RoelofVos
SaeedFarokhi
Introduction
to Transonic
Aerodynamics
Series editor
Andr Thess, German Aerospace Center, Institute of Engineering
Thermodynamics, Stuttgart, Germany
Founding Editor
Ren Moreau, Ecole Nationale Suprieure dHydraulique de Grenoble,
Saint Martin dHres Cedex, France
Introduction to Transonic
Aerodynamics
123
Roelof Vos
Faculty of Aerospace Engineering
Delft University of Technology
Delft
The Netherlands
Saeed Farokhi
Department of Aerospace Engineering
The University of Kansas
Lawrence, KS
USA
ISSN 09265112
ISSN 22150056 (electronic)
Fluid Mechanics and Its Applications
ISBN 9789401797467
ISBN 9789401797474 (eBook)
DOI 10.1007/9789401797474
Library of Congress Control Number: 2015930202
Springer Dordrecht Heidelberg New York London
Springer Science+Business Media Dordrecht 2015
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or
dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained
herein or for any errors or omissions that may have been made.
Every effort has been made to contact the copyright holders of the figures and tables which have been
reproduced from other sources. Anyone who has not been properly credited is requested to contact the
publishers, so that due acknowledgment may be made in subsequent editions.
Printed on acidfree paper
Springer Science+Business Media B.V. Dordrecht is part of Springer Science+Business Media
(www.springer.com)
Preface
viii
Preface
of each subject. The book contains 60 examples and more than 200 practice
problems.
This textbook is intended primarily for senior undergraduate students or graduate
students with prior knowledge in aerodynamics. Although we repeat the fundamental equations and flow characteristics in the beginning of the book, we do
assume that the student has had a course on subsonic aerodynamics and is familiar
with its fundamentals. Even though knowledge of transonic aerodynamics is
important for many internal flow applications (e.g., turbo machinery, engine
intakes, exhausts, etc.) the present textbook primarily focuses on external aerodynamics with limited applications to internal flows. Examples are targeted mainly
toward wings and bodies exposed to a transonic flow eld. Many of the examples
reference real aircraft or wing components. Therefore, a strong connection is
present between the content of this textbook and the subject of transonic aircraft
design. To understand why a modern highsubsonic aircraft is designed the way it
is, requires one to understand the subject matter of this textbook.
Acknowledgments
Writing this book has been a privilege. Naturally, we would not have been able to
do so without the support of our respective universities. We therefore would like to
thank the University of Kansas and Delft University of Technology for providing us
with the time and resources to write this book. We would also like to thank the
publisher, who, based on a 135page summary, trusted that we would extend it to its
current form. We would also like to acknowledge Dr. Luca Guadani and Dr. Ali
Elham who proofread various chapters and performed some of the numerical calculations to support the examples in this book. A special thank you to emeritus
professor Egbert Torenbeek for providing us with meticulous feedback on the
content of various chapters. Finally, we would like to thank the following persons
on both sides of the Atlantic, who over the course of multiple years, aided in the
preparation of this document: Mr. Thomas Statsny, Ms. Lisanne van Veen,
Mr. Maarten Broekhuizen, Mr. Kevin Haagen, Ms. Maaike Weerdesteyn, and
Mr. Amool Raina.
Delft, The Netherlands, 2014
Lawrence, KS, USA
Roelof Vos
Saeed Farokhi
Contents
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
4
5
8
14
16
18
.....
.....
.....
21
21
22
.....
22
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
29
33
35
38
38
40
41
42
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
42
43
44
46
48
50
ix
Contents
2.5
.
.
.
.
.
.
.
.
.
.
.
.
51
51
53
56
60
62
..
62
.
.
.
.
.
.
.
.
.
.
.
.
.
.
65
67
69
69
72
75
79
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
81
81
82
85
89
99
103
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
106
114
117
120
123
126
127
130
134
144
ShockExpansion Theory. . . . . . . . . . .
4.1 Introduction . . . . . . . . . . . . . . . .
4.2 Lift and Wave Drag . . . . . . . . . .
4.3 BiConvex Airfoil . . . . . . . . . . . .
4.4 Axisymmetric and Slender Bodies.
4.5 Examples and Applications . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
145
145
169
173
183
188
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Contents
xi
4.5.1
.
.
.
.
188
192
194
209
........
........
........
211
211
212
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
222
232
251
256
256
256
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
258
258
263
265
273
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
275
275
279
279
284
287
...
291
.
.
.
.
.
.
.
.
.
.
.
.
294
296
300
302
303
313
...
...
319
324
...
...
325
327
Method of Characteristics . . . . . . . . . . . . . . . . . . . . . .
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 2D Irrotational Flows . . . . . . . . . . . . . . . . . . . . .
5.3 Design of a 2D Supersonic MinimumLength
Nozzle (MLN) . . . . . . . . . . . . . . . . . . . . . . . . . .
5.4 WaveField Method Versus LatticePoint Approach.
5.5 Axisymmetric Irrotational Flows . . . . . . . . . . . . . .
5.6 Examples and Applications . . . . . . . . . . . . . . . . .
5.6.1
Design of a Family of Supersonic Nozzles .
5.6.2
Deflecting Jet . . . . . . . . . . . . . . . . . . . . .
5.6.3
Nonuniform Inlet Condition: Example
of Source Flow. . . . . . . . . . . . . . . . . . . .
5.6.4
Streamlines and Ducts . . . . . . . . . . . . . . .
5.6.5
Curved Shocks . . . . . . . . . . . . . . . . . . . .
5.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
xii
Contents
6.5.3
Turbulent Boundary Layer . . . . . . . . . . . .
6.5.4
Boundary Layer Transition. . . . . . . . . . . .
6.6 Interference Drag . . . . . . . . . . . . . . . . . . . . . . . .
6.6.1
Interference Drag in Subsonic Conditions .
6.6.2
Interference Drag in Transonic Conditions .
6.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
331
343
349
350
352
356
364
Airfoil Aerodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.2 Pressure Distribution About Airfoils . . . . . . . . . . . . . .
7.3 LaminarFlow Airfoils . . . . . . . . . . . . . . . . . . . . . . . .
7.4 Supercritical Airfoils . . . . . . . . . . . . . . . . . . . . . . . . .
7.4.1
ShockFree Supercritical Airfoil . . . . . . . . . . .
7.4.2
Supercritical Airfoils with Shocks . . . . . . . . . .
7.4.3
Sonic Rooftop Airfoils . . . . . . . . . . . . . . . . .
7.4.4
Effect of Trailing Edge Geometry . . . . . . . . . .
7.5 LowSpeed Stall . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.5.1
Qualification of LowSpeed Stall . . . . . . . . . .
7.5.2
Reynolds Effect on Maximum Lift Coefficient .
7.5.3
Mach Effect on Maximum Lift Coefficient. . . .
7.5.4
HighLift Devices . . . . . . . . . . . . . . . . . . . . .
7.6 HighSpeed Stall. . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.6.1
Flow Separation at the Shock Foot . . . . . . . . .
7.6.2
Transonic Buffet . . . . . . . . . . . . . . . . . . . . . .
7.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
367
367
368
372
375
379
383
390
390
393
394
397
399
403
405
406
411
419
425
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
427
427
428
430
433
441
449
450
453
461
........
........
........
461
464
470
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Contents
8.5
xiii
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
474
476
481
490
491
.........
.........
.........
496
499
509
513
521
527
529
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
545
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
551
Chapter 1
Abstract Airliners and jet fighters are often designed for Mach numbers close to
the speed of sound. At those speeds the flow domain is termed transonic and contains both subsonic and supersonic flow. Typical aerodynamic characteristics in this
domain are shock waves, drag divergence, and transonic buffet. To suppress these
adverse effects or postpone them to higher Mach numbers, multiple design measures
can be taken. A supercritical airfoil can be designed that postpones the onset of
shock waves. In addition, wing sweep can be introduced or the airplane can be area
ruled to further reduce the adverse transonic effects. This chapter presents some of
these practical measures and their effect. It also discusses the definition of transonic
aerodynamics and why it is beneficial to fly at Mach numbers where transonic flow
exits. Furthermore, it demonstrates the challenges associated with simulating transonic flow and transonic windtunnel experiments. Finally, an outline of the present
textbook is given.
Fig. 1.1 Bell X1, the first aircraft to officially fly supersonic (Photo USAF)
generation of jetpowered fighter aircraft were designed for (low) supersonic speeds
and featured typical design characteristics (such as an arearuled fuselage) to reduce
their peak drag coefficient as the airplane surpassed Mach 1.
It took more than 25 years after the first supersonic flight for the first transport
aircraft to fly supersonic. The Russian Tupolev 144 and the AngloFrench Concorde
were the first aircraft that cruised beyond Mach 2 and were thought to initiate a new
era in aviation: supersonic flight [3]. As is generally known now, there never was
a true supersonic era. Concorde flew between 1976 and 2003, but when it retired
there was no other supersonic aircraft to replace it. Environmental issues (supersonic
boom) as well as fuel cost had limited the feasibility of the aircraft. Instead, highsubsonic jet transport aircraft have dominated the civil aviation market for the past
60 years and their number increased rapidly over the 1970s and 80s. Being relatively
fuel efficient, fast and quiet (no supersonic boom), they have proven to be the best
solution for past and presentday airlines.
Jet transports fly in the lower transonic regime, also known as the highsubsonic
regime. It is therefore of interest to look more closely to the flow phenomena that
are taking place at these speeds and how this effects the design and operation of the
aircraft. To be able to justify a particular aircraft design we need to have a strong
knowledge about the flow phenomena and how we can influence them to meet the
requirements set by the costumers while satisfying the constraints of the aviation
authorities. One of the general goals of aerodynamic design is to maximize the
range parameter (Mach number times lifttodrag ratio, M L/D). According to
the Brguet range equation, the maximum cruise range (Rmax ) for a given thrusttoweight ratio is found when the product of Mach number (M) and lifttodrag ratio
L/D) is maximized [17]:
Rmax =
L
D
max
a
ln
SFC
Wbegin cruise
Wend cruise
(1.1)
where SFC is the specific fuel consumption, a the ambient speed of sound, and W
the gross weight of the aircraft. Assume that we can find an airplane shape that
25
74
B74
73
74
B
10
1960
1970
A3
11
MD
R
7E
ER
76
00
B
A3
00
6
IM
P
01
15
00
L1
30
10
DC
10
DC
L1 101
0
01
11
00
8
30
30
DC

B74
B70
15
71
30
00
00
20
00
DC
1980
1990
2000
40
Mdd Douglas definition
Mdd Boeing definition
C130H
F106
C5 A
30
727
20
10
0
0.4
0.5
0.6
0.7
0.8
Mach Number, M ()
0.9
1.0
Fig. 1.3 Typical compressibility drag behavior and definition of the dragdivergence Mach number
(modified from Ref. [16])
to have a weak shock terminating this supersonic region such that the wave drag is
minimized. It was already shown in the 1960s that by careful design of the airfoils
upper surface it is possible to decelerate supersonic flow from as high as M = 1.4
isentropically without having a shock, thereby minimizing the wave drag [14].
5
upper transonic
lower transonic
shockwave
sonic line
Mcrit
supersonic
sonic line
supersonic
domain
bow shock
subsonic
domain
1.0
subsonic
domain
Freestream
Mach number
transonic flows
Fig. 1.4 The basic classes of external transonic flows (after Ref. [13])
conditions. When the freestream Mach number is even further increased, the shock
waves on either side of the airfoil start to move aft as more of the body is immersed
in supersonic flow. As the freestream Mach number exceeds unity, a detached bow
shock forms in front of the body. Only part of the flow behind the bow shock is
decelerated to subsonic conditions. The flow behind the oblique branches of this
bow shock remains supersonic downstream. When the freestream Mach number is
increased even more, the subsonic domain becomes smaller and can be deemed to be
sensibly empty. The Mach number at which this occurs marks the boundary between
transonic and supersonic flow. Even though there is no physical discontinuity when
the flow transitions between supersonic and transonic, the upper freestream Mach
number for the flow to be termed transonic is generally set at 1.3 [6].
A breakthrough came in the late 1940s (a few years after the Bell X1 had made
its maiden flight). Slotted test sections were introduced, where thin, rectangular slots
were made such that the displaced airflow could move out of the tunnel into a socalled plenum chamber that surrounded the test section. By doing this, the blockage
problem was essentially solved and the flows could be accelerated through Mach 1
without any problems. These slotted tunnels were essentially the first transonic wind
tunnels and became operational in the early 1950s. Figure 1.5 shows the 16foot
transonic wind tunnel that is operated by NASA Langley. Later it was found that the
improvement is even greater when the walls are not slotted but perforated [21].
Apart from windtunnel blockage, transonic wind tunnel tests faced another problem: reflected shock waves from the walls. This was particularly an issue for transonic
flows because the shock on the airfoil was closeto normal to the surface of the body.
Therefore, the reflected shock wave interfered with the flow over the body, rather
than with the flow downstream of the body as would be the case for oblique shock
waves [15]. The requirement on the wind tunnel walls was therefore to effectively
cancel the reflected shock waves by producing streamlines downstream of the shock
that corresponded to the flow downstream of the shock. In this respect, porous walls
proved to be superior to slotted walls in which the higher pressure in the viscous
slots extended upstream of the shocks thereby distorting the flow.
In addition to the problem of shockwave reflection, the growth of the boundary
layer on the walls of the test section proved to be a source of inaccuracy. The thickening of the boundary layer was caused by small windtunnelwall irregularities. It was
found that a slightly converging wind tunnel wall resulted in a much thinner boundary layer compared to a parallel or slightly diverging wall. The wall inclination was
therefore a parameter that could be changed to reduce the boundarylayer thickness
growth [7]. This problem was targeted in the 1970s. One transonic wind tunnel
incorporated theoretical correction models that used input from a series of test runs.
Fig. 1.5 The presentday slottedwall test section of the NASA Langley 16foot transonic tunnel [4]
It was then calculated how the walls needed to be oriented such that the wall interference was minimized and the most accurate results were obtained [22].
Finally, it was found that considerable turbulence formed in the wind tunnel that
modified the boundary layer on the test article in an unpredictable way. Even though
the walls were positioned such that their influence was minimized, this phenomenon
had a considerable influence on the accuracy of the aerodynamic measurements. It
was found that the source of the turbulence came from acoustic disturbances from the
slots or perforations. A practical approach yielded a solution to this problem. Simple
splitter plates in each of the holes diminished the acoustic waves for the perforated
walls. Analogously, a thin wire mesh that covered the slots had the same effect in
case of slotted walls.
Wind tunnel tests are conducted on scale models of aerodynamic bodies that
resemble the true model very accurately. All the conditions in the tunnel are set such
as to match realistic conditions as well as possible. However, the Reynolds number
effect is one characteristic that is hard to simulate when scaling down to wind tunnel
sizes. Since the Reynolds number is dominant in the determination of friction drag,
boundary layer thickness, transition point, and shock position, it is evident that testing
at lower Reynolds number can have severe impact on the predicted aerodynamic
performance of the test object.
For that reason, the idea arose in the early 1970s to modify the test gas such as to
match the Reynolds number encountered by the wind tunnel model to the Reynolds
number of the fullscale body. By increasing the density of the test gas the Reynolds
number increased linearly. To increase the density, three different measures were
taken: the air in the tunnel was replaced with a heavier gas (e.g., carbondioxide), the
gas was cooled by injecting liquid nitrogen into the flow, and the air was pressurized.
Combining these measures resulted in a Reynolds number which could be twenty
times as high as for conventional wind tunnels [8].
Combine all these efforts with the measures that were described in the previous
paragraphs and it becomes clear that transonic cryogenic wind tunnels are among the
most complex aerodynamic facilities to test aerodynamic bodies. Currently, there are
only three large scale facilities in the world: one in the United States (National Transonic Facility, NTF) and two in Cologne, Germany (European Transonic Windtunnel,
ETW, and Kryo Kanal Kln). The ETW can cool the flow to 110 K in combination
with pressures up to 450 kPa and a Mach number of 1.3 [8]. Advantages of this
tunnel are that compressibility effects, friction effects and deformation effects can
be studied independently by varying Mach number, dynamic pressure and Reynolds
number independently of one another. A photograph taken in the test section of the
ETW is presented in Fig. 1.6.
Fig. 1.6 ETW wind tunnel with aircraft model (Photo ETW; printed with permission)
(b)
(a)
M >1
M < 1
M< 1
(c)
M >1
M= 1
(d)
M >1
M >1
M >1
M >1
M >1
Upper
Surface
M
Lower
Surface
0
M < 1
M < 1
M =1
M > 1
Fig. 1.7 Four flow patterns and corresponding Mach distributions representative of transonic flow
over an airfoil (after Ref. [12]). a M < 1, b M < 1, c M = 1, d M > 1
Physical Laboratory (UK) and Whitcomb [23] at NASA in the early 1960s. The
pressure distributions on a conventional (NACA 6series) airfoil and a supercritical
airfoil are shown in Fig. 1.8. The strong shock that abruptly transitions the supersonic
flow to subsonic on the suction surface of the conventional airfoil (in transonic Mach
number) is replaced by a relatively weak shock that appears further aft on the airfoil.
The corresponding pressure distributions show the difference between a smooth
transition and an abrupt transition to subsonic flow on the two airfoils. We also note
that the supercritical airfoil has a concave curvature near the trailing edge on its
pressure surface. This curvature reversal on the aft pressure surface causes a higher
loading on the airfoil and thus a larger contribution to lift.
We have superimposed a critical pressure coefficient line on the data of Ref.
[11] in Fig. 1.9 to see (or separate) the supersonic zone on the airfoil. The pressure
distribution on a supercritical airfoil at Mach 0.8 with a lift coefficient of 0.54 and a
chord Reynolds number of three million is shown. Noteworthy features are significant
aft loading on the airfoil, as well as a relatively smooth compression on the suction
surface at the end of supersonic bubble. A footprint of flow separation, i.e., a flat C p
profile downstream of the shock, on the suction surface is visible in the data. We also
note that the pressure surface never reaches sonic flow. This pressure distribution is
typical for airfoils that were developed for high subsonic Mach numbers.
Another measure to postpone the drag divergence is the application of swept
wings. The first incarnations of sweptback wings date back to the early 1900s,
when John William Dunne (an Irish aeronautical engineer) designed and built various
tailless biplanes with aftswept wings. In Fig. 1.10 the Dunne D.8 is shown taking off
from the airfield at the Farnborough air show. Dunne applied wing sweep to shift the
neutral point (i.e., the aerodynamic center of the aircraft) further aft, such that it would
lay behind the airplanes center of gravity. This ensured a positive static stability
margin, which is an important prerequisite for having acceptable flying qualities for
manually controlled aircraft. Many other tailless aircraft relied on wing sweep to
10
(a)
Relatively
strong
shock
(c)
M>1
M>1
Relatively
weak
shock
M>1
(d)
(b)
Cp
Cp
()
Cp, crit
()
Cp, crit
(+)
(+)
Fig. 1.8 Pressure distribution on a conventional (NACA 6series) airfoil and a supercritical airfoil
(after Ref. [23]). a Supersonic domain over NACA 6series airfoil. b Pressure distribution over
NACA 6series airfoil. c Supersonic domain over supercritical airfoil. d Pressure distribution over
supercritical airfoil
Weak shock
Separation (footprint)
1.0
Supersonic
0.8
Reattachment
0.6
C p, crit =  0.435
0.4
0.2
Subsonic
Cp
0
0.2
0.4
0.6
0.8
0.0
0.2
0.4
x
c
0.6
0.8
1.0
Fig. 1.9 Pressure distribution on a supercritical airfoil at Mach 0.80, cl = 0.54, Rec = 3.0 106
(Flagged symbols are for pressure surface, data after Ref. [11])
11
12
mence. After the war, the concept of forwardswept wings was almost completely
abandoned due to the high weight penalty that was introduced due to adverse aeroelastic effects. The aftswept wing, on the other hand, was embraced by both military
and civil airplane manufacturers in Europe and the United States. It was Boeing that
first introduced aftswept wings on a large bomber aircraft. This aircraft, the B47
Stratojet (Fig. 1.11), also featured podded jet engines that were suspended below
the shouldermounted wing. This configuration with sweptback wings and engines
mounted in nacelles below the wing is still the dominant aircraft configuration for
highsubsonic airliners.
With the flight of the X1 in 1947, the supersonic domain had opened up for
military aircraft. However, the jet engines at the time were not yet powerful enough
to overcome the drag rise around Mach one. Therefore, first generation of jet fighters
flew at high subsonic speeds. New specifications from the US Airforce in 1950
dictated an aircraft that could fly with a top speed of Mach 1.2. It was Convair that was
awarded the contract to design and build a prototype aircraft. The resulting aircraft
featured a delta wing and no horizontal tail plane. Test flights showed disappointing
results: the airplane could only achieve M = 0.98, far below the requirement. A major
redesign was initiated to reshape the fuselage such that the peak drag coefficient
around Mach one reduced. This reshaping, commonly referred to as area ruling,
proved to be very effective. The new prototype was able to fly beyond the speed of
sound and met the topspeed requirement of Mach 1.2. This airplane (designated
F102A) became quite successful with 1,000 copies built. In Fig. 1.12, a comparison
between the first prototype and the second prototype of the aircraft is shown.
The arearule was initially proposed by Richard Whitcomb as a rule of thumb
as he called it. This rule of thumb essentially stated that the wave drag of an arbitrary
body at sonic conditions is solely dependent on the crosssectional area distribution
of that body in the direction of the free stream. He first proposed this rule in 1952
13
Fig. 1.12 Area ruling applied to Convair F102 (Photos NASA). a YF102 prior to area ruling. b
YF102A with arearuled fuselage
based on experimental results that he had obtained in the NACA wind tunnels and the
theory of stream pipes that had been introduced the year before by Adolf Bsemann.
However, already during the second world war the effect of crosssectional area distribution on transonic drag was acknowledged by Otto Frenzl. Working at Junkers, he
introduced socalled displacement bodies that were strategically positioned such
as to reduce the wave drag a highsubsonic conditions. The idea of a favorable crosssectional area distribution was to be implemented in 1945 by Waldemar Voigt on the
Messerschmidt P112, which was never completed. In 1946, Dietrich Kchemann
introduced the pinched fuselage, a reduction of fuselage diameter at the location
of the wing. This modification was later dubbed the Kchemann Cokebottle design
and is one of the modifications found in the YF102A. A mathematical description of
the ideal crosssectional area distribution was presented in two separate articles by
Wolfgang Haack (1941) [9] and William Sears (1947) [19]. For a given combination
of length and volume, the SearsHaack body possesses the crosssectional area distribution with the least amount of wave drag at Mach one. For speeds beyond the speed
of sound, a different area distribution should be found according to the supersonic
area rule. This rule was first introduced by Wallace Hayes in his PhD dissertation of
1947 [10].
14
15
1.
1.
.8
.8
.6
.6
Cp
.4
Cp
.2
0
.4
.2
.2
.2
.4 .6
x/c
.4
= 0.26
.8
1.
M = 0.86
.4
1.
1.
1.
.8
.8
.8
.6
.6
.6
.4
.4
Cp .4
Cp .2
Cp .2
.2
.4
.4
.6
x/c
.8
1.
.8
1.
= 0.89
.6
.6
.2
.2
.2
inviscid
coupled with b.l.
.2
.4 .6
x/c
= 0.42
.8
1.
0
.2
.2
.4
.4 .6
x/c
.8
.2
.2
.4
= 0.60
.4
.6
x/c
= 0.73
.6
.6
.6
1.
Fig. 1.13 Impact of viscous coupling with inviscid flow solution on a finite wing in transonic flow
(after Ref. [20])
0.8
Cp 0.4
Euler
Euler + Coupled
Boundary Layer
Test
Cp
0.8
0.4
0.0
0.0
0.4
0.4
M = 0.84
= 2.08
0.8
0.8
Cp 0.4
Cp 0.4
0.0
0.0
0.4
0.4
Fig. 1.14 Impact of viscous coupling on pressure distributions on a winglet in transonic flow (after
Ref. [20])
16
from the shortcomings of the boundary layer simulation of complex viscous flows.
Therefore, such problems demand the use of fully viscous solvers that are based
on the (ReynoldsAveraged) NavierStokes equations rather than inviscid solutions
coupled with a boundary layer solver.
17
methods include the supersonic flow through nozzles and the supersonic exhaust
plume of an underexpanded jet.
Chapter 6 presents the sources of drag that are encountered by a nonlifting body
that is moving at transonic Mach numbers. A qualitative characterization of laminar
and turbulent boundary layers is presented along with the concept of transition. In
addition, calculation methods are presented to estimate the friction drag of bodies
subjected to a pressure distribution. Furthermore, a method for the calculation of the
pressure drag over axisymmetric bodies is presented based on the linear potential
flow equation. It is shown how the pressure drag is a function of the crosssectional
area distribution of the body. The concept of area ruling is presented and examples
are shown of practical implementations of this knowledge.
In Chap. 7 we present the transonic aerodynamics about liftproducing airfoils. It
is shown how various design parameters influence the velocity distribution and the
shock formation over an airfoil. The development of supercritical airfoils is explained
from a historic perspective. The concept of shockboundary layer interaction, related
shock stall, and shock oscillation are also explained. Furthermore, it is demonstrated
that transonic flow limitations can play a major role in the maximum lift coefficient
of (multielement) airfoils at Mach numbers as low as 0.25. Examples demonstrate
how lowspeed and highspeed stall limit the flight envelope of aircraft that travel at
transonic Mach numbers.
In the final chapter (Chap. 8) we discuss the aerodynamics of swept wings in
transonic flow. It is shown through simplesweep theory why a swept wing allows for
a higher dragdivergence Mach number and what its effect is on the drag coefficient
in the transonic regime. Apart from these advantages the chapter also presents the
challenges that aerodynamic designers face to reduce the adverse effects of wing
sweep: tip stall, early formation of tip shocks, form drag at the wing root, an early
onset of boundarylayer transition, aerodynamic twist, and aileron reversal. Apart
from aftswept wings, also the merits and drawbacks of forwardswept wings are
presented.
Each of the chapters includes a set of examples and a set of problems to practice
with the theoretical concepts that are presented. We hope that through the addition
of these examples and problems, the reader can familiarize himself or herself with
important aspects that are related to transonic flow. Answers to selected problems
are given in the back of this book (p. 527 and onwards). A comprehensive list of
symbols and nomenclature is also added (starting at p. 543) such that the reader can
check the meaning and units (all in SI) of the variables that are used throughout the
chapters. Each chapter also lists a variety of references for further reading. Naturally,
if the reader is interested to learn more about particular topics that are presented in
this book we encourage him/her to consult the original sources. They often contain
much more detail about the particular topic at hand and provide a valuable link to
more material on the subject.
18
Problems
1.1 Consider an aircraft with a maximum takeoff weight (MTOW) of 560 metric
tons. Assume that at the beginning of the cruise phase this aircraft has burned some
fuel and weighs 96 % of its maximum takeoff weight. At the end of the cruise phase
assume that the aircraft weighs only 75 % of its MTOW.
(a) For a cruise altitude of 35,000 ft (10.67 km) calculate the speed of sound, a, using
a temperature lapse rate of 6.5 K/km, a sealevel temperature of 288 K, and a
gas constant of 287 J/kg/K.
(b) For a cruise Mach number of 0.85 and a specific fuel consumption of SFC =
0.000186 [N/N/s] calculate the range, R, when L/D = 16, L/D = 19, and
L/D = 22. Use the Brguet range formula (1.1).
1.2 The maximum range of an airplane is achieved when V L/D is maximized.
In the low subsonic realm the drag coefficient can be approximated as the sum of
the zerolift drag coefficient, C D0 and the induced drag coefficient, C Di , with C Di =
kC L2 , k being a constant. For a business jet in clean configuration C D0 = 0.021,
k = 0.038 and the wing loading is W/S = 3 kN/m2 . Using a spreadsheet program
solve the following problems:
(a) Graph the relation between C L (vertical axis) and C D (horizontal axis).
(b) Now, calculate the lifttodrag ratio (C L /C D ) for this business jet for a lift
coefficient ranging from 0 to 1.7. Graph the relation between the lift to drag ratio
(vertical axis) and lift coefficient (horizontal axis).
(c) From your graph, estimate the maximum L/D and the lift coefficient this
occurs at.
(d) Express the velocity as a function of the wing loading, density, and lift coefficient.
(e) Now, calculate the product of velocity and lifttodrag ratio (V C L /C D ) for
a lift coefficient ranging from 0 to 1.7. Graph this relation by putting the lift
coefficient on the horizontal axis. Assume a value of 0.3 kg/m3 for the density.
(f) From your graph, estimate the maximum V C L /C D and the lift coefficient
this occurs at.
References
1. Anderson, J.: Modern Compressible Flow With Historic Perspective, 3rd edn. McGraw Hill,
New York (2003)
2. Bensberg, W., Cranz, C.: Uber eine photographische Mehode Zur Messung von
Geschwindigkeiten und Geschwindigskeitverlusten bei Infanteriegeschossen. Artillerische
Monatshefte 41, 333346 (1910)
3. Burgess, E.H.: Concorde inaugurates the supersonic era. In: Proceedings of the 9th Annual
Meeting and Technical Display. AIAA197316, Washington, DC (1973)
4. Cornelliussen, S.T.: The transonic wind tunnel and the NACA technical culture, Chap. 4. From
Engineering Science to Big Science, pp. 91133. NASA, Washington (1998)
References
19
5. Denning, R., Armstrong, J.A., Allen, J.E.: The broad delta airliner. Aeronaut. J. 107(1075),
547558 (2003)
6. Ferrari, C., Tricomi, F.G.: Transonic Aerodynamics. Academic Press, New York (1968)
7. Goethert, B.H.: Transonic Wind Tunnel Testing. No. 49 in AGARDograph. Pergamon Press,
Oxford (1961)
8. Green, J., Quest, J.: A short history of the European Transonic Wind tunnel ETW. Prog. Aerosp.
Sci. 47, 319368 (2011)
9. Haack, W.: Geschossformen kleinsten Wellenwiderstandes. Bericht 139 der LilienthalGesellschaft, pp. 1428 (1941)
10. Hayes, W.D.: Linearized Supersonic Flow. Ph.D. thesis, California Institute of Technology
(1947)
11. Hurley, F.X., Spaid, F.W., Roos, F.W., Stivers, L., Bandettini, A.: Detailed Transonic Flowfield
Measurements about a Supercritical Airfoil Section, NASA TMX3244 (1975)
12. Kuechemann, D.: The Aerodynamic Design of Aircraft. Pergamon Press, Oxford (1978)
13. Moulden, T.H.: Fundamentals of Transonic Flow. Wiley, New York (1984)
14. Pearcey, H.H.: The aerodynamic design of section shapes for swept wings. Adv. Aeronaut. Sci.
3, 277322 (1963)
15. Ramm, H.J.: Fluid Dynamics for The Study of Transonic Flow. Oxford University Press,
New York (1990)
16. Roskam, J.: Airplane Design, Part 6: Preliminary Calculation of Aerodynamic. Thrust and
Power Characteristics. DARcorp, Lawrence (2006)
17. Ruijgrok, G.J.J.: Elements of Airplane Performance. Delft University Press, Delft (1989)
18. Scholz, D., Ciornei, S.: Mach number, relative thickness, sweep and lift coefficient of the
wingan empirical investigation of parameters and equations. DGLR Jahrbuch 2005 (2005)
19. Sears, W.R.: Projectiles of minimum wave drag. Q. Appl. Math. 4(4), 361366 (1947)
20. Tinoco, E.N.: CFD applications to complex configurations: a survey. In: Henne, P. (ed.) Applied
Computational Aerodynamics, Chap. 15. AIAA, Washington (1990)
21. Tokaty, G.A.: Fluid Dynamics. W & J Mackay & Co Ltd., Chatham (1971)
22. Vidal, R.J.: Wall Interference Effects in Transonic Flows. Calspan Corp., Buffalo (1976)
23. Whitcomb, R.T., Clark, L.R.: An Airfoil Shape for Efficient Flight at Supercritical Mach
Numbers, NASA TM X1109. Langley (1965)
Chapter 2
2.1 Introduction
Transonic flow conditions are encountered by the majority of todays jet aircraft.
Almost all major airlines use highsubsonic jet transports on their medium to long
haul flights. In addition, business jets, fighter aircraft, and military UAVs such as the
Northrop X47 are also confronted with transonic effects. To comprehend the physics
of transonic flow, this chapter presents a review of physical and mathematical topics
that form a basis for the subsequent chapters of the book. For further reading on these
topics, the reader is referred to the reference list at the end of this chapter, that lists a
number of text books that adequately explain the subject matter in more detail. After
this short introductory section, we will present basic reviews on two mathematical
topics: partial differential equations and vector algebra. Since mathematics is often
referred to as the language of physics, a good understanding is mandatory for the
Springer Science+Business Media Dordrecht 2015
R. Vos and S. Farokhi, Introduction to Transonic Aerodynamics,
Fluid Mechanics and Its Applications 110, DOI 10.1007/9789401797474_2
21
22
(2.1)
23
T2
T1
x x+x
Directing our attention to the vertical direction we know that the following force and
acceleration balance should hold over the part x:
T1 sin + T2 sin = x
2u
t 2
(2.2)
x 2 u
T t 2
(2.3)
Since tan = u/xx and tan = u/xx+x we can now write a partial
differential equation where both derivatives with respect to x and t appear:
1
x
u
x
x+x
u
x
=
x
2u
T t 2
(2.4)
If we subsequently take the limit as x 0 we obtain the well known onedimensional wave equation:
2
2u
T
2 u
=
c
with c2
2
2
t
x
(2.5)
The wave equation is a second order (=the highest power of one of the differential
terms), linear partial differential equation. Not only does this equation describe the
transverse motion of a vibrating string it also describes the wave motion of a plane
wave (e.g. acoustic wave) [16]. In that case we replace the excitation of
the string by
the sound pressure, p (x, t) and c becomes the speed of sound, c = RT , where
is the ratio of specific heats, R is the gas constant, and T is the static temperature
in the calorically perfect gas.1
1
24
We can now proceed with a solution procedure of the wave equation. There are
several approaches to derive a closed form solution of the problem, one of which
is by using DAlemberts approach. This approach relies on a transformation to the
characteristic coordinates and according to:
= x + ct,
= x ct
(2.6)
2u
=0
(2.7)
(2.8)
If we assume that two arbitrary functions, f (x) and g(x) describe the initial
position and velocity, respectively, we can express the functions (x) and (x) in
terms of the these initial conditions. We have:
u(x, 0) = f (x) = (x) + (x)
u t (x, 0) = g(x) = t + t
(2.9)
(2.10)
(2.11)
25
If we integrate both sides of (2.11) between x0 and x and divide both sides by c
we obtain:
(x) (x) =
1x
g(s)ds with k(x0 ) = (x0 ) (x0 ).
c x0
(2.12)
Now, to get (x) we simply add (2.12) to (2.9) and divide both sides by two. Similarly,
to obtain (x) we subtract (2.12) from (2.9) and divide both sides by two. The
following results are found:
1
f (x) +
2
1
(x) = f (x)
2
(x) =
1
1 x
g(s)ds + k(x0 )
x
2c 0
2
1 x
1
g(s)ds k(x0 )
2c x0
2
(2.13)
(2.14)
Substituting (x + ct) and (x ct) in the above expressions and inserting everything back into (2.8) results in the following solution to this initial value problem:
u(x, t) =
1
1 x+ct
g(s)ds.
[ f (x + ct) + f (x ct)] +
2
2c xct
(2.15)
For a brief moment, let us return to interpretation of this result with respect to
the original problem of the vibrating string and assume that the initial velocity,
u t (x, 0) = g(x) = 0. To come up with a valid solution that satisfies the boundary
conditions (u(0) = u(L) = 0), the following should hold:
1
[ f (ct) + f (ct)] = 0
2
1
u(L , t) = [ f (L + ct) + f (L ct)] = 0
2
u(0, t) =
(2.16)
(2.17)
From (2.16) we learn that the initial shape function f (x) is odd. In other words it
is antisymmetric with respect to the uaxis. Combining this with (2.17) we obtain
that f (L + ct) = f (L + ct), which shows that the initial function should have a
period of 2L.
Example 2.1 Calculate the solution to the homogeneous wave equation (2.5) for
u(x, 0) = f (x) = sin(2x), u t (x, 0) = g(x) = 0, and c = 1.
Solution:
Using DAlemberts solution we can directly substitute f (x) = sin(x) and g(s) =
0 in to (2.15). We obtain:
u(x, t) = sin 2(x + t) + sin 2(x t)
A graphical representation of this string is sketched over the interval [0, 1] between
time t = 0 and t = 0.25 with time increments t = 0.05 (Fig. 2.2).
26
t=0
t=0.05
t=0.1
u (x,t)
0.5
t=0.15
t=0.2
t=0.25
0.5
0.2
0.4
0.6
0.8
(2.18)
where f (x, t) is the disturbance function. We assume that the string is initially
undisturbed, i.e.:
(2.19)
u(x, 0) = u t (x, 0) = 0
Using the characteristic coordinates as in (2.6), we can rewrite u(x, t) as follows:
u(x, t) = u
+
,
2
2c
= v(, )
(2.20)
Similarly, we can transform the forcing function: f (x, t) = G(, ). Note that when
t = 0 we have = and = x. Therefore, we can write the first initial condition:
u(x, 0) = u(, 0) = v(, ) = 0
(2.21)
To transform the second initial condition we write the first partial derivative of v with
respect to in terms of u x and u t by employing the chain rule (see Problem 2.5):
v =
1
1
u x + ut
2
2c
(2.22)
1
1
u x (, 0) + u t (, 0)
2
2c
(2.23)
27
Now, we know that the second part of the righthand side of the equation is zero
due to the initial condition u t (x, 0) = 0. And, since u(x, 0) = 0 we know that
1/2u(x, 0)x= must also be zero. Therefore, we now have the second initial condition
in transformed coordinates:
v (, ) = 0
(2.24)
1
G(, )
4c2
(2.25)
We are interested in the excitation at a particular time and location, u(x0 , t0 ), which
corresponds to v(0 , 0 ). Therefore, we need to integrate the lefthand side of (2.25)
twice with appropriate boundary conditions:
0
0
v (, )dd
Solving the inner integral by using the fundamental theorem of calculus (FTC)2
results in:
v (, )d = v (, ) v (, 0 )
0
We note that the first term on the righthand side must equal zero, due to the initial
condition (2.24). The second integration step thus results in:
0
0
v (, 0 )d = v(0 , 0 ) v(0 , 0 )
Again, due to the initial condition (2.21) for = 0 , the first term on the righthand
side equals zero, which leaves us with the negative of what we were initially looking
for: v(0 , 0 ). Now, by performing the same integration on the righthand side
of (2.25) and bringing the minus sign to the other side of the equality sign (which
cancels) we have a solution in characteristic coordinates:
v(0 , 0 ) =
1 0
G(, )dd
4c2 0 0
(2.26)
To change back to the original coordinate system, the determinant (for explanation
of the determinant see Sect. 2.3) of the Jacobian, det(J ), needs to be calculated, since
dd = det(J )dxdt. It is left to the reader (Problem 2.6) to show that this equals 2c.To
The fundamental theorem of calculus specifies the relationship between the two central operations
of calculus: differentiation and integration [11].
28
(b)
(a)
( 0 ,0)
(x 0 ,t 0 )
( 0 ,0)
(0,0)
0
x0  ct0
x0+ct0
Fig. 2.3 Transformation of the area of integration between the characteristic and the original coordinate system. a Region of integration in the (, )plane. b Region of integration in the (x, t)plane
transfer the limits of the integrals it is convenient to look at the region over which we
are integrating G(, ). This region is displayed in Fig. 2.3a. If we look at the vertices
of this triangle, and what they represent in the physical (x, t)plane, we note that:
(0 , 0 )
(0 , 0 )
(0 , 0 )
(x0 , t0 )
(x0 + ct0 , 0)
(x0 ct0 , 0)
The region of integration in the (x, t)plane that corresponds to these vertices is
shown in Fig. 2.3b. We therefore have:
u(x0 , t0 ) =
1 t0 x0 +ct0
f (x, t)dxdt
2c 0 x0 ct0
(2.27)
The region of integration in the physical domain can be used to see whether a
point in the domain is influenced by a particular disturbance. When the area under
the graph of the disturbance function and the area of the integration region (partially)
overlap, then the point in the domain is influenced by the disturbance. However,
if there is no overlap at all, the disturbance function does not affect that particular
point in the domain. The characteristic lines ( = x + ct and = x ct) therefore
mark the boundaries of the socalled domain of dependence. The solution, therefore,
only depends on a disturbance that happens within this domain of dependence. Any
disturbance outside this region does not have any effect on the excitation of this point.
In Example 2.2 this is demonstrated.
29
t
f(x,t)
1
(x,t)
0 xt
1
1 x+t
u(x, 0) = 0
u t (x, 0) = 0
Calculate u(x, t) explicitly for (x, t) satisfying: x > t, x < 1, and x + t > 1.
Solution:
To solve this problem we use (2.27) with c = 1. Due to the simple
forcing function,
f (x, t), we can directly see that the solution u(x, t) = 1/2 (x,t) f (x, t)dxdt,
where (x, t) represents the region of integration. This region is schematically shown
in Fig. 2.4. Using simple geometry we calculate the area of this region (x, t) =
t2
x2
1
2 2 xt x t 2 . The solution then becomes:
u(x, t) =
1 2
(t x 2 2xt + 2x + 2t 1)
4
x=L
Fig. 2.5 Model problem for the onedimensional heat equation: a bar of length L
30
and homogeneous material and that it is perfectly insulated, such that heat flows only
along the xdirection.
The derivation of the heat equation starts with Fouriers law of heat conduction
that states that the heat flux, q, is negatively proportional to the temperature gradient,
T /x, according to (in onedimensional space):
q = kT /x,
(2.28)
where k is the thermal conductivity of the material. The negative sign indicates
the direction of heat flow is from hot to cold, i.e. in the opposite direction to the
temperature gradient. In the absence of work done, a change in internal energy per
unit volume in the material, E, is proportional to the change in temperature, T .
That is,
E = cT
(2.29)
where c is the specific heat capacity and is the mass density of the material. Choosing
zero energy at the absolute zero temperature, this can be rewritten as
E = cT.
(2.30)
(2.31)
t t t + t
(2.32)
is given by
c
x+x
xx
[T (, t + t) T (, t t)] d = c
t+t x+x T
dd , (2.33)
tt xx
where the FTC was used. Additionally, with no work done and the absence of any
heat sources or sinks, the change in internal energy in the interval [x x, x +x]
is accounted for entirely by the flux of heat across the boundaries. By Fouriers law,
this is
t+t x+x 2 T
t+t T
T
(x + x, )
(x x, ) d = k
k
dd
tt
tt xx 2
x
x
(2.34)
again by the fundamental theorem of calculus. In higher dimensions, the divergence
theorem (Sect. 2.3.3) is used instead. By conservation of energy (Sect. 2.5.3),
t+t x+x
tt
xx
[cT kT ] dd = 0.
(2.35)
31
k
Tx x ,
c
T
k
=
t
c
2 T
x 2
(2.36)
(2.37)
(2.38)
which is the heat equation. The coefficient k/(c) is called thermal diffusivity and
is denoted by for convenience.
To solve the heat equation (2.38), we use the separationofvariables technique [18] where we substitute
T (x, t) = F(x)G(t)
(2.39)
(2.40)
(2.41)
(2.42)
Equation (2.41) is second order in space and Eq. (2.42) is first order in time.
To find the eigenfunctions of the heat equation we need a set of boundary equations
that give us some details about the heat transfer conditions of the rod at the ends.
In the present case we choose to keep the ends of the rods at zero temperature:
T (0, t) = T (L , t) = 0. A general solution to (2.41) is:
F(x) = A cos(x) + B sin(x)
(2.43)
nx
L
for n = 1, 2, . . .
(2.44)
32
(2.45)
G n (t) = Cn en t for n = 1, 2, . . . ,
(2.46)
where Cn are arbitrary constants. This results in the following final solution of this
problem:
Tn (x, t) = Fn (x)G n (t) = Dn sin
nx
L
en t for n = 1, 2, . . .
2
(2.47)
where Dn are arbitrary constants. These are the eigenfunctions of the problem and
n = cn
L are the eigenvalues.
In order to solve the heat equation we need an initial condition, f (x). This initial
condition is prescribing the temperature distribution along the bar at time t = 0
must comply with the boundary conditions. We know that the sum of each of the
solutions in (2.47) is also a solution to (2.38) and we consider the following series
of eigenfunctions:
nx
2
en t
Dn sin
(2.48)
T (x, t) =
L
n=1
Since at t = 0 we need to satisfy the initial condition the following relation must be
satisfied:
nx
=
f (x)
(2.49)
T (x, 0) =
Dn sin
L
n=1
nx
2 L
dx
f (x) sin
L 0
L
(2.50)
Example 2.3 Solve the heat equation, Tt = Tx x on the interval 0 < x < 1 with
boundary conditions T (0, t) = Tx (1, t) = 0 and initial condition T (x, 0) = 1.
Solution:
Using the same methodology as above we separate the temperature function in the
product of F(x) and G(t). We have
F(x) = A cos(x) + B sin(x)
33
Again, we state that due to the first boundary condition A = 0. The second boundary
condition implies: B cos(x) = 0. This equation is satisfied when
1
for n = 0, 1, 2, . . .
= n+
2
This results in the following general solution:
1
Fn (x, t) = Bn sin x n +
2
Substituting
this in (2.39), with G n (t) being the same as in (2.46) and n =
n=0
2
1
2 n+ 21 t
e
Dn sin x n +
2
(2.51)
1
sin x m +
0
2
2
m + 21
This gives the following final solution (switching back from m to n for convenience):
T (x, t) =
n=0
2
1
2
2 n+ 21 t
e
sin
x
n
+
2
n + 21
(2.52)
In Fig. 2.6, we see a graphical interpretation of the solution. Note how the initial
condition is satisfied by showing almost a complete horizontal line at t = 0. The
boundary condition at x = 0 is also satisfied, while we can see that the slope of the
line at x = 1 is zero, indicating that we have also satisfied this boundary condition
correctly.
34
t=0
T (x,t)
0.8
0.6
0.2
0.4
0.4
0.6
0.8
0.2
1.0
0
0.2
0.4
0.6
0.8
Example 2.4 To appreciate the difference between the conservative and nonconservative form of a partial differential equation we consider the onedimensional heat
equation (2.38) without the assumption that , c, or k are constant with x. In other
words, the thermal diffusivity, is a function of x: = (x). First this equation is
formulated in conservative form:
T
T
=
(2.53)
t
x
x
The nonconservative form of the same equation reads:
T
2 T
T
= 2 +
t
x
x x
(2.54)
Note how in (2.53) the coefficient, , can vary with position (x) but its derivative does
not appear in the equation. Therefore, (2.53) is in conservation form. In (2.54) the
derivative on the RHS of (2.53) has been expanded and now contains the derivative
term /x, which is a nonconservative term in the equation. Therefore we deem
(2.54) to be in nonconservative form.
The same logic that is used for the onedimensional heat equation can be expanded
to the equations of motion that describe fluid flows (Sect. 2.5). These governing
equations must hold at any distinct point in the flow. To formulate these equations
we take a Eulerian approach where we consider a fixed control volume through which
the fluid passes. The conservation form for each PDE at such a point also allows us
to formulate finitedifference representations that provide a good approximation to
the PDEs. These, in turn, can be used in an iterative numerical code to obtain the
global properties of a flow field. The conservative formulation is therefore often used
to solve partial differential equations by numerical methods.
35
(2.55)
(2.56)
b b2 4ac
dy
=
dx
2a
(2.57)
Depending on the term under the radical in (2.57), the family of characteristic curves
displays different behavior. When b2 4ac > 0 there are two distinct families of real
characteristic curves. In this case the PDE is termed hyperbolic. When b2 4ac = 0
a single family of real characteristic curves is found and the PDE is called parabolic.
Finally, when b2 4ac < 0 the RHS of (2.57) is complex and no real characteristic
curves exist. The PDE is termed elliptic in this case.
Example 2.5 Consider the wave equation (2.5) (using for wave speed to avoid
confusion in notation):
u tt 2 u x x = 0
Determine whether this equation is hyperbolic, parabolic, or elliptic, and determine,
if possible, its characteristic curves.
Solution:
The coefficients of the wave equation are: a = 1, b = 0, and c = 2 . We calculate
the discriminant to be 42 and since is real we know that 42 > 0 and that the
wave equation is therefore hyperbolic. Substituting the coefficients in (2.57) gives
the two characteristic curves: dt/dx1,2 = 1 .
Every PDE that is represented in an orthogonal coordinate system can be transformed to its canonical form. A coordinate transformation takes place and every
point in the original coordinate system is mapped onto the arbitrary coordinate system. In a twodimensional space, this would imply mapping the x and y coordinates
onto the plane: (x, y) (, ), which we also used for the wave equation. How
do we find these characteristic coordinate transformation? Well, that is quite simple:
we employ (2.57) to find the slope of the characteristic lines and integrating this
results in an expression for the characteristic curves.
36
Example 2.6 Find the characteristic curves for the wave equation of Example 2.5.
Solution:
For (2.5) we have dt/dx1,2 = 1 . Integrating this gives:
1
x +
1
t2 = x + ,
t1 =
where, and are integration constants. Rearranging these equations results in the
relations of (2.6).
In this arbitrary (orthogonal) coordinate system, the hyperbolic, parabolic, and
elliptic PDEs assume a particular canonical (natural) form. Let us examine these
canonical forms for each class of second order PDEs. We assume that the solution
u(x, t) is written in the transformed coordinates as (, ).
Hyperbolic PDE
(a)
(b)
t
region of
influence
t
region of
influence
characteristics
(x0 ,t0 )
domain of
dependence
(c)
region of =
influence
characteristic
(x0 ,t0 )
domain of
dependence
(x0 ,t0 )
domain of
dependence
Fig. 2.7 Domain of dependence and region of influence for the three different classes of second
order PDEs. a Hyperbolic. b Parabolic. c Elliptic
Parabolic PDE
either:
37
Similar to hyperbolic PDEs, the parabolic PDEs have a region of influence. However, this region is now unbounded by characteristics and therefore spans the entire
space (can also be time) beyond the point where the initial disturbance takes place
(see Fig. 2.7b). In addition, the entire region before this point influences this point
in the flow. Parabolic PDEs describe steady boundary layer equations along with
parabolized viscous flows.
Elliptic PDE The canonical form of an elliptic PDE can be written in its characteristic coordinates as follows:
+ = h 5 ( , , , , )
Any disturbance in a flow field that is described by elliptic PDEs is felt throughout
the entire flow. In other words, the entire flow domain forms the region of influence.
Conversely, every point in the flow domain influences any other point in the flow
domain. Therefore, the entire flow can also be viewed as the domain of dependence
(see Fig. 2.7c). Elliptic PDEs describe the steady subsonic, inviscid flow along
with an incompressible inviscid flow field.
Consider the twodimensional potential equation for steady, irrotational, inviscid
flow (we will derive this equation in Sect. 2.7.2):
y2
x y
x2
1 2 x x + 1 2 yy 2 2 = 0
c0
c0
c0
(2.58)
The velocity potential function is denoted with (x, y), with the velocities, x (x, y)
and y (x, y), as well as the speed of sound, c0 . The three coefficients of interest can
be identified as follows:
a =1
x2
c02
b=
2x y
c02
c =1
y2
c02
(2.59)
x2 + y2 c02
c02
= M 2 1,
(2.60)
where M is the Mach number of the flow. For subsonic flows M < 1 and the
potential equation is elliptic. When the flow is supersonic M > 1 and the equation
38
39
b1
n
..
ai bi
a b = (a1 an ) . =
bn
(2.61)
i=1
(2.62)
where is the measure of the smaller angle between a and b (0 180 ), a and
b are the vector magnitudes of vectors a and b, and n is a unit vector perpendicular
to the plane containing a and b in the direction given by the righthand rule. If the
vectors a and b are parallel (i.e., the angle between them is either 0 or 180 ), by
the above formula, the cross product of a and b is the zero vector 0. In the present
text the cross product in the threedimensional Euclidian space is considered and
the vectors a and b therefore each have three entries. This allows us to calculate the
entries for v using the thirdorder determinant:
i j k
v = a1 a2 a3 = (a2 b3 a3 b2 )i + (a3 b1 a1 b3 )j + (a1 b2 a2 b1 )k,
b1 b2 b3
(2.63)
where i, j, and k are the unit vectors in the respective directions x, y, and z.
Example 2.7 For the vectors a = (1, 2, 3) and b = (4, 5, 6), calculate the inner
product, the outer product and the vector product of a with b and b with a.
Solution:
The inner product is calculated according to (2.61):
a b = 1 4 + 2 5 + 3 6 = 32
40
4 5 6
(1, 2, 3)T (4, 5, 6) = 8 10 12
12 15 18
To calculate the vector product we use (2.63):
a b = (2 6 3 5, 3 4 1 6, 1 5 2 1) = (3, 6, 3)
b a = (a b) = (3, 6, 3)
In vector calculus we have two kinds of functions: scalar functions and vector functions. The output of a scalar function at a particular point P is a scalar:
f = f (P)
Vector function, g, is based on the vectorial function evaluation at a particular point P:
g = g(P) = (g1 (P), g2 (P), g3 (P))
The output of the vector function is a vector. The domain where a vector function is
defined is a region of space (can also be a surface or a curve in space). Within this
region (or on that surface or line) we say that this vector function defines a vector
field. In Cartesian coordinates a vector field can be denoted by
g(x, y, z) = (g1 (x, y, z), g2 (x, y, z), g3 (x, y, z)),
while a scalar function in Cartesian coordinates can be written as f (x, y, z).
41
i+
j+
k
x
y
z
(2.65)
We can define a scalar function whose gradient forms the three components of the
velocity field for irrotational, inviscid flow. This socalled potentialflow equation
reduces the system of equations to a single equation, thereby simplifying the problem
of finding the velocity field considerably. For example, the velocity distribution in a
fluid flow can be represented by the vector field V. At any point, P, throughout the
fluid this vector field can be related to the gradient of a scalar field, , at this point:
V(P) = (P). Since this is true throughout the entire physical domain we define
the velocity potential function as:
V
(2.66)
Example 2.8 Consider a fluid flow where the velocity vector field is given by V =
(2x, yz 2 , zy 2 ). Determine the velocity potential function that describes this flow, if
it exists.
Solution:
To find (x, y, z) find the primitive function of each of the components of the vector
field:
= 2xdx = x 2 + C1 (y, z)
y2 z2
= yz 2 dy =
+ C2 (x, z)
2
z2 y2
+ C3 (x, y)
= zy 2 dz =
2
It is possible to combine the above equations into a single expression for the following
potential flow function: = x 2 + 21 y 2 z 2 + C4 , where C4 is a constant. When we
employ (2.66) to , we find V . This confirms that exists and is indeed the potential
function of V .
g2
g3
g1
+
+
x
y
z
(2.67)
42
Example 2.9 For the vector field of Example 2.8 calculate the divergence.
Solution:
Employing (2.67) we obtain the following:
div V = V = 2 + z 2 + y 2
g3
g2
y
z
i+
g1
g3
z
x
j+
g2
g1
x
y
(2.68)
k
In a velocity field, the curl is called the vorticity vector and is related to the rotation
of the flow. Rotation of the flow is introduced due to the viscosity of the gas and leads
to turbulent conditions. We will see later that the rotation of a flow field is directly
tied to the formation of curved shock waves in supersonic flow via Croccos theorem.
Example 2.10 Determine the curl of the vector field of Example 2.8.
Solution:
We can directly apply (2.68):
curl V = V = (2zy 2zy)i + (0 0)j + (0 0)k = 0
In Example 2.10 we see that the specified vector field is irrotational. This agrees with
the fact that we can define a potential equation. This confirms that we can define a
potential function only when the vector field is irrotational.
43
through the surface equals the divergence of that vector field inside the enclosed
volume. For a velocity vector field, V , this theorem can be written as follows3 :
V dS =
V dV
(2.69)
The gradient theorem states that the gradient of a scalar field, p, integrated over
the control volume, equals the scalar field integrated over the surface vector of the
control volume:
pdS =
pdV
(2.70)
V
where dS is the element of the surface vector in the normal direction to the surface
pointing outward. Finally, we can relate the curl of a vector field over a control surface,
S, to the closed line integral of the vector field in a counterclockwise direction over
its boundary, C. This is the KelvinStokes theorem (often referred to as Stokes
theorem):
V dC =
( V ) dS
(2.71)
C
This theorem applies to any vector field, not only V but also mass flow or force fields.
44
(2.72)
where R is the gas constant for a specific gas. In terms of specific volume, v = 1/,
(2.72) can also be expressed as pv = RT . A new thermodynamic state variable is
defined as the sum of internal energy, e, and the product of pressure and specific
volume and is called specific enthalpy:
h = e + pv = e + RT
(2.73)
Even though specific enthalpy is introduced here without any physical motivation,
we will see later that it is indeed a fundamental state variable for aerothermodynamic
analysis. Both specific energy and specific enthalpy (which also has unit J/kg) are
both related to the temperature for a perfect gas:
e = e(T )
(2.74)
h = h(T )
(2.75)
In transonic flow we assume that the gas is perfect and we can consequently find
de = cv dT
(2.76)
dh = c p dT
(2.77)
where cv is the specific heat at constant volume and c p the specific heat at constant
pressure. These coefficients can be assumed to be constant for air up to temperatures
of 1,000 K, which is the basis for the calorically perfect gas assumption. Based on
this assumption we can now define the following relations between the specific heat
coefficients, cv and c p , the specific internal energy and enthalpy, e and h, the gas
constant, R, and the ratio of specific heats, :
45
e = cv T, h = c p T, =
cp
R
R
, cp =
,
, cv =
cv
1
1
R = c p cv
(2.78)
Using the above relations, the temperature and pressure can now be defined in terms
of the independent state variables e and :
p = ( 1)e, T =
( 1)e
R
(2.79)
For dry air at temperatures that are normally encountered during flight, the gas
constant equals R = 287.04 J/kg/K.
The viscosity () and thermal conductivity (k) are properties of a gas that are
both dependent on the temperature. The viscosity of the fluid is responsible for the
momentum transport on molecular level. The dynamic viscosity of an ideal gas is:
1
c
3
(2.80)
Here is the mean free path between molecules and c is the average velocity of the
molecules. Examining this equation we can see that the viscosity increases with the
average velocity and therefore with temperature. Sutherlands equation states that the
viscosity of air is dependent on its temperature according to the following relation:
=
0
T
T0
3/2
T0 + S
T +S
(2.81)
where 0 and T0 are a reference viscosity and temperature, respectively, and S is the
Sutherland temperature which is S = 110 K for air. For ISA (international standard
atmosphere) conditions these are 0 = 1.7894 105 kg/m/s and T0 = 288.16 K.
The thermal heat conductivity was already introduced in Fouriers law (2.28) and is a
property of the gas. For air k = 0.024 W/m/K at a temperature of 273 K. The Prandtl
number relates the viscosity and the thermal conductivity of a fluid according to:
Pr =
cp
(2.82)
Since k, , and c p are dependent on the temperature, the Prandtl number is dependent
on temperature as well. However, for air over a substantial temperature range (up to
600 K) we can assume the Prandtl number remains constant at 0.71 [3]. In Fig. 2.8
the viscosity and conductivity of air are presented as a function of temperature, along
with the assumed values for the constants that have been used. We see that there is
an appreciable change in conductivity and viscosity with temperature, which can be
important if we model highspeed fluid flows.
46
T = 288K, = 17.8mg/m/ s
Pr = 0.71 cp = 1003.5J/kg/ K
Viscosity, (mg/m/s)
30
25
20
15
10
200
300
400
500
Temperature, T (K)
600
45
40
35
30
25
20
15
200
300
400
500
Temperature, T (K)
600
Fig. 2.8 Graphical representation of (2.81) and (2.82) for dry air
(2.83)
Contrary to that, the heat and work depend on the path between the two states.
Therefore, their changes are represented with a in (2.83). To demonstrate this
process dependency Fig. 2.9 shows how the total amount of work changes with the
path taken from point 1 to point 2. Note that this figure shows expansion processes
where pV n = constant. This is a polytropic process, with p and V representing
the pressure and volume of the gas, respectively. The exponent, n, depends on the
process. For example if n = 0 we have an isobaric process, when n = 1 we have an
isothermal process (using the perfect gas law). Of course, we can define an infinite
number of ways to get from state 1 to state 2, each described by a different value of n.
If work is done by a system under adiabatic conditions (meaning no heat addition
or extraction, hence Q = 0) the internal energy (read: temperature) of the system
47
(2.85)
For a process that is reversible (no dissipative phenomena occur), we can calculate
the specific work (w) according to:
w=
2
1
pdv =
p2 v 2 p1 v 1
1n
p1 v1 ln vv21
(2.86)
Substituting the perfect gas law for pv in (2.86), we get the following expression for
nonisothermal processes:
R(T2 T1 )
(2.87)
w=
n1
Since we assume that the change in energy equals the work done on the system and
using (2.76), we can write:
e = cv (T2 T1 ) = w
(2.88)
Combining (2.87) and (2.88) leads, after simplification, to the following statement:
n=
cp
=
cv
(2.89)
which demonstrates that for an adiabatic and reversible process the following is true:
pv = constant
(2.90)
Let us recap the result of this process. We have made two assumptions in the
previous derivation: (1) no heat addition or extraction; (2) no dissipative phenomena
occur. Finally, we implicitly assumed that the total amount of mass within the system
48
Q rev
+ dsirrev
T
(2.92)
This statement says that ds has two parts, one is reversible and the other is irreversible
due to dissipation phenomena. The dissipative phenomena within the system always
increase the entropy of the system:
dsirrev 0
(2.93)
49
ds
Q
T
(2.94)
(2.95)
When a process is isentropic (literally meaning equal disorder) the entropy of the
system and its surroundings remains constant from state 1 to state 2.
Combining the first and second law of thermodynamics, (2.83) and (2.91), respectively, produces Gibbs equation:
T ds = de + pdv
(2.96)
If we use the chain rule we can find the change in enthalpy (2.73) to be dh =
de + pdv + vd p. Combining this with (2.96) we can now write two alternative forms
for the first law of thermodynamics, one in terms of specific energy and one in terms
of specific enthalpy:
T ds = de + pdv
T ds = dh vd p
(2.97)
(2.98)
By inserting the relations between temperature and specific energy (2.76) and specific
enthalpy (2.77), respectively, and by utilizing the state law (2.72) we can find two
expressions for ds: one in terms of specific volume and one in terms of specific
pressure (see Problem 2.22). Integrating both relations between the state properties
at state 1 and state 2 results in the following expressions for the change in entropy
of a calorically perfect gas:
T2
v2
+ R ln
T1
v1
T2
p2
s2 s1 = c p ln
R ln
T1
p1
s2 s1 = cv ln
(2.99)
(2.100)
50
Solution:
We calculate the pressure ratio to be p2 / p1 = 1.512 and enter this in (2.100) to find
s2 s1 = 0.746 J/kg/K. Rewriting (2.99) we get the following:
(s1 s2 ) + cv ln TT21
2
v1
=
= exp
1
v2
R
Substitution of the appropriate values and constants gives 2 /1 = 1.34.
The example above is actually a demonstration of how the state properties of the
flow change when it goes through a normal shock wave. Pressure, density, temperature, and entropy all increase depending on the Mach number of the flow in front of
the shock wave. In the above example this Mach number is 1.2 and the entropy change
is still relatively small. Notice, though, that even at this relatively low Mach number
the shock wave creates a relatively large pressure, temperature, and density rise.
T2
T1
T2
T1
cv /R
(2.101)
c p /R
(2.102)
=
and
=
R
1
R
1
With these substitutions and subsequent manipulations (see Problem 2.23) we can
now state the isentropic relations:
p2
=
p1
2
1
=
T2
T1
(2.103)
51
Now that we have the isentropic relations we ask ourselves: why are they so
important? We know that in reality a gas flow is viscous and thermally conducting
and therefore dissipative phenomena always occur. However, we also know that these
phenomena are only dominant in small regions in the flow: the boundary layer and
inside the shock waves. Outside of these regions the dissipative phenomena are so
small that they can often be neglected when we want to calculate the local state
properties. However, within these regions the isentropic relations do not hold and
we cannot relate the state properties in this simple way. This becomes evident in the
subsequent section and chapters of this text.
V l
(2.104)
where l is a characteristic length, for example the chord length of an airfoil or the
diameter of a pipe. A second nondimensional number to characterize a fluid flow is
its Mach number. The Mach number is defined as the ratio of the local flow velocity
to the local speed of sound:
V
M=
(2.105)
a
The local speed of sound, a, is a function of the state properties of the fluid. For a
perfect gas the speed of sound is solely dependent on the temperature of the gas:
a=
=
p
= RT
(2.106)
52
of the mass inside the control volume. We use the empirical law that mass can neither
be destroyed nor created during this process. Mass can only exit the control volume
through its enclosed surface, S. The resulting mass flowing out of the control volume
can be written as follows:
V dS
(2.107)
S
where V is the velocity vector. Note that the vector dS has a magnitude of dS and
a direction n, which is perpendicular to the surface of dS. We can express the time
rate of increase in mass inside V using the following volume integral:
dV
t
(2.108)
The change in mass within the control volume should be in balance with the net mass
flux through the control surface. This conservation principle results in the following
continuity equation in integral form:
S
V dS +
dV = 0
t
(2.109)
Equation (2.109) is the conservation of mass in integral form. Applying the divergence theorem to the surface integral combines both terms within one volume integral
(see Problem 2.24). From this integral the conservation form of the continuity equation can be found. The conservation of mass law in differential form is:
+ (V ) = 0
t
(2.110)
Let us briefly analyze this equation. It is a first order partial differential equation.
At a particular point in space it describes the change in density with time and the
divergence of the product of density and velocity. We can simplify this equation
even further by introducing the substantial derivative operator, D/Dt. The substantial derivative combines the local derivative (/t) and the socalled convective
derivative (V ), to represent the total derivative:
D
=
+V
Dt
t
(2.111)
=
+ (V ) =
+u
+v
Dt
t
t
x
y
(2.112)
53
This equation physically states that the change in density is due to temporal and
spatial variations. Using the definition of the substantial derivative (2.111) we can
rewrite the continuity equation (2.110) in substantialderivative form:
D
+ ( V ) = 0
Dt
(2.113)
When homogeneous, incompressible flow is considered, the first term on the LHS
can be dropped and only have V = 0. However, generally speaking this is only a
valid approximation at Mach numbers below M = 0.3 where density variations are
less than 5 %. In transonic flows with typical Mach numbers in the range of 0.81.3
we cannot use the incompressible flow assumption.
d
(mV )
dt
(2.114)
Let us again consider a fluid passing through a finite control volume. The forces
on the fluid in this control volume can be divided into forces that are acting on the
fluid (such as gravity) and forces that are acting on the control surface, S (pressure and shear forces). In Fig. 2.10 it is schematically shown how the body force,
pressure and stress vectors act on a control volume in Cartesian coordinates. Note
that ij is the stress on the surface of the control volume with normal vector i in
y
22
21
23
32
p
13
11
x
33
31
12
Fig. 2.10 Schematic control volume with body force (left), surface pressure (center) and shear
stresses (right)
54
the direction j. The stress components for which i = j are shear stress components, while the stress components normal to the surface (i = j) are associated with
the thermodynamic pressure. The thermodynamic pressure can be perceived as the
force exerted on the control volume walls as fluid molecules coincide with it during
their random movement. The latter becomes important only for such effects where
fluid compressibility is essential. Examples would include shock waves and sound
propagation. For incompressible flows the thermodynamic pressure is zero.
If we denote the body forces per unit volume by f we can represent the total
body force by:
f dV
(2.115)
body force =
V
The most common body force per unit volume is the gravitational force for which
f = g. The surface force due to pressure always acts perpendicular to the surface
of the control volume:
pdS
(2.116)
surface force due to pressure =
S
Note that the pressure force is negative, since the pressure acts in the opposite direction to the surface normal vector, n. Finally, the surface force due to friction can be
written as follows:
ij dS
(2.117)
viscous surface force =
S
where ij is the viscous shear stress tensor. In three dimensions this tensor is expressed
as follows:
11 12 13
ij = 21 22 23
(2.118)
31 32 33
A tensor can be perceived as a multidimensional vector. The inner product of a
tensor with a vector therefore results in a vector, rather than a scalar as we noted in
Sect. 2.3. In three dimensions, this tensor consists of three vectors, which form its
columns. The inner product should be applied to each of these vectors. The entries
of the resulting vector correspond to the scalar result of each of these operations. We
can combine each of the force components and form the LHS of (2.114):
F=
V
f dV
S
pdS +
ij dS
(2.119)
Now, let us turn our attention to the RHS of (2.114) and look at the time rate of
change of linear momentum. Similar to the components in the continuity equation
we distinguish two contributions to the time rate of change of momentum: first due
to the momentum change with time of the fluid inside the control volume and second
55
due to momentum entering and leaving the control volume. These components can
be defined as follows:
d
(mV ) =
V dV +
(V dS)V
dt
t
V
(2.120)
We can now combine (2.119) and (2.120) to obtain a first version of the fluidflow
momentum balance in integral form:
V dV + (V dS)V =
f dV
pdS +
ij dS (2.121)
t
V
We can expand the shear stress terms in this equation in terms of state variables
and fluid constants, but let us first evaluate the differential form of the momentum
equation.
By applying the gradient and divergence theorems (Sects. 2.3.2 and 2.3.3, respectively) appropriately to each of the individual terms in (2.121) we can rewrite this
equation only in terms of volume integrals, which allows us to evaluate only the
integrand. This results in the following differential form of the momentum equation:
(V ) + (V V ) = f p + ij
t
(2.122)
Note that V V is the outer product of V with itself, leading to the secondorder
velocity tensor. For Newtonian fluids (stressrate of strain relation is linear) the
relation between the shear stress tensor and velocity the components can be written
as follows:
u j
u k
u i
+ ij
+
i, j, k = 1, 2, 3
(2.123)
ij =
x j
xi
xk
where is the dynamic viscosity and is second (or bulk) coefficient of viscosity
of the fluid. Furthermore, u i is the velocity component of V in the direction of i and
ij is the Kronecker delta.4 Following Stokes hypothesis:
2
=
3
(2.124)
u j
u i
+
x j
xi
2 u k
ij
3 xk
1 for i = j
0 for i = j
(2.125)
56
By expanding the LHS of (2.122) using the chain rule and subsequent substitution
of the continuity equation (see Problem 2.25), we can write the momentum equation
in substantialderivative notation according to:
DV
= f p + ij
Dt
(2.126)
u j
u i
2
DV
u k
ij
+
= f p +
Dt
x j
x j
xi
3
xk
(2.127)
We emphasize that (2.127) is a set of three secondorder partialdifferential equations (if threedimensions are considered). These equations are often referred to as
the NavierStokes equation named after ClaudeLouis Navier and George Gabriel
Stokes.
(2.83)
Note that (2.83) is the balance between the rate of change in internal (or specific)
energy with the sum of the rate of specific heat added to the fluid and the rate of
specific work done by the fluid. In order to express the first law in energy per unit
volume we need to multiply (2.83) by the density:
de + w = Q
(2.128)
Let us expand each of those terms in terms of state variables of the fluid. We start
with the LHS of (2.128), which is the change of internal energy. Remember, that this
law is written for a gas in stationary condition. Since we are considering a moving
fluid, the kinetic energy per unit mass is to be added. If we denote the total energy
per unit volume with E t we have:
V2
Et = e +
2
(2.129)
Since the subject of this text is aerodynamics, we have intentionally omitted a potential energy term in (2.129). The time rate of change of total energy in the control
volume can now be written as:
57
E t dV
t
V
(2.130)
Since fluid is entering and leaving the control volume, the kinetic and internal energy
changes. Each particle that leaves the control volume has a volumespecific energy
density of E t . The total rate of volumespecific total energy leaving the control
volume therefore equals the volumetric flux over the boundary dS times E t : E t V dS.
Integrating over the entire control surface yields the following:
E t V dS
(2.131)
time rate of energy transfer across control surface =
time rate of change of total energy inside control volume =
Now, let us turn our attention to the heat term in (2.128). First, we consider the
time rate of change of heat inside the control volume. Similar to the energy flux, we
define two components: the time rate of change of heat inside the control volume
and the heat flow going in and out of the volume via the control surface. We switch
in nomenclature from specific heat ( Q, unit: J/kg) to heat per unit volume ( Q,
unit: J/m3 ). The time rate of change of heat inside the control surface is given by:
time rate of change of heat inside control volume =
QdV
t
(2.132)
By using Fouriers law of heat conduction (see also (2.28) in Sect. 2.2.2), the heat
flow (q, unit: J/s/m2 ) is linearly related to the temperature gradient:
q = kT
(2.133)
where k is the coefficient of thermal conductivity and T is the temperature of the gas.
The heat flow into the control volume is in opposite direction to the control surface
vector dS. Therefore, the rate of heat addition is negatively related to the heat flux
over the control surface dS:
q dS
(2.134)
heat flow across control surface =
S
Having examined the energy and heat flux in the control volume, the last part of
(2.128) is the rate of work being done on the fluid in the control volume. We consider
three different contributions that are related to the three different forces we defined
in Sect. 2.5.2. First we have the time rate of change in work due to the body forces.
We know that work can be defined as the force times a displacement. Hence, the
time rate of change of work can be expressed as the force multiplied by the velocity.
Therefore, we have the following:
time rate of change of work due to body forces =
V
f V dV
(2.135)
58
Secondly, we have the pressure on the control surface that does work on the fluid
inside the control surface. We take the inner product of the RHS of (2.116) with V
to obtain the work flux due to this component:
time rate of work done due to pressure on the control surface =
pdS V
(2.136)
Finally, we look at the rate of work done due to the shear stress on the control surface.
Similar to the pressureinduced work flux, the work flux due to the shear stress can
be found by taking the inner product of the RHS of (2.117) with V :
time rate of work done due to shear stress on the control surface =
( ij dS) V
(2.137)
Remember that ij is the stress tensor as defined in (2.118).
We have now defined the time derivative of each of the components of (2.128) and
we can formulate the first version of the energy balance between the control volume
and its surroundings:
E t dV +
E t V dS =
QdV
q dS +
f V dV
t
t
S
S
V
V
V
pdS V +
( ij dS) V
(2.138)
S
We can cast (2.138) in differential form by applying the gradient and divergence
theorems to the appropriate terms and changing everything to volumetric integral
formulation. The integrand of that equation yields the following energy balance:
Et
(Q)
+ Et V =
q + f V ( p V ) + ( ij V ) (2.139)
t
t
The first term on the LHS of (2.139) represents the rate of increase of E t in the control
volume, while the second term represents the total energy lost due to convection
through the control volume. The first term on the RHS is the amount of heat produced
by external factors. The second term is the rate of heat lost by means of conduction.
The third term represents the work done by body forces while the last two terms
represent the work done by normal and shear stresses on the surface, respectively.
By subsequently employing the continuity and momentum equation (see Problem
2.26) we can write the energy equation in a convenient substantialderivative form:
(Q)
De
+ p( V ) =
q + ( ij V ) ( ij ) V
Dt
t
(2.140)
The sum of last two terms in this equation is termed the dissipation function, , and
represent the rate at which mechanical energy is expended due to viscosity when
59
the fluid is deformed. The LHS of (2.140) can be rewritten in terms of enthalpy
by employing (2.73) and the continuity equation (see Problem 2.27). The following
energy balance results:
D p (Q)
Dh
=
+
q+
Dt
Dt
t
(2.141)
(2.142)
D(e + V 2 /2)
= ( p V )
Dt
(2.143)
Here, we have assumed the body forces, f , to be negligible. Using the continuity
equation (2.113), we can rewrite this equation as follows:
p
D(h + V 2 /2)
=
Dt
t
(2.144)
V2
= H = constant
2
(2.145)
where H is the definition of the total enthalpy. If we use the relation between enthalpy
and static temperature (2.77), we can write:
cpT +
V2
= c p Tt = constant
2
(2.146)
where Tt is the total temperature. For a flow where all the streamlines emanate from
the same uniform freestream, the temperature and velocity are therefore uniquely
correlated through (2.146). We can manipulate (2.146) and include the definition of
the speed of sound (2.106) to obtain:
1 2
Tt
=1+
M
T
2
(2.147)
60
Equation (2.147) states that there exists a unique relationship between the Mach
number and the static temperature in the flow. For a given total temperature, Tt , in
the freestream, Eq. (2.147) shows that the static temperature decreases with Mach
number. If we combine (2.147) with the isentropic relations (2.103) we can express
the temperature, density and pressure as a function of the Mach number. We have
tabulated this relation for Mach numbers ranging from 0 through 10 in Appendix A.
d(Q)
=0
dt
(2.148)
Let us rewrite the three equations that represent the conservation of mass (2.110),
momentum (2.122) and energy (2.139), respectively, and apply the assumptions of
(2.148):
+ (V ) = 0
(2.110)
t
(V ) + (V V ) + p ij = 0
t
Et
+ E t V + q + p V ( ij V ) = 0
t
(2.149)
(2.150)
This shortened form permits us to explicitly rewrite all five equations in their conservation form:
(u) (v) (w)
+
+
+
=0
(2.151a)
t
x
y
z
x y
(u) (u 2 ) (uv) (uw) p
x x
x z
+
+
+
+
+
+
= 0 (2.151b)
t
x
y
z
x
x
y
z
yy
yz
x y
(v) (uv) (v2 ) (vw) p
+
+
+
+
+
+
= 0 (2.151c)
t
x
y
z
y
x
y
z
yz
(w) (uw) (vw) (w2 ) p
x z
zz
+
+
+
+
+
+
= 0 (2.151d)
t
x
y
z
z
x
y
z
61
qz
qx
+
(ux x + vx y + wx z )
+
x
y
z
x
+ (ux y + v yy + w yz ) +
(ux z + v yz + wzz ) = 0
y
z
(2.151e)
(2.152)
where:
u
U = v
w
Et
v
uv + p x y
F=
v2 + p yy
vw yz
(E t + p)v ux y v yy w yz + q y
u
u 2 + p x x
uv x y
uw x z
(E t + p)u ux x vx y wx z + qx
w
uw x z
vw yz
G=
2
w + p yz
(E t + p)w ux z v yz wzz + qz
E=
(2.153)
The first and last row in these vectors represent the continuity and energy equation,
respectively. The three middle rows represent the three components of the momentum
equation. The form above is often used because it is easier to code in numerical form.
It still represents the full equations of motion. Note that this set in the subsequent
text is referred to as the NavierStokes (NS) equations.
62
v = v + v
h = h + h
w = w + w
T = T + T
= +
H = H + H
(2.154)
(2.155)
1 t
u dt = 0
t 0
(2.156)
The equations of motion of Sect. 2.5 are modified by substitution of the flow parameters that we have defined in (2.155). Subsequently, each of the equations is time
averaged. The result yields a set of equations that have averaged fluctuating terms.
Each of the terms in the equations of motion that have averaged fluctuating terms
63
are subsequently set to zero. This yields a more compact version of these equations,
including the fluctuating terms. We briefly discuss each of the governing equations
below.
When substituting the relations of (2.154) in the continuity equation (2.110) the
following averaged equation can be obtained:
u + u j = 0
+
t
x j
(2.157)
For incompressible flow, the fluctuating and unsteady density terms can be dropped
from this equation. Furthermore, = = constant. What we are left with is a much
reduced continuity equation:
u j
=0
(2.158)
x j
Starting from the NavierStokes equation (2.127) and neglecting body forces,
the Reynoldsaveraged momentum equation can be obtained by substitution of the
relations (2.154) and applying subsequent time averaging:
u i + u j +
u i u j + u i u j
t
x j
p
=
ij u j u i j u i u j u i u j
+
xi
x j
where
ij =
u j
u i
+
x j
xi
2 u k
ij
3 xk
(2.159)
(2.160)
p
ij u i u j
u i u j =
+
(u i ) +
t
x j
x j
x j
(2.161)
64
D u i
Dt
p
=
xi
Particle acceleration
of mean motion
Mean pressure
gradients
ij lam
x j
Laminarlike
stress gradient
for the mean motion
ij turb
x j
(2.162)
This shows how the Reynolds stresses appear as an addition to the mean flow parameters that are very much comparable to the original NS equations. The Reynoldsaveraged stress is related to the fluctuating velocities according to:
ij turb = u i u j
(2.163)
T
u j H + u
H H +
j H + u j H + u j H + u j H k
t
x j
x j
u j
p
2
u k
u i
=
+ u i
+
u i ij
+
t
t
3
xk
xi
x j
!
u j
u
2
k
i
ij u i
+ u i
+ u i
t 3
xk
xi
x j
(2.164)
For incompressible flows the energy equation can be written as follows:
H
T
u j H + u j H k
+
t
x j
x j
!
u j
u
u j
p
u i
+ u i
+
=
+
+ u i i
u i
t
x j
xi
x j
xi
x j
(2.165)
Since the last bracketed term within the square brackets on the RHS of (2.165) is often
small compared to the other terms within the square brackets, it is usually neglected.
What remains is an energy equation comparable to the original energy equation
(2.140) except for the fluctuating term u j H . Since this term shares its brackets
with heat flux terms it is often termed the Reynolds heat flux. A similar analysis can
be done for the energy equation where the Reynolds heat flux component is:
( q)turb =
c
p T u j c p T u j u j c p T
x j
(2.166)
65
The Reynoldsaveraged equations form a set of five partial differential equations with
more than five unknowns. As was briefly explained, additional fluctuating terms need
to be related to mean flow properties via appropriate turbulence models. We discuss
a typical turbulence model in Sect. 2.6.3.
v =
w =
h
h =
T
T =
H
H =
(2.167)
v = v + v
h = h + h
w = w + w
T = T + T
= +
H = H + H
(2.168)
where the primed parameters stand for time dependent deviations from the average
t
1
which are zero when integrated over time (i.e. u = t
0 u dt = 0). The averages
of the doubly primed fluctuating quantities are not zero. Instead the time average
of the doubly primed fluctuations multiplied by the density equals zero (i.e. u =
t
1
t 0 u dt = 0).
We substitute the averaged flow parameters (2.168) in the equations of motion
of Sect. 2.5. Substituting the averaged flow parameters into the continuity equation
(Eq. 2.110) and subsequently time averaging the equation yields:
+
u j = 0
t
x j
j = 1, 2, 3
(2.169)
66
p
u j +
ij u i u j
i, j = 1, 2, 3 (2.170)
u i u j =
+
t
x j
xi
x j
where, neglecting viscosity fluctuations, ij is as follows:
ij =
u j
u i
+
xi
x j
!
u j
u i
2 u k
2 u k
+
ij
ij
+
3 xk
xi
x j
3 xk
(2.171)
A brief look at the expression reveals a more complex expression than the original,
nonaveraged expression of Eq. (2.125). In practice, however, the viscous terms with
doubly primed fluctuations are likely candidates for being neglected based on their
magnitude compared to the massaveraged variables. By substituting the Reynoldsaveraged continuity equation (2.169) in the Reynoldsaveraged momentum equation
(2.170) and employing the substantial derivative yields the following [19]:
p
=
xi
D u i
Dt
Particle acceleration
of mean motion
Mean pressure
gradients
ij lam
x j
ij turb
(2.172)
x j
Laminarlike
stress gradient
for the mean motion
Note that the Reynoldsaveraged momentum equation (above) has the same form
as the original NavierStokes equation, Eq. (2.127), with the addition of a turbulent
stress term. Explicitly, the laminar and turbulent stress terms are given by:
ij lam =
ij turb = u i u j +
u j
u i
+
xi
x j
u j
u i
+
xi
x j
2 u k
ij
3 xk
!
2 u k
ij
3 xk
(2.173)
(2.174)
To arrive at the Reynolds form of the energy equation, the massweighted variables
(2.168) are substituted in (2.140). Subsequent elimination of terms that go to zero
yields the following Reynolds energy equation in massweighed variables:
p
T
u j H + u j H j k
=
H +
+
u i ij + u i ij
t
x j
x j
t
x j
(2.175)
A similar analysis can be performed on the energy equation. Apart from the laminar and turbulent stress terms a laminar and turbulent heat flux term can be
defined. The new apparent turbulent stresses and flux terms that appear in both the
Reynoldsaveraged momentum and energy equation should be treated as new variables. Therefore, additional equations are required that make assumptions regarding
the apparent turbulent quantities and the mean flow variables.
67
u k
2
= 2T Sij ij T
+ k
3
xk
(2.176)
1
2
u j
u i
+
x j
xi
(2.177)
a transport equation is
To predict the value of the turbulent kinetic energy, k,
developed from the Reynoldsaveraged NavierStokes equations. Using Boussinesqs
assumption for eddy viscosity (2.176) this transport equation reads (in substantialderivative form):
D k
=
Dt
x j
k
u i
T
k 3/2
2
+
+ 2T Sij kij
(2.178)
CD
Prk x j
3
x j
l
A derivation of this equation is beyond the scope of this text but can be found in
texts on turbulence modeling (i.e. [22] or [6]). In (2.178) the Prandtl number for
turbulent kinetic energy (Prk ) appears as a closure constant (Prk = 1.0). C D is the
dissipation coefficient and has been experimentally shown to be C D = 0.164. The
The
term on the LHS of this equation represents the particle rate of increase of k.
terms on the RHS of Eq. (2.178) represent the diffusion rate, the generation rate
respectively. Here l is a characteristic length, referred
and the dissipation rate of k,
to as the mixing length. The mixing length can be interpreted as follows: a fluid
parcel will conserve its properties for a characteristic length, l, before mixingwith the
68
surrounding fluid. Prandtl [14] described that the mixing length may be considered
as the diameter of the masses of fluid moving as a whole in each individual case; or
again, as the distance traversed by a mass of this type before it becomes blended in
with neighboring masses
The dissipation rate of k which is embedded in the last term in (2.178) is often
3/2
a transport equation
represented with : = C D k l . Similar to the analysis of k,
developed from the Reynoldsaveraged NavierStokes equations can be established
for :
u i
T
2
2
D
+
+ C1
=
2T Sij kij
C2
Dt
x j
Pr x j
3
x j
k
k
(2.179)
The term on the LHS of Eq. (2.179) represents the particle rate of increase in dissipation, while the terms on the RHS represent the diffusion, generation and dissipation
rates of . In terms of k and the terms in Eqs. (2.178) and (2.179) are as follows:
l = CD
k 3/2
T = C
k 2
4/3
C = C D
(2.180)
T
x j
(2.181)
Analogous to the laminar heat conduction coefficient, k, the turbulent heat conductive
heat coefficient is defined according to:
kT =
c p T
PrT
(2.182)
where PrT is the turbulent Prandtl number that is most commonly takes on the value
of PrT = 0.9. The combination of (2.181) and (2.182) form the closure for the
energy equation. That is, if the last bracketed term in (2.175) is neglected because it
is small compared to the other terms within the square brackets. Table 2.1 displays
model.
the constants that are to be used in the k
Table 2.1 Model constants
model [19]
for k
C
0.09
C1
1.44
C2
1.92
Prk
1.0
Pr
1.3
PrT
0.9
69
The kepsilon method that has been presented is tailored towards closure of the
incompressible Reynoldsaveraged equations of motion. Similar models have been
developed for the closure of the Reynoldsaveraged, massweighted equations of
motions that were developed in Sect. 2.6.2. The closure of the compressible Reynolds
equations is beyond the scope of this text. Even though in transonic aerodynamics the
flow is compressible, the incompressible turbulence models have been proven to give
method for compressible flow is treated
good predictions up to Mach 5 [17]. The k
extensively by Launder and Spalding [12] and Mohammadi and Prionneau [13].
= p
(2.183)
Dt
70
Neglecting the viscous and heattransfer terms in the energy equation (2.140) results
in the following:
De
+ p( V ) = 0
(2.184)
Dt
Alternatively, this equation can be written in terms of enthalpy, h, by modifying
(2.141):
Dp
Dh
=
(2.185)
Dt
Dt
Equations (2.113), (2.183), and (2.184) are generally known as the Euler equations.
The compressible Euler equations can be written in conservation form according
to (2.152). However, the vector representation is now simpler than for the full NS
equations:
U=
v
w
Et
u
u 2 + p
E=
uv
uw
(E t + p)u
v
uv + p
2
F=
v + p
vw
(E t + p)v
w
uw
G=
vw
w2 + p
(E t + p)w
(2.186)
V2
2
V
(2.187)
V2
2
1
V = p
(2.188)
The vorticity can be related to the specific entropy, s, according to Croccos equation:
V
V = T s
t
V2
2
(2.189)
This equation can be derived from the first and second laws of thermodynamics,
(2.83) and (2.94), respectively. Note that the LHS of (2.189) is also present in the
Lagrange form of the momentum equation (2.188). A physical interpretation of
71
2.0
1.5
1.0
0.5
0.0
0.5
1.0
RAE 2822
M =0.750
Re =6.20010 6
=2.734
CL =0.7431
CD =0.02284
CM =0.0941
L/D =32.54
Experiment
3.19
0.743
0.0242
0.106
72
=3.0
Cl=0.60
1.2
Cl=1.27
M=1.0
0.8
0.4
0.0
M=0.68
Expected
Inviscid
1.2
M=1.0
0.8
0.4
0.0
0.4
0.4
0
Cl=0.60
1.6
1.6
2.0
.2
.4
.6
.8
1.0
.2
.4
.6
.8
1.0
Fig. 2.12 Comparison between predicted and measured pressure distribution at constant angle of
attack (left) and constant lift coefficient (right) (after Ref. [5])
the same angle of attack as in the experiment (left) and at the same lift coefficient
as in the experiment (right). None of these predictions gives an accurate result.
This shows that we must be extremely careful when assessing the results from a
purely inviscid solver. Given the short computation time of viscousinviscid solvers
such as MSES, it is therefore advised to include the presence of the boundary layer
when predicting the pressure distribution over a body. In particular at lowReynolds
numbers and transonic Mach numbers this is an important prerequisite for obtaining
reliable results.
(2.191)
73
In Cartesian coordinates, we align the xaxis with the freestream velocity, V , and
represent the local velocity components, u, v and w by
= x
x
= y
v=
y
w=
= z
z
u=
(2.192a)
(2.192b)
(2.192c)
If we assume the flow to be steady, the continuity equation (2.113) reduces to:
(V ) = 0
(2.193)
(x ) +
( y ) +
(z ) = 0
x
y
z
(2.194)
Note in this equation that the partials x , y , and z appear. We will find an expression
for each of these density derivatives by considering the potential form of the steady,
inviscid, irrotational momentum equation.
If we assume that the flow is steady (i.e. /t = 0) the Lagrange form of the
momentum equation (2.188) can be reduced to:
V2
2
1
= p
(2.195)
V2
d p = d
2
= d
x2 + y2 + z2
2
!
(2.196)
In isentropic flow the speed of sound (a) is given by (2.106). We can substitute the
expression for the speed of sound (2.106) in (2.196) and obtain:
x2 + y2 + z2
d = 2 d
a
2
!
(2.197)
x2 + y2 + z2
2
!
(2.198a)
74
y = 2
a y
x2 + y2 + z2
z = 2
a z
x2 + y2 + z2
!
(2.198b)
!
(2.198c)
x y x y
y z yz
x z x z
2
2
=0
a2
a2
a2
(2.199)
1.5
1.5
1.0
1.0
Cp*
0.5
0.0
0.5
2y/b = 0.206000
1.0
0.2
0.4
0.6
Cp*
0.5
0.0
0.5
2y/b = 0.521000
1.0
MatricsV
Flight test
1.5
0.0
In the late 1980s and early 1990s the analysis code MatricsV was developed by
the National Aerospace Laboratory of the Netherlands to estimate the aerodynamic
forces on wingfuselage combinations (threedimensional) in highsubsonic conditions. This code relies on a viscousinviscid formulation, where the full potential
equation (2.199) is numerically solved in the inviscid domain outside the boundary
layer. In the boundary layer the boundary layer equations are numerically solved.
Subsequently, an interaction algorithm is used to match the flow properties at the
0.8
1.0
1.5
0.0
MatricsV
Flight test
0.2
0.4
0.6
0.8
1.0
Fig. 2.13 Comparison of predicted and measured pressure distributions about two sections of the
Fokker 100 wing (from [20]). Note that this is a twodimensional representation of results that were
obtained from threedimensional calculations (MatricsV)
75
edge of the boundary layer to the inviscid flow outside the boundary layer [20]. This
analysis code was specifically developed to investigate the flow about wingbody
combinations. The objective was to develop a code that could produce results in less
than week of computation time. Due to the increase in computational power, this has
now (2015) reduced to less than five minutes on a personal computer, including the
generation of the mesh. In Fig. 2.13 a comparison between the predicted and measured pressure distribution at two spanwise wing stations of the Fokker 100 can be
seen. We see that the prediction of the pressure distribution is in excellent agreement
with the measured data at the same angle of attack. The accurate prediction in combination with the short computation time make this type of code a viable candidate
in the early stages of preliminary wing design.
2.8 Summary
In the preceding sections we have briefly reviewed the fundamental tools that the
reader needs to be familiar with in order to properly understand the material that
is presented in the subsequent chapters. We have done a very basic review of partial differential equations, their solution methods, and their classification. We have
also reacquainted ourselves with the mathematical operations in vector algebra. We
have explained the fundamental laws of thermodynamics: the state law that relates
the thermodynamic state variables to each other, the first law that balances work,
heat and internal energy, and the second law that tells us the direction in which a
thermodynamic process takes place. Finally, we have applied all this knowledge in
defining the equations of motion of fluid flow. When expanded we have shown that
the equations of motion are a coupled system of five partial differential equations
[(2.110), (2.127) (3 components), and (2.139)] with five unknowns: , e, and the three
velocity components, u, v, w. These equations are often referred to as the NavierStokes equations and they can be cast in different forms: integral form, substantialderivative form, and conservativederivative form. Because solving these equations
requires vast computational resources we have presented derivatives of these equations. In the Reynoldsaveraged NavierStokes equations the flow is decomposed into
a mean flow with fluctuating terms. A turbulence model can then be used to relate
those fluctuating terms to the mean flow parameters. A further simplification of the
NavierStokes equations is obtained through the assumption of having an inviscid
and nonconducting flow. The resulting equations are often referred to as the Euler
equations and they can provide an accurate prediction of the flow about a body, in
particular when they are used in parallel with the boundarylayer equations. If also
the assumption is made that the flow is irrotational, the Euler equations reduce to
the fullpotential equation. Solving the threedimensional fullpotential equation in
combination with the boundarylayer equations can still produce accurate predictions
of transonic flow about bodies, as long as the shock strength is relatively weak.
76
Problems
Review of Partial Differential Equations
2.1 Show that (2.7) results from (2.5) by using the chain rule and substituting =
x + ct and = x ct.
(xct)
+
sin
is a solution to the
2.2 Demonstrate that u(x, t) = sin (x+ct)
L
L
following problem:
u tt = c2 u x x
x
u(x, 0) = 2 sin
L
u(0, t) = u(L , t) = 0
2.3 Consider a string of steel wire, measuring 1 m in length and weighing 0.5 N. It
is stretched by a force of 100 N. What is the corresponding speed c of the transverse
waves?
2.4 Consider the wave equation (2.5) with the following boundary conditions:
u(0, t) = u x (L , t) = 0 .
(a) Use separation of variables technique to calculate the eigenvalues, eigenfunctions
and general solution.
(b) Now, assume L = and c = 1. With initial conditions u(x, 0) = 0 and
u t (x, 0) = 1, calculate the solution for u(x, t).
(c) With initial conditions u(x, 0) = sin(x/2) and u t (x, 0) = 2 sin(x/2)
3 sin(5x/2) calculate the solution for u(x, t).
2.5 Demonstrate that (2.22) can be derived from u t = 0 by using the chain rule and
the transformation (2.6).
2.6 Show that the Jacobian for the change in characteristic coordinates (2.6)
equals 2c.
2.7 For the same problem as in Example 2.2, calculate the solution for:
(a) x < t, t > 0, x + t < 0, and check u tt u x x = 1.
(b) x > 1, t < x < t + 1, and check u tt u x x = 0.
2.8 Show that (2.43) is indeed a solution of (2.41).
2.9 Demonstrate that from (2.49) we can obtain (2.50) by using the method outlined
in the text.
2.10 Using the separationofvariables technique find a solution to the following
problem:
u t = ku x x
(0 x L)
2.8 Summary
77
u(0, t) = 0
u x (L , t) = 0
u(x, 0) = f (x)
(t > 0)
(t > 0)
(0 < x < L)
2.11 For Problem 2.10, provide calculate Dn in closed form for L = and
f (x) = x.
2.12 Consider the homogenous heat equation, c2 u x x u t = 0. Determine whether
this equation is hyperbolic, parabolic, or elliptic.
2.13 Consider the potential equation on p. 74 (2.199). When M > 1, determine the
characteristic coordinates and write the equation in its canonical form.
Review of Vector Algebra
2.14 Let a = (2, 5, 9) and b = (4, 2, 7). Calculate the following:
(a)
(b)
(c)
(d)
(e)
a+b
ab
ab
ba
ab
2.15 For the following velocity vector fields, find the potential function if it exists.
Check that (2.66) holds in each case. (x, y, z): (a) (x, 2y, 3z), (b) (yz, x z, x y),
(c) (ye x , e x , 1), (d) (2y, 5x, 0).
2.16 For the vector fields of Problem 2.15, calculate the divergence.
2.17 For the vector fields of Problem 2.15, calculate the curl. Compare your results
to those of Problem 2.15. What do you notice?
Review of Thermodynamics
2.18 With c p as given in Fig. 2.8 and R as given in the text, calculate cv , , e, and
h for air at an altitude of 10 km under ISA conditions.
2.19 For the values given in Problem 2.18 also calculate the viscosity of the air,
and the thermal conductivity, k.
2.20 Calculate the change in specific internal energy, e, for air that is being compressed isentropically. Assume that the initial temperature is 288 K and the final
temperature, due to compression, 340 K.
2.21 Show that (2.97) and (2.98) can be derived by combining the first and second
law of thermodynamics, (2.83) and (2.94), respectively.
78
2.22 Derive the following relations by starting from (2.97) and (2.98), respectively.
dv
dT
+R
T
v
dT
dp
ds = c p
R
T
p
ds = cv
(2.200)
(2.201)
2.23 Demonstrate that (2.103) can be derived by combining (2.101) and (2.102) and
applying the substitutions as described in the text.
Equations of Fluid Motion
2.24 Use the divergence theorem (2.69) to prove that (2.109) and (2.110) are mathematically identical.
2.25 Consider the LHS of (2.122). Show that by using (2.110) this can be simplified
to DV
Dt .
2.26 Consider the energy equation (2.139).
(a) By employing the continuity equation (2.110) show that the following is true:
Et
D(E t /)
=
+ Et V
Dt
t
De
DV
D(E t /)
=
+
V
Dt
Dt
Dt
(c) Use (2.122) to show that the second term in the equation above can be written as:
DV
V = f V p V + ( ij ) V
Dt
(d) Using the three equations above, demonstrate that (2.140) is identical to (2.139).
2.27 Consider the energy equation in substantialderivative form (2.140).
(a) Show, by employing the continuity equation, that the following identity is true:
Dh
Dp
De
+ p( V ) =
Dt
Dt
Dt
(b) Explicitly write out the dissipation function, , in terms of the velocity components in threedimensional Cartesian coordinates. Use the assumption that
= 23 .
2.28 Write out (2.110), (2.149), and (2.150) in threedimensional Cartesian coordinates and demonstrate that you can write this as (2.152).
2.8 Summary
79
References
1. Anderson, J.: Computational Fluid Dynamics: The Basics and Applications, 1st edn. McGraw
Hill, New York (1995)
2. Anderson, J.: Modern Compressible Flow with Historic Perspective, 3rd edn. McGraw Hill,
New York (2003)
3. Anderson, J.: Fundamentals of Aerodynamics, 5th edn. McGraw Hill, New York (2010)
4. Anon.: Direct numerical simulation (DNS) wikipedia. http://en.wikipedia.org/wiki/Direct_
numerical_simulation (2012)
5. Blackwell, J.A.: Scale effects on supercritical airfoils. In: Proceedings of ICAS, pp. 370283
(1978)
6. Cebeci, T.: Analysis of Turbulent Flows. Elsevier, Amsterdam (2004)
7. Cook, P.H., McDonald, M.A., Firmin, M.C.P.: Aerofoil RAE 2822pressure distributions,
and boundary layer and wake measurements. In: AGARD138 (1979)
8. Drela, M., Giles, M.B.: Viscousinviscid analysis of transonic and low Reynolds number airfoils. AIAA J. 25(10), 13471355 (1987). doi:10.2514/3.9789
9. Farokhi, S.: Aircraft Propulsion, 2nd edn. Wiley, Chichester, UK (2014)
10. Ferrari, C., Tricomi, F.G.: Transonic Aerodynamics. Academic Press, New York (1968)
11. Keener, J.P.: Principles of Applied Mathematics. Westview Press, Boulder (1999)
12. Launder, B.E., Spalding, D.B.: Lectures in Mathematical Models of Turbulence. Academic
Press, London (1972)
13. Mohammadi, B., Pironneau, O.: Analysis of the Kepsilon Turbulence Model. Wiley, Chichester, UK (1994)
14. Prandtl, L.: ber die ausgebildete turbulenz. In: Proceedings of the Second International
Congress of Applied Mechanics, pp. 6275. Orell Fssli Verlag, Zurich (1927)
15. Reynolds, O.: On the dynamical theory of incompressible viscous fluids and the determination
of the criterion. Philos. Trans. R. Soc. Lond. 186, 123164 (1895)
16. Ruijgrok, G.J.J.: Elements of Airplane Acoustics. Eburon, Delft (2003)
17. Schlichting, H., Gestern, K.: Boundary Layer Theory, 8th edn. Springer, Berlin (1999)
18. Strauss, W.A.: Partial Differential Equations: An Introduction. Wiley, Chichester (1992)
19. Tannehill, J.C., Anderson, D.A., Pletcher, R.H.: Computational Fluid Mechanics and Heat
Transfer, 2nd edn. Taylor & Francis, Philadelphia (1997)
20. Van Muijden, J., Broekhuizen, A.J., Van der Wees, A.J., Van der Vooren, J.: Flow analysis
and drag prediction for transonic transport wing/body configurations using a viscousinviscid
interaction type method. In: Proceedings of the 19th ICAS Congress. Anaheim, California
(1994)
21. Van Wylen, G., Sonntag, R.: Fundamentals of Classical Thermodynamics. Wiley, Chichester,
UK (1973)
22. Wilcox, D.: Turbulence Modeling for CFD, 2nd edn. DCW Industries, Canada (1998)
Chapter 3
Abstract This chapter starts the foundation with the small disturbance theory
applied in the subsonic and supersonic flows. Both of these regimes are linear and
thus fail at sonic speed, which is at the heart of transonic flow. Nonlinear small
disturbance theory for the transonic regime is subsequently derived and discussed.
With limited analytical options for the transonic nonlinear differential equation, special attention is directed to the development of similarity laws. These laws connect
nondimensional parameters such as pressure coefficient, lift and drag coefficients on
families of affinely related bodies of different thickness or slenderness ratio and angle
of attack. Brief discussion of hodograph transformation where switching between
the dependent and independent variables causes the governing equation to be linear is presented. Finally, semiempirical models on lift curve slope in the transonic
regime and the approximate location of detached shocks in front of blunt bodies are
developed and compared to experimental results. This chapter contains 12 examples
and concludes with 39 practice problems.
3.1 Introduction
On a phenomenological basis, transonic flow is often characterized by the appearance of shocks, boundary layer separations, loss of lift and rise in drag. On the
mathematical basis, transonic flow is characterized by a dominant nonlinearity in its
governing equations. Even in the case of small disturbances that are caused by planar and slender bodies at small angle of attack, the transonic perturbation potential
equation is nonlinear. This, i.e., the nonlinearity in the governing differential equations, limits the options of solution techniques. The question then arises whether it is
possible to transform the transonic differential equation such that it holds the same
nondimensional solution on bodies that are similar in the range of transonic Mach
numbers? The transformed bodies are then called affinely related to each other. We
shall further discuss affinelyrelated bodies in this chapter. The same solution of the
transformed differential equation can only be guaranteed when the coefficients are
held constant, from one transonic flow to another transonic flow. It is therefore the
coefficients of transformed differential equations that reveal the nature of transonic
Springer Science+Business Media Dordrecht 2015
R. Vos and S. Farokhi, Introduction to Transonic Aerodynamics,
Fluid Mechanics and Its Applications 110, DOI 10.1007/9789401797474_3
81
82
similarity parameters, as we shall see. The utility of such concepts is in the power of
generalization inherent in the similar solutions in transonic regime. This allows us to
extrapolate the aerodynamic behavior of affinely related bodies at different transonic
Mach numbers and different thickness, or slenderness ratios and angles of attack. It is
interesting to note that another nonlinear aerodynamic regime in the context of small
disturbance theory, namely hypersonic flow, also lends itself to the development of
similarity parameters on affinelyrelated bodies. In this chapter we examine similarity solutions from subsonic to supersonic flows including the transonic regime. The
monumental work of Shapiro [9] is a key reference to this classical topic.
(3.1)
where the u,
v and w are perturbation velocity components in x, y and z directions
respectively and they are small compared to V . Obviously, this approximation
excludes the neighborhood of stagnation point where the change in velocity is large.
Therefore, excluding the neighborhood of stagnation point, we can now introduce a
perturbation potential function that describes the perturbation velocity components
in the flow according to:
u = x
v = y
(3.2)
w = z
Therefore, the full velocity potential, and the perturbation velocity potential, are
related by the following set of equations:
x = V + x where x /V = O()
y = y where y /V = O()
z = z where z /V = O()
(3.3)
83
Now, for steady, inviscid, irrotational and isentropic flow, we have the full potential
equation, namely
y2
z2
x2
1 2 x x + 1 2 yy + 1 2 zz
a
a
a
2
x y x y
y z yz
x z x z
2
2
=0
2
2
a
a
a2
(2.199)
This is a second order, nonlinear partial differential equation that is best solved numerically. Also, note that the local speed of sound, a, is related to velocity components
following the energy equation in adiabatic flows according to:
a2
V2
a2
V2
+
= +
1
2
1
2
(3.4)
a 2 = a
1 2
2
V V
2
(3.5)
(3.6)
We then obtain:
2
a 2 = a
a =
2
2
a
1
2V u + u 2 + v 2 + w 2
2
(3.7)
u
1 2
u 2
v 2
w 2
( 1) 2
2
2
V V = a
V 2
+ 2 + 2 + 2
2
2
V
V
V
V
(3.8)
We note that the last two terms in the parenthesis on the righthandside of (3.8)
are secondorder terms, i.e., O(2 ), and may be neglected; therefore the local speed
of sound may be approximated, to first order, by
a2
2
1 ( 1)M
2
a
x
V
+ O(2 )
(3.9)
84
The full potential equation (2.199), in the limit of small perturbations, where we
neglect the squares and products of small parameters, i.e., the second order terms,
as O(2 ), and for freestream Mach numbers that exclude transonic and hypersonic
flows, may be written as
2
(1 M
)x x + yy + zz = 0
(3.10)
This is the classical linear differential equation that governs the dynamics of gas
in steady subsonic and supersonic flows around bodies that are thin (planar), or
slender (3D) and are at small angles with respect to the flow. The impact of small
perturbations in velocity is also felt at the pressure level. We introduce the pressure
coefficient, C p , which is a nondimensional coefficient for the pressure differential
at a particular location in the flow:
Cp
p p
2
=
2
q
M
T 1
p
2
1 =
1
2
p
M
T
(3.11)
V2
V2
= T +
2c p
2c p
(3.12)
x
V
+ O(2 )
(3.13)
Applying binomial expansion to (T /T ) 1 where the temperature ratio is represented by (3.13) and keeping only the linear term, i.e., O() term, we get
T
T
1
x
x
2
2
+ O(2 ) (3.14)
1 ( 1)M
1 M
V
V
x
V
(3.15)
The solid surface boundary condition demands that the local flow slope and the local
body slope to be matched at all points along the body surface. This is also known
as the flow tangency condition to the solid surface. An alternative description of
solid surface boundary condition calls for the normal component of the flow to the
solid surface to vanish at all points on the solid surface. For a 2D body in the xy
85
V+ u
V
y = f(x)
plane, the body may be mathematically represented by y = f (x) with a local body
slope of dy/dx or d f /dx. This is schematically shown in Fig. 3.1. A representative
flow velocity vector is also shown in Fig. 3.1. The corresponding flow slope may be
expressed in terms of a general angle, , which is also valid on the body, i.e.,
tan =
y
v
v
=
V + u
V
V
(3.16)
dx
V
(3.17)
2
1 M
(3.18)
A transformation of the function (x, y), is proposed that converts Eq. (3.18) in
),
the xy plane (known as the physical plane) to the Laplaces equation in (,
which governs the incompressible flow, in the transformed plane. Consider the
following transformation:
=x
= y
=
(3.19)
Then, by using the chain rule, we calculate the terms of Eq. (3.18) in the new variables,
for example
86
1
=
=
1
2
1
=
=
x x =
=
2
x
x
x
=
= =
y =
y
y
2
=
=
yy =
=
y2
y
y
x =
(3.20)
(3.21)
(3.22)
(3.23)
By substituting Eqs. (3.21) and (3.23) in (3.18), we get the transformed governing
equation in (, ) plane, i.e.,
2
1
+ = 0
+ = 0
(3.24)
d
V
(3.25)
But since = y following (3.22), we note that the transformed and the physical
bodies have the same slope,
df
dg
=
(3.26)
d
dy
Thus we conclude that the proposed transformation as described by (3.19), which
is known as the PrandtlGlauert transformation, does not change the shape of the
body. This is an important result, as it is most useful. The pressure coefficient in the
transformed plane is
C p M =0 2
(3.27)
V
Note that we have added a subscript M = 0 to C p in order to signify the Machzero or incompressible condition that prevails in the transformed plane. Replacing
by x from (3.20) in (3.27), we relate the compressible and incompressible
pressure coefficients, namely
Cp M
=0
x
x
= Cp M
= 2
= 2
V
V
V
(3.28)
Or alternatively,
87
C p M =0
C p M =0
=
Cp M =
1 M
(3.29)
1
C p dx
c
(3.30)
We conclude that the lift coefficient in subsonic compressible flow also is magnified
by the same factor, (1/) from the incompressible lift coefficient that appeared in
C p , namely
(cl ) M =0
(cl ) M =0
(3.31)
=
(cl ) M =
2
1 M
Similarly, the lift curve slope, in the subsonic compressible plane is related to the
incompressible lift slope, according to
dcl
d
dcl
d
M =0
dcl
d
M =0
2
1 M
(3.32)
This equation also indicates that the lift curve slope is magnified in the subsonic
compressible flow as compared to the incompressible, by the factor 1/. To complete
the discussion of twodimensional lift in the compressible domain, we address the
issue of circulation, , around an airfoil. Since the local strength of the bound vortex
sheet, representing an airfoil, is equal to the tangential velocity jump across the sheet,
which is amplified by 1/ factor in the compressible plane, the vortex sheet strength
is increased by the same factor. Therefore the circulation, which is the integral of the
vortex sheet strength along the mean camber line, is amplified by the same factor.
Another view that leads to the same conclusion is based on static pressure jump and
its proportionality to vortex sheet strength. Therefore, we may conclude that
( ) M =
( ) M =0
( ) M =0
=
2
1 M
(3.33)
Finally, we note that the KuttaJoukowski theorem relating lift to circulation will
remain valid in the compressible subsonic flow as well, i.e.,
L M = V ( ) M
(3.34)
88
xx + yy = 0
2
C p 2
y V
x
V
dys
dx
C p,0 2
d s
d
ys
Fig. 3.2 An airfoil in the physical and the transformed planes following PrandtlGlauert transformation
Figure 3.2 shows the same body in two planes, one subsonic compressible and the
other incompressible, following the PrandtlGlauert transformation that preserves
the shape of the transformed body. Finally, the pitching moment coefficient on an
airfoil, say about the leading edge, is related to the integral of the moment of the
pressure coefficient according to:
cm, LE =
1
xC p dx
c2
(3.35)
=
cm, LE M =
1 M
(3.36)
The hallmark of all of these compressibility correction results that are derived from
the linear theory, is the singularity of the solutions at M = 1. The failure to predict
a finite pressure, lift, drag, pitching moment coefficients in the vicinity of sonic flow
is due to the linearized nature of differential equation, which does not represent the
physics of transonic flow, i.e., dominated by the nonlinearity. We will examine the
suitable transonic small perturbation equation after we review a few examples from
the linear theory.
Example 3.1 The minimum (or peak suction) pressure coefficient on an airfoil at 2
angle of attack, in incompressible flow, is C p,min = 0.95. Calculate the pressure
coefficient at the same point on the airfoil, when freestream Mach number is 0.5.
Solution:
Applying the PrandtlGlauert compressibility correction to this airfoil, we get:
89
(C p, min ) M =0
0.95
C p, min M =
=
1.097
2
1 M
1 (0.5)2
Example 3.2 The lift curve slope, Cl , of an airfoil in incompressible flow is 6 rad1 .
Calculate the lift curve slope for the same airfoil at Mach 0.6.
Solution:
The lift curve slope is also amplified by the 1/ factor according to PrandtlGlauert
compressibility correction formula, namely (3.29), therefore
dcl
d
dcl
d
=
M
M =0
2
1 M
6 [rad1 ]
=
7.5 [rad1 ]
2
1 (0.6)
90
C p M =0
Cp M =
1 M L2
(3.37)
The local Mach number on the airfoil in the compressible domain is noted by M L in
(3.37). The effect of this improvement is to create higher suction on airfoils where
PrandtlGlauert consistently underpredicted the experimental data. Laitone related
M L to the freestream Mach number via an isentropic relation and expressed the
revised PrandtlGlauert compressibility correction for subsonic flow as:
C p M =0
Cp M =
1 2
2
2
2
1 M + M 1 + 2 M /2 1 M C p M
(3.38)
=0
Cp
M =0
2
(C p ) M =0
2 +
M 2
1 M
2
1+
(3.39)
1M
1.8
1.6
1.4
Laitone
Cp
KarmanTsien
1.2
PrandtlGlauert
1.0
Experiment
0.8
0.2
0.4
0.6
0.8
91
Tsien corrections apply to the local pressure coefficient only and do not extend to the
force coefficients, as in the PrandtlGlauert correction. However, based on point wise
correction of local pressure coefficient in the subsonic compressible domain, we may
proceed to integrate the pressure jumps and their moment to calculate lift and moment
coefficients on the airfoil using Laitone or KarmanTsien compressibility corrections.
In subsonic compressible flow over an airfoil, we define a new parameter, called
the critical Mach number, Mcrit . The flight Mach number that corresponds to the
first appearance of a sonic flow on the airfoil is called the critical Mach number. The
first appearance of the sonic point occurs at peak suction, i.e., the point of minimum
pressure, C p, min , or the point of maximum velocity, Vmax on the airfoil, as expected.
Since the pressure coefficient at the sonic point on the airfoil is defined as
C pcrit =
2
2
Mcrit
p
p
1
(3.40)
And we may relate the static pressure ratio, p / p , to the Mach numbers, Mcrit and
Mach 1 (for sonic point) using an isentropic flow connection (2.147) between the
flight and the sonic point on the airfoil, i.e.,
p
=
p
1+
1 2
2 Mcrit
+1
2
(3.41)
We can express the critical pressure coefficient as a function of the critical Mach
number as:
5
4.5
Sonic Boundary
4
3.5
3
Supersonic
Flow
2.5
2
1.5
Subsonic
Flow
1
0.5
0
0
0.1
0.2 0.3
0.4
0.5
0.6
0.7 0.8
0.9
92
C pcrit
2 1+
=
Mcrit
1 2
2 Mcrit
+1
2
(3.42)
The graph of (3.42) for C pcrit , as is customary, is shown in Fig. 3.4. We note that
C pcrit is the sonic boundary with the region to the right of the curve denoting supersonic and the region to the left of C pcrit curve identifying the subsonic regime. We use
the knowledge of critical Mach number on an airfoil at certain angle of attack, in
low speed, to mark the onset of transonic flow phenomenon, or the onset of shock
appearance. We can graphically construct the critical Mach number for an airfoil
where the knowledge of peak suction C pmin exists in the incompressible limit.
Example 3.4 The airfoil of Example 3.1 had a C pmin = 0.95 at the incompressible
limit. First graph the C pmin for the same airfoil, in the subsonic compressible flow,
using the PrandtlGlauert compressibility correction and second find the intersection
of this curve and the critical pressure coefficient described by (3.42).
Solution:
We first generate
a spreadsheet of the PrandtlGlauert compressibility correction
applied to C p M=0 = 0.95 for a range of Mach numbers (below 1) and then
graph the two functions, representing C p , as shown in Fig. 3.5.
Example 3.5 Develop a family of compressibility correction curves using KarmanTsien approach for the range of incompressible C p s between 1, corresponding to
incompressible stagnation point and 3 corresponding to strong suction at angle of
attack. Also, graph critical pressure coefficient representing the sonic point on the
airfoil.
Incompressible
C p ,min =  0.95
5
4.5
C p,crit
4
3.5
Prandtl  Glauert
Compressibility
Correction
3
2.5
2
1.5
1
0.5
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
93
Solution:
We apply KarmanTsien compressibility correction to the incompressible pressure
coefficients, described in the problem, and graph them using a spreadsheet program.
The result is shown in Fig. 3.6. The first appearance of sonic flow on the airfoil is
marked by the critical pressure coefficient, C pcrit curve. The Mach numbers beyond
critical, which is also called supercritical Mach numbers, will no longer be reached
isentropically.
Some classical experimental data, shown in Fig. 3.7 reveal the lift coefficient, k L
as a function of angle of attack (here called angle of incidence) for two airfoils.
Since k L in these graphs represents lift coefficient based on double dynamic pressure
2 ), it represents 1 of how we conventionally define lift coefficient, c , i.e., based
( V
l
2
2 ). The U in these graphs represents V and a represents
on dynamic pressure ( V
1
a , therefore a freestream Mach umber range 0.25 < M < 1.70 is investigated.
To relate the lift curve slopes between different subsonic Mach numbers, we may use
Suction
4.5
C p,crit
4
3.5
3
2.5
2
1.5
1
0.5
0
0.5
Compression
Incompressible
Stagnation Point
(Cp )
1
1.5
2
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
94
0.6
0.6
kL
kL
0.4
0.2
0.4
U=0.5a1
U=0.25a1
U=0.7a1
U=0.7a1
U=0.5a1
U=0.25a1
0.2
U=0.8a1
U=1.7a1
U=1.7a1
0.05
0
5
10
Angle of Incidence (deg)
0.0
0
5
10
Angle of Incidence (deg)
5
0.2
0.2
Fig. 3.7 Experimental data on airfoil lift in subsonic compressible flow on two airfoils (Definition
of lift coefficient k L is based on double dynamic pressure, therefore it is 21 cl ), after [11]
dcl
d
dcl
d
M1 =
M2
1 M22
(3.43)
1 M12
The lift curve slopes as a function of Mach number follow a near perfect match
with the PrandtlGlauert prediction for the circular arc airfoil (i.e., between 0.25 and
0.50). Although, the lift curve slope increase to Mach 0.7 is less accurate, but still
it is in the right direction, i.e., slope increases with compressible subsonic Mach
number. The accuracy is however reduced in the RAF31a airfoil. In both cases,
the main departure in the subsonic flow occurs when we cross the critical Mach
number boundary and enter the transonic regime on the airfoil. As evidenced by
data, RAF31a has entered the transonic zone at freestream Mach number of 0.5
and the circular arc airfoil enjoys a higher critical Mach number, in excess of 0.7.
The example of subsonic compressible flow over a wavy wall is both instructive
in its own right, as well as serving as a building block for wind tunnel testing in
transonic flow. Here, we use an example to study the flow over a wavy wall.
Example 3.6 Consider a subsonic compressible flow over a wavy wall of infinite
extent, in the xdirection, with amplitude, h and the wavelength, . The wall may
be described by a cosine function according to:
yw = h cos
2x
(3.44)
95
Calculate the perturbation potential function, (x, y), as well as pressure distribution
on the wall. Also, determine the similarity parameter based on h/.
Solution:
In the limit of small h/, we may use
2 x x + yy = 0
(3.45)
As the governing equation for , subject to the flow tangency boundary condition on
the wall and total decay of the perturbations at infinity. The separation of variables
suggests that
(x, y) = F(x)G(y)
(3.46)
Therefore,
x x = F (x)G(y)
yy = F(x)G (y)
(3.47a)
(3.47b)
(3.48)
We set the pure function of x equal to a constant and the pure function of y equal to
the negative of that constant to satisfy Eq. (2.50), namely
F
= k 2
F
1 G
= k2
2 G
(3.49a)
(3.49b)
(3.50)
The periodic function in x matches the periodic wall and boundary condition in x,
i.e., the wavy wall. The solution to (3.49b) is
G(y) = Ceky + De+ky
(3.51)
The coefficients A, B, C and D are as yet arbitrary and need to be determined from
our boundary conditions. The exponentially decaying part of the solution in lateral
direction matches the boundary condition at infinity for subsonic flows and is thus
physical. However, the exponentially growing term is unphysical and we thus need
to set D = 0. The wall boundary condition demands
96
y
h
dyw
= 2 sin(2x/)
V
dx
(3.52)
(3.53)
h
k
[A cos(kx) + B sin(kx) = 2 sin(2x/)
V
We may satisfy the above equation, by inspection, which sets A = 0, k = 2/ and
k
h
B = 2
V
Which gives the only unknown coefficient, B, as
B=
hV
hV
1
2
M
2x
sin
2
2 y
1M
(3.54)
2x
4 (h/)
cos
2
1 M
(3.55)
Comparing the pressure coefficient at the wall to the cosine function that described
the wall, we conclude that the pressure is 180 out of phase (or Cp is inphase)
with the wall and is symmetric with respect to its peaks and valleys; therefore, the
net axial force, i.e., the (inviscid) drag, on the wall is zero. This is DAlembert
paradox. We also note that the impact of Mach number on the pressure coefficient is
inversely proportional to , which was predicted by PrandtlGlauert compressibility
correction. We finally note the attenuation of pressure disturbance away from the wall
(i.e., the exponential decay function) depends on M through the term, which
approaches zero as M approaches 1 (see Eq. (3.54)). This is another hallmark of
transonic flow, namely the disturbances that are caused by the body propagate away
from the body without significant decay or attenuation.
97
yw = H
M
H
y
yw = h cos 2 x
x
h
The general expression for pressure coefficient, invoking the wave number
k = 2/, is
2hk
Cp
cos(kx)ek y
(3.56)
Therefore, the similarity parameter, f , for the wavy wall involving the pressure
coefficient, freestream Mach number and the nondimensional wall parameter is:
2
C p 1 M
= f (kx, k y)
hk
(3.57)
98
We again approach the problem through separation of variables, as before, and note
that the presence of the upper wall imposes a new boundary condition and thus
remove the necessity to set the coefficient of the exponentially growing term equal
to zero. The new solution, again in terms of the wave number k, is:
(x, y) =
h
V
sin(kx)eky 1 + e2k(yH )
2k
H
1e
(3.58)
The pressure coefficient follows the linear theory and at the wall, we approximate
the wall location by y = 0, or the xaxis, to get:
C pw
2hk 1 + e2k H
cos(kx)
1 e2k H
(3.59)
Comparing (3.59) that corresponds to the windtunnel environment to the wavy wall
problem in unbounded space, i.e.,
C pw
2hk
cos(kx)
(3.60)
And the dimensionless length scale is k H . The wind tunnel interference is defined
as the change in C pw (or local speed on the model) caused by the presence of the
windtunnel wall, expressed as a percentage, namely
C pw W T C pw f light
C pw
1 + e2k H
2e2k H
=
=
1
=
1 e2k H
1 e2k H
C pw f light
C pw f light
=
2
e+2k H 1
(3.61)
99
2 1
M
(3.62)
(3.63)
where F is any function of its argument, i.e., (x y). Also, any function of (x + y)
is a solution to the wave equation, namely
(x, y) = G(x + y)
(3.64)
Mach waves
M
M
1
dy/dx = +1/
M2 1
Mach waves
dy/dx = 1/
100
M
x
Fig. 3.10 Wave patterns generated by a thin biconvex airfoil in supersonic flow as viewed by
linear theory
x
F
= 2
V
V
(3.65)
Here, F is the derivative of F with respect to its argument (x y). Now, we relate
F to y via
(3.66)
y = F
Therefore, the pressure coefficient is
F
2
=
C p = 2
V
y
V
(3.67)
The solid surface boundary condition demanded that the flow and the body (locally)
possess the same slope (see 3.17), namely
y
df
dx
V
(3.68)
In which the body is represented with the function f. The body surface slope is the
tangent of the body surface angle, which for thin bodies may be approximated by
the local body angle, , in radians, i.e.,
2
(3.69)
Note that a similar result would be produced had we chosen the function G(x +y)
instead of F(x y). The plus and minus signs in (3.69) reflect the possibilities of
compression or expansion on a surface with the local angle in radians, in supersonic
flow, respectively. The determination of whether a surface is undergoing compression
or expansion by the flow is obvious as we note whether the flow is locally turning
into itself or is it locally turning out of itself respectively. Here we note that at a
Cp
101
point on the body, as the (supersonic) Mach number increases the pressure coefficient
is reduced, by the 1/ rule in the linear theory. We also note the singularity in C p as
freestream speed approaches sonic.
Now, let us apply the theory of linearized supersonic flow to a few examples.
Example 3.8 A flat plate is at an angle of attack, , in a supersonic flow, as shown.
Apply the linear theory to this problem to derive an expression for the wave drag and
lift coefficient.
y
M >1
Flat Plate
Solution:
From the linear theory, the upper surface experiences a suction pressure and the lower
surface experiences a compression, both proportional to local wall angle, i.e., ,
C pu
2
2
and C pl +
(3.70)
Therefore, the flat plate experiences a normal force with the coefficient cn according to
4
(3.71)
Cn
Then, we may resolve the normal force in the lift and wave drag components through
cosine and sine of the plate angle of attack respectively and in the limit of small angles,
the cosine is replaced by 1 and the sine is replaced by (in radians) to get the final
result, i.e.,
4
( in radians)
(3.72)
cl
cd, w
42
( in radians)
(3.73)
The flat plate, according to linear theory, creates a lifttodrag ratio in supersonic flow
that is inversely proportional to angle of attack, and is singular at = 0, namely
L
1
Dw
(3.74)
102
Example 3.9 Apply the supersonic linear theory to the wavy wall problem (of
Example 3.6). Examine the behavior of the solution (x, y) and the wall pressure
coefficient, C pw (x). Also, calculate the wave drag coefficient for one wavelength
section of the wall.
Solution:
The equation for the wall is
yw = h cos(kx) where k
2
(3.75)
dx
V
(3.76)
At y = 0, as before, to get
hk sin(kx) =
F (x)
V
(3.77)
hV
cos(kx)
(3.78)
hV
cos[k(x y)]
(3.79)
x (y = 0)
hk
= 2 sin(kx)
V
(3.80)
Note that the pressure coefficient on the wall varies as the sine function whereas the
wall is described by the cosine function. Therefore there is a 90 phase shift between
the wall and its pressure distribution, as shown below.
103
( /2hk) C p
y w /h
1
0.5
00
kx
1
0.5
1
1
=
C pw
0
dyw
dx
hk
1 2
2 sin(kx) [hk sin(kx)] d(kx)
dx =
2 0
(3.81)
Or
Cd,w =
(hk)2
(3.82)
Note that hk in the numerator of (3.82) is proportional to h/, which measures the
wavy wall thickness ratio. Therefore, wall wave drag coefficient scales as the square
of the wall thickness parameter, hk.
The wave drag is the sum of the wave drag of a symmetrical profile at zero angle of
attack and the wave drag of a flat plate at an angle of attack.
cd, w = cd, th + cd, AoA = cd, th +
42
(3.83)
In fact, we may even add the friction drag coefficient, c f , to (3.83) to produce the
overall drag coefficient of a thin symmetrical profile in supersonic flow at angle of
attack, namely,
104
cd = c f + cd, th +
42
(3.84)
Example 3.10 Use the principle of superposition in the linear theory to develop
the dragpolar of a 10 % thick, symmetrical doublediamond airfoil, as shown, at
M = 2. Vary the angle of attack between 2 and 12 and assume the friction
drag coefficient for the profile is 0.005. Also by drawing a tangent from the origin of
c L c D coordinate system to the drag polar curve, identify the angle of attack that
corresponds to the maximum L /D . Graph L /D as a function of angle of attack
and verify this angle.
nose
2
4
t/c =10%
M = 2
Solution:
The lift coefficient is calculated based on a flat plate, i.e., (3.72). The drag coefficient
is the sum of three drag coefficients, as in (3.84). The thickness contribution to drag
is calculated from the pressure coefficients on the four panels of the symmetrical
airfoil, namely
cd, th
t
t
1
= C p1 C p2
C p1 + C p3 C p2 C p4
=
2
c
c
(3.85)
Here we used the symmetry of the profile and lumped the contribution of the lower
half of the airfoil with its upper half. The pressure coefficients C p1 and C p2 are
related to the angles in radians that those panels make with respect to xaxis, i.e.,
21 vertex angle, respectively. The halfvertex angle, nose , is
nose = tan1 (0.1) 0.09967 (rad) or 5.7135 (deg)
Therefore,
2(0.09976)
0.19934
21
2(0.09976)
0.19934
C p2 =
21
t
0.039868
cd, th = C p1 C p2
c
A simple spreadsheet calculation produces the following table (Table 3.1). and
the associated graphs.
C p1 =
105
Table 3.1 Lift and drag characteristics of a diamond airfoil in supersonic flow [linear theory]
(deg)
(rad)
cd
cl
L/D
2
0
2
4
6
8
10
12
0.034917
0.000000
0.034917
0.069813
0.104726
0.139636
0.174533
0.209440
0.1396
0.0000
0.1396
0.2793
0.4189
0.5585
0.6981
0.8378
0.04974
0.04487
0.04974
0.06436
0.08873
0.12285
0.16671
0.22033
2.8070
0.0000
2.8070
4.3387
4.7207
4.5463
4.1876
3.8023
The tangent to the drag polar from the origin finds the maximum L/D, on the drag
polar curve, which from our graph is 6 .
LiftDrag Polar
10% Thick, DoubleWedge Airfoil
1
0.8
Lift Coefficient, cl
(L/D) max
0.6
0.4
0.2
0
0.00
0.05
0.10
0.20
0.15
0.25
0.2
Drag Coefficient, cd
The liftdrag ratio is also plotted versus the angle of attack in the following figure,
which confirms the maximum L/D for a 10 % thick symmetrical double wedge
occurs at 6 .
LifttoDrag Ratio
10% Thick, DoubleWedge Airfoil
L/D
2
1
4
2
0
0
2
4
6
10
11
12
106
(3.86)
Q
x2
y2
z2
Q
4r
(3.88)
where Q is the source strength, i.e., the volume flow rate from the 3D (incompressible) source. The flowfield associated with a source at the origin is thus purely radial
with
Q
=
(3.89)
Vr =
r
4r 2
The denominator of (3.89) is the surface area of a sphere of radius r and the product
of the area and the normal velocity component (Vr ) then constitutes the volume flow
rate, from the source as expected
Figure 3.11 shows a slender body of revolution with a pointednose in uniform
flow. We define slenderness and fineness ratios for the body shown in Fig. 3.11.
y
x
V
z
107
These are:
d
slenderness ratio
F=
fineness ratio
d
=
(3.90)
(3.91)
(3.92)
We may divide the governing equation by 2 (and absorb the in the y and zderivatives) to get:
(3.93)
x x + ( y)( y) + ( z)( z) = 0
A solution to this equation, analogous to the incompressible solution, is known as
the subsonic source solution, namely,
(x, y, z) =
Q
x2
+ 2 y2
+ 2 z2
4 x 2 + 2 r 2
(3.94)
3 D Source
at origin
Circle,
y 2 +z 2 = r 2
r
x
M
z
(3.96)
108
Q
x2
2 y 2
2 z 2
Q
= (x, r )
4 x 2 2 r 2
(3.98)
where r = y 2 + x 2
The constant potential surfaces for supersonic source are described by:
x 2 2 r 2 = const.
(3.99)
f ()d
M >1
(3.100)
4 (x )2 2 r 2
Mach Cone
dr/dx =1/
r
Zone of
x action
z
A hyperboloid of revolution
describes a constant potential
surface , =const. for a
supersonic source
Fig. 3.13 A Mach cone downstream of a source in a supersonic flow and a constant potential
surface, (x, r )
109
Source/sink Distribution
Body of Revolution
(x, r)
r
x
d
x
Note that a minus sign is now absorbed in the general description of the source/sink
distribution, f (). The parameters in (3.100) are shown in Fig. 3.14.
The overall contribution of the source/sink distribution along the xaxis is obtained
by integrating (3.100) from the leading edge (i.e., x = 0) to the vertex of the Mach
cone on the xaxis that intercepts the point (x, r ), as shown in Fig. 3.15.
Hence the potential at (x, r ) is
(x, r ) =
xr
0
f ()d
(3.101)
4 (x )2 2 r 2
The streamwise perturbation of the flow due to the source distribution along the
xaxis is u(x,
r ) = x (x, r ), or
u(x,
r) =
xr
0
f ()d
(3.102)
4 (x )2 2 r 2
xr f ()
=
0
r
4
x
r
(3.103)
(x )2 2 r 2
Mach Cone
Intercepting
(x, r)
(x, r)
r
x
(x r)
Fig. 3.15 The segment (x r ) contributes to the potential at (x, r ) in a supersonic flow
110
u
x
= 2
V
V
(3.104)
f ()d
1 xr
2V 0
(x )2 2 r 2
(3.105)
C p 2
Therefore, the pressure coefficient is
Cp =
The flow tangency condition on the body (with body radius R(x)) demands the body
slope and flow inclination to match, namely,
dR
f ()d
v
v
1 xR x
=
=
(3.106)
dx
V + u
V
4V 0
R
(x )2 2 r 2
In the limit of slender body, R is much smaller than x , and also x, hence the
integrand and the limit of the integral simplify to:
x
dR
1
f (x)
f ()d =
dx
4 RV 0
4 RV
(3.107)
Hence the strength of the source/sink distribution f (x) is related to the slope of
the body, d R/dx, the body radius, R and the freestream velocity, V . We may cast
(3.107) in terms of the body cross sectional area variation, dS/dx, as
f (x) 4V R
dS
dR
= 2V
dx
dx
(3.108)
Note that the negative sign that we absorbed in the definition of f (x) is now recovered
in (3.107) or Eq. (3.108). Based on the slender body approximation, we may express
the perturbation potential, the streamwise and the lateral velocity perturbations as
well as the pressure coefficient in the flowfield of a slender body in supersonic flow as:
d
V xr dS
2 0
d (x )2 2 r 2
V xr d2 S
d
u(x,
r) =
2
2 0
d
(x )2 2 r 2
2
d
V xr d S x
v (x, r ) =
2
0
2
d
r
(x )2 2 r 2
d
1 xr d2 S
C p (x, r ) =
0
d 2 (x )2 2 r 2
(x, r ) =
(3.109)
(3.110)
(3.111)
(3.112)
Now, let us apply these results to a slender cone in supersonic flow. The cone of
semivertex angle , has a body radius R(x) according to:
111
R(x) = x tan
= x
R(x)
(3.113)
Therefore the cross sectional area variation S(x) along the axis of the cone is calculated from:
S(x) x 2 2
(3.114)
This is a quadratic in x; hence the second derivative of cross sectional area of the
cone is constant, namely
d2 S
(3.115)
= 2 2
d 2
Now, substitute the second derivative of area in the pressure coefficient integral
(3.112) to get:
C p, cone 2 2
xR
= 2 2 ln
(x )2 2 R 2
x+
x 2 2 R 2
(3.116)
C p, cone
R
2 ln
2x
2
= 2 ln
2
2 1
M
C p, cone
2
(3.117)
2 ln
2
Here we conclude that slender cones in supersonic flow obey the following similarity
rule:
C p, cone
= f ()
(3.118)
2
This is essentially the Gthert rule for axisymmetric bodies, where is replaced by
the characteristic thickness ratio of the body, . The similarity parameter in (3.118) is
. Therefore, slender cones with different semivertex angles, 1 and 2 in different
supersonic flows, with M1 and M2 , will have their cone surface pressures related
by
2
C p, cone1
= 12
(3.119)
C p, cone2
2
If the similarity parameter is held constant in the two flows, namely
2 1=
2
1 M1
2 M2 1
(3.120)
112
We may also apply the linear supersonic slender body theory to a parabolic body of
revolution, as a second example. The geometry of the parabolic body of revolution,
in the meridian plane, is shown in Fig. 3.16.
The equation for body radius, R, as a function of body axis coordinate, x, is
R(x) =
2 2
(x 1/4)
F
(3.121a)
dR
4x
=
dx
F
(3.121b)
d2 R
4
= = const.
dx 2
F
(3.121c)
Here F is defined as the ratio of body length to maximum diameter, or the fineness
ratio, i.e., the inverse of slenderness ratio. Note that the body length is normalized to
one, which serves as a reference length scale. The cross sectional area of the body is
that of a circle or S = R 2 , which differentiates into
dR
dS
= 2 R
dx
dx
2
dR 2
d S
d2 R
= 2
+R 2
dx 2
dx
dx
(3.122)
(3.123)
Now, if we substitute the body area distribution in (3.112) for the pressure coefficient and integrate the equation we get a closed form solution for the C p distribution.
Note that the new limits of the integral should begin at x = 0.5 instead of zero that
we had in our derivation, which requires x to be replaced by x + 0.5 to represent a
shift to the left on the xaxis, namely
C p (x, r ) =
d
1 x+0.5r d2 S
2
0.5
d
(x + 0.5 )2 2 r 2
(3.124)
Again, upon substitution for d2 S/d 2 in (3.124) from (3.123), we can evaluate the
integral to get the surface pressure coefficient on a parabolic body of revolution in
supersonic flow, namely
r
Parabola
x
0.5
+0.5
=1
C ps =
113
4
2
2 2
1 x + 0.5
12x
cosh
1
+
6
R
F2
R
4
+ 2 6 [(x + 0.5) 4x] (x + 0.5)2 2 R 2
F
(3.125)
A graph of (3.125) for a parabolic body of revolution for a fineness ratio of 10 and
freestream Mach number of M = 1.4 is shown in Fig. 3.17. Also the pressure
distribution on a 2D parabolic body is also graphed in Fig. 3.17 for comparative
purposes.
The twodimensional supersonic linear theory predicts:
Cp =
d R/dx
2
2
(3.126)
(3.127)
0.4
Tw
o
me
ns
0.2
ion
al,
Cp 0
0.2
0.4
Three Dimensional
M = 1.4
Di
=1
.4
Three Dimensional
M = 0
0.5
0.5
Fig. 3.17 Pressure distribution on two and threedimensional bodies with parabolic contours
114
Also the bodies sharp nose and the trailing edge in incompressible flow provide for
singularity at the sharp corners, as indicated in the graph of Fig. 3.17 for M = 0.
x
( + 1)
x x
V
(3.130)
Note that the term (in 3.129) can be made arbitrarily small when the freestream Mach
number is made arbitrarily close to 1. Consequently, the small perturbation equation
for steady, threedimensional irrotational flows assumes the following form in the
transonic limit,
x
2
2
( + 1)
x x
)x x + yy + zz = M
(3.131)
(1 M
V
The nonlinear term on the righthand side of (3.131) speaks to the nature of transonic
flow. Combining the lefthand side term with the right (in x x ), we get the transonic
small disturbance equation:
2
2 x
(1 M ) ( + 1)M
x x + yy + zz = 0
V
(3.132)
Figure 3.18 shows the pressure coefficient approaching the sonic flow from the
subsonic side and from the supersonic side, where linear theory was deemed valid.
115
 Cp, 0
Fig. 3.18 Variation of pressure coefficient with Mach number according to linear theory
yy + zz = 0 as M 1
(3.133)
Equation (3.133) in the limit of sonic flow shows a lack of dependence on the streamwise (i.e., x) direction. This indicates that the disturbance caused by a body in sonic
flow propagates laterally and stays confined in the streamwise direction. This picture of near sonic flow helps us understand an important character of transonic flow.
Figure 3.19 shows a cartoon of a body in sonic flow where the disturbance caused
by the body laterally extends to large distances whereas the axial extent is limited to
the body length.
The transonic small disturbance equation (3.132) does not lend itself to analytical
solution; therefore, we seek to establish similarity laws that would hold on bodies that
are affinelyrelated, but at different transonic Mach numbers and different thickness
or slenderness ratios (and different angles of attack). First, let us define affinelyrelated bodies. Figure 3.20a shows a body of maximum thicknesstochord ratio, 1
Stream
surfaces
y
M =1
Fig. 3.19 A slender body in sonic flow showing extensive lateral and limited axial extent of disturbance
116
(a)
y
(b)
1 = t1 /
ys(x)
t1
ys(x)
t2
2 = t2 /
Fig. 3.20 Affinelyrelated profiles with different thickness/slenderness ratios. a Definition sketch
for a body with thickness ration 1 . b An affinelyrelated body to (a) but with thickness ratio 2
and in Fig. 3.20b we note that the same body is stretched in the ydirection and it
has attained a thicknesstochord ratio of 2 .
We define the local body slope as the product of a dimensionless function h(x/)
and the body thickness (or slenderness) ratio, , as
dys
= h(x/)
dx
(3.134)
(3.135)
In addition, the angle of attack, , in affine transformations is adjusted by the thickness (or slenderness) ratio according to:
1
2
=
1
2
(3.136)
Although the example profile shown in Fig. 3.20 is symmetrical, i.e., it has zero
camber, affine transformation of cambered profiles will change camber in proportion
to thickness ratio as well. If we define a profiles camber in the xy plane as yc /
(recall that camber in twodimensional profiles is customarily defined as z c /c, in x
z plane, where c is the chord length), then affine transformation adjusts or changes
camber according to (using our notation in xy plane):
(yc /)1
(yc /)2
=
1
2
(3.137)
117
(3.138)
y=
y 1/3
(3.139)
z=
z 1/3
(3.140)
x=
V 2/3
(3.141)
Note that the dimensionless lateral coordinates, y and z are scaled down by the factor,
1/3 , as compared to the streamwise coordinate, x, with the intention of placing them
all on the same order of magnitude. The choice of dimensionless perturbation potential yields the same governing equation and boundary conditions for affinelyrelated
bodies, as we shall see. For a rigorous proof of these choices of transformed coordinates, the reader may consult Ref. [9]. The transonic small disturbance equation is
now expressed in terms of the transformed coordinates and perturbation potential as:
2 )
(1 M
2
(
+
1)M
x x x + y y + zz = 0
2/3
(3.142)
(3.144)
118
K ( + 1)x x x + y y + zz = 0
(3.145)
The boundary condition on the solid body is the flow tangency condition, which
matches the body to flow slope, namely
v
dys
dx
V
(3.146)
(3.147)
Now the righthand side of (3.146) may be expressed in terms of the transformed
coordinates as
( )( y )( y/ y)
V 2/3 ( y )( 1/3 /)
y
v
=
=
=
= ( y )
V
V
V
V
(3.148)
y = h(x)
(3.149)
x
V
(3.150)
119
( )(x )(x/x)
x
V 2/3 (x )(1/)
= 2
= 2
= 2 2/3 x
V
V
V
(3.151)
This produces the transonic similarity rule for the pressure coefficient, C p , namely
Cp
2x = f (K , x, y, z)
2/3
Cp
= f (K , x, y, z)
2/3
(3.152)
C p2
2/3
(3.153)
On two bodies that are affinelyrelated in transonic flow, i.e., at the same x, y and z
locations, when the two transonic flows have K 1 = K 2 and the angle of attacks on
the two bodies follow (3.136), namely
2
1
=
1
2
Now, let us put the transonic similarity rule, (3.153), to use in the following example.
Example 3.11 Consider an airfoil with 10 % thicknesstochord ratio in M = 0.9
(transonic) flow at 4 angle of attack. The peak suction pressure coefficient is
C p, min = 0.55. What does transonic similarity rule say about an affinelysimilar
airfoil in Mach 0.95? Also, calculate C p, crit corresponding to M = 0.9 and
M = 0.95.
Solution:
First, let us calculate the transonic similarity parameter, K , namely
K1
2 )
(1 M1
2/3
0.88190
Now, let us consider an affinelyrelated airfoil with thickness 2 in Mach 0.95 flow
at an angle of attack, 2 . In order to have the same transonic similarity parameter,
K 2 = K 1 , the freestream Mach number has to be:
K2
2 )
(1 M2
2/3
0.88190
120
2 = 1
2
1
=4
0.03676
0.10
1.47
C p2 = C p1
2
1
2/3
0.03676
= 0.55
0.10
2/3
0.2822
Therefore the pressure coefficient at the peak suction point in Mach 0.95 flow is
0.2822. The following graph shows the schematic drawing of two transonically
similar flows (not to scale).
y
K=0.88190
= 0.10
o
= 4
M=0.90
K=0.88190
0.03676
1.47 o Cp, min 0.2822
M= 0.95
2 )
(1 M
2
( + 1)M x x x + y y + zz = 0
2/3
There are different forms of transonic similarity parameter that are possible, which
are all identical at M = 1, but differ at Mach numbers slightly below and above
2 in the
the sonic point. For example, as we noted earlier, the product of ( + 1)M
121
bracket in (3.142) may be replaced with ( + 1), in the limit of sonic flow and that
led to the definition of the transonic similarity parameter K in (3.143). Here, we first
introduce the transonic similarity parameter that Von Krmn proposed, namely
K
2
2
1 M
+1 2
2 M
2/3
(3.154)
2
2 ) in the numerator of the Von Krmn parameter and write
What if we take (1 M
it as the product of (1 M ) and (1 + M ) and then replace (1 + M ) by 2?
This is certainly a valid approximation in transonic flow; therefore another similarity
parameter emerges, namely
1 M
(3.156)
K
2/3
+1
It may be shown that the pressure coefficient then takes on the form
C p [( + 1)/2]1/3
= f 1 (K , x, y, z)
2/3
(3.157)
Although the derivation of the expression (3.157) for pressure coefficient is beyond
the scope of this book, the reader may consult Ref. [9] for detailed derivation. We also
note that (3.157) is more general than the basic expression (3.152), as it accounts for
the effect of different gases through ratio of specific heats, . The two expressions for
similarity pressure coefficients share the same functional dependency on slenderness
ratio, , namely 2/3 .
We are now ready to examine the similarity principles in force and moment coefficients on two dimensional thin/slender bodies in transonic flow. The lift coefficient
is written as
1
C p dx = C p dx
(3.158)
cl =
We divide both sides of (3.158) by 2/3 to get
cl
2/3
C
p
dx = f 1 (K )
2/3
(3.159)
122
2/3
cl [( + 1)/2]1/3
= f 2 (K )
2/3
(3.160)
The drag coefficient may be written as
1
Cp
cd =
dys
dx
dx =
C p h(x)dx
(3.161)
cd [( + 1)/2]1/3
= f 3 (K )
5/3
(3.162)
From the transonic similarity expression for 2D wave drag coefficient, we note
that drag coefficient on twodimensional thin bodies in transonic flow is proportional
to 5/3 . The 2D wave drag then rises as 5/3 , which indicates the advantage of thin
profiles in transonic flow. The 2D drag/lift ratio is proportional to as well, which
again points to the advantage of thin bodies in transonic flight.
The pitching moment coefficient about the leading edge is
cm, LE =
1
C
xdx
=
C p xdx
p
2
(3.163)
xdx = f 4 (K )
2/3
2/3
cm, LE [( + 1)/2]1/3
= f 4 (K )
2/3
(3.164)
Example 3.12 The lift coefficient of a 10 %thick airfoil at the angle of attack of 3
in Mach M = 1. air stream ( = 1.4) is cl = 0.12. What does transonic similarity
theory predict as the lift coefficient of an affinely similar airfoil (also called the same
airfoil family in airfoil theory) that is 15 % thick and is in transonic flight through
air (with = 1.4)?
123
Solution:
First, we establish the transonic similarity parameter, K, according to (3.156), namely
K =
1 M1
2/3
1
1 1.175
0.8123
0.12/3
M2 = 1 K 2
= 1 + 0.81230.152/3 1.229
2
1
2/3
Now, we need to examine the angle of attack on the second airfoil, as it is adjusted
by affine transformation, i.e.,
1
2
=
2
1
2 = 1
2
1
= 3 (1.5) = 4.5
0.1
0.15
M
1.175
1.229
3
4.5
cl
0.12
0.1572
2
1 M
( + 1) 2
(3.165)
124
(3.166)
1
1
1
dy
dy
C p (2 y) dx 2 2
C p y dx 2 2
C p x[ h(x)]dx
A 0
dx
0
dx
0
(3.167)
After cancelling the 2 from numerator and denominator of (3.167) and taking the
2 in the integrand to change x to x, we get
CD
1
0
C p h(x)xdx
1 C
CD
p
h(x)xdx = f (K )
0 2
2
CD
= f 6 (K )
2
(3.168)
(3.169)
(3.170)
Therefore, the wave drag of a projectile rises as the fourth power of its diameter in
transonic regime, which strongly supports the notion (or necessity) of slender bodies
in transonic flight.
For a planar body, e.g., a wing, the transonic similarity parameter is:
1 M
K =
2/3
+1
(3.171)
The pressure coefficient is the same as the 2D flows, but affine transformations
change the wing aspect ratio, A, in addition to the angle of attack and thus the
following three parameters are kept constant on two wings that are affinelyrelated,
namely:
1 M2
1 M1
=
(3.172a)
2/3
[(1 + 1)1 ]
[(2 + 1)2 ]2/3
1 /1 = 2 /2
(3.172b)
A1 [(1 + 1)1 ]1/3 = A2 [(2 + 1)2 ]1/3
(3.172c)
Thus the pressure coefficient for a wing of thickness ratio, , aspect ratio, A, span
of b, ratio of specific heats, and angle of attack, is:
125
4
Predicted theoretical
critical Mach Number
5/3
(t/c)
(CD p)min
AR(t/c)1/3= 3.0
2.5
2.0
1.5
1.0
.5
.25
0
2.0
1.6
1.2
.8
.4
M2 1
.4
.8
1.2
2/3
(t/c)
Fig. 3.21 Curves of the generalized drag coefficient for symmetrical wings for = 1.4 (after
Ref. [6])
C p / ; A(+1)1/3 1/3 =
2/3
f 7 (K , x, y, z, z/b)
( + 1)1/3
(3.173)
2/3
f 8 (K )
( + 1)1/3
(3.174)
2/3
f 10 (K )
( + 1)1/3
(3.175)
5/3
f 9 (K )
( + 1)1/3
(3.176)
The transonic zerolift wave drag coefficient for unswept wings with a symmetrical
NACA 63 profile is presented in Fig. 3.21, as presented by McDevitt [6]. We recognize
2 )/ 2/3 in the graph as well as the
the similarity parameters A(t/c)1/3 and (1 M
5/3
drag coefficient divided by . For a given thickness to chord ratio, aspect ratio, and
freestream Mach number, this graph allows us to calculate the wave drag coefficient.
A similar graph can be found in the USAF Datcom [3] for general airfoils with
rounded leading edges. We will show in Chap. 8 how this method can be extended
to calculate the wave drag of swept wings at transonic speeds.
126
2
v
( + 1)M
u
u
+
=
u
x
y
V
x
(3.177)
=0
y
x
(3.178)
y)
v = v (x, y)
(3.179)
du =
(3.180)
(3.181)
(3.182)
127
Alternatively, we may express the two independent as dependent variables according to:
x = x(u,
v )
y = y(u,
v )
(3.183)
(3.184)
A comparison between the two sets of Eqs. (3.182) and (3.184) shows that
v y
u y
and xv =
v x
vx
and yv =
yu =
xu =
(3.185)
Now, if we substitute these in (3.177) and (3.178), we get the new governing equation
set as:
2
x
( + 1)M
y
2 y
(1 M
+
=
(3.186)
)
u
v
u
V
v
y
x
=0
v
u
(3.187)
The new set of governing equations, known as transonic hodograph equations are
now linear in x and y, as independent variables where the new dependent variables
are u and v . This clever mathematical exercise has two impediments in real life
applications. First, the physical boundary conditions are always expressed in terms
of x and y as independent variables and not the u and v . Second, the solution can only
exist if the determinant of the coefficient matrix
= 0. The determinant may only
be zero in supersonic flows and the loci of all the points with = 0 is called the limit
line. Except for simple boundaries and geometries (as in a wedge), the application
of hodograph transformation to transonic problems is not practical. In addition, with
the advances in Computational Fluid Dynamics (CFD), the nonlinearity of transonic
flow poses no problem to achieving highfidelity simulation or solution. For more
details on hodograph transformation such classical texts as Shapiro [9] and Liepmann
and Roshko [5] may be consulted.
128
Subsonic 2D
Theoretical
CL =
Supersonic 2D
Theoretical
4
CL =
M 2 1
1M 2
10
Typical Unswept
High Aspect Ratio Wings
9
8
Thin Airfoil
Thick Airfoil
Typical Swept Wing
7
6
5
4
3
2
1
0
0
.5
1.0
1.5
2.0
2.5
3.0
129
4.
B2707300 threeviews:
3.
Experiment
2.
Computation
Linear Potential
1.
0.
0.
0.5
1.0
1.5
2.0
2.5
aerodynamic center,
xac/cr (%)
Aerodynamic Center
65
60
Computation
Linear Potential
55
Experiment
50
0.
0.5
1.0
1.5
2.0
2.5
(a)
(b)
Detached shock
M>1
Sonic line
Detached shock
M<1
M<1
M > 1
M>1
Sonic line
M > 1
Fig. 3.24 Two examples of a blunt body in supersonic and upper transonic flow. a Blunt body with
a sharp corner (Discontinuous curvature). b Blunt body with continuous curvature
continuous curvature however, as in Fig. 3.24b, there is no simple rule that identifies
the sonic boundary on the body. However, there are two empirically supported ideas
on the location of the sonic boundary on twodimensional and axisymmetric bodies.
These are (see [7] or [9]):
The location on the 2D body where b = max is approximately the sonic boundary
on the body. Here, b is the body angle and max is the maximum turning angle,
or flow deflection angle, corresponding to an attached plane oblique shock at the
freestream Mach number, M .
The location on the axisymmetric body where b = max is approximately the
sonic boundary on the body. Here, max is the maximum turning angle, or flow
deflection angle, corresponding to an attached conical shock at the freestream
Mach number, M .
130
(a)
(b)
90o
Approximate
Location of Sonic
Boundary
b = max
M> 1
M > 1
max
Fig. 3.25 Definition sketch of the approximate location of the sonic point on a blunt body [ is
the shock wave angle, is the flow turning angle and max is the maximum turning angle]. a Shock
wave angledeflection diagram boundary. b The approximation location of the sonic boundary
Asymptote
tan1
y = (tan ) x; Equation of
asymptote
y =(tan ) x 2 x 02 ; Equation of
hyperbola
or
M > 1
y = x 2 x02
x0
where:
sin1 (1/ M )
M2 1
and
x0 is the position of wave vertex
Fig. 3.26 Hyperbola as the approximate shape of a detached shock wave in a supersonic bluntbody
flow
131
Sonic
streamtube
S
yS
M > 1
1
( s + max )
2
L= xSB x0
max
SB
yc= 2y S /3
M=1
ySB
yc= yS /2
0
xSB
x0
Fig. 3.27 Definition sketch used in the calculation of the location of detached shock waves
132
yc
1 ys
1
V ys2 /2
ys
y dm =
y( V ) dy =
=
m
m 0
V ys
2
(3.188)
1 ys
1
2 V ys2 /3
2ys
y dm =
y(2 V )y dy =
=
2
m
m 0
ys V
3
(3.189)
These positions for the mass centroid are labeled in Fig. 3.27.
By applying the continuity equation to the sonic streamtube upstream of the
detached shock and the sonic flow through the sonic line downstream of the detached
shock, we establish the sonic flow area S S B (normalizing with respect to the
reference length scale of the problem, namely, y S B in plane 2D flows and reference
area scale, y S2 B in axisymmetric flows) and thus the length, L, where the vertex of
the detached shock is located upstream of the sonic boundary x S B . Stepbystep
calculation procedure based on the theoretical development of Ref. [7] is outlined in
Shapiro [9], and is not repeated here, for brevity. Once the position of the detached
shock is established, we may perform a momentum balance, in the streamwise, i.e.,
the xdirection, between the sonic streamtube upstream of the shock and the sonic
flow through the sonic line downstream of the shock. This results in an estimation
of nose pressure drag, upstream of the sonic line on the blunt body. Alternatively,
the momentum deficit in the xdirection between upstream and downstream of the
bow shock is equal to the pressure drag of the blunt nose, up to the sonic boundary.
Such calculations were performed by Ferri [2] and Moeckel [7], among others. The
momentum balance method as described here may also be applied to supersonic
inlets that operate in subcritical mode, i.e., with external spillage (see for example,
Ref. [1]). The subcritical mode corresponds to the operation of a supersonic inlet
where the normal (or terminal) shock is pushed outside the inlet and thus forms a
bow shock ahead of the inlet cowl lip. The momentum balance method yields inlet
cowl pressure drag in subcritical operation, or alternatively, the spillage drag of a
supersonic inlet.
Now, we examine some results of the theoretical prediction of the detached shock
position in bluntbody supersonic and upper transonic flows and the experimental
data. Figure 3.28 (from Refs. [7, 9]) shows the detached shock position in supersonic
flow (i.e., Mach 1.7) in front of a cone, sphere and projectile.
First, we note that the shock position is well captured by the theory. Second,
we note that the location of the sonic line is reasonably well predicted, however,
the experimental data point to a curved sonic line, as compared to the straight line
assumption in the theory of Ref. [7]. Finally, we note that the location of the sonic
boundary on the three blunt bodies is nearly independent of the body shape upstream
of the sonic point. Note that the normalizing length scale is yS B , as discussed earlier.
133
3.6
M = 1.7
Shock
3.2
2.8
Shock
2.4
M=1
2.0
Ladenburg
(Experimental)
Moeckel
(Theoretical)
ySB
1.6
Projectile
1.2
Sonic
Line
Sphere
0.8
Cone ( , )
Sphere ( , )
0.4
Projectile ( )
0
0.4
0.8
1.2
1.6
2.0
2.4
xx0
ySB
134
(b) 1.5
M
d
(in.)
1/4
1/2
1
d
L
1.0
(a) 18
Theoretical
16
Experimental
14
12
} TwoDimensional
0.6
1.0
AxiSymmetric
(c)
1.5
1.2
d
(in.)
1/4
9/16
3/4
9/16
3/4
9/16
1/4
1/2
1.1
L
ySB
10
Theoretical
1.0
8
L
6
TwoDimentional
(Theoretical)
0.9
Theoretical
0.8
0.7
Axially Symmetric
(Theoretical)
2
0
1.0
0.6
1.9
L
d
(deg)
35
35
35
45
45
60
90
90
0.5
1.4
1.8
2.2 2.6
M
3.0
3.4
3.8
0.4
1.0
1.5
1.9
Fig. 3.29 Detached shock position ahead of a axisymmetric and planar objects, b spheres and c
cones (after [9])
3.9 Summary
Transonic flow is more a physical description of the flow than a mere Mach number
range near sonic. From the lower end, i.e., the subsonic side, transonic flow is entered
at the critical Mach number. On the upper end, i.e., the supersonic side, transonic
flow persists until the profile is submerged entirely in supersonic flow. It is indeed
the mixture of subsonic and supersonic flow with the appearance of shocks on the
body that characterizes and dominates transonic aerodynamics. We learned that this
flow regime is inherently nonlinear and its analysis does not lend itself to full linearization of the perturbation potential equation. Therefore, the solution techniques
are numerical in nature and the classical theoretical developments lead to a group of
transonic similarity parameters and similarity laws. These laws allow for extrapolation of experimental data at different transonic flow conditions on affinelyrelated
bodies.
Problems
3.1 The peak suction pressure on an airfoil at 5 angle of attack is C p, min = 1.25
in lowspeed flow. Calculate the pressure coefficient at the same point on the airfoil, when the freestream Mach number is 0.7 using PrandtlGlauert compressibility
correction.
3.9 Summary
135
3.2 The lift curve slope, cl , of an airfoil in incompressible flow is 1.95 rad1 .
Calculate the lift curve slope for the same airfoil at Mach 0.75, using PrandtlGlauert
correction.
3.3 The lift coefficient of an airfoil at 5 angleofattack is cl = 0.5 in lowspeed,
i.e., incompressible flow. Calculate the lift coefficient of the same airfoil, at the same
angleofattack and at Mach 0.75 using PrandtlGlauert compressibility correction.
Also, calculate the percent change in circulation caused by the compressibility effects
at M = 0.75.
3.4 Calculate the critical pressure coefficient, C p, crit for a range of Mach numbers
starting at 0 and extending to Mach 1. Graph the critical pressure coefficient versus
Mach number.
3.5 An airfoil has a C p, min = 1.01 in low speed. Extend this result to the compressible domain, using three compressibility corrections:
(a) PrandtlGlauert compressibility correction
(b) Laitone compressibility correction, and
(c) KarmanTsien compressibility correction.
Also, use the graphical technique to determine the critical Mach number for this
profile and flow condition.
3.6 In a lowspeed WT test, i.e., M = 0, we measured C p, min on an airfoil.
M
Cp,min
=1.4
(a) At what freestream Mach number, M , will the minimum pressure coefficient
double?
(b) Calculate C p, crit , if the Mcrit = 0.72
(c) Calculate the lift curve slope at Mcrit if the lowspeed lift curve slope is a0 =
2 rad1 .
3.7 A thin airfoil is in a Mach 0.35 flow and develops peak suction, C p, min of 2.4
and a lift coefficient, cl of 0.78.
Calculate:
Cl.q .c
M
Cp, min
136
2x
yw = h cos
Calculate:
(a) the streamwise velocity at (2, h)
(b) the velocity vector at (, 2h)
(c) the minimum and maximum pressure coefficient, C p, min and C p, max on the wall
3.9 A twodimensional airfoil is in subsonic flight. The flight Mach number is M =
0.5 and C p, min = 3.5.
Cp, min =  3.5
M=0.5
Calculate:
(a)
(b)
(c)
(d)
3.10 A thin airfoil has a minimum pressure coefficient of C p, min = 0.5 at low
speed, i.e., incompressible limit.
y
M
=1.4
Cp, min
Calculate:
(a)
(b)
(c)
(d)
3.9 Summary
137
Calculate:
(a)
(b)
(c)
(d)
Calculate:
(a)
(b)
(c)
(d)
y = h cos(2x)
h <<1
x
0
1/4
1/2
3/4
138
3.9 Summary
139
Calculate:
(a) the pressure coefficient at the stagnation point
(b) the shock Mach number, Ms
(c) the pressure coefficient downstream of the shock
3.21 The pressure coefficient, C p , is measured at two locations on the surface of a
biconvex airfoil, as shown.
y
CpA =0.2
CpB =0
M =2.0
x
Calculate:
(a) p A / p B
(b) Ratio of freestream dynamictostatic pressure, q / p
3.22 A 12 % thick airfoil experiences C p, min, 1 = 0.75 at M1 = 0.85 at 2
angle of attack (i.e., 1 = 2 ). Using transonic similarity rule, we wish to relate
M2 , 2 and C p, min, 2 on an affinelyrelated airfoil that is 6 % thick, to airfoil 1.
Therefore, calculate M2 , 2 and corresponding C p, min, 2 for the 6 % thick airfoil,
as shown.
Cp, min, 1=  0.75
M 1=0.85
1=2o
t/c =12%
Cp, min, 2
t/c=6%
M2
3.23 A 2D body of 5 % thickness ratio, i.e., 1 = 0.05 is in transonic flow with the
flow Mach number of M1 = 0.90. The pressure coefficient at point A on this body
is C p1= 0.12. Now consider an affinelyrelated body that has a thickness ratio and
is in a Mach 0.95 flow, i.e., M2 = 0.95.
Calculate:
(a) the transonic similarity parameter, K
(b) the thickness ratio of the second (affinelyrelated) body, 2
(c) the pressure coefficient, C p2 , on the second body that corresponds to point A on
the first body
M1
M2
140
3.24 Use the transonic similarity rule on wave drag coefficients (on twodimensional bodies) and compare the wave drag coefficients of two affinelyrelated airfoils with 5 and 10 % thickness ratios, respectively. Assume the transonic similarity
parameter, K , remains constant between the two airfoils.
3.25 The lift coefficient of a 5 %thick airfoil at the angle of attack of 2 in Mach
M = 1.12 air stream ( = 1.4) is cl = 0.10. Apply transonic similarity theory to
predict the lift coefficient of an affinely similar airfoil that is 10 % thick in transonic
flight through air (with = 1.4)?
3.26 A thin symmetrical halfdiamond airfoil is in supersonic flow, as shown. Using
linearized supersonic flow theory; we are allowed to breakdown the wave drag problem of the airfoil into a separate thickness problem and a separate angleofattack
problem (on a flat plate). The lift is then purely due to the angle of attack on a flat
plate. Based on this approach to linearized supersonic aerodynamics, calculate:
(a) Airfoil wave drag coefficient due to thickness, cd, th
(b) Airfoil wave drag coefficient due to angleofattack, cd, Ao A
(c) Airfoil lift coefficient, cl
t/c =0.10
nose
= 6o
2
3
c
M = 5
3.27 An airfoil in Mach 0.80 flow at 2 AoA experiences the pressure distribution,
as shown below (not to scale). Calculate:
(a)
(b)
(c)
(d)
(e)
Cp
Cp, crit
1.0
x/c
(+)
0.35
Cp, stag
M >1
0.70
T.E.
3.9 Summary
141
3.28 For a family of slender axisymmetric bodies (that are affinelyrelated) in Mach
1 flow, use transonic similarity rule to compare their wave drag coefficients as a
function of their base diameters while keeping their length constant.
3.29 Consider a finite wing of aspect ratio A R1 = 8 and thickness ratio 1 = 0.10
is in transonic flight at Mach 0.9 (with = 1.4) at an angle of attack of 6 . First,
calculate the transonic similarity parameter, K , for this wing. Next, consider an
affinelytransformed wing of 5 % thickness, (i.e., 2 = 0.05), with = 1.4 and the
same similarity parameter, K 2 = K 1 . Using transonic similarity laws, calculate:
(a)
(b)
(c)
(d)
(e)
(f)
3.30 A detached shock is formed ahead of a cylinder in a Mach 1.2 flow. Estimate
the location of the sonic boundary on the cylinder. Assume = 1.4.
3.31 A detached shock is formed ahead of a sphere in a Mach 1.2 flow. Estimate the
location of the sonic boundary on the sphere. Assume = 1.4.
3.32 Calculate the coordinate, y S /y S B , of the sonic point, S, on a (3D) hyperbolic
detached shock ahead of a sphere. The sonic point on the shock, S, is where the
flow emerges from the shock at sonic speed. The freestream Mach number is 2.0
and = 1.4 [see Fig. 3.27 for definition sketch]. Also, calculate the flow angle S
downstream of the sonic point on the shock.
3.33 A detached shock is formed ahead of a 2D blunt object in a Mach 1.6 flow.
Assuming the shape of the detached shock is described by a hyperbola; calculate
the shock angle, c on the streamline passing through the mass centroid of the sonic
stream tube. Assume = 1.4.
Streamline passing y
through the mass
centroid in 2 D flow
yS
S
SB
s M=1
max
yS /2
ySB
x0
xSB
3.34 Consider a transonic flow where the freestream Mach number is:
M = 1 + where 1
142
Assuming a detached shock is formed, show that the Mach number downstream
of the normal shock is:
M2 1
M = 1 +
M2 1 
3.35 To demonstrate that weak shocks that appear in transonic flow are nearly isentropic, we choose the freestream Mach number to a normal shock to be:
M = 1 + where 1
Use the normal shock relations to show that nondimensional entropy rise across
the normal shock, s/R is of order of 3 , i.e.,
s
O(3 )
R
3.36 Concept of rotationality of the flow downstream of the curved shocks is derived
from Croccos theorem (for steady, inviscid flows), i.e.,
V ( V) = h t T s
where h t is the flow stagnation enthalpy. However, stagnation enthalpy remains
constant in gas dynamic flows involving shocks (of any shape) therefore Croccos
theorem reduces to:
V ( V) = T s
Use the result of Problem 3.34 to show that the vorticity field downstream of curved
shocks in transonic flow is negligibly weak and thus the flow may be modeled as
irrotational.
3.37 A 2D wedge of nose angle is in a supersonic flow. Apply the linearized
theory to the wedge and show that the expression for pressure coefficient may be put
in the transonic similarity form:
+1 2/3
1/3
[( + 1) /2] C p
2
=
2
2 1
2/3
M
3.9 Summary
143
3.38 Show that from linear theory in supersonic flow, a thin symmetrical biconvex
airfoil creates wave drag according to:
Cd, w
2
t
=
2
c
3 M 1
16
(Cd, w )diamond
2
t
= (Cd, w )min =
2
M 1 c
t
= const.
for
c
4
The above expression represents the minimum wave drag. You may use linear theory
in your proof.
Hint: Wave drag coefficient for a general thin airfoil shape using linear theory is
derived to be (see for example LiepmannRoshko, or Shapiro):
Cd, w =
4
2 1
M
A + B + 02
where
1c
c 0
1c 2
(x) dx
c 0 c
dh
dx
dx
144
References
1. Farokhi, S.: Aircraft Propulsion, 2nd edn. Wiley, Chichester, UK (2014)
2. Ferri, A.: Elements of Aerodynamics of Supersonic Flows. Macmillan Company, New York
(1949)
3. Hoak, D.E., Anderson, R., Goss, C.R.: USAF Stability and Control Datcom, Air Force Wright
Aeronautical Laboratories. Wright Patterson Air Force Base, Ohio (1978)
4. Laitone, E., Pardee, O.: Location of Detached Shock Wave in Front of a Body Moving at
Supersonic Speeds. NACA RM A7B10 (1947)
5. Liepmann, H.W., Roshko, A.: Elements of Gas Dynamics. Wiley, New York (1957)
6. McDevitt, J.B.: A correlation by means of transonic similarity rules of experimentally determined characteristics of a series of symmetrical and cambered wings of rectangular plan form.
NACATR1253 (1955)
7. Moeckel, W.E.: Approximate Method for Predicting Form and Location of Detached Shock
Waves Ahead of Plane or Axially Symmetric Bodies. NACA TN 1921 (1949)
8. Raymer, D.P.: Aircraft Design: A Conceptual Approach. AIAA, Washington (1989)
9. Shapiro, A.H.: The Dynamics and Thermodynamics of Compressible Fluid Flow. Ronald Press,
New York (1953)
10. Stack, J., Lindsey, W.F., Littell, R.E.: The Compressibility Burble and the Effect of Compressibility on Pressures and Forces Acting on an Airfoil. NACA TR 646 (1939)
11. Stanton, T.E.: A high speed wind channel for tests on aerofoils, Technical Report for the
Aeronautical Research Committee for the year 19271928, T&M 1130 (1928)
12. Tinoco, E.N.: CFD applications to complex configurations: a survey. In: Henne, P. (ed.) Applied
Computational Aerodynamics, Chap. 15. AIAA, Washington (1990)
Chapter 4
ShockExpansion Theory
Abstract This chapter develops the foundation of inviscid nonlinear shockexpansion theory on pointed bodies in supersonic flows. Fundamental equations
of shock and expansion waves are derived and discussed. To extend the analysis
to threedimensional bodies, conical shock theory is derived. Examples from diamond and biconvex airfoils are used to highlight the shockexpansion techniques in
developing lift and wave drag on twodimensional pointed bodies. Drag polar and
the aerodynamic performance parameter, (L/D)max are discussed. Axisymmetric
and slender bodies are the basis of 3D shockexpansion application. Approximate
methods based on the method of local cones is introduced to estimate the effect
of 3D expansion, using conical shocks associated with smaller nose angle. The
problem of asymmetrical flows is analyzed through the method of local cones. The
shockexpansion theory is extended to the upper transonic regime. Here, the principle
of stationary local Mach numbers near sonic freestream flow is demonstrated and
applied to wave drag of axisymmetric bodies at sonic speed. This chapter contains 9
examples and 42 practice problems at the end of the chapter.
4.1 Introduction
Supersonic flow over bodies involves the creation of waves of compression and
expansion to achieve flow turning. Blunt bodies, for example, create a detached, or
bow shock that forms ahead of the blunt body and the flow downstream of the shock is
a mixed subsonicsupersonic flow. Such mixed flows are not amenable to analytical
solutions. However, there is a class of problems in inviscid supersonic flow over
pointed bodies that lend itself to exact analytical solution (known as shockexpansion
theory). This class of inviscid supersonic flow problems requires the oblique shock
(either plane or conical) to be attached to the body in flight at the leading edge/nose.
This necessitates first the body to be pointed and second the turning imposed by the
body at the leading edge not to exceed the maximum turning angle corresponding
to a plane oblique (or conical) shock at the designated flight Mach number. Given
these stipulations, shockexpansion theory may be applied to any inviscid flow above
sonic, which includes the upper transonic regime.
Springer Science+Business Media Dordrecht 2015
R. Vos and S. Farokhi, Introduction to Transonic Aerodynamics,
Fluid Mechanics and Its Applications 110, DOI 10.1007/9789401797474_4
145
146
4 ShockExpansion Theory
1
M
tan =
(4.1a)
1
(4.1b)
M2 1
1
1
= tan1
= sin1
M
M2 1
(4.1c)
For example in a Mach2 flow, the Mach angle is sin1 21 , which is 30 . In sonic
flow at Mach 1, the Mach angle is 90 , i.e., normal to the flow and for Mach 1.1, the
Mach angle is reduced to 65.38 with respect to the local flow direction. However a
single Mach wave due to its infinitesimal strength is unable to provide finite turning
of the supersonic flow that is needed to make room for a body (of finite thickness) in
flight. Consequently, finite compression and expansion waves are created to turn the
supersonic flow. The coalescence of compression Mach waves creates shocks and
Mach wave
M>1
M=1
M 1
4.1 Introduction
(a)
147
(b)
Oblique
Shock
Compression
Mach waves
Streamline
Expansion
Mach waves
2
1
2
M1 > 1
M2 > 1
M2 < M 1
M2 > 1
M2 > M1
M1 > 1
1
Fig. 4.2 Waves on concave and convex surfaces in supersonic flow. a Waves generated on a concave
surface. b Waves generated on a convex surface
finite expansion waves are created through many expansion Mach waves, known as
the PrandtlMeyer expansion waves. Figure 4.2a shows the coalescence of compression Mach waves on a concave surface that culminates in the shock formation and
Fig. 4.2b shows the spreading of the expansion Mach waves on a convex surface that
create finite turning in the flow at supersonic speeds.
We note that the compression Mach waves on a concave surface coalesce, as
shown in Fig. 4.2a, and form an oblique shock wave whereas the expansion Mach
waves on a convex surface (Fig. 4.2b) spread/diverge and thus never form a surface of
discontinuity of rarefaction. The coalescence of compression Mach waves that form
the shock thus creates a region of very steep pressure and temperature gradients in the
gas where normal viscous stress and thermal conductivity of the gas play an important
role in establishing the equilibrium thickness for the shock wave. It is therefore the
dominance of viscous effects within the shock that creates an entropy increase and
thus nonisentropic flow conditions across the shock. We also know from kinetic
theory that the thickness of the shock is proportional to the meanfree path of the gas
at reference sonic temperature condition (e.g., see classical textbooks of Shapiro [6],
LiepmannRoshko [5], among others). The steep gradients in pressure, temperature,
density, normal component of velocity, entropy and stagnation pressure across the
shock are depicted graphically in Fig. 4.3.
The structure of a normal shock in standard sea level pressure and temperature conditions with upstream Mach number of 1.1 is shown in Fig. 4.4 (from [6]). Subscripts
x and y denote the flow conditions upstream and downstream of the shock, respectively. The streamwise coordinate x shows the region where the steep gradients exist.
Therefore, noting that the xaxis has a 105 inch scale, we appreciate the thickness,
or rather thinness of the shock in standard air. To put these numbers in perspective,
we recall that the meanfree path in air at standard conditions is 3.7 105 cm
(or 1.46 105 in.), which makes the thickness of the shock between 35 times
the meanfree path in air at standard conditions. The static pressure and temperature
across the normal shock are shown in ratios and the normal velocity component in
units of ft/s. The density ratio is not shown in the graph, but it rises as well as the
gas is compressed across the shock. It is thus the product of density and temperature
148
4 ShockExpansion Theory
(a)
u1
(b)
(c)
T2
p2
2
u2
(d)
p1
T1
(e)
2
(f)
s2
pt
1
1
s1
pt
Fig. 4.3 Schematic drawing of flow gradients across a shock, with shock thickness, (nottoscale).
a Normal component of velocity. b Static pressure. c Static temperature. d Density. e Entropy.
f Total pressure
Fig. 4.4 Variation of flow properties within a normal shock (from [6]). Note that x indicates the
conditions ahead of the shock wave, while y indicates the conditions behind the shock wave
4.1 Introduction
149
ratios following the perfect gas law that produces the static pressure ratio across the
shock, as shown in Fig. 4.4. To demonstrate the variation of normal shock thickness
with Mach number at standard pressure and temperature, we examine Fig. 4.5 (from
[6]). The normal shock thickness on a loglog scale indicates an exponential drop in
shock thickness as shock strength or Mach number increases. As a result of shock
thinning with increasing Mach number (at a given altitude), the gradients increase
and thus total pressure losses mount.
The normal shock relations are derived through the application of the great conservation laws of nature, namely conservation of mass, linear momentum and the
energy across the shock. The conservation of mass applied to a normal shock with
stations 1 and 2 identifying up and downstream of the normal shock, gives birth to
the first constant of motion, related to normal shock flows, namely
1 u 1 = 2 u 2 = const.
(4.2)
The application of the Newtons second law of motion to a control volume shown
in Fig. 4.6, gives birth to the second constant of motion for a flow through a normal
shock, namely
(4.3)
p1 + 1 u 21 = p2 + 2 u 22 = const.
Since the flow across a shock is adiabatic, i.e., there is no heat exchange between
an external source and the fluid in the control volume, and since the fluid within the
control volume undergoes no chemical reaction, the law of conservation of energy
for nonchemically reacting, steady flows across a shock produces the third constant
of motion, namely
150
4 ShockExpansion Theory
Normal shock
p2
T2
p1
T1
1
u1
s1
pt
1
ht
Control
Volume
u2
s2
pt
2
ht
h t1 = h t2 = const.
(4.4)
a 2
a 2
a 2
a 2
= cpT +
=
+
=
2
2
1
2
+1
1
a 2
= h t = const. (4.6)
2
(4.7)
2h t = const.
(4.8)
Any of the constant speeds, at , a or Vmax can serve as our reference speed in gas
dynamic calculations dealing with normal (or oblique) shocks. For example a has
served as the reference speed in shock polar and hodograph solutions in gas dynamics
and Vmax is used in TaylorMaccoll formulation of conical shocks, as we will briefly
review. Also a is used in the definition of a characteristic Mach number, M using
the following definition:
4.1 Introduction
151
u
a
(4.9)
The characteristic Mach number, is related to the Mach number through the law of
conservation of energy equation, namely,
( + 1) M 2
2 + ( 1) M 2
(4.10a)
2
( + 1) /M 2 ( 1)
(4.10b)
M 2 =
Or inversely,
M2 =
Note that subsonic flows with M < 1, have M < 1, sonic flow with M = 1, has
M = 1 and supersonic flow with M > 1 has M > 1. The added advantage of M
over M is that M is bounded whereas M may tend to infinity. This is seen in the
following limit:
+1
as
M
(4.10c)
M
1
In summary, primary constants of motion for a normal shock are:
u = const.
p + u 2 = const.
h t = const.
at = const.
a = const.
Vmax = const.
The simultaneous solution of the conservation laws yields the celebrated Prandtls
relation relating the product of gas speeds across a normal shock to the square of the
(constant) speed of sound at sonic state, a , following
u 1 u 2 = a 2
(4.11)
The immediate result of the Prandtls relation is the inverse relationship between the
upstream and downstream characteristic Mach numbers, namely
M1 =
1
1
or M2 =
M2
M1
(4.12)
152
4 ShockExpansion Theory
upstream Mach number, M, we may express the jump conditions across a normal
shock as follows:
( + 1) M12
2
u1
=
=
(4.13)
1
u2
2 + ( 1) M12
p2
2 2
M1 1
=1+
(4.14)
p1
+1
2 + ( 1) M 2
T2
2 2
1
M1 1
= 1+
(4.15)
T1
+1
( + 1) M12
2 + ( 1) M 2
2 2
s2 s1
1
M1 1
=
ln 1 +
R
1
+1
( + 1) M12
2 2
ln 1 +
M1 1
(4.16)
+1
1
2+(1)M12
2 + ( 1)
2 (1)
2
M
s
pt2
2
1
M12 1
= e R = 1 +
pt1
+1
2 + ( 1) M12
(4.17)
In addition, downstream Mach number is related to upstream Mach number through
the use of Prandtls relation, namely
M22 =
1 + [( 1) /2] M12
M12 ( 1) /2
(4.18)
The graphical depiction of the flow properties across a normal shock is very instructive and the levels and scales of these parameters may be used as an instructional
tool. For example in Fig. 4.7 we have plotted the static pressure, temperature and
density ratios across a normal shock for a caloricallyperfect gas with the ratio of
specific heats, = 1.4. Note that the upstream Mach number range is intentionally
limited to the supersonic range of up to Mach 5.
We note that the level of pressure jump is higher than both density and temperature
ratios, since in fact it is the product of the two jumps (in density and temperature)
that create the pressure jump. In the transonic range, we note that the behavior of
these jumps in static pressure, temperature and density is nearly linear. Limiting the
Mach number range to 1.3 and replotting the ratios of pressure, temperature and
density in Fig. 4.8 we can show this linear behavior.
The Mach number downstream of a normal shock (for = 1.4) is shown in
Fig. 4.9. We note that the initial drop in downstream Mach number is rapid and it
gradually plateaus with higher Mach numbers. In fact, there is a limiting downstream
4.1 Introduction
153
31.00
26.00
p2/p1
21.00
16.00
T2/T1
11.00
6.00
1.00
1
1.5
2.5
3.5
4.5
1.2
1.25
1.80
p2/p1
1.70
1.60
1.50
1.40
T2/T1
1.30
1.20
1.10
1.00
1
1.05
1.1
1.15
1.3
Mach number that can be deduced from expression (4.18) in the limit of upstream
Mach number tending to infinity, namely
M2
1
as M1
2
(4.19)
154
1.000
4 ShockExpansion Theory
0.900
0.800
M2
0.700
0.600
0.500
0.400
1.000
Downstream
Mach Number, M2(~)
2
3
4
Upstream Mach Number, M1 (~)
0.950
0.900
0.850
0.800
0.750
1
1.05
1.1
1.15
1.2
1.25
1.3
in Fig. 4.11. The labels for the zones of action and silence signify the space where
disturbance is felt and the space corresponding to forbidden signal propagation, or
zone of silence. The Mach cone downstream of the moving source thus represents
the zone of action and the vast zone upstream of the Mach cone represents the zone
of silence.
The motion of finite disturbances at supersonic speeds, as in bodies with finite
thickness moving faster than the speed of sound, involves the creation of oblique
shocks. The analysis of plane oblique shocks is based on the application of conservation principles in the normal and tangential directions to the shock. The schematic
drawing of an oblique shock in a supersonic flow, shown in Fig. 4.12, identifies the
wave angle , the flow turning angle, , and the normal and tangential components
of flow with respect to the oblique shock wave.
4.1 Introduction
155
(a)
(b)
Mach Wave
a. t
Mach Wave
V=a
V>a
V. t
Zone of
Silence
Zone of Action
Zone of Silence
Zone of Action
Fig. 4.11 Delineation of zones of action and silence for a moving source (of small disturbance) in
a compressible medium. a Source moving at the speed of sound. b Source moving at supersonic
speed, V > a
(a)
O.S.
Mn1 > 1
Mt1
Mn1
O.S.
w1
M2
M1
(b)
u1
V2
V1
w2
Mt2
Mn2
u2
Mn2 < 1
w2 = w1
Fig. 4.12 An oblique shock with its normal and tangential flow components. a The components
of flow Mach number across an oblique shock. b The structure of flow velocities across an oblique
shock
(4.20)
However, as the static temperature and thus the speed of sound are both raised across
the shock, the tangential Mach numbers up and downstream of the oblique shock
are unequal. Examining part (a) and part (b) of Fig. 4.12 shows the constancy of w,
but a drop in Mt across the oblique shock, while maintaining the same shock wave
angle, and the deflection angle, in the two graphs. Therefore oblique shocks
enjoy an additional constant of motion in comparison to normal shocks and that is
the tangential velocity. In summary the primary constants of motion for an oblique
shock are as follows:
156
4 ShockExpansion Theory
w = const.
u = const.
p + u 2 = const.
h t = const.
at = const.
a = const.
Vmax = const.
Since the normal component of conservation equations are the same for the
oblique and normal shock, all the jump conditions across an oblique shock are thus
solely determined from the normal component of upstream Mach number, namely,
M1 sin . We will thus replace the upstream Mach number, M1 , in normal shock equations (4.13)(4.17) by M1 sin to establish the jump conditions across an oblique
shock of angle . These are summarized as follows:
( + 1) M12 sin2
2
u1
=
=
(4.21)
1
u2
2 + ( 1) M12 sin2
2 2 2
p2
M1 sin 1
=1+
(4.22)
p1
+1
2 + ( 1) M 2 sin2
T2
2 2 2
1
M1 sin 1
= 1+
(4.23)
T1
+1
( + 1) M12 sin2
2 + ( 1) M 2 sin2
2 2 2
s2 s 1
1
M1 sin 1
=
ln 1 +
R
1
+1
( + 1) M12 sin2
(4.24)
2
M12 sin2 1
ln 1 +
+1
2 2 2
pt2
s
M1 sin 1
=e R = 1+
pt1
+1
1
2+(1)M12 sin2
2 + ( 1) 2 M 2 sin2 (1)
(4.25)
2
2
2 + ( 1) M1 sin
From the velocity triangles up and downstream of the oblique shock, we get the
following two equations:
4.1 Introduction
157
tan =
u1
w1
(4.26a)
tan ( ) =
u2
w2
(4.26b)
The ratio of these two equations eliminates the constant tangential velocity and
replaces the normal velocity ratio by inverse density ratio, to produce:
( + 1) M12 sin2
tan
=
tan ( )
2 + ( 1) M12 sin2
(4.27)
The range of the wave angle lies between the Mach angle, and 90 that
correspond to a normal shock. The flow turning angle, , thus starts from zero when
the oblique shock is an infinitesimalstrength Mach wave to a maximum turning
angle, max , which is shown in Fig. 4.13 (from [2]).
Example 4.1 The flow upstream of a normal shock is at Mach M1 = 2.0, in air with
= 1.4. The stagnation speed of sound is at = 300 m/s, calculate:
(a) Downstream Mach number, M2
(b) The ratio of speeds of sound, a2 /a1 across the normal shock
(c) The change in gas speed, u, across the shock in m/s
Solution:
We may use M22 =
1+[(1)/2]M12
M12 (1)/2
158
4 ShockExpansion Theory
90
M1=
2010
8
6 5
4.5
4.0
3.8
3.6
3.4
3.2
3.0
80
70
2.8
2.8
3.0
3.2
3.4
3.6
3.8
4.0
4.5
5
6
8
10
20
Weak Shock
50
40
Shockwave angle,
(deg)
60
30
20
shock wave
M2
10
streamline
12
16
20
Deflection angle, (deg)
Fig. 4.13 Oblique shock charts for air are from Ref. [2]
24
28
32
4.1 Introduction
159
90
80
70
2.2
Shockwave angle,
(deg)
60
Weak Shock
50
40
30
shock wave
M2
streamline
20
10
0
26
30
34
38
42
46
Deflection angle, (deg)
50
54
58
160
4 ShockExpansion Theory
strong solution is not used in the calculations; therefore it is shown as dashed line. In
part (b) of Fig. 4.14 we are showing an attached oblique shock (the weak solution) to
a 2D wedge or ramp where the nose angle is less than the maximum turning angle
at M1 . We also show an abrupt turning of the supersonic flow across the plane 2D
oblique shock in Fig. 4.14b. In part (c) of Fig. 4.14 we note the relative position of
the oblique shock and the Mach wave at M1 . It shows that the wave angle for the
oblique shock, , is greater than the Mach angle, , as expected. Finally, in part (d) of
Fig. 4.14, we show a detached shock formation upstream of a blunt body with a nose
angle greater than the maximum turning angle, max , allowed for an attached oblique
shock at freestream Mach number, M1 . We may think of bow shock formation as
the wave that is needed to create higher entropy rise across the shock in response to
higher aerodynamic drag of a blunt body.
The conical shocks, which are created on cones in supersonic flow at zero angle
of attack, create a new class of flows known as conical flowfields. Since the flowfield downstream of an attached conical shock lacks length scale, the flow variables
become a pure function of the (cone) angle in spherical coordinate system, as shown
in Fig. 4.15. The flow variables, e.g., p, T , , V , M, are thus constant on constant
cones in between the conical shock and the cone. These surfaces are indeed compression Mach cones that provide isentropic compression and continuous turning of
the flow downstream of a conical shock to accommodate the changing flow area in
the downstream direction. The changing flow area downstream of a conical shock
(a)
Oblique Shock
(strong solution)
(b)
Oblique Shock
(weak solution;
acceptable)
Oblique Shock
(weak)
strong
M1 > 1
weak
M1 > 1
<
<
max
Oblique
Shock
(c)
>
(d)
max
Mach
wave
M1 > 1
M1 > 1
<
max
90
nose
>
max
Fig. 4.14 Oblique and detached shock formation on pointed bodies of different nose angle in
supersonic flow. a Two theoretically possible oblique shock angles. b Attached oblique shock on
a 2D compression ramp. c The relative positions of oblique shock and Mach wave in supersonic
flow. d The formation of a bow shock ahead of a body in supersonic flow with nose > max
4.1 Introduction
161
Compression Mach cones
Conical Shock
Conical Shock
r
shock
M1 > 1
cone
Cone at
zero angleofattack
Compression Mach
Cones
Fig. 4.15 Formation of the conical shock on a cone at zeroangleofattack in supersonic flow
Mach
Cone
Conical
Shock
>
M1 > 1
Zone of Silence
Zone of Action
Fig. 4.16 Relative position of the conical shock and the Mach cone
is a new feature compared to the constant flow area downstream of a plane oblique
shock (see Fig. 4.14b). It is exactly this feature that leads to 3D relieving effect, as
we shall see. A graphical depiction of the continuous bending of a streamline is also
shown in Fig. 4.15.
Here again, we emphasize that the angle of the conical shock is greater than the
Mach angle corresponding to M1 , as expected. To emphasize this point and also
show the zones of action and silence on a cone, we graph Fig. 4.16.
162
4 ShockExpansion Theory
V =
dVr
d
(4.28)
V
+ V
=0
(4.29)
2Vr + V cot +
The density is further eliminated from the above equation by the application of Euler
momentum equation and the local speed of sound in the flow downstream of the
conical shock. The result is a secondorder nonlinear ordinary differential equation
in Vr that is known as the TaylorMaccoll equation:
d2 Vr
1
dVr 2
dVr
2
2
2Vr +
cot +
Vmax Vr
2
d
d
d2
dVr d2 Vr
dVr
dVr
=0
+
Vr
d
d
d
d2
(4.30)
Note that the only independent variable is and the only dependent variable is Vr . The
Vmax term is the reference speed that we introduced earlier in this chapter, as a constant
of motion, and is used in conical flowfields, following TaylorMaccolls theoretical
formulations. The Eq. (4.30) is still dimensional and may be nondimensionalized
3 . The resulting nondimensional equaby dividing both sides of the equation by Vmax
tion is:
2
dVr
d2 Vr
1
dVr
2
cot +
2Vr +
1 Vr
2
d
d
d2
dVr d2 Vr
dVr
dVr
+
=0
(4.31a)
Vr
d
d
d
d2
where
Vr
Vr
Vmax
(4.31b)
4.1 Introduction
163
From energy equation, V is related to the local Mach number according to:
V =
1/2
2
+
1
( 1) M 2
(4.31c)
There are many good references on the subject of conical flowfields and conical
shocks and the reader may refer to them to fill in the blanks in the above derivation.
An excellent contemporary reference is by Anderson [1] that is recommended for
further reading. The TaylorMaccoll equation (4.31a) is solved numerically using the
standard numerical approach of RungeKutta fourth order technique starting from the
shock at with the known Mach number and flow angle immediately downstream of
the shock. These are used to get Vr and V that serve as the initial conditions for the
RungeKutta solution of the TaylorMaccoll equation. Starting from the immediate
downstream of the conical shock, we march in the negative theta direction by taking
small incremental steps of . Per step in the numerical procedure, we calculate
the new Vr and V until we capture the cone at V = 0. In this approach, we are
finding the cone that supports the conical shock at . The numerical solution of
TaylorMaccoll equation has been used to develop conical shock charts shown in
Fig. 4.17. In part Fig: 3.17a, only the weak solution is plotted and we clearly note
the maximum cone angles that support attached conical shocks. In part 4.17b, the
ratio of cone surface static pressure to freestream stagnation pressure is plotted.
Part 4.17c is a graph of the surface Mach number on the cone for different cones
in supersonic flow. In part 4.17d, we have the ratio of stagnation pressures across
a conical shock for different cone angles and freestream Mach number. Part 4.17e
highlights the threedimensional relieving effect by comparing a cone to a wedge and
their corresponding wave angles. The wave angle for the wedge is greater than the
wave angle for the cone and thus the surface of the wedge experiences higher (static)
pressure than the (static) pressure on the corresponding cone. This is the essence
of threedimensional relieving effect. The cone wave drag coefficient, referenced to
base area, is also shown in part 4.17f.
Example 4.2 Consider a 20 (2D) ramp in a Mach 2 flow of air with = 1.4. Also,
there is a cone of semivertex angle of 20 in the same supersonic flow condition.
Calculate and compare:
(a) the wave angles on the 2D ramp and the cone
(b) the surface pressure on the ramp and the cone.
Solution:
From the plane oblique shock charts (Fig. 4.13) we read the weak shock wave angle
corresponding to Mach 2 freestream and a turning angle of twenty degrees as
53.4 . From the conical shock charts (Fig. 4.17a) we read the corresponding conical
shock angle of s 38 . Comparison of these two angles reveals the weaker 3D
shock as compared to the 2D shock.
164
4 ShockExpansion Theory
(a)
(b)
0.7
pS
0.6
, deg.
30
20
10
0.2
20
10
0.1
1.0
or
0.3
40
M1
0.4
in
50
pS
p1
M1 min or
2.0
3.0
4.0
1.0
2.0
1.0
0
10
20
30
40
2.0
0.8
M1 min or
0S
0.2
max
2.0
0.6
0.4
50
1.0
3.0
p01
(f)
ne
ge
Co
32
ed
24
Shock wave attached
to nose of both
cone and wedge
16
8
0
1.0
1.5
2.0
2.5
ma
45
50
3.0
4.0
M1
CD , (Referred to Base
Area of Cone)
or
max
, deg.
40
or
2.0
M1
(e) 48
in
M1
1.0
4.0
M
1m
10
15
20
25
30
35
40
0
1.0
4.0
3.0
0, 5
(d)
p0S/p01
MS
3.0
50
40
M1
MS
M1
30
M1
(c) 4.0
p1
0.5
M1
90
80
70
60
50
40
30
20
10
0
M1
3.0
M1, min or M1
3.5
4.0
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
1.0
40
M1
20
15
10
5
2.0
3.0
4.0
M1
Fig. 4.17 Conical shock charts for = 1.4 from Shapiro [6]. a Shock angle versus approach Mach
number, with cone angle as parameter. b Ratio of surface pressure to free stream stagnation pressure
versus free stream Mach number. c Surface Mach number versus free stream Mach number, with
cone angle as parameter. d Ratio of surface stagnation pressure to free stream stagnation pressure,
with cone angle as parameter. e Regions of shock attachment and detachment for cone and wedge.
f Pressure drag coefficient based on projected frontal area
4.1 Introduction
165
The surface pressure on the ramp is calculated based on the normal component of
the flow to the oblique shock, i.e., M1 sin = 2 sin(53.4 ) 1.6, which produces a
shock (static) pressure ratio of 2.82 downstream of the 2D shock. Using Fig. 4.17b,
we read the ratio of the cone surface static pressure to the freestream stagnation
pressure of 0.25. The stagnation and static pressures are related by (the isentropic
relation):
1 2 1
pt
= 1+
M
= 7.824 for M = 2
p
2
Therefore the ratio of cone surface pressure to freestream static pressure is: pc / p1
0.25(7.824) = 1.956 The counterpart of this ratio for the 2D ramp was calculated to
be 2.82. Here again we reveal the 3D relieving effect as evidenced by this example,
i.e., lower cone surface pressure.
Example 4.3 Compare the maximum turning angles of Mach2 flow of air in 2D
and 3D through plane oblique and conical shocks, respectively.
Solution:
Referring to the plane oblique shock charts, we note that max 23 in 2D. The
implication of this angle is that in twodimensional flow a plane oblique shock in
Mach 2 can at most cause 23 turning.
For the case of 3D flows, we refer to conical shock charts. From Fig. 4.17e, we
note that maximum turning at Mach 2 is nearly 40 . The implication is that conical
shocks can stay attached to cones of up to 40 semivertex angle in Mach 2 flow,
whereas in 2D the ramp angle above 23 causes the shock to be detached. This
too is due to 3D relieving effect.
We now direct our attention to flow expansion. The counterpart of compression
waves is the expansion waves, which cause flow acceleration and turning in supersonic flow. The direction of flow turning (with respect to local upstream flow) determines whether an expansion or a compression wave is needed to cause the necessary
flow deflection. One description that says turning the flow into itself is caused by
compression waves and turning the flow out of itself is accomplished by expansion
waves is useful when we consider right and leftrunning waves. We identify rightand leftrunning waves in supersonic flow through the eyes of an observer who stands
on the wave and is facing downstream. The rightrunning wave (RRW) appears on
the right side of the observer and the leftrunning wave (LRW) appears on the left
side of the observer. The expansion waves that are centered at a corner are referred to
as expansion fans and are known as PrandtlMeyer expansion waves. Let us examine some compression and expansion corners in supersonic flow and the waves that
cause the flow deflection in Fig. 4.18. The oblique shocks are marked as O.S. and
the PrandtlMeyer expansion fans are abbreviated to PM E.F. in Fig. 4.18. We also
have defined the flow angle, , with respect to horizontal, or xaxis and have assigned
positive values to when the flow direction is above horizontal and have assigned
a negative value to when the flow points in the downward (i.e., y) direction. We
can also discern the flow turning into itself and flow turning out of itself from
compression and expansion corners that are schematically shown in Fig. 4.18.
166
4 ShockExpansion Theory
(a)
(b)
LRW: LeftRunning Waves
LRW
O.S.
> 1
2
y
X
2
M1 > 1
M2
M2
M1 > 1
2
RRW
O.S.
< 1
2
(c)
(d)
LRW
PM E.F.
< 1
2
y
X
M1 > 1
1
M2
2
M2
2
X
M1 > 1
RRW
PM E.F.
> 1
2
Fig. 4.18 Definition sketch of compression and expansion corners and the corresponding waves that
are generated. a Compression corner, oblique shock, LRW. b Compression corner, oblique shock,
RRW. c Expansion corner, PrandtlMeyer expansion fan, LRW. d Expansion corner, PrandtlMeyer
expansion fan, RRW
Mach Wave
w
V
w
)
V
)
d
V + dV
d )
Fig. 4.19 An exaggerated view of infinitesimal turning across an expansion Mach wave
4.1 Introduction
167
Expansion waves, as noted earlier are Mach waves, which cause an infinitesimal
turning of the supersonic flow. The analysis of an expansion Mach wave can be
carried out using the definition sketch in Fig. 4.19. Note that we had to exaggerate
the infinitesimal turning in order to see the geometrical relations involving flow
angles and velocities across the Mach wave in Fig. 4.19.
By applying the law of sines to the triangle with sides V and V + dV , we may
relate the infinitesimal turning angle, d, and incremental speed change, dV , in the
following equation:
V
V + dV
=
(4.32)
sin 2 +
sin 2 d
The Eq. (4.32) may be simplified by expanding the sine terms to get:
cos
cos
1
dV
=
=
=
V
cos ( + d)
cos (d) sin
1 (d) tan
1 + (d) tan
1+
(4.33)
By taking isentropic flow conditions across the Mach wave; we may relate incremental
turning angle, d, to the incremental Mach number change, dM, according to:
d =
M 2 1 dM
1 + 1 M 2 M
(4.34)
Equation (4.34) may be integrated from zero turning, corresponding to Mach one
(sonic flow), to a general turning angle (M) corresponding to supersonic Mach
number, M. The angle (M) is called the PrandtlMeyer angle and has a closed
form solution, i.e.,
(M) =
+1
tan1
1
1 2
M 1 tan1 M 2 1
+1
(4.35)
In addition to a table for the PrandtlMeyer function and Mach angle, we graph these
two functions in terms of Mach number in Fig. 4.20. In Appendix C the relation
between M, and are tabulated up to a Mach number of 18.
Fig. 4.20 PrandtlMeyer
and Mach angles, in degrees,
in supersonic flow with
= 1.4
140
120
100
80
60
40
20
0
max
(M)
(M)
10
Mach Number, M
100
= 130.5
168
4 ShockExpansion Theory
+1
1
[rad]
1
2
(4.36)
2
w
= 22
Expansion
ramp
Head Mach
wave
1
M1 = 2.0
Tail Mach
wave
30
2
20.6
M2 = 2.836
4.1 Introduction
169
Solution:
The PM function in region 1 is 1 26.5 . The PM function on the expansion ramp is thus: 2 = 1 + 26.5 + 22 = 46.5 Therefore the corresponding Mach number is M2 = 2.836. The leading Mach wave makes an
angle of 1 = sin1 (1/2) = 30 and the tail Mach wave makes an angle of
2 = sin1 (1/2.836) 20.6 Finally, the ratio of static pressures on the two ramps
is related to their respective Mach numbers following the isentropic flow condition
up and downstream of the corner, namely
p2
=
p1
1 + ( 1) M12 /2
1 + ( 1) M22 /2
Upon substitution for the up and downstream Mach numbers (along with = 1.4),
we get: p2 / p1 0.2729. Therefore, flow expansion around a 22 corner in Mach
2 upstream flow caused a flow acceleration to Mach 2.836 on the downstream ramp
and the static pressure was reduced to 27.3 % of upstream value, i.e., p2 = 0.2729
atm., stagnation pressure, pt , remained constant.
L = N cos A sin
D = N sin + A cos
(4.37)
170
4 ShockExpansion Theory
y
A
D
N
L
x
M >1
x=c
Fig. 4.21 A pointed 2D body in supersonic flight with the resultant force (per unit span), R
resolved in normalaxial (N and A ) directions and liftdrag (L and D ) directions
N =
c
0
( pl pu ) dx
(4.38)
And the axial force (per unit span) is related to the integral of pressure times the
body slopes (along the chord) according to:
A =
c
pu
dyu
dx
pl
dyl
dx
dx
(4.39)
In terms of normal and axial force coefficients, cn and ca , we may express Eqs. (4.38)
and (4.39) in terms of the nondimensional pressure coefficients on the upper and
lower surfaces, namely
1c
C p C p dx
c 0
l u
dyu
dyl
1c
C pu
C pl
dx
ca =
c 0
dx
dx
cn =
(4.40)
(4.41)
Now, we are ready to apply these principles to a typical supersonic profile, namely,
a diamond airfoil.
Example 4.5 A symmetrical diamond airfoil is shown at 5 angle of attack (not to
scale). Use shockexpansion theory to calculate:
(a)
(b)
(c)
(d)
(e)
171
y
3
M = 2.4
15
15
15
15
Solution:
Freestream has to turn an additional 10 up to reach panel #1 (i.e., from 5 to 15 ).
This is the case of flow turning into itself. Therefore, the turning has to occur via an
oblique shock wave. From M chart for a plane oblique shock, we get:
OS
M = 2.4 and = 10
NS
M sin 1.306
33
p
1
p
to get:
C p1 0.1996
Also, the Mach number on panel 1 is:
M1 =
0.786
Mn 1
2.01
sin ( )
sin (33 10 )
OS
44.2
Again the normal component of flow establishes the shock jumps, namely
M sin 1.67
NS
0.6458
Mn 2
1.71
sin ( )
sin (44.2 20 )
172
4 ShockExpansion Theory
p
1
p
to get:
C p2 0.5273
To march to ramps 3 and 4, we recognize that the net turning angle is twice the
15 (i.e., 30 ) and since the flow is turning out of itself, we expect an expansion fan
to cause the turn. We start with 1 which corresponds to M1 and from PrandtlMeyer
table, we get: 1 26.38 and thus
PM
3 = 1 + 30 = 56.38
M3 3.35
The isentropic conditions between panels 1 and 3 can be used to establish static
pressure, p3 ,
1 + ( 1) M12 /2 1
p3
=
0.1272
p1
1 + ( 1) M32 /2
We use chain rule to write
p3 / p = ( p3 / p1 ) ( p1 / p ) = 0.1272 1.805 = 0.2296
Therefore, the pressure coefficient follows as
2
Cp =
2
M
p
1
p
or
C p3 0.1911
Now, marching onto panel 4 from panel 2, we note that the net turning angle is
again 30 . The PM angle for panel 2 is 2 18 , which gives the PM angle for
panel 4 to be 4 48 . From PM tables, we get M4 2.9. The isentropic conditions
between panels 2 and 4 can be used to establish static pressure, p4
p4
=
p2
1 + ( 1) M22 /2
1 + ( 1) M42 /2
0.15861
173
p
1
p
or
C p4 0.1251
The normal force coefficient is:
cn = 0.5 C p2 + C p4 C p1 C p3
Therefore, we get the normal force coefficient as cn 0.1968. To calculate the axial
force, we first establish the airfoil thicknesstochord ratio as t/c = tan 15 . This
gives t/c 0.2678 (a rather thick profile for supersonic flow!). Therefore, the axial
force coefficient may be written as:
ca =
t
1
C p1 + C p2 C p3 C p4
2
c
Fig. 4.22 Definition sketch of a general (asymmetrical) biconvex airfoil in supersonic flow at
angle of attack
174
4 ShockExpansion Theory
yu (x) and yl (x) respectively. The nose angles on the upper and lower surfaces, nu
and nl , are also identified. The chordwise position of the maximum thickness, xtmax ,
and the maximum thickness itself, tmax , are shown. The upstream supersonic flow is
at angle (angle of attack) with respect to chord.
First, we need to identify the waves that are needed to turn the flow on the upper
and lower surfaces at the leading edge (nose). Depending on the relative values of
the upper surface nose angle and the angle of attack, we may have an oblique shock
or a PrandtlMeyer expansion fan at the leading edge on the upper surface. This is
easily established by the following rules at the leading edge (for the case of positive
angle of attack, i.e., > 0):
Case 1: < nu The wave is an oblique shock, which turns the supersonic flow
by (nu ) angle (note that the flow turns into itself ).
Case 2: = nu A Mach wave appears at the leading edge on the upper surface
with zero net turning (since = nu ) imposed across the Mach wave.
Case 3: > nu A PrandtlMeyer expansion fan will be formed at the leading
edge on the upper surface, with a net turning of ( nu ). Note that the flow turns
out of itself.
(a)
(b)
(c)
Fig. 4.23 The wave pattern on a biconvex airfoil at positive angle of attack, . a Case 1: with
< nu (an oblique shock will form at the leading edge on the upper surface). b Case 2: with
= nu (upstream flow is aligned with the upper surface at the leading edge). c Case 3: with
> nu (PrandtlMeyer expansion fan is formed at the leading edge on the upper surface)
175
The leading edge of the lower surface will encounter an oblique shock, which
turns the flow by (nl + ) angle. The identification of the leadingedge waves will
allow us to march across the waves and onto the upper and lower surfaces of the
biconvex airfoil. The waves that are formed on the airfoil beyond the leading edge
are all expansion Mach waves that cause the flow to accelerate and the static pressure
to drop. In the vicinity of the airfoil, we will have the wave patterns corresponding
to Case1, Case2 and Case3, shown in Fig. 4.23 where solid lines show oblique
shocks and the dashed lines are expansion Mach waves.
The waves at the trailing edge (on the upper and lower surfaces) are formed to
equalize the static pressure and flow direction downstream of the airfoil. In supersonic
airfoil theory, we define the slipstream as a (free) vortex sheet that is formed at the
trailing edge of the airfoil that separates the upper and lower flows with flow directions
on two sides coinciding with the slipstream and the static pressure being continuous
across the slipstream. Although the flows on two sides of the slipstream are parallel,
they are not equal in magnitude, i.e., V4 = V5 . It is indeed the jump in tangential
velocity across the slipstream that is called the vortex sheet strength. Also, since the
airfoil is in the zone of silence of the waves at the trailing edge, the flow on the airfoil
is not affected by the waves at the trailing edge. However, these waves impact the
formation of slipstream (i.e., vortex sheet), which may impact aircraft components
downstream of the wing (e.g., engine inlet). We show the slipstream downstream
of a supersonic airfoil at high angle of attack and its two boundary conditions in a
definition sketch in Fig. 4.24. We have (arbitrarily) chosen to show an oblique shock
wave as well as a PrandtlMeyer expansion fan at the trailing edge of the airfoil in
Fig. 4.24. However, it is possible to have two oblique shock waves at the trailing
edge, if the flow environment downstream of the airfoil (i.e., static pressure and
flow direction) demands it. Indeed, the nature of these waves at the trailing edge is
established by the two boundary conditions on the slipstream, namely the equality
of static pressure ( p4 = p5 ), which is known as the pressure equilibrium condition,
and the flow angles (4 = 5 ).
Once, we established the nature of the waves at the leading edge, we may proceed
to calculate the pressure and Mach number immediately downstream of the leading
Boundary Conditions
PM E.F.
O.S.
M4
M1 > 1
4
1
p4 = p5
= 5=
4
p4
Slipstream
(Vortex sheet)
M4
M5
3
O.S.
PM E.F.
M5
p5
5
Fig. 4.24 The slipstream downstream of a supersonic airfoil at angle of attack with possible trailing
edge waves
176
4 ShockExpansion Theory
edge on the upper and lower surfaces of the airfoil. The flow downstream of the
leading edge on a biconvex airfoil undergoes a continuous acceleration on both
upper and lower surfaces (as shown in Figs. 4.23 and 4.24). Thus, the flow expands
isentropically on both sides of the airfoil downstream of the leading edge where
oblique shocks may appear. The difference between the upper and lower surfaces is
therefore in the initial conditions that are set at the nose by the type and the strength of
the waves that are formed at the leading edge. Otherwise, the solution methodology
is identical on the upper and lower surfaces of the biconvex airfoil.
Solution Methodology
We start at the leading edge, calculating the nose angle on the upper surface, according to:
(4.42)
nu = tan1 (dyu /dx)
x=0
Depending on the angle of attack, , we may have one of the three cases that we
described earlier as determining the wave at the leading edge on the upper surface.
We then march across the wave at the leading edge and establish:
1. Mnu , Mach number of the flow immediately downstream of the nose on the upper
surface
2. pnu , static pressure immediately downstream of the nose on the upper surface
We then propose to march downstream on the upper surface by creating a table of
values for the surface pressure (and Mach number). For this purpose, we may divide
the airfoil into (n + 1)chordwise calculation stations, from leading edge to trailing
edge, similar to the graph shown in Fig. 4.25. Here, we used the chord length, c,
as the normalizing length scale in the problem and graphed y/c versus x/c, as the
nondimensional coordinates. Counting the leading and trailing edges as calculation
stations, we have (arbitrarily) chosen 11 positions along the airfoil upper surface
in Fig. 4.25, which divides the airfoil into ten segments/panels along the chord, as
shown.
Next, we calculate the surface (inclination) angle at every calculation station,
namely at n = 2 to 11 [note that we have already calculated the leading edge angle,
i.e., n = 1 in Eq. (4.42)]. For the ith station, we use:
y/c
Panel number 7
n=1
n=2
n=3
n=4
n=5
n=6
n=7
n=8
n=9
0.3
0.4
0.5
0.6
0.7
0.8
n=10 n=11
0.0
0.1
0.2
0.9
1.0
x/c
Fig. 4.25 The upper surface of a general biconvex airfoil divided into 11 equallyspaced calculation
stations and 10 panels
177
u(i) = tan1 (dyu /dx)
(4.43)
x=xi
The Mach number immediately downstream of the nose that we calculated earlier
is now used to determine the PrandtlMeyer angle at the nose, i.e., n or 1 . Since
the flow is continuously expanding on the biconvex airfoil, the PrandtlMeyer angle
continually increases on the upper surface, from one station to the next, by the net
flow turning angle between those stations, namely
u(i+1) = u(i) + u(i+1) u(i)
(4.44)
The absolute value in Eq. (4.44) shows that the PrandtlMeyer angle increases as we
march towards the trailing edge on the biconvex airfoil. Now, we can determine the
local Mach numbers, Mu(i) based on the PrandtlMeyer angles, u(i) . The isentropic
flow condition on the upper (and lower) surface can be used to calculate the static
pressure according to isentropic rule:
1
2
M
1 + 1
u(i)
2
=
1
2
1 + 2 Mu(i+1)
pu(i+1)
pu(i)
(4.45)
We have now created the pressure distribution on the upper surface of the airfoil at
discrete points along the airfoil, from the leading edge to the trailing edge. By defining
the effective panel areas for each calculated pressure as the halfway points between
the adjacent calculation stations (or panels), we can proceed with the calculation
of the incremental force, per effective panel area. A definition sketch is shown in
Fig. 4.26 as an aid to understand the effective panel area and the calculation of the
incremental force and pitching moment based on pu(i) . The increment of normal
components of force (per unit span) on the upper and lower surfaces due to the static
pressures pu(i) and pl(i) are:
Nu(i) = pu(i) xi+1/2 xi1/2
Nl(i) = + pl(i) xi+1/2 xi1/2
(4.46a)
(4.46b)
Effective Panel
for pu(i)
pu(i)
y/c
i1
yu/c
i+1
x/c
i1/2
i+1/2
178
4 ShockExpansion Theory
The increments of axial component of force (per unit span) on the upper and lower
surface due to static pressures pu(i) and pl(i) are:
Au(i) = + pu(i) yu(i+1/2) yu(i1/2)
Al(i) = pl(i) yl(i+1/2) yl(i1/2)
(4.47a)
(4.47b)
The increments of pitching moment about the leading edge (per unit span) on the
upper and lower surfaces are:
Mu(i) = pu(i) xi+1/2 xi1/2 xi + pu(i) yu(i+1/2) yu(i1/2) yu(i) (4.48a)
Ml(i) = pl(i) xi+1/2 xi1/2 xi pl(i) yl(i+1/2) yl(i1/2) yl(i) (4.48b)
Finally, we add all the increments to get the normal and axial force components
(per unit span) on the upper surface and proceed similarly on the lower surface.
Equation (4.37) will then be used to express lift and wave drag (per unit span) in terms
of the normal and axial force and the angle of attack, . We will now demonstrate
the shockexpansion theory applied to an asymmetric, parabolic profile biconvex
airfoil with negative camber in the following example.
Example 4.6 Consider a 5 % thick biconvex airfoil of parabolic arc upper and lower
profiles with negative camber, as shown in the schematic drawing. The upper surface
contributes 2 % to thickness at 50 % chord and is thus defined by:
yu (x) = 0.08x (1 x)
The lower surface contributes 3 % thickness to the airfoil at mid chord and is thus
defined by:
yl (x) = 0.12x (1 x)
By virtue of negative camber, a stronger oblique shock appears at the leading edge
on the lower surface than the oblique shock on the upper surface. Therefore, we
expect this airfoil to produce lift (albeit small) even at zero angle of attack. Use
shockexpansion theory to first calculate the pressure coefficient on the upper and
lower surfaces, then calculate the lift, wave drag and pitching moment (about leading
edge) coefficients for this airfoil at the freestream Mach number of M = 1.4 and
at zero angle of attack. Graph the airfoil and its upper and lower surface pressure
distributions along the chord.
y
0.02c
nu
M = 1.4
x
0.03c
= 1.4
c
nl
NottoScale
179
Solution:
First, we calculate the leading edge angles for the upper and lower surfaces.
dyu
= (0.08 0.16x)x=0 = 0.08
dx x=0
Therefore, the nose angle on the upper surface is nu 4.57 . Similarly, we calculate
the leading edge angle for the lower surface,
dyl
= (0.12 + 0.24x)x=0 = 0.12
dx x=0
nl 6.84
In both cases, the supersonic flow turns into itself and thus requires an oblique shock
to form at the leading edge on both upper and lower surfaces. Using oblique shock
charts (Fig. 4.13), we get:
M = 1.4 and nu 4.6
OS
u 52.2
NS
OS
l 56.2
NS
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
1.4
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
1
0
0.004
0.007
0.010
0.013
0.015
0.017
0.018
0.019
0.020
0.020
0.020
0.019
0.018
0.017
0.015
0.013
0.010
0.007
0.004
0
0
0.006
0.011
0.015
0.019
0.023
0.025
0.027
0.029
0.030
0.030
0.030
0.029
0.027
0.025
0.023
0.019
0.015
0.011
0.006
0
0
0.001
0.002
0.003
0.003
0.004
0.004
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.004
0.004
0.003
0.003
0.002
0.001
0
4.57
4.12
3.66
3.21
2.75
2.29
1.83
1.37
0.92
0.46
0.00
0.46
0.92
1.37
1.83
2.29
2.75
3.21
3.66
4.12
4.57
6.84
6.16
5.48
4.80
4.12
3.43
2.75
2.06
1.37
0.69
0.00
0.69
1.37
2.06
2.75
3.43
4.12
4.80
5.48
6.16
6.84
Table 4.1 Geometric and flow calculations on a biconvex airfoil using Shockexpansion theory
M
u
l
x
yu
yl
ycamber
u ( )
l ( )
4.2525
4.7082
5.1645
5.6212
6.0783
6.5358
6.9936
7.4516
7.9098
8.3681
8.8264
9.2848
9.7431
10.2013
10.6593
11.1170
11.5745
12.0316
12.4884
12.9446
13.4003
u ( )
2.1848
2.8635
3.5440
4.2260
4.9094
5.5939
6.2795
6.9658
7.6527
8.3401
9.0276
9.7151
10.4024
11.0893
11.7757
12.4612
13.1458
13.8291
14.5112
15.1916
15.8703
l ( )
Mu
1.227
1.245
1.262
1.279
1.296
1.313
1.329
1.346
1.362
1.378
1.394
1.41
1.426
1.441
1.457
1.473
1.488
1.504
1.519
1.535
1.55
Ml
1.141
1.171
1.199
1.226
1.253
1.278
1.303
1.328
1.353
1.377
1.401
1.425
1.448
1.472
1.495
1.518
1.542
1.565
1.588
1.611
1.634
180
4 ShockExpansion Theory
1
0.9773
0.9554
0.9338
0.9126
0.8917
0.8723
0.8521
0.8334
0.8150
0.7969
0.7792
0.7617
0.7456
0.7288
0.7122
0.6970
0.6811
0.6664
0.6510
0.6369
1
0.9626
0.9283
0.8960
0.8645
0.8359
0.8081
0.7809
0.7544
0.7296
0.7055
0.6819
0.6600
0.6377
0.6169
0.5966
0.5761
0.5570
0.5384
0.5204
0.5029
1.2610
1.2324
1.2048
1.1775
1.1507
1.1244
1.1000
1.0745
1.0509
1.0277
1.0049
0.9825
0.9605
0.9403
0.9190
0.8981
0.8789
0.8588
0.8403
0.8209
0.8031
1.4124
1.3596
1.3111
1.2655
1.2210
1.1807
1.1413
1.1030
1.0655
1.0305
0.9964
0.9632
0.9321
0.9006
0.8712
0.8427
0.8137
0.7867
0.7604
0.7350
0.7102
0.190
0.169
0.149
0.129
0.110
0.091
0.073
0.054
0.037
0.020
0.004
0.013
0.029
0.044
0.059
0.074
0.088
0.103
0.116
0.131
0.143
0.301
0.262
0.227
0.194
0.161
0.132
0.103
0.075
0.048
0.022
0.003
0.027
0.049
0.072
0.094
0.115
0.136
0.155
0.175
0.193
0.211
0.0010
0.0005
0.0002
0.0001
0.0000
0.0000
0.0002
0.0004
0.0007
0.0005
0.0078
0.0051
0.0030
0.0011
0.0006
0.0021
0.0035
0.0048
0.0058
0.0034
dca,u
0.0007
0.0055
Table 4.2 Geometric and flow calculations on a biconvex airfoil using Shockexpansion theory
pu / pnose
pl / pnose
pu / p
pl / p
C p, u / p
C p, u / p
dcn
dca,l
0.0012
0.0017
0.0010
0.0005
0.0001
0.0000
0.0001
0.0005
0.0012
0.0022
0.0017
0.0072
0.0105
0.0071
0.0041
0.0017
0.0002
0.0015
0.0022
0.0022
0.0015
dcm,u
0.0106
0.0157
0.0108
0.0066
0.0030
0.0001
0.0019
0.0031
0.0032
0.0023
dcm,l
182
4 ShockExpansion Theory
y/c
l
Now, from the equation of the slope of the lower surface, dy
dx = 0.12 + 0.24x,
we calculate the surface angle at increments of 10 % chord. This procedure is easily
implemented in a spreadsheet program (e.g., Excel). Table 4.1 shows the spreadsheet
results pertaining to the geometry, PrandtlMeyer angle and Mach numbers on the
upper and lower surfaces of the airfoil. Tau is the thickness ratio for the upper and
lower surfaces, 0.02 and 0.03 respectively.
From local Mach number calculations, we determine the static pressure and thus
static pressure coefficient on the airfoil using isentropic relation (Eq. 4.45). The increments of force and pitching moment are also calculated using Eqs. (4.46)(4.48).
These calculations are performed using a spreadsheet program (see Table 4.2).
0.03
0.01
0.01
0.03
0.05
(a)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
x/c
(b) 0.3
0.2
Cp
0.1
Cp, upper
0.1
0.2
Cp, lower
0.3
0
0.2
0.6
0.4
0.8
x/c
Fig. 4.27 Application of shockexpansion theory to a 5 % thick biconvex airfoil with negative
camber. a The 5 % thick parabolic biconvex airfoil with negative camber (2 % upper thickness and
3 % lower thickness). b Pressure distribution on the upper and lower surfaces
183
The normal force coefficient is the sum of dcn column (in Table 4.2): cn
0.00233. The axial force coefficient is the sum of dca, u and dca, l columns (in
Table 4.2): ca 0.01453. The pitching moment coefficient about the leading edge
is the sum of dcm, u and dcm, l (in Table 4.2): cm, LE 0.0132.
Since the airfoil is at zero angle of attack, the normal and axial force directions
coincide with the lift and drag directions respectively. The shape of the airfoil, including the mean camber line is shown in Fig. 4.27a and the pressure distribution is shown
in Fig. 4.27b, from the Excel files.
Fig. 4.28 Cone at an angle of attack (and zero yaw) in a supersonic flow showing an asymmetrical
disposition of a conical shock attached to the cone at the vertex. a An axisymmetric cone at an angle
of attack. b Meridian angle, at the base of the cone
184
4 ShockExpansion Theory
z
Spiraling streamlines
on a cone with angle
of attack and yaw
Conical Shock
Leeward side
=0
M >1
M >1
Windward side
=270
=
90
=180
Cone
Fig. 4.29 Spiraling streamlines on a cone in supersonic flow at angle of attack and yaw
(a)
(b)
Curved shock
(Attached)
Local cone
M >1
M >1
Fig. 4.30 A (pointed) slender body of revolution in supersonic flow at zero angle of attack. a Slender
body of revolution at zero angle of attack. b Local cone (tangent) to point A has semivertex angle,
185
Example 4.7 Consider a slender body of revolution at zero angle of attack in supersonic flow with M = 2.0. The semivertex angle is 0 = 30 . Calculate the pressure
coefficient at the vertex and estimate the pressure coefficient at point A, where the
body angle is = 10 using the method of local cones.
A
M = 2.0
= 10
0
0
= 30
x
Solution:
From conical shock charts (Fig. 4.17), for Mach 2.0 and a cone of 30 semivertex
angle, we get the conical shock angle at the vertex to be 0 48 . The normal
component of flow to the shock is therefore:
M sin 0 2 sin 48 = 1.486
Therefore the static pressure ratio at the nose is calculated from
pN
2 2
M sin2 0 1
=1+
p
+1
To be p N / p = 2.4106. Assuming the pressure at point A corresponds to the cone
pressure of 10 semivertex angle in Mach 2 flow, we use Fig. 4.17b (the conical shock
charts) to get the ratio of the cone surface static pressure to the freestream stagnation
pressure (from Fig. 4.17b) to be: pc / pt 0.17. For Mach 2, pt / p = 7.824
and therefore
pA
p A pt
=
(0.17) (7.824) 1.330
p
pt p
We note the static pressure at point A is significantly less than the pressure at the
nose, as expected. The static pressure at point A is 55 % of the static pressure at
the nose.
The same approximation applies to a slender axisymmetric body in angle of attack,
. For a general axisymmetric body at an angle of attack, we define an equivalent
cone angle, c that is tangent to point A and the freestream direction is the axis of the
equivalent cone, as shown in Fig. 4.31. The body angle at A is , the nose angle is 0
and the equivalent cone angle is c . The meridian angle is also shown in Fig. 4.31,
which affects the equivalent cone angle according to (4.49).
186
4 ShockExpansion Theory
Equivalent cone (for point A on the
body) that creates a semivertex
angle, with respect to M vector
= 0
270
90
M >1
A
180
Fig. 4.31 Definition sketch used in the method of local cones for a slender body in angle of attack
2
cot sin2
2
(4.49)
Example 4.8 Consider a slender axisymmetric body, similar to the body shown in
Fig. 4.31, with the semivertex angle, 0 = 20 , in a supersonic flow with M = 2
and at 10 degrees angle of attack, i.e., = 10 . Apply the method of local cones to
point A on the body where the local body angle is 10 ; and A is on the windward
side of the body with = 180 . Estimate the pressure coefficient at A using the
method of local cones.
Solution:
The nose angle with respect to the flow direction on the windward side is the sum of
20 for the nose and 10 for the angle of attack (this may also be deduced from
Eq. (4.49)). Therefore, the conical shock angle corresponding to M = 2 and
equivalent cone angle of 30 is (the same as Example 4.7): 0 48 on the windward side. The rest of the calculations immediately downstream of the nose on the
windward side follow the Example 4.7, namely
M sin 0 2 sin 48 = 1.486
and the static pressure ratio at the nose (on the windward side) is calculated to be:
2 2
p N
2
M
=
1
+
sin
1
2.411
0
p =180
+1
To get the equivalent cone angle for point A, we substitute for the body angle at A,
namely = 10 and the meridian angle, = 180 as well as the angle of attack,
= 10 in Eq. (4.49). We get:
c = cos +
187
2
cot sin2
2
c = 20
Therefore the corresponding (equivalent) conical shock produces the ratio of the cone
surface pressure to freestream stagnation pressure of pc / pt 0.25 at M = 2
(from Fig. 4.17b). From isentropic table, we get pt / p = 7.824 for Mach 2, and
therefore we estimate p A / p to be:
pA
p A pt
=
(0.25) (7.824) 1.956
p
pt p
In Example 4.8, we assumed that the body surface pressure was the same as
the equivalent cone surface pressure. However, within the framework of method of
local cones; there are two hypotheses that are advanced on the flow conditions, i.e.,
pressure and velocity, at point A as compared to an equivalent cone that is tangent
to A, namely:
Hypothesis 1: pressure at A is identical as on the equivalent cone with c , good
for high velocities
Hypothesis 2: velocity at A is identical as on the equivalent cone, which is better
for low velocities
2.0
= 6
No. 1
1.0
No. 1
1.6
= 0
R6.5
1.2
No. 2
1.0
CD
= 6
5
= 0
R6.5
= 6
0.4
No. 3
No. 3
= 0
0
3.5
1.5
No. 2
0.8
1.0
2.5
M
Fig. 4.32 The effect of angle of attack and body shape on nose wave drag (after Ref. [3])
188
4 ShockExpansion Theory
We will examine these hypotheses more closely in the next section (Examples and
Applications). In addition, the base pressure is assumed to be equal to the freestream
pressure, i.e., pb = p , which is a reasonably accurate assumption for most applications of truncated bodies in supersonic flow. Once the pressure distribution on the
slender axisymmetric body is obtained [using (4.49) for an equivalent cone] and
further assuming that the base pressure is equal to the ambient pressure, we may
approximate the aerodynamic forces and moments on the body, namely, lift, wave
drag, side force and three moment coefficients for the pointed slender body in supersonic flow. The effect of angle of attack and body shape (i.e., nose bluntness) on
the nose wave drag coefficient is shown through experimental data in Fig. 4.32 (from
Krasnov 1970). Body number 1 is a flathead cylinder, whereas the bodies designated
as number 2 and 3 represent a streamlined nose of 1.5 and 2.5 (nose) fineness ratios,
respectively. Note that the lowerend range of freestream Mach numbers dips into the
transonic zone. Here, the effect of angle of attack is seen as an upward shift in C D .
(4.51a)
(4.51b)
(4.51c)
where R is the radius of curvature of the body (i.e., a constant for a tangent ogive).
In terms of the nondimensional coordinates, the equation for the (circular arc) ogive
is described by:
1/2
FN2
r = 1 2R 1 1 2 (1 x)2
(4.52)
189
The nose fineness ratio may be related to the nondimensional radius of curvature
via:
(4.53)
FN = R 1/4
The local body slope is obtained by differentiating r (x) with respect to x, i.e.,
1/2
FN2
1 dr
FN
dr
2
=
=
tan =
(1 x) 1 2 (1 x)
dx
2FN dx
R
R
(4.54)
FN
R
FN2
FN
+ 1/4
(4.55)
Based on the method of local cones described in Sect. 4.4, we propose to examine
the two hypotheses as they apply to a tangent ogive, namely:
1. the pressure on the body is equal to the pressure on the equivalent cone, of the
same local angle , and the stagnation pressure corresponding to 0 cone (at the
nose)
2. the velocity on the body is equal to the velocity on an equivalent cone, of the same
local angle, and the stagnation pressure corresponding to a cone of , semivertex
angle (i.e., for higher total pressure recovery than the first hypothesis)
(x)
dr
dx
r(x)
0
d
x
lN
R
0
190
4 ShockExpansion Theory
The first hypothesis assumes the same pressure results in a pressure coefficient, C pp
and the second method assumes the same surface velocity produces C pv . These
pressure coefficients are related to each other via [3]:
C pv = C pp +
2 ( 1)
2
M
(4.56)
where,
0
(4.57)
pt 0
0
; Stagnation pressure ratio on a cone with the nose angle 0 (4.58)
pt
pt 
; Stagnation pressure ratio on a cone with the nose angle
(4.59)
pt
The second hypothesis, i.e., C pv proves to be a more accurate estimate of the local
pressure coefficient than the C pp . The C pv leads to a negative pressure coefficient
on the aft sections of the ogive, i.e., near the shoulder and the shadow region that is
borne by both the Method of Characteristics as well as the experimental data. The
graphs of C pp and C pv on tangent ogives at Mach numbers 2 and 5 are shown in
Figs. 4.34 and 4.35 respectively (from Ref. [3]). Note that the pressure coefficient
C pv near the shoulder region of the tangent ogive where 0, dips into the negative
territory for tan 0 > 0.35, or semiapex angle of >19.3 .
Fig. 4.34 Pressure
coefficient on a tangent ogive
based on competing
hypotheses [3]
0.8
M =2
0.65
0.7
Cpp
0.60
0.55
0.6
0.50
0.5
0.45
0.4
ta
n
Cp
0.40
0.3
Cpv
0.35
0.30
0.2
0.25
0.20
0.15
0.1
0.10
0
0.1
0.05
0
0.2
0.4
tan
0.6
0
0.8
191
0.8
M =5
0.7
Cpp
A
0
0.65
0.6
0.60
Cp
0.5
0.55
ta
n
0.50
0.4
0.45
0.3
0.40
Cpv
0.35
0.30
0.25
0.2
0.1
0.15
0
0
0.1
0.2
0.4
0.6
tan
0.8
196FN2 16
14 (M + 18) FN2
(4.60)
Solution:
From Eq. (4.53), we establish the nondimensional radius of curvature of the nose
to be:
R = FN2 + 1/4 = 8.23
From Eq. (4.55) we calculate the semivertex angle at the nose:
0 = sin1 FN /R 20
192
4 ShockExpansion Theory
For the body angle of 11.2 and the semiapex angle of 20 we calculate the tangents
to be: tan 0.20 and tan 0 0.365. We use these values in Fig. 4.34 to get the
two pressure coefficients: C pp 0.13 and C pv 0.12 Finally, the pressure drag
coefficient of the tangent ogive is estimated from the semiempirical correlation
(4.60). But first, we need the pressure coefficient on the cone of the semiapex angle
20 , in Mach2 flow (from Fig. 4.17f, noting that the cone pressure coefficient and
the wave drag coefficients are the same since the base pressure is assumed to be equal
to p ), C pc 0.325
196FN2 16
C D = C pc 1
0.0998 0.1
14 (M + 18) FN2
(4.61)
This is the principle of stationary local Mach numbers near sonic freestream flow
conditions on slender bodies. This condition can be used as an extrapolation tool to
(a) 1.0
(b)
0.30
0.4
c
M =1
0.2
0
=1
0.6
0.20
(CDp)M
(CDp )M
=1
0.8
0.10
M =1
0
0
30
60
c
(deg)
90
30
0
60
90
(deg)
Fig. 4.36 Wave drag coefficient, C D p , at sonic flow (after Ref. [3]). a Cone. b Ogive
193
extend the results of force and moment coefficients from subsonic and supersonic
sides through the transonic region. We will apply this principle to a slender cone and
an ogive in transonic flow. First, the experimental results for the wave drag coefficient
for cones and ogives are shown in Fig. 4.36a, b respectively at sonic freestream Mach
number [3]. Note that the cone semiapex angle of 90 is the case of a flathead
cylinder and the ogive with the semivertex angle of 90 is the case of hemisphere.
Applying the principle of stationary Mach numbers to slender axisymmetric bodies,
e.g., cone or ogive, where transonic similarity applies, produces:
2
2 +1
1
2 M
M
CD =
(C D ) M =1 +
2
2
( + 1) M
( + 1) M
(4.62)
This expression is valid for axisymmetric slender bodies that include both cones
and ogives. Therefore the experimental data on C D cone at M = 1, as shown in
Fig. 4.36a substituted in Eq. (4.62) allows for the extension of wave drag in transonic
range, i.e., from subsonic to supersonic. In particular the slope of the drag coefficient
near sonic flow is determined. The proof of the match between the theory and the
experimental results on a cone in the transonic range is shown in Fig. 4.37a [3]. The
rising dashed lines through sonic Mach number are produced from the theoretical
model, i.e., Eq. (4.62). Similarly, the wave drag coefficient on an ogive in the vicinity
of sonic flow may be explored using Eq. (4.62) in conjunction with the ogive wave
drag coefficient for sonic flow, i.e., Fig. 4.36b. Comparison with experimental results
shown in Fig. 4.37b is in good agreement with the theoretical model.
dmid
(a)
(b)
M
lN
1.2
= 45
= lN /dmid
0.12
35
0.8
CD
CD
0.08
25
0.4
20
5
6
0.04
10
0
0.8
1.2
1.6
2.0
0.8
1.2
1.6
2.0
Fig. 4.37 Experimental data on wave drag coefficient in transonic flow (after Ref. [3]). Note that
N is the nose fineness ratio, FN in our notation. a Cone. b Ogive
194
4 ShockExpansion Theory
4.6 Summary
Shockexpansion theory is a powerful tool in analyzing aerodynamic forces and
moments on pointed bodies in steady supersonic flow where the oblique shock(s) at
the leading edge are attached. This method produces exact results within the confines
of inviscid flow theory. The method breaks down however when the shocks become
detached where the flow at the leading edge exceeds the maximum turning angle
supported by straight oblique shocks. The twodimensional flow problems are easily
analyzed using plane oblique shocks and expansion waves. The threedimensional
problems are analyzed using conical shocks. The 3D expansion through conical
Mach waves is best accomplished using the method of characteristics, i.e., the subject
of Chap. 5. Approximate method of local cones is introduced to estimate the effect
of 3D expansion, using conical shocks associated with a smaller nose angle. The
problem of asymmetrical flows is also analyzed through the method of local cones.
The extension of aerodynamic forces through sonic velocity is accomplished through
the principle of stationary local Mach numbers. The wave drag coefficient on cones
and ogives in transonic flow are in good agreement with the theoretical predictions.
Problems
4.1 A twodimensional projectile with a sharp nose is exposed to a Mach 3 flow, as
shown. Assuming the base pressure is p , calculate:
(a) p1 / p
(b) p2 / p
(c) wave drag coefficient referenced to chord, Cd D /q c.
p1
M = 3.0
p
5
10
p
p2
4.2 Calculate the stagnation pressure measured by a Pitot tube on an inclined ramp
in a supersonic flow, as shown.
4.6 Summary
195
10
M = 2.0
Pitot
p = 100 kPa
p3
MT.S. = 2.0
5
y
p2
p1
x
4.4 A supersonic flow expands around a sharp corner, as shown. Calculate the following parameters:
(a) Downstream Mach number, M2
(b) Flow area ratio, A2 /A1
(c) The angle of PrandtlMeyer fan envelope, .
Streamline
A1
M1 = 2.0
= 30 A
2
196
4 ShockExpansion Theory
4.5 A symmetrical halfdiamond airfoil has a nose angle nose = 30 and is exposed
to a supersonic flow. A Pitot tube is installed on each of the three surfaces, as shown.
Calculate the Pitot tube readings on the airfoil surfaces at zero angle of attack.
Pitot tubes
2
M1 = 3.0
pt1 = 100 kPa
3
30
30
4
Pitot tube
4.6 Calculate the wave drag coefficient of a 2D sharpnosed projectile, as shown,
assuming the base static pressure is equal to ambient, i.e., pbase = p . The 2D
wave drag coefficient here is defined based on b, i.e., Cd D /q b. Assume the
angle of attack is zero.
20
M = 2.0
4.7 A PrandtlMeyer centered expansion wave is visualized with the wave angles
as shown. Calculate:
(a) the flow turning angle, i.e., wall angle, w
(b) velocity ratio across the expansion waves V2 /V1 .
V1
30
16.6
w
V2
4.6 Summary
197
tmax
M = 2.4
15
4.9 Calculate the 2D lift and drag coefficients of a twodimensional wave rider, as
shown. Assume the base pressure is equal to the ambient, p4 = p1 . Shock wave
angle is 1 = 45 . The upper surface is aligned with the flight direction (surface 2).
Surface 3 is the lower surface and surface 4 is the base.
c
4
2
p4 = p1
M1 = 3.0
1
45
Oblique
shock surface
1
M = 2.0
= 1.4
10
20
198
4 ShockExpansion Theory
t
2
M1 = 5.0
= 1.4
4
10
10
10
10
4.12 Consider a 10 % thick biconvex airfoil of parabolic arc upper and lower profiles
with negative camber. The upper surface contributes 3 % to thickness at 50 % chord
and the lower surface contributes 7 % thickness to the airfoil at mid chord. First,
write the equations yu (x) and yl (x) for the upper and lower profiles. Next, use
shockexpansion theory to calculate:
(a) the pressure coefficient at the leading edge, and increments of 10 % chord until
trailing edge
(b) the lift coefficient
(c) the (wave) drag coefficient
(d) the pitching moment coefficient about the leading edge.
y
yu(x)
0.03c
M =2
x
0.07c
= 1.4
c
yl(x)
4.13 A hexagonal airfoil is a suitable shape for supersonic flow. Assuming the airfoil
shown is 10 % thick and it is at zero angle of attack in Mach 2.4 flow, use shockexpansion theory to calculate:
(a) pressure coefficients on all six panels
(b) the airfoil wave drag coefficient.
y/c
x/c
0.05
1.0
0.25
0.75
4.14 Identify the waves and draw the wave pattern around the twodimensional
symmetrical body at points A, B and C, as shown. For expansion waves, calculate
the wave angle of the head and tail waves and with the oblique shocks calculate the
wave angle. Note that the body is symmetrical and is set at zero angleofattack.
4.6 Summary
199
M1 = 3.0
30
A
10
4.15 A supersonic flow approaches an expansion corner, as shown. The static pressure ratio p2 / p1 is 0.25. Calculate:
(a) Mach number on the expansion ramp, M2
(b) Wall angle, w (in degrees).
p1
M1 = 2.5
p2
w
4.16 Two compression ramps in supersonic flow create two plane oblique shocks,
as shown. Calculate:
(a)
(b)
(c)
(d)
M1n = 1.2
1
M1 = 2.4
M2n = 1.2
2
1
OS1
M2
OS2
200
4 ShockExpansion Theory
4.17 A sharp hexagonal airfoil is placed in a supersonic flow. The Mach numbers
on its upper three surfaces are as indicated. Calculate:
(a)
(b)
(c)
(d)
M2 = 3
M1 = 2
M3 = 4
T.E.
L.E.
y
t/c = 0.10
M = 1.8
= 1.4
x
x=0
x=1
4.19 Apply shockexpansion theory to the diamond airfoil in supersonic flow (as
shown), to calculate:
(a)
(b)
(c)
(d)
(e)
(f)
1
12
2
M = 2.0
t1
14
t2
12
14
4
2
c
4.6 Summary
201
1
3
2
Write a computer program or use the exact shockexpansion formulas in a spreadsheet, to calculate and graph p3 / p1 for a range of Mach numbers, M1 = 1.5, 3, 5, 10
and a range of ramp angles, , from the corresponding wave angles, , to near maximum turning angles. Do we recover the flow static pressure p1 when we get to
region 3?
4.21 An oblique shock is described by its upstream flow components (u and w are
normal and parallel to the shock, respectively) as shown. Calculate:
(a) Upstream speed of sound, a1
(b) Characteristic Mach number, M2 .
O.S.
1
u1 = 360 m/s
w1 = 410 m/s
*
= 280 m/s
= 1.4
R = 287 J/kg.K
w2
V1
u1
u2
V2
w1
4.22 A flat plate is in Mach 1.2 flow at 2 angle of attack. Use shockexpansion
theory to predict its lifttowave drag ratio. The ratio of specific heats is = 1.4.
4.23 A biconvex airfoil has a nose angle, nose = 10 . For an angle of attack of 5
and freestream Mach number, M = 2.4, calculate:
202
4 ShockExpansion Theory
y
= 2
x
M = 2.8
= 1.4
= 1.4
M = 3.0
0
A
= 40
0
= 20
x
4.6 Summary
203
4.26 Consider a slender axisymmetric body, as shown, with the semivertex angle,
0 = 30 , in a supersonic flow with M = 2.4 and at 5 degrees angle of attack, i.e.,
= 5 . Apply the method of local cones to point A on the body where the local body
angle is 10 ; and A is on the meridian plane with = 120 . Estimate the pressure
coefficient at A.
= 5
0
= 30
x
A
M = 2.4
= 1.4
= 10
4.28 Use the wave drag coefficient for a cone of semivertex angle of 45 at M =
1.0, from Fig. 4.36a to create the pressure drag behavior of the cone in the transonic
range from M = 0.8 to M = 1.5. Graph your results and compare it to the
experimental data in Fig. 4.37a.
204
4 ShockExpansion Theory
M
=0
M
=15
M
=30
M
=40
1.0
0.8
0.6
0.4
0.2
0.0
0
10
20
30
40
(deg)
Finally, if we install a Pitot tube on the 30 cone in the Mach 3 flow, calculate the
stagnation pressure that the Pitot tube reads, as a fraction of flight stagnation pressure.
4.30 Use the wave drag coefficient for an ogive of fineness ratio 3.0 at M = 1.0,
from Fig. 4.36b. You first need to calculate the semivertex angle for a tangent ogive
that has a fineness ratio of 3.0 before you can use Fig. 4.36b. Next, calculate the
pressure drag behavior of the ogive in the transonic range from M = 1.0 to M =
1.4. Graph your results and compare it to the experimental data in Fig. 4.37b for the
ogive with fineness ratio 3.0.
4.31 A cone of 30 semivertex angle is in supersonic flow at M = 2.0 and zero
angle of attack. Use conical shock charts to find:
4.6 Summary
205
4.32 Consider a 30 (2D) ramp in a Mach 3 flow of air with = 1.4. Also, there is
a cone of semivertex angle of 30 in the same supersonic flow condition. Calculate
and compare:
(a) The wave angles on the 2D ramp and the cone
(b) The surface pressure on the ramp and the cone
(c) Explain the 3D relieving effect as evidenced by this problem.
30
M = 3.0
b
3
l
2
206
4 ShockExpansion Theory
30
2
M1 = 2.4
p1 = 30 kPa
T1 = 40 C
R = 287 J/kg.K
= 1.4
4.35 An oblique shock intersects a wall at a corner, as shown (not to scale). What
kind of wave(s) will form at the corner to accommodate the turning of the flow?
Calculate the Mach number in region 3.
M1 = 2.4
10
30
10
M = 1.6
= 1.4
y
x
O.S.
PM E.F.
4.6 Summary
207
4.37 In the method of local cones, we introduced a function for the cone semiapex
angle, c that was related to the angle of attack (in radians), local body angle and the
meridian angle following:
c = cos +
2
cot sin2
2
We wish to graph c as a function of x for a tangent ogive, for = 10 and the local
ogive angle varying between the nose angle, 0 = 45 to the shoulder value, where
= 0 around the circumference of the ogive, i.e., with varying between 0 and
360 . First make a table for versus x for a tangent ogive in increments of x = 0.1
along the axis of the ogive, i.e.,
x
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
( ) 45
0
Then, calculate (e.g., using a spreadsheet) and graph c versus at each x. Note
that there is a singularity at the shoulder where = 0 [due to cot () term], that
corresponds to x = 1. To avoid the singularity at the shoulder; we limit x to 0.99.
Identify the shadow region on the ogive at angle of attack, where c is negative.
4.38 Consider a flat plate at angle of attack in supersonic flow. Use shockexpansion
theory to calculate and graph the drag polar (i.e., graph of Cl vs. Cd ) for the airfoil
in the range of from 0 to 12 for constant freestream Mach numbers of M = 2,
3 and 5.
4.39 A symmetrical diamond airfoil has a semivertex angle of 20 in a Mach 2 flow
in air ( = 1.4).
(a) At what angle of attack, will the leadingedge shock detach?
(b) At what angle of attack will the flow downstream of the leadingedge shock be
sonic?.
4.40 Consider a 15 (2D) ramp in a Mach3 flow of air with = 1.4. Also, there is
a cone of semivertex angle of 15 in the same supersonic flow condition. Calculate
and compare:
(a)
(b)
(c)
(d)
208
4 ShockExpansion Theory
Compression Mach Waves
Free Boundary (constant pressure)
pa
1
60
y
x
pa
4
3
M1 = 2.6
pa
pa
Centerline
C
E
B
Cp, crit
Cp, stag
Cp, min
References
209
References
1. Anderson, J.: Modern Compressible Flow with Historic Perspective, 3rd edn. McGraw Hill,
New York (2003)
2. Anon.: Equations, Tables, and Charts for Compressible Flow. NACA TM 1135, Moffett Field
(1953)
3. Krasnov, N.F.: Aerodynamics of Bodies of Revolution. Elsevier Publication, New York (1970)
4. Liepmann, H.W., Bryson, A.E.: Transonic flow past wedge sections. J. Aeronaut. Sci. 17(12),
745 (1950)
5. Liepmann, H.W., Roshko, A.: Elements of Gas Dynamics. Wiley, New York (1957)
6. Shapiro, A.H.: The Dynamics and Thermodynamics of Compressible Fluid Flow. Ronald Press,
New York (1953)
7. Taylor, G.I., Maccoll, J.W.: The air pressure on a cone moving at high speed. In: Proceedings
of Royal Society, vol. 139. London (1933)
Chapter 5
Method of Characteristics
Abstract The governing equations of motion for two and threedimensional inviscid
irrotational supersonic flows are derived. Although these equations are nonlinear,
there are unique curves in the physical space, known as characteristics that turn
the governing partial differential equations into ordinary differential equations that
may be integrated along the characteristics. Therefore, the method of characteristics (MOC) is an exact solution technique that is graphical in nature and since it is
nonlinear, it applies to upper transonic flow when the local Mach number is sonic or
supersonic. The theory of MOC is developed and applied to 2D irrotational flows.
Within MOC, two competing techniques known as wavefield method and latticepoint approach are presented. The first example that is detailed is the minimumlength
supersonic nozzle design. The principle of wave cancellation at a solid boundary is
used in supersonic nozzle design and it is well described. The next application is
the wave pattern in supersonic exhaust plume. Here, wave reflection from a free
boundary, i.e., the free shear layer, is introduced and analyzed. The cases of underand overexpanded exhaust plumes are considered and MOC is applied to capture
the spatial evolution of the jet. MOC is extended to axisymmetric irrotational flows.
Family of supersonic nozzles, deflecting jets, nonuniform inlet condition, streamlines and ducts and curved shocks are the applications of MOC that are treated in
this chapter. This chapter contains 9 MOC application examples and 21 practice
problems at the end of the chapter.
5.1 Introduction
The governing partial differential equations in fluid mechanics are nonlinear, even in
the case of 2D inviscid, irrotational flows. Although, linearization as an approximate
method is successfully developed for thin, slender bodies at small angle of attack
in subsonic and supersonic flows, the accuracy of solution is diminished when perturbations are no longer small. In addition, we learned that some flow regimes, e.g.
transonic, may not be linearized even for slender bodies. Interestingly, in supersonic
flow where the governing partial differential equation is hyperbolic, there are unique
curves in physical space called characteristics that turn the governing nonlinear
Springer Science+Business Media Dordrecht 2015
R. Vos and S. Farokhi, Introduction to Transonic Aerodynamics,
Fluid Mechanics and Its Applications 110, DOI 10.1007/9789401797474_5
211
212
5 Method of Characteristics
(5.1)
For irrotational flows where the curl of the velocity field vanishes, i.e.,
V =0
v
u
=
y
x
(5.2)
The mathematical implication of the irrotationality condition (i.e., Eq. (5.2)) is that
a scalar potential function, exists such that:
V =
(5.3)
(5.4a)
(5.4b)
(5.5)
213
u
1 p
u
+v
=
x
y
x
(5.6a)
v
1 p
v
+v
=
x
y
y
(5.6b)
Since the flow is isentropic, we may relate the pressure gradient to the density gradient
and the local speed of sound according to:
p
= a2
x
x
(5.7a)
= a2
y
y
(5.7b)
Combining Eqs. (5.6) and (5.7) substituting them for the density terms in Eq. (5.5)
we get the following equation in velocity components u and v and the local speed of
sound:
u 2 u
2uv u
v2 v
1 2
2
+ 1 2
=0
(5.8)
a
x
a y
a
y
Replacing the velocity components in Eq. (5.8) with partial derivatives of the scalar
potential function, , we get the full potential equation in steady twodimensional,
irrotational flows:
y2
2x y
x2
1 2 xx
xy + 1 2 yy = 0
a
a2
a
(5.9)
214
5 Method of Characteristics
v
velocity derivative uy (or x
) is indeterminate. We form a system of three equations
and three unknowns by noting that:
du =
u
2
2
u
dx +
dy =
dy
dx +
2
x
y
x
x y
(5.10a)
dv =
v
v
2
2
dx +
dy =
dx + 2 dy
x
y
x y
y
(5.10b)
Therefore, the system of three equations for the three unknowns, x x , x y and yy
are written as:
u2
2uv
v2
1 2 xx 2 xy + 1 2 yy
(5.11)
a
a
a
dxx x + dyx y = du
(5.12)
dxx y + dy yy = dv
(5.13)
u
y
or
v
x ,
(1 u 2 ) 0 (1 v2 )
a2
a2
dx
du
0
0
dv
dy
xy =
(1 u 2 ) 2uv (1 v2 )
a2
a2
a2
dx
dy
0
0
dx
dy
(5.14)
The condition for the indeterminate x y is satisfied by demanding the numerator and
denominator of Eq. (5.14) to be simultaneously zero. First, by setting the determinant
in the denominator equal to zero, we get the equation for the characteristics, namely:
v2
u2
2uv
dy
dx = 0
1 2 (dy)2 dx
a
a2
a2
(5.15)
Dividing both sides of Eq. (5.14) by (dx)2 and rearranging, we get a quadratic equation for the slope of the characteristic curve, dy/dx:
2
u2
dy
v2
2uv dy
1 2
+ 1 2 =0
+ 2
a
dx
a
dx
a
(5.16)
215
dy
dx
=
uv
a2
uv
a2
uv
a2
characteristic
u2
a2
+
u2
a2
2
1 ua 2 1
2
1 ua 2
v2
a2
uv
a2
v2
a2
M2 1
2
1 ua 2
(5.17)
We note that the square root of (M 2 1) written at the end of Eq. (5.17) holds
the key to the nature of the characteristics, namely for subsonic flows the square
root becomes imaginary and thus we conclude that no (real) characteristics exist in
subsonic flows. In supersonic flows however, the square root yields real numbers and
thus we conclude that two (real) characteristics exist. At sonic flow, the square root
identically vanishes and thus one real characteristic exists. These in the language
of classifications of the second order partial differential equations (see Sect. 2.2.4)
are called elliptic, hyperbolic and parabolic respectively, corresponding to subsonic,
supersonic and sonic flows. Figure 5.1 shows the definition sketch for the velocity
vector in x y plane, its components and the flow angle, that it makes with respect
to xaxis. Based on this, we express the velocity components in terms of the velocity
magnitude and the angle, , according to:
dy
dx
characteristic
u = V cos
(5.18)
v = V sin
(5.19)
M 2 sin cos M 2 1
=
1 M 2 cos2
(5.20)
We express Mach number in terms of the Mach angle, in Eq. (5.20), to get
dy
dx
=
characteristic
sin cos
cot
sin2
2
1 cos
sin2
(5.21)
y
V
u
x
216
5 Method of Characteristics
2
2
1 ua 2 0 1 av 2
(5.23)
dx
=0
du
0
0
dv
dy
Expanding the determinant, we get:
u2
v2
1 2 dudy + 1 2 dvdx
a
a
(5.24)
Using the slope of the characteristics dy/dx (from Eq. (5.17)) in Eq. (5.24) and the
substitutions (5.18) and (5.19), we simplify the compatibility equation to the following form:
1
dv
=
du
1
u2
a2
v2
a2
dy
dx
characteristic
1
d (V sin )
=
=
d (V cos )
1
u2
a2
v2
a2
auv2
M2 1
u2
a2
(5.25)
C+ Characteristic with
/ dx)A = tan( A + A )
Tangent to C+ characteristic at A
(LeftRunning Mach Wave)
( dy
VA (> a)
A
A
Tangent to C characteristic at A
(Right  Running Mach Wave)
C Characteristic with
(dy / dx)A = tan( A A )
x
Fig. 5.2 Local characteristics are right and leftrunning Mach waves in twodimensional supersonic flow
217
d (V sin )
M 2 sin cos M 2 1
=
d (V cos )
1 M 2 sin2
(5.26)
(5.27)
(5.28)
(5.29a)
(5.29b)
Hence from an initial data line in 2D supersonic flows, we establish both local characteristic directions as well as the constants (K + and K ) along those characteristics.
Although characteristic directions change as we march in the flow direction, the constants that we established from the initial data line, (K + and K ), remain the same.
We had introduced the zone of action and zone of silence in Chap. 4 to lie within the
Mach cone downstream of a point in supersonic flow and the entire space upstream of
the Mach cone, respectively. In this chapter, the two characteristics that emanate from
a point in supersonic flow in the downstream direction are indeed consistent with the
boundaries of the zone of action. We can now extend this concept to upstream and
define a domain of dependence that impacts the flow at a point in supersonic flow that
lies within the upstream characteristics passing through the point. In this context, we
may regard the domain of dependence upstream of a point as the past events that
influenced the state of flow at point A and the region of influence in the downstream
direction to represent the future of the flow that is impacted by the presence at
point A. These physical arguments impact the finite differencing strategy (of information flow) in computational fluid dynamics. Figure 5.3 is a definition sketch that shows
the concepts of domain of dependence and region of influence in supersonic flow.
The construction of the characteristic network relies on three unit processes. These
unit processes help calculate flow properties at internal points, wall points and at
shock points. First, the internal point is created at the intersection of two characteristics that still lies within the flowfield. Consider points 1 and 2 in the x y plane
218
5 Method of Characteristics
y
C+ Characteristic
Domain of Dependence
of point A
VA
Region of Influence
of point A
C Characteristic
x
Fig. 5.3 Definition sketch of intersecting C+ and C characteristics and their physical significance
in supersonic flow
y
C+1
M1
y1
C1
y3
y2
M3
C3
C+2
C+3
M2
C2
x
x2
x1
x3
(see Fig. 5.4) for which we know the flow inclination/direction, 1 and 2 as well as
the Mach numbers, M1 and M2 . From local Mach numbers, we calculate the Mach
angles, 1 and 2 as well as the PrandtlMeyer (PM) angles, 1 and 2 . These angles
are sufficient to establish the characteristic directions for C+1 , C1 , C+2 and C2
as well as the compatibility constants K +1 , K 1 , K +2 and K 2 along those characteristics respectively. The intersection of C1 and C+2 creates point 3, which shares
the same compatibility constant with K 1 and K +2 that form two equations with
two unknowns, i.e.,
(5.30a)
3 + 3 = 1 + 1 = K 1
3 3 = 2 2 = K +2
(5.30b)
The solution to Eqs. (5.30a) and (5.30b) yields the flow angle 3 and (indirectly,
through PM angle, 3 ) Mach number at point 3:
219
3 =
K 1 + K +2
2
(5.31a)
3 =
K 1 K +2
2
(5.31b)
To graphically construct point 3, from the known points 1 and 2, we graph the
C+2 characteristic at the average angle corresponding to points 2 and 3 (along a C+
characteristic), i.e., C+2 is graphed at the angle of
2 + 3
2 + 3
+
2
2
(5.32)
2
2
(5.33)
Note that we are averaging the angles of the characteristics between points 1 and
3 and 2 and 3 and not the slopes (i.e., the tangent of the angles). Also, since we
are drawing straight lines for the characteristic network between the adjacent grid
points, in general the accuracy will improve if the points are closer to each other.
Figure 5.5 shows the average angles used in connecting C and C+ characteristics
to create point 3.
We next treat the unit process that involves the intersection of a characteristic and
an adjacent wall. Consider a point in the flowfield (e.g., point 1 in Fig. 5.6) that has
its local flow angle, 1 and the local Mach number, M1 known. Based on these, we
can establish the angles 1 and 1 from the Mach number M1 and thus calculate the
K +1 and K 1 corresponding to point 1 in the flow. Now, consider an adjacent wall
to point 1, as shown in Fig. 5.6.
y
1
y1
1 + 3 1 + 3
2 2
(C )13
y3
(C+ ) 23
2 + 3 2 + 3
2 2
2
y2
x
x2
x1
x3
220
Fig. 5.6 The unit process
involving the intersection of
a characteristic and an
adjacent wall
5 Method of Characteristics
y
y1
C +1
M1
C1
2
y2
w2
x1
x2
(5.34)
The flow angle at point 2 coincides with the wall angle at point 2 (from flow tangency condition), which means that we have an estimate for 2 , based on the C1
characteristic line that was drawn at the angle of (1 1 ) from point 1 towards
point 2. The estimate of 2 , from the known wall angle w at (x2 , y2 ), substituted in
Eq. (5.34) yields an estimate for PM angle 2 , according to:
2 = K 1 w2
(5.35)
The PM angle 2 from Eq. (5.35) then yields M2 and 2 . Now based on the new
information at the wall, namely 2 , we can redraw the C1 characteristic for the
average angle of
1 + 2
1 + 2
(5.36)
2
2
The intersection of the new (C )12 characteristic and the wall will result at a new
point 2, with its corresponding w2 . Depending on the correction of the wall angle
in our calculations, i.e., w2 w2 , we may have to iterate to reach the desired level
of accuracy. The factors that influence the correction angle of the wall are the wall
curvature and the proximity point 1 to the wall.
The next unit process involves the intersection of a characteristic, which is a
pressure wave, and an oblique shock. The drawing in Fig. 5.7 serves as a definition
sketch for this interaction.
At point 1 in the flowfield, we know the local Mach number, M1 and the flow
angle, 1 . Therefore, we know K +1 , which remains constant on C+1 characteristic.
The point 2 is at the intersection of the C+1 characteristic and the oblique shock
wave, as shown in Fig. 5.7. The interaction of a characteristic, i.e., a Mach wave
or pressure wave, with an oblique shock causes the shock strength to change, and
a characteristic, i.e., a pressure wave, to be reflected from the shock wave. These
are schematically shown in Fig. 5.7 where the angle of the oblique shock, has
changed to 2 after the C+1 characteristic has interacted with the wave. Also note
that a C2 wave is reflected from the oblique shock. Since point 2 is on the same C+
221
C2
C+1
M1
1
1
C1
Oblique Shock
Fig. 5.7 Definition sketch of a C+ characteristic (pressure) wave interaction with an oblique shock
wave
(5.37)
The flow angle at 2 is the turning or deflection angle across the oblique shock at 2,
which is related to M and 2 via oblique shock relations. The strategy is to guess a
flow deflection angle, 2 and then find the corresponding wave angle 2 from oblique
shock relations establish M2 and thus 2 . Then check Eq. (5.36) to see whether 2 2
is indeed equal to K +1 . We then need to iterate on our choice of flow angle at 2 to
converge on a consistent solution. Depending on the application, we may set the level
of acceptable accuracy at for example 0.01 . The classical textbooks of Liepmann
and Roshko [5] and Shapiro [7] as well as the contemporary book by Anderson on
compressible flow [1] should be consulted.
Example 5.1 We have experimentally measured the flow directions and Mach numbers at two points (1 and 2) in a supersonic flow, as shown. Calculate:
(a) The flow angle and Mach number at the intersection of the C 1 and C+2 , i.e.,
point 3.
(b) The angle of the C characteristic between the points 1 and 3.
(c) The angle of the C+ characteristic between the points 2 and 3.
(d) The coordinates of point 3.
y
y1 =1
M1
(C )1 3
y3
M3
M1
(C+ )23
y2 =0
2
x1 = x 2 =0
M2
x
x3
(deg )
M2
+5
1.8
(deg )
0
222
5 Method of Characteristics
Solution:
Based on the Mach number and the flow angle specified for points 1 and 2, we
complete the table:
Node M (deg) (deg) (deg) K (deg) K + (deg)
1
2 5
26.38 30
31.38
21.38
2
1.8 0
20.73 33.75 20.73
20.73
24.94
2
2
and the angle of the C+ characteristic connecting nodes 2 and 3 follows Eq. (5.32), i.e.,
2 + 3
2 + 3
+
34.64
2
2
The coordinates x3 , y3 are calculated from the intersection of two straight lines, with
known angles, i.e., slopes. We find:
x3 0.8652
and
y3 0.5977
Note that the fineness of the characteristic net is set by the closeness of the points
along the initial data line.
(a)
(b)
LRW or
C+ Characteristics
y
V
V> a
=0
Counter +
clockwise
Direction
223
V>a
= 0
Clockwise
Direction
V
RRW or
C Characteristics
x
Fig. 5.8 Basic expansion corners that create two families of characteristics that change the flow
angle to a Encountering RightRunning expansion Mach waves increasing the flow angle, .
b Encountering LeftRunning expansion Mach waves decreasing the flow angle,
224
5 Method of Characteristics
y
w, max
th
C+
C+
Mth =1.0
th =0
, R
Uniform flow
in the throat
section
C+
Cx
Centerline
C
C
Me>1
e=0
, R
C
Fig. 5.9 Centered expansion waves at the throat of a choked nozzle with uniform flow that form a
network of C+ and C characteristics
sonic line is straight and the flow at the throat is uniform, as shown in Fig. 5.9.
The condition of uniform flow is reasonable for nozzles with slow areavariation
convergent sections. However, if the area variation in the converging section prior to
the throat is large, it will affect the throat curvature and thus the sonic line is curved,
which causes flow nonuniformity. For additional discussion on this topic the reader
may consult Shapiro [7] and Ferri [3].
We take advantage of flow symmetry with respect to the xaxis and use the MOC
to design only the upper contour (or upper half) of the nozzle. The lower contour is the
reflection of the upper contour through the mirror of the xaxis. In essence since the
xaxis is a streamline, we may treat it as a solid wall. The reflected C+ characteristics
are then treated as reflections from a solid wall. Note that the symmetry argument
allowed the number of the nodes in the characteristic network to be reduced by
half. Figure 5.9 shows a sonic throat with uniform flow and four C characteristics
when we include the first, i.e., the = 90 , characteristic. The reflection of the
90 C characteristic from the xaxis is upon itself, as expected. However, it is
the next C characteristic with a prescribed incremental flow turning angle, initial
that creates a reflection as C+ characteristic, as shown in Fig. 5.9. The subsequent
C characteristics emanating from the sharp corner cause additional turning and
reflections that create the network of C and C+ characteristics. Returning to the
four characteristics shown in Fig. 5.9, we conclude that the characteristic that sets off
the MOC is the one with a nonzero initial . Let us use an expanded view of Fig. 5.9
and label the characteristic nodes, for further analysis (see Fig. 5.10). Note that the
exit Mach number, Me , is considered to be a nozzle design parameter, i.e., known.
Consequently, the PrandtlMeyer angle at the nozzle exit, e , is also a known design
parameter. The ratio of specific heats of the gas is and is treated as constant in our
application of MOC. In reality, the expansion process in the nozzle causes a reduction
in the gas temperature and thus chemical composition of the gas and change. The
same arguments apply to the gas constant, R, which changes with gas composition.
However, we treat both and R as constant, which means that we assume frozen
chemistry in the nozzle.
A
2
225
e
Ae
Me > 1
e = 0
e = f (Me)
= const.
R = const.
x
Fig. 5.10 The grid points in a characteristic network in a 2D supersonic nozzle
According to Fig. 5.10, the flow angle at node 8 is zero, 8 = 0, since it lies on
the xaxis (i.e., plane of symmetry). The C characteristic that passes through node
8 and the sharp corner at point A has to maintain the same K value, namely
8 + 8 = A + A
(5.38)
Node 8 is the last node on the xaxis; therefore its PrandtlMeyer angle is the same
as the exit value, e . Also, since the flow turning angle started from zero at the sonic
condition the value of the wall angle is the same as the corresponding PrandtlMeyer
angle, as discussed in Chap. 4. Therefore, Eq. (5.38) may be rewritten as
8 + 8 = e = A + A = 2w, max
(5.39)
We have now established a geometrical relation between the maximum wall angle,
w, max and the PrandtlMeyer angle at the nozzle exit, i.e., a nozzle design parameter
corresponding to the exit Mach number, Me . We have shown:
w, max =
e
2
(5.40)
Since we started the MOC with the selection of a small initial angle, node 1 attains
that angle. However node 1 is on the xaxis and its flow angle should be identically
zero. This indeed is the small discrepancy that our MOC introduces in the design
of the supersonic nozzle. The characteristic that connects points A and 1 is a C
characteristic that is also a Mach wave. We first determine the Mach number and the
corresponding Mach and PrandtlMeyer angles at node 1. The flow angle was the
initial that we had chosen to begin the MOC procedure, i.e.,
1 = initial
(5.41)
226
5 Method of Characteristics
Since the flow expanded from the sonic throat condition and turned by the initial
angle, the PrandtlMeyer angle attains the same value, namely
1 = initial
(5.42)
(5.43a)
K +1 = 1 1 = 0
(5.43b)
(5.44b)
(5.45)
Equation (5.45) is one equation with two unknowns, 2 and 2 . To establish a second
equation involving the same unknowns, we note that node 2 is on the C+ characteristic
that passes through node 1, therefore, it shares the same K + value with node 1, i.e.,
K +2 = 2 2 = K +1 = 0
(5.46)
(5.47)
From PrandtlMeyer angle 2 we get the flow Mach number M2 and subsequently
the Mach angle, 2 . Now, we have two straight characteristics to graph, one from A
to node 2 and one from node 1 to node 2. But this process is identical to the worked
Example 5.1.
The third C characteristic takes the flow turning at a through an additional
turning angle, . Following the same arguments as node 2, we get the angles 3
and 3 and the corresponding K values as
227
3 = 3 = initial + 2
(5.48a)
K 3 = 3 + 3 = 2(initial + 2)
(5.48b)
K +3 = 2 3 = 0
(5.48c)
e
2
K 3 = e
(5.49a)
(5.49b)
The connection between the design exit PM angle and the steps in our MOC is
revealed in Eq. (5.49a). Namely, for a give design e , we divide the PM angle by
two and select the initial turning angle and the subsequent steps in . In the simple
example of Fig. 5.10, we have taken two steps beyond initial .
Continuing with our calculations, we note that the flow properties at node 4 and
3 are the same, i.e.,
4 = 3 = w, max and 4 = 3 = w, max
(5.50)
We graph the wall angle between the two nodes on the wall at the average flow angles
at the two nodes, which in the case of A and 4, is w, max . Now, let us proceed to
node 5 on the xaxis. We conclude that the flow angle at 5 has to be zero and it lies
on the same C characteristic as node 2, therefore
5 = 0
(5.51a)
K 5 = K 2
(5.51b)
We get the PrandtlMeyer angle at node 5 from Eqs. (5.51a) and (5.51b) to be:
5 = 2 (initial + ) = e 2
(5.52)
The C+ characteristic that is reflected form the plane of symmetry has the value
K +5 = 5 + 5 = e + 2
(5.53)
K 3 + K +5
=
2
(5.54a)
228
5 Method of Characteristics
6 =
K 3 K +5
= e =
2
(5.54b)
K 6 = e
(5.54c)
K +6 = e + 2
(5.54d)
(5.55a)
7 = 6 = e
(5.55b)
The wall that connects nodes 4 and 7 is drawn at the average of flow angles at 4 and
7, namely
w, max +
4 + 7
=
(5.56)
w 47 =
2
2
Proceeding to node 8, we note that it lies on the xaxis and it is the last node, as we
had discussed earlier. Therefore,
(5.57a)
8 = 0
8 = e
(5.57b)
K 8 = e
(5.57c)
K +8 = e
(5.57d)
The flow properties at node 9 on the wall are the same as those at node 8. Therefore
the wall angle connecting nodes 7 and 9 is the average of the two flow angles at 7
and 9, namely
7 + 9
=
(5.58)
w 79 =
2
2
Equation (5.58) that states that the angle of the last wall segment is one half of our
step size, , is a general result. This means that regardless of the number of steps
in the selection in MOC, the last wall segment makes an angle of /2 with
respect to the xaxis.
Based on our graphical construction, the ratio y9 /ya in the x y plane represents
the ratio of exittothroat height, which for twodimensional nozzles, it is the ratio of
exittothroat area. This should theoretically be equal to the Ae /A corresponding
to Me and . However, there will be a discrepancy that is attributed to our nonzero
initial angle as well as the finite number of steps, i.e., s, which we took to reach
e /2. A reduction in initial has the largest impact on reducing the discrepancy in
area ratios followed by the size of steps. Now, we have completed the design of
a 2D, minimumlength isentropic nozzle with uniform inlet condition using MOC,
albeit with only three C characteristics to demonstrate the principle.
229
31.730
1.730
7.730
13.730
19.730
25.730
31.730
31.730
7.730
13.730
19.730
25.730
31.730
31.730
13.730
19.730
25.730
31.730
31.730
19.730
25.730
31.730
31.730
25.730
31.730
31.730
31.730
31.730
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
7.730
7.730
7.730
7.730
7.730
7.730
13.730
13.730
13.730
13.730
13.730
19.730
19.730
19.730
19.730
25.730
25.730
25.730
31.730
31.730
15.865
0.865
3.865
6.865
9.865
12.865
15.865
15.865
0.000
3.000
6.000
9.000
12.000
12.000
0.000
3.000
6.000
9.000
9.000
0.000
3.000
6.000
6.000
0.000
3.000
3.000
0.000
0.000
15.865
0.865
3.865
6.865
9.865
12.865
15.865
15.865
7.730
10.730
13.730
16.730
19.730
19.730
13.730
16.730
19.730
22.730
22.730
19.730
22.730
25.730
25.730
25.730
28.730
28.730
31.730
31.730
1.630
1.070
1.210
1.330
1.430
1.530
1.630
1.630
1.360
1.460
1.560
1.660
1.770
1.770
1.560
1.660
1.770
1.870
1.870
1.770
1.870
1.980
1.980
1.980
2.080
2.080
2.200
2.200
(deg)
37.862
69.195
55.764
48.778
44.393
40.834
37.862
37.862
47.356
43.252
39.889
37.061
34.418
34.418
39.889
37.061
34.418
32.344
32.344
34.418
32.344
30.350
30.350
30.350
28.750
28.750
27.049
27.049
1
1.398
1.5436
1.6256
1.6877
1.7443
1.7979
2.3072
1.807
1.9521
2.0646
2.1691
2.2707
3.3647
2.1538
2.2935
2.4348
2.5741
4.2023
2.4655
2.6414
2.8173
5.0641
2.8552
3.0722
6.0637
3.3346
7.2537
1
0
0.30955
0.43882
0.52717
0.60498
0.67787
1.3713
0
0.15439
0.2711
0.37912
0.485
1.6334
0
0.11698
0.2365
0.35702
1.7885
0
0.1226
0.2496
1.9019
0
0.13055
1.9805
0
2.0117
230
5 Method of Characteristics
A suitable choice may be initial = 0.865 and the remaining 15 divided into five
steps of 3 each, i.e., let = 3 .
There is a rule on the number of characteristic nodes in a MinimumLengthNozzle
(with uniform inlet condition), which states that:
No. of nodes = [2 + (n + 1)]
n
2
(5.59)
Nodes
(dy/dx) char
Nodes
(dy/dx) char
A to 1
A to 2
A to 3
A to 4
A to 5
A to 6
2 to 8
3 to 9
4 to 10
5 to 11
6 to 12
9 to 14
10 to 15
11 to 16
12 to 17
15 to 19
16 to 20
17 to 21
20 to 23
21 to 24
24 to 26
2.51235
1.27408
0.897
0.68756
0.5307
0.40374
1.17509
0.87118
0.67937
0.53173
0.40801
0.84059
0.67344
0.53676
0.42178
0.68017
0.55126
0.44178
0.57345
0.4671
0.49611
1 to 2
2 to 3
3 to 4
4 to 5
5 to 6
6 to 7
8 to 9
9 to 10
10 to 11
11 to 12
12 to 13
14 to 15
15 to 16
16 to 17
17 to 18
19 to 20
20 to 21
21 to 22
23 to 24
24 to 25
26 to 27
2.126231
1.576156
1.424069
1.373924
1.360623
1.361322
1.064162
1.037232
1.033785
1.04338
1.049892
0.837755
0.845641
0.865011
0.879234
0.696655
0.721989
0.735424
0.601679
0.618436
0.51031
231
Nodes
(dy/dx) char
Nodes
(dy/dx) char
A to 7
7 to 13
13 to 18
0.284045
0.247946
0.185243
18 to 22
22 to 25
25 to 27
0.131585
0.078662
0.026173
A
1
=
A
M
1+
1 2
2 M
+1
2
+1
2(1)
(5.60)
The area ratio for an exit Mach number of 2.2 and a gas with = 1.4 is
(A/A )isentropic = 2.004975. We note that the area ratio based on the MOC (2.0117)
is within 0.33 % of the theoretical value. This level of accuracy is remarkable when
we consider that we only did a 27point calculation.
In Fig. 5.11, the wall point at the throat is at (1, 1) whereas the nozzle exit coordinate at the wall, i.e., grid point 27, is at:
x27 = 7.2537 and y27 = 2.0117
The geometric implication of these coordinates is that a minimumlength nozzle
with uniform inlet condition and exit Mach number of 2.2 has a physical length
(downstream of the throat) that is 6.2537 times its throat halfwidth.
The last grid point on the axis of symmetry is number 26. The xcoordinate for
point number 26 is:
x26 = 3.3346
Since xthroat is at 1, the physical distance of the last C characteristic that intersects
the axis of symmetry is at 2.3346 (units of throat halfwidth) downstream of the throat.
The implication here is that the flow accomplishes its exit condition on the axis of
symmetry at only 2.3346/6.2537 or 37.3 % of the nozzle length. The flow continues
to evolve away from the axis up to the wall. Therefore the condition of uniform flow
is accomplished after the entire flow has accelerated to the exit condition. Figure 5.12
shows the evolution of Mach number along the centerline of the supersonic nozzle.
2.5
2
13
1.5
27
25
22
18
7
A
6
12
0.5
1
0
0
0.5
17
21
24
1.5
2.5
26
3.5
4.5
5.5
6.5
7.5
Fig. 5.11 The outer contour of a minimumlength 2D isentropic nozzle with uniform inlet condition designed for exit Mach number of 2.2 using MOC ( = 1.4)
232
5 Method of Characteristics
2
M 1.5
1
x
Fig. 5.12 The evolution of Mach number along the nozzle centerline ( = 1.4)
Note that the waviness of the Mach distribution curve along the nozzle axis is caused
by sparse (only 6 in this example) data points along the centerline (remember that
we used 6 C characteristics to accelerate the flow at the throat). Increasing the
number of the characteristics, or equivalently, reducing the turning angle, , of
each characteristic will smooth the Mach distribution curve along the nozzle axis.
LRW
1
M1
=0
V
LRW
Streamline
M3
2
1
2
LRW
M2
3
3
2
3
M4
4
4
Fig. 5.13 Simplewave problem in supersonic expansion flow over a convex surface with LRWs
233
when the flow crosses from field 1 to field 2 in Fig. 5.13, the flow angle is reduced
by the turning at the wall at point 1. The PrandtlMeyer angle, , also jumps across
the waves. In an expanding flow, such as the convex wall in Fig. 5.13, the PrandtlMeyer angle jumps up by the amount of wall turning angle, since the Mach number
in flow increases along a streamline that crosses the waves emanating from corners
1, 2 and 3.
For the general case where the flow upstream of corner 1 makes an angle, 1 , with
respect to the xaxis, the flow angles in regions 2, 3 and 4 are:
2 = 1 1
(5.61a)
3 = 1 1 2
(5.61b)
3 = 1 1 2 3
(5.61c)
The PrandtlMeyer angles in regions 2, 3 and 4 are related to the PM angle in
region 1 and the wall turning angles, following:
2 = 1 + 1
(5.62a)
3 = 2 + 2
(5.62b)
4 = 3 + 3
(5.62c)
From Eqs (5.62a), (5.62b), and (5.62c) we get the Mach numbers in regions 2, 3 and
4 respectively. The waves at the corners, as noted earlier, are C+ characteristics that
are drawn at the average Mach and flow angles according to definition:
dy
= tan( + )
(5.63)
dx C+
where the mean angles are defined as:
1 + 2
1 =
2
(5.64a)
2 + 3
2 =
2
(5.64b)
2 + 3
3 =
2
(5.64c)
1 = sin1
2
M1 + M2
(5.64d)
2 = sin1
2
M2 + M3
(5.64e)
234
5 Method of Characteristics
3 = sin1
2
M3 + M4
(5.64f)
Figure 5.14 shows the definition sketch of the physical Mach waves and the mean
Mach waves in an expansive supersonic flow over a convex surface.
It is important to note that the concept of mean Mach wave, as described by
Eqs. (5.64d)(5.64f), is mainly used as a tool that facilitates the numericalgraphical
solution to supersonic flows using wavefield method. It is used to replace the head
and tail Mach waves that envelope the PM expansion fans. However, these are all
physical waves, including the mean Mach wave that exist in nature and are fundamental to supersonic flows (see Fig. 5.14).
The counterpart of a convex wall is a concave surface, which in supersonic flow
creates compression Mach waves to turn the flow. The PM angle across a (leftrunning) compression Mach wave is thus reduced by the turning angle of the wall. The
reduced PM angle across compression Mach waves creates a reduction in flow Mach
number and an increase in Mach angle. Consequently, the compression Mach waves
on a concave wall converge, in contrast to the expansion Mach waves that diverge.
Figure 5.15 shows the simple wave problem in supersonic flow over a concave wall
and a representative streamline.
The flow deceleration across (the leftrunning) compression Mach waves as shown
in Fig. 5.15 satisfy the PM relations:
2 = 1 (2 1 )
(5.65a)
3 = 2 (3 2 )
(5.65b)
4 = 3 (4 3 )
(5.65c)
Note that the leftrunning compression Mach waves cause an increase in flow
angle, i.e., 4 > 3 > 2 > 1 , therefore PM relations (5.65a)(5.65c) result in a
____
y
x
4
2
3
1
3
2
M4
Convex wall
Fig. 5.14 Definition sketch for the physical and mean Mach waves in supersonic flow over expansion corners
235
M1
4
3
y
Streamline
Concave wall
Fig. 5.15 Supersonic flow over a 2D concave surfaceA simple wave problem with compression
Mach waves
reduction in PM angle, i.e., 4 < 3 < 2 < 1 , which in turn cause a reduction
in Mach number, M4 < M3 < M2 < M1 . To generalize the principle, we now use
rightrunning compression Mach waves on a concave surface a shown in Fig. 5.16.
The flow angle is reduced across the rightrunning compression Mach waves,
which is the opposite of the leftrunning waves of the same type. However, the flow
deceleration still applies, namely M4 < M3 < M2 < M1 . Therefore to generalize
the PM relations that we used in the case of leftrunning compression Mach waves,
Eq. (5.65a)(5.65c), may be cast in terms of the absolute value of flow turning angle
according to:
2 = 1 2 1  Compression Mach wave
(5.66a)
(5.66b)
(5.66c)
Concave wall
Streamline
1
2
3
Fig. 5.16 Turning of supersonic flow over a 2D concave surface by rightrunning compression
Mach waves
236
5 Method of Characteristics
The generalized Eq. (5.65a)(5.65c) apply to both types of compression Mach waves,
i.e., RRW as well as LRW. By using the same principle, the generalized case of a
convex surface in supersonic flow where expansion Mach waves cause a flow turning
and acceleration follow the PM relations:
2 = 1 + 2 1  Expansion Mach wave
(5.67a)
(5.67b)
(5.67c)
Figure 5.17 is a definition sketch for an expansive flow over a convex surface with
rightrunning (Mach) waves.
The nonsimple wave problem in supersonic flow involves the creation and interaction of both types of waves, i.e., leftrunning and rightrunning waves. We observed
an example of nonsimple wave in the design of minimumlength supersonic nozzle
where the throat created C characteristics (as in Fig. 5.10) and the C+ characteristics were created by the reflection of C characteristics from the xaxis, i.e., the
nozzle centerline. Now, we are ready to apply the wavefield method to isentropic
supersonic flows with waves of both families.
The unit processes in wavefield method are: wave generation, wave reflection
from a solid boundary, wave reflection from a free boundary and wave cancelation.
The method is best learned through an example. Wavefield approach is applied to
design of a minimumlength 2D nozzle as shown in Fig. 5.18. The exit Mach number
is 2.06, which requires 14 of flow turning to accelerate a sonic flow to Mach 2.06
flow. Subsequently, seven waves of each 2 strength are created at the sharp corner,
A (see Fig. 5.18). The numbers in cells are characteristic numbers. Once we learn the
rules for assigning the characteristic numbers, the process becomes easy. The first
rule is that the sum of the two numbers, 600 and 400, at the sonic cell adds up to
1,000. In subsequent cells the sum is less than 1,000. In the first cell after the sonic
throat, where the flow has crossed a 2 turn (LeftRunning) Mach wave, the sum of
the two numbers is 998 (i.e., 598 and 400). Therefore the difference between 1,000
Convex wall
RRW
3
4
2
1
RRW
Streamline
y
x
RRW
1
RRW
Fig. 5.17 A convex wall in supersonic flow with expansion Mach waves (RRW)
237
MinimumLength Nozzle
B
Wave Reflection
598
400
600
400
598
398
586
386 586
388
586
400
586
398
Wave Generation
Wave Cancellation Region
Fig. 5.18 A 2D minimumlength nozzle (MLN) designed using wavefield method (adapted from
[3])
and the sum of the numbers in the cell constitute the PM angle, , which is equal to
in magnitude. Now we present the rule about the difference of the two numbers.
The difference between the numbers in the cell minus 200 is the flow angle in the
cell. Note that at the sonic throat the difference between 600 and 400 is exactly 200,
which means that the flow angle at the throat is zero. In the next cell, the difference
between the numbers 598 and 400 is 198, which taking away 200 leaves a 2 of
flow angle in that cell. The next cell where the numbers 598, 398 are written mean
the following: the sum is 996, which means the PM angle in that cell is 4 . The
difference between the characteristic numbers is 200, which means that the flow
angle is 0. Note that the cell in discussion is adjacent to the centerline, where flow
has to attain zeroangle due to flow symmetry. In addition the flow has crossed a
LRW and its reflection, i.e., a RRW; therefore the flow angles of 2 encountered
by a LRW is canceled by the +2 induced by the RRW that was reflected from the
centerline. The reason for the PM angle to be 4 is due to crossing of two 2 Mach
waves, which add up to 4 . The cell that has the maximum wall angle is designated
with two numbers, 586 and 400. The sum of the numbers is 986, which means that
the PM angle in that cell is 14 . The difference between the two numbers is 186,
which implies the flow angle in the cell is 14 . Note that the flow angle crossing
the LRW is reduced; therefore we expect the flow angle to increase as we cross
the RRW, as in the reflection of the LRW from the centerline. The last cell that we
examine is characterized by two numbers, 586 and 386. Their sum is 972, which
implies that the PM angle is 28 . This angle corresponds to the exit Mach number
of 2.06. So far we have experienced the wave generation at the throat where seven
LRWs created the initial expansion. We also experienced the wave reflection from a
solid boundary, namely the centerline behaves as solid boundary. The rule was that
the refection from a solid boundary is in likemanner where expanding Mach waves
reflect as expanding Mach waves, except the reflections are of the opposite family,
namely the RRWs in this case.
238
5 Method of Characteristics
Table 5.4 Characteristic parameters that are used in wavefield approach ( = 1.4)
N
(deg)
M
(deg)
p/ p
1,000
999
998
997
996
995
994
993
992
991
990
989
988
987
986
985
984
983
982
981
980
979
978
977
976
975
974
973
972
971
970
969
968
967
966
965
964
963
962
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
1.000
1.081
1.133
1.176
1.219
1.258
1.294
1.330
1.365
1.400
1.435
1.469
1.503
1.537
1.571
1.605
1.638
1.672
1.706
1.741
1.775
1.810
1.845
1.880
1.915
1.950
1.986
2.024
2.060
2.097
2.132
2.173
2.211
2.249
2.288
2.329
2.369
2.411
2.453
90.000
67.640
61.988
58.217
55.153
52.665
50.617
48.763
47.114
45.585
44.176
42.901
41.701
40.595
39.546
38.551
37.620
36.730
35.876
35.061
34.283
33.542
32.828
32.143
31.483
30.847
30.233
29.617
29.041
28.484
27.972
27.405
26.897
26.407
25.918
25.428
24.965
24.508
24.063
1.000000
0.907248
0.851025
0.804867
0.761912
0.723304
0.688964
0.655897
0.624887
0.594836
0.566083
0.539221
0.513255
0.488759
0.465116
0.442326
0.420759
0.399939
0.379809
0.360539
0.342107
0.324543
0.307674
0.291536
0.276111
0.261340
0.247249
0.233236
0.220336
0.208035
0.196900
0.184785
0.174144
0.164089
0.154276
0.144668
0.135835
0.127330
0.119294
(continued)
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
239
(deg)
p/ p
2.495
2.537
2.581
2.626
2.671
2.718
2.765
2.812
2.860
2.911
2.961
3.012
3.065
3.119
3.174
3.226
3.287
3.346
3.407
3.468
3.530
3.595
3.661
3.728
3.798
3.870
3.943
4.018
4.095
4.175
4.256
4.340
4.427
4.517
4.610
4.704
4.802
4.905
5.010
23.632
23.211
22.794
22.387
21.985
21.591
21.202
20.833
20.465
20.090
19.736
19.389
19.042
18.701
18.367
18.06
17.71
17.39
17.07
16.76
16.46
16.15
15.85
15.56
15.27
14.98
14.69
14.41
14.13
13.86
13.59
13.32
13.06
12.79
12.53
12.27
12.02
11.76
11.51
0.111723
0.104549
0.097667
0.091182
0.084987
0.079144
0.073585
0.068507
0.063651
0.058907
0.054618
0.050599
0.046760
0.043166
0.039813
0.03684
0.03373
0.03097
0.02836
0.02597
0.02378
0.02172
0.0198
0.01803
0.01639
0.01485
0.01346
0.01218
0.01099
0.0099
0.00891
0.008
0.00717
0.0064
0.00571
0.00509
0.00452
0.004
0.00354
(continued)
240
5 Method of Characteristics
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
(deg)
p/ p
5.119
5.231
5.348
5.472
5.599
5.730
5.865
6.007
6.152
6.307
6.472
6.642
6.821
7.008
7.202
7.407
7.622
7.853
8.092
8.343
8.616
8.902
9.210
9.535
9.881
10.260
10.665
11.088
11.552
11.27
11.02
10.78
10.53
10.29
10.05
9.817
9.583
9.355
9.123
8.888
8.659
8.43
8.204
7.981
7.759
7.539
7.316
7.099
6.884
6.665
6.45
6.233
6.02
5.809
5.593
5.38
5.174
4.966
0.00312
0.00274
0.00241
0.0021
0.00183
0.00159
0.00138
0.00119
0.00103
0.00088
0.00075
0.00064
0.00054
0.00045
0.00038
0.00032
0.00027
0.00022
0.00018
0.00015
0.00012
9.6E05
7.7E05
6.1E05
4.8E05
3.8E05
2.9E05
2.2E05
1.7E05
The reflected waves from the centerline are still responsible for more flow acceleration in subsequent cells, as their expansive nature is preserved in a likemanner
reflection. We learn that the opposite is true for the reflection of waves from free
boundaries when we discuss underexpanded nozzle flows. The wave cancellation is
accomplished at the wall where we turn the wall to meet the flow angle upstream of
the wave, i.e., before it meets the wall. To demonstrate this point let us examine the
cell next to the maximum wall angle cell where the characteristic numbers are 586
and 398. The sum is 984 which means that the flow PM angle is now 16 , which
is 2 more than the previous adjacent cell (with 586 and 400 numbers). The difference between the numbers is 188, which means that the flow angle in the cell is
241
Outer wall is a streamline
Intermediate
Streamlines
Intermediate
Streamline
Wave Reflection
Nozzle Length
Centerline is a streamline,
thus behaves as a solid wall
Fig. 5.19 Wavefield method is used to design a 2D supersonic nozzle with an extended initial
expansion region that produces longer nozzle than MLN (adapted from [3])
12 . Since the flow angle in the wall cell and the wall angle have to match, the
wall angle is 12 . The cell where the maximum wall angle is a boundary had a
flow and wall angle of 14 ; therefore the wall has turned in the positive direction, i.e., in counterclockwise direction by 2 . This is the wave cancellation step in
wavefield method. We note that the wave strength does not change in reflections,
i.e., the 2 that we started the initial expansion persists through reflections. In this
example, we learned about the characteristic numbers, the wave generation step, the
wave reflection from a solid boundary step and the wave cancellation step at the wall.
To facilitate the numericalgraphical calculation, a suitable table that has a column
for characteristic number, N , which is the sum of the two numbers in a cell, is added
to a PM table that includes Mach angle and p/ p as additional columns. Table 5.4
shows the characteristic parameters that are used in the wavefield approach in MOC
for a gas with = 1.4. These parameters and more detail may be found in NACA
Report 1135 [2].
The wave generation step in Fig. 5.18 was concentrated at a point that led to the
design of minimumlength nozzle (MLN). However the concentrated expansion at the
throat is very abrupt which may affect the flow quality at the nozzle exit. The MLN
is suitable for flight due to its reduced weight/cost, but in stationary applications,
such as supersonic wind tunnels, the weight or cost advantage is overshadowed by
the flow quality advantage of accomplishing uniform, wavefree flow condition in
the test section. As a result, supersonic nozzles with an arbitrarily stretched wavegeneration region are designed for stationary applications. The example shown in
Fig. 5.19, which is adapted from Ferri [3], shows a spread of the wave generation near
the throat, followed by wave reflections from the centerline and wave cancellations
at the wall similar to the previous MLN case. In this example Ferri addresses the
case of a nozzle with fixed length and then works in the reverse order to establish
the waves that are necessary in its design. Our purpose was however to show that the
wave generation region may be spread to accomplish higher quality flow in the exit
plane/test section of a supersonic wind tunnel.
242
5 Method of Characteristics
Incident pressure
wave (expansion)
Reflected pressure
wave (expansion)
LRW
RRW
y
x
Solid Wall
Reflected pressure
Incident pressure
wave (compression) wave (compression)
LRW
y
RRW
x
2
1
So lid W all
Fig. 5.20 Incident and reflected pressure waves (of expansion and compression types) from a solid
wall
Example 5.3 Show that the strength of the reflected (isentropic) pressure wave from
a solid wall is the same as the strength of the incident wave, i.e., show:
()reflected = ()incident
()reflected = ()incident
Solution:
We consider two cases of expansion and compression waves, as shown in Fig. 5.20.
The proof of the first part, namely flow turning of the incident wave from a solid wall
is canceled by the reflected wave is imposed at the solid surface. Mathematically, we
state that:
1 = 3 = w
(5.68)
2 = 1
(5.69)
243
In PrandtlMeyer flow theory, we have shown that the change in PM angle is
equal to the magnitude of the flow turning angle. The expansion waves will cause
flow acceleration and thus PM angle increases by the absolute value of the turning
angle. In this context, the LRW or RRW both cause an increase in PM angle for an
expansion wave. The compression waves will cause flow deceleration and thus PM
angle decreases upon encountering of such waves. This may be stated mathematically
for expansion waves as:
2 = 1 +  Expansion waves
(5.70a)
3 = 2 +  Expansion waves
(5.70b)
(5.71a)
3 = 2  Compression waves
(5.71b)
Therefore, the change in PM angle across the incident expansion wave and its
reflection is:
(5.72a)
2 1 == ()incident = 
3 2 == ()reflected = 
(5.72b)
The change in PM angle across the incident and reflected compression waves is:
2 1 == ()incident = 
(5.73a)
3 2 == ()reflected = 
(5.73b)
The Eqs. (5.72) and (5.73) prove the second assertion that ()reflected = ()incident .
The case of wave reflection from a free boundary is of interest when we encounter
underexpanded (or overexpanded) nozzles. The problem statement is that a supersonic stream at Mach, Me , and static pressure, pe , that is discharged in an atmosphere
with static pressure, pa , where pe = pa has to adjust its pressure to match the ambient pressure at the free boundary separating the jet from the ambient. We know that
the interface between the two fluids, i.e., the free shear layer, cannot sustain a static
pressure jump. The case of pe > pa is called underexpanded flow and the opposite is true for overexpanded flows. In underexpanded flows, there is an excess of
static pressure at the nozzle exit that needs to be (abruptly) relieved when it meets
the ambient fluid. The abrupt static pressure relief mechanism in supersonic flow is
through centered expansion fans, which are known as PM expansion Mach waves.
The case of overexpanded flows where the nozzle exit pressure is lower than the
ambient static pressure, i.e., subatmospheric, the mechanism for abrupt rise in static
244
5 Method of Characteristics
pressure in supersonic flow is through shock waves. Figure 5.21 shows a definition
sketch of an underexpanded jet and its wave train in the exhaust plume. Representing
the centered expansion fan at the nozzle lip by only two (finite) expansion waves,
we arrive at field numbers 112. The jet centerline is the plane of symmetry, which
is a streamline and may thus be replaced by a solid wall. The reflection of expansion
waves from the centerline (i.e., a solid surface) is in likemanner; therefore further
expansions below atmospheric occur upon reflection. The jet in field number 3 is
perfectly expanded, which means that p3 = pa thereby maintaining the constant
pressure condition at the free boundary. To compensate for over expansion of the
flow in fields number 4 and 5, the expansion waves reflect as pressure waves from
the free boundary to create fields number 7 and 8. This is the principle of unlike
reflection at a free boundary. Compression waves that are generated at the boundary
reflect from the centerline (that behaves like a solid wall) as compression waves and
thus create fields with static pressure that are in excess to ambient pressure (in fields
number 10 and 11). The process of unlike wave reflections from the free boundary
and like reflections from the centerline repeats to form a periodic exhaust plume
shape with a spatial wavelength of , as depicted in Fig. 5.21.
The periodicity of the wave structures is noted in Fig. 5.21 by repeating cell
numbers 2, 3, 4 and 5 towards the end of the plume, where in theory the flowfield
is identical to their counterparts near the nozzle lip. The case of wave pattern in
an overexpanded nozzle is shown in Fig. 5.22. The waves that are formed at the
nozzle lip are shock waves (here are drawn as oblique shocks) and the subsequent jet
___
Streamline
pa
ConstantPressure
Free Boundary
pa
pa
pe > p a
, R
y
8
11
6
4
9
5
2
12
10
Centerline
Wavelength,
pe = p1
pa
pa
Me
p2
p11
p 3 = pa
p5 < pa
p7 = pa
p8 < pa
p12 = pe
p9 = pa
Fig. 5.21 The wave pattern in the exhaust plume of an underexpanded jet and the pressure distribution along a representative streamline (not to scale)
Oblique Shock
Free Shear layer
Expansion Waves
Compression Waves
Streamline
pa
pa
___
Streamline
ConstantPressure
Free Boundary
pa
pa
Me
pa
pa
pa
pa
pe < p a
, R
245
pa
pa
pa
y
Centerline
Wavelength,
pa
pa
pa
pe < pa
Fig. 5.22 The exhaust plume of an overexpanded nozzle with a representative Streamline and its
pressure distribution along the jet (not to scale)
contraction. It is through the process of like and unlike reflections from the centerline
and the free boundary respectively that gives the exhaust plume a periodic structure.
We recall that the strength of the shock wave at the lip depends on the severity of overexpansion. It is even possible to have a normal shock inside the nozzle of severely
overexpanded flows. Figure 5.22 shows the case of an overexpanded nozzle where
the pressure rise across the oblique shock at lip is sufficient to reach the ambient
static pressure, pa .
Example 5.4 Calculate the initial exhaust plume angle for an underexpanded jet
with Me = 2.2, pe / pa = 2.0 with the ratio of specific heats, = 1.4.
Solution:
Since the centered expansion waves are isentropic, stagnation pressure remains constant, i.e., pt2 = pt1 . From isentropic tables we get:
Me = M1 = 2.2
pte / pe = 10.693
pt2
pte
pte pe
=
=
= (10.693) (20) 21.38
p2
pa
pe pa
From the ratio of pt2 / p2 and isentropic table we get M2 2.64. The flow turning
angle across PM waves is = , therefore from 1 = 31.73 and 2 = 42.31,
we get 2 42.31 31.73 10.58 .
246
5 Method of Characteristics
2
2
M2
1
1
Me = M 1
ConstantPressure
Free Boundary
D
7
yA=1
F
5
Me= 1.435
pe= 1.347pa
= 1.4
C
6
10
11
12
Solution:
The wavefield approach in MOC is best suited for supersonic problems involving a
free boundary. The flow parameters at the exit are:
Me = 1.435
e = 10 , e = 44.176 and
pte
3.344
pe
Since the expansion process at the nozzle lip is isentropic, the total pressure at the
nozzle exit remains constant along the jet, therefore
pt3
pte pe pa
=
= 3.344 1.347 1 = 4.504
p3
pe pa p3
M3 = 1.639
3 = 16
The flow turning angle across a centered expansion fan is equal to the change of PM
angle, across the fan, namely
3 1 = 3 1 = 6
Assuming that the jet axis coincides with the xaxis, we use the flow angle at the
nozzle exit equal to zero (i.e., as reference), and thus the flow angle in field number
247
Field number
(deg)
(deg)
(deg)
1
2
3
4
5
6
7
8
9
10
11
12
0
3
6
0
3
0
0
3
6
0
3
0
10
13
16
16
19
22
16
19
16
16
13
10
44.176
40.595
37.620
37.620
35.061
32.828
37.620
35.061
37.620
37.620
40.595
44.176
248
5 Method of Characteristics
C
C
C+
C+
C
C+
C+
C
C
C+
C
C
C+
C+
1.5
4.5
1.5
4.5
1.5
1.5
1.5
1.5
1.5
1.5
4.5
1.5
4.5
1.5
42.39
39.11
39.11
36.34
36.34
33.94
36.34
36.34
33.94
36.34
36.34
39.11
39.11
42.39
43.89
43.61
40.61
40.84
37.84
35.44
34.84
37.84
32.44
34.84
31.84
37.61
34.61
40.89
( ) (deg)
Slope of waves
40.89
34.61
37.61
31.84
34.84
32.44
37.84
34.84
35.44
37.84
40.84
40.61
43.61
43.89
0.8658
0.6900
0.8573
0.8644
0.6961
0.7118
0.6961
0.6961
0.7118
0.6961
0.8644
0.8573
0.6900
0.8658
we attribute an average and that correspond to the two neighboring fields to the
wave. We use these average angles in the construction of the characteristic network.
With the known slopes and the starting coordinate, e.g., A, we start our march
in the streamwise direction by writing the equations of straight lines (to represent
waves). For example, we start with the wave AB.
yAB = 1 + 0.86578x
xb = 1/0.86578 = 1.155
With the coordinate of point B known in the previous step, we write the equations
for the waves AC and BC.
yAC = 1 0.69005x
y BC = 0.857326 (x 1.155)
From the intersection of AC and BC, we get the coordinates of point C, namely
xC = 1.2862 and yC = 0.11246. We continue with the remaining waves and their
intersections to establish the coordinates of the nodes.
yCD = 0.11246 + 0.8644 (x 1.2862)
yAD = 1 + tan(6 )x 1 + 0.105x
From the intersection of AD and CD, we get the coordinates of point D, namely,
x D = 2.6327 and y D = 1.2764.
yCE = 0.11246 + 0.69606 (x 1.2862)
249
250
5 Method of Characteristics
A
B
C
D
E
F
G
H
I
J
K
0.000
1.155
1.286
2.633
1.448
2.940
3.248
4.433
4.594
5.881
4.726
1.000
0.000
0.112
1.276
0.000
1.062
1.276
0.000
0.112
1.000
0.000
1.000
0.000
0.112
1.276
0.000
1.062
1.276
0.000
0.112
1.000
0.000
Compression Waves
1.5
1.0
Node
+6
D
F
o
J
10
11
0.0
6
0.5
Expansion Waves
12
K
0.5
1.0
1.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
x
Wavelength of the periodic (inviscid) plume structure
Fig. 5.23 The exhaust plume of an underexpanded supersonic jet discharging into a stationary
ambient using wavefield approach in MOC [Uniform exit flow with Me = 1.435, pe = 1.347 pa
and = 1.4 = constant]
Wavefield approach is more suitable for plane flows where wave strength remains
constant, even after interaction with other waves. Also the wave method is advantageous when there is a free boundary to be determined, as in underexpanded nozzle
flows. But in 3D or axisymmetric flows, wave strength varies along the wave and
the method is less useful. The latticepoint MOC is readily applied to 2D as well as
3D flowfields.
251
(5.80)
u
v
+
x
r
+u
+v
+
=0
x
r
r
Axisymmetric
Shock
x = constant planes
tan = dr/dx
r = y2 + z 2
r
r
M >1
(5.81)
nose
Axisymmetric
Body
Fig. 5.24 A pointed axisymmetric body at zeroangleofattack and yaw in supersonic flow
252
5 Method of Characteristics
(5.82)
V2
d
2
u 2 + v2
= d
2
(5.83)
The pressure term in Eq. (5.82) may be related to the square of the speed of sound
and density perturbation since the flowfield is isentropic, namely
d p = a 2 d
(5.84)
a
2
a
(5.85)
In partial differential form, Eq. (5.85) may be written as two equations, namely
1
1
= 2
x
a
1
1
= 2
r
a
u
v
u
+v
x
x
u
v
u
+v
r
r
(5.86a)
(5.86b)
We may now eliminate the density form the continuity Eq. (5.81), to get:
v v
u
u
+
+ 2
x
dr
r
a
u
u
v
+v
x
x
v
a2
u
v
u
+v
=0
r
r
(5.87)
(5.88)
Irrotational flow demands that the curl of the velocity field to vanish, namely:
e re e
1 r x
V = r x = 0
r
v rw u
(5.89)
e
253
v
u
x
r
=0
u
v
=
x
r
(5.90)
(5.91)
The remaining two complimentary equations in du and dv are written similar to the
2D case as:
du =
u
u
u
v
dx +
dr =
dx +
dr
x
r
x
x
(5.92)
v
v
dx +
dr
x
r
(5.93)
dv =
Now, we have formed the system of three Eqs. (5.91)(5.93) and three unknowns
(u/x, v/x and v/r ) in cylindrical polar coordinates. We solve for v/x
using Cramers rule:
(1 u 2 /a 2 ) v/r (1 v2 /a 2 )
dx
du
0
0
dv
dr
N
v
=
=
2
2
2
2
2
x
D
(1 u /a ) 2uv/a (1 v /a )
dx
dr
0
0
dx
dr
(5.94)
By setting the denominator equal to zero, we get the equations for the characteristics,
namely:
(1 u 2 /a 2 )dr 2 dx
2uv/a 2 dr 1 v2 /a 2 dx = 0
(5.95)
Note that Eq. (5.95) is identical to (5.15) for 2D flow except coordinate y is
replaced with coordinate r , therefore the slope of the characteristics in 3D axisymmetric flow which is dr/dx follows (Eq. (5.22)):
dr
dx
= tan( )
(5.96)
characteristic
As expected, the characteristics are Mach waves that lie above and below the local
velocity vector. Since we are dealing with 3D flows, the two characteristics are
indeed the projections of the Mach cone in the meridian plane. The C characteristic
makes an angle and the C+ characteristic makes an angle + with respect
to the xaxis in the meridian plane.
254
5 Method of Characteristics
By setting the numerator, N , in Eq. (5.94), equal to zero, we get the compatibility
conditions along C and C+ characteristics in 3D axisymmetric flows.
1
dv
=
du
1
u2
a2
v
a2
dr
dx
v dr
r du 2
characteristic
1 av 2
(5.97)
By substituting for the characteristics slope, dr/dx, in Eq. (5.97), and replacing velocity components u and v by the velocity magnitude and flow angle according to:
u = V cos and v = V sin
(5.98)
d = M 2 1
2
V
M 1 cot r
(5.99)
Since the first term in Eq. (5.99) is the differential of the PrandtlMeyer angle, d, we
can combine it with d on the LHS to arrive at the simplified compatibility equation
along C and C+ characteristics in axisymmetric irrotational flows as:
d( + ) =
dr
1
Along C characteristic
M 2 1 cot r
d( ) =
dr
1
Along C+ characteristic
M 2 1 + cot r
(5.100)
(5.101)
Note that the compatibility conditions in 3D axisymmetric flows is no longer the
algebraic equations = constant that we derived in 2D irrotational flows.
The compatibility equations are now differential equations that have to be solved
numerically. By taking small steps in r in an x = constant plane, we replace the
differential of r by r and the corresponding differential of , by ( )
and create finite differences connecting two points along the characteristics network.
Here we show an example of axisymmetric irrotational flow that is solved by the
MOC.
Example 5.6 Apply the MOC to a cone with 30 halfvertex angle that is matched
with a circular arc ogive (from [3]) at point 1 (i.e., x = 6), as shown in Fig. 5.25.
The freestream Mach number is 3.07.
Solution:
This graphicalnumerical solution is worked out by Ferri [3]. The conical shock angle
at the nose is established by the TaylorMaccoll equation [9]. It makes an angle of
255
y
45
37
36
8
7
6
5
4
2
4
3
2
1
s
47
48
38
30
31
17 29
16a 22
24 25
26
16
18 19
20
12
13 14
11
7
15
8 9
10
6
3
5
1
39
40
32
33
27
21
= 30o
46
9 10 11 12 13 14 15 16
Fig. 5.25 MOC applied to a sharpnosed axisymmetric body of revolution in Mach 3.07 (Adapted
from [3])
39.42 with respect to the xaxis. The C+ characteristics at the nose form a simple
wave region and are Mach cones with different vertex angles . Corresponding to
each is a flow angle, that is established by TaylorMaccoll equation. The angular
space between the cone and the conical shock is divided into 5 incremental steps. The
start of the calculation is at point 1 (on the coneogive) where the C+ characteristic
from the cone intersects the 6 characteristics (compression Mach waves) emanating
from the nose at points 2, 4, 7, 11 and 16.
The table of values produced by Ferri is shown in its original form in Table 5.8.
The fractional notation for angle is expressed in minutes (where there are 60 min in
1 ). Also note that the ycoordinate is the same as r in our notation and W is the
nondimensional velocity, which is the ratio of local velocity V , to the maximum
velocity, Vmax (that we had defined in Chap. 3). Krasnov [4] is recommended for
further reading on aerodynamics of bodies of revolution.
Table 5.8 The MOC
calculations corresponding to
axisymmetric coneogive
example by Ferri [3]
Point
1
2
4
7
11
16
30
32
34
36
38
39 25
30
28
26 15
24 40
23
21
0.6400
0.6414
0.6438
0.6477
0.6516
0.6560
32 28
32 20
32 6
31 44
31 22
30 58
6.000
6.220
6.509
6.876
7.390
7.902
3.462
3.880
4.380
4.982
5.753
6.473
256
5 Method of Characteristics
M=2.91
M=2.37
600
598 596 594
382 382
382 382
596
398 594 592 584
394 392 384
598
596 592 590 588 586
398
396 394 390 388 386
600
400
598
400
592
382
590
382 588
382 586
382
584
382
582
382
(b)
580
380
578
378
576
376
576
378
576
380
576
382
578
580
382
582
382
582 384
582 386
576 576
576 576 584
594 594
400 398 390 582 388
386 384
576
390
400 398
388
576
576 576
594
396 576
392 390
396
394
(a)
Fig. 5.26 Wavefield approach is used in the design of a Mach 2.37 and Mach 2.91 isentropic
nozzles (from [3])
(a)
257
582
390
582
382
590
390
582
390
582
382
Horizontal Axis
Streamline
582
390
590
390
(b)
582
390
598
390
598
390
Streamline
Horizontal Axis
598
390
598
390
598
390
Fig. 5.27 Wavefield method applied to a nozzle with its exit cross section inclined with respect to
the jet axis (from [3]) a Expansion b Compression
from the horizontal axis as the exhaust plume evolves. We note that the asymmetry
in the exit plane of supersonic nozzle, as shown in Fig. 5.27, presents a case for thrust
vectoring, albeit of fixed and weak form.
r = const.
C+
Virtual Source
r1
CC
=0
C+
C
3 7
2 6
1 5
M=const.
C+
C+
9
M>1
4 8
+
C+
CC
r2
r3
r4
Fig. 5.28 Supersonic flow in a straight 2D diverging duct with a few representative C characteristics
258
5 Method of Characteristics
1+
1+
1 2
2 M2
1 2
2 M1
+1
2(1)
For a given wall angle in region 2, 2 , we calculate the flow Mach number in
region 2 by using the PM expansion theory (presented in Chap. 4, shockexpansion
theory). The lengths of the C+ characteristics along the head and tail Mach waves
that are cut by the streamline are l1 and l2 respectively. Since these lengths are the
hypotenuse of a right triangle with respective heights h 1 and h 2 , they are related
according to:
l1 = h 1 / sin 1
(5.103a)
l2 = h 2 / sin 2
(5.103b)
259
1+
1+
1 2
2 M2
1 2
2 M1
+1
2(1)
1+
1+
1 2
2 M2
1 2
2 M1
+1
2(1)
(5.104)
We have established the orientation of line BF, i.e., the tail characteristic, as well as its
length, l2 . For intermediate C+ characteristics, e.g., BE, we choose an intermediate
Mach number between M1 and M2 , which immediately produces its corresponding Mach and PM angles. The corresponding flow turning angle is the difference
between the two PM angles (i.e., the difference between the intermediate PM angle
and the upstream value). The length BE is related to the length BD according to
Eq. (5.104) where M2 is replaced by the Mach number that we chose for the intermediate C+ characteristic. In the Cartesian coordinates attached to point B in Fig. 5.29,
we establish the coordinates of the bending streamline, DEF by noting that
x D = 1 cos 1 and y D = 1 sin 1
(5.105a)
(5.105b)
(5.105c)
Example 5.7 A parallel uniform flow at Mach 1.6 approaches a sharp corner that
turns the flow in the clockwise direction by 21.89 , similar to the schematic drawing
of Fig. 5.29. Calculate the coordinates of the bending streamline with h 1 = 1 and
graph the streamline. Assume that the flow is planar 2D with = 1.4.
C+ characteristics or (leftrunning) Mach waves
Streamline
M1
1
h1
1 =0
l1
M2
l2
(2
2)
x
A
2
Streamline
h2
260
5 Method of Characteristics
M2 = 1.8
M1 = 1.6
M=1.6
C+1
=0
C+2
M3 = 2.0
C+3
M4 = 2.2
C+4
M5 = 2.4
C+5
M=2.4
=21.89
Fig. 5.30 Definition sketch for the five C+ characteristics in the expansion corner
Solution:
The first C+ characteristic is the head Mach wave that makes Mach angle with respect
to local upstream flow:
1 = sin
1
M1
= sin
1
1.6
= 38.68
The PM angle corresponding to the incoming flow is 1 = 14.86 . The PM angle
after 21.89 of wall turning is the sum of the upstream PM angle and the turning
angle, i.e., = 36.75 . The corresponding Mach number is M = 2.4. We divide
the centered expansion fan into finite steps in Mach number, e.g., steps of 0.2. This
choice produces 5 C+ characteristics as shown in Fig. 5.30.
Since the 5 C+ characteristics are chosen based on Mach number, we immediately
associate a Mach angle and PM angle with each characteristic. The flow angle, ,
starts from zero (based on our choice of the coordinate system) and the steps in
are the difference between the local PM angle and the incoming PM angle. The
length ratio along different characteristics associated with an intersecting streamline
is calculated from Eq. (5.104). The coordinates for the bending streamline follow
Eq. (5.105). These are summarized in table form from Excel calculations (Tables 5.9,
5.10 and 5.11).
The coordinates of the wall are summarized in table form:
The coordinates of the streamline are also summarized in table form:
The graph of the expansion corner and the corresponding streamline is shown in
Fig. 5.31.
We may now select a second streamline with a different h 1 value than the
Example 5.7, say h 1 = 2.0 The second streamline bends similar to the first and
the two streamlines, as an example, form the boundaries of a bending duct that
accelerate a Mach 1.6 flow to the exit Mach number of 2.4.
261
y
Streamline
1.5
C+1
Wall
Streamline
C+2
C+3
0.5
C+4
C+5
0
2
1
0.5 0
1
Wall
1.5
2
Fig. 5.31 Expansion corner of Example 5.7 and the graph of a representative streamline ( = 1.4)
2
2.5
2
1.5
1
0.5
0
0.5 0
1
1.5
2
10
Fig. 5.32 The expansion corner of Example 5.7 and the graph of the second streamline ( = 1.4)
Example 5.8 Calculate and graph a second streamline for the expansion corner of
Example 5.7 with h 1 = 2.0.
Solution:
We follow the same procedure that we outlined in Example 5.7 and produce the
following tables in a spreadsheet calculation. Note that the flow characteristics, M,
, and remain unchanged on the second streamline (Tables 5.12 and 5.13).
Figure 5.32 shows the second streamline, with h 1 = 2. We may use the two
streamlines as the boundaries of a curved duct and is shown in Fig. 5.33.
The curved duct of Example 5.8 is a supersonic nozzle with an exit flow angle
of 21.89 with respect to the entrance flow direction. By incorporating a second
expansion fan of opposite family, it is possible to expand the flow and accomplish
Table 5.9 Characteristics in
an expansion corner
M
(deg)
(deg)
(deg)
L/L 1
x
C+1
C+2
C+3
C+4
C+5
1.6
14.86
38.68
0.00
1.00
1.25
1.8
20.73
33.75
5.86
1.29
1.83
2
26.38
30.00
11.52
1.69
2.56
2.2
31.73
27.04
16.87
2.21
3.47
2.4
36.75
24.62
21.89
2.88
4.61
262
5 Method of Characteristics
x
y
x
y
Streamline 2
Streamline 1
2
2.5
2
1.5
1
0.5
0
0.5
1
1.5
2
2
0
2.00
1.00
1.25
1.00
0
0
1.83
0.97
4
1.606869
2.56
0.86
3.47
0.62
4.61
0.22
y
Curved Duct
10
Fig. 5.33 The shape of a curved duct that accelerates a uniform, parallel Mach 1.6 flow to a uniform
and parallel Mach 2.4 flow ( = 1.4)
an axial exit flow. Figure 5.34 shows a schematic drawing of an Sduct supersonic
nozzle. The inlet flow is uniform and parallel, in field #1 and is in the xdirection.
Field #2 is a continuous (expansive) turning of the flow that is associated with a
centered expansion fan, which creates a nonuniform and nonparallel flowfield. The
flow in field #3 is again uniform and parallel, i.e., nonaccelerating or decelerating.
Field #4 is a continuous expansion process associated with a centered expansion fan
of opposite family. Therefore, the turning of the flow is in the counterclockwise
direction. Field #5 is the uniform parallel exit flow, which is in the axial direction.
Table 5.12 Characteristic
values and streamline
coordinates
M
(deg)
(deg)
(deg)
L/L 1
x
y
x
y
2.00
2.00
C+1
C+2
C+3
C+4
C+5
1.6
14.86
38.68
0.00
1.00
2.50
2.00
1.8
20.73
33.75
5.86
1.29
3.66
1.94
2
26.38
30.00
11.52
1.69
5.12
1.71
2.2
31.73
27.04
16.87
2.21
6.95
1.25
2.4
36.75
24.62
21.89
2.88
9.22
0.44
6.95
1.25
9.22
0.44
2.50
2.00
3.66
1.94
5.12
1.71
263
y
x
2
3
264
5 Method of Characteristics
(a)
(b)
G
MC
C
C+
A
M1 > 1
A
Reflected Ccharacteristic
(neglected)
C
B
A
Fig. 5.35 The waves on a slender body in supersonic flow a Interaction of Mach waves and an
oblique shock (simplified view), b expanded view of the flow and shock angles
Example 5.9 A 2D pointed body at zero angle of attack is in a Mach 2 flow, as
shown. Its leading edge angle on its upper surface is A = 10 . Point B on the body
has an angle of B = 8 . Calculate:
(a)
(b)
(c)
(d)
(e)
C
M1 = 2.0
=8
C+
A
B
=8
= 10
Solution:
With M1 = 2 and the nose angle of 10 , we use the oblique shock charts (Fig. 4.13)
to get the wave angle, A 39.5 . Since the flow angle downstream of the shock at
C is estimated to be 8 , i.e., the same as the body at B, we use the oblique Shock
charts to get c 37.2 .
We march across the O.S. at the nose, by using the following relations:
M1n = M1 sin A = 2 sin 39.5 1.27
M2n 0.80 from NS table
M2 = M A =
265
M2n
1.63
sin( A A )
5.7 Summary
Method of characteristics (MOC) is a numericalgraphical technique that allows
supersonic flowfields (including upper transonic flows) with large perturbations to
be analyzed. We showed that two real characteristics exist in supersonic flows and
we demonstrated that these characteristics are indeed local rightrunning and leftrunning Mach waves. With unique compatibility conditions that hold along the characteristics, we march in the flow direction and establish flow properties along a
characteristic network. We introduced two equivalent approaches to MOC, namely
wavefield and latticepoint methods where the former approach was more suitable
for problems with free boundaries than the latter method. Although our analysis was
based on potential flows, the rotational flows, e.g., with stronglycurved shocks, may
also be analyzed with a suitable MOC that include flow vorticity (e.g., see [7] or [3]).
For thin bodies at small angle of attack, potential flow characteristics may be used
to establish the shape of the curved shock.
In the applications section, we learned that supersonic nozzle/wind tunnel contours
may be designed using the MOC. The wavefield method captures the flowfield of
under or overexpanded exhaust plumes. We learned to follow a streamline in supersonic flows and form possible supersonic passages, e.g., an Sshaped nozzle. These
applications may be extended to include turbomachinery blade design in supersonic
flows. Additional original contributions to MOC may be found in PrandtlBusemann
[6] and ShapiroEdelman [8].
Problems
5.1 A C+ characteristic intersects the wall of a supersonic nozzle, as shown. The
wall angle is w = +15 . Calculate the Mach number at 2 (on the wall) and the slope
of the C+ characteristic, i.e., dy/dx, between points 1 and 2.
266
5 Method of Characteristics
w=
C
C+
+15o
M1 =2.0
o
1 =+10
5.2 In a steady, 2D, irrotational, isentropic supersonic nozzle flow, we have the
flow Mach number and direction at point 1, as shown. The nozzle wall is straight and
makes an angle, w = 9.5 with respect to xaxis. Calculate the flow Mach number
on the nozzle wall that corresponds to the C characteristic that passes through point
1 (i.e., point 2, M2 ). Also calculate the slope of the C characteristic, dy/dx, between
points 1 and 2.
M1 =2.2
o
1 = 4
C
x
2
w
5.3 Based on the initial data line in a diverging section of a 2D supersonic nozzle,
as shown, M1 = M3 = 1.6, 1 = 3 = 15 , M2 = 1.8, 2 = 0 . Use Method
of Characteristics (MOC) for 2D irrotational flow to calculate:
(a) M5 and 5
(b) M7 and 7
8
y
3
x2 5
7
1 4
6
5.7 Summary
267
M = 3.2
2
2
M = 3.2
M2
2 = 12o
C+ characteristic
y
x
1
M1 = 2.2
1 = 9o
C+ C+
y
C
C+
C+
C
Me=2.8
x
The last C+
characteristic
Me=1.2
2
1
Calculate:
(a) K 1 in degrees
(b) K +3 in degrees
x
Centerline
268
5 Method of Characteristics
Nodes
2
M
(deg)
2.0
0.0
1.8
4.0
1.5
8.0
y
x
10
12
13
3
2
1
5 7
4
9
6
11
Centerline
Calculate:
(a) Flow angle and Mach number at node 5, i.e., 5 and M5
(b) Wall angle connecting nodes 3 and 8 on the wall, w 38
(c) Mach number, M11
5.8 We wish to design an isentropic minimumlength, twodimensional, supersonic
nozzle for air ( = 1.4) with uniform inlet flow condition at the choked throat. The
exit Mach number is to be Me = 1.8, as shown in Fig. 5.36. We use the Method of
Characteristics, MOC, to design the nozzle and choose the initial turning angle of
initial = 0.365 and five subsequent turns of 2 each from the sonic throat with
uniform parallel flow, as shown in the schematic drawing.
Calculate:
(a) The maximum wall angle, w, max
(b) The flow angle, 7
(c) The flow angle, 13
5.9 A 2D minimumlength nozzle is designed using MOC. There are 5 C characteristics at the throat and each turn the flow equally by 2 (Fig. 5.37). Calculate:
(a)
(b)
(c)
(d)
5.7 Summary
269
22
18
13
7
th
M e=1.8
A
w,max
12
Centerline
1
14 19
23
26
3
e
Me
1
y
B
Centerline
x
Me = 2.06
pe=1.621 p
a
pa
5.10 Calculate the initial exhaust plume angle for an underexpanded jet with Me =
2.6, pe / pa = 3.0 with the ratio of specific heats, = 1.4. Assuming the flow in
the nozzle was isentropic; calculate the nozzle pressure ratio, NPR that is defined as
pt0 / pa where pt0 is the total pressure at the nozzle entrance.
5.11 A symmetrical twodimensional underexpanded nozzle has an exit Mach number of Me = 2.06 and exit static pressure that is 162.1 % of the ambient pressure,
i.e., pe = 1.621 pa (see Fig. 5.38). Assuming that the nozzle exit flow is uniform and
the gas specific heat ratio is = 1.4 = constant, use MOC (wavefield method) to
map out the (inviscid) flowfield in the exhaust plume and establish the jet boundary.
What is the wavelength of the periodic plume structure, as compared to nozzle exit
halfwidth?
270
5 Method of Characteristics
5.12 A symmetrical twodimensional overexpanded nozzle has an exit Mach number of Me = 2.0 and exit static pressure that is 50 % of the ambient pressure, i.e.,
pe = 0.5 pa . Assuming that the nozzle exit flow is uniform and parallel with the gas
specific heat ratio of = 1.4 = constant, calculate:
(a)
(b)
(c)
(d)
(e)
Me= 2.0
pe=0.5 pa
B
A
C
Inviscid constant  pressure jet boundary
5.13 The initial data line in a 2D straight diverging duct is a circular arc with
radius of curvature R, where R = 1 m. The flow Mach number is M = 2.0 =
constant in magnitude on the initial data line. The straight, radial diverging duct has
its two sidewalls symmetrical with respect to the axis of the duct and their angles are
w = 15 , as shown. There are five points on the initial data line. Point number 1
is on the lower wall. Point number 3 is on the centerline. Point number 5 is on the
upper wall. Points 2 and 4 are halfway between 1 and 3 and 3 and 5 respectively.
Fill out the table below for the coordinates and flow parameters at the 5 initial data
points.
R =1m
w=
Parameters
+15
w=
15
1 2 3 4 5
x (m)
y (m)
(deg)
(deg)
(deg)
5.14 A parallel uniform flow at Mach 1.4 approaches a sharp corner that turns the
flow in the clockwise direction. The flow downstream of the corner reaches Mach
number of 2.8. Calculate the wall turning angle and use the wavefield method to
5.7 Summary
271
calculate the coordinates of the bending streamline with h 1 = 1 (see Fig. 5.29 for
definition sketch) and graph the streamline. Assume that the flow is planar 2D with
= 1.4.
5.15 Apply wavefield method to the incidentreflected expansion wave problem
from a solid wall, as shown. The Mach number in field #1 is 2.0 and the flow angle is
0 . The incident pressure wave turns the flow 2 in the clockwise direction. Assuming
= 1.4, calculate:
(a)
(b)
(c)
(d)
Reflected pressure
wave (expansion)
LRW
RRW
y
x
Solid Wall
y
x
C+
C+
3
C
B
5.17 A parallel uniform flow at Mach 2.0 approaches a sharp corner that turns
the flow in the clockwise direction by 10.37 . First calculate the flow Mach number
downstream of the expansion corner. Second, calculate the coordinates of the bending
streamline with h 1 = 3.0 and graph the streamline. Assume that the flow is planar
2D with = 1.4.
5.18 To accelerate a Mach 2.0 stream to Mach 2.4 isentropically, an expansion fan
that turns the flow by 10.37 will be formed, as shown schematically by lines AC and
272
5 Method of Characteristics
AE. This is the same problem as 5.16 with a single streamline that passes through
h 1 = 3.0. Now, construct a second streamline with h 1 = 4.0, which along with
the first streamline may form the walls of a supersonic nozzle. Calculate and graph
a second streamline with h 1 = 4.0. What is the area ratio, A2 /A1 , of this nozzle?
Assume flow is planar 2D and = 1.4.
h
4.0 M =2.0
1
3.0
M 2=2.4
5.19 A 2D Sshaped supersonic nozzle is designed based on two centered expansion
fans of opposite family, ABC and DEF, following the schematic drawing shown. The
flow at the entrance to the Sshaped nozzle is parallel and uniform in the xdirection
with M1 = 2.0. The nozzle is divided into three fields. The field #2 represents the
flow that has emerged from the first bend. It has attained a flow angle of 2 = 20 .
The field #3 represents the flow that has emerged from the second bend and is purely
axial, i.e., 3 = 0 . Calculate:
(a) The Mach number in field #2, M2
(b) The Mach number in field #3, M3
(c) The nozzle exittoentrance area ratio, A3 /A1
D
x 1
2
3
A
E
F
5.20 We apply wavefield theory to the half diamond airfoil in supersonic flow, as
shown. Assuming that the flow downstream of the shock is irrotational, we may use
the irrotational MOC at the shoulder with two C+ characteristics, AB and AC. Mach
number in region 2 is M2 = 2.0 and the flow angle is 2 = +4 . The flow angles in
regions 3 and 4 are 3 = 0 and 4 = 4 respectively. Calculate:
(a) Mach numbers in regions 3 and 4
(b) The angles that C+ characteristics make with respect to the xaxis
(c) Flight Mach number, M1
5.7 Summary
273
C
B
y
3
C+
C+
2
A
4
x
5.21 Apply the latticepoint method to a curved (convex) body ABDF to estimate
the shock angles for the segments, AC, CE and EG. The body angles are known to be:
Body Angle (deg) A
B
8
12
D
6
G
F
4
E
C
1
M1=2.4
= 1.4
A
H
x
References
1. Anderson, J.: Modern Compressible Flow with Historic Perspective, 3rd edn. McGraw Hill,
New York (2003)
2. Anon: Equations, Tables, and Charts for Compressible Flow. NACA TM 1135, Moffett Field,
CA (1953)
3. Ferri, A.: Elements of Aerodynamics of Supersonic Flows. Macmillan Company, New York
(1949)
4. Krasnov, N.F.: Aerodynamics of Bodies of Revolution. Elsevier Publication, New York (1970)
5. Liepmann, H.W., Roshko, A.: Elements of Gas Dynamics. Wiley, New York (1957)
6. Prandtl, L., Busemann, A.: Nherungsverfahren zur Zeichnerischen Ermittlung von Ebenen
Strmungen mit berscallgeschwindigkeit. Stodola Festschrift p. 499 (1929)
7. Shapiro, A.H.: The Dynamics and Thermodynamics of Compressible Fluid Flow. Ronald Press,
New York (1953)
8. Shapiro, A.H., Edelman, G.M.: Method of characteristics for twodimensional supersonic flow
graphical and numerical procedure. J. Appl. Mech. 14(2), A154 (1947)
9. Taylor, G.I., Maccoll, J.W.: The air pressure on a cone moving at high speed. Proc. R. Soc. Lond.
139, 298311 (1933)
Chapter 6
Abstract This chapter presents the sources of drag that are encountered by a
nonlifting body that is moving at transonic Mach numbers. In inviscid conditions
drag can be produced if the integrated pressure over a body has a component in the
opposite direction to the flow (pressure drag). First, the reader is introduced to the
relation between the geometry of a body and the pressure distribution over the body.
Subsequently, a method for the calculation of the pressure drag over axisymmetric
bodies is presented based on the linear potential flow equation. It is shown how the
pressure drag is a function of the crosssectional area distribution of the body. The
concept of area ruling is presented and examples are shown of practical implementations of the area rule. If the flow is assumed viscous, other drag sources arise: friction
drag and drag due to boundarylayer separation. A qualitative characterization of laminar and turbulent boundary layers is presented along with the concepts of transition
and separation. Also, the interaction between shock and boundary layer (both weak
and strong) is further detailed and it is shown how this influences drag divergence in
transonic conditions. In addition, calculation methods are presented to estimate the
boundarylayer properties of laminar and turbulent boundary layers along with their
transition region for bodies subjected to an external pressure and velocity distribution.
This chapter contains 8 examples and concludes with 29 practice problems.
6.1 Introduction
There exist two types of drag: friction drag and pressure drag. Friction drag is mostly
caused by the boundary layer between the body and the outer flow and is a function of
the viscosity of the fluid. A nonlifting body in transonic flow can experience pressure
drag through two mechanisms. One is due to a wake that is formed when the flow
streamlines separate from the body surface (drag due to flow stagnation and separated
flow). The second cause is through the formation of shock waves (wave drag). Wave
drag is an aerodynamic phenomenon which is unique to supersonic flow. This type
of drag is associated with the energy which is radiated away from the vehicle in
the form of pressure waves. In the present chapter we focus on these three causes
for drag. The aim of the chapter is to explain to the reader how a nonlifting body
produces drag and what parameters are of influence.
Springer Science+Business Media Dordrecht 2015
R. Vos and S. Farokhi, Introduction to Transonic Aerodynamics,
Fluid Mechanics and Its Applications 110, DOI 10.1007/9789401797474_6
275
276
control surface, S
u1
C
B
F
body
F
body
D
2D control volume
control surface,
Fig. 6.1 Body force generated by pressure and shear stresses on the body and their relation to the
velocity deficit in the wake
In Fig. 6.1 a body is shown in a control volume. It is assumed that the upper and
lower boundary of the volume are formed by streamlines of the flow. The pressure
and shear forces that act on the body result in a total body force, F body . By Newtons
third law, the body exerts an equal and opposite force on the fluid, F body . Let us
investigate the effect of this body force on the momentum in the flow. We start from
Newtons second law:
d
(2.114)
F = (mV )
dt
Remember that the righthand side (RHS) of (2.114) physically represents the time
rate of change in momentum that is moving through the control volume. Its representation has been given in the LHS of (2.121):
F=
V dV +
(V dS)V
t
V
(6.1)
S+
The integral over the second control surface, is always zero (regardless of control
volume shape) due to the fact that the inner product of velocity and surface area
vanishes on the body surface. If we neglect the small body forces on the fluid ( f = 0),
the external force F can be broken down into two parts: the reaction force from the
body, F body , and the force exerted across the outer boundary:
F=
pdS F body
(6.2)
S
(V dS)V
pdS
S
(6.3)
6.1 Introduction
277
In (6.3), the force exerted on the body is coupled to properties of the flow inside
the control volume. We would like to determine the xcomponent of the body force
from the parallel velocity components at the beginning of the control volume, u 1 , and
at the end of the control volume, u 2 . We write the xcomponent of (6.3) as follows:
Fbodyx =
(V dS)u
pdSx
(6.4)
First we notice that when the boundaries of the control volume are chosen at an
infinite distance from the body the pressure, p, should be constant on this boundary.
For a constant pressure, the integral yields:
pdSx = 0
(6.5)
Since the upper and lower surface of the control volume are streamlines of the flow,
the velocity vector and the control surface are locally aligned. This means that the
inner produce V dS = 0 along those boundaries. Therefore, we can write for Fbodyx :
Fbodyx =
B
1 u 21 dy
C
2 u 22 dy
(6.6)
By employing the continuity equation in integral form (2.109) over the entrance
and exit of this control volume and subsequent substitution yields the following
expression:
C
(6.7)
Fbodyx = 2 u 2 (u 1 u 2 )dy
D
278
52
=0.723
M=0.84
CL=0.40
51
1.0
50
49
Loss in total
pressure due to
boundary layer
48
47
46
0
.2
.4
.6
.8
1.0
1.2
1.4
pT (psi)
279
(6.8)
We can derive a relation between the local pressure in the fluid and the local velocity.
We can integrate both sides of (6.8) between two arbitrary points along the streamline.
If we assume that the flow is incompressible (i.e. 1 = 2 = ) we obtain Bernoullis
equation (see Problem 6.3):
1
1
p1 + V12 = p2 + V22
2
2
(6.9)
(6.10)
The second term in (6.10) is the dynamic pressure, often denoted with q. Similar to the
static pressure, p, the dynamic pressure is a scalar. Bernoullis equation literally states
that the local velocity is inversely proportional to the square root of the local pressure.
The pressure coefficient may be written as
Cp =
p p
q
(3.11)
Note that the subscript implies that we are dealing with a point in the flow at
an infinite distance from the measuring point. This is referred to as the freestream
condition. Substituting (6.10) and assuming incompressible conditions yields the
following expression for the pressure coefficient:
Cp = 1
V
V
2
(6.11)
280
This relation directly ties the pressure coefficient to the local velocity. We can
approximate this equation by assuming that the local velocity is a relatively small
deviation from the freestream velocity: V = V + V and V /V 1. In that
case, (3.11) reduces to:
V
(6.12)
Cp = 2
V
We now have a very simple linear relationship between the supervelocity, V , and
the pressure coefficient, Cp . Next, we have to correlate the pressure coefficient to the
local curvature of the streamlines.
Perpendicular to the streamline we can derive a relation between the pressure on
a fluid particle and its velocity along a curved streamline. A schematic drawing of
such a streamline is shown in Fig. 6.3. We define the direction perpendicular to the
streamline as n, and along the streamline with s. The freebody diagram (FBD) shows
the pressure forces that are acting on this fluid particle in the ndirection, in inviscid,
irrotational flow, while the kinetic diagram (KD) shows its centripetal acceleration.
If we assume the dimensions of the fluid particle to be s in length and r in width,
then Newtons second law reduces to:
mV 2
= ( p + p p)s
r
(6.13)
p+p
n
s
m
r
m
p
FBD
m
mV
r
KD
Fig. 6.3 Forces and accelerations on a fluid particle perpendicular to a curved streamline
281
the pressure at an infinite distance normal to the body should equal the freestream
pressure, p , we can deduce that the pressure at the wall, p, should be lower than
p . In other words, a convex shape leads to a reduction in pressure and consequently
a negative pressure coefficient, Cp . A similar argument can be made for the concave
shape. However, in this case the direction normal to the body wall, is in opposite
direction to the radius vector, r . Therefore, the pressure gradient in this direction is
negative and the pressure continuously decreases when we move further away from
the wall. Consequently, we can deduce that the pressure at the wall should therefore
be higher than the pressure infinity: p > p . The pressure coefficient at the body
for a concave shape is therefore positive.
We have now qualitatively determined that streamlines along convex curves have
decrease Cp and streamlines along concave curves increase Cp . In addition, we can
see from (6.14) that the there is an inverse relation between the radius of curvature,
r , and the pressure gradient, d p/dn. From this relation we can deduce that when
r 0, the pressure gradient must tend to infinity, i.e. d p/dn . We also see that
when limr d p/dn 0 and that V V . The effect of the radius of curvature
of the body, r = R, can be evaluated on a qualitative basis. We can plot the pressure
coefficient Cp as a function of the distance from the center of curvature r , using the
aforementioned relations at the respective boundaries at r = 0 and r . This is
shown in Fig. 6.5. We can observe that the position of the wall has an effect on the
local pressure, p, at r = R. Whenever the wall is positioned close to the center of
concave wall
R
dp/dn < 0
dp/dn > 0
Cp< 0
Cp
wall
Cp > 0
concave shape
r
convex shape
p>p
p<p
p
8
p
convex wall
282
curvature (i.e. it has a large curvature itself) the pressure difference between pr = R
and p is large and we have a high Cp . Conversely, when R is large (i.e. the wall
has little curvature) the pressure difference is much smaller and Cp tends to zero.
Hence, on a purely qualitative basis we can state that the radius of curvature of the
body is negatively related to the pressure coefficient. In other words, the larger the
curvature, the higher the absolute value of the pressure coefficient.
We can combine the relation between curvature and Cp and between Cp and
V in a straightforward manner. We know that when a geometry displays a convex
curvature the pressure coefficient, Cp < 0. Consequently, employing (6.12) we note
that the supervelocity, V > 0. A convex curvature, therefore, leads to local speeds
that surpass the freestream velocity. Similarly, when the geometry has a concave
curvature, the pressure coefficient is positive and V < 0. Concave shapes therefore
reduce the speed of the fluid near the body. Finally, we can state that the absolute
curvature (whether convex or concave) has a magnification effect on V . A body
with sharp corners, for example, yields a large speed fluctuations at these corners,
while a body with little curvature only induces mild speed variations.
Now that we have established the effect of local curvature on the pressure coefficient, we take a look at the distribution of the pressure coefficient over a body. The
value of the pressure coefficient at various locations on the body is often termed the
pressure distribution. Strictly speaking this is incorrect. However, we have used the
term pressure distribution in relation to the Cp distribution and we will continue to
do so in this chapter. In subsonic flow, the pressure at any point A has an effect on the
pressure at an arbitrary point B (point B is in point As region of influence). Every
change in pressure coefficient due to the presence of curvature at point A therefore
also affects the pressure coefficient at any other arbitrary point B on that body. This
makes the intuitive prediction of the pressure coefficient somewhat more difficult.
However, we can still relate the geometric features of a given body to its pressure
distribution by relating the change in Cp to the local curvature. For example, if the
flow encounters a convex shape with a small radius of curvature we expect a sharp
increase in local suction. Conversely, if the flow encounters a concave shape with a
large radius of curvature, we expect a gradual increase in local pressure.
This is shown in Fig. 6.6 where two axissymmetric bodies of identical thicknesstolength ratio are placed in an axial flow field. The ellipsoid has a blunt leading
edge, which leads to a high local curvature. The Cp therefore increases rapidly.
Because the curvature over the ellipsoid is low over the center section of the body,
the pressure coefficient shows little variance. The absence of the blunt leading edge
on the paraboloid makes for a more gradual increase of the pressure coefficient.
However, due to the higher curvature in the center section, the pressure coefficient
reaches a higher peak.
In Fig. 6.7 we show the pressure distribution over an important part of the fuselage: the cockpit. In this case, the aircraft in question has been modified with the
addition of a large nose fairing (radome) to house a radar. The radome has a significant impact on the pressure distribution over cockpit. The pressure distribution over
the top of the fuselage shows the most erratic behavior. Initially the flow is accelerated from near stagnation pressure (Cp = 1) at the nose towards a minimum of about
283
0.08
Paraboloid
0.06
II
0.04
CP
0.02
Ellipsoid
0
0.02
0.04
0
0.2
0.4
0.6
0.8
1.0
X= l
R
I
II
Fig. 6.6 Pressure distribution over an ellipsoid and paraboloid of identical maximum thickness
and length at incompressible axial flow (after Ref. [49])
0.6
PRESENT METHOD
EXPERIMENTAL DATA
TOP
BOTTOM
0.4
0.2
0
50
100
150
200
300
350
400
CP
0.2
0.4
0.6
0.8
1.0
BASIS FUSELAGE
Fig. 6.7 Pressure distributions on a C135 fuselage with a large radome at zero angle of attack
(after Ref. [6])
284
Cp = 0.24. Then, the flow decelerates once more to a value close to stagnation at the
kink between the radome and the cockpit window. Due to the high convex curvature
over the window, the flow accelerates rapidly beyond the kink. At the bottom curve
we also see the presence of the concave kink between fuselage and radome in the
pressure distribution. Before the kink the velocity on the uncurved part of the radome
is close to its freestream value (Cp 0). At the kink the flow decelerates (Cp > 0)
after which it accelerates once more over the convex part of the lower fuselage.
Example 6.1 Below an axissymmetric body is shown. Assuming inviscid, incompressible flow, sketch the notional pressure distribution about this body.
ellipsoid
cylinder
ellipsoid
Solution:
We first identify some of the key characteristics of the body under consideration.
Since we consider inviscid conditions we have a stagnation point at the leading edge
and at the trailing edge where the pressure coefficient is 1. This body has a blunt nose,
so we expect a sharp increase in suction over the first few percent of the body. Then the
curvature decreases, and so does the flow velocity. Hence the slope dCp /dx reduces.
Once the flow reaches the cylinder it does not see any change in curvature. Therefore,
the flow decelerates towards freestream velocity. The Cp will therefore increase to
a value close to 0. Even further downstream a convex shape is present again which
accelerates the flow and creates a second suction peak. Finally, the flow decelerates
to stagnation over the last few percent of the body. Due to the symmetric shape of
the body we also expect a symmetric shape in pressure distribution. Applying this
line of reasoning results in the notional pressure distribution drawn in Fig. 6.8.
Cp
x/l
285
compressible pressure coefficient, Cp M is related to the incompressible pressure
(6.15)
Cp M =
2
1 M
In other words, the pressure coefficient increases with the freestream Mach number
up to infinity at M = 1. We know from practice that in reality this theoretical value
is not reached. However, it does show qualitatively that the pressure coefficient and
thus the supervelocities over a given body in highsubsonic flow are affected by the
compressibility of the air.
Let us clarify this qualitative statement through the relationship between the pressure coefficient and Mach number in steady isentropic flow. If we employ the relation
between Mach number and static temperature in isentropic conditions (2.147) and
the relation between pressure and temperature (2.103), we have:
1 2 1
1 2 1
= p 1 +
M
M
pt = p 1 +
2
2
(6.16)
This relation shows that under the assumption of isentropic compression there is a
relation between the local Mach number, M, and the Mach number at infinity, M ,
via the static pressure. We use the following definition of the pressure coefficient:
2
Cp =
2
M
p
1
p
(3.11)
Combining (6.16) and (3.11) results in the following relation between pressure coefficient, local Mach number, and freestream Mach number:
1 2
M
1
+
2
2
(6.17)
1
M2 =
1
1
1
2
1 + 2 M Cp
For air flow about curved bodies this equation can be used up to a local Mach number
of 1.58. This is the (theoretical) maximum supersonic speed that can be achieved over
a body in subsonic flow. It will be derived in Sect. 7.5. A graphical overview relating
the local Mach number to the pressure coefficient is shown in Fig. 6.9. It can be seen
from this graph that high negative pressure coefficients (suction) in combination
with moderate to high subsonic Mach numbers lead to a local Mach number well
beyond 1.
If we define the increase in local Mach number as M = M M then we can
also plot the relative increase in Mach number M/M . This is shown in Fig. 6.10.
If we take this graph and we compare the relative increase in Mach number for
Cp = 1 then we can see that for M = 0.1 the increase is around 42 %, while
for M = 0.8 the increase is around 62 %. In other words, there is a progressive
increase in local Mach number with increasing freestream Mach number.
286
1.6
0.8
0.7
0.6
1.2
M=
tr e
es
0.5
0.4
am
0.3
0.2
r, M
0.6
be
um
hn
ac
0.8
0.4
0.1
0.2
0
2
, =1.4
3
fre
1.4
5
10
15
Pressure Coefficient, C p
8 0.8
0.7
0.8
0.6
0.4
0.2
0.7
0.5
0.3
0.1
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.5
1.5
Pressure Coefficient, C p
We have now established that the curvature in a body is directly responsible for
the supervelocities about that body and the related pressure coefficient. In addition,
we have shown that the local Mach number is strongly affected by the pressure
coefficient and value of the freestream Mach number. This should give the reader
some indication about the relation between the geometry of the body, the freestream
Mach number and the expected pressure coefficient. In addition to the effect close to
the body, the body has a larger effect on the streamlines further away from the body
when the Mach number approaches 1. As can be seen in Fig. 6.11, at a Mach number
of 0.8, region of influence is much larger that at a Mach number that is wellbelow 1.
This is a second effect of compressibility that is important for transonic aircraft.
287
M << 1
M 0.8
As was established in Chap. 3, this effect plays an important role when wind tunnel
tests are carried out to predict the aerodynamic coefficients of the airplane.
The displacement effect caused by the body on the streamlines far away from
it continues to grow with Mach number. In supersonic conditions, the linearized
potential theory shows that the streamlines even at an infinite distance from the body
are affected. In fact, they follow exactly the same path as the wall itself. In supersonic
conditions, it is the slope of the surface with respect to the freestream velocity vector
that determines the local pressure coefficient [see (3.69)].
(6.18)
(6.19)
288
u
A
(6.20)
(6.21)
Furthermore, we have already seen that the speed of sound can be written as:
a=
dp
d
(2.106)
If we combine (6.19), (6.21), and (2.106), we can show that (see Problem 6.8):
du
dA 2
= M 1
A
u
(6.22)
See a text book on calculus such as Ref. [2] for details on logarithmic differentiation.
289
sweptwing configuration with its equivalent body (1) and a deltawing configuration
with its equivalent body (2). The equivalent bodies are axisymmetric and have the
same crosssectional area distribution as the sweptwing configuration and deltawing configuration, respectively. The graph shows us the variation in zerolift drag
coefficient of the four wind tunnel models over the Mach range. We see that when the
Mach number is increased beyond M = 0.9 all models show a significant increase
in drag. This drag is attributed to the formation of shock waves and is therefore
0.25
0.20
equivalent body 1
0.15
0.10
equivalent body 2
0.05
1.0
0.9
1.1
0
20
15
10
5
10
M21
K=
2 2
( +1)Mb
Fig. 6.13 Zerolift drag coefficient of wingbody combinations and their equivalent axisymmetric
bodies (data from Refs. [8, 9])
290
termed wave drag. This wave drag can be quantified by the momentum deficiency
over the shock wave as we have seen in Sect. 6.4.3. If we compare the curves for
the deltawing configuration and its equivalent body, we see that their wave drag
in the transonic domain is very similar. This is in line with Whitcombs area rule.
However, when we look at the comparison between the sweptwing configuration
and its equivalent body, we notice that there is a significant difference between the
wave drag of the two bodies.
291
Let us investigate why there is such a strong difference between the sweptwing
configuration and its equivalent body. It is expected that the formation of shock waves
is an important reason for discrepancy. Since pressure waves, such as shock waves or
Mach waves, induce density gradients in the flow, schlieren optics is an often used tool
to visualize these waves and the flow field. Schlieren (German, meaning streaks)
optics captures density gradients in compressible flows. The physical principle behind
such visualization techniques as schlieren or shadowgraph is based on light beam
deflection as it propagates through a medium with density gradients. For details
of these flow visualization techniques, their distinguishing features, limitations and
applications, the reader may refer to classical gas dynamic books such as Shapiro [50],
Liepmann and Roshko [37], Ferri [20] or books on wind tunnel testing principles.
Figure 6.14 shows schlieren images of the two wind tunnel models at M = 0.98. We
see that both models show shock waves in the same longitudinal position. However,
the shock wave over the wing of the sweptwing model is much stronger than the
corresponding shock wave over the equivalent model. The shock wave over the
sweptwing model causes boundarylayer separation at the outer portion of the wing
which causes additional drag with respect to the equivalent body. This explains the
larger drag over the sweptwing configuration compared to its equivalent body.
Based on the observations above, we conclude that the area rule works well for
deltawing aircraft of low aspect ratio but works less well for sweptback wingbodies
with moderate aspect ratio. The area rule is therefore limited to lowaspectratio
thinwing body combinations. Viscous effects, such as shockwaveboundarylayer
interaction, prevent a correct application of the area rule to wingbody combinations
with higher aspect ratio or thicker wings.
292
In this equation ndS = dS. Under the same assumptions, the continuity equation in
integral form (2.110) can be written as follows:
(V n)dS = 0
(6.24)
S
V
v=
=
i
(6.25)
V
V
Here V is the freestream velocity in the xdirection and i is the unit vector parallel to
the xaxis. Note that this perturbation function v represents the normalized perturbation to the velocity field in x, y, and z direction, assuming that the freestream is aligned
with the xaxis. Substituting (6.25) into (6.23) we obtain the following expression:
2
pn + V
(6.26)
Fbody =
(v + i) (v n + i n) dS
S
(6.27)
We can employ the continuity (6.27) to show that (6.26) can be simplified. If we
assume that S is an enclosed surface we can introduce the freestream static pressure,
p , and reduce the momentum equation to (see Problem 6.9):
2
Fbody =
v (v n + i n)dS
(6.28)
( p p ) n + V
S
Equation (6.28) is a compact equation that relates the total body force to the properties
of the freestream flow ( p , V ) and the local perturbation velocities (v), density ()
and pressure p under the assumption of an inviscid and steady state flow.
We are interested in calculating the wave drag of a threedimensional body in
supersonic flow. An example of such a body in its cylindrical control volume is
shown in Fig. 6.15. The surface area, S of the control volume is divided into three
parts marked S1 S3 . Let us first extract the drag force, D, from (6.28):
2
u (v n + i n) dS
(6.29)
D = Fbodyx =
( p p ) i + V
S
x
= x
r
r
(6.30)
293
S2
R1
S3
S1
(6.31)
S1
( p p )dS3
S3
S2
2
V
x (x + 1)dS3
(6.32)
S3
We can simplify this expression by acknowledging that in supersonic flow the pressure and velocity at S1 must remain unperturbed by the body (i.e. p = p and
n x = 0). Similarly, we assume that at S3 is sufficiently far behind the body such
that the flow has become essentially twodimensional in the yz plane (Trefftz
plane). Therefore, x = 0 such that the last integral in (6.32) vanishes. In addition,
we employ Bernoulli (6.9) to relate the pressure to the velocity disturbance velocity
potential in cylindrical coordinates:
1 2 2
p p = V
r + 2 x + 2x
2
(6.33)
S2
V 2 2
( y + 2z )dS3
x r dS2 +
2
S3
(6.34)
294
This equation expresses the total pressure drag experienced by the body. In supersonic
flow, the pressure drag stems from two sources: the wave drag and the vortex drag.
The second integral in (6.34) gives the vortex drag, which is identical to the induced
drag for subsonic flow. The first integral represents the wave drag:
2
Dwave = V
x r dS2
(6.35)
S2
This is a remarkably simple expression for the wave drag in terms of the derivatives
of the perturbation velocity potential. We must note that the present result is only
valid for bodies that are pointed in the rear.
Q
4r V
(6.36)
where r is the distance measured at any point to the origin of the source.
Since we know that the superposition of elementary sources can represent the
shape of any axissymmetric body we would like to know what the wave drag would
be of a line of sources. Therefore, we introduce a socalled lineal source distribution:
a straight line of sources of strength Q = V f (x). Here, f (x) represents the variable source strength along the line. Without derivation we show that the associated
perturbation velocity potential in supersonic flow is [4]:
xr
f (x1 )dx1
1
=
2
(x x1 )2 2 r 2
0
(6.37)
where = M 2 1 and r is the distance from any point in space to the origin of
the yz plane. We assume that f is continuous and that f (0) = f (l) = 0. We can
substitute (6.37) in (6.35) to find a closedform expression of the wave drag. The
derivation involves complex algebraic operations that are beyond the scope of this
295
book. Reference [4] can be consulted for details on the exact derivation. We merely
present the reader with the result:
Dwave =
2 l l
V
f (x1 ) f (x2 ) ln x2 x1  dx1 dx2
4
(6.38)
0 0
x
2 l
V
=
f (x)dx f (x1 ) ln(x x1 )dx1
2
0
(6.39)
If we compare (6.37) with (6.39) we notice that the drag coefficient has become
independent of provided that the body ends in a point (as in Fig. 6.15) or ends in
a cylindrical portion. Finally, the tangentiality condition at the surface of the bodies
relates the source strength to the first derivative of the crosssectional area, S:
f (x) = S (x)
(6.40)
Example 6.2 We consider a body of revolution with the following radius distribution:
1 2
1
for 0 x 1
R(x) = x
4
2
(a) In one figure, plot the distribution of R, S, and S as function of x
(b) Calculate the drag coefficient of this body with respect to the unit square as
reference area.
Solution:
(a) We first calculate the crosssectional area distribution that corresponds to this
radius distribution:
2 2
1
1
x
S(x) = R 2 (x) =
4
2
Subsequently, we determine the derivative of the crosssectional area distribution
with respect to x:
2
1
1
1
x
x
S (x) = 4
4
2
2
The results are plotted in Fig. 6.16.
296
0.6
R(x)
S(x)
dS/dx
0.4
0.2
0
0.2
0.4
0.6
0
0.2
0.4
0.6
0.8
x/l
(b) To calculate the wave drag coefficient with respect to the frontal area (/16) we
rewrite (6.39) accordingly:
C Dwave =
l
x
16
f
(x)dx
f (x1 ) ln(x x1 )dx1
2
0
1
f (x) = S (x) = 12 x
2
2
1
When examining (6.39) we notice that for x = x1 the integrand of the second
integral tends to negative infinity. This means that we cannot evaluate this integral using standard quadrature methods. We therefore replace the integral by a
double series:
I
i1
16
C Dwave 2
f (x j ) ln(xi x j )x
f (xi )x
i=2
j=1
297
l
(1 + cos )
2
(6.41)
With this transformation the nose of the body is located at = , while the base is
situated at = 0. We can now express f as a Fourier series (see Sect. 2.2.2 for an
introduction on Fourier series or consult Ref. [30]) in according to:
f () = l
An sin (n)
(6.42)
n=1
(6.44)
2
sin (n + 1) sin (n 1)
sin (2)
l
+
=
An
A1 +
4
2
n+1
n1
S() =
n=2
(6.45)
If we integrate the area distribution between 0 and we obtain an expression for the
volume of the body in terms of the first two Fourier coefficients:
l 3
l
S() sin d =
V =
2
8
1
A1 A2
2
(6.46)
Finally, we can obtain an expression for the wave drag in terms of the Fourier coefficients by substituting (6.42) in (6.38):
2 l 2 0 0
V
Dw =
An n cos(n1 )
Am m cos(m2 )
16
n=1
m=1
298
l
ln (cos cos 1 ) sin 1 sin 2 d1 d2
2
2 l2
V
=
n A2n
8
(6.47)
n=1
16V
l 3
(6.48)
64V 2
2
V
l 4
(6.49)
In order to find the ideal area distribution for this body one can substitute A1 = 0,
A2 = 16V /l 3 into (6.45) and let all remaining An = 0. We then obtain the
following equation:
1
4V
(6.50)
sin sin 3
S() =
l
3
This is is the crosssectional area distribution that yields the lowest wave drag. The
axissymmetric body that it represents is termed the SearsHaack body after the
English scientist William Sears and the German scientist Wolfgang Haack, who
independently derived (6.50). In Fig. 6.17 one can see a SearsHaack body for a
fineness ratio of approximately 6.8.
To minimize the zerolift wave drag on any body in transonic flow, the crosssectional area distribution should be as close as possible to the SearsHaack body.
299
0.08
R(x)
S(x)
R(x), S(x)
0.06
0.04
0.02
0
0.2
0.4
0.6
0.8
x/l
The actual shape of the aircraft is fairly independent of its wave drag according to the
theory of Bsemann and Whitcomb. It is therefore important that large volumetric
additions to the body (e.g. wings, vertical and horizontal tail planes) are positioned
such that a crosssectional area distribution results that matches as closely as possible
to the SearsHaack body. Small deviations from the ideal SearsHaack area distribution do not lead to excessive amounts of wave drag. However, discontinuities in area
distribution can cause a large increase in wave drag and should therefore be avoided.
6.3.3.2 Von Krmn Ogive
The SearsHaack body gives minimum drag for a body with a pointed nose and rear.
A second shape of interest is the optimum nose cone shape for a straight body of
revolution (e.g. a rocket, or bullet). If one applies different boundary conditions to
(6.45) one can find the area distribution for which the nose cone produces minimum
drag. For a nose cone with a given base area, S(l), at x = l we have = 0. Substituting
= 0 in (6.45) we obtain:
4S(l)
(6.51)
A1 =
l 2
Since A1 is nonzero, the wave drag is minimized when An = 0 for n 2. Substituting
(6.51) into (6.47) one obtains the expression for the minimum wave drag of the nose
cone:
2 S(l)2
2 V
(6.52)
Dw =
l2
If we substitute the expression for A1 into (6.45) we obtain the area distribution for
the minimumdrag nose cone:
1
S(l)
(6.53)
+ sin 2
S=
2
This resulting body is called the Von Krmn ogive and gives the minimum drag for
a given base area and length. Figure 6.18 shows an example of a Von Krmn ogive.
300
R(x), S(x)
0.4
R(x)
S(x)
0.3
0.2
0.1
0
0
0.2
0.4
0.6
0.8
x/l
(b)
min
Minimum drag
coefficient, CD (~)
.04
prototype
(a)
.03
revised
improved
nose
.02
.01
0
.6
.7
.8
.9
1.0
1.1
1.2
prototype
10,000
8,000
thrust available
6,000
revised
4,000
improved nose
2,000
0
.6
.7
.8
.9
1.0
1.1
1.2
Fig. 6.19 Effect of crosssectional area distribution on performance of the F102 prototype at
35,000 ft (after Ref. [17]). a Drag coefficient and area distribution. b Drag an thrust versus Mach
number
301
In the late 1960s when commercial transports started to emerge, it became evident
that aircraft needed to be designed with the area rule and SearsHaack body in mind.
This resulted in some particular features of commercial transports that are still very
obvious this day. An example of how the crosssectional area distribution was altered
on the Boeing 747 is shown in Fig. 6.20. It can be seen that by extending the upper
deck the area distribution became more smooth, which resulted in a higher dragdivergence Mach number.
Another famous example of transonic area ruling is found in the antishock bodies
that were developed by Whitcomb and Kchemann. The antishock bodies were
positioned near the trailing edge of the wing to smoothen the crosssectional area
distribution of the airplane. The effect of these bodies was a reduction in shock
strength over the upper surface of the wing, resulting in a lower wave drag. This was
demonstrated by experiments at NASA, which are shown in Fig. 6.21. Notice the
more aft position of the shock wave due to the application of the antishock bodies.
Even though these bodies increased the wetted area of the airplane (i.e. more friction
drag), their benefit in terms of wave drag reduction outweighed this penalty. The
socalled Kchemann carrots found their way onto several aircraft such as the Convair
990 (fastest highsubsonic transport airplane to date) and the Handley Page Victor.
Yet another aircraft that benefited from the research of Whitcomb and Kchemann
was the Soviet Unions revolutionary Tu95 bomber. The Tu95 demonstrated the
ability to cruise at jetlike speeds of Mach 0.85 while using propellerdriven engines.
One of the reasons for this aircrafts exceptional performance was the use of large
engine fairings aft of the inboard engines. These pods brought the external shape
of the Tu95 closer to the SearsHaack body to reduce drag and increase maximum
speed.
Extended
cab
(b)
Wing
W/B fairing
Body
Crosssection aera
0.040
Basic cab
Extended cab
0.032
C L = 0.5
0.024 CL = 0.4
CL = 0.3
0.016
Body station
0.80
0.90
1.00
Fig. 6.20 Example of transonic area ruling on the Boeing 747 [24]. a Extended cab resulted in a
smoother area distribution. b Dragdivergence Mach number increased substantially
302
303
304
(a)
(b)
Fig. 6.22 Flow visualization of flow around a thin plate attached to a cylinder (after Ref. [1]).
a Potential flow streamlines. b Streamlines of water flowing past the object
the boundaries of the body Fig. 6.22b. We see that the streamlines do not follow the
curvature of the body anymore but separate from the surface near the top and bottom
of the cylinder.
If we consider the flow past a simple symmetric airfoil and we assume that the
Reynolds number is relatively large and the airfoil is aligned with the direction of the
freestream, then the effect of viscosity is confined to a very narrow region close to
the body. The streamlines about the the airfoil resemble very much the potential flow
streamlines (see Fig. 6.23a). Now, when we place this same airfoil at a significant
incidence angle with respect to the freestream flow direction, the viscous effects
become very pronounced and the flow field about the airfoil is completely changed
(Fig. 6.23b). The airfoil is said to be in a stalled condition. The boundary layer has
separated from the surface and a large wake is present between the streamlines and
the body. Separation is caused by a large adverse pressure gradient on the body
surface which causes separation of the boundary layer.
We first look at the development of a boundary layer in the absence of any pressure
gradient. In Fig. 6.24 the velocity profile of the flow about a flat plat is represented by
white areas. We clearly see that there is a noslip condition at the wall: the fluid is at
(a)
(b)
Fig. 6.23 Flow visualization of flow past an airfoil (after Ref. [1]). a At small angle of attack. b At
high angle of attack
305
a standstill there. With increasing distance from the wall, y, the velocity, u, increases
progressively until a point where it matches the freestream value, U . We typically
define the thickness of the boundary layer, , as the y coordinate where the velocity
vector reaches 99 % of the freestream value. If we look progressively downstream
we see that the thickness of the boundary layer increases, while the velocity gradient,
du/dy, decreases. The velocity gradient at the wall is directly related to the shear
stress, w , the fluid is exerting on the wall:
w =
du
dy
(6.54)
where is the dynamic viscosity of the fluid. This relation was originally proposed by Isaac Newton and confirmed by experiments carried out by Poiseuille
in 1849 using liquid flow in tubes. The local friction coefficient is the wallstress
nondimensionalized by the freestream dynamic pressure according to:
cf =
w
1
2
2 V
(6.55)
Based on our observations of the velocity profiles in Fig. 6.24 and Eqs. (6.54) and
(6.55), we can deduce that the local friction coefficient is highest near the leading
edge and that it reduces further downstream.
To understand why the boundary layer is growing in thickness downstream we
look at the time history of vorticity within the boundary layer. We know that for a
given control surface, the vorticity vector ( V ) is related to the velocity around
the contour of the control surface:
( V ) dS = V dS
(2.71)
S
The vorticity enclosed within the contour of the control surface is called the circula
tion, :
= V dS
(6.56)
C
Let us try to estimate the circulation in the boundary layer. In Fig. 6.25 we show a
contour drawn about the velocity profile in the boundary layer. Due to the noslip
306
U0
condition, the velocity at the wall is zero. Furthermore, we assume that locally the flow
is aligned with the wall and that the perpendicular velocity component is therefore
zero. Evaluating the circulation per unit length about this contour therefore results in:
=
V dS = 0 + 0 U0 + 0 = U0
We can conclude that at any station, the circulation per unit length equals U0 . The
vorticity is constant at any streamwise station along the plate: U0 times a unit
length. The total amount of vorticity within each contour remains the same. Because
the flow upstream of the plate has no vorticity, we can conclude that the vorticity is
introduced at the leading edge of the plate due to the noslip boundary condition.
Even though the circulation per unit length is constant at any streamwise station
on the plate, the distribution of vorticity normal to the plate does change. Due to the
viscosity of the fluid the vorticity is spread transversely as it travels downstream. In a
laminar boundary layer, this goes according to the principle of molecular diffusion.
The local boundary layer thickness represents the distance the vorticity has diffused
away from the plate. This growing process is a function of time, t, and the diffusivity of momentum. The latter quantity is more commonly known as the kinematic
viscosity:
(6.57)
=
=
(6.58)
x
U0 x
Rex
This relationship is only valid at high Reynolds numbers where /x 1 and for
laminar boundary layers. We can note from (6.58) that with higher freestream velocity
the boundary layer becomes thinner at any station along the plate. This is simply
because the boundary layer has had less time to grow. Bodies exposed to high2
307
Reynolds number flows therefore have a thinner boundary layer with respect to their
length.
u
1
dy
0
e u e
(6.59)
If we assume that the flow in the boundary layer is incompressible its displacement
thickness can be reduced to:
u
1
dy
(6.60)
i =
0
ue
The incompressibility assumption is often used in practice, since it has shown to give
good correlation to experimental results.
Similar to the displacement thickness we can also define a thickness that accounts
for the missing momentum in the boundary layer. This thickness is termed the
momentum thickness, , and can be found from:
u
u
1
dy
(6.61)
=
0 e u e
u e
Again, if we assume that in the boundary layer the flow is incompressible we have:
u
u
i =
1
dy
(6.62)
0 ue
ue
Finally, we introduce a shape factor, H , that relates the displacement thickness to
the momentum thickness according to:
H=
(6.63)
The shape factor is a function of the shape of the velocity distribution within the
boundary layer, hence its name. Notional velocity profiles for both high and low
shape factor are shown in Fig. 6.26. Notice that for a small H the velocity gradient at
the wall is higher compared to a high H . The shape factor will prove its importance
in predicting, for example, transition and separation (Sect. 6.4).
308
1.0
1
y/
high H
low H
u/u e
1 l
cf dx
l 0
(6.66)
(6.67)
This equation demonstrates once more that an increase in the Reynolds number
decreases the friction drag coefficient of a surface wetted by an airflow. If we relate
that to airplanes, we can conclude from this simple derivation that both an increase in
size as well as an increase in cruise velocity decreases the average friction coefficient
of the aircraft. However, we must note that the effect of surface roughness has not
yet been taken into account and that we solely speak of laminar boundary layers. It
should also be noted that absolute drag does increase with velocity and size.
309
(a)
(b)
40
M = 12
32
32
=
M
M=1
24
M = 6
16
12
M = 8
24
=
M
16
8
=6
M M = 4
y
Normal Distance Parameter, x Re (~)
40
M = 4
8
M = 2
M = 0
0
0.2
0.4
0.6
0.8
1.0
10
M = 2
20
30
Fig. 6.27 Mach number and temperature distribution across a laminar boundary layer on an insulated flat plate for various Mach numbers. Calculations carried out by Van Driest [59] assuming
Pr = 0.75 and S/Tw = 0.505. a Mach number distribution. b Temperature distribution
For compressible flows the density is not constant and the assumption of a constant viscosity and heat transfer coefficient through the boundarylayer thickness are
no longer valid. The friction between the flow and the wall generates a temperature
difference between the wall and the freestream. The resulting temperature distribution in the boundary layer has an effect on the velocity distribution and therefore
on the velocity gradient at the wall. At the same time, the temperature has an effect
on the viscosity through Sutherlands formula (2.81). Both effects contribute to a
change in friction coefficient.
For a laminar boundary layer and adiabatic wall conditions, Van Driest [59] calculated the change in velocity profile with freestream Mach number using Croccos
method.3 Assuming adiabatic wall conditions and a constant heat transfer coefficient,
the temperature and Mach number profiles within the boundary layer could be determined for various freestream Mach numbers. The Mach number and temperature
distribution is shown in Fig. 6.27.
If we examine Fig. 6.27a, we can observe that an increase in Mach number has
several effects. First of all, it increases the boundary layer thickness. Secondly, the
Mach number gradient at the wall decreases with Mach number. It can be shown that
also the velocity profile follows this trend [59]. This results in a lower shear stress
at the wall, which causes a reduction in friction coefficient with Mach number. The
sonic line is also shown in Fig. 6.27a. With increasing Mach number, the sonic line
moves away from the wall. However, if we compare the location of the sonic line to
the thickness of the boundary layer, we conclude that the position of the sonic line
only changes slightly. For example, for M = 2 the sonic line lies at approximately
3
310
0.40, where has been defined as the distance where u/V = 0.995. For M = 8
the sonic line is located at approximately 0.35.
The temperature distribution in Fig. 6.27b shows the effect of friction on the wall
temperature. Note that at M = 2 the wall temperature is 40 % higher than the freestream temperature. At M = 4 there is more than a factor of 3 between the wall
temperature and the freestream temperature. We can therefore deduce that for adiabatic conditions, the temperature increase over at high supersonic Mach numbers
results in temperatures that are higher than conventional aluminum can sustain without degrading material properties. Therefore, supersonic aircraft that fly faster than
M 2.5 have to rely on more heat resistant materials, such as titanium. In the
hypersonic realm (M > 5) the wall temperatures become extremely high, which
requires aerospace vehicles that operate in this regime to have an ablative heat shield
or a thermalsoak heat shield. In the transonic domain, aerodynamic heating is comparatively small and has little effect on the velocity gradient and friction coefficient.
Therefore, the incompressible boundarylayer equations can often be successfully
used to predict the boundary layer properties in the transonic flow domain.
Using the method of Van Driest, the compressible friction coefficient can be
correlated to the incompressible friction coefficient. An approximation of the relation between compressible and incompressible friction coefficient was derived by
Johnson and Rubesin [27]:
Cf
1
=
0.12
C fi
1 + 0.1305M 2
(6.68)
Both the approximation of Johnson and Rubesin as well as the calculation of Van
Driest is shown in Fig. 6.28. We see a gradual decay of the friction coefficient with
increasing Mach number. We also note that in the transonic flow domain the decrease
in friction coefficient is relatively small. Using (6.68), a reduction of 1.5 % at M = 1
can be computed. The change in turbulent friction coefficient is also presented in
Fig. 6.28 and is discussed in more detail in Sect. 6.4.2.
6.4.1.3 Flow Reversal in Laminar Boundary Layers
In the preceding paragraphs we have investigated the development of boundary layers over flat plates with negligible pressure gradients (i.e. p/x = 0). We now
investigate how the pressure gradient affects the velocity profile in the boundary
layer, the friction drag, the transition point, and the separation point.
We start our discussion with the presence of a favorable pressure gradient. In
practice this means that the further downstream one travels, the lower is the pressure
(i.e. p/x < 0). In this case, the flow outside the boundary layer accelerates
continuously while the favorable pressure gradient is present. This effect is illustrated
in Fig. 6.29 where we see the velocity profile of a fluid in a converging channel (twotoone contraction ratio). The flow accelerates in the converging part of the channel
such that it reaches twice its velocity in the narrower part of the channel.
(~)
311
Coles 1
Chapman and Kester 2
Monaghan and Cooke 1
Rubesi, Maydew and Varga 1
Brinich and Diaconis 1
Korkegi 1
Lobb, Winkler and Persh 1
M=0
0.8
Pr = 0.75
0.6
LAMINAR
(1+0.1305M 2 ) 0.12
(1+0.2M 2 )0.467
0.4
Sommer & Short method
0.2
nic
ic
son transo
sub
TURBULENT
supersonic
hypersonic
10
Fig. 6.28 Variation of skin friction coefficient with Mach number. Data from Refs. [25, 27, 37, 39,
46, 53, 58, 59]. 1: Mean skin friction values. 2: Local skin friction values, for Reynolds number
Re = 8,000 based on momentum thickness
(a)
(b)
h =.53
=.66
h
h
UPSTREAM
DOWNSTREAM
Fig. 6.29 Flow visualization of velocity profiles in a converging channel (after Ref. [1]). a Velocity
profile in a converging channel. b Comparison between upstream and downstream velocity profiles
When looking at the velocity profile in the boundary layer we see that the boundary
layer thickness reduces considerably. Most of this reduction can be explained by
the twotoone decrease in flow area. However, using the distance to the nearby
streamline, h, as a reference dimension, we can see that also the relative thickness
/ h has decreased. If we consider the vorticity within the layer up to h we can deduce
that the vorticity increases linearly with the increase in freestream velocity. This new
vorticity is added to the vorticity already present in the boundary layer at the wall.
It is as if a new boundary layer is being formed at every increment in the converging
duct. The combined profile at the exit of the contraction is relatively thinner because
the vorticity has had little time to diffuse. In the narrow part of the duct we now have
a boundary layer with a larger percentage of vorticity close to the wall. This results
in a higher velocity gradient, shear stress, and friction coefficient.
312
In a divergent channel, the opposite effect occurs. When the diffuser angle is
relatively low, the flow velocity in the channel will decelerate and an unfavorable (or
adverse) pressure distribution is present: p/x > 0. The pressure downstream is
higher than the pressure upstream. The result is a thickening of the boundary layer
both in the absolute and in the relative sense. In other words, h /x > 0. The reduced
velocity slope at the wall reduces the shear stress and the friction coefficient. The
velocity profile becomes less full: less vorticity is concentrated near the wall and more
is concentrated near the edge of the boundary layer. This can be seen by comparing
the velocity profiles at the first two stations of Fig. 6.30b.
If the deceleration of the flow continues to take place the velocity gradient (u/ y)
near the wall becomes smaller and smaller. Eventually the shear stress goes to zero,
followed by a reversal of the flow direction near the wall. This is visualized in the
3rd station of Fig. 6.30b. We see that the flow at the second station is still directed
to the right. While the flow at the third station is directed to the left. Somewhere in
between those two stations the flow at the wall has come to a complete stand still.
The corresponding shear stress at this point is zero. We call this point the separation
point. Beyond this point, the fluid, which was in contact with the wall in the upstream
boundary layer, has separated from the wall by a region of reversed (or recirculating)
flow. This is visualized in Fig. 6.31.
We have now established that the development of the boundary is influenced by
the pressure gradient. Given the fact that any body in a flow field generates a pressure
distribution, we can deduce that the boundary layer development is influenced by
the shape of the pressure distribution. In addition, the separation point (if there is
any) is affected by the pressure distribution. The shape of the velocity profile at any
point in the boundary layer is also dependent on the conditions of the boundary layer
downstream. For the simple case of a flat plate we know that the boundary layer
transitions from laminar to turbulent and subsequently grows in thickness the further
we move downstream. A thicker boundary layer has a lower velocity gradient at the
wall and is therefore more prone to separate whenever an adverse pressure gradient
is present. If we recall the lessons we learned from Sect. 6.2 we can state that in order
to prevent separation we need to modify the outer shape of the body such that the we
obtain relatively small pressure (adverse) gradients in regions where the boundary
layer is relatively thick.
(a)
(b)
Fig. 6.30 Flow visualization of velocity profiles in a diverging channels of different diffuser angle
(after Ref. [1]). a Diverging channel with small diffuser angle. b Diverging channel with larger
diffuser angle
313
dp
>0
dx
u
u
u
Point of
separation
u
>0
y
u
=0
y
u
<0
y
In this context, the word critical is unrelated to the flow properties at sonic conditions.
314
Fig. 6.32 The classical dye
experiment in pipe flow as
described by Reynolds in
1883 (after Ref. [45]). a Low
speed: laminar flow, b high
speed: turbulent flow, c spark
photograph of condition (b)
(a)
(b)
(c)
T/S
waves
Threedimensional
vortex
breakdown
Turbulent
spots
Fully
turbulent
flow
Stable
laminar
flow
x
0
Spanwise
vorticity
Edge
contamination
Recrit
(x)
Laminar
Transition length
Turbulent
Re tr
for which any disturbance of arbitrary wave length causes the boundary layer to
become unstable. Based on a theoretical investigation of compressible boundary
layer stability over a flat plate, Lees [35] concludes that for subsonic conditions any
laminar boundarylayer flow is unstable at sufficiently high Reynolds numbers. The
minimum critical Reynolds number is largely determined by the product of mean
density and the mean vorticity across the boundary layer. The addition of heat to the
fluid through the solid surface reduces the minimum critical Reynolds number and
therefore hastens the onset of transition. Cooling the wall has the opposite effect.
For adiabatic conditions, Laufer and Vrebalovich [34] concluded that the Mach
number has no effect on the stability of the laminar boundary layer but is solely
dependent on the displacementthickness Reynolds number, Re . In addition, the
amplification factor is also shown to be independent of the Mach number. However,
it is shown that the disturbances propagate faster with increasing Mach number. Consequently, a disturbance propagating to a certain xposition will not be amplified as
much in a higher Mach number flow because its time available to grow is less, even
315
though its amplification coefficient is the same. Therefore, the adiabatic compressible boundary layer is shown to be more stable with respect to small disturbances
than the incompressible one.
Once the boundary layer has transitioned from laminar to turbulent, its velocity
profile changes significantly. Whereas the laminar boundary layer showed a smoothly
changing velocity profile, the large amount of vorticity in the turbulent boundary layer
creates a more rugged velocity profile. In Fig. 6.34 the instantaneous velocity of a
laminar and turbulent boundary layer are shown on opposite sides of a flat plate. The
ruggedness of the velocity profile in the turbulent boundary layer is evident. When we
superimpose a large number of these instantaneous velocity profiles over time, we can
distinguish a mean velocity profile for both the turbulent and the laminar boundary
layer. It can be seen from Fig. 6.34 that the deviation from the mean velocity profile
is much larger in the turbulent boundary layer than it is in the laminar boundary layer,
due to the randomness of the vortices that are created in the turbulent boundary layer.
In Fig. 6.35 the two velocity profiles are compared. When superimposing them
we see that the turbulent boundary layer shows a larger velocity gradient at the wall.
The turbulent boundary layer therefore generates more shear stress as per (6.54).The
incompressible friction coefficient at a particular station x on a flat plate can be
approximated by the power law [66]:
cf =
0.027
1/7
(6.69)
Rex
Note that this relation only holds for low subsonic conditions. For turbulent boundary
layers the incompressible friction coefficient therefore decays much slower with
1/7
1/2
while cf Rex
for laminar boundary layers. In
Reynolds number: cf Rex
Fig. 6.36 the local friction coefficient is plotted for both a turbulent and a laminar
boundary layer. In addition, a typical transition line is shown which shows the rapid
increase in local friction coefficient when the boundary layer transitions from laminar
to turbulent.
The circulation is the same in both the laminar and turbulent boundary layers.
However, the distribution of circulation is different. For the turbulent boundary layer
(a)
(b)
Fig. 6.34 Boundary layer around a flat plat: flow is turbulent on upper surface and laminar
on lower surface (after Ref. [1]). a Instantaneous velocity profiles. b Superposition of many
instantaneous velocity profiles
316
(a)
(b)
LAMINAR
TURBULENT
Fig. 6.35 Comparison between velocity profiles in laminar and turbulent boundary layers (after Ref.
[1]). a Comparison of mean laminar (solid) and turbulent (dashed) velocity profiles. b Comparison
of velocity gradients at the wall for laminar and turbulent boundary layers
0.005
Exact
Powerlaw theory
0.004
0.003
0.002
Typical
transition
curve
Laminar flow:
0.664
(Blasius)
Re
0.001
0
105
106
107
108
109
1010
more vorticity is concentrated near the plate even though some vorticity has also
spread farther from the plate. The thickness of the turbulent boundary layer on a flat
plate at point x can be approximated according to [66]:
0.16
1/7
x
Rex
(6.70)
Comparing this to the laminar boundary layer we see that the laminar boundary layer
grows according to x 1/2 , while the turbulent boundary layer grows much faster
according to x 6/7 .
The distribution of momentum is also different between the two boundary layers.
The unsteady random rotary motions associated with vorticity that is aligned with the
flow transports highmomentum fluid towards the plate while lowmomentum fluid
is transported away from the plate. This extra momentum close to the wall enables
the turbulent boundary layer to withstand much larger adverse pressure gradients
317
Fig. 6.37 Gloster Javelin FAW.9 with three rows of vortex generators on the wing to delay the
onset of boundary layer separation. Photo Adrian Pingstone
before separating from the wall. This is the reason why vortex generators are often
seen on airplane wings. They generate additional vorticity in the direction of the
flow that transports momentum towards the wing surface and hence delays the onset
of separation. Rows of vortex generators can be seen on various modern transonic
airplanes. A historic example is the Gloster Javelin having three rows of vortex
generators on its outer wing (Fig. 6.37).
6.4.2.1 Effect of Reynolds Number and Mach Number on Friction Drag
We have already noted that the effect of the Reynolds number on the friction coefficient is less strong in turbulent boundary layers than in laminar boundary layers.
There exist many empirically derived relations that correlate the turbulent friction
coefficient, Cf , of a hydraulically smooth plate to the freestream Reynolds number.
Schlichting [48] proposed the following interpolation formula:
Cf =
0.472
(log10 Re)2.58
(6.71)
Note that the two coefficients in (6.71) are empirically determined. Von Krmn and
Schoenherr used the following equation to fit the empirical data:
0.242
= log10 Cf
Cf Re
(6.72)
Equation (6.72) needs to be solved iteratively to find a solution for Cf . Note that the
Von KrmnSchoenherr relation relies on a single empirical coefficient. In Fig. 6.38
the two relations are shown together with various experimental results. Some of the
experiments date back to Froude in the late 19th century. There exists quite some
scatter between the various experimental data. According to Ref. [40] the experimental data from Smith and Walker (1959) is most reliable due to the accurate
measurement techniques. It can be seen from Fig. 6.38 that both theoretical lines fol
318
4.0
x 10
3.8
3.6
3.4
3.2
3.0
2.8
2.6
Schlichting
2.4
2.2
2.0
3
10
20
30
x 10 6
Fig. 6.38 Variation of turbulent skin friction coefficient with Reynolds number. Data selected
from [40]
low approximately the same trend. It depends on the experimental data set to judge
which theoretical line gives a better prediction.
It is of practical interest to investigate the effect of the Mach number on the
friction coefficient of a turbulent boundary layer. We would like to know if the friction
coefficient in transonic domain is influenced by the Mach number. Figure 6.28 on
p. 309 shows experimental results for the compressible, turbulent friction coefficient
up to a Mach number of 10. In addition, two theoretical lines are plotted, which
are shortly discussed hereafter. It can be observed that the friction coefficient of the
turbulent boundary layer has a much stronger dependency on the Mach number than
the friction coefficient of a laminar boundary layer.
Rubesin et al. [46] base their prediction of the compressible, turbulent friction
coefficient on the interpolation curve of Schlichting (6.71):
Cf =
0.472
0.467
2
(log10 Re)2.58 1 + 1
M
(6.73)
319
(6.74)
(6.75)
(6.76)
where the viscosity ratio is given by the Sutherland equation (2.81). The relationship
between Cf and Re is given by the Von KrmnSchoenherr relation (6.72):
0.242
= log10 Cf
Cf Re
(6.77)
Finally, the compressible skin friction coefficient is obtained from Cf according to:
Cf =
Cf
T
T
(6.78)
If we compare the two theoretical lines we can conclude that the method of Sommer
and Short is most consistent with experimental results up to M = 10. However, in the
transonic regime equation (6.73) gives an approximation that is sufficiently accurate
given the scatter in experimental results. For M = 1, the compressible friction
coefficient is about 8 % less than its incompressible counterpart. If we compare that
to the laminar boundary layer (1.5 % at M = 1), we can conclude that the turbulent
boundary layer is more sensitive to changes in Mach number and that the change in
friction coefficient due to compressibility effects is significant in the transonic regime.
320
components intersect (see Sect. 6.6.2). Minimizing supervelocities on these components to prevent the onset of supersonic flow and associated shock waves can be a
design goal to prevent such a drag penalty.
Figure 6.39ad show shockboundary layer interaction in transonic speeds (from
Ref. [36]). The first three flow visualizations show the effect of successively increasing
flow Mach number while the boundary layer is laminar. The last figure shows shock
boundary layer interaction for a turbulent boundary layer in transonic speeds. Note
that the weaker shocks that are formed in lower flow Mach numbers of 0.795 and
0.843 interact with the boundary layer, but do not cause significant separation (see
Fig. 6.39 Shock
boundarylayer interaction at
transonic speed for
knifeedge airfoil normal to
flow direction (from
Ref. [36]); Reprinted with
permission of the American
Institute of Aeronautics and
Astronautics, Inc. a Flow at
Mach 0.795,
Rec = 8.33 105 (laminar).
b Flow at Mach 0.843,
Rec = 8.45 105 (laminar).
c Flow at Mach 0.895,
Rec = 8.77 105 (laminar).
d Flow at Mach 0.895,
Rec = 1.75 106 (turbulent)
(a)
Flow
(b)
Flow
(c)
Flow
321
(d)
Flow
Fig. 6.39 (continued)
Fig. 6.39a, b). This is referred to as weak interaction of the shock with the boundary
layer. For a slightly higher Mach number, namely 0.895, the shocks are strong enough
to cause boundary layer separation (see Fig. 6.39c). We call this strong interaction.
The turbulent eddies are visible at the interface of the separated shear layer and the
subsequent pressure waves that emanate from the wake. The effect of boundary layer
type, i.e., laminar versus turbulent, on shockboundary layer interaction is qualitatively shown in Fig. 6.39c, d. Note that the flow Mach number is maintained between
the two cases (both are at 0.895) but the Reynolds number is doubled. The laminar
boundary layer shows earlier separation on the airfoil than the turbulent boundary
layer case, as expected. Consequently, the wake downstream of the airfoil in the
laminar boundary layer case is thicker than in the turbulent boundary layer case. We
also note the lambda shock formation at the trailing edge of the airfoil in Fig. 6.39d.
322
compression waves
M e <1
M e >1
M>1
large
eddy
region
M<1
M>1
sonic line: M=1
merging layer
viscous sublayer
Fig. 6.40 Detail of weak shockwave boundarylayer interaction for a turbulent boundary layer
(after Ref. [41])
the streamlines move away from the wall. The resulting curvature in the streamlines
causes the formation of (isentropic) compression waves. These compression waves
start at the sonic line and allow the supersonic flow to change the flow direction
(remember that in supersonic conditions the direction of the flow can only be altered
through the formation of waves as we explained in Chap. 4). The fan of compression
waves merge into a shock wave outside of the boundary layer. Close to the boundary
layer edge the shock wave is still slightly oblique which reduces the flow velocity
behind the shock to low supersonic. Within the boundary layer this is also the case:
the flow behind the compressions waves can still be (slightly) supersonic. This gives
rise to a pocket of supersonic flow behind the shock wave that is termed the supersonic tongue. Within the supersonic tongue the flow decelerates isentropically to
subsonic speeds. The size of the supersonic tongue is casedependent and is merely
notional in Fig. 6.40.
The profile of the velocity distribution in the boundary layer is altered in the short
interaction region. The outer velocity decreases substantially and the boundary layer
itself thickens. This results in a higher value for the shape factor (H ) and a lower
velocity gradient at the wall.
The outer flow through the shock wave experiences an almost instantaneous rise in
static pressure. In other words: dCp /dx . Based on our previous discussion on
the effect of the pressure distribution on boundary layer separation we might therefore
wrongfully conclude that the boundary layer would invariably separate from the surface. However, a closer look into the shockwave boundarylayer interaction reveals
that the pressure increase is smeared out over a larger streamwise region. This smearing property is due to the formation of the compression waves within the boundary
layer. The effective pressure gradient within the boundary layer is therefore much
smaller and merely results in an increase in shape factor and boundary layer thickness.
Finally, we note that even in the case of a weak interaction the boundary layer
properties are adversely affected. Beyond the shock wave the boundary layer can
323
slip line
oblique shocks
M e <1
M>1
M>1
M e >1
M>1
separation line
separation
bubble
M<1
Fig. 6.41 Detail of strong shockwave boundarylayer interaction for a turbulent boundary layer
(after: [16])
324
larger distance such that the triple point is moved upwards and the integrated total
pressure loss over the shock system decreases.
The velocity profile of the boundary layer changes substantially over this interaction. Initially, the profile shows a high pressure gradient near the wall. However, due
to the large adverse pressure gradient under the oblique shock, the velocity at the wall
quickly reduces to zero at point S. The flow separates and a bubble of recirculating
flow forms. The boundary layer shows a negative velocity gradient at the wall. The
reverse flow aids significantly in the upstream transmission of the postshock pressure rise that causes the separation of the boundary layer in the first place. Beyond
the second oblique shock the deceleration of the subsonic flow in the boundary layer
causes the streamtube to expand. This aids in the reattachment of the boundary layer.
At the reattachment point (R) the velocity gradient at the wall is again zero. Over
the distance of the separation bubble the sonic line gradually moves away from the
wall towards the boundary layer edge. Further downstream of the bubble the velocity
profile has significantly thinned compared to the initial velocity distribution due to
the lower edge velocity and thicker boundary layer.
The onset of separation at the foot of the shock is dependent on the development of
the boundary layer in front of the shock and the surface curvature at the shock. However, for general purposes we can assume that on a flat plate separation starts at shock
Mach numbers (Me ) in the range of 1.31.35 [5]. In many practical cases the flow
behind the shock wave is further decelerated towards the trailing edge of the body.
Under influence of such an adverse pressure gradient the flow might separate close to
the trailing edge (trailing edge separation). Increasing the shock Mach number pushes
the shock further aft. In addition, the separation bubble grows, which also moves the
reattachment point further aft. Finally, the reattached boundary layer is now more
likely to separate under the adverse pressure gradient. This moves the trailingedge
separation point more forward. At some Mach number the reattachment point gets
close enough to the point of trailing edge separation such that the flow does not reattach anymore. This generates a large wake that originates at the foot of the shock.
This shockinduced separation is an important contributor to the exponential increase
in drag (drag divergence) that occurs at the drag divergence flight Mach number.
For more information on the interaction between the normal shock and the boundary layer the reader is referred to Refs. [5, 16].
325
(6.79)
u
v
u
p
u
u
+ v
=
+
+
x
y
x
y
x
y
(6.80a)
xmomentum equation
ymomentum equation
u
v
v
v
p
+ v
=
+
+
x
y
y
x
x
y
(6.80b)
energy equation
(ux y ) (vx y )
V2
2 T
2 T
V e +
+
= k 2 + k 2 pV +
2
x
y
y
x
(6.81)
326
v
u
+
=0
x
y
(6.82)
xmomentum equation
u
ymomentum equation
u
u
dpe
+ v
=
+
x
y
dx
y
u
y
(6.83a)
p
=0
y
(6.83b)
2
h
u
h
2 T
d pe
u
+ v
=k 2 +u
=
x
y
y
dx
y
(6.84)
energy equation
Note that subscript e represents the conditions at the boundarylayer edge. Also
note that the ymomentum equation basically states that the pressure stays constant throughout the boundary layer. The pressure is therefore solely dependent on
x. Hence, from now on we replace its partial derivative operator () with a total
derivative operator (d).
By defining the boundary values at a solid surface, this set of equations can be
seen as a boundary value problem (BVP). To assess this BVP, consider a coordinate
system which is bound to a solid object, with the yaxis perpendicular to its surface
and the xaxis aligned with the flow. The wall boundary values usually imply the
noslip condition:
(6.85)
u(x, 0) = 0, v(x, 0) = v0
Here, v0 is the velocity perpendicular to the wall. In most conditions v0 = 0. However,
in case boundarylayer suction is applied v0 < 0. In addition, the temperature at the
wall is defined as follows:
(6.86)
T (x, 0) = Tw (x)
In the general case, Tw is a function of time, i.e. Tw = Tw (x, t). In the case of
adiabatic wall conditions the heat flow to the wall, qw is zero and the following
boundary condition results:
T
=0
(6.87)
y y=0
If adiabatic wall conditions exist, the adiabatic wall temperature (Tw = Taw ) is a
part of the solution rather than a set boundary condition. The steadystate adiabatic
wall temperature can, however, be computed by using (6.74). At the outer edge
(subscript e) of the boundary layer, the following boundary conditions exist:
lim u(x, y) = u e (x),
(6.88)
327
An estimate of the values of u e (x) and Te (x) can be obtained by calculating the
flow around an object using for example the Euler equations. The five boundary
conditions together with the coupled equations of motion [(6.82)(6.84)] form the
basis for the analysis of laminar boundary layer flow. A significant simplification with
respect to the original equations of motion can be observed. It is stressed that the
above equations are derived for laminar boundary layers, assuming incompressible
flow and a thin boundary layer. A more complex orderofmagnitude analysis can
be carried out for boundary layers which exhibit a significant thickness [48, 55].
However, this is beyond the scope of the present text.
When we integrate the boundarylayer momentum Eq. (6.83a) from y = 0 to
y = and also employ (6.82) we obtain the Von Krmn momentum integral relation,
(also known as the momentum integral equation):
cf
du e
w
d
+ (2 + H )
= 2 =
dx
u e dx
u e
2
(6.89)
1
ff =0
2
(6.90a)
subject to
f = f = 0
on = 0
f 1 as
(6.90b)
(6.90c)
328
where
=y
f () =
V
x
u
V
(6.90d)
(6.90e)
The resulting velocity gradient at the wall has been quoted in (6.64). The displacement
thickness of a typical Blasius velocity profile (incompressible conditions) can be
shown to be a function of the Reynolds number according to:
1.72x
=
Rex
(6.91)
(6.92)
2x=c
c
(6.93)
(6.94)
329
2 du e
dx
(6.95)
Subsequently, we correlate the shear stress parameter [(S()] and the shape factor
[H ()] to using Thwaites empirical relations [66]:
S() =
w
( + 0.09)0.62
u e
(6.96)
(6.97)
with z = (0.25 ). Note that when S 0 the shear stress goes to zero. In other
words, the boundary layer separates when < 0.09. Substituting (6.95) and (6.96)
in (6.94) reduces it to:
dx
d
= 2 [S (2 + H )] = F()
(6.98)
ue
dx
du e
Thwaites shows that F() = 0.45 6. We can therefore rewrite (6.98) to the
following differential equation [employing (6.95)]:
d
ue
dx
2
+6
du e
= 0.45
dx
(6.99)
2 6
u
e
= 0.45
(6.100)
Integrating this differential equation and assuming that the integration constant is
zero results in the following relation between boundarylayer edge velocity and
momentum thickness:
0.45 x
u 5e dx
(6.101)
= 6
ue
0
330
used to obtain estimates of the displacement thickness and shear stress at any point
on the body. By doing a number of iterations of inviscid and displacementthickness
calculations, a converged value of the displacement thickness results. It can be shown
that the displacement thickness is about 1/3 of the boundary layer thickness [48].
The following example illustrates this process.
Example 6.3 Consider a plate with chord length of c = 1 m, which is exposed to
a flow with V = 100 m/s at 0 m ISA conditions. Consider the following velocity
distribution at the edge of a boundary layer:
x
u e = V 1
c
(6.102)
Calculate:
(a) The chordwise position of separation, xseparation .
(b) The friction coefficient, cf between x = 0 and xseparation .
Solution:
(a) The analytic nature of the velocity distribution allows us to compute from
(6.101):
=
x
x
5
0.45
5 1
V
dx =
6
6 1 x
c
V
0
c
0.075
x
6
c
1
1
V
c
x
6
1
c
331
(6.103)
We nondimensionalize the velocity in the boundary layer and the boundary layer
thickness by introducing two additional new variables:
u+ =
u
u
w /w
yu
y =y
=
w
(6.104)
(6.105)
30
0.047
0.0151
0.20
7.49
1.22
0
0.026
0.038
0.0000 0.0047 0.0097
0.22
0.22
0.21
14.84
9.85
2.42
1.61
S
w
cf 103
mm
()
()
N/m2
()
50
0.064
0.0271
0.18
4.92
0.80
60
0.071
0.0338
0.17
4.06
0.66
70
0.078
0.0410
0.15
3.35
0.55
80
0.085
0.0488
0.14
2.73
0.45
90
100
110
120
0.092
0.099
0.106
0.114
0.0571 0.0662 0.0760 0.0866
0.12
0.10
0.07
0.03
2.17
1.63
1.08
0.42
0.35
0.27
0.18
0.07
332
6 Aerodynamics of Nonlifting Bodies
333
u+ = y+ y+ < 5
(6.106)
The velocity in the merging layer (sometimes called overlap layer or blending
layer) between the laminar sublayer and the largeeddy region can be predicted using
a formula provided by Spalding [54]:
+
y =u +e
(u + )2
(u + )3
u +
+
e
1 u
2
6
From experiments, the Von Krmn constant is found to be = 0.41 and B = 5.0
for a smooth wall. This equation is termed the law of the wall.
In the largeeddy region the velocity changes logarithmically with distance from
the wall. The following approximation by Coles and Hirst [12] is often used:
u+ =
1
ln y + + B with = 0.41 and B = 5.0 y + > 30
(6.108)
The logarithmic region in the boundary layer extends to 1020 % of the total boundary
layer thickness, depending on the value of Re . In the wake component of the
boundary layer, the local pressure distribution has a significant effect on the velocity
distribution. Therefore, Coles [11] proposed to add an additional quantity, , to
account for the local pressure distribution. The resulting law of the wake is written
as follows:
u+
1
2 y
ln y + + B +
f
y + > 30
(6.109)
It was shown that Coles wake parameter, , could be related to the displacement thickness of the boundary layer by integrating (6.109) over the boundary layer
thickness:
1+
with =
(6.110)
cf
The value of is a measure for the local skin friction coefficient and is useful in
the analyses which follow. Note that this is not the same as the we used in the
analysis of the laminar boundary layer with Thwaites method (Sect. 6.5.2). Finally,
the wake function f (y/) is of the form [11]:
f
= sin2
y
2
(6.111)
This implies that the wake function is zero at the wall and unity at the edge of the
boundary layer. This function allows for the Sshape we expect to see in the velocity
profile in the wake of the boundary layer whenever there is a pressure gradient present
(see Sect. 6.4). At the edge of the boundary layer, we can now relate the edge velocity
334
45
< 100
40
=6
=4
=2
30
25
viscous
sublayer
merging
layer
=0
u = u/u
35
20
15
logarithmic
region
10
largeeddy
layer
5
0
0
10
outer
region
10
10
10
10
y = y u /
Fig. 6.42 Velocity distribution in a turbulent boundary layer for various values of , assuming
+ = 10,000
ue
1 Re
2
= ln
+B+
u
(6.112)
The velocity distributions as presented in this section are summarized in Fig. 6.42
for various values of . If we connect the prediction of Spalding to that of Coles and
Hirst, we observe the typical Sshape in the velocity profile of the boundary layer flow.
(6.89)
In order to use this equation we assume that an outer velocity distribution is known
and that we need to obtain values for cf , , and H . In order to do so we require at
least two additional equations. One is given by Coles wallwake law: (6.112). Based
335
on this relation the following algebraic equation due to Felsch et al. [19] describes
the relation between H , Re and cf :
cf = 0.058Re0.268 (0.93 1.95 log10 H )1.705
(6.113)
The third relation stems from a large body of experimental research where is
correlated to Clausers equilibrium parameter which is defined as follows [10]:
=
du e
d pe
= 2 H
w dx
u e dx
(6.114)
where d pe /dx is the pressure gradient at the edge of the boundary layer and is as in
(6.110). Clausers equilibrium parameter is directly related to the pressure gradient.
Whenever < 0 we have a favorable pressure gradient (d pe /dx < 0). Vice versa,
when > 0 we have an adverse pressure distribution (d pe /dx > 0). The proposed
correlation between and stems from Ref. [15]:
= 0.4 + 0.76 + 0.42 2
(6.115)
H 2 + 3.2 + 1.5 2
H 1
(1 + )
(6.116)
Let us review this method and put this in a practical perspective. We would like to
calculate the properties of the boundary layer at various stations along a plate under
the presence of a pressure distribution. To that extent we intend to solve the ordinary
differential equation (6.89) with known boundary conditions. These boundary conditions could for example stem from an assumed value of the momentum thickness
or Reynolds number at x = 0. Subsequently, we use the algebraic equations (6.113)
(6.116) to close the problem.
Example 6.4 Assume we have a pressure distribution over a curved plate of unit
length that can be represented by the following polynomial:
Cp = bx(ax 1)
where a and b are coefficients. Let us further assume that the boundary layer over the
curved plate is made fully turbulent at x = 0 and that the initial Reynolds number
based on momentum thickness, Re = 100. Furthermore, the freestream velocity is
V = 50 m/s and 0 m ISA conditions apply. Consider the following cases:
336
1. a = 21 and b = 1
2. a = 1 and b = 21
3. a = 2 and b = 2
For these three cases perform the following tasks:
(a) Plot the pressure distribution, velocity distribution and distribution of the velocity
gradient for each of the aforementioned cases.
(b) Plot the distribution of the following boundary layer parameters for the three
aforementioned cases: H , , , cf , , and .
(c) Briefly reflect on the results that have been plotted.
Solution:
(a) First we employ (6.11) to rewrite the equation of Cp for u e :
u e = V 1 + bx(1 ax)
Subsequently we take the derivative of u e with respect to x to find du e /dx:
b(1 2ax)
V
du e
=
dx
2
1 + bx(1 ax)
We substitute each combination of a and b in the above equations and plot the
results in Fig. 6.43. A quick evaluation of these three graphs shows us that for
case 1 we have a completely favorable velocity gradient. For cases 2 and 3 we
have a velocity gradient that is favorable until x = 0.5. Beyond this point the
pressure gradient is negative, which implies that the outer flow is decelerated.
This effect is four times larger for case 3 than four case 2.
(b) To answer the second part of the question we must solve the first order differential
equation (6.89). We rewrite (6.89) according to:
F=
cf
cf
1 2ax
du e
b
d
=
(2 + H )
=
(2 + H )
(6.117)
dx
2
u e dx
2
2 1 + bx(1 ax)
0.5
65
50
0.2
60
0.3
du /dx (1/s)
ue (m/s)
Cp (~)
0.4
55
a=0.5, b=1
a=1, b=0.5
a=1, b=2
0.1
0
0.5
50
0.5
50
0.5
337
ue
G(1) = 0.058
0.268
2
du e
=0
H
cf u e dx
H 2 + 3.179 + 1.5 2
2
G(3) =
=0
H 1
(1 + )
cf
Notice that we have three equations with three unknowns: H , cf , and . The
Matlab routine fsolve is used to compute the roots of this system of equations.
Having computed H and cf we can now evaluate the RHS of (6.117) and compute n+1 . Finally, we use this new value of in combination with H to compute
For details on the RungeKutta method the reader is directed to an introductory text on numerical
methods for differential equations, which is also covered in Ref. [30].
338
x 10
5
* (m)
(m)
x 10
5
3
2
1
0
2
1
0.2
0.4
0.6
0.8
0.2
0.4
0.6
0.8
0.01
3
2
cf (~)
H (~)
a = 0.5, b =1
a = 1, b = 0.5
a = 2, b = 2
0.005
1
0
0.2
0.4
0.6
x (m)
0.8
0.2
0.4
0.6
0.8
x (m)
Fig. 6.44 Distribution of boundary layer parameters according to three different pressure
distributions
via (6.63). We repeat this procedure at every xn until xn = 1. In Fig. 6.44 the
results are plotted for the three cases.
(c) If we look at case 1, a favorable pressure gradient is present. This favorable
pressure gradient results in an almost constant value of the shape factor. If we
compare this to the second case we see the effect of the adverse pressure gradient
on the momentum thickness and displacement thickness. We also see that the
friction coefficient becomes smaller due to a slightly higher value of the shape
factor. Remember that this corresponds to a lower velocity gradient at the wall
(see also Fig. 6.26). Finally, we have case 3 where the shape factor rises much
quicker due to the larger adverse pressure gradient. If we examine the friction
coefficient we see that it tends towards zero around x = 0.62. We know that this
is the condition of separation. In this example the flow separates from the surface.
We see an exponential increase in and and we notice that H goes to a value
close to 3.0. Beyond x = 0.62 the calculation of the roots of the functions G gives
a complex solution and is therefore neglected in the generation of these plots.
339
parameter that determines the significance of the roughness is the ratio between the
sand grain height (k) and the boundarylayer thickness. We distinguish two cases:
1. Hydraulically smooth surface. If k is smaller than the height of the viscous sublayer, i.e. k < v , k does not affect the boundary layer significantly. In other
words: the friction coefficient is unaffected by the presence of roughness. We
can imagine each sand grain to be a bluff body that is immersed in a very low
Reynoldsnumber flow. In such flow (see Fig. 6.22a) the streamlines are attached
over the complete body. The drag that results in this flow (Stokes flow) is due to
friction and therefore has the same dependency on Reynolds number as a perfectly
smooth surface.
2. Hydraulically rough surface. If k is larger than the height of the viscous sublayer, i.e. k > v , the sand grains create small wakes behind them due to flow
separation. The pressure drag that is generated at each sand grain results in
the approximate friction drag. Because the pressure drag is less dependent on
the Reynolds number than the friction drag, the friction coefficient over a rough
surface also has much lower dependency on the Reynolds number. The value of
the friction coefficient is dominated by the relative size of the sand grains with
respect to the boundary layer thickness: k/.
If we assume the sand grains to have an average height of y = k, we can introduce
a nondimensional sand grain height according to (6.105):
k+ =
ku
(6.118)
From experiments (e.g. Clauser [10]) it is known that the layer where velocity changes
logarithmically with wall distance still exists when roughness is present. However,
the intercept of the logarithmic curve moves downward with progressively larger
values of k + . This alters (6.108) as follows:
u+ =
1
ln y + + B B(k + )
(6.119)
1
ln(1 + 0.3k + ) for k + > 60
(6.120)
The result is the following expression for the nondimensional velocity profile in a
boundary layer that is disturbed by roughness:
u+ =
1
1
ln y + + B ln(1 + 0.3k + )
(6.121)
340
From the experiments it is know that for values below k + < 4 roughness has no
effect on the velocity profile in the boundary layer. Between 4 < k + < 60 there
is a transitional roughness regime. Finally, above k + > 60 the flow has become
hydraulically rough. Based on (6.121) it can be shown that as the sand grains become
larger, the effect of viscosity becomes negligible and the friction coefficient therefore
becomes independent of the Reynolds number (see Problem 6.23):
u+ =
1 y
+ 7.9
ln
(6.122)
We can insert (6.121) into (6.83a) under the assumption of an absent pressure
gradient. Integrating the result over the thickness of the boundary layer and the length,
x, results in a relation between the friction coefficient, cf , the Reynolds number, Rex
and the grain size, k [66]:
Rex = 1.73(1 + 0.3k + )e Z Z 2 4Z + 6
0.3k +
(Z 1)
1 + 0.3k +
(6.123)
where Z = . This relation, even though it is implicit, is valid for uniform sandgrain roughness over the complete range of hydraulically smooth, transitional, and
fully rough walls in turbulent flat plate flow. If we write k + = ku = Rex (k/x)
, we
can plot the local friction coefficient as a function of the local Reynolds number for
various values of x/k. This is shown in Fig. 6.45. The effect of the roughness is evident
from this figure. The larger the grains compared to the length x the higher the friction
coefficient. In addition, we see that the friction coefficient of a smooth plate decays
with Reynolds number (as expected). However, we also see that the decaying effect
of the Reynolds number diminishes for roughened plates above a certain Reynolds
number. The smaller the sand grains compared to the size of the distance x, the larger
the Reynolds number at which the Reynolds number ceases to decrease the friction
coefficient. This Reynolds number is called the cutoff Reynolds number. This is in
10
x
=10
k
10
100
1,000
104
105
106
10
10
10
10
10
10
341
.0030
Critical
values
l/k s
.0025
JETSTAR
C141
.0020
3.2 x 10
6.5 x 10
9.1 x 10
C5
.0015
.0010
20
30
40
50
60
80
100
200
good agreement with classical experiments carried out by Nikuradse on pipe flow
with roughened walls [42].
The results plotted in Fig. 6.45 can be directly translated to the friction drag coefficient of full scale aircraft and wind tunnel models. In Fig. 6.46 the variation in friction
coefficient with Reynolds number is shown for three airplanes of the same generation
and built by the same company (Lockheed). It could be assumed that the manufacturing techniques for all three airplanes was of the same level leading to identical
values for the equivalent sand grain roughness of the skin of the airplane. What is
shown is the relative sand grain height, scaled with respect to a characteristic length,
l, of the airplane. Similar to the theoretical curves in Fig. 6.45, the largest airplane
(C5) demonstrates to have the largest cutoff Reynolds number (100 million). This
advocates the statement that large aircraft benefit from their size when it comes to
the friction drag coefficient.
Example 6.5 Assume we have a flat plate with an equivalent sand grain roughness of
0.010 mm. Calculate using (6.123) the friction coefficient (cf ) at x = 2.5 m. Assume
that we have 0 m ISA conditions and a freestream velocity of 100 m/s. Also calculate
the wall stress, w and the velocity gradient at the wall, (u/ y)w .
Solution:
We first calculate the Reynolds number at x = 2.5 m:
Rex=2.5 =
17.89 106
342
Because (6.123) cannot be solved directly, we use a for loop in Matlab according to
the following routine. We first establish a set of possible values for . Subsequently,
for each value of we calculate the value for k + , Z , and Rex :
for
= [1, 1.001, 1.002, . . . , 50]
do
Rex (k/x)
k+ =
Z =
+
0.3k +
Z 4Z + 6
(Z 1)
1 + 0.3k +
end
Finally, we compare the outcome of Rex to the value that we computed above:
Rex = Rex=2.5 Rex 
We subsequently find one value of for which Rex is closest to zero. This results
in = 26.3. Substituting this in (6.110) results in:
cf =
2
= 0.00290
2
From the friction coefficient we can calculate the local shear stress according to
(6.55):
1
1
w = cf U02 = 0.00290 1.225 1002 = 17.7 N/m2
2
2
The local velocity gradient can be found by employing (6.54):
u
y
=
w
m/s
w
= 0.994
Example 6.6 Considering the same roughened plate as in Example 6.5, calculate the
value of the total friction coefficient over the entire plate. Assume the plate has a
length of l = 3 m and that the boundary becomes turbulent at x = 0.10 m. Ignore
the contribution of the laminar boundary layer over the first 10 cm to the friction
coefficient.
Solution:
The friction coefficient of the plate is given by (6.66) with the upper bound x = 3 and
the lower bound x = 0.10. We evaluate this integral numerically by dividing the plate
343
x+x
x
x
cf (x)dx = cf (x) + cf (x + x)
2
The total friction drag over the plate is the sum of the incremental contributions:
1
C f j = 0.0030
l
29
Cf =
j=1
dp
d T u
u
2u
+
fx
= v0
2
y
y
dx
dT y y
(6.124)
Note that we have split the partial derivative / y(u/ y) and have used / y =
(/T )(T / y). The lefthand side of (6.124) is a measure for the curvature of the
boundary layer profile near the wall. Generally speaking, a fuller boundary layer
corresponds to a convex velocity profile, i.e. a more negative value of 2 u/ y 2 . A negative pressure gradient (d p/dx < 0) therefore results in lower disturbance growth,
while an adverse pressure gradient (d p/dx > 0) increases disturbance growth. One
can also heat or cool the wall. Wall cooling in air (T / y > 0, /T > 0) stabilizes the boundary layer, while heating it (T / y < 0, /T > 0) destabilizes the
boundary layer. Alternatively, one can apply suction perpendicular to the boundary
layer (v0 < 0) in order to reduce the growth of disturbances. This has been successfully demonstrated to postpone transition in the wind tunnel and in flight [7, 44].
Finally, one can apply a body force to the flow in the direction of the velocity vector.
This can be done by locally ionizing the air using an electric field. By employing
an asymmetric electrode arrangement, a body force can be exerted on the resulting
plasma [13]. It has been shown that a periodic body force is effective in attenuating
the TollmienSchlichting waves [29]. Both suction and plasma control are examples
of active laminar flow control (LFC).
344
In the early 20th century William Orr and Arnold Sommerfield independently
derived the linear homogeneous ODE describing the infinitesimal laminar flow
instability based on the steady continuity equation (2.113) and the unsteady linear
momentum equation (2.126). In contrast to our earlier assumption, Orr and Sommerfield allowed infinitesimal velocity perturbations perpendicular to the flow direction
within the boundary layer. They assumed that these perturbations changed harmonically with time and could grow exponentially in time and space depending on the
frequency of the wave, the wave propagation speed, the kinematic viscosity, and
the shape of the velocity profile. For negligible viscosity, the solution of the OrrSommerfield equation showed that only velocity profiles with an inflection point in
the velocity profile would amplify the instabilities. Additionally, the second derivative of the velocity with respect to y (distance perpendicular to the surface) needed
to be smaller than zero. This led to the mistaken assumption that a favorable pressure gradient could stabilize the boundary layer up to an infinite Reynolds number.
Prandtl demonstrated, however, that viscosity can indeed be destabilizing for certain
wave numbers at a finite Reynolds number.
A more elaborate discussion of the the stability of laminar boundary layers can
be found in the text of White [66]. A selection of the results for boundary layers over
smooth flat plates is listed below:
1. The minimum or critical Reynolds number for initial instability is Rex crit
91,000.
2. The smallest unstable wavelength is min 6. Thus, unstable TollmienSchlichting waves are long compared to boundary layer thickness.
3. Compared to Rex crit 91,000, the point of final transition to turbulence is at
about Rex tr 3 106 , or about 30 times further downstream.
In practice, the transition process is influenced by many factors: surface roughness,
freestream disturbances, pressure gradients, vibration, sound, etc. Therefore, the
exact Reynolds number (or position, for a given fluid and speed) can vary with time
in a random way. Placing objects in the boundary layer results in a swift amplification
of the instabilities in the boundary layer and hastens the onset of transition. Such trip
strips are often used in wind tunnel experiments to fix the point of transition at a fixed
percentage of the wing chord. Other sources of instabilities stem from the attachment
line of the boundary layer on the leading edge of objects that are placed oblique to
the flow (such as the leading edge of a swept wing). In addition, instabilities arising
from threedimensional cross flow (also found on swept wings) hasten the onset of
laminar to turbulent transition. This will be further discussed in Chap. 8 when we
treat the aerodynamics of swept wings.
In the derivation of the critical Reynolds number a Blasius6 profile was assumed
for the velocity distribution, u(y), in the boundary layer. However, it was shown that
the critical Reynolds number can be 90 times higher than this value when the velocity
6 The Blasius velocity profile satisfies the nonlinear ordinary differential equation that can be derived
from the boundary layer equations under specific assumptions, see (6.90a).
345
y
u (y)
dp
> 0 (adverse)
dx
Point of
inflection
dp
=0
dx
dp
< 0 (favorable)
dx
profile in the boundary layer satisfies the condition: 2 u/ y 2 < 0. Such velocity
profiles are very stable and postpone transition to a much higher Reynolds number.
Such a stable velocity profile can be formed when a favorable pressure gradient
is present as is shown in Fig. 6.47. The boundary layer profile that is subjected to
an adverse pressure gradient shows an inflection point, which means that 2 u/ y 2
changes sign in this point. In this case the critical Reynolds number (and the transition
Reynolds number) is decreased.
The critical Reynolds number can be shown to correlate to the shape factor of the
boundary layer at the point where instabilities start to occur. The shape factor can,
in turn, be calculated by the velocity distribution in the xdirection outside of the
boundary layer (as was shown through Thwaites method on pp. 305306). In Fig. 6.48
this correlation is experimentally demonstrated through a range of experiments. Both
the critical Reynolds number as well as the transition Reynolds number are shown.
10 9
10 8
10 7
Re x, tr
10 6
Re x, crit
10 5
10 4
10 3
2.0
2.2
2.4
2.6
*
Shape factor, H =
2.8
(~)
3.0
346
The curve that is fitted through transition data points corresponds to the following
curvefitted equation [64]:
log10 (Rex, tr ) 40.4557 + 64.8066H 26.7538H 2 + 3.3819H 3
(6.125)
By using Thwaites method for the calculation of the local shape factor and combining that with (6.125), a swift prediction of the transition Reynolds number can be
obtained.
Example 6.7 Consider the flow problem of Example 6.3. Use the criterion of (6.125)
to calculate where transition is expected to occur.
Solution:
We need to calculate the local Reynolds number at each xposition, Rex . Subsequently, we need to evaluate whether the following inequality holds:
Rex < Rex, tr
We employ (6.125) to compute the transition Reynolds number, Rex, tr . We have
tabulated these values up to the position where separation is predicted (Table 6.2).
We can observe that the inequality holds until x = 70 mm. This means that transition
takes place between 60 and 70 mm. Note that the boundary layer calculations beyond
this point should therefore consider a turbulent boundary layer.
Another popular method to predict boundary layer transition is based on linear
stability theory (LST). The e9 method as pioneered by van Ingen [60] and Smith and
Gamberoni [52] assumes that transition occurs when the most unstable TollmienSchlichting waves are amplified by a certain factor. Here, e 2.71828 is the base
of the natural logarithm. The transition amplification factor is usually taken as e9
8,100, although a different value than 9 can be assigned to the exponential depending
on the disturbances in the freestream flow [61]. Therefore, the method is currently
referred to as the e N method, where N depends on the turbulence intensity in the
free stream. van Ingen [61] uses the following relating between the value of N and
the turbulence intensity, Tu:
N = 2.13 6.18 log10 (Tu) for 0.1 % < Tu < 2 %
(6.126)
40
50
60
70
80
90
Rex 106
() 0.00 0.07 0.13 0.20 0.26 0.32 0.38 0.44 0.49 0.55 0.60 0.66 0.71
Rex, tr 106 () 4.59 3.72 2.91 2.19 1.55 1.02 0.61 0.32 0.14 0.05 0.02 0.01 0.00
347
factor at the beginning of transition. For turbulence levels below Tu < 0.1 % the
experimental data shows too much scatter to warrant a simple relation as (6.126).
The turbulence intensity (in percentages) is defined as follows:
Tu = 100
RMS(u )
u
(6.127)
dN
(Hk ) Re Re0 (Hk )
dRe
(6.128)
dN
dRe
dN
= 0.01 [2.4Hk 3.7 + 2.5 tanh(1.5Hk 4.65)]2 + 0.25 (6.130)
dRe
1.415
20
3.295
log10 Re0 =
0.489 tanh
12.9 +
+ 0.44
Hk 1
Hk 1
Hk 1
(6.131)
In the above equations the parameter Hk is the kinematic shape factor, which is
calculated using the following empirical relation assuming a constant density across
the boundary layer [67]:
H 0.290Me2
Hk =
(6.132)
1 + 0.113Me2
with Me being the Mach number at the boundarylayer edge and H being the shape
factor from (6.63).
The method described above predicts the onset of transition. The laminar
boundarylayer variables can be calculated using (6.59), (6.61), and (6.63). If the
outcomes are combined with the properties at the boundarylayer edge (u e , pe , Te ),
the above equations can be used to calculate the value of N . If N surpasses a predefined threshold (e.g. N = 9), the boundary layer transitions from laminar to turbulent.
Example 6.8 Assume we have curved plate of length c = 1m, that supports a pressure distribution according to:
348
Cp = 8
!
x
x
1 +1
c
c
The flow over this plate can be assumed incompressible and adiabatic. The freestream
velocity of this flow is V = 50 m/s and the plate is at mean sealevel conditions
of the international standard atmosphere (i.e. 0 m ISA). Carry out the following
assignments:
(a) Use Thwaites method to calculate the position of laminar separation and the
momentum thickness distribution.
(b) For Tu = 0.1 % and Tu = 2 % compute the logarithm of the amplification factor,
N , at which transition takes place.
(c) Use the method presented prior to this example to compute the distribution of
the logarithm of the amplification factor, N .
(d) For each turbulence level, compute the position of transition point and mark
these points in the plot of the reduced amplification factor.
Solution:
(a) We follow Example 6.3 to answer the first question. The most convenient way
to do this is to implement all the relevant equations in a spreadsheet program
or other computational software such as Matlab. We first convert the pressure
distribution to a velocity distribution by employing (6.11):
u e = V 1 Cp
Subsequently, we take the derivative of u e with respect to x and we employ
(6.101) to compute the distribution of and up to = 0.09. The resulting
separation point is (x/c)sep = 0.61.
(b) We employ (6.126) to compute that for Tu = 0.1 % we have N 8.3 and for
Tu = 2 % we have N 0.3.
(c) We need to solve (6.128). To that purpose, we implement (6.130)(6.132) into our
spreadsheet program. Furthermore, we need the momentumthickness Reynolds
number (6.129), the equation of state (2.72), and the equation for the speed
dN
, and Re0 . The resulting values are
of sound (2.106) to calculate Re , dRe
Logartithm of amplification
factor, N (~)
349
15
e9
10
Tu = 0.1%
destabilizing
Tu = 3.0%
0
stabilizing
5
xtr
10
15
0.2
0.4
0.6
0.8
x (m)
Fig. 6.49 Distribution of logarithm of amplification factor: N = ln e N belonging to Example 6.8
that small disturbances in the boundary layer are damped. This corresponds to the
pressure distribution, which features a favorable pressure gradient up to x = 0.5. We
see that when the pressure gradient becomes less negative, the value of N starts to rise
quickly. Moreover, when the pressure gradient is positive (beyond x = 0.5) a very
rapid increase in N can be seen. This implies that initial disturbances in the boundary
layer are rapidly amplified under an adverse pressure gradient and therefore cause a
quick onset of transition. When we discuss naturallaminar flow airfoils in Sect. 7.3,
we will see how an airfoil can be designed in order to sustain laminar flow over a
large percentage of the chord.
350
the vertical tail can cause interference drag. For the explanation of the concept of
interference drag it is not important whether the interfering bodies are designed to
generate lift or not.
0.25
interference
CD
ReC= 4 .105
0.20
0.15
0.10
t
c = 3t
0.05
0
Fig. 6.50 Interference drag between two streamlined bodies (after Ref. [26])
351
The example above illustrates that two bodies do not need to intersect in order
for their mutual interference to cause separation of the boundary layer. This type
of interference is mainly based on the flow outside the boundary layer, which is
essentially inviscid. However, the magnified acceleration and deceleration of the flow
outside the boundary layer also has an effect on the development of the boundary
layer, as we have seen in Sect. 6.4.2. Therefore, the local friction coefficient and
shape factor of the boundary layer are directly affected by the interference of the two
bodies. The increase in adverse pressure gradient for the bodies in Fig. 6.50 increased
the shape factor of the boundary layer which made the boundary layer more prone
to separation than without the interference of the two bodies.
The effect of interference on the boundary layer is even greater when two bodies intersect. A relevant example is shown in Fig. 6.51a where a streamlined strut
intersects with a flat plate. Both the strut and the flat plat generate a boundary layer.
The boundary layer on the plate is, generally speaking, a lot thicker than the one on
the strut due to the longer distance it has traveled and therefore the time it has had
to develop. This results in a relatively small du/dywall . At the intersection of the
two components the boundary layers join. At the leading edge (stagnation) and over
(a)
0.15
increment
CD
0.10
l
t = .43
c
sectional drag
0
0.05
20
40
strut angle,
(b)
60
(deg)
Profile NACA 0015
d
= 0.5
c
Separation
U
Separation
Fig. 6.51 Interference drag due to the intersection of a curved body with a flat wall. a Interference
between strut and wall (after Ref. [26]). b Interference causing separation between cascades and
wall (after Ref. [22])
352
the rear of the airfoil, the boundary layer is invariably retarded, which often leads to
separation (du/dywall < 0). This is illustrated in Fig. 6.51b where the interference
between airfoil cascades and the wall is shown. Near the wall the separated region
on the cascades expand more upstream. In addition, a separated region starts to form
near the leading edge of the cascades. The separated flow, in turn, causes a wake
which results in additional pressure drag. As can be seen in Fig. 6.51a the angle
between the strut and the wall is an important factor for the value of the interference
drag coefficient. When the angle between the strut and the wall becomes smaller
than 90 the interference drag also increases. This is due to increased separation in
the narrower one of the two corners.
In order to minimize interference drag between components we need to make
sure that the boundary layer remains attached at all times. The unfavorable addition
of pressure distributions should be avoided. One way to achieve this is to apply
fillets between the two surfaces. This fillet locally reduces the supervelocities at
the intersection of the two bodies. Thereby it reduces the magnitude of the adverse
pressure gradient, which shifts the point of separation further downstream or prevents
the formation of separation all together.
353
Fig. 6.52 Leadingedge fairing at the wing root of an A340300 (photo A Pingstone)
0.1
0.3
0.4
0.5
Cp = 0.0
0.2
upper surface
0.6
Cp = 0.0
lower surface
Fig. 6.53 Theoretical isobar patterns on area ruled empennage for M = 0.80 and C L h = 0.117
The first example is the wingbody fairing which smoothes the angle between the
wing and the fuselage. Without the presence of such a fairing the boundary layer
that has developed over the fuselage side is likely to separate. An example of such a
fairing is presented in Fig. 6.52.
Another example is the junction between the horizontal and vertical tail plain in
a Ttail (see Fig. 6.53). Often, an additional fairing (acorn) is positioned to reduce
the added supervelocities that are caused by interference. Adding this body allows
the designer to locally change the area distribution of the intersection such that
interference is minimized.
The position of the nacelle with respect to the wing is a third example of how
interferencedrag considerations dominate this geometric feature. The positioning
of the nacelle under the wing implies a strong interference between the pressure
distribution over the nacelle, pylon, and lower wing surface. An interesting example
in this respect is the development of the Convair 990. The specification for this
airplane included a very high cruise Mach number (Mcruise = 0.89). During early
experimentation it surfaced that supersonic flow accompanied by strong shock waves
354
were present on the inboard sides of the pylons. The drag rise7 of the airplane
coincided with the drag rise of the nacelles.
To remedy this problem and postpone the onset of drag divergence to a higher
Mach number local area ruling was applied. A representative streamtube was chosen by including (part of) the crosssectional area of nacelle, pylon and wing. This
streamtube is shown in Fig. 6.54a. A target area distribution was defined that had
Treated as solid boundaries
(a)
1.5 Rnacelle
Sect. AA
(b)
1.5 Rnacelle
max
Sect. CC
Sect. BB
1200
post
modification
1000
max
800
nacelle exit
600
added
area
400
200
0
40
80
120
160
200
240
280
320
360
400
Fig. 6.54 Local area ruling between nacelle, pylon and wing of Convair 990 (after Ref. [31]).
a Streamtube boundaries. b Area distribution prior and post modifications. c Drag curve prior and
post modification. d Proposed modifications to nacelle and pylon
7
Drag rise is a term that describes the exponential increase in drag coefficient with Mach number.
355
(c)
Change in aircraft drag
coefficient, D (cts)
Initial configuration
Area modification
.0060
.0040
.0020
0
.80
.82
.84
.86
.88
.90
.92
(d)
A
WING
LE
B
Inboard
nacelle
A
forward pylon
fairings
A
Outboard
nacelle
A
terminal
fairing
B
Sect. AA
Sect. BB
aft pylon
fairings
Wing
LE
B
Sect. AA
Sect. BB
to reduce the local supervelocities. Modifications to the crosssectional area distribution were made by thickening the pylon, modifying the nacelle aft body, and
modifying the terminal fairing [32]. This combined effort on all four pylonnacelle
combinations significantly postponed the drag rise onset and resulted in a lower drag
coefficient at the design Mach number (Fig. 6.54c). The resulting geometry of the
pylons is shown in Fig. 6.54d. Notice the typical arearuling characteristics of the
pylon: close to where the leading edge of the wing intersects the crosssectional area
of the pylon reduces. It is as if the pylon gives the flow additional space to go around
the wing.
The examples presented above illustrate the importance of considering the aerodynamic interference between the various components on the airplane. The present discussion is not limited to nonlifting components of the airplane. On the contrary, when
integrating bodies that are designed to generate forces the supervelocities are generally larger and the formation of strong shock waves in combination with separation
356
is even more likely to occur. The integration of winglets to the tip of the wing is an
example where the interference of two forceproducing bodies is likely to produce
interference drag. Creating a fillet with a large radius can prevent the onset of strong
shocks in the corner of the two lifting surfaces. For more examples on interference
drag created by lifting and nonlifting components the reader is referred to the text
of Obert [43].
6.7 Summary
There are several parts of a highsubsonic airplane that do not contribute to the
production of lift during the cruise phase of the mission (e.g. fuselage, fin, nacelle).
These surfaces do provide a significant contribution to the total drag that is produced
by the airplane. We have investigated three important sources of drag: friction drag,
drag due to separation, and wave drag.
Wave drag that is associated with the formation of shock waves. A simple method
has been presented to calculate the wave drag of an arbitrary nonlifting body. This
method, which is based on linearized supersonic potential theory, uses an equivalent
axissymmetric body with the same crosssectional area distribution as the body
under investigation (transonic area rule). It has been shown how the crosssectional
area should be distributed along the length of the body in order to minimize the wave
drag. Several examples of the application of area ruling have been presented for both
civil and military aircraft.
Friction drag is associated with shear forces that are generated in the boundary
layer between the outer flow and the airplane surface. It has been shown that the shape
of the velocity distribution in the boundary layer has an effect on the local friction
coefficient of both a turbulent and laminar boundary layer. In addition, it affects the
streamwise position at which the laminar boundary layer transitions to a turbulent
boundary layer. A turbulent boundary layer has a higher velocity gradient at the wall
and produces more friction drag than a laminar boundary layer. In addition, the turbulent boundary layer is much thicker than the laminar boundary layer which affects the
outer pressure distribution through the displacement effect. The shape of the velocity
profile in the boundary layer, in turn, is affected by the outer pressure distribution.
The shape of the body (which is responsible for the pressure distribution) is therefore
of pivotal importance for the total amount of friction drag that is generated. While
the turbulent boundary layer is less sensitive to changes in Reynolds number than the
laminar boundary layer, it is much more sensitive to changes in Mach number. The
presence of surface roughness has an effect on the transition point (shifts forward)
as well as on the local value of the friction drag (becomes larger). Additionally,
it is shown that the friction coefficient becomes smaller with increasing Reynolds
and Mach number. However, for roughened surfaces there exists a cutoff Reynolds
number above which the friction coefficient remains constant with Reynolds number.
6.7 Summary
357
(a) Show that the
continuity equation (2.109) can be rewritten according to: intake
1 u 1 dy = exhaust 2 u 2 dy.
(b) Show that (6.7) can be derived from (6.6) and the relation shown under (a).
6.2 Consider the flow about the body in the control volume of Fig. 6.1.
(a) If we assume that the velocity defect behind the wing (u 1 u 2 ) is small compared
C
to the free stream velocity, V , show that D V (V u 2 ) dy.
D
(b) Show that, under the same assumptions, the drag can also be derived from the
C
pt pt2 dy.
totalpressure defect in the wake of the wing: D
D
(c) Based on the items above, how would you measure the drag of a body at section
CD?
358
(a)
paraboloid
cylinder
paraboloid
(b)
ellipsoid
cylinder
paraboloid
(c)
paraboloid
paraboloid
6.5 In the figure below you find a sideview of a British experimental aircraft for
radar surveillance (Early Warning), the British Aerospace Nimrod AEW. Sketch the
relation of the static pressure coefficient at the lower and upper crown line of the
forward fuselage in the plane of symmetry.
1
Cp 0
6.6 (a) Derive (6.17) based on the expression for the pressure coefficient and the
isentropic relation between total pressure and static pressure.
(b) If M = 0.8 and Cp = 2, calculate the local Mach number.
(c) If M = 0.8 and Cp = 3, calculate the local Mach number.
(d) Calculate for what value of the Cp the theoretical local Mach number tends to
infinity if M = 0.8.
6.7 Summary
359
6.7 For the free stream Mach numbers ranging from 0.1 to 0.8, graph the increase in
Mach number, M, as a function of pressure coefficient. Use a range of Cp between
2 and 0.
Wave Drag
6.8 Show that (6.22) can be derived from (6.19), (6.21), and (2.106).
6.9 Consider the perturbed momentum equation in integral form (6.26).
(a) Show that, by employing the continuity equation (6.27), this equation can be
simplified according to:
F body =
!
2
pn + V
v (v n + i n) dS.
(b) Argue that the expression above is identical to (6.28) because we integrate over
an enclosed surface, S.
6.10 Consider (6.50) in terms of the Glauert variable .
(a) Rewrite (6.50) in terms of x.
(b) If R(x) is the distribution of the radius of the SearsHaack body, express R(x)
in terms of the maximum radius, Rmax , and x.
6.11 Consider the body described in Example 6.2.
(a) Calculate the volume, V , and maximum radius, Rmax corresponding to the aforementioned body.
(b) Calculate the wave drag of a SearsHaack body with the same length and volume
as the aforementioned body.
(c) Calculate the relative decrease in wave drag coefficient (in percentages) that can
be attained when modifying this body to the SearsHaack body with identical
volume and length.
6.12 The aft fuselage of the Airbus A320 exhibits a tailored waist at the location
of the horizontal tail. In addition, the vertical tail is moved forward relative to the
horizontal tail. What are the explanations for their relative positioning?
horizontal tail
vertical tail
360
6.13 The figure below reveals the large wingfuselage fairing of the Airbus A380.
This fairing is so large that it received the nickname the bathtub. Associated with it
are a large wetted area and a weight penalty. Carefully explain the reason of existence
of a wingfuselage fairing in general, despite the drawbacks mentioned.
6.14 The Convair 990 (see below) was designed to be a jet transport with a cruise
Mach number of M = 0.89. In 1958 this was quite challenging due to the limited
computational resources. Wind tunnel tests on the initial design demonstrated that the
drag coefficient was too large to attain the desired cruise Mach number. A somewhat
unconventional solution was proposed and eventually adopted in the final design.
The solution was to position large bodies near the trailing edge of the wing on the
top surface.
(a) Carefully explain how the addition of these bodies can result in a lower drag
coefficient at the design Mach number.
(b) Name two disadvantages of these bodies.
(c) Why do modern transport aircraft lack these additional bodies?
6.7 Summary
361
Photo NASA
362
6.19 Derive (6.103) by using the definition of u + and the definition of w in (6.54).
6.20 Plot the velocity profile according to the law of the wall. On the vertical axis
show u + , ranging between 0 and 25. On the horizontal axis, show y + , ranging
between 1 and 1,000.
6.21 To investigate the effect of the Coles wake paramter on the velocity profile,
plot (6.109) for = 0, 1, 2.5, 5, 10, 15. On the vertical axis show u + , ranging from
0 to 100. On the horizontal axis, show y/ between 0 and 1.0.
6.22 Show that (6.112) can be derived from (6.109) by substituting y = in (6.109).
6.23 Show that when k + we can reduce (6.121) to (6.122).
6.24 Assume we have a flat plate with an equivalent sand grain roughness of
0.020 mm. Assume that we have 0 m ISA conditions and a freestream velocity of
100 m/s.
(a) Calculate using (6.123) the friction coefficient (cf ) at x = 2 m.
(b) Calculate the wall stress, w .
(c) Calculate the velocity gradient at the wall, (u/ y)w .
6.25 Consider the plate and flow conditions of Problem 6.24 and the method of
Example 6.6. Imagine we have 5 plates with lengths: 1, 3, 10, 20, and 50 m. Assume
that on each plate the turbulent boundary layer starts at x = 0.10 m. Calculate
(a) The Reynolds number, Rel , for each plate
(b) The friction coefficient, Cf , for each plate.
(c) Make a plot of the Reynolds number versus the friction coefficient.
6.26 Consider a flat plate of 10 m long. Assume that we have 0 m ISA conditions,
a freestream velocity of 100 m/s, and that the turbulent boundary layer starts at
x = 0.10 m. We would like to evaluate the effect of various sand grain sizes: k = 0,
0.025, 0.050, 0.075, 0.10 mm.
(a) Plot the value of the local friction coefficient, cf as a function of the position x
along the length of the plate for the five roughness heights.
(b) Plot the value of the total friction coefficient, Cf , as a function of the roughness
height, k.
Interference Drag
6.27 In the figures below you see pictures of two versions of a British Navyfighter,
the Hawker, later ArmstrongWhitworth Sea Hawk. The later version, powered
by an engine with 5400 lb static thrust at sea level, has a fairing at the junction of
the horizontal and vertical tailplane. The earlier version, with a 5000 lb static thrust,
lacked this fairing. Explain the (aerodynamic) cause for the presence of the fairing
on the later version.
6.7 Summary
363
6.28 The picture below shows the Tupolev, Tu154 in landing configuration. Visible
is the pointed cone at the intersection of the horizontal and vertical tailplane. Explain
the aerodynamic function of that cone using sketches of pressure distributions.
Photo A. Pingstone
6.29 The Lockheed P38 Lightning (below) was in many ways innovative. An undesirable characteristic of this aircraft was heavy horizontal tail buffeting at high speed.
Initially it was expected that it was classical flutter of the horizontal tail itself and
could be cured using mass balances. However, that did not solve the problem at all.
Only after a fairing was added at the root of the wings leading edge, where it meets
the fuselage, did the buffet disappear. What can be concluded about the precise cause
of this phenomenon?
Photo USAF
364
References
1. Abernathy, F.H.: Fundamentals of boundary layers. In: Illustrated Experiments in Fluid
Mechanics. National Committee for Fluid Mechanics Films. Cambridge (1970)
2. Adams, R.A.: Calculus: A Complete Course, 4th edn. Addison Wesley Longman Ltd., Reading
(1999)
3. Anderson, J.: Fundamentals of Aerodynamics, 5th edn. McGraw Hill, New York (2010)
4. Ashley, H., Landahl, M.: Aerodynamics of Wings and Bodies. Addison Wesley, Reading (1965)
5. Babinsky, H., Dlery, J.: Transonic shockwaveboundarylayer interactions. In: ShockWave
BoundaryLayer Interactions, pp. 87136. Cambridge University Press, Cambridge (2011)
6. Bauer, A.B., Smith, A.M., Hess, J.L.: Potential flow and boundary layer theory as design tools
in aerodynamics. Can. Aeronaut. Space J. 16(2), 5369 (1970)
7. Braslow, A.: A History of SuctionType Laminarflow Control with Emphasis on Flight
Research. NASA, Washington (1999)
8. Chan, Y.: An Experimental Study of the Transonic Equivalence Rule with Lift. Aeronautical
report LR609. National Aeronautical Establishment. National Research Council of Canada,
Ottawa (1982)
9. Chan, Y.: An Experimental Study of the Transonic Equivalence Rule with Lift, Part II. Aeronautical report LR614. National Aeronautical Establishment. National Research Council of
Canada, Ottawa (1984)
10. Clauser, F.H.: The turbulent boundary layer. J. Aeronaut. Sci. 21, 91108 (1954)
11. Coles, D.E.: The law of the wake in the turbulent boundary layer. J. Fluid Mech. 1, 191226
(1956)
12. Coles, D.E., Hirst, E.A.: Computation of turbulent boundary layers. In: Proceedings of the
AFOSRIFP Stanford Conference, vol. 2. Stanford University, Stanford (1968)
13. Corke, T.C., Post, M.L., Orlov, D.M.: SDBD plasma enhanced aerodynamics: concepts, optimization and applications. Prog. Aerosp. Sci. 43(78), 193217 (2007). doi:10.1016/j.paerosci.
2007.06.001
14. da Dosta, A.L.: Application of computational aerodynamics methods to the design and analysis
of transport aircraft. In: Proceedings of ICAS, vol. 2, pp. 261269 (1978)
15. Das, D.K.: A simple theory for calculating turbulent boundary layers under arbitrary pressure
gradients. Int. J. Eng. Fluid Mech. 1, 8399 (1988)
16. Delery, J.: Shockwave boundarylayer interactions. In: Babinsky, H., Harvey, J. (eds.) Physical
Introduction, Chap. 2, pp. 586. Cambridge University Press, New York (2011)
17. Donlan, C.J.: An Assessment of the Airplane Drag Problem at Transonic and Supersonic
Speeds. NACARML54F16, Washington (1954)
18. Drela, M., Giles, M.B.: Viscousinviscid analysis of transonic and low Reynolds number airfoils. AIAA J. 25(10), 13471355 (1987). doi:10.2514/3.9789
19. Felsch, K. O., Geropp, D., Waltz, A.: Method for turbulent boundary layer prediction. In:
Proceedings of the Stanford Conference on the Computation of Turbdent Boundary Layers,
pp. 170176 (1968)
20. Ferri, A.: Elements of Aerodynamics of Supersonic Flows. Macmillan Company, New York
(1949)
21. Fick, A.: Ueber diffusion. Annalen der Physik 170(1), 5986 (1855)
22. Gersten, K.: AGARD Report 299: Corner interference effects (1959)
23. Gleyzes, C.J., Cousteix, Bonnet, J.L.: Theoretical and experimental study of low Reynolds
number transitional separation bubbles. In: Conference on Low Reynolds Number Airfoil
Aerodynamics, pp. 137152 (1985)
24. Goodmanson, L., Gratzer, L.: Recent advances in aerodynamics for transport aircraft. In: 9th
Annual Meeting and Technical Display, Annual Meeting. American Institute of Aeronautics
and Astronautics (1973). doi:10.2514/6.19739
25. Hicks, R.M., Hopkins, E.J.: Effects of Spanwise Variation of LeadingEdge Sweep on the Lift,
Drag, and Pitching Moment of a WingBody Combination at Mach Numbers from 0.7 to 2.94,
NASA TN D2236, Moffett Field (1964)
References
365
26. Hoerner, S.F.: Fluid Dynamic Drag. Hoerner Fluid Dynamics (1965)
27. Johnson, H.A., Rubesin, M.W.: Aerodynamic heating and convective heat transfersummary
of literature survey. Trans. ASME 71, 447456 (1949)
28. Kegelman, J.T., Mueller, T.J.: Experimental studies of spontaneous and forced transition on an
axisymmetric body. AIAA J. 24(3), 397403 (1986)
29. Kotsonis, M., Giepman, R., Hulshoff, S., Veldhuis, L.: Numerical study of the control of
TollmienSchlichting waves using plasma actuators. AIAA J. 51(10), 23532364 (2013).
doi:10.2514/1.J051766
30. Kreyszig, E.: Advanced Engineering Mathematics. Wiley, New York (1988)
31. Kutney, J.T., Piszkin, S.P.: Reduction of drag rise on the Convair 990 airplane. J. Aircr. 1(1),
812 (1964)
32. Kutney Sr, J.: The inside story of the Convair 990 the fastest subsonic airliner in the world. In:
Proceedings of AIAA Joint Propulsion Conference and Exhibit, July, pp. 118. AIAA20075338 (2007)
33. Kuzmin, A.G.: BoundaryValue Problems for Transonic Flow. Wiley, Chichester (2002)
34. Laufer, J., Vrebalovich, T.: Stability and transition of a supersonic laminar boundary layer on
an insulated flat plate. J. Fluid Mech. 9(2), 257299 (1960). doi:10.1017/S0022112060001092
35. Lees, L.: The Stability of the Laminar Boundary Layer in a Compressible Fluid. NACA TN
1360. Langley Field, Virginia (1947)
36. Liepmann, H.W.: The interaction between boundary layer and shock waves in transonic flow.
J. Aeronaut. Sci. 13(12), 103138 (1946)
37. Liepmann, H.W., Roshko, A.: Elements of Gas Dynamics. Wiley, New York (1957)
38. Little Jr., B.H.: Advantages and problems of large subsonic aircraft. In: AGARDLS37High
Reynolds Number Subsonic Aerodynamics. AGARD (1970)
39. Lobb, R.K., Winkler, E.M., Persh, J.: Experimental investigation of turbulent boundary layers
in hypersonic flow. J. Aeronaut. Sci. 22(1), 19 (1955)
40. MacWilkinson, D.G., Blackerby, W.T., Paterson, J.H.: February correlation of fullscale drag
predictions with flight measurements of the C141A aircraftphase II. Wind Tunnel Test,
Analysis, and Prediction Techniques, NASA CR2333, Marietta (1974)
41. Moulden, T.H.: Fundamentals of Transonic Flow. Wiley, New York (1984)
42. Nikuradse, J.: Stromungsgesetze in rauhen Rohren. Forsch. Arb. Ing.Wes, p. 631 (1933)
43. Obert, E.: Aerodynamic Design of Transport Aircraft. IOS Press, Delft (2009)
44. Pfenniger, W.: Laminar flow controllaminarization. In: Special Course on Concepts for Drag
Reduction, pp. 3.13.75. AGARD R654 (1977)
45. Reynolds, O.: On the experimental investigation of the circumstances which determine whether
the motion of water shall be direct or sinuous, and the law of resistance in parallel channels.
Philos. Trans. R. Soc. Lond. Ser. A 174, 935982 (1883)
46. Rubesin, M.W., Maydew, R.C., Varga, S.A.: An Analytical and Experimental Investigation of
the Skin Friction of the Turbulent Boundary Layer on a Flat Plate at Supersonic Speeds. NACA
TN 2305. Moffett Field, California (1951)
47. Saric, W.S., Reed, H.L., White, E.B.: Stability and transition of threedimensional boundary
layers. Annu. Rev. Fluid Mech. 35(1), 413440 (2003). doi:10.1146/annurev.fluid.35.101101.
161045
48. Schlichting, H., Gestern, K.: Boundary Layer Theory, 8th edn. Springer, Berlin (1999)
49. Schlichting, H., Truckenbrot, E.: Aerodynamik des Flugzeuges. Springer, Heidelberg (1960)
50. Shapiro, A.H.: The Dynamics and Thermodynamics of Compressible Fluid Flow. Ronald Press,
New York (1953)
51. Shapiro, A.H.: Pressure fields and fluid acceleration. In: Illustrated Experiments in Fluid
Mechanics. National Committee for Fluid Mechanics Films. Cambridge (1962)
52. Smith, A., Gamberoni, N.: Transition, Pressure Gradient, and Stability Theory. Douglas Aircraft
Company, Rept. ES 26388 (1956)
53. Sommer, S.C., Short, B.J.: FreeFlight Measurements of TurbulentBoundaryLayer Skin Friction in the Presence of Severe Aerodynamic Heating at Mach Numbers from 2.8 to 7.0. NACA
TN 3391, Moffett Field (1955)
366
54. Spalding, D.B.: A single formula for the law of the wall. J. Appl. Mech. 28, 455457 (1961)
55. Tannehill, J.C., Anderson, D.A., Pletcher, R.H.: Computational Fluid Mechanics and Heat
Transfer, 2nd edn. Taylor & Francis, Philadelphia (1997)
56. Thwaites, B.: Approximate calculation of the laminar boundary layer. Aeronaut. Q. 1, 245280
(1949)
57. Torenbeek, E., Wittenberg, H.: Generalized maximum specific range performance. J. Aircr.
20(7), 617622 (1983)
58. Van Driest, E.R.: Turbulent boundary layer in compressible fluids. J. Aeronaut. Sci. 18(3),
145160 (1951)
59. Van Driest, E.R.: Investigation of Laminar Boundary Layer in Compressible Fluids Using the
Crocco Method. NACA TN 2597, Washington (1952)
60. van Ingen, J.L.: A Suggested SemiEmpirical Method for the Calculation of the BoundaryLayer Transition Region. TU Delft, Report VTH74, Delft (1956)
61. van Ingen, J.L.: The eN method for transition prediction. Historical review of work at TU Delft.
In: 38th Fluid Dynamics Conference and Exhibit, pp. 149. Seattle (2008)
62. von Krmn, T.: Uber laminare und turbulent Reibung. Zeitschrift fur Angewandte Mathematik
und Mechanik 1(4), 233252 (1921)
63. Wallace, L.E.: The Whitcomb area rule: NACA aerodynamics research and innovation. In:
Mack, P.E. (ed.) From Engineering Science to Big Science, Chap. 5. National Aeronautics and
Space Administration, History Office, Washington (1998)
64. Wazzan, A.R., Gazley Jr. C., Smith, A.: HRx method for predicting transition. J. Aircr. 19(6),
810812 (1981)
65. Whitcomb, R.T.: A study of the zerolift dragrise characteristics of wingbody combinations
near the speed of sound. NACA TR 1273 (1956)
66. White, F.M.: The stability of laminar flows. In: Viscous Fluid Flow, 2 edn. New York (1991)
67. Whitfield, D.: Integral Solution of Compressible Turbulent Boundary Layers Using Improved
Velocity Profiles, Arnold Engineering Development Center AEDCTR7842. Arnold Air Force
Station, Tennessee (1978)
68. Whitford, R.: Chapter 3: fuselage design. In: Design for Air Combat, pp. 148160. Janes,
London (1987)
69. Wickens, R.H.: Aerodynamic design of lowdrag fuselages. Can. Aeronaut. Space J. 36(4),
189201 (1990)
Chapter 7
Airfoil Aerodynamics
7.1 Introduction
Experimental research into transonic aerodynamics was initiated before the first
world war. Although initially tailored towards munition, at the end of WWI transonic
effects were showing up in the increased tip speeds of propellers. It was found that
thicker sections at the tip yielded much less thrust and a much higher drag coefficient.
Soon it was found that when these sections were made thinner, these adverse effects
were not encountered [2]. Even though the physical understanding of transonic aerodynamics was still nonexisting at the time, practical solutions to the adverse effects
were already conceived. Experimental research in transonic aerodynamics always
prevailed above the theoretical research because of the many challenges associated
with modeling of transonic flow.
With the introduction of the jet engine on transport aircraft in 1949 on the de
Havilland Comet, the operating Mach numbers for passenger aircraft increased
Springer Science+Business Media Dordrecht 2015
R. Vos and S. Farokhi, Introduction to Transonic Aerodynamics,
Fluid Mechanics and Its Applications 110, DOI 10.1007/9789401797474_7
367
368
7 Airfoil Aerodynamics
substantially. It was soon found that the profile shapes of the popular naturallaminar
flow airfoils produced significant wave drag and were therefore not suitable for this
new type of aircraft. The application of such airfoils required relatively large sweep
angles in order to increase the Mach number at which strong shock waves started to
cause excessive wave drag. Even though the swept wing solution worked well from
a cruiseperformance point of view, it posed drawbacks in the lowspeed regime as
well as in weight and aeroelastic performance. Better airfoil sections were therefore
pivotal in the development of contemporary wings on highsubsonic aircraft.
We have seen in Chap. 6 that any body submersed in a flow field produces a
pressure distribution. When the integral of the pressure distribution over the entire
body is nonzero, a resultant force is produced. We showed that for nonlifting bodies this resulting force is comprised only of the drag force. Such conditions only
apply for (axis) symmetric bodies at zero angle of attack. Whenever either of these
requirements is violated a resultant force component perpendicular to the freestream
direction results. This discrete force (often referred to as the lift) has a point of application somewhere on the body. The resultant force is now a combination of the lift
force and the drag force, as shown in Fig. 6.1. When this body is twodimensional
it reduces to an airfoil. For many highaspect ratio wings, the flow over the center
section of the wing can be treated as twodimensional. Airfoil shape therefore plays
an important role in the lift and drag characteristics of the wing.
This chapter explains the typical transonic flow characteristics about twodimensional airfoil sections (commonly referred to as airfoils). We use a historical approach to take the reader on a quest to discover the relevant aerodynamic
effects related to transonic flow about airfoils. We demonstrate how experimental
researchers such as Pearcey and Whitcomb in the late 1940s and early 1950s came
up with airfoils designed for supercritical conditions. They showed how sections
could be shaped to operate at high Mach numbers and at high lift coefficients without a large (wave) drag penalty. We present the key differences between laminarflow
airfoils and supercritical airfoils. We also discover that transonic flow about airfoils
is not limited to highsubsonic freestream Mach numbers. On the contrary, the maximum lift coefficient is shown to be highly dependent on the freestream Mach number.
Especially multielement airfoils are shown to yield lower maximum lift coefficients
due to transonic effects at Mach numbers as low as 0.2. Finally, we present the concept of transonic buffeting: a highfrequency, selfsustained oscillation of the shock
wave that severely limits the maximum lift coefficient of an airplane at its operating
Mach number and beyond.
369
thickness, t
camber
leading edge
radius
chord line
chord, c
370
7 Airfoil Aerodynamics
(a)
2.0
1.5
1.0
c p 0.5
NACA 0006
M = 0.2000
Re = 10E6
=6
c l = 0.6821
cmc/4 = 0.014
(b)
2.0
1.5
1.0
c p 0.5
0.0
0.0
0.5
0.5
1.0
1.0
NACA 0018
M = 0.2000
Re = 10E6
=6
cl = 0.7020
cmc/4 = 0.000
Fig. 7.2 Comparison of pressure distribution between two airfoils of different thickness as simulated by Xfoil version 6.94. a NACA 0006. b NACA 0018
have been plotted in Fig. 7.2. The aerodynamic coefficients of each of the airfoils are
displayed next to the pressure distribution.
Let us start with the similarities between the two airfoils. First of all, they are
both symmetrical, which means their camber is identical. Classical airfoil theory
tells us that the lift curve slope for a symmetrical airfoil should equal dcl /d = 2.
Since we know that for this airfoil cl = 0 at = 0 we expect cl to be close to 0.66
for = 6 . Comparing that to the value shown in Fig. 7.2a, b we see that the lift
coefficient for both airfoils are close to this value, meaning that both airfoils generate
almost the same amount of lift. Secondly, we look at the moment coefficient about
the quarterchord point. Assuming classical airfoil theory this value should be zero
for both airfoils. We see that this is indeed the case for the NACA 0018, but that the
NACA 0006 displays a small noseup pitching moment about the quarter chord.
We have established that both profiles have a similar overall effect on the lift
and moment coefficient of the airfoil. Yet their pressure distributions are somewhat
different. The thinner profile of the NACA 0006 causes a much higher suction peak
over the leading edge of the airfoil. This large suction peak is followed by a relatively sharp adverse pressure gradient that gradually lessens in magnitude further
downstream. The thick airfoil is much more blunt. The radius of curvature at the
nose of this airfoil is therefore larger resulting in a lower suction peak. The lower
suction peak is followed by a shallower adverse pressure gradient. The magnitude of
the suction peak can directly be translated to the magnitude of the supervelocities.
A high suction peak yields high supervelocities that can become supersonic at relatively low freestream Mach numbers. At M = 0.2 the critical pressure coefficient is
16.3 according to Eq. (3.42).
In the previous chapter we have seen that the value of the adverse pressure gradient in combination with Reynolds number determines whether or not the boundary
layer separates. Even though that information cannot be extracted from the analysis
carried out above, the shape of the pressure distribution can tell us qualitatively where
we should expect flow separation to start. For the thinner airfoil the large adverse
pressure gradient near the leading edge will become so large with increasing angle of
371
1.5
M = 0.74
M = 0.68
M = 0.2
0.5
0.5
NACA 0012
(b)
1.5
pressure coefficient, C p (~ )
(a)
pressure coefficient, C p (~)
attack that the boundary layer is likely to separate from the surface close to the leading edge (even though the boundary layer itself is relatively fresh). Leading edge stall
is therefore to be expected on this profile. The thicker airfoil shows a more shallow
adverse pressure gradient which is likely to separate the flow starting at the trailing
edge, where the boundary layer shape factor (H ) has its highest value [see (6.63)
on p. 305 for the definition of the shape factor]. Typically when H is between 2.4
and 2.6 boundary layer separation occurs. With increased angle of attack this point
is likely to move forward leading to a relatively gentle onset of stall. These deductions on the stall behavior of the airfoil are purely conjectural based on the pressure
distribution. However, experimental results from Ref. [1] support these conclusions.
We will elaborate more on the separation of the boundary layer in lowspeed and
highspeed conditions in Sects. 7.5 and 7.6, respectively.
The effect of compressibility on the local pressure coefficient has been discussed
qualitatively in Sect. 3.2. The PrandtlGlauert compressibility correction magnifies
the value of the pressure coefficient. It has been shown that this correction (or derivatives thereof) can accurately predict the value of the pressure coefficient by simply multiplying this correction factor to the pressure coefficient obtained from an
incompressibleflow calculation (see Chap. 3). However, when Mach numbers are
increased beyond Mcrit the compressibility correction loses its accuracy. In Fig. 7.3
it is shown why this correction cannot be applied in transonic conditions. The same
airfoil (NACA 0012) is evaluated with two different prediction tools for the same
freestream conditions (a constant angle of attack and varying Mach number). The
prediction in Fig. 7.3a is based on a vortex panel method (Xfoil) with an embedded
compressibility correction. We see that the pressure distributions at low Mach numbers are merely amplified. The graphs in Fig. 7.3b are generated using an Euler solver
M = 0.74
M = 0.68
M = 0.2
0.5
0.5
NACA 0012
1
0
0.2
0.4
0.6
0.8
0.2
0.4
0.6
0.8
Fig. 7.3 Comparison between results from two prediction tools for the flow over a NACA 0012
airfoil with = 2.1 and Rec = 10 million. a Vortex panel method + compressibility correction
(Xfoil v. 6.94). b Euler approximation (MES v. 3.04)
372
7 Airfoil Aerodynamics
(MSES2 ). The graphs of Fig. 7.3b show a distinct change in pressure distribution
when the Mach number is M = 0.68 or higher. We see how the adverse pressure gradient of Fig. 7.3a is replaced by a more gradual increase in negative C p in Fig. 7.3b.
The steep drop in (negative) C p is indicative of a shock wave, which is also absent in
Fig. 7.3a. This comparison demonstrates the limitation of compressibility corrections
in the transonic regime. We will discuss this change in pressure distribution in more
detail in Sect. 7.4.
MSES is a viscousinviscid analysis program that couples the numerical solution of the Von Krmn equation (6.89) in the boundary layer to the numerical solution of the steady state conservative
Euler equations (2.183) outside the boundary layer. More information can be found in Refs. [10, 15].
3 This prediction is carried out by Xfoil 6.94 using an e N method for transition prediction with
Ncrit = 10.
373
(b)
0.5
NACA 2412
x 10
lower surface
0
0.2
0.4
0.6
0.8
0.5
x 10
upper surface
4
0.5
0.5
(a)
upper surface
lower surface
0.2
0.4
0.6
0.8
Fig. 7.4 Predicted pressure and shape factor distribution for cl = 0.3, M = 0.2, and Rec = 6106 .
a NACA 2412 at = 0.4 . b NACA 661 212 at = 0.9
both of them produce the same section lift coefficient. If each airfoil would be used
in a separate wing, the structural weight of each wing for a given planform shape
would be virtually identical. Even though the integrated difference of the pressure
coefficient between upper and lower surface is identical for both wings, their pressure
distributions are quite different. The thickest point of the 2412 airfoil is located
(by default) at 30 % chord. On the upper surface we see the start of the adverse
pressure gradient at approximately x/c = 0.2. Transition to a turbulent boundary
layer occurs around x/c = 0.45. On the lower surface the adverse pressure gradient
is slightly steeper and transition occurs closer to the leading edge than on the upper
surface. Notice that in this simulation transition can be seen in the jump in the friction
coefficient. This indicates where the boundary layer has transitioned from laminar to
turbulent. In this particular simulation the step change in displacement thickness that
is caused at the transition point can also be seen in the pressure distribution curve by
means of a little bump.
The thickest point of the 661 212 has been shifted beyond the 50 % chord. This
creates an almost flat plateau in the pressure distribution over the suction side of the
airfoil up the 60 % chord. On the lower surface there is a favorable pressure gradient
pressure up to x/c = 0.6. Transition occurs on both sides beyond the 60 % chord.
Notice the effect of the favorable pressure gradient on the lower side: transition is
shifted more aft. We know that the average friction coefficient, C f is correlated to
the integral of the local friction coefficient, c f , over the chord length. The total area
under the two lines in the frictioncoefficient diagram are therefore representative
for the total amount of friction drag of the profile. Based on that knowledge we can
directly see that the total friction drag produced by the 661 212 is considerably lower
374
7 Airfoil Aerodynamics
.024
NACA 2414
NACA 661212
.020
.016
with roughness
.012
.008
smooth
.004
drag bucket
0
1.6
1.2
0.8
0.4
0.4
0.8
1.2
1.6
than the friction drag produced by the 2412. The predicted drag coefficient is 0.0031
or 31 drag counts4 for the 661 212 against 51 counts for the 2412.
Comparison with experimental data of Ref. [1] shows that drag coefficient is
somewhat underestimated. In Fig. 7.5 the experimental wind tunnel measurements
are presented for a Reynolds number of 6 million. It can be seen that at cl = 0.3 the
drag coefficients is around 34 counts for the 661 212 versus 63 counts for the 2412.
We notice a 45 % decrement in drag when switching from a 2412 to a 661 212. The
lifttodrag ratio of the airfoil is increased with more than 80 % from 48 to 88.
As can be seen from Fig. 7.5, the naturallaminarflow (NLF) airfoil has a limited
range of lift coefficients where the friction drag is low. As a matter of fact, the
designation of the airfoil already shows this:
position of minimum pressure
thickness in percent chord
661 212
design lift coefficient in tenths
lowdrag range of lift coefficient in tenths
The design lift coefficient is 0.2 and we can expect a low drag coefficient (also known
as the drag bucket) between cl = 0.1 and cl = 0.3. Outside of the drag bucket the drag
of the NLF airfoil is either close to, or higher than the drag coefficient of the NACA
2412. In addition, when the skin is roughened (using NACA standard roughness) the
drag coefficient at cl = 0.3 increases to 95 cts while the drag of the 2412 increases to
100 cts, making the difference between the two airfoils very small. This illustrates the
sensitivity of both profiles to skin roughness. In practice, roughness can stem from
irregularities on the skin, rivets, or waviness in the skin surface. In addition, steps
4
375
cd
M
cl =constant
= 0.10
(7.1)
This exponential increase in drag was attributed to the formation of strong shock
waves on the wing surface in combination with separation at the shock foot. It
was found that the natural laminar flow airfoils produced a significant amount of
wave drag whenever they were exposed to Mach numbers beyond the critical Mach
number (see Fig. 7.6a). This limited the maximum Mach number of early jet fighters
(b)
1.5
1.5
c l = 0.7
c l = 0.5
c l = 0.3
(a)
Cp, crit
0.5
0
c l = 0.7
c l = 0.5
c l = 0.3
0.5
Cp, crit
0.5
0.5
NASA SC(2) 0412
1
0
0.2
0.4
0.6
0.8
0.2
0.4
0.6
0.8
Fig. 7.6 Predicted pressure distribution for cl = 0.3, cl = 0.5, and cl = 0.7 at M = 0.75,
Rec = 20 106 . a NACA 661 212. b NASA SC(2) 0412
376
7 Airfoil Aerodynamics
that used NACA 6series airfoil sections. One effort to tackle this problem was
to investigate whether airfoil shapes could be developed that would have a higher
critical Mach number in combination with lower wave drag in supercritical (i.e.
Mlocal > 1) conditions. In the United Kingdom it was H.H. Pearcey that investigated
new airfoil shapes that would allow for supersonic flow on the upper surface of the
airfoil without a strong terminating shock wave [35]. In the Unites States under the
supervision of R.T. Whitcomb a new family of supercritical airfoils was developed
specifically designed to produce low wave drag at high Mach numbers [17]. The
pressure distribution of one of these airfoils is shown in Fig. 7.6b.
If we compare the pressure distributions between the NLF airfoil and the supercritical (SC) airfoil of identical thickness, we see a large difference on both the upper
and lower surface. In the graph for the NLF airfoil we already recognize the presence
of a shock wave at cl = 0.3 (remember that this cl is still the lowspeed design range
for this airfoil). We recognize the shock wave by the sharp increase in C p close to
x/c = 0.6. If we look at the SC airfoil we see that the suction over the upper surface
is much lower at cl = 0.3. Most of the lift for this airfoil stems from the difference in
pressure on the aft part of the airfoil. The concave shape of the lower surface causes
an increase in pressure on the lower surface which makes a large contribution to the
total lift coefficient. We call this aft loading and it is one of the means to lower the
required supervelocities on the upper surface of the airfoil for a given lift coefficient
(see Problem 7.2). When the lift coefficient is increased to 0.5 the shock wave on
the NLF airfoil increases in strength (larger pressure rise). On the SC airfoil we also
see the formation of a shock wave around x/c = 0.35. However, the latter shock
wave is much weaker than the one over the NLF airfoil as can be observed from the
smaller pressure increase. Finally, at cl = 0.7 the shock over the NLF airfoil has
shifted forward, while its strength has remained approximately the same. However,
the trailingedge C p has now become substantially more negative, indicating that
the boundary layer has separated from the trailing edge. On the SC airfoil we see
that the shock has moved aft and has increased in strength. There is no indication of
boundary layer separation.
In Fig. 7.7 the corresponding curves for the drag coefficients are shown. Notice
that the drag coefficient (on the vertical axis) has been expressed in terms of drag
counts. The drag coefficient consists of two fundamental components: the friction
drag coefficient, cd f , and the pressure drag coefficient, cd p . In this analysis the pressure drag coefficient has been found by integrating the pressure over the section and
projecting the resulting force vector onto the direction of the flow.5 If we compare
the drag coefficient development between the two airfoils, we immediately see that
the NLF airfoil shows a sharp increase in drag coefficient between cl = 0.3 and
cl = 0.5. Since the friction drag is almost constant, this increase in drag is almost
solely due to the formation of a strong shock wave on the upper surface. The SC
airfoil shows almost a constant value for the total drag coefficient over this range. At
cl = 0.5 the drag of the SC airfoil is therefore 37 % lower than for the NLF airfoil.
At cl = 0.7 the shock has become significantly stronger, which results in an increase
5
This calculation procedure for obtaining the pressure drag is often called a nearfield analysis.
160
140
d
df
cd
180
120
100
80
60
40
df
140
dp
120
100
80
60
40
20
0
0.3
160
(b)
180
(a)
377
20
0.4
0.5
0.6
0
0.3
0.7
0.4
0.5
0.6
0.7
Fig. 7.7 Predicted drag coefficients at M = 0.75, Rec = 20 106 . Note cd p indicates pressure
drag, while cd f indicates friction drag. a NACA 661 212. b NASA SC(2) 0412
in pressure drag. However, the flow remains attached and the pressure drag does not
rise as fast as for the NLF airfoil.
For a given airfoil shape, the thickness ratio (t/c) and lift coefficient are often
the most important parameters that influence the drag divergence Mach number.
Figure 7.8 shows how for supercritical and NACA airfoils the thickness ratio influences the dragdivergence Mach number. According to this graph, for supercritical
airfoils, the dragdivergence Mach number decreases linearly according to the following statistical relation:
0.88
Mdd
t
for cl = 0.5
= 0.92 1.16
c
G805046
c = 0.5
l
0.84
Stateoftheart
Supercritical Airfoils
64A406
0.80
0.76
(7.2)
Conventional
Airfoils
0.72
NA Rockwell
NASA
NACA
Northrop
0.68
66210
64A211
64A412
65215
0.64
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
378
7 Airfoil Aerodynamics
The decrease in drag divergence Mach number can be explained by the fact that the
thicker airfoils create higher supervelocities over the upper and lower surface than
a thin airfoil. A thicker airfoil will therefore encounter critical conditions at a lower
free stream Mach number than a thin airfoil. In general, this also results in an earlier
onset of shockinduced boundarylayer separation and therefore drag divergence.
In Fig. 7.9 we show an example of the effect of lift coefficient on the drag divergence Mach number. The drag divergence Mach number in this graph is defined as
the point on the line where cd / M = 0.10. We can observe that with increasing lift
coefficient the drag divergence Mach number decreases. This can be explained by
the fact that the increase in lift coefficient is a result of higher supervelocities on the
suction side of the airfoil, which causes an earlier formation of the shock wave and
associated shockinduced separation. A rudimentary equation to compute the drag
divergence Mach number that includes both the thicknesstochord ratio and the lift
coefficient is given by Korn:
Mdd + t/c + 0.10 =
(7.3)
where is a technology factor that amounts to 0.95 for supercritical airfoil sections.
Torenbeek [44] derived a modified version of Korns equation based on empirical
data of second generation supercritical airfoils [17]:
Mdd + t/c + 0.10cl1.5 = M with M = 0.935
(7.4)
It should be emphasized that a different airfoil shape can highly influence the dependency of the drag divergence Mach number to the lift coefficient. It was already
shown by Gthert in 1944 that careful modification of a NACA 0012 could yield a
constant drag divergence Mach number of 0.78 for a lift coefficient ranging from 0
to 0.4 [16]. Blackerby and Johnson show in Ref. [7] that changing the forward 12 %
of the airfoil can have a profound effect on the drag divergence Mach number.
.040
.038
d
= 0.10
.036
cl = 0.55
.034
.032
cl = 0.50
.030
cl = 0.45
.028
.026
.70
.72
.74
.76
.78
.80
379
Compression
Shock Wave
Expansion
Sonic
boundary
M
 M
Fig. 7.10 Notional reflection of a single expansion wave emanating close to the leading edge (after
Ref. [35])
380
7 Airfoil Aerodynamics
the expansion wave. The flow behind the reflected compression wave experiences a
decrement in Mach number of equal strength. The resulting Mach number behind
this compression wave would therefore be equal to the Mach number in front of the
first expansion wave.
When the compression wave reaches the airfoil surface it is being reflected. If the
surface is flat the compression wave is reflected as a compression wave (reflection in
like sense). However, if the surface has sufficient convex curvature, the flow tangency
condition requires the streamline to bend towards the surface and the incident compression wave is reflected as an expansion wave. Alternatively, the surface can have
exactly the right amount of convex curvature such that there is no reflection from the
wall. To decelerate the flow it is beneficial to reduce the convex curvature such that
the reflected wave is also a compression wave. This has been implicitly assumed in
the notional sketch of Fig. 7.10. The flow behind this second compression wave has
a lower Mach number than the flow in front of the first expansion wave (M in
Fig. 7.10). The compression waves therefore aid in the reduction of the Mach number
further downstream over the airfoil. Because this compression process is isentropic
it is termed isentropic recompression.
In reality the single expansion wave should be replaced by an infinite number of
expansion waves that emanate from the convex surface close to the leading edge and
possibly also further downstream on the airfoil. This expansion fan is reflected from
the sonic boundary as a compression fan. The expansion waves and compression
waves are intersecting. This is visualized in Fig. 7.11 where the expansion waves are
shown as characteristics. We see the expansion characteristics as dashed lines and
the compression characteristics as solid lines. In the lower graph the Mach number
distribution is shown, where is the PrandtlMeyer angle corresponding to the local
Mach number of the flow. In this graph the upper dashed line shows the Mach
number distribution if only the expansion waves would be taken into account. Due to
the reflection of the compression waves from the sonic line and subsequently from
the crest, it can be seen how the Mach number is lowered.
In Ref. [39] it is described that the formation of a shock is the result of a coalescence
of the recompression characteristics. If the reflected compression waves coalesce,
they merge to form a shock wave. If even a short segment of the profile is straightsided (instead of convex) a shock will be formed as the result of a coalescence of
the compression waves reflected from this segment. In Fig. 7.12 we see two airfoils
and their pattern of characteristics. The airfoil in Fig. 7.12a has been designed to be
shock free. In other words, the recompression waves emanating from the sonic line
do not coalesce in the physical plane. This can only happen if the shape of the sonic
line and the flow deflection are carefully tailored to each other [39]. If only a small
disturbance travels downstream along the reflected Mach waves it will result in their
coalescence: a shock (Fig. 7.12b). This shows how carefully the airfoil geometry
must be arranged in order to obtain shockfree supersonic flow.
Pearcey demonstrated experimentally that even subsonic flows that locally accelerate up to M = 1.4 can be decelerated isentropically without the formation of a
shock. To achieve these properties the airfoil had to have the following properties [35]:
381
Sonic
boundary
Expansion from convex surface
First reflected compression
} Subsequent reflections
} Limiting characteristic
and reflection
Freestream
P
direction
Crest
Surface distribution
1.0
Fig. 7.11 Sketch of Machwave pattern in a region of supersonic flow over an airfoil and the Mach
number distribution resulting from the supersonic position of the first simple wave, 1 , and the
compressive effect, 2 (after Ref. [33])
(a)
(b)
Fig. 7.12 Pattern of characteristic waves over two airfoils (after Ref. [39]). a Shockfree. b Coalescence causes shock
(a) a sharp suction peak needs to be present at the leading edge during subsonic flow
conditions at a particular angle of attack
(b) behind the suction peak rapid deceleration has to occur in subsonic flow, and
(c) behind the leading edge the curvature distribution of the upper surface has to be
such that the expansion waves in the supersonic region, the sonic line, and the
reflected compression waves are formed behind the suction peak.
382
7 Airfoil Aerodynamics
(a)
(b)
CP
CP
1.0
1.0
CP*
CP*
0.5
0.5
inviscid theory
0.5
0.5
1.0
1.0
x/c
x/c
Fig. 7.13 Pressure distribution about a two different shockfree airfoils. a M = 0.709, = 3.6
(modified from Ref. [8]). b M = 0.765, cl = 0.58, Rec = 21 106 (after Ref. [23])
Due to the sharp suction peak and high adverse pressure gradient in the subsonic
pressure distribution these airfoils are often termed peaky airfoils. Pearcey showed
that the significant amount of isentropic recompression could result in shockfree airfoils or airfoils with a weak normal shock wave terminating the supersonic domain.
Other researchers (e.g. [8, 23, 39]) demonstrated the potential and limitations of
shockfree supercritical airfoils. Figure 7.13 shows the pressure distribution for two
of such airfoils. In both graphs the critical C p has been indicated with a dashed line.
We see that both airfoils show a smaller adverse pressure gradient allowing for a
larger supersonic bubble. According to Ref. [32] one can expect that for given values
of lift coefficient and thickness ratio, the design Mach number for shockfree airfoils
will be generally somewhat lower for the peaky designs than for the nonpeaky
designs of Fig. 7.13. However, this does not necessarily mean that these nonpeaky
airfoils also have a higher drag divergence Mach number. On the contrary, they have
been shown to be more sensitive to offdesign conditions than peaky airfoils.
We have now shown that shockfree airfoils have the possibility of producing
relatively high lift coefficients (around cl = 0.6) without producing any wave drag.
In theory this is an ideal solution. However, in practice these airfoils are hardly used
on highsubsonic transport aircraft. One of the main reasons for this is their sensitivity
to changes in the flow condition. These could stem from a change in Mach number,
angle of attack, or even Reynolds number. This is illustrated for the GarabedianKorn
airfoil of Fig. 7.13b in Ref. [21] where the airfoil is predicted6 to be shock free for
M = 0.751 and cl = 0.625 while for M = 0.750 and cl = 0.629 it is predicted to
have two shocks. In reality the displacement effect of the boundary layer also plays
6
The prediction was made by assuming the flow to be inviscid (Euler solution).
383
an important role in whether the shockfree condition can be attained. Given the fact
that the displacement thickness is dependent on the Reynolds number it is very well
possible that an airfoil shows to be shockfree in the wind tunnel while having a
weak shock during flight for the same Mach number and lift coefficient.
384
7 Airfoil Aerodynamics
P
sonic line
Q
M<1
M>1
M<1
weak shock wave
G
rounded nose
relatively flat central upper surface
B
NASA SC(2) 0412
the (flap) structure. Figure 7.14 shows the geometric characteristics of a 12 % thick
supercritical airfoil.
In addition to the geometric characteristics, Fig. 7.14 also shows key aerodynamic
properties of a supercritical airfoil. A notional boundary layer thickness has been
drawn for reference. We see that the sonic line, which separates the two flow domains,
forms a continuous boundary around the supersonic bubble. We see that the shock
wave is slightly curved. The tip of the shock (point P) is in the supersonic domain.
The curved part of the shock (oblique with respect to the local flow direction) causes
the flow decelerate to a Mach number higher than unity, causing a small supersonic
region behind this part of the shock. The flow decelerates isentropically to subsonic
conditions in this region. The sonic line therefore attaches to the shock at point Q.
Below point Q the shock is normal to the local flow direction and the flow decelerates
to subsonic conditions. If we move towards the foot of the shock (point F) we see
that the shock is formed by the coalescence of compression waves that smear the
pressure increase over a finite interval. Due to this pressure increase, the boundary
layer thickens (as is explained in Sect. 6.4.3).
Most modern jet transports employ supercritical airfoils or use some form of supercritical geometry in their airfoil design. For example, all the Airbus aircraft employ
supercritical airfoils, as do the latest Boeing aircraft (777, 787). Other Boeing aircraft (e.g. 737700 and 767) also include airfoils that are modified using supercritical
airfoil technology [4].
Example 7.1 Consider an aircraft that performs an accelerated climb from M = 0.50
to M = 0.77 while the lift coefficient of the aircraft remains constant. At 70 % semispan the wing has a DSMA523a supercritical airfoil with a thicknesstochord ratio
385
of 11 %, developed by McDonnel Douglas (see below). To investigate the aerodynamic phenomena at that spanwise location, you are asked to perform a twodimensional aerodynamic analysis for a constant lift coefficient of cl = 0.85. This
includes the pressure distributions over the airfoil at the Mach numbers M = 0.50,
M = 0.73 and M = 0.77 and the corresponding moment coefficients and drag
coefficients.
DSMA523a
Solution:
First we need to chose a suitable solver for this question. Since we know that the
flow about a supercritical airfoil in highsubsonic conditions will be mixed (both
subsonic and supersonic domains) we choose a numerical approximation of the Euler
equations. In this case we use a finitevolume approximation of the Euler equations
on an unstructured grid. A detailed description of this approximation can be found
in Ref. [9]. Since these calculations are based on the assumption that the flow is
inviscid, we will not be able to deduce any viscous effects such as separation. In
addition, the exact shock position is most likely to be somewhat inaccurate due to
the lacking shock waveboundary layer interaction (SWBLI). However, it does give
a good insight into the development of the shock wave and its position and gives
a reasonable prediction for high Reynoldsnumber flows, such as found over highsubsonic transport aircraft. It is therefore found to be a suitable tool to determine
the pressure distributions about this airfoil. The results for the prescribed conditions
are found in Fig. 7.15. Note that these plots show the input values above the graph:
Mach number and angle of attack. The output in terms of aerodynamic coefficients is
displayed within the plot. To keep the lift coefficient constant at each Mach number
we performed several iterations where we changed the angle of attack, until the result
showed a lift coefficient close to the target of cl = 0.85.
M = 0.50, = 2.05deg
M = 0.73, = 0.25deg
M = 0.77, = 0.60deg
1.5
1.5
1.5
0.5
0.5
c = 0.854
l
= 0.00336
c
d
cm, c/4 = 0.131
0.5
1
CP*
0.5
c = 0.852
l
= 0.00638
c
d
cm, c/4 = 0.163
0.5
CP *
0.5
0.5
c = 0.857
l
= 0.00384
c
d
cm, c/4 = 0.203
0.5
0.5
Fig. 7.15 Development of pressure distribution with constant lift coefficient and increasing Mach
number over a DSMA523a airfoil. Note this is the result of an inviscid calculation, hence cd = cd p
386
7 Airfoil Aerodynamics
Let us analyze the changing pressure distribution of Example 7.1. First we consider
the effect of the increasing Mach number on the pressure distribution over the airfoil
while the lift coefficient remains constant (see Fig. 7.15). In the left graph we see
the expected peaky behavior of the supercritical airfoil at lowspeed conditions. The
suction peak at the leading edge is followed by a relatively sharp adverse pressure
gradient, due to the lack of curvature on the top side of the airfoil. We also identify
the aft loading, which is caused by the concave shape of the lower side of the airfoil
over the aft 40 %c. When the freestream Mach number goes beyond the critical
Mach number, a plateau is formed in the pressure distribution. This represents the
supersonic domain of the flow and is terminated by a sharp pressure increase: the
shock wave. As we increase the freestream Mach number even more the shock
wave starts to move backwards towards the trailing edge, while the crest of the
pressure distribution lowers. Notice that by only increasing the Mach number by
0.05, we now have more than 85 % of the top surface emersed in supersonic flow. Also
note that the pressure distribution over the bottom side of the airfoil hardly changes.
The flow is fully subsonic there and we only distinguish a magnification of the
pressure coefficient between M = 0.5 and M = 0.73 due to compressibility effects.
Furthermore, we see that the drag coefficient decreases when the Mach number
increases from 0.73 to 0.77. This large decrease is rather unrealistic and is attributed
to the absence of viscosity in the flow solver. Finally, we would like to point out
the rapid change in moment coefficient about the quarter chord point. In subsonic
conditions we would expect the moment coefficient to stay constant about this point.
It is clear from the values for cm, c/4 that the nosedown pitching moment increases
substantially. This phenomenon is often referred to as Mach tuck or tuck under. To
compensate for this and balance the aircraft around its centerofgravity, we need to
apply a larger noseup pitching moment. For a conventionally configured airplane
this is achieved by providing more downforce on the horizontal stabilizer, which
increases the trim drag.
Example 7.2 Consider an aircraft that is flying at a constant Mach number of M =
0.732. The airfoil at 70 % semispan is again the DSMA523a supercritical airfoil.
A coordinated turn (no sideslip) at constant Mach number and altitude requires a
change in lift coefficient of the wing, and therefore on the wing section. To investigate
the aerodynamic phenomena, you are asked to do a twodimensional analysis. This
includes the calculation of the pressure distribution about the DSMA523a airfoil at
a constant Mach number of M = 0.73 and the following lift coefficients: cl = 0.5,
cl = 0.9, and cl = 1.1. In addition, you are asked to calculate the moment coefficient
about the quarter chord and the wave drag coefficient.
Solution:
Having done the previous example (Example 7.1), we use the same numerical approximation of the steady Euler equations, keeping in mind that we assume the flow to be
inviscid. We evaluate the airfoil at the prescribed lift coefficients. However, since the
lift coefficient is an output of the calculation (by summing the pressure difference
387
M = 0.73, = 1.50deg
M = 0.73, = 0.30deg
M = 0.73, = 1.00deg
1.5
1.5
1.5
0.5
Cp
0.5
cl = 0.498
c = 0.0018
d
cm, c/4 = 0.178
0.5
1
0.5
0.5
0
cl = 0.889
c = 0.00754
d
cm, c/4 = 0.161
0.5
0.5
cl = 1.08
c = 0.0163
d
cm, c/4 = 0.163
0.5
1
0.5
Fig. 7.16 Development of pressure distribution with increasing lift coefficient and constant Mach
number over a DSMA523a airfoil. Note this is the result of an inviscid calculation, hence cd = cd p
between the suction and pressure surface) we need to do this iteratively by varying
the angle of attack in the input file. The result is displayed in Fig. 7.16.
In the example above we can distinguish the effect that changing lift coefficient
has on the pressure distribution over the wing. We see that for a relatively low lift
coefficient, the main contribution to the lift coefficient comes from the aft loading of
the airfoil, which, in turn, produces quite a large nosedown pitching moment about
the quarterchord point. When the lift coefficient increases (through the change in
angle of attack), we see that a plateau is formed in the pressure distribution. This
plateau rises and becomes longer when the lift coefficient is even further increased. At
the same time we see that the shock wave becomes stronger, yielding more wave drag
with higher lift coefficient. A quick analysis of the results shows that the average
lift curve slope of this airfoil at M = 0.73 is 14.1(1/rad), which is more than
twice the value of what we expect from this airfoil at low speeds, which is around
2 = 6.28 (1/rad). This demonstrates that in the present conditions this airfoil is
much more responsive to changes in angle of attack, than at low speeds. In practice
this means that atmospheric gusts can have a relatively large impact on the pressure
distribution over the wing and can hence introduce large fluctuations in wing lift. In
the cabin we feel this as if the aircraft is driving on a road dotted with potholes. We
will see in Chap. 8 that the introduction of sweep reduces the liftcurve slope of the
aircraft and, therefore, partially compensates for this large sensitivity.
The results shown in Examples 7.1 and 7.2 are predictions of an inviscid Euler
code. In Fig. 7.17 a Mach and cl sweep is shown about a supercritical airfoil. Both
experimental results obtained in a wind tunnel (Rec = 14.5 106 ) and numerical results obtained by an Euler code with boundary layer approximation (viscousinviscid prediction) are shown. If we look at the wind tunnel test results we see
a similar behavior as in the results of the aforementioned examples. With increasing Mach number and constant lift coefficient the shock wave moves aft while the
minimum pressure coefficient decreases somewhat. When the Mach number is kept
388
7 Airfoil Aerodynamics
inviscid calculation:
2.0
Cp
2.0
Cp
2.0
Cp
M =0.499
CL =0.770
1.5
1.5
M =0.690
CL =0.927
1.5
1.0
1.0
1.0
0.5
0.5
0.5
20
0.5
40 60
x/c (%)
80
0
20
0.5
1.0
40 60
x/c (%)
80
M =0.732
CL =0.898
20
40
60
80
0.5
1.0
1.0
const.
const. CL
1.0
M =0.732
CL =0.645
0.5
Cp 0
Sym
Ref.
RC x 10
14.5
Douglas / Garabedian
14.5
20
Lift
coefficient
sweep
60
80
1.0
1.0
0.5
Cp 0
DSMA 671
40
0.5
6
0.5
1.0
20
40 60 80
x/c (%)
M =0.732
CL=0.499
Fig. 7.17 Mach number and cl sweep for a 14 % thick supercritical airfoil (after Ref. [22])
constant while the lift coefficient is increased, a shock wave starts to form. The
plateau in the lift distribution rises with increasing lift coefficient, while the shock
position remains almost constant.
Finally, we note that the prediction of the pressure distribution of the viscousinviscid flow solver is quite accurate. Both position of the shock and the supercritical
C p is accurately predicted. We see in the top right graph of Fig. 7.17 an Euler prediction (inviscid) for two conditions: constant and constant cl . This means that the
equations are being solved respectively for the same and cl as in the wind tunnel
experiment. We see that the Euler prediction is unable to properly capture the position
of the shock wave. Due to the absence of the displacement thickness of the boundary
layer, the higher effective curvature of the upper side of the airfoil generates a higher
supercritical Mach number which pushes the shock wave more aft. For the same
lift coefficient (constant cl ) and higher Mach number the airfoil is positioned under
a lower angle of attack, resulting in an underprediction of the supercritical Mach
number. For the evaluation at constant angle of attack, the prediction is even more
off: the supercritical speeds are over predicted and the shock wave is too far aft. This
combination results in an over estimation of the lift coefficient. The large discrepancy
between the Euler prediction and the measured data is evidence that the inclusion
of the boundary layer is important to accurately capture the pressure distribution
(a)
389
.8
0
.6
2
1
0
1
2
4
6
8
10
12
.4
.2
.2
.4
NACA
.6
.2
.4
.3
(b)
Section drag coefficient, cd (~)
.5
.6
.7
.8
.9
1.0
2
1
0
1
2
4
6
.04
.02
NACA
0
.2
.3
.4
.5
.6
.7
.8
.9
1.0
Fig. 7.18 The measured variation of section properties with Mach number of a NACA 64A010
airfoil for various angles of attack [18]. a Lift coefficient. b Drag coefficient
about supercritical airfoils. The results of Examples 7.1 and 7.2 can therefore not be
perceived as an accurate prediction, but merely serve as examples.
The change in lift and drag coefficient with Mach number is shown in Fig. 7.18
for a modified NACA 6series airfoil of 10 % thickness. In Fig. 7.18a we observe
that initially the lowsubsonic lift coefficient is amplified by the compressibility
effect, in accordance with the compressibility corrections that were introduced in
Chap. 3. However, at higher Mach numbers, the formation of shock waves break the
amplification trend and with ever higher Mach number, shockinduced separation
(see also Sect. 7.6) causes the lift coefficient to drop. At the same time, the drag
coefficient due to the presence of the shock wave and shockinduced separation starts
to rise. With increased angle of attack, the onset of the drag rise occurs at ever lower
Mach numbers, showing that the dragdivergence Mach number is a function of both
390
7 Airfoil Aerodynamics
Mach number and angle of attack. This behavior is not unique to this specific airfoil
but applies to most supercritical airfoils. The higher the required lift coefficient, the
higher the angle of attack needs to be to achieve that. With higher angle of attack,
higher supervelocities on the upper surface of the airfoil are encountered, resulting
in an earlier onset of shock waves and shockinduced separation.
(a) 2
M = 0.72,
c = 0.636,
l
391
= 1.206 deg
cm = 0.082
(b) 2
cd = 0.0027, cd = 0.0054
p
f
M = 0.72,
= 0.015 deg
c = 0.500, cm = 0.078
l
cd = 0.0017, cd = 0.0054
p
1
1
CP, crit
CP
Cp, crit
Cp
(c) c
t/c = 0.110
.012
.010
Mdd
supercritical
.008
sonic rooftop
.6
.7
.75
Fig. 7.19 Comparison between a supercritical and a sonic rooftop airfoil of similar but different
geometry (after Ref. [33]). Note cd f indicates friction drag while cd p indicates pressure drag. a
Supercritical airfoil. b Sonic rooftop airfoil. c Drag coefficient versus Mach number
Finite angle
V2
a
V1
At point a: V1 = V2 = 0
Cusp
a
V1
V2
At point a:= V1 V2 0
Fig. 7.20 Trailing edge shapes relating to the Kutta condition (after Ref. [3])
geometric measures: a cusp trailing edge, and a finite thickness of the trailing edge.
We will discuss these two measures subsequently.
In Fig. 7.20 we show two generic trailing edges. The first trailing edge has a finite
angle, while the upper and lower surface of the second trailing edge are parallel
at point a. In the theoretical analysis of incompressible, inviscid flow the Kutta
condition is applicable (see for example Ref. [3]). This simply implies that the flow
smoothly leaves the trailing edge. If we apply this condition to the trailing edge with
finite angle, we see that it can only be satisfied for the upper and lower surface when
the velocities there are zero. This implies a stagnation point (C p = 1) located at
point a. If we consider the cusped trailing edge, we notice that the Kutta condition
can be satisfied while the upper and lower velocity are larger than zero C p < 1.
392
7 Airfoil Aerodynamics
The only condition that applies is that upper and lower velocity are equal. This
implies that C p < 1 for a cusp trailing edge.
In practice we do not have inviscid conditions, and the presence of the boundary
layer allows trailingedge pressure coefficients to be lower than 1 (as is shown in
many of the figures in this chapter). However, we still see that airfoils with cusped
trailing edges allow for lower trailing edge pressure coefficients without separation
than airfoils with a finite trailingedge angle. Therefore, cusped trailing edges have
a smaller adverse pressure gradient over the aft part of the airfoil, which postpones
separation. Many supercritical airfoils therefore employ cusped trailing edges (e.g.
SC(2)0412 and DSMA523a that have featured in this chapter).
One of the major disadvantages of having a cusped trailing edge is the lack of
physical space in the trailing edge to make a structure. In this region of the wing
often flaps or ailerons are positioned. Due to the diminishing trailing edge thickness, the second moment of area is relatively small. To give the trailing edge its
required stiffness (to prevent aeroelastic deformation) a relatively heavy structure
is required. Obviously, this introduces weight penalties, which eventually translate
into an increase in induced drag (see Problem 7.6). To remedy this problem, aerodynamicists investigated the effect of cusped trailing edges with a finite thickness.
It was shown that blunt trailing edges reduce the adverse pressure gradient on the
upper surface by utilizing offsurface pressure recovery [41]. In practice this means
that the pressure coefficient of the lower surface could be lower than the pressure
coefficient on the upper surface.
Let us try to imagine what the effect of such a blunt trailing edge would be. First
of all, we foresee an increase in base drag: the low pressure a the trailing edge acts
on an aftfacing surface creating additional pressure drag. In subsonic conditions we
therefore expect an increase in drag coefficient for a given lift coefficient (hence,
a reduction of the lifttodrag ratio). However, with increasing Mach number and
the formation of a shock wave, the boundary layer is exposed to an ever increasing
adverse pressure gradient. Due to the lower adverse pressure gradient of the airfoil
with blunt trailing edge, the momentum thickness will be smaller (see also Example 6.4) reducing the total momentum loss in the boundary layer. So we have two
effects at transonic conditions: an increase in pressure drag due to the blunt base of
the airfoil, and a decrease in pressure drag due to the reduced momentum thickness
at the trailing edge. Which of these two phenomena dominates depends on the relative thickness of the trailing edge compared to the airfoil chord. Investigations at
NASA showed that the optimum trailingedge thickness varied with the maximum
thickness of the airfoil and was somewhat below 0.7 %c [17]. Figure 7.21 shows the
effect of a blunt trailing edge on the drag coefficient development of an 11 % thick
supercritical airfoil at cl = 0.7. It becomes clear from this figure that at subsonic
conditions the drag coefficient due to the blunt trailing edge is higher, while at transonic conditions (between M = 0.68 and M = 0.79) the blunt trailing edge reduces
the drag coefficient.
393
.018
10
y/c (%)
sharp
.016
blunt
10
10
20
30
40
50
60
70
80
90
100
x/c (%)
.014
.012
.010
(t/c)TE in percent:
0 (sharp)
1.0 (blunt)
.008
.006
.60
.64
.68
.72
.76
.80
.84
Fig. 7.21 Effect of blunt trailing edge on drag coefficient for an 11 % thick airfoil at cl = 0.7 (after
Ref. [17])
394
7 Airfoil Aerodynamics
(b)
Moment coefficient, Cm (~)
(a)
Lift coefficient, CL (~)
CL, max
CL, max
25
Anlge of attack
CL, max
(deg)
25
Anlge of attack
(deg)
Fig. 7.22 Typical stall behavior of a passenger aircraft during flight. Data obtained from Ref. [12]
for a Dornier 328. a Lift coefficient. b Moment coefficient
length of the aircraft. Therefore, many aircraft employ highlift devices to increase
the lift coefficient of the wing such that they can have a lower stall speed and therefore
meet field length requirements. Typical highlift devices are slotted flaps at the trailing
edge and flaps, slats, or drooped noses at the leading edge. Fighter aircraft also employ
their highlift devices during combat maneuvers, which often occur in the transonic
speed realm. At any lift coefficient the highlift devices are automatically positioned
such that maximum lifttodrag ratio is achieved. This is an important parameter for
the minimum turn radius or maximum turn rate that can be achieved.
Back to the maximum lift coefficient at low speeds and why it is important to treat
this in a text on transonic aerodynamics. In the following subsections we discuss the
various types of stall and the effect of Reynolds number and Mach number on the
maximum lift coefficient of an airfoil. Furthermore, we present the aerodynamics of
highlift devices. We also show that transonic effects can often be the limiting factor
for the maximum lift coefficient that can be achieved, even at low subsonic speeds.
395
pulled inwards, the jet moves to the body instead. Gasses experience the same effect
around an airfoil. The flow is accelerated due to the presence of the body. Its pressure
reduces (Bernoulli) and the flow sticks to the body. The further away from the body,
the lower the velocity increase. This principle has been illustrated in Sect. 6.2.1. The
fact that the flow attaches to the body can therefore be attributed to the lower pressure
of the fluid around it. When this pressure rises due to an adverse pressure gradient in
the streamwise direction, the velocity of the air around the airfoil decreases and the
Coanda effect diminishes. At the same time, the boundarylayer momentum reduces
and reverse flow within the boundary layer marks the beginning of boundary layer
separation (Chap. 6). When the angle of attack increases further, the boundary layer
separates from the surface and becomes a shear layer between a turbulent wake and
the laminar outer flow: the wing has stalled.
We distinguish three types of stall: leadingedge stall, trailingedge stall, and thinairfoil stall. The latter one is merely an artifact of lowReynolds number flows and,
contrary to what its name suggests, not only pertains to thin airfoils. At a certain
angle of attack the laminar boundary layer separates due to the high adverse pressure
gradient right after the nose of the airfoil. Further downstream the laminar boundary
layer transitions to a turbulent boundary layer and reattaches to the surface. The
bubble of reverse flow that is formed between the separation and reattachment point
is called the laminar separation bubble. When increasing the angle of attack, this
bubble expands downstream. Beyond a certain angle of attack the boundary layer
does not reattach and a turbulent wake is present over the entire aircraft. This means
an instant loss in lift and a large increase in drag. Most aircraft that are designed for
highsubsonic speeds actually fly at higher Reynolds numbers, making thinairfoil
stall almost never seen in practice. However, scaled versions of these aircraft which
are tested in at a lower Reynolds number in the wind tunnel could develop thinairfoil
stall. Thinairfoil stall can be prevented by forcing transition to occur prior to the
point where the boundarylayer separates from the surface.
Leadingedge stall occurs when the boundary layer is not capable of negotiating
the large adverse pressure gradient that follows the suction peak on the nose. Before
the stall angle of attack is reached as small laminar separation bubble forms just aft
of the suction peak. This bubble may span as little as 1 % of the chord length and
forms behind the suction peak under the steep adverse pressure gradient. The separated laminar boundary layer transitions over the bubble to turbulent and reattaches.
Increasing the angle of attack moves this bubble forward and shortens it until the
laminar boundary layer is incapable of reattachment. The result is that flow separates
and that a wake is created starting at the nose of the airfoil (see also Fig. 6.23b).
Alternatively, when a turbulent boundary layer is capable of reattachment it could
still separate at a short distance behind the bubble due to the adverse pressure gradient. When either of these two events happens the suction peak over the leading edge
collapses and a relatively constant pressure coefficient over the airfoil surface results.
In the lift curve of the airfoil we see a step decrease in lift over a very small increment
in angle of attack. A wake over the entire airfoil is created which creates a sudden
increase in drag. In Fig. 7.23 leadingedge stall is shown for a 10 %thick airfoil.
396
7 Airfoil Aerodynamics
angle of attack, (deg)
2.0
8
4
12
20
.01
.02 .03
.04 .05
1.6
1.2
0.8
0.4
0
12
8
Cp
4
0
Fig. 7.23 Leading edge stall: notional lift and drag curves (top), typical pressure distributions
(middle), and uppersurface stream line (bottom). Data based on Ref. [28] for Re = 4.0 106 ,
M = 0.17
Trailingedge stall, on the other hand, is more gradual. Once the boundary layer
has survived the steep adverse pressure gradient behind the leading edge, it progressively grows thicker and gets a higher shape factor. This means that the momentum
of the boundary layer close to the wall is diminishing. At a certain angle of attack
the boundary layer starts to separate at the trailing edge (usually when H is between
2.4 and 2.6), creating a small wake. Increasing the angle of attack further increases
the adverse pressure gradient over the aft part of the airfoil and moves the separation
point upstream. At the same time a laminar separation bubble could occur closer to
the leading edge of the airfoil. With increasing angle of attack the size of the wake
from the rear separation point increases and the pressure coefficient behind the separation point remains relatively constant. Increasing the angle of attack even further
could merge the two separated regions creating a complete wake over the airfoil.
Figure 7.24 visualizes the interaction between separation, pressure distribution and
lift coefficient.
In the previous discussion we have treated leadingedge and trailing edge stall as if
they were two independent types of stall. In practice often mixed types of stall occur.
In such cases a short laminar separation bubble occurs after which the boundary layer
reattaches. Further downstream the turbulent boundary layer separates indicating
trailingedge stall. The laminar separation near the leading edge can have a profound
effect on the thickness of the boundary layer as is noted in [30]. The increase in the
397
angle of attack, (deg)
2.0
8
4
12
16
.01
.02 .03
.04 .05
1.6
1.2
0.8
0.4
0
8
Cp 4
0
Fig. 7.24 Trailing edge stall: notional lift and drag curves (top), typical pressure distributions
(middle), and uppersurface stream line (bottom). Data based on Ref. [31] for Re = 6.3 106 ,
M = 0.15
extent of an existing bubble increases the initial thickness of the turbulent boundary
layer and, therefore, increases the tendency for the latter to separate. In this way the
laminar separation bubble promotes the onset of turbulent stall near the trailing edge.
If we compare Figs. 7.23 and 7.24 we see that the difference in stall behavior is
a function of the airfoil shape. Important parameters that determine whether leading
or trailingedge stall occurs include the thicknesstochord ratio, the leadingedge
radius, and the leadingedge camber. Typically, blunt and thick airfoils are more
likely to suffer from trailingedge stall, while sharp and thin airfoils suffer from
leadingedge stall. In addition, the Reynolds number plays an important role whether
the leadingedge or trailingedge stall prevails.
398
7 Airfoil Aerodynamics
1.6
trailingedge stall
1.2
leadingedge stall
0.8
thinairfoil stall
0.4
0
0
Reynolds number:
0.7 million
3.0 million
9.0 million
10
15
20
number effect that is instrumental for the maximum lift coefficient: the position of
the transition region. An increase in Reynolds number could also imply a switch
from leadingedge stall to trailingedge stall. In Fig. 7.25 this is demonstrated for a
moderately thick airfoil, which stall behavior is evaluated at three different Reynolds
numbers. Note that this airfoil shows thinairfoil stall at low Reynolds numbers even
though it is not considered a thin airfoil. This confirms that thinairfoil stall is not
limited to thin airfoils. We explain the various Reynoldsnumberrelated effects on
stall in the subsequent paragraphs.
As the Reynolds number increases the position where the boundary layer transitions from laminar to turbulent shifts forward (see Sect. 6.5.4). As the region of
transition has a large effect on the boundary layer development over the airfoil it has
an impact on when (at what angle of attack) and where (right after reattachment
or at the trailing edge) the turbulent boundary layer separates. For a low Reynolds
number the transition point lies downstream of the suction peak. If a small laminar
separation bubble is present the boundary layer transitions from laminar to turbulent over the bubble and reattaches. The reattached boundary layer at the end of the
separation bubble has a significantly higher thickness, which still grows in thickness
downstream of the reattachment point. Increasing the Reynolds number shifts the
transition region forward, which reduces the size of the separation bubble and consequently reduces the thickness of the boundary layer downstream of the bubble. The
downstream boundary layer therefore separates at a higher angle of attack, which
increases the maximum lift coefficient. When the transition region moves in front of
the separation point, the effect of a higher Reynolds number is limited to the thickness decrease of the turbulent boundary layer, which could increase the maximum
lift coefficient even further.
lowdrag airfoils
conventional airfoils
max
(~)
1.3
399
1.2
1.1
P63A
1.0
P51B
YP80A
0.9
P39N
0.8
F8F3
0.7
0.6
0.2
P38F
0.3
0.4
0.5
0.6
0.7
0.8
400
7 Airfoil Aerodynamics
NACA 4series airfoil
4
Cp
4
crit
2
2
Cp
Cp
M = 0.40
CL = 1.23
CL = 0.96
max
max
Cp *
M = 2
4
4
2
2
Cp
Cp
M = 0.50
CL = 1.06
CL = 0.93
max
Cp
2
Cp
Cp
max
M =
+3
2
crit
2
Cp
M = 0.63
CL = 0.77
CL = 1.08
max
max
2
2
Cp
Cp
M = 0.68
0
CL = 0.68
max
0
CL = 1.03
max
Fig. 7.27 Typical pressure distributions about NACA 4series airfoils (left) and NACA 6series
airfoils (right) at maximum lift coefficient (after Ref. [40])
The same theoretical boundary has been shown in Fig. 6.9 where the dashed
line shows a theoretical limit in local Mach number for supersonic flow over a
convex surface embedded within
subsonic flow. It has been shown that a characteristic
Mach number of M = 2 is the highest Mach number for which Mach waves
can exist for which the curvature is monotonic.7 This corresponds to a situation
where the streamline curvature is still continuous and the flow is isentropic and
7
A monotonic curvature means that the Mach wave is either entirely convex or entirely concave.
If Mach wave curvature is not monotonic it means an inflection point is present somewhere on the
Mach wave.
401
uniform [27]. Via Eq. (4.10b) we calculate that this corresponds to a maximum local
Mach number of:
2
1.58 for = 1.4
(7.5)
M=
3
We can compare this maximum Mach number to the maximum Mach number that
can be achieved in the throat of a supersonic convergentdivergent channel. In that
case the flow is said to be choked. Whatever we do upstream or downstream of the
throat, the Mach number can never exceed M = 1. In our case we have a similar
condition only this time we consider external flow and limit the freestream Mach
number to values below unity. The theoretical value of 1.58 is limited to flow over
convex surfaces that provide monotonic curvature to the Mach waves. The absolute
maximum is attained when discontinuities in streamline curvature are allowed.
In
4
2.50
3
(7.6)
For practical cases (7.5) is the most likely maximum value to be approached in
most cases, according to Laitone [24]. Substituting (7.5) in (6.17) and rewriting this
for C p yields a theoretical minimum local pressure coefficient as a function of the
freestream Mach number:
1
3
1 2
2
M
1+
1
(7.7)
C p M =2 =
M
+1
2
This minimum has been indicated in Fig. 7.27 with the dashed line. Note that the
minimum pressure comes very close to this value but never exceeds it. This confirms
the statement that (7.7) is a practical limit on the minimum pressure coefficient
and that (7.5) is a limiting Mach number. This analysis also demonstrates that even
at speeds well below the speed of sound, transonic effects dictate maximum lift
coefficient.
The theoretically obtained minimum pressure coefficient has been compared to
experimental data obtained from a series of wind tunnel tests by Axelson [5]. As
can be seen in Fig. 7.28 at Mach numbers beyond 0.5, considerably lower pressure
coefficients can be obtained than those predicted by (7.7). An alternative prediction
by Mayer comes closer to the measured values. Mayer based his predictions on a
series of wind tunnel tests between M = 0.4 and M = 2.2 [29]. He proposed a
statistical relation between the minimum value of the pressure coefficient and the
freestream Mach number:
1
(7.8)
C pmin = 2
M
402
7 Airfoil Aerodynamics
leading edge
3.5
min
(~)
ahead of shock
3
2.5
2
1
Cp = 2
min M
1.5
Pressure coefficient, Cp
Mlim=
0.5
Mlim=
0
0.3
0.4
0.5
2
3
+3
2
0.6
0.7
0.8
8
0.9
(~)
It can be shown (see Problem 7.9) that this results in a static pressure ratio of approximately 0.30 between the suction peak and the free stream. In other words, it is
predicted to achieve 70 % of pure vacuum under the suction peak. At Mach numbers
higher than 0.5 it can be seen from Fig. 7.28 that Mayers prediction comes closer to
the measured values published in [5], but still underestimates in certain cases.
If we return to Fig. 7.27 we observe that at M = 0.63 and M = 0.68, a shock
wave is present for the lowdrag airfoil. In addition, the dashdotted line indicates a
theoretical maximum local Mach number that cannot be exceeded in front of a shock
wave without creating separated flow [24]:
M=
+3
1.48 for = 1.4
2
(7.9)
In Ref. [24] it is demonstrated that for constant stagnation conditions ahead of the
shock and a Mach number of 1.48 the maximum absolute value of p2 (static pressure
behind the shock) is achieved. In other words, for this limiting Mach number the
maximum magnitude of static pressure recovery is achieved (although this is usually
less than the free stream static pressure). If we imagine the local velocity to increase
beyond this Mach number and form a shock wave further downstream, the static
pressure recovery would be less ( p2 would be lower). Therefore, the higher subsonic
free stream static pressure would push the normal shock wave back upstream until
the maximum absolute static pressure would be produced. Substituting (7.9) in (6.17)
and rewriting this for C p yields a theoretical limiting local pressure coefficient in
front of the normal shock wave:
1
2
2
1 2
2
1+
M
C p +3 =
1
(7.10)
2
M=
M
+1
2
2
403
This is confirmed by the graphs for M = 0.63 and M = 0.68 for the NACA
6series airfoil in Fig. 7.27. We observe that the pressure coefficient is increasing
between the suction peak and the shock. This indicates that the local Mach number
is reducing from a value close to 1.58 to alower value in front of the shock.
It is demonstrated in [11] that M = ( + 3)/2 is the only Mach number for
which a curved streamline can become straight when passing through a normal shock
wave. Consequently, it would be the most likely Mach number in front of a shock
wave whenever boundary layer separation occurs at the foot of the shock. If we
combine these observations we can conclude that on convex profiles the maximum
Mach number in front of a normal shock wave never exceeds 1.48 and that it is
likely that this shock causes boundary layer separation. Evidence of the accuracy of
this predicted pressure coefficient is presented in Fig. 7.28. The experimental data
consistently shows that the value of M = 1.48 ahead of the shock is never exceeded
in any of the experimental tests.
404
7 Airfoil Aerodynamics
boundarylayer flow wake flow
slat wake
slat wake
wing wake
wing boundary layer
shockwaveboundarylayer interaction
transition
transition
cove separation
cove separation
supersonic bubble
SLAT
WING
FLAP
Fig. 7.29 A320 airfoil in landing configuration (modified from Ref. [13])
the slat merges with the boundary layer of the wing. This is termed boundary layer
confluence. This happens again over the upper surface of the flap where the wake of
the wing surface merges with the flap boundary layer. In this example, there exists
reversed flow over the flap, which indicates the onset of stall. Over the slat we also
observe a local bubble of supersonic flow that is terminated with a shock wave.
Under the shock wave the flow transitions from laminar to turbulent resulting in a
thickening of the boundary layer. We can imagine how, with increasing Mach number,
the strength of the shock wave increases, which eventually results in boundarylayer separation at the shock foot (see also Sect. 7.6.1). It is this phenomenon that
effectively limits the maximum lift coefficient as a function of the freestream Mach
number.
To substantiate these claims, consider Fig. 7.30. Here we have a typical multielement airfoil consisting of a slat and a doubleslotted flap system. The graph shows
the minimum pressure coefficients measured at the suction peak on the leading edge
of the slat. In addition, the statistical limit by Mayer (7.8) as well as the theoretical
limit of (7.7) are shown. Note that the theoretical limit predicts a lower pressure
coefficient than the statistical prediction by Mayer on this lowsubsonic domain.
The symbols indicate various combinations of flap and slat deflection. Weclearly
see that the boundary is formed by the limiting Mach number of M = 2/ 3 ,
regardless of the flap or slat setting. We also observe a decrement in minimum
pressure coefficient before the local Mach number reaches the limiting value. In the
cases where the slat is deployed the decrement in lift coefficient between M = 0.19
and M = 0.28 for this particular airfoil is between 0.3 and 0.4 [46]. This causes
a 10 % decrease in maximum lift coefficient due to the transonic choking of the
flow over the slat leading edge.
405
(~)
24
Mlim=
22
Cp
2
3
1
18
10
35
10
16
10
35
35
10
35
min
2
8
Pressure coefficient, Cp
min
26
20
14
12
10
0.18
0.2
0.22
0.24
0.26
0.28
0.3
(~)
Fig. 7.30 Minimum pressure coefficients at the leadingedge suction peak for Re = 6.9 106
(after Ref. [46])
406
7 Airfoil Aerodynamics
total separation (stall)
shock
attached flow
shock
total separation
CL, max
buffet boundary
boundary for
critical freestream
conditions
CL
CL, max
CL = const.
design point
subsonic
M
transonic
supersonic
1.0
Fig. 7.31 Separation boundary at various Mach numbers and lift coefficients (after Ref. [42])
In addition, this combination of Mach number and lift coefficient effectively limits
the altitude an airplane can reach at a given Mach number when flying in steady,
symmetric, horizontal flight (see Problem 7.11).
407
Fig. 7.32 Schlieren images of the flow around a 6 % thick RAE 104 airfoil at 2 angle of attack
(from Ref. [34]). a M = 0.84. b M = 0.86. c M = 0.88. d M = 0.90
7 Airfoil Aerodynamics
408
0.6
0.5
0.4
bc
d
0.3
0.2
0.1
0
0.5
0.6
0.7
0.8
0.9
1.0
on the upper surface creates an inclined ramp for the flow outside the boundary layer
resulting in a more oblique shock wave terminating the supersonic domain.
The effect of shockinduced separation on the lift coefficient is shown in Fig. 7.33.
The letters a through d refer to the respective images in Fig. 7.32. We see that the
maximum lift coefficient at a constant angle of attack is reached around M = 0.87.
When the boundary layer fully separates from the surface the lift coefficient starts to
decrease rapidly. Most of this decrease is caused by the increase in suction over the
lower surface due to the decrease in trailingedge pressure when the boundary layer
has separated. With increasing Mach number, the shock wave on the lower surface
moves faster towards the trailing edge than the upper shock wave creating more
suction over the lower surface and hence reducing the lift coefficient. A minimum
is obtained when the lower shock reaches the trailing edge. Increasing the Mach
number even further results in a higher value of the lift coefficient that corresponds
with both shocks having reached the trailing edge.
In Fig. 7.34 we schematically repeat this development of the shock pattern over
a generic airfoil of low thickness. The same effects as in the wind tunnel are seen
as can be verified by comparison to Fig. 7.32. In Fig. 7.34 the strength of the shock
has been indicated by the thickness of the shock line. Note how the shock strength
is highest at the surface, while its strength decrease further away from the body.
Intuitively, we can verify this because the supervelocities are highest close to the
surface. Furthermore, the relative size of the shock wave increases with higher Mach
number. At M = 1.4 the shocks at leading and trailing edge even exceed the bounds
of this figure. Finally, the notional point of separation is shown along with the growth
of the wake over the transonic regime. Note how the wake grows up to M = 0.9 and
decreases again above that value.
409
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.6
Fig. 7.34 Notional development of section lift coefficient over a symmetric airfoil at = 2
Fig. 7.35 Schlieren images of the flow around a 6 % thick RAE 104 airfoil at M = 0.75 (from Ref.
[34]). a = 2.7 . b = 3.7 . c = 4.7 . d = 5.7
410
7 Airfoil Aerodynamics
Let us follow Pearcey [34] in his investigation into the behavior of shockinduced
separation at constant Mach number and increasing angle of attack. In Fig. 7.35 we
see schlieren images of the same airfoil as in Fig. 7.32. The Mach number has been
fixed at M = 0.75 while the angle of attack increases progressively from = 2.7
through = 5.7 . It is important to note that the transition has been artificially
induced at the nose and that the shock wave interacts with a turbulent boundary
layer at any of the aforementioned angles of attack. We see that somewhere between
= 2.7 and = 3.7 the boundary starts to separate at the shock foot. The shock is
still quite close to the leading edge and the boundary layer reattaches downstream.
With increasing angle of attack the shock moves aft and the separation bubble expands
rapidly towards the trailing edge. This causes the trailing edge pressure to decrease
and the shock to halt before it starts to move forward. At even higher angles of attack
the shock moves further forward together with the separation point.
In Fig. 7.36 the position of the shock foot is plotted for a range of angles of attack
and Mach numbers. We distinguish uniform behavior at the various Mach numbers:
initially the shock moves aft up to the point that the separation bubble reaches the
trailing edge. Beyond that angle of attack the shock wave and the separation point
move forward. With decreasing Mach number the angle of attack at which shockinduced separation starts becomes progressively higher. Simultaneously, the shock
position at which separation starts becomes closer to the leading edge. It becomes
difficult to judge at what Mach number the separation ceases to be induced by the
shock instead of being induced by the adverse pressure gradient behind the suction
peak, as seen at low speeds.
Let us take a look at the effect of shockinduced separation on the lift coefficient.
In Fig. 7.37 we see the lift coefficient development over the RAE 104 airfoil at two
Mach numbers. In 7.37a we see that lift coefficient remains closeto constant beyond
the point where the separation bubble reaches the trailing edge. For M = 0.80 we
expect the separation to start at a lower lift coefficient. This is indeed the case as
can be seen in Fig. 7.37b. However, here we have somewhat of a different case.
xsh
(~)
c
0.8
M
0.95
0.6
0.90
0.4
0.85
0.2
0.80
0.70
0.75
Angle of attack,
(deg)
0
0
10
411
(a)
(b)
1.2
RAE 104
1.0
1.4
0.8
0.6
0.4
0.2
1.0
0.8
0.6
0.4
0.2
0
0.2
0
Angle of attack,
(deg)
10
Angle of attack,
10
(deg)
Fig. 7.37 Lift versus angle of attack for a 6 % thick RAE 104 airfoil at two transonic Mach numbers
(after Ref. [34]). a M = 0.70. b M = 0.80
The lift coefficient increases even when the shockinduced separation covers the
upper surface from shock foot to trailing edge. Such behavior is attributed to offsurface pressure recovery and occurs beyond the angle of attack where the pressure
coefficient at the trailing edge falls below the critical pressure coefficient. When
this is the case, the pressure coefficient at the trailing edge has no effect anymore
on the development of the flow over the lower surface. With increasing angles of
attack, the pressure coefficient on the lower surface begins to rise again. At the same
time, the pressure coefficient on the upper surface decreases even further below the
critical pressure coefficient allowing the shock wave to move backwards resulting in
an increase in suction over the upper surface. The combined effect is an increase in
lift coefficient beyond the angle of attack of shockinduced separation. This effect
is dependent on the presence of offsurface pressure recovery and therefore partially
on the shape of the airfoil trailing edge.
412
7 Airfoil Aerodynamics
(a)
(b)
2.0
2.0
Cp
1.5
1.5
1.0
1.0
.5
Cp
.5
= 3.5o
M = 0.723
.6
.6
x/c
R = 4.2106
.2
R = 4.2106
1.0
1.0
0
= 3.5o
M = 0.723
.4
.6
.8
1.0
.2
x/c
.4
.6
.8
1.0
Fig. 7.38 RA16SC1 airfoil. Envelope of the pressure distributions at buffet onset (after Ref. [25]).
a Calculation. b Experiment
(a)
(b)
Vs
M1
M
M1
Vs
Wake deflected
upwards
Weak shock
(attached flow)
(c)
(d)
Vs
M1>1.3
Strong shock
Weak shock
M
Wake
M1>1.3
Vs
Fig. 7.39 Shock wave development over a 14 % thick circular arc airfoil with increasing subsonic
freestream Mach number (after Ref. [14]). a M1 < 1.2 (no separation). b M1 > 1.2 start periodic
flow. c 1/2 cycle after (b). d M1 > 1.3 shock too strong for reattachment
structure. This phenomenon is termed transonic buffet and its mechanics is further
explained in this section.
Before we look at a realistic case of transonic buffet on a supercritical airfoil
it is instructive to consider the shock motion over a circular arc airfoil. The shock
development over a 14 % thick circular arc airfoil at symmetric flow conditions
is presented in Fig. 7.39. In Fig. 7.39a we see shocks of equal strength appearing at
both the upper and lower surface. In Fig. 7.39b we assume that a pressure disturbance
upstream of the shock on the upper surface causes the shock to move more forward.
The instantaneous shock Mach number (Ms ) is given by:
Ms = M1 +
1 dxs
a dt
(7.11)
413
where xs is the location of the shock and dxs /dt = Vs is the velocity of the shock
wave with respect to the airfoil. As the shock moves forward Ms increases beyond the
Mach number at which separation is triggered. So as the shock moves forward it also
shifts the separation point forward creating an asymmetric wake. At the same time, the
trailing edge pressure decreases which moves the shock on the lower surface closer
to the trailing edge. In moving backward, Ms decreases which prevents the boundary
layer from separating. As the upper shock moves forward over the convex surface it
encounters a slower supersonic region. Therefore, its strength weakens as it moves
upstream. The flow attaches at a point where the shock becomes too weak to initiate
separation. On the lower side the reverse is happening. As the shock moves closer
to the trailing edge its strength is increasing and at some point it initiates separation.
This causes the wake to move to deflect downward as can be seen in Fig. 7.39c. This
process is now repeating itself on the lower surface. The deflected wake essentially
causes the shock on the lower surface to move forward, while the shock on the
upper surface starts to move backward. Even though there is a time lag between the
downward deflection of the wake and the motion of the shock wave, the flow can
maintain a selfsustained periodic motion. Finally, when M1 > 1.3 (Fig. 7.39d) the
shock is strong enough at all times to cause separation. No reattachment can take
place and the periodic motion of the shock wave ceases.
Close to the surface of the airfoil we can approximate the flow as one dimensional.
We follow Ref. [26] to relate the oscillating shock position to the oscillating strength
of the shock wave. The latter one is given as the ratio of static pressure over the
shock wave. We denote the properties of the flow ahead of the shock wave with a
subscript 1 and behind the shock with subscript 2. The shock strength would then
be quantified as p2 / p1 . We assume that the flow has a varying velocity, u, a varying
static pressure, p, and a varying speed of sound, a, in the streamwise (x) direction.
The displacement of the shock in time is given by x. The pressure and velocity in
front and behind the shock can be written as follows:
dP1
x + p 1
dx
p2 = P2 + p 2
dU1
x
u 1 = U1 u s + u 1 +
dx
u 2 = U2 u s + u 2
p1 = P1 +
(7.12)
(7.13)
(7.14)
(7.15)
where p and u are perturbation pressures and velocities, respectively, and P and U
are the steadystate pressures and velocities, respectively. We know that the pressure
ratio over the shock wave is given by the RankineHugoniot relation:
2
p2
=1+
p1
+1
u 21
a12
1
(4.14)
414
7 Airfoil Aerodynamics
1
the velocities of u,
u s , and dU
dx x are an order of magnitude smaller than U1 . We
therefore only retain the multiples