, or S-shaped,
curve, which approaches a minimum and maximum value at the asymptotes. If the
sum of the weighted inputs is larger than the threshold value, the neuron generates
a signal; otherwise, no signal is red. Neural networks can operate in feedforward
or feedback mode.
10
Complex systems, for which dynamical equations may not be
known or may be too difcult to solve, can be modeled using neural nets.
For ow control, neural networks provide convenient, fast, nonlinear adaptive
algorithms to relate sensor outputs to actuator inputs via variable-coefcient func-
tions and nonlinear, sigmoid saturation functions. With no prior knowledge of the
pertinent dynamics, a self-learning neural network develops a model for those dy-
namics through observations of the applied control and sensed measurements. The
network is by nature nonlinear and can therefore better handle nonlinear dynam-
ical systems, which is a difcult task when classical (linear or weakly nonlinear)
control strategies are attempted. The feedforward type of neural network acts as a
nonlinear lter forming an output from a set of input data. The output can then be
compared with some desired output, and the difference (error) is typically used in a
back-propagation algorithm that updates the network parameters.
The number of researchers using neural networks to control uid ows is growing
rapidly. Herein, we provide only a small sample. Using a pretrained neural network,
Fan et al. (1993) conducted a conceptual reactive-ow-control experiment to delay
laminar-to-turbulence transition. Numerical simulations of their ow-control system
demonstrate almost complete cancellation of single and multiple articial wave dis-
turbances. Their controller also successfully attenuated a natural disturbance signal
with developing wave packets from an actual wind-tunnel experiment.
Jacobson and Reynolds (1993b, 1995, 1998) used neural networks to minimize
the boundary velocity gradient of three model ows: the one-dimensional stochastic
Burgers equation, a two-dimensional computational model of the near-wall region
of a turbulent boundary layer, and a real-time turbulent ow with a spanwise array
of wall actuators together with upstream and downstream wall-sensors. For all three
problems, the neural network successfully learned about the ow and developed
into procient controllers. For the laboratory experiments, however, Jacobson and
Reynolds (1995) reported that the neural network training time was much longer and
the performance was no better than a simpler ad hoc controller they had developed.
Jacobson and Reynolds emphasized that alternative neural net congurations and
convergence algorithms may, however, greatly improve the network performance.
Using the angle of attack and angular velocity as inputs, Faller, Schreck, and
Luttges (1994) trained a neural network to model the measured unsteady surface
pressure over a pitching airfoil (see also Schreck, Faller, and Luttges 1995). Following
training and using the instantaneous angle of attack and pitch rate as the only inputs,
their network was able to predict the surface pressure topology accurately as well as
10
Note that this terminology refers to the direction of information through the network. When a neural
net is used as a controller, the overall control loop is, however, a feedback, closed loop: the self-learning
network dynamically updates its various parameters by comparing its output to a desired output, thus
requiring feedback information relating to the effect of its control.
the time-dependent aerodynamic forces and moments. The model was then used to
develop a neural network controller for wing-motion actuator signals, which in turn
provided direct control of the lift-to-drag ratio across a wide range of time-dependent
motion histories.
As a nal example, Kawthar-Ali and Acharya (1996) developed a neural network
controller for use in suppressing the dynamic-stall vortex that periodically develops in
the leading edge of a pitching airfoil. On the basis of the current state of the unsteady
pressure eld, their control system specied the optimum amount of leading-edge
suction to achieve complete vortex suppression.
In this chapter, I have emphasized the frontiers of the eld of ow control, reviewing
the important advances that took place during the past few years and providing a
blueprint for future progress. In two words, the future of ow control is in taming
turbulence by targeting its coherent structures: reactive control. Recent developments
in chaos control, microfabrication, and soft computing tools are making it more
feasible to perform reactive control of turbulent ows to achieve drag reduction,
lift enhancement, mixing augmentation, and noise suppression. Field applications,
however, have to await further progress in those three modern areas. Other less
complex control schemes, passive as well as active, are more market-ready and are
also witnessing a resurgence of interest.
The outlook for reactive control is quite optimistic. Soft computing tools and
nonlinear dynamical systems theory are developing at a fast pace. Microelectro-
mechanical systems technology is improving even faster. The ability of Texas In-
struments to produce an array of one million individually addressable mirrors for
around 0.01 cent per actuator is a foreteller of the spectacular advances anticipated in
the near future. Existing automotive applications of MEMS have already proven the
ability of such devices to withstand the harsh environment under the hood. For the
rst time, targeted control of turbulent ows is in the realm of the possible for future
practical devices. What is needed is a focused, well-funded research and development
program to make it all come together for eld application of reactive ow-control
systems.
In parting, it may be worth recalling that a mere 10% reduction in the total drag
of an aircraft translates into a saving of one billion dollars in annual fuel cost (at
1999 prices) for the commercial eet of aircraft in the United States alone. Contrast
this benet to the annual worldwide expenditure of perhaps a few million dollars
for all basic research in the broad eld of ow control. Taming turbulence, though
arduous, will pay for itself in gold. Reactive control, as difcult as it seems, is neither
impossible nor a pie in the sky. Besides, lofty goals require strenuous efforts. Easy
solutions to difcult problems are likely to be wrong, as implied by Henry Menckens
words of wisdom quoted at the beginning of this last chapter.
A parting verse from William Shakespeares The Taming of the Shrew:
Hortensio (a gentleman of Padua): Now go thy ways, thou hast tamd a curst shrew.
Lucentio (a gentleman of Pisa): Tis a wonder, by your leave, she will be tamd so.
Epilogue
To be conscious that we are perceiving or thinking is to be conscious of
our own existence.
(Aristotle, 384322 B.C.)
Minds are like parachutes. They only function when they are open.
(Sir James Dewar, 18421923)
The eld of ow control is broad, practically very important, and rich in scientic
and technological challenges. Though the eld is as old as man himself, its potential
for improving our lives will keep it going strong for yet another millennium. In this
book, I have made a modest attempt to place the eld in a unifying framework
and to categorize the different control strategies properly. At a minimum, I hope
to have provided a useful navigation tool through the colossal literature in the eld
of ow control and its intricately related subelds such as control theory, chaos,
microelectromechanical systems, soft computing, and so forth. Review articles were
favored for citation because they themselves usually contain many useful citations.
As is clear fromall the previous chapters, there is no lack of ow-control methods
to achieve a particular goal for free-shear as well as wall-bounded ows across the
entire range of Mach and Reynolds numbers. Ranging from simple to complex, from
inexpensive to expensive, from passive to active to reactive, and from market-ready
to futuristic, a great variety of control devices are at the disposal of the uids engineer.
Flow control is most effective when applied near the transition or separation points;
in other words, near the critical-owregimes where owinstabilities magnify quickly.
Therefore, delaying or advancing laminar-to-turbulence transition and preventing or
provoking separation can readily be accomplished. To reduce the skin-friction drag
in a nonseparating turbulent boundary layer, where the mean ow is quite stable, is a
more challenging problem. In all cases, because ow-control goals are often adversely
interrelated, tough design compromises are always in the forefront. When designing
a ow-control device to achieve a particular goal such as skin-friction reduction, the
engineers foremost task is to ensure a minimum and most benign trade-off.
Market-ready techniques include passive and predetermined active control. Shap-
ing, suction, heating or cooling, Lorentz body force, and compliant coatings can be
used to delay transition by an order of magnitude in Reynolds number, and, except
for the latter technique (at least at this time), can also be used to prevent boundary
359
360 EPILOGUE
layer separation. The use of polymers, microbubbles, riblets, and LEBUs can lead to
skin-friction reduction in turbulent boundary layers. Numerous other techniques are
available to reduce form drag, induced drag, and wave drag. Remaining issues for
eld application of market-ready techniques include cost, maintenance, and reliabil-
ity. Potential further improvements in classical ow-control techniques will perhaps
involve combining more than one technique to aim at achieving a favorable effect
that is greater than the sum. Examples include combining suction or polymer in-
jection with riblets for increased effectiveness and savings. Owing to its obvious
difculties, synergism has not been extensively studied in the past but deserves future
consideration.
Classical control techniques, though spectacularly successful in the past, have
been extended to near their physical limits. Conventional strategies are often ineffec-
tive for turbulent ows. Substantial gains are potentially possible, however, when re-
active ow-control methods are used to target specic coherent structures for modu-
lation. Particularly for turbulent ows, reactive control requires large numbers of
sensors and actuators and will not become practical until the technology for manu-
facturing inexpensive, robust microsensors and microactuators becomes available.
Autonomous control algorithms and associated computers to handle the required
colossal data in real time must also be developed. Further research is needed in dynam-
ical systems theoryparticularly chaos control, microelectromechanical systems, al-
ternative neural network congurations, convergence algorithms, and distributed
control. The difculties are daunting, but the potential payoffs are enormous.
Did we benet from ow control thus far? The simple act of streamlining repre-
sents one of the oldest and most spectacular achievements of ow control. Stream-
lined spears were perhaps the difference between life and death to our Stone Age
ancestors about 400,000 years ago (see Chapter 1). Pressure drag due to separa-
tion has virtually been eliminated from modern airplanes and submarines, making
it possible to y or sail as well as saving countless resources. Automobiles to date
have signicantly lower aerodynamic drag than they did a generation ago. Sweeping
back the wings of all large airliners made it possible for them to y just below the
speed of sound without encountering the tremendous wave-drag penalty associated
with shock-wave formation, and thus also saving countless resources. Dilute polymer
solutions can reduce the skin-friction drag by up to 80% in wall-bounded ows of
most liquids of practical interest. In fact, polymers are in actual use in several pipe
ows, including the crude oil trans-Alaskan pipeline and the IraqTurkey pipeline.
Jet airplanes would not have been possible without the signicant noise reduction
made possible by clever design. High-lift devices are common on all sort of aircraft,
making it possible for jumbo jets to take off and land at safe, moderate speeds and to
use relatively short runways. Mostly laminar wings are very close to being viable for
commercial aircraft, promising further savings in fuel consumption. In my view, the
most signicant recent advance in the eld of ow control is its gradualperhaps
even not so gradualshift from being an art to being a science. More control devices
are now designed using rst principles. This trend will undoubtedly accelerate as
computer power continues to increase exponentially and, with the advent of molec-
ular electronics (see Section 1.3), seemingly indenitely.
As for the future? The French pilot, warrior, and writer Antoine de Saint-Exup ery
wrote in Citadelle (The Wisdom of the Sands): As for the future, your task is not to
foresee, but to enable it. Despite certain practical difculties mostly related to the
EPILOGUE 361
lowprice of fuel and the high price and volume of polymers, there is little fundamental
reason not to be able to benet from polymer injections to reduce dramatically the
skin-friction drag of the turbulent boundary layers developing around underwater
vehicles and the submerged portion of ship hulls. A stated goal of NASA is to double
the lift-to-drag ratio for the United States commercial eet of aircraft in the next two
decades. This lofty goal will obviously best be achieved by simultaneously lowering
the drag and increasing the lift. In the future, reactive control will target certain
coherent structures to tame turbulence in practical vehicles and industrial devices.
Drag will be reduced to near the lowest level allowed by the laws of nature, and
thus valuable resources will be saved and a better environment will be preserved for
future generations. My personal wish list includes the utilization of advanced ow-
control strategies to make practical such seeming pies in the sky as the commercial
submarine, the reusable spaceplane, and the personal skycarall briey mentioned
in Section 1.3. In fact, prototypes of the latter two dreams were scheduled for testing
during 1999 (Ashley 1998; May 1999).
Now my friends it is time to end, at least for now, the adventure that we have
started. But where do we go from here? I leave you with the words of the mathe-
matician Charles Lutwidge Dodgson (18321898), who created the two memorable
stories Alices Adventures in Wonderland and Through the Looking-Glass merely to
amuse the young daughterAliceof an acquaintance, and whose nom de plume
was Lewis Carroll.
Cheshire Puss, she began, rather timidly, as she did not at all know whether it would
like the name: however, it only grinned a little wider. Come, its pleased so far, thought
Alice, and she went on, Would you tell me, please, which way I ought to walk from
here?.
That depends a good deal on where you want to get to, said the Cat.
I dont much care where said Alice.
Then it doesnt matter which way you walk, said the Cat.
so long as I get somewhere, Alice added as an explanation.
Oh, youre sure to do that, said the Cat, if you only walk long enough.
(From Lewis Carrolls Alices Adventures in Wonderland, 1865)
Bibliography
Abbott, I.H., and von Doenhoff, A.E. (1959) Theory of Wing Sections, Dover, New York.
Abergel, F., and Temam, R. (1990) On Some Control Problems in Fluid Mechanics, Theor.
Comput. Fluid Dyn. 1, pp. 303325.
Acarlar, M.S., and Smith, C.R. (1987a) A Study of Hairpin Vortices in a Laminar Boundary Layer.
Part 1. Hairpin Vortices Generated by a Hemisphere Protuberance, J. Fluid Mech. 175, pp. 141.
Acarlar, M.S., and Smith, C.R. (1987b) A Study of Hairpin Vortices in a Laminar Boundary Layer.
Part 2. Hairpin Vortices Generated by Fluid Injection, J. Fluid Mech. 175, pp. 4383.
Acharya, N., Sen, M., and Chang, H.-C. (1992) Heat Transfer Enhancement in Coiled Tubes by
Chaotic Mixing, Int. J. Heat Mass Transfer 35, pp. 24752489.
Achia, B.U., and Thompson, D.W. (1977) Structure of the Turbulent Boundary Layer in Drag-
Reducing Pipe Flow, J. Fluid Mech. 81, pp. 439464.
Ackeret, J., Ras, M., and Pfenninger, W. (1941) Verhinderung des Turbulentwerdens einer Gren-
zschicht durch Absaugung, Naturwissenschaften 29, pp. 622623.
Adkins, R.C. (1975) A Short Diffuser with Low Pressure Loss, J. Fluids Eng. 97, pp. 297302.
Adkins, R.C. (1977) Diffusers and Their Performance Improvement by Means of Boundary Layer
Control, in AGARD Special Course on Concepts for Drag Reduction, AGARD-R-654, paper
no. 6, Rhode-Saint-G en` ese, Belgium.
Adkins, R.C., Mathaus, D.S., and Yost, J.O. (1980) The Hybrid Diffuser, ASME Paper No.
80-GT-136, New York.
Adleman, L.M. (1994) Molecular Computation of Solutions to Combinatorial Problems, Science
266, no. 5187, 11 November, pp. 10211024.
Afzal, N. (1976) Millikans Argument at Moderately Large Reynolds Number, Phys. Fluids 1,
pp. 600602.
Afzal, N., and Bush, W.B. (1985) AThree-Layer Asymptotic Analysis of Turbulent Channel Flow,
in Proc. Indian Acad. Sci. (Math. Sci.) 94, pp. 135148.
Agarwal, R., Yun, K., and Balakrishnan, R. (1999) Beyond Navier Stokes: Burnett Equations for
FlowSimulations in Continuum-Transition Regime, AIAAPaper No. 99-3580, Reston, Virginia.
Ahuja, K.K., and Burrin, R.H. (1984) Control of Flow Separation by Sound, AIAA Paper
No. 84-2298, New York.
Ahuja, K.K., Hayden, W., and Entrekin, A. (1993) Mixing Enhancement and Noise Reduction
Through Taps Plus Ejectors, AIAA Paper No. 93-4347, Washington, D.C.
Ahuja, K.K., Whipkey, R.R., and Jones, G.S. (1983) Control of Turbulent Boundary Layer Flows
by Sound, AIAA Paper No. 83-0726, New York.
Akhavan, R., Jung, W.J., and Mangiavacchi, N. (1992) Turbulence Control of Wall Bounded Flows
by Spanwise Oscillations, Appl. Scientic Res. 51, pp. 299303.
Alder, B.J., and Wainwright, T.E. (1957) Studies in Molecular Dynamics, J. Chem. Phys. 27,
pp. 12081209.
Alder, B.J., and Wainwright, T.E. (1958) Molecular Dynamics by Electronic Computers, in Trans-
port Processes in Statistical Mechanics, ed. I. Prigogine, pp. 97131, Interscience, New York.
Alder, B.J., and Wainwright, T.E. (1970) Decay of the Velocity Auto-Correlation Function, Phys.
Rev. A 1, pp. 1821.
Aleyev, Yu.G. (1977) Nekton, Dr. W. Junk b.v. Publishers, The Hague, The Netherlands.
363
364 BIBLIOGRAPHY
Alfredsson, P.H., and Johansson, A.V. (1984) On the Detection of Turbulence-Generating Events,
J. Fluid Mech. 139, pp. 325345.
Allan, J.F., and Cutland, R.S. (1953) Wake Studies of Plane Surfaces, Trans. North East Coast
Institute of Engineers and Shipbuilders 69, pp. 245266.
Allen, M.P., and Tildesley, D.J. (1987) Computer Simulation of Liquids, Clarendon Press, Oxford,
Great Britain.
Alrefai, M., and Acharya, M. (1995) Controlled Leading-Edge Suction for the Management of
Unsteady Separation over Pitching Airfoils, AIAA Paper No. 95-2188, Washington, D.C.
Alshamani, K.M.M., Livesey, J.L., and Edwards, F.J. (1982) Excitation of the Wall Region by
Sound in Fully Developed Channel Flow, AIAA J. 20, pp. 334339.
Alvarez-Calderon, A. (1964) Rotating Cylinder Flaps of V/STOL Aircraft, Aircraft Eng. 36,
pp. 304309.
Alving, A.E., Smits, A.J., and Watmuff, J.H. (1990) Turbulent Boundary Layer Relaxation from
Convex Curvature, J. Fluid Mech. 211, pp. 529556.
Amato, I. (1998) Formenting a Revolution, in Miniature, Science 282, no. 5388, 16 October,
pp. 402405.
Anders, J.B. (1990a) Boundary Layer Manipulators at High Reynolds Numbers, in Structure of
Turbulence and Drag Reduction, ed. A. Gyr, pp. 475482, SpringerVerlag, Berlin.
Anders, J.B. (1990b) Outer-Layer Manipulators for Turbulent Drag Reduction, in Viscous Drag
Reduction in Boundary Layers, eds. D.M. Bushnell and J.N. Hefner, pp. 263284, AIAA,
Washington, D.C.
Anders, J.B., and Watson, R.D. (1985) Airfoil Large-Eddy Breakup Devices for Turbulent Drag
Reduction, AIAA Paper No. 85-0520, New York.
Anders, J.B., Hefner, J.N., and Bushnell, D.M. (1984) Performance of Large-Eddy Breakup Devices
at Post-Transitional Reynolds Numbers, AIAA Paper No. 84-0345, New York.
Anders, J.B., Walsh, M.J., and Bushnell, D.M. (1988) The Fix for Tough Spots, Aerospace Amer-
ica 26, January, pp. 2427.
Anderson, J.L., and Quinn, J.A. (1972) Ionic Mobility in Microcapillaries, J. Chem. Phys. 27,
pp. 12081209.
Andreopoulos, J., Durst, F., Zaric, Z., and Jovanovic, J. (1984) Inuence of Reynolds Number on
Characteristics of Turbulent Wall Boundary Layers, Exp. Fluids 2, pp. 716.
Angell, J.B., Terry, S.C., and Barth, P.W. (1983) Silicon Micromechanical Devices, Faraday Trans-
actions I 68, pp. 744748.
Antonia, R.A., Bisset, D.K., and Browne, L.W.B. (1990) Effect of Reynolds Number on the Topol-
ogy of the Organized Motion in a Turbulent Boundary Layer, J. Fluid Mech. 213, pp. 267286.
Antonia, R.A., Fulachier, L., Krishnamoorthy, L.V., Benabid, T., and Anselmet, F. (1988) Inuence
of Wall Suction on the Organized Motion in a Turbulent Boundary Layer, J. Fluid Mech. 190,
pp. 217240.
Antonia, R.A., Rajagopalan, S., Subramanian, C.S., and Chambers, A.J. (1982) Reynolds-Number
Dependence of the Structure of a Turbulent Boundary Layer, J. Fluid Mech. 121, pp. 123
140.
Antonia, R.A., Teitel, M., Kim, J., and Browne, L.W.B. (1992) Low-Reynolds-Number Effects in
a Fully Developed Turbulent Channel Flow, J. Fluid Mech. 236, pp. 579605.
Antsaklis, P.J. (1993) Control Theory Approach, in Mathematical Approaches to Neural Net-
works, ed. J.G. Taylor, pp. 123, Elsevier, Amsterdam.
Anyiwo, J.C., and Bushnell, D.M. (1982) Turbulence Amplication in Shock WaveBoundary
Layer Interaction, AIAA J. 20, pp. 893899.
Aref, H. (1984) Stirring by Chaotic Advection, J. Fluid Mech. 143, pp. 121.
Aref, H., and El Naschie, M.S., editors (1995) Chaos Applied to Fluid Mixing, Pergamon Press,
Oxford, Great Britain.
Arkilic, E.B. (1997) Measurement of the Mass Flow and Tangential Momentum Accommoda-
tion Coefcient in Silicon Micromachined Channels, Ph.D. Thesis, Massachusetts Institute of
Technology, Cambridge, Massachusetts.
Arkilic, E.B., Schmidt, M.A., and Breuer, K.S. (1995) Slip Flow in Microchannels, in Rareed
Gas Dynamics 19, eds. J. Harvey and G. Lord, Oxford University Press, Oxford, Great Britain.
Arkilic, E.B., Schmidt, M.A., and Breuer, K.S. (1997a) Gaseous Slip Flowin Long Microchannels,
J. MEMS 6, pp. 167178.
BIBLIOGRAPHY 365
Arkilic, E.B., Schmidt, M.A., and Breuer, K.S. (1997b) TMAC Measurement in Silicon Microma-
chined Channels, in Rareed Gas Dynamics 20, ed. C. Shen, 6 pages, Beijing University Press,
Beijing, China.
Aroesty, J., and Berger, S.A. (1975) Controlling the Separation of Laminar Boundary Layers in
Water: Heating and Suction, RAND Corporation Report No. R-1789-ARPA, Santa Monica,
California. (Also available from U.S. NTIS; Document Number AD-AO20026.)
Aronson, D., and L ofdahl, L. (1994) The Plane Wake of a Cylinder: An Estimate of the Pressure
Strain Rate Tensor, Phys. Fluids 6, pp. 27162721.
Aronson, D., Johansson, A.V., and L ofdahl, L. (1997) A Shear-Free Turbulent Boundary Layer
Experiments and Modeling, J. Fluid Mech. 338, pp. 363385.
Ashino, I., and Yoshida, K. (1975) Slow Motion between Eccentric Rotating Cylinders, Bulletin
JSME 18, no. 117, pp. 280285.
Ashkenas, H.I., and Riddell, F.R. (1955) Investigation of the Turbulent Boundary Layer on a
Yawed Flat Plate, NACA Technical Note No. TN-3383, Washington, D.C.
Ashley, S. (1996) Getting a Microgrip in the Operating Room, Mech. Eng. 118, September,
pp. 9193.
Ashley, S. (1998) Bringing Launch Costs Down to Earth, Mech. Eng. 120, no. 10, October,
pp. 6268.
Aslanov, P.V., Maksyutenko, S.N., Povkh, I.L., Simonenko, A.P., and Stupin, A.B. (1980) Turbulent
Flows of Solutions of Surface-Active Substances, Izvestiya Akademii Nauk SSSR, Mekhanika
Zhidkosti i Gaza, no. 1, pp. 3643.
Atassi, H.M., editor (1993) Unsteady Aerodynamics, Aeroacoustics, and Aeroelasticity of Turbo-
machines and Propellers, SpringerVerlag, New York.
Atassi, H.M., and Gebert, G.A. (1987) Modication of Turbulent Boundary Layer Structure by
Large-Eddy Breakup Devices, in Proc. Int. Conf. on Turbulent Drag Reduction by Passive Means,
vol. 2, pp. 432456, Royal Aeronautical Society, London, Great Britain.
Atta, R., and Rockwell, D. (1990) Leading-Edge Vortices Due to Low Reynolds Number Flow
Past a Pitching Delta Wing, AIAA J. 28, pp. 9951004.
Atwood, B.T., and Schowalter, W.R. (1989) Measurements of Slip at the Wall During Flow of
High-Density Polyethylene through a Rectangular Conduit, Rheologica Acta 28, pp. 134
146.
Au, D., and Weihs, D. (1980) At High Speeds Dolphins Save Energy by Leaping, Nature 284,
no. 5756, 10 April, pp. 548550.
Aubry, N. (1990) Use of Experimental Data for an Efcient Description of Turbulent Flows,
Appl. Mech. Rev. 43, pp. S240S245.
Aubry, N., Holmes, P., Lumley, J.L., and Stone, E. (1988) The Dynamics of Coherent Structures
in the Wall Region of a Turbulent Boundary Layer, J. Fluid Mech. 192, pp. 115173.
Auerbach, D. (1994) Controlling Extended Systems of Chaotic Elements, Phys. Rev. Lett. 72,
pp. 11841187.
Auerbach, D., Grebogi, C., Ott, E., and Yorke, J.A. (1992) Controlling Chaos in High Dimensional
Systems, Phys. Rev. Lett. 69, pp. 34793482.
Ayers, R.F., and Wilde, M.R. (1956) An Experimental Investigation of the Aerodynamic Charac-
teristics of a Low Aspect Ratio Swept Wing with Blowing in a Spanwise Direction from the Tips,
College of Aeronautics, Note No. 57, Craneld, Great Britain.
Bahi, L., Ross, J.M., and Nagamatsu, H.T. (1983) Passive Shock Wave/Boundary Layer Control
for Transonic Airfoil Drag Reduction, AIAA Paper No. 83-0137, New York.
Baker, G.L., and Gollub, J.P. (1990) Chaotic DynamicsAn Introduction, Cambridge University
Press, Cambridge, Great Britain.
Bakewell, H.P., and Lumley, J.L. (1967) Viscous Sublayer and Adjacent Wall Region in Turbulent
Pipe Flow, Phys. Fluids 10, pp. 18801889.
Ball, K.O. (1971) Flap Span Effects on Boundary Layer Separation, AIAA J. 9, pp. 20802081.
Bandyopadhyay, P.R. (1980) Large Structure with a Characteristic UpstreamInterface in Turbulent
Boundary Layers, Phys. Fluids 23, pp. 23262327.
Bandyopadhyay, P.R. (1982) Period Between Bursting in Turbulent Boundary Layers, Phys. Fluids
25, pp. 17511754.
Bandyopadhyay, P.R. (1983) Turbulent Spot-Like Features of a Boundary Layer, Annals New
York Acad. Sci. 404, pp. 393395.
366 BIBLIOGRAPHY
Bandyopadhyay, P.R. (1986a) Drag Reducing Outer-Layer Devices in Rough Wall Turbulent
Boundary Layers, Exp. Fluids 4, pp. 247256.
Bandyopadhyay, P.R. (1986b) ReviewMean Flow in Turbulent Boundary Layers Disturbed to
Alter Skin Friction, J. Fluids Eng. 108, pp. 127140.
Bandyopadhyay, P.R. (1987) Rough-Wall Turbulent Boundary Layers in the Transition Regime,
J. Fluid Mech. 180, pp. 231266.
Bandyopadhyay, P.R. (1988) Resonant Flow in Small Cavities Submerged in a Boundary Layer,
Proc. R. Soc. London A 420, pp. 219245.
Bandyopadhyay, P.R. (1989) Effect of Abrupt Pressure Gradients on the Structure of Turbulent
Boundary Layers, Proc. Tenth Australasian Fluid Mechanics Conf., eds. A.E. Perry et al., vol. I,
pp. 1.11.4, University of Melbourne, Melbourne, Australia.
Bandyopadhyay, P.R. (1990) Convex Curvature Concept of Viscous Drag Reducion, in Viscous
Drag Reduction in Boundary Layers, eds. D.M. Bushnell and J.N. Hefner, pp. 285324, AIAA,
Washington, D.C.
Bandyopadhyay, P.R. (1991) Comments on Reynolds Number Effects in Wall-Bounded Shear
Layers, AIAA Paper No. 91-0231, New York.
Bandyopadhyay, P.R. (1992) Reynolds Number Dependence of the Freestream Turbulence Effects
on Turbulent Boundary Layers, AIAA J. 30, pp. 19101912.
Bandyopadhyay, P.R., and Ahmed, A. (1993) Turbulent Boundary Layers Subjected to Multiple
Curvatures and Pressure Gradients, J. Fluid Mech. 246, pp. 503527.
Bandyopadhyay, P.R., and Balasubramanian, R. (1993) A Vortex Model for Calculating Wall
Pressure Fluctuations in Turbulent Boundary Layers, in ASME Symposium on Flow Noise Mod-
eling, Measurement and Control, eds. T.M. Farabee, W.L. Keith, and R.M. Lueptow, NCA-Vol.
15/FED-Vol. 168, pp. 1324, New York.
Bandyopadhyay, P.R., and Balasubramanian, R. (1995) Vortex Reynolds Number in Turbulent
Boundary Layers, Theor. Comput. Fluid Dynamics 7, pp. 101117.
Bandyopadhyay, P.R., and Balasubramanian, R. (1996) Structural Modeling of the Wall Effects of
Lorentz Force, J. Fluids Eng. 118, pp. 412414.
Bandyopadhyay, P.R., and Hussain, A.K.M.F. (1984) The Coupling Between Scales in Shear
Flows, Phys. Fluids 27, pp. 22212228.
Bandyopadhyay, P.R., and Watson, R.D. (1988) Structure of Rough-Wall Turbulent Boundary
Layers, Phys. Fluids 31, pp. 18771883.
Bandyopadhyay, P.R., Breuer, K.S., and Blechinger, C.J., editors (1994) Application of Microfabri-
cation to Fluid Mechanics, FED Vol. 197, New York.
Banks, S.P. (1986) Control Systems Engineering, PrenticeHall International, Englewood Cliffs,
New Jersey.
Barenblatt, G.I. (1979) Similarity, Self-Similarity, and Intermediate Hypothesis, Plenum Press, New
York.
Barenblatt, G.I. (1993) Scaling Laws for Fully Developed Turbulent Shear Flows. Part 1. Basic
Hypothesis and Analysis, J. Fluid Mech. 248, pp. 513520.
Barenblatt, G.I., and Prostokishin, V.M. (1993) Scaling Laws for Fully Developed Turbulent Shear
Flows. Part 2. Processing of Experimental Data, J. Fluid Mech. 248, pp. 521529.
Barenblatt, G.I., Chorin A.J., and Prostokishin, V.M. (1997) Scaling Laws for Fully Developed
Turbulent Flows in Pipes, Appl. Mech. Rev. 50, pp. 413429.
Barger, J.E., and von Winkle, W.A. (1961) Evaluation of a Boundary Layer Stabilization Coating,
J. Acoustical Soc. of America 33, p. 836.
Barker, R.A. (1986) The Aerodynamic Effects of a Serrated Strip Near the Leading Edge of an
Airfoil, M.S. Thesis, Royal Air Force College, Report No. ETN-87-99480, Cranwell, Great
Britain.
Barker, S.J., and Gile, D. (1981) Experiments on Heat-Stabilized Laminar Boundary Layers in
Water, J. Fluid Mech. 104, pp. 139158.
Barnwell, R.W., and Hussaini, M.Y., editors (1992) Natural Laminar Flow and Laminar Flow
Control, SpringerVerlag, New York.
Barnwell, R.W., Bushnell, D.M., Nagamatsu, H.T., Bahi, L., and Ross, J. (1985) Passive Drag
Control of Airfoils at Transonic Speeds, U.S. Patent No. 4,522,360.
Bar-Sever, A. (1989) Separation Control on an Airfoil by Periodic Forcing, AIAA J. 27, pp. 820
821.
BIBLIOGRAPHY 367
Bart, S.F., Tavrow, L.S., Mehregany, M., and Lang, J.H. (1990) Microfabricated Electrohydro
dynamic Pumps, Sensors & Actuators A 2123, pp. 193197.
Bastedo, W.G., Jr., and Mueller, T.J. (1985) Performance of Finite Wings at Low Reynolds Num-
bers, in Proc. Conf. on Low Reynolds Number Airfoil Aerodynamics, ed. T.J. Mueller, pp.
195205, University of Notre Dame, Notre Dame, Indiana.
Bastedo, W.G., Jr., and Mueller, T.J. (1986) Spanwise Variation of Laminar Separation Bubbles on
Wings at Low Reynolds Numbers, J. Aircraft 23, pp. 687694.
Batchelor, G.K. (1967) An Introduction to Fluid Dynamics, Cambridge University Press, Cambridge,
Great Britain.
Batchelor, G.K., and Green, J.T. (1972) The Determination of the Bulk Stress in a Suspension of
Spherical Particles to Order c
2
, J. Fluid Mech. 56, pp. 401427.
Bau, H.H. (1994) Transport Processes Associated with Micro-Devices, Thermal Sci. Eng. 2,
pp. 172178.
Bauer, S.X.S., and Hernandez, G. (1988) Reduction of Cross-FlowShock-Induced Separation with
a Porous Cavity at Supersonic Speeds, AIAA Paper No. 88-2567, New York.
Baullinger, N., and Page, V. (1989) High Altitude Long Endurance (HALE) RPV, AIAA Paper
Bechert, D.W., Hoppe, G., and Reif, W.-E. (1985) On the Drag Reduction of the Shark Skin,AIAA
Paper No. 85-0546, New York.
Becker, S., Durst, F., and Lienhart, H. (1999) Laser Doppler Anemometer for In-Flight Velocity
Measurements on Airplane Wings, AIAA J. 37, pp. 680687.
Benjamin, T.B. (1960a) Effects of a Flexible Boundary on Hydrodynamic Stability, J. Fluid Mech.
9, pp. 513532.
Benjamin, T.B. (1960b) Fluid Flowwith Flexible Boundaries, in Proc. Eleventh Int. Cong. of Appl.
Mech., ed. H. G ortler, pp. 109128, SpringerVerlag, Berlin.
Benjamin, T.B. (1963) The Threefold Classication of Unstable Disturbances in Flexible Surfaces
Bounding Inviscid Flows, J. Fluid Mech. 16, pp. 436450.
Benjamin, T.B. (1966) Fluid Flowwith Flexible Boundaries, in Proc. Eleventh Int. Cong. of Appl.
Mech., ed. H. G ortler, pp. 109128, SpringerVerlag, Berlin.
Bergles, A.E. (1978) Enhancement of Heat Transfer, in Proc. Sixth Int. Heat Trans. Conf., vol. 6,
pp. 89108, Hemisphere, Washington, D.C.
Bergles, A.E. (1981) Principles of Heat Transfer Augmentation, in Heat Exchangers, Thermal-
Hydraulic Fundamentals and Design, eds. S. Kakac, A.E. Bergles, and F. Mayinger, pp. 819842,
Hemisphere, New York.
Bergles, A.E. (1985) Techniques to Augment Heat Transfer, in Handbook of Heat Transfer Ap-
plications, second edition, eds. W.M. Rohsenow, J.P. Hartnett, and E.N. Gani c, pp. 3.13.80,
McGraw-Hill, New York.
Bergles, A.E., and Morton, L.H. (1965) Survey and Evaluation of Techniques to Augment Con-
vective Heat Transfer, Report No. EPL 5382-34, Department of Mechanical Engineering,
Massachusetts Institute of Technology, Cambridge, Massachusetts.
Bergles, A.E., and Webb, R.L. (1985) A Guide to the Literature on Convective Heat Transfer
Augmentation, in Twenty-Third National Heat Transfer Conf.: Advances in Enhanced Heat
Transfer, Denver, Colorado.
Berkooz, G., Fisher, M., and Psiaki, M. (1993) Estimation and Control of Models of the Turbulent
Wall Layer, Bull. Am. Phys. Soc. 38, p. 2197.
Berkooz, G., Holmes, P., and Lumley, J.L. (1991) Intermittent Dynamics in Simple Models of the
Turbulent Boundary Layer, J. Fluid Mech. 230, pp. 7595.
Berman, N.S. (1978) Drag Reduction by Polymers, Ann. Rev. Fluid Mech. 10, pp. 4764.
Berman, N.S., and George, W.K. (1974) Time Scale and Molecular Weight Distribution Con-
tributions to Dilute Polymer Solution Fluid Mechanics, in Proc. Heat Trans. Fluid Mech.
Inst., eds. L.R. Davis and R.E. Wilson, pp. 348364, Stanford University Press, Stanford,
California.
Bernard, J.J., and Siestrunck, R. (1959)
Echanges de Chaleur dans les

Ecoulements Pr esentant des
D ecollements, in Proc. First Int. Cong. Aero. Sci., eds. Th. von K arm an et al., pp. 314332, Adv.
Aero. Sci., vol. 1, Pergamon Press, London, Great Britain.
Bertelrud, A., Truong, T.V., and Avellan, F. (1982) Drag Reduction in Turbulent Boundary Layers
Using Ribbons, AIAA Paper No. 82-1370, New York.
368 BIBLIOGRAPHY
Beskok, A. (1994) Simulation of Heat and Momentum Transfer in Complex Micro-Geometries,
M.Sc. Thesis, Princeton University, Princeton, New Jersey.
Beskok, A. (1996) Simulations and Models of Gas Flows in Microgeometries, Ph.D. Thesis,
Princeton University, Princeton, New Jersey.
Beskok, A., and Karniadakis, G.E. (1994) Simulation of Heat and MomentumTransfer in Complex
Micro-Geometries, J. Thermophys. & Heat Transfer 8, pp. 355370.
Beskok, A., and Karniadakis, G.E. (1999) A Model for Flows in Channels, Pipes and Ducts at
Micro and Nano Scales, Microscale Thermophys. Eng. 3, pp. 4377.
Beskok, A., Karniadakis, G.E., and Trimmer, W. (1996) Rarefaction and Compressibility Effects
in Gas Microows, J. Fluids Eng. 118, pp. 448456.
Betchov, K. (1960) Simplied Analysis of Boundary-Layer Oscillations, J. Ship Res. 4, pp. 3754.
Bewley, T.R., Moin, P., and Temam, R. (1997) Optimal and Robust Approaches for Linear and
Nonlinear Regulation Problems in Fluid Mechanics, AIAAPaper No. 97-1872, Reston, Virginia.
Bewley, T.R., Temam, R., and Ziane, M. (1998) AGeneral Framework for Robust Control in Fluid
Mechanics, Center for Turbulence Research No. CTR Manuscript 169, Stanford University,
Stanford, California.
Bhatnagar, P.L., Gross, E.P., and Krook, M. (1954) A Model for Collision Processes in Gases. I.
Small Amplitude Processes in Charged and Neutral One-Component Systems, Phys. Rev. 94,
pp. 511524.
Bhatt, W.V. (1971) Flight Test Measurement of Exterior Turbulent Boundary Layer Pressure Fluc-
tuations on Boeing Model 737 Airplane, J. Sound Vibr. 14, pp. 439457.
Bhattacharjee, S., Scheelke, B., and Troutt, T.R. (1985) Modication of Vortex Interactions in a
Reattaching Separated Flow, AIAA Paper No. 85-0555, New York.
Bird, G.A. (1963) Approach to Translational Equilibrium in a Rigid Sphere Gas, Phys. Fluids 6,
pp. 15181519.
Bird, G.A. (1965) The Velocity Distribution Function within a Shock Wave, J. Fluid Mech. 30,
pp. 479487.
Bird, G.A. (1976) Molecular Gas Dynamics, Clarendon Press, Oxford, Great Britain.
Bird, G.A. (1978) Monte Carlo Simulation of Gas Flows, Ann. Rev. Fluid Mech. 10, pp. 1131.
Bird, G.A. (1994) Molecular Gas Dynamics and the Direct Simulation of Gas Flows, Clarendon
Press, Oxford, Great Britain.
Biringen, S. (1984) Active Control of Transition by Periodic Suction-Blowing, Phys. Fluids 27,
pp. 13451347.
Biringen, S., and Maestrello, L. (1984) Development of Spot-Like Turbulence in Plane Channel
Flow, Phys. Fluids 27, pp. 318321.
Black, T.J. (1966) Some Practical Applications of a New Theory of Wall Turbulence, in Proc.
1966 Heat Transfer & Fluid Mechanics Institute, eds. M.A. Saad and J.A. Miller, pp. 366386,
Stanford University Press, Stanford, California.
Black, T.J. (1968) An Analytical Study of the Measured Wall Pressure Field under Supersonic
Turbulent Boundary Layers, NASA Contractor Report No. CR-888, Washington, D.C.
Blackwelder, R.F. (1978) The Bursting Process in Turbulent Boundary Layers, in Workshop on
Coherent Structure of Turbulent Boundary Layers, eds. C.R. Smith and D.E. Abbott, pp. 211
227, Lehigh University, Bethlehem, Pennsylvania.
Blackwelder, R.F. (1988) Coherent Structures Associated with Turbulent Transport, in Transport
Phenomena in Turbulent Flows, eds. M. Hirata and N. Kasagi, pp. 6988, Hemisphere, NewYork.
Blackwelder, R.F. (1989) Some Ideas on the Control of Near-Wall Eddies, AIAA Paper No. 89-
1009, New York.
Blackwelder R.F. (1998) Some Notes on Drag Reduction in the Near-Wall Region, in Flow Con-
trol: Fundamentals and Practices, eds. M. Gad-el-Hak, A. Pollard, and J.-P. Bonnet, pp. 155198,
SpringerVerlag, Berlin.
Blackwelder, R.F., and Eckelmann, H. (1978) The Spanwise Structure of the Bursting Pheno-
menon, in Structure and Mechanisms of Turbulence I, ed. H. Fiedler, pp. 190204, Springer
Verlag, Berlin.
Blackwelder, R.F., and Eckelmann, H. (1979) Streamwise Vortices Associated with the Bursting
Phenomenon, J. Fluid Mech. 94, pp. 577594.
Blackwelder R.F., and Gad-el-Hak M. (1990) Method and Apparatus for Reducing Turbulent Skin
Friction, United States Patent No. 4,932,612.
BIBLIOGRAPHY 369
Blackwelder, R.F., Gad-el-Hak, M., and Srnsky, R.A. (1987) Method and Apparatus for Control-
ling Bound Vortices in the Vicinity of Lifting Surfaces, U.S. Patent No. 4,697,769.
Blackwelder, R.F., and Haritonidis, J.H. (1983) Scaling of the Bursting Frequency in Turbulent
Boundary Layers, J. Fluid Mech. 132, pp. 87103.
Blackwelder, R.F., and Kaplan, R.E. (1976) On the Wall Structure of the Turbulent Boundary
Layer, J. Fluid Mech. 76, pp. 89112.
Blackwelder, R.F., and Kovasznay, L.S.G. (1972) Time-Scales and Correlations in a Turbulent
Boundary Layer, Phys. Fluids 15, pp. 15451554.
Blackwelder, R.F., and Swearingen, J.D. (1990) The Role of Inectional Velocity Proles in Wall
Bounded Flows, in Near-Wall Turbulence: 1988 Zoran Zari c Memorial Conference, eds. S.J.
Kline and N.H. Afgan, pp. 268288, Hemisphere, New York.
Blair, M.F. (1983) Inuence of Free-StreamTurbulence on Turbulent Boundary Layer Heat Transfer
and Mean Prole Development, Part IIAnalysis of Results, J. Heat Transfer 105, pp. 41
47.
Blair, M.F., and Werle, M.J. (1980) The Inuence of Free-Stream Turbulence on the Zero Pres-
sure Gradient Fully Turbulent Boundary Layer, United Technologies Research Center Report
No. R80-914388-12, Hartford, Connecticut.
Blake, W.K. (1986) Mechanics of Flow-Induced Sound and Vibration, Academic Press, New York.
Blake, W.K., and Gershfeld, J.L. (1989) The Aeroacoustics of Trailing Edges, in Frontiers in
Experimental Fluid Mechanics, ed. M. Gad-el-Hak, pp. 457532, SpringerVerlag, New York.
Blick, E.F., and Walters, R.R. (1968) Turbulent Boundary-Layer Characteristics of Compliant
Surfaces, J. Aircraft 5, pp. 1116.
Bogdevich, V.G., and Malyuga, A.G. (1976) The Distribution of Skin Friction in a Turbulent
Boundary Layer of Water Beyond the Location of Gas Injection, in Studies on the Boundary Layer
Control (in Russian), eds. S.S. Kutateladze and G.S. Migirenko, p. 62, Institute of Thermophysics,
Novosibirsk, U.S.S.R.
Bogdevich, V.G., Evseev, A.R., Malyuga, A.G., and Migirenko, G.S. (1977) Gas-Saturation Ef-
fect on Near-Wall Turbulence Characteristics, in Second Int. Conf. on Drag Reduction, Paper
No. D2, BHRA Fluid Engineering, Craneld, Great Britain.
Bolton, J.S., editor (1988) Noise Control Design: Methods and Practice (Noise-Con 88 Proceedings),
Noise Control Foundation, Poughkeepsie, New York.
Bouchon-Meunier, B., Yager, R.R., and Zadeh, L.A., editors (1995a) Fuzzy Logic and Soft Com-
puting, World Scientic, Singapore.
Bouchon-Meunier, B., Yager, R.R., and Zadeh, L.A., editors (1995b) Advances in Intelligent
ComputingIPMU94, Lecture Notes in Computer Science, vol. 945, SpringerVerlag, Berlin.
Bradley, R.G., and Wray, W.O. (1974) A Conceptual Study of Leading-Edge-Vortex Enhancement
by Blowing, J. Aircraft 11, pp. 3338.
Bradshaw, P. (1967) Inactive Motion and Pressure Fluctuations in Turbulent Boundary Layers,
J. Fluid Mech. 30, pp. 241258.
Bradshaw, P. (1969) A Note on Reverse Transition, J. Fluid Mech. 35, pp. 387390.
Bradshaw, P. (1994) Turbulence: The Chief Outstanding Difculty of Our Subject, Exp. Fluids
16, pp. 203216.
Bradshaw, P., and Huang, G.P. (1995) The Law of the Wall in Turbulent Flow, Proc. R. Soc.
London A 451, pp. 165188.
Bradshaw, P., Ferriss, D.H., and Atwell, N.P. (1967) Calculations of Boundary-Layer Development
Using the Turbulent Energy Equation, J. Fluid Mech. 28, pp. 593616.
Bradshaw, P., Ferriss, D.H., and Johnson, R.F. (1964) Turbulence in the Noise-Producing Region
of a Circular Jet, J. Fluid Mech. 19, pp. 591624.
Bragg, M.B., and Gregorek, G.M. (1987) Experimental Study of Airfoil Performance with Vortex
Generators, J. Aircraft 24, pp. 305309.
Braslow, A.L., Burrows, D.L., Tetervin, N., and Visconti, F. (1951) Experimental and Theoretical
Studies of Area Suction for the Control of Laminar Boundary Layer, NACA Report No. 1025,
Washington, D.C.
Bremhorst, K., and Walker, T.B. (1974) Spectral Measurements of Turbulent Momentum Transfer
in Fully Developed Pipe Flow, J. Fluid Mech. 61, pp. 173186.
Brendel, M., and Mueller, T.J. (1988) Boundary Layer Measurements on an Airfoil at a Low
Reynolds Number in an Oscillating Freestream, AIAA J. 26, pp. 257263.
370 BIBLIOGRAPHY
Breuer, K.S., Bandyopadhyay, P.R., and Gad-el-Hak, M., editors (1996) Application of Silicon
Microfabrication to Fluid Mechanics, DSC-Volume 59, ASME, New York.
Breuer, K.S., Haritonidis, J.H., and Landahl, M.T. (1989) The Control of Transient Disturbances
in a Flat Plate Boundary Layer through Active Wall Motion, Phys. Fluids A 1, pp. 574582.
Briedenthal, R.E., Jr., and Russell, D.A. (1988) Aerodynamics of Vortex Generators, NASA
CR-182511, Washington, D.C.
Brookes, M. (1997) On a Wing and a Vortex, New Scientist 156, 11 October, pp. 2427.
Brown, A.C., Nawrocki, H.F., and Paley, P.N. (1968) Subsonic Diffusers Designed Integrally with
Vortex Generators, J. Aircraft 5, pp. 221229.
Brown, G.L., and Roshko, A. (1971) The Effect of Density Difference on the Turbulent Mix-
ing Layer, in Turbulent Shear Flows, pp. 23.123.12, AGARD-CP-93, Rhode-Saint-G en` ese,
Belgium.
Brown, G.L., and Roshko, A. (1974) On Density Effects and Large Strucure in Turbulent Mixing
Layers, J. Fluid Mech. 64, pp. 775816.
Brown, G.L., and Thomas, A.S.W. (1977) Large Structure in a Turbulent Boundary Layer, Phys.
Fluids 20, no. 10, part II, pp. S243S252.
Brunauer, S. (1944) Physical Adsorption of Gases and Vapours, Oxford University Press, Oxford,
Great Britain.
Bryzek, J., Peterson, K., and McCulley, W. (1994) Micromachines on the March, IEEE Spectrum
31, May, pp. 2031.
Buckingham, A.C., Hall, M.S., and Chun, R.C. (1985) Numerical Simulations of Compliant
Material Response to Turbulent Flow, AIAA J. 23, pp. 10461052.
Burd, J.E. (1981) Flow Control for a High Energy Laser Turret Using Trapped Vortices Stabilized
by Suction, M.Sc. Thesis, Naval Postgraduate School, Monterey, CA. (Also available from U.S.
NTIS; Document Number AD-A115263.)
Burns, T.F. (1981) Experimental Studies of Eppler 61 and Pfenninger 048 Airfoils at LowReynolds
Numbers, M.Sc. Thesis, University of Notre Dame, Notre Dame, Indiana.
Burton, T.E. (1974) The Connection Between Intermittent Turbulent Activity Near the Wall of a
Turbulent Boundary Layer with Pressure Fluctuations at the Wall, Massachusetts Institute of
Technology Report No. 70208-10, Cambridge, Massachusetts.
Busch-Vishniac, I.J. (1998) Trends in Electromechanical Transduction, Physics Today 51, July,
pp. 2834.
Bush, V. (1945) ScienceThe Endless Frontier, Government Printing Ofce, Washington, D.C.
Bushnell, D.M. (1983) Turbulent Drag Reduction for External Flows, AIAA Paper No. 83-0227,
New York.
Bushnell, D.M. (1989) Applications and Suggested Directions of Transition Research, in Fourth
Symp. on Numerical and Physical Aspects of Aerodynamic Flows, Long Beach, CA, 1619
January.
Bushnell, D.M. (1990) Supersonic Aircraft Drag Reduction, AIAA Paper No. 90-1596, New
York.
Bushnell, D.M. (1994) Viscous Drag Reduction in Aeronautics, in Proc. Nineteenth Cong. of
the Int. Council of the Aeronaut. Sci., vol. 1, pp. XXXIIILVI, paper no. ICAS-94-0.1, AIAA,
Washington, D.C.
Bushnell, D.M. (1998) Frontiers of the Responsibly Imaginable in Aeronautics, Dryden Lecture,
AIAA Paper No. 98-0001, Reston, Virginia.
Bushnell, D.M., and Donaldson, C.D. (1990) Control of Submersible Vortex Flows, NASA Tech.
Memo. 102693, Washington, D.C.
Bushnell, D.M., and Greene, G.C. (1991) High-Reynolds-Number Test Requirements in Low-
Speed Aerodynamics, in High Reynolds Number Flows Using Liquid and Gaseous Helium,
ed. R.J. Donnelly, pp. 7985, SpringerVerlag, New York.
Bushnell, D.M., and Hefner, J.N., editors (1990) Viscous Drag Reduction in Boundary Layers,
AIAA, Washington, D.C.
Bushnell, D.M., and Malik, M.R. (1988) Compressibility Inuences on Boundary-Layer Transi-
tion, in Symp. on Physics of Compressible Turbulent Mixing, 2527 October, Princeton, New
Jersey.
Bushnell, D.M., and McGinley, C.B. (1989) Turbulence Control in Wall Flows, Ann. Rev. Fluid
Mech. 21, pp. 120.
BIBLIOGRAPHY 371
Bushnell, D.M., and Moore, K.J. (1991) Drag Reduction in Nature, Ann. Rev. Fluid Mech. 23,
pp. 6579.
Bushnell, D.M., and Trimpi, R.L. (1986) Optimum Supersonic Wind Tunnel, AIAA Paper
Bushnell, D.M., Cary, A.M., and Holley, B.B. (1975) Mixing Length in Low Reynolds Number
Compressible Turbulent Boundary Layers, AIAA J. 13, pp. 11191121.
Bushnell, D.M., Hefner, J.N., and Ash, R.L. (1977) Effect of Compliant Wall Motion on Turbulent
Boundary Layers, Phys. Fluids 20, no. 10, part II, pp. S31S48.
Bushnell, D.M., Malik, M.R., and Harvey, W.D. (1989) Transition Prediction in External Flows
via Linear Stability Theory, in IUTAM Symp. Transsonicum III, eds. J. Zierep and H. Oertel,
pp. 225242, SpringerVerlag, Berlin.
Bushnell, D.M., Yip, L.P., Yao, C.-S., Lin, J.C., Lawing, P.L., Batina, J.T., Hardin, J.C., Horvath,
T.J., Fenbert, J.W., and Domack, C.S. (1993) Reynolds Number Inuences in Aeronautics,
NASA Technical Memorandum No. TM-107730, Washington, D.C.
Bussmann, K., and M unz, H. (1942) Die Stabilit at der laminaren Reibungsschicht mit Absaugung,
Jahrb. Dtsch. Luftfahrtforschung 1, pp. 3639.
Butler, K.M., and Farrell, B.F. (1993) Optimal Perturbations and Streak Spacing in Wall-Bounded
Shear Flow, Phys. Fluids A 5, pp. 774777.
Butter, D.J. (1984) Recent Progress on Development and Understanding of High Lift Systems, in
AGARD Conference on Improvement of Aerodynamic Performance Through Boundary Layer
Control and High Lift Systems, AGARD-CP-365, pp. 1.11.26, Brussels, Belgium.
Calarese, W., Crisler, W.P., and Gustafson, G.L. (1985) Afterbody Drag Reduction by Vortex
Generators, AIAA Paper No. 85-0354, New York.
Cantwell, B.J. (1981) Organized Motion in Turbulent Flow, Ann. Rev. Fluid Mech. 13, pp. 457
515.
Carlson, H.A., and Lumley, J.L. (1996) Flow over an Obstacle Emerging from the Wall of a
Channel, AIAA J. 34, pp. 924931.
Carmichael, B.H. (1974) Application of Sailplane and Low-Drag Underwater Vehicle Technology
to the Long-Endurance Drone Problem, AIAA Paper No. 74-1036, New York.
Carmichael, B.H. (1981) LowReynolds Number Airfoil Survey, NASACR-165803, Washington,
D.C.
Carpenter, P.W. (1988) The Optimization of Compliant Surfaces for Transition Delay, in Turbu-
lence Management and Relaminarisation, eds. H.W. Liepmann and R. Narasimha, pp. 305313,
Carpenter, P.W. (1990) Status of Transition Delay Using Compliant Walls, in Viscous Drag Re-
duction in Boundary Layers, eds. D.M. Bushnell and J.N. Hefner, pp. 79113, AIAA, Washington,
D.C.
Carpenter, P.W. (1993) Optimization of Multiple-Panel Compliant Walls for Delay of Laminar-
Turbulent Transition, AIAA J. 31, pp. 11871188.
Carpenter, P.W. (1997) The Right Sort of Roughness, Nature 388, 21 August, pp. 713714.
Carpenter, P.W. (1998) Current Status of the Use of Wall Compliance for Laminar-Flow Control,
Exp. Thermal & Fluid Sci. 16, pp. 133140.
Carpenter, P.W., and Garrad, A.D. (1985) The Hydrodynamic Stability of Flow over Kramer-Type
Compliant Surfaces. Part 1. TollmienSchlichting Instabilities, J. Fluid Mech. 155, pp. 465510.
Carpenter, P.W., and Garrad, A.D. (1986) The Hydrodynamic Stability of Flow over Kramer-Type
Compliant Surfaces. Part 2. Flow-Induced Surface Instabilities, J. Fluid Mech. 170, pp. 199232.
Cary, A.M., Jr., Weinstein, L.M., and Bushnell, D.M. (1980) Drag Reduction Characteristics
of Small Amplitude Rigid Surface Waves, in Viscous Flow Drag Reduction, ed. G.R. Hough,
pp. 144167, AIAA, New York.
Castro, I.P. (1984) Effects of Free StreamTurbulence on LowReynolds Number Boundary Layers,
J. Fluids Eng. 106, pp. 298306.
Cebeci, T. (1989) Calculation of Flow over Iced Airfoils, AIAA J. 27, pp. 853861.
Cebeci, T., and Chang, K.C. (1978) Calculation of Incompressible Rough-Wall Boundary-Layer
Flows, AIAA J. 16, pp. 730735.
Cebeci, T., and Egan, D.A. (1989) Prediction of Transition due to Isolated Roughness, AIAA J.
27, pp. 870875.
Cercignani, C. (1988) The Boltzmann Equation and Its Applications, SpringerVerlag, Berlin.
372 BIBLIOGRAPHY
Chambers, J.R. (1986) High-Angle-Of-Attack Aerodynamics: Lessons Learned, AIAA Paper
No. 86-1774-CP, New York.
Chan, D.Y.C., and Horn, R.G. (1985) Drainage of Thin Liquid Films, J. Chem. Phys. 83,
pp. 53115324.
Chang, H.-C., and Sen, M. (1995) Application of Chaotic Advection to Heat Transfer, in Chaos
Applied to Fluid Mixing, eds. H. Aref and M.S. El Naschie, pp. 211231, Pergamon Press, Oxford,
Great Britain.
Chang, P.K. (1970) Separation of Flow, Pergamon Press, Oxford, Great Britain.
Chang, P.K. (1976) Control of Flow Separation, Hemisphere, Washington, D.C.
Chapman, S., and Cowling, T.G. (1970) The Mathematical Theory of Non-Uniform Gases, third
edition, Cambridge University Press, Cambridge, Great Britain.
Cheeseman, I.C., and Seed, A.R. (1967) The Application of Circulation Control by Blowing to
Helicopter Rotors, J. R. Aeronaut. Soc. 71, pp. 451467.
Chen, C.-C., Wolf, E.E., and Chang, H.-C. (1993) Low-Dimensional Spatiotemporal Thermal
Dynamics on Nonuniform Catalytic Surfaces, J. Phys. Chemistry 97, pp. 10551064.
Chen, C.P., Goland, Y., and Reshotko (1979) Generation Rate of Turbulent Patches in the Laminar
Boundary Layer of a Submersible, in Viscous FlowDrag Reduction, ed. G.R. Hough, pp. 7389,
AIAA, New York.
Chen, D., and Shi Ying, Z. (1989) Control of Separation in Diffusers Using Forced Unsteadiness,
AIAA Paper No. 89-1015, New York.
Cheng, H.K. (1993) Perspectives on Hypersonic Viscous Flow Research, Ann. Rev. Fluid Mech.
25, pp. 455484.
Cheng, H.K., and Emmanuel, G. (1995) Perspectives on Hypersonic NonequilibriumFlow, AIAA
J. 33, pp. 385400.
Chiwanga, S.C., and Ramaprian, B.R. (1993) The Effect of Convex Wall Curvature on the Large-
Scale Structure of the Turbulent Boundary Layer, Exp. Thermal & Fluid Sci. 6, pp. 168176.
Choi, H., Moin, P., and Kim, J. (1992) Turbulent Drag Reduction: Studies of Feedback Control
and Flow Over Riblets, Department of Mechanical Engineering Report No. TF-55, Stanford
University, Stanford, California.
Choi, H., Moin, P., and Kim, J. (1993) Direct Numerical Simulations of Turbulent Flow over
Riblets, J. Fluid Mech. 255, pp. 503539.
Choi, H., Moin, P., and Kim, J. (1994) Active Turbulence Control for Drag Reduction in Wall-
Bounded Flows, J. Fluid Mech. 262, pp. 75110.
Choi, H., Temam, R., Moin, P., and Kim, J. (1993) Feedback Control for Unsteady Flow and its
Application to the Stochastic Burgers Equation, J. Fluid Mech. 253, pp. 509543.
Choi, K.-S. (1989) Near-Wall Structure of a Turbulent Boundary Layer with Riplets, J. Fluid
Mech. 208, pp. 417458.
Choi, K.-S., editor (1991) Recent Developments in Turbulence Management, Kluwer, Dordrecht,
The Netherlands.
Choi, K.-S., and Clayton, B.R. (1998) The Mechanism of Turbulent Drag Reduction with Wall
Oscillation, in Proc. Int. Symp. on Sea Water Drag Reduction, ed. J.C.S. Meng, pp. 229235,
July 2223, Naval Undersea Warfare Center, Newport, Rhode Island.
Choi, K.-S., and Graham, M. (1998) Drag Reduction of Turbulent Pipe Flows by Circular-Wall
Oscillation, Phys. Fluids 10, pp. 79.
Choi, K.-S., DeBisschop, J.-R., and Graham, M. (1998) Turbulent Boundary-Layer Control by
Means of Spanwise-Wall Oscillation, AIAA J. 36, pp. 11571163.
Choi, K.-S., Yang, X., Clayton, B.R., Glover, E.J., Atlar, M., Semenov, B.N., and Kulik, V.M. (1997)
Turbulent Drag Reduction Using Compliant Surfaces, Proc. R. Soc. London A 453, 2229
2240.
Chow, C.Y., Chen, C.L., and Huang, M.K. (1985) Trapping of Free Vortex by Airfoils with Surface
Suction, AIAA Paper No. 85-0446, New York.
Chu, H.H., and Blick, E.F. (1969) Compliant Surface Drag as a Function of Speed, J. Spacecraft
& Rockets 6, pp. 763764.
Ciccotti, G., and Hoover, W.G., editors (1986) Molecular Dynamics Simulation of Statistical Me-
chanics Systems, North Holland, Amsterdam, The Netherlands.
Cichy, D.R., Harris, J.W., and MacKay, J.K. (1972) Flight Tests of a Rotating Cylinder Flap on a
North American Rockwell YOV-10A Aircraft, NASA CR-2135, Washington, D.C.
BIBLIOGRAPHY 373
Clauser, F.H. (1954) Turbulent Boundary Layers in Adverse Pressure Gradient, J. Aeronaut. Sci.
21, pp. 91108.
Clauser, F.H. (1956) The Turbulent Boundary Layer, Adv. Appl. Mech. 4, pp. 151.
Coles, D. (1956) The Law of the Wake in the Turbulent Boundary Layer, J. Fluid Mech. 1,
pp. 191226.
Coles, D. (1978) A Model for Flow in the Viscous Sublayer, in Workshop on Coherent Structure
of Turbulent Boundary Layers, eds. C.R. Smith and D.E. Abbott, pp. 462475, Lehigh University,
Bethlehem, Pennsylvania.
Coles, D.E. (1962) The Turbulent Boundary Layer in a Compressible Fluid, Rand Corporation
Report No. R-403-PR, Santa Monica, California.
Coles, D.E. (1969) The Young Persons Guide to the Data, in Proc. 1968 AFOSR-IFP-Stanford
Conference on Computation of Turbulent Boundary Layers, eds. D.E. Coles and E.A. Hirst,
vol. 2, pp. 145, Thermosciences Division, Stanford University, Stanford, California.
Colin, P.E., and Williams, J. (1971) Assessment of Lift Augmentation Devices, AGARD-LS-43-71,
AGARD, NATO, Rhode-Saint-G en` ese, Belgium.
Coller, B.D., Holmes, P., and Lumley, J.L. (1994a) Control of Bursting in Boundary Layer Models,
Appl. Mech. Rev. 47, no. 6, part 2, pp. S139S143.
Coller, B.D., Holmes, P., and Lumley, J.L. (1994b) Control of Noisy Heteroclinic Cycles, Physica
D 72, pp. 135160.
Collier, C.P.,Wong, E.W., Belohradsk y, M., Raymo, F.M., Stoddart, J.F., Kuekes, P.J., Williams, R.S.,
and Heath, J.R. (1999) Electronically Congurable Molecular-Based Logic Gates, Science 285,
no. 5426, 16 July, pp. 391394.
Collins, F.G. (1979) Boundary Layer Control on Wings Using Sound and Leading Edge Serrations,
Collins, F.G. (1981) Boundary Layer Control on Wings Using Sound and Leading-Edge Serrations,
AIAA J. 19, pp. 129130.
Collins, F.G., and Zelenevitz, J. (1975) Inuence of Sound upon Separated Flow over Wings,
AIAA J. 13, pp. 408410.
Compton, D.A., and Johnston, J.P. (1991) Streamwise Vortex Development by Pitched and Skewed
Jets in a Turbulent Boundary Layer, AIAA Paper No. 91-0038, New York.
Comte-Bellot, G. (1963) Contribution ` a l
Etude de la Turbulence de Conduite, Doctoral Thesis,

University of Grenoble, France.
Comte-Bellot, G. (1965) Ecoulement Turbulent entre Deux Parois Paralleles, Publications Sci-
enifques et Techniques du Ministere de lAir No. 419, Paris, France.
Cook, W.L., Mickey, D.M., and Quigley, H.G. (1974) Aerodynamics of Jet Flap and Rotating
Cylinder Flap STOL Concepts, AGARDFluid Dynamics Panel on V/STOL Aerodynamics, paper
no. 10, Delft, The Netherlands.
Cooke, J.C., and Brebner, G.G. (1961) The Nature of Separation and Its Prevention by Geometric
Design in a Wholly Subsonic Flow, in Boundary Layer and Flow Control, ed. G.V. Lachmann,
vol. 1, pp. 144185, Pergamon Press, Oxford, Great Britain.
Cooper, A.J., and Carpenter, P.W. (1995) The Effects of Wall Compliance on Instability in Rotating
Disc Flow, AIAA Paper No. 95-2257, Washington, D.C.
Cooper, A.J., and Carpenter, P.W. (1997a) The Stability of Rotating-Disc Boundary-Layer Flow
over a Compliant Wall. Part 1. Type I and II Instabilities, J. Fluid Mech. 350, pp. 231259.
Cooper, A.J., and Carpenter, P.W. (1997b) The Stability of Rotating-Disc Boundary-Layer Flow
over a Compliant Wall. Part 2. Absolute Instability, J. Fluid Mech. 350, pp. 261270.
Cooper, A.J., and Carpenter, P.W. (1997c) The Effect of Wall Compliance on Inexion Point
Instability in Boundary Layers, Phys. Fluids 9, pp. 468470.
Corino, E.R., and Brodkey, R.S. (1969) A Visual Investigation of the Wall Region in Turbulent
Flow, J. Fluid Mech. 37, pp. 130.
Corke, T.C., and Cavalieri, D.A. (1997) Controlled Experiments on Instabilities and Transition to
Turbulence in Supersonic Boundary Layers, AIAA Paper No. 97-1817, Washington, D.C.
Corke, T.C., Cavalieri, D.A., and Matlis, E. (1999) Boundary Layer Instability on a Sharp Cone
at Mach 3.5 with Controlled Input, AIAA J., to appear.
Corke, T.C., Glauser, M.N., and Berkooz, G. (1994) Utilizing Low-Dimensional Dynamical Sys-
tems Models to Guide Control Experiments, Appl. Mech. Rev. 47, no. 6, part 2, pp. S132
S138.
374 BIBLIOGRAPHY
Corke, T.C., Guezennec, Y., and Nagib, H.M. (1980) Modication in Drag of Turbulent Boundary
Layers Resulting fromManipulation of Large-Scale Structures, in Viscous FlowDrag Reduction,
ed. G.R. Hough, pp. 128143, AIAA, New York.
Corke, T.C., and Mangano, R.A. (1989) Resonant Growth of Three-Dimensional Modes in Tran-
sitioning Blasius Boundary Layers, J. Fluid Mech. 209, pp. 93150.
Corke, T.C., Nagib, H.M., and Guezennec, Y. (1981) A New View on Origin, Role and Manip-
ulation of Large Scales in Turbulent Boundary Layers, NASA Contractor Report No. 165861,
Washington, D.C.
Corrsin, S. (1957) Some Current Problems in Turbulent Shear Flow, in Symp. on Naval Hy-
drodynamics, ed. F.S. Sherman, pp. 373400, National Academy of Sciences/National Research
Council Publication No. 515, Washington, D.C.
Corrsin, S. (1959) Outline of Some Topics in Homogenous Turbulent Flow, J. Geophys. Research
64, pp. 21342150.
Corrsin, S. (1974) Limitations of Gradient Transport Models in Random Walks and Turbulence,
in Turbulent Diffusion in Environmental Pollution, eds. F.N. Frenkiel and R.E. Munn, pp. 2560,
Academic Press, New York.
Crabtree, L.F. (1954) The Formation of Regions of Separated Flow on Wing Surfaces, Part I: Low-
Speed Tests on a Two-Dimensional Unswept Wing with a 10% Thick RAE101 Section, Royal
Aircraft Establishment Report Aero. 2528, Farnborough, Great Britain.
Crabtree, L.F. (1957) Effects of Leading-Edge Separation on Thin Wings in Two-Dimensional
Incompressible Flow, J. Aeronaut. Sci. 24, pp. 597604.
Crighton, D.G. (1984) Long-Range Acoustic Scattering by Surface Inhomogeneities Beneath
a Turbulent Boundary Layer, J. Vibration, Acoustics, Stress & Reliability in Design 106,
pp. 376382.
Currie, I.G. (1993) Fundamental Mechanics of Fluids, second edition, McGrawHill, New York.
Cutler, A., and Bradshaw, P. (1986) The Interaction Between a Strong Longitudinal Vortex and a
Turbulent Boundary Layer, AIAA Paper No. 86-1071, New York.
Cutler, A., and Bradshaw, P. (1989) Vortex/Boundary-Layer Interactions, AIAA Paper No. 89-
0083, New York.
Daniel, A.P., Gaster, M., and Willis, G.J.K. (1987) Boundary Layer Stability on Compliant Sur-
faces, Final Report No. 35020, British Maritime Technology Ltd., Teddington, Great Britain.
Darby, R.A. (1954) An Aircraft Manufacturer Looks at Boundary Layer Control, Report No.
RR-1, Fairchild Aircraft Division, Fairchild Engine and Airplane Corp., Hagerstown, Maryland.
Davidson, C.J. (1985) The Experimental Investigation of the Effects of Roughness upon Aerofoil
Characteristics at Low Reynolds Numbers, M.Sc. Thesis, Craneld Institute of Technology,
Craneld, Great Britain.
Davies, C., and Carpenter, P.W. (1997a) Numerical Simulation of the Evolution of Tollmien
Schlichting Waves over Finite Compliant Panels, J. Fluid Mech. 335, pp. 361392.
Davies, C., and Carpenter, P.W. (1997b) Instabilities in a Plane Channel Flow Between Compliant
Walls, J. Fluid Mech. 352, pp. 205243.
Davis, L., editor (1991) Handbook of Genetic Algorithms, Van Nostrand Reinhold, New York.
Deane, A.E., and Sirovich, L. (1991) A Computational Study of Rayleigh-B enard Convection.
Part 1. Rayleigh-Number Scaling, J. Fluid Mech. 222, pp. 231250.
Debye, P., and Cleland, R.L. (1959) Flowof Liquid Hydrocarbons in Porous Vycor, J. Appl. Phys.
30, pp. 843849.
Decken, J.V. (1971) Aerodynamics of Pneumatic High-Lift Devices, in AGARD Conference on
Assessment of Lift Augmentation Devices, eds. P.E. Colin and J. Williams, AGARD-LS-43-71,
paper no. 2, Rhode-Saint-G en` ese, Belgium.
DeCourtye, D., Sen, M., and Gad-el-Hak, M. (1998) Analysis of Viscous Micropumps and
Microturbines, Int. J. Computat. Fluid Dyn. 10, pp. 1325.
De Gennes, P.-G. (1990) Introduction to Polymer Dynamics, Cambridge University Press,
Cambridge, Great Britain.
Delery, J.M. (1985) Shock Wave/Turbulent Boundary Layer Interaction and Its Control, Prog.
Aero. Sci. 22, pp. 209280.
Delville, J., Cordier, L., and Bonnet, J.-P. (1998) Large-Scale-Structure Identication and Control
in Turbulent Shear Flows, in Flow Control: Fundamentals and Practices, eds. M. Gad-el-Hak,
A. Pollard, and J.-P. Bonnet, pp. 199273, SpringerVerlag, Berlin.
BIBLIOGRAPHY 375
DeMeis, R. (1986) Sounding a Happy Note for Lift, Aerospace America 24, August, pp. 1011.
Den, L.M. (1990) Issues in Viscoelastic Fluid Mechanics, Ann. Rev. Fluid Mech. 22, pp. 13
34.
Dennell, R. (1997) The Worlds Oldest Spears, Nature 385, 27 February, pp. 767768.
Despard, R.A., and Miller, J.A. (1971) Separation in Oscillating Laminar Boundary-Layer Flows,
J. Fluid. Mech. 47, pp. 2131.
Didden, N., and Ho, C.-M. (1985) Unsteady Separation in a Boundary Layer Produced by an
Impinging Jet, J. Fluid Mech. 160, pp. 235256.
Dillner, B., May, F., and McMasters, J.H. (1984) Aerodynamic Issues in the Design of High-
Lift Systems for Transport Aircraft, in AGARD Conference on Improvement of Aerodynamic
Performance Through Boundary Layer Control and High Lift Systems, AGARD-CP-365, paper
no. 9, Brussels, Belgium.
Dimotakis, P.E. (1991) Turbulent Free Shear Layer Mixing and Combustion, in High-Speed Flight
Propulsion Systems, eds. S.N.B. Murthy and E.T. Curran, pp. 265340, AIAA, Washington, D.C.
Dimotakis, P.E. (1993) Some Issues on Turbulent Mixing and Turbulence, GuggenheimAeronau-
tical Laboratory Report No. FM93-1a, California Institute of Technology, Pasadena, California.
Dinkelacker, A. (1966) Preliminary Experiments on the Inuence of Flexible Walls on Boundary
Layer Turbulence, J. Sound Vibr. 4, pp. 187214.
DiPrima, R.C., and Swinney, H.L. (1985) Instabilities and Transition in Flow Between Concentric
Rotating Cylinders, in Hydrodynamic Instabilities and the Transition to Turbulence, eds. H.L.
Swinney and J.P. Gollub, second edition, pp. 139180, SpringerVerlag, Berlin.
Ditto, W.L., and Pecora, L.M. (1993) Mastering Chaos, Scientic American 269, August,
pp. 7884.
Ditto, W.L., Rauseo, S.N., and Spano, M.L. (1990) Experimental Control of Chaos, Phys. Rev.
Lett. 65, pp. 32113214.
Dixon, A.E., Lucey, A.D., and Carpenter, P.W. (1994) The Optimization of Viscoelastic Compliant
Walls for Transition Delay, AIAA J. 32, pp. 256267.
Domaradzki, J.A., and Metcalfe, R.W. (1987) Stabilization of Laminar Boundary Layers by Com-
pliant Membranes, Phys. Fluids 30, pp. 695705.
Donnelly, R.J. (1991) High Reynolds Number Flows Using Liquid and Gaseous Helium, Springer
Verlag, New York.
Donohue, G.L., Tiederman, W.G., and Reischman, M.M. (1972) Flow Visualization of the Near-
Wall Region in a Drag-Reducing Channel Flow, J. Fluid Mech. 56, pp. 559575.
Donovan, J.F., and Selig, M.S. (1989) Low Reynolds Number Airfoil Design and Wind Tun-
nel Testing at Princeton University, in Low Reynolds Number Aerodynamics, ed. T.J. Mueller,
D oring, C., Grauer, T., Marek, J., Mettner, M.S., Trah, H.-P., and Willmann, M. (1992) Microma-
chined Thermoelectrically Driven Cantilever Structures for Fluid Jet Deection, in Proc. IEEE
Micro Electro Mechanical Systems 92, pp. 1218, 47 February, Travem unde, Germany.
Dougherty, N.S., and Fisher, D.F. (1980) Boundary Layer Transition on a 10-Degree Cone, AIAA
Dowling, A.P. (1983) Flow-Acoustic Interaction Near a Flexible Wall, J. Fluid Mech. 128,
pp. 181198.
Dowling, A.P. (1985) The Effect of Large-Eddy Breakup Devices on Oncoming Vorticity, J. Fluid
Mech. 160, pp. 447463.
Dowling, A.P. (1986) Mean FlowEffects on the Low-Wavenumber Pressure Spectrumon a Flexible
Surface, J. Fluids Eng. 108, pp. 104108.
Dowling, A.P., and Ffowcs Williams, J.E. (1983) Sound and Sources of Sound, Ellis Horwood,
Chichester, Great Britain.
Drazin, P., and Reid, W. (1981) Hydrodynamic Stability, Cambridge University Press, Cambridge,
Great Britain.
Dreyden, H.L. (1948) Recent Advances in the Mechanics of Boundary Layer Flow, in Advances
in Applied Mechanics, eds. R. von Mises and Th. von K arm an, vol. 1, pp. 140, Academic Press,
Boston, Massachusetts.
Driver, D.M. (1989) Experimental Study of a Three-Dimensional Shear-Driven Turbulent Bound-
ary Layer with Streamwise Adverse Pressure Gradient, Ph.D. Thesis, Stanford University,
Stanford, CA.
376 BIBLIOGRAPHY
Duncan, J.H. (1986) The Response of an Incompressible, Viscoelastic Coating to Pressure Fluctu-
ations in a Turbulent Boundary Layer, J. Fluid Mech. 171, pp. 339363.
Duncan, J.H. (1987) A Comparison of Wave Propagation on the Surfaces of Simple Membrane
Walls and Elastic Coatings Bounded by a Fluid Flow, J. Sound Vibr. 119, pp. 565573.
Duncan, J.H. (1988) The Dynamics of Waves at the Interface Between a Two-Layer Viscoelastic
Coating and a Fluid Flow, J. Fluids Structures 2, pp. 3551.
Duncan, J.H., and Sirkis, J.S. (1992) The Generation of Wave Patterns on Isotropic Coatings by
Pressure Fluctuations in a Turbulent Boundary Layer, J. Sound Vibr. 157, pp. 243264.
Duncan, J.H., and Hsu, C.C. (1984) The Response of a Two-Layer Viscoelastic Coating to Pressure
Disturbances from a Turbulent Boundary Layer, AIAA Paper No. 84-0535, New York.
Duncan, J.H., Waxman, A.M., and Tulin, M.P. (1985) The Dynamics of Waves at the Interface
Between a Viscoelastic Coating and a Fluid Flow, J. Fluid Mech. 158, pp. 177197.
Durbin, P.A., and McKinzie, D.J. (1987) Corona Anemometry for Qualitative Measurement of
Reversing Surface Flow with Application to Separation Control by External Excitation, in Proc.
Forum on Unsteady Flow Separation, ed. K.N. Ghia, pp. 1518, ASME, New York.
Dussan, E.B. (1976) The Moving Contact Line: the Slip Boundary Condition, J. Fluid Mech. 77,
pp. 665684.
Dussan, E.B. (1979) On the Spreading of Liquids on Solid Surfaces: Static and Dynamic Contact
Lines, Annu. Rev. Fluid Mech. 11, pp. 371400.
Dussan, E.B., and Davis, S.H. (1974) On the Motion of FluidFluid Interface Along a Solid
Surface, J. Fluid Mech. 65, pp. 7195.
Dutton, R.A. (1955) Experimental Studies of the Turbulent Boundary Layer on a Flat Plate With
and Without Distributed Suction, Ph.D. Dissertation, Cambridge University, Cambridge, Great
Britain.
Dutton, R.A. (1960) The Effects of Distributed Suction on the Development of Turbulent Boundary
Layers, ARC R&M No. 3155, Aeronautical Research Council, London, Great Britain.
Eckelmann, H. (1974) The Structure of the Viscous Sublayer and the Adjacent Wall Region in a
Turbulent Channel Flow, J. Fluid Mech. 65, pp. 439459.
El ena, M. (1975) Etude des Champs Dynamiques et Thermiques dun Ecoulement Turbulent en
Conduit avec Aspiration ` a la Paroi, Th` ese de Doctorat des Sciences, Universit e dAix-Marseille,
Marseille, France.
El ena, M. (1984) Suction Effects on Turbulence Statistics in a Heated Pipe Flow, Phys. Fluids 27,
pp. 861866.
El ena, M., and Dumas, R. (1978) Champs Dynamiques et Thermiques dun Ecoulement Turbulent
en Conduit avec Aspiration ` a la Paroi, in Sixth Int. Heat Trans. Conf., vol. 5, pp. 239244,
Hemisphere, Washington, D.C.
Ellington, C.P., van den Berg, C., Willmott, A.P., and Thomas, A.L.R. (1996) Leading-Edge Vor-
tices in Insect Flight, Nature 384, 19/26 December, pp. 626629.
Ely, W.L., and Berrier, F.C. (1975) Performance of Steady and Intermittent Blowing Jet Flaps and
Spanwise Upper Surface Slots, Air Force Flight Dynamics Laboratory Report No. AFFDL-TR-
75-128, Wright-Patterson Air Force Base, Ohio.
Emmons, H.W. (1951) The Laminar-Turbulent Transition in a Boundary Layer. Part I, J. Aero.
Sci. 18, pp. 490498.
Eppler, R., and Somers, D.M. (1985) Airfoil Design for Reynolds Numbers Between 50,000 and
500,000, in Proc. Conf. on Low Reynolds Number Airfoil Aerodynamics, ed. T.J. Mueller,
pp. 114, University of Notre Dame, Notre Dame, Indiana.
Epstein, A.H., and Senturia, S.D. (1997) Macro Power from Micro Machinery, Science 276, 23
May, p. 1211.
Epstein, A.H., Senturia, S.D., Al-Midani, O., Anathasuresh, G., Ayon, A., Breuer, K., Chen, K.-S.,
Ehrich, F.F., Esteve, E., Frechette, L., Gauba, G., Ghodssi, R., Groshenry, C., Jacobson, S.A.,
Kerrebrock, J.L., Lang, J.H., Lin, C.-C., London, A., Lopata, J., Mehra, A., Mur Miranda, J.O.,
Nagle, S., Orr, D.J., Piekos, E., Schmidt, M.A., Shirley, G., Spearing, S.M., Tan, C.S., Tzeng,
Y.-S., and Waitz, I.A. (1997) Micro-Heat Engines, Gas Turbines, and Rocket EnginesThe
MIT Microengine Project, AIAA Paper No. 97-1773, AIAA, Reston, Virginia.
Ericsson, L.E. (1967) Comment on Unsteady Airfoil Stall, J. Aircraft 4, pp. 478480.
Ericsson, L.E. (1971) Unsteady Airfoil Stall and Stall Flutter, NASA CR-111906, Washington,
D.C.
BIBLIOGRAPHY 377
Ericsson, L.E. (1988) Moving Wall Effects in Unsteady Flow, J. Aircraft 25, pp. 977990.
Ericsson, L.E., and Reding, J.P. (1971) Unsteady Airfoil Stall, Review and Extension, J. Aircraft
8, pp. 609616.
Ericsson, L.E., and Reding, J.P. (1984) Unsteady Flow Concepts for Dynamic Stall Analysis,
J. Aircraft 21, pp. 601606.
Ericsson, L.E., and Reding, J.P. (1986) Fluid Dynamics of Unsteady Separated Flow. Part I. Bodies
of Revolution, Prog. Aero. Sci. 23, pp. 184.
Ericsson, L.E., and Reding, J.P. (1987) Fluid Dynamics of Unsteady Separated Flow. Part II. Lifting
Surfaces, Prog. Aero. Sci. 24, pp. 249356.
Eriksson, L.J., Allie, M.C., Bremigan, C.D., and Gilbert, J.A. (1988) Active Noise Control and
Specications for Fan Noise Problems, in Noise Control Design: Methods and Practice (Noise-
Con 88 Proceedings), ed. J.S. Bolton, pp. 273278, Noise Control Foundation, Poughkeepsie,
New York.
Erm, L.P., Smits, A.J., and Joubert, P.N. (1987) LowReynolds Number Turbulent Boundary Layers
on a Smooth Flat Surface in a Zero Pressure Gradient, in Fifth Symposium on Turbulent Shear
Flows, eds. F. Durst et al., pp. 2.132.18, SpringerVerlag, Berlin.
Esashi, M., Shoji, S., and Nakano, A. (1989) Normally Closed Microvalve Fabricated on a Silicon
Wafer, Sensors & Actuators 20, pp. 163169.
Falco, R.E. (1974) Some Comments on Turbulent Boundary Layer Structure Inferred from the
Movements of a Passive Contaminant, AIAA Paper No. 74-99, New York.
Falco, R.E. (1977) Coherent Motions in the Outer Region of Turbulent Boundary Layers, Phys.
Fluids 20, no. 10, part II, pp. S124S132.
Falco, R.E. (1980) The Production of Turbulence Near a Wall, AIAA Paper No. 80-l356, New
York.
Falco, R.E. (1983) New Results, a Review and Synthesis of the Mechanism of Turbulence Produc-
tion in Boundary Layers and its Modication, AIAA Paper No. 83-0377, New York.
Falco, R.E. (1991) A Coherent Structure Model of the Turbulent Boundary Layer and Its Ability
to Predict Reynolds Number Dependence, Phil. Trans. R. Soc. London A 336, pp. 103129.
Faller, W.E., Schreck, S.J., and Luttges, M.W. (1994) Real-Time Prediction and Control of Three-
Dimensional Unsteady Separated FlowFields Using Neural Networks, AIAAPaper No. 94-0532,
Washington, D.C.
Fan, L.-S., Tai, Y.-C., and Muller, R.S. (1988) Integrated Movable Micromechanical Structures for
Sensors and Actuators, in IEEE Transactions on Electronic Devices, vol. 35, pp. 724730.
Fan, L.-S., Tai, Y.-C., and Muller, R.S. (1989) IC-Processed Electrostatic Micromotors, Sensors
& Actuators 20, pp. 4147.
Fan, X., Hofmann, L., and Herbert, T. (1993) Active FlowControl with Neural Networks, AIAA
Paper No. 93-3273, Washington, D.C.
Farabee, T.M., and Casarella, M.J. (1991) Spectral Features of Wall Pressure Fluctuations Beneath
Turbulent Boundary Layers, Phys. Fluids A 3, pp. 24102420.
Farge M. (1992) Wavelet Transforms and Their Applications to Turbulence, Ann. Rev. Fluid
Mech. 24, pp. 395457.
Favre, A., Dumas, R., Verollet, E., and Coantic, M. (1966) Couche Limite Turbulente sur Paroi
Poreuse avec Aspiration, J. M ecanique 5, pp. 328.
Feir, J.B. (1965) The Effects of an Arrangement of Vortex Generators Installed to Eliminate Wind
Tunnel Diffuser Separation, UTIAS Technical Note No. 87, Institute for Aerospace Studies,
University of Toronto, Toronto, Canada.
Ferziger, J.H., and Peri c, M. (1996) Computational Methods for Fluid Dynamics, SpringerVerlag,
New York.
Feynman, R.P. (1961) Theres Plenty of Roomat the Bottom, in Miniaturization, ed. H.D. Gilbert,
pp. 282296, Reinhold Publishing, New York.
Ffowcs Williams, J.E. (1965) Sound Radiation from Turbulent Boundary Layers Formed on Com-
pliant Surfaces, J. Fluid Mech. 22, pp. 347358.
Ffowcs Williams, J.E. (1977) Aeroacoustics, Ann. Rev. Fluid Mech. 9, pp. 447468.
Ffowcs Williams, J.E. (1984) Anti-Sound, Proc. R. Soc. London A 395, pp. 6388.
Fiedler, H.E. (1986) Coherent Structures, in Advances in Turbulence, eds. G. Comte-Bellot and
J. Mathieu, pp. 320336, SpringerVerlag, Berlin.
Fiedler, H.E. (1988) Coherent Structures in Turbulent Flows, Prog. Aero. Sci. 25, pp. 231269.
378 BIBLIOGRAPHY
Fiedler H.E. (1998) Control of Free Turbulent Shear Flows, in Flow Control: Fundamentals and
Practices, eds. M. Gad-el-Hak, A. Pollard, and J.-P. Bonnet, pp. 335429, SpringerVerlag, Berlin.
Fiedler, H.E., and Fernholz, H.-H. (1990) On Management and Control of Turbulent Shear Flows,
Prog. Aero. Sci. 27, pp. 305387.
Fiedler, H.E., and Head, M.R. (1966) Intermittency Measurements in the Turbulent Boundary
Fiedler, H.E., Glezer, A., and Wygnanski, I. (1988) Control of Plane Mixing Layer: Some Novel
Experiments, in Current Trends in Turbulence Research, eds. H. Branover, M. Mond, and Y.
Unger, pp. 3064, AIAA, Washington, D.C.
Fischer, M.C., and Ash, R.L. (1974) A General Review of Concepts for Reducing Skin Friction,
Including Recommendations for Future Studies, NASA Technical Memorandum No. X-2894,
Washington, D.C.
Fisher, D.H., and Blick, E.F. (1966) Turbulent Damping by Flabby Skins, J. Aircraft 3, pp. 163
164.
Fitzgerald, E.J., and Mueller, T.J. (1990) Measurements in a Separation Bubble on an Airfoil Using
Laser Velocimetry, AIAA J. 28, pp. 584592.
Flatt, J. (1961) The History of Boundary Layer Control Research in the United States of America,
in Boundary Layer and Flow Control, ed. G.V. Lachmann, vol. 1, pp. 122143, Pergamon Press,
New York.
Flettner, A. (1924) Die Anwendung der Erkenntnisse der Aerodynamik zum Windantrieb von
Schiffen, Jb. Schiffbautech. Ges. 25, pp. 222251.
Fontaine, A.A., Deutsch, S., Brungart, T.A., Petrie, H.L., and Fenstermacker, M. (1999) Drag
Reduction by Coupled Systems: Microbubble Injection with Homogenous Polymer and Surfactant
Solutions, Exp. Fluids 26, pp. 397403.
Foreman, J.E.K. (1989) Sound Analysis and Noise Control, Van Nostrand Reinhold, New York.
Fortuna, G., and Hanratty, T.J. (1971) The Inuence of Drag-Reducing Polymers on Turbulence
in the Viscous Sublayer, J. Fluid Mech. 53, pp. 575586.
Fowler, T.B. (1989) Application of Stochastic Control Techniques to Chaotic Nonlinear Systems,
IEEE Trans. Autom. Control 34, pp. 201205.
Francis, M.S., Keesee, J.E., Lang, J.D., Sparks, G.W., and Sisson, G.E. (1979) Aerodynamic Char-
acteristics of an Unsteady Separated Flow, AIAA J. 17, pp. 13321339.
Fraser, L.A., and Carpenter, P.W. (1985) A Numerical Investigation of Hydroelastic and Hydro-
dynamic Instabilities in Laminar Flows over Compliant Surfaces Comprising One or Two Layers
of Viscoelastic Material, in Numerical Methods in Laminar and Turbulent Flow, eds. C. Taylor
et al., pp. 11711181, Pineridge Press, Swansea, Great Britain.
Frick, C.W., and McCullough, C.B. (1942) Tests of a Heated Low Drag Airfoil, NACA ARR,
December, Washington, D.C.
Frisch, U. (1995) Turbulence: The Legacy of A.N. Kolmogorov, Cambridge University Press, Cam-
bridge, Great Britain.
Frisch, U., and Orszag, S.A. (1990) Turbulence: Challenges for Theory and Experiment, Phys.
Today 43, no. 1, January, pp. 2432.
Fuhr, G., Hagedorn, R., M uller, T., Benecke, W., and Wagner, B. (1992) Microfabricated Electro-
hydrodynamic (EHD) Pumps for Liquids of Higher Conductivity, J. MEMS 1, pp. 141145.
Fujimasa, I. (1996) Micromachines: A New Era in Mechanical Engineering, Oxford University
Fung, K.-Y., editor (1994) Symp. on Aerodynamics & Aeracoustics, World Scientic, Singapore.
Gabriel, K.J. (1995) Engineering Microscopic Machines, Scientic American 260, September,
pp. 150153.
Gabriel, K.J., Jarvis, J., and Trimmer, W., editors (1988) Small Machines, Large Opportunities:
A Report on the Emerging Field of Microdynamics, National Science Foundation, published by
AT&T Bell Laboratories, Murray Hill, New Jersey.
Gabriel, K.J., Tabata, O., Shimaoka, K., Sugiyama, S., and Fujita, H. (1992) Surface-Normal
Electrostatic/Pneumatic Actuator, in Proc. IEEEMicro Electro Mechanical Systems 92, pp. 128
131, 47 February, Travem unde, Germany.
Gadd, G.E. (1960) Boundary Layer Separation in the Presence of Heat Transfer, AGARD Report
No. 280, AGARD, NATO, Rhode-Saint-G en` ese, Belgium.
Gadd, G.E., Cope, W.F., and Attridge, J.L. (1958) Heat-Transfer and Skin-Friction Measurements
BIBLIOGRAPHY 379
at a Mach Number of 2.44 for a Turbulent Boundary Layer on a Flat Surface and in Regions of
Separated Flow, ARC R&M 3148, Aeronautical Research Council, London, Great Britain.
Gad-el-Hak, M. (1986a) Boundary Layer Interactions with Compliant Coatings: An Overview,
Appl. Mech. Rev. 39, pp. 511524.
Gad-el-Hak, M. (1986b) The Response of Elastic and Viscoelastic Surfaces to a Turbulent Boundary
Layer, J. Appl. Mech. 53, pp. 206212.
Gad-el-Hak, M. (1986c) The Use of the Dye-Layer Technique for Unsteady Flow Visualization,
J. Fluids Eng. 108, pp. 3438.
Gad-el-Hak, M. (1987a) Compliant Coatings Research: A Guide to the Experimentalist, J. Fluids
Structures 1, pp. 5570.
Gad-el-Hak, M. (1987b) Unsteady Separation on Lifting Surfaces, Appl. Mech. Rev. 40, pp. 441
453.
Gad-el-Hak, M. (1988a) Review of Flow Visualization Techniques for Unsteady Flows, in Flow
Visualization IV, ed. C. V eret, pp. 112, Hemisphere, Washington, D.C.
Gad-el-Hak, M. (1988b) Visualization Techniques for Unsteady Flows: An Overview, J. Fluids
Eng. 110, pp. 231243.
Gad-el-Hak, M. (1989) Flow Control, Appl. Mech. Rev. 42, pp. 261293.
Gad-el-Hak, M. (1990) Control of Low-Speed Airfoil Aerodynamics, AIAAJ. 28, pp. 15371552.
Gad-el-Hak, M. (1992) Splendor of Fluids in Motion, Prog. Aero. Sci. 29, pp. 81123.
Gad-el-Hak, M. (1993) Innovative Control of Turbulent Flows, AIAA Paper No. 93-3268,
Washington, D.C.
Gad-el-Hak, M. (1994) Interactive Control of Turbulent Boundary Layers: A Futuristic Overview,
AIAA J. 32, pp. 17531765.
Gad-el-Hak, M. (1995) Questions in Fluid Mechanics: Stokes Hypothesis for a Newtonian,
Isotropic Fluid, J. Fluids Eng. 117, pp. 35.
Gad-el-Hak, M. (1996a) Compliant Coatings: ADecade of Progress, Appl. Mech. Rev. 49, no. 10,
part 2, pp. S1S11.
Gad-el-Hak, M. (1996b) Modern Developments in Flow Control, Appl. Mech. Rev. 49, pp. 365
379.
Gad-el-Hak, M. (1997) The Last Conundrum, Appl. Mech. Rev. 50, no. 12, part 1, pp. 12.
Gad-el-Hak, M. (1998a) Fluid Mechanics from the Beginning to the Third Millennium, Int. J.
Engng. Ed. 14, pp. 177185.
Gad-el-Hak, M. (1998b) Frontiers of FlowControl, in FlowControl: Fundamentals and Practices,
eds. M. Gad-el-Hak, A. Pollard, and J.-P. Bonnet, pp. 109153, SpringerVerlag, Berlin.
Gad-el-Hak, M. (1999) The Fluid Mechanics of MicrodevicesThe Freeman Scholar Lecture,
Gad-el-Hak, M., and Bandyopadhyay, P.R. (1994) Reynolds Number Effects in Wall-Bounded
Flows, Appl. Mech. Rev. 47, pp. 307365.
Gad-el-Hak, M., and Blackwelder, R.F. (1985) The Discrete Vortices from a Delta Wing, AIAA
J. 23, pp. 961962.
Gad-el-Hak, M., and Blackwelder, R.F. (1987a) Simulation of Large-Eddy Structures in a Turbulent
Boundary Layer, AIAA J. 25, pp. 12071215.
Gad-el-Hak, M., and Blackwelder, R.F. (1987b) Control of the Discrete Vortices from a Delta
Wing, AIAA J. 25, pp. 10421049.
Gad-el-Hak, M., and Blackwelder, R.F. (1987c) ADrag Reduction Method for Turbulent Boundary
Layers, AIAA Paper No. 87-0358, New York.
Gad-el-Hak, M., and Blackwelder, R.F. (1989) Selective Suction for Controlling Bursting Events
in a Boundary Layer, AIAA J. 27, pp. 308314.
Gad-el-Hak, M., Blackwelder, R.F., and Riley, J.J. (1981) On the Growth of Turbulent Regions in
Laminar Boundary Layers, J. Fluid Mech. 110, pp. 7395.
Gad-el-Hak, M., Blackwelder, R.F., and Riley, J.J. (1984) On the Interaction of Compliant Coatings
with Boundary Layer Flows, J. Fluid Mech. 140, pp. 257280.
Gad-el-Hak, M., Blackwelder, R.F., and Riley, J.J. (1985) Visualization Techniques for Study-
ing Transitional and Turbulent Flows, in Flow Visualization III, ed. W.J. Yang, pp. 568575,
Hemisphere, Washington, D.C.
Gad-el-Hak, M., and Bushnell, D.M. (1991a) Status and Outlook of Flow Separation Control,
380 BIBLIOGRAPHY
Gad-el-Hak, M., and Bushnell, D.M. (1991b) Separation Control: Review, J. Fluids Eng. 113,
pp. 530.
Gad-el-Hak, M., Davis, S.H., McMurray, J.T., and Orszag, S.A. (1984) On the Stability of the
Decelerating Boundary Layer, J. Fluid Mech. 138, pp. 297323.
Gad-el-Hak, M., and Ho, C.-M. (1985) The Pitching Delta Wing, AIAA J. 23, pp. 16601665.
Gad-el-Hak, M., and Ho, C.-M. (1986a) Unsteady Vortical Flow Around Three-Dimensional
Lifting Surfaces, AIAA J. 24, pp. 713721.
Gad-el-Hak, M., and Ho, C.-M. (1986b) Unsteady Flow Around an Ogive-Cylinder, J. Aircraft
23, pp. 520528.
Gad-el-Hak, M., and Hussain, A.K.M.F. (1986) Coherent Structures in a Turbulent Boundary
Layer. Part 1. Generation of Articial Bursts, Phys. Fluids 29, pp. 21242139.
Gad-el-Hak, M., Pollard, A., and Bonnet, J.-P. (1998) Flow Control: Fundamentals and Practices,
Gad-el-Hak, M., and Sen, M. (1996) Fluid Mechanics in the Next Century, Appl. Mech. Rev. 49,
no. 3, pp. IIIIV.
Gadetskii, V.M., Serebriiskii, I.A.M., and Fomin, V.M. (1972) Investigation of the Inuence
of Vortex Generators on Turbulent Boundary Layer Separation, Uchenye Zapiski TSAGI 3,
pp. 2228.
Gallagher, J.A., and Thomas, A.S.W. (1984) Turbulent Boundary Layer Characteristics over
Streamwise Grooves, AIAA Paper No. 84-2185, New York.
Galperin, B., and Orszag, S.A., editors (1993) Large Eddy Simulations of Complex Engineering
and Geophysical Flows, Cambridge University Press, Cambridge, Great Britain.
Gampert, B., Homann, K., and Rieke, H.B. (1980) The Drag Reduction in Laminar and Turbulent
Boundary Layers by Prepared Surfaces with Reduced Momentum Transfer, Israel J. Technology
18, pp. 287292.
Garcia, E.J., and Sniegowski, J.J. (1993) The Design and Modelling of a Comb-Drive-Based Mi-
croengine for Mechanism Drive Applications, in Proc. Seventh Int. Conf. on Solid-State Sensors
and Actuators (Transducers 93), pp. 763766, Yokohama, Japan, 710 June.
Garcia, E.J., and Sniegowski, J.J. (1995) Surface Micromachined Microengine, Sensors & Actu-
ators A 48, pp. 203214.
Garnkel, A., Spano, M.L., Ditto, W.L., and Weiss, J.N. (1992) Controlling Cardiac Chaos,
Science 257, pp. 12301235.
Garrad, A.D., and Carpenter, P.W. (1982) ATheoretical Investigation of Flow-Induced Instabilities
in Compliant Coatings, J. Sound Vibr. 85, pp. 483500.
Gartling, D.K. (1970) Tests of Vortex Generators to Prevent Separation of Supersonic Flow in a
Compression Corner, Applied Research Laboratory Report No. ARL-TR-70-44, University of
Texas, Austin, TX. (Also available from U.S. NTIS; Document Number AD-734154.)
Gaster, M. (1988) Is the Dolphin a Red Herring?, in Turbulence Management and Relaminari-
sation, eds. H.W. Liepmann and R. Narasimha, pp. 285304, SpringerVerlag, Berlin.
Gates, B., Myhrvold, N., and Rinearson, P. (1995) The Road Ahead, Viking, New York.
Gebert, G.A. (1988) Turbulent Boundary Layer Modication by Streamlined Devices, Ph.D.
Thesis, University of Notre Dame, Notre Dame, Indiana.
Gedney, C.J. (1983) The Cancellation of a Sound-Excited TollmienSchlichting Wave with Plate
Vibration, Phys. Fluids 26, pp. 11581160.
Gee, M.L., McGuiggan, P.M., Israelachvili, J.N., and Homola, A.M. (1990) Liquid to Solidlike
Transitions of Molecularly Thin Films Under Shear, J. Chem. Phys. 93, pp. 18951906.
George, W.K., and Castillo, L. (1993) Boundary Layers with Pressure Gradient: Another Look at
the Equilibrium Theory, in Near-Wall Turbulent Flows, eds. R.M.C. So, C.G. Speziale, and B.E.
Launder, pp. 901910, Elsevier, Amsterdam.
George, W.K., Castillo L. (1997) Zero-Pressure-Gradient Turbulent Boundary Layer, Appl. Mech.
Rev. 50, no. 12, part 1, pp. 689729.
George, W.K., Castillo, L., and Knecht, P. (1992) The Zero-Pressure Gradient Turbulent Boundary
Layer Revisited, in Thirteenth Symposiumon Turbulence, ed. X.B. Reed, 2123 April, University
of Missouri, Rolla, Missouri.
Layer, in William C. Reynolds Anniversary Symposium on Turbulence, 2223 March, Asilomar,
California.
BIBLIOGRAPHY 381
Layer, unpublished manuscript submitted to Phys. Fluids.
Gerbeth, G., editor (1997) in Proc. Int. Workshop on Electromagnetic Boundary Layer Control
(EBLC) for Saltwater Flows, 78 July, Forschungszentrum Rossendorf, Dresden, Germany.
Germano, M., Piomelli, U., Moin, P., and Cabot, W.H. ( 1991) A Dynamic Subgrid-Scale Eddy
Viscosity Model, Phys. Fluids A 3, pp. 17601765.
Ghosh, S., Chang, H.-C., and Sen, M. (1992) Heat-Transfer Enhancement due to Slender Recir-
culation and Chaotic Transport Between Counter-Rotating Eccentric Cylinders, J. Fluid Mech.
238, pp. 119154.
Gillis, J.C. (1980) Turbulent Boundary Layer on a Convex, Curved Surface, Ph.D. Dissertation,
Stanford University, Stanford, California.
Glanz, J. (1995) Computer Scientists Rethink Their Disciplines Foundations, Science 269,
no. 5229, 8 September, pp. 13631364.
Gleick, J. (1987) ChaosMaking a New Science, Viking, New York.
Goldberg, D.E. (1989) Genetic Algorithms in Search, Optimization, and Machine Learning,
AddisonWesley, Reading, Massachusetts.
Gold Feld, M.A., and Zatoloka, V.V. (1979) On the Improvement of Separating Properties of
a Turbulent Boundary Layer as a Result of the Effect of a Shock Wave, Izvestiya Sibirskogo
Otdeleniya, Akademii Nank 3, pp. 4047.
Goldstein, M.E. (1984) Generation of Instability Waves in Flows Separating from Smooth Sur-
faces, J. Fluid Mech. 145, pp. 7194.
Goldstein, M.E., and Hultgren, L.S. (1989) Boundary-Layer Receptivity to Long-Wave Free-Stream
Disturbances, Ann. Rev. Fluid Mech. 21, pp. 137166.
Goldstein, S. (1948) On Laminar Boundary Layer Flow Near a Position of Separation, Q. J.
Mech. Appl. Math. 1, pp. 4369.
Goodman, W.L. (1985) Emmons Spot Forcing for Turbulent Drag Reduction, AIAA J. 23,
pp. 155157.
G ortler, H. (1955) Dreidimensionales zur Stabilit atstheorie laminarer Grenzschichten, ZAMM
35, pp. 362363.
Grant, H.L. (1958) The Large Eddies of Turbulent Motion, J. Fluid Mech. 4, pp. 149190.
Granville, P.S. (1977) Drag and Turbulent Boundary Layer of Flat Plates at Low Reynolds Num-
bers, J. Ship Research 21, pp. 3039.
Granville, P.S. (1979) Drag of Underwater Bodies, in Hydroballistics Design Handbook, vol. 1,
pp. 309341, SEAHAC TR 79-1, Naval Sea Systems Command, Washington, D.C.
Grappin, R., and L eorat, J. (1991) Lyapunov Exponents and the Dimension of Periodic In-
compressible Navier-Stokes Flows: Numerical Measurements, J. Fluid Mech. 222, pp. 61
94.
Grass, A.J. (1971) Structural Features of Turbulent Flow over Smooth and Rough Boundaries,
Gratzer, L.B. (1971) Analysis of Transport Applications for High Lift Schemes, in AGARDCourse
on Assessment of Lift Augmentation Devices, eds. P.E. Colin and J. Williams, AGARD-LS-43-71,
paper no. 7, Rhode-Saint-G en` ese, Belgium.
Gravesen, P., Branebjerg, J., and Jensen, O.S. (1993) MicrouidicsA Review, J. Micromech.
Microeng. 3, pp. 168182.
Gray, J. (1936) Studies in Animal Locomotion. VI. The Propulsive Powers of the Dolphin, J. Exp.
Biol. 13, pp. 192199.
Gregory, N., Stuart, J.T., and Walker, W.S. (1955) On the Stability of Three-Dimensional Boundary
Layers with Applications to the Flow due to a Rotating Disk, Phil. Trans. R. Soc. London A
248, pp. 155199.
Grigson, C.W.B. (1992) An Accurate Smooth Friction Line for Use in Performance Prediction, in
Proc. Roy. Inst. Nav. Arch. W5, pp. 19.
Gupta, A.K., and Kaplan, R.E. (1972) Statistical Characteristics of Reynolds Stress in a Turbulent
Boundary Layer, Phys. Fluids 15, pp. 981985.
Gutmark, E.J., and Grinstein, F.F. (1999) Flow Control with Noncircular Jets, Ann. Rev. Fluid
Mech. 31, pp. 239272.
Gutmark, E.J., and Ho, C.-M. (1986) Visualization of a Forced Elliptical Jet, AIAA J. 24,
pp. 684685.
382 BIBLIOGRAPHY
Gutmark, E.J., Schadow, K.C., and Yu, K.H. (1995) Mixing Enhancement in Supersonic Free
Shear Flows, Ann. Rev. Fluid Mech. 27, pp. 375417.
Guvenc, M.G. (1985) V-Groove Capillary for LowFlowControl and Measurement, in Microma-
chining and Micropackaging of Transducers, eds. C.D. Fung, P.W. Cheung, W.H. Ko, and D.G.
Fleming, pp. 215223, Elsevier, Amsterdam, the Netherlands.
Gyr, A., and Bewersdorff, H.-W. (1995) Drag Reduction of Turbulent Flows by Additives, Kluwer,
Dordrecht, the Netherlands.
Haight, C.H., Reed, T.D., and Morland, B.T. (1974) Design Studies of Transonic and STOLAirfoils
with Active Diffusion Control, Advanced Technology Center Report No. ATC-B-94300/4CR-
24, Dallas, Texas. (Also available from U.S. NTIS; Document Number AD-A011928/9.)
Haile, E., Delabroy, O., Durox, D., Lacas, F., and Candel, S. (1998) Combustion Enhancement by
Active Control, in Flow Control: Fundamentals and Practices, eds. M. Gad-el-Hak, A. Pollard,
and J.-P. Bonnet, pp. 467499, SpringerVerlag, Berlin.
Haile, J.M. (1993) Molecular Dynamics Simulation: Elementary Methods, Wiley, New York.
Halsey, T.C., and Martin, J.E. (1993) Electrorheological Fluids, Scientic American 269, October,
pp. 5864.
Hama, F.R. (1947) The Turbulent Boundary Layer Along a Flat Plate, Parts I and II, Reports of
the Institute of Science and Technology 1, pp. 1316 and 4950, University of Tokyo, Tokyo,
Japan.
Hama, F.R. (1954) Boundary-Layer Characteristics for Smooth and Rough Surfaces, Trans. Soc.
Nav. Archt. Marine Engrs. 62, pp. 333358.
Hancock, P.E. (1980) The Effect of Free-StreamTurbulence on Turbulent Boundary Layers, Ph.D.
Dissertation, Imperial College, London, Great Britain.
Hancock, P.E., and Bradshaw, P. (1983) The Effect of Free-StreamTurbulence on Turbulent Bound-
ary Layers, J. Fluids Eng. 105, pp. 284289.
Hanratty, T.J. (1990) A Conceptual Model of the Viscous Wall Region, in Near-Wall Turbulence:
1988 Zoran Zari c Memorial Conference, eds. S.J. Kline and N.H. Afgan, pp. 81103, Hemisphere,
New York.
Hansen, R.J., and Hunston, D.L. (1974) An Experimental Study of Turbulent Flows over Com-
pliant Surfaces, J. Sound Vibr. 34, pp. 297308.
Harder, K.J., and Tiederman, W.G. (1991) Drag Reduction and Turbulent Structure in Two-
Dimensional Channel Flows, Phil. Trans. R. Soc. London A 336, pp. 1934.
Haritonidis, J.H., Senturia, S.D., Warkentin, D.J., and Mehregany, M. (1990a) Optical Micro-
pressure Transducer, United States Patent number 4,926,696.
Haritonidis, J.H., Senturia, S.D., Warkentin, D.J., and Mehregany, M. (1990b) Pressure Trans-
ducer Apparatus, United States Patent number 4,942,767.
Harley, J.C., Huang, Y., Bau, H.H., and Zemel, J.N. (1995) Gas Flow in Micro-Channels,
J. Fluid Mech. 284, pp. 257274.
Harris, C.D., and Bartlett, D.W. (1972) Wind-Tunnel Investigation of Effects of Underwing Leading
Edge Vortex Generators on a Supercritical-Wing Research Airplane Conguration, NASATMX-
2471, Washington, D.C.
Harvey, J.K., and Perry, F.J. (1971) Floweld Produced by Trailing Vortices in the Vicinity of the
Ground, AIAA J. 9, pp. 16591660.
Harvey, W.D. (1986) Low-Reynolds Number Aerodynamics Research at NASA Langley Research
Center, in Proc. Int. Conf. on Aerodynamics at LowReynolds Numbers, vol. II, pp. 19.119.49,
Royal Aeronautical Society, London, Great Britain.
Harvey, W.D., Bushnell, D.M., and Beckwith, I.E. (1969) On the Fluctuating Properties of Turbu-
lent Boundary Layers for Mach Numbers up to 9.0, NASA TND-5496, Washington, D.C.
Hayakawa, L., and Squire, L.C. (1982) The Effect of the Upstream Boundary Layer State on the
Shock Interaction at a Compression Corner, J. Fluid Mech. 122, pp. 369394.
Hayes, S., Grebogi, C., and Ott, E. (1994) Communicating with Chaos, Phys. Rev. Lett. 70,
pp. 30313040.
Hayes, S., Grebogi, C., Ott, E., and Mark, A. (1994) Experimental Control of Chaos for Commu-
nication, Phys. Rev. Lett. 73, pp. 17811784.
Hazen, D.C. (1984) Aeronautics Technology Possibilities for 2000: Report of a Workshop. Aero-
nautics and Space Engineering Board, Commission on Engineering and Technical Systems, U.S.
National Research Council, Washington, D.C.
BIBLIOGRAPHY 383
Head, M.R., and Bandyopadhyay, P.R. (1978) Combined Flow Visualization and Hot-Wire Mea-
surements, in Coherent Structure of Turbulent Boundary Layers, eds. C.R. Smith and D.E.
Abbott, pp. 98129, Lehigh University, Bethlehm, Pennsylvania.
Head, M.R., and Bandyopadhyay, P.R. (1981) New Aspects of Turbulent Boundary-Layer Struc-
ture, J. Fluid Mech. 107, pp. 297338.
Heckl, M. (1988) The Use of Mathematical Methods in Noise Control Design, in Noise Con-
trol Design: Methods and Practice (Noise-Con 88), ed. J.S. Bolton, pp. 2738, Noise Control
Foundation, Poughkeepsie, New York.
Hefner, J.N. (1988) Dragging Down Fuel Costs, Aerospace America 26, January, pp. 1416.
Hefner, J.N., and Sabo, F.E., editors (1987) Research in Natural Laminar Flow and Laminar Flow
Control, Part 1, NASA CP-2487, Washington, D.C.
Hefner, J.N., Anders, J.B., and Bushnell, D.M. (1983) Alteration of Outer Flow Structures for
Turbulent Drag Reduction, AIAA Paper No. 83-0293, New York.
Hefner, J.N., Weinstein, L.M., and Bushnell, D.M. (1980) Large-Eddy Breakup Scheme for Turbu-
lent Viscous Drag Reduction, in Viscous Flow Drag Reduction, ed. G.R. Hough, pp. 110127,
AIAA, New York.
Hendricks, E.W., and Ladd, D.M. (1983) Effect of Surface Roughness on the Delayed Transition
on 9:1 Heated Ellipsoid, AIAA J. 21, pp. 14061409.
Hendricks, E.W., Lawler, J.V., Horne, M.P., Handler, R.A., and Swearingen, J.D. (1989) Experi-
ments in Drag-Reducing Polymer Flows, in Advances in Fluid Mechanics Measurements, ed. M.
Gad-el-Hak, pp. 535568, SpringerVerlag, New York.
Henry, J.R., Wood, C.C., and Wilbur, S.W. (1956) Summary of Subsonic Diffuser Data, NACA
RML-56F05, Washington, D.C.
Hess, D.E. (1990) An Experimental Investigation of a Compliant Surface Beneath a Turbulent
Boundary Layer, Ph.D. Dissertation, The Johns Hopkins University, Baltimore, Maryland.
Hess, D.E., Peattie, R.A., and Schwarz, W.H. (1993) ANonintrusive Method for the Measurement
of Flow-Induced Surface Displacement of a Compliant Surface, Exp. Fluids 14, pp. 7884.
Hinch, E.J. (1977) Mechanical Models of Dilute Polymer Solutions in Strong Flows, Phys. Fluids
20, no. 10, part II, pp. S22S30.
Hinze, J.O. (1975) Turbulence, second edition, McGrawHill, New York.
Hirschfelder, J.O., Curtiss, C.F., and Bird, R.B. (1954) Molecular Theory of Gases and Liquids,
Wiley, New York.
Ho, C.-M. (1986) An Alternative Look at the Unsteady Separation Phenomenon, in Recent
Advances in Aerodynamics, eds. A. Krothappalli and C.A. Smith, pp.165178, SpringerVerlag,
New York.
Ho, C.-M., and Gutmark, E.J. (1987) Vortex Induction and Mass Entrainment in a Small Aspect-
Ratio Elliptic Jet, J. Fluid Mech. 179, pp. 385405.
Ho, C.-M., and Huang, L.-S. (1982) Subharmonics and Vortex Merging in Mixing Layers,
J. Fluid Mech. 119, pp. 443473.
Ho, C.-M., and Huerre, P. (1984) Perturbed Free Shear Layers, Ann. Rev. Fluid Mech. 16,
pp. 365424.
Ho, C.-M., and Tai, Y.-C. (1996) Review: MEMS and Its Applications for FlowControl, J. Fluids
Eng. 118, pp. 437447.
Ho, C.-M., and Tai, Y.-C. (1998) Micro-Elctro-Mechanical Systems (MEMS) and Fluid Flows,
Ann. Rev. Fluid Mech. 30, pp. 579612.
Ho, C.-M., Tung, S., Lee, G.B., Tai, Y.-C., Jiang, F., and Tsao, T. (1997) MEMSA Technology
for Advancements in Aerospace Engineering, AIAA Paper No. 97-0545, Reston, Virgina.
Hoffmann, J.A. (1981) Effects of Free StreamTurbulence on Diffuser Performance, J. Fluids Eng.
103, pp. 385390.
Hoffmann, J.A., Kassir, S.M., and Larwood, S.M. (1988) The Inuence of Free Stream Turbu-
lence on Turbulent Boundary Layers with Mild Adverse Pressure Gradients, NASA CR-184677,
Washington, D.C.
Hogan, H. (1996) Invasion of the Micromachines, New Scientist 29, June, pp. 2833.
Holland, J.H. (1992) Adaptation in Natural and Articial Systems, MIT Press, Cambridge,
Massachusetts.
Holmes, B.J. (1988) NLF Technology is Ready to Go, Aerospace America 26, January, pp. 16
20.
384 BIBLIOGRAPHY
Holmes, P., Lumley, J.L., and Berkooz, G. (1996) Turbulence, Coherent Structures, Dynamical
Systems and Symmetry, Cambridge University Press, Cambridge, Great Britain.
Holstein, H. (1940) Messungen zur Laminarhaltung der Grenzschicht an einemFl ugel, Lilienthal-
Bericht S10, pp. 1727.
Hooshmand, A., Youngs, R., Wallace, J.M., and Balint, J.-L. (1983) An Experimental Study of
Changes in the Structure of a Turbulent Boundary Layer Due to Surface Geometry Changes,
Hopkins, E.J., Keener, E.R., and Polek, T.E. (1972) Hypersonic Turbulent Skin Friction and Bound-
ary Layer Proles on Nonadiabatic Flat Plates, AIAA J. 10, pp. 4048.
Horstmann, K.-H., and Quast, A. (1981) Widerstandsverminderung durch Blasturbulatoren,
DFVLR Report No. FB-81-33, Braunschweig, West Germany.
Hough, G.R., editor (1980) Viscous Flow Drag Reduction, AIAA, New York.
Howard, F.G., and Goodman, W.L. (1985) Axisymmetric Bluff-Body Drag Reduction Through
Geometrical Modication, J. Aircraft 22, pp. 516522.
Howard, F.G., and Goodman, W.L. (1987) Drag Reduction on a Bluff Body at Yaw Angles to 30
Degrees, J. Spacecraft & Rockets 24, pp. 179181.
Howard, F.G., Hefner, J.N., and Srokowski, A.J. (1975) Multiple Slot Skin Friction Reduction,
Howard, F.G., Quass, B.F., Weinstein, L.M., and Bushnell, D.M. (1981) Longitudinal Afterbody
Grooves and Shoulder Radiusing for Low Speed Bluff Body Drag Reduction, ASME Paper
No. 81-WA/FE-5, New York.
Howarth, L. (1938) On the Solution of the Laminar Boundary Layer Equations, Proc. R. Soc.
London A 164, pp. 547579.
Hoyt, J.W. (1972) Turbulent Flow of Drag-Reducing Suspensions, Report No. TP 299, Naval
Undersea Center, San Diego, California.
Hoyt, J.W. (1979) Polymer Drag ReductionA Literature Review, in Second Int. Conf. on Drag
Reduction, Paper No. A1, BHRA Fluid Engineering, Craneld, Great Britain.
Hsiao, F.-B., Liu, C.-F., and Shyu, J.-Y. (1990) Control of Wall-Separated Flowby Internal Acoustic
Excitation, AIAA J. 28, pp. 14401446.
Hu, H.H., and Bau, H.H. (1994) Feedback Control to Delay or Advance Linear Loss of Stability
in Planar Poiseuille Flow, Proc. Roy. Soc. London A 447, pp. 299312.
Huang, L.S., Maestrello, L., and Bryant, T.D. (1987) Separation Control Over an Airfoil at High
Angles of Attack by Sound Emanating from the Surface, AIAA Paper No. 87-1261, New York.
Huang, P.G., Bradshaw, P., and Coakley, T.J. (1993) Skin Friction and Velocity Prole Family for
Compressible Turbulent Boundary Layers, AIAA J. 31, pp. 16001604.
Huberman, B. (1990) The Control of Chaos, in Proc. Workshop on Applications of Chaos, 47
December, San Francisco, California.
Huberman, B.A., and Lumer, E. (1990) Dynamics of Adaptive Systems, IEEE Trans. Circuits
Syst. 37, pp. 547550.
H ubler, A., and L uscher, E. (1989) Resonant Stimulation and Control of Nonlinear Oscillators,
Naturwissenschaften 76, pp. 6769.
Huerre, P., and Monkewitz, P.A. (1990) Local and Global Instabilities in Spatially Developing
Flows, Ann. Rev. Fluid Mech. 22, pp. 473537.
Huffman, G.D., and Bradshaw, P. (1972) A Note on von K arm ans Constant in Low Reynolds
Number Turbulent Flows, J. Fluid Mech. 53, pp. 4560.
Hunt, J.C.R., Carruthers, D.J., and Fung, J.C.H. (1991) Rapid Distortion Theory as a Means
of Exploring the Structure of Turbulence, in New Perspectives in Turbulence, ed. L. Sirovich,
Hunter, P.A., and Johnson, H.I. (1954) A Flight Investigation of the Practical Problems Associated
with Porous Leading-Edge Suction, NACA TN-3062, Washington, D.C.
Hurley, D.G. (1961) The Use of Boundary Layer Control to Establish Free Streamline Flows, in
Boundary Layer and Flow Control, ed. G.V. Lachmann, vol. 1, pp. 295341, Pergamon Press,
Husain, H.S., and Hussain, A.K.M.F. (1983) Controlled Excitation of Elliptic Jets, Phys. Fluids
26, pp. 27632766.
Husain, H.S., and Hussain, F. (1991) Elliptic Jets. Part 2. Dynamics of Coherent Structures: Pair-
ing, J. Fluids Mech. 233, pp. 439482.
BIBLIOGRAPHY 385
Husain, H.S., and Hussain, F. (1993) Elliptic Jets. Part 3. Dynamics of Preferred Mode Coherent
Structures, J. Fluids Mech. 248, pp. 315361.
Hussain, A.K.M.F. (1983) Coherent StructuresReality and Myth, Phys. Fluids 26, pp. 2816
2850.
Hussain, A.K.M.F. (1986) Coherent Structures and Turbulence, J. Fluid Mech. 173, pp. 303356.
Hussain, A.K.M.F., and Hasan, M.A.Z. (1985) Turbulence Suppression in Free Turbulent Shear
Flows Under Controlled Excitation. Part 2. Jet-Noise Reduction, J. Fluid Mech. 150, pp. 159
168.
Hussain, A.K.M.F., and Reynolds, W.C. (1975) Measurements in Fully Developed Turbulent Chan-
nel Flow, J. Fluids Eng. 97, pp. 568580.
Hussain, F., and Husain, H.S. (1989) Elliptic Jets. Part 1. Characteristics of Unexcited and Excited
Jets, J. Fluids Mech. 208, pp. 257320.
Hussaini, M.Y., editor (1997) Collected Papers of Sir James Lighthill, vols. IIV, Oxford University
Huyer, S.A., Robinson, M.C., and Luttges, M.W. (1990) Unsteady Aerodynamic Loading Produced
by a Sinusoidally Oscillating Delta Wing, AIAA Paper No. 90-1536, New York.
Iglisch, R. (1944) Exakte Berechnung der laminaren Reibungsschicht an der l angsangestr omten
ebenen Platte mit homogener Absaugung, Schr. Dtsh. Akad., Luftfahrtforschung 8B, pp. 151.
Illingworth, C.R. (1954) The Effect of Heat Transfer on the Separation of a Compressible Laminar
Boundary Layer, Quart. J. Mech. Appl. Math. 7, pp. 834.
Inger, G.R., and Siebersma, T. (1988) Computational Simulation of Vortex Generator Effects on
Transonic Shock/Boundary Layer Interaction, AIAA Paper No. 88-2590, New York.
Isomoto, K., and Honami, S. (1989) The Effect of Inlet Turbulence Intensity on the Reattachment
Process over a Backward-Facing Step, J. Fluids Eng. 111, pp. 8792.
Israelachvili, J.N. (1986) Measurement of the Viscosity of Liquids in Very Thin Films, J. Colloid
Interface Sci. 110, pp. 263271.
Israelachvili, J.N. (1991) Intermolecular and Surface Forces, second edition, Academic Press, New
York.
Itoh, N. (1987) Another Route to the Three-Dimensional Development of TollmienSchlichting
Waves with Finite Amplitude, J. Fluid Mech. 181, pp. 116.
Izakson, A. (1937) Formula for the Velocity Distribution Near a Wall, Zh. Eksper. Teor. Fiz. 7,
pp. 919924.
Jacobs, J., James, R., Ratliff, C., and Glazer, A. (1993) Turbulent Jets Induced by Surface Actua-
tors, AIAA Paper No. 93-3243, Washington, D.C.
Jacobson, S.A., and Reynolds, W.C. (1993a) Active Control of Boundary Layer Wall Shear Stress
Using Self-Learning Neural Networks, AIAA Paper No. 93-3272, AIAA, Washington, D.C.
Jacobson, S.A., and Reynolds W.C. (1993b) Active Boundary Layer Control Using Flush-Mounted
Surface Actuators, Bull. Am. Phys. Soc. 38, p. 2197.
Jacobson, S.A., and Reynolds, W.C. (1994) Active Control of Transition and Drag in Boundary
Layers, Bull. Am. Phy. Soc. 39, p. 1894.
Jacobson, S.A., and Reynolds, W.C. (1995) An Experimental Investigation Towards the Ac-
tive Control of Turbulent Boundary Layers, Department of Mechanical Engineering Report
No. TF-64, Stanford University, Stanford, California.
Jacobson, S.A., and Reynolds, W.C. (1998) Active Control of Streamwise Vortices and Streaks in
Jaffe, N.A., Okamura, T.T., and Smith, A.M.O. (1970) Determination of Spatial Amplication
Factors and Their Application to Predicting Transition, AIAA J. 8, pp. 301308.
James, R.D., Jacobs, J.W., and Glezer, A. (1994) Experimental Investigation of a Turbulent Jet
Produced by an Oscillating Surface Actuator, Appl. Mech. Rev. 47, no. 6, part 2, pp. S127
S1131.
Jang, J.-S.R., Sun, C.-T., and Mizutani, E. (1997) Neuro-Fuzzy and Soft Computing, PrenticeHall,
Upper Saddle River, New Jersey.
Jeffery, E. (1940) Nothing Left to Invent, J. Patent Ofce Society 22, pp. 479481.
Jeffery, G.B. (1920) Plane Stress and Plane Strain in Bipolar Co-ordinates, Phil. Trans. R. Soc.
Ser. A 221, pp. 265289.
Johansen, J.B., and Smith, C.R. (1986) The Effects of Cylindrical Surface Modications on
Turbulent Boundary Layers, AIAA J. 24, pp. 10811087.
386 BIBLIOGRAPHY
Johansson, A.V., and Alfredsson, P.H. (1982) On the Structure of Turbulent Channel Flow,
J. Fluid Mech. 122, pp. 295314.
Johansson, A.V., and Alfredsson, P.H. (1983) Effects of Imperfect Spatial Resolution on Measure-
ments of Wall-Bounded Turbulent Shear Flows, J. Fluid Mech. 137, pp. 409421.
Johansson, A.V., Her, J.-Y., and Haritonidis, J.H. (1987) On the Generation of High-Amplitude
Wall-Pressure Peaks in Turbulent Boundary Layers and Spots, J. Fluid Mech. 175, pp. 119142.
Johari, H., and McManus, K.R. (1997) Visualization of Pulsed Vortex Generator Jets for Active
Control of Boundary Layer Separation, AIAA Paper No. 97-2021, Washington, D.C.
Johnson, R.P. (1980) Review of Compliant Coating Research of M.O. Kramer, RPJ Associates
Technical Report No. RPJA-TR-0955-001, Palos Verdes Peninsula, California.
Johnson, W.S., Tennant, J.S., and Stamps, R.E. (1975) Leading-Edge Rotating Cylinder for Bound-
ary Layer Control on Lifting Surfaces, J. Hydronaut 9, pp. 7678.
Johnston, J., and Nishi, M. (1989) Vortex Generator JetsAMeans for Passive and Active Control
of Boundary Layer Separation, AIAA Paper No. 89-0564, New York.
Johnston, J., and Nishi, M. (1990) Vortex Generator JetsMeans for Flow Separation Control,
AIAA J. 28, pp. 989994.
Jones, B.M. (1934) Stalling, J. R. Aero. Soc. 38, pp. 753770.
Joslin, R.D., Erlebacher, G., Hussaini, M.Y. (1996) Active Control of Instabilities in Laminar
Boundary LayersOverview and Concept Validation, J. Fluids Eng. 118, pp. 494497.
Joslin, R.D. (1998) Aircraft Laminar Flow Control, Ann. Rev. Fluid Mech. 30, pp. 129.
Joslin, R.D., and Morris, P.J. (1992) Effect of Compliant Walls on Secondary Instabilities in
Boundary-Layer Transition, AIAA J. 30, pp. 332339.
Jung, W.J., Mangiavacchi, N., and Akhavan, R. (1992) Suppression of Turbulence in Wall Bounded
Flows by High-Frequency Spanwise Oscillations, Phys. Fluids A 8, pp. 16051607.
Juvet, P., and Reynolds, W.C. (1989) Entrainment Control in an Acoustically Controlled Shrouded
Jet, AIAA Paper No. 89-0969, New York.
Kachanov, Y.S., Koslov, V.V., and Levchenko, Ya.V. (1974) Experimental Study of the Inuence of
Cooling on the Stability of Laminar Boundary Layers, Izvestia Sibirskogo Otdielenia Ak. Nauk
SSSR, Seria Technicheskikh Nauk, Novosibirsk, no. 8-2, pp. 7579.
Kailasnath, P. (1993) Reynolds Number Effects and the Momentum Flux in Turbulent Boundary
Layers, Ph.D. Thesis, Yale University, New Haven, Connecticut.
Kamal, M.M. (1966) Separation in the FlowBetween Eccentric Rotating Cylinders, Trans. ASME
Ser. D 88, pp. 717724.
Kandil, O.A., and Chuang, H.A. (1988) Unsteady Vortex-Dominated Flows Around Maneuvering
Wings Over a Wide Range of Mach Numbers, AIAA Paper No. 88-0317, New York.
Kandil, O.A., and Chuang, H.A. (1990a) Unsteady NavierStokes Computations Past Oscillating
Delta Wing at High Incidence, AIAA J. 28, pp. 15651572.
Kandil, O.A., and Chuang, H.A. (1990b) Computation of Vortex-Dominated Flow for a Delta
Wing Undergoing Pitching Oscillation, AIAA J. 28, pp. 15891595.
Kannberg, L.D. (1988) The Urgency Will Return, Mechanical Engineering 110, p. 33.
Kaplan, R.E. (1964) The Stability of Laminar Incompressible Boundary Layers in the Presence
of Compliant Boundaries, Sc.D. Thesis, Massachusetts Institute of Technology, Cambridge,
Massachusetts.
Karim, M.A., and Acharya, M. (1994) Control of the Dynamic-Stall Vortex Over a Pitching Airfoil
by Leading-Edge Suction, AIAA J. 32, pp. 16471655.
Karlsson, R.I., and Johansson, T.G. (1986) LDV Measurements of Higher Order Moments of
Velocity Fluctuations in a Turbulent Boundary Layer, in Proc. Third Int. Symp. on Applications
of Laser Anemometry to Fluid Mechanics, eds. D.F.G. Dur ao et al., paper no. 12.1, July 79,
Lisbon, Portugal.
Karniadakis, G. Em (1999) Simulating Turbulence in Complex Geometries, Fluid Dyn. Res. 24,
pp. 343362.
Karniadakis, G. Em, and Orszag, S.A. (1993) Nodes, Modes and Flow Codes, Physics Today 46,
no. 3, March, pp. 3442.
Kastrinakis, E.G., and Eckelmann, H. (1983) Measurement of Streamwise Vorticity Fluctuations
in a Turbulent Channel Flow, J. Fluid Mech. 137, pp. 165186.
Katz, Y., Nishri, B., and Wygnanski, I. (1989a) The Delay of Turbulent Boundary Layer Separation
by Oscillatory Active Control, AIAA Paper No. 89-0975, New York.
BIBLIOGRAPHY 387
Katz, Y., Nishri, B., and Wygnanski, I. (1989b) The Delay of Turbulent Boundary Layer Separation
by Oscillatory Active Control, Phys. Fluids 1, pp. 179181.
Kawthar-Ali, M.H., and Acharya, M. (1996) Articial Neural Networks for Suppression of the
Dynamic-Stall Vortex Over Pitching Airfoils, AIAA Paper No. 96-0540, Washington, D.C.
Kays, W.M., and Crawford, M.E. (1993) Convective Heat and Mass Transfer, third edition,
McGrawHill, New York.
Keefe, L.R. (1993a) Two Nonlinear Control Schemes Contrasted in a Hydrodynamic Model,
Phys. Fluids A 5, pp. 931947.
Keefe, L.R. (1993b) Drag Reduction in Channel Flow Using Nonlinear Control, AIAA Paper
No. 93-3279, Washington, D.C.
Keefe, L.R. (1996) A MEMS-Based Normal Vorticity Actuator for Near-Wall Modication of
Turbulent Shear Flows, in Proc. Workshop on Flow Control: Fundamentals and Practices, eds.
J.-P. Bonnet, M. Gad-el-Hak, and A. Pollard, pp. 121, 15 July, Institut dEtudes Scientiques
des Carg` ese, Corsica, France.
Keefe, L.R., Moin, P., and Kim, J. (1992) The Dimension of Attractors Underlying Periodic Tur-
bulent Poiseuille Flow, J. Fluid Mech. 242, pp. 129.
Keith, W.L., Hurdis, D.A., and Abraham, B.M. (1992) A Comparison of Turbulent Boundary
Layer Wall-Pressure Spectra, J. Fluids Eng. 114, pp. 338347.
Kendall, J.M. (1970) The Turbulent Boundary Layer over a Wall with Progressive Surface Waves,
Kennard, E.H. (1938) Kinetic Theory of Gases, McGrawHill, New York.
Kenteld, J.A.C. (1985a) Drag Reduction of Controlled Separated Flows, AIAA Paper No. 85-
1800, New York.
Kenteld, J.A.C. (1985b) Short, Multi-Step, Afterbody Fairings, J. Aircraft 21, pp. 351352.
Kidd, J.A., Wikoff, D., and Cottrell, C.J. (1990) Drag Reduction by Controlling Flow Separation
Using Stepped Afterbodies, J. Aircraft 27, pp. 564566.
Kim, H.T., Kline, S.J., and Reynolds, W.C. (1971) The Production of Turbulence Near a Smooth
Wall in a Turbulent Boundary Layer, J. Fluid Mech. 50, pp. 133160.
Kim, J., and Moin, P. (1986) The Structure of the Vorticity Field in Turbulent Channel Flow. Part 2.
Study of Ensemble-Averaged Fields, J. Fluid Mech. 162, pp. 339363.
Kim, J., Moin, P., and Moser, R.D. (1987) Turbulence Statistics in Fully-Developed Channel Flow
at Low Reynolds Number, J. Fluid Mech. 177, pp. 133166.
Kim, J.H., and Stringer, J., editors (1992) Applied Chaos, Wiley, New York.
Kind, R.J. (1967) A Proposed Method of Circulation Control, Ph.D. Dissertation, Cambridge
University, Cambridge, Great Britain.
Kirsch, J.W. (1974) Drag Reduction of Trucks with S
3
Air Vane, in Proc. Second Symp. on
Aerodynamics of Sports and Competition Automobiles, pp. 251263, AIAA, New York.
Klebanoff, P.S. (1954) Characteristics of Turbulence in a Boundary Layer with Zero Pressure
Gradient, NACA Report No. R-1247, Washington, D.C.
Klebanoff, P.S., and Diehl, Z.W. (1952) Some Features of Articially Thickened Fully Developed
Turbulent Boundary Layers with Zero Pressure Gradient, NACAReport No. 1110, Washington,
D.C.
Klebanoff, P.S., Schubauer, G.B., and Tidstrom, K.D. (1955) Measurements of the Effect of
Two-Dimensional and Three-Dimensional Roughness Elements on Boundary-Layer Transition,
J. Aero. Sci. 22, pp. 803804.
Klebanoff, P.S., Tidstrom, K.D., and Sargent, L.M. (1962) The Three-Dimensional Nature of
Boundary Layer Instability, J. Fluid Mech. 12, pp. 134.
Klewicki, J.C., Murray, J.A., and Falco, R.E. (1994) Vortical Motion Contributions to Stress
Transport in Turbulent Boundary Layers, Phys. Fluids 6, pp. 277286.
Kline, S.J. (1967) Observed Structure Features in Turbulent and Transitional Boundary Layers,
in Fluid Mechanics of Internal Flow, ed. G. Sovran, pp. 2779, Elsevier, Amsterdam, The Nether-
lands.
Kline, S.J., and Afgan, N.H. (1990) Near-Wall Turbulence: 1988 Zoran Zari c Memorial Conference,
Hemisphere, New York.
Kline, S.J., and Runstadler, P.W. (1959) Some Preliminary Results of Visual Studies of the Flow
Model of the Wall Layers of the Turbulent Boundary Layer, J. Appl. Mech. 26, pp. 166
170.
388 BIBLIOGRAPHY
Kline, S.J., Reynolds, W.C., Schraub, F.A., and Runstadler, P.W. (1967) The Structure of Turbulent
Knight, J. (1999a) Cunning Plumbing, New Scientist 161, no. 2172, 6 February, pp. 3237.
Knight, J. (1999b) Dust Mites Dilemma, New Scientist 162, no. 2180, 29 May, pp. 4043.
Knudsen, M. (1909) Die Gesetze der Molekularstr omung und der inneren Reibungsstr omung der
Gase durch R ohren, Annalen der Physik 28, pp. 75130.
Koga, D.J., Reisenthel, P., and Nagib, H.M. (1984) Control of Separated Flowelds Using Forced
Unsteadiness, Fluids & Heat Transfer Report No. R84-1, Illinois Institute of Technology,
Chicago, Illinois.
Kogan, M.N. (1969) Rareed Gas Dynamics, Nauka, Moscow. Translated from Russian, ed. L.
Trilling, Plenum, New York.
Kogan, M.N. (1973) Molecular Gas Dynamics, Ann. Rev. Fluid Mech. 5, pp. 383404.
Kolmogorov, A.N. (1941a) The Local Structure of Turbulence in Incompressible Viscous Fluid for
Very Large Reynolds Number, Dokl. Akad. Nauk SSSR 30, pp. 301305. (Reprinted in Proc.
R. Soc. London A 434, pp. 913, 1991.)
Kolmogorov, A.N. (1941b) On Degeneration of Isotropic Turbulence in an Incompressible Viscous
Liquid, Dokl. Akad. Nauk SSSR 31, pp. 538540.
Kolmogorov, A.N. (1941c) Dissipation of Energy in Locally Isotropic Turbulence, Dokl. Akad.
Nauk SSSR 32, pp. 1618. (Reprinted in Proc. R. Soc. London A 434, pp. 1517, 1991.)
Kolmogorov, A.N. (1962) A Renement of Previous Hypothesis Concerning the Local Structure
of Turbulence in a Viscous Incompressible Fluid at High Reynolds Number, J. Fluid Mech. 13,
pp. 8285.
Koplik, J., and Banavar, J.R. (1995) Continuum Deductions from Molecular Hydroynamics,
Korkan, K.D., Cross, E.J., and Cornell, C.C. (1986) Experimental Aerodynamic Characteristics
of an NACA 0012 Airfoil with Simulated Ice, J. Aircraft 22, pp. 130134.
Kosecoff, M.A., Ko, D.R.S., and Merkle, C.L. (1976) An Analytical Study of the Effect of Surface
Roughness on the Stability of a Heated Water Boundary Layer, Final Report PDT 76-131,
Dynamics Technology, Inc., Torrance, California.
Koshland, D.E., Jr. (1995) The Crystal Ball and the Trumpet Call, Science 267, 17 March,
p. 1575.
Koskie, J.E., and Tiederman, W.G. (1991) Turbulence Structure and Polymer Drag Reduction
in Adverse Pressure Gradient Boundary Layers, School of Mechanical Engineering Report
No. PME-FM-91-3, Purdue University, West Lafayette, Indiana.
Kostelich, E.J., Grebogi, C., Ott, E., and Yorke, J.A. (1993a) Targeting from Time Series, Bul.
Am. Phys. Soc. 38, p. 2194.
Kostelich, E.J., Grebogi, C., Ott, E., and Yorke, J.A. (1993b) Higher-Dimensional Targeting,
Phys. Rev. E 47, pp. 305310.
Kovalnogov, S.A., Fomin, V.M., and Shapovalov, G.K. (1987) Experimental Study of the Possi-
bility of Passive Control of Shock-Boundary Layer Interactions, Uchemye kZapiski TSAGI 18,
pp. 112116.
Kovasznay, L.S.G. (1970) The Turbulent Boundary Layer, Ann. Rev. Fluid Mech. 2, pp. 95112.
Kovasznay, L.S.G., Kibens, V., and Blackwelder, R.F. (1970) Large-Scale Motion in the Intermittent
Region of a Turbulent Boundary Layer, J. Fluid Mech. 41, pp. 283325.
Krall, K.M., and Haight, C.H. (1972) Wind Tunnel Tests of a Trapped Vortex-High Lift Airfoil,
Advanced Technology Center Report No. ATC-B-94300/3TR-10, Dallas, Texas. (Also available
from U.S. NTIS; Document Number AD-762 077.)
Kramer, M.O. (1957) Boundary-Layer Stabilization by Distributed Damping, J. Aeronaut. Sci.
24, pp. 459460.
Kramer, M.O. (1960a) Boundary-Layer Stabilization by Distributed Damping, J. Aero/Space Sci.
27, p. 69.
Kramer, M.O. (1960b) Boundary Layer Stabilization by Distributing Damping, J. Am. Soc. Nav.
Engrs. 72, pp. 2533.
Kramer, M.O. (1960c) The Dolphins Secret, New Scientist 7, 5 May, pp. 11181120.
Kramer, M.O. (1961) The Dolphins Secret, J. Am. Soc. Nav. Engrs. 73, pp. 103107.
Kramer, M.O. (1962) Boundary Layer Stabilization by Distributed Damping, J. Am. Soc. Nav.
Engrs. 74, pp. 341348.
BIBLIOGRAPHY 389
Kramer, M.O. (1965) Hydrodynamics of the Dolphin, in Advances in Hydroscience, ed. V.T.
Chow, vol. 2, pp. 111130, Academic Press, New York.
Kramer, M.O. (1969) Die Widerstandsverminderung schneller Unterwasserk orper Mittels
k unstlicher Delphinhaut, in Jahrbuch 1969 der Deutschen Gesellschaft f ur Luft- und Raum-
fahrt, eds. H. Blenk and W. Schulz, pp. 19, Cologne, Federal Republic of Germany.
Kravchenko, A.G., Choi, H., and Moin, P. (1993) On the Relation of Near-Wall Streamwise
Vortices to Wall Skin Friction in Turbulent Boundary Layers, Phys. Fluids A 5, pp. 33073309.
Kreplin, H.-P., and Eckelmann, H. (1979) Behaviour of the Three Fluctuating Velocity Components
in the Wall Region of a Turbulent Channel Flow, Phys. Fluids 22, pp. 12331239.
Kudva, A.K., and Sesonske, A. (1972) Structure of Turbulent Velocity and Temperature Fields
in Ethylene Glycol Pipe Flow at Low Reynolds Number, Int. J. Heat Mass Transfer 15,
pp. 127145.
Kuethe, A.M. (1973) Boundary Layer Control of Flow Separation and Heat Exchange, U.S.
Patent No. 3,741,285.
Kukainis, J. (1969) Effects of Three-Dimensional Boundary Layer Control Devices on a Quasi-
Two-Dimensional Swept Wing at High Subsonic Speeds, Arnold Engineering Development
Center Report No. AEDC-TR-69-251, Arnold Air Forc Base, Tennessee.
Kundu, P.K. (1990) Fluid Mechanics, Academic Press, New York.
Kutateladze, S.S., and Khabakhpasheva, E.M. (1978) Structure of Wall Boundary Layer (Forced
Flow, Thermal Convection), Institute of Thermodynamics, Siberian Branch of the USSR
Academy of Science, Novosibirsk, USSR.
Kwong, A., and Dowling, A. (1993) Active Boundary Layer Control in Diffusers, AIAA Paper
No. 93-3255, Washington, D.C.
Laadhari, F., Skandaji, L., and Morel, R. (1994) Turbulence Reduction in a Boundary Layer by a
Local Spanwise Oscillating Surface, Phys. Fluids 6, pp. 32183220.
Lachmann, G.V., editor (1961) Boundary Layer and Flow Control, vols. 1 and 2, Pergamon Press,
Ladd, D.M., and Hendricks, E.W. (1988) Active Control of 2-D Instability Waves on an Axisym-
metric Body, Exp. Fluids 6, pp. 6970.
Lai, Y.-C., and Grebogi, C. (1993) Synchronization of Chaotic Trajectories Using Control, Phys.
Rev. E 47, pp. 23572360.
Lai, Y.-C., Deng, M., and Grebogi, C. (1993) Controlling Hamiltonian Chaos, Phys. Rev. E 47,
pp. 8692.
Lai, Y.-C., Grebogi, C., and T el, T. (1994) Controlling Transient Chaos in Dynamical Systems,
in Towards the Harnessing of Chaos, ed. M. Yamaguchi, Elsevier, Amsterdam, The Netherlands.
Lai, Y.-C., T el, T., and Grebogi, C. (1993) Stabilizing Chaotic-Scattering Trajectories Using Con-
trol, Phys. Rev. E 48, pp. 709717.
Lakshmanan, B., Tiwari, S.N., and Hussaini, M.Y. (1988) Control of Supersonic Interaction Flow
Fields Through Filleting and Sweep, AIAA Paper No. 88-3534, New York.
Lamb, H. (1895) Hydrodynamics, second edition, Cambridge University Press, Cambridge, Great
Britain.
Landahl, M.T. (1962) On the Stability of a Laminar Incompressible Boundary Layer Over a Flexible
Surface, J. Fluid Mech. 13, pp. 609632.
Landahl, M.T. (1967) A Wave-Guide Model for Turbulent Shear Flow, J. Fluid Mech. 29,
pp. 441459.
Landahl, M.T. (1972) Wave Mechanics of Breakdown, J. Fluid Mech. 56, pp. 775802.
Landahl, M.T. (1973) Drag Reduction by Polymer Addition, in Proc. Thirteenth IUTAM
Congress, eds. E. Becker and G.K. Mikhailov, pp. 177199, SpringerVerlag, Berlin.
Landahl, M.T. (1977) Dynamics of Boundary Layer Turbulence and the Mechanism of Drag
Reduction, Phys. Fluids 20, no. 10, part II, pp. S55S63.
Landahl, M.T. (1980) A Note on an Algebraic Instability of Inviscid Parallel Shear Flows, J. Fluid
Mech. 98, pp. 243251.
Landahl, M.T. (1990) On Sublayer Streaks, J. Fluid Mech. 212, pp. 593614.
Landau, L.D., and Lifshitz, E.M. (1987) Fluid Mechanics, second edition, Pergamon Press, Oxford,
Great Britain.
Landweber, L. (1953) The Frictional Resistance of Flat Plates at Zero Pressure Gradient, Trans.
Soc. Nav. Arch. Mar. Eng. 61, pp. 532.
390 BIBLIOGRAPHY
Landweber, L., and Siao, T.T. (1958) Comparison of Two Analyses of Boundary-Layer Data on a
Flat Plate, J. Ship Research 1, pp. 2133.
Lang, T.G. (1963) Porpoise, Whales, and Fish: Comparison of Predicted and Observed Speeds,
Nav. Eng. J. 75, pp. 437441.
Lange, R.H. (1954) Present Status of Information Relative to the Prediction of Shock-Induced
Boundary Layer Separation, NACA TN-3065, Washington, D.C.
Lankford, J.L. (1960) Investigation of the Flow over an Axisymmetric Compression Surface at
High Mach Numbers, Report No. 6866, U.S. Naval Ordnance Laboratory, Corona, California.
Lankford, J.L. (1961) The Effect of Heat Transfer on the Separation of Laminar Flow Over Ax-
isymmetric Compression Surfaces. Preliminary Results at Mach No. 6.78, Report No. 7402,
U.S. Naval Ordnance Laboratory, Corona, California.
Lauchle, G.C., and Daniels, M.A. (1987) Wall-Pressure Fluctuations in Turbulent Pipe Flow,
Phys. Fluids 30, pp. 30193024.
Lauchle, G.C., and Gurney, G.B. (1984) Laminar Boundary-Layer Transition on a Heated Under-
water Body, J. Fluid Mech. 144, pp. 79101.
Laufer, J. (1951) Investigation of Turbulent Flow in a Two-Dimensional Channel, NACA Report
No. R-1053, Washington, D.C.
Laufer, J. (1954) The Structure of Turbulence in Fully Developed Pipe Flow, NACA Report
No. R-1174, Washington, D.C.
Laufer, J. (1975) New Trends in Experimental Turbulence Research, Ann. Rev. Fluid Mech. 7,
pp. 307326.
Launder, B.E., and Spalding, D.B. (1974) The Numerical Computation of Turbulent Flows,
Comput. Methods Appl. Mech. Eng. 3, pp. 269289.
Lee, C.S., Tavella, D.A., Wood, N.J., and Roberts, L. (1986) Flow Structure of Lateral Wing-Tip
Blowing, AIAA Paper No. 86-1810, New York.
Lee, C.S., Tavella, D.A., Wood, N.J., and Roberts, L. (1989) Flow Structure and Scaling Laws in
Lateral Wing-Tip Blowing, AIAA J. 27, pp. 10021007.
Lee, D.G. (1974) Subsonic Force Characteristics of a Low Aspect Ratio Wing Incorporating a
Spinning Cylinder, DTNSRDC Report No. ASED-329, Bethesda, Maryland. (Also available
from U.S. NTIS; Document Number AD-AOO11135.)
Lee, M., and Ho, C.-M. (1989) Vortex Dynamics of Delta Wings, in Frontiers in Experimental
Fluid Mechanics, ed. M. Gad-el-Hak, pp. 365428, SpringerVerlag, New York.
Lee, M., and Ho, C.-M. (1990) Lift Force of Delta Wings, Appl. Mech. Rev. 43, pp. 209221.
Lee, M., and Reynolds, W.C. (1985) Bifurcating and Blooming Jets, Thermosciences Divi-
sion Report No. TR-22, Department of Mechanical Engineering, Stanford University, Stanford,
California.
Lee, R.E., Yanta, W.J., and Leonas, A.C. (1969) Velocity Prole, Skin Friction Balance and Heat
Transfer Measurements of the Turbulent Boundary Layer at M 5 and Zero Pressure Gradient,
National Ordnance Laboratory Report No. NOL-TR69-106, White Oak, Maryland.
Lee, T., Fisher, M., and Schwarz, W.H. (1993a) Investigation of the Stable Interaction of a Passive
Compliant Surface with a Turbulent Boundary Layer, J. Fluid Mech. 257, pp. 373401.
Lee, T., Fisher, M., and Schwarz, W.H. (1993b) The Measurement of Flow-Induced Surface Dis-
placement on a Compliant Surface by Optical Holographic Interferometry, Exp. Fluids 14,
pp. 159168.
Lee, T., Fisher, M., and Schwarz, W.H. (1995) Investigation of the Effects of a Compliant Surface
on Boundary-Layer Stability, J. Fluid Mech. 288, pp. 3758.
Lee, T., Fisher, M., and Schwarz, W.H. (1997) An Experimental Study of the Boundary-Layer
Nonlinear Instability Over Compliant Walls, unpublished manuscript originally submitted to
J. Fluid Mech.
Lee, W.K., Vaseleski, R.C., and Metzner, A.B. (1974) Turbulent Drag Reduction in Polymeric
Solutions Containing Suspended Fibers, AIChE J. 20, pp. 128133.
Lees, L. (1947) The Stability of the Laminar Boundary Layer in a Compressible Fluid, NACA
Report No. 876, Washington, D.C.
Legner, H.H. (1984) A Simple Model for Gas Bubble Drag Reduction, Phys. Fluids 27,
pp. 27882790.
Lerner, E.J. (1999) Next-Genration Lithography, Industrial Physicist 5, no. 3, June, pp. 18
21.
BIBLIOGRAPHY 391
Lewkowicz, A.K. (1982) An Improved Universal Wake Function for Turbulent Boundary Layers
and Some of Its Consequences, Z. Flugwiss. Weltraumforsch. 6, pp. 261266.
Liandrat, J., Aupoix, B., and Cousteix, T. (1986) Calculation of Longitudinal Vortices Embedded
in a Turbulent Boundary Layer, in Fifth Symposium on Turbulent Shear Flows, eds. F. Durst,
B.E. Launder, F.W. Schmidt, and J.H. Whitelaw, pp. 7.177.22, SpringerVerlag, New York.
Liang, W.J., and Liou, J.A. (1995) Flow Around a Rotating Cylinder Near a Plane Boundary,
J. Chinese Institute of Engineers 18, pp. 3550.
Libby, P.A. (1954) Method for Calculation of Compressible Laminar Boundary Layer with Axial
Pressure Gradient and Heat Transfer, NACA TN No. 3157, Washington, D.C.
Liebeck, R.H. (1978) Design of Subsonic Airfoils for High Lift, J. Aircraft 15, pp. 547561.
Liebeck, R.H., and Camacho, P.P. (1985) Airfoil Design at Low Reynolds Number with Con-
strained Pitching Moment, in Proc. Conf. on Low Reynolds Number Airfoil Aerodynamics,
ed. T.J. Mueller, pp. 2751, University of Notre Dame, Notre Dame, Indiana.
Liepmann, H.W. (1952) Aspects of the Turbulence Problem. Part II, Z. Angew. Math. Phys. 3,
pp. 407426.
Liepmann, H.W. (1962) Free Turbulent Flows, M ecanique de la Turbulence, Int. Symp. Nat. Sci.
Res. Centre, pp. 211227, Marseille 1961, CNRS, Paris, France.
Liepmann, H.W. (1979) The Rise and Fall of Ideas in Turbulence, American Scientist 67, no. 2,
pp. 221228.
Liepmann, H.W. (1997) Boundary Layer Transition: The Early Days, Appl. Mech. Rev. 50, no.
2, pp. R1R4.
Liepmann, H.W., and Fila, G.H. (1947) Investigations of Effects of Surface Temperature and
Single Roughness Elements on Boundary Layer Transition, NACAReport No. 890, Washington,
D.C.
Liepmann, H.W., Brown, G.L., and Nosenchuck, D.M. (1982) Control of Laminar Instability
Waves Using a New Technique, J. Fluid Mech. 118, pp. 187200.
Liepmann, H.W., and Nosenchuck, D.M. (1982) Active Control of LaminarTurbulent Transi-
tion, J. Fluid Mech. 118, pp. 201204.
Lighthill, M.J. (1952) On Sound Generated AerodynamicallyI. General Theory, Proc. Roy. Soc.
London A 211, pp. 564587.
Lighthill, M.J. (1954) On Sound Generated AerodynamicallyII. Turbulence as a Source of
Sound, Proc. Roy. Soc. London A 222, pp. 132.
Lighthill, M.J. (1962) The Bekerian Lecture, 1961Sound Generated Aerodynamically, Proc.
Roy. Soc. London A 267, pp. 147182.
Lighthill, M. J. (1963a) Introduction. Boundary Layer Theory, in Laminar Boundary Layers,
ed. L. Rosenhead, pp. 46113, Clarendon Press, Oxford, Great Britain.
Lighthill, M.J. (1963b) Introduction. Real and Ideal Fluids, in Laminar Boundary Layers, ed. L.
Rosenhead, pp. 145, Clarendon Press, Oxford, Great Britain.
Lighthill, M.J. (1973) On the WeisFogh Mechanism of Lift Generation, J. Fluid Mech. 60,
pp. 117.
Lighthill, M.J. (1975) Aerodynamic Aspects of Animal Flight, in Swimming and Flying in Nature,
eds. T.Y. Wu, C.J. Brokaw, and C. Brennen, vol. 2, pp. 423491, Plenum, New York.
Ligrani, P.M., and Bradshaw, P. (1987) Spatial Resolution and Measurements of Turbulence in the
Viscous Sublayer Using Subminiature Hot-Wire Probes, Exp. Fluids 5, pp. 407417.
Lin, C.C. (l945) On the Stability of Two-Dimensional Parallel Flows, Parts I, II and III, Q. Appl.
Maths. 3, pp. 117142, 218234, 277301.
Lin, J.C., and Ash, R.L. (1986) Wall Temperature Control of Low-Speed Body Drag, J. Aircraft
23, pp. 9394.
Lin, J.C., and Howard, F.G. (1989) Turbulent Flow Separation Control Through Passive Tech-
niques, AIAA Paper No. 89-0976, New York.
Lin, J.C., Howard, F.G., Bushnell, D.M., and Selby, G.V. (1990) Investigation of Several Passive
and Active Methods for Turbulent Flow Separation Control, AIAA Paper No. 90-1598, New
York.
Lin, J.C., Howard, F.G., and Selby, G.V. (1990) Control of Turbulent Separated Flow over a
Rearward-Facing Ramp Using Longitudinal Grooves, J. Aircraft 27, pp. 283285.
Lin, J.C., Robinson, S.K., McGhee, R.J., and Valarezo, W.O. (1992) Separation Control on High
Reynolds Number Multi-Element Airfoils, AIAA Paper No. 92-2636, Washington, D.C.
392 BIBLIOGRAPHY
Lin, J.C., Weinstein, L.M., Watson, R.D., and Balasubramanian, R. (1983) Turbulent Drag Char-
acteristic of Small Amplitude Rigid Surface Waves, AIAA Paper No. 83-0228, New York.
Lindner, J.F., and Ditto, W.L. (1995) Removal, Suppression and Control of Chaos by Nonlinear
Design, Appl. Mech. Rev. 48, pp. 795808.
Linke, W. (1942)

Uber den Str omungswiderstand einer beheizten ebenen Platte, Luftfahrt-
forschung 19, pp. 157160.
Lipkin, R. (1993) Micro Steam Engine Makes Forceful Debut, Science News 144, September,
p. 197.
Lissaman, P.B.S. (1983) Low-Reynolds-Number Airfoils, Ann. Rev. Fluid Mech. 15, pp. 223239.
Lissaman, P.B.S., and Harris, G.L. (1969) Turbulent Skin Friction on Compliant Surfaces, AIAA
Liu, C.K., Kline, S.J., and Johnston, J.P. (1966) Experimental Study of Turbulent Boundary Layer
on Rough Walls, Department of Mechanical Engineering Report No. MD-15, Stanford Univer-
sity, Stanford, California.
Liu, J., Tai, Y.C., Lee, J., Pong, K.C., Zohar, Y., and Ho, C.M. (1993) In-Situ Monitoring and
Universal Modeling of Sacricial PSG Etching Using Hydrouoric Acid, in Proc. IEEE Micro
Electro Mechanical Systems 93, pp. 7176, IEEE, New York.
Liu, J., Tai, Y.C., Pong, K., and Ho, C.M. (1995) MEMS for Pressure Distribution Studies of
Gaseous Flows in Microchannels, in Proc. IEEEMicro Electro Mechanical Systems 95, pp. 209
215, IEEE, New York.
Liu, J.T.C. (1988) Contributions to the Understanding of Large-Scale Coherent Structures in De-
veloping Free Turbulent Shear Flows, in Advances in Applied Mechanics, eds. J.W. Hutchinson
and T.Y. Wu, vol. 26, pp. 183309, Academic Press, Boston, Massachusetts.
Loeb, L.B. (1961) The Kinetic Theory of Gases, third edition, Dover, New York.
L ofdahl, L., and Gad-el-Hak, M. (1999) MEMS Applications in Turbulence and Flow Control,
L ofdahl, L., Glavmo, M., Johansson, B., and Stemme, G. (1993) A Silicon Transducer for the
Determination of Wall-Pressure Fluctuations in Turbulent Boundary Layers, Appl. Scientic Res.
51, pp. 203207.
L ofdahl, L., K alvesten, E., and Stemme, G. (1994) Small Silicon Based Pressure Transducers for
Measurements in Turbulent Boundary Layers, Exp. Fluids 17, pp. 2431.
L ofdahl, L., K alvesten, E., and Stemme, G. (1996) Small Silicon Pressure Transducers for Space-
Time Correlation Measurements in a Flat Plate Boundary Layer, J. Fluids Eng. 118, pp. 457
463.
L ofdahl, L., Stemme, G., and Johansson, B. (1989) A Sensor Based on Silicon Technology for
Turbulence Measurements, J. Phys. E. Sci. Instum. 22, pp. 391393.
L ofdahl, L., Stemme, G., and Johansson, B. (1991) Reynolds Stress Measurements Using Direction
Sensitive Double-Chip Silicon Sensors, Meas. Sci. Technol. 2, pp. 369373.
L ofdahl, L., Stemme, G., and Johansson, B. (1992) Silicon Based Flow Sensors Used for Mean
Velocity and Turbulence Measurements, Exp. Fluids 12, pp. 270276.
Long, R.R., and Chen, T.-C. (1981) Experimental Evidence for the Existence of the Mesolayer
in Turbulent Systems, J. Fluid Mech. 105, pp. 1959.
Looney, R.W., and Blick, E.F. (1966) Skin Friction Coefcients of Compliant Surfaces in Turbulent
Flow, J. Spacecraft & Rockets 3, pp. 15621564.
Loose, W., and Hess, S. (1989) Rheology of Dense Fluids via Nonequilibrium Molecular Hydro-
dynamics: Shear Thinning and Ordering Transition, Rheologica Acta 28, pp. 91101.
Lowell, R.L., and Reshotko, E. (1974) Numerical Study of the Stability of a Heated Water Bound-
ary Layer, Case Western University, Report No. FTAS/TR-73-93, Cleveland, Ohio.
Lucey, A.D., and Carpenter, P.W. (1993) The Hydroelastic Stability of Three-Dimensional Distur-
bances of a Finite Compliant Panel, J. Sound Vibr. 165, pp. 527552.
Lucey, A.D., and Carpenter, P.W. (1995) Boundary Layer Instability over Compliant Walls: Com-
parison between Theory and Experiment, Phys. Fluids 7, pp. 23552363.
Luchik, T.S., and Tiederman, W.G. (1986) Effect of Spanwise Probe Volume Length on Laser
Velocimeter Measurments in Wall Bounded Turbulent Flows, Exp. Fluids 3, pp. 339341.
Ludwig, G.R. (1964) An Experimental Investigation of Laminar Separation from a Moving Wall,
Lumley, J.L. (1969) Drag Reduction by Additives, Ann. Rev. Fluid Mech. 1, pp. 367384.
BIBLIOGRAPHY 393
Lumley, J.L. (1973) Drag Reduction in Turbulent Flow by Polymer Additives, J. Polym. Sci.:
Macromol. Rev. 7, pp. 263290.
Lumley, J.L. (1977) Drag Reduction in Two Phase and Polymer Flows, Phys. Fluids 20, no. 10,
part II, pp. S64S71.
Lumley, J.L. (1978a) Two-Phase and Non-Newtonian Flows, in Topics in Applied Physics,
vol. 12, ed. P. Bradshaw, second edition, pp. 289324, SpringerVerlag, Berlin.
Lumley, J.L. (1978b) Computational Modeling of Turbulent Flows, Adv. Appl. Mech. 18,
pp. 123176.
Lumley J.L. (1981) Coherent Structures in Turbulence, in Transition and Turbulence, ed. R.E.
Meyer, pp. 215242, Academic Press, New York.
Lumley, J.L. (1983) Turbulence Modeling, J. Appl. Mech. 50, pp. 10971103.
Lumley, J.L. (1987) Turbulence Modeling, in Proc. Tenth U.S. National Cong. of Applied Me-
chanics, ed. J.P. Lamb, pp. 3339, ASME, New York.
Lumley, J.L. (1991a) Order and Disorder in Turbulent Flows, in New Perspectives in Turbulence,
ed. L. Sirovich, pp. 105122, SpringerVerlag, Berlin.
Lumley, J.L. (1991b) Control of the Wall Region of a Turbulent Boundary Layer, in Turbu-
lence: Structure and Control, ed. J.M. McMichael, pp. 6162, 13 April, Ohio State University,
Columbus, Ohio.
Lumley, J.L. (1992) Some Comments on Turbulence, Phys. Fluids A 4, pp. 203211.
Lumley, J.L. (1996a) Turbulence and Turbulence Modeling, in Research Trends in Fluid Dynam-
ics, eds. J.L. Lumley, A. Acrivos, L.G. Leal, and S. Leibovich, pp. 167177, American Institute of
Physics, Woodbury, New York.
Lumley, J.L. (1996b) Control of Turbulence, AIAA Paper No. 96-0001, Washington, D.C.
Lumley, J.L., Acrivos, A., Leal, L.G., and Leibovich, S. (1996) Research Trends in Fluid Dynamics,
American Institute of Physics, Woodbury, New York.
Lumley, J., and Blossey, P. (1998) Control of Turbulence, Ann. Rev. Fluid Mech. 30, pp. 311327.
Lumley, J.L., and Kubo, I. (1984) Turbulent Drag Reduction by Polymer Additives: ASurvey, Sib-
ley School of Mechanical and Aerospace Engineering Report No. FDA-84-07, Cornell University,
Ithaca, New York.
L uscher, E., and H ubler, A. (1989) Resonant Stimulation of Complex Systems, Helv. Phys. Acta
62, pp. 544551.
Luttges, M.W. (1989) Accomplished Insect Fliers, in Frontiers in Experimental Fluid Mechanics,
ed. M. Gad-el-Hak, pp. 429456, SpringerVerlag, New York.
Luttges, M.W., Somps, C., Kliss, M., and Robinson, M. (1984) Unsteady Separated Flows: Gen-
eration and Use by Insects, in Unsteady Separated Flows, eds. M.S. Francis and M.W. Luttges,
pp. 1271136, U.S. Air Force Academy, Colorado Springs, Colorado.
Mabey, D.G. (1979) Inuence of the Wake Component on Turbulent Skin Friction at Subsonic
and Supersonic Speeds, Aeronaut. Quart. 30, pp. 590606.
Mabey, D.G. (1988) Design Features Which Inuence Flow Separations on Aircraft, Aero. J. 92,
pp. 409415.
Mabey, D.G., Meier, H.U., and Sawyer, W.G. (1976) Some Boundary Layer Measurements on
a Flat Plate at Mach Numbers from 2.5 to 4.5, Royal Aeronautical Establishment Technical
Report No. RAE-74127, London, Great Britain.
MacCurdy, E. (1938) The Notebooks of Leonardo da Vinci, vols. I and II, Reynal & Hitchcock,
New York.
Macha, J.M., Norton, D.J., and Young, J.C. (1972) Surface Temperature Effect on Subsonic Stall,
Madavan, N.K., Deutsch, S., and Merkle, C.L. (1984) Reduction of Turbulent Skin Friction by
Microbubbles, Phys. Fluids 27, pp. 356363.
Madavan, N.K., Deutsch, S., and Merkle, C.L. (1985) Measurements of Local Skin Friction in a
Microbubble-Modied Turbulent Boundary Layer, J. Fluid Mech. 156, pp. 237256.
Maestrello, L., Badavi, F.F., and Noonan, K.W. (1988) Control of the Boundary Layer Separation
About an Airfoil by Active Surface Heating, AIAA Paper No. 88-3545-CP, New York.
Magarvey, R.H., and McLatchy, C.S. (1964) The Disintegration of Vortex Rings, Can. J. Phys.
42, pp. 684689.
Magnus, G. (1852) On the Deection of a Projectile, Abhandlung der Akademie der Wissens-
chaften, Berlin, Germany.
394 BIBLIOGRAPHY
Majumdar, A., and Mezic, I. (1998) Stability Regimes of Thin Liquid Films, Microscale Thermo-
phys. Eng. 2, pp. 203213.
Majumdar, A., and Mezic, I. (1999) Instability of Ultra-Thin Water Films and the Mechanism of
Droplet Formation on Hydrophilic Surfaces, in Proc. ASME-JSME Thermal Engineering and
Solar Energy Joint Conference, San Diego, California, 1519 March. Also to appear in J. Heat
Transfer.
Malik, M.R., Weinstein, L.M., and Hussaini, M.Y. (1983) Ion Wind Drag Reduction, AIAA
Malkus, W.V.R. (1956) Outline of a Theory of Turbulent Shear Flow, J. Fluid Mech. 1, pp. 521
539.
Malkus, W.V.R. (1979) Turbulent Velocity Proles from Stability Criteria, J. Fluid Mech. 90,
pp. 401414.
Maltby, R.L. (1962) Flow Visualization in Wind Tunnels Using Indicators, AGARDograph
No. 70, NATO Advisory Group for Aerospace Research and Development, Rhode-Saint-G en` ese,
Belgium.
Mangalam, S.M., Bar-Sever, A., Zaman, K.B.M.Q., and Harvey, W.D. (1986) Transition and
Separation Control on a Low-Reynolds Number Airfoil, in Proc. Int. Conf. on Aerodynamics
at Low Reynolds Numbers, vol. I, pp. 10.110.19, Royal Aeronautical Society, London, Great
Britain.
Mansour, N.N., Kim, J., and Moin, P. (1988) Reynolds-Stress and Dissipation-Rate Budgets in a
Turbulent Channel Flow, J. Fluid Mech. 194, pp. 1544.
Marchman, J.F., Manor, D., and Plentovich, E.B. (1980) Performance Improvement of Delta Wings
at Subsonic Speeds Due to Vortex Flaps, AIAA Paper No. 80-1802, New York.
Markoff, J. (1999) Tiniest Circuits Hold Prospect of Explosive Computer Speeds, The New York
Times CXLVIII, no. 51,585, 16 July, p. A1 and p. C17.
Maskell, E.C. (1955) Flow Separation in Three Dimensions, RAE Report Aero. 2565, Royal
Aircraft Establishment, Farnborough, Great Britain.
Mastrangelo, C. (1993) Integration, Partition, and Reliability of Microelectromechanical Sys-
tems, invited oral presentation at AIAA Third Flow Control Conference, 69 July, Orlando,
Florida.
Mastrangelo, C., and Hsu, C.H. (1992) A Simple Experimental Technique for the Measurement
of the Work of Adhesion of Microstructures, in Technical Digest IEEE Solid-State Sensors and
Actuators Workshop, pp. 208212, IEEE, New York.
Maureau, J., Sharatchandra, M.C., Sen, M., and Gad-el-Hak, M. (1997) Flow and Load Charac-
teristics of Microbearings with Slip, J. Micromech. Microeng. 7, pp. 5564.
Maxwell, J.C. (1879) On Stresses in Rareed Gases Arising from Inequalities of Temperature,
Phil. Trans. R. Soc. Part 1 170, pp. 231256.
Maxworthy, T. (1979) Experiments on the WeisFogh Mechanism of Lift Generation by Insects
in Hovering Flight. Part 1. Dynamics of the Fling, J. Fluid Mech. 93, pp. 4763.
Maxworthy, T. (1981) The Fluid Dynamics of Insect Flight, Ann. Rev. Fluid Mech. 13, pp. 329
350.
May, M. (1999) Im Just Flying Down to the Supermarket, New Scientist 162, no. 2188, 29 May,
pp. 2427.
McComb, W.D., and Chan, K.T.J. (1979) Drag Reduction in Fibre Suspensions: Transitional
Behavior Due to Fibre Degradation, Nature 280, pp. 4546.
McComb, W.D., and Chan, K.T.J. (1981) Drag Reduction in Fibre Suspension, Nature 292,
pp. 520522.
McComb, W.D., and Rabie, L.H. (1979) Development of Local Turbulent Drag Reduction Due
to Nonuniform Polymer Concentration, Phys. Fluids 22, pp. 183185.
McCormick, M.E., and Bhattacharyya, R. (1973) Drag Reduction of a Submersible Hull by Elec-
trolysis, Nav. Eng. J. 85, pp. 1116.
McCroskey, W.J. (1977) Some Current Research in Unsteady Fluid Dynamics, J. Fluids Eng. 99,
pp. 839.
McCroskey, W.J. (1982) Unsteady Airfoils, Ann. Rev. Fluid Mech. 14, pp. 285311.
McCullough, G.B. (1955) The Effect of Reynolds Number on the Stalling Characteristics and
Pressure Distributions of Four Moderately Thin Airfoil Sections, NACA TN-3524, Washington,
D.C.
BIBLIOGRAPHY 395
McCullough, G.B., and Gault, D.E. (1951) Examples of Three Representative Types of Airfoil-
Section Stall at Low Speed, NACA TN No. 2502, Washington, D.C.
McCune, J.E., and Kerrebrock, J.L. (1973) Noise fromAircraft Turbomachinery, Ann. Rev. Fluid
Mech. 5, pp. 281300.
McInville, R.M., Hassan, H.A., and Goodman, W.L. (1985) Mixing Layer Control for Tangential
Slot Injection in Turbulent Flows, AIAA Paper No. 85-0541, New York.
McLachlan, B.G. (1989) Study of a Circulation Control Airfoil with Leading/Trailing-Edge Blow-
ing, J. Aircraft 26, pp. 817821.
McLean, J.D., and Herring, H.J. (1974) Use of Multiple Discrete Wall Jets for Delaying Boundary
Layer Separation, NASA CR-2389, Washington, D.C.
McManus, K., and Magill, J. (1996) Separation Control in Incompressible and Compressible Flows
Using Pulsed Jets, AIAA Paper No. 96-1948, Washington, D.C.
McManus, K., and Magill, J. (1997) Airfoil Performance Enhancement Using Pulsed Jet Separation
Control, AIAA Paper No. 97-1971, Washington, D.C.
McManus, K., Ducharme, A., Goldey, C., and Magill, J. (1996) Pulsed Jet Actuators for Suppress-
ing Flow Separation, AIAA Paper No. 96-0442, Washington, D.C.
McManus, K.R., Legner, H.H., and Davis, S.J. (1994) Pulsed Vortex Generator Jets for Active
Control of Flow Separation, AIAA Paper No. 94-2218, Washington, D.C.
McMasters, J.H., and Henderson, M.L. (1980) LowSpeed Single Element Airfoil Synthesis, Tech.
Soaring 6, pp. 121.
McMichael, J.M. (1993) MEMS and Challenges of Flow Control, invited oral presentation at
AIAA Third Flow Control Conference, 69 July, Orlando, Florida.
McMichael, J.M. (1996) Progress and Prospects for Active Flow Control Using Microfabricated
Electromechanical Systems (MEMS), AIAA Paper No. 96-0306, Washington, D.C.
McMichael, J.M., Klebanoff, P.S., and Meese, N.E. (1980) Experimental Investigation of Drag on
a Compliant Surface, in Viscous Flow Drag Reduction, ed. G.R. Hough, pp. 410438, AIAA,
New York.
McMurray, J.T., Metcalfe, R.W., and Riley, J.J. (1983) Direct Numerical Simulations of Active
Stabilization of Boundary Layer Flows, in Proc. Eighth Biennial Symp. on Turbulence, eds. J.L.
Zakin and G.K. Patterson, paper no. 36, University of Missouri, Rolla, Missouri.
Mehregany, M. (1993) Overview of Microelectromechanical Systems, invited oral presentation
at AIAA Third Flow Control Conference, 69 July, Orlando, Florida.
Mehregany, M., DeAnna, R.G., and Reshotko, E. (1996) Microelectromechanical Systems for
Aerodynamics Applications, AIAA Paper No. 96-0421, Washington, D.C.
Mehta, R.D. (1985a) Aerodynamics of Sports Balls, Ann. Rev. Fluid Mech. 17, pp. 151
189.
Mehta, R.D. (1985b) Effect of a Longitudinal Vortex on a Separated Turbulent Boundary Layer,
Mehta, R.D. (1988) Vortex/Separated Boundary-Layer Interactions at Transonic Mach Numbers,
AIAA J. 26, pp. 1526.
Meng, J.C.S., editor (1998) Proceedings of the International Symposium on Sea Water Drag Re-
duction, 2223 July, Naval Undersea Warfare Center, Newport, Rhode Island.
Merkle, C.L., and Deutsch, S. (1985) Drag Reduction by Microbubbles: Current Research Status,
Merkle, C.L., and Deutsch, S. (1989) Microbubble Drag Reduction, in Frontiers in Experimental
Fluid Mechanics, ed. M. Gad-el-Hak, pp. 291336, SpringerVerlag, New York.
Metcalfe, R.W. (1994) Boundary Layer Control: A Brief Review, in Computational Fluid Dy-
namics 94, eds. J. Periaux and E. Hirschel, pp. 5260, Wiley, New York.
Metcalfe, R.W., Battistoni, F., Ekeroot, J., and Orszag, S.A. (1991) Evolution of Boundary Layer
Flow over a Compliant Wall During Transition to Turbulence, in Proc. Boundary Layer Tran-
sition and Control Conference, pp. 36.136.14, Royal Aeronautical Society, Cambridge, Great
Britain.
Metcalfe, R.W., Rutland, C.J., Duncan, J.H., and Riley, J.J. (1986) Numerical Simulations of
Active Stabilization of Laminar Boundary Layers, AIAA J. 24, pp. 14941501.
Miau, J.J., Chen, M.H., and Chow, J.H. (1988) Flow Structures of a Vertically Oscillating Plate
Immersed in a Flat-Plate Turbulent Boundary Layer, in Proc. Eleventh Symp. on Turbulence,
paper no. A28, University of Missouri, Rolla, Missouri.
396 BIBLIOGRAPHY
Mickley, H.S., and Davis, R.S. (1957) Momentum Transfer for Flow over a Flat Plate with Blow-
ing, NACA Technical Note No. TN-4017, Washington, D.C.
Migay, V.K. (1960a) Diffuser with Transverse Fins (English translation from Russian), Energo-
mashinostroenie, no. 4, p. 31.
Migay, V.K. (1960b) On Improving the Effectiveness of Diffuser Flows with Separation (English
translation from Russian), Mekhanika i Mashinostroyeniye, no. 4, pp. 171173.
Migay, V.K. (1961) Increasing the Efciency of Diffusers by Fitting Transverse Fins (English trans-
lation from Russian), Teploenergetika , no. 4, pp. 4143.
Migay, V.K. (1962a) The Efciency of a Cross-Ribbed Curvilinear Diffuser (English translation
from Russian), Energomashinostroenie, no. 1, pp. 4546.
Migay, V.K. (1962b) The Aerodynamic Effectiveness of a Discontinuous Surface (English transla-
tion from Russian), Inzhenerno-Fizicheskiy Zhurnal 5, pp. 2024.
Migay, V.K. (1962c) Investigating Finned Diffusers: Effects of Geometry on Effectiveness of Finned
Body Diffusers (English translation from Russian), Teploenergetika, no. 10, pp. 5559.
Migun, N.P., and Prokhorenko, P.P. (1987) Measurement of the Viscosity of Polar Liquids in
Microcapillaries, Colloid J. of the USSR 49, pp. 894897.
Millikan, C.B. (1939) A Critical Discussion of Turbulent Flows in Channels and Circular Tubes,
in Proc. Fifth Int. Cong. Appl. Mech., eds. J.P. Den Hartog and H. Peters, pp. 386392, Wiley,
New York.
Milling, R.W. (1981) TollmienSchlichting Wave Cancellation, Phys. Fluids 24, pp. 979981.
Milne-Thomson, L.M. (1968) Theoretical Hydrodynamics, fth edition, Macmillan, London, Great
Britain.
Modi, V.J., Fernando, M., and Yokomizo, T. (1990) Drag Reduction of Bluff Bodies Through
Moving Surface Boundary Layer Control, AIAA Paper No. 90-0298, New York.
Modi, V.J., Mokhtarian, F., Fernando, M., and Yokomizo, T. (1989) Moving Surface Boundary
Layer Control as Applied to 2-D Airfoils, AIAA Paper No. 89-0296, New York.
Modi, V.J., Sun, J.L.C., Akutsu, T., Lake, P., McMillan, K., Swinton, P.G., and Mullins, D. (1980)
Moving Surface Boundary Layer Control for Aircraft Operations at High Incidence, AIAA
Modi, V.J., Sun, J.L.C., Akutsu, T., Lake, P., McMillian, K., Swinton, P.G., and Mullins, D. (1981)
Moving Surface Boundary Layer Control for Aircraft Operation at High Incidence, J. Aircraft
18, pp. 963968.
Moffatt, H.K. (1964) Viscous and Resistive Eddies Near a Sharp Corner, J. Fluid Mech. 18,
pp. 118.
Moin, P., and Bewley, T. (1994) Feedback Control of Turbulence, Appl. Mech. Rev. 47, no. 6,
part 2, pp. S3S13.
Moin, P., and Kim, J. (1982) Numerical Investigation of Turbulent Channel Flow, J. Fluid Mech.
118, pp. 341377.
Moin, P., and Kim, J. (1985) The Structure of the Vorticity Field in Turbulent Channel Flow. Part 1.
Analysis of Instantaneous Fields and Statistical Correlations, J. Fluid Mech. 155, pp. 441464.
Moin, P., and Mahesh, K. (1998) Direct Numerical Simulation: A Tool in Turbulence Research,
Mokhtarian, F., and Modi, V.J. (1988) Fluid Dynamics of Airfoils with Moving Surface Boundary-
Layer Control, J. Aircraft 25, pp. 163169.
Mokhtarian, F., Modi, V.J., and Yokomizo, T. (1988a) Effect of Moving Surfaces on the Airfoil
Boundary Layer Control, AIAA Paper No. 88-4337-CP, New York.
Mokhtarian, F., Modi, V.J., and Yokomizo, T. (1988b) Rotating Air Scoop as Airfoil Boundary-
Layer Control, J. Aircraft 25, pp. 973975.
Monin, A.S., and Yaglom, A.M. (1971) Statistical Fluid Mechanics, vol. I, MIT Press, Cambridge,
Massachusetts.
Moon, F.C. (1992) Chaotic and Fractal DynamicsAn Introduction for Applied Scientists and
Engineers, Wiley, New York.
Moore, F.K. (1958) On the Separation of the Unsteady Laminar Boundary Layer, in Boundary-
Layer Research, ed. H. G ortler, pp. 296310, SpringerVerlag, Berlin.
Moran, M.J., and Shapiro, H.N. (1995) Fundamentals of Engineering Thermodynamics, third
edition, Wiley, New York.
Morduchow, M., and Grape, R.G. (1955) Separation, Stability, and Other Properties of
BIBLIOGRAPHY 397
Compressible Laminar Boundary Layer with Pressure Gradient and Heat Transfer, NACA TN
No. 3296, Washington, D.C.
Morkovin, M.V. (1969) Critical Evaluation of Transition from Laminar to Turbulent Shear Lay-
ers with Emphasis on Hypersonically Traveling Bodies, Air Force Flight Dynamics Laboratory
Report No. AFFDL-TR-68-149, Wright-Patterson AFB, Ohio.
Morkovin, M.V. (1984) Bypass Transition to Turbulence and Research Desiderata, in Transition
in Turbines Symp., NASA CP-2386, Washington, D.C.
Morkovin, M.V. (1988) Recent Insights into Instability and Transition to Turbulence in Open-Flow
Systems, AIAA Paper No. 88-3675, New York.
Moroney, R.M., White, R.M., and Howe, R.T. (1991) Ultrasonically Induced Microtransport, in
Proc. IEEE Micro Electro Mechanical Systems 91, pp. 277282, Nara, Japan, IEEE, New York.
Morrison, W.R.B. (1969) Two-Dimensional Frequency-Wavenumber Spectra and Narrow-Band
Shear Stress Correlations in Turbulent Pipe Flow, Ph.D. Thesis, University of Queensland,
Queensland, Australia.
Morrison, W.R.B., Bullock, K.J., and Kronauer, R.E. (1971) Experimental Evidence of Waves in
the Sublayer, J. Fluid Mech. 47, pp. 639656.
Mueller, T.J. (1985a) Low Reynolds Number Vehicles, AGARDograph No. 288, AGARD,
NATO, Neuilly-sur-Seine, France.
Mueller, T.J., editor (1985b) Proceedings of the Conference on Low Reynolds Number Airfoil
Aerodynamics, University of Notre Dame, Notre Dame, Indiana.
Mueller, T.J. (1985c) The Inuence of Laminar Separation and Transition on Low Reynolds Num-
ber Airfoil Hysteresis, J. Aircraft 22, pp. 763770.
Mueller, T.J., editor (1989) Proceedings of the Conference on LowReynolds Number Aerodynamics,
University of Notre Dame, Notre Dame, Indiana.
Mueller, T.J., and Burns, T.F. (1982) Experimental Studies of the Eppler 61 Airfoil at LowReynolds
Numbers, AIAA Paper No. 82-0345, New York.
Muirhead, V.U., and Saltzman, E.G. (1979) Reduction of Aerodynamic Drag and Fuel Consump-
tion for Tractor-Trailer Vehicles, J. Energy 3, pp. 279284.
Muller, D.A., Sorsch, T., Moccio, S., Baumann, F.H., Evans-Lutterodt, K., and Timp, G. (1999)
The Electronic Structure at the Atomic Scale of Ultrathin Gate Oxides, Nature 399, no. 6738,
24 June, pp. 758761.
Mullin, T., Greated, C.A., and Grant, I. (1980) Pulsating Flow over a Step, Phys. Fluids 23,
pp. 669674.
Muntz, E.P. (1989) Rareed Gas Dynamics, Ann. Rev. Fluid Mech. 21, pp. 387417.
Murlis, J., Tsai, H.M., and Bradshaw, P. (1982) The Structure of Turbulent Boundary Layers at
Low Reynolds Numbers, J. Fluid Mech. 122, pp. 1356.
Nadolink, R.H., and Haigh, W.W. (1995) Bibliography on Skin Friction Reduction with Polymers
and Other Boundary-Layer Additives, Appl. Mech. Rev. 48, pp. 351459.
Nagamatsu, H.T., Dyer, R., and Ficarra, R.V. (1985) Supercritical Airfoil Drag Reduction by
Passive Shock Wave/Boundary Layer Control in the Mach Number Range .75 to .9, AIAA Paper
Nagamatsu, H.T., Trilling, T.W., and Bossard, J.A. (1987) Passive Drag Reduction on a Complete
NACA 0012 Airfoil at Transonic Mach Numbers, AIAA Paper No. 87-1263, New York.
Nagano, Y., and Tagawa, M. (1990) AStructural Turbulence Model for Triple Products of Velocity
and Scalar, J. Fluid Mech. 215, pp. 639657.
Nagel, A.L., Alford, W.J., Jr., and Dugan, J.F. (1975) Future Long-Range TransportsProspects
for Improved Fuel Efciency, NASA Technical Memorandum No. X-72659, Washington,
D.C.
Naguib, A.M. (1992) Inner- and Outer-Layer Effects on the Dynamics of a Turbulent Boundary
Layer, Ph.D. Thesis, Illinois Institute of Technology, Chicago, Illinois.
Naguib, A.M., and Wark, C.E. (1992) An Investigation of Wall-Layer Dynamics Using a Combined
Temporal Filtering and Correlation Techniques, J. Fluid Mech. 243, pp. 541560.
Nakagawa, S., Shoji, S., and Esashi, M. (1990) A Micro-Chemical Analyzing SystemIntegrated on
Silicon Chip, in Proc. IEEE: Micro Electro Mechanical Systems, Napa Valley, California, IEEE
90CH2832-4, IEEE, New York.
Nakayama, W. (1986) Thermal Management of Electronic Equipment, App. Mech. Rev. 39,
pp. 18471868.
398 BIBLIOGRAPHY
Narasimha, R., and Kailas, S.V. (1986) Energy Events in the Atmospheric Boundary Layer, Indian
Institute of Science Report No. 86-AS-8, Bangalore, India.
Narasimha, R., and Kailas, S.V. (1987) Energy Events in the Atmospheric Boundary Layer, in
Perspectives in Turbulent Studies, eds. H.U. Meier and P. Bradshaw, pp. 188222, Springer
Verlag, Berlin.
Narasimha, R., and Kailas, S.V. (1990) Turbulent Bursts in the Atmosphere, Atmospheric Envi-
ronment 24A, pp. 16351645.
Narasimha, R., and Liepmann, H.W. (1988) Introduction, in Turbulence Management and Re-
laminarisation, eds. H.W. Liepmann and R. Narasimha, pp. xiixx, SpringerVerlag, Berlin.
Narasimha, R., and Ojha, S.K. (1967) Effect of Longitudinal Surface Curvature on Boundary
Narasimha, R., and Sreenivasan, K.R. (1973) Relaminarization in Highly Accelerated Turbulent
Narasimha, R., and Sreenivasan, K.R. (1979) Relaminarization of Fluid Flows, in Advances in
Applied Mechanics, vol. 19, ed. C.-S. Yih, pp. 221309, Academic Press, New York.
Narasimha, R., and Sreenivasan, K.R. (1988) Flat Plate Drag Reduction by Turbulence Manipu-
lation, S adh an a 12, pp. 1530.
Nayfeh, A.H., and Balachandran, B. (1995) Applied Nonlinear DynamicsAnalytical, Computa-
tional, and Experimental Methods, Wiley, New York.
Nayfeh, A.H., Ragab, S.A., and Al-Maaitah, A. (1986) Effects of Roughness on the Stability of
Boundary Layers, AIAA Paper No. 86-1044, New York.
Nelson, C.F., Koga, D.J., and Eaton, J.K. (1987) Control of the Unsteady Separated Flow Behind
an Oscillating Two-Dimensional Flap, AIAA Paper No. 89-1027, New York.
Nelson, C.F., Koga, D.J., and Eaton, J.K. (1990) Unsteady, Separated Flow behind an Oscillating,
Two-Dimensional Spoiler, AIAA J. 28, pp. 845852.
Nelson, M.M., and Illingworth, W.T. (1991) A Practical Guide to Neural Nets, AddisonWesley,
Reading, Massachusetts.
Neuburger, D., and Wygnanski, I. (1988) The Use of a Vibrating Ribbon to Delay Separation on
Two-Dimensional Airfoils: Some Preliminary Observations, in Proc. Workshop II on Unsteady
Separated Flow, ed. J.M. Walker, pp. 333341, Frank J. Seiler Research Laboratory Report
No. FJSRL-TR-88-0004, U.S. Air Force Systems Command, Colorado Springs, Colorado.
Newman, G.R. (1974) An Experimental Study of Coherent Structures in the Turbulent Boundary
Layer, Post-Graduate Study Dissertation, Cambridge University, Cambridge, Great Britain.
Nickerson, J.D. (1986) A Study of Vortex Generators at Low Reynolds Numbers, AIAA Paper
Nikuradse, J. (1932) Gesetzm assigkeit der turbulenten Str omung in glatten R ohren, Forschg.
Arb. Ing.-Wes., no. 356, Germany.
Nikuradse, J. (1933) Str omungsgesetze in rauhen R ohren, Forschg. Arb. Ing.-Wes., no. 361,
Germany.
Nisewanger, C.R. (1964) FlowNoise and Drag Measurements of Vehicle with Compliant Coating,
U.S. Naval Ordnance Test Station Report No. 8518, NOTS No. TP-3510, China Lake, California.
Noor, A., and Jorgensen, C.C. (1996) A Hard Look at Soft Computing, Aerospace America 34,
September, pp. 3439.
Norman, J.R., and Fraser, F.C. (1937) Giant Fishes, Whales and Dolphins, Putnam, London, Great
Britain.
Nosenchuck, D.M., and Lynch, M.K. (1985) The Control of Low-Speed Streak Bursting in Tur-
bulent Spots, AIAA Paper No. 85-0535, New York.
Novak, C.J., Cornelius, K.C., and Roads, R.K. (1987) Experimental Investigations of the Circular
Wall Jet on a Circulation Control Airfoil, AIAA Paper No. 87-0155, New York.
OConnor, L. (1992) MEMS: Micromechanical Systems, Mechanical Engineering 114, February,
pp. 4047.
Odell, G.M., and Kovasznay, L.S.G. (1971) A New Type of Water Channel with Density Strati-
cation, J. Fluid Mech. 50, pp. 535543.
Offen, G.R., and Kline, S.J. (1974) Combined Dye-Streak and Hydrogen-Bubble Visual Observa-
tions of a Turbulent Boundary Layer, J. Fluid Mech. 62, pp. 223239.
Offen, G.R., and Kline, S.J. (1975) A Proposed Model of the Bursting Process in Turbulent Bound-
ary Layers, J. Fluid Mech. 70, pp. 209228.
BIBLIOGRAPHY 399
Ogorodnikov, D.A., Grin, V.T., and Zakharov, N.N. (1972) Boundary Layer Control of Hyper-
sonic Air Inlets, NASA TTF-13927, Washington, D.C.
Oldaker, D.K., and Tiederman, W.G. (1977) Spatial Structure of the Viscous Sublayer in Drag-
Reducing Channel Flows, Phys. Fluids 20, no. 10, part II, pp. S133144.
Oran, E.S., Oh, C.K., and Cybyk, B.Z. (1998) Direct Simulation Monte Carlo: Recent Advances
and Applications, Ann. Rev. Fluid Mech. 30, pp. 403441.
Orszag, S.A. (1971) Accurate Solution of the Orr-Sommerfeld Stability Equation, J. Fluid Mech.
50, pp. 689703.
Orszag, S.A. (1979) Prediction of Compliant Wall Drag Reduction, NASA Contractor Report
No. 3071, Washington, D.C.
Ott, E. (1993) Chaos in Dynamical Systems, Cambridge University Press, Cambridge, Great Britain.
Ott, E., Grebogi, C., and Yorke, J.A. (1990a) Controlling Chaos, Phys. Rev. Lett. 64, pp. 1196
1199.
Ott, E., Grebogi, C., and Yorke, J.A. (1990b) Controlling Chaotic Dynamical Systems, in Chaos:
SovietAmerican Perspectives on Nonlinear Science, ed. D.K. Campbell, pp. 153172, American
Institute of Physics, New York.
Ottino, J.M. (1989a) The Kinematics of Mixing: Stretching, Chaos, and Transport, Cambridge
University Press, Cambridge, Great Britain.
Ottino, J.M. (1989b) The Mixing of Fluids, Scientc American 260, January, pp. 5667.
Ottino, J.M. (1990) Mixing, Chaotic Advection, and Turbulence, Ann. Rev. Fluid Mech. 22,
pp. 207253.
Ouellette, J. (1996) MEMS: Mega Promise for Micro Devices, Mechanical Engineering 118,
October, pp. 6468.
Ouellette, J. (1999) Electronic Noses Sniff Out New Markets, Industrial Physicist 5, no. 1,
pp. 2629.
Owen, F.K., and Horstman, C.C. (1972) On the Structure of Hypersonic Turbulent Boundary
Owen, F.K., Horstman, C.C., and Kussoy, M.I. (1975) Mean and Fluctuating Flow Measurements
of a Fully-Developed, Non-Adiabatic, Hypersonic Boundary Layer, J. Fluid Mech. 70, pp. 393
413.
Oyler, T.E., and Palmer, W.E. (1972) Exploratory Investigation of Pulse Blowing for Boundary
Layer Control, Report No. NR72H-12, Columbus Aircraft Division, North American Rockwell
Corp, Columbus, Ohio. (Also available from U.S. NTIS; Document Number AD-742 085.)
Paiziz, S.T., and Schwarz, W.H. (1974) An Investigation of the Topography and Motion of the
Turbulent Interface, J. Fluid Mech. 63, pp. 315343.
Panton, R.L. (1990a) Scaling Turbulent Wall layers, J. Fluid Eng. 112, pp. 425432.
Panton, R.L. (1990b) The Role of Dimensional Analysis in Matched and Composite Asymp-
totic Expansions, in Ocean Waves Mechanics, Computational Fluid Dynamics and Mathemat-
ical Modelling, ed. M. Rahman, pp. 363379, Computational Mechanics Publications, Boston,
Massachusetts.
Panton, R.L. (1990c) Effects of a Contoured Apex on Vortex Breakdown, J. Aircraft 27,
pp. 285288.
Panton, R.L. (1990d) Inner-Outer Structure of the Wall-Pressure Correlation Function, in Near-
Wall Turbulence: 1988 Zoran Zari c Memorial Conference, eds. S.J. Kline and N.H. Afgan,
pp. 381396, Hemisphere, New York.
Panton, R.L. (1991) The Effects of Reynolds Number on Turbulent Wall Flows, in Proc. Eighth
Symposium on Turbulent Shear Flows, paper no. I-15, 911 September, Technical University of
Munich, Munich, Germany.
Panton, R.L. (1996) Incompressible Flow, second edition, WileyInterscience, New York.
Panton, R.L., editor (1997a) Self-Sustaining Mechanisms of Wall Turbulence, Computational Me-
chanics Publications, Southampton, Great Britain.
Panton, R.L. (1997b) A Reynolds Stress Function for Wall Layers, J. Fluids Eng. 119, pp. 325
330.
Panton, R.L. (1998) On the Wall-Pressure Fluctuations under a Three-Dimensional Boundary
Layer, J. Fluids Eng. 120, pp. 407410.
Panton, R.L., Goldman, A.L., Lowery, R.L., and Reischman, M.M. (1980) Low-Frequency Pres-
sure Fluctuations in Axisymmetric Turbulent Boundary Layers, J. Fluid Mech. 97, pp. 299319.
400 BIBLIOGRAPHY
Panton, R.L., and Linebarger, J.H. (1974) Wall Pressure Spectra Calculations for Equilibrium
Panton, R.L., and Robert, G. (1993) The Wall-Pressure Spectrum under a Turbulent Boundary
Layer: Part 2, Theory and Results, in ASME Fluid Engineering Conference, eds. M.J. Morris
and B.F. Carrol, FED-Vol. 155, pp. 4348, ASME, New York.
Panton, R.L., and Robert, G. (1994) The Wavenumber-Phase Velocity Representation for the
Turbulent Wall-Pressure Spectrum, J. Fluids Eng. 116, pp. 477483.
Papell, S.S. (1984) Vortex Generating Flow Passage Design for Increased Film-Cooling Effective-
ness and Surface Coverage, NASA TM-83617, Washington, D.C.
Patel, V.C., and Head, M.R. (1968) Reversion of Turbulent to Laminar Flow, J. Fluid Mech. 34,
pp. 371392.
Patera, A.T. (1986) Spectral Element Simulation of Flow in Grooved Channels: Cooling Chips
with Tollmien-Schlichting Waves, in Supercomputers and Fluid Dynamics, eds. K. Kuwahara,
R. Mendez, and S.A. Orszag, pp. 4151, SpringerVerlag, Berlin.
Patera, A.T., and Miki c, B.B. (1986) Exploiting Hydrodynamic Instabilities. Resonant Heat Trans-
fer Enhancement, Int. J. Heat Mass Transfer 29, pp. 11271138.
Patterson, G.K., Zakin, J.L., and Rodriguez, J.M. (1969) Drag Reduction. Polymer Solutions,
Soap Solutions, and Solid Particle Suspensions in Pipe Flow, Indus. & Eng. Chem. 61, pp. 22
30.
Paula, G. (1996) MEMS Sensors Branch Out, Aerospace America 34, September, pp. 2632.
Pearcey, H.H. (1961) Shock Induced Separation and Its Prevention by Design and Boundary Layer
Control, in Boundary Layer and Flow Control, ed. G.V. Lachmann, vol. 2, pp. 11661344,
Pergamon Press, Oxford, Great Britain.
Pearson, J.R.A., and Petrie, C.J.S. (1968) On Melt Flow Instability of Extruded Polymers, in
Polymer Systems: Deformation and Flow, eds. R.E. Wetton and R.W. Whorlow, pp. 163187,
Macmillian, London, Great Britain.
Perrier, P. (1998) Multiscale Active Flow Control, in Flow Control: Fundamentals and Practices,
Perry, A.E., and Abell, C.J. (1975) Scaling Laws for Pipe-Flow Turbulence, J. Fluid Mech. 67,
pp. 257271.
Perry, A.E., and Abell, C.J. (1977) Asymptotic Similarity of Turbulence Structures in Smooth- and
Rough-Walled Pipes, J. Fluid Mech. 79, pp. 785799.
Perry, A.E., and Chong, M.S. (1982) On the Mechanism of Wall Turbulence, J. Fluid Mech. 119,
pp. 173217.
Perry, A.E., Henbest, S.M., and Chong, M.S. (1986) A Theoretical and Experimental Study of
Wall Turbulence, J. Fluid Mech. 165, pp. 163199.
Perry, A.E., Li, J.D., Henbest, S., and Marusic, I. (1990) The Attached Eddy Hypothesis in Wall
Turbulence, in Near-Wall Turbulence: 1988 Zoran Zari c Memorial Conference, eds. S.J. Kline
and N.H. Afgan, pp. 715735, Hemisphere, New York.
Perry, A.E., Lim, T.T., and Teh, E.W. (1981) A Visual Study of Turbulent Spots, J. Fluid Mech.
104, pp. 387405.
Peters, H. (1938) A Study in Boundary Layers, in Proc. Fifth Int. Cong. Appl. Mech., eds. J.P.
Den Hartog and H. Peters, pp. 393395, Wiley, New York.
Petersen, I.R., and Savkin, A.V. (1999) Robust Kalman Filtering for Signals and Systems with Large
Uncertainties, Birkh auser, Boston, Massachusetts.
Petroski, H. (1997) Development and Research, American Scientist 85, no. 3, pp. 210213.
Pfahler, J. (1992) Liquid Transport in Micron and Submicron Size Channels, Ph.D. Thesis,
University of Pennsylvania, Philadelphia, Pennsylvania.
Pfahler, J., Harley, J., Bau, H., and Zemel, J.N. (1990) Liquid Transport in Micron and Submicron
Channels, Sensors & Actuators A 2123, pp. 431434.
Pfahler, J., Harley, J., Bau, H., and Zemel, J.N. (1991) Gas and Liquid Flowin Small Channels, in
Symp. on Micromechanical Sensors, Actuators, and Systems, eds. D. Cho et al., ASME DSC-Vol.
32, pp. 4960, ASME, New York.
Pfeffer, R., and Rosetti, S.J. (1971) Experimental Determination of Pressure Drop and Flow Char-
acteristics of Dilute Gas-Solid Suspensions, NASA Contractor Report No. 1894, Washington,
D.C.
Pfenninger, W. (1946) Untersuchungen uber Reibungsverminderung an Trag ugeln, insbesondere
BIBLIOGRAPHY 401
mit Hilfe von Grenzschichtabsaugung, Reports of the Inst. of Aerodynamics, no. 13, ETH
Z urich, Z urich, Switzerland.
Pfenninger, W., and Vemuru, C.S. (1990) Design of Low Reynolds Number Airfoils, J. Aircraft
27, pp. 204210.
Phillips, O.M. (1979) The Last Chance Energy Book, Johns Hopkins University Press, Baltimore,
Maryland.
Phillips, W.R.C. (1987) The Wall Region of a Turbulent Boundary Layer, Phys. Fluids 30,
pp. 23542361.
Phillips, W.R.C., and Ratnanather, J.T. (1990) The Outer Region of a Turbulent Boundary Layer,
Phys. Fluids A 2, pp. 427434.
Piekos, E.S., and Breuer, K.S. (1996) Numerical Modeling of Micromechanical Devices Using the
Direct Simulation Monte Carlo Method, J. Fluids Eng. 118, pp. 464469.
Piekos, E.S., Orr, D.J., Jacobson, S.A., Ehrich, F.F., and Breuer, K.S. (1997) Design and Analysis
of Microfabricated High Speed Gas Journal Bearings, AIAA Paper No. 97-1966, AIAA, Reston,
Virginia.
Pinebrook, W.E., and Dalton, C. (1983) Drag Minimization on a Body of Revolution Through
Evolution, Computer Methods in Appl. Mech. and Eng. 39, pp. 179197.
Pister, K.S.J., Fearing, R.S., and Howe, R.T. (1990) A Planar Air Levitated Electrostatic Actuator
System, IEEE Paper No. CH2832-4/90/0000-0067, IEEE, New York.
Plesniak, M.W., and Nagib, H.M. (1985) Net Drag Reduction in Turbulent Boundary Layers
Resulting from Optimized Manipulation, AIAA Paper No. 85-0518, New York.
Poisson-Quinton, Ph. (1950) On the Mechanism and the Application of the Control of the Bound-
ary Layer of Airplanes, in Int. Colloquium on Mechanics, vol. II, Poitiers, France.
Pollard, A. (1997) Passive and Active Control of Near-Wall Turbulence, Prog. Aero. Sci. 33,
pp. 689708.
Pollard A. (1998) Near-Wall Turbulence Control, Flow Control: Fundamentals and Practices,
Pomeau, Y., and Manneville, P. (1980) Intermittent Transition to Turbulence in Dissipative Dy-
namical Systems, Commun. Math. Phys. 74, pp. 189197.
Pong, K.-C., Ho, C.-M., Liu, J., and Tai, Y.-C. (1994) Non-Linear Pressure Distribution in Uni-
form Microchannels, in Application of Microfabrication to Fluid Mechanics 1994, eds. P.R.
Bandyopadhyay, K.S. Breuer, and C.J. Belchinger, ASME FED-Vol. 197, pp. 4752, ASME, New
York.
Pope, S.B. (1996) Reacting Flows and Combustion, in Research Trends in Fluid Mechanics, eds.
J.L. Lumley, A. Acrivos, L.G. Leal, and S. Leibovich, pp. 229243, American Institute of Physics,
Woodbury, New York.
Porodnov, B.T., Suetin, P.E., Borisov, S.F., and Akinshin, V.D. (1974) Experimental Investigation
of Rareed Gas Flow in Different Channels, J. Fluid Mech. 64, pp. 417437.
Povkh, I.L., Bolonov, N.I., and Eidelman, A.Ye. (1979) The Average Velocity Prole and the
Frictional Loss in Turbulent Flowof an Aqueous Suspension of Clay, Fluid Mech. Soviet Research
8, pp. 118124.
Prandtl, L. (1904)
Uber Fl ussigkeitsbewegung bei sehr kleiner Reibung, in Proc. Third Int. Math.
Cong., pp. 484491, Heidelberg, Germany.
Prandtl, L. (1925a) Bericht uber Untersuchungen zur ausgebildeten Turbulenz, Z. angew. Math.
Mech. 5, pp. 136139.
Prandtl, L. (1925b) Magnuseffeckt und Windkraftschiff, Naturwissenschaften 13, pp. 93108.
Prandtl, L. (1935) The Mechanics of Viscous Fluids, in Aerodynamic Theory, ed. W.F. Durand,
vol. III, pp. 34208, SpringerVerlag, Berlin.
Praturi, A.K., and Brodkey, R.S. (1978) A Stereoscopic Visual Study of Coherent Structures in
Turbulent Shear Flows, J. Fluid Mech. 89, pp. 251272.
Preston, J.H. (1958) The Minimum Reynolds Number for a Turbulent Boundary Layer and the
Selection of a Transition Device, J. Fluid Mech. 3, pp. 373384.
Preston, J.H., and Sweeting, N.E. (1944) The Velocity Distribution in the Boundary Layer of a
Plane Wall at High Reynolds Numbers with Suggestions for Further Experiments, Aeronautical
Research Council Report No. ARC-FM-671, London, Great Britain.
Pretsch, J. (1942) Umschlagbeginn und Absaugung, Jahrb. Dtsch. Luftfahrtforschung 1, pp. 54
71.
402 BIBLIOGRAPHY
Prudhomme, R.K., Chapman, T.W., and Bowen, J.R. (1986) Laminar Compressible Flow in a
Tube, Appl. Scientic Res. 43, pp. 6774.
Purohit, S.C. (1987) Effect of Vectored Suction on a Shock-Induced Separation, AIAA J. 25,
pp. 759760.
Purshouse, M. (1976) On the Damping of Unsteady Flow by Compliant Boundaries, J. Sound
Vibr. 49, pp. 423436.
Purshouse, M. (1977) Interaction of Flow with Compliant Surfaces, Ph.D. Thesis, Cambridge
University, Cambridge, Great Britain.
Purtell, L.P., Klebanoff, P.S., and Buckley, F.T. (1981) Turbulent Boundary Layer at Low Reynolds
Number, Phys. Fluids 24, pp. 802811.
Puryear, F.W. (1962) Boundary Layer Control: Drag Reduction by Use of Compliant Coatings,
David Taylor Model Basin Report No. 1668, Bethesda, Maryland.
Qin, F., Wolf, E.E., and Chang, H.-C. (1994) Controlling Spatiotemporal Patterns on a Catalytic
Wafer, Phys. Rev. Lett. 72, pp. 14591462.
Radin, I. (1974) Solid-Fluid Drag Reduction, Ph.D. Thesis, University of Missouri, Rolla,
Missouri.
Radin, I., Zakin, J.L., and Patterson, G.K. (1975) Drag Reduction in SolidFluid Systems, AIChE
J. 21, pp. 358371.
Ragab, S.A., and Nayfeh, A.H. (1980) AComparison of the Second-Order Triple-Deck Theory and
Interacting Boundary Layers for Incompressible Flows Past a Hump, AIAA Paper No. 80-0072,
New York.
Raghunathan, S. (1985) Passive Control of Shock-Boundary Layer Interaction, Prog. Aero. Sci.
25, pp. 271296.
Rajagopalan, S., and Antonia, R.A. (1984) Conditional Averages Associated with the Fine Structure
in a Turbulent Boundary Layer, Phys. Fluids 27, pp. 19661973.
Rajagopalan, S., and Antonia, R.A. (1993) RMS Spanwise Vorticity Measurements in a Turbulent
Boundary Layer, Exp. Fluids 14, pp. 142144.
Rao, D.M. (1979) Leading-Edge Vortex Flap Experiments on a 74-Deg. Delta Wing, NASA
CR-159161, Washington, D.C.
Rao, D.M., and Kariya, T.T. (1988) Boundary-Layer Submerged Vortex Generators for Separation
ControlAn Exploratory Study, AIAA Paper No. 88-3546-CP, New York.
Rao, K.N., Narasimha, R., and Badri Narayanan, M.A. (1971) The Bursting Phenomenon in a
Turbulent Boundary Layer, J. Fluid Mech. 48, pp. 339352.
Ras, M., and Ackeret, J. (1941)

Uber Verhinderung der Grenzschicht-Turbulenz durch Ab-
saugung, Helv. Phys. Acta 14, p. 323.
Raspet, A. (1952) Boundary-Layer Studies on a Sailplane, Aeronaut. Eng. Rev. 11, pp. 5260.
Raupach, M.R., Antonia, R.A., and Rajagopalan, S. (1991) Rough-Wall Turbulent Boundary
Layers, Appl. Mech. Rev. 44, pp. 125.
Rayleigh, Lord (1880) On the Stability, or Instability, of Certain Fluid Motions, Proc. London
Math. Soc. 11, pp. 5770.
Rayleigh, Lord (1887) On Waves Propagated along the Plane Surface of an Elastic Solid, Proc.
London Math. Soc. XVII, pp. 411.
Reed, H.L., and Nayfeh, A.H. (1986) Numerical-Perturbation Technique for Stability of Flat-Plate
Boundary Layers with Suction, AIAA J. 24, pp. 208214.
Reed, H.L., and Saric, W.S. (1987) Stability and Transition of Three-Dimensional Flows, in
Proc. Tenth U.S. Nat. Cong. of Applied Mechanics, ed. J.P. Lamb, pp. 457468, ASME, New
York.
Reed, H.L., and Saric, W.S. (1989) Stability of Three-Dimensional Boundary Layers, Ann. Rev.
Fluid Mech. 21, pp. 235284.
Reischman, M.M., and Tiederman, W.G. (1975) Laser-Doppler Anemometer Measurements in
Drag-Reducing Channel Flows, J. Fluid Mech. 70, pp. 369392.
Reisenthel, P.H., Nagib, H.M., and Koga, D.J. (1985) Control of Separated Flows Using Forced
Unsteadiness, AIAA Paper No. 85-0556, New York.
Reisenthel, P.H., Xiong, Y., and Nagib, H.M. (1991) The Preferred Mode in an Axisymmetric Jet
With and Without Enhanced Feedback, AIAA Paper No. 91-0315, Washington, D.C.
Render, P.M. (1984) The Experimental and Theoretical Aerodynamics of Aerofoil Sections Suitable
BIBLIOGRAPHY 403
for Remotely Piloted Vehicles, Ph.D. Thesis, Craneld Institute of Technology, Craneld, Great
Britain.
Render, P.M., Stollery, J.L., and Williams, B.R. (1985) Aerofoils at Low Reynolds Numbers
Prediction and Experiments, in Numerical and Physical Aspects of Aerodynamic Flows III,
ed. T. Cebeci, pp. 155167, SpringerVerlag, New York.
Reshotko, E. (1976) Boundary-Layer Stability and Transition, Ann. Rev. Fluid Mech. 8, pp. 311
349.
Reshotko, E. (1979) Drag Reduction by Cooling in Hydrogen-Fueled Aircraft, J. Aircraft 16,
pp. 584590.
Reshotko, E. (1985) Control of Boundary Layer Transition, AIAA Paper No. 85-0562.
Reshotko, E. (1987) Stability and Transition, How Much Do We Know?, in Proc. Tenth U.S.
National Cong. of Applied Mechanics, ed. J.P. Lamb, pp. 421434, ASME, New York.
Reynolds, G.A., and Saric, W.S. (1986) Experiments on the Stability of the Flat-Plate Boundary
Layer with Suction, AIAA J. 24, pp. 202207.
Reynolds, O. (1883) An Experimental Investigation of the Circumstances Which Determine
Whether the Motion of Water Shall be Direct or Sinuous, and of the Law of Resistence in Parallel
Channels, Phil. Trans. Roy. Soc. London A 174, pp. 935982.
Reynolds, O. (1895) On the Dynamical Theory of Incompressible Viscous Fluids and the Deter-
mination of the Criterion, Phil. Trans. Roy. Soc. London A 186, pp. 123164.
Reynolds, W.C. (1976) Computation of Turbulent Flows, Ann. Rev. Fluid Mech. 8, pp. 183208.
Reynolds, W.C. (1993) Sensors, Actuators, and Strategies for Turbulent Shear-Flow Control,
invited oral presentation at AIAA Third Flow Control Conference, 69 July, Orlando, Florida.
Reynolds, W.C. (1996) Control of Turbulent Flows, in Research Trends in Fluid Mechanics,
eds. J.L. Lumley, A. Acrivos, L.G. Leal, and S. Leibovich, pp. 253262, American Institute of
Physics, Woodbury, New York.
Reynolds, W.C., and Carr, L.W. (1985) Review of Unsteady, Driven, Separated Flows, AIAA
Reynolds, W.C., Eaton, J.K., Johnston, J.P., Hesselink, L., Powell, D.J., Roberts, L., and Kroo, E.
(1988) Flow Control for Unsteady and Separated Flows and Turbulent Mixing, AFOSR-TR-
89-0232, Washington, D.C. (Also available from U.S. NTIS; Document Number AD-A205989.)
Rice, E.J., and Raman, G. (1993) Mixing Noise Reduction for Rectangular Supersonic Jets by
Nozzle Shaping and Induced Screech Mixing, AIAA Paper No. 93-4322, Washington, D.C.
Richardson, L.F. (1920) The Supply of Energy from and to Atmospheric Eddies, Proc. R. Soc.
London A 97, pp. 354373.
Richardson, L.F. (1922) Weather Prediction by Numerical Process, Cambridge University Press,
Richardson, S. (1973) On the No-Slip Boundary Condition, J. Fluid Mech. 59, pp. 707719.
Richter, A., Plettner, A., Hofmann, K.A., and Sandmaier, H. (1991) A Micromachined Electro-
hydrodynamic (EHD) Pump, Sensors & Actuators A 29, pp. 159168.
Riley, J.J., and Gad-el-Hak, M. (1985) The Dynamics of Turbulent Spots, in Frontiers in Fluid
Mechanics, eds. S.H. Davis and J.L. Lumley, pp. 123155, SpringerVerlag, Berlin.
Riley, J.J., Gad-el-Hak, M., and Metcalfe, R.W. (1988) Compliant Coatings, Ann. Rev. Fluid
Mech. 20, pp. 393420.
Ringleb, E.O. (1961) Separation Control by Trapped Vortices, in Boundary Layer and Flow
Control, ed. G.V. Lachmann, vol. 1, pp. 265294, Pergamon Press, Oxford, Great Britain.
Ritter, H., and Messum, L.T. (1964) Water Tunnel Measurements of Turbulent Skin Friction on
Six Different Compliant Surfaces of 1 Ft Length, British Admiralty Research Laboratory Report
No. ARL/N4/GHY/9/7, ARL/G/N9, London, Great Britain.
Ritter, H., and Porteous, J.S. (1964) Water Tunnel Measurements of Skin Friction on a Compliant
Coating, British Admiralty Research Laboratory Report No. ARL/N3/G/HY/9/7, London, Great
Britain.
Robert, G. (1993) The Wall-Pressure Spectrum under a Turbulent Boundary Layer: Part 1, Ex-
periments, in ASME Fluid Engineering Conference, eds. M.J. Morris and B.F. Carrol, FED-Vol.
155, pp. 3742, ASME, New York.
Robinson, E.Y., Helvajian, H., and Jansen, S.W. (1996a) Small and Smaller: The World of MNT,
Aerospace America 34, September, pp. 2632.
404 BIBLIOGRAPHY
Robinson, E.Y., Helvajian, H., and Jansen, S.W. (1996b) Big Benets from Tiny Technologies,
Aerospace America 34, October, pp. 3843.
Robinson, M.C., and Luttges, M.W. (1984) Unsteady Separated Flow: Forced and Common Vor-
ticity About Oscillating Airfoils, in Unsteady Separated Flows, eds. M.S. Francis and M.W.
Luttges, pp. 117126, U.S. Air Force Academy, Colorado Springs, Colorado.
Robinson, S.K. (1990) A Review of Vortex Structures and Associated Coherent Motions in Tur-
bulent Boundary layers, in Structure of Turbulence and Drag Reduction, ed. A. Gyr, pp. 2350,
Robinson, S.K. (1991) Coherent Motions in the Turbulent Boundary Layer, Ann. Rev. Fluid
Mech. 23, pp. 601639.
Robinson, S.K., Kline, S.J., and Spalart, P.R. (1989) A Review of Quasi-Coherent Structures in
a Numerically Simulated Turbulent Boundary Layer, NASA Technical Memorandum No. TM-
102191, Washington, D.C.
Rogallo, R.S., and Moin, P. (1984) Numerical Simualtion of Turbulent Flows, Ann. Rev. Fluid
Mech. 16, pp. 99137.
Romeiras, F.J., Grebogi, C., Ott, E., and Dayawansa, W.P. (1992) Controlling Chaotic Dynamical
Systems, Physica D 58, pp. 165192.
Roos, F.W. (1996) Microblowing for High-Angle-of-Attack Vortex Flow Control on a Fighter
Aircraft, AIAA Paper No. 96-0543, Washington, D.C.
Roos, F.W., and Kegelman, J.T. (1986) Control of Coherent Structures in Reattaching Laminar
and Turbulent Shear Layers, AIAA J. 24, pp. 19561963.
Rosenhead, L. (1963) Laminar Boundary Layers, Clarendon Press, Oxford, Great Britain.
Roshko A. (1976) Structure of Turbulent Shear Flows: ANewLook, AIAAJ. 14, pp. 13491357.
Roshko, A. (1992) Bluff-Body and Other Shear Flows, in Proc. NUWC Division Newport Sem-
inar Series on Turbulence and Its Control, compiled by P.R. Bandyopadhay and J.C.S. Meng,
pp. 4.14.34, Naval Undersea Warfare Center Technical Memorandum No. NUWC-NPT TM
922089, Newport, Rhode Island.
Rott, N. (1956) Unsteady Viscous Flow in the Vicinity of a Stagnation Point, Q. Appl. Math. 13,
pp. 444451.
Rotta, J.C. (1951) Statistische Theorie nichthomogener Turbulenz, Z. Phys. 129, pp. 547572.
Rotta, J.C. (1962) Turbulent Boundary Layers in Incompressible Flow, Prog. Aeronautical Sci.
2, pp. 1219.
Rotta, J.C. (1970) Control of Turbulent Boundary Layers by Uniform Injection and Suction
of Fluid, in Seventh Cong. of the International Council of the Aeronautical Sciences, paper
no. 70-10, ICAS, Rome, Italy.
Runstadler, P.G., Kline, S.J., and Reynolds, W.C. (1963) An Experimental Investigation of Flow
Structure of the Turbulent Boundary Layer, Department of Mechanical Engineering Report
No. MD-8, Stanford University, Stanford, California.
Runyan, L.J., and Steers, L. L. (1980) Boundary Layer Stability Analysis of a Natural Laminar
Flow Glove on the F-111 TACT Airplane, in Viscous Flow Drag Reduction, ed. G.R. Hough,
pp. 1732, AIAA, New York.
Ryskin, G. (1987) Turbulent Drag Reduction by Polymers: A Quantitative Theory, Phys. Rev.
Lett. 59, pp. 20592062.
Sabadell, L.A. (1988) Effects of a Drag Reducing Additive on Turbulent Boundary Layer Struc-
ture, M.Sc. Thesis, Princeton University, Princeton, New Jersey.
Saddoughi, S.G., and Veeravalli, S.V. (1994) Local Isotropy in Turbulent Boundary Layers at High
Reynolds Number, J. Fluid Mech. 268, pp. 333372.
Saffman, P.G. (1978) Problems and Progress in the Theory of Turbulence, in Structure and Mech-
anisms of Turbulence II, ed. H. Fiedler, pp. 273306, SpringerVerlag, Berlin.
Saffman, P.G., and Baker, G.R. (1979) Vortex Interactions, Ann. Rev. Fluid Mech. 11, pp. 95
122.
Sajben, M., Chen, C.P., and Kroutil, J.C. (1976) A New, Passive Boundary Layer Control Device,
Saric, W.S., and Reed, H.L. (1986) Effect of Suction and Weak Mass Injection on Boundary-Layer
Transition, AIAA J. 24, pp. 383389.
Saripalli, K.R., and Simpson, R.L. (1980) Investigation of Blown Boundary Layers with an Im-
proved Wall Jet System, NASA CR-3340, Washington, D.C.
BIBLIOGRAPHY 405
Sasaki, K., and Kiya, M. (1985) Effect of Free-Stream Turbulence on Turbulent Properties of a
Separation-Reattachment Flow, Bulletin JSME 28, pp. 610616.
Sasaki, K., and Kiya, M. (1991) Three-Dimensional Vortex Structure in a Leading-Edge Separation
Bubble at Moderate Reynolds Numbers, J. Fluids Eng. 113, pp. 405410.
Savas,

O, and Coles, D. (1985) Coherence Measurements in Synthetic Turbulent Boundary Lay-
ers, J. Fluid Mech. 160, pp. 421446.
Savins, J.G. (1967) A Stress-Controlled Drag-Reduction Phenomenon, Rheologica Acta 6,
pp. 323330.
Savu, G., and Trifu, O. (1984) Porous Airfoils in Transonic Flow, AIAA J. 22, pp. 989991.
Sawyers, D.R., Sen, M., and Chang, H.-C. (1996) Effect of Chaotic Interfacial Stretching on
Bimolecular Chemical Reaction in Helical-Coil Reactors, Chem. Eng. J. 64, pp. 129139.
Sawyers, D.R., Sen, M., and Chang, H.-C. (1998) Heat Transfer Enhancement in Three-
Dimensional Corrugated Channel Flow, Int. J. Heat Mass Transfer 41, pp. 35593573.
Schaaf, S.A., and Chambr e, P.L. (1961) Flow of Rareed Gases, Princeton University Press, Prince-
ton, New Jersey.
Schewe, G. (1983) On the Structure and Resolution of Wall-Pressure Fluctuations Associated with
Turbulent Boundary-Layer Flow, J. Fluid Mech. 134, pp. 311328.
Schildknecht, M., Miller, J.A., and Meir, G.E.A. (1979) The Inuence of Suction on the Structure
of Turbulence in Fully Developed Pipe Flow, J. Fluid Mech. 90, pp. 67107.
Schilz, W. (1965/66) Experimentelle Untersuchungen zur Akustischen Beeinussung der Str omung-
sgrenzschicht in Luft, Acustica 16, pp. 208223.
Schlichting, H. (1959) Einige neuere Ergebnisse uber Grenzschichtbeeinussung, in Proc. First
Int. Cong. Aero. Sci., eds. Th. von K arm an et al., pp. 563586, Adv. Aero. Sci., vol. 2, Pergamon
Press, London, Great Britain.
Schlichting, H. (1979) Boundary-Layer Theory, seventh edition, McGrawHill, New York.
Schlichting, H., and Pechau, W. (1959) Auftriebserh ohung von Trag ugeln durch kontinuierlich
verteilte Absaugung, ZFW 7, pp. 113119.
Schlichting, H., and Ulrich, A. (1940) Zur Berechnung des Umschlages laminar-turbulent, Jahrb.
Dtsch. Luftfahrtforschung 1, pp. 835.
Schmitz, F.W. (1967) Aerodynamics of the Model Airplane, Part I, Airfoil Measurements, NASA-
TM-X-60976, Washington, D.C.
Schneider, P.E.M. (1980) Sekund arwirbelbildung bei Ringwirbeln und in Freistrahlen, Z. Flug-
wiss. 4, pp. 307318.
Schoeld, W.H. (1985) Turbulent Boundary Layer Development in an Adverse Pressure Gradient
After an Interaction with a Normal Shock Wave, J. Fluid Mech. 154, pp. 4362.
Schoppa, W., and Hussain, F. (1998) A Large-Scale Control Strategy for Drag Reduction in Tur-
bulent Boundary Layers, Phys. Fluids 10, pp. 10491051.
Schreck, S.J., Faller, W.E., and Luttges, M.W. (1995) Neural Network Prediction of Three-
Dimensional Unsteady Separated Flow Fields, J. Aircraft 32, pp. 178185.
Schubauer, G.B., and Skramstad, H.K. (1947) Laminar Boundary-Layer Oscillations and Stability
of Laminar Flow, J. Aero. Sci. 14, pp. 6978.
Schubauer, G.B., and Skramstad, H.K. (1948) Laminar Boundary-Layer Oscillations and Transi-
tion on a Flat Plate, NACA Report No. 909, Washington, D.C.
Schubauer, G.B., and Spangenberg, W.G. (1960) Forced Mixing in Boundary Layers, J. Fluid
Mech. 8, pp. 1032.
Schulz, M. (1999) The End of the Road for Silicon, Nature 399, no. 6738, 24 June, pp. 729730.
Scott, M.R., and Watts, H.A. (1977) Computational Solution of Linear Two-Point Boundary Value
Problems via Orthonormalization, J. Numerical Analysis 14, pp. 4070.
Sears, W.R. (1956) Some Recent Developments in Airfoil Theory, J. Aeronaut. Sci. 23, pp. 490
499.
Sears, W.R., and Telionis, D.P. (1972a) Unsteady Boundary-Layer Separation, in Recent Research
on Unsteady Boundary Layers, ed. E.A. Eichelbrenner, vol. 1, pp. 404442, Presses de lUniversite
Laval, Quebec, Canada.
Sears, W.R., and Telionis, D.P. (1972b) Two Dimensional Laminar Boundary Layer Separation for
Unsteady Flow or Flow Past Moving Walls, Considering Singularity Due to Bifurcating Wake
Bubble, in Recent Research on Unsteady Boundary Layers, ed. E.A. Eichelbrenner, vol. 1,
pp. 443447, Presses de lUniversite Laval, Quebec, Canada.
406 BIBLIOGRAPHY
Sears, W.R., and Telionis, D.P. (1975) Boundary-Layer Separation in Unsteady Flow, J. Appl.
Math. 28, pp. 215235.
Seidl, M., and Steinheil, E. (1974) Measurement of Momentum Accommodation Coefcients on
Surfaces Characterized by Auger Spectroscopy, SIMS and LEED, in Rareed Gas Dynamics 9,
eds. M. Becker and M. Fiebig, pp. E9.1E9.2, DFVLR-Press, PorzWahn, Germany.
Selby, G.V., and Miandoab, F.H. (1990) Effect of Surface Grooves on Base Pressure for a Blunt
Trailing-Edge Airfoil, AIAA J. 28, pp. 11331135.
Sen, M. (1989) The Inuence of Developments in Dynamical Systems Theory on Experimental
Fluid Mechanics, in Frontiers in Experimental Fluid Mechanics, ed. M. Gad-el-Hak, pp. 124,
SpringerVerlag, New York.
Sen, M., Wajerski, D., and Gad-el-Hak, M. (1996) A Novel Pump for MEMS Applications,
J. Fluids Eng. 118, pp. 624627.
Sen, P.K., and Arora, D.S. (1988) On the Stability of Laminar Boundary-Layer Flow over a Flat
Plate with a Compliant Surface, J. Fluid Mech. 197, pp. 201240.
Service, R.F. (1999) Organic Molecule Rewires Chip Design, Science 285, no. 5426, 16 July,
pp. 313315.
Shah, D.A., and Antonia, R.A. (1989) Scaling of the Bursting Period in Turbulent Boundary Layer
and Duct Flows, Phys. Fluids A 1, pp. 318325.
Shah, R.K., and Joshi, S.D. (1987) Convective Heat Transfer in Curved Ducts, in Handbook of
Single-Phase Convective Heat Transfer, eds. S. Kakac, R.K. Shah, and W. Aung, pp. 5.15.46,
WileyInterscience, New York.
Sharatchandra, M.C., Sen, M., and Gad-el-Hak, M. (1997) NavierStokes Simulations of a Novel
Viscous Pump, J. Fluids Eng. 119, pp. 372382.
Sharatchandra, M.C., Sen, M., and Gad-el-Hak, M. (1998a) Thermal Aspects of a Novel Micro-
pumping Device, J. Heat Transfer 120, pp. 99107.
Sharatchandra, M.C., Sen, M., and Gad-el-Hak, M. (1998b) A New Approach to Constrained
Shape Optimization Using Genetic Algorithms, AIAA J. 36, pp. 5161.
Shercliff, J.A. (1965) A Textbook of Magnetohydrodynamics, Pergamon Press, Oxford, Great
Britain.
Sherman, F.S. (1990) Viscous Flow, McGrawHill, New York.
Shih, J.C., Ho, C.-M., Liu, J., and Tai, Y.-C. (1995) Non-Linear Pressure Distribution in Uniform
Microchannels, ASME AMD-MD-Vol. 238, New York.
Shih, J.C., Ho, C.-M., Liu, J., and Tai, Y-.C. (1996) Monatomic and Polyatomic Gas Flowthrough
UniformMicrochannels, in Applications of Microfabrication to Fluid Mechanics, eds. K. Breuer,
P. Bandyopadhyay, and M. Gad-el-Hak, ASME DSC-Vol. 59, pp. 197203, New York.
Shiloh, K., Shivaprasad, B.G., and Simpson, R.L. (1981) The Structure of a Separating Turbu-
lent Boundary Layer. Part 3: Transverse Velocity Measurements, J. Fluid Mech. 113, pp. 75
90.
Shinbrot, T. (1993) Chaos: Unpredictable Yet Controllable? Nonlinear Science Today 3, pp. 18.
Shinbrot, T. (1995) Progress in the Control of Chaos, Adv. Physics 44, pp. 73111.
Shinbrot, T. (1998) Chaos, Coherence and Control, in FlowControl: Fundamentals and Practices,
Shinbrot, T., Bresler, L., and Ottino, J.M. (1998) Manipulation of Isolated Structures in Experi-
mental Chaotic Fluid Flows, Exp. Thermal & Fluid Sci. 16, pp. 7683.
Shinbrot, T., Ditto, W., Grebogi, C., Ott, E., Spano, M., and Yorke, J.A. (1992) Using the Sensitive
Dependence of Chaos (the Buttery Effect) to Direct Trajectories in an Experimental Chaotic
System, Phys. Rev. Lett. 68, pp. 28632866.
Shinbrot, T., Grebogi, C., Ott, E., and Yorke, J.A. (1992a) Using Chaos to Target Stationary States
of Flows, Phys. Lett. A 169, pp. 349354.
Shinbrot, T., Grebogi, C., Ott, E., and Yorke, J.A. (1993) Using Small Perturbations to Control
Chaos, Nature 363, pp. 411417.
Shinbrot, T., Ott, E., Grebogi, C., and Yorke, J.A. (1990) Using Chaos to Direct Trajectories to
Targets, Phys. Rev. Lett. 65, pp. 32153218.
Shinbrot, T., Ott, E., Grebogi, C., and Yorke, J.A. (1992b) Using Chaos to Direct Orbits to Targets
in Systems Describable by a One-Dimensional Map, Phys. Rev. A 45, pp. 41654168.
Shinbrot, T., and Ottino, J.M. (1993a) Geometric Method to Create Coherent Structures in Chaotic
Flows, Phys. Rev. Lett. 71, pp. 843846.
BIBLIOGRAPHY 407
Shinbrot, T., and Ottino, J.M. (1993b) Using Horseshoes to Create Coherent Structures in Chaotic
Fluid Flows, Bul. Am. Phys. Soc. 38, p. 2194.
Sigal, A. (1971) An Experimental Investigation of the Turbulent Boundary Layer over a Wavy
Wall, Ph.D. Thesis, California Institute of Technology, Pasadena, California.
Sigurdson, L.W., and Roshko, A. (1985) Controlled Unsteady Excitation of a Reattaching Flow,
Silhanek, V. (1969) On Aircraft Longitudinal Motion after Boundary Layer Control SystemFailure
During Take-off and Landing, Summary Report No. Z-13, Aeronautical Research and Test
Institute, Prague, Czechoslovakia.
Simpson, R.L. (1970) Characteristics of Turbulent Boundary Layers at Low Reynolds Numbers
With and Without Transpiration, J. Fluid Mech. 42, pp. 769802.
Simpson, R.L. (1976) Comment on Prediction of Turbulent Boundary Layers at Low Reynolds
Numbers, AIAA J. 14, pp. 16621663.
Simpson, R.L. (1989) Turbulent Boundary-Layer Separation, Ann. Rev. Fluid Mech. 21, pp. 205
234.
Simpson, R.L., and Shivaprasad, B.G. (1983) The Structure of a Separating Turbulent Boundary
Layer. Part 5: Frequency Effects on Periodic Unsteady Freestream Flows, J. Fluid Mech. 131,
pp. 319339.
Simpson, R.L., Chew, Y.-T., and Shivaprasad, B.G. (1981a) The Structure of a Separating Turbulent
Boundary Layer. Part 1: Mean Flow and Reynolds Stresses, J. Fluid Mech. 113, pp. 2351.
Simpson, R.L., Chew, Y.-T., and Shivaprasad, B.G. (1981b) The Structure of a Separating Turbulent
Boundary Layer. Part 2: Higher-Order Turbulent Results, J. Fluid Mech. 113, pp. 5373.
Simpson, R.L., Shivaprasad, B.G., and Chew, Y.-T. (1983) The Structure of a Separating Turbulent
Boundary Layer. Part 4: Effects of Periodic FreestreamUnsteadiness, J. Fluid Mech. 127, pp. 219
261.
Singer, J., Wang, Y.-Z., and Bau, H.H. (1991) Controlling a Chaotic System, Phys. Rev. Lett. 66,
pp. 11231125.
Sirovich, L., and Deane, A.E. (1991) A Computational Study of RayleighB enard Convection.
Part 2. Dimension Considerations, J. Fluid Mech. 222, pp. 251265.
Sirovich L., and Karlsson S. (1997) Turbulent Drag Reduction by Passive Mechanisms, Nature
388, 21 August, pp. 753755.
Smagorinsky, J. (1963) General Circulation Experiments with the Primitive Equations, Mon.
Weather Rev. 91, pp. 99164.
Smith, A.M.O. (1957) Transition, Pressure Gradient, and Stability Theory, in Actes IX Congr` es
International de M ecanique Appliqu ee, vol. 4, pp. 234244, Universit e de Bruxelles, Bruxelles,
Belgique.
Smith, A.M.O. (1974) High Lift Aerodynamics, AIAA Paper No. 74-939, New York.
Smith, A.M.O. (1975) High-Lift Aerodynamics, J. Aircraft 12, pp. 501530.
Smith, A.M.O. (1977) Stratfords Turbulent Separation Criterion for Axially-Symmetric Flows,
J. Appl. Math. & Phys. 28, pp. 929939.
Smith, A.M.O., and Gamberoni, N. (1956) Transition, Pressure Gradient and Stability Theory,
Douglas Aircraft Company Report No. ES-26388, El Segundo, California.
Smith, A.M.O., and Kaups, K. (1968) Aerodynamics of Surface Roughness and Imperfections,
Society of Automotive Engineers Paper No. SAE-680198, New York.
Smith, A.M.O., and Thelander, J.A. (1973) The Power ProleA New Type of Airfoil, McDon-
nell Douglas Corporation Report No. MDC-J6236, Long Beach, CA. (Also available from U.S.
NTIS; Document Number AD-773 655/6.)
Smith, A.M.O., Stokes, T.R., Jr., and Lee, R.S. (1981) Optimum Tail Shapes for Bodies of Revo-
lution, J. Hydronautics 15, pp. 6773.
Smith, C.R., and Metzler, S.P. (1982) AVisual Study of the Characteristics, Formation, and Regen-
eration of Turbulent Boundary Layer Streaks, in Developments in Theoretical and Applied Me-
chanics, vol. XI, eds. T.J. Chung and G.R. Karr, pp. 533543, University of Alabama, Huntsville,
Alabama.
Smith, C.R., and Metzler, S.P. (1983) The Characteristics of Low-Speed Streaks in the Near-Wall
Region of a Turbulent Boundary Layer, J. Fluid Mech. 129, pp. 2754.
Smith, C.R., and Schwartz, S.P. (1983) Observation of Streamwise Rotation in the Near-Wall
Region of a Turbulent Boundary Layer, Phys. Fluids 26, pp. 641652.
408 BIBLIOGRAPHY
Smith, D.W., and Walker, J.H. (1959) Skin-Friction Measurements in Incompressible Flow, NACA
Report No. R-26, Washington, D.C.
Smith, R.L., and Blick, E.F. (1966) Skin Friction of Compliant Surfaces with Foamed Material
Substrate, J. Hydronautics 3, pp. 100102.
Smith, R.W. (1994) Effect of Reynolds Number on the Structure of Turbulent Boundary Layers,
Ph.D. Thesis, Princeton University, Princeton, New Jersey.
Smits, A.J. (1990) New Developments in Understanding Supersonic Turbulent Boundary Layers,
in Proc. Twelfth Symp. on Turbulence, eds. X.B. Reed, Jr., et al., pp. IL4.1IL4.19, University of
Missouri, Rolla, Missouri.
Smits, A.J., and Dussauge, J.P. (1996) Turbulent Shear Layers in Supersonic Flows, American
Institute of Physics, Woodbury, New York.
Smits, A.J., and Wood, D.H. (1985) The Response of Turbulent Boundary Layers to Sudden
Perturbations, Ann. Rev. Fluid Mech. 17, pp. 321358.
Smits, A.J., Spina, E.F., Alving, A.E., Smith, R.W., Fernando, E.M., and Donovan, J.F. (1989) A
Comparison of the Turbulence Structure of Subsonic and Supersonic Boundary Layers, Phys.
Fluids A 1, pp. 18651875.
Smits, J.G. (1990) Piezoelectric Micropump with Three Valves Working Peristaltically, Sensors &
Actuators A 2123, pp. 203206.
Sniegowski, J.J., and Garcia, E.J. (1996) Surface Micromachined Gear Trains Driven by an On-
Chip Electrostatic Microengine, IEEE Electron Device Letters 17, July, p. 366.
So, R.M.C., and Mellor, G.L. (1973) Experiment on Convex Curvature Effects in Turbulent Bound-
ary Layers, J. Fluid Mech. 60, pp. 4362.
Soderman, P.T. (1972) Aerodynamic Effects of Leading Edge Separation on a Two-Dimensional
Airfoil, NASA TMX-2643, Washington, D.C.
Soo, S.L., and Trezek, G.J. (1966) Turbulent Pipe Flow of Magnesia Particles in Air, I&EC
Fundamentals 5, pp. 388392.
Sovran, G., Morel, T., and Mason, W.T., Jr. (1978) Aerodynamic Drag Mechanisms of Bluff Bodies
and Road Vehicles, Plenum Press, New York.
Spaid, F.W. (1972) Cooled Supersonic Turbulent Boundary Layer Separated by a Forward Facing
Step, AIAA J. 19, pp. 11171119.
Spalart, P.R. (1986) Direct Simulation of a Turbulent Boundary Layer up to R
= 1410, NASA
Technical Memorandum No. TM-89407, Washington, D.C.
Spalart, P.R. (1988) Direct Simulation of a Turbulent Boundary Layer up to R
= 1410, J. Fluid
Mech. 187, pp. 6198.
Speziale, C.G. (1991) Analytical Methods for the Development of Reynolds-Stress Closures in
Turbulence, Ann. Rev. Fluid Mech. 23, pp. 107157.
Spina, E.F., Donovan, J.F., and Smits, A.J. (1991) On the Structure of High-Reynolds-Number
Supersonic Turbulent Boundary Layers, J. Fluid Mech. 222, pp. 293327.
Squire, H.B. (1933) On the Stability for Three-Dimensional Disturbances of Viscous Fluid Flow
Between Parallel Walls, Proc. R. Soc. London A 142, pp. 621628.
Sreenivasan, K.R. (1988) AUnied Viewof the Origin and Morphology of the Turbulent Boundary
Layer Structure, in Turbulence Management and Relaminarisation, eds. H.W. Liepmann and
R. Narasimha, pp. 3761, SpringerVerlag, Berlin.
Sreenivasan, K.R. (1989) The Turbulent Boundary Layer, in Frontiers in Experimental Fluid
Mechanics, ed. M. Gad-el-Hak, pp. 159209, SpringerVerlag, New York.
Sreenivasan, K.R., and Antonia, R.A. (1977) Properties of Wall Shear Stress Fluctuations in a
Turbulent Duct, J. Appl. Mech. 44, pp. 389395.
Sreenivasan, K.R., and White, C.M. (1999) The Onset of Drag Reduction by Dilute Polymer
Additives, and the Maximum Drag Reduction Asymptote, J. Fluid Mech., to appear.
Sreenivasan, K.R., Ramshankar, R., and Meneveau, C. (1989) Mixing, Entrainment and Fractal
Dimensions of Surfaces in Turbulent Flows, Proc. R. Soc. London A 421, pp. 79108.
Sritharan, S.S., editor (1998) Optimal Control of Viscous Flow, SIAM, Philadelphia, Pennsylvania.
Stanewsky, E., and Krogmann, P. (1985) Transonic Drag Rise and Drag Reduction by Active/
Passive Boundary Layer Control, in Aircraft Drag Prediction and Reduction, AGARD R-723,
pp. 11.111.41, AGARD, NATO, Rhode-Saint-G en` ese, Belgium.
Staniforth, R. (1958) Some Tests on Cascades of Compressor Blades Fitted with Vortex Gener-
ators, NGTE Memorandum No. M.314, National Gas Turbine Establishment, Farnborough,
Great Britain. (Also Aeronautical Research Council, CP-487, London, Great Britain.)
BIBLIOGRAPHY 409
Steinheil, E., Scherber, W., Seidl, M., and Rieger, H. (1977) Investigations on the Interaction of
Gases and Well-Dened Solid Surfaces with Respect to Possibilities for Reduction of Aerodynamic
Friction and Aerothermal Heating, in Rareed Gas Dynamics, ed. J.L. Potter, pp. 589602,
AIAA, New York.
Stollery, J.L., and Dyer, D.J. (1989) Wing-Section Effects on the Flight Performance of a Remotely
Piloted Vehicle, J. Aircraft 26, pp. 932938.
Stratford, B.S. (1959a) The Prediction of Separation of the Turbulent Boundary Layer, J. Fluid
Mech. 5, pp. 116.
Stratford, B.S. (1959b) An Experimental Flow with Zero Skin Friction Throughout Its Region of
Pressure Rise, J. Fluid Mech. 5, pp. 1735.
Strazisar, A.J., Reshotko, E., and Prahl, J.M. (1977) Experimental Study of the Stability of Heated
Laminar Boundary Layers in Water, J. Fluid Mech. 83, pp. 225247.
Strykowski, P.J., and Niccum, D.L. (1991) The Stability of Countercurrent Mixing Layers in
Circular Jets, J. Fluid Mech. 227, pp. 309343.
Strykowski, P.J., and Sreenivasan, K.R. (1990) On the Formation and Suppression of Vortex
Shedding at Low Reynolds Numbers, J. Fluid Mech. 218, pp. 71107.
Stuart, J.T. (1963) Hydrodynamic Stability, in Laminar Boundary Layer Theory, ed. L. Rosen-
head, pp. 492579, Clarendon Press, Oxford, Great Britain.
Stull, F.D., and Velkoff, H.R. (1975) Flow Regimes in Two-Dimensional Ribbed Diffusers,
Swanson, W.M. (1961) The Magnus Effect: A Summary of Investigations to Date, J. Basic Eng.
83, pp. 461470.
Swearingen, J.D., and Blackwelder, R.F. (1984) Instantaneous Streamwise Velocity Gradients in
the Wall Region, Bull. Am. Phys. Soc. 29, p. 1528.
Swearingen, J.D., and Blackwelder, R.F. (1987) The Growth and Breakdown of Streamwise Vor-
tices in the Presence of a Wall, J. Fluid Mech. 182, pp. 255290.
Swift, J. (1726) Gullivers Travels, 1906 reprinting, J.M. Dent and Company, London, Great Britain.
Szodruch, J., and Schneider, H. (1989) High Lift Aerodynamics for Transport Aircraft by Interac-
tive Experimental and Theoretical Tool Development, AIAA Paper No. 89-0267, New York.
Tai, Y.-C. (1993) Silicon Micromachining and Micromechanics, invited oral presentation at AIAA
Third Flow Control Conference, 69 July, Orlando, Florida.
Tai, Y.-C., and Muller, R.S. (1989) IC-Processed Electrostatic Synchronous Micromotors, Sen-
sors & Actuators 20, pp. 4955.
Tam, C.K.W. (1995) Supersonic Jet Noise, Ann. Rev. Fluid Mech. 271, pp. 1743.
Tang, J., and Bau, H.H. (1993a) Stabilization of the No-Motion State in Rayleigh-B enard Con-
vection Through the Use of Feedback Control, Phys. Rev. Lett. 70, pp. 17951798.
Tang, J., and Bau, H.H. (1993b) Feedback Control Stabilization of the No-Motion State of a
Fluid Conned in a Horizontal Porous Layer Heated from Below, J. Fluid Mech. 257, pp. 485
505.
Tang, W.C., Nguyen, T.-C., and Howe, R.T. (1989) Laterally Driven Polysilicon Resonant Mi-
crostructures, Sensors & Actuators 20, pp. 2532.
Tani, I. (1964) Low-Speed Flows Involving Bubble Separations, in Prog. Aeronautical Sci. 5,
eds. D. K uchemann and L.H.G. Sterne, pp. 70103, Pergamon Press, New York.
Tani, I. (1969) Boundary-Layer Transition, Ann. Rev. Fluid Mech. 1, pp. 169196.
Tani, I. (1986) Some Equilibrium Turbulent Boundary Layers, Fluid Dyn. Res. 1, pp. 4958.
Tani, I. (1987) Turbulent Boundary Layer Development over Rough Surfaces, in Perspectives in
Turbulent Studies, eds. H.U. Meier and P. Bradshaw, pp. 223249, SpringerVerlag, Berlin.
Tani, I. (1988) Drag Reduction by Riblet Viewed as Roughness Problem, Proc. Japan Acad. B
64, pp. 2124.
Tani, I., and Motohashi, T. (1985a) Non-Equilibrium Behavior of Turbulent Boundary Layer
Flows. Part I. Method of Analysis, Proc. Japan Acad. B 61, pp. 333336.
Tani, I., and Motohashi, T. (1985b) Non-Equilibrium Behavior of Turbulent Boundary Layer
Flows. Part II. Results of Analysis, Proc. Japan Acad. B 61, pp. 337340.
Tavella, D.A., Lee, C.S., and Wood, N.J. (1986) Inuence of Wing Tip Conguration on Lateral
Blowing Efciency, AIAA Paper No. 86-0475, New York.
Tavella, D.A., Wood, N.J., Lee, C.S., and Roberts, L. (1986) Two Blowing Concepts for Roll and
Lateral Control of Aircraft, Department of Aeronautics and Astronautics Report No. TR-75,
410 BIBLIOGRAPHY
Tavella, D.A., Wood, N.J., Lee, C.S., and Roberts, L. (1988) Lift Modulation with Lateral Wing-
Tip Blowing, J. Aircraft 25, pp. 311316.
Taylor, G. (1972) Low-Reynolds-Number Flows, in Illustrated Experiments in Fluid Mecha-
nics, pp. 4754, National Committee for Fluid Mechanics Films, MIT Press, Cambridge,
Massachusetts.
Taylor, G.I. (1923) Stability of a Viscous Liquid Contained between Two Rotating Cylinders,
Phil. Trans. R. Soc. London A 223, pp. 289343.
Taylor, G.I. (1935) Statistical Theory of Turbulence, Proc. Roy. Soc. London A151, pp. 421478.
Taylor, H.D. (1948a) Application of Vortex Generator Mixing Principles to Diffusers, Research
Department Concluding Report No. R-15064-5, United Aircraft Corporation, East Hartford,
Connecticut.
Taylor, H.D. (1948b) Design Criteria for and Applications of the Vortex Generator Mixing Princi-
ple, Research Department Report No. M-15038-1, United Aircraft Corporation, East Hartford,
Connecticut.
Telionis, D.P. (1979) ReviewUnsteady Boundary Layers, Separated and Attached, J. Fluids Eng.
101, pp. 2943.
Telionis, D.P., and Werle, M.J. (1973) Boundary-Layer Separation from Downstream Moving
Boundaries, J. Appl. Mech. 40, pp. 369374.
Tennant, J.S. (1973) A Subsonic Diffuser with Moving Walls for Boundary Layer Control, AIAA
J. 11, pp. 240242.
Tennant, J.S., Johnson, W.S., Keanton, D.D., and Krothapalli, A. (1975) The Application of
Moving Wall Boundary Layer Control to Submarine Control Surfaces, University of Tennessee
Report No. MAE-75-01210-1, Knoxville, Tennessee. (Also available from U.S. NTIS; Document
Number AD-AO23536.)
Tennekes, H., and Lumley, J.L. (1972) A First Course in Turbulence, MIT Press, Cambridge,
Massachusetts.
Theodorsen, Th. (1952) Mechanism of Turbulence, in Proc. Second Midwestern Conf. on Fluid
Mechanics, pp. 118, Ohio State University, Columbus, Ohio.
Theodorson, Th. (1955) The Structure of Turbulence, in 50 Jahre Grenzschichtsforschung
(Ludwig Prandtl Anniversal Volume), eds. H. G ortler and W. Tollmien, pp. 5562, Friedr. Vieweg
und Sohn, Braunschweig, Germany.
Thibert, J.J., Reneaux, J., and Schmitt, V. (1990) ONERA Activities on Drag Reduction, in
Proc. Seventeenth Cong. of the International Council of the Aeronautical Sciences, vol. 1, paper
no. 90-3.6.1, pp. 10531064, ICAS, Washington, D.C.
Thieme, H. (1997) Lower Palaeolithic Hunting Spears from Germany, Nature 385, 27 February,
p. 807.
Thomas, A.S.W. (1983) The Control of Boundary-Layer Transition Using a Wave Superposition
Principle, J. Fluid Mech. 137, pp. 233250.
Thomas, A.S.W., and Bull, M.K. (1983) On the Role of Wall-Pressure Fluctuations in Deterministic
Motions in the Turbulent Boundary Layer, J. Fluid Mech. 128, pp. 283322.
Thomas, F.O., Liu, X., and Nelson, R.C. (1997) Experimental Investigation of the Conuent
Boundary Layer of a High-Lift System, AIAA Paper No. 97-1934, Washington, D.C.
Thomas, F.O., Nelson, R.C., and Liu, X. (1999) The Conuent Boundary Layer of a High-Lift
System: Experiments and Flow Control Strategies, AIAA J., to appear.
Thomas, L.B., and Lord, R.G. (1974) Comparative Measurements of Tangential Momentum and
Thermal Accommodations on Polished and on Roughened Steel Spheres, in Rareed Gas Dy-
namics 8, eds. K. Karamcheti, Academic Press, New York.
Thomas, M.D. (1992a) On the Resonant Triad Interaction in Flows over Rigid and Flexible Bound-
aries, J. Fluid Mech. 234, pp. 417442.
Thomas, M.D. (1992b) On the Nonlinear Stability of Flows over Compliant Walls, J. Fluid Mech.
239, pp. 657670.
Thompson, P.A., and Robbins, M.O. (1989) Simulations of Contact Line Motion: Slip and the
Dynamic Contact Line, Nature 389, 25 September, pp. 360362.
Thompson, P.A., and Troian, S.M. (1997) A General Boundary Condition for Liquid Flowat Solid
Surfaces, Phys. Rev. Lett. 63, pp. 766769.
Thwaites, B. (1949) Approximate Calculations of the Laminar Boundary Layer, Aero. Quart. 1,
pp. 245280.
BIBLIOGRAPHY 411
Tichy, J., Warnaka, G.E., and Poole, L.A. (1984) A Study of Active Control of Noise in Ducts,
J. Vibration, Acoustics, Stress, and Reliability in Design 106, pp. 399404.
Tiederman, W.G., Luchik, T.S., and Bogard, D.G. (1985) Wall-Layer Structure and Drag Reduc-
tion, J. Fluid Mech. 156, pp. 419437.
Tien, N.C. (1997) Silicon Micromachined Thermal Sensors and Actuators, Microscale Thermo-
phys. Eng. 1, pp. 275292.
Tison, S.A. (1993) Experimental Data and Theoretical Modeling of Gas Flows Through Metal
Capillary Leaks, Vacuum 44, pp. 11711175.
Tobak, M., and Peake, D.J. (1982) Topology of Three-Dimensional Separated Flows, Ann. Rev.
Tokhi, M.O., and Leitch, R.R. (1992) Active Noise Control, Clarendon Press, Oxford, Great Britain.
Tollmien, W. (1935) Ein allgemeines Kriterium der Instabilit at laminarer Geschwindigkeits-
verteilungen, Nachr. Wiss. Fachgruppe G ottingen, Math.-Phys. Klasse 1, pp. 79114.
Toms, B.A. (1948) Some Observations on the Flow of Linear Polymer Solutions through Straight
Tubes at Large Reynolds Numbers, in Proc. First Int. Cong. Rheol., vol. 2, pp. 135141, North-
Holland, Amsterdam.
Townsend, A.A. (1956) The Structure of Turbulent Shear Flow, Cambridge University Press,
Townsend, A.A. (1961) Equilibrium Layers and Wall Turbulence, J. Fluid Mech. 11, pp. 97120.
Townsend, A.A. (1970) Entrainment and the Structure of Turbulent Flow, J. Fluid Mech. 41,
pp. 1346.
Townsend, A.A. (1976) The Structure of Turbulent Shear Flow, second edition, Cambridge Univer-
sity Press, Cambridge, Great Britain.
Tritton, D.J. (1988) Physical Fluid Dynamics, second edition, Clarendon Press, Oxford, Great
Britain.
Truckenbrodt, E. (1956) Ein einfaches N aherungsverfahren zum Berechnen der laminaren Rei-
bungsschicht mit Absaugung, Forschg. Ing.-Wes. 22, pp. 147157.
Tsahalis, D. Th., and Telionis, D.P. (1974) Oscillating Laminar Boundary Layers and Unsteady
Separation, AIAA J. 12, pp. 14691476.
Tsai, H.M., and Leslie, D.C. (1990) Large Eddy Simulation of a Developing Turbulent Boundary
Layer at Low Reynolds Number, Int. J. Num. Methods in Fluids 10, pp. 519555.
Tsao, T., Jiang, F., Miller, R.A., Tai, Y.-C., Gupta, B., Goodman, R., Tung, S., and Ho, C.-M.
(1997) An Integrated MEMS System for Turbulent Boundary Layer Control, in Technical
Digest (Transducers 97), vol. 1, pp. 315318.
Tu, B.J., and Willmarth, W.W. (1966) An Experimental Study of the Structure of Turbulence
Near the Wall Through Correlation Measurements in a Thick Turbulent Boundary Layer, De-
partment of Aerospace Engineering Report No. 02920-3-T, University of Michigan, Ann Arbor,
Michigan.
Tuckermann, D.B. (1984) Heat Transfer Microstructures for Integrated Circuits, Ph.D. Thesis,
Tuckermann, D.B., and Pease, R.F.W. (1981) High-Performance Heat Sinking for VLSI, IEEE
Electron Device Lett. EDL-2, no. 5, May.
Tuckermann, D.B., and Pease, R.F.W. (1982) Optimized Convective Cooling Using Micromachined
Structures, J. Electrochem. Soc. 129, no. 3, C98, March.
Ueda, H., and Hinze, J.O. (1975) Fine Structure Turbulence in the Wall Region of a Turbulent
Boundary Layer, J. Fluid Mech. 67, pp. 125143.
Ueda, H., and Mizushina, T. (1979) Turbulence Structure in the Inner Part of the Wall Region
in a Fully Developed Turbulent Flow, in Proc. Fifth Biennial Symp. on Turbulence, eds. G.K.
Patterson and J.L. Zakin, pp. 357366, Science Press, Princeton, New Jersey.
Ulrich, A. (1944) Theoretische Untersuchungen uber die Widerstandsersparnis durch Laminarhal-
tung mit Absaugung, Schriften Dtsch. Akad. Luftfahrtforschung B 8, p. 53.
Vakili, A.D. (1990) Review of Vortical Flow Utilization, AIAA Paper No. 90-1429, New York.
Vakili, A.D., Wu, J.M., and Bhat, M.K. (1988) High Angle of Attack Aerodynamics of Excitation
of the Locked Leeside Vortex, Society of Automotive Engineers Paper No. SAE-88-1424, New
York.
Vakili, A.D., Wu, J.M., Liver, P., and Bhat, M.K. (1985) Flow Control in a Diffusing S-Duct,
412 BIBLIOGRAPHY
Van den Berg, H.R., Seldam, C.A., and Gulik, P.S. (1993) Compressible Laminar Flow in a Cap-
illary, J. Fluid Mech. 246, pp. 120.
Van Dyke, M. (1964) Perturbation Methods in Fluid Mechanics, Academic Press, New York.
Van Ingen, J.L. (1956) A Suggested Semiempirical Method for the Calculation of the Boundary-
Layer Transition Region, Department of Aerospace Engineering Report No. V.T.H.74, Institute
of Technology, Delft, The Netherlands.
Van Ingen, J.L., and Boermans, L.M.M. (1986) Aerodynamics at Low Reynolds Numbers: A
Review of Theoretical and Experimental Research at Delft University of Technology, in Proc.
Int. Conf. on Aerodynamics at Low Reynolds Numbers, vol. I, pp. 1.11.40, Royal Aeronautical
Society, London, Great Britain.
Van Laere, L., and Sas, P. (1988) Principles and Applications of Active Noise Cancellation,
in Noise Control Design: Methods and Practice (Noise-Con 88 Proceedings), ed. J.S. Bolton,
pp. 279284, Noise Control Foundation, Poughkeepsie, NY.
Van Lintel, H.T.G., Van de Pol, F.C.M., and Bouwstra, S. (1988) APiezoelectric Micropump Based
on Micromachining of Silicon, Sensors & Actuators 15, pp. 153167.
Vargo, S.E., and Muntz, E.P. (1996) A Simple Micromechanical Compressor and Vacuum Pump
for Flow Control and Other Distributed Applications, AIAA Paper No. 96-0310, AIAA,
Washington, D.C.
Vasilyev, O.V., Yuen, D.A., and Paolucci, S. (1997) Solving PDEs Using Wavelets, Computers in
Physics 11, no. 5, pp. 429435.
Verollet, E., Fulachier, L., Dumas, R., and Favre, A. (1972) Turbulent Boundary Layer with Suction
and Heating to the Wall, in Heat and Mass Transfer in Boundary Layers, eds. N. Afgan, Z. Zaric,
and P. Anastasijevec, vol. 1, pp. 157168, Pergamon Presss, Oxford.
Vidal, R.J. (1959) Research on Rotating Stall in Axial-Flow Compressors: Part IIIExperiments
on Laminar Separation from a Moving Wall, Wright Air Development Center Technical Report
No. 59-75, Wright-Patterson Air Force Base, Ohio.
Viets, H. (1980) Coherent Structures in Time Dependent Shear Flows, in Turbulent Boundary
Layers, AGARD CPP-271, paper no. 5, AGARD, NATO, Neuilly-sur-Seine, France.
Viets, H., Ball, M., and Bougine, D. (1981) Performance of Forced Unsteady Diffusers, AIAA
Viets, H., Palmer, G.M., and Bethke, R.J. (1984) Potential Applications of Forced Unsteady Flows,
in Unsteady Separated Flows, eds. M.S. Francis and M.W. Luttges, pp. 2127, U.S. Air Force
Academy, Colorado Springs, Colorado.
Viets, H., Piatt, M., and Ball, M. (1979) Unsteady Wing Boundary Layer Energization, AIAA
Viets, H., Piatt, M., and Ball, M. (1981a) Forced Vortex Near a Wall, AIAA Paper No. 81-0256,
New York.
Viets, H., Piatt, M., and Ball, M. (1981b) Boundary Layer Control by Unsteady Vortex Genera-
tion, J. Wind Eng. & Industrial Aerodynamics 7, pp. 135144.
Vijgen, P.M.H.W., van Dam, C.P., Holmes, B.J., and Howard, F.G. (1989) Wind-Tunnel Investi-
gations of Wings with Serrated Sharp Trailing Edges, in Low Reynolds Number Aerodynamics,
ed. T.J. Mueller, pp. 295313, SpringerVerlag, Berlin.
Vincenti, W.G., and Kruger, C.H., Jr. (1965) Introduction to Physical Gas Dynamics, Wiley, New
York.
Virk, P.S. (1975) Drag Reduction Fundamentals, AIChE J. 21, pp. 625656.
Virk, P.S., Merrill, E.W., Mickley, H.S., Smith, K.A., and Mollo-Christensen, E.L. (1967) The
Toms Phenomenon: Turbulent Pipe Flow of Dilute Polymer Solutions, J. Fluid Mech. 30,
pp. 305328.
Viswanath, P.R. (1988) Shockwave-Turbulent Boundary Layer Interaction and Its Control: A
Survey of Recent Developments, S adh an a 12, pp. 45104.
Viswanath, P.R. (1995) Flow Management Techniques for Base and Afterbody Drag Reduction,
Von Doenhoff, A.E. (1938a) A Method of Rapidly Estimating the Position of the Laminar Sepa-
ration Point, NACA TN No. 671, Washington, D.C.
Von Doenhoff, A.E. (1938b) A Preliminary Investigation of Boundary-Layer Transition Along a
Flat Plate with Adverse Pressure Gradient, NACA TN No. 639, Washington, D.C.
Von K arm an, Th. (1930) Mechanische

Ahnlichkeit und Turbulenz, Nachr. Ges. Wiss. G ottingen,
Math.-Phys. Klasse, pp. 5876.
BIBLIOGRAPHY 413
Von Smoluchowski, M. (1898)
Uber W armeleitung in verd unnten Gasen, Annalen der Physik

und Chemie 64, pp. 101130.
Von Winkle, W.A. (1961) An Evaluation of a Boundary Layer Stabilization Coating, Naval
Underwater Systems Center, Tech. Memo. No. 922-111-61, New London, Connecticut.
Wadsworth, D.C., Muntz, E.P., Blackwelder, R.F., and Shiett, G.R. (1993) Transient Energy
Release Pressure Driven Microactuators for Control of Wall-Bounded Turbulent Flows, AIAA
Paper No. 93-3271, AIAA, Washington, D.C.
Waggoner, E.G., and Allison, D.O. (1987) EA-6B High Lift Wing Modications, AIAA Paper
No. 87-2360-CP, New York.
Wagner, R.D., and Fischer, M.C. (1984) Fresh Attack on Laminar Flow, Aerospace America 22,
March, pp. 7276.
Wagner, R.D., Bartlett, D.W., and Maddalon, D.V. (1988) Laminar Flow Control is Maturing,
Aerospace America 26, January, pp. 2024.
Wagner, R.D., Fischer, M.C., Collier, F.S., Jr., and Pfenninger, W. (1990) Supersonic Laminar Flow
Control on Commercial Transports, in Proc. Seventeenth Congress of the International Council
of the Aeronautical Sciences, vol. 1, pp. 10731089, paper no. 90-3.6.3, ICAS, Washington, D.C.
Wagner, R.D., Maddalon, D.V., and Fischer, M.C. (1984) Technology Development for Laminar
Boundary Control on Subsonic Transport Aircraft, AGARD CP-365, paper no. 16, AGARD,
NATO, Rhode-Saint-G en` ese, Belgium.
Walker, D.T., and Tiederman, W.G. (1990) Turbulent Structure in a Channel Flow with Polymer
Injection at the Wall, J. Fluid Mech. 218, pp. 377403.
Walker, J.D.A. (1978) The Boundary Layer due to Rectilinear Vortex, Proc. R. Soc. London A
359, pp. 167188.
Walker, J.D.A., and Herzog, S. (1988) Eruption Mechanisms for Turbulent Flows Near Walls, in
Transport Phenomena in Turbulent Flows, eds. M. Hirata and N. Kasagi, pp. 145156, Hemi-
sphere, New York.
Wallis, R.A., and Stuart, C.M. (1958) On the Control of Shock Induced Boundary Layer Separa-
tion with Discrete Jets, Aeronautical Research Council Current Paper No. 494, London, Great
Britain.
Walsh, M.J. (1980) Drag Characteristics of V-Groove and Transverse Curvature Riblets, in Vis-
cous Flow Drag Reduction, ed. G.R. Hough, pp. 168184, AIAA, New York.
Walsh, M.J. (1982) Turbulent Boundary Layer Drag Reduction Using Riblets, AIAA Paper
Walsh, M.J. (1983) Riblets as a Viscous Drag Reduction Technique, AIAA J. 21, pp. 485486.
Walsh, M.J. (1990) Riblets, in Viscous Drag Reduction in Boundary Layers, eds. D.M. Bushnell
and J.N. Hefner, pp. 203261, AIAA, Washington, D.C.
Walsh, M.J., and Lindemann, A.M. (1984) Optimization and Application of Riblets for Turbulent
Drag Reduction, AIAA Paper No. 84-0347, New York.
Walsh, M.J., Sellers, W.L., III, and McGinley, C.B. (1989) Riblet Drag at Flight Conditions,
Walsh, M.J., and Weinstein, M. (1978) Drag and Heat Transfer on Surfaces with Small Longitu-
dinal Fins, AIAA Paper No. 78-1161, New York.
Wang, Y., Singer, J., and Bau, H.H. (1992) Controlling Chaos in a Thermal Convection Loop,
J. Fluid Mech. 237, pp. 479498.
Wannier, G.H. (1950) A Contribution to the Hydrodynamics of Lubrication, Quart. Appl. Math.
8, pp. 132.
Wark, C.E., and Nagib, H.M. (1991) Experimental Investigation of Coherent Structures in Tur-
bulent Boundary Layers, J. Fluid Mech. 230, pp. 183208.
Warkentin, D.J., Haritonidis, J.H., Mehregany, M., and Senturia, S.D. (1987) A Micromachined
Microphone with Optical Interference Readout, Proc. Fourth Int. Conf. on Solid-State Sensors
and Actuators (Transducers 87), June, Tokyo, Japan.
Warnaka, G.E. (1982) Active Attenuation of Noise: The State of the Art, Noise Control Eng. 18,
pp. 100110.
Warner, J.V., Waters, D.E., and Bernhard, R.J. (1988) Adaptive Active Noise Control in Three
Dimensional Enclosures, in Noise Control Design: Methods and Practice (Noise-Con 88 Pro-
ceedings), ed. J.S. Bolton, pp. 285290, Noise Control Foundation, Poughkeepsie, New York.
Wazzan, A.R., Okamura, T.T., and Smith, A.M.O. (1968) The Stability of Water Flowover Heated
and Cooled Flat Plates, J. Heat Transfer 90, pp. 109114.
414 BIBLIOGRAPHY
Wazzan, A.R., Okamura, T.T., and Smith, A.M.O. (1970) The Stability and Transition of Heated
and Cooled Incompressible Boundary Layers, in Proc. Fourth Int. Heat Transfer Conf., eds.
U. Grigull and E. Hahne, vol. 2, FC 1.4, Elsevier, New York.
Webb, R.L. (1987) Enhancement of Single-Phase Heat Transfer, in Handbook of Single-Phase
Convective Heat Transfer, eds. S. Kakac, R.K. Shah, and W. Aung, pp. 17.117.62, Wiley
Interscience, New York.
Webb, R.L. (1993) Principles of Enhanced Heat Transfer, Wiley, New York.
Webb, R.L., Camavos, T.C., Park, E.L., Jr., and Hostetler, K.M., editors (1981) Advances in En-
hanced Heat Transfer, ASME HTD-18, New York.
Wei, T., and Willmarth, W.W. (1989) Reynolds-Number Effects on the Structure of a Turbulent
Channel Flow, J. Fluid Mech. 204, pp. 5795.
Weiberg, J.A., Giulianettij, D., Gambucci, B., and Innis, R.C. (1973) Takeoff and Landing Perfor-
mance and Noise Characteristics of a Deected STOL Airplane with Interconnected Propellers
and Rotating Cylinder Flaps, NASA TM X-62,320, Washington, D.C.
Weis-Fogh, T. (1973) Quick Estimates of Flight Fitness in Hovering Animals, Including Novel
Mechanisms for Lift Production, J. Exp. Biol. 59, pp. 169230.
Wells, C.S., editor (1969) Viscous Drag Reduction, Plenum Press, New York.
Wells, C.S., Jr., and Spangler, J.G. (1967) Injection of a Drag-Reducing Fluid into Turbulent Pipe
Flow of a Newtonian Fluid, Phys. Fluids 10, pp. 18901894.
Went, F.W. (1968) The Size of Man, American Scientist 56, pp. 400413.
Werle, M.J., Paterson, R.W., and Presz, W.M., Jr. (1987) Trailing-Edge Separation/Stall Alleviation,
AIAA J. 25, pp. 624626.
Wheeler, G.O. (1984) Means for Maintaining Attached Flow of a Flow Medium, U.S. Patent
No. 4,455,045.
Whitcomb, R.T. (1956) A Study of the Zero-Lift Drag-Rise Characteristics of Wing-Body Combi-
nations Near the Speed of Sound, NACA Report No. 1273, Washington, D.C.
White, A., and Hemmings, J.A.G. (1976) Drag Reduction by Additives: Reviewand Bibliography,
BHRA Fluid Engineering, Craneld, Great Britain.
Whitehead, A.H., and Bertram, M.H. (1971) Alleviation of Vortex-Induced Heating to the Lee
Side of Slender Wings in Hypersonic Flow, AIAA J. 9, pp. 18701872.
Whites, R.C., Sudderth, R.W., and Wheldon, W.G. (1966) Laminar Flow Control on the X-21,
Astronautics & Aeronautics 4, pp. 3843.
Wickerhauser, M.V. (1994) Adapted Wavelet Analysis from Theory to Software, A.K. Peters Ltd.,
Wellesley, Massachusetts.
Wieghardt, K. (1943)
Uber die Wandschubspannung in turbulenten Reibungsschichten bei ver a

nderlichem Aussendruck, Kaiser Wilhelm Institut f ur Str omungsforschung, no. U&M-6603,
G ottingen, Germany.
Wilcox, D.C. (1993) Turbulence Modeling for CFD, DCW Industries, Los Angeles, California.
Wilkinson, S.P. (1988) Direct Drag Measurements on Thin-Element Riblets with Suction and
Blowing, AIAA Paper No. 88-3670-CP, Washington, D.C.
Wilkinson, S.P. (1990) Interactive Wall Turbulence Control, in Viscous Drag Reduction in Bound-
ary Layers, eds. D.M. Bushnell and J.N. Hefner, pp. 479509, AIAA, Washington, D.C.
Wilkinson, S.P., Anders, J.B., Lazos, B.S., and Bushnell, D.M. (1987) Turbulent Drag Reduction
Research at NASALangley-Progress and Plans, in Proc. Int. Conf. on Turbulent Drag Reduction
by Passive Means, vol. 1, pp. 132, Royal Aeronautical Society, London, Great Britain.
Wilkinson, S.P., Anders, J.B., Lazos, B.S., and Bushnell, D.M. (1988) Turbulent Drag Reduc-
tion Research at NASA Langley: Progress and Plans, Int. J. Heat and Fluid Flow 9, pp. 266
277.
Wilkinson, S.P., and Balasubramanian, R. (1985) Turbulent Burst Control Through Phase-Locked
Surface Depressions, AIAA Paper No. 85-0536, New York.
Wilkinson, S.P., and Lazos, B.S. (1987) Direct Drag and Hot-Wire Measurements on Thin-Element
Riblet Arrays, in Turbulence Management and Relaminarization, eds. H.W. Liepmann and
R. Narasimha, pp. 121131, SpringerVerlag, New York.
Williams, D.R., and Amato, C.W. (1989) Unsteady Pulsing of Cylinder Wakes, in Frontiers in
Experimental Fluid Mechanics, ed. M. Gad-el-Hak, pp. 337364, SpringerVerlag, New York.
Williams, D.R., and Papazian, H. (1991) Forebody Vortex Control with the Unsteady Bleed Tech-
nique, AIAA J. 29, pp. 853855.
BIBLIOGRAPHY 415
Williams, J.C. (1985) Singularities in Solutions of the Three-Dimensional Laminar-Boundary-Layer
Equations, J. Fluid Mech. 160, pp. 257279.
Williams, J.C., III (1977) Incompressible Boundary-Layer Separation, Ann. Rev. Fluid Mech. 9,
pp. 113144.
Williams, J.C., III, and Johnson, W.D. (1974a) Semisimilar Solutions to Unsteady Boundary-Layer
Flows Including Separation, AIAA J. 12, pp. 13881393.
Williams, J.C., III, and Johnson, W.D. (1974b) Note on Unsteady Boundary-Layer Separation,
AIAA J. 12, pp. 14271429.
Williams, T.I. (1987) The History of Invention, Facts on File Publications, New York.
Willis, G.J.K. (1986) Hydrodynamic Stability of Boundary Layers over Compliant Surfaces, Ph.D.
Thesis, University of Exeter, Exeter, Great Britain.
Willmarth, W.W. (1959) Space-Time Correlations and Spectra of Wall Pressure in a Turbulent
Boundary Layer, NASA Memorandum No. 3-17-59W, Washington, D.C.
Willmarth, W.W. (1975a) Structure of Turbulence in Boundary Layers, Adv. Appl. Mech. 15,
pp. 159254.
Willmarth, W.W. (1975b) Pressure Fluctuations Beneath Turbulent Boundary Layers, Ann. Rev.
Willmarth, W.W., and Bogar, T.J. (1977) Survey and New Measurements of Turbulent Structure
Near the Wall, Phys. Fluids 20, no. 10, part II, pp. S9S21.
Willmarth, W.W., and Sharma, L.K. (1984) Study of Turbulent Structure with Hot Wires Smaller
than the Viscous Length, J. Fluid Mech. 142, pp. 121149.
Willmarth, W.W., and Tu, B.J. (1967) Structure of Turbulence in the Boundary Layer Near the
Wall, Phys. Fluids 10, pp. S134S137.
Wilson, C.E. (1989) Noise Control, Harper & Row, New York.
Wiltse, J.M., and Glezer, A. (1993) Manipulation of Free Shear Flows Using Piezoelectric Actua-
tors, J. Fluid Mech. 249, pp. 261285.
Wimpenny, J.C. (1970) Vortex Generators, U.S. Patent No. 3,525486.
Winant, C.D., and Browand, F.K. (1974) Vortex Pairing: The Mechanism of Turbulent Mixing
Layer Growth at Moderate Reynolds Numbers, J. Fluid Mech. 63, pp. 237255.
Wood, C.J. (1961) A Study of Hypersonic Separated Flow, Ph.D. Thesis, University of London,
London, Great Britain.
Wood, N.J., and Nielsen, J.N. (1985) Circulation Control AirfoilsPast, Present, Future, AIAA
Wood, N.J., and Roberts, L. (1986) Experimental Results of the Control of a Vortical Flow by
Tangential Blowing, Joint Institute for Aeronautics and Acoustics Report No. JIAA TR-71,
Stanford University, Stanford, CA.
Wood, N.J., and Roberts, L. (1988) Control of Vortical Lift on Delta Wings by Tangential Leading-
Edge Blowing, J. Aircraft 25, pp. 236243.
Wood, N.J., Roberts, L., and Celik, Z. (1990) Control of Asymmetric Vortical Flows over Delta
Wings at High Angles of Attack, J. Aircraft 27, pp. 429435.
Wooldridge, C.E., and Muzzy, R.J. (1966) Boundary-Layer Turbulence Measurements with Mass
Addition and Combustion, AIAA J. 4, pp. 20092016.
Wortmann, A. (1987) Alleviation of Fuselage from Drag Using Vortex Flows, Department of
Energy Report No. DOE/CE/15277-T1, Washington, D.C.
Wu, J., and Tulin, M.P. (1972) Drag Reduction by Ejecting Additive Solutions into a Pure-Water
Boundary Layer, ASME J. Basic Eng. 94, pp. 749756.
Wu, J.M., Vakili, A.D., and Chen, Z.L. (1983) Investigation on the Effects of Discrete Wingtip
Jets, AIAA Paper No. 83-0546, New York.
Wu, J.M., Vakili, A.D., and Gilliam, F.T. (1984) Aerodynamic Interactions of Wingtip Flow with
Discrete Wingtip Jets, AIAA Paper No. 84-2206, New York.
Wuest, W. (1961) Survey of Calculation Methods of Laminar Boundary Layers with Suction
in Incompressible Flow, in Boundary Layer and Flow Control, ed. G.V. Lachmann, vol. 2,
pp. 771800, Pergamon Press, New York.
Wygnanski, I., Sokolov, M., and Friedman, D. (1976) On a Turbulent Spot in a Laminar Boundary
Yager, R.R., Zadeh, L.A., editors (1992) An Intorduction to Fuzzy Logic Applications in Intelligent
Systems, Kluwer Academic, Boston, Massachusetts.
416 BIBLIOGRAPHY
Yajnik, K.S. (1970) Asymptotic Theory of Turbulent Shear Flows, J. Fluid Mech. 42, pp. 411427.
Yajnik, K.S., and Acharya, M. (1978) Non-Equilibrium Effects in a Turbulent Boundary
Layer Due to the Destruction of Large Eddies, in Structure and Mechanisms of Turbulence,
ed. H. Fiedler, vol. 1, pp. 249260, SpringerVerlag, New York.
Yakhot, V., and Orszag, S.A. (1986) Renormalization Group Analysis of Turbulence. I. Basic
Theory, J. Scientif. Comput. 1, pp. 351.
Yang, T., Ntone, F., Jiang, T., and Pitts, D.R. (1984) An Investigation of High Performance, Short
Thrust Augmenting Ejectors, ASME Paper No. 84-WA/FE-10, New York.
Yeo, K.S. (1988) The Stability of Boundary-Layer Flow over Single- and Multi-Layer Viscoelastic
Walls, J. Fluid Mech. 196, pp. 359408.
Yeo, K.S. (1990) The Hydrodynamic Stability of Boundary-Layer Flow over a Class of Anisotrpic
Compliant Walls, J. Fluid Mech. 220, pp. 125160, 1990.
Yeo, K.S. (1992) The Three-Dimensional Stability of Boundary-Layer Flow over Compliant Walls,
J. Fluid Mech. 238, pp. 537577.
Yeo, K.S., and Dowling, A.P. (1987) The Stability of Inviscid Flows over Passive Compliant Walls,
J. Fluid Mech. 183, pp. 265292.
Young, A.D. (1953) Boundary Layers, in Modern Developments in Fluid Dynamics: High Speed
Flow, ed. L. Howarth, vol. 1, pp. 375475, Clarendon Press, Oxford, Great Britain.
Young, A.M., Goldsberry, J.E., Haritonidis, J.H., Smith, R.I., and Senturia, S.D. (1988) A Twin-
Interferometer Fiber-Optic Readout for Diaphragm Pressure Transducers, in IEEE Solid-State
Sensor and Actuator Workshop, 69 June, Hilton Head, South Carolina.
Yun, W. (1993) System Considerations for Integration of Microsensors and Electronics, invited
oral presentation at AIAA Third Flow Control Conference, 69 July, Orlando, Florida.
Zagarola, M.V. (1996) Mean Flow Scaling in Turbulent Pipe Flow, Ph.D. Thesis, Princeton
University, Princeton, New Jersey.
Zagarola, M.V., Smits, A.J., Orszag, S.A., and Yakhot, V. (1996) Experiments in High Reynolds
Number Turbulent Pipe Flow, AIAA Paper No. 96-0654, Washington, D.C.
Zakin, J.L., Poreh, M., Brosh, A., and Warshavsky, M. (1971) Exploratory Study of Friction
Reduction in Slurry Flows, in Chem. Eng. Prog. Symp. Seri., no. 67, vol. 111, pp. 8589,
AIChE, New York.
Zakkay, V., Barra, V., and Hozumi, K. (1980) Turbulent Boundary Layer Structure at Low and
High Subsonic Speeds, in Turbulent Boundary LayersExperiments, Theory and Modelling,
AGARD Conference Proceedings No. 271, pp. 4.14.20, Rhode-Saint-G en` ese, Belgium.
Zaman, K.B.M.Q., Bar-Sever, A., and Mangalam, S.M. (1987) Effect of Acoustic Excitation on
the Flow over a Low Re Airfoil, J. Fluid Mech. 182, pp. 127148.
Zaman, K.B.M.Q., and Hussain, A.K.M.F. (1980) Vortex Pairing in a Circular Jet Under Con-
trolled Excitation. Part 1. General Response, J. Fluids Mech. 101, pp. 449491.
Zaman, K.B.M.Q., and Hussain, A.K.M.F. (1981) Turbulence Suppression in Free Shear Flows by
Controlled Excitation, J. Fluids Mech. 103, pp. 133159.
Zaman, K.B.M.Q., and McKinzie, D.J. (1989) Control of Laminar Separation over Airfoils by
Acoustic Excitation, AIAA Paper No. 89-0565, Washington, D.C. (Also NASA TM-101379,
Washington, D.C.)
Zaman, K.B.M.Q., and Potapczuk, M.G. (1989) The Low Frequency Oscillation in the Flow over
a NACA0012 Airfoil with an Iced Leading Edge, in Proc. Conf. Low Reynolds Number
Aerodynamics, ed. T.J. Mueller, pp. 117128, University of Notre Dame, Notre Dame, Indiana.
Zang, T.A., Hussaini, M.Y., and Bushnell, D.M. (1984) Numerical Computations of Turbulence
Amplication in Shock Wave Interactions, AIAA J. 22, pp. 1322.
Zaric, Z. (1972) Wall Turbulence Studies, Adv. Heat Transfer 8, pp. 285350.
Zhang, F., and Sheng, C. (1987) A Prediction Method for Optimum Velocity Ratio of Air Jet
Vortex Generator, J. Aerospace Power 2, pp. 5560.
Zhuk, V.I., and Ryzhov, O.S. (1980) Formation of Recirculation Zones in the Boundary Layer on
a Moving Surface, Fluid Dyn. 15, pp. 637644.
Zilberman, M., Wygnanski, I., and Kaplan, R.E. (1977) Transitional Boundary Layer Spot in a
Fully Turbulent Environment, Phys. Fluids 20, no. 10, part II, pp. S258S271.
Zukauskas, A.A. (1986) Heat Transfer Augmentation in Single-Phase Flow, in Proc. Eighth
International Heat Transfer Conference, vol. 1, pp. 4757, Hemisphere, Washington, D.C.
Index
acoustic efciency, 258
acoustic excitations, 181
active control, 28, 334
active momentum addition, 178
Alaskan pipeline, 217, 360
antisound, 267
base drag, 209
blind men and the elephant, 34, 42
Boltzmann equation, 293
boundary conditions, 282
higher-order slip, 286
higher-order temperature-jump, 286
bubble ramp, 203
bursting process, 51, 323
bursts, 52, 323
period between, 91
buzz instabilities, 268
chaos control, 343, 345
OGY strategy, 345
chaotic advection, 244, 246
chaotic mixing, 244, 246
egg-carton, 248
heat transfer, 247
mass transfer, 247
Chapman-Enskog expansion, 295
closed, feedback control loop, 29, 334
Coanda effect, 178
coherent structures, 39, 260, 322
denition of, 42
Falco eddies, 46, 49, 323
free-shear ows, 43
history of, 39
large eddies, 7, 46
low-speed streaks, 51, 323
memory effects, 99
near-wall eddies, 46, 323
photographs of, 3651
wall-bounded ows, 44, 46, 90, 323
coiled tubes, 230, 241, 242
chaotic, 247
chemical reaction, 247
mass transfer, 247
more twists, 242
combustion instabilities, 268
commercial submarine, 6, 361
compliant coatings
dispersion curves, 126
dolphins secret, 140
free-surface waves, 125
funding for, 147
future of, 146
instabilities of, 127
linear stability theory, 131
model for, 131
noise suppression, 263
optimization of, 133
practical examples, 139
prior to 1985, 122
turbulent ows over, 141
compressibility, 17, 280
compressible ow equations, 15
computational uid dynamics, 186
continuity equation, 11
continuum approximation, 11, 277
control goals, 25
interrelation between, 26, 196, 259
tradeoffs, 26, 259
control of turbulence, 319
Crabtree criterion, 195
crossow instabilities, 107
crossow vortices, 174, 237
da Vinci, Leonardo, 36
Dean number, 242
delta wings, 183
dipole, 255
direct simulation Monte Carlo, 296
DNA computers, 8
dolphins secret, 140
drag, 207
induced drag, 210
pressure, 207
417
418 INDEX
drag (cont.)
reduction of, 209
skin-friction, 207, 211
wave drag, 210
drag reduction, 209
counterrotating vortices, 223
bers, 217
geometrical modication, 218
LEBU, 218
mechanisms of, 216
polymers, 217
riblets, 218
rough surface, 222
spanwise oscillations, 223
surfactants, 217
synergism, 225
wall jets, 223
wavy walls, 221
dynamical systems theory, 343
Eckert number, 16
energy equation, 13
equations of motion, 11
turbulent ows, 22, 162
eras of ow control, 3
Euler equations, 16
Falco eddies, 46, 49, 323
bers, 217
eld versus laboratory Re, 61
FISI, 128
aps, 182
ow control
classication schemes, 27
denition of, 2
ve eras of, 3
funding for, 4, 6, 357, 360
future of, 4, 6, 357, 360
high Re, 101, 329
suction, 320
trade-offs, 26
turbulent ows, 319
ow visualization
rst picture, 37
gallery of, 3651
ow-induced surface instabilities, 128
uid-structure instabilities, 127
classes of, 127
forces and moments
on body in relative motion, 206
free-shear ows, 30
inectional velocity proles, 104
turbulence enhancement, 239
free-surface waves, 125
funding for research, 4, 6, 147, 357, 360
future of ow control, 4, 6, 357, 360
fuzzy logic, 352, 354
genetic algorithms, 352, 354
Grashof number, 16
Grays paradox, 140
heat transfer enhancement, 240, 241
history of the universe, 5
incompressible ow equations, 16
inectional velocity proles, 104
injection, 213
inner-outer interaction, 54
instability
absolute, 33
convective, 33
Iraq-Turkey pipeline, 217, 360
jet noise, 264
screech tones, 265
shock-associated noise, 265
shock-cell pattern, 265
subsonic ows, 265
supersonic ows, 265
KarhunenLo` eve decomposition, 344
K arm an integral equation, 23
Knudsen number, 275
regimes of, 276
Kolmogorov scales, 7, 308, 325327
Kramer, 121, 122, 129, 130, 140
MHD, 340
laminar ow control, 111
laminar separation, 165, 190, 192
large eddies, 7, 39, 43, 46, 49, 53, 54
LEBU, 218
LennardJones potential, 299
LFC, 111
Liebeck airfoil, 164
lift-to-drag ratio, 164, 189
Lighthills tensor, 257
applied to uid-structure system, 131
Liouville equation, 292
liquid ows
LennardJones potential, 299
MEMS, 296
molecular dynamics simulations, 299
non-Newtonian, 298
slip ow, 299
local shear layer
unsteady separation, 159
log-law
alternatives to, 78
infallibility of, 59
rise and fall of, 76
Lorentz force, 339
low Re airfoils, 191
INDEX 419
low-speed streaks, 51, 323
Lyapunov exponent, 245
Mach number, 18, 32, 275
regimes of, 32
magnetohydrodynamics control, 338
Magnus effect, 169
mass transfer enhancement, 247
mechanisms of drag reduction, 216
MEMS, 271, 349
boundary conditions, 282
compressibility, 280
continuum model, 273, 277
direct simulation Monte Carlo, 296
ow physics, 272
intermolecular forces, 306
liquid ows, 296
manufacturing of, 271
molecular-based models, 289
slip effects, 282284
statistical mechanics, 290
stiction, 304
surface phenomena, 303
surface-to-volume ratio, 272, 303
temperature-jump, 284
thermal creep, 284
turbulence measurements, 308
MHD, 338
K arm an integral equation, 340
momentum equation at wall, 340
reactive control, 341
microactuators, 349
microbearings, 313
microbubbles, 214, 217
microelectromechanical systems, 271, 349
manufacturing of, 271
microfabrication, 349
micropumps, 310
microsensors, 308, 327, 349
microturbines, 312
mixing, 230
as a dynamical system, 234, 244
enhancement, 230, 235
kinematics, 233
mechanisms, 231
molecular transport, 231
range of Reynolds numbers, 230
transition advancement, 235237
turbulence, 235237
mixing enhancement, 230, 235
using chaos, 244, 246
molecular computing, 8, 9, 360
molecular electronics, 8, 9, 360
molecular-based models, 289
moletronics, 8, 9, 360
momentum equation, 12
written at the wall, 20, 161, 340
momentum equation at wall
MHD, 340
momentum-defect law, 68
monopole, 255
MooreRottSears criterion, 151, 154, 155, 157,
158, 161, 170
moving walls, 168, 221
moving-wall separation, 154, 168
natural laminar ow, 111, 198
utility of, 202
neural networks, 352, 355
NLF, 111, 198
noise, 251
coherent structures, 260
exible surface, 258
jets, 264
presence of rigid wall, 258
subsonic ow, 258
supersonic ow, 258
suppression, 259
turbomachinery blades, 266
noise pollution, 251
noise suppression, 259
active, 267
antisound, 267
compliant coatings, 263
jets, 264
method classication, 259
passive devices, 261
reactive, 267
sonars, 260
trade-offs, 259
Nusselt number, 232
open control loop, 29, 334
open, feedforward control loop, 29, 334
particles, 217
passive contaminant, 233
passive control, 28, 334
passive scalar, 233
passive suction, 166
passively transported tracer, 233
P eclet number, 16, 232
polymers, 214, 217
Prandtl
father of ow control, 2, 3
separation criterion, 163
Prandtl number, 18, 232, 233, 281, 285
Prandtl, Ludwig, 2, 3, 39, 59, 67, 113, 153, 154,
165, 166, 169
predetermined control, 28, 334
420 INDEX
pressure drag, 207
displacement effects, 207
due to separation, 207
induced drag, 207
wave drag, 207
proper orthogonal decomposition, 344
turbulence, 236
quadrupole, 256
reactive control, 28, 329, 334
adaptive, 30, 334, 355
dynamical systems-based, 30, 334, 335
EMHD, 341
MHD, 338
neural networks, 355
optimal, 30, 334, 335
physical-model-based, 30, 334
required sensors and actuators, 336
relaminarization, 223
acceleration, 224
dissipation, 224
favorable p-gradient, 225
heating or cooling, 225
suction, 225
work done against external force, 224
Reynolds number, 16, 31, 62, 232, 275
ying, range of, 189
low range of, 190
range for mixing problems, 230
regimes of, 31
turbulence, 235
Reynolds number effects, 80, 81, 102
ow control, 98
Reynolds stress, 85
peak location, 87
Reynolds, Osborne, 38
Reynolds-averaged equations, 22
riblets, 218
scales of turbulence, 324
screech tones, 265
selective suction, 331
sensors
requirements for turbulent ows, 327
separation
control of, 151, 197
criterion for, 163
MooreRottSears criterion, 170
moving-wall, 154
phenomenon of, 150
provocation, 183
steady, 153
three-dimensional, 154, 172
turbulent boundary layer, 158
unsteady, 154, 170
vortex-induced, 158
separation bubble, 165, 190192
breakdown of, 194
bursting of, 194
conditions for formation, 192
Crabtree criterion, 195
short, 193
TaniOwenKlanfer criterion, 195
separation control, 151, 197
acoustic excitations, 181
active, 178
blowing, 178
aps, 182
future of, 188
injection, 178
moving-walls, 168
passive suction, 166
periodic forcing, 182
shaping, 163, 197, 198
suction, 165
transpiration, 165
tripping, 178
turbulators, 173
velocity prole modiers, 163
vortex generators, 176
wall heating/cooling, 168
wall jets, 179
separation criterion, 163
separation in turbulent ow, 158
separation provocation, 183
shaping, 114, 163
Sherwood number, 232
shock-associated noise, 265
shock-cell pattern, 265
skin-friction
injection, 213
injection of foreign substance, 214
reduction of, 211, 212
reduction of near-wall momentum,
212
turbulent ows, 212
skin-friction reduction
mechanisms of, 216
skycar, 6, 361
slip-ow, 282284
softcomputing, 352
sonars, 260
signal-to-noise ratio, 260
sound
dipole, 255
equation of propagation, 254
levels of, 251
Lighthills tensor, 257
monopole, 255
pseudosound, 255
quadrupole, 256
INDEX 421
sources of, 251, 254
speed of, 253
spaceplane, 6, 361
stability modiers, 110
shaping, 114
suction, 112
wall heating or cooling, 116
wall motion, 111
steady separation, 153
stiction, 304
Stone Age, 3, 360
Stradford closure, 164
Stratford closure, 213
streamlined spears, 3, 360
suction, 112, 165, 320
surfactants, 214, 217
swept-back wings, 360
synergism, 225
Taming of the Shrew, 318, 319, 357
TaniOwenKlanfer criterion, 195
targeted control, 329, 330
TaylorG ortler vortices, 107, 174, 237
thermal creep, 284
three-dimensional separation, 154,
172
TollmienSchlichting instabilities, 128
tracer, 233
trade-offs, 26, 252, 259, 260
transition, 105
advancement, 235237
critical Reynolds number, 109
delay, 104, 110
routes to, 105
stability modiers, 110
wave cancellation, 118
transition advancement
free-shear ows, 237
wall-bounded ows, 236
transition delay, 104, 110
transpiration, 165
TSI, 128
turbulators, 173, 202
turbulence
as a dynamical system, 40, 343
computer requirement, 7, 21
degrees of freedom, 345
equations of motion, 22
scales of, 324
statistical view of, 39
taming of the shrew, 318, 357
the last conundrum, 40
turbulence enhancement, 238, 239
wall-bounded ows, 239
turbulence mixing noise, 265
turbulent boundary layer
asymptotic state, 80
comparison to free-shear ows, 69
constant-stress region, 66
ow regimes, 63
higher-order statistics, 82
log-law, 74
mean ow, 73
outer region, 68
power-law, 78
Reynolds stress, 85
root-mean-square, 83
self-preservation, 81
separation of, 158
viscous region, 64
turbulent ow separation, 158
typical eddies, 49, 323
universal equilibrium theory, 325
universal wake function, 69
unsteady separation, 154, 170
archetypes of, 156
local shear layer, 159
velocity prole modiers, 163
velocity-defect law, 68
viscous scales, 49, 63, 309, 325
visualization
rst picture, 37
gallery of, 3651
vortex generators, 176, 230
vortex-induced separation, 158
vorticity ux
spanwise, 30, 211
streamwise, 30
Voyager aircraft, 201, 210, 268
wall heating or cooling, 116, 168
wall jets, 179
wall motion, 111, 168
wall-bounded ows, 31, 105
comparison to free-shear ows, 69
constant-stress region, 66
ow regimes, 63
viscous region, 64
wave cancellation, 118
wavlets, 41
wavy walls, 221

Passive Active Ve Reactive. Flow Controlpdf

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Passive Active Ve Reactive. Flow Controlpdf

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Passive Active Ve Reactive. Flow Controlpdf

Uploaded by

Copyright:

Available Formats

Flow Control

PASSIVE, ACTIVE, AND REACTIVE FLOW MANAGEMENT

c; per unit span)

c; per unit span)

propagation velocity for longitudinal waves

(T) dimensionless function that characterizes viscosity variation with

1); also used as a sufx in tensor notations

longitudinal wave number

viscous length scale (=,u

ratio of boundary-layer thickness (or pipe radius) to viscous length scale

St Strouhal number (f L,v

longitudinal tension of thin plate

viscous time-scale (given by the inverse of mean wall vorticity)

viscous velocity scale; also called friction velocity (=

another symbol for friction velocity

) universal wake function

pressure drop normalized using a viscous pressure scale (Lp4a

dissipation function for incompressible ows

dimensionless variable; also postcollision value

(T) is a dimensionless function that characterizes the viscosity variation

are the velocity and pressure uctuations

angle of attack and the wake behind the tanker

= 742, and the boundary layer is again

= 725. Flat plate towed in

= 625 [after Gad-el-Hak

, is given by the inverse of

, is determined by the characteristic distance by which

) follows directly from the time and length scales

has been identied by Kline et al. (1967) and

=3006500, see Smith and

is consistent with the notion that this is

) and may also have signicant shear in

=50017,500, allowing them to unambiguously clarify Reynolds number ef-

>5000). This is sketched

near the outer edge of the boundary layer but at

to the ow direction. Head and Bandyopadhyay (1981) have observed such

(> 5000). It is possible to arrive precisely

a,; that is the channel half-width (or boundary-layer thickness) expressed

indicates a Reynolds number associated with the

increases monotonically with Re

and never reaches an asymptote.

= constant = 2.0 (5.10)

= constant = 1.0 (5.11)

= constant = 1.4 (5.12)

= 2000, H = 1.41, and at Re

= 10,000, H = 1.33. The effect is even more

. However, in deriving the log-law, Eq. (5.8), the presumably

at which a log-region is rst established? Coles (1962) analysis of the wake

600. The data of Bandyopadhyay and

= 425. These can be regarded as two indications of a minimumvalue supporting

has been chosen to insure that it

- 6000. After excluding all

>15,000. The drop cannot

= 6000, is 2.7 for subsonic

>15,000, the gradual departure of LU

asymptote suggests that some new effects are gradually appearing in

= 50,000, but the boundary layer might also continue to change

=4655100 showed clear Reynolds number ef-