Krishnamurty Karamcheti
Professor of Aeronautics and Astronautics Stanford University
Preface
The aim of this book is· to explain the basic principles and analyticaJ methods underlying the theory of the motion of an ideal fluid (an inviscid incompressjble fluid) and the role of the theory in describing and predicting the flows associated with the motion' of certain bodies of aerodynamic: interest suc:h as wHigs and bodies of revolution. I have attempted to describe idealfluid aerodynamics, although restricted to certain problems, . .as a branch of theoretical physics. The subject is' developed from basic: principles showing clearly the complementary features of physical understanding and the mathematical handling of the theory.. The intention is to show the role of physical understanding in mathematical formulation, .to ~ring out the motivation for the mathematical language and, methods employed and the necessity for a!,plying a certain,amount of mathematical rigor in arriving at physically appealing solutions. The book is written to serVe as a sel~contained text at the senior under.. graduate or firstyear graduate level. The idea is not to give inadequately explained solutions to many special problems, but rather to present, for~ few selected practical problems, a unified treatment leading from ba.ic principles to practically meaningful results. A large part of the book deals with general concepts and mathematical methods, always related, however, to the solution of problems. In this way I hope that the book will perform the valuable function of teaching subject matter related to a broader methodology that will lead logically to more advanced topics and methods in fluid mech.anics; it should be of interest to students in various disciplines, such as applied mathematics, physics, and engineering. This book has grown out of lectures on aerodynamic theory which I have offered for the last decade and which have been received with considerable enthusia>m. It is because of the students' encouragement that I venture to publish thnm. I am greatly indebted to Professor Irmgard Fliigge Lotz for reviewing the manuscript and fOr many valuable suggestions and discussions. My special thanks are due to Dr. Maurice L. Rasmus~n who read the manu· script and offered valuable critic'sm. Many students have helped me enthusiasticaUy with the preparation of the book, and my deep appreciation goes to all of them.
ali
face
I am very grateful to Professor O. O. Tietjeos for furnishing me with original prints of many of the fto.w photographs. . The original source of the photographs for the plates and 9 is the National Physical Laboratory, Engl~nd, and I am gready obJiged to its Director for permission to reproduce the photographs which are Crown copyright. The original photographs for plates 3, 4, 6a, and 7 are all from prewar Germanpubiication$, and J wish to r~ord my indebtedness to their respective sources. Plates 3 and 4 are after F. Homann, Forschrmg auf dem Gebiete des Ingenieurwesens, 7 (1936). Plate 68 is after L. Prandtl, Handbucli cler Experimentalphysik, 4, Part } (Leipzig. J931). Plate 7 is after Piandtl, The Physics of Solids and Fluids (London, 1~30) The typing was capably handled by' Mrs. Katherine Bradley, Miss Gail Lemmond, and Mrs.' Elaine Morris. My sincere thanks to them, Finally I wish to express my appreciation to John Wiley and Sons for the understanding, patience, and encouragement they have extended me over the years. Krishnamurty Karamcheti
a
Contents
1. INTRODUCTION. . . . . . . • .
..  ......  . , ..
1
Stanford, Caltf0miD August 1966
1.1 Fluid as a Continuous Medium . . . . . . . . • • • . • 2 1.2 Properties of a Fluid at Rest: Thermodynamic properties; Compressibility.; Incompressible fluid; : Heat conduction and the coefficient of thermal conductivity . . . . • • . . • • . 2 1.3 Properties of a Fluid in Motion: Friction or viscosity; Coefficient of viscosity; Compressibility; Heat transfer. 4 1.4 Laminar and turbulent motioos . . . . . . . . . . . • . 10 1.5 Some Relevant Parameters: Relative magnitude of the foroes, Froude number, ReynOlds number, and M~h number; Par.ameters characterizing compressibility; PrandtJ number; Parameters on which force and heat transfer depend . . • . 13 1.6 Range of Some Parameters . . . . . : . . . . • ,23 1.1 Conditiong for Neglecting Compressibility' Effects; Case of liquids; Case of gases. •. . . , . . .', . . . • • . 23 1.8 Conditions for ~eglecting Gravity EffeCts . • • . . " . . 25 1.9 Nature of the Problem when Compressibility Effects are Negligible. . . . . . . . . . . . . . . . . . . • . . . 25 1.10 Variation of Fl(>w Patterns with Reynolds Num~; Flow past bluff bodies; Flow past stteamlined bodies . . . . . . ". 26 1.11 Variation of Flow Pattern with Mach number . . . . . . . 3S J.l2 Effects of Viscosity at High Reynolds Numbers:, The Boundary layer: Boundary layer concept; Some characteristiCs of the lami.nar boundary layer; Turbulent boundary layer; separauon; Wakes . . . . . . '. . . . . . . . . . . . '. . 1.13 Consequences of the BoundaryLayer Concept .. ' 1.14 Ideal Fluid Theory . . . . . . . . ' . . . . . . . . . .
2. ELEMENTS OF VEcrOR ALGEBRA AND <;:ALCULUS
56
57
"
2.1 2.2 2.3 2.4 2.S 2.6 2.7
, Representation of a Vector. Addition and Subtraction . . Definition of a Vector . . . Multiplication by a Number . Unit Vector . . . . . . . . Zero Vector . . . . . . . . ScalarProduct of Two Vectors
S8
60 60
61 61 61
.'
ix
x
Contents
2.8 Vector Product of Two Vectors . . . . . . . 2.9 Plane Area as a Vector 2.10 Velocity of a Point of a Rotating Rigid Body 2.11 Polar and Axial Vectors. , . . . . . ' . . 2.12 Multiple Products: S¢alar· triple product; Vector iriple product . . " . . . . 2.13 Components of a Vector . . . . . . . . . . . . . 2.14 Specification.9f a Vector • . . . . : . . . . . . . ~.15 Cartesian Coordinates !Dd the i, j, k System of Unit Vectors 2.16 Notion of Curvilinear Coordinates . . . . • . . . . . . . 2.17 Orthogonal Curvilinear Coordinates: ExampJcs.Cylinchical a.nd spherical coordinates ..• . . ... . . . • . • • 2.18 Products of Vectors in Terms of Their Components . 2.19 Functions Involving Vectors and Scalars • • • • • • 2.20 Scalar and Vector Helds . . . . . . . . . . '.' . . 2.21 Differentiation of a VeCtor Function of a Scalar Variable . 2.22 Changes in the· Unit' Vectors of CylindricIIJ and Spherical Coordinates. . . . . .' . . . . . '. . • . . . . . . . . 2.23 Frames of Reference . . . . . . . .'. . . . . : • . . . 2:24. Differentiation of a Scalar Function of a Vector: Concept of a gradient . . . . . . . . . . . . . . . . . . . . . ' . . 2.25 Differentiation of a Vector Function of a Vector: COncept of tensor gradie,nt of Ii vector; R~lationof divergence, strain, rotation, and curl to the tensor gradient . . 2.26 Del, the Vector Differential Operator .. . . . 2.27 Integration of a Vector Function of a Scalar . 2.28 line Integrals: Circulation . . . . 2.29 Surface Integrals . . . '.' . . . . 2.30 Volume Integrals . . . . . . . . 2.31 Integral Definition of the,Gradient . 2.32 Divergence 0(' a Vector Field . . . 2.33 Curl of a Vector Field,. . . . . . 2.34 Components of a Curl as Circulation. 2.35 Some Related Remarks. . .,' . . .,.' 2.3,6 Relations Between Surfacr. and Volume Integrals,: Gradient theorem; Divergence theorem. . 2.37 Theorem of Stakes . 2.38 Further 0p'lations . 2.39 Lap~ Operator ., 2.40 Green~ Theorem . 2.41 Irrotatiomll Field: Scalar potential . .' 2.42 SQlenoidal Field: . Vector potential . 2.43 Laplace's Equation. .'. . . . . . 2.44 Poisson's Equation. . . . . . , . 2.45 Expressions in General Orthogonal Curvilinear Coordinates: Unit vectors: Infinitesim'll dist.ance between two neighboring
r • • • • • • • • •
Contents points; Differential volume and surface elements; Gradient; Di~ergence; Curl; Laplacian 2.46 Some Useful Relations 3. STRESS IN A FLUID,
xl
'.
62 65 66 69 69 72 74 75 77 79 '83 84 87 89 94 96 98
140 146
.148
3.1 Surface Forces and Body Forces. 3.2 Concept of Stress and the Specification of Stress at a Point 3.3 Stress in a Fluid at Rest: Hydrostatic pressure 3.4 Stress in a Fluid in Motion 3.5 Stress in a NonViscous Fluid in Motion: PreSsure 3.6 Pressure Distribution in a Fluid at Rest. 3.7 Concluding R e m a r k s , .
148 149 153 154 155 155 157 IS8
I ~8 159
4.
DESCRIPTION OF FLUID MOTION.
4.1 4.2
4.3
4.4
103 108
III III
113
115 116
4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 5.
Lagrangian Method Eulerian Method . Connection between the Lagrangian and Eulerian Descriptions Steady and Ur.steady Mvtions . Path Line, Streamlines Stream Surfaces and Stream Tubes. Reference Frame and Streamline Pattern Stream Function3 Stream Function For Twodimensional Flow Stream Function For Axisymmetric Motion , Stagnation Points
I SO 162 162 162 164 165 165 16'8 170 172
119
120 125 t.28
EULERIAN LQUATIONS FOR THE MOTION OF AN IDEAL FLUID,
5.1 5.2 5.3
175
17S 1'7$
130 131 133 133 t34 136 137
138
13~
5,4 5.5 5,6 5.7 5.8 5.9 5.10 5,11 5, J2
Local, Convective and Material Derivatives Euler's Equation. Equation of Conservation of Mass: Change of volume of a fluid element Equation of Energy , , . . , . , Equation of State . . . . . . . .. Equations for an Inviscid Compressible Fluid Condition of Incompressibility, Consequences of Incompressibility Equations for an Ideal Fluid Inilial Conditions Boundary Conditions for an Ideal Fluid: Condition at a solid fluid b,~undary: Condition at a free surface Conditions at Infinity.
181 184 186 187 187 188 190 190 190 194
Contents '5.13 Stream Functions for .Iocompu'uible Flow • • . • '. . 194 S.t4 VectQr'PQtential for· Incompressible Flow and Its Relation in the Stream Functions. ~ . . • . . . . . . . . . . . ',' 196 S.IS Elimination of the Body Forces from the Equation of M~tion for a Certaiil Inc:ompreSsible Flow Problem. ' 197 '" ALTERNATE FORMS OF THE 'EQUATIONS 198
Contents of Change of Circulation~ Kelvin's Theorem • Irrotational Motion . . . . . . • . . : • . . '. Velocity Potential . • . . .'. . • •. . . . . • Equations for Irro:ational Motion of an Ideal FlWcl . . Irrotational Motion as an Impulsiwly Gencra~ MotioO: Velocity Potential as the Potential of an Impulse: • • . • . 9.10 Boundary ConditiQns: Condition' at a. solid8uid boundary; Conditions at ,free surface • . . . • . . . ~ . . • • . . 9.11 Problems of Concern.' . . • . • • . . . . • '., . . • . '. 9.12 Some Topological Notions: Connectivity; Reconcilable' a~d irreconcilable paths; Reducible and irrcduc:ible circuits; Reconcilable and irreconcilable circuits; Simply connected region; Doubly connected region; Multiply connected region; Barriers . . . . . . . . • . .., ." . . 9.13 lrrotational Motion in a Simply Connected Region. 9.14 Irrotational Motion in a Doubly Connected Region. 9.15 Summary. . . . . . . . . . 9.16 Conditions at Infinity. . . . . . . . . . 9.17 Velocity Components at Infinity . . . . . . . . . .. 9.18 Some Further Properties of IrrotAtional Motion: Simply connected region; Doubly connected region. . . . . . . . 9.19 Stream Functions and the Velocity Potential; Twodimensional motion; Axisymmetric motion
9 ..S 9.6 9.7. 9.8 9.9
Ra~e
239 244 244 US
246
249
250
6.1 Equation of Change • . • . 198 6.2 Conservation of Mass . • • 199 6.3 Conservation of Momentum. 199 6.4 Conservation of Energy • . • .. 201 6.S Integral Form of the Equations From the Point of View of a Fixed Region of Space . . . . . . . . . . ;'. . . . . . 20~ 6.6 IiltegraI Form of the Equations From the Point of View of a Finite Fluld Region . . . .' . . . . . . . . . . . . . . 203 6.7 Rate of Change of a QUantity Following a Fluid ~n • • 20S 6.8 Equations of Change ocari IdCal Fluid •• . . • .. ", . . , 207 6.9 Rate ofChangc of It Q\aU)tity Following ~ Moving Regionol Space . . . . . . . . • • . • . . • • . . . . . . • • ,207
2S2'
258
1S9 263 261
268 '
269 276
7. EQUATIONS OF DISCONTINUOUS, MOTION. '. . . "
.
210
7.1 A Stationary Dilcontinuity ,in a Steady Flow . . • . • • • 7.2 A Moving Discontinuity in the Unsteady Flow of an I~ 'Fluid ~ . . . . • •.• ,. . . . • • . . . . • . • • • . . 7.3 DiscOntinui.ty in the Flow ofaalnliornogeneouiIncompressible Fluid . . 7.4 Remarks . . . . . .'. • '. . . . . . .
8.
.210
213
10.
UNSTEADY ACYCLIC MOTION.
10.1 10.2 10.3 10.4
278
217 219
INTEGRATION OF EULER'S EQUATION
IN SPECIAL CASES. . . . . . . , . . . .
Mathematical CharaCter of the Equations . . . . . . . . • Integration of Euler's Equation in Steady RotationalMoti~ Spatial Variation of Hs . . . . . • • . , . . . . . Integration of EUler'i Equation in Irrotational Motion 8.' Remarks on an Irrotationa,i Forc:c Field 8:6 Remarks on,Bernu,ulli'sEquation 8.1 8.2 8.l 8.4
221 221 223 224
226
to.5 10.6 10.7 10.8
10.9 10.10 10,11 10.12
2i7
,229
231
Mathematical Problem . . . ~ Expanding Sphere . .... . . Problem for a Translating Body in Terms of Reference Frame. . . . . . . . '.' . . . . Translating Sphere . . . . . . . . . . . . . Force on a Translating Body of Arbitrary Shape Impulse . , . . . . . . . . The Apparent Mass Tensor . . Kinetic Energy and Impulse . . Moment on a Translating Body Uniform Transiation . Per,manent Translation Remarks . . . . . .
278
. . . 279 Body Fixed . 281 . . 284 .. 291 ' 297
297
JOI 304
309 311 311
,. IRROTATIONAL MonoN,•. '.' •
9.1 Most General Motion 9f a Fl,uid ElelQent. " 9,2 Rotation and Vorticity • . . . . ~' . . . . 9;3' Cin:uIation and Vorticity. , ~ • . . ' . . . 9.4~ 0( CbanpofVoi'dclty: HelmhOltz's 1boomil
'.
11. STEADY ACYCLIC MOTION'
,
'312 312 313 317
.  ..
"
231 233 235
236
11.1 Statement of the Problem . 11.2 Simple Polynomial Solutions 11.3 The Source Potential . . . , ; ~ 11.4 Source i!l a Uniform Flow (Axisymmetric Flow over a Semiinfinite Body of Revolution), . . . . . .~' _ . . . . .'.
Contents Contents
Il.S Source and Sink in a Uniform Flow (Axisymmetric Flow over . . a OosedBody of Revolution) . 11.6 Line Distribution of Sources and Sinks in a Uniform Flow: Axisymmetric Flow over Slender Bodie~ of Revolution 11.7 The Doublet Potential' . . .. 11.8 Doublet in a Uniform Stream; Flow over a Sphere 11.9 Line Distribution of Doublets in a Uniform'Stream: Lateral and Axisymmetric Flow Past a Body of Revolution 11.10 Flow Past Arbitrary Bodies of Revolution. ILl! Flow Past an Arbitrary Body 11.12 Pressures 11.13 Discussion. , 11.14 Force on an Arbitrary Body: d'Aiembert's Paradox. 11.1 S (;irculation as the Agency of Force . .. 12. STEADY TWODIMENSIONAL ACYCUC MOTION
12.1 12.2 12.3 12.4 12.S 12.6 12.7 12.8 12.9 12.10 14.2 14.3 14.4 14.5 14.6 14.7 14.8 14.9 14.10 14.11 14.12 14.13 14.14 14.1S 14.16 14.17 14.18
327 331 333 339
342 343 344
348
3S0
354
3SS 3S9 3S9
361 363 364 365 367 369 371 372
Recapitulation. , . . ... , . . Further Considerations Relating to the Stream Function Problem in Terms of the Stream Function. Uniform Stream. Source Flow '. Combination of a Source and a Sink of Equal Strength . Doublet . . . ~' Source and Sink of Equal Strength in a Uniform Stream Doublet in Uniform Stream:' Flow over a Circular Cyliiider Flow ,Past an Arbitrary Cylinder. '.'
14.19
14.20 14.21
14.22
i4.23
Nomenclature and Algebra of Complex Numbers Geometrical Interpretation Polar and Exponential Forms of a Complex Number . Function of a Complex Number . Analytic Function ' ',' ~auchyRiemann Conditions Some Conseque~ of CauchyRiemann Equations Remarks . . •' . . . . . . . . . . . . . . . Some Analytic Functions, . . . . . . . . . . . Geometrical Significance of a Complex Function.: Notion of Mapping Some Simple Transformatio~ ; .Conformal Transformation: Transformation ~y Analytic Functions. Critical Point of a Transformation Complex Integrals The Cauchy Integral Theorem. Integration in Multiply Connected 'Regions Some Simple Integrals, . The Cauchy Integral Formula . .' . Unlimited Differentiability of an Analytic Function' Taylor Series Laurent Series. . . . . . . . . . . . . . Integration of Function with Singularities:. ~e Residue Theorem ,
404
40S
.407
409
410
411
413 414 41S 416 417 420
422
425
426
428
431
435
437
438
a
440
441
375 376
376
13. CIRCULATION AND'LIFi' FOR AN INFINITE WING IN STEADY FLOW
13.1 13Z 13,3 13.4
15. TWODIMENSIONAL MOTION' AND THE COMPLEX VARIABLE .
IS.1 Complex Potential and Complex Velocity I S.2 Flows Represented by Some Simple Functions .
444 444 446 450 451
13.S
13.6 13.7
13.8
13.9
13.10
14.
Circulatory Flow with Constant VortiCity. Circulatory Flow without Rotation: Vortex Flow Circulation as the Strel"gth of a Vortex Flow . . Stream and Potential Functions for a Vortex Flow Uniform Flow Past It Circular Cylinder with Circ~ati6n Flow with Circulation Past an Arbitrary Cylinder . . . KuttaJoukowski Tlieorem and the Problem oftheJ~irculation Theory of Lift '. r Airfoils, Circulation, and the Kutta Condition ',' The Generation of Circulation. " Mathematical Problem
378 381 382
382
IS.3 Circulation and Source Strength .
IS.4 Flow Past an Arbitrary Cylinder. 15.S Flow Past a Circular Cylinder . 15.6 Complex Representation of Forces and Moments Acting on
452
453
389
389 390 393
an Arbitrary Body: Blasius'Relations
15.7 Force and Moment on an Arbitrary Cylinder:
15.8
Kutta454 456 458
395
ELEMENTS OF THE THEORY OF FUNCTIONS OF A COMPLEX VARIABLE
14.1 ,Geneqil Solution of Laplace's Equation in Two DimenSions:
402 402
Joukowski Theorem .' Mapping of Flows. 15.9 Transformation of Circulation and Source Strength 15.10 Transformation of Flow Past an Arbitrary Cylinder into that Past a Circular Cylinder: Transformation; Conplex potential and complex velocity for the flow past the circle; Velocity fi~ld in the plane of the arbitrary cv:indcr; The pressure field; Force and moment on the arbitrary cylinder: Remarks
458
Introduction of the Complex Variable
Contents 16. PROBLJ1,:M OF THE. AIRFOIL .
16.1 16.2 16.3 16.4 16.5
16~6
Contents 19. ELEMENTS OF FINITE WING THEORY
19.1 19.2 19.3 19.4 19.5 19.6 19.7
nil
535 538 548 548 554 562 564 564
466
16.7 lc>..8 16.9 16.10 16.11 16.12
Nomenclature. . . . . . . Mapping· of the Trailing Edge KuttaConditlon and the Value of Circulation: Lift ontheAirfoil . . . , . . . . . . . . . . . Moment on the Airfoil: Moment at nolift; Aerodynamic center and the moment about the aerodynamic center. . . Velocity and Pressure Distributions on the Airfoil Surface. Transformation of a Circle into an Airfoil. The 10ukowski Transformation . TIre 10ukowski Airfoils , .' . . ... . . Properties ofJoukowski Airfoils.. . . . Other Airfoils: KarmanTrefftz airfoils. Theodocsen's Method tor the Arbitrary Airfoil
466 467· 468 470 471 474 475 477 479
Prandtl's Theory, . . . . . . . . . . Problems of Inte~ . . . . . . . . . Elliptic Lift DistribUtion: Elliptic Wing Solution for the Arbitrary Wing: Trefftz's Method. Forces and Moments on an Arbitrary Wing'. . . . ~tion of the Smallest Drag. . . . . . . . . . Determination of the Coefficients A.. : Methods of and Irmgard Lotz . . .~. . . . . .
. . . Glauret
20.
484
487 490 492 492 494 495 496 499
ELEMENTS OF THE nlEORY FOR THE FWW PAST A SLENDER BODY OF REVOLurION. . .
20.1 20.2
568 568
17. ELEMENTS OF 'i'HIN AIRFOIL THEORY.
17.1 17.2 17.3 17.4 17.5 17.6
.
Fonn~lation of the Problem in Terms oftbe Perturbation Field
20.3 20.4 20.5 20.6 20.7
Simplification of the Boundary Condition . Transfer of the Boundary Condition .' Frame Work of the Theory of the Thin Airfoil Pressure R~lation in the Simple Theory . ; . . Symmetrical Airfoil at IZero Angle of Attack: SoJuti9n by 500 So~ Distribution . . . . . . . . . . . . i7.7 Cambered Airfoil of Zero Thickness at Zero Angle of Attack: . Solution by Vortex Distribution . ' 506 11..8 flat Plate Airfoil at an Angle of Attack: Solution by Vortex 514 Distribution . . '. 516 17.' Aerodynamic Characteristics of a Thin Airfoil .
Formulation of the Problem in Terms of the Perturbation Field . . . . . . . . . . . . . . . . . . . . . . . '.' Boundary Con~ion for a Slender Body of Revolution: Axisymmetric Flow Past a Slender Body of Revolution; Cross Flow or Latera. Flow Past a.Slender ~y of Revob';on Axisymmetric Flow Past a Slender Body of Revolution: Solution by Source Distribution . . . . . . . . . . . . . Cross. Flow Pa:ot a Slender Body of Revolution: Solution by Doublet Distribution . ; . . • . . . . . . . . ; . . . . Pressure Distribution; Axial Flow; Lateral'Flow; Flow at an Angle of Yaw . . . . . . . . . . . Forces on the Body of Revolution . . Moment on the Body of Revolution'
572
574 578 581 584 587 589 601 605
607
.
Selected Problems References .• .
Some
Books
ii. SOME FEATURES OF FLOW WITH VORTICITY.
18.1 18.2 18.3 18.4 18;5 18.6' 18.7 18.8 18.9 18.10
518 518 519 ·520 520 . 523 524 526 528 530 532
Appendix A: Appendix B:
Recapitulation. . . . . . . . . . . . 'io.rtex Line, Surface, Tube and Filament Vorticity Field is a Divergenceless Field. Spatial Conservation of Vorticity: Strength of a Vortex Tube Consequences of the Theorems of Helmholtz and Kelvin . . Velocity Field Que to a Vortex Distribution in an Incompressible F1ui<j.. . . .'. . . . . . . . . . . . . . Velodty Field of a Vortex Filament: BiotSavart La " Simple Applications . . . . . . Vortex Sheet .' . . . . . . . . Solution for the Vector Potential.
Theorems of Linear and Angular Momentum. Characteristics of the Flow Fields of TwoDimensional Source and Vortex Distributions Appendix C:· Poisson's Integral Formulas Appendix Q: Conjugate Fourier Series . Appendix E: SOme: Integrals '
Index . . .
612. 617 621 624 629
Chapter 1
Introduction
Aerodynamics is one branch of fluid mechanics, the science of the motion of liquids and gases, both of ~hich together are known as fluids~ Fluid mechanics is'a very extensive subject encompassing widely diverse topics such as the motion of airplanes and missiles t!trough the atmosphere, satellites through the outer atmosphere, submarines and ships through water, .the' swimming of microscopic organisms, the flow of liquids and gases through ducts, the tran .fer of heat and mass by fluid motion, propagation of sound throus'" gases and liquids, the study of ocean waves and tides, the study of air r .!&Sses in the atmosphere, and many astrophysical, geophysical, anei meteorological problems. Aerodynamics deals primarily with the motion of air, or; ~ore generally, of any gas.'" The science dealing with the motion of water, or, mote generally, of any liquid is called hydrodynamiCS. A great deal of aerodynamics is concerned with the phenomena on which flight (such as that of an airplane) depends, and aerodynamics is usually thought of as the science offlight. The possibility of flight rests on the nature of the force experienced by a body moving through air. This force on the body is usually resolved into two components: one called lift, in a directio~ normal to the flight direction and a fixed direction in the body, and the other called drag, in a direction op'tx>sed to that of motion. In the practical problem of flying lift is desirable to sustain or lift the body against its own weight, whereas drag is undesirable because it hinders the motion of the body, forcing us to expend energy (by means of a propulsive device) to compensate for it. The principal requirement of mechanical flight is to have a body that experiences a large lift and a low drag. Fortunately, there are certain bodies, such as the wings of an airplane, which are capable of prooucing more lift than drag. The study of lift and drag is an important part of aerodynamics. The scope of this book is limited to certain aspects of the study oflift and drag. In. particular. we shall be concerned mainly with the study of flows related to the motion of a wing or an airshiplike body moving with a constant velocity througr. otherwise undisturbed air, the
.. It is then referred to as gasdynamics.
1
IdealFluid Aerodynamics
Introduction
3
magnitude of the velocity being sufficiently small compared to the speedjf sound in the undisturbed air. As is known from observation, essential differences exist between the phenomena underlying the motion of a body through air at a sufficiently low speed and those underlying the motion of the body at a high speed as compared to t~ speed of sound in the undisturbed air. We now describe b!iefly some of the properties of a fluid that are called into play in determining flows pertaining to our objective and outline the approximate basis on which we shall analyse these flows. 1.1 Fluid As a Continuous Medium Fluids, like all matter, are made up of molecules. Thus the properties of fluid motion such as are observed may be studied on the basis of the mechanics of the molecules composing the fluid. Although such a procedure may appear feasibll! in principle, it will indeed be a formidable task to achieve solutions of practical problems. Apart from this consideration, we are generally not interested in the details of the mechanics of the molecules. What we wish to do is to establish relations between various macroscopically observable quantities pertaining to a fluid at rest or in motion. Such observabie properties are called macroscopic or bulk properties. They are mean values in space and time obtained by taking the average over a sufficiently large volume containing a considerable number of molecules and a sufficiently long time compared to a certain time related to the mechanics of the molecules. From the macroscopic point of view this will mean" in many practical flow problems, such extremely small volumes and short time intervals that the variations in the bulk properties of the fluid, whether at rest or in motion, could hardly be observed within those volumes and time intervals. For instance. at 'normal temperature and pressure. a ~olume of 1011 cc (a cube of width 1/1000 mm) Y'ill".contain about 2.7 x 107 molecules. a large number. This being the case, It IS a reasonable approximation to regard a fluid, whether at rest or in motion, as a continuous distribution of matter. We then speak of the fluid as a continuous medium or as a continuum. With such a picture of the fluid, we call an infinitely !lOlall fluid element a fluid particle. We now consider some of the bulk properties of a fluid. In doing so we restrict ourselves to a fluid in which chemical reactions, electromagnetic' processes. radiation, and the like are absent. 1.2 Properties of a Fluid at Rest
ThermodYlUlmic Properties_ The 'properties density. pressure. and temperature at a point in a liquid or a gas in static equilibrium are well
known. To fix our ideas about the nature of a fluid, let us recall the notion of pressure~ At any point it is ~he magnitude of the normal stress acting on an ele~entalplane arc:a passing through that point (see Chapter 3 for more de.tads on the concept c)f stress, pressure, and so forth). It is known from experience that in a fluid at rest only normal stresses occur and thilt in .general. they are "compressive" in nature. When only normal stresse~ ~ur~ it can ~ shown easily that at any point they should be equal in all dlr~..ons..It IS ~l~o .a matter of experience that the force exerted by a flUid m static eqwhbnum on a solid body submerged in it is due to only normal stres.ses acting on the surface of the body. . !he factthat.only normal stresses occur for a fluid in static equilibrium IS ~n contrast With the state of affairs for a solid in static equilibrium (Doth bemg under the action of external forces). It is known that for a solid in static equilibrium both normal and tangential stresses occur in general. ~e absence of tangential stresses in a state of static equilibrium is what distinguishes a flUid from a solid and may be considered as the property that defines a fluid. For a fluid in static equilibrium; besides pressure. density, and temperature, .t~e properties such as internal energy, enthalpy, and entropy are also fa~har concepts. These various thermodynamic properties and others h~e them a~e not all.indepe?dent of each other. As shown by observations, functional relations eXIst between them. Such relations are known as' characte~istic equarions or equations of state. For example, p == p~T, where R IS a. co?stant, is an equation of state for a perfect gas. Equations of state for hqulds and other gases are not of such simple forn.
Ali fluids undergo changes in volume under changes and tem~rature. For fluids in static ~quilibrium the changes l~ pressure result from the ap~lied external forces and the changes in tempera~ure result from nonumform heating of the fluid. The ability for changes In voTUI'M of a mass offluid is known as compressibility. It is well known that gases are more easily compressed than liquids. When a fixed mass of a flui~ undergoes ch~nges in volume, its density also changes. !bus the capacity for chan~s m the density of a mass element of a fluid IS also known as compressibility. Under normal circumstanCes, in liquids, ,the changes in density due to pressure chan~es are practically unobservable. Density changes due to temperature d~fferences are, however, not negligible in general. If the t~m~tature differences are sufficieotly small, the density changes in a liqUid are almost nil and the liquid may then be regarded as an incompressible fluid. An incompressible fluid is one whose elements undergo no changes in volume or density.
~f pressure
Compres~.
IdealFluid Aerodynamics Compressibility is easily noticeable in gases. In certain circumstances, howeve,r, the changes in volume or density of an element of a gas may be negligibiy small. In such a case, as a reasonable approximation, the gas may be regarded as an incompressible fluid.
Introduction
Hellt Co_ctio.. When a fluid in static equilibrium is heated nonuniformly, heat may be transferred (without causing motion of the fluid) from points a1 which the temperature is high to those at which it is low by what is called thermal conduction. Consider a surface element situated at some point in the fluid. Observations show that under usual circumstances the heat flux (which is the amount of heat transferred across the surface element per unit time per unit area) in the direction of the normal to the element is proportional to the spatial rate of change of ~he temperature at that point in the direction of the normal The heat flow occurs in the direction of decreasing temperature. Thus, if q.. denotes tbe heat flux and aT/an denotes the rate of increase of temperature with distance in the direction of the normal, we have
q
..
Frktio" or V_os;ty. It is a mattet of experience that even smoothly shaped bodies· moving with a constant velocity through an otherwise undisturbed fluid encounter a resistance to' their motion. Similarly, a fluid flowing through a pipe offers resistance. These observations suggest that for a fluid in motion tan!ential stresses in addition to normal stresses
10.0 9.0
8.0
7.0
6.0
Thermal conductivity for air at 1 atm I )C lOS (calorie secI emI 0C )
==
kan
aT
.if'
(1.1)
5.0 4.0
3.0
Thetmal conductivity for water at 1 atm
. . . . . )C
lal (calorie secI em 1oC I )
where k is a proportionality factor known as the coefficient of thermal conductivity or simply as the thermal condu(,.tivity. It is a material property of the fluid, and thus its value differs from fluid to fluid. The thermal conductivity of a fluid is always positive and, in general, a function of temperature and pressure. Iri the context of macroscopic considerations k has to be known from experimental observations. The temperature variation of k at atmospheric pressure for water and air is 'shown in Fig. 1.1. It is found that the static equilibrium of a fluid in wl,ich the temperature is not constant is unstable unless certain conditions are fulfilled. This instability leads to the appearance of convection currents in the fluid. which tend to mix the fluid in such a way that the temperature is equalized. t.3 Properties or a Fluid in MotioD The density of an element of a fluid in motion is defined in the same way as a fluid at rest. The concepts of pressure, temperature, and other properties as known in thermodynamics are assumed to apply equally well to a fluid in motion. It is further assumed that the equations of state obtained for a fluid in thermodynamic equilibrium are equally valid for fluids in motion .. The reasonableness of these assumptions rests on the fact that for many flow problems they lead to results that are in satisfactory agreement with observations.
2.0
1.0
°O~~~~~~~~_ll~I
200 . 250 300
Temperature rC)
Fig. 1.1 Thermal conductivity of water and air.
occur on an,)' elemental plane passing through a point in the fluid. The appearance of these tangential stresses only when the fluid is in motion constitutes the phenomenon of internal friction or viscosity in a fluid. Such stresses give rise to a resistance to nonuniform motion of a fluid.
• For example, a bo<!y of revolution moying in the direction of its axis or a thin flat plate moving in its {llane.
6
IdealFluid Aerodynamics
Introduction
7
Nonuniform motion refers to the situation in whic::h the velocity differs from point to point in the tluid. In general, viscosity gives rise not only to resistive tangential stresses but also to resistive normal stresses. Such normal and tangential stresses are called frictional or viscous stresses. They appear only when the tluid is in, nonuniform motion and disappear when the non uniformities disappear.. They do not occur at aU when the fluid' is in static equilibrium. For a fluid in motion it is assumed that the viscous stresses occur over and above the normal stresses associated with pressure. For a wide range of tlow situations observations tell us that, because of the phenomenon of viscosity, there can be no relative velocity at all between a moving tluid and a solid body at their surface of contact. Coe.fliekllt of J'ucosity. Consider the following experiment. * A, tluid is contained between two parallel plates of indefinite extent that are placed at a small distance h (see Fig. 1.2). One of the plates is at rest
visCosity of the fluid and is called the "oefJicient of shear viscosity or simply 1M coefficient of viscosity. Equation (1.2) states that the viscous stress is proportional to the average spatial rate of change of the velocity in, theftuid over the distance h. Now ,consider a similar situation in which the fluid is moving in parallel plane layers with the same direction everywhere. Let y denote distance from a fixed point measured perpendicular t,o the layers and let' u .denote the velocity ofa fluid layer at a distance y. With u as a function of.y, the fluid is said to be in simple shearing motion. There is a shearing or tangential stress between adjacent layers of the fluid. In light of the interpretation of Eq. (1.2), wcrassert ~hat the sheantress T at any point of the fluid for the motion un~er consideration is given by , du (1.3) '1"p'dy where ,A, as before, ill the' coefficient of viscosity for the ftuid in question. In generalizing (1.3) for a tluid in a more general motion,it is assumed that in a wide range of flow conditions the viscous stresses are linearly related to the rates of strain in the fluid. These rates of strain are given by certain combinations of the spatial derivatives of the velocity components. In these relations there appear two factors of proportionality, both of which are referred to as coefficients of viscosity. One of thefu is the shearviscosity cOefficient p already discussed, and the other is expressible in terms of p and the bulk modulus (also knownas .thecoejJicient of bulk viscosity) of the fluid. In many investigations of fluid flow it is assumed that the bulkviscosity coefficient is zero. For motions in which the flui~ may be regarded as incompressible it turns out that ·because there are no vo'lume changes the viscous stresses do not contain any terms involving the bulk viscosity coefficient. Thus in many fluidflow problems only one viscosity coefficient ,occurs, and this is p. Fluids characterized by a relation of the form (1.3) are called Newumian fluids. The coefficient of viscosity" which henceforth shall mean only the shear viscosity p. has different values for different fluids and fora particular fluid is, in general, a function of both temperature and pressure. In the context of macroScopic considerations, viscosity, just like thermal conductivity, has to be known, from experimental observations. Within a given range of temperatures and pressures the dependence of I' on temperature is markedly noticeable, whereas that on pressure is hardly seen. For gases I' is an increasing function of temperature; for liquids it is a dec:.reaslng function of temperature. 'For both liquids and gases p tncreases slightly with pressure. theclJanges in p being very small for pressure changes that are likely to occur in many circumstances.
Fla. 1.2 Illustrating definition of ¥iscosity.
while the other is moving, with a constant velocity l/ parallel to itself. Because of viscosity, the fluid will also be in motion, its velocity at the moving plate being U and that at the stationary plale being zero. Resistive tangential stresses occur in the fluid and at the plates. Experiment shows that over a wide range of conditions the tangential ctress T acting on either of the plates i,s proportional to the relative velocity between the plates and inversely proportional to the distance h. Thus we have
U '1"= P h
(1.2)
wbere I' 'is a factor of proportionality that is inde~ndent of U and h arid depends only oli the nature of the fluid. This factor is a measure of the
• Such an ~xperiment is perhaps not easily realizable, but it affords the ~implest example for our purposes. A realizable experiment analogous to the one under consideration is that o~ the ,motion of a fluid contained in the narrow annular space formed by two concentric cylinders, one of which rotates at a constarit speed and the otheris stationary.
,
IdealFluid Aerodynamics
Introduction
5.0
9
'.
Table 1.1 gives the values of p for some selected fluids, at 15°C and atmospheric pressure. For air and water, which are of particular interest to us the variation of p with temperature at atmospheric pressure is shown in Figs. 1.3·and 1.4. As we shall see later, the effect of viscosity on the motion of a fluid is determined by the ratio of p to tfte density p rather than by p alone. This ratio pIp is known as the kinematic viscosity and is usually denote.d by". Table 1.1 and Figs 1.3 and 1.4 include the corresponding values of"..
4.0
3.0
Table 1.1. Viscosity p and Kinematic Viscosity II, in CGS Units,jor Gases and Liquids at 15°C and Atmojp~eric Pressure
II x
2.0
AJr
Kinematic viscosity I'X 10 (cm' secI) viscosity IA x 1()4 (dyneseccm 2)
10'
II
x 10
1.5
Air Nitrogen N. Oxygen O. Hydrogen H. Helium He Neon Ne Argon A Carbon dioxide CO. Water H.O Mercury Hg Paraffin oil Olive oil Glycerine Castor oil Pitch
0.18 0.17 0.20 0.09 0.20 (l.31 0.22 0.14 11.4 16 ..0 200 1000 13,000 15,000
~101l
I.S I.S 10.5 1.2 3.7
1,3
OL__
100
~~
__
~
____
~~
__
~~
__
~
0
Temperature (Oe)
Fla. 1.3 Viscosity.and kinematic viscosity of air.
0.77 0.1.4 0.012
2.0
2.S
10 100 ISO
~IOll
1.5
Water Kinematic viscosity I' x 102 (em' secI) viscosity IA x 102 (dyne sec cm 2)
Compressibility. Density and volume changes for an element of Ii fluid in motion a(ise, as in a fluid at rest, from temperature and pressure changes~ When a fluid is in motion, these changes occur because of various factors in addition to external forces and nonunifor~ heating of the fluid. These factors are those that affect the motion of a fluid element. Pressure changes would result from changes of mdftlentum ofthe element and the action of viscous stresses, in addition to the action of the external forces. Temperature changes would result .from exchanges between kinetic and internal energies, which are due in turn to work done by external forces, pressure forces, and viscous forces, in addition to nonuniform heating. The situation in general is rather complicated. Later we shall examine the parameters that characterize the compressibility of a
1.0
0.5
Temperature (·C~
Fig. 1.4 Viscosity. and kin:matic viscosity of wat~r.
10
IdealFluid Aerodynamics
Introduction
11
fluid element in motion. At that time we shall also see under what conditions the c0mpressibility of the element may be neglected. fleat Transfer. It is a matter of common experience that a heated body immersed in a moving fl~id cools considerably more rapidly than one in a fluid at rest in which heat transfer is accomplished only by conduction: in a moving fluid heat transfer is accomplished by the combined action of conduction and comJection. By 'convection we mean the motion of a nonuniformly heated fluid. If the motion is caused by the action of gravity alone, it is known as free con,.ection. If the motion is caused by agencies other than gravity, it is known as forced COfll'ection. The velocities occurring in free convection are small. Generally. forced convection is the governing process for heat transfer in a flowing fluid. When heat transfer occurs between a flowing fluid and a body immersed in it, convection is absent at the surface of the body it'ielf. The transfer of heat between the fluid and the body takes place primarily by heat conduction ihrough a thin layer close to the surface. U!'dcr normal circllmstances it is found that at the surface of the body the temperature of the fluid is equal to the temperature of the .body. t.4 Laminar aDd Turbulent Motion llis. a matter of observation that there are two fl,lndamentally different types oftluid flow, which we call/aminor and turbulent flows. T!1e difference between the two flows is readily illustrated by the famous· experiment of Reynolds (1883) .. He observed the nature of the flow of water tbrough a long glass .tube connected to a reservoir. The observation was m~e by introducing a dye at the entrance of the tube .. At small velocities the dye forms a thin, straight thrl'ad paraliel to the a.xis of the tube. indicating that the flow is steady and orderly in nature. This t)pe of flow is referred to as lamin~r flow. As the velocity js graduaU)· increased at a certain velocity. (depending on the dimensions of the tlJbe), the flow suddenly changes in character; the dye thread becomes violently agitated. and th.: dye soon spreads over the whole tube. The flow becomes one of an irregular character. We call such a ftow.turbulent ftow. These observations are ilh;strated in Plate 1. Regular laminar motion is ex~:eptional in nature: turbulert( flow ·is much more common. This is true for the flow ofwati=r in a river. for.a movtn!; stream of gas, as for the atmosphere which. as a .whote., may be at feSt. Turbulent motion. unlike laminar motion, is cllllracterized by rahdom fluctuatio~s at a fixed point in the Ruid' quantities such as velocity and pressure. Such fluctuations result in pro.counced mixing. ;Fluctuations in ftuid velocity may be detec~ed by what is called the hOH..ire anemometer
Reynolds No. 2000
Reynolds No. 1500
ReynOlds No. 3000
Plate 1 Illustrating Reynolds' observations of laminar and turbulent flows of water through a glass tube. Courtesy of Professor F. J. Bayl~y and WileyInterscience. Figure 6.2 of F. J. Bayle)·: ¥ Introduction 10 Fluid Dynamics, 1958.
Introduction
13
,
I
j
(0)
(sec, for instance, Dryden and Kuethe, 1929). When fluctuations exist, a mean velocity may be defined by taking an average over a time sufficiently large .compared to the time scale of the fluctuations. By vl:locity in a turbulent flow we mean such a mean velocity. The turbulent fluctuations are generally composed of a wide range of frequencies. The magnitude of the fluctuations can range from extremely small, almost immeasurable fractions. of the mean speed to as high as 30 to 50 per cent of the mean speed. The random character of turbulent flow is illustrated in Plate 2, which shows records of the velocity in two different parts, laminar and turbulent, of a thin rectangular air jet issuing from a slit. When such a jet impinges on a wedfe placed a short distance from the slit, a sound tone, known as the edge tone, of discrete frequency is generated. In such a situatiqn the fluctuations in the fluid velocity are of discrete frequency. A record of these fluctuations is also included in the Plate 2. The analysis of turbulent flow requires special methods. In our studies here we will have no occasion to enter into tbe details. I.S Some Relevant Parameten We now seek the parameters ~hat characterize the effects of external forces, viscosity, ·compressibility, and heat transfer in a fluid 'in ,motion. For this purpose and to fix our ideas we consider specifically the problem of a rigid body moving with a constant velocity through, an infinitely extending fluid which is initially in a state of static equilibrium under the action of gravity. Gravity forces are the only external forces acting on tbe fluid. We assume'that in the undisturbed state the temperature and the internal energy (measuted per unit mass) are constant throughout the fluid. W~ denote this temperature and internal energy by T and e, respectively. The pressure .and density in the undisturbed state would vary from point to point of the fluid. Let p and p denote some reference values ofthe pressure and density in the undisturbed state. The geometric sbape of the surface of the body is fixed. The surface is then specified by a length I characteristic of the body. Let V denote the magnitude of the constant, velocity with which the body is moving. When a body is set into motion with a constant velocity through an otherwise undisturbed fluid, after a sufficient length of time the surface of the body would attain a temperature different from that of the undisturbed fluid. This is so even if·the body and the fluid have the same temperature before the body is set into motion. For the following considerations we. assume that, when the body is moving with constant velocity tl)e temperature of its surface is maintained at a constant tctmperature T", different from the temperature T of the undisturbed fluid.
(b)
(e) Plate 1 Nature of velocity ftuctuations: (a) in the laminar region of an air jet; (b) in tbe turbulent region of an air jet; (e) in the jet region of tbeftow field of \In edge tone of di!ICrete frequency. Time along the horizontal axis and speed along the vertical axis.
12
IdealFluid Aerodynamics
Introduction an elemept as shown 'in Fig. I.S and the component of the pressure force in the zdirection.The magn"itude of this component IS equal to
cp dz dy dz == op In
Relilt;ue Mllg,,;tudes of tile Force.. Fr'tJIIM NlUllber, ReYlIDlIb NIIIrIber, IIIIIl Mllc" Number. We, now make an estimate of the relative magnitudes of the various forces acting on a fluid element in motion. According to Newton~s second law of motion, we have
the time rate of change of momentum of a fluid element == the gravJly force + the, pressure force + the viscous force acting on the element (1.4)
,
oz
oz
Assuming that all components of the pressure force are of about equal magnitude, we say that the pressure force on the element is prt)porJionalro
tl1
s
~In
I
where lJp is a representative change in pressure. d ~ h FJa. 1.6 Viscous force on an To ~timate the magnitu e 01 t evis element. cous force, we proceed in a similar way. As shown in Fig. 1.6, we consider only the contribution to the zcomponent of the viScous fon;:e due to the appropriate shear stresses acting on the yfa~§ (the faces normal to y) of the element. Thus the, zcomponent of the viscous force is given by
,
Fla. 1.5 Pressure force on an element.
The time rate of change of momentum df the element is the negative of the inertiol force acting .on the element. The equation of motion thus. expresses simply the fact that the result of all the forces acting on the element is zero .. We obtain an estimate of the relative mapitudes' of these forces as foilows.· Consider a fluid element of volume 6T. The speed of the element is proportional to V, the speed of the body. .The change in momentum of tbe element is proportional to' pV6r, Where p is the density in the undisturbed state. We may say that this change would occur in a time interval that is proportional to tl}e time II V taken by the body to move £. distance equal to'the characteristic length I. Thus the rate of change of momentum of the element or the inertial force acting on it ;$ proporti01lll1 to
. pV6. == pV InIJV I
I
~ dz dy dz _ a,. lJi 81/
ar
Now, we may $8Y that the,spatj8irate of change a."lo, is proportio~al to TIl. where ." is the shear stl~ss 'at the' point the element is situated. Furtherinou:, we say that
.,.. 8it
whc.re u is the velocity in thczdirection, and that
811 T_U
.au . V 011 I
We then have
let g denote the force due to gravity, measured per unit mass. Then the gravity force Qctingon the flUid element ;s I!tpmf to pg~T. To estimate the pressure force, we examine the order of magnitude of one of the components of that foret. For this purpose we consider
• In the followiog di..ccussion we are concemecl only with the magnitudes of the various quantities under consideration.
811
a,. In,...., IL f In
,.. II
We assum~ that all components of the viscous force 'on the element are of about the' same magnitude, and the visco.w force on the element may be said to be proporti01lll1 to (PJlIJl) 6'1'.. .
16
IdealFluid Aerodynamics
InirocauctiOo
and
1'1
The r~lative magnitudes of the various forces on the element are then
a~lows:
inertial force: gravity force: pressure force: viscous force" =
pVI. pg.lJP.1' ~
V· V· """'pIp a l
The nond~mnrsional parameter VIa, where a is a characteristic sound speed in .the undisturbed state (P, p), ;s known as the Mach number and denoted by M. We thus have
(1.9) and
I'
. I . II
(1.5)
The ratio of any 'one of the forces to another is a nondimensional number. The various nondimensional numbers obtainert by taking the ratio of the inertial force to each of the others play an important part in fluid mechanics. They are known by special names after the scientists who have initially pointed out their importance. The ratio of the. inertial force to the gravity force is, given by the parameter valgl, whose square root is usually known as the Froude nUDlber and denoted by Fr. Thus we have Fr=and
V
M' _1___ V· __ inertial force lJp/p' lJplp pressure force
For a perfect gas (for which p ~ pRT) with constant specific he~ts it is 'found that '
.fil
(1.6)
a l =yE
p
and tliat therefore
(1.7)
FcI inertial force gravitl force
Vi =yM' pIp where y is the ratio of the specific constant volume.
h~t
, The ratio of the, inertial force to the viscOus force is given by the parameter pVIII'. This is known as the Reynolds number and usually denoted by R. We thus have R
at constant pressure to that at
==
pVI I'
= VI __ in~rtialforce
" viscous force ,
(1.8)
where ." is the kinematic viscosity. The ratio of the inertial to the 'pressure forces is given by
=lJpI p
VI
V·
1
pi p lJplp
It can be shown that the speed of propagation of small (soundlike) disturbances through' a. fluid that is initially at rest and in a uniform state. is proportional to (PIp)"', wbere p and p. are the pressure and density in the undisturbed state. Such speed of propagation is known as ' the speed of so~nd. Denoting this speed by a, we have
P~lIIMt., c.""'mmzbtg Compr(uibiIity. The changes in volume or denSity of a ftwd element, as. already pointed out, arise out of pressure and temperature changes brought about by the motion. Thus the factors that characterize compressibility arc those on which pressure and temperature changes depend: We shall now examine these factors. On the basis of the equation of motion' (1.4) and the c~nsideration thai the. pressure fon:e per unit volume of the element is proportional to' the ratio. of the ch.ange lJp in the pres~ure to a distance, we' may say that lJp/~ IS proportional to each of the forces, inertia; gravity. and viscous.: actmg on the clement. Thus, we have
lJp,...., pV',
pgl,
For the relative change in pressure, lJplp, we therefore have
l!. "" a'
p
• This implies the absence of any .external fon:es.
,
gl pIp
~V
Ip
(1.10')
1.
6p ,....,M 1 p ,
IdealFluid. Aeroclynam~_
If
where If is the thermal conductivity and T. is the temperature. With (T  T.> being the temperature dUl'ereIlCe between tbe undistUrbed·f1uid· . and the body. we may say that is proportiOll8l to (T  T.>/f. This being the case, the heat low through the .•faCes and thuS. ,'" rat. 0/ h«lt additio" tot'" element by heat C01UIuctio,. may be IIlid to, be prupor·
Introducing the nondimensional parameters Fr, M, and R, we niay state ~ I I
'M FrI'
M R
(1.11)
arIas
To seek the factors influencing temperature changes we note tHat sucb changes arc proportional to tbe changes in internal energy of the element. The 'factors influencing the changes in internal energy are. exhibited by the equation of energy, which states the rate of change of internal energy of an element + th~ rate of change of its kinetic energy  the rate at which work is done on the element by the gravity force + the rate at which work is done on the element by the viscous forces + the rate at which work is..done on the element bi the pressure forces + the rate at which heat is added to the element by· heat conduction (1.12) Denoting by ·tie the change in internal energy per unit mass of an element of volume 61' ~nd assuming that the various changes takc:piace' in a time interval proportional to IfV as before, we may say that the rate :of change of internal energy ,...., tie the rate of change of kinetic energy,..."
~/~
.
Ic(T T.) tw
,.
f.
1/~
7~/t
f
V 61'
/I VI
the rate at which work. is done by the gravity force ,...., g Vp 6r the rate at which work is done by the pressure force ,....,
.
•
,.. 1.7 ..Heat floW into ... elemeftt.
tbe rate at which work is done by the viscous force I I 6r To eitima~ the rate of heat addition by c:Onductiob we COilsider the ~~t shoWn in Fig. ':7 and the beat flow through its z·faces. OeDoting by 'I. the·8wt through the face situated at z; the amount of heat added tbe element per unit time is given by
For thC relative changes of the internal energy or,equivalently, of the temperature of the elem~t we therefore have' .
liT lie V· or ,....,T e 2e'
Introducing the speed o( sound a·.andthc nondiDlellsional parameters Fr, M~ and R, we may state that !1 T or lie,,:MI at
k(T T.. ). pVle (1.14)
to
~dzdyd% _ aqe6r
a~
az
We ~y that 8q,,/az is proportional to '1./1. According to the equation Df heat conduction (1.1). we have q  k
uy
T
e ' 2 e'
e
•
aT az
Since the relati~ changes, ApIp. in the denSity of the. el,ement arc . proportional to the relative changes in pressure and temperature. we
IdealFluid Aerodynamics
Introduction Prfllllltl NlIlIIber. may write
21
For any ftuid, introducing the Prandtf. number a, we
k(T TID) . 1 CpTT TID =pYle aR e ' T
obtain by using (1.11) and (1.14)
M'
Fri"
Mla l ;. k(T  T.) Re pYle (US)
These. then, are the parameters that ciUiracterize the compressibility of a ftuid ciement. Among them are only five independent parametersMI. MI/Frl, MI/R; al/e, k(T  T ..)/pVle. For a perfect gas with constant specific heats we have
The Prandtl number, like the viscosity and the thermal conduc (vity, is a material property and it thus varies from lIuid to lIuid. For any particular lIuid it is in general a function of temperature and presSur~. For gases it is of the order of unity; for liquids a varies more widely. For very viscous liquids it may be very large. Some representative values of a at 20"C and I atm pressure for various substances are as foUows: Air Water Mercury GUycerine
0.733 6;75,' 0.044 7250
a l == "(,,  l)e
and
e
CpT
where C. is the specific heat at constant pressure. We may then write
"
 == "(,,  1) e
and
a
l
For air at atmospheriC pressure a varies from 0.719 at _50°F to 0.722 at 600°F. For water at atmospheric pressure a varies from 6.88 at 68°F to 1.7 at 212°F.
k(T T.,) . pYle
,,k
I' T T., I'C. pVI T k 1 T Tio =)tI/.C.,R T
The ratio I'Cp . /k is neodimensional and is known as the Prandtlnumber, • usually denoted by (1, ~ntroducing a, for a perfect gas W1Ut constant specific heats we have k(T T..) 1 T T.. = ,,pYle aR T Thus for a perfect gas with constant specific heats we may state that
PulUMter."II W1tici Force _'Heat TrlllU/er Depellll. Let us deno.t~ by Fthe magnitude of the force acting on the body and by Q the magnitud'e' of the heat transferred to the body per unit time. We may regard F as meaning the total force or a compdnent of the total force in some chosen direction. We expect the force and the heat transfer to depend on the propertiesg,p, P. T, e, 1', k, V, I, T., tpe shape and orientation ofthe Body. We reprd the geometric shape of the boundary of the body 'as fixed. The shape js then completely specified by the characteristic length I. The orientation of the body is specified' by two angles, for exampk~ IX and P. which the velocity vector forms...with the axes of a coordinate system fixed in the body~ The force and heat transfer may then be expressed in the following funct~onal form:
F
and
= F1(a., p, I,
V,T.. , g,p,p, T, e, 1', k)
(L17)
dp,...., M', p
MI Fr2'
M2
R'
(" 1) Y MI . 2'
MI "(,,  1) F ' I r (1.16)
Q == F.(IX, p, I, V, TID' g,p, p, T, e., 1', k)
By using the methods of dimensional analysiS· we may further express these functional relations in nondimensional form. To describe the force and the 'heat transfer and the 12 quantities on which they depend we employ the four fundamental dimensions: mass, length, time, and
• For the methods of dimensional analysis seethe cited references.
L.. T TID
aR
T
For such a gas the five independent parameters characterizing compressibility are
:; , F r
y(y .... l)
11
IdealFluid Aerodynamics
Introduction
lJ
temperature. This means that according to dimcnsional analysiS$he force and h~t transfer, suitably nondimensionalized, may each be cxprcssed as a function of eight independcnt nondimensional parafletcrs formed from thc 12 variables «, p, I, V, T., g, p, p, T, e, JJ, and k. Since the combination p VIII has thc dilflensions of a force, we introduce the parameter F/ p VIII in place of F. Simi.larly, because Ik(T  T.. ) has the dimensions of encrgy per unit timc, the same as those of Q, wC introduce the paramcter Q/lk(1'  T..) in place of Q. Thc cight independe~t parameters on which F/pVlII and Q/lk(T  T.) depend may be given in different forms. Wc chooSe those (orms that would lead us to the parameters which have already appeared in our considcrations. The rcsults may be cxprcssed as follows:
1.6 RaDge of Some Parameters The problem we have been considering, namely that of Ule motion of a body with a oolIStant speed through a fluid, covers a widc range of lengthscales, speeds, and fluids. Thc lcngthscales may vary from about 103 mm to about 300 meters. Thc speeds may rapge frQrn It few centimeters per second (or even less) to thousands. Fluids,with widely differing values of kinematic viscosity may be of concern. This being the case, the range of Reynolds nu~bers is indeed very wide. Rcynclds numbers from about 103 to about 107 to 108 occur in practice •. The motion of aircraft through the atmosphere involves lengthscales of about I to 50 meters, speeds of about 30 meters/sec to about JOOO meters/sec, and Reynolds numbers of about I" or 10' to J07 or 10 8 • The motion of submarines and torpedOes through water in~olvcs lengthscales of about 5 to 100 meters, speeds of about 10 meters/sec, and Reynolds numbers of about LO·. Our concern in'this book is with problems of this type and thus with motions involving high Reynolds numbers, R> 10". With speeds and !engthscales varying over a wide range, the ra;'ge of Froude number, Fr = V/J gl, occurring in practice i~ also wide. For fluid motions involving aircraft the Froude numbers may have values of 1 to 100 or more. For submari'ne and torpedo motions Fr may have values in the range of 0.3 to 1.5. We now consider the raoge of Mach numbers. Under normal conditions the speed of sound in gases is 'about 300 to 400 meters/sec. In liquids the speed of sound is much higher. For instance, 1t normal temperature, the speed of sound in air' is 370'fneters/sec, whereas in water it is about 1600 meters/sec. Speeds of bodies mo_ving through liquids are usually limited to small val~es ..Possible maximum speeds are in the range of 20 to 30 meten/sec. This bemg the case, the Mach numbers in liquid flows are extremely smallthe order of 0.01. Speeds of bodies moving ~rough gases, on the <?ther hand, rna' range from considerably small values to large values compared to the speed of sound. This means that the Mach numbers for gas flows may va'ty from values much less than unity to values much greater than unity. 1.7 Conditions for Neglecting Compressibility Effects We have seen that the changes in density of a fluid element come about from pressure and temperature changes arising out of the motion 0f the fluid. Density changes caused by pressure changes would be negligible if
• See Lighthill (1963).
pyl ,. = 11
F
( «,
V pYI. V p kT T. ') p, Jil' ; , JpJp' pe' JJe '~
Q _~( lk(T  T.)  I «,
P~
'Jr/'
pYI
JJ
'JPTP' pe , JJe '
l
~1. kT T.)
T
(1.18)
Introducing Fr, R, M,
CI,
a and C p , we,may write (1.18) as
a 1ST T.. ) p, Fr, R, M, ,, T
F .;; = /1{ «, pY r ·
e
CI
e
.
Q
lk(T T.)
== II ( «, p, Fr, R, M,!. ,~.CpT, T.)
e
CI
I '
(1.19)
e
T
The determination of the functions J,. and Ji is the central problem of many theoretical and experimental studies in aerodynamiCS qJ1d hydrodynamics. For a' perfect gas with constantspecifi~ h~ts wc have
a  == e
l
y(y  1)
CpT , =y e
In such a case (1.19) may be expressed in the following form: F ;; == /1 ( «, p. Fr, R, M, y, CI, T.. )
pVI
T
lk(T T.)
Q
=11(<<' p, Fr, R. M, y, (f~ Till) T
(1.20)
24
IdealFluid Aerodynamics
Introduction of a fluid clement. The motion then is essenti!llly one of an incompressible fluid. C. . 0/ GtIH.. In the motion of gases the Mach number may rsoge from values much less than unity to values much greater than unity, and, in general, density changes arise from both pressure and temperature changes. However, when the Mach number is much less than unity, that is, wherl the speed is small compared to the speed of sound, and there is no heilt transfer between the fluid and the body, the changes in the den~ity of an clement would be negligible. Under such circumstances the motion of a gas may also be regarded as that of an incompressible fluid. 1.8 Coaditioas for NeaiectiaI GnYity mef;ts The force due to ~vity on a fluid element may be neglccted in comparison with the inertial force, the pressure force, and the viscous force when the following conditions arc fulfilled: Frl _ Vi __ inertial force» 1 gl gravity force ~ Frl _ a l ~ __ pressure force» 1 p. MI gl p gravity force
the following conditions were satisfied (refer to relation 1.10):
L",!::'
.JiTP
M«l
_ MI
a
..l.!.. "" gl
«1
pIp a l Fr' Y M' ~"" .« 1 Ip R
(1.21)
Density changes caused by temperature changes ~ould be negligible if the following conditions were fulfilled (refer to relatIon 1.1~):
__ VI
e
e p pe
MII~«i.
I
e
l
gl _ M' a «1
Fr' e
l
a "'« 1 e '" V = M' a « 1
l
pie
Re
k(T  T",) == .!. C,TT  T. pYle aR e T
«I
Fr' _ v V ;.." viscous force» 1 R gil gravity force The first two conditions are satisfied whenever the lcngthscale and the speed arc such that both V and a are much greatcr than Sinee g is l , the speeds Vand a should be greater than say 3.2 ,!i about 9.8 meters/sec meters/sec. The condition that a » is one of the conditions for neglecting the compressibility effccts due to gravity, the other being that e »gl.
By putting these conditions together we ~ay stat~ ~hat compressibility effects would be negligible under the followmg conditIons: M« 1
Jii.
« 1 Frl
aR e
yl
« 1 R
T
MI
J'ii
(1.22)
!.« 1
e
I
.!. C.,TT T.« I
1.9
Nature of the Problem wbeo Compressibility Eft'ects are Negligible
It may be noted that the parameter C.,T/e is of the order of unity.
ClUe 0/ LitpIids. In the motion of a liquid the Mach number is very small, M ~ 0.01. The parameters gl/a 2 and '" Vjlp are also extremely small. Consequently, in the motion of a liquid the density cJtanges caused by pressure changes af'e indeed negligible. This, however, does not mean that there are no den.:ty changes, for they may result from temperature changes. Temperature changes occur primarily in heat transfer between the fluid and the body. In the absence of heat transfer we ~y SUite t~at the motion of a liquid occurs without any noticeable changes m the denSity
We have seen that when the Mach number is much less than unity, Fr and R arc greater than unity, and that when there is no heat 'transfer compressibility effects are negligible. Under such circumstances the problem of a body moving with a constant speed through afluid mf.y be regarded as that of a body moving through an incompressible fluid. The. main objective of the problem, then, is that of determining the force on the body .. The functional dependence of the force is represented as follows~
~
F = ft«(J., pV I
p,
Fr, R)
(1.23) .
Introduction When tbe motion can' be·regarded as incompresaible, it is possible (as we shall sec later) to separate ~ effect of gravity fro~ the dynainical problem. of thcmotioJ,l of the fluid. Gravity e~ may be ~unted lor by considcriR'g the cortcsponding stalic:situation. This bein8 the case, the Froudc number may be ~ from consideration, and'the problem reduces to that of detcnn:iniog the Coilowing fgnctional relation for the force: unbounded fluid the streamline pattern appears the same at all times if observed from a frame fixed with respect to the body. In fact, it is, equivalent to that due·to steady.ftow past a body that is kept stationary, and the velocity far ahead oCthe body is equal to the negative of the vel9City of the. body in the initial problem (see Fig. 1.8). By steady flow we mean that
F p~"

" f(a.,'p, R)
(1.24)
•
Our concern is with flows at high Rcynolds numbers, that is~ for R ~ lOt. For such flows and for certain types of body that 'arc of practical interest, and for certain orientations of these bodies, the effects of viscosity arc generally confined to a very thin layer surrounding the body. ' In the major' part of the flow the viscous effects are negligible~ In such a situation it is, possible to analyze certain imPQrtant aspectS of the problem by assuming that the viscosity is zero or equivalently that .the fluid is frictionless or iIIvisciti. The Reynolds numbc:r then does not eriter the problem and the functional depe~"ence for the force is simply pyip== f(a.,
.,~,' ~..
.
.
.
.
.. ...... . ," _..
. .

",
,
Fla. 1.8 Steadr flow past a fixed body.
the velocity at every point is constant in time. In the following we refer to our problem as that of steady flow past a stationary body. In steady flow' streamlin~s and paths described by ftuid elements are the same. In describing the variation of flow patterns with Reynolds number we concern ourselves with situations in which cOQlpressibility effects are negligible, for this is our main concern. Later we describe some q1,lalitative features of the effects of Mach number on the flo~ pattern at high Reynold~ number.
F
p)
(1.2S)
The central problem in this book"'is the deteonination of the function f for ,certain types of body such as wings.!lDd 'airshiplike bodies. . AlthoughundCr certain circumstances the role of viscosity and that of the Reynolds nuniber ~uld be, dropped fr()m direct con~iderations, it is necessary alwa~ to bear in mind the important role played by visCosity or, more significantlyI the Reynolds number (for as we h'ave ~n, what ma~ is not Yiscosity per se but the n()ndimensiorull parameter Vlfp),jn detenpiniJ,lg the nature of the flow arid the typo of ameaningful,approximate treatment that may be attempted fOf the problem. To·~ some appreciation in this d~on we describe, briefly'in tJtc next section some ql,la1itative features of certain flOlfS a~ different Reynolds numbers.
Flo", Past Bluff Bodies. We consider twodimensional' flow past a circular cylinder (Fig. 1.9). By this we mean that the flow pattern is
v
1.1l' Variatioir of Flow Patter. witbRe,..,... N..... To, dcscri~ theflQ)! .patterns. q~t8ti'(• • . Use illllSuatioos of,the stt~amliM$. of th:4ow. A' streamline is a line with the property that at each point of the line the dfrcction ofthe line is the same as the dfrcction of thc\(Clqcity vector of ~he flow at tha,t, pOint~· Suitable experimental_ methOds permit visualization and photography of the streamline patterns'
offtuid Oow.· . For. a rigid body, translating with, a constant velocity through an
~
....1. 1.9 Flow past a circular cyhr.der.
• kalClel:encea eitecl at tbC ea4 of 'tile boo~
identical in all planes normal to the axis of the cylinder. Such a How may be reali~ed approximately over the central regron of a long cylinder~ Plate" 3 and 4 show the flow patterns for various Reynolds numbers. (In these examples. R = Ud/l'.whc:re d is the diameter of the cylinder.)
R =3·9
R = 18·6
R = 33·5
Plate 3
Flow past a circular cylinder at small Reynolds number~. after Homann U936).
Courtesy of Oxford University Press. Plate 31 of Goldstein (1938).
18
19
IdealFluid AerodynamicsFor very small Reynolds numbers (for R les~ than about 4) the streamlines near the surface are closely parallel to' it. _The flow pattern is almost symmetrical about the diameters of the cylinder parallel and normal to the undisturbed I,tream.. As R increa~es, the, streamlines althe rear of the cylinder widen out mote and more to form It ~losed region behind it. This regIon is known as.a seP4rationregion 01; seJ1(l,ration bubble. It is followed by a layer known as a Wake (see also ~te·4). The separation bubble is separated from the main flow by a pa.ir of sepa~ation linu originating from ~omepoint on each side of the rear of the cylinder. The points are referred to as separaiioll points; .The flo\. in the·separat;ion bu~e eonsists of tw.o regions in which the 6,uidis in circulatory Plotion, and each of·these regions is known as a vortex .. Thus the separation bubble consists Bf a vortex pair. As R increaSc:s, the separq,tlOJl 1:>ubble becomes more .and more elongated jn the direction of the main "tream. This. happens until a certain Value of R (about 40) is rea.ched and instab.ility sets in. In the'range below this R, the separation point moves forward from ~ rcai as R increases . . As'R rises above this critical value, instabilities setiJ1 and the vortices become asymmetri~ shortly after the beginning of ~he mOtion, leave the cylinder, and move downstream (see Plates 4 and 5). Vortices form, grow, and leave the cylinder periodically. As they move downstream, they form a regular pattern consisting of a double row of alternating vortices known as Karman's vortex street (see Plate 5). A distinct vortex street, as shown in Plate 5, occurs for Reynolds numbers between 40 and about 700. For larger Reynolds n11IQ.bers the proccssegbcconie morc. and more irregular and complicated as R increases. The variation with R shown by' tbo'. flow pattern ~ro1ind a circular cylinder is characteristic a1~o of twoqimeilsionad steady 'flow past cylinders of some other shapes. Bodies that show such flow ~~acteristics are known as bluff1Jodies. We· may describe a bl11ff body as one for which separation of the flow fJ'q.m. the surface takes place well ahead 'Of the rear p,art leading to a large.wake. It should be noted that whether or not a b.ody i$ bluff depends not:only on the shape of the body but also on its orientation. This is illustrated by Plate 6, which shows the Dow past an elliptic cylinder. When the cylinder is oriented·with its major axiS normal to the stream, it behaves like a: bhiff bQdy. When the cylinder is' oriented with its major a.xis parallel·.to the stream, it behaves .unlike· ~ btuff bod)!. Thevariatioli with Reynolds number of the flow patterns aC9und threedimensional bluff bodies, such as a sphere, is in some ways sin ilar to that around twodimensional bodies. There are, however, certa.in lifferences. but there is les,s information on threedimerisional bluff bodie,
l'Iow Past Streamlined Bodies. A streamlined body may be described asone for which ~I'aration of flow, ifit occurs, does so near the rear of the
Introduction
J!
(b) as viewed with respect to the undisturbed fluid.
Karman vortex street R = 250: (0) as viewed with respect to the cylinder; Courtesy of Professor O. G. Tietjens; Plate 24 of Prandtl and Tieljens (1934). Also, O. G. Tietjens: Stromungslehre Vol. I, SpringerVerlag, I Q60. Plate 5
31
Ideal Fluid Aerodynamics
Introduction
JJ
body and' the consequent wake is vcry narrow. An example. shown in . Plate 6, is the elliptic cylinder With its major axis in the direction. of the stream. Ex~iments reveal that to achieve a streamlined \lody the body must be. well rounded and slender and elongated in the direction of motion and that the surface of the body should come to a point or an edge at the rear with a gentle curvature. A slender body of revolution is another example of a streamlined body. So also is a cylinder of the "airfoil" shape, which is employed fo.rlifting wings. At very low Reynolds numbers the flow patte~ 'due to steady twodimensional flow past a streamlined cylinder, such as the slender elliptical cylinder shown in Plate 6 is no different from that for flow pa'>t a bluff body.
i
l.eld ll8
ed&e\''t:,..,.:::::::===::\======T=r=Ii:lin=';ed::;:':.1'l
'Chord'
~
~
~
~~
Fig. 1.10 Airfoil nomenclature.
Plate 6 Fl.) .... p,,;t .In eHipl:c cylinJcr: (J) wito th.: major axis in the direclion of the undhturbed slr'.,,:n; (b) With ,he:: r.1;ljlH ax,s nJrmal to the stream. (a) Courtesy of Oxford UniverSity Pre,s. Plate II of Gultlskin (1938). (b) Courtesy of Professor O. G, Tietjc'1s. Figure 611. Plate 26 of Prandti anJ Tietjens (1934).
Significant differences. however, are observed at high R~ynolds numbers for which the flow pattern depends on the orientation of the body with respect to the main stream. We consi'tter the case of flow past an airfoil (see Plate 7). The orientation of the airfoil with respect to main stream is known as the angle of attack, denoted by at. It is measured between the direction of the undisturbed main stream lied a ~eference line, known as the chord, draw~ from the leading edge to the trailing edge (see Fig. 1.10). At small values of at the streamlines near the body follow the airfoil surface closely, right to the trailing edge (see a of Plate 7). There is a very narrow wake. As the angle IX increases. changes in the flow pattern occur, primarily on the upper surface of the airfoil. Separation begins at the rear on the upper side and moves forward as a. increases. Correspondingly, a wake is generated which grows as 0( increases. At a I.:ertain value of:x the flow separates near the leading edge, giving rise to a large wake, ju;;t as in the flow past a bluff body (see b of Plate 7). In practical applications we are interested in simations at small :x. The flow pattern for finite wings large in a direction normal to that of the main stream is in many ways similar, at least over a major portion of the wing, to that for an airfoil. Complications develop for shorr wings.
IdealFluid Aerodynamics
Introduction
Jj
For steady flow past a slender body of revolution with its axis in the 'direction of the main stream 'We find that separation occurs cioseto the rear and that the wake is very thin. Our concern would be wjth steady flow at high Reynolds numbers past airfoils and large finite wings at small angles of attack and past bodies of revolution with their 'axes parallel to the stream. Furthermore, we are interested only in speeds for which the compressibility effects are negligible. To have some appreciation of the effects of compressibility we give in the next section a brief description of the effect of Mach number on the flow pattern, . at high Reynolds numbers around an airf~l and a body of revolution'. 1.11 Variatioa of Flow Pattern with Mach Number We may consider st~ady flow at high Reynolds number past an airfoil at a gentle angle of attack. Plate 8 shows the variation of the flow pattern with increasing Mach number M = U/a"", where U is the speed of the undisturbed stream far ahead of the airfoil and 0"" is the speed of sound in the undisturbed stream. The photographs offlow were obtained by what is called the shadowgraph method, which responds to variations of density (strictly to the second spatial derivative) in the flow field.· These photographs show that the flow pattern changes conaiderably as the Mach number is increased. The flow fields forM below a certain value are completely different from those for M above th~t value~ Above this value, which may be referred to as the critical Mach number, the flow patterns are characterized by the appearance of narrow regions through which considerable spatial variations of density occur. Such regions are known as shocks. t As we proceed in the direction of the main stream, the density increases, almost abruptly, through the shocks. As M increases from the critical value but still remains less than unity, shocks first appear on the forward part of the airfoil and move back wit~ increasing strength towards the trailing edge. As M exceeds unity but is still dose to it, a detached shock wave appears in front of the leading edge and the othenhocks appearing on the surface of the airfoil occur at or near the trailing edge. AsM continues to increase, the leading edge shock moves closer to that edge. If the nose of the airfoil is sharp, as is common for "supersonic airfoils," the leading edge shock is attached to that edge when M exceeds a certain value greater than unity. On the basis of the flow patterns exhibited at various Mach numbers, it is usual to divide the whole range of possible mach numbers into several parts, each part'being associated with a different type offlow. The range of
Plate 7 Flow past an airfoil: (a) at a gentle angle of attack: (b) at a large angle of ,attack. Note the separation on the top $Urface. Courtesy of Oxford Universily I'n:ss. Plate 12 of Goldstein (1938).
t
• See cited references for a description of this and other methods. For a detailed description of the origin of shocks consult cited references ....
36
IdealFluid Aerodynamics
Direction of Flow
introduction
37
Direction of Flow
M,
= 0·628
M, = 0·683
Ml = 0732
M\ = 0784
M, = 0835
M, = 0876 Plate 8 Shadowgraphs of subsonic flow past an airfoil at a gentle angle of attack _(critical Mach number is 0.695). Courtesy of Oxford UniversIty Press. Plate 16 of Modern Developments in Fluid Dynamics: High Speed Flow, L. Howarth, editor, 1953.
.18
Idea.lfluid Aerodynamics
. Introduction
19
....'." "t!,.\~,
~
. ''';''''
,', :
;~
.... ~.
•"'......
~.,
.
..L'
.J1'~'
M
= 0·86
'~ ~':'~,*!.~~~\
,
:~
/.
,.
M  176
•..'4
,.
_ ..
M 1·01
\\
; ,.
M = 2·51
/
;"
/
/
M 1·26
Plate 9 Shadowgraphs at various Mach numbe~ of a projectile in Bight. Courtesy efOxford University Press. Plates 21 and 22 of Mod~rn DeL'~/opmenls in Fluid Dy"amics: High Speed F101tl, L. Howarth, editor. 1953.
40
If;1ealFluid Aerodynamics
Introduction
41
M extending from zero up to the critical Mach number is known as the subsonic range. In this range the flow is characterized by the absence of any shoc~. It exhibits all the features of incompressible flow, although gradual density variations take place as M increases. In this range the local· Mach number (the ratio of the local speed to' local speed Of sound) is subsonic at every point of the flow field. At subsonic Mach numbers compressible flow pa~t a slender airfoil may be related to an equivalent incompressible flow problem. • The range of M extending from the critical value to some value above unity but close to unity is usually referred to as the transonic range. In this range the flow field is characterized by the appearance of shocks on the surface bf the body, afld a detached leading edge shock when M is close to unity. In such a range. the rather complicated flow field consists of partly (locally) subsonic and partly Oocally) supersonic regions. The range in which M i5 grea,ter than the value for which there are only leading edge and trailing edge shocks with no others on the airfoil surface (except perhaps near the trailing edge) is known as the supersonic range. In this range the flow almost everywhere is,supersonic. A small locally subsonic flow region may, however, exist near the leading edge, particularly if the airfoil has a round nose. The range of Mach numbers exceeding a certain supersonic Mach number (which depends on the situation at Qand) is referred to as the hypersonic range. . Steady flow past a body of revolution exhibit. similar' features at different Mach numbers. Plate 9 shows shadowgra 'h pictures of a projectile in free flight. 1.12 ElI'ects of Viscosl(y at BoundarY Layer
because of the effect of viscosity, no matter how small, the layer of fluid immediately adjacent to the surface is 1t rest. Away from the wall the fluid is in motion with a certain velocity. This means th9:t as the solid surface is approached fluid layers are retarded. The retardation arises out of the actfon of the viscous forces. Depending on the viscosity of the fluid, the retarding effect may ~xtend only to short distances or to large distances from the body. It is ob$erved that for fluids with small viscosity, sucb as water and air (strictly speaking, in flows at large Reynolds numbets), the retardation effects are confined only to a very thin region close to the body. In such a region the 'Velocity rises rapidly from zero at the wall to its value in the main stream. In that region the spatial rate of change of velocity (velocity gradient) in a direCtion normal to ~h.e body is ~arge, a~d consequently the viscous forces would not be neglIgible even If the VIScosity were small. Outside that region the velocity gradients are s!Dall and the viscous forces there would be negligible. The thin region close to the body in which the viscosity effects are confined is called the retardation layer or, more popularly, the bOundary layer. Outside the b~un~a~ la~er the fluid may be regarded to a high degree of accuracy as an mVlscld flUid. Both theory and experiment have supported the correctness of Prandtl's boundarylayer concept.
Hi.
Reyaolds ~ber: The
We now return to the consideration of ~e problem of steady flow at high Reynolds number past a fixed rigid streamlined body. Although we. are primarily concerned with flows with negligible compressibility effects, the following considerations are equally applicable in their essentials when such effects are present. When compressibility effects exist, account must be taken of energy exchanges ano temperatu,re differences. In transonic and sl>personic flows complications develop from the appearance of shocks.
Boundary lAyer Co"cept. In 1904 Prandtl introduced a farreaching idea, now known as the boundary layer concept, and showed the way to treat satisfactorily the flow past a streamlined body at high Reynolds numbers (Prandtl, 1904). He showed that at high Rfynolds numbers the eff'ecte of viscosity are confined to a very thin layer close to the body and a thin wake extending from the body. When a fluid flows past a fixed body,
SoIM CluullCtnUtk, D/II LtImiIfIIr BoIIIIIIiIry lAyne The flow inside a boundary layer may be laminar or turbulent or partly laminar and partly turbulent. We now consider the characteristics of a laminar boundary layer. For the sake of clarity we consider steady twodimen~ional flow past a thin fiat plate (Fig. 1.11). We introduce the coordinates z and y as sho.wn. According to the concept of the boundary layer, we expect the followmg: just ahead of the leading edge tbe fluid approaches the velocity U, the velocity of the undisturbed stream, at all distances normal to the plate; at any section z lying on the plate the velocity will be zero at the plate, y = 0, and rise to the value V a short distance, say y = 6, from the plate. This situation is illustrated in Fig. 1.11, in ~hic? the distribution of u, the velocity component parallel to the plate, With distance y from the plate IS shown. The scale for y is greatly exagge~ated. . We call 6 the boundary layer thickness. It vanes along the plate, bemg zero at the leading edge; thus (1.26) 6 = 6(x)
We now wish to esfimate.6(x). We do this on the basis of the consideration that within the boundary layer the viscous forces are of the same order of magnitude as the inertial forces. In estimating these forces for a fluid element in the boundary layer, we have to bear in mind the fact that within
IdealAuid Aerodynamics the layer the changes in the ydirection occur in a much smaller distance than in the xdirection, so that the flow situation in the boundary layer at any section x does not depend sensibly on what happens behind that section but mostly on what happens ahead. This means that within. the ooundary layer the characteristic length fot changes in the xdirection is the distance x itself of the section unde'r c.onsideratioll.!rom the leading edgt of the plate. The charqcter;st;c length for change in the y·direction ;s the boundarylayer thickness 6(x) at the section .under consideration.
y
Introduction
43
We note that the boundary layer grows as the square root of the distance from the leading edge. The boundarylayer thickness is proportional to the square root of the kinematic viscosity J'. Thus the thickness IS decreases with ". with 6  0 as "  O. Since b(x)lx is proportional to .j lIRe, then
6(x)
x
«1
if R.,» 1
8(%)
Fig. 1.11 Schematic representation of the boundary layer on a flat plate.
Now consider an element situated at the section x. Using considerations similar to those given in Section I.S and noting that x is the characteristic length for changes in the xdirection, we find that the xcomponent of the inertial force acting qn the element per unit volume is proportional to pUSlx. Similarly, we find that the xcomponent of the viscous force on the element per unit volume is proportional to 'T0I6 or equivalently to /JUllJ 1 , where 'To is the shear stress on the plate at the section x; 'To == 'To(X}. Since the viscous and inertial forces .are of the same ma~itude, we have . .
Using this relation and introducing a ,Reynolds number R., defined by . R", = Uxlll, we obtain (1.27) (1.28) (1.29)
Thus, if R., becomes very large, lJlx becomes very small, and the flow over the plate with its very thin boundary layer becomes nearly that of a ftuid without viscosity. However,.no matter how smalllJlx is, there is always the boundary layer with a decisive effect on the flow at the plate. Consider now the behavior of the viscous stress 'To at the plate. We observe that 'Tot put varies as .jI/R.,; thus the stress becomes smaller as R., becomes larger. Even at large values of RIO there is a finite, although small, viscous stress on the plate. The stress 'To is proportional to .j; and \s smaller, the smaller the kinematic viscosity of the ftuid. However, no matter how small." is; there is a finite viscous stress and consequently a viscous drag on the plate. We now consider the variation of the velocity and pressure within the· boundary layer. Let u and v denote the x and ycomponents of the velocity. In the boundary layer, at any x, u is a function of 1/, with u = 0 aty == Oandu = Uaty = 6. Theoretically,uapproaches Uasymptotically and therefore the definition of 6 cannot be made precise. We may, however, define {} as the distance within which u reaches a certain percentage of U, say for instance 99 per oent of U. Now, since {} varies with x, u must also vary with x; thus u == u(x, y). Since 'Tcris equal top(ou/Oylat the plate and 'To decreases with x, the slope iJuf~.at the plate decreases with x. Using this result and the fact that 6increases with x, we may represent the variation of u Schematically (see Fig. 1.12). We note that at a given distance from the plate u decreases with the distance from the leading edge. SinCe u is a function II. x and y, there will be a v that will also bca function of x, y. This is SC1 . in the following manner. Consider the rectangular box shown in Fig. I.l.~; the .thickness of the box is unity in the direction normal to the xy~plane. Two sides of the box are normal to the xdirection and situated at an irrftnitesimal distance dx. The other two sides are normal .to the 1/direction, one of them being the surface of the plate and the other located at 11 == y. Auid flows into and out of the box through its sides. However, there is no accumulation of mass in the box. We therefore require the rate of flow of mass into the box to balance the rate of flow of mass out of the box. Fluid flows into the box through the xface (i.e., normal to xdirection) situated at x and ftows out through the xface
~....

"
IdealFluid 'Aerodynamics
u u u u u~Ed;of
boundary
Introduction
15
rate of outflow of mass through the yface at y is equal to
Dt>
(x, I'Y) dx
Equating this rate of outflow to the n~t inflow through the xfaces, we have
v(x, y) dx =
layer
f
[u(x, y)  u(x
+ dx, y)J dy
(au) dx ax z.~
We may set, neglecting higher order term!>,
u(x
+
dx, y) = u(x, y)
~
+
au
Fla. 1.12 Schematic representation of .the IIvclocity distribution in' the boundary
layer over a flat plate.
We then have
v(x, y) = 
situated at x
+ tIz.
The rate of inflow through the face at x is equal to
p
LVu(x, y} dy
+ dx is equal to + dx, y) dy
Note that au/ox is negative, At the plate v is zero. At the edge of the boundary layer y = 6(x) the vcomponent does not vanish but has a vaiue 'given by
v(x, b(x»
lo ax dy
whereas the rate of outflow through face at x
pLVu(x
=
l
'CZ)
o

au
ax
dy
(1.30)
On the basis of this relation we may state that
v(x, b) __ U 6(x) x
SinCe u(x + tIz, y) is less than u{x, y), the outflow does not balance the inflow. This means that becl\use there is no accumulation of mass in the box, there must be Ii flow of fluid through the yfaces of the box to balance the net inflow through its xfaces. There must therefore be a v velocity. However, there is no flow through the yface at y == 0, which is the plate itself. If v(z, y) is the ycomponent of the velocity at the point x, y, the
or
(1.31)
We thus conclude that the vvelocity is small, ,the ratio vi U being of the same order of magnitude as blx. We next consider the variation of pressure across the boundary layer. The pressure, force in the ydirection, aplay per unit. volume, on a fluid element in the boundary layer is of the same order of magnitude as that of the inertial force, pv2/b per unit volume, on the element in that direction. • We therefore have
(1.32)
The'magnitude of the pressure change ll.p across the boundary layer is therefore given by [ b(x)] I U2 ll.p  pv' __ put ' __ 12(1.33)
x
FIa.l.13 Illustrating the need for the vvelocity in the boundary la)~r.
R.,
We conclude that in the c:>undary layer l'patial variation of pressure in the
• The viscous force in that direction is of the same order of magnitude.
46
IdealFluid Aerodynamics
Introduction
47
ydirection and the actual change in pressure across the layer are negligibly 5mall. As we shall see later, this conclusion has farreaching implications. The solution for the problem of steady incompressible flow in a laminar boundary layer alonga flat plate, as formulated by Prand~l, was obtained by Blasius in 1908. His solution shows that" (defined as the value ofy when u is within 0.6 per cent of U) is gi\l'en by
The subscript e. signifies that the Q\l8iltity refers to the edge of the boundary l@yer.
r",.bIIk", lloIuuItuy Lqu. ObserYatiOQs shoW that at high Reynolds pumbers the flow in a boundary layer does not remain laminar all along the surface of a solid body but usu&lly becomes turbulent at some distance
b(z) ,5.2 ; = ..jR"
The local skin friction coefficient is given by
(1.34)
6.0
C = 'To(x) _ ~
1
ipU
I

..jR"
(1.35)
5.0
R¥
+ 1.(8)( lOS
ct 1.82 )( 105 03.64)( 105 • 5.46)( 10& .728)( 105
The distribution of the velocity across the boundary layer is similar at different xstations along the plate. This being the case, va:iation ofu with x and y may be represented by a single curve. Such a curve, as obtained by Blasius, is !>ilo\\ '1 in Fig. 1.14, where ulV is plotted against the parameter 7] = y..j V{vx. Als( shown in the figure are experimental results obtained by N!kuradse (194: t. It is seen that the theory and the experiment are in excellent agreeme' t. This agreement is also found for the local skin friction coefficie' .t, shown in Fig. 1.15. The experimental results are obtained by Liepmann and Dhawan (1951, 1953) by direct measurement of the skin frictio!", coefficients. So far we have co~sld,ered tbe features ofa laminar boundary layer along a plane surface. Similar results also hold for the boundary layer along a curved surface if certain conditions are met. It is necess"ry that the radius of curvature of the surface is everywhere large compared to the boundarylayer thickness and that there is nOne of the variation in curvature that would occur near sharp edges. Under such conditions the flatplate results may be applied to the prol'!em of boundary layer along a curved surface if. we now regard the coordinates x, y as defining a suitable system of curvilinear coordinates given by curves parallel to the given surface and straight lines normal to the surface. The surface itself is giyen by y = 0 (see Fig. 1.16). For flow past a curved surface the velocity component parallel to the wall at the edge bfthe boundary layer is not a constant, as it is' in the case of the flat plate, but a function of ~he distance x. Similarly the pressure at the edge of the boundary layer is also a function of 2:. We therefore introduce the following notation:
~: 4.0
""
•
.
3.0
2.0
1.0
1.0
I'll. 1.14 Velocity distribution in the laminar boundary layer on a
Experimental re,suhs from Nikuradse (1942).
ftat plate.
V, = V.(x) = u(x, <Xx» P.
from the leading ed~. The transition froql laminar to turbulent motion originates from the instability (in some form or other) of the laminar motion and depends on various factors.· The tUlOsition takes place over a region and not at a po~nt. The value of the Re)'Jlold~.number R .. at which transition to turbulent flow be,gins is knownas tbeGfitieal Reynolds number denoted by R •. crlt • Under normal circumstances ~.cl'U for the bounda{y
• For some details on this subject (which is still not fully unct.rstood) see, for example, ,K.uethe and ~hetzer (1959).
== P.(x) =
p(x, <Xx»
(1.36)
48
0.010
Ideal·Fluid Aerodynamics
, . Theory PrardtJ!"
0.005
0.003 0.002

j'
die oulas the solid :s.urface isappto.ac~, ther~ is ,always alayer of fluid next to the surface u,at is laID.in4dike. 10 ,ucb a la;y~r~ wbjch 41 known as the ·/am;1ftU sub/oyer, th" velocity riR;$ rapidly from zero· at the wall ·toa certain value. In the .sublayertbc velocity gradient normal to the wallis $everer than it would.be if the boundafytaycr were wholly laminar. This being the. case. the local shoat stress on the· solid surface js much greater than it would be if the boundarylaycr were laminar. This is iUusuateclin Fig. 1.15, in w.bK:h ~asured frictien ~fficiCDtl,l for a tloll'bulent boundary layer along a plate are also shown. The velocity distribution across a
0.001
1.0
0.0005
;y
0.8
Turbulent
C.0003
'i 0.6 _
0.4
0.0002
0.3 10
2
Us R,,,
o
0.8 1.0
.Fig. 1.IS Local skin friction on a flat plate. Direct measurements by Liepmann and Dh.lwan (1951, 1953).
t
Fla. 1.17 Comparison Q(laminar and turbUicnt boundary layer profiles.
turbulent boundary layer is shown schematically in Fig.. 1.17, in which the Blasius ciistribution for laminar layer)s also included for comparison. The mechan~~m oftae ftow ir. a turbulent boundary layer is complicated and ditreJ'(:nf from that of the flow in a laminar boundary layer. ConbCquentJy, th~ order of magnitude estimates given for the laminar layer do not apply directly to the turbulent layer; There. is ~till no detailed theory for the ftow in a turbulent bOundary layer. There are, however, .semiempirical solutions for the turbulent boundarylayer thickness and local friction coefficients that are useful. It is found that for Reynolds number!> in the range 105 to loe the boundary~layer thickness and the locial' shear stress To show the foIl owing dependence on the Reynolds number R,.:
~x) TO(X) ;,...., oU2
,....,
iayer along a flat plate is in the range of about 105 to 10 • This me~ns that for the problems that will concern us later the boundary layerls lIkely to be turbulent ovcr most parts of the body. '. . . Since turbultnt motion results in pronounccd mlx.mg of the fl~ld, the retardine: effect of a solid surface on the fluid spreads farther in the rl~rectlOn normal the surface in turbulent motion that in laminar ~otlOn .. In generaL however. the retardation elleet still takes ~la~e ~f()mment1y III a layer that is very thin in c()fl1parison to) thl! c~ara~tenstl~ dIstance along the body, and the cencept qf it boundary layer IS stilI applicable .. Even when the boundary layer is turbulent, since the turbulent fluctuatIOns should
8
to
;"
"\
~.~
... 
 _',l;
l.l~.
...
1
J
1 RO ...
(1.37'
.;.
~.l
/

Recall that in the laminar case 6/x and TO(X)/ pUs vary as R;o·6 .
'"
::..
&plU'tllio.. An important <;baracteristic of the boundary layer is that under certain 1:jrcumstances it lea~ toa reverse ftowover some region close to the wall (tha( is the solid surface). By reverse flow we m~n fi<>w in
" .] F Ig. I •16 Boullda"'' layer coordinales for flow over a curved surface.
50
IdealFluid Aerodynamics
Introduction
51
a direction opposite to that of the main stream outside the layer. When reverse flow appears, it usually leads to separation of the boundary layer and consequently of the main stream from the body. It preduces a .flo~ pattern that is completely different from the pattern th.at would eXIst If there were no separation. Illustrations of separated flows have already been given. We shall not enter here into a description of the time development of the reverse flow and the consequent separation; for such a description reference may be made to Prandtl (l935) or Prandtl and Tietjens (1934) or Sclilichting (1955). In Fig. 1.18 we show schematically the steady flow pattern in the vicinity of separation after separation is completed. Also included in the figure
The separation point, B in Fig. 1.18, is not the point. of minimum pressure for which op.lax == O. The minimum p.ssure OCCurs at somepoint A ahead of B. The greater the adverse pressure gradient the shorter the distance AB. . In the presence of an adverse pressure gradient a l~minar l:oundary layer maY'either separate or first become turbulent and then separate. A
E.+c __
 .
....
Fig. 1.18 Flow near separation.
are distributions of u, the velocity component parallel to tbe wall, with y the distance normal to the. wall, the yscale 'being exaggerated. The figure shows that the streamline branching from B divides the flow coming from the left and right. Such a streamline is known as 'file divid~g streamline or as mentioned before as the separation streamline. The pomt B is the separation point. We see that at the separation point ou/ay = o. At the wall ahead of the separation, ou/ay is positive; downstream of the separation it is ne'gative. .. . . In th~ separated region, elose to the wall, the flow I~ 10 the ~everse direction, and consequently the pressure there must mcrease 10 t~e direction of the outside main stream. Thus, to produce the state of affaIrs shown in fig. U8 i1;1itially the flow in the boundary layer close to the wall i~ subjettCd to an increasing pressure in the forward direction. ~ow, according to the boundarylayer concept; when the boundary layer IS not separated from the wall, the pressure at the wall is approximately equal to p.(i), the pressure at the edge of the layer. We therefore co.nelude that t~e onset of reverse flow at the wall and the subsequent separatron of the maIO flow should have resulted from an increasing pressure in the forward direction, that is, from a situation in which the gradient op./Ox is positive. Such a gradient is referred to, tor obvious reasons, as an adverse pressure gradient. We may thus !itate that separation of the boundary layer results from an adverse pressure gradient.
u.
u.
FIg. 1.19 IIIustratiDg the wake behiDd a flat plate.
u.
~urbulent boundary layer, because of the intense mixing due to turouience, IS better able to resist se~ration than a laminar layer. Boundarylayer considerations do not apply at and downstream of separation. In particular, the assumption that the pressure gradient op/ay no~mal to the ~all is small does not hold any longer. Additional complications also anse, for the flow in the extended separated region is usually unsteady and turbulent.
~""e.f. Consider steady flow, at high Reynolds number, past a statIonary flat plate of finite extent at zero angle of attack (Frg. 1.19). ~ehind the trailing edge of the plate, the boundary layers along the two Sides of the plate leave the plate and coalesce into one region behind the plate. Such a region is known as a wake, a region where Viscous effects are still dominant. With increaSing distance downstream the width of the wake increases while its mean velocity decreases. A similar wake appears in the case of steady flow at higb ReynOlds numbers past a stationary streamlined body at a gentle angle of attack
51
IdealFluid Aerodynamics
Introduction
53
(See picture a of Plate ;). It is fbund that the ftow in the wake is usually turbulent even though it might have. originated from laminar boundary layers along the body. For streamlined bodies at gentle angl.e of attack, in high Reynolds number ftows, the wakes are still thin compared to the characteristic distance along the mainstream direction. Thin wakes appear only in two instances; either the boundary layer is not separated from the body or, if separated, remaining very close to the rear end of the body. When the boundary layer separates, as it does for bluff bodies and streamlined bodies at "adverse" angles of attack, an extended or a thick wake involving vortices and turbulence appears in the ftow (See Plates 4, S, 6, and 7). Thinwake ftows, as contrasted with thickwake ftows, are not only practically important but also are amenable to approximate calculations. We see in the next section how this may hi" done.
Fla. 1.21 Inner flow corresponding to Fig. 1.20.
1.13 Coasequences of the BoundaryLayer Coacept
Consider steady high Reynolds number flow past a stationary.streamlined body which is at, gentle angle of attack (Fig.. 1.20). On the basis of
Edae ofWlke
In view ~fthese features, the flow field can be divided for the f ma.thematlcal analysis into two retions·'I) An I'n '. pur~~ h bo C> ., ner regton conslstmg of t e un~ary laye.r ~ow along the body and the wake behind it, and (2) an outer regt~n conslstmg of the main flow. The inner flow is said t . ( . 0 occur under the Impre sed (F I 21 s pressure P. x)'. WIth a velocity U.(x) at the edge <Xx) I,gs:. and 1.22). I~ consts:uctmg the solution for the inner flow the flUI~ IS to be treated as VISCOUS, taking into account the simplificc.tions'that res~ t from the features ~f the boandaryIayer flow. The solution shOUld satIsfy .th~ Socalled noslip condi/;on u = 0, v = 0 Jlt the wall The lOner boundary of the outer flow is the edge of the bo~nda I and the wake. In constructing the solution for the outer flow the fI rtd ayer f • UI may be treated to a h' h d sa;isfy the _ The( )SOlutiobn lOner boundary. ~ • p  P. x at t e
~hould
l~ounefa~ ~on~~~~::c~,:s;~~V~~~u~.
:
:

,... 1.20 Flow past a streamlined body at high Reynolds number.

 ..
",   .... \_  .~
"."."'"
......

the boundary layer concept and the characteristics of the· flow within such
a layer, ","e may summarise the main features oft~ flow' as follows:
The viscous effects are confined to the boundary layer and the wake behind the body. In the ftow outside the boundary lay~r the viscous effects are negligible. In the boundary layer the velocity component parallel to the wall rises rapidly from zero at the wall to the mainstream value U,(z) at,the edge of the layer. Thcvelocity component normal to the wall is very small throughout the layer. At the edge of the layer it ~ finite but still very small. The change in pressure acrOss the layer is negligibly small. The pressure at the wall is, therefore approximately the pressure at. the edge of the boundary layer:
p(z,O)  p{z, 6{x»  P.(x)
~
We say t/1at the outer ftow "impresses" its pressure on the ooundary layer.

JJ
FIc. 1.22 Ou~ flow corresponding 10 Fig. 1.20.
Introduction IdealFluid Aerodynamics We note that a priori neither the edge of the boundary layer, nor U.(x), nor P.(x) are known. However, on the't;asis of the characteristics of the boundarylayer flow; the two problems outlined above IQay be solved approximately as follows: First :we solve the problem of the ou~er flow. To do. so we usume that the inner boundary of the outer flow is coincident with the sunace of the body and a surface behind the body, such a surface being obtained by letting the thic~ of the :ake go to zero while retaining the effect of the wake on the main flow. The problem then is that of the steady flow of an inviscid Suid past the given body with the effects of tlte wake properly taken into aCcoUDt. The boundary .condition then is that the flow be tangential t.> tbe wall, the socalled slip condillon. Some appropriate condition should also ':Ie satisfied on the wake surface. Having determined the outer flow, we attack he problem of the inner flow. We USunle that the conditions at the edge of the inner flow, namely U.(x) and ,.(x), are appro~imately equal to the corresponding values at the wall and the wake surface given by the inviscid solution .for the outer problem. The procedure may be repeated if so desired. Under normal circumstances. one evaluation of the outer and inner solutions is satisfactory. The inviscid solution gives the pressure distribution over the body and the force and moment resulting from the pressure distribution. The solution for tbe inner problem gives the boundarylayer thickness, the wake, and most importantly the frictional force, the socalled skin{riction drag, on the body. Naturally, the inviscid solution cannot fprnish any answers for the viscolfs drag. Our concern in tG'1S book is with the analysiS of the outer problem. We have seen that this is the first step in analysing .high Reynolds number flow past a streamlined body, a problem of considerab~e practical interest. Prandtl's boundaryIayer concept is a landmark in the science of fluid mechanics. Its role in fluid mechanic~ rlln be summed up as follows. The concept furnishes a method for relating the motion of viscous fluid past a body with the corresponding motion of a nonviscous fluid past that body. It shows that for flow at high Reynolds numbers past streamlined bodies, viscosity does not influence the pressure field. The flow pattern around such bodies, the pressure fields, and the pressure forces on the bodies can, therefore, be computed to a high degree of satisfaClion on the basis of inviscid fluid flow. It shows that the first step 1" the analysis of the motion of j\ viscous fluid past a body is that of ,solving the corresponding in:viscid motion past the body. In formulating the inviscld problem, proper account must be taken of tile effects of the wake on the ,main flow. In this way the boundaryIa yef concept has brought out the important role played by i.nviscid flow theory and has lent credence to the exten~ive use of such a theory in hydrodynamics a~d aerodynamics. The concept leads to a visc<'us flow theory known as the "boundarylayer theory" valid for flows at
SJ
~~ Reynolds numbeR. Such a theory furniabea a method for computing akin fnction and also ~ transfer. ~hen comp~biJity c8'ecta qQCUl'. The concept explains the ongm of separation and the Sow patteral associatccl with bluff bodies. It shows that under an adverse pressure gradient a bo clary layer would separate; . . UD In this way it emphasizes that solutiODI and Sow patterns fumisbec1 by inriIc:id Sow theory are unreaUatic when they in',olve severe adverse nft'lUI ..........r ..... along solid sudacea present in the fluid. ' r ~. ___ta On the other hand,· on the basis of the inviscid solution, boundaryIa . yer theory wo~d indicate the ~ble location of separation.
i
Before the. introduCtion of th~ boundary:layer concept, a larF part of was a disturbingly mysterious sUbject. Theoretical studies were entirely based on inviscid motion and bad ·th no con?ection problems or no appreciation for con~~tlOn where It eXisted. On the other hand there was a large body of empI~ly.~llected knowledge, under the name of hydraulics, which enabled lOdJV~dual treatment of various practical problems. Hydraulics lacked an unified theory. With the appearance of the boundarylayer concept and the boundarylayer theory, the two subiects th . ti· I hydr ad od . d· . J' eoreca 0 n ~er ynam.Jcs, an hydraulics, have been brought together under an umfied theory.
hydrod~am.Jcs ~d aerod~cs
w~th ~racticaJ
~uc~
1.14 Ideal Fluid Theory
In this book ~e are concerned with inviscid flow where compressibility are pra~tically. negligible. This means that we may regard the fluid as IIIcompreSSlble beSIdes being inviscid. An incompressible inviscidfl id is known as an ideal fluid. II N~t"rally"n? real ~uid is an ideal fluid. However, on the basis orall our prevlOu.s ~nsI(i~ratJons, we may state that under many racticall ' InterestIDg SituatIOns, the motion of a real fluid may be analyse! to a hi accuracy by regarding the fluid as an ideal fluid. The .stud apart from its practical utility, is a suitable to the ~nnclp.les and methods unperlying the. general science of fluid mechamcs which encompasses many diverse and complex pheaomena. W,e now pr~ to the development of the theory of ideal fluid flow wIDgs and bodies of revolution. past
eff~ts
~egree ~f Idealfluld.th~ory,
~f introdu~on
l
Elemeata of Vector
AJ&ebra
and CalcuJus
Chapter 2
. Elements of Vector Algebra and Calculus
In' the analysis of fluid motion, as in the anAlysis of many physical phenomena, we· are. concerned with quantities that may be classified according to the information needed to sPecify them completely. Quanti. ties such as mass, density, and tenipcrature need (after a cboice of units) specifiCation of their magnitudes only, thai ·is, a single number is all that is ncCtssary to specify each of them. Such qUantities arc called scalar quantities or simply scalars. A quantity such as force or velocity requires the specification of a magnitude anil a direction, that is, of a directed magnitude. Quantities 'ofthis type are called r>ector quantities. Vector q~antities that obey certain 'rules (s'lCh as the parallelogram law of addition) are defined as·vectors. As we shall sec later, not all vector quantities are vectorS: Quantities that require specification of mwe information than needed for vectors also occur in physical problems. For example, to desCribe a quantity such as stress we need to give a force (i.e., a directed magnitude) and asurfacc (i.e., the orientation or direction of 14e surface) on whiCh the force acts. SuCh quantities are known as tensors. There are various kinds of tensors, a~d, generally speaking, vectors and scalars are degenerate tensors. Operations of algebra and calculus, such as are known for scalar quantities, arc also developed for vectors and tensors. Algebra, applied to vectors, is known as vector algebra, while calculus applied to vectors is known as vector ~alculus or vector analysiS. Similarly, we have tensor algebra and tensor calculus. In analyzing a physical phenomenon, we set up interrelations between the various quantities that characterize the phenoillenon by using laws' of nature (such as the laws of Newton, the law of energy conservation). To write a natural law, one introduces a coordinate system in a chosen frame of reference and expresses the various physical quantities involved by means of measurements made with respect to that system. When we choose such a procedure, the expression for the law contains terms that are dependent on the chosen coordinate, system and CQnsequently
appears differently in different systems. But the laws of nature are independent of the artifidal cboice of:a coordi1Iate' system. Therefore ,,=,C may' seek. to expreu the 'uturallaws in' a form not related to a pat_ ticular coordinate system. A wayot doing this is ptoVided by vector or tensor analysis. Vector notation exhibits quantities such as displiu:ements, velocities, foices, accelerations, moments, ·and angular velocities their natural color, that is, as quantities that possess directions besides magnitudes. When vector notation is used, a coOrdinate system need not be introduced. Thus use of,' vector' notati.on· in formulating pbysica1'1aws leaves them inimJarimlt /orm (i.e., ia a form independent of ~rdinate d~pti.on~. St~dying ~ ph~c:al phCII()IDeDon' by means of equations written ID Invanant form often leads: ,to a deeper understanding of tbe phenomenon. Besides, the use of vector: notation .brings , consid~ble simplicity into the analysis of problems. It is our desire to develop the equations of fluid motion and· obtain ~any of tbe ~ic results of our ~tudjes in a form not related to any partIcular coordmate system. To this end we shaU.employ vector notation, vector a1gcbrar and vector analysis. We sball, therefore, begin our studies by acquainting ourselves in this chapter with the clements of vector algebra and calculus. Vector methods are adequate to treat the aspects of fluid motion presented in this book. Hence we shall not concern ourselves with tensor algebra or calculus_· The concepts of vector analysis are closely assQCiated witl~ the concepts of fluid mechanics.
m
2.1 RepreseatatieD of • Vector
, If P and Q ~ any two points in space, the directed straight~line se~ent frOlll p t~ Q loca~es the position of point Q with respect to the ~lDt P. Such a dlrect¢l.ine segment is called a pOSition vector. It is the slmp~est example of a vector quantity. Graphically we represent the position vector from_ P to Q by a straight arrow running from P to Q
as shown in Fig. 2.1. The length of the arrow gives the magnitude of the distance Q from P to Q, while· the sen~ of the arrow indicates tpe direction fromPto Q. Following· the example of the position vector we represent any vector quantity (e.g., velocity, force) by an arrow pointing in the same directio~ as the vector. The length of the P arrow is uwIt proportional to the magni F". 1.1 ~epl'CSClltation ,of • tude of the quantity. vector.'
• In discussing the mec:WUCS of • viscous ftuid, tet)Sor &naIytis becomes utefuL
58
IdealFluid Aerodynamics
Elements of Vector Algebra and C1ilculu5
59
A suitable conventJon lS.adOpted to denote a vector quantity. In printed work it is usuaIlysymbolized by a boldface letter; For example. the letter r may be u~ for the position vector. the letter V for the velocity vector and so on. In writing it is custon:uuy to place an arrow or a bar over the letter that denotes a: vector quantity. Thus for a position vector we write I. for a velocity vector V. and so on. If we wish to show that a di~ted line segment from a po~nt P to a point Q represeRts a certain vector quantity. we sometimes use the notation PQ to denote the vector. The magnitude ofa vector A is denoted by IAI or simply by the letter A. Two vectors A and B are equal if the magnitude of A is equal to the magnitude of B and if the direction of A is the same as the directiop of B. Thus a vector is not changed. if it is moved parallel to itself. This means that generally the position of a vector in spJCC may be chosen arbitrarily; In certain applications, however <as in the calculation of the ~oment of a force) the actual point of location of a vector may be important. A vector when associated with a particular point is known as a localized or bound vector; otherwise it is known as afree veet.or. When two or more vectorS are parallc;l to the same line, they are said to be collinear. When two or more vectors are oarallel to the same plane, they are said to be coplanar.
2.2 AdditioD aDel SubtractioD
~. ~
This notion of addition is readily applicable to vector quantities otber than position vectors. If A and B are any two vectors, we can represent them by arrows drawn such that the initial point of B coincides with the terminal point of A (Fig. 2.3a). Then the vector sum A + B is given by the vector C, which extends from the initial point of A to the terminal
Z +..
II
__B=_.• ,.
+
(II)
Let P andQ be two points in space and let OP and OQ ,be the respective
~
position vectors froin a reference point 0 <Fig. 2.2}. PQ denotes the vector from P to Q. From 0 the point Q may be reached along the
~ •
(e)
•
Fig. 2.3 Addition of two
vecto~.
Fig. 2.2 Additioo of position vectors.
vector OQ or, alternatively. along the \I.ector OP to " and ~en al~g vector fiQ to Q. We define that 6Q is the sum of the vectors OP and PQ. A~cordingly we write that
'>
~
OQ == OP+PQ
"
~
"
point of B. In the same manner A may be added to B and the sum B + A obtained as shown in Fig. 2.3b. Putting together Figs. 2.3a and 2.3b (which arc equal triangles). we obtain a parallelogram as shown in Fig. 2.3c. The vectors A and B drawn from a common origin form the sides of the parallelogram. The diagonal C drawn from the common origin represents the sum A + B or B + A. Thus we say that vectors are added according to the "parallelogram law" of addition. Repeated application of the parallelogram law determines the sum of any number of 'lectors. Since A +B = B + A, vectors may be added in any ordet Whatsoever. We, therefore, say vector addition is commutlltiml.
Elements of Vector Algebra and Calculus
61
1dea!·F1uid Aerodynamics
1.5 Unit Vector
~~~.
•

~.......;.~
(6)
A vector of unit length (i.e., of unit magnitude) is called a unit vector. Considering any vector A, form the product
The result is simply a unit· vector in the direction of A. Denoting this unit vector by eA , we write A eA ==A or
A == AeA
.
(c)
.
== A + (B)
That. is, any vector may be represented as the product of its magnitude and a umt vector. A unit vector is used to designate a direction.
JI'Ic. .2.4 Subtraction of two vectors.
Subtraction of vectors is carried out on the same lines as their addition. To form the vector difference A  B we write
1.6 Zero Vector
A vector of zero magnitude is called a zero vector. It· has no definite direction ass<><:iate<i with it.
1.7 Scalar ProcIud of Two Vecton . Besides addition, subtraction, and multiplication by a number two further algebraic operations, known as the scalar product and the ~;ctor product, can be defined for vector . quantities. To introduce the scalar product, we recall the concept of work. When a force F acts on a mass point, and if under its action the mass experiences an infinitesimal displacement 5, we define the work done by the force as equal to the orthogonal projection of the force along the direction of the displacement times the magFig. 2.5 Concept of work. nitude of the displacement (Fig. 2.5). If 8 is the angle between F and 5, we express the work done by
W = IFI cos 8 151 = Fs cos 8
A B
and reduce the operation of subtraction to one of addition {Fig. 2.4). The negative vector  B has the same magnitude as B but points in a direction opposite to that of B.
1.3 Deftnidoll of • Vector
We now define a vector 4S a quantity that possesses both a magnitude and a direction and obeys the parallelogram law of addition. Obeying this law is important, for there arc quantities that have both magnitude and direction but do not add according to the par~lelogram law. A DJlite rotation of a rigid body,althougb possessing magnitude and direction', is not vector, for such rotations.do notobcy the para11clogram l~w. On the other hand, an infinitesimal rotation of a rigid body is a vector. The reader should verify these statemen~
•
a
1.4 Multl.,llcatinD by • Number
If a vector A is multiplied by a number m, we obtain another vector the magnitude of which is m times the magnitude of A, and the direction of whi<;h is t~same as tbat of A.
where F a~d s denote, respectively, the magnitudes of F and s. The work done, W, IS a scalar quantity that is obtained by a certain kind of product
ImAI
=?
m IAI == mA
IdealFluid AerodynamicS
Elements of Vector Algebra and Catc:ulus
6J
between two v~ors,. namely F and.. Such an operation may be given a name and defined for any two vectors. Since the result of the plod\lct is. a scalar, it is called the sr.alar product or two vectors. !t is
o~ration
defined tIS the scalar quantity .:>quIIl to tire i'r,?~ct of the' magni~s of thE rJ4edors times the cosine of the ,ngle between their directiollS. If A and B are any two vectors, thCir' scala.' product is denoted by A· B and read as .A dot B. Thus we
wnr.e
A· B • IAIIBI cos 8
.ABcos8
(2. I} Fla. 1.'
~t
where 8 ;s the angle between the vectors. The scalar produ~ is also known as the dotpr.t or as the inner product. By using the notath)n of the scalar product the "'orle done by a force F during an infinitesimal displacement I may be represented by
of a moment due to a force.
WF·.
simple results follovdmmediately frt.m the definition (2.1) of the scalar product. . . Since A • B .. B .A, the scalar product is ~mmutative. If the vectorS A and B are JKtpendicular to each other, their scalar .. product is zero, since 8 "'" .,,/2 and cos 8  O CoBverselY, if A· B'" 0, it (ollows that either the vectors ~. mutually. perpendicular ~r at least' one ,~ them is zero. If two vectors A 'and B are parallel to each other, their scalar product becomes simply equal to the, product of their magnitudes (i.e., AB), since 8 "'" 0 and cos 8 "'" 1. The scalar produ~t of a vector by il$Clf is equal to the square of its magnitude. Thus we have A.A "'" AA "'" AI The product A· A is sometimes denoted by A'. The orthogonal proj~tion of a vector A in any direction e is give.by the product A • e. 1.8 Vector Product of, Two Vectors
of action of the force (i.e., the lever arm). Denoting these quantities, respectively, by M, F, and I we have MFI
~
A few
If r denotes the vector OP and 8 the angle measured from r to F such that 0 ~ 8 ~." (see Fig. 2.6), the magnitude of the moment becomes MFrsin8 The direction of the moment is that of a rotation about 0 in the plane formed 'by the vectors r and F. Drawing the vectors rand F from the common origin 0, we observe that the d,irection of rotation due to the moment tellds to bring the vector r into the vector F (Fig. 2.7). To express these ideas symboli<:a1ly we first set up at 0 an axis of rotation such.that it is perpendicular to the plane rand F and points is the direction in which a righthanded screw would advance when turned in the direction of rotation due to the moment (i.e., from r to, F): Along this axis of rotation Vie then draw a unit vector e", and ~gree .that it represents the direction of the': moment vectorM (Fig. 2.8). Thus we write (2.2) M == Fr sin Oe", and represent it as shown in Fig. 2.8.
10 introduce this product we consider the concept of the moment due to a force. Suppose we wish to describe the moment about a point 0 of a force F acting .at the point P (Fig. 2.6). To describe the moment com~ pletely we must give a magnitude and a direction, that is, we must specify a vector quantity. Let us, therefore, denote the moment by M. By definition the magnitude of the moment is equal to the product of the magnitude of the force and the shortest distance from the reference poilU to the line
Direction of
moment FI,. )..7 Direction of the moment.
IdealFluid Aerodynamics
ElemeDti
According to (2.2) the moment vector may be looked on as resulting from a certain type of product operation between two other vectors. Thus Eq. (2.2) may be made the basis for defining such a product between any two vectors. Since the result" of such a product is a vector, it may be called the vector product ana defined as foHows. The vector product of the vectors A and B is a vector C whose magnitude is equal to the produqt of the magnitudes of A and B times the sine of the
Allis of rotation
or Vector Alpbta and Calculus
65
WesbaltnQw·state a. few simplerosult$ that;follovnudily from the definition of the vector product. The products A x B a.nd B x A are not equal. In fact, we have
A x B  B x A
This means that tile vector product is 1IOt cOmnlutatir; It ii, therefore necessary to preserve the order of the vectors when dea!...g with operations that involve vector produtts.
.... 2.9 Vector product.
Fla. 2.1 Representation of a moment vector.
angle (J measured from A to B such that 0 ~ (J ~ 'Fr, and whose direction is specified by the condition that C is perpendicular to the plane of the vectors A and B and points in the direction in which a righthanded screw advances when turned so as to bring A toward B .. The vector product is usuaHy denoted by writing the vectors with a cross between them as AxB
If two vectors A a.nd B.are parallel to each other, their cross product is zero. Conversely, if A )( .B  0, the vectors A and B are either parallel or at least one of them is zero. It follows that the vector product of a vector int itself is zero.
l.9 PIne Area u • Vector
The magnitude of the vector A x B is equal to the area. of the parallelo
gram formed by the vectors A a.nd B. In fact,. the vector A x B may
be conSidered to represent both in magnitude and direction the area of thc parallelogram whose sideS arc A and B, if a planc ai'eacan be re~n,.; sented as a ·vector. Now, any plane area may be regarded as possess~ng a direction besides· a magnitudc, thc directional character arising out of the need to specifythc orientation in space of the plane area. It is customary to· denote the direction gf a plane area by means of a uni! v~or drawn in the direction of thc·normal to that plane. To fixthc dIrectIon of the normal wc fitst assign a certain sense ~f travel along the contour that fo~s the boundar; of the plane arca in qucstion. Thcn thc direction of the normaI is takcn (by convcntion) ~ that in which a righthanded screw advances as it is rotated according to thc sensc of travcl along thc boundary curve or contour. Thus if t:j is a curvc enclosing an area S on thc plane P and the direction along t:jis assigned as shown in Fig. 2.10, the direction of the area is given by the unit vector II normal to the plane
and read A cross B. For this reason it is also called the cross product.· . If A. and B are the respective magnitudcs of A and B and if e denotes thc direction of Ax B, we write A
xB == eAB sin (}
(2.3)
and represent it geometrically as shown in Fig. 2.9. . . By using the notation of the vector product (2.2) can now be abbreViated to the form M = rx F
• Skew product and outer product are also used.
66
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
67
and directed as shown in the figure.· The area itself can now be represented by a vector S whose magnitude is S and whose direction :s n. Thus, symbolically, we write
S = Sn
figure, is perpendicular to the radius and to the axis of rotation. Denoting the direction of V by the unit vector e, we have
V = wae
(2.4a)
Let 0 be any point on the axis of rotatiC''l and let r denote the position According to these ideas, the vector product A x B represents both In
S=Sb
vector
OF (Fig. 2.11).
If (J is the angle between r and the axis (measured
Fig. 2.10 Plane area as a vector quantity.
magnitude and direction the area of the parallelogram whose sides are A and B.
2.10 Velocity ofa Point of a Rotating Rigid Body
The vector product of two vectors has many geometrical and physical applications. The description of the moment of a force about a point is one important example. The descriptior. of the velocity of a point of a rotating rigid body is another, and we shall consider it here. S.uppo~e a rigid body is rotating with an angular speed w about a certam aXIs. We wish to describe the veloc~ty of a point P of the body. Let the vector V denote the velocity of the point. Each point of the body describes a circle that lies in a plane perpendicular to the axis and with its center on the axis. The radius of the circle is the perpendicular distance from the point of interest to the axis (Fig. 2.11). The magnitude of th~ velocity of the point is simply equal to the product of the angular vel0clty wand the radius, say a, of the circle. The velocity V, directed as shown in the
• We have used the common notation n instead of e•. Fig.2.11 Velocity of a point of a rotating rigid body.
such that 0 ~ (J ~ 11), the radius a is equal to r sin 6, r being the magnitude of r. Then from (2.4a) we obtain V
= wre sin (J
(2.4b)
Now, angular velocity, like the moment' due to a force, is a vector quantity possessing a direction and a magnitude and, as can be verified, obeys the parallelogram law of addition. !t is thus a vector and we denote it by w. The direction of w, as in the case of the moment, is a sense of rotation about a certain axis. To represent it by means of a unit vector we adopt the convention (or the rule) of the righthanded screw. According to this rule, the direction of the angular velocity is given by a
IdealFluid Aerodynamks
Elements of Vcctor Algebra and Calculus
69
unit vector draWn along the axis·ofrotation and pointing in that direction in which a righthanded screw would. advance when it is turned in the direction of rotation about the axis. Denoting this unit vector bye.... we write
W'"
we....
and represent it as shown in Fig. 2.12. Now. if c. is a unit vector in the dir.ection of r we observe that
c... x cr = e sin ()
With this relation. Eq. (2.4b) may be written as
FJa. %.Il Vector ~tation of ~gular velocity.
v = cure...
x er
= we", x rer
=W
x r
(2.5)
This states that the velocity of a point of a rigid boQy rotating about an axis is given by the vector product of the angular velocity and the position vector drawn from any point on the axis of rotation to the point under consideration (see Fig. 2.13). 1.U Polar aDd Axial Vectors
It might have been noticed during the preceding considerations that there is a certain difference between vectors such as angular velocity and the moment of a force and .vectors such as velocity. force, and displacement. The difference between the two types of vectors lies in the way they are represented by 'directed line segments (i.e., arrows). In the case of vectors such as fcirce and velocity the direction of the arrow is the true direction of the vector it represents. Vectors that can be represented in this way are called polar vectors. In quantities Sl!ch as angular velocity and moment the direction of the arrow is not the actual direction of the quantity it represents, for the actual direction in this case is that of a ro~ation shout an axis; and whit we have done is to .choose to represent this direction 'of rotation by means of a directed segment along the axis of rotation. To specify the direction of that segment we have adopted, arbitrarily of course, the Tighthand rule (i.e., the convention of the righthanded screw). Vectors tha~ art represented this way are called axial
vectors.
1.12 Multiple Products
FJa. %.13 Velocity of a point of a rotating rigid boGy as a vector product.
We return to the products of vectors. Products between three vectors are called triple products,' If A, B, Care ahy three vectors, triple pr,oducts of the form A(B • C); A· B xC; A )t (ll x C) are easily defined.
70
IdealFluid Aerodynamics Elements of Vector Algebra and Calculus
71
The product A(8 • C) is simply a multiplication of the vector A by a scalar that is equal to 8 • C.
Sca/iu Tripk Pro_to The result of the product A· 8 x C is a scalar. Therefore such a product is called a scalar triple product. Drawing the vectors A, 8, C from a common origin, we readily see that the produ~ A • 8 x C is equal to the volume of the parallelepiped formed by the vectors A, 8:C (Fig. 2.14). The following simple results· are important in applications. In a triple scalar product, the dot and cross. can be interchanged without changing the value of the result. Symbolically, we h,4lve
by 8 and C and is perpendicular to the vector A (Fig. 2.1S). In such a case A x (8 x C) can be expressed as a linear combination of the vectors 8 and C (see 2.13). Thus we write
Ax(BxC)=mB+IIC
(2.9a)
A•8 x C  A x B •C
(.2.6)
c
B
Fla. 1.14 Scalar triple product as the volume of a parallelepiped.
A cyclical pern:.utation of the order of the vectors in a scalar triple product leaves the product unchanged. 'This is expressed by
A·8 x C *= B· C x A  C· A x 8
A
Flg.2.15 Vector triple product.
where m and n are numbers. It can be shown that
m =A·C
(2.7)
It further follows that
A· B x C
n = A·8
=;
=
A· C x B = C· B x A
B· A x C
(2.8)
We thus have
A x (8 x C) = (A • C)8  (A • 8)C
(2.9b)
Vector Tripk Product. The result of the product A x (8 x C) is a vector. Therefore such a product is called a vector triple product. The vector A x (8 x C) is normal to the plane formed by A and (B x C). The vect,or 8 x C is, however, perpendicular to the plane formed by 8 and C. This means that the vector A x (8 x C) lies in the plane formed
• Proof of these results is left as an exercise. 
Since the vector product of two vectors changes sign when the order of the vectors is changed, it follows that
Ax~xC)=Ax~xm=~xmxA=~~C)xA
Products involving more than three vectors can be readily evaluated in terms of triple products and others we have already considered.
72
IdealFluid Aerodynamics
Elements of Vector Algobra and Calculus
73
2.13 Components of a Vector
The study, so far, of vector algebra has proceeded on the basis of the geometrical description of a vector as a directed line segment. We now proceed toward an analytical description of a vector and of the: operations on it: Such a description yields a connection between vectors and ordinary numbers and relates the operations (of algebra and calculus) on vectors with those, on numbers. The analytiwl description of a vector is based on the notion of components of a vector. From the idea of addition it follows that any vector may be represented as the sum of a number of arbitrarily chosen noncop/anar, vectors. When this is done we say that a certain vector is r~solved into a number of compon~nt I vectors. I ! Let us consider, for simplicity, a vector I i A in a plane. The minimum number of" I nonparallel vectors into which A may bi; I I resolved are two. Thus let us designate I I as shown in Fig. 2, t6, two nonparallel I directions by the unit vectors e~ and b • ~ oe. Choosing suitably two l"umbers a and b we can draw vectors ae.. and be h such Fig. 2.16 Decomposition of a that their sum is represented by the vector vector in two dimensions. A (see figure). Thus ae.. and be b become the Pecto;' components of A in the directions e.. and eb. To exhibit analytically, the component form of A we write
The numbers a. b, and c, which may be positive or negative, are cal~ed the scalar components· of A or the measure numbers oC the respc:ctlve vector components of A, It is usual to refer to the number\a. b, c Simply as the components of A; it being understood that they are the scalar components of A in the respective, directions e,G' e., and e•. To show the utility of expressmg vectors m component fo~m. let us first set up a basic system of three noncoplanar vectors. Then, With respect
e
Fig.l.17
Decomposition of a vector.
A:;:: ae..
+ beb
If it" is so desired 'we could decom~se A into a number of nonparallel vectors. But; selecting any two of these vector components. 'we can express each of the rest in terms of the selected two. Thus the number of, independent components that are necessary and sufficient to decompose a vector in a plane is two. In a similar manner, any vector in space (i.e., in threedimensional space) can be resolved into vector components that are noncoplanar. Now. the number of inJependent components that' are necessary and sufficient to decompose a vector in space is three. Thus if we designate, as shown in Fig. 2.17, three non coplanar' directions by the u.nit vcctors e... e., ee' any vector A may be represented as made up of the component vcctors ae.. , be., ce" where a. b. c are suitably chosen numbers. Thus the component form of a threedimensional vector A is expressed by
to this basic system, let all vector quan~ities.be ex~ressed in their co~~ nent forms: Once this is done. all operations IOvolvmg the vector quantities will reduce to operations involving their scalar components. For example. consider equality of two vectors At and A z· . . If ea. eb • and e. denote a basic system of umt vectors, and If a lo br.. C I are the respective components of Al and b c, are the respect". " components of A,. we have for' Al = A, (2. 11 a)
a,.
the component relation
alea
+ bl~ + cxe. =
aze..
+ b,eb + c,eo
(2.llb)
This further reduces to three scalar equations
(2. 11 c)
(2.11d)
Cl
= c,
(2.11e)
As another example consider the addition of two vectors Al and A z· • Note that unless e•• e•• and e, are mutually perpendiCular. a '" A • e•• b " A • tt. and c '" A • e,.
A, = aea
+ beb + ceo
(2.10)
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
75
If the sum Al + AI is denoted by the vector Aa and its components by aa, ba, Ca, we have for A . • == A'+ A I I (2.12a) the form
ale,.
+ bae" + cle. == ale,. + ble" + cleo + ate,. + b.2e" + c~e.
a. ==
(2.l2b)
may require that the three numbers 'representing a vector obey such a transformation rule . According to these ideas, we define a vector, analytically, as an ordered set of three numbers that obey certain specific rules. If the set of numbers q10 q., ql represent a vector A, we express this symbolically by writing A
This can be rewritten as three scalar equations
== (q1o q., q.)
+ a. b. == bl + b, CI  CI + <;1
01
(2.12c) (2.12d) (2.12e)
1.15 Cartesian Coordinates and i, j, k System of Uait Vedon
The three unit vectors that constitute a reference system for the various vector quantities under consideration are usually taken along the directions of the axes of a coordinate 'system used to define a point in space. A
Incidentally, the foregoing examples show how the use of vector notation leads to simplicity of expression. A single vector relation such as (2.11a) is equivalentto three scalar relations such as (2.11c) to (2.11e). Similarly, relation (2.12a) js represented by three equations, (2.l2c) to (2.12e).
z
p. (s, y, z)
I
I
Is
1.14 Specification of a Vector
In the preceeding section we have seen that if two vectors are equal, their components with. respect to a chosen reference system of three noncoplanar unit vectors are also equal. Conversely, if the corresponding ~ompon~nts oftwo.vect~rs are ~qual, the vectors must themselyes be equal 10 magnitude and direction. This means a vector is uniquely determined by a set of three numbers that form the components of the vector. To specify a vector, quantities other than its components can also be used. To illustrate this consider a vector A. Let a, b,.c denote its como. ponents with respect to the unit vectorse,., eli' ~, and let ex, p, ydenote the angres A makes with e,., eli' e•. Then each of the sets of numbers (a, b, c), (a, p, y), (b, y, IX), (c, ex, P), (a, b, y), etc. determines completely the direction and magnitude of A. We may thus describe a vector analytically as a set of three numbers that, in some fashion, are related to a chosen r~ference system of unit vectors, say e h e 2, ea. The set of numbers is ordered in the sense that the first number in the set corresponds to the direction e l , the second to e l , and the third to ea. Furthermore, the three numbers representing a vector must obey certain specific rules, for not every ordered set of three numbers expresses a vector. To see how such rules may be set up, consider any vector and describe it in two different reference systems. As we pass from one reference system to another, the vector itself remains unch~nged, whIle the sets of numbers that describe it will be different in the two systems. However, since both the sets describe the same vector, we can set up a rule to express one set of numbers in terms of the other. Then we
~+I"'" /....,/~.;,..~ Y
1 ______ .... 1. / / /
I
I
x
1
Fla. 1.18 Cartesian coordinates.
familiar example of a coordinate system is the rectangular Cartesian coordinate system. In this system a point in space is specified by three cpordinates measured with respect to three mutually perpendicular axes. We denote the axes by X, Y, Z and the coordinates by x, y, z (Fig. 2.18). The coordinate axes may be designated X, Y, Z arbitrarily, but it is necessary to adopt a consiste~t convention. As shown in Fig. 2.19 there are two distinct ways of disposing the axes. The method shown in Fig.
z
/~y
yJ
or
~y
z
x
(0;
x
(b)
FIa.1.19 (a) Righthanded axes; (b) lefthanded axes.
76
Idealfluid AerodynamiCs
Elements 01 Vector Algebra and Calculus 77
2.190 may be characterized by the convention that a rightJranded rota.ti()n 0 about the positive dircction of the Zaxis through 90 brings the positive Xaxis into the positive· Yaxis. A system of axes oriented according to this rule of righthanded rotation is kaown as a righthanded system. On the same lines the method of orienting the ues as shown fjg. 2.19b may be characterized' by a lefthanded rotation about the axi~ Z. In our work here we shall use righthanded systems of axes exclusively. Let P be a point in space designated by the coordinates x,y, % referred to
in
Now. the sume system ofi, J. k unit vectors may be set up at any point P and uSl!d to describe the component form of any vector assodated with the point. Thus if A is such a vector quantity and if a:r' aw' a, denote the components of A with respect to i, j, k, We wnte A
= (0."
A:=
a w a.) '
=
a",i
+ awl + o,k
(2.16)
The magnitude of A is gtven by
IAI = ..../a"'2 + a. 2 + a,2
(2.17)
f
2.16 Notion of Curvilinear Coordinates In analyling many physical probl~ms it is often advantageous to use coordinates of greater generality than the Cartesians. We shall now see how such general coordinates may be introduced and characterized. In the Cartesian system various points in space are defined by a.ssigning different values to the coordinates x, y, z. In such an X·YZ.6pace, consider a system (If three independent functions expressed by
pelf"~, z)
qJ,. =
q~
(/1(x, y, z)
qa
Fig. 1.20 The I, J, k system of uDit vectors.
""' q.(x, y, z) =q.(x, y, z)
(2.18)
a Cartesian coordinate system XYZ. Let 0 be the origin of coordinates and let r denote the position vector OP. To describe the compone~t fonn of r we choose three unit vcctors al~ng the positive directions of the axes X,Y, Z. We denote these unit vectors by I, j, k, respectively· (Fig. 2.20). They form a righthanded system oforthogonal (i.e, mutually perpendicular) vectors. The components of r with respect to the i, j, k system are simply. x, 1/, z. Thus we write
~ ~
r = OP
== (x, y, %) = xi + yj + %k
= ";za
(2.13)
(2.14)
Since i, j, k are mutually perpendicular, the magnitude of r is given by
r == Irl
+ yl + Zl
If Q;, p, yare the angles r makes,.respectively, with i, j, k, the direction cosines of r are given according to the following relations: x =r • i = r cos Q; (2.15) y = r·j = rcosp z=r·k=rcosy
., Jt
;~
such that there is a unique correspondence between (x, y, z) and (qlt q2. q,). Dy means of these functions we can determine for any point. P with the coordinates x. y, z a set of three new numbers ql' q2' qa. Conversely. if ql. qt, q. are chosen, a point with the coordinates x, y, z can be determined. This means that the position of a poil1.t P in the XYZ space may be specified eit~r by the set of numbers (x, y, %) or by (qh qa. q3)' Thus to each point P(x, y, z) we can assign the' <;orresponrling values lq). qz. qs) as a set of new coordinates. In this sense the system of functions expressed by (2.18) may be interpreted as defining a Transformation of coordinates. * 'The coorJinates ql' q2' qs are ~nown as the general coordi11Qtes of 0 point. Note that qlt Q2. q3 are coordinates and that they need not necessarily possess the dimensions of length. In other words, they are not necessarily fhe compone~ts of the position vector describing the point P. , , Let us now recall.the geometrical significance of a function onhe form f(x, y, ::) = const. Such a function represents a family of surfaces, with each surface of th~family corresponding to a different value of the .constanLWe concern ourselves with cases where only one surface of the family will pass through a chosen point: Consider now the system of
• Another·interpretation of these equations is that they define a mapping .,f the XYZ space on to the .pace of q.. q.. q. c:oordinates. For a detailed discussion on "systems of flinc:tions, transformations, and mappings," see Courant (1934).
customary. to use I, j, k instead of ~,e., e ••
7.
equations expressed by
ql ... ql(%' y, z) == const.
IdealAuid AerodynamiCS
Elements of Vector Algebra and
Calcul~
q. == q.(%, y, z) == const. (2.19) q. == q.(%, y, z) == const. They represent three independent families of surfaces such that~ in general, one surface of each f~mily passes through a chj)ser, point. Then any point in space: may' be located as the point of intersection of three independent surfaces represented by a system of equations such as (2.19). The values of
At the point P we draw a tangent to each of the coordinate· lines. These tangents are taken as the coordinate axes at tlie point P (see Fig. 2.21). The axes are chosen positive in the direction in which ql> q., q. increase from the point P. Along the coordinate axes thus formed we mark out from the point P three unit vectors el> e., e. (see Fig. 2.2l').· This system of unit vectors at the point P can then be used as a reference system for all vector quanfities associated with that point. It should be readily noted that in a system of curvilinear coordinates, the axes and the reference system of unit vectors are not, in general, of fixed' directions in space. Their directions change from point to point in space. We should bear in mind this particular aspect of curvilinear coordinates. 1.17 <>rtIIopDaI, Cun1liDear Coordinates In many problems, when curvilinear coordinates are used, one chooses the coordinates in such a way that the coordinate surfaces intersect at right angles at each point in space. Such coordinates are called orthogonalcurvilinear coordinates. In our studies here we shall be concerned only with orthogonal systems. As examples of such systems we shall consider the following two speCial cases. t
ql surface
I I
I
~
Cy/iNlrical Coortlilflltel.
In ~his system a point in space is located by the
ql r
coordinates .
%
Iz I I I
1//
lJl
//
== (J
shown in Fig. 2.22.' (Note that here i is not the magnitude of the position vector GP.)· The transformation between Cartesian coordinates snd: cylindrical coordinates is expressed by the equations
...v y
Fig. 2.21
Curvilinear coordinates.
r'" ql(%' y, z) (J
.JzI + 1/
arc tan 
1
ql, q2, qa, which belong to the three surfaces passing through a point, are then assigned to the point as coordinates. These are nothing but the general coordinates previously expressed by the functions (2.18). The surfaces (curved in general) described by the equations (2.19) are called curvilinear coordinate surfaces. The coordinates ql; q., q. are therefore also known as curvilinear coordinate~. The coordinate surfaces passing through any point P(ql> q., qJ intersect in pairs ana give rise to three space curves passing through that point. These curves of intersection are called the coordinate lines: The surfaces q2 = const. al'd q3 = const. intersect in a curve along which the c0ordinate ql alone varies. Thus this curve is called the qlIine or the qlcurve. Similarly, we have a q2line and a q.line (see Fig. 2.21).
== qM,%, y, z) z == q.(%, y, z) 
'y
%
z
or, inversely, by
%rcos(J
y == r sin (J zz
• In Fig. 2.21 the naming of the coordinate surfaces is initially so ~osen as to make
elt el, el a righthanded system.
t For further examples and for a discussion on nonortbogonal systems see Margenau and Murphy (1956).
EleJDenta of Vector Algebra and Calculus
IdealFluid Aerodynamics
81
The coordinate surfaces given by (I) T . . const. arc cylinders coaxial with the Zaxis, (2) 8 .. const. arc half plancs through the Za~s, (3) const. are plancs perpendicular to the Zaxis (Fig. 2.23). At any point PC', 8, z) the vectors cr. e" e~ donote the reference system of unit vectors drawn respectively in the directions of increasing ,,8, and z (Fig. 2.22). ,The unit vectotsare orthogonal to one another and form, in the order e r , e" e" a righthanded system. These unit vectors except for e. are generally of different directions at different points in space. Let R denote the position vector OP front the origin 0 to the point PC"~ 8, z). The component form of R is then expressed by
*
~
~~;~y
Rrer+ze.
If A is any vector associated with the point PC', 8, z), and if A r , A., A. are the components of A with respect to the unit vectors er , e., e. at P, we write A .. (Art A" 4,)  Arer + A,e, + A.e.
x
Fla. 2.%2 Cylindrical ,coordinates.
z
SpllniuJ CDO'__~.. In this system a point in space is located by the c')ordinates ' ql  ,
q.  8 q." 9J
as shown in Fig. 2.24. The transformation between Cartesians and spherical coordinates is expressed by the equations
, ~r + 1/' + z·
8arccos
.Jr + Y' +"z'
, :B
z "_arcsin .. arc: Sin
•
~~'+tI Jr + 1/' + z·
9J  arc cos
z
Y
=constant
or, inversely. by
~r+tI ~r+tI ~  'sin 8 cos rp
y,sin8$inrp
%
= 'cos (J
Fig.2.23
Cylindrical(T""din?te surfaces.
The coordinate surfaces given by (I) , == const. are co~centric spheres about the origin, (2) (J:= const. are circular cones with vertex· at the origin and axis along the Zaxis, (3) rp  const. are half planes through the , Zaxis (Fig. 2.25). At any point per, (), '1') the vectors er , e., e. denote the reference system of unit vectors drawn, respectively, in the directions of increasing'. 8, and rp
11
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus are the components of A with respect to the unit vectors I r • • " p. we write A == (A r, A" A,,) == Arlr + A,I, + A.e. 1.18 Prodaca of Vedon In T .... of TheIr CompoaeDts
I.
IJ
set up at
z
.£.+"'1'
x
Fla. 1.14 Spherical coordinates.
We shall now express in component form the various products we have previously considered. For this purpose we choose an orthogonal righthanded system of unit vectors. We denote the unit vectors by e1' e •• I. and the components of any vector A with respect to these by AI' A., A •• • First we consider the scalar and vector products between the unit vectors. A scalar product of the form e1 • e1 is equal to unity. A scalar product of the form ~1 • e. is zero. We thus have .
(2.20a)
and
(Fi . 2.24). The unit vectors are orthogonal to one another ~nd form, in thegorder e , e" e", a righthan~ed sy~tem: These vectors are, tn general, of r different directions at differential potnts tn space. ~. The component form of the position vector r = OP IS expressed by r = rer
(2.20' )
A vector product of the form e1 X e1 is zero. A vector produE:t of the· form e1 x e. is equal to e., and of the form I. x e1 is e.. We thus have (2.2Ia) and el x e. x ea x e. = e. x e 1 = e, e a = Is x e. == e1 e l = el X e l = I.
If A is any vector associated with the point P(r, 8, 9') and if A r , A" A"
z
(2.21 b) (2.2Ic) (2.2Id)
We now consider the produ,cts between two vectors A and B. For the ~Iar product A. B we have
A • B = (Aiel
+ A,c2 +
A.e.)· (B1el
+ B.e. + Ble.)
(2.22)
USing .the relations (2.20a) and (2.2Ib), this becomes
<p
=constant
A • B = A1Bl
+ A.B. + A.Ba
r= cohstllnt
which states that the scalar product of two vectors is equal to the sumo! the products of their corresponding components. For the vector product A x B we have ...
A x B = (Aiel
+ A,c2 + A,e,) + (AaBl 
x (Ble l
+ B 2e 2 + B.e.) + (A1B2
 A 2 Bl)e.
By using the relations (2.2Ia) through (2.2Id), this becomes A)( B = (A2Ba  A aB 2)el
A1Ba)e2
x
Fig. ~.lS Sphericalcoordinate surfaces.
• e .. e" e. may represent the system of unit vectors set up at a point described by any orthogonal, curvilinear, coordinate system.
u
This may be written in the determinant form
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
,s
vector A, we say that the vector A is a function of the scalar' I and write
el
e.
ei
(2.23)
A  A(t)
A )( B 
Al A. A.
Bl B.
B.
whi~h is more cqnvenient to remember.
We now consider triple products betw~ the vectors ~, B, C. The ~esult of the triple scalar product A· B )( C may be written m the det~rmlDant form
Suppose we want to describe the temperature at every point pf a heated body. To do this we specify each point of the body by means of a position vector drawn from an arbitrarily chosen reference point and .y that the temperature T is a function of the position vector r. Symbolica:liy~ we write T ... T(r) Here we have an example of a scalar as a function of a vector. Similarly, if for each· value of a vector r there corresponds a certain value of a scalar t/I, we say that the scalar ~ is a function of the vector r and write
Al A. A.
A • B )( C 
BIB.
B.
(2.24)
Cl
C. C.
",  c/l{r)
When r denotes the position vector, we say that; is a scalar function of position. The distributions of pressure, density, and temperature in the atmosphere are examples' of scalar funct'ions of positioh, Let us consider next a rigid lJody rotating with a constant angular velocity w. As seen in Section 2.10, the ~Iocity of a point of the body is given by where r is the position vector from a reference point taken on the axis of rotation. Different values of r yield the velocities of the different points of the body. We say that tlte velocity is a function of the position vector and write symbolically v  V(r) This is an example of a vectoras a function of another vector. If for each value of a vector r there corresponds a certain value of another vectot.A, we say that A is a function of r and write
A  A(r)
po~itioll.
which is easy to remember.· . '. The result of the triple vector product A )( (B )( C) may be. written 10 the determinant form·
e.
e.
A.
(B1C I  /fIC.)
(2.25)
This concludes the essentials of vector algebra. "!Ie now paSs on to the elements of vector calculus.
1.19 FuaetIoas iDvolvID. Vecton aDd Scalan
We are familiar with the conceptS of calculus as applied to scalars ~hat are functions of other scalars. Extension of such concepts to functions involving vectors and scalars fonils vector. analysis. ~e..shall first look a~ the various types of functions that are likely to arise In the vector de scription of physical problems. . . . d so If we wish tl) describe the motion of a mass pomt 1R s~ce, we Will 0 by specifying at different instances of time its position ~Ith respect t.o so~: point fixed in a chosen frame of reference. T~at IS, we descrlb~ t I position vector r a~ a function of time t, a scalar vanable. Such a functlOna relation is symbohcally expressed by
r  r(t)
When r signifies the position v~tor, we say that A is a vector function of The gravitational fotce experienced by one body in the presence of another ~s an example of a vector function of position. Similarly. the Coulomb force acting on one electrically charged body in the presence of another charged body is a vector function of position. The functional relations we have introduced above are only specia: forms of the more general relations expressed by
This is an example of a vector as a function of a scalar. Si~ilarly, if for c;ach value of a scalar variable t there corresponds a certam value of a
• Proof is left as an exercise,
", =
~r. t)
(a scalar as a function of another scalar and a vector) and A = A(r, t)
Elements of Vector Algebra and Calculus IdealFluid Aerodynamics
2.20 Scalar &ad Vector Fields
11
(a vector as a function of another vector and a scalar). When rand t signify position and time, respectively, we say that tf, is a scalar function ~f position and time and that A is a veclor function ofposition .nd lime. If 10 the case of a heated body the temperature at any point of the body varies with time, we say that the temperature' Tis a scalar function pos.it~on and time and write T ... T(r, I). Similarly, if in the case of a rotating Mgtd body the angular velocity changes with time, the velocit~ at any point ~f the body varies with time, and we say that the velocity V IS a vector function of position and time and write V = V(r, t). . Using the principle of decomposit~on of a. vect~r mt~ scalar co~ ponents, we can interpret the preceding funC!lons lD~olvlDg v~ors ID terms of scalar functions of scalar variables. Such an mterpretatlon sets up a correspondence between the operations of vector calculus with those of scalar (or ordinary) calculus. Consider fint the function A = A(t). Let A .. A., A. be the components of A with respect to a fixed system ofthree un'it vectors denoted by el, e., ea· We thus write A Aiel + A,e. + A,e. (AI' A., A.)
or
=
=
With this represent,ation we can interpret the function A(t) as equivalent to the three functions Al = Al(/), A, A.(t), Aa = Aa(t)
A scalar or a vector function of position assigns to each point of a portion of space a definite value of a scalar or a vector quantity. The various points of the given region together with the corresponding values of the quantity, scalar or vector, form what is called afield. If the quantity concerned is a scalar, the field is called a scalar field; if the quantity is a vector. the field is called a vector field. If we are dealing with a scalar or vector function of position and. time, the values of the scalar or vector quantity at the various po~.nts of the region change from instant to insnnt and the field becomes an unsteady or nonstationary field. If we are dealing with a scalar or vector function of position only, the field retains the same structure for all times and we say that it is Ii steady or a stationary field .. The concept of a field helps us to show what is happening simultaneously at all points of a region of space. An arbitrary poi~t in a field is usually called afield point. Gonsider a scalar field represented by a singlevalued function rf, = .fJ(r). In such a field we can draw a family of surfaces such that each surface passes through .all those points that have the same value of the sc.alar quantity rf,. The surfaces are thus surfaces of constant rf, and are represented by rf,  .fJ(r)= const. with the constant taking a different value for each surface. Such surfaces are commonly referred to as level surfaces. Surfaces of constant density or of constant temperature, or of constant pressure in the atmosphere are all examples of level surfaces. If a scalar field is steady.. its level surfaces remain constant with time. If the scalar field is unsteady, the level surfaces change from instant to inlltant. One may picture a vector field by imagining arrows placed at various points of the region of space, each arrow pointing in the di~ectionof the vector quantity associated with the point and having a length proportional to the magnitude of the quantity. As an example. the velocity field of a rigid body rotating with a constant angular velocity is shown in F~g. 2.26. In a vector field one can draw a system of curves such that each curve is tangent at each point on it to the direction of the vector quantity associated with that point. Such curves are called field lines. If the vector field is a force field. the field lines are known as the lines offorce; if the field is the velocity field of a flowing fluid, they are known as streamlines. A familiar example of field lines is the picture of curves formed by iron filings in the presence of a magnet. The :1eld line~ of the velocity field of a rigid body rotating with a constant angular velocity are shown in Fig. 2.27. If the vector field we are dealing with is stationary, the picture of its field
=
Thus a vector function Qf a scalar variable is equivalent to a system of three independent scalar functions of the same scalar variable. Consider next the function tf, = .fJ(r). If ql' q., q. are the components of r with respect to a system of unit vectors el, e., e. we can interpret .fJ(r) as equivalent to the function. tf, =' tf,(q., q., q.) This means that a scalar function of a vector is equivalent to a scalar function of three independent scalar variables. 'Consider now the function A = A(r). If, as before, AI' A., A. are t?e components of A and q.. q.,q. are the components of r, we can write A = A(r) as equivalent to the system of functions expt:essed by
Al = A 1(r) = A.(ql> q., qa)
A2
=
A2(r)
= A 2(ql' Q2, Q3)
q2. qa)
A3 = A3(r) = A 3 (QI>
Thus a vector function of a vector is equivalent to a system of three , independent sealar functions of three scalar variables. In a similar maAner, the functi,ons rf, ~ rf,(r. t) a~d A = A(r, t) can be express~d in terms of scalar functions of scalar vanables.
"
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
are parallel vectors. Recalling that the vector product of two parallel vectors is zero, we write tis )( A == 0 (2.2~) This, the~, is the (differential) equation that determines (at any instant) field lines of the vector field A  A(r, I). In determining the streamhnes of the velocity field of a fluid in motion we shall have occasion to study the integration of an equation oftbe type (2.26). Using the p,inciple of decomposition of a vector into components, Eq. (2.26) can be readily specialized to any chosen coordinate system. Since the function A  A(r, t) is equivalent to three scalar functions of position and time, any vector field may be regarded as equivalent to three scalar fields.
t~e
"
FI~. 2.26 Velocity field of a rotating rigid body .as.seen in a plane normal to the axis
of rotation.
2.21
DHrenatiatioo of'. Vect6r Faac:tioa of. Scalar Variable
If a vector A ehanges from a value A, to a value A., the increment in A denoted by &A is simply the vector difference between A. and Ai' That is,
&A::oo A.  Ai
A change in ~ ~cct~r ~y be brought about by a change in its magnitude or by a change ID Its dm:ctlon or by a change in both magnitude and direction. If a vector A is a function of a scalar variable'l, the increment M in A corresponding to an increment &1 in I from I to I + &t is given by
lines remains unchanged with time; otherwise tbe picture changes from instant to instant. To cOnstruct analytically the field lines of a vector field" :: A(r, I), we proceed as follows. Consider the field line passing through a point r at some instant oftime. Let tis ~enote an element of the line through r. By definition tis has the same direction as that of the vector '" associated with the point r at the instant considered. That is, tis and A
M  A(I
+
&/)  A{I)
If the ratio &A/ &1 (i.e., the average variation of A with respect to t in the iQterval &1) tends to a limit when &t tends to zero, that limit is called the d'erivative of A with respect to I (compare the definition of the derivative of a scalar function of a scalar variable). Fo1l9wing the usual convention of differential calculus. we denote this derivative by dA/dt and write
dA(r) = lim ~A = lim A(t dt ~IO ~t . ~I"'O
+ ~t) ~t
A(t)
(2.27)
Let us ·now look at the geometrical interpretation of this derivative. If we represent the various values of the continuously varying ~ector A by means of arrows drawn from a common origin, denoted by 0, the terminus of the vector will describe a curve, denoted by ~; in space (seeFig. 2.28a). i..et 'Qp represent A at time t and OQ represent it at time t + M (Fig. 2.2gb). Then the increment ~A is represented by the vector
chollliQ ef the curve~. Thus we have
~A
.... 1..l7 Field lines of the. veloci'Y field of Fig. 2;26.

chord PQ
~t
"
90
Ideal.F1uid Aerodynamics
Elements of Vector Algebra and Calculus
'1
and
point P. Denoting bye, a unit tangent vector· at P (sec figure). we write
~
dA .Ii M.. I' PQ   m   Imdt A'''O fl.t A'''O fl.t
'
(2.28) With this, Eq. (2.28) becomes
A.~O
lim PQ
as
== e.
To interpret the'limit we proceed as follows. A point such as P .or Q on the curve t'6 can be specified by giving either the v~tor A or the distance s measured along the curve from some initial point taken as ~lerence (sec . figure). As t varies. $ will change justas A doeS; so $  $(1). and A may be
dA dt
expre~sing
==  e.
ds dt
(2.29)
the derivative as the product of a magnitude and a direction. As an example of the above consloerations, let us consider the motion of a mass particle. At any instant let its position be denoted by r == r(1) measured from a point fixed in a chosen frame of reference. The path (or the trajectory) of the particle is given by the curve «j traced by the vector r as r varies. The velocity V of the particle at any instant (i.e., at the position r) is given by the derivative dr/dr. Thus we have
v == dr(t)
o
(t;J) (6)
(b)
dt
= ~e dt'
==
Ve
•
FI¥o 1.21 (a) Space curve traced by A(t):
illustrating the differentiation oC A(t).
considered as depending on s. Let /).S denote the increment in s'from P to Q. Therefore fl.s = length of the arc PQ Introducing /).s, we rewrite Eq. (2.28) as
dA "' lim PQ = (lim (lim /).s) dl 4t~O fl.B fl.t 410 /).s At~O /).1
~
~~
PQ)
= ds lim PQ dt A.O as
Now, PQ/fl.s is a vector along iQ with a magni!'\) ~ equal to
'
(2.28a)
stating that the velocity is tangential to the tnrjectOlY at the instant considered and that the magnitude V of the velocity is equal to the rate of change of distance along the trajectory (i.e., to the speed). As a simple result following Eq. (2.29) we note that the direction of thl! derivati,'e dAidt when A is of constant length but of changing direction is perpendicular to the rector A. We consider next the differentiation of the sums and products of vector functions a!l of which depend on the same scalar variable. In all such case!. the formal methods of differentiation as employed in scalar calculus. are equally Hpplicable except that in cases involving vector products, Ihe order of the vectors must be preserved. This is. of course, a natural consequence of the fact that vector products are not commutati,Ve. ACL'Ordingly, we have the following results. . The higher derivatives of the function A = A(:) uc cl'nstructed by successive differcliti.ilion just as in scalar calculus. If U =~ 11(t) is the sum of two fUnCli,)ns such that
tJ( t) == A(t)
length of the (:hord PQ length of the arc PO As /).S + 0,
+
B(f)
we JlIl.\':JU dA dB =_.1
I~Il
and the direction of PQ. becomes that of the tangent to the curve ~ at the
• The
dircctio,~
tit
of the tangent vector is
ril' d t
t;~en
in the direction of increasing s.
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
93
If, however, the unit vectors are also changing with the scalar t, we have
If u = U(/) ami A = A(/), we have
d dA du (uA) = u  +A
dt . dt dt
dA dAI A del dA 2 4 de2  = el + 1  + ez + 2 i
dt
dt
dt
dt
dt
+ dA3 e3 + dt
A dea 3 dt
(2.3Oc)
Here the order of the factors involved need not be preserved. If A = A(t) and B = B(t), we obtain·
dt (A • B)  A •
d
di + di . B
dB
dA
Since the scalar product is commutative, the order of the vectors in this differentiation need not be preserved. The derivative of the cross product' A(t) )( B(t) is given by
dt
~ (A )(
B) = A )( dB dt
dB dt
+ dA
dt
)( B
To illustrate the case in which the reference unit vectors are also chanJlOg let us consider the description in cylindrical coordinates of the motion of a mass ·particle. Accordingly, we denote at any instant the position of the particle by r, 0, z and bye" e" e. the corresponding unit vectors. We ask for the .velocity, V, of the particle at the instant considered. By definition, the velocity of the particle is equal to the rate of change of its position. Thus if R = R(t) gives the position of the particle with respect to a fixed point, we have dR V = Vet) = dt
and is not equal to
In cylindrical coordinates
A )( 
dA + B x ', etc.
dt
Therefore we obtain
V =  (re, dt
Since the vector product is not commutative, here the order of the vectors should be preserved. Considering triple products, we have
d
+ ze.)
d dA dB de  (A • B)( C) =  . B )( C +A •  )( C + A . B )( d
lit dt . dt t
Since the direction of the unit vector e, changes with change of location of the nass particle, this equation expands to
and
~ [A x (8 x C)] = dt
dA x (B x C) + A x (dB x
dt dt
C) + A x (B x ddC ),
. t
dr V=e dt'
+r+e
de, dt
dz. dt'
Here again the order of the vectors has to ~ pr~served. , In concluding this section we rdate the derivative of the vector A(t) Wlt~ the derivatives of its components. To do this we choose a system of UOit vectors el> ea, e, and express
A(t) = A!el
To evaluate> the rate of change of the unit vector e, we proceed in the same way as in deriving Eq. (2.29) and obtain de, dO =e. dt dt With this relation, the velocity expressed In cylindrical coordinates becomes dr dO dz . V =  e + r  e +  e. (2.31) dt' dt' dtIf we did not recognize that e, is changing. we woutd have arrived at the incorrect result that the velocity of the particle is equal to
+ A 2e Z + Ases'
dt
We therefore obtain dA d  =  (AIel)
dt dl
d +
(A i e 2)
d + d (A 3e3)
t
(2.30a)
If the unit vectors are constant,· this reduces to
~~
dt
= dAl e
at
1
+ dA
dt
2
e2
+ dAs e3
dt
(2.30b)
e
T
• }.. .:hange in a unit vector <:an ~ brought about by only a change in its direction, for by definition it. magnitude i'!l always unIty,
dr dt
+ dz e dt
Z
94
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
95
1.22 Cbanps In tile Volt Vectors of CyUadrlcal and Spherleal CoonIJnates When we move .from one point to another in cylindrical or spherie&l coordinates, changes occur in the directions of the reference unit vectors. We' shall now determine these changes.
infinitesimal rotation (2.32) of the system e" e" e.. Then the change in any of the unit vectors is given
by
(2.33) where the subscript j m;ry be r, 8, or z. Using relations (2:32) and (2.33) we obtain the changes in the unit vectors as
Cyli1Ulrictd Coor"'e'. infinitesimal distance
ds
We move frpm a point It .. (r, IJ, z) over an
== dr e, + r d6 e, + th e.
in some direction. Here e" e" e. are the unit vectors associated with the point It. Let e:, e.', e.' be the unit vectors associated with the point R + ds. As shown' in Fig. 2;290, the system e:, e,', e,' results from an
de, == dB e. x e, == d8 e. de, == dB e. x e, == dB e,
de.
(2.34)
== d8 e. x e.
== 0
SPMriCtl' Coort/i",,'el, Now we denote a point in space by r We move.in some directio~. over an infinitesimal distance
ds
== (r, 8, cp).
== dr e, + ,dB e. +
r sin Odcp e.,
where e" e" e., are the unit vec ors associated with the point r. Denote by e/, e.', e.,' the unit vectors associated with the point r + ds. We observe, as indicated in Fig. 2. 29b, that the system e/, e,', e.,' results from an infinitesimal rotation (2.35)
(a)
of the system er , e., el , where e is a unit vector in the direction of the axis from which 0 is measured. Expressing e in terms of e, and e. by the relation e == cos Oer " sin Oe, we write Eq. (2.35) as
d+
== dcp cos Oe, 
dcp sin Oe,
In
+ dOe.,
(2.36)
By using the relation for d+, the changes determined from the equation
the unit vectors may be
.
where the subscript i may be r. 0, I)r f1i. WeJhus obtain
de r = dO e8
des
(b)
+ d~ sin (Je.. ==  dO e + dfli' cos Oe.,
T
(2.3 7)
de,/,
(b) spherical
= dcp sin Oer 
drp cos Oe8
~imilar
Fig. 2.29 Changes in unit vecfurs: ;:oordinates.
(a) eylindrical coordinates;
In con.::\uding this section we' wouid like to mention that in a
96
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus Expressing A in component form we write
97
way one can determine the changes in the unit vectors of any orthogonal, curvilinear coordinate system. See Section 2.45.
2.23
Fram~
of Reference
K'dA dt
 ==
K'd

In the preceding considerations the vector quantities were described with respect to a chosen origin. Such an origin is a poiht fixed in soITIe frame of reference. ay a frame of reference we mean a frame in space and time that will enable us, by suitable measurements, to describe physical phenomena such as the position of mass points and the passage of time. Although we may postulate the existence of a reference 'frame that is absolutely fixed in space and time, we are obliged, for practical reasons, to deal with reference frames that are in relative motion with each other. This being the case we should recognize that operations performed in one frame of reference on functions involving vectors and scalars will yield results that are, in genual, different from those obtained by similar op'crations in another frame of reference. This means that an explicit mention of the reference frame in which the operations are being performed is essential unless it is understood that once and for all a particular reference frame is all that is employed. • We shall now illustrate some of these ideas by working out a particular problem. Suppose that K and Ko are two,frames of reference such that the frame K rotates with an angular velocity w = W(/) with respect to the frame Ko, 1 being time. Consider now a vector function of time, say A = A(/). Our problem is to find the relation between the derivative of A with respect to t as computed in the frame K and the similar derivative computed in 'he frame Ko. First of all we must introduce an explicit notation to distinguish the derivative operations in the two frames. Accordingly, we shall denote by
K
dt
(aiel + ate! + a3e a)
= ( el
K'da K'da K'da ) l+e I+e·· dt I dt I dt K'de _2 'dt
+ ( a1 K'del + a dt
+ a. Kode3) _
dt
(2.38)
. ConsIder the term Kodal . S·mce t he d ' . 0 f . a sca1 f t ' envatIve ar unc Ion 0 f a dt sealar variable does not depend on the reference frame, we note that KOdal Kdal dal = =dt dt dt
We can, therefore, write
Kadal
el
Tt = e Tt =
1
Kdal
Kd(ale l ) dt
(2.39a)
since e l is independent of t in the frame 1(. Similarly, we have
Ktdal
e,
and
Tt =
dt 
Kd(a2et) dt
(2.39b)
e Ktdtl ___ e) _' Kd(a 3_3_
• dt
Combjning the relations (2.39), we arrive at the result
(2.39c)
0i!. the derivative operation in frame
dt
Ko
and by
K
el
K
Kada
_1 + e z
dt
Kada Kada Kd  ' + ea ~ =  (aiel + a2el + aaea) dt dt dt
=
i!. the derivative operation in frame
dt
(2.40)
dt
Let e l , e 2 , e a represent a system of unit vectors fixed in the system K, and let ai' a2> aa be the respective scalar components of A. We observe tnat the unit vectors e b e 2 • e a are not functions of the variable 1 in the frame K. whilc.> they are functions of t in the frame Ko. The scalar components are simply scalar functions of a scalar valiable in either frame of reference; in this case the distinction between the reference frames becomes irrelevant.
• The differentiatIOn of vectors v.ith particular emphasis on the importance of reference frames is extensively discussed by Kane (1961).
Consider next the term
Kadel
dt
Since e l is a fixed vector in the frame K that is rotating with angular velocity W(/) with respect to frame Ko, it can be . erified that
KOde l dt
=W
x el .
98
Then we can write
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
99
(2.41a) Similarly, we obtain (2.4tb) and
completely the variation at r of a fUDction 4>(r), we need a set of three numbers. As will be seen, this set of numbers represents a vector. Such a vector is called a gradient vector or simply a gradient. To discuss the nature and significance of a gradient, let us first choose Cartesian coordinates x, y, z and write r == (x, y, z) == xi
+ yj + zk
as
KOde3  == w x a 3e , dt
(2.4lc)
z
Combining the relations (2.41), we arrive at the result
al
•
a 2e 2
xOde l dt
+
a2
KOde2 dt
+ as
KOde
d~1 == w x (aiel
==wxA
+
+ a,e,)
(2.42)
By using the relations (2.40) and (2.42), Eq. (2.38) may be rewritten as
.JLy o
x
FIg. 2.30. Gradient.
KOdA KdA == +wxA dt dt
(2.43)
This gives the required relation between the derivatives of A(I) in the reference fra!1les K and Ko. 2.24 Differentiation of a Scalar Function of a Vector:. Concept of a Gradient and
t/>
Let us denote by
== t/>(r) =
t/>(x, y, z)
Let us consider specifically a scalar function of position t/> :: 4>(r) and ask for its spatial variation. The conclusions we arrive at for this function., are equally applicable to other scalar functions of a vector variable. Since the independent variable in the function 4>(r) :s a vector, we have the choice of an infinite number of directions in which to take the increment ~r or. equivalent.y, the differential increment dr. The differential increment in 4> corresponding to dr would, in general, be different in different directions. This means lhat in describing the variation of <P we must specify the direction in which the variation is taken. We thus talk about the 'spatial derivative of 4> in a particular direction and refer to it as a directional deriL"!1til'e. According to these ideas it would appear that to describe completely the spatial variation at any puint of a function 4> = 4>(r), we may have to specify the derivatives of cb in all possibie directions at thElt point. Fortunately, however, this is not necessary. It turns out, as we shali sce, that all that is necessary is to give the derivatives of 4> in three indepcnuent directions. These three derivatives are then sufficient to d€terminc the variation of <P ifl any ether diredicn. Thus to ~pecify
ds == dse
a small increment in r in some direction e. Hereafter we adopt the notation ds for dr so as to avoid confusion between Idrl and dr = d Ir!. which are not the same. With this notation
Idtl == Idsl
We observe that
== ds
ds
= (dx, ely, dz) = dxi + dyj + dzk
ds is given by
(2.44 )
(See Fig. 2.30.) The change in ~ over the directed distance
dt/> =
o~  dx
ox
o~ o~ + oy dy + oz dz
\)f
Recalling that the scalar product of two vectors is the sLim
the products
1()(}
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
101
of their corresponding components. we rewrite Eq. (f44) as
dcp = (acp i ax ax
+
ay
acp j
+
acp
az
k) . (dxi + . ds
dyj
+ dzk)
(2.45)
acp . acp . at/J) = ( I+J+k ay
where hI> h 2 • ha may be referred to as scale factors. These factors may vary from point to point; in other words they are. in general, functions of position. They can be determined in terms of ql, q2' q3' This is shown in Section 2.45 where further details are given. We may now write Eq. (2.47) as
az
dt/J = = =
.!. ot/J hi dql + 1. at/J hi dql + !. at/J ha dq3
hi Oql _hi oql
ha aqa'
Dividing Eq. (2.45) by ds we obtain
dcp
ds
= (acp i + at/J j + ot/J k) .ds
ax oy
az
.!. at/J l5s1 + .!.. ot/J
hi aql hi OQl
1
ds
hi aq,
h z aql
l5s1
+ 1. ot/J bS 3
ha aqs
= (at/J i
ax
+ at/J j + ot/J k) .e
oy oz
(2.46)
(1. at/J e + .!.. ot/J ez + .!. :ot/J es )
ha 0'13
• (!5s l e l
+
!5s ze z + oSse 3 ) (2.48)
We thus see that the directional derivative of t/J(r) in any chosen direction is equal to the component in that direction of a particurar vector. The components of that vector are the partial derivatives of t/J with respect to distances along the three coordinate axes (see Fig. 2.30). We arrive at the same conclusions if instead of Cartesians we choose any orthogonal, curvilinear coordinates. To see this, let us denote a point P i,l space by a set of orthogonal, curvilinear coordinates ql> qz, qs (see Section 2.17). Then the scalar function of position t/J is expressed as t/J = t/J(r) = t/J(q1> qa. q3) Note that qh q2, qa are coordinates and are not necessarily the components of r. As before, let ds = dse denote a differential increment in r in some direction e from the point P. If dq1> dq2, dq3 are the corr~sponding increments in the coordinates, the differential increment de/> over the divided distance ds is given by (2.47) The partial derivatives appearing in Eq. (2.47) are not derivatives with respect to distances. Also, dq1> dql' dqs are not components of ds. To express Eq. (2.47) in a·form similar to Eq. (2.45), we proceed as follows. Let S1> S2. Sa denote distances measured from P along the ql' qz, qs curves respectively. Let OSlo !5s., bs, denote the distances along ql' qz, q3 curves corresponding to the increment's dql' dql, dq3' Let us say that the dq's and the !5s's are related as follows
where e1• ea. e a are the unit 'leCtors at P in the directions of the coordinates ql. qa, qa (See Section 2.17 an.d Fig. 2.21). Now since I5S1> bs 2• !5s s are differential lengths along the ql, qz, q3 curves, we have ds = !5s1e l
+ bsze z + oSSe 3
ez +
(2.49)
By using this relation, Eq. (2.48) may be rewritten as
dt/J =
(.!. ot/J
hi aql
e1
+ 1. ot/J
hI aq,
.!.. acb
h3 aq3
e3 )
•
d"
(2.50)
which has the same form as Eq. (2.46). Dividing Eq. (2.50) by ds we obtain the spatial'derivative of t/J in the direction e as'
(2.51 )
We thus see that to every scalar function of position there corresponds at each poin~ a particular vector which determines at that point the spatial variation of the scalar function. We call such a vector the gradient of the function and denote it by the word grad. With this notation Eq. (2.50) can be put in the form (2.52) dcp = grad cp • ds and (2.51) in the form
d cp = grad cp • f ds
bS I = hi dql bS a = hI dqz bS 3
(2.53)
== h. dqa
The gradient is then.defined by stating that in any orthogonal, curl'ilinear
101
IdealFluid Aerodynamics
Elements of Vector Algebra anJ Calculus
I(JJ
coordinate system the components of the gradient of a scalar function of position are simply the partial derivatives of the function with respect to distances in the directions of the respective coordinate axes. Symbolically, we write·
grad t/> _ (at/> e l
OSI
+ at/> ~ + at/> ~)
as. ass
but the field of the gradient of the scalar function in question. With this field interpretation, an interesting result follows from Eq. (2.53). Consider any level surface ofthe scalar field t/>(r). By definition", is a constant on such a surface. That is, at any point on a level surface
=
(1. at/>
hi Qql
el
+ 1. at/> ~ + 1. at/>
h. oq.
hi oqa
ea)
(2.54)
dt/> 0 ds
for every direction lying in the surface. Then from Eq. that
(i. 53)
it follows
The components of grad t/> in Cartesians are given by
at/> at/> at/>
ox' oy' 0%
in cylindrical coordinates r, 8,
%
by
•
at/> lot/> at/> ar '; a8 '
az
and in spherical coordinates r, 8,
qJ
by q, =constant
surface
at/> 1 at/> 1 at/> or ' ; 08 ' r sin () aqJ
Further significance of a gradient can be gathered from Eq. (2.53). If, at any point, the components ofgrad t/> are constructed in different directions, the component that is numerically the greatest will be in the direction of the gradient itself and will be of the same magnitude as that of the gradient. This means (from Eq. 2.53) tb.at the greatest value of the derivative dt/>Ids at a given point occurs in the direction of grad t/> and equals the magnitude of the gradient. Conversely, we may state that at any point the gradient of a scalar function of po~ition t/>, is equal in magnitude and direction to the greatest derivative of t/> with respect to distance at that point. In general, the gradient ofa scalar function of position varies from point to point of the region of space in which the function is defined. Thus. to every scalar function of position t/>(r) there corresp<;mds a certain· vector function of position. grad t/>(r), which describes the spatial variation of rP at all points of the space concerned. ' Since a scalar function of position describes a scalar field, the concept of a gradient is directly applicable to a scalar field. Thus the spatial variation of a scalar field is given by a certain vector field, which is nothing
• Although for convenience we use the representation given in the top line of (2.54), we always mean by that lhe repre~litaticn given in the lower line of (2.54).
o
Fla. 2.31 Level surface.
when e lies in a level surface. Since a vector has no component in a direction normal to itself, we conclude that grad t/> at any point is normal to the lhel surface passing throQ'gh tha, point (Fig. 2.31). If we m~p the scalar field t/> by means of its level surfaces and draw at the. same time the field lines of grad t/>, we find that the field lines. intersect the level surfaces orthogonally. We have defined here the gradient by means of a differential operation. A definition of the gradient by means of an integral operation will be given in Section 2.31.
I
2.25 Differentiation of a Vector Function of a Vector
Consider a vector function of position A = A(r). Here again the independent variable is a vector. Therefore, when we speak a~o~t the spatia! variation of A, we must say in which direction that vanatlon IS constructed. To see what is required to describe completely the spatial rate
101
IdealFluid Aerodynamics
·Elements
or Vector
Algebra and Calculus
105
of change at any point of the vector function A(r), let us choose Cartesians and write A(r) == A) + A.j + A.k As before, let ds == dse be a differential increment in r in some direction e. Let dA be the differential increment in A over the directed distance ds, and let dA",. dAw. and dJlf. be the corresponding increments in the scalar components of A. We therefore write (2.55) dA = dA) + dA.J + dA)I Since A." Av. A. are scalar functions of position, according to Eq. (2.52), the increments dA,., etc., are given by
dA,
If instead of Cartesians we choose a system of orthogonal, curvilinear coordinates, we would arrive at the same conclusion. In this case, however, the elements of the tensor describing the spatial variation of A(r) are not simply the partial derivatives of the components of A with respect to distances along the coordinate axes. They now include additional terms that arise due to the fact that the directions of the reference unit vectors change with change of position. To see this, we choose some orthogonal, curvilinear coordinates"!t> q., q. and express A(r) as A(r) == Aiel + A.el + A.e.
wnere elt el, e. are' the reference unit vectors associated with the point r andA It AI' A. are the components of A with respect to the system e l. et, e,. The compOnents and the unit vectors .are functions of position. The differential increment in~A over a directed distance tis == dse is then given by dA  {(dAI)e1
== ds • grad A,
(2.56)
where the subscript rmay.be z, y, or z. By using (2.~6), Eq. (2.55) may be rewritten as dA = (tis. grad A~)i
+ (tis. grad A.,)j + (ds· grad A.)k
(2.57)
+ (dAJe. + (dA.)e.} + {AI(del ) + A.(de,J + A.(de,J}
(2.59)
Dividing this equation by ds we obtain, in Cartesians, the derivative of A with respect to distance in any direction eas
dA = (c. grad A.,)i ds
where dA I , etc., are the changes in the components over the distance cis, and del' etc., are the corresponding changes in the unit vectors. Now, as before, we can write (dAI)el
+ (e. grad A.)j + (e· grad A.)k
(2.58)
+ (dAJe. + (dA.)e.
== (tis· grad AI)e1
Equations (2.51) and 0.58) show that to determine the sPatial variation of A(r) in any direction from a given point, we need to kn~w at that point a set of three vectors associated with 4., namely the gradients of A. z' A A v' '" Equivalently we need to know a set of nine numbers that constitute these three vectors. In Cartesians, the set of nine numbers is given by the array
+ (tis • grad AJe. + (tis· grad A,)e.·
+ (cis· XJe.+ (tis· XJe.
(2.60)
After evaluating del' etc., (see Section 2.45) it is possible to ·..rite AI(deJ
+ A.(de,J + A.(de,J
.. (tis. Xl)el (2.61)
eM,. aA,. oA.,.
ax oA. ox. oA. ox
oy
oz
oy az oA. oA.. oy oz
~ ~
where Xi, X., X. are vectors. involving AI' A., A. and components of the vectors del' de., and de~. Verify this for cylindrical and spherical coordinates. Combining the relations (2.60) and (2.61) and introducing the notation
WI
== grad Al + Xl
(2.62)
+ X2 WI == grad As + X,
W t := grad A2
The elements of the array are the various partial derivatives of the components of A with respect to distances along the X, Y. Z axes. The elements obey the same rules. as the elements of whai is known as a secondorder tensor. We may, therefore, describe the array of nine numbers as a secondorder tensor and state that the spatial variation of A(r) is specified completely by a secondorder tensor. .
equation (2.5 0 ) may be or, dividing by ds, as
~ewritten
as
(2.63)
dA = (ds· WI)e l
+ (ds· Wz)e;! + (ds· W,)e,
dA  = (e· W1)eI + (e· W 2 )e2 + (e· Wa)ea
ds
(2.64)
106
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
1(17
where e is the direction of tis. Thus we see tnat in any orthogonal, curvilinear, coordinate system the spatial variation at any point of A(r) is given by three vectors or by a second order tensor. As seen from relation (2.62), the elements of the tensor are not simply the partial derivatives of the components of A with respect to distances along the coordinate axes. The tensor that specifies completely the spatial variation of A(r) is known as the tensor ~radient of A, usually denoted by the symbol grad A. We thus have (2.65)
This tensor is usually known as the 'rotation of A. As seen, it is antisymmetrical about the diagonal formed by the elements that are zero. The strain of A, being symmetr:ic, actually contains six independent elements. The rotation of A, being antisymmetric, actually cont~ins three independent elements. In other words, the rotation of A is actually specified by a set of three numbers. This set of three numbers, as may be verified, obeys the same rules as a set specifying a vector. This means the rotation of A, although it is a secondorder antisymmetric tensor, oan be represented by a vector. Denoting ~uch a vector by B, its components in the directions of the reference unit vectors el' e a, e l are given by e l • B = I(WII

Wu )
WIl)
where the element ~; (i andj may take any of the values 1,2, or 3) denotes the jth component (i.e., m the, direction of e;l of the vector WI' again i being I, 2, or 3. The vectors etc., are defined by the relation (2.62), In general, the tensor gradient varies from point to point. Hence we say that the spatial variation of a veetor field is given by a tensor field. In the vector description of physical problems~ we are not directly concerned with the complete tensor gradient of A(r). Only certain combinations of the elements of the tensor are significant. Three such combinations are particularly important. One of them is ~ scalar quantity obtained by summing the diagonal terms Wu , W n • and Was. This sum is known as the divergence of A and is denoted by div A. We thus have
ez ' B
=:
I(Wl l
WI'
e.· B = l(WIi  Wu ) For reasons that will become evident later, a vector equal to 2B is known as the curl of A denoted by curl A. The physical significance of the names divergence, rotation, and curt will become apparent later on when we shall define divergence and curl of a vector by means of certain integral Qperations. For the significance of the name strain, reference may be made to Section 9.1. The reader n!ay verify that in Cartesians the follo,wing results are obtained:
div A = Wu
+
WII
+
Was
The other two combinations are secondorder tensors. One of them is given by the array
WlI
2 oA,!
ox
strain of A = 
oA", oy
+ oA.
ox oy
oA", oz oA. OZ
+ oA.
ox
+ +
WZl
W+ WII)
lI
2Wu Wu Wu
Wu
+
Wit
loA. 2 ox oA. ox.
oAr +. oy OZ
2 oA.
+ oA.
oy
2Wu
and
+ oA",
oA. oy
j
+ oA.
oz
k
2 oA. OZ
This tensor is usually known as .the strain of A. As seen, it is symmetrical about the diagonal formed by the elements Wao W 22 , and Wu. The other secondorder tensor is given by the array
curl A =
0
0
0
OZ
ox oy
o
Rotation of A is simply
A", A. A.
1 curl A.
101
2.16 Del, the Vector DUl'ereatial Operator
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus and
109
Consider the expression (2.54) for the .gradient of #,r) in any orthogonal, cllrvilinear, coordinate system: grad
t/>  e1 ~aa + e. at/> + e. at/>
SI
as.
as,
This expression may be rewritten as
This means to
'a a a t/>= ( ela + e'a + e. a ) t/> SI s. s. obtain the gradient of t/> we operate on t/> by the
grad
In carrying out these operations in coordinate systems other than the Cartesians, we should take proper account of the fact that the system of reference units el , e t , e. 'changes with change of position. It may be verified that in Cartesians we obtain the .following results:
operator
and
V • A _ oA", + oA. + oAr. Ox 011 0%
el

+ e.. os. + e.OSI os,
000
t
VxA==
This is a vedor differential operator and is usually denoted b)' the symbol V, called del. We thus define·
ox
0
011
a
az
0
V .== el  0
oSI
+' e· + e.a a a
==
el

1 a 1 a + e. +hi Oql . h. aq. h.aq. 1 0
os.
.A", A. A.
(2.66) These results are identical, respectively, with the divergence and curl of the vector A (see Section 2.25). Thus it is usual to set V.A and V x A defined as
S.
The expression for e.:t'in Cartesians is
== divA == curl A
(2.69) (2.70)
V _,I.!
in cylindrical coordinates r, 0,
r % is
ax
+ J1. + j! oy a%
r
Tile Operator B • V. If B is any vector. an operator B. V can be
o 1a a V=e +e,+ear ao •0%
and in spherical coordinatesr, 0, «pis
B . V == (el BI + e.B. + e.B.) . ( el .i. +
OSI
e• .i.
OSt
+ e3.i.)
010 V == er  + e,ar r· 00
1 a + e.r sin 0 a«p
a 0 0 == Bl  + B, + B.as, as.
OSI
os.
(2.71)
TluOperators V • tUUl V x. Since deLis a vector operator, the operators V. and V x may be introduced and applied on any vector field. If A(r) is any vector field, the scalar product V • A and the vector product V x A are formed as follows:
It is seen that B· V is a scalar differential operator. ,In Cartesians we have
~+B .B·V=B'" ox •1.+B! oy , oz
Applying the operator to a scalar function #.,r) we have
V· A = (el
OSI
0 + a + e. a • (e AI + e.A, + e.A.) 0)
el
as,
Sa
i
(2.67)
• Recall that although for convenience we use the representation in the top line of (2.66), we always mean by that the. representation in the lower line of (2.66).
(2.72)
110
Applying it to a vector function A(r) we have
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
III
(B. V)A
==
(Bl ,:>0+ B, ,:>0 + B. ,:>0_ '(Aiel + Aat, + AA)
tlSI tiS, tlsJ.
(2.73)
In carrying out the operation, account must be taken of the tact that the unit vectors el , e t , e. change with change of position. Of particular significance, are the operators V and e· V, where ds == dse is, at any given point, a small increment in some direction e of the pOsition vector r .. We recall [see Eq. (2.52)] that the differential change in a function t/>(r) over the directed distance tis ftom a given point is given by
as.
with respect to distance)in the direction e. In applying Eqs. (2_76) and (2.77) to coordinate systems other than .Ca~sians we should take note of the fact that the reference unit vectors change with c~ange of position. . The utility of the operator V lies in the fact that in working out problems we can treat V formally as a vector and apply to it the rules of vector al~ebra and· calculus. In doing this, however, we should bear in mind that V is an operator and not an actual vector. Therefore, in the formal application of vector rules to V, it is necessary. to preserve the order in which del appears with respect to the otJ:!er factors involved. For instance, even though A· B = B • "A, the operation V· A is not equal to the operator A • V. 2.27 Integration of • Vector Function of • Scalar
dt/> == ds· grad t/>
This may be rewritten as
dt/> == (ds • grad)t/>
.== (ds • V)t/>
(2.74)
If a vector A is a function of a scalar variable t, we can form the socalled indefinite integral
This means that the o~rator ds· V, when applied to a scalar function t/>(r), yields the differential chan,ge in t/> over the distance ds. Similarly, we have
f
A(t) dt_
(2.78)
dt/> == (e. V)t/> ds
(2.75)
showing that the operator e . V when applied to t/>(r) yields the derivative of t/> with respect to distance in the direction e. Consider now the expression (2.57) we have obtained, in Cartesians, for the change dA in the function A(r) over the directed distance ds = dse from a given point. We have . dA
in the same. manner as is done in scalar integration (i.e., integration of a ~calar function of a scalar variable). The result of the integration (2.7.8) IS another vector function of the scalar t and is determined to within an additive crostant which, in general, is a vector. We thus write
fAct) dt =
It follows that
B(t)
+C
==
(ds· grad AJi
+ (ds. grad Aw)l + (ds· grad A.)k
dB
dt
== A(t)
This may be rewritten (noting that i, j, k are fixed directions) as dA
== (ds • grad)(A) + (ds • grad)(AJ) + (ds· grad)(A.k)
= (ds • grad)(A)
If the variable t changes ~ntinuously from a particular value I to another particular value .t~, the integral 1
.
+ Awj + A.k)
(2.76)
= (ds • grad)A = (ds. V)A
i
ts
A(t) dr
11
is the definite integral of A between the limits 11 and I,. 2.28 Uoe Integrals: Circulation Consider a scalar function of position t/> = t/>(r) and the field 'described by it. In such. a field let'" r:epresent a space curve drawn from point a to another pomt b. We asSign a direction to the curve as that of travel along the curve from a to b. Let r denote the position from some origin' of a point P on the curve and ds an element of length along the curve froln
This means the operator ds· V when applied to a vector function A(r) yields,just as wher applied to a scalar function t/>(r), the differential change in A over the directed distance ds. Similarly, it can be seen that dA =(e·V)A ds
(2.77)
a
showing that the operator e • V applied tC' A(r) yields the derivative of A
111
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
113
the point P (sec Fig. 2.32). If e, is a unit vector tangential to the curve at the point P, we have tis == dse,. The integral
or, equivalently,
f f
c/>(r) d.
of the integral (2.79) depends only on the endpoints and, becomes independent of the path that joins them. We shall look into those conditions later. We note that another line integral or A(r) may be formed as follows: fA(r) )( d. or fbA(r»( e. ds
c/>(r)e, ds The result of this integral is a vector.
Circulation. Line integrals of the type des~ribed above may be formed around closed curves. Of particular interest is the integral
taken along the curve ~ is called the line integral of t/> along the path ~. The value. of this integral is a vector.
b
fA. ds or
fA. e,d, or
fA. cos at ds
around a closed space curve~. Such an integral is known as the circulation of the vector A ~round the curve~. In general, the value of the circulation is nonzero and depends on the function A(r) aDd the closed curve~. In certain circumstances, however, the circulation vanishes and becomes independent of the curve. We shall look into these details later ..
2.29 Surface Integrals
Fic. 1.31 Line integral.
Consider an open surface S drawn in the field described by a scalar function of position c/>(r). Let the surface be divided into a number 0.( infinitesimal elements. Each of the su.face elements may be considered as a plane area and denoted as a vector
dS
ConMter now a vector runctlon of position A(r). I(~ is a.spaec curve as before, we:can form the ~ine integral
== odS
'c'A. ds 1.
or
fA. e, ds •
(2.79)
along the curve ~ between the given endpoints (sec Fig. 2.32). th.e integral is simply the, integral along ~ of the component of A tangential to the curve_ 1f, as shown in tile figure, at is the angle between e, and A. the integral (2.79) may bewrittelJ as
o is a unit vector normal to tlfe surface element (sec Fig. 2.33). The unit vcC"+or 0 is drawn arbitrarily from one side or the other of the surface S. If, however~ a direction of travel is first assigned along the boundary curve ~ of the surface, the direction of 0 is chosen according to the right,. hand rule with respect to the direction of travel along~. Using these notationswe·form the integral
II
.f]
t/>(r) dS
or
II
8
t/>(r)o ds
fb A cos
at
ds.
'where A is the magnitude of A. The result ~f th.is integral is a scalar. In general, this line integral, like any other hne ~nte~ral, d~pends on the 'function A(r) the path 'alol1g which the integration IS camedout and 011 the endpoint; of the path. Under certain conditions. however, the value
over the entire surface S. Such an integral is called the surface integral of t/> over the'surface S. The result of the integral is a vector. Consider next a vector funqtion of position A(r) and let S be an open surface drawn in its field. Then the integral
(2.80)
s
8
111
IdealFluid Aerodynamics
Elements of V~or Algebra and Calculus For a scalar field ~ = t/>(r}, we have the integral
lIS
taken over the surface S is called the surface integral of A over the surface S. Since A • 0 is the component of A in the direction of the normal to the surface element (see Fig. 2.33). the integral (2.80) is simply the surface integral of this component. The value of the integral is thus a scalar. The quantity A • dS or A • 0 dS is usually callt"d the outflow of vector A through the surface element dS. By outflow of A we me .t the flow of A
ff~ dS
s
A.
or
ff~n dS
8
(2.81)
The result of this integral is a vector.
Il
FII. 2.33 Surface integral.
FII. l.34 Integral over a closed surface.
into the region that contains the normal to dS. This is the case when the component A • 0 is positive. With this interpretation, the surface integra! (2.80). is called the outflow of vector A through the surface S .. Since A· 0 may be positive at some points and negative at other points of the surface S, by outflow of A through S we mean actually the net outflow of A . . For the vector field A(r}, w.e can form another surface integral expressed by
For a vector field A = A(r), we have two integrals .. On.e of'them is
fjA.dS
8
or
fj A.odS
8
the result of which is a scalar. The other integral is
II
8
A(r} x dS
or
II
s
if
s
Ax dS or
if
s
A x
0
dS
(2.83)
A(r) x odS
the result of which is a vector. The three integrals (2.81), (2.82), and (2.83) appear frequently in.the analysis of physical problems. The int~gral (2.82) is called the outflow of A through the surface S. It actually gtves the net outflow of A through S from the region enclosed by S.
The result of such an integral is a vector. Surface integrals of the type described above may be formed also with closed surfaces. As shown in Fig. 2.34, let S bea closed surface and, as before, let dS = 0 dS denote an element of S. For a closed surface, we shall always draw the normal so /IS to point outward from the region enclosed by the surface and refer to it as the outward normal. Using this convention we form the following surface integrals.
2.30 Volume Integrals
Consider a regi~n of spa~e.1I in. the field of a scalar function of position ep(r}. Let the regIOn be dlvlced IOto a number of infinitesimal volume
116
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
117
elements of magnitude dT. Then the integral
III4>(r)
91
To see, the significance of the vcctor U, we ~k at the point P the component of U in some direction e. We have
dT
u· elim 1. Jt4>(D • e) dS
.'''04T Jj' , AS
(2.85)·
taken throughout the volume 91 is known as the volume integral of cP over the region 91. The result of the integral is a sCalar. Similarly, if 91 is a region of space in the field of a vector function of position A(r), the integral
since e is ~ given direction. The right side 'of Eq. (2.85) is evaluated as follows. Sancethe shape of the volume element dT is arbitrary, we set it
IIf
III
A(r)dT
is known as the volume integral of A over the region fJI. The result of such an integral is a vector. In subsequent chapters we will find many examples of the line, surface, and volume integrals introduced in the preceding tJtree sections. There are certain transformation relations that enable us to convert these integrals into one another. These relations follow directly from the integral definitions of gradient, divergence. r.lld curl that will be given in the following section's:, The operations involved in those definitions are the ones that usually arise in the setting up of physical problems. 1.31 IDtegral DefiDitiOD of·the ·GradieDt Consider a point P in a scalar field described by cp = 4>(r). Surround the point by a small closed surface dS. ,Let dT be the volume enclosed by dS (see Fig. 2.35). The shape or the volume element is arbitrary. If dS = D dS is an elemental area on the surface dS (D, according to our convention, is the outward normal), the quantity cpa dS is a vcctor at the element dS of magnitude t/J dSpointing in the direction D. We form the integral
FI8. 2..15 Illustrating the integral definilion of a gradient.
up at ~he point P as a cylinder with axis along e, base du, and length 6.s (see Fig. 2.36). Then fonhe cylinder we hav'!' jfc/>{D. e) dS
==
II
c/>{D • e) dS
+
If
4>(0. e) dS
+
If
4>(D • e) dS (2.86)
divide it by the volume dT and take the limit of the resulting ratio for vanishing dT. This limit, when it exists, represents a certain vector associated with the point P. The vector, as obtained, is derived from the scalar functio.n c/>(rj. Denoting this, vector tentatively by U we write U = Jim.! i=(t/Jo dS ArO dTlf
A8
A8
wall
race 1
ral:e I
Since, at every point of the cylinder wall D is normal to e, n • vanishes on the wall,. and consequentJy
e
(2.84) On the face I,
D
.aU
II
tp(n • e) ds = 0
(2.87)
= :e. We assume that cp is uniform OVf'r the face and is
118
IdealFluid
A~rodynamics
Elements of Vector Algebra and Calculus Using Eqs. (2.91) and (2.85) we have U •e
119
equal to the value of t/> at the point
P, that is, at r. We then have
(2.88)
If
face 1
== lim ..!... d</J AT
.:1......0
t/>(n·
e) dS =  t/>(r) Au
ATds
On facc 2, n == e and assume that t/>(r
t/> is uniform over the racc with the value
t/>(r)
(2.92) . This s~ow~ that ~he vector at !he point P is such that its component any. duc:ctlon e gIves th~ derIvative at P of t/> with respect to distance in that dlrec:tJon. Such a vector has been named previously the gradient of t/> (see S~ctlon 2.24). Ther.efore we identify grad t/> with U and give the followmg integral definition:
10
+ Ase) =
+ dt/> As
ds
(2.89)
l!
e
grad
t/> == lim 1.. t( t/>o d.f:
A, ..
oATlf
AS
.
(2.93)
In closing this section we draw attention to the fact that small like AT (see Eq. 2.91);
If
t1S
t( t/>n dS
. IS
2.32 Divergence of a Vector Field
Consider next a vector field A == A(r). Let P be a point in that field and let tlS be a small .closed .surface surrounding the point and enclosing a volu~e AT (see ~Ig. 2.37). As explained before (see Section 2.29), the quantIty A • 0 dS IS the outflow of A through the elemental area 0 dS of
Fit. 2.36 Volume element to compute the gradient.
where dt/>/ds is the derivati.e at P of t/> with respect to distance in the direction e. In relation (2.89) terms involving higher derivatives are neglected. This and the assumption of uniformity of t/> over the faces are permissible in view of the ensuing limit .Jperation. Using (2.89) we obtain
II
<$(0' e) dS = t/>{r) Au
+ ~~ As Au
(2.90)
face!
By using relations (2.87), (2.88), ~nd (2.90), Eq. (2.86) may ..... ..duced to
AS
lr
!J.S
j:( <$(0' e) dS == d<$ As Au
ds
d<$ ==AT ds
(2.91)
Fig.2.37 Illustrating the integral definition of divergen~e.
since As Au is the volume of the cylinder.
110
I~Fluid·
Aerodynamics
Elements of Vector Algebra and Calculus
121
M and the integral
if
A.• D dS is the net outflow ()f vector A tMough the
. !a •• dS we observe that the limit of this
M ~ surf~ I1S. Forming (1/I1T)lJ'
AS
ratio for I1T  0
is the lIel OlitflOW at P of A per unit volume of the regi".•
enclosing the point. In other words, ~he limit denotes simply the divergence
of thevcctor A from the point considered. Accordingly, sUch a limit is
called. the divergence of the vector field A and denoted by div A. We thus
define div A l!i lim
. • .. 0
..!..,4t .... dS
IlT If
4B
(2.94)
To obtain the .expression for the divergence in any orthogo~, curvi· linear coordinate system, we carry out, clioosingsuitably thcvolume element I1T, the operation shown on the right . . ofEq. (i94). In this way we obtain the following results· (1) in Cartesians %,:y, ~ div A _
aA~ + aA. + iA.
(2) in cylindrical coordmates
r,. 8,
1
a%
a,
a.
(2.95)
Z
a 1 a.4. . iM. divA('~)+ + , ar·· , atJ ih
.
Fig.2.38 Curl of velocity as twice the angular velocity.
(2.96)
an angular velocity w. According to Eq.(2.~), we have
{3) .in sl:'beric:al coordinates r, 8, rp
A(r) = V(r) = w x r
div A 2.33
.!.! (,1..4,.) + ~ ..!{.4. iin 6) + _1_ a.4.
r' a,
r sin 8 a6
. r sin 8'arp
(2.97)
Then the vector B associated with V is given by
B
c.t of· •. v__
~
==
lim .!. Ie V ",0 t17 It'
.1S
X 0
dS
The divergence operation, as seen abQve, derives a scalar from a vector. We can define, .in an ·ana1ogous manner, an operation that Will derive a vector from a veCtor, To 40 this, we consider,as before (see Fig. 2.~7),. point P in a vector .field A(I') and form the Iimlt
= lim.!.
.1,0
~7 It'
.1S
If (w
x r) x
0
dS
(2.99)
(see Fig. 2.38.) Since
(w x r) ~ 0 = (0' w)r  (0· r)w
lim  1 ..ir.. o I1T
vecto~B.
if
4B
A )(;.dS
.
.
(2.98)
Eqllatl0"
~2.99)
may be written as
B = lim
4,0
The result of this limit is a vector~ whiCh we shall denote tentatively by the . TQ see the physical .~igr.ificance of the vector B we choose the vector A i..the limit (2.98) to be the velocity field V(r) of a rigid body rotating with
~7
J..:. Ie [(0' w)r ~ (0' r)w] dS.
J1'
foS
(2.100)
To evaluate the integral in Eq. (2.100) we proceed as follows. At the point
111
P we choose
Ideal· Fluid Aerodynamics
Elements of Vector Algebra and Calculus
113
.aT as a circular cyhnder with its axis parallel to w, its base
!l.q equal to 1J'a' and its length equaJ to !l.s (see Fig. 2.39). We may then '; write Eq. 2.1(0) as
Now, on the wall of the cylinder n is normal to w. Therefore o· w vanishes on the wall of the cylinder and we have
B=
A~i~ ~T{ ",it(o' w)r ~.ff
dS
~yUnder
(0' w)r dS
~l£
If
wall
(0' w)r dS == 0
(2.102a)
(0' w)r dS
On the face I, 0,' w is equal to w, and on the f~ 2"it is equal to w. Therefore we obtain
II
To evaluate the various integrals in Eq. (2.101) we assume (which is permissible in view of the ensuing limit) ,that ' r on face I is uniform and eq~1 to, r at P which we denote by r(P); r on face 2 is uniform and equal to rep) + !l.seGl , the value ofr at the center of that face. Here e", is a unit vector in the direction of w; r on the wall of the cylinder is uniform and equal to r(P) + a.
(D' w)r dS
+
II
(D' w)r dS
lace 1
Ia.ce 2
= wr(P)!l.u
+ w[r(P) + ~se",] ~u
(2.102b)
= w !l.s'!l.u
= W!l.T
since t::.T = t::.s':\u is the volume of the cylinder. Next we observe that
II
",all
n· r dS ==
II ~ . == r~) II
wall
[r(P)
+ a) dS +
•
a dS
II
wall
a :a dS
(2.102c)
Now,
II
waU
wall
a dS is zero. Further, we have
/~,
\.
........
_ _Ii
 
==
a
If d~=
1'rQ 2)
a2f7D.As
,,'&11
= 2(
t::.s
•
= 2t::.aAs = ,)Il.
Therefore Eq, (2,102c) bcc(1f;Jcs
FJc. 1.3f Volume ~IelDeDt to compute the curl ohelcx:ity.
If ,~ .
r dS =
~bT
(2102d)
IdealFluid Aerodynamics
For the integral
Elements of Vector Algebra and Calculus
us
fin . r dS over the face 1 and 2 we have
+
ff
face 1
o· r dS
If
o· r dS == e.. · r(P)ds
==dsdu ==dT
+ e", . [r(P) + e
CD
ds] du
(2.102e)
face I
curl A in any direction e. This task at tbe same time establishes a relation between the curl of a vector field and the circulation of the vector field. Such a relation, as we shall learn, is of great importance in the study of many aspects of fluid motion. 2.34 Component ·of Curl as CircaladOn Consider a point P in a vector field A(r), and let e denote a unit vector at P in some direction. Associated with A'there is a vector curl A at the point P (see Fig. 2.40). We seek the component of A in the direction e.
Using Eqs. (2:102), Eq. (101) may be reduced to
B
or
== lim
A' .... O
dT
..!.. {w dl"  w[2dT
+ dTl}
(2.103)
== 2w B2w
In terms of Eq. (2.99), since V x
0
•
==
0
curiA
x V, we have
(2.104)
A, .... odT
lim..!.. I( 0 x V dS == B == 2w
It' AS
This shows that for the velocity field V(r) of a rigid body rotating with the angular velocity w, the limit in Eq. (2.104) derives a vector, namely B, whic~ is simply equal to twice the angular velocity or the rotation associated with V(r). For this reason. we speak of the derived vector,  B. as the curl or the rotation ofV. Extending this no~ion, the limit in Eq. (2.104) for any vectot field A(r) is called the curl of A (or rotation of A) and denoted by curiA (or rot A). Thus we define curl A == lim..!.. Ie 0 x A dS A, .... odT It'
AS
FIt. 2.40 A vector and ita curl..
Using the definition (2.105), We have
e • curl A  lim ..!.. Ie e • 0 A, .... odT Jj'
AS
)(
A dS
(2.106)
since e is a given vector. Noting that
e •0 x A
(2.105) Eq. (2.106) may be written as e· curl A
== A· e x D
or curl A ==lim..!.. A, .... odT It'
AS
Ie 
A x odS
= lim A, ....
odT
1M
AS
A •e x
D
dS
(2.107)
When we are dealillg with the velocity field of a rigid body, or of a deformable body such as a fluid or a solid, the curl of the velocity field is known as the vortex vector or the vorticity. In the general motion of a deformable body, the curl of the velocity at any point is equal to twice the angular velocity at that point (see Sections 9.I·and 9.2). To derive the expressions for curl A in any orthogonal, ('~rvilinear coordinate system, we choose the volume element dT suitably md carry out the limit spe(..ified in Eq. (2. t05). To express curl A in a ~ iven coordinate system means giving the componentsof curl A in the direl.(ions of the three reference unit vectors. We shall now obtain the component of
To evaluate the integral in Eq. (2.107) we choose the volume element dT as a cylinder with its axis parallel to e, its base equal to du, and its height equal to 6.h (see Fig. 2.41). Then we have
if
I>S
A • e x ntiS =
if
wall
A· e x n dS
+
If
I
A· e
X
n dS
+
ff
A· e
X
n as (2.108)
face
faee"
116
IdealFluid Aerodynamics
Elements of Vector Algebra anJ Calculus value at h = O. Equation (2.109) then becomes
117
Since the normals on the faces 1 and 2 are parallel t~ e, the product X D v~nishes on these faces; conseqtJently the integrals in Eq. (2.108) over the faces also vanish. To evalWl.te the remaining integral over the wall, we introduce h the distance measured along e, and s the distance measured along C., the curve of intersection of the cylinder with the plane
e
~ A • e x D dS 48
Ah fA. e, ds
O.
(2.110)
Using Eq. (2.110) with Eq. (2.107) we obtain
e
e • curl A == lim Ah,( A· e, ds 4,0 AT 1 • 0
Since AT = Ah A(1, this takes the form
e • curl A == lim
. oVo A(1
.!.,(
10. A· e. ds
(2.111)
e
e,
D
F'II.1.41 Volume element for showing the relation between curl and circulation.
normal to e. The direction along the curve C. is assigned according to the rule of righthand rotation about e (see Fig. 2.41). We then have
~ A • e x n dS =
48
IJ
wall
A • ex .. dS
Fig. 2.42 Component of curl as the limit of circulation.
(2.109)
We observe that e x n does not change with h and is equal to a unit vector e, ip the direction of curve C. at the point under consider tinn. We further assume (which is permissible in view of the ensuing limit) that at a given s the vector A is uniform over the height ~h and is equal to the
A . e. ds is the circulation of A around curve C" we see • c, that Eq. (2.111) gives the component of curl A in terms of the circulation of A around a certain curve. Equation (2.11'1) is an alternatil'e definition for curl A. It states that curl A is a vector whose component in any direction e at the point P is equal to .the limit of the circulation per unit area 0/ A around a small curve enclosing P and lying in the plane e (see Fig. 2.42). Recalling that
i
118
IdealFluid A!rodynamics
Elements of Vector Algebra and Calculus element, these equations can be written in the approximate forms: grad c/> ==
119
The direction of integration of A around the curve is according to the rule of righthand rotation about the given direction e. Using Eq. (2.111 )we can obtain in any orthogonal, curvilinear coordinate system the expression for curl A. In Cartesians x, y, z we have j
k
1. ..A: c/>D dS
d'T11'
AS
(2.116)
div A (2.112) curl A
= 1...A: A· 0 dS
d'T11'
AS
(2.117)
curl A ==
a a
have
0
ox oy 0% A" A. A,
~ 1. ..A: A
dT11'
AS
x
R
dS
(2.118)
In cylindrical coordinates r, (J,
% we
er
curl A
re,
where now !:is denotes the surface of the differential element dT. Equivalently (2.116), (2.117), and (2.118) may also be expressed in the forms: e'l (2.113)
d1' grad c/>.==
== 
1
r or
a a .£1
o(J
0% Ar rA, A,
I
0
In spherical coordinates r, (J, qJ we have er curiA 1 ==   
re,
0
iJ(J
r sin (Je~ oqJ
(2.114)
fj c/>o dS dT div A = fj A· dS d'T curl A == fj A x dS
AS
(2.116a) (2.117a) (2.118a)
0
AS
0
0 rl sin 8 or
AS
Ar rA, r sin 8A~
1.35 Some Related Remarks
1. The expressions for grad c/>, div A and curl A in Cartesian, cylindrical, and spherical coordinates show that in terms of the differential operator V we have grad c/> = Vc/> div A == V·A and curl A == V x A
2. Using the operator V, the integral definitions (2.93). (2.94), arid (2.105) for the gradient, divergence, and curl, respectively, may all be grouped in the form
The approximation implied in (2.116), (2.117), and (2.118) may be made as close as we please, !>ince dT may be taken as small as we please. 4. Consider a surface element 0 !:is in a vector field A(r). Let en denote the boundary curve of the surface element, and let ds denote a differential element of the Curl A curve. The direction of ds is that of righthand rotation about n (see Fig. 2.43). According to' the integral definition (2.111) for curl a A, we have
o • curl A == hm
ASo!:iS
J.. ,(
Y A· ds
CA
(2.119)
v(.:)
X
If the surface element n dS is taken equal to n uS, a differential surface element, Eq. (2.119) can be written in the approximate form
n • curl A =
= lim
A,O
A
J_..A:( :. !:iT 11'
AS
)0
dS
(2.115)
A x
dS .Je.
~~
ds
A· ds
(2. L20)
Fig. 2.43 Outflow of curl' thr0ugh a surface element is equal tu the cin;uiJ tion around its boundary.
or in the form
3. If the volume element in Eqs. (2.93), (2.94), and (2.105) defining the gra tient, divergence, and curl is taken equal to dT, a differential 'volume
ndS.curlA=~
A·ds (2.120a)
... en
130
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus grad q, in the region 91. We thus have lim I(grad q,h {not
l: ... co
131
The approximation implied in Eqs. (2.120) may be made as close as we please, since dS may be taken as small as possible. 2.36
Relat~oDS
between Surface and Volume Integrals
The integral definitions of grad q" div A, and curl A give rise to some important relations between surface and volume integrals that occur frequently in the analysis of scalar and vector fields. We'shall now derive these relations. Let us first consider a scalar field q,(r). In this field let 91 denote the volume of a finite region of space enclosed by a closed surface S, Subdivide 91 into a number of small volume elements. Let the volume of element k be denoted by aT" and its surface by 6.S". Then, fcr the clement OT", according to Eq. (2.116a), we have
"1
=
III •
grad q,'dT
(2.124)
With the relations (2.123) and (2.124) Eq. (2.122) becomes
if
q,n dS =
III
R
grad q, dT
(2.125)
if
rP o" dS" = (grad q,)" OTI;
(2.121)
which gi:ves a relation between surface and volume integrals in a scalar field. Equation (2.125) is sometimes called the gradient theorem. . Consider next a vector field A(r) and, as before, let S be a closed surface enclosing a finite region of space, denoted by 91. Using relations (2.117a) and (2.118a), and proceeding on the same lines as above, we arrive at the following relations:
where 0" dS~ is an element of area on the surface 6.S", and (grad <Ph is grad rP taken at an arbitrary point within the element OT". The approximation implied in Eq. (2.121) becomes closer as OT" becomes smaller. We sum the Eq. (2.121) for all elt:ments in the region and proceed to the limit as the number of elements becomes indefinitely large: We thus have (2.122) In carrying out the summation in the left side of Eq. (2.122), we oliserve that on the common surface between any two adjoining elements, the normal for one elemrnt is opposite to that for the other element. ThiS means the sum of the integrals IS 4>0" dS" over [he common surface for the two elements vanishes. In this way, considering all the elements, we will be left for the sum with contributions from only the surface elements that lie on the surface S. This result is independent of the way the region f!l is divided into elements OTk • We therefore obtain
and
if if
s
S
A· n dS =
.III
R
div A d
(2.126)
A x n dS
=
III
R
curl A dT
(2.127)
The relation (2.126) is usually known as the divergence theorem or the theorem of Gauss. It states that the outflow of a veeier field A through a closed surface S is equal to rhe volume integral of the divergence of the vector field over the region enclosed by S. Using the operator V, the integral relations (2.125), (2.126), and (i.127) may be grouped in the form
°
if ( :.
S
)ndS =
A x
III v(.:)'
R
dT
(2.128)
x A
2.37
Theorem of Stokes
=
Jj
s
1>0 dS
(2.123)
The sum ill the right side of Eq, (2.122) is by definition the integral of
Equation (2.120a) gives rise to a relation between a line integral and a surfa,;e integral in a vector field. Let ~ be a closed curve in a vector field A(r) and let 5 be an arbitrary (in general, curved) surface bounded by the curve (see Fig. 2.44). We assign arbitrariiy one or the other side of th~ surface as the positive side and divide that side of the surface into a largl! number of elements oS" by a network of small curves C". At each element the normal Ok is set up from the positive side of S. A direction of travel along any Ck is chosen according to the rule of righthand rotatiull about
131
IdealFluid Aerodynamics
Elements of Vector Algebra and Caklulus
13J
the corresponding normal Ok. This procedure also fixes the direction of travel along the curve~. For the surface element Dt tJSk , according to Eq. (2.120a), we have "
jet A· dSt = (curl A)k' Ok tJSk
(2.129)
where ds k is a differential length along Ck , a.nd (curl A)k is curl A taken at an aOrbitrary point within the element tJSk • The approximation in Eq. (2.129) becomes closer as tJSk becomes smaller.
This is known as Stokes' theorem. It states that the circulQtion of a vector A around a curve ~ is equal to the outflow of curl A through an arbitrary surface S bounded by the curve~. Note that if we consider different surfaces drawn with the same boundary curve, the outftow of curl A js the same through all the surfaces. In terms offtuid ftow, if A represents the velocity field V, curl A becomes the vorticity and Eq. (2.131) states that the circulation around a curve ~ is equal to the out flow of vorticity through an arbitrary surface S bounded by
~o
A simple result tbat follows immediately from Eq. (2.131) is that if we consider a closed surface, the integral taken over the boundary curve vanishes giving
o
if
B
curl A • D dS
=0
(2.132)
1.38 Further Operations
It may be recalled .that gradient. divergence, and curl are operations that involve firstorder partial derivatives with respect to space. Repeated application of gradient, divergence, and curl lead to expressions that involve spatial derivatives of an orde, higher than the first. We shall now look into those cases th8t involve only the secondorder derivatives. Consider first a scalar field r/l{r). Associated w~th ", there is only one firstorder differential expression .. namely grad",. Since grad", is a vector field, we can form from it ihe following two secondorder differential expressions: div (grad",) or V· V", and curl (grad ",) or V)( V ~ Consider n~xta vector field A(r). The firstorder differential expressions associated with A are only two, the div A and the curl A. From these we can form the following secondorder differential expressions:
VI
Fig. 2.44 Illustrating the theorem of Stokes.
We sum (2.129) for all the elements on the surface S and proceed to the limit as their number becomes indefinitely large. We thus have (2.130) It is easily verified that lim
6S..
koo
1, A. dS k 1:1 Yet
i
=
krl
±Yet 1,
A.
USk
=
1, A. ds YW'
grad (div A) or Vrv· A)
div (curl A) or V· V )( A and curl (curl A) 9r V)( (V )( A)
where ds is an elemental length along "C. The right of Eq. (2.130) is, by definition, the integral of curl A • n over the surface S. Therefore Eq. (2.130) becomes
£.
A • ds =
II
S
1.39 Laplace Operator·
In terms of the del operator, for div grad", we can write divgrad ",
curl A . n dS
(2.131)
== V • Vt/> = (V • V)t/>
134 The operator
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
135
v . V == div grad
VI
is a scalar secondorder differential operator. It is known as the Laplace operator or Laplacian and denoted by the symbol VI. We thus have
mathematical physics. Its applications are numerous and may be found in the theory of gravitational fields, fluid mechanics, electrodynamics and optics, and in the theory of differential and integral equatioJls. To derive Green's theorem we start with the divergence theorem, Eq. (2.126):
==
V. V
== div grad
(2.133)
The expression for the Laplacian in general orthogonal, curvilinear coordinates is given in Section 2.4~ In Cartesians x, y, z it has the form
I
IfI
R
div A dT =
fJ
8
A • n dS
V
= ox. + oy. + Oz.
o·
a·
a·
Let 'I' = 'I'(r) and 4> = t/>(r) be two scalar functions of position. Form the vector field
tp
(2.134)
grad tP,
== 'I'V4>
In cylindrical coordinates r, (), z it has the form
and substitute it for the vector field A(r) in the divergence theorem (2.126).
VI=~~(r~) +l~+~
r
or
Or
rl O()I
oz'
(2.135)
We thus have
Iff
R
div(lp grad
4» dT =
~ 11' grad 4>.n dS
8
2.S
(2.138)
In spherical coordinates r, (), rp it has the form VI =
l[~(rlsin ()~) + ~(sin ()~) + 0 sin () ~)J ~(_1 0 r2 sin () or or o() o()
The integrand in the left side of Eq. (2.138) may be expanded (2.136) div (11' grad
4» = 11' div grad 4>
+ grad 11"
grad·4>
The Laplacian being a scalar operator may be applied to a vector field, A(r), and \te can speak of VIA. In obtaining the expression for VIA !O an orthogonal, curvilinear, coordinate system, account must be taken of the fact that not only the components 'of A but also the reference unit vectors are functions of position. In Cartesians, since the unit vectors are constant, the components of V2A are simply the Laplacians of the corresponding components of A (i.e., V2A." etc.). This, however, is not true in the case of a general orthogonal, curvilinear system. To avoid any possible misinterpretation, in general, of the components of VIA as the Laplacians of the ~orresponding components, we express VIA in a form involving grad, dlV, and curl only. Such a form is the vector identity
~ay
With this, Eq. (2.138) takes the form
fff
R
('I' div grad 4>
+ grad 11' • grad 4»
dT =
dT =
if
8
'I' grad 4> • n dS
or, using the Laplacian, in the form
fff
R
('I' yl4>
+ grad 'I' • grad 4»
If
8
(2.139)
'I' gra4 4> • n dS
Introducing o4>/on to denote the derivative of 4> with respect to distance in the direction of the outward normal n, we have grad 4>.• n =
VIA = grad div A  curl (curl A) = V(V· A) V )( (V x A) (2.137) This id~ntity may be readily. verified by expansion. in Cartesians or by expandmg V x (V x A) according to the formula for a vector triple product.
04> on
Equation (2.139) therefore may also be written in the form
Iff
R
('I' V I 4>
+ grad 11' • grad 4»
dT =
if
S
'I'
:~ dS
2.40 Green's Theorem
Having introduced the Laplacian, we shall now derive another integral relation involving it and known as Green's theorem. Green's theorem occupies an important position in mathematics and in various branches of
Equation (2.139) is known as Green's theorem in the first form. Now, consider the vector function
11' grad 4>  4> grad 'I'
136
IdealFluid Aerodynamics Elements of Vector Algebra and Calculus
137
and substitute it for the vector A in the divergence theorem (2.126). We thus have
III
R
div('P grad t/>  t/> grad, 'P)dT ==
if
8
('I' grad t/>  t/> grad ,'1'). n dS
(2.140) The integrand in the I~ft side of Eq. (2.140) may be expanded and shown to be equal to 'I' Vlt/>  t/> VItp Therefore Eq. (2.140) takes the form
identity (2.143) may be demonstrated by working out the differential operation V x Vt/> in any orthogonal, curvilinear coordinate. In particular, the identity is easily verified in Cartesians. Now, suppose that a vector field is such that in certain regions of space its curl vanishes. Then in those regions, on the basis of Eq. (2.143), the vector field may be represented as the gradient of a scalar field. Thus when curl A
== 0
(2.144)
we can write
A=gradt/>' grad t/>  t/> grad '1'). D dS (2.141) ('I' o¢,
JJJ<'P Vlt/>  t/> VI'P) dT ==
R
ff(tp
8 8
or, equivalently, the form
III(tp
R
Vlt/>  t/> Vltp) dT =
if
on
 t/>
0,,) dS on
This equation is known as Green's theorem in the second form .•
2.41 Irrotatlooal Field
Let 0 dS be a differential surface element containing a point P in a scalar fieldtMr). Consider the vector field grad ~andapplytoit, at P, Eq. (2.120a). We thus have
DdS· curl (grad t/» ==
t.
where t/> == t/>(r) is some scalar field. . If A is known, ~ can be determined from Eq. (2.144), which is a firstorder partial differential (vector) equation. When A is not known, an equation for ~ is to be developed from the equations that· govern A. In physical problems, replacing an unknown irrotational vector field by an unknown scalar field reduces the number of scalar unknowns from' three to one and consequently introduces some simplification in the analyses of the problems. If the curl of a vector field vanishes in certain regions of space, we say it is an irrotational field in those regions. The scalar field, the gradient of which represents an irrotational vector field, is usually known as a scalar potential or simply a potential. This name results from the fact that the scalar field representing an irrotational force fiel~ is simply (except perhaps for the sign) the potential energy of the force field.
grad t/> • ds
(2.142)
2.42 Solenoidal Field
Let dr be a differential volume element containing a point P in a vector field V = VCr). Consider the vector field curl V and apply to it, at P, Eq. (2.117a). We thus have
where e" is the boundary curve of the surface element· 0 tiS. Since grad t/>. ds is equal to d.p, the total differential along e", we have
f
since
c.
grad t/> • d.
==
i Yea dt/> =
=0
0
dT div (curl V)
==
if
curl V • n dS
(2.145)
en
4B
is a closed curve. Therefore Eq. (2.142) becomes odS· curl (grad~)
where /l.S is the surface of the vrlume element dT. According to Eq. (2.132), the right side of Eq. (2.14:;) is zero. Therefore Eq. (2.145) becomes
Since this equation is true fo" any arbitrary surface eJement.o ds through P, it follows that curl (grad~)
dT div (curl V)
=0
=0
(2.143)
This states that the curl or rotation of any gradient vector is zero. The
• The 'nomenclature regarding the first or second form is not uniform. The nomenclature used here seems to be preferred in America.
Since this relation is true for any arbitrary volume element dT enclosing P, 'it follows that div (curl V) = 0 (2.146) This states that the divergence of any curl vector is zero. The identity (2.146) may be demonstrated by working out the differential operation V • (V x V) in any orthogonal curvilinear coordinates. In particular, it is easily verified in Cartesians.
138
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
139
Now, suppose that A(r) is a vector field such that in certain regions of space its divergence vanishes. Then in those regions, on 1he basis of Eq. (2.146), A(r) may be represented as tile curl of some other vector field, say B(r). Thus when div A = 0 we can write A = curl B (2.147) In the analysis of physical problems, replacing a divergenceless vector A by curl B leads to certain advantages. A vector field whose divergence is zero is known as a solenoidal field. In analogy with a scalar potential a vector such as B, defined by Eq. (2,147), is known as' a vector potential.
2.44 Poisson's Equation
Let us consider the situation' where the rotation of a vector field A(r) is zero but its divergence is not zero. In fact..,. 'letdiv A be described by a scalar field q(r). We thus have curl A = 0 div A = q(r) We set A = grad </> (2.1S7) thus identically satisfying Eq·. (2.1SS). Using Eqs. (2.1S6) and (2.1S7) we obtain (2.1S8) 'VI</> == div grad </> = q(r) as the equation to determine </>(r). Equation (2.158) is an inho£Pogeneous Laplace equation. Such an equation is called poisson's equation. Consider next a vector field VCr) whose dinergence is zero but whose rotation is not zero. Let curl V be described by a vector l1eld nCr). Thus V(r) is characterized by the equations divV = 0 curl V = nCr) We set V = curl B (2.161) (2.159) (2.160) (2.1SS) (2.1S6)
2.43 Laplace's Equation
Suppose A(r) is a vector field whose divergence and curl are both zero. We then have div A = 0 (2.148) curl A = 0 (2.149) On the basis of Eq. (2.148) we can represent A as the gradient of a scalar function, say </>(r). Alternatively, on the basis· of Eq. (2.149), we may represent A as the curl of a vector function B(r). First let us write A = grad </> (2.1S0) satisfying identically Eq. (2.149). Substituting Eq. (2.1S0) into Eq. (2:148) we have 'VI</> =<'div (grad </» = 0 (2.1SI)
~~tiM.
as the equation to determine </>. Such an equation is called lAplace's • In stead of setting A equal to grad </>, suppose we write A = curl B (2.1S2)
thus satisfying identically Eq. (2.159). Using Eqs. (2.160) and (2.161) we obtain curl (curl B) = n (2.162) On using the vector identity (2.137) and stipulatinl! that div B is equal to iero, Eq. (2.162) becomes (2.163) This is Poisson's equation governing the vector field B(r). Recalliilg that the Laplacian is a secondorder differential expression, the equations of Laplace and Poisson are secondorder partial differential equations. They govern many physical phenomena, for example, steady .heat conduction, electrostatics, magnetostatics, gravitational fields, flow of an ideal fluid. In many technicalIy interesting problems involving these phenomena, the source term [i.e., q(r) or nCr) in Eq. (2.158) or (2.163) in Poisson's equation] vanishes outside certain limited regions of space. In such a case Poisson's equation holds only inside those regions, whereas Laplace's equation holds outside those regions.
satisfying identically Eq. (2.148). Substituting Eq. (2.IS2) into Eq. (2.1.49) we have curl (curl B) = 0 (2.1S3) According to the vector identity (2.137) curl (curl B) = grad div B  'V'B Since div B is unspecified, we stipulate that div B is zero. Then Eq. (2.153) takes the form (2.IS4) which, again, is Laplace's equation.
14(}
IdealFluid AerodynamicS
Elements of Vector Algebra and Calculus
141
The theory of the motion of an ideal fluid, as we shaH see, is identical with the pursuit of constructing solutions to the equations of Laplace and Poisson. l.4S Expressions in Genenl Ortbogonal, Curvilinear Coordinates
Similarly. considering the q2 and q.curves, we obtain (2.166)
and,
(2.167) where Sl and s. are distances measured, respectively, aiong the q. and q3 curves. Equations (2.165), (2.166), and (2.167) enable us to determine e1• and so forth and 6s1 /6q1. etc., once we know what or/oq1' etc .• are. The derivatives ar/oql' and so forth, are obtained as follows in ten~s of the transformation relations (2.164) and the i, j, k unit vectors. Introduce the notation
We now complete our study of vector analysis with the expressions in orthogonal, curvilinea'r coordinates for the directed distance between two neighboring points, the differential operatorsgrad, div, curl, and Laplacianand the changes in the reference unit vectors. Let q1' q2' qa be the general curvilinear coordinates ofa point P in space. They are defined by the transformation
ql == Ql(X, y, %)
q2 == q2(X, y. %)
q~
== qs(x, y, %)
or, inversely, by
x = X(qh q2' q8)
(2.168) (2.164) and
h _ 6s, .==6q,
y =' Y(qhQ2' q8)
%
= %(q1' q2' q8)
where I, Y, % are, as usual, the Cartesian coordinates of P. Let el> e 2 • es denote. as always, the system of reference \Jnit vectors, at P, corresponding to the coordinates qh q2. q3 (see Section 2.16). We first dc;termine the unitt vectors from the transformativn relations (2.164). It should be noted that until the unit vectors are determined it is not possible to say whether or not they (or equivalently the chosen curvilinear coordinates) are orthogonal.
Equations (2.165), (2.166) and (2.16 7 ) may be grouped in the convenient form Or II e = (2.169) '" '" oq.". where the subscript m may be 1, 2, or 3. Expressing the position vector as
r == X(qh q2' q.)i
Unit Vectors. Consider the q1curve th;ough the point P(see Section 2.16). Let r denote the position vector to P from a fixed origin and SI the distance along the q1curve (being positive when measured in the direction ofincreasingq1)' Then. according to Eq. (2.17) and the definition of e1, we have
+ Y(q1' q2,q,)j + %(qr, q2' q.)k
oqm oqm oqm
(2.170) (2.171)
we obtain
ox. oY. h..e mar =  =  I +  J +0% k =
oq",
where, again, the subscript m may be I, 2, or 3. The relations (2.171) determine completely the hm's and the em's. In fact, from (2.171) we have
h= '" oqm
and (2.165)
1 ( ax )1 + (oy )1 + (Oz )1 oq",
~q",
(2.172)
1 oy. 0%) e m =  ' i+J+k h m oq", oq", oqm
(ax
(2.173)
142
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
143
Equations (2.172) and (2.173) hold whether or not the curvilinear ~ ordinates, defined by the relations (2.164), are orthogonal. Havmg determined the em's, if we find .that the products
From Eq. (2.175), for the square of the infinitesimal distance ds, We obtain (dS)' E cis· ds
I:
vanish at all points, then the unit vectors or, equivalently, the curvilinear coordinates are orthogonal.
",1,.1
I I
I
•
(h ..e", • h",e,.) dq", dq,.
(2.177)
Il1ji,,;tesimal Distanee betwee" Two Neig"b"ring Poi"ts. neighboring point of P, and let
Let Q be a
This equation is true for any curVilinear ooordinate system, orthogonal or not. For an orthogonal system we have the result·
e", • eft  0
ql
+ dql' q. + dql, q. + dq.
x
when
m '" n
be the curvilinear coordinates of Q while
 1 whenm  n Therefore, for an orthogonal, curvilinear, coordinate system, Eq. (2.177) becomes (ds)' = (hi dtfJI + (hi dqJ' + (h. dqJ'
+ dx,y + dy,z + dz are its Cartesian coordinates. We denote by ds = dse. the directed distance PQ. If dS ds 2 , dS a are differential arc lengths along the qlt qt, qa curves,
lt
== (dsJ' + (dSJI + (dsJ'
(2.178)
respectively, we have
ds = e1 dS l
+ e 2 dS 2 + e3 dS a
(2.174)
This relation does not assume tHat the ooordinates are orthogonal. The arc lengths ds 1 ,_etc., are determined as follo~s~ .. . As always ds is identical with dr, where r IS the POSition vector of pomt
P. Now, since
we obtain ds
==
or, dr =  Oql Oql
Or Or + ; dq, +; dq. (lq. (lq.
dq.
DifferelltW J10lllIM tuul SIII'/flCe EklMllts. We concern ourselv~ hereafter with an orthogonal, curvilinear, coordinate system only. Consieiw. two points P and Q such that the coordinates of Pare ql' ql, q. and those ef Q are ql + dtflt q. + dq! and q. + dtf.. The coordinate surfaces pasSing through P and Q deacri~ a volume element that We shall denote by d.,.. The sides of the element are curvilinear, and, in general, the element is not a parallelepiped (see Fig. 2.45). For illustration, the volume elements in cylindrical and spherical coordinate~ are shown in Fig. 2.45. To the first order the volume of the eleme,nt in curvilinear, orthogonal coordinates is given by dT  ds1 ds , ds.
= e1 hl dql + e.h l
+ eah. dq.
(2.175)
where the h's and e's are given by Eqs. (2.172) and (2.113). From Eqs. (2.174) and (2.175) it follows that
= h1 h",. tiql dtfl dq. where the scale factors hi, hi, h. refer to the point qb ql' ql'
(2.179)
dS I dS 2
= hI dql

The area of the surface element forming theql face (i.e., the face normal to ql coordinate) of the volume element d.,. is given by (2.180) Similarly, the areas of the ql and respectively, by and
ds.
9
h. dq: h8 dq3
(2.176)
q. faces of the element d.,. are given,
(2.181)
giving the differential distances along the coordinate curves in terms of the coordinates. From Eq. (2.176) we see that hi, h" h3 are .of the ~ature of scale factors which relat~ the differential distances to the dIfferentials oft~e coordinates. These scale factors vary from point to point and thus are, In general, functions of position.
(2.182)
• For a JtOJtOrtMgo1lll1 SYSlClJl e,. .
e,." 0 when m "
11.
144
IdealFluid Aerodynamics
Elements of Vector Algebra and Calculus
us
The del operator is given by
1 i 1 0 1 a VeeI·+e. +e,h~ 41q.
hI 41ql
h.41q.
The expression for grad r/> may be derived also on the basis of the integral definition (2.93). Thi~ is left as an exercise.
(112
+ dqz) face
Divergenee. The expression for the divergence of a vector field A(qt> q•• q,) is conveniently obtained by using the integral definition (2.94).
As may be verified we obtain
(a)
z
C",I. The expression for the curl of a vector. field A(ql, ql. q.) is conveniently obtained by constructing its components in the direCtions of the reference unit vectors el • e•• e. on the basis of the definition (2.111). As may be verified. we obtain a result that can be exhibited as
hiel
Curl£*
hie. Q aq.
}riA.
hae.
1 0 hlh.h. 41ql hiAl
a
41q. h.A.
(2.1J4~
(r + dr)dI
r sin 'dIP
z
Laplacian. The expression for the Laplacian of a scalar function (f position r/>(ql' q., q.) is readily obtained by substituting grad r/> for A ar d the corresponding compon~nt~ of grad r/> for At> AI. and AI in the relatk'l (2.183). We thus obtain
V·r/> =
di~ grad
r/> =
_._l_[~(hlh. or/»
hlh.ha 41ql hI 41ql
(2.185)
(II)
(c) (a)
+ l..(hahlO.r/» + .E..(hlh.Or/»]
oq. h. oq. oa. ha oqs
It follows that the Laplacian opera4lr is given by V·
FIe. 1.45 Volume element in curvilinear coordinates:
cylindrical coordinates; (c) spherical coordinates.
general coordinates; (b)
Gradient. The expression for the gradi,ent?f a ~caHu function +(qh q •• q.) is conveniently obtained as deSCrIbed m SectIOn (2.24). We
==
div grad
have
 hlhzh s OqI
__l_[.i.(hzha.i.) + .i.(hshl.i.) + 1... (hlh2 .!_)l
hI Oql
oqz
hz Oq2
OQ3
113 iJqa
J
Clulnges in the Referenee LJnit Vectors. The relation (2.173) t. nr bles us to determine the change in any of the unit vectors Cl • e 2 , e 3 n::~ult:'\g \fom
1.
'1. ' •. "
ldalFluid Aerodynamics
Elements of Vector Algebra and Ca1culus
141
• change in the coordinates qlt q•• q•. Such a determination. however. leaves the result in terms of the i. j, k systems of unit vectors, but we are usually iriterested in having the result in terms· of the coordiaates qh q•• q. and the corresponding unit vectors. Such a result is readily obtained by «he following method. Denote by e l' •••' •••' the unit vectors at the point ql + &il' q. + lJq .. and q. + &i•• where &il' &i •• 6q. are infinitesimal changes (for clarity we use temporarily the symbol 6 instead of 4). Denote by 6. the change in position corresponding to the coordinate changes &ilt &il' &i•. The unit vectors e,.' •••' •••' may be regarded as resulting froin a translation of the over the directed distance 6s and a rotation of them vectors through an infinitesimal.angle ~ about Ii certain axis passing through their common origin .. ~ote that we are concerned here with orthogonal curvilinear coordinates only. Let ~ represent vectorially the infinite~imal angular rotation: Once ~ is known. the changes in the unit vectors are given (since translation causes no change)
and vectors are all functions of posit:on.
V(f1tI;) III: ",V", + ",VV' V • (",A)  ",V • A + V",. A V x (",A) ." ",V )( A + V", x A V(A • B) .. (A • ·V)B + (B. V)A + A x (V x B) + B )( (V )( A) V • (A )( B)  B • V )( A  A ·V x B . V x (A x B) = A(V • B) + (B. V)A  B(V • A)  (A • V)B V x (V x A)  V(V. A)  V'A
'1. '., '.
~  e,.'  e,.  ~ )( e,.
".  e.'  e.  ~ x e. 6..  e.'  '" ." ~ X ••
(2.187)
Now, we know that the angular displacement"" and the associated displacement h are related. Iii faci, we have (2.188) To express ." in tenns of the COOrdinatesqb q.,.f. and the unit vectors
e,., el, e. we write

h .. 6s1e,.
+ 6s~. +. 6s... hi 6qle,. + h.6q ... + he &i ...
hie,. hae.
Using this and (2.184). we express relation (2.188) as
h.e.
~~
1
2h t h.h.
aql
hi· 6ql
a
aq.
a
aq.
6q.
a
(2.189)
h.' 6q. hal
1046 Some V.1uI R.latloas 'Jlte following formulas are of frequent use .in applications. They can be verified either by expansion in cartesian coordinates or by treating ·v as If vector (while retaining its actual meaning as an operator) in the appropriate vector formulas for products. In the following the scalars
Stre,ss in a Fluid
149
Chapter 3
Stress in a Fluid
Theory of the motion of a fluid, like the theory of the motion of a system of masses, is based on Newton's laws of mechanics. Similarly, the study of the state 'of rest of a fluid is based on the laws of static equilibrium as applied to a mechanical system of masses. In adapting these laws to a fluid we usually choose as our syste~ a certain finite or elemental region of the fluid. Thus for a fluid in motion we require, according to Newton's second law, that the rate of change of momentum of the fluid contained within ·any chosen region he equal to the resultant of all the forces acting on it. If the fluid is at rest, the momentum is zero and we require that the resultant of all the forces acting on any portion of the fluid be zero. If the fluid is in a state of uniform motion (i.e., all fluid elements have the same velocity for all timeli), the rate of change of its momentum is zero. Therefore the laws of static equilibrium also apply to a'state of uniform motion. To the law of change of momentum we add the companion law of moments that the sum of all the moments of the forces acting on any part of a fluid at rest is zero. The moments are all taken with respect to a single reference point. As a preliminary step in the formulation of these laws, we shall consider in this chapter the nature of the fort:es that act on any region of a fluid and the method of their specification. Thili involves the concept of stress in a fluid. FollOWing fnese considerations, we shall develop the law of static equilibrium for a fluid. To formulate completely the law of motion for a fluid we must first decide about a method for describing fluid motion. This we do in Chapter 4. Finally,~n Chapter 5, we shall formulate the law of motion along with other equatibns necess!!~y for the analysis of fluid motion.
forces are in the nature of actions and reactions. Such forces are commonly referred to as internal forces. Since they act across a surface that is imagined to ~par~te the fluid, they are also called surface forces. The co~plete specification of these forces is based on the concept of stress, which we shall take up presently (see Section J.2). In addition to the surface forces the fluid may, in general, be SUbjected to forces that act throughout the body of the fluid as such. The simplest ex~mple of such a force is the force due to gravity, that is, the socalled weight ~f the fluid. Forces o!" this type are called body forces. They are proportIOnal to the volume or mass of the fluid considered. Thus the body for~es can be specified as so much per unit volume or per unit mass of the fl~ld, that ~s, as an inte~sity. In general, the body force may vary from pomt to pomt of the fluid, and at any point it may have different value~ at differ~~t instances of time. Thus the body force is a vector function of position and time. .
3.2 Concept of Stress and the Spedficatioa of Stress at • Point
The concept of stress and the method of Its specification, as described below, are applicable to any continuous deformable medium whether it be a fluid or a solid. ~ . ~t us con~ider a plane surface S drawn through some point F in a flUid. The flUid may be either in motion or at rest. We denote by n the nor~al to the su.rface .and. consider t~e forces exerted across S by the portion of the flUid whIch hes on the Side of n. These internal forces are not, in. general, distributed uniformly across S. We represent these fo~Ces as equivalent to a force F acting at the point P and a moment M about some axis t.hrough P (Fig. 3. Ca). If we gradually shrink the area of the surface to the point P, both F and M will tend to zero. However for a vani~hingly small area the equivalent lorce .,' may be assumed to ~ pro, portlo,nal to t~e a,:a: Then the ratio of force to area may be assumed to tend t~ a defimte hmlt as the area shrinks to a point. It can be shown that the ratIO ofth~ moment t.o the area vanishes as the area goes to zero. This means the action of th~ Internal forces in the immediate vici.nity of point p. ~cross the surface n (I.e., .Mrmal to n) can be specified by the limit of the" ratio of force to area, that IS, by a force per unit area at that point. This is called the stress at the point P across a plane rl. It is a vector and its direction is, in general, different from that of n. Denoting such' a stress rector by O'n we define symbolically
0'"
3.1 Surface Forces and Body Forces
In the space occupied by a fluid, whlch is in motion or at rest, let us imagine a sbrface enclosing some part of the fluid. The portions of the fluid close to the surface on its two sides exert forces on each other. These
148
. F = I1m s.~o Sn
(3.n
• In this connection see Love (1944).
IJO
IdealFluid Aerodynamics
St~ss
in a Fluid
where Sit is a plane area normal to' D. The definition of a stress vector involves two directioMthat of the normal to the surface and that of the stress itself. Now if dS is an elemental area normal to D, t~ force exerted across dS by the ftuid that is on the side of D is (Fig. 3.lb) .
G"ds
lSI
The stress at a point P across an elemental surface D can be specified either by a stress vector G" or, equivalently, by its three components
(II)
~
""dS dS
(a)
•
t1"
Fie. 3.2 . Decomposition ora stress vector.
(b)
FIt. 3.1
Representation of internal forces in a ftuid.
forc: s at a pO\~t P across only a certain plane D. For the complete s cificatlOn of the ,mtcrnal forces at P we should know the stress vector': P across al~ the planes, infini.te i?number, that can be drawn through P. It IS easily seen, by consldenng the equilibrium of an infinitesimal ~etrahedron_su~h as shown in Fig. 3.3, that if the stress I:ector across three
tension ~r a ~ile stress. The components 0'", and 0'". are known as the tangential st~~sses at P on the D surfate. They are shearing stresses. The defirlltJ~ri (3.1) of a stress vector enables us to specify the internal
referred to a system of three unit vectors. Denoting the unit vectors by elo e., ea and the respective components of a" by.a nlo a ... , a"a, we write
The meaning of the double subscript is apparent. The first subscript denotes the plane across which the stress component acts, wher::as the second subscript denotes the direction of the component. The decomposi~ion of a stress vector is usually done with respect to an orthogonal system of unit vectors. Generally the unit vectors selected are those associated with the (orthogonal) coordinate system employed (Fig. 3:2a). Another way of choosing the unit vectors is to select one of them in the direction of the normal D and the other two in two mutually perpendicular directions lying'in the surface D. Denoting the latter directions L¥ e, ~nd e. we write (3.3) 'fhe comoonent 0' nn is known as the normal stress at P on the D surface • (FiS' 3.2b). Its positive direction is that of n. It is then known as a
pomt IS "omplet~ly specified by glvmg a set of three stress vectors. Equivalently we can give the components of these stress vectors. Thus if a C'm, (1 n Jenote the stress vectors at P across the planes e e e d ,If' " m' n' an I
Independent planes passing through a point 'is given, the stress vector across any other plane p~ssing through that p'1int is determined. It thus follows that . th~ stat: of the mterna~ forces, ~l~o known as the '3late of stress, at any
Fig. 3.3 Equilibrium of a tetrahedron.
Stress in a Fluid
152
151
IdealFluid Aerodynamics
eb et, e. are a set of unit vectors at P, the stress components can be given
in the form
(0'11 0'"
0'...
O"S)
0'",3
O'"d
0',,1
(3.4)
For the complete specification of the state of stress in the region of interest of a ftuid we must give at each point the stress tensor. The tensor will generally be different at different points, and at any point it may vary from instant to instant. This means thelt the state of stress within a ftuid is specified by a tensor function of position and time, that is, by a tensor field. 3.3 Stress in • Fluid at Rest: Hydrostatic: Pressure The stress tensor for a ftuid takes on a particularly simple form when the ftuid is at rest. It is a fact of experiellce that tangential stresses do not exist in a fluid at rest. This means the stress vector at any point of the fluid at rest is wholly normal to any surface element passing through that point (see Eq. 3.3). Symbolically we write that (3.7) In such a case the stress tensor takes the form
0'",
O'"s
For purposes of calculation it is convenient to take the unit vectors el> et , e3 as those defined by the (orthogonal) coordinate system used and the
(3.8).
FIa 3A
State of stress at a point.
o
ea·
With For such a state of stress we can draw a fundamental cl~nclusion. Considering the equilibrium of an infinitesimal tetrahedron we can show that when tlTT! stress vector at a point is wholly normal in all directions, its magnitude is lhe same for all elemental planes passing through the point. • In such a case all that is required to specify the ~tress at a pomt is simply a single number. Denoting this number b) (J we write
(3.9)
I planes e" em, eft as the corresponding coordinate panes eb e., this notation the stress components (3.4) become
(::: ::: ::)
0'11
(3.5)
O"SI
O'ss
This array of nine numbers is known as a stress tensor: It specifi~ completely the state of stress or simply the stress ~t a pomt. !he diagonal terms an, etc., represent the normal stresses, whI~e the nondiagonal terms 0"11' ·etc., represent tangential or shear stres~s (~lg ..3.4). The equilibrium of the moments on an mfimteslmal cube shows that
0'11 '123 0'11
for all directions of n. If c: is positive, an represents a tensile ~l ess. But from experi~nce we find t,hal' no tensile stresses occur in the inrerio.· :)f a fluid. t This nlean:; i~ would be morl! appropriate to rcpre~ent e~ ;,$ a compression. This we do by replacing the number rJ by a neg2!ive number, say po Then IhF slrt!ss ill a fluid at rest is represented by
a"
= 0'81
0'13
(3.6)
=
pn
0"31 
That is, tbe stress tensor is symmetrical about its diagonal. This ~eans to specifY'completely the state of stress at a point we actually need SIX stress components.
... This conclusion 15 known as Pascats fa..... rn this Cilse the str,!ss vector des.::ribes a sphere. t Tensile ~tresses are exhib!ted at the socalled "free surfaces" and in "thin flu:d film,,"
154
IdealFluid Aerodynamics
Stress in a Fluid
Jjj
for all directions ofn. The corresponding stress tensor is given by
(7 ~p ~) o
0
p
(3.11)
The scalar p is called tho pressure or, more preciseiy, the hydrostatic· pressure at the point considered. Equation (3.10) or (3.11) can be taken as the definition of a ftuid. Note that pis a positive number. The state of s~ress within the whole region of a ftuic;l at rest is specified by giving the prdssure as a function of position and time, that is, by a scalar field,.denoted by F == p(r, t). The hydrostatic pressure is generally related to the density and temperature orthe ftuid. For instance, we know that in gases that obey Charles' and Boyle's laws the pressure, the density p, and the temperature T are connected by the relation p == p]l..T, where. R is a constant for the gas conSidered.
they disappear when the rates of strain disappear, thus leaving the stress tenso~ as that of.a ~niform pressure in all directions. Similarly, if the velOCity of the ftuld IS zero everywhere, the viscous stresses become zero and we realise again uniform pressure at a point. Because of these inter~ pr~tations, the splitting of the stress tensor in a mOVing ftuid into ~ umform pressure p and the viscous stresses. is convenient. A~rdi~g ~o these ideas, the state of stress within the whole region of a movmg ftuld IS. expressed·by the combination of a scalar field and a tensor field. In developi~g the. theo~es of ~uid motion we assume generally that t~e~odynamlc co.nslderatlor.s, which are strictly applicable to equilibrium SituatIons, are valId for moving ftuids and uSe them. In such a case we must talk about a thermodynamic pressure that is related to the density and temperature at a point in the ftuid. It is then assumed that the thermodynamic pressure is the same as the pressure p occurring in the stress tensor. The justification for these assumptions is that the theories based on them seem to give results that are in good agreement with experiments..
3.4 Stress in a Fluid in MotionWhen a ftuid is in motion the phenomenon of viscosity or of inte~l friction manifests it:;e)f, and at any point; in tbe fluid both tangential and normal.stresses occur. In this case we deal with the complete stress tensor and express itas made up of two paJ1sOne that represents viscous or fricti01lll1 stresses only, and the other thaJ represents compressive stresses equal ioall directions, that is, a pressure and thus not related to friction. This pressure, also denoted by p, is sin:tilar but not identical to the hydlo~ static; pressure. The viscous stresses, which occur both as la,ngential and normal compenents, a~ usually denoted by Tlb r1l, etc. Thus the state of stress, at any point in a moving ftuidis g\ven in the form
3.5 State of Stress in a NoaUscous Fluid
ia
Modoll
. A large part of the theory .offtuid motion, as pointed out in Chapter I. IS developed on the assumptIon that the fluid is frictionless or nonviscous.In such a case the coefficients of viscosity, and conSlequently the viscous stresses, are set t~ zero ~nd. one ~akes the state of stress as that of a uniform. pressure. The u.ltlmate justIficatIOn for such an assuD"ption lies again in the comparISon of ItS consequences with experiments. The study of 'fluid motion treated in ihis book, as previously stated, it based on the assumption of a nonviscous fluid,
3.6 Pressure, Distribution in a Fluid at Rest
Forthe static e~uilibrium of a fluid, the sum of all the forces acting on an~ part o~ the ~UI~ should. be zero. To apply this law we consider at any
(7 ~p ~) + (:: ::: ::) o
0
p
'ral ral
TU,
(3.12)
pOlDt r an mfimteslmal regIOn of the fluid and write, that the resultant of the boc!y forces a~ting on the region+the resultant of the surface forces acting on it == 0
The viscous stresses are assumed to be proportional to the Nites of straint occurring at the point considered. The proportioJ;lality constants, known ~s viscosity coefficients, depend on the nature of the fluid. For any ,given fluid the viscous stresses are small when the rates of :>tr ain are small, and
• "HydrostatiC" signifies a ftuid at
rest~
(3.13) Let .Or denote the volume of the region considerc;.d and bS the su rlace t'. . enc IoSlng It. If r = f(r) represents the distribution of the body forces . f h fl 'd . per umt mass 0 t e UI ,the body force acting on the element OT is equal to
pC b1' (3.14)
t The rates of strain at a point arc in turn given by certain combinations of the partial
derivatives of the components of the velocity. at that point (sec Section 9_1).
where p = per) IS the density of the fluid. The density may vary fror') point to point of the fluid.
{56
Ideal· fluid Aerodynamics
Stress in a Fluid
157
T~ determine the resultant of the surface forces let us first consider dn infinitesimal area a dS on the surfa~ lJS (Fig. 3.5). Since the state of stress in a fluid at rest is given by a pressure p = p(r), the surface force acting on the·fluiJ within lJS across the surface element a dS is
Equivalently we may wrik this equation as gradp = pC
(3.17)
podS
The resultant of the surface forces is then obtained simply by adding vectorially the pressure forces acting on all the ele.mental areas of the surface lJS. It is· thus equal to
fi
18.
This equation is thus the analytical form of the condition for static eqUIlibrium of a fluid. Therefore it is the basis for amllysing the statics of incompressible and compressible fluids. The content of Eq. (3.17) is significant. It states that static equilibrium of a fluid is possible only if the body force (pf) per unit L'olume can be expressed as the gradient of a scalar function. This implies that pf should be irrotational. If the density of the fluid is uniform throughout space, we have grad (
po dS
(3.15)
,p
l!.) = f
Equation (3.17) assures us that the resultant of the forces acting on any part of a fluid at rest is zero. For the fluid to be at rest the condition that the resultant of the. moments acting on a fluid element should also be zero. That this condi.tion is automatically satisfied may be verified by the reader. 3. 7 Concluding Remuks
To pass on from the s_tatics to the dynamics of a fluid, we may consider a certain portion of the fluid arid apply to it the second law of Newton. Thus we write that the rate of change of momentum of considered = the resultant of the body region + the resultant of the pressure surface + the resultant of the viscous surface: the region of fluid forces acting on the forces acting at its forces acting at its
Fig. 3.5 Illustrating the calculation
o~
resultant pressure.
(l.W)
for an infinitesimal volume element lJ.,. we have, according to Eq. (2.1 16a),
fj
68
po
dS = !5.,. grad p
This shows that the resultant of the pressure forces acting on the surface of an infinitesimalftuid element <S.,. is given by
~(hgradp'
(3.16)
With the expressions (3.14) and (3.16), the condition (3. d) becomes
For the complete analytical formulation of this law we must first d~k e about a method fo:, describing the motion of the flllid, whether we wish talk about the fate of each individual element of the fluid or about the whole fluid as such. Thi~ we shall do in the next chapter. Fquation (3.18) involves the density, the pressure, and thevdodty as unknowns if we a5sume that the body forces :lie given and that the viscous stres~s are related to the dc:ri"atives of the velocity. It IS 'thus apparent that additiOlial equations have to be formulated. This we do in Chapter 5.
to
pf d.,.  b.,. grad p = 0
or, expressing it
per unit vqlume,
p( 
gradp
=0
Description of Fluid Motion
159
Chapter 4
Description of Fluid Motion
element with respect to any chosen coordinate system. Since the fluid elements are continuously distributed, the values that the parameters a, b, c will assume for the various elements are continuous. Now, the motion of the fluid.can be described by giving as a function of time, t, any quantity Q that is associated with each element a, b, c. Considering the whole fluid, Q becomes a function of time and the particle parameters a, b, c. Thus we write . Q == Q(a, b, c, t) (4.1) For instance, the position r of the various elements at any time t is expressed by the functional form
We shall now begin to set up analytically t~e problem of fluid mo. tion. The fluid, as remarked. before, is regarded as matter distributed continuously. At various points of the distribution, we may choose infinitesimally small regions of fluid and regard them as possessing indiViduality. We may thus refer to them as fluid elements or fluid particles To describe fluid motion, two methods are possible. One possibility is to descnbe the fate of each individual fluid particle, that is, to adopt a particle,point of view. The other possibility is to forget about the i~divid \lality of the various fluid elements And concern ourselves only With the state of motion in the space filled by the fluid, that is, to adopt a socalled field point of view. The particle description is known as the Lagrangian method (after Lagrange, 17361813), while the field description is known as the Eulerian method (after Euler, 17071783).. In this chapteT we shall discuss these two methods and also consider additional concepts related lO the description of fluid motion.
r
== r(a, b, c, t)
(4.2)
If X', y, z are the Cartesian components of r, the scalar form of (4.2) is shown by x == x(a, b, c, t)
y
==
I
yea, b, c, t)
Z ==.~{a, b, c, t)
4.1 Lagrangian Method
In this method we ask about what happens to the individual fluid eleinents in the course of time, what are their paths, their velocities, densities, or any other properties associated with them. Thus in this case the fluid motion is described by describing the fate of each individual fluid p~rticle. Such a procedure follows the practice usually adopted in the mechanics of mass particles where one seeks the trajectories of thQ individual particles. Since in this method we are concerned with each particle, we must, so to speak, label the various elements (i.e., name them) in ord::r tn distinguish between them. We do this by assigning, at some instant of t'f'1e, three scalar parameters, say a, b, c, !o each element. A convcni::nt ~hoice or' these parameters, for instance, is to give the position coordin, tes of an
• Historically.
equdti0nS
Thus in Lagrangian descriptiOl the particle parameters a, b, c, and time t are the independent varia!>ks. The unknown variables art: the position coordinates (or, equivalently, the velocity or acceleration ~omponents) of an element and other quantities such as the density, giving its state. Tq determine the unknowns we set up a necessary number of equations between them by applying to each fluid particle natural laws, such as Newton's second law of motion, and conservation of mass. We shall not enter into the derivation of these equations, which are usually known' as the Lagrangian equations of flUid motion. For these equations reference may be made to Lamb (1932), or Sommerfeld (1950). Alth<YUgh the Lagrangian description appears W be a natural way to set up problems of fluid motion, generally it is not as convenient and meaningful as the Eulerian description, which we shall follow exclusively_in our studies here. The Lagrangian method gives more information than one needs. for often one is not interested in the fate of each fluid particle. There are,. however, specific instances, such as certain onedimensional (involving one space coordinate) problems, where the Lagrangian point of view is fruitful.
4.2
Eulerian Method
of fluid motio:1 'J3,d
,1:;
Lalrangiarl :n,:tr.0d
01.CIIf In
Euler's
pers.
In this method we focus our attention on the various points of the space filled by the flowing fluid and what is happening at each of these points as time goes on. What is happening is. of course, to be given in terms of quantIties such as veiocity.density, and pressure, which are of interest in the· rnction of the flUid. Thus, in the Eulerian point of view, we give the
158
160
IdealFluid Aerodynamics
Description of Fluid Motion
161
values that a fluid quantity Q assumes at various points of the space at different instances of time. In other words, Q is described as a function of position and time. If r denotes the position of a space point with respect to a chosen coordinate system and t denotes the time, we write
With (4.6), Eq. (4.4) may be expressed in terms of the Eulerian variables :::, y, z, t. We write
Q = Q[a = gl(X, y,
Z,
t), b = g2(X, y, z, t), c = ga(x, y,z, t), t]
(4.7)
'>0
Q = Q(r, t)
(4.3)
The transition from the Eulerian to the Lagrangian description is not simple. In terms of Eulerian variables we have
Recalling the nl,tion of a field we observe that in l:ulerian description fluid motion is spl!ciflCd by various scalar and vector fields. Thus we talk about wloeit)' field, acceleration ficld, density field, and so on. In sllch a description. the identity of the fluid particle:s that occupy the various points in space at various times is irrelevant. At any imtant eaeh point of the regiC'n may be associated with a fluid element for which the velocity, density, etc., may be taken as those occurring at the point at the instant considered. In the Eulerian description, lhe independent variables are the position coordinates (I.e .. componeRts of r) of a point in space and time t. The dependent variables are the fields of velocity, density, etc. To solve for these fields we set up a necessary system of equations by again using laws such as Newton's second law of motion and conservation of mass. The derivation of these Eulerian equations will occupy us in the next chapter.
Q = Q(x, y, z, t)
(4.8)
pass to the Lagrangian description we must ex?ress x, y, Z in terms of the particle parameters a, b, c. This we do as follows. Let the velocity components at the pOint x, y, z at time t be given (in tetpts of Eulerian description) by u = F 1(x, y, z, t)
to
v = F1(x, y, z, t)
w ~=
(4.9)
Fa(x, y, z, t)
In terms Qf the Lagrangian d .i\:ription we shall have
4.3 Connection between the I.agrangiao and Eulerian Descriptions
To relate these two descriptions we need to know the identity of the fluid element that occupies a certain position in space at a given time. In othei words. we need the relations between points in space': and the particle parameters. We consider first the transilion from the l.agrangian to the Eulerian description. Let some quantity Q be given in terms of the Lagrangian "?riah\cs a, b, c, t. We then have
where x, y, Z (4.10) both describe the velocity components of a fluid dement. Combining them we obtain  = F 1(x, y, z, t) at
at are functions of the variables a, b, c, t. Equations (4.9) ar.d
u
b =  ,
at
v
ay = ,
at
w=  ,
~
(4.10)
ox
at
oy  = FI(x, y, z, t)
(4.11)
az  = F,(x, y, z, t)
at
These are (first order) differential equations for the position coordinates of an element as described by the Lagrangian method. Integration of these equations leads to solutions of the form
Q = Q(o, b, c, t)
(4.4)
To pass to the Eulerian description we must express 0, b, c in terms of the coorJinates, say x, y, z, of a point in space. In the Lagrangian method, they are given by relations cf the form
:r: = /1(0,
x = /1(X O' Yo. lo, t)
1)
= /2(X O, Yo, zo, t)
(4.12)
b, c, t)
h. r, 1)
,'I,
if = fAa,
The~.:
;: =
(4.5)
z. and t. We shall
Yf
Ilx o, Yo. Zo, t)
: = 13(0. h, c. I)
may be then have
sl~lved
to obtain a. b, c in terms of r.
"'here x I';, " , 0' the constants ot integration, are chos~n as the initial valul!s :;11)' of r, y, 7 at an initial instant t = to. We may set the particle parameters a, b, c' equal to Xo. Yo, z~ re:;pectively and rewrite Eq. (4.12) as
x = /l(a, b. c, t)
( l.6)
L'
!I
= 12(a, h, c, t)
(4.13)
=
g .•i.r. !I.
~. t)
z = /3(a, b, c, t)
161
IdealFluid Aerodynamics
Description of Fluid Motion Carrying out the cross product, we obtain
163
Equation (4.13) enables us to express (4.8) in terms of the Lagrangian variables. Thus we obtain
J
dx dy u v
k
dz = i(w dy  v dz)
w
Q
= Q[z = flea, b, c, t), y =/.(a, b, c, t), Z = /3(a, b, c, t), t]
(4.14)
+ j(u dz 
w dx)
+ k(v dx 
u dy)
4.4 Steady and Unsteady Motioos
If at various points of the flow field all quantities (such as velocity, density, pressure) associate~ with the fluid flow remain u'lchanged with time, the motion is said to be steady; otherwise it is called unsteady. Thus in steady motion time drops out of the independent variables, and the various field quantities simply become ,functions of the space coordinates.
(4.l7b)
=0
This vector equation is equivalent to three scalar equations
wdy  v dz == 0
udzwdx==O vdxudy=O
(4.17c)
(4.17d) (4.l7e)
4.5 Path Line
The curve described in space by a moving fluid element is'known as its trajectory or path line. Suci! a line is obtained by giving the position of an element as a function of time.
v
Streamline
4.6 Streamlines
We can describe another set of curves in the space filled by a flowing fluid.· At a certain instant of time, mark out the direction of velocity at each point of space. Then draw a family of curves such that each curve is tangcnt at each point to the velocity direction at that point. Such curves, as mentioned before, are called streamlines. They are nothing but the field lines of the vector field of velocity (see 2.20). To express analytically the equations for the streamlines we proceed as follows. Considering a ce(tain instant of time, at any point r, let ds be an element of the streamline passing through the point and let V denote the velocity ve~tor ,at that point at that instant (Fig. 4.1). Then, in view of the definition 'of a streamline, we state that the direction of ds is the same as that of V, that is, (4. is) ds is parallel to V Since the cross product of two parallel vectors is zero, we express (4.IS) by writing (4.16) ds"V==O which then is the differential 'equation for a streamline. Equation (4.16) , can 'be readily specialized to imy coordinate description. Choosing Cartesians, for instance, let the components of ds be denoted t;>y dx, dy, dz, and those of V by u, v, w. Then (4.16) becomes
(idz
FJa. 4.1
Definition of a streamlinc.
which can further be put in the symmetric form
dz _ ~ == dz w v u
(4.18)
+ j dy + k dz)
x (iu
+ jv +
kw) = 0
(4,173)
• For detailcd considerations regarding the relation between path lines, streamlines, and the socalled streak lines consult Prandtl and Tietjens (1934).
Note that u, v, ware functions of z, y, z, andt. Equation (4.17) or (4.18) is the Cartesian form of the differential equations for a streamline. We consider the integration of such equations in Section 4.9. Thus at each instant of time we can construct a picture of the streamlines. If the motion is unsteady, the streamline picture will change from instant to instant. If the motion is steady, the picture remains the same for all times. In this casepath lints and streamlines are identical. A picture of the streamlines helps us to see, as it were, the flow field and therefore plays an important part in the analysis and understanding of fluid t10w problems.
164
IdealFluid Aerodynamics Description of Fluid Motion
/65
Experimentally, various methods are available by which we can make fluid dow visiBle and obtain photographs of the streamlines. We cannot enter into a discussion of these techniques here.· Examples of experimentally obtained streamline pictures have already been given. In our studies here we shall compare such pictures with those obtained analytically. This would help us to evaluate our theoretical ideas in the light of experimental facts.
everywhere. For steady flow, a stream tube behaves like one with rigid boundaries inside which fluid flows. It follows that in steady flow, an impermeable solidfluid boundary, across which there cannot be any flow of fluid, would be a stream surface.
4.8 Reference Frame and Streamline Pattern
It is to be noted that a streamline pattern would appear differently from different reference frames. This is illustrated in Plates 12 and 13 for flow around an airfoil that is moving through an otherwise undisturbed fluid. The streamline pattern shown in (12) is obtained in a reference frame fixed with respect to the airfoil, while the pattern shown in (13) is obtained in a reference frame fixed with respect to the undisturbed fluid far away from the airfoil.
4.7 Stream Surfaces and Stream Tubes
If at any instant of time we draw an arbitrary line in the fluid region and draw the streamlines passing through the line, a surface is formed. Such a surface is called a streamline surface or simply a stream surfar.:e (Fig. 4.2).
4.9 Stream Functions
We now consider the integration of the Eq. (4.16) for a streamline
dsxV=O
Although V = V(r, t), the integration of (4.16) involves only the space variables. As such, in what follows, we shall not exhibit explicitly the dependence on time. Bear in mind, however, that the results obtained hold at every instant of time. First consider the Cartesian form of (4.16) as given by (4.17).
Fig. 4.2 Stream surface.
If we consider a closed curve and draw all the streamlines passing through it, a tube called a stream tube'is formed (Fig. 4.3). In unsteady motIL'n, the shape of a stream surface drawn through an arbitrarily chosen curve and the shape of a stream tube drawn through a chosen closed curve will change with time. lD steady motion stream surfaces and stream tubes once drawn remain unchanged. No jfuid can· cross a stream surface or the walls of a stream tube, for the walls and the stream sllrface are always parallel to the fluid velocity
v(x, y, z) dx  u(x, y, z) dy = 0 w(x, y, z) dy  to(x, y, z) dz = 0 w(x, y, z) dx  II(X, y, z) ch
(4.19)
=0
These are a set of differential equations for the variables x, y, z. As may be readily verified, (4.19) is actually a set of two iridependent equations. We may, therefore, represent (4.19) by two equations of the form
a 1(x, y, z) dx a2(x, y, z) dx
+ Mx, y, z) ely + cix, y, z) dz = + b,(x, y, z) dy + c.(x, y, z) dz =
0 0
(4.20)
Fig. 4.3 Stream tube, • Consult, for instance Pankhurst and Holder (1952) or Prandtl and Tietjens (1934).
Equations of this type are known as PfalJian diflerential equations. For the theory of such equations, reference may be made to any suitable book, for instance that by Margenau and Murphy (1956) or by Sneddc·n (1957). The solution of each equation in (4.20) may be represented by an equation of the form (4.21) f(x, y, z) = c where c is a constant. As is well known, Eq. (4.21) describes a oneparameter family of surfaces. We thus conclude that the solution of
166
IdealFluid Aerodynamics
Description of Auid Motion
167
(4.20) or equivalently of the differential equation for the streamline is given by two independent functions
'f'l(X, 'f'2(X,
y, z) ==
Cl C2
y, z) =
(4.22)
where C1 and C2 are constants. Such a conclusion could have been reached immediately f~om the observation that a line in space, such as a streamline, may be described as the curve of intersection of two surfaces. The functions '1'1 and '1'2 should naturaUy be related to the velocity
,.he function p(r) is arbitrary except that it should satisfy a certain condition that follows directly from Eq. (4.24). As may be verified it is a vector identity that if h(r) and f.(r) are two scalar functio'1.s of position, then div (gradh X grad fa) = 0
It, therefore, follows that p(r) should be such that it satisfies the condition
div (I'V) == 0 To sum up, we state that the solution of the equation dsxV==O for a streamline is given by two functions lpl(r) ==
Cl
.(4.25)
1p.(rt= c.
where C1 and c. are constants. Tl.ese functions and the velOcity are related by I' V == grad '1'1 X grad 11'. where p is a scalar function of position that satisfies the condition
Fig.4.4 Stream functions and streamline.
div I'V == 0 The functions '1'1 and '1'2 are known as stream functions.  If the function I'(r) is known, then the vector field V(r) may be replaced by two scalar fields 'f'l(r) and 'f'2(r) gaining possible mathematical advantage. In general, a fun~tion I' satisfying the condition (4.25) without at the same time imposing physically untenable conditions on ·the velocity .field, is not readily obtained .. Under certain circumstances, however, independent considerations relating' to the fluid motion lead to relations of the form divf(r)V = 0 where f is a certain scalar function of position. In such cases we identify p(r) with f(r) and replace V(r) by 1p1(r) and 'f'2(r) by use of Eq. (4.24). Examples of. such situations ar. the motion of an incompressible fluid and the steady motion of a compressible fluid. As we shall see later, for an jncompressible fluid we shall have divV = 0 then we can set I'(r) = 1
• Note that we have suppressed showing explicitly the dependence on time.
components u, v, w. We now. establish the relation. For generality and convenience we now switch to vector representation. We write.
V = VCr)
'1'1 = 'P1(r) = (4.23) '1'1 = 'f'2(r) = Now, along the streamline, the velocity vector V lies in both the surfaces '1'1. = C1 and '1'2 = c. (Fig. 4.4). This being the case, along the streamline V IS ~ormal to both grad '1'1 and grad '1'., the gradients being parallel to the normals to their respective surfaces; Symbolically we have
c.
'"I
V·grad'f'l ==0 V • gra~ '1'2 == 0 This shows that V is normal to the plane formed by the vectors grad '1'1 and grad '1'2' In other words V is paraUel to the cros, product grad 'fI1 x grad If'z Hence we write p(r)V = grad '1'1 X grad ~L'~ where per) is a scalar function of position.
(4.24)
168
For steady compressible flow, we shall have div p(r)V = 0 where p(r) is the density field. Then we can set ,u(r) = p(r)
IdealFluid Aerodynamics Description of Fluid 'Motion We then have
169
j
k
Solving fluid flow problems by use of two stream functions when 'such use is possible has not received much attention, and at this stage no general comments can be made about it. In certain problems where only two velocity components appear, the other being zero, there would be only one single unknown stream function instead of two. We now consider such situations.
,u(x, y)V = a" a" o ax ay
0
Equivalently, we obtain
0
,uv =  ax
The function
a",
(4.26)
4.10
Stream Function for TwoDimensional Flow
,u
should he such that the following relation is satisfied:
When the motion of a fluid is such that the flow pattern and the various flow quantities are indepel'\den\ of distance along a certain fixed direction, the motion is said to be twodimensional or planar. Thus, if we designate such a direction as the Z axis, we shall have
a,uu + a,uv = ax oy
y
div,uV = 0
(4.27)
.E.. ( ) == ~
0%
v
V
a%
of any quantity = 0
The motion in all planes normal to Z will appear the same. Introducing Cartesians x, y, %, we write cis = (dx, dy, dz) V = (u,v, 0) U = u(x, y) v = v(x, y)
~~~
u
_ _ _ _ L_ _ _ _ _ _
,X
We consider the equation for a streamline in the form
dx dy d% ==u(x, y) v(x, y) 0 It immediately follows that dz = 0 or % = constant Thus, one of the stream functions is simply z. The other is the only un . known function. It is a function of x and y only. Denote it by ",(:e, y). To relate ",(x, y) to the velocity components u, v we use the Eq. (4.24). Set* = ",(x, y)
Z
Fig.4.5 Velocity components in twodimensional flow.
We now express the results in terms of cylindrical polar coordinates r,
0, z (see Fig. 4.5). We write
ds = (dr, rdO, dz)
V = (u r , "8' 0)
Ur
"'1
=
ur(r, 0)
"'2
=
Z
U
8
= u8(r, 0)
• We may as well set '1'. = z and '1'2 = ",. The choice made leads to the usual form of
the
re~u:ls.
We then have
dr rdO dz ==urCr, 0) us(r, 0) 0
170
IdealFluid Aerodynami<:s
,Descrirtion of Fluid Motion
171
As ~fore, one of the stream functions is z. For the other, say '"  tp(r,8) 'we have, using Eq. (4.24),
In terms of the spberical coordinates R, 8, cp,we write .... (dR, R dB,
V UR"" (UR' U"
er A,.
e,
r 08
e.
·.0
R sin 8 dfP)
0)
per, O)V _ a", ! otp
uJiR, 9) u, .. u,(R, 8)
o
or, equivalently,
o
1
(4.28)
Consider the equation for a streamline in the. form dR uJiR, 8) .
It immediately follows that
'(4.29) or
a", pu, ....  or ![o'(rpuJ + .!(Pu,)] r Or
'08
divpV  0
The fuhction
per, 8) should satisfy the relation
 R d8 u,(R, C)
R sin 8 dfP 0
cp  constant Thus, one of t~e stream functions is simply cp. The other, say tp .... tp(R, 8), is the only unknown function. To relate tp(R,8) to the velocity components uR, U" we set '1'1  .,(R, 8) and use Eq. (4.24). We· thus have
4.11
Stream Fanctioo for Axisymmetric MotioD
When the motion of a fluid is such that the flow pattern and the flow quantities at corresponding points are the saJtle in all planes passing
z
eR
p(R, 8)V 
e,
oR R 08
a" .! 0"
o
o
o
1
R sin 8
Equivalently, we have
PUR
1 0", R1 sin 8 08
1 0", R sin 8 oR
Fla. 4.6 Coordinates (or axisymmetric motion.
(4.30)
pu,....;;~~[..! (RI sin 8pUR) + !
RI SID 8 oR
through a certain fixed axis, the motion is ~aid to be axisymmetric. Introduce spherical coordinates R, 8, fP or cylindrical coordinates r, fP, x as shown in Fig. 4.6. Let the motion be axisymmetric about the X axis. We then have
The function· peR; 8) should satisfy the relation
(R sin 8p
1. ( ) _ ..! of any quantity OfP OfP· '
0
08
u,)] == div pV
.... 0
(4.31)
172
IdealFluid Aerodynamics
Description of Fluid Motion
173
We now ellpress the results in terms of cylindrical coordinates r, fIJ, x (see Fig. 4.6). We have ds =: (dx, dr, r dq;) V
u,
ax
=:
(u". "r'O)
u,,(x, r) ur(x, r)
dr
U'" =: =:
=:
",,(x, r)
ur(x, r)
",(x, r)
q;
=0
r
dcp
"'I =
"'2 =
fl(x, r)V
=:
e e, .
or
o o
e~1
0
0", 0",
ax
r
(a) Nodes
1 flU = '" r
~U
a", or
1 0",
.
,

r'Ox
(4.32)
(4.33)
The stream function for allisymmetric flow is known as the Stokes stream junction.
4.11 Stagnation Points
(b) Saddle
(c) Spiral
P~ints in the flow field where the velocity becomes zero are known as stagnation. points. In terms of the compC'nents of the velocity, this ,means that at a stagnation point all the velocity componentS are zero. Consider the differential equations for a streamline in the Cartesian form dy = vex, y, z) dx u(x, y, z) (4.34) dz w(x, y, z) dx u(x, y, z)
(d) Vortex
Fig. 4.7
Examples of streamline patterns at a stagnation point in twodimensionarllow.
174
At a stagnation point these equations become
IdealFluid Aerodynamics
Chapter 5
(4.35)
0  ==dz 0 dz 0 =dz 0
dy
Eulerian Equations for the MotiQn of an Ideal Fluid
Points at which differential equations of the type (4.34) take the form (4.3::) are called singular points of the equations. Thus stagnation points are such singular points. We shall not enter here into the theory of integration of Eqs. (4.34) at the singular. points. For such theory, reference may be made to bo.oks on differential equations, for instance see Goursat (1959). Without explicit analysis we state that the streamline pattern at a stagnation point is not the simple picture of a single str~amline passing through" a given point. Examples of possible streamline patterns at a stagnation point in twodim~nsional flow are given in Fig. 4.7. In this context, reference may be made ~o Karman and Biot (1940) where analysis is given for determining the shape of the integral curves for the equation
=dx
dy
f(z, y) g(z, y)
at its singular points.
The Eulerian method of description defuies, for any particular value of time, the state of motion at all points of the space occupied by the fluid, while for a given position the method gives the history of what goes on at that place. Thus in this point of view the fluid flow is characterized by the fields of velocity, pressure, density, and so ori, and a fluid element or particle occupying a certain point at aeertain" time assumes for its proDerties the values that are appropriate to that point at that instant. The object of our investigations in fluid flow problems; then, IS to determine these fields. We attempt this by first formulating possible relationships that should be satisfied between the field quantities. These relationships are esta1?lished on the basis of certain natural laws such as N~wton's second law of motion. We refer to the relation!'hips betWeen the field quantities as the equations offluid motion. The equations, may be set up either in differential form .in in integral form. Furthermore, they may be developed either from t"~· point of view of a certain" "fluid region" that contains the same fluid elements (or element) for all times or from the point of .View of a "fixed volume in space'" through which different fluid elements flow through. We shall concern ourselves first with the derivation of the equations in the differential form from the point of view of an infinitesimal ftuiQ region. We shall then naturally be involved with the calculation of the ~ime rate.of change of any quantity following·a fluid element. We shall discuss this in the next section and then formulate the basic equations. 5.1 Local, Convective, and Material Deriyatives Consider the fluid element situated at the point r at time t. Let Q(r, t) denote some fluid property Q (density. velocity, etc.) associated with the point r at that instant. The ftuidelement ~ituated there will, therefore, assume for its corresponding property Q the value Q(r, t). In a short time
175
116
IdealFluid Aerodynamics
Eulerian Equations for the Motion of an Ideal Fluid
177
interval fl.t the element moves through a directed distance ~s == V /}.t, where V i; its velocity at rand t (Fig. 5.1). In the new situation the element will assume for Q the value appropriate to the positien (r + V fl.t) at the time (t + fl.1). This we denote as Q(r + V ~t, t + fl.t) .. T~en the change in Q for that moving element in the time mterval fl.t IS given by
fl.Q
==
Q(r
+ V fl.t, t + fl.t)
 Q(r, t)
Path of element
Fla 5.1
lllu_strating the local, convective and material derivatives.
and the rate of change of Q following the element is simply)he limit of fl.Q/fl.t as fl.t vanishes. This rate of change is usually denoted by DQ/Dt. Therefore we have
DQ Dt
== lim fl.Q = lim Q(r + V~t, t + fl.t)  Q(r, t)
AIO
(5.1)
fl.t
AIO
fl.t
This rate of change can be expressed as made up of two parts, o~e a change due to local variation with time of th~ .fluid pro~rty ~t a gIVen position and the other due to a change of POSition at ~ given time. Formally we write Q(r
+ V fl.t, 't + fl.t) == Q(r, t) +
(~Q\ fl.t + (~.~\r.t(fl.t)1 + ... (It Jr.t (It J
'V fl.t
+ (oQ\
as Ir.t
+ (Ol~)
os . r.t
(V fl.t)Z
+ .. ,
(5.2)
where s denotes distance in the direction of the vefocity V at the point r at time t. UsiJ,g the Taylor expansion (5.2) in relation (5.1) we obtai!}
where the derivatives are ev~luated at rand t_ The derivative oQlot denotes iocal variation with time at a given position, while the derivative oQlos denotes v3riation with change of position at a given time. The term (oQlos)v itself denotes the time rate of change of Q due to change of position. A physical interpretation of (5.3) is as follows. Consider a flow field which at any given instant is uniform throughout the space but varies from instant to instant. If a fluid element moves in such a field from the point r, the change in any property Q for that element in a small time interval 6t i! (oQlot) 6t (correct to the first order in 6t), where oQlot denotes the rate at which Q is changing locally at the point r. This change is called the local change. Suppose the flow field is steady but not uniform, that is, the fluid property Q though varying from point to point does not ~hangt with time at any point. Now, if in such a flow field an element moves from a· . point r to a new point r + V 6t, a .change in Q must take place for that element so as fo adjust the elemtnt to the new location. This change, which results from the fact that the element is getting into a new environment of the flow field, is called the convective change. This change is equal to (oQlos)V 61 (to the first order in V 6t), the change in Q over the directed distance 6s equalto V 6t. When a flow field is neither steady nor uniforin, the change in any property Q for a particular fluid element will be made up of both the local and convective ch~nges. and is. equal to [(oQ/ot) lIt.+ (oQ/os)V 6t). Hence it follows that the rate of change of Q following a fluid element is given by Eq. (5.3). The local rate of change oQlol is known as the lucal derivative, and the convective rate of change (oQ/os)V is known as the.convective derivative. The total rate of change DQ/ Dt is usually known as the substantial derivative. It is sometimes referred to as the particle or material derivative. This is a more descriptive name, for the derivativ!= is constructed following a certain fluid element. Equation (5.3) may be used to compute the material derivative of any quantity whether it be a scalar field, a vector field, or a more general tensor field .. For scalar and vector fields, the convective derivatiye in (5.3) may be put, as follows, into a more explicit form. We recall that the operator e· grad (or e • V) applied to a scalar orvector field yields the derivative of that field with respect to distance in the direction e (see 2.75 and 2.77). Now, in (5.3) the term oQ/os represents the derivative of Q with respect to distance in the direction of the velocity. Denoting this direction bye.. we may therefore write
DQ _ oQ Dt at
+ oQ V
os
(5.3)
17.
IdealAuid
A~rodynamics
Eulerian Equations for the Motion of, an Ideal Auid
179
and aQV::v.gradQ
we have
as
(p~V)
D
DI
= F
(5.6)
whether Q is a scalar or a vector. Then for the material derivative we have
as the equation of motion ,of the fluid.
As the elelDCnt moves, its shape, volume, and density wi'!. in general,
DQ :: iJQ
DI
qt
+ V • grad Q
(5.4)
change. II's mass, how~ver, remains constant. Thus (5.6) takes the form
"p~=F
If the quantity Q is a vector field, denoted, say, by A) the convective derivative of A may be further expanded aCcording to the following formula: v· grad A = l[grad (Y. A)  V x curl A A x curl V  curl (Y x A)
DV
DI
(5.1)
where DV/ DI is the acceleration of the fluid element situated at r at time
+V(div A) 
•
_=ad8
A(div V)]
(5.5)
We will now set up the basic equations that govern the motion of an inviscid incompressible fluid. At the start, however, we will not regard the fluid as incompressible. The condition of inco~pressibi~ity will be introduced at the ,appropriate time so as to bring out clearly the role of incompressibility in the formulation offluid problems. Initially, we regard as our unknowns the velocity field V(r, I), the pressure field p(r, I), and the density field per. I). We seek to establish relationships between these fields by applying to a certain ftuid element the basic laws of natu~: Newton's
Surface of element _ &8 Vo/ume of element. &T
second law of mOlion, law of con.rervalion of mass, law of conservation of energy. It might tum out that these laws are actually inadequate in setting
sufficient relations between the unknowns. In such a case one must invoke additional relations suggested by experience.
u~
FIt. 5.2 Fluid element.
Equation (5.7) expresses the usua1.acccleration form of Newton's law: mass times acceleration is equal to force. The acceleration DV/ Dt may be determined by use of Eqs. (5.1) or (5.4). According to (5.4) we have
I.
1.2 lAder's Eqaatloll
The fundamental equation governing the motion of any mechanical system expresses Newton's second law of motion, which states that at any
wiant, the role of clronge of momentum of a system 4 equal 10 the force acting on il at lhal Wiant. By force is understood the resultant of all the forees that are acting. To obtain the equation governing the motion of a ftuid we apply Newton's ~ond law to a moving fluid element. 'Let us consider an infinitesimally small clement of fluid situated at the position r at time t (Fig. 5.2). If V and p are tbe velocity and density, respectively, at rand t, and if 6T denotes the volume of the clement, the mass and momentum of the element are p:6T and p 6TV~ respectively. We denote by F the force acting on the element at that instant and equate it, according to Newton's law, to the rate of change of momentum of the elelDCnt, which is simply the material deiivative of the momentum. Thus
Iii" = al + V • grad V
Using (5.5) we obtain V"· grad V  grad VI
DV
aV
(5.8)
2
 V x curl V
"
It then follows that the acceleration' "'ayalso be expressed in the form
DV aV  = DI
VI a, + grad 2 
V x curl V
(5.9)
This form, which may be easily specialized to any orthogonal, curviliQea,r coordinate system, is very useful when we investigate the integration at: the
Eula:ian Equations for the Motion of an I~l Fluid
110
IdealFluid Aerodynamics
111
equations of fluid motion (see Section 8.4). In. Eq. (S.9) th~ vorticity, namely curl V, which is twice the angular velOCity of the Owd. appears explicitly. . " . In determining the total force acting on the fluid eleme~t we dlstmgulsh between the socalled body forces and surface forces (Sectlons 3.2 and 3.7). We denote by f the body force per unit mass of the fluid. The body force acting o~ the jluid element is, therefore, P ~".(. To obtain the resultant of the surface forces acting on the· element we note that, since we are concerned only with a non~iscous fluid, the surface forces are simply pressu("e forces that act no~mal to the ~urface of the element. Let us denote by p the pressure, by ~S the total su~face area of the fluid element, and by D dS an elemental area. on ~S, D bemg an outward normal (Fig. S.2). The resultant oftM pressure forces acting on the element is then equal to
This equation of motion is one of the fundamental equations of OUid dynamics. It was first obtained by Euler in 17SS and is called Euler's Equation. In deriving this equation no account has been taken of the viscous nature. of a fluid. Therefore it holds good only for the motions of fluids in which viscosity is assumed to be zero, that is,' for the motions of an inviscid fluid. Euler's equation (S.I1) represents a system of three scalarequaq 'ns for· the five unknownsthe three Sc:aIar components of the velocity, the pressure, and. the density. Hence it is necessary to obtain· additional equations. We do this by the application of the laws of conservation of mass and energy. We derive first the equation that. expresses the conservation of mass.
5.3 F.qaatioa of CaMemadoa of M_
We assert that the mass ofanyfluid element remains constant as it moves about even though, in general, its shape, volume, and density may change". Consider a fluid element of volume h that is Situated at the point r at tiine t. The mass of the element is ph.· Since·the mass of the element is a constant, it follows that p. h is a CQlUttRIt if one fo/~ows the same element. In other words, the material derivative of phis zero.
(ph)"'; 0 Dt .
§padS
IS
According to the integral definition of the gradient of a scalar function (Eq. 2.1161') we have
ffpa
18
dS
= ~6Tgrad p
D·
.
(S.12)
This shows that in an in viscid jlu.id,  grad P represents the resultant surface force acting on unit volume of t"! jluid.. The total force on the fluid element Is thus gIVen by
This is the equation of conservation of mass in its simplest fonn. Equation (5.12) may ~ rewritten as
F = p 6",(  6i grad p
The equation of motion of the fluid (Eq. S.6) now takes the form
(5.10)
h!!£
Dt
+ p!!..(6T) .. 0
Dt .
(S.13)
The material derivative of the density is given by
p lJT DV = p M  lJT grad p
Dt
Ee = iJp + V • grad p Dt iJt
(5.11a)
(S.14)
or
p
DV
Dt
= pf  grad p
The material derivative D(~)I Dt of the volume of the element may be expressed in terms of the velocity field. To do this we .observe that
when expressed per unit volume of the fluid. Using Eqs. (5.8) or (S.9) for the acceleration we may write (S.1la) in the following alternate form:
.Q. (~T)
Dt
= lim 61(t + dt) 110
61(t)
In
p(~~ + V. grad V)
or
= I'f  grad
p
p
(5.11b)
p(OV
ot
+ grari V
2 _
V x curl
2
v)
= pf  grad
(S.11c)
where ~T(t) is the volume of the element at time t and 6T(t + 6T) is its . volume after a small time interval ~t. The change in volume of the element during the time intervallJ! may be determined as follows. Let us consider the element as situat'ed in the position rat tiine t and assume! for c1_arity.
1.2
IdealFluid Aerodynami<:s
Eulerian Equations for the Motion of an Ideal Fluid definition of the divergence of a vector (Eq. 2.117) we have
that at that instant the velocities at the various points on surface of~he eiement are all directed outward, that is, toward the re810n that contains the outward normal (see Fig. 5.3). Then, as shown in the figure, !he surface of the element t~t is tJS at time t grows into the ~urface tJS1 dun~g the interval·6t. Both the surfaces tJS and tJS1 contain the same flul~ element but at different times. .The change in volume of the elemen~ IS simply equal to ;the volume swept by the surface of the element dunng
!he
ffv .
.8
D
dS  tJT div V
Using this relation we obtain
.P.. (tJT) == tJT div V
Dt This may be rewritten as div V
_1.
D· (tJT)
tJT Dt
(5.")
This equation shows that t~ divergence of the velocity field give, tire rtIU ofchDnge ofvolume ofa fluid element per unit volume. It thus represents the rate of volumetric strain. It is also known as the dilatation or exteruion. From Eqs. (5.12).and (SolS) we obtain
6r Dp
Dt or
+ p tJT div V _
0
Dp .'  + pdivVO Dt
(5.17)
when expressed per unit volume of (he fluid. This is the required equation of conservation of mass or, simply, the equation of mass~ It is a relation
Fig: 5.3 Illustrating the chan~ in volume of a fluid element.
between the velocity and density fields only. A relation that does not involve any dynamical quantities,such as forces and pressures, is known as a kinematical relation. Thus the equation of c()nservati~n ofmass is purely a kinematical relation. Consequently, it holds good for the motion of all
~~
the time {Jt. Now, if the velocities at various points ofthe .surface of the element are not all directed outward, the change in volume IS equal t~ the. net volume swept outward by the surface of the eleqtent. If.D dS IS an elemental area on the surfau tJS, n bei~\g the outward normal (Fig. 5.3), t~e net volume swept outward by {JS in time {Jt is given (to the first order m
.
Substituting from Eq. (5.14) for the material derivative of the density we may rewrite Eq. (5.11) as
 +
or as
ap
tJt) by
where V denotes velocity. Th~s it follows that
ffv ,8
at
V • grad p
.
+ p div V == 0
0
(5.18)
{Jt· n dS
ap .  + d·IV PV. =
at
.P.. (tJT) ,;",JtV • n dS Dt If 68
Now, for an infinitesimally small volume element according to the integral
This equation has an interesting interpretation which we shall see later (Section 6.2). The equation of conservation of mass is also known as the equation ofcontinUity. By continuity we mean physical continuity, implying that the fluid always remains a continuum, that is, as continuously distributed .matter.
184
5.4 Equation of Energy
IdealFluid Aerodynamics
Eulerian Equations for the Motion of an Ideal Fluid pressure forces. Accordingly, we ha>«:
WI ==' JfPD dS· V ==
18
1"
(S.22)
The law of conservation of energy expresses the balance of energy that take place between a system yd its surroundings. In applying the law to a fluid' in motion we should regard the fluid as a thermodynamic system, assume the existence of the usual thermodynamic variables and take into account all the several processes that may contribute to the exchange of energy between the different fluid elements themselves and between the fluid and its surroundings. Some examples of such processes are: work on the element by the body and surface forces, heat conduction, chemical reaction, radiation and electromagnetic action. For our purposes we shall assume that the 'fluid ,is a nonheatconducting medium
e~hanges
ffpv. DdS
18
where 6S, as before, is the surface of the element. We have assumed, as before, that the fluid is inviscid and, consequently, that the surface forces are only pressure forces. For an infinitesimally small volume element Eq. (5.22) becomes WI == ~ div pV
and that no processes of energy excbange other than the work by body and surface forces take place. Under these assumptions the law of conservation
of energy for a fluid element may ije expressed as rate of increase of energyt: of a fluid element = rate WI at which work is done on the ele,ment by the body forcel> + rate WI afwhich work is done on the element, at its surface, by the surfaCe forces This is expressed symbolically as =W1+WI
Dt
== lJT (grad p • V + P div V) (5.23) This equation shows that the rate at which work is done by the surface forces is equal to div pV per unit volume of the fluid and that this rate of work is made up of two partsQne part given by grad p' V represents the rate, per unit volume, at which work is done by the resultant of the surface forces, and .the other part given by p div V represents the rate, per unit volume, at which work is done on the element due to an increase in its volume. : Recall that div V represents the dilatation. Using Eqs. (5.19), (5.20), (5.21), and (5.23), we obtain
:,[p~(e+ ~I)J =p~f.V....,lJrgradp.V6TpdivV
(5.19) or, since the material derivative of p lJT is zero,
DE
The energy of the fluid is the sum of two parts, one the kinetic energy, a mechanical quantity, and the other the internal ent:rgy, a thermodynamic quantity. We specify the internal energy of the fluid by the scalar field e = .e(r, t), which denotes the internal energy per unit mass at any point r at time i. The kinetic energy of the fluid per 'unit mass is given by the scalar field yt/2. Thus the energy of the fluid per unit mass is equal to
Vi e+2
Plh!2.(e + yt) = Dt 2
p
pM·V 
~gradp·V 
6TpdivV
Expressing it per unit volume, we may write this equation as
!2.(e + VI) = Dt· 2
pf. V  grad p. V 
P V div
(5.24)
This is the required equation of conservation .of energy or, simply, the equation .of energy. It is also referred to as the e~'!lation of total energy. EquatIOn t5.24) may further be reduced to a simpler forin as follows. We form the scalar product of the equation of motion (5.11a) with obtain
p
V
and'
Consequently. the energy of the fluid element is given by
E== p 6T(e + ~I)
The rate at which work is done by the body forces is given by WI::;: p 6Tf· V
(5.20)
!2.(V?) = f. V Dt 2
grad p' V
(5.25)
(5.21)
This equation, known as the equation of mechanical energy. states that the rate of change of kinetic energy of a l1uid element is equal t~ the rate at which wolil is done by the body forces and the resultant of the surface forces. Subtracting (5.25) from (5.24) we obtain
De p  = p'd'IV V Dt
(5.26)
TQ obta.in the rate at which work is done by the surface forces we assume dtat the ftuid is inVi~d and,. consequently, that the surface forees are only
186
IdealFluid Aerodynamics
Eulerian Equations for the Motion of an Ideal fluid
1.7
which states that the rate of increase of internal energy of a fluid element is equal to the rate at which work is done on the element during an increase in its volume. This simple form of the energy equation is an expression of the first law of thermodynami.cs applied to a system undergoing an adiabatic change of volume, that is, without any heat addition. EquatiQn (5.26) is thus known as the thermodynamic form of the energy equation. It holds good only for a nonviscous, nonheatconducting fluid. The energy equation introduces the internal energy as an additional unknown in the formulation of the equations of fluid motion. The. list of unknowns is thus increased from five to six, namely V, p, p, e, while the number of equations is brought up to five. It therefore follows that an additional relation between some of these unknowns is needed. Such a relation is provided by the socalled equation of state of the fluid.
where T is the temperature. Since we are presently concerned with the variables e, p, and P, we shaH choose
e = e(p, p)
as the equation of state.
(5.27)
5.6 Equations for an Inviscid Compressible Fluid
Summarizing the precediDg consideratiolls we may state that the motion of an inviscid, non heatconducting, compressible fluid, which is assu:ned to be a simple thermodynamic system, involves three scalar fields p, p, e and one vector field V as the unknown functions and is governed by the following system of equations:
(l) Equation of motion
p
5.5 Equation of State
The thermodynamic state of a system is described by a ceJ,"tain number of thermodynamic yariables, such as pressure, density, temperature, internal energy, enthalpy, and entropy. Such variables depend only on the state of the system and are called variables of state. For the complete description of the thermociynC1\u1c state only a few, instead of all, of the state variables are necessary. The number of variables required depends on the nature of the system: it may be a simple system composed of a single fluid or a homogeneous mixture of inert fluids, or it may be a nonsimple system composed of a mixture of reacting fluids. From measurements we realize that for a silnple system the thermodynamic state is completely described by two state variables, whereas for a nonsimple ~ystem a greater number of variables is required. Once the number of variabl~ required for a complete thermodynamic description is known, we can express ~ny of the state variables as a function of a few others chosen as independent variables. The number of these independent variables is simply the number required for complete thermodynamic description~ Such functional relations between s\llte variables are called equations of state. A material medium may be characterized by its equations of state. The actual functional relation implied in an equation of state cannot be determined froni. thermodynamics. It is usually obtained from measurements. For our purposes we shaH assume that the fluid in motion is a simple thermodynamic system. In such a case the fluid may be characterized by equations of state such as
DV
Dt
= pf  grad p
(5.lla)
(2) Equation of conservation of mass
Dp
Dt
(3) Equation of energy
+ pdiv V =
0
(5.17)
De p  = p d' V IV
Dt
(5.26)
(4) Equation of state e = e(p, p) or (5.27)
p = p(e, p)
These equations are known as the equations of inviscid compressible flow. The integration of these equations forms the subject matter of the mechanics of an indscid compressible fluid. Our business is with the mechanics of an inviscid fluid that is incompressible. We shall now introduce the condition of incompressibility and deduce the equations that govern the motion of an inviscid incompressible fluid.
S.7 Condition of Incompressibility
e = e(p, p) p
= p(e, p)
A fluid is said to be incompressible if the volume of every element of the fluid is a constant for all times. Thus, if br denotes, as before, the volume of a fluid element, the condition of incolllpressibility may be expressed in the differential form as
R(br)
p  p(p, T)
DI
=
(hdiv V
=
0
118
or simoly as divV = 0
IdealFluid Aerodynamics (5.28)
Eulerian Equations for the Motion of an Ideal Fluid For an incompressible fluid, the energy equation (5.26) becomes
119
This equation, which is purely a kinematical relation, states that the divergence of the velocity field i~ zero.
De ae  =  + V • grad e = 0
Dt at element is
(5.32)
S.8 Conseqaeaces of Ineompressibillty
The assumption Of incompressibility gives ~se to much simplification of the ~uations of fluid ,mechanics. We observe that if the fluid is assumed incompressible, the condition of incompressibility may be added to the equations of motion and mass conservation to, form a system of eqWltions sufficient to solve for the unknown fun~tions. p, p, and V. This means we need not invoke any thermodynamical considerations and additional equations, s,uch as the equation of energy and the equation of state. It thus appears that the assumption of incompressibility reduces the problem of the motion of an inviscid fluid from one in mechanics and thermodynamics to one in mechanics only. As a consequence of Eq. (5.28), the equation of conservation of mass (5.17) becomes Dp ap  =  + V • grad p = 0 (5.29)
which states that in an. incompressible fluid the internal energy of any fluid 0 cOllstant for all times. From (5.32) it follows that if at some initial time e is nonunif~rm in space (whence we say the fluid is' again inhomogeneous), then for aU other times there will exist both local and spatial variations of e although the material derivative of e is zero. Howev~r, if at some instant e is uniform in space (whence we say that the fluid is homogeneous), then for all other times both the locai and spatial variations of e vanish independently and (5.32) reduces to
e = const.
(5.33)
Bt:cause of the' condition of incompressibility (5.28) and th~ consequent energy equation (5.32), the total energy equation (5.24) becomes
D V2 p   = pl· V  grad p • V Dt 2
which is simply the equation of mechanical energy. We see that for an inviscid, incompressible, inhomogeneous fluid we may solve for p, p, and V from the condition of incompressibility (5.28). the equation of mass conservation (5.29), and the equation of motion (5.lla) independently of the equation of energy (5.32), Onc~ V is known •. we may sol~e the energy equation (5.32) to obtain e. Equations (5.28), (5.29), and (5.32) are purely mechanical in nature, and their solution does not involve any thermodynamical considerations. This situation becomes strikingly dear in the case of a homogeneous fluid. For a homogeneous fluid the density is constant and drops out of the unknown functions, thus reducing the initial list from p. p. V to p and V only .. We may solve for these from the condition of incompressibility (5.28) and the equation of motion (5.31). For a homogeneou~ fluid the internal 'energy is' also a constant for all tim~s, Consequently, the internal energy and along with it all thermodynamic considerations disappear. as it were, from ,the problem of the motion of an inviscid. incompressible. homogeneous fluid. Once the fluid is assumed incompressibl~. the pressure becomes purely a mechanical variable, and no thermodynamic significance. may be given to it. This is clearly seen in the case of the homogeneous ,fluid. For an inviscid. incompressible, homogeneous fluid both the density and the int~rnal energy are constant although the pressure varies both in time and in space. There is thus no fllOctional relation between the variable p and the density and the internal energy. For an illuminating explanation of the
Dt
at
.
This equation expresses the fact that, if! an incompressible fluid, the density of .'ny fluid element is a constant for all times. The equation does not imply that the density is a constant throughout the flow field f~r all times. From Eq. (5.29) it follows that if, at some initial instant, the fluid is inhomogeneous (i.e., one whose fluid elements are of different densities), it will remain inhomogeneous for all oth::r tirTtes and that if the fluid is homogeneous (i.e., one Whose elements are all of the same density) at some initial instant, it will remain homogeneous for all other times. Thus for an incompressible inhomogeneous fluid there will be both local and spatial variations of density although the material rate of change of dej1sity is zero. For an incompressible homogeneous fluid, however, the local and spatial variation vanish independently, and the density is constant in time as well as in space. Thus for an incompressible homogeneous flUid Eq. (:).29) takes the simple form p = const. (5.30) We consider next Euler's equation (5.1la). For an inhomogeneous fluid, the assumption of incompressibility leav,,~ Euler's equation unchanged .. For a homogeneous fluid, however, the ~ssumption of incompressibility. allows one to set p equal ta a constant and to wlte Euler's equation in the simpler form
Q!:::. = I  grad
Dt
(p.)
p
(5.3\)
190
IdealFluid Aerodynamics
Eulerian Equations for the Motion of an Ideal Fluid
/9/
concept of pressure in an incompressible ftUid, the reader may consult Sommerfeld (1950).
5.9 Equations for. an Ideal Fluid
Our concern here is with th,e study of the motion of:an inviscid, incompressible, homogeneous fluid, the socalled ideal fluid. Naturally no real'fluid has t~e properties of an ideal fluid. However, as pointed· out in Chapter 1, .!lnder certain circumstances many aspects of the motion of a real fluid may be analyzed on the basis of an ideal fluid. As we have.seen, the motion of an ideal flui~ is characterized by pressure and velocity fields only and is governed by the following equations: Condition of incompressibility divY= 0 and Euler's equation
DV =f _ grad Dt
on the nature of the differential equations that are assumed to govern the motion of the fluid. In this sense, for instance, differences exist between the boundary conditions fora viscous fluid and a nonviScou's fluid. We shall concern ourselves with only an ideal fluid.
COIUlitioIl lit II Sou.jb1i4 BoUlUltuy. We assume that the fluid is bounded by an impermeable so1i4 and require that no fluid should therefore penetrate the solid surface. Since, in general, each Clement of the solid surface may be in motion relative t.o the fluid, this requirement may be expressed as the following condition:
At each point of the solidfluid surface, at every instant, the component normal to the surface of the relative velocity between . the fluid and the solid must vanish. If at any point on the surface V R represents the relative velocity and n the normal to the surface, this condition may be written symbolically as VR
•D
(l!) p
(5.31)
whl"re
DV aV = Dt at cV + V • grad V = at
= 0
on a solidfluid surface
(5.34)
+ grad Vi  V x curl V 2
If V and Vs denote, respectiveJ " the velocities of the fluid and the surface, condition (S.34) takes the feCIr.
(VVs)·D=O or
These equations are the basis of all investigations in the mechanics of an ideal fluid. To specify a given problem we must add to these differer).tial equations the socalled initial and boundary conditions. We shall now see how to formulate s~ch cQnditions. 5.10 Initial Conditions Initial conditi<;ms are conditions that describe completely th~ state of the fluid at some instant of time which is referred to as the initial time . .For an ideal fluid a complete set of initial conditions is obtained if the .velocity V and the pressure p are specified at the initial time. .
5.11
V • D = VS
• D
on a solidfluidsurface
(5.35)
Equation (S.3S) states that at each point of a solidfluid boundary. at every instant, the normal component of the velocity of the fluid is equal to the normal component of the velocity of the boundary. If the boundary is formed by a fixed (stationary with respect to a frame fixed in space). rigid solid surface. the velocity of the surface at every point is zero and we obtain V • D = 0 on a fixed, rigid solidfluid surface (5.36) Now. consider a solidfluid surface. each element of which is in motion relative to the fluid.· We represent the surface by the equation
F(r. t) = 0
Boundary Conditions for an Ideal Fluid
Physical conditions that .should be satisfied 6n given boundaries of the fluid are known as boundary conditions. There are several types of boundaries and as such there are various possibilities for the boundary conditions. We consider.two types ofb9I,lOdari.es. (I) a solidjiuid boundary. where the fluid is bounded by a solid surface' and (2) afluidjiuid boundary. where the fluid is bounded by another fluid or the same fluid in a different state of motion; this boundary is usually referred to as a free s Irface or a free buulldary. The nature and !lumber of the boundary conditIOns depend, 1;0 on the assumptions made with regard to the nature of the fluid. more specificaHy
where F(r, t) is a scalar function of position and time. The normal to the surface at any point oh the surface is then given by
D
=
± grad F Igrad FI
• For example. such a surface may be that of a deformable solid body which IS either stationary or moving through a flowing fluid. Similarly. the surface may be that of a rigid solid body moving through a fluid;
191
IdealFluid Aerodynamics
D,
Eulerian Equations for the Motion of an Ideal Fluid
193
With this relation for
the boundary condition (5.35) becomes _ (5.37)
V • grad F = V s . grad F on F(r, t) = 0
This condition (5.37) may be transformed further irlto a more convenient form. To do this we first show that Vs· grad Fis equal to oF/ot, the local rate of change of the function F(r, t). Consider the scalar field F = F(r, t) and imagine the surface particles of the solid to move in such a field. At eacn Instarit of time, the surface of the solid is given by the surface formed by those points that have a value of F equal to zero at that instant.· Thus if we move with the surface particles of the solid, we observe no Change in the function F(r, t). In other words, the change or, equivalently, the total time rate of change in F(r, t) foilowing a surface particle ofihe solid is zero. Since V s is the velocity of any surface particle of the solid, we therefore obtain
on the tangential component of the relative velocity between the ftuid and a solid surface bounding the ftuiq. This means in an inviscid fluid, at a solidftuid boundary, the tangential component of the relativfvelocity may assume any value that is consistent with the solution ~f the flow field obtained on the basis of other specified conditions. The fluid thus may slip past the solid boundary. This is in contrast to the situation in a real viscous fhlid. For a viscous fluid, on the basis of experience, we require (for the most part) that at a solidftuid boundary, the relative velocity should be zero at a.1l times, that is, both the normal and tang~ntial components of the relative velocity should vanish.
CoNlitioll .t tJ Free S",/tICe, Kinematical conditions similar to those applicable at a solidftuid boundary should be satisfied also at any surface that fOrulS the boundary between two different immiscible ftuids or between two different states of motion of the same ftuid. We describe such a surface (which may be deformable and moving), as before, by the equation
or
 + V s • grad F = or
V s . grad F =  
of
0
Fr, t) = 0
(5.38)
Let V 8 denote the velocity uf a y element of the surface, and let V 1 and V t denote the velocities of the aUld on either side of the surface. Requiring that there be no fluid flow across the surface, we obtain the condition
Vl
•
of ot
Using this relation, we express the condition (5.l7) as
D
= V8' D
= VI· D at each point on the surface F(r, t)
= 0(5.41,>
of = Dt ot
DF
+ V • grad F =
0
on
F(r, t) = 0
(5.39)
If the solidfluid boundary is formed by the surface of a stationary rigid solid, the equation. of the boundary becomes
F(r) = 0
where D is the normal to the surface. Expressing n in terms of grad F, we have Vl • grad F = Vs· grad F = VI· grad F on F(r, t) = 0 (5.42) By using relation (5.38), tl1is condition may be expre.ssed by the equations
and the boundary condition (5.39) reduces to
V • grad F
of
(5.40)
at + V grad F = O}
l •
=0
on
F(r) = 0
and
on
F(r, t) = 0
(5_43)
The boundary condition at a solidfluid boundary is purely a kinematical condition. The condition that we have formulated refers only to the normal component of tile relative velocity between the fluid and the solid. Nothing has been said about the tangential component of the relative velocity. For an inviscid fluid, the one we have assumed, there is no a priori physical requirement that may be used to stipulate a condition
According to the function F(r, t), a vallJe of F is attached to each point of space at any gi~n in'stant. However. only certain points at that instant will ha~!! a value of F equal t_o;z\!ro_ A surface drawn through such points gives at that instant the position of the surface of the' solid.
of + VI' grad F =
at
0
If the surface is fixed in space, V 8 is zero, and this surface condition reduces to (5.44) V l • D = V 2 • Q = 0 on F(r) = 0 !n addition to this kin~matic condition, a dynamic condition has to be satisfied at a free surface. This condition is formulated as follows. Consider any infinitesimal element DdS of the surface and let PI and pz denote the pressures on either side of the element. We assume that the fluid is inviscid. as before. and that there are no cohesil'e forces (such as
194
Idealfluid Aerodynamics
Eulerian Equations for the Motion of an Ideal fluid The function '" should be such that div ",V = 0 Now, according to the condition of incompressibility we have
195
those arising from surface tension). Then, since the surface element is of zero thickness, the condition for dynamic equilibrium of the element becomes It therefore follows that at a free surface
divV
PI = P2
(5.45)
",(r, t)
and express V
=0 =
I
at every point of the surface at any time. The. kinematic condition at a free surface refers, on either side ,of the surface, only to tne normal component of the relative velocity between the fluid and the surface. As in the case of the solidfluid boundary no a priori condition can be stipulated for the tangential component of the relative velocity on either side of a free surfate. It thus follows that for a inviscid fluid the tangential compohent of the fluid velocity at any point on one side of a free surface is not equal to tlte tangential component, at that point, of the fluid velocity on the other side of the surface, and that neither of these components is equal to the tangential component of the velocity of the surface at that point. This means the tangential component of the flUid velocity is discontinuous across a free surface. In this sense, a free surface is known as a surface of discontinuity, specifically, a surface of tangential discontinuity. The possibility of such a surface of discontinuity arises solely from the assumption of an inviscid fluid. In a real viscous fluid, although there may be narrow regions of space where the fluid velocity may change very rapidly, a strict discontinuity is impossible.
It therefore follows that in incompressible flow we may set
= grad '1'1
X grad
'1'1
(5.46)
Thus, in incompressible flow, the vector field V(r, t) may be replaced by two scalar fields. tpl(r, t) and tp2(r, t). It is readily seen that for twodimensional incompressible flow, Eqs. (4.26) and (4.28) become respectively
u(x, y, t) =
.E.. tp(x, y, t)
oy
(5.47) ox
o vex, y, t) =   tp(x, y, t)
and
u(r, 8, t)
1 =  0 tp(r, 0, t) roO (5.48)
o v(r, 0, t) =  or tp(r, 0, t)
5.12
Conditions at Infinity
The corresponding relations for axisymmetric 'incompressible flow are obtained from Eq~. (4.30) and (4.32) by setting '" = I. In terms of spherical coordinates R, 0, rp we have
In this took we shall consider mostly fluid flows related to the motion of a body through a fluid which is otherwi<;e undisturbed. In such flows, the boundaries of the flow which are at large distances from the body are regarded as being at infinity. Boundary conditions at infinity are usuallv given by prescribing the velocity and pressure at infinity. We may requir~, for instance, that the velocity due to the disturbance of the body be zero, or at least finite at infinity. A more precise specification of the condition depends on the nature of the j!quations governing the fluid motion under consideration.
uR(R, 0, t) =
useR, 0, t) =   .   tp(R, 0, t) R SIO oR
In terms of cylindrical coordinates x, r, rp we have
0 tp(R,O, t) R SIO 00 1 0
1
I'
°

(5.49)
°
0
S.B Stream Functions for Incompressible Flow
We had seen that a flow field may be described !:y two ,tream functions 'PI(r, I) and tp2(r, I) that are connected to the velocity field V(r, I) by the relation (see Section 4.9) fl(r, I)V = grad 'PI x grad 'P2
u.,(x, ur(x,
T,
10 t) =   tp(x, r, t) r or
~5.50)
T,
t) =   
1
tp(x, r, t)
TOX
196
S.14
IdealFluid Aerodynamics Vector PG'leotial Cor Incompressible Flow and Its RelatioD to tbe Stream Functions
Eulerian Equations for the Motion of an Ideal Fluid In terms of cylindrical coordinates x, r,
fjJ
197
we have t5.56)
Since the dive.rgence of any curl vector is identically zero, we may express the velocIty field in an incompressible flow for which div V
1 A(x, r, t).= e.  'f(x, r, t)
r
== 0
(5.51)
S.lS
in terms of another vector field A(r, I) such that
Elimination of tbe Body Force from the Equation of Motion for a Certain Incompressible Flow Problem
V
== curl A
We speak of A as a "vector potential" (see Section2.42). Such a vector potential is naturally related to the streain functions. From Eqs. (5.46) and (5.51) we have Curl A
=
V
= grad 'fl x
grad 'f2
(5.52)
Consider twodimensional motion. Using Cartesians we have
V = (u(x, y, t), v(x, y, 1),0]
Consider the problem offlow due to a body moving through an infinitely extending incompressible homogeneous fluid, ,a problem of most interest to us. Let uc; assume that initially the fluid with the body immersed in it is at rest under the action of body forces f per unit mass. The fluid is later set into motion by lettiog the body move through it. The body forces continue to act on the fluid. Denote by Pia the pressure when the fluid is at rest and by p the pressure when the fluid is in motion. For static equllibrium of t~e fluid we require that·
• Note that Eq.
(~.57)
shows that r should be irrotational.
'PI = 'f(x, y, t)
V'z = z
pf = gradp" For dynamic equilibrium of the fluid Vie require that DV.  gra d p p= I Dt
f +"
(5,57)
L.et the components of A be denoted by A"" Av ' ~z. Equation (5.52) then YIelds the relations
(5.58)
oAw 0'1' ==U=oy oz oy oA", oA. 0'1' oz  ox = v =  ox oA  oA", 0 = ox oy
~Az
II
(5.53)
where, for generality, we hav~ included a term f" to represent the yiscous force on the fluid element. Using Eq. (5.57) and noting that p does not vary when the flUId moves, we may eliminate the body force from Eq. (5.58) and obtain
p
EY. =
Dt
_ grad (p  p,,)
+~
satIsfied by having
It is readily verified that since A = A(x, y, t) only, the system t 5. 53 ) is
"
(5.59)
A(x, y, t)
== (0,0, V'(x, y, t)) =kY;
=
kV'(r, 0, t)
= gradp'
(5.54)
+ f"
In cylindrical coordinates r, 0" and z, Eq. (5.54) takes the form
A(r, 0, I)
J'l a similar way the relation between the vector potential and the stream function in axisymmetric incompressible flow rna)' be established. In terms spherical coordinates R, 0, rp., we obtain
o.f
where p' is the difference of the pressure in the dynamical situation from the hydrostatic pressure Pli' It thus follows that for the problem under consideration the body force may be dropped from the equation of motion if the pressure in the resulling equalion is regarded as the d(fference of the actual pressure from the hydrostatic. The effect of the body force may be ,accounted for independently by solving the corresponding static problem. It should be borne in mind that the present conclusions are strictly valid for the particular problem considered here.
1 A(R, 0, t) = e" '. V'(R, 0, t)
R
Sin
0
(5.55)
Alternate Forms of the Equations
199
Chapter 6
An equation such as (6.1) is usually referred to as the equation of conse~ vation of the quantity G. We shall now specialize this ~quation to ob~m the differential form of the socalled conservation equations (or the motIon of an inviscid fluid.
6.1 CODSenation of ~. . We choose a small element of volume tJ'r fixed in space afld encl~sed y the surfaCe 6S. To set up the equation of change for the· mass contamed 10 tJT we assume that the changes in that mass are brought about solely by inflow of mass through tJS. Then Eq. (6.1) takes the form
Alternate Forms of the Equations
h.
In the preceding chapter the differential equations that govern the motion of an inviscid fluid were developed from the point of viCl1W of a moving fluid element. We shall now derive an alternate but equivalent form of these equations from a different point of view, namely that of a definite region of space enclosed by a surface fixed in the flowing fluid. Following this, we 'shall obtain, from the two points of view, integral forms of the equations. We begin by formulating an equation for the change of any fluid quantity contained irta fixed region of stlace.
6.1 Equation of Change
ap~ __ ... Jj'pV. n dS
.It
at .
.s
~ee:~aoe
V is the socalled mass flux .vector and D dS is an elemental area on 6S, n being t~e outward normal: ~in~ 6T is a fixed v~lume 't may be taken out "of t differentlatton 10 the lefthand SIde of eIemen t , I , I (6.2). Also, for an infinitesimal volume element, the s~rface mtegra on the right~hand side may be replaced by lJT div pV. With these changes (6.2) may be rewritten as lJT..1!. "'" lJT div pV
h
a
at
Consider a certain region R fixed in the space occupied by a flowing fluid and bounded by a surface S. Let G denote the amount of any fluid quantity, such as mass, mOl\1entum, energy, angular momentum, contained in the region R at any time t. We set up an equation for the change in G by equating aGlat, the rate of increase of Gin R, to the sum of all the changes per unit time in G due tt> different causes. One of the causes by which a change in G may occur is that of production of G in R. When this happens we sa): that there are sources of G in R. We shall assume that there are no sources of G. Another cause by which an increase in G will occur is that of net transport of G into the region R by the fluid flowing into and out of R through the bounding surface. We shall refer to this increase in G as the (net) rate ofinflow of G into R. Changes in G may be brought about by. causes other than transport by the flowing fluid. Depending on the actual quantities under consideration, s~ch changes are a.ccounted for by certain laws ofmec;hanics and thermodynamics. Calling such changes as "other changes" we express the equation of change for G as
oG
or, writing it per unit volume, as
ap =
As we have already Seen (Section 5.3), this .l"orm of the equation of conservation of mass is equivalent to the equatIon Dp
Dt
at
_ div pV
(6.3)
+ p divV =
0
which was obtained from the point of view of a moving fluid element.
at
=
rate of inflow of G into R
6.3 CODSenation of Momeatum To write the equation of change for the momentum of the ~uid, besides the rate of change,due to inflow of momentum, one must take,mto ac~ount the rate of change of momentum of the fluid contained at any Insta.nt In the fixed volume 6'T due :to the body and surfaces acting on that flUId at th.e instant considered. We denote, as before, by f the body force per umt mass of the fluid and assume that the fluid is inviscid. Then the total force acting on the fluid in O'T is simply
pf O'T 
+ other changes of G per unit time
198
(6.1)
if
6S
pn dS
100
IdealFluid Aerodynamics
Al~ate
Forms of the Equations
101
As before, p is the pressure. The net rate of inflow of momentum into the volume 6T is given by  ffV(pV n) dS
0
The vector div (pVV) may further be expressed in a form that contains only familiar vector operations. Thus we observe that div (pujV) From this we obtain div (pVV)
(6.4)
== ==
pV grad U i
0
+ uidiv pV + V div pv.
(6.7)
IS
Therefore, for the momentum of the fluid. the equation 'of change (6.1) takes the form
pV grad V
0
apv «57' t( at = lrV(pV a) dS
0
it + pr «57' .lr pn dS
18
(6.5)
Using relation (6.6) and remembering that for an infinitesimal volume element
68
The term. in (6.5) representing the rate of inflow of moinentum may be expressed in different forms. To show this we introduce a Cartesian system with el • el, es as the reference unit vectors and form the ith component of the term (b.4). W: thus obtain C,offV(pV n) dS ffUj(pV n) dS
0 0
ffpn dS
.8
;=
«57' grad p
Eq. (6.S) may be expressed~ per unit volume of the fluid. as
apv ....div (pVV) at
+ pf " grad p
(6.8)
.8
.S
!f(PUiV) a dS
0
18
where U j is the ith component of V. For an infinitesimal volume element we have
This equation is usually referred to as the equation of conserva~ion of momentum. The dyad4c pVV is called the momentum flux tensor. USIng the relation (6.7) and the equation of change for mass (6.3), Eq. (6'8) may Qe reduced to the form
,{f(PU;V) adS  «57' div (pu/V)
0
p(~~ + V
0
grad V)
=
pf  grad
p
•8
which is the equation of motion for a moving fluid element .
It then follows that
fiV(pV
18
0
0)
dS = fJT[e l div (pu1V)
+ ~ div (pu.V) + ~ div (PIl'aV)]
Introducing the notation that div(pVV) represents a vector· such that c/o div (pVV) we write ffV(pV a) dS = «57' div (pVV)·
0
== div (pujV)
(6.6)
6.4 CODSe"ation of Energy To write the equation of change for the energy of ~he fluid, besides the rate of inflow of energy, we must account for additIOn of energy to the fluid in the fixed volume {JT due to several other pr~cesses .. We assume, as before that the fluid is nonviscous and nonheatconductIng and that no proces~s other than the work by body and surface forces take place. T~e rate at which work is done by the body and surfaces forces on the flUId contained in 6'1' is
6'1'pf
0
68
• A product such as AB between two vectors is known as a dyadic prodMct. A dyadic product ~B is a tensor. the elements of which are given by
A,BJ
V ,{fPV
68
0
n dS
The rate of inflow of energy into the volume is
where A, and BJ are ith and jth components of A and B, respectively. The subscripts refer to the directions of the axes of the chosen coordinate system and may take any of the values I. 2. 3. The divergence of a dyadic product is a vector.
202
IdealAuid Aerodynamics
Alternate Forms of the Equations
20J
where e, as before, is the internal energy per unit mass of the fluid. Therefore, for the energy of the fluid, the equation of change (6.1) takes the fomi
(3) Energy
:, p( e + ~I) ~ == {fp( e + ~I)V' D dS
's
:t fff p(
R
e
+ ~I) dT ==
{fp( e + ~I)V' n dS
8
+ ~pf. V 
+
{f pV • 0 dS (6.9)
fff
R"
pf. V dT  {fPV' n dS
8
(6.13)
Recalling that for an infinitesimal volume lJT {fA. D dS lJTdiv A
'8
's
Eq. (6.9) may be exp~ssed, per unit volume of the fluid, as
a (VI) . ( VI) a; pe + 2" == dlv pe +. 2" V + pf. V 
Equations (6.11) to (6.13) are referred to as the integral form of tire conservation equations for an Inviscid fluid. In the lefthand side of each of these equations, the partial derivative may be taken inside the integral'sign for· R is a fixed ·volume,. The integral relations (6.11) to (6.13) may be reduced to the differential form by assu~ing that the functions involved !have sufficiently many derivatives in the region R and letting the volume of the region R tend to zero or, as is usually done, by applying the general integral relations
a/at
div pV
(6.10)
This equation is usually referred to as the· equation of conseroation of energy. The vector pee + (VI/2)]V is known as the energy flux vector. Using the equation of change for mass (6.1), Eq. (6.10) may be reduced to the form
{f~ dS
S
8
iJA. DdS
"Iff == fff
R
R
grad 4> dT div A dT
pDt + VI) == pf. V ..... div pV '!!"(e 2
wh.ich is the equation of energy derived from the point of view ofa moving flUId element. . 6.5 . Integral Fonn of the Equations ftom the Point of View of a Fixed Region of Space In the preceding sections we obtained the differential form of the equations of change for mass, momentum, and energy by considering. a small fixed volume of space in the flowing fluid. Instead of a small region we now consider a finite region R of space enclosed by a surface S fixed in the flowing fluid. The equations of change then take the following form: (I) Mass
6.6 Integral Form of the Equations from the Point of View of a FInite Fluid Region In the last chapter we derived the equations of mass, motion, and energy from the point of view of a moving fluid element. If, instead of a fluid element, we follow a finite fluid region, we obtain these equations in an integral form. By a fluid region we mean a region that is composed of the same fluid elements for aI/ times. Sl,lppose that at some instant of time we mark out a certain region R of the fluid as that enclosed by a surface S. If the surface moves with the fluid such that it always passes through the same fluid elements, it will always enclose the same fluid elements. Thus the surface S is a fluid surface and the region R it encloses is a fluid region. In general, the shape and size of a fluid. region change as the region moves with the fluid. Now, applying the law of constancy of mass, the second law of Newton, and the law of energy conservation to a fluid tegion R enclosed by a fluid surface S, we obtain the following equations:
(I) Mass
:, fff
R
S
p
3T ==
(2) Momentum
{fPV' DdS s
(6.11)
fr
fII
R
pV ciT
=
{fV(PV' D) tiS f..
Iff
R
pf tiT
{fPO
S
dS
(6.12)
(6.14)
1fU
IdealFluid Aerodynamics
Alternate Fonns of the Equations
}05
(2) Mo;;on
~t
(3) Energy
Rill
IIf
pV dT ==
Rill
III d~ ffpn
pf
8111
dS
(6.15)
where, as before. Q == Q(r. t) denotes per 'nit mass of the fluid any quantity associated with the fluid. Another way, which is more direct and simple. is based on the foHowing interpretation of:the material derivative of
fff
R(I)
pQdT.
~tJII p( e + ~I) dT ==
Rill
III pf. ffpv.
V dT
Rill 8(1)
DdS
(6.16)
6.7 Rate of Change of'. Quantity Following a Fluid Region
Let 5 denote the bounding surface of a certain fluid region at ttie time (Fig. 6. I). In a short time intervallJI thereafter the different fluid elements composing the fluid region ~ove to new locations. Let 5' denote the bounding surface of fhe fluid region ·at time I + {)t (Fig. 6.1). Denote by
I
The notation R(I) and 5(1) are used to emphasize that we are dealing with a fluid region and a fluid surface. In each of these equations the lefthand side represents the rate of change of the total amount of a quantity associated with the fluid that is contained in the fluid surface 5 at some instant; the rate of change is computed following that portion of fluid, that is, following the flu~d surface 5 or, equivalently, the fluid region R. Such a rate of change may generally be represented as
:~III pQ dT
Rill
(6.17)
where Q == Q(r, I) is any fluid quantity (scalar or vector) specified per unit mass of the fluid. Now, although the volume R(I) changes with time, the mass contained in'that volume is a constant for all times. Therefore we temporarily change the volume integral in (6.17) into an integral over the mass by writing p dT == tOn and obtain .!!.fffpQ dT
R(1l
•
Dt
==.!!.
Dt f
Q dm ==
mu.
mu.
f J!. Dt
(Q dm)
Fjg. 6.1 Illustrating the computation of the rate of Change of a quantity following a fluid region.
inR(t)
inR(t>
Since the mass dm of a fluid element is a constant for all times, this relation becomes (6.18)
R the volume enclosed by the surface S and by R' the volume enclosed by S'. If Q == Q(r, I) denotes per unit mass any quantity (scalar or vector) associated with the fluid, by definition we have
~tIfI pQ dT == 1:~~'[fIf pQ dT  fff pQ dTJ
RII) R' R
By using this relatiol.1, the lefthand side of each of the equationc: (6.14) to (6.16) may be represented as a volume integral. The equations obtailled in thiS section are equivalent to the equations obtained from the point of view of a fixed region of space (see Section 6.5). One way to show this is by means of the !eneral relation I p DQ
where. as before, R(I) denotes the volume of the moving fluid region. We introduce temporarily the notation
G(R,
Dt == .! ot
(pQ)
+ div PQV
(6.19)
G(R', I
t) == Iff + 1St) == Iff
R
pQ dT pQ dT
R'
IdeaPFluid Aerodynamics with the obvious meaning that G represents the total amount of the quantity cO:lcerned in the volume under consideration. With this meaning for G, and referring to Fig. 6.1, we write
G{R',I
Alternate Forms
or the
Eq~tions
201
6.8 Eqaatioas of CIIMge for _ Ideal fluid
For an ideal fluid, the density and internal energy are constant for all
times and the equations of change take, the following forms:
(1) Differe1ltial form of the equations (see Eqs. 6.3, 6.8,6.10, respectively)
Mass: div V  0
+ .5/) == G(R, t + .51) + G{A, 1 + t51) G{R, t
G{B,I ... .5t)
Now, to the first order in .51, we have
+ .5t) == G{R, t) + [aG{R, t)]
and
G(A, t
at
dt
Rflxed
Momentum:
p av
+ .5t) 
G{B; t
+ .5t) == the net outflow, during the time .5t, or G
through the surface S fixed' in the flow field
at
... _ p div (VV) + pf  grad p
_  pV grad V
0
+ pf 
grad p
Energy:
=.5t It therefore follows that
If
p 'VI _ pdiv '(VI V) 
at
a
2
2
. + pfoV dlv pV
0
'
Q{pV n) dS
0
Sflxed
_ pV grad 0
Vi
2',
+ pf. V 
V grad p
~tIII pQ dT == :rIII pQ dT +
RW R fIxed
=
III a~;
R.
dT
If + If
S fixed
(2) 11Itegral form of the~fIUltions (see Eqli. 6.11; 6.12, 6.13, respectively)
Q(pV. n) dS
Mass:
Q{pV • n)
as
(6.20)
Momentum:
8
This equation expresses the significant result tha.t the material rate of change of the total amount of any quantity associated with the fluid which at any instant is contained in a surface S fixed in space is equal to the sum of two parts, (I) a part that is the local rate of change of the total amount of the quantity in the fixed volume R enclosed by the surface Sand (2) a part that is the net outflow per unit time of the quantity concerned through the surface S fixed in the flowing fluid. The differential form of (6.20) is simply
p
III ~~
R
d" 
p§V(V. D) dS
B
+
pIIf, ifpa
dT R B
dS
Energy:
p
III :t(~I)
R
dT  p
~ ~I V 'DdS
B
+
6.9 Rate of
pffffoVdTR
if
Ij
,V.adS
p DQ
Dt
== apQ + div pQV
at
CIIaaae of a QaaatIt.y F....... a Mmag Rep. of s,.c.
which is Eq. (6.19} Using (6.19) and (6.20) we may readily show that the equations derived from the point of view of a moving fluid element or fluid region are equivalent to the equations of change (the socalled conservation equations) derived fr.om the p0int of view of a fix~d, infinitesimal or finite, volume of space. In particu!3r, we observe that the equation of Change for the momentum is indeed an expression of Newton's second law of motion.
In Section 6.7 we considered the totaltilht rate of cIlange of. a quantity following. a fluid region. We nOw inquire' into the total ti~e of change of a quantity foUowi8g a region that niovts, with an arbitrary velocity different from that of th& fluid. In other words, let, us consider the situation in which the moving region is not a fluid region. , Let S denote the surface of a region R v.:hich moves through a scalar 0,' vector field A == A(r, t). Let the velocity of an element of the surface S tw.
rate
,IdealFluid Aerodynamics
Alternate Forms of the Equations
denoted by~. In general ~ varies over S. The region R and the enclosing surface S are functions of time, and they do l10t in general retain their shape or size: R == R(/) and 5 ~ 5(/)
to first order in
~t.
With these results it follows that
A dT
:t III
, RId
== ==
III ~~dT + if A~
Y. S
•• d5
~e ask. fo~ the total time rate of change of the quantityff.JA(r, I)
Tbjs rate
IS
glvea by
dT.
for the material
B(I)
ifI ~~
R
dT
+ § A~ •• d5
S
(6.22)
If the region is a fluid region, the surface Moves With the fluid; ~ equals the fluid velocity V. Equation 6.22 then redUces, as it should, to (6.20)
RW
RUHII
RW
derivativ~ of
III
A
dT.
Denote by V. the volume of space occupied by R at time I and by VI that occupied by R at time I + Ill. We have
B(I)
RUI
_
U
Illustrating ,be rate
III A  Ifr~(r, dT dT IIf Ad,. III A(r, t + 6t) dT
t)
Y.
PI
quaDtity rollowing • moviDg regiOD.
or cban&e or a
RCc+'I)
For simplicity, suppt>se that at time tthc velocity ~ is directed outward all around the surface of R; then VI encloses V. as shown in Fig. 6.2. We write
Iff A(r, t + Ilt) dT  III A(r,
VI Y.
1
+
Ilt) dT
+
IIf ...4(r, t + Ill) dT
YlY'
Furthermore, we have
IIf
V.
VIV,
A(r, t
+
Ill) dT ..
IfI[
V.
A(r, I)
+
~~ 61] dT
tofust order in 61 and A(r i ,
+ Ill) dT _
.
fIJ

the ~mount .0fA swept by the surface of R dunng the time Ilt
Ilt times the rate at which the surface of R sweeps the field anhe . mstant t
D
... IJt(ffA~. dS)
S
Equatioos of DiIcontinuous Motion
Jll
Chapter 7
Equations of Discontinuous Motion
In Chapter S we saw that at afree surface the tangentialcomponent of the luid velocity is discontinuous across the surface. In this sense the motion is discontinuous and the free surface is a discontinuity surface. In general,
a discontinWty surface (or silnply a disc01,llinuity)ls any surface across which the motion is discontinuous, that is, «ross which jumps '"to!. occUr in any 0/ fire fluid properties. We now' investigate in a g~neral way the type of
discontinuous motion that is possible in the low of an ideal fluid. For this, purpose we assume that a discontinuity is present in the flow and inquire into the relations that must be satisfied by the fluid properties across the discontinuity io that the motion obeys the basic Iaw~ f fluid flow. The discontinuity surface may be stationary (i.e., fiXed ltlative to a fixed observer) or it may be in motion, the velocity of any element of the surface being in general different from that of the luid on either side Qf it. We shall consider both tf!ese cases. In the following we treat first the motion of an ideal fluid, and in that context we sball assume that the density and the internal. energy are.:onstant throughoUt the flow field for all times. The case of an inhomogeneous incompressible inviscid fluid is then considered.
.... 7.1
s~ diaioDtiDbity iD 1teaCl111oW.
, For this purpose we c~se. as shown in Fig. 7.2. a region R that is fixed in space ano contains the discontinuity such that the discontinuity separa~ R into two parts. According to the integral fon'll of the con.: servation equations for an iclea1 fluid (see Section 6.5),we bve
Conservation olmass
,j} V • D'dS 8
0
(7it)
7.1 A Stad*" DIIcoatIuIty Ia • SteMy Flow Let the sdce D denote a stationary dUcontinuity in a steady flow field
(see FIg. 7.1). We designate one side of the surface as the positive side and the other as the negative side and intrQCiuce the following notation
Dl denotes the positive side of the surface D, denotes the negative side of the surface .. dS denotes any surface element on Dl Ii.. dS denotes the corresponding surface element on D, VI and PI denote velocity and pressure at all)' point on Dl V, and p, denote the velocity and pressure ,at the corresponding point
Conservation of momentum
p,{fV(V. n) dS;'" if 'PD dS + pIffr d'T
8 8 8
Co~ervation
(1.2)
of energy
pff~' (V. D) dS ,8
the
if
8 .
pV· D dS
+
p f· V d'T Iff
R
(7.3)
onD,
. ~ problem is to derive, according to thebasic laws of fluid motion, the ,relations between VI,PI and V.. P'~
where S is the fixCcl sunace enclosing R. , The boundary of the discontinuity marks out the surface S into two distinct parts, whic •• ~e denote by '~l and ~. ' Introd~cing t& and 1:.. we express each of tile surface integral'ln the Eqs. (7.1) to (7.3) as the sum of
'lTO
IdealFluid Aerodynamics two surface integrals:
Equations of Discontinuous Motion
llJ
fi(
B
Now, because the relations (7.4) to (1.6) hold for.any arbitrary portion of the surface D, they reduce to the following co~ditions: )dS 
ff(
1:1
)dS
+ ff(
1:,
)dS
COffServatlon' o/mass
(VI  Va> ·n,z 0
or' (7.1)
The conservation .equations are valid nO matter ~ow small ~he regio~ R is. We therefore perform a limiting process by lettIng the regIOn R shrmk to the discontinuity surface D;. In such a limiting process ~1 tends to D1 ,
VI·D. 
V  .t.say,. ....
(7.8)
Conservation Conservation
0/ momentum 0/ energy
2
PA(VI  Va>  (P.  P~ AV.I)  (i.P.  API)
e(lVI •

(7.9)
TheSe conditions must be satisfied at eau:h point of the discontinuity. Equation (7.7) states that (V I  Va> is perpendicular to D. whereas Eq. (7.8) states that if A" O. (VI  VJ is parallel to D". Therefore it follows that the solution of (7.7) apd (7.8) is
1 VI' a. == V.· .. = 0
PI·P.
(7.10)
(7.11)
Equation (7.9) is then automatically satisfied. From (7.10) we infer that
Flg.7.2
Fluid region containing a ui~ontinuity.
there is no jlow acron the sur/ace ulid thilt the tangential compone,!t' of the .fluid velocity is discontinllOus arross ~t; in other words, the surface D is a tangential discontinuity. We, thus conclude that only a tangential discontinuity is possible in the steady jlow of an idealJIuid.
7.2 A MoYing
~i tends to D 2 , and the volume in.tegrals in (7.2) and (7.3) tend to zero.
. Thus we 'obtain the following relations:
DiscoDtiaaity iD
tile :UDSteady
Flow of aD Ideal Flaid
Consert'4tion of ma.~s
ff(V 1  V2)· 0lt dS = 0
Tl
(7.4)
Conservation of momentum
p
If
[V1(V 1 • °tl)
(v 2(V f' Dd)] dS =

f/(Pl  P~)Od dS
(7.5)
.Jl C;onS4Na1irm
ilJ
p
of ellergy
CV1J('\I; • tid) 
V22!V~. Olt)] dS =  ff(P1V,  'P2Vt}' na dS
v
(7.6)
We ·now take up the more general case where the discontinuity surface is mov;"g through the ftow field which is unsteady. Let l; denote the velocity of any element of the surface; This velocity may, in general, be nonuniform over the surface. We.'use the same scheme as before to de~ribe tile two SIdes of the discontinuity and the ftuid quantities on its' two sides. We choose, as shown in Fig. 7.2, ajluld region· R that contains the discontinuity D such that it separates R into two distinct parts, Rl and R I • Also, the boundary of D marks out the surface S enclosing K into two distinct parts, l:1 and ~.. We note that R, R 1 , R., S'~I; ~. are all functions of time. Now, according to the integral form of the basic equatiDns
• Note that now R is not fixed in space~ being a ftuid region it moves with the.
EquatioDSol Discontinuous MOtiOD
aoveming the motion of an idtill fluid (see Section 6.6), we have
velocity of any element. olthesurfacc 0(/). each of the Eqs. (7.1.2) to (7.14) is of the form (7.12) of cha~ge. of we wrIte
dS
Now. the left.handside of
Condition o/l!rcompreuibillty
.
:tflldTO
Rill
:tlffQdT
ftW
MIUI
where Q may be I or pV or P(V1/2). This is· silJlply the total time rate
:t Iff p p IfI :t
dT Rill RUI
dT  0
nu. reduces to Eq. <7~2).
Motion
p~f.IJv dT Dt
..RCtI
,IfffdT if padS .  pIIf +fl dS ~ ff". dS
RCCI Bu)
Using Eq.
f dT
pa
(7.13)
Rtf)
EJUl
Elcd
E1wrV.
:t IffQ :t fIfQ + : fffQ (7. If» S1(t) :tIfIQ = III~~ +II Q~. dS = Iff ~~ +If dS II Q1~ ..... dS(7.17)
Q dT
fff ~t fff
B(t)
Q dT following the timedependent region R(t). Hence
==
dT 
dT
dT
Rill
. Rill·
RIft)
t Raltl
(7.l~
and denoting by
the surface enclosingR~ we.obtain
dT
dT
D
RlltI
R1W
BIIII
dT
QV • D
.~ :tJff Vi PTPIfff· VdT 11'"
_pIlI :tIII(
RUI
Bu)
if dS
pV· D
Rllt)
Il(1I
DIIII
where Q1 signifies a fluid quantity on the positive side D of the .discontinuity. Similarly: we obtain 1 . pV·.
f· V dT :
II
BW
pV • D
dS II
dS
(7.14)
,,~ IIf Q dT = Iff. ~; dT +fI QV •
RaW Rai II I.W
D
dS +IIQ.t; ..... dS
Dtltl
(7.1~)
cUt
mil ~II ~d now Wish to express the lefthand sideo.f each o( these equations in terms of volume and surface integrals related to the regions Rl and R I • For this purpose we note that
We
whe~e ~I signifies. ~ 'fluid quantity on the negative side D. of.tho contlDulty. ComblDlDg (7.17) and (7.18), we obtain the result ..
).dT"
:tIII< .)
RIUI
dT
+
:tfII(
RaU)
~tIIfQ dT  Iff ~~ dT +fI QV· D dS ff Ql~ ...... dS
RCII
R(II II(/)
).dT
/., a1thgugh th. region R is a fluid region, the .regions Rl and ~I are ·not.
This is SO because the surface Dl (or DJ. whIch forms a portIon of the surface that ellcloses the region Rl (or R.). is not a fluid surface.. ~n vie~ of this we proceed as follows. We recall (see Section 6.9J that If pet) IS a (m.oving) timedependent region of space in a scalar or vector field Q(r, t), the total time rate .of change of the integral the rqion P(t) is given by
• Although
+
II dS +IIQI~ dS
QV • D
.....
DI(I)
(7.19)
IaW
~
IfI
R(t)
QdT"
~ IfI QdT
R(t)
ffI
1'(1)
Q(r, t) dT following
:, fff J4:, fff
QdT
nl(!)
QdT
R,(t)
~ffI Q dr = Iff ~~ dT + ~ Q~. DdS
1'( II
/'U)
,,(II
f;:>r RI(t) is not a fluid region. Similarly.
.
(7.15)
~ Iff QdT ~
RI(I)
:t ffI
RI(I)
QdT
where air) is the timedependent surface of the region pet) and { is the
216
. IdealFluid Aerodynamic:s
Equations of Discontinuous Motion and that
211
Now, we rewrite the left':'band sides of (7.12) to (7.14) in the form expressed by (7.19) and perform a limiting proce&s in which the region R is allowed to shrink to the discontinuity surface D. The surface~1 tends to D 1, the surface ~.tends to D., ~nd the volume integrals tend to zero. In this way we obtain the following relations:
Condition of incompreuibility
(7.27)
II[(V1  ~ "~(V. Motion'
p
~.~] dS == 0
(7.20)
.D
Equation (7.25) is then automatically satisfied. Equations (7.26) and (7.27) show that there is no ftow across the discontinuity surface, that the tangential component of the fluid velocity . is discontinuous across this surfaCe and is" different on either side of the surface from the tangential component of the velocity of the surface. Thus the discontinuity is a tangential discontinuity. We therefore conclude that in the motion of an ideal fluid only a tangential discontinuity (whether stationary or moving) is all that is possible. For a stationary discontinuity, that is, for ~ equal to zero, (7.26) reduces to
VI· Del
II {VI[(VI 
~ .....] 
V.[(Vi, 
~. aJ} ds
.a
I?
Energy
If<Pa  ,va. as
D
== V.· nel == 0
(7.21)
which is precisely Eq. (7.10) derived in the previous section. Equation (7.10) applies to a moving discontinuity if we replace the fluid velocity V by the relative velocity (V  ~ between the fluid and the moving dis. continui.y.
7.3 Disc.ontinuity In "the Flow of an Inhomogeneous Incompressible Fluid
~ If {Vll[(VJ  ~ ·.1 D
V.·[(V. 
~). aJ} dS
==
fI(P.V.  P1
r;
V l).
~ dS
(7.22)
Since the relations (7.20) to (7.22) hold for any arbitrary portion of the surface D, they reduCe to the following conditions:
Condition of incompressibility (Vl ~).~ .. (l1:a ~.~  A, say
or
(Vl V.)·.DeI Motion Energy

(7.23)
0
(7.24) (7.25)
Wt; now investigate the possibility of discontinuou~ motion in the flow of an inhomogeneous, incompressible, inviscid fluid. We consider a moving discontinuity in the flow of such a fluid and adopt the same notation as in the foregoing section to describe the necessary details of the flow field (see also Fig. 7.2). In the motion of an inhomogeneous, incomp'ressible, inviscid fluid, the density and the Internal energy vary, in general, with time and position although the volume of any fluid region is preserved for all times (recall Sectton 5.8). The equations governing the motion of such a fluid region R(t) containing the discontinuity (Fig. 7.2) are
Condition of incompressibility
~tJJJdT == 0
R(t)
Mass
These conditions 'must be satisfied at each point of the discontinuity. From (7.2~) and (7.24) we infer that
1 <':1
or

~.
Del 
(VI  ~. ~
== 0
Motion
(7.26)
Ii; JjJpV dT
R(t)
=
fff pf ff
dT Ru)
S(f)
po
dS
11'
E1ee.rIJ
1deaJ.;F1ul4' AerOdy1Wnk:S
Equations of
7.4
Discontinu~us
Motion
119
Remarks
!tfff
BUl
p(e
+
~') d.,
JfJ
B(t)
pf·
Vd~ffpV ·adS
8(j1
Wet, ~te
the Icfthand sioe of eac;h of th* equauons 4n the form
~pr:es!'.ed byJiq. (7.1~).ancl~9.nn. the limiting proc:edu"'e. ~. before. in whic:h. the: region .R is allowed to~,* tp the diSCQntiDuily surface D. We thus .obtain the iollowiDg' relations:
In view of the considerations given in this cbap~r we conclude that in the flow of an inviscid, incompressible fluid, the only possible type ofa f~ surface (i.e., a fluidfluid boundary)' is one of tangentialdiscon~nuity. Sharp discontinuities as envisaged here are a conseql,lence of the assumption that the fluid is invi~d (Le., the viscosity if zero) and, therefore, should not be expected to appear in the flow of a viscous fluid, When
Corulilion of InttmIpreuibility
ff[(Vl
Mass
p
D

~)
...... 
(\'.
o· a"J"dS ==0'
112
fJ[Pl(Vl ~,... ';"P.(V. :~)· ....1dS.= 0 .
Motion
/
V'
l
/
N.rrow rIlion thrOUlh wIlich veloclty of the streamch'lI&es from Out ult
_u
IflPi:VJ,(.V~"'" ~60~
D
Plv.(V...
~~)DdJdS"""ff(P. ~.~1I.S
D
III
/
/
/
Energy
rffpl(e + ¥)«V ·'t;)·...,1 PI(" + ~)[(Y.... ~)~ailJas Sf(PIVI : plV1) ·D4 dS
1
1
Fig. 7.3
Example of a region of vorticity in a real ftuid.
Since these relations hold. for any arbitraryponion of the ,slid'ace D.*hey ~educe to the.followingcondiqons:
Condition of incompressibility
(VI  ~). n" = (V,  Q. D"
or
=.
0
Motion
PI "'Pl
The equation!; of mass and energy are automatiCally satisfied. We thus conclude that in the motion of an inviscid incompressible .fluid:" 'hetherit ;s
homogeneous or inhomogeneous), a. tangential di$~ontinuity (whether stationary' or mov;,,; viis all tlrat ;., possible.. The. conditions to be satisfied across the discontimiity are giveD in general by Eqs. (7.26) and (7.27)
finite viSCOSity, no matter how small, is taken into account, .a sharp discontinuity is impossible. In the motion of a viscous fluid one, however, observes regions through which the velocity changes rapidly. An example of such a region is illustrated in Fig. 7.3. The region is a region of rotational motion or, equivalently, a region of voticity. In the formulatIOn of theoretical models for real·fluid flows involving such regions of vorticity, one generally idealizes them as sharp discontinuities. In this sense, a discontinuity is a surface of concentrated vorticity and we refer to it as a . wrtex surface or l"ortex sheet. In ideal fluids ['ortex sudaces and discontinuities art! in in fact identic,!l. Vortex surfaces are employed in the analysis of many physical problems, in partIcular, of the problem of lift. We ~hail thus be concerned a great deal with flows involving vortex sheets. There are many properties as~ociated with such flows and we shall learn about them in their proper context (see Chapters 17 to 19). In real flows the narrow regions 0f vorticity originate generally 'thl;
220
IdealFluid Aerodynamics
solid surfaces present in the flow. The socalled boundarylayer region close to the solid surface and the extension of that region after it leaves the surface art usually narrow regions of vorticity. The region of vorticity that extends from the solid surface does not. retain its shape permanently. It usually rolls up and forms socalled discrete vortices. The appearance of such vortices changes the flow field from' thllt associated with the presence of only a narrow vortex region. This should be borne in mind when one formulates the theoretical models of real flows It can be shown· that a discontinuity surface in an ideal fluid is unstable to any small disturbance. By this we mean that if a small ~isturbance, ~n the shape of the surface for instance t is introduced, the disturbance wIll continue to grow with time and thus completely alter the shape of the discontinuity. . We CQnclude this chapter with a lew remarks about the. type of .dlScontinuity surface possible in the motion of a r:t0nviscous, but compressible, fluid. The investigation in this case may be carried out in the same manner as done for the incompressible fluid. We then find that there are two possibie type:; of discontinuities, (1) a tangential disconti?uity as defined before and (2) a ."ormcl disco"tinuity, where the tangential component of the fluid velocity is continuous across the disconti~uity whi~e the normai component of the velocity, the pressure, a~d the denstty (as well as the other thermodynam\c quantities) are discontinuous. Furthermore, we fic.d that, so as to satisfy the second law of thermodynamics, the normal discontinuity can occur only in a certain range of flow speeds (namel~ in the range C'f the socalled supersonic speeds) and only as a ~mp~ss~on diSC()ntinuity, t.hat is, as the socalled shock. Across a shock dlscontinul~, the pressure and density increase in the direction of the flow velOCity measured relative to the di~ontinuity while the normal component of the relative velocity decreases.
Chapter 8
Integration of Euler's Equation in Special Cases
b t!lis chapter we shall consider bricOy the question of integration of the equations that govern the motion of an ideal fluid. First we shall outline some of the mathematical features of the equations and then shall see how under certain drcumstances it is possible to·form an integral of Euler's equation. We shall find that there are two cases when this is possible. (1) When the booy forces are irrotational and th.e motion is steady, integration is possible along a streamline. (2) When the body forces are irrotational and the motion iuelfls without rotation, integration is possible in any direr.tion and for unsteady motion. These results naturally suggest that we inquire into the conditions under which fluid moti9ns are irrotational. This we shall do in the next chapter.
8.1 Mathematical Character of the Eqaatioas
The differential equations governing the motion of an ideal fluid are
(1) Condition o/incompressibiljty
divV = 0 which in Cartesians takes the form (8.ta)
(2) Euler's equation
(8.1)
 +
oV
ot
grad 
Vi
2
~ V x curl V
== f   grad p p
221
1
fB..2)
111
IdealFluid Aerodynamics
Integratiop pi
EII~'I
Equation in Special Cases
113
which in Cartesians is expressed by the three scalar equations:
014
OV av OV Of, 1 OD  + u:...... + v + w It __ .:L
ot
+ u ou "t ,V~ + ,,...f! =1. _!.~
0:1: 011
az
p 0:1:
p
ow a,,· ow iw 1  + u +v + w =1. __ aD .:L
ot
iJx
as
az
•
01/
(8.2a)
ot
0:1:
011
Or
p
az
where u. v. ware the components of V and h.!..J. are the compollents off. The velocity field V(r, t) and the 'pressure field p(r. t) constitute four scalar unknowns. To determine these the system of four equations expressed by (8.1) and (8.2) are to be solved simultaneously together with specified initial and boundary conditions. We characterize these equations with the folloWing remarks. Since the'equations involve mote thanone.independent variablt, they forot a system of partial' differential ~ti(}ns. The iricompressibility condition'invol~$ thTee dependent variables {i.e., the velocity 'components) and three independent variables (the space coordiriates), while the equati'on of motion involves four dependent variables (Le., the three velocity com}Jon~ts and 'flie pressure) with four independent variables (the space coordinates and titne)~ The ihtegratibnot partial differeritialequations is a more difficult problem mathematicaUy tha. that of ordinary differential equations. The order of a differential equatiohis given by the order of the highest derivative that appears in the eauation. Thus all four equations we have are of the first order. A differential equation is said to be I;n~ar if no nonlinear terms, that is, :products or powers of the unknowns and lor their derivatives, occur in the equation. if such products and powers appear, the equation is nonlinear. We notice that the incompressibility condition (8.1) is linear, whereas the Euler's equation (8.2) is nonlinear. The nonlinearity of this ~quation is due to the socalled convection terms in the acceleration: grad or in Cartesians
independent solutions is also a solJltion, and that, therefore. we can build a complete solution by simply superposing various particular solutions. Thisprinciple oJsuperposition cannot be used incase ofa nonliQear equation. Thus ,there is no general way by which, we may atfempt to solve the equations (8.1) and (8.2). In such a situation we may adopt alternative methods involving numerical inteltratioJl or transformation of variables or approximations leading to linearization of the eqtlations,. These pro' cedures, however, will not concern us in our present studies. Fortunately, the problem of motion of an ideal fluid takes on a different aspect than that of solying simultaneously the. condition of incompressibility and the nonlinear Euler's rquatlon. It turns out, as we shall show in the next chapter, that for an ideal fluid under the actIOn of rotation free forces all motions started from a state of rest are permanently irrotational. This is very significant mathematically. When the angular velocity field is permanen~ly zero, curl V is zero and the term V x curl V in the acceleration vanishes leaving the e4'lation of motion in a form that can be integrated once and for all (see equation below). The velocity field when irrotational may be replaced by a scalar field; <I> = <I>(r, t) say, by setting V = grad <1>. This permits the integration of the equation of motion, reducing the problem to that of determining <1>. A single scalar equation is all that is then necessary to solve for <1>, and this turns out to be, using the incompressibility condition (8.1), nothing but Laplace's equation
V!(J) = 0
This means th~t in the irrotational motion of an ideal fluid we can solve for the velocity field (i.e., for <1» independently of the equation of motion! Another significant factor is that Laplace's equation is linear. This means that in the analysis of idealfluid motion mathematical difficulties associated with nonlinear partial differential equations drop out of the picture. In fact, a great amount of mathematical knowledge already gathered becomes available. 8.1 Integration of Euler's Equation in Steady Rotational Motion
(~I)
 V x curl V
and so forth
For steady motion Euler's equation becomes grad (
V2 p) 2" +; 
V x curl V
=f
(8.3)
ou au ou u+v+w ' . ax oy az
mathemati~al
The integration of nonlinear differential equations is a very difficult problem. In, fact, there is no general theory yet available for such equations. In case ef a linear eqwttion we find that the sum of two
We observe that integration of this equation becomes possible if it is carried out along a streamline and if the body force f is assumed irrotational. The vector given by the term V x curl V in the acceleration is normal to V, Since the direction of a s:reamlin~ at any point is that of the velocity vector at that point, it foHews that (V x curl V) . ds is zero if ds is taken
124
IdealFluid Aerodynamic:s
[.1
Integration of Euler's Equation in Special Cases
115
along a streamline. Thus, taking the scalar product of (8.3) with ds element of a streamline, we obtain grad
(~I + ;) . cis 
I. cis . along a streamline
(8.4)
We assume that the body force is an irrotational/orce field and set
the varwtion pf H, is related ~o curl V, the vorticity (or, equivalently, the an ular velocity) field of the fluid motion. From this r.quation i~ follows that grad H, is zero if curl V is zero or if curl V is parallel t9 V. In ,either case H, becomes a true constant throughout the flow field. To obtain the variation 0/ H,/rom streamline to streamline, let us consider a streamline and form the component of (8.8) in'a direction normal to
curl V
1= grad
U
Streamline
where fj is a scalar function of position. In the problems we shall deal with, the body force consists of only gravity force and this is irrotational. Using (8.4) and (8.5) we obtain grad This shows that grad ( "2+;U
(~I :+ e.  iJ) . tis p, •
0 along a streamline
(8.6)
VI
p
)
Fla.
I..
Illustrating the spatial variation
or H•.
is a vector normal t.o the streamline. Equation (8.6) integrates to
V· p  +
2
p
U .... const.  H..

the streamline. Thus, if D denotes the normal to the streamline at any point on it (see Fig. 8.1), we obtain (8.7)
 ' .... D •
say, along a streamline
This is known as the Bernoulli's equation along a streamline. The constant H, is not a constant that has the same value for the entire space filled by the fluid. In general, it changes from one streamline to another. In this sense H, is a/unction 0/ poSition, although along a streamline it is a' constant. We now obtain an eq~ation that gives the change of H, in space.
oH on
grad H,
=D•V
x curl V
(8.9)
.... D X
V • curl V
8.3 Spatial VariatioD 01 H,
For steady motion, and assuming as before· that the body force is irrotational, Euler's equation may be written as grad ( or as grad H, = V,x curl V where H, (8.8)
where aH,/on is "the spatial rate of change of H, in the direction D. The unit v<=Ctor D is normal to V. Therefore the magnitudeof D x V is simply V. Choo ing a unit vector b such that the vectors V. p, and b in that order form ~ righthand system (Fig. 8.1), we write
DXV='Vb
Equation(8.9) then takes the form
2" + p 
VI
p
U .... V x curl V
_),
aH, __ V(b. curl V)
(In
(8.10)
Since the angular velocity w is half of curl V, this equation may also be expressed as
="2 +; 
V·
P
U
aH, on
=
2V(b. w)
(8.11)
,From equation (8.8) it immediately follows that H, is a constant along a streamline and is indeed the constant introduced in (8.7). Equation (8.8) is the equation that determines the spatial variation of H,. It shows that
This shows that there is no variation of H, from streamline to streamline if the angular velocity is zero, that is, if the motion is irrotational..' In other words, if the motion is irrotational, .he Euler equation may be Integrated
116
Ideal· Fluid
Acr.:9dynaroi~
JDtcgrttion of Eulcr's Equation in Special Cases
117
once and for all. As we shall sec in the next section, this is tru~ even for unsteady motion.
8.4 Integration of Euler's EquatiQn in Irrotational Motion
As we shall learn in the next chapter, for the motion of a fluid ~o be irrotational it is necessary that the bod} force r be irrotational. Thus we assume that r is irrotational and set it, as before, equal to grad V. We now assume thai the motion is irrotalional, that is, that curl V = 0 and, consequently, that the velocity may be expressed as
V = grad 11>
(S.12)
aynamicsaltd jll u~ f()r the Solution' of numer~us technical problems in aerodynamics; ~draulics, hydraulic machinery, and so On. We wish to point out at this sUlJ!e that the foregoing discussions in Sections 8.2 to 8.4 have been carried out without any recourse to a coordinate description, and tbe cnnclU~UlnS drawn are, therefore, independent of the choice of a coordinate system. This has been pOSSible because of the use of vector notation and the associated concepts of scalar product, vector product, and the gradient. We recommend that the reader work out the preceding results by using Canesians, for instanc,c, and gain for himself an appreciation for the role of vector methods.
(S.13)
8.5 Remarks on an Irrotational Force Field
A force field F is 'said to be irrotational in a certain region of space when in that region' curl F ~anishes. Further, we can then represent the force field as the gradient of a scalar field. Denoting this by fj we write, as before,
where 11> = l1>(r, t) is a scalar function of position and time. With the assumption (E.12) and the relation (8.13), Euler's equation becomes grad where
( ot
~
Vi +  + P  V) 2
IJ
F = grad il
(S.IS)
= 0
(S.14) (S.15)
VI = (grad 11»1 = grad <11 grad <I> This equation readily integrates to
~
In Cartesians, .for instance, the component form of this equation is expressed by oil F == of) F == ail (S.19) F,,= ay'
•
ax
•
Oz
 + 2 + p 01
Vi
P
_ U = const. == F(t),
.
a function of time
(8.16)
Because the integration is only with respect to space, the constant is independent of space coordinates, whereas it may, in general, still depend on time. Hence it is expressed as F(t). At any instant of tune F(t) has a uniform value for all points of the fluid. If the motion is steady, variations with time do not exist and equation (S.16) reduces to Vi P . 2 +   U = canst. = H (S.17) p This is the famou" Bernqulli equation. It was obtained by paniel Bernoulli, before the discovery of Euler's equation, by considerations simi1~r to the modern principle of energy conservation. The constant H in (8.17) is now truly a constant, that is, independent of both time and space. It has the same Value for all points o'f the fluid for all times. Equation (8.16) is a genualization of Bernouili's equation for unsteady motions. Accordingly, it is referred to as the unsteady Bernoulli's equation. Bernoulli's equation is the most important relation in eiemcntary fluid
where F", F", F. arc the components ,of F. Relation (S.lS) forms the basis tor determlmng Uif F is known or, conversely, for determining F if Dis known. As the simplest example, let us ~onsider the case of our "gravity" field. We choose a Cartesian cooratnaxe system with the Zaxis pointing upward, that is, pointing in a direction opposite to that 01 the acceleration due to gravity. The force on a body of ro~ss m in this fo'rce field is then given by
F = (F", F", F.)
== (0,0,
mg)
(8.20)
Curl F vanishes everyWhere. Thus· gravity field is an irrotational force field. Tp determine the scalar functinn U, the gradient of'which may represent 1", we must solve the differential eauation (S.18) or, equivalently, the component equations (8.19) ""romfS.19) and (8,20) we obtain
.ax
af) == 0, au = 
ay
U
0, 
e}(J
oz
==
mg
(S.21)
which may readily be integlated' to yield
==
~mgz
(8.22)
fJ is determined only to within an. additive constant which we have here set equal to zero so that fj is zero at z == O.
22.
IdealFluid Aerodynamics
Integration of Euler's Equation in Special Cases
229
In ca.se of .an irrotational force field, such as the gravity field we have been dISCUS~lOg, one can attach a certain physical significance to the scalar function that represents the force field. To do this, consider the motion of a particle acted on by a force field F. The work done by th~ force field on the particle during its motion from a point Po to a POlOt P alon~ a path rt' is expressed by
W
f
P•.
P
F· cis
p
(8.23)
)
~/.
Po
Fig. 8.2
scalar function, 0, the gradient of which represents the force field. Thus, in an irrotational force field, the scalar function fj is physically equivalent to the work done by the force field. In problems of mechanics, the negative of the work done by an irrotational force field is defined as the "potential energy" 'or the "potential" of the system on which the force field acts. According to this definition a scalar function that represents an irrotational force field is essentially the negative of the potential energy.· In light of these considerations it is customary to refer to our scalar function, such as fj inEq. (8.18), as a ''force potential" or simply as a "potential." This nomenclature is fui!her extended to all irrotational vector fields, and it is common to talk about "velocity poiential," "acceleration potential;' etc. On the sarrie basi~ we refer to irrotatjonai vector fields as "potential fields." For a mechanical system subjected to only irrotational forces, the lotal energy of the system (i.e., the sum of the kinetic and potential energies) is conserved, that is, remains a constant for all times. Hence an irrotational force neldis also sometimes called a "conservative field." In concluding this section we add a few remarks about the independence of the integral
Illustrating the work done by
3 (OIU.
~h~re ds is an ele~ent of length on the path rt' (Fig. 8.2). The symbol f IOdlcates that the IOtegration is to be carried out along the given path rt'. Generally, the 'Jalue of the integral
integral to be independent of the path and to be dependent only on the endpoints, we require that th~ value of U(P)  fj(P o) be the same no matter how we approach the. points Po and P, that is,' by what path or direction. This means that the function fj should assume at any poi!lt a definite single value irrespective of the path chosen to alTive at the point.
dfi to In independent of the path connecting the ~ . endpoints P and Po, [j should be a ~inglevalued function ofposition This is Thus, for the integral
f dU
p.
op
with respect to the path of integration. For this
l
P
fP
jP.
F· ds
wou/~ ~epend.on the endpOints and the path connecting them. If the force
field.:s Irrotatlonal, F may be replaced by the gradient of a scalar function, say V. Then F ·ds becomes an exact or a total differential of fj and the integral of (F· ds) between any two endpoints Po and P becomes independent of the path connect.ing them. Thus for the work done by an irrotational force field we write
wltat we'~ave considered here. In the analy!;is of phy.sical problems, one does meet with multiple valued potentials for irrotational vector fields. In such a case an integral such as (PdO need not have the same vnlue for
ali paths connecting Po and P.. We shall learn in our studies that a multiplevalued potential plays an important part in the theory of lifting bodies.
8.6 Remarks on Bernoulli's Equation
.
Jp.
w=JP F·ds =JPgradU'ds=fP dU
Po
Po
Po
=
U(P)  U(Po)
(8.24)
In light of ~he preceding considerations we may inTerpret BernouW's equation as simely a statement of the conservation oj mergy. For steady;
• If we had initiaJly written F .. grad iJ instead of F = grad jj as done in equation (8.18) U would have become identical with the potential energy.
This ~tates that. t.he work done by an irrotationai force field is simply a functton of position of the endpoints. This function is nothing but the
2JO
IdealFluid
~erodynamics
irrotational motion of an ideai fluid we have from Eq. (8.17)
P
V·
2
+ p  U == const.
Chapter 9
Here p( va/2) is the kinetic energy of the fluid per unit volume and  fj is, again per unit volume, the potential energy related to the body for~s which arc: c&S5umed irrotational. To interpret p let 'us recall that the field of the resultant pressure force, specified again per unit volume, is grad p. This means the resultant pressure force is irrotational and p is its potential. It follows immediately that pis tbe potential energy (of course pel" unit volume again) related to the'resultant pressure force. With the preceding interpretations, the term (p  U) is simply the potential energy of the fluid. Thus Bernoulli's equation (8.17) reads the kinetic energy
lrrotational Motion
+ the potential energy = canst.
It is not surprising that Bernoulli's equation is simply a statement of the law of conservation of energy. We have seen (refer to Section 5.8) that for an incompressible, inviscid fluid the equation of energy conservation (Eq. 5.24) reduces to the equation for mechanical energy (Eq. 5.25). In such a case, when the equation of mechanical energy is integrated (that is what we do to obtain Bernoulli's equationkone should obtain that the total energy of the system is conserved. Before concluding this chapter it is interesting to put Bernoulli's equation (8.17) in the form COQ'lmonly used in hydraulics. Considering the body force to be that due to gravity alone, we have from (8.22) [j = pgz for
In the last chapter we saw that great mathematical simplicity could be achieved in the analysis of fluid motion if the motion can be treated as irrotational. We shall now examine the conditions under which the motion may be treated as irrotational. W~ begin with ~n. a"nalY5i~ of the , general motion of a fluid elemen~ with a vle~ ~o recogDlzmg particularly that the angular velocity ·of a flUid element IS ~~deed half ~he curl of the velocity. which has been referred to as the vorticity. We then develop the important theorems of Helmholtz and Kelvin which describe the fate ~f vorticity and circulation in an inviscid fluid. These theorems show that If the motion of an ideal fluid is once irrotaV + dV tional, it is always irrotational. We then take up certain properties of irrotational motion: 9.1 Most General Motion of ~ Fluid
Element
U in (8.17).
Thus we obtain
Vi p
2
+ p + pgz = const.
p
or writing per unit mass of the fluid
 +  + gz =
2
p
Vi
const.
..,hich is the form commonly met in hydraulics.
In the most general case, the motion of V a fluid element consists of a translation, a rotation. and a deformation. We show this by considering the relative motion between two infinitely close points of a fluid element. At a certain instant of time t, let P and Q denote any two such points and let, rand r + dr be their respective positions (Fig. 9.1). Fig. 9.1 IIIuslrating the general In a general motion' of the element, the displacement of a ftuideloment. points P and Q both experience changes o f . .. ,position. Let ~ denote the displacement of In a small. tIme mter~al nd ~ the corresponding displacement of Q. If V(r, t) IS the, veloc~ty a • .he point P at the time t, the velocity at Q at the same moment IS expressed to the first order by
P
d!
VCr
+ dr, t) = VCr, t) + dV
(9.1)
1J2
IdealFluid Aerodynamics
Irrotational Motion
1JJ
where tIV is the change in V over the directed distance dr. Since dr is infinitesimal, tIV is given by (see Section 2.26, Eq. 2.76)
tIV = (dr· grad)V
where V = V(r, t). For the displacements of P and Q we have
(9.2)
and where JV I and JVI are velocity vectors given by the first and second terms respectively on the righthand side of Eq. (9.6). Since qJ/j is antisymmetric, as qJjj is equal to CPu, the vector JV t becomes
t;o = VCr, t) dt ~ = VCr + dr, t) dt
=~+ JVdt
(9.3)
(9.4)
We notice immediately that in the displacement of the point Q, there is a part that is fhe same as that of P. This part, denoted by l;o, is the same for al\ points of the fluid element and, therefore, corresponds to a translation of the element as a whole. The component tIV, which is the relative velocity between P and Q, can be shown to be made up of. rotation and a deformation. We carry out the proof in Cartesians. Accordingly, we denote the reference unit vectors by e" e2 , e, and write.
r = (Xl'
XI,
Equation (9.9) shows that dVlls the velocity a. the poi~t Q due to a rig~d body rotation of the fluid element as a w.hol.e about an mstantaneous aXIs through the point P. The angular velocity IS equa~ to w. The veltx:ity JV I is characterized by the set of OIne numbers
X,)
dr = (dx dxl , dXa)
(9.11)
We then obtain
V == (U I , U., u,) JV = (du" du l , du,)
dV = (dr. grad)V
"
= II I
iI
('I
_i
1102: 1
au dX ) e, I
(9.5)
where the partial derivatives are evaluated at the point r at the time t. We now express each of the partial derivatives as the sum of a symmetric and an antisymmetric term and write .
tIV =
f{i =}: (f
iI .\
i [l(aU 11 2 ax,
€il 11
+ OU I ) + l(au;  ~)J dxl}e;
oXi
obtained by specializing €iJ by giving i and j all the possible values fro~ I, 2, 3. This array forms a secohdorder symmetnc tensor, symmetric because €ij is equal to €u. The diagonal.elements of this tensor represent rates of normal strains, while the nondlagonal. terms represent rates of shear or tangential strains. * The tensor (9.11) IS therefore known as the rale of strain tensor. Hence it foHows that the velocity JV 1 represents a velocity due to defQrmation of the fluid element. . . We have thus shown that the general motion of a flUId element IS made up of three partsa translation, a rotation, and a deformation.
9.2 Rotation and Vorticity
2
.ox 1
ax;
i
dxl)e;
+·i (± dxJ)e
iI. j~1
ljJil
.
where
=JV1
+ dVI
€il
1jJ.
(9.6)
=
aU i + OU I ) 2 ax} . Ox;
!(
(9.7) (9.8)
"
= l(au i _ au;) 2 ax I OXi '
From the preceding considerations it follows that at any p.~int in the flow field of a fluid, associated with the velocity vector, there IS a rate of strain tensor (denoted above by €jj) and a rate of rotation tensor (denoted above by ljJiJ). Recalling the splitting of t~c tensor gradie~t of a vector field into a symmetric and an antisymmetnc part (see SectIOn 2.25), ~e observe that the symmetric part of the tensor gradient of the velOCity measures the (instantaneaus) rate of strain while the antisymmetric part measures the (instanta'1euus) rate of rotation at the point consideied.
• Proof .is left out as anexercise..
• We adopt this convention for simplicity, which will become apparent in the follOwing.
231
IdealFluid Aerodynamics
Irrotational Motion
235
The rate of strain tensor being symmetric requires actually six components for its complete specification. The rate of rotation tensor being antisymmetric requires actually a set of three components for its complete specification. It can, therefore, be represented by a vector. This vector is nothing but the angular velocity vector w introduced in Eq: (9.9). The
the rate of rotation) at any point of the flowing fluid is equal to half the curl of the velo.city at that point at the instant considerea. The vector curl V is called the vortex veCtor or simply the vorticity of the
fluid at the point considered. Denoting the vorticity by 0, we write
o
It then follows that
that is,
i!!!!
curl V
(9.14) (9.15)
wiO
angular velocity =
i
vorticity
We thus see that associated· with thc velocity field V(r, t) of a flowing fluid there is a vorticity field n{r, trand an angular velocity field w(r, t). A schematic representation of thc vclocity, vorticity, and angular velocity at any point is'shown in Fig. 9.2.
9.3 CirculatioD and Vorticity
We shall now apply to thc velocity ficld of a flowing fluid the relations that exist between .circulati~n and corl of any vector field. Following the definition for any vector field (see Section 2.28), wc d~finc the circulation of
FIa.9.1 Angular velocity aDd vo.rticity.
relation between the elements of the rotation tensor and those' of given by Eq. (9.10), which may be expressed as
(a)
is
(9.12) is the component of (a) in the direction e j and where 'i, j, k takc (a) maybe expressed in Cartesians as whe.re
W,
successive values from I, 2, 3. By using Eq. (9.8), the vector
Fla. 9.3 Circulation.
FIa.9.4 Vorticity and circulation &r:>und an infinitesimal surface e1cment
the velocity field as the line integral of V around any closed curve ~. Wccall this simply cirfUlatio!1 and denote it by r or r .. ~ Wc thus have The determinant in this relation is nothing but the curl of the velocity vector. thus it follows that the vector representing the rate of rotation tensor is given by (9.13) (a) = ! curl V which may be specialized easily to any orthogonal, curvilinear coordinate system. This eQu~tion states that' the instantaneoUs a!'.'GUlar L'elociJy (or
r .. 
f.
V· ds
(9.16)
whcre ds is an elemcnt of thc curve ~ (see Fig. 9.3). Consider an infinitesimal surface element D dS ~ituated at any point P. Let e" denote the boundary curve of thc element and drc. the circulation around it, the direction of integration along thc curve being that given by the rule of righthand rotation about D (see Fig. 9.4). Then, according to
136
the relation (2.120a), we have dr c. ='
IdealFluid Aerodynamics
Irrotational Motion
237
f
c.
V· ds = curl V • D dS
(9.17)
= Sl· n dS
where S1 i~. the vortex vector at the point P. This equation states that .the circulation arQund any infinitesimal surface element D dS is equal to the flux of vorticity through that element.
fluid element or, equivalently, the fate 'of circulation around a fluid curve. This examination may be carried out by .working directly either with the vorticity or with the circulation. Here we work with the vorticity; in the next section we shall work with the circulation. The equation governing the rate of change of vorticity of a fluid element may be obtained directly. from the equation of motion. For this purpose, considering only an idea1 fluid, we take the curl of Eq. (5.31), which is
w  + grad ~ .v )( n _ r ~
2
grad f p
()
(9.19)
and.obtain
n
aSl   curl (V )( Sl)  curl r
at
s
Expanding the second tenn on the lefthand side, we have curl (V )( Sl)  (div Q)V  V· grad Sl + = V· grad n +.Sl.grad V
n· grad V 
(div V)Sl (9.20)
since the divergence of the vorticity is always zero, anel since the fluid is assumed incompressible, the divergence of the velocity is zero. Substituting (9.20) into (9.19), we obtain
Fig. 9.5 Circulation around '{; is equal to
~he
outflow of vorticity through S.
an or
at + V • gradn 
Sl • grad V  curl r
Consider now a finite closed curve re and let S represent any surface bounded by re. Then, according to Stokes' theorem (Eq. 2.131), we obtain
:on Dt
r" =
=
f
"
s
n • grad V = curl r
V· ds
~ ff curl V· n dS
s
(9.18)
We assume that the body force is ;rrotational and set curl r  0 Thus we obtain
ffn.Dds
This equation states that the circulation around a closed curve re is equal to the outjiow of vorticity through any surface S whose boundary is formed by re (Fig. 9.5).· It also shows that if we consider different surfaces drawn with the same boundary curve, the outflow of vorticity is the same through all these surfaces. 9.4 Rate of Change of Vorticity We now investigate the conditions under which the motion of an ideal fluid is irrotational. For this purpose we examine the fate of vorticity ofa

Dn =
Dt
n.grad V
(9.21)
which gives the rate of change of. vorticity of an element of aD ideal fluid as it moves under the action of an irrotational body force~ Recalling the . meaning ofa term such as B· grad A (see Section 2.26, Eqs. 2.76 and 2.77), we see that n • grad V = where
n (the rate of change of V with respect to distance
in the direction of Sl)
a is the magnitude of n.
231
IdealFluid Aerodynamics
Irrotatiooal Motion
239
Equatiort (9.21) states that the material rate of change of the vorticity is zero whenever the vorticity is zero. On the basis of this result, we should not, however, proceed immediately to the conclusion that if the vorticity of a fluid ekment is zero at some time it is zero at all times. It is of course true that (9.21) leads to such a conclusion. To show this rigorously we first cire the following kinematical relation, * which is true whether the ftuid is compressible or incompres~ible:
D (n . DdS) =
boundary Crt .of the element DdS, we have
A·ndS ..... dr c"
Therefore Eq. (9.22) takes therorm
.!!.. (drc ) == 0
Dt •
(9.23)
Dt
[on ot
curl (V )( n) • D dS
]
==
[~~  n . grad V + (div V)O
J.•
dS
where D dS is a surface element moving with the fluid. If the fluid is incompressible, we have
.!!.. (n . DdS) = [Dn  n . grad v] . D dS
DtDt
Wilh this relation (9.21) reduces to
.!!..(~I. D dS) .. 0
Dt
(9.22)
This equatio'l expresses the important result: In the motion of an l'deal fluid subjected to irrotational body forces, the material rate of change of the outflow of vorticity through any surface element moving with the fluid is permanently zero, or, equivalently, the outflow .of vorticity through any surface element moving with the fluid remains a constant for all times. In this sense vorticity is convected with rhe fluid. We may now draw the conclusio~ that if the vorticity of any surface element is zero at some time, it will remain so for all times as the element moves with the fluid, for according to (9.22) the outflow of vorticity throltgh any surface element moving with the fluid should remain permanently zero if it is zero at some time. EquatiOQ (9.22) may readily be expressed in terms of the circulation around the surface element. Denoting by drc" the circulation around the
DdS moving with
This expresses the result: In the motion of an ide,!1 fluid subjected tn irrotational body forces, the material rate of change of the circulation around any surface element moving with the fluid is permanently zero,or,equivalently, the circulation around any surface element moving with the fluid remains a constant for all times. . Equation (9.21), or, equivalently, (9.22), is referred to as Helmho/;z's theorem. * If forms the basis for the derivation of Hemlholtz's theorem of vortex ,motion, which we shali take up in the Chapter on vortex motion. TIlt abov~ derivation of Helmholtz's theorem, which, following Helmholtz, start'! from Euler's equation, seems to indiCate that any conclusions drawn from the theorem are valid ill the motion of only an ideal fluid, thai. is, of an incompressible inviscid fluid of constant density. We can, however, show that such cortclusions hold also in the motion of a compJusihle inviscid fluid. This can be done readily if we start, foliowing Kelvin (I 869), from the concept of circulation. This 'We do in the next section. We close this section with the observation that in twodimensional and axisymmet,ic flows the term o· grad V is Uk"t{cally zerot and (9.21) therefore rcdMces to
·~ .... o
Dt .
{9.Z4)
This equation expresses the result: in twodimensional and axisymmetrice,l flows oj an ideal Jluid under the action of ;rrotational body forces, tile oorticity of a fluid element remains a constant for aU times.
~
Rate
or Change 01 ClreaJatioa
• If A is any vector field, the rate of change or outflow or A through any surface element ito., fluid may be expressed as
DI
D (A. D ciS) _
[iI.t. + V div A ill
curl (V )( A)] . DdS
We consider a closed fluid curve (i.e., a curve consis~ing of the same fluid particles at all times) and seek the rate ofchange of circulatiOll around such a curve as it moves with the fluid. Let ~ represent any closed curve drawn at the instant t and let r denote the circulation around <jf at that instant. During a time interval fl.t the different fluid particles that make up <jf move into new positions. Consequently, in the interval I1t the curve ~ will assume a different shape and occupy a new location. Let <if! be the
For a proof of this fonnuIA see Sommerfeld (1950).
t The reader shoul~ verify this as an exercise.
• See Helmholtz (18S8).
IdealFluid Aero;dynamiC:S
Jrrotatiooal Motion
From the quadri1atera1 abb1al (see Fig. 9.6) we see that
aal+~"+~
ur
or (9.29) Denoting the velocity at the point a by V aDd that at b by V + dV, where dV is the change in V from a over the directed distance ... the displacements 881 and bbl can be expressed as "IVAt and
.
,It time t
bitt 
(V
+ dV)~
(9.30)
I
With Eqs. (9.28), (9.29)~ and (9.30) we obtaiD
+ III
.... 9" IUUltratiDs the cJwl&c in circulation around. fluid curve.
E.. .. _ dV
Dt
(9.31)
CUfVC at t + At ud let r l denote the circulation around ~I (soc Fig. 9.6). 'The rate of change incircWatioD aloDg ~ followiDg its motion is then given
It theD follows that
by the material ~vative
1 
'
Dt a,~.' At iii terms of the line'iDtegrals of the velocity this equation takes the form
J>r I' r r   Im
(9.25)
(9.32) With this reladoD Eq; (9.27) becomes
E..1. V ... lim l.["VI . . . .  " V .ds] (9.26) ,Dt Dt:r.. ",~oAt j", , SinCe wea~ followiDg the same set of fluid particle~, the limit on the right
Dr •
1" '
Dr ...
Dt
1. DV • tis _ j .. Dt.
r"
1.
8 ...
(9.33)
side of this equation is the same as the integral
f".(DIDt)(V ~'ds) along the
cUrve ~ at the instaDt t.' Therefore (9.26)'becomes
Dr
Dt
'V.E... (9.27) Dt Dt , where DVI Dtis the acceleratioD at theiDstant t at the point considered on 'the curve~ .. To iDterpret the expression (DI Dt) ds we proceed as follows. If .. repr~~nts the element ab of the curv~ ~ and ... represents the correspon~ing elemeDt alht of~1 (see Fig. 9.6), we have
._" ~~ .. +" , 1.. :r"
Dtj" Dt
A,"'O
_.Q1.
V ... _ " E.. (V • ds) j" Dt
where 8 deDOtes the acceleratioD vector~ Equation (9.33) states thOt at any instant the rate of change of circulation around any fluid CftIrve ;j tqual to the line integral of the acctleration around that CJUVe taken at the instant considered. AD immediate consequence of Eq.(9.33)is that if theacceleration is expressible as the gradieDt of some (singlevalued) scalar function, that is, if the acceleratioD field is aD irrotational vector field, then the integral
f"
8· . .
is
zer()
aDd consequeDtly Dr/Dt is zero.
The considerations leading to Eq. (9.33) are still kinematical. To proceed further we have to bring in dynamical consideratioDs. To evaluate the iDtegral
£
8 . . . we
restrict ourselves
to an inviscid fttJid and use Euler'S
equation (5.11). Thus 'We obtain
E.. .. _ lim _"'~_ds_
At
(9.28)
Dr  
Dt"
'f
a· .. 
f
.f· .. 
"
f
gradp . lis "P
(9.34)
IdealFluid Aerodynamics
InotationaJ Motion
143
In this equation the density p is Plot assumed to be a constant. We now assume that the body force is ;rrolalio"o/ and set f.grad fj where have
V is a sinBlevalued scalar function of position and time.
Then we
fy f· cis  fy dU  0
Equation (9.34) consequently reduces to
motion .of .. inviscid fluid, the rate of change of circulation ~d any fluid curve Is permanently zero if the body forces are irrotational and if there is a Singlevalued pressuredensity relation,or. equivalently under these conditions, the circulation around a fluid CIII'r¥ remains a cons,ant lor al/ limes as the auve moves with the fluid. This theorem is known as Kelvin'. circulation tlworem. Consider now any infinitesimal surface element a tIS that moves with the fluid. "Let dr c.' denote the circulation around the boundary C.. of
such a surfaCle element. Then Eq. (9.36) may be put in the form
Dr ... _" Dt
i..
~.cIs
p
(9.35)
D· Dt(drc.)'" 0
which is the same as Eq. ~9.23). Furthermore, since
If the fluid is assumed to be incompressible aDd homogeneous so that p is constant, the integral on the right side of the above equation vanishes.
a
1. I!!!l. cis ! i .. p p
t
y
dr c • g·adS·
we readily obtain the equatiOn
dp ,,'0
If theftuid is assumed to be either compressible or incompressible and inhomogeneous $0· that· p is a variable, the integral no longer vanishes automatically for any arbitrary curve~. However, if there is a singlevalued relation between. the density and the pressure so that one can
($l.adS)  0 Dt
which is exactly Helmoltz's theorem, namely, Eq. (9.22). Thus we see that the contents of the theorems of Helmholtz and Kelvin are identical. .Kelvin's derivation shows clearly that the theorems are applicable not only to incompressible homogeneous fluids but also to compressible fluids. in which there is a singlevalued p  p relatima. Considering at any instant a finite fluid curve ~, draw an arbitrary surface S such that <G is the bOundary of S . . Then. since
D
.
express
. P
I!!!.l_ pad P
where P is a singlevalued scalar function' of position and time, the integral again vani$bes for any arbitrary curve 'I. Then we have
I. p i .. I!!!.l. cis 
t 8~d
..
P • tis
r ...
f
..
V • lIS
~ n . D dS
tl
II
 £.dPO
. 1 hus, when we assume that p is a constant or that there is a singlevalued relation between p and p, Eq. (9.35) becomes
Eq. (9:36) may be put in the form
:,ffn ~.
tl
dS .. 0
(9.37)
Dr 0 Dt
This equation expresses the theo"m
(9.36)
This equation states that under the conditions considered in its derimtion. the outflow of vorticity tlrrough any surface bounded by Q c/osed'jluid curl:i
0/ conservation 0/ circullltion:
In the
remains constant for all. times as tire CUrt'e mt)l'es with the fluid. In concluding this section we note tlult the conditions under which the
theorems of Helmholtz a.nd Kelvin are valid are precisely· the conditions under which an accele,ation pOtential can be defined .
 For iristance, in the motion of an invilcid compressible fluid of constant entropy, . .d pI p is simply grad h. where h is the enlhalpy per unit IIIQS.
IdealFluid Aeroclynamic:a
Irro..tioDa1 Motion
'.6 Irrotatloal Motioa
Several important results may be deduced from the theorems of Helmholtz and Kelvin. They are, for iilstance, the startingequations for developing the theorems of vortex motion that we shall consider in a later cllapter. At present we use the tbeorems to deduce 'the conditions under which the motion of fluid is irrotational. Cohsider an inViScld ftgjd and let the body forces be irrotational. Also, let the fluid be such that there ill a singlevalued presSuredensity relation. Such a fluid is known as a baro,tropiefluid. An incompressible homogeneous ftuid is a special case of a barotropic ftuid. In ~ state of rest or of uniform motion the circulation around every infinitesimal surface element of the ftuidis zero; simibarly, the vorticity (or equivalently the rotation) of every element of the ftuid is also zero.. If the ftuid is then brought into a ditferent state of motion, the circulation around every surface element and the vorticity of every element, aCCC?rding to Kelvin's theorem, are still zero. This means that under the action ofpotential body forces 01/ motions
Sinc:e. the gradient of a· constant is zero, ~ is de~ined .nly up to an addItive cons~nt .. The scalar function ~ is known as the velocit, potential.
'.s'
TIle Equtloasfoi
Irrotatioaai ModoII of _ , ..........
a
. The motion ofan ideal ftui~ is cbaractcrized by the velocity ,nd pressure fields ·as. the unknowns and IS governed by ·(S.28) and (S.31).Thcsc are
.(I) Condition of ;ncompressibilJtY
(2). Equation of motion
DV
divV  0
Iii  at + srad 2· V
iN
. .p.
.
(.
)c.
C\1rl V·  f  arad
!)
ofan inviscid barotropie jluit:! set up from a state of rest or of uniform motion are permanently rotation1ree, tIuit is, tMy are permanently irrotational.
Since irrotational motion implies the existence of a velocity potential, such motions said to be potential motions. Our main cOncern in this book is with the motion of'an ideal fluid on which only irrotational body forCes act. Genendly· the motions we shall consid4r originate from a state of rest or of uniform Motion and are conieqlJlmdy irrotational. Therefore our concern hereafter will be with the irrotational motion of an Ideal ftuid. We now proceed to set up the equations that govern such a motion.
are
If the motion is irro.tational, we ,add. to ~ the condition of irrotationality (~.38) and ~place; according to tbedefinitioD(9.39), the velocity by ~he .ve~oclty potential. Furthermore, as we ha~ ~~. for the motion to be Irrota~C)nal the ~Y force must ~ .irrotati.onal. Tl\erefore we $Ct f equal to grad U, w~ereU(r,.t) is.thescalarpolenti&.loftbe body fo~. In,this way We obtam th~ followltlg as the equations governing fM irrotational
motion. of an idealjluid. (I) Condition of incomptessibility
or (2) .Equation of motion div (gtad ~)  div V  0
(9.40)
'.7 VeIodty PetatIaI
Thr(lIughout the region in which the fluid motion is irrotational we have
or
a . v· at (srad~) + grad.2 == grad . U
sra~(~ + ~ +1 
grad
(.)
!.
(9.41)
g == curl V .... 0
(9.38)
where
at
2
p
if)··... ,0
which is the condition ·of irrotationalhy. Equation (9.38) ensures the existence of a scalar function of position and time, say
VI II;' (arad ~)I
This equation imin~ateJY'fnteglWtes to
cD ==
s·Jch. that
~(r, t)
v == grad <I>
(9.39)
o<l>.I{ JJ a + 2 grad ....... p .....) + £. ,.

Uf(t)

.
(9.42)
• For sake ef continuity and convenience we gather in the following two sections ideas and results that have been introduced on previous occasions. See, for instance, Sections 8.1 and 8.4.
where itt) is. a funct~on ~ time that is URifn tJiroughout spa~ at any msta~t. ThiS equation IS tbe fllfSmuJy BemouJ/i's equtltiiIR. Which was obtamed before (see Eq. 8.16).
146
IdealFluid ActoclyDainics
Irrotational Motion
147
, We see that whereas the general (i.e., rotational) motion of an ideal fluid is characterized by four scalar unknowns, namely, the pressure and the three components of tt e veiocity. the irrotational motw,n of the fl~id involves only two sCalar unknowns, namely, the pressure and the veloc~ty potential. In the general motion of the fluid, to determine the unknowns we must solve simultaneously the condition of incompressibility and the equation of motion. In the irrotational motion, however, this is no longer necessary, for we may solve for the potential from the condition of incompressibility independently of the equation of ~otion. The equati~n ~f motion itself may be integrated once and for all. Once tbe potential IS determined, the pressure is ~dily found from'the integrated form of the equation of motion, namely, BernouUl's equation. Also, we may then obtain the velocity field, if desired, from Eq. (9.39). We thus conclude that the entire problem of irrotational motion ofan ideal fluid amounts to that of solving lAplace's equation, which ~w expresses the condition. of incompressibiliry. This is a great mathematical advantage, for there is considerable knowledge already available with regard to the solutiqns of Laplace's equation, which governs many physical phenom.ena. , It is instructive to recdgnite that Eq. (9.40) 'could have been written down by stipulating that the fluid in motion is incompressible and that the motion is irrotational; in other words, by specifying that the velocity field V is such that its divergence and curl are both zero (see Section 2.43). Then we can introduce a potential cf) on the basis of the fOnditiori that curl V is equal to zero and determine cf) frdm the condition
VIcf) == div V
Suppose a mechanical system is subjected to very large forces for a very short time. Finite changes of velocity and momentum of the system will then OCCur suddenly. ~uch changes are said to be brought about impulsively, and the forces producing them are known as impulsive forces. Let the momentum of a systt;m be changed impulsively at any instant to from Po to P. Then the ,relation between the sudden change in momentum at to and F, the impulsive force causing the change, is given by
ISP(to) == P  Pc == lim
610
f
~II
F dt
to
The right member of this, relation is known as the impulse. Now consider an element of an ideal fluid in motion and let its momentum be changed impulsively at any instant to by meaos of impulsive body forces and pressure forces. Impulsive pressure forces are generated by sudden, changes in the boundary conditions as, for example, when a solid body i~mersed in the fluid is suddenly set in motion. The impulsive change m mo~entum of the fluid element is given by
] ISm[V  V o =
110
l~.n [f"HIlSmf dt J.'k+I'c5T grad Pdt] J~
(9.43)
:= 0
w~re ISm is the mass of the element, ,1ST its volume, f the body force per unit ~ass~ a~d V  V~ is ~he impulsive change in the velocity at to. This e~uatlon IS Simply an mtegral of the equation of motion (Eq. 5.11) with time over the interval 1St, which is allowed to go to zero. Since the mass, the volume, and the density of the fluid element are constant, Eq. (9.43) takes the form
V  Vo = lim
ItO
In such a case the velocity is, determined without invoking any dynamic and thermodynamic considerations. Such a velocity field, we say, is only ki'1ematically feasible. As we have seen, if in addition to the conditions on thc velocity field we further assume that the fluid is homogeneous and inviscid, all kinematically feasible veloci!l fieldS are also dynamIcally feasible. 9.9 Irrotational Motion as an Impulsively Genelated Motion: Velocity Potential as the Potential an Impulse
[i
"HI
t.
1 f dt  p
1'+61grad P dtJ
to
(9.44)
We assume that the body forces are i,rotationo/ and set f equaJ to grad Equation (9.44) may then be expressed as
U.
V  Vo
= grad X = grad
1  grad 111" p
(9.45)
or
The instantanevus state of irrotational motion of an ideal fluid or a chai'lIze in such a state may be interpreted as br()l!;~ht about ll,1rollgh the actio~ of suddenly, or impulsively, applied forceso Such an interr~etatio,n iii not only illuminating but is useful in applico.tions:· To, bUIld ~hls interpretation we first recall basic definitions related to Impubve nwllon.
• See, for instance, Chapter JO on un,tcdody mcti(l[)f,.
(X  ;)
lo'HI
where
X == lim
and
1%1
dtO
f
to
[j dt
to
== lim
dtO
f
tordl
pdt
Irrotatiooal Motion
IdealFluid Aerodynamics
Gr~ Xis the impulse of the body forces and <grad 1Irlp) isthe.impulse of the pressure forces, both· ~pulses being measured perunjt mass of the fluid. We may call i ~ PQtential of the impulse of the body forces, .. . the potential oftheimpulae of the pressure forces, and lx ~ (1Ir/p» the
.potential of the impUlse, diat is, ot: the total impulse. The potential .. is known as the impulsive preSSlll'e. If we set V. as equal to zero,that is, jf we consider that the state of motion just before the application of the impulse .is a state of rest,· Eq. (9.45) t,akes the form .
v
grad
(X ~) p.
(9.46)
From· this c,quation it imrnc:diately follows that curl.V is zero, that is, the motion imlJlClijately after ~ impulse is irroiational and that tbc¥elocity potentialt of this motion is ~veD by fIlXP
1Ir
is a singlevalued potential:· The state of motion at some instant of time is independent of the state of motion at any other instant of time. The state of motion·at·any given instant is determined solelyby the State of the boundary conditions at that instant. In this sense the motion has no memory. Any changes in the flow field as brought about by cJumges in the boundary conditions arc immediately felt throughout. the flow field. One says that all changes are propagated instantaneously,. or with infinite speed in aU directions. This means that at each instant the state of motion at any point of the flow field is intimately conncc;ted with the state of motion at another point at that instant. .SillQC the flow has no memory, the notion of initial conditions becomes meaningless. In other words, boundary conditions are the only' conditions that may be given to specialise the solution of the governing differential equation, which in this case is Laplace's equation. In this sense the mathematical problem representing the motion is a socalled Ixnmdiuy value problem.
(9.47)
9.10 Bomadary CooditioDS
We now express the bou~dary conditions (see Section 5.11) for an ideal fluid in terms of thevelooity potential.
Furthermore, since both i and .. arc'Single":valued functions, the'velocity potential is also a 'singl~valued functi,on. Equations (9.46) and (9.47) show that if impulsive irrotational forces (in the form of body and pressure forces) arc applied on an ideal fluid that is initially in a state of rest, the fluid is instantaneously set into an irrotational motion and the, resulting velocity potential is singlevalued and equal to the potenti.al of.the impul.se. Equivalently, any state of irrotational motion of an id~al flwd for which there is a sin~valucd potential may be brought to restmstantaneously by the application of suitable impulsive irrota~onal.forces. ~e have ~hus, shown that any given instantaneous state of m:otat,onal motion of an ,deal fluid, fo, which ·there is a sing/~valued potential, may. be .interpret~d as
ColfIIititM ", • &lUJFIMid~. The kinematical condition at such a boundary is expressed by Eq. {5.35) or,equivaiently'by Eq. (5.39). Equation (5.35) now becomes ' .
grad fIl· D ==
~ == V, • D on the solidfluid surface on
(9.48)
Here ~lon is the spatial derivative of fIl in the direction of the outward normal. Equation (5.39) takes the form
DF Dt  of + gradfll. grad F == 0 01 .
generated impubively from a state of rest by the appllcat,on of SUitable impulsive irrotationalforces and that the velocity potential is the.potential of ihe impulse. Any changes in the state of an irrotational motion for which there is a singlevalued potential may also ~ interpreted as generated by impulsive pressure forces and forces that are irrotati~naL . Equations (9.46) ancl (9.47) show that the pr~lDg co~cluslons hold equally when'there are no body forces, and the Impulse IS ..,nly due to impulsive pressure forces. . ' . On the basis of the foregoing conclusions we mfer the followmg Important properties of the irrotational motion of an ideal fluid for which there
• A state of rest is possible only if the body forces are irrotational. t An arbitrary constant may always be added to the potential without altering Eq. (9.46) and any oftbe results that follow from it.
on
F(r, t) == 0
(9.49)
Collllitiou .t • Free SlIr/.ce. Let q>l denote. the potential on one side of the surface and fila the potential on the other side of the surface. Then the kinematic condition expressed by Eq. (S.40) may be written as
grad fill • D = V,, D = grad 412 • n on the surface (9.50)
ECluations (5.42), which express this condition in an equi ..alent rnanner,
• It is important to bear this in mind. for as it will be apparent later that important differences e~ist between ftows with singlevalued potentials and flows with multivah.. ed potentials.
1S9
now~me
IdealFluid Aerodynamics
aF and 'aF
at + grad ~1 at
•
grad F
;=
O}
on
F(r, t)
== 0
(9.5t)
 + grad~I' grad F·= 0
R
The dynamic· condition expressed by Eq. (5.44) requires that the pressure be continuous across the free surface. Using the unsteady Bernoulli's equation (9.42), we may write'this dynamic condition as
Fla· 9.7 A body moving with a constant velocity through a fluid.
Pl[~1 +1 (grad ~1)1  Ut)] == at 2
== Pt[~1 + ! (grad ~J. 
PI
==
PI
on
at
2
II(t)J
F(r, t): .0 (9.52)
We have yet to specify the conditions aUnfinity. do this we need first to consider the probl~ms we shall be interested in and some of the general properties C?f the solutions to such problems. 9.11 PrOblems
:0
should also satisfy an infinity condition that has yet to be completelY specified. The other problem of interest is that of uniform steady flow past a fixed rigid body· (Fig. 9.8). The mathematical problem is again to determine the velocity potential ~2(r) as the solution of the equation (9,SS) in the region R exterior to the body such that the solution satisfies the condition

or Coacern
The problem. of irrotational motion of an .ideal ~uid, a.~ .we have .sc:en, reduces to tbatof solving Laplace's equation With specified· aUXIliary conditions. Solutions of Laplace's equation are known as har.monic functions . . We shall, therefore, be concerned with harmonic f~nctions and their properties. In learning about !>ome of these properties we shall con:.;der specifically problems that are of most interest to us, There are two such problems which, although appearingdistinct, are actually equivalent. One of the problems concerns ~he .motion. arising out of a solid body mo~ing through an infinitely extendmg Ideal flUid. The motion of the ft~id is entirely due to that of the body (Fig. 9.7). T,he mathematical problem is to determme the velocity potential <1>1(r, t) as the solution of the equation (9.53) in the regitJn R exterior to the body, such that the solution satisfies the boundary condition
0cI>% = grad "'2 • D == 0 ~
an
on
s
(9.56)
and an infinity condition yet to be specified. Now the surface 8 is not a function of time. The two problems are of equivalent form. To see this, let us write grad <1>2 = U where, <p = <p(r)
IS
+ grad <p
(9.57)
another scalar function of position defined by the
R
0cI>1 = grad ~1 • n = Ub(r, t) • D
an
on
s
(9.54)
Fig.9.8
Steady foow past
d fi.\;.j
body.
where 8 == 8(#) denotes the surface of the body, D the outward normal, and U.(r, t) Ole wlocity of the surface points of the body. Tbe solution
• Note that the uniform flow could as well N: time depcnde!1! and the following considerations are equally applic'lble to that situation ..
IdealFluid Aerodynamics
Irrotational Motion
Eq. (9.57). Equations (9.55) and (9.56) now take the form
V~O
(9.58)
and
a. _ an
grad ••• _ U .•
on
s
(9.59)
R_COIICiJdk I11III 1"_ctnlCildk Patlu. A path is any line joining two points of a connectcc;t region. Consider the region exterior or interior to Ii finite closed surface (Fig. 9.9). Let P and Q be any two points ,and let Cl • Ct. C. be any three paths lying in that region and connecting P and Q. Any two of the paths can be made to coincide with each other by continuous variation of the paths concerned. and without ever leaving the
Same form as (9.S3). and (9 ..54). We rdAY. therefore, state that the problem is to determine cz, as the solution of We equation (9.60) VIcJ»== 0
~ equations are of the
in the region R exterior to the body such that the· solution satisfies the boundary C9ndition·
s
(a'
~
an
grad cz, ...... !(r, t)
on
s
(9.61)
and certain condi~ions at infinity. For the first problem we have and
j(r, t) .... U.(r, I) ••
For the second problem we have
(9.62)
R
cz, := ~r)
and
j(r;I):=
U··.
(9.63)
The mathematical proble~ represented by (9.60) and (9.61) is known .as the NeUl7ftli.,n exterior problem. We shan consid~r so~e general pro~rtles of the solution. (oJ such a problem and the speclficatlon. of the condltl~ns at infinity. To proceed we must'first 'ntroduce a few SImple topolgglcal notions.
9.11 Some fopoluclcal Notions
CtflfMcti"ity. A region oj space is.said to be c~nnecled ifany two points o that s ace can be connecte.d bya continuous Ime (or p~th) that ~oes not if h P . J ' s ", the region For example the regions I.ltenor and leave t e bounuarre VJ • ';' T h I exterior to a closed surface, individually, are connected regJOr~! h~ w ~ e region of space consisting tog.ether of the interior and extefl"i regions IS, however, nota connected regIon.
$
Q
(b,
FiI.9.9 Paths defined'by a finite closed surface; (a) interior paths; (6) exterior paths.
Tne function [in the following should not be confused with [(I) in the Bernoulli
equation.
region under consideration. The paths are th~n said to be reconcilobl,. We note that all paths between any two points in the region exterior '" interior to a finite closed surface aie reconcilable. Now consider the regions exterior and interior ~o the surface of an infinite cylinder (Fig. 9.10). The cylinder extends to infinity in both the directions normal to the plane of the figure. It i~ immediately seen that in the interior region all paths are reconcilable. Let P and Q be any two points in the exterior regi')n R and as shown let C!. C~, C. be any t!uee paths lying in that region and connecting"P and Q. We see that the paths C1 and C2 can be mad!' to coincide by ~Jntinuous variation withot.\t e.vet
154
IdealFluid Aerod)"Wlli<:s
R
Irrotational Motion
255
The circuits are then said to be reducjble. We note that all circuits in the region exterior or interior to it finite closed surface are reducible. Now consider the regions exterior and interior to the surface of an infinite cylinder (Fig. 9.12). It is immediately see~ that all circuits in the interior region are reducible. Let ~l and ~I (or ~a and ~.) be, as shown, any two circuits, one of ~h~. nrmely ~h. enclosing the cylinder, the other, namely ~I. not enclosing the cylinder. The latter is r.educible. The circuit ~l cannot be contracted to a point without leaving the region R.
Fig. 9.10. Paths in the region exterior to the surface of an infinite cylinder.
leaving the region R, and they are therefore reconcilable. The paths C1 and C 3 • or equivalently C. and Ca, can never be made to coincide without leaving the region R. The best we can do without leaving the region is, as shown by dotted lines in Fig. 9.10, to approach the boundary of the cylinder with C1 or· Cz brought to one side and C, to the other side. The p:lths C1 and Ca, and C2 and C 3 are said to be irreconcilable. We note that there are only two irreconcilable paths in the region exterior to the infinite cylinder. . Summing up, we state that any two paths in a connected region are reco'lcilable if they can be made to coincide by continuous variation without ever passing out of that region. They are irreconcilable if they cannot be made to co:ncide withollt leaving the region. Reducible and Irreducible Circ"ita. A circuit is any closed line in a con. ected region. Consider the region exterior or interior to a finite closed surface (Fig. 9.11). Let '~'t and <'C z be, as shown, any two circuits. By continuous v~riation (or deformation) each of the circuits can J>e contracted to a p~int without ever leaving the region under consideration.
Fig.9.1l . Circuits in the region exterior to the surface of an infinite cylinder.
c~,
Fig. 9.11
The best that can be done is to contract it, as shovyn by the dotted lines in the figure, to the boundary of the cylinder. The circuit 't'1 (or <'C ,) is, therefore, said to be "reducible. We note that in the region exterior to the cylinder there is only one irreducible circuit. namely the circuit enclosing the cylinder. Summing up, we state that any circuit in a connected region is reducible if the circuit' can be contracted to a point without ever leaving that region. A circuit that cannot. be contracted to a point without leaving the region of interest is irreducible. We note that if P and Q are any two points in a connected region then any two reconcilable paths joining the points form a reducible circuit. Any two irreconcible paths joining the points form an irreducible circuit. Reconcilable and Irreconcilable Circuits. Consider the region exterior or interior to a closed finite surface and any two circuits '&'1 and <tf2 in that region (Fig. 9.11). The circuits can be made to coincide by continuous variation without ever leaving the region under considerafion. We say that
Circuits in regions defined by a finite closed surface.
156
IdealFluid Aerodynamics
IrrotatioDaI Motion
p
151
the circuits are reconcilable. We note that all circuits are reconcilable in the region exterior or interior to a finite closed surface. Now consider the regions exterior and interior to the surface of an infinite c;ylinder (Fig. 9.12). We note immediately that all circuits in the interior region are reconcilable. Any two circuits such as ~. and ~ , in the exterior region R can ~ made to coincide by continuous deformatio'n without ever leaving ~the region R. Similarly, circuits such as ~1 and ~a can be made to coincide with,each other. We say that ~1 and ~J' and ~I and ~, form reconcilable circuits. The circuits ~ 1 and ~2 (or equivalently ~ J and ~ ,) cannot be made to coincide without passing out of the region 'R. We say that ~1 and ~1 are irreconcilable circuits. Summing up we ~tate that any two circuits in a connected region are reconcilable if they can be made to coincide by continuous variation without ever leaving that region. Two circuits are irreconcilable if they cannot be made.'o coincide without le~ing the region under consideration. We note that if ~1 and ~. are any two reconcilable circuits lying in a connected region, they can beconoected by a continuous surface lying wholly in that repon, the circuits forming the boundaries of the surface. For irreconcilable circuits no such surface is possible.
Simply Co".ctetl Rqitl& A cOllMCted region in which all patlu connecting any two points 0/ the region are reconcilable is said to be simply connected. In Such a region all circuits are reducible and reconcilable. The regions exterior and interior to a closed finite surface and the region  interior to the surface of an infinite cylinder are simply connected regions. The region exterior to the surface of an infinite cylinder is not a simplyconnected region. ' Do.llly Couected RqiD& Consider the region exterior to the surface of an infinite cylinder. Of all the paths that can be drawn connecting any two points of the region there are, as we had seen, only two paths (such as C1 and Cs of Fig. 9.10) that are irreconcilable with each other. The others are either reconcilable with one another or with one of the two irreconcilable paths. In view of this situation we speak of the region exterior to the surface of the infinite cylinder as doubly connected. We state that a connected region in which there are only two irreconcilable paths and n"o more is a doubly connected ;egion. In such a regio~, there is only one irreducible circuit. In the case of the infinite cylinder it is the circuit enclosing the cylinder. Multiply Connected', Region. A connected region in which there are irreconcilable paths or." 'equil'alently, irreducible circuits, is said to be a multiply connected regian. If there are n irreconcilable paths or, equivalently, n  I irreduci.bfe circuits, the region is said to be np\y connected.
Q
FIt 9.13 A triply c:onoec:ied region.
The region exterior to the surfaces of two infinite cylinders is a triply connected region (Fig. 9.13). A multiply coo',.:.ctecl regioo may be rendered simply connectedby introducing suitJf. Ie barriers or boundaries .that may ~ot ~ crossed. Consider the ooubly connected region extenor to an mfiOlte cylinder. Insert a barrier, as shown in Fig. 9.14a, extending ~rom the cylinder to infinity and agree not to cross it. Then in ~h~ ~gton thus modified (exterior to the shaded part in the figure) all paths JOIOlng any two
1lIIniD,.
(a)
(b}
Fig.9.14 Barriers: (0) doubly connected region; (b) triply connected region.
258
IdealFluid Aerodynamics
Irrotational Motion the region R. Hence it follows that
points are reconcilable, and all circuits that can be drawn in that region are red ucible. The modified region is, therefore, simply connected. In a similar way by inserting a suitable number of barriers, a multiply connected region may be rendered simply con'lected. A triply connected region is made simply connected by inserting two barriers (see Fig. 9.14b).
r =
£.
del>
=
£.
V • cis
=0
(9.65)
9.13 Irrotadonal Motion in a Simply Connected Region
The important rol~ played by the preceding notions in problems of ikotational motion will become apparent in this and succeeding sections.
R
We conClude that the circulation around every circuit in a simply connected region is zero. Thus, the irrotational motion of a fluid in a simply connected region is motion without any circulation. We refer to it as acyclic motion. Now consider any two points P and Q in the repon 1(. Let C1 and C 2 be any two paths joining Q to P. Since the circulation around the circuit formed by C1 and C1 is zero, we h a v e "
f
p
V·cIs
fP
V·ds(\
QvlaCI
QvlaC.
In terms of the potential this becomes
f
Fig. 9.15 Illustrating the properties of irrotational motion in a simply connecte<l region.
p del> Q via C'I
fP
Q
del> ::.: 0
(9.66)
via c.
[el>{P) : 'P(Q)1vtacl = [el>(P)  el>(Q)1vlac.
We C9nclude that the integrai in (9.66) is independent of the path connecting Q to P and that consequently el>( Q) and el>(P) should be single valued. We state that the irrotational motion of a fluid in a simply connected region is characterized by a single valued velocity potential.
Consider irrotational motion in a simply connected region R such as that exterior to a finite body moving through an infinitely extending fluid (Fig. 9:15). Let «I be any circuit drawn in that region. The circulation around 't' is given by
9.14 Irrotational Motion in • Doubly Couuected Region
Consider the irrotational motion in a doubly connected region R such as that exterior to the surface of an infinite cylinder moving through an infinitely extending fluid, the generators of the cylinder being normal to the plane of motion.· Let ~l be any reducible cir.cuit (Fig. 9.16). A !!urface 0"1 whose boundary is ~1 can always be drawn such that 0"1 lies wholly in the region R (i.e., the fluid) where the vorticity n is eql;al to zero. Therefore, for the circulation around the circuit ~ 1 we obtain
r =
£.
V· cis
=
£ .
tel>
(9.64)
where, to recall, V is the velocity and el> is the velocity potential. Let 0" be any open surface lying in the:: region R (i.e., in the fluid) such that the circuit «I forms the boundary of 0". Since the region is simply connected; it is always possible to draw such a surface. Now according to Stokes' theorem we have
r~, = v· cis = n· n dS = 0
~I
"1
f
II
(9.67)
f
~
V· ds =
=
If fIn
~
curl V· n dS
·ndS
:5
where n = curl V is the vorticity. But the vorticil;'
"
We conchide that in a doubly connected region the circulation around all reducible circuits is ze;o. Let 1f2 be an irreducible circuit. Now, it is impossible to draw a surface whose boundary is ~2 such that the surface will lie entirely in the region R. This means we can no longer determine by application of Stokes' theorem whether I'~ vanishes or not. Furthermore, if it does not vanish there is
• We are considering twodimensional motion.
zero everywhere:: in
•
160
IdealFluid Aerodynamics
IrrotatioDal Motion
161
no way of telling what its value is. We thus conclude that in a doubly conne.cted region, the circulation around any irreducible circuit may not vanish and its value remains undetermined. Conside~ next anot~er irreducible circuit such as ~I shown in Fig. (9.16). It IS now poSSible to draw a surface, say a, whose boundaries
We now consider the properties of the velocity potential in a doubly connected region. Let P and Q be any two points in the region R (Fig. 9.17). Let C1 and C. be any two reconcilable paths joining Q and P. Then since the circulation around the reducible circuit formed by C1 and C. is zero, we obtain
f
p
QvlaCa
y·tIs
IP
OvIaC.·
V·dsO
C3
Fla· 9••6 D1ustratmg the properties of irrotatiolUll motion in a doubly c:onncc:ted region.
Q
are the circuits ~I and ~I such that the surface lies wholly in the region R (Fig. 9.16). We then hav!'
Fla. 9.17 Illustrating the multivaluedness of the potential in a doubly connected
region.
where the directions of integration along ~., ~I are consistent with the indicated direction for the normal D. It follows that
v.
f
V • cis 
f
Y.
y • tis 
IIA •
•
D
dS .. 0
(9.68)
In terms of the potential we have
i
p
OvlaCa
d<'P 
iP d<'P == 0 ovlaC.
(9.70)
r Y. ==
f V. f
tis ~.
[<'P(P)  <'P(Q)JVIaCa  [<'P(P)  <J>(Q)]vla o.
We conclude that the integral
v.
V· cis ==
r Y.
(9.1'9)
Io d<J>
P
has the same value for all rtconcilable
We conclude thaI the circulation around all irreducible circuits has the same value. Similar results hold also for a multiply connected region. We shall not enter into a derivation of those results. It is thus seen that the irrotational motion of a fluid in mUltiply connected regions may be a motion characterized . by n?nze!"J) circulations around irreducible circuits. The present consideratIOns leave the values of the circulations undetermined. Additional independent considerations need to be invoked if the circulations "are to be determined. lrrotational motion with nonzero circulations around irreducible circuits is referred to as cyclic motion.
paths connecting Q and P. It' further follows that along reducible circuits the velocity potential <'P is single valued. The situation is different for irreconcilable paths and irreducible circuits. Let Ca be a path th~t is irreconcilable with C1 and consequently also with Ca. Now the circulation around the irreducible circuit formed by C 1 and Ca need not be zero; for example, it may be r. Then we obtain (taking proper account of the direction of integration for I')
( p
JQvlac.
d<'P 
iP
QvlaOI
d<f>
=
!'
(9.71 )
[<J>(P)  <J>(Q)]vlao.  [<'P(P)  <J>(Q)]viacI =
r
161
We conclude that the integral
P
IdealFblid Aerodynamics
Irrotational Motion' then have
161
reconcilable paths connecting P and Q. It further follows that along irreducible circuits the velocity potential el> ma, be multivalued. FroI:JI Eqs.
(9.70) and (9.71) we obtain
Jo
t del>' may have different values along ir
(9.75) where r is the circulation around an irreducible circuit. We thus conclude that across the barrier the potentilll jumps by an amount r.
[el>(P)  el>( Q)Jvla C.
From (9.71) we have
[el>(P)  el>( Q)]VIa C. =
r
(9.72)
[el>(P)]vla C.
= [el>(P)]VIa C, + r + [el>( Q)]v1a C.

[el>(Q)]vla C,
Now consid~r another path, irreconcilable to C 1 , such as C4 in Fig. 9.18. The paths C 3 and C 4 form an irreducible circuit. Hence, we obtain
[el>(P)  Cl>( Q)]vla C.

[el>(P)  Cl>( Q)1vla C. =
r
(9.73)
Fig.9.19 Illustrating the jump in the potential across a barrier.
9.15
Summary
Fig. 9.18 Illustrating further the multivaluedness of the potential in a doubly connected region.
From (9.72) and (9.73) it follows that
[$(P)]viac. = [$(P)]viac,
+ 21' + [$(Q)]Yiac.
 [$(Q)]vlac,
(9.74)
In this, depending on the path from Q to P, the potential at P may Jlssume many values that will differ by integer multiples of the circulation·' r. Similar results hold for irrotational motion in multiply connected regions. We shall not enter into a derivation of those results. Refer~nce may be made to Lamb (1932). Consider again the doubly connected region R. Let us now render the region into a simply connected region by inserting a barrier as shown in Fig. 9.19. Consider two adjacent points P and Pion either side of the barrier and the integral of d$ along any path C connecting P to Pl' We
• Assume that ~(Q) is zero for all paths.
We summarize briefly the main results of the preceding two sections. For the irrotational motion of a ftuid in a simply connected region, the circulation around every circuit is zero and the velocity potential is single vlllued. For the irrotational motion of a ftui" in a doubly connected region the following results hold: the circula:ion around any reducible r:ircuit is zero; the circulation around an irreducible circuit mayor may not be zero, it remains unknown;' the circulation has the same value for all irreducible circuits; the velocity potential may be many valued over irreducible circuits; the various values of the potential differ by multiples of the circulation; if a barrier is inserted to ma~e the region simply connected there is a jump in the potential across the barrier; the jump being equal to the value of the circulation. 9.16 Conditions at Infinity
We now turn to the specif.cation of the conditions at infinity for tht" Neumann exterior p'toblem. To recail, the problem concerns the irrotational motion of fluid in the region R exterior to the surfat;e S of a body (sec Section 9.11) ..We are to determine the potential <J> as the solution of the equ3tlO!'I VIet> = ('
Idealt=luid Aerodynamics such that the solution satisfies the condition :  grad cz, • n 
Irrotational Motion for a body moving through a fluid and
165
fer, t)
on S
fifer, t)dS  u·
s
.... 0
if
s
n dS
and some conditions at infinity. To determine the nature of the conditions at infinity we proceed as follows. Let~. be ah arbitrary surfaCl" drawn such that it encloses the
(9.79)
for steady flow past a ji~d rigid body. Consider the case ora body moving through the flurd. Suppose that the motion of the body is pureiy a rigid body motion; that is, the motion consists of translation and rotation only but no deformation. Let V(t) denote the velocity of translation and w(t) the angular velocity of the body, the axis of rotation passing·through some point of the body. With r measured from a point on the axis of rotation we have U.(r, t) Equation (9.78) then becomes
== Vet) + w(t)
x r
Fic. 9.18 Illustrating t~ determination of infinity conditions.
iffer,
S
t) dS  Vet) .
if
S
DdS
+ w(t) .
if
8
r x DdS
(9.80)
surface S of the body (see Fig. 9.20)~ Now, according to the divergence theorem we have
0
If the body is also deforming. the integral
if .... cz, ... as if
d
E
lI'ad cz, • b
as
if
S
U.(r, t) • D as
IfJ
R
R
divgrad
~ dT·
(9.76)
S
where R is the volume of the region includedbetw=n the surfaces S and ~ , and, as shown in Fig. 9.20, D denotes the outward norm~l on the surfaces S and~. It follows that
need not vanish, there being in general a contribution due to the deformation part ofthe body motion. The contribution is nonzero only if the deformation is such that the volume of the body is alti!,ing. With these colisiderations in mind, we shall distinguish two cases: one that of a rigid body, whether in motion through the fluid or fixed in a flow past it, and the other that ofa deforming body. In the latter case by deformation we shall mean only volume changes. We conclude that
iffer,
s
It follows that
t) dS == 0,
for a rigid body for a deforming body (9.81 )
If
t
grad <l» • n dS 
if
8
'grad cz,
~ D as ...
fi fer,
8
t) dS·
(9.77)
" 0,
Now from Eqs: (9.62) and (9;63) we obtain
if
t
grad cz, • D dS = 0,
for a rigid body
off fer,
'"
t) ciS =
fi
S
U.\r, t). n as
(9.78)
(9.82) # O. for a deforming body The integral over ~ will therefore vanish if the body is rigid or will be finite and equal to a definite function of time if the body is undergoing
166
IdealFluid
Aer~ynamics
Irrotational Motion
267
volume changes. Now, the integral over ~ must be independent of the shape and location of the arbitrary surface~. This means grad <II should behave 'in a certain way, with distance from the body such that Eq. (9.82) is fulfilled. To examine this behavior, consider first a finite threedimensional body. The region exterior to the body surface S is then simply connected. Let us choose 1: to be the su~face of a sphere with its center at some point close to the body. Choose the center of the sphere as the origin of coordinates and introduce spherical coordinates r, 8, qJ. Equation (9.82) then becomes
where, as shown in the figure, D is the normal to rt' and ds is an element of the circuit rt'. Now choose <;' to be a circle with center at some.point d05C to the body, the center as the origin of coordinates and introduce cylindrical coordinates r, 8 in the plane of motion. Equation (9.86) then becomes
J.
a<II r dO = 0,
ar
~O,
for a rigid cylinder but finite, for a deforming body .(9.87)
cln:1e
sphere
J! or
i( iJ<J> r2 sin 0 dO dqJ = 0,
jII!
for a rigid body but finite, for a deforming body (9.83)
0,
It follows that O<f>lor should behave at least as follows as r goes to infinity: as r ....... 00 if the body is rigid (9.84) O<f> .,,1 if the body is undergoing rl • volume changes
or
It can be shown, by the application of the curl theorem, Eq. (2.127), to the vector field grad <II in the region R included between the surfaces Sand ~, that the derivatives 0<11100 and O<f>liJqJ should go to zero as I/r 2 as r goes (0 infinity: O<f> O<f> 1 (9.85) """' asroo
JIIi. 11.11
lUustratiDg the determination or infinity conditions for twodimensional flow exterior to an iDftnite cyliDder. .
It follows that iJ<I>lar should behave at least as follows as r goes to infinity:
as r
00
OqJ r' This is true for rigid and deforming bodies.
iJO '
if the cylinder is rigid (9.88) iJ<I>l
R.ecall that in the problem of the body moving through ttie fluid the conditions (9.84) an<U9.85) apply on the actual velocity potential, while in the problem of flow past a fixed body they apply on the potential t/J defined by the relation grad <II = grad t/J + u (see Section 9.11). Consider now the two dimensional flow in the region exterior to an infinite cylinder. Let ~o denote the boundary curve of the surface S of the cylinder, the curve being obtained by the intersection of a plane of motion with the cylinder (Fig. 9.21). Similarly, let rt' denote the boundary curve of the arbitrary surface 1:. Equation (9.82) then takes the form
"""'or r'
if the cylinder is undergoing volume changes
To examine the behavior of the derivative M>loO, we consider the circulation arouDd the circuit rt'. Since rt'. is an irreducible circuit (th~ region exterior to ~. is doubly connected) the circulation around rt' may Dot vanish. We bave)r~ Jr~. Now, as before, we choose a circle for rt' and obtain
...( grad4> ...
_1..
grad<ll ..... r
(9.89)
r"
I
grad <II • n ds
== 0,
for a rigid cylinder but. finite, for a deforming cylinder (9.86)
leads to the same result.
i
cI~le
1 a<II   rdO =
r 00
r
(9.90)
• Note that the application or the curl theorem to grad ~ in the region between .... and ...
161
IdealFluid
.Aerod~
Irrotational Motion we have the following results: (u,  U,),
(u .
169
We conclude that iJ4>/08 should behave as follows:
as"
00
'iJ4> 1  .......,08 r'
,...., finite,
if
r
== 0
(9.91)
1 u • ••)• r
(9.94)
iff' F 0
. where U., U" U. are the components of thf! uniform stream V, and u., u" u are the components of the actual velocity. For ftow past an infinite • cylinder the corresponding results are:
•
We note that in the problem of the cylinder moving through the ftuid the conditions' (9.88) and (9.91) apply on the actual velocity potential, while in the problem of ftow past a fixed body they apply on the potential t/>. 9.17 Velocity Components at Infinity The behavior of the velocity components at infinity may now be stated. Consider first the motion arisillg out of a body mo!ing thr'ough the fluid. For a finite body we have the following result: if body is rigid
.
.
1 (u,  U,),....,1 ,....,r
rl
ifr .. 0 if r " 0
(9.95)
"" ' I
r
if body undergoes volume changes
(9.92)
where U. arid U,. are the components of the uniform stream V, and u.and u, are the oomponents of the actual velocity. We conclude that for :the steady flo,,!, post 0 fixed rigid body the fluid velocity at infinity is equal to the original uniform ve,,,city. 9.18 SomeFartber Properties of Irrotational ,Motloit
Simply COlUWcted Regio". We shall now record some further pro~rtieS of irrotational motion that are of particular in!ttcst to us. Consider first rilotion in a simply connected region. . (1) The potential 4» can neither be a maximum nor minimum in the interior
and . 1
"",.". :1 r
where u., .u,. u. are components of the velocity. For an infinite cylinder the corresponding results are: if body is rigid
of the fluid.
Consider a pointP in the interior of the fluid. Let 6T be an infinitesimal volume element surroundingP. Let 6S be the surface of the clement. We then have
1 ,....,_.,
r
if the body unQ~rgoes volume changes
(9.93)
if:
IS
dS 
~ grad 4» ;. dS
.S
.. 6T di" (grad 4»)  0 This shows that in the immediate· neighborhood of P t1te spatial derjvative iJ4>/on for all directions of the outward normal n can neither be v/holly positive nor negative. Hence we conclude that 4» can neither be a minimum nor a maximum at P. . ' It follows .that the maximum and minimum values of 4» will occur only on the boundary of the motion.
u,  , r'l
~;
1
if r ifr
== 0
~O
1
r
where u., u, are the components of the velocity. We conclude that in the problem of a body moving through a fluid the fluid velocity at infinity is zero. Consider next the steady ftow past a fixed rigid body. For a finite body
(2) The spatial dcrivativesof4» are also harmonic functions, that is they
270
IdealFluid Aerodynamics
Irrotational Motion
211
satisfy lAplace's equation. We have
or
V = grad~ div V == VI(J) == 0
curl V == 0 From the
las~
~
i
8
grad CI>•• DdS
=fff (grad CI»ldT
B
relation we obtain
Such a relation also holds for each o(the derivatives 04>/iJz, 04>/ar, 04>lih since they are also harmonic. Adding the Corresponding re!ationsJor the derivatives, we obtain
curl (curl V) = grad (div V)  VSV = vsv
Hence it follows that or,
~ff grad Vi ~ D dS
8
Iff
R
[(grad
U)I
+ (grad vt + (arad w)l] dT
(9.%)
>0
where,
V· (grad CI» == 0
In Cartesians, we have
u
aCI>
VI(:) == 0 va(:) == 0
V
1 (:) 
OZ
04> v. 011 .04> w,
y ...
0
ul + Vi +'wl
az
Thus the first derivatives of CI> are also harmonic functions. In the same way, we can show that the higher derivatives of CI> are also hatmoni,. (3) The spatiol derivatives of CI> can neither be maximum no" minimum in the interior of the fluid. This follows from (1) if instead of :> we consider its derivatives. (4) The velocity c;omponents can neither be a maximum nor a minimum in the interior of the fluid. This follows from (3). (5) The magnitude of the velocity cd'nnot be a maximum in the interior of the fluid. If~ and CI> are two scalar functions of position, Green's theorem (2.139) states that
Now CODSloer a point Plin the interior pf tile duid. Let dS ~ the surface of an infinitesimal volume element surrounding P, Equation (9.~) then tak~ ,the form
j
a:
dS _
if
18
grad yt. DdS
(9.97)
IS
>0
if
S
'Y grad CI> • D dS
= fff<'y V· CI> + grad 'Y • grad CI»
R
dT
substitute CI> for 'Y. We then obtain, noting that VI(J) = 0,
ff
S
CI> grad CI> • D dS =
IIf
R
This shows. that in, the immediate neighbOrhood of P the ~patial derivative avalon for all directiOns of the outward normal Ii can never be wholly negative. Hence we conclude that J'I or equivalently' the magnitude of the velocity can never be a maximum at a point in the interior of die fluid. It follows that the maximum of VI will occur on the boundary oftbe motion. It may be noted that no Conclusion isorawn here regarding the possibility of the magnitude of the velocity attaining a minimum within the fluid. (6) The pressure attains its minimum at tire boundary of the fluid. The pressure at any point is given by the Bernoulli equation '
(grad CI»I dT
,   p ( '
aCI>
va), at + ~ + f(t) 2
m
IdealFluid, Aerodynamics .
Irrotatiooai Motion
113
where the effect of the body forces is not included. Consider as before a .point P in the interior of the fluid and let lJS be the surface of an infinitesimal volume clement lJT surrounding P. We then have
This being the case. we can readily show that cJ)1  $, m~t be a constanL Using Green's theorem (2.139) we obtain .
fJ ~:
18
dS
=
==
.s
fJ
_p
IfI
R
[srad(cJ)l  cJ).)J' dT .....
i
~ ~
(cJ)1 :.... cJ).) grad
(~1 
cJ).)'.ia dS
grad P • D dS
I
~p.! Ie grad (J). D dS  e t( grad,V
otJtr .s
2Jtr 'fl
l
•D
dS
:=
:= 
at ., pIe oVa 2Jtr Tn dS
•s
E. fIJv~ dT  e t( pv
if
III
or or cI>.  cJ)1
R
 Jtr (cJ)1 Jt .
B
$.) 1.. (cJ)1 
on
«.) dS
(cJ)1  $.) grad (cJ)1  cJ).) ... d$'
2Jtr on .Il
dS
where l: is an arbitrary surface enclOsing S. The integral over S vanishes on account of the boundary condition: If we let the arbitrary sUrfaCe l: go to infinity, the integral over l: also vuishcs on accOunt ofthc behavior of the integrand at infinity. Hence'we obtain . [grad (<<I 
<0
The last step follows from (9.97).' . ., 'This shows that in the immediate neighborhood of P the spatial denvative for al~ directions of the outward normal D can never.~ wholly positive. Hence wo conclude that the pressure can never ~. a mlDlmum. at a point in the interior of the fluid. It follows that the miniMUm pressure will occur on the boundary of the motion. . (7) The solution of the Neumann exterior problem, in a ~mply connected
cJ)~)r dT !;Ii: 0
aplon
+ a constant
(9.98) ,
The constant may depelld'on time. TJiisis the dcsitcd result.
region is IIIIit:pIe up to an additive constant. To recall. the pr081em is to solve the equation
V~O
DodIy Co. .~4 RqitnI. ,The properties described in (I) to (6) above arc also valid' for irrotational motion in a doubly connected region.
The uniqueneSs property given·in (7) ~uircs modification. Consider the Neumann problem in the doubly connected region exterior to an infinite cylinder. Let tt0 denote the boundary curve of the Cylinder (Fig. 9.22). The problem'is to det.ermine cJ) the ~Iution of
in the'region R exterior to the surface S with the condition
(JcI) _
on
fer, t)
as
.
Qn
S
VIcJ)  0
in. the region R exterior t~ tl. with the condition
and certain conditions on the derivatives of ~ at infinity. as given by Eqs. (~.84) and (~.85). Suppose that there arc two sol~tions cJ)l, and cI>. satisfying the equation and th~ auxiliary conditions. ConSider the difference cJ)1  ~.. We have V'(<<l  cI>.)  0
:
!(r,'O
on
~o
.! (cI>l 
on
cI>.) 0 on S
The behavior of the derivativ~ of cJ) as r + 00 is as given by Eqs.··(9.81S, and (9.91) . Suppose that there arc two'solutions cJ)1 and cJ)" Then the di1fcrcncc ".  cJ)1  cI>. is a solution. of
Furtherritore, the derivatives of (cJ)1  cJ)1) behave in a certain way at
infi~ity.
174
IdealFluid Aerodynamics
Irrotational Motion
17$
in Rand, satisfies the condition
Now we have on·
U7
a'Y __ 0 an
'''0
'1'.+  (cI>J.+  (cI>J.+ b_ = (cI>I)._  (cI>.>._
'Y
Furthermore, as r _:00 the derivatives of 'I' behave in the way required by Eqs. (9.88) and (9.91). To e~amine the nature of 'I' in R, we wish to apply, as before,: Green's theorem (2.139). This theorem applies to singlevalued functions only. However, the region R is doubly connected, and
are single valued.
Although cI>1 and cI>1 are multivatued r the derivatives This meanS that
acI>Ja"l ai;rd acI>alanl
a'Y) (onl
It therefore follows that
) a
Ir+
~ onl.~
(0'1') ,
ff (gr~d ,,)' dS ;:: I {[(cI>I)L barrier
(cI>I).J ' [(cI>J._
+ (~i>.+]} ':: tis
I
The difference in the potential across the barrier is simply equal to the circulation around an irreducible circuit. Let the circulation correSponding to cI>1 be denoted by r l. and that corresponding to cI>. by r,. We then have 'fJ(grad '1')' dS =
•
, '
(rI ' r I)
ar.rrler
f a'Y ds on
1
.... 9.11 Illustrating 'the uniqueness proof for the Neumann problem in a doubly c:oanected region. '
or,' in terms of cI>1 and cI>.,
therefore cI>1' cI>" and 'I' may be many valued. We can make th~m single valued by inserting a barrier as shown in Fig. 9.22. Let't. be an arbitrary circuit enclosing i",. Applying Green's theorem (~.139) to 'I' in the region (I ~een i"o and ~'. we obtain
ff [grad (cI>1 ' cI>JJI dS = (rl •
r l)
f
'o~
1
(cI>1  cI>1) ds
(9.99)
barrier
. 'fo proceed further we must specify the circulations are equal, then <1>1 =: cI>1 a constant
r 1 and rl'
If they
f.J(grad
."
'1')1 dS 
I
'I' 0'1' ds
anI
+
I'f' a'Y an
+
ds
ODAB
 f 'I' a'Y ds f'Y 0'1' ds on v, on' ,
oileD
I
wbere 8 1 and 8 respectively are 'the normals on the barrier and the'circuits i"o and~, and tis is an element oflength on the boundary of (1. The integral over ~o vanishes sincea'Y/on is zero on ~o. We now let ~ go to infinity. In view of the behavior of 'I' and as r  00, we conclude that the integral over ~ vanishes as r ~ 00. Hence we obtain
a'Y/on
11+
.
f.J
•
<grad '1')1 dS  /.
f'Y 0'1' ds +f'Y a'Y ds anI anI
b
whr:re b+ denotes the CD side of the barrier arid b_ the "B side of the oarrier.
If r I and 1'1 a~e not equal, ,then cI>. and cI>1 are not equal. We thus conclude that the solution to the Neumann 'exterior pwolem in a doubly connected region is uniquely determined (up to a" additive constant) only when the circulation is specified. For the same b.oundary and infinity conditions, different values of the circulation yield d!fferent solutions. The uniqueness theor,ems for the solution of the l Neumann exterior problem are of considerable importance in aerodynamic theory., Without the theorems it would never' be possible to 'assert that the flow pattern ,calculated for certain boundary conditions is the correct one. For a doubly connected region the theorem shows that the circulation must be specified in order to obtain a unique solution. It must be noted that the value of the circulation cannot be specified on 'he hasis of the cOl/sideratimu giuen so far. The specification of the circulation must lher~fore rest on other considerations such as those derived from a physical u~.derstanding of the real features of the flow problem at hand.
Irrota1ional Motion
m
1 a(0'Y) V''Y,. r Or A, '1 OI'Y +0 ' 1 ,1 06
IdealFluid 'Aerodynamics in Cartesians, and 9.19 We now consider ~he relation between the velocity potential and the stream functions. andtbe equations' governing the stream functions in irrotational motion. Foi' 'incompressible motiQn the velocity field is related to the stream functions by v  grad 'I"1 x grad 'Y~
(9.105)
in cylindrical 'coordinates. Thus ,in twodimensional irrotational motion of an incompressible ftilid. the stream function obeys Laplace's equation.
AXUy1lUfteW Mftio". For such a motion we' have in spheri.:ar ordinates r, 6, rp (see Section 4.11)
c0
If the motion is also irrotational and
'Y1
It then follows that
(9.100)

'Y(r,
6)
V..,; grad"
we have
This then is the relation between the velocity potential and tbe stream functions. An equation governing 'Y1 and 'Y1 in irrotational motion is given by the condition of irrotationality :
cutl (grad'l"1
(see Section 4.10)
X
f 0'Y A, r r1sin606 10. 1 a'Y'  aB  u,     ,, ' , sin (} or
u    
a.
(9.106)
Similarly, in cylindriCal coordinates r, rp, x we have
grad '1".)  0
(9.101)
,and
, TffHJ~ M.... For s~ .. motion we have in Cartesians
'Y == 'Y(r, x) '1'. 'rp
acI» u    a,' r ,ax
It tbeo fonows that
t"
(9.107)
v'a, a,
Similarly. in cylindrical coordinates we have
'a. i1'Y ,UaX oz a. , iJ'Y
u.ax or
~
\
to'Y
'r
(9.102)
Then equation (9.101) becomes
_1_ O"Y
sin (}
or
+! i.(_1_0'Y)
rl iJ9 sin 6 '06
_ 0
l
(9.108)
in spherical Coordinates r, 9, " ~nd
'Y1  '1"(" 9) 'Y.%
and
O"Y =!o'l° +~=o
or'
r
a,
or
(9.109)
a.
1 0cJ)
  ur   09 A, ,
1O'Y
; 00 The eqlijltiori (9.101) become:
u,  
0'Y iJr
(9.103)
in cylindrical coordinates, r. ". x. We note that the equation for the stream function in axisymmetric inotational incompressible motion is not Laplace·sequation.
V"Y ~r
o;r:
O"Y
O!l:!
0
(9.104;
. Unsteady Aqdic. Motion
Chapter 10
shouldv~·in a certain way:With distanCe from the body iii that.distuc:e tends to infinity. Henceforth. we shall refer to this i'equimrient 'as tile
infinity condition..
Unsteady AcyClic Motion
fu~
Note that the \)ouodary condition (10.2) may be expressed also in the . .
or
.!!  grad.~ •• un
VCr. i) ••. on
F(r. t)  0
(10.3)
grad~'gradF .U"gradF on F(r,t)·O
We now begin the considerations lor ConstruCting the ~llItions of flow problems associated with the motiod of a solid body through an ideal ftuid. Our tiftal aim is to determine the pressure distribution over the surface of the body and the forw and moment .aing on the body. In this chapter we shall concern ourselves with a few simple example$ and some general results concerning the acyclic motion resulting from the motion of a rigid solid body translating through an in6nitely extending ideal fluid.
This states that at every point on the surface of the body the compoQeDt normal to the body oftht fl~ velocity is equal to the normal <.ompOnent of the body velocity. Once the velocity potential is .d~ed. the pressure at any point is given by
Per. t) 

pr~
+:i <.... ~,] +
d
f(t}"
(10.4)
tC.l Matltematid1 ProbIeat
Our problem is to determine the velocity potential.for the ftuid'motion resulting from the motion of a solid body through an infinitely extending ftuid that was initially in a state of rest. We choose a space fixed reference frame and describe the surface of the body as .
The effect of the body forces on the pnssure is not included in this relation. Such effect. as explained before, may be accounted for by calculating the hyclrostatic.p~ure under the action ·of the body forces and adding it to the pressure calculated by (10.4). D~note the pressure at.infinity.which is a constant, by Pio. Then the relation for the pressure may be "'"tten as
p(r. t)  p ...  p [ ~
JOlr. t)
 0
a.l.
,he
and the velocity of an element of the surface of the body as V. In general. the body may be translating. rotating. and deforming. Consequently. V may in general be a function of position on the surface of the body anei time. If the body is rigid and is in translatory motion only. then V is a function of time but uniform over the surface of the body. For the' problem under consideration we shall denote the velocity potential by ~ rather than by ~ which :las been employed generally so far for the velocity potential. The mathematical problem is to dete~ine '" as solution of the equation VI", _ 0 ( 1 0 . 1 )
1 .. :1 + i Wad ;)I . J
(lO.S)
10.1 Expudiag Sphere _ . We no~ ·consider. a simple. example of unsteady motion. A sphere Immersed 10 the flUId expands unifo~lyin all directions, We wish to . determine the pressure on the surface of the sphere. We choose the center of the sphere as the origin of coordinates and introduce spherical coordinates r. 8. tp. If a denotes the radius of the sphere the surface of the sphere is described by
r = a(t)
such dial tie solution satisfies the boundary condition
or by
(10.6)
(l0.7)
aF + grad~. gradF 0
at
on F(r. t)  0
(10.2)
F(r, t)
:::!
r  a(t) :;:: 0
furthermdre;thc components
ot the velocity
Vgrade/>
The boundary condition (10.2) then takes the form .
=
O,p
Or
da
dt
on r  a(t)
(Ul.8)
278
Unsteady Ac:ycJic: Motion
IdealFluid Aerodynamic:s
111
T~
In Iipt of the boundary ~ndition we rea1ize that ~ is a function of rand t only:· .
~
10.3 ProWem for a Tpmlatlna. Body .. Referaa Frame
of • Body
lb.
== t/J..r, t}
== 0
Then the equation for ~ becomes
.!.(,.s ot/J)
or or
o~
Henceforth we shall be concerned with the problem of .fluid motion resulting from the motion of a rigid body Iranslating through the fluid. For such a situation the velocity of the body does not change with position on the surface of the body although the translatory velocity may be a function of time: (10.14) U = U(t) only
For a rigid body the function specifying the surface ohhe body becomes independent of time if described from a reference frame fixed in the body. From a space fixed frame the function describing the body surface is time dependent even if the body is rigid and· undergoing only translatory motion with a .C<?nstant velocity. These observations combined with the fact that time does not appear explicitly in the governing equation for the velocity potential suggest that it will be:advantageous to consider the analysis of the problem in terms of a reference frame fixed with respect to the body. We shall therefore use such a reference frame. Consequently we shall now express the problem, which has so far been expressed in terms of a space fixed reference frame, in terms ofmeasur;:nents made from a body fixed frame. . Denote by Kl a referena:· frame fixed with respect to the moving body . We shall denote by the subscr:ipt I all measurements and operations made with respect to the frame K I • The spac;e fixed reference frame shan be denoted by K, and the measurerI)ents and operations made with respect to X shall be denoted without any subscript. We now set up the connection between the deScriptions in the two frames (Fig. 10. i). Since Xl is translating with a velocity U(.t) with respect to K we have, assuming that the two frames are coincident at time zero, rl
Integration yieldli
 .. A(I} or ,.s
+ const
(l0.9) .
and
t/l(r,t) _  .,.(t) r
The constant, which may depend on time, lIlay be set equal" to zero. We determine A(t) on the basis of the,condition ClO.8). We thus obtain
A(t) _ 41't) da. . dt and
f/,{r, ,) __
•
a',)
r
~
dt
r
. (10.10)
The velocity IS gaven by
V  JI'8d .,. For the pressure at
.l.·
""7 dt e
a'(,)da
(10.1n
any
point~e obtain
p(r, t)  Pe 
[~ + ! (grad t/J),J . ot. 2
_ Pe
+ e[.!(a l da) ~ . r dt dt ~
!(a' da)rlJ d,
.1
== r 1(r, t) == r

(10.12)
L~U(T)dT
(10.1 S)
Th~ pressure on the surface of the sphere is
given by
(10.13)
t For the potential t/J~. in Kl corresponding to cfo in K we have
~1(rl> 11)
'1  ',(r, I) 
p[a"(t), t] ,Pe + ~[3e;J + 2a ~;:]
= ~l[rl(r, I), , 1(ra, t)]
..
 t/>(r, I)
(10.16)
The pressure is uniform over the surface of the sphere and consequently' the fluid. exerts no· resultant foree or moment on the sphere. Work, however. has to'be done in expanding ~~ sphere against the pressure f~rces. Theexpancling sphuc is a simple example of a body undergoing volume cha~ges. For ~tlch a body, w~ recall, the velocity compone~ts should vaaish at least as IJrI as r  00. This is bom: out by the solution (10.9).
Similarly for the function Fl describing the body surfaccin Kl and for the pressure PI in [(1 we have
Ft(r t , t 1) = F1[rt(r. t), tl(r, t)] = FCr, t} PIeri' 11) = p,[rl(r, I). 11(r, r)] = per, t)
(10.17)
(10.18)
l.Z
IdealFluid Aerodynamics
Unsteady Acyclic Motion
It follows that
or (10.23) We conclude that in terms of the frame Xl the problem is tPI(r1> t l ) as the solution of the equation . V IltPl
10 determ;n~
== 0
on £I(r l)
(10.24)
such that the solution satisfies boundary condilion
VltPl . D == lI(t). D
== 0
(10.25)
Furthermore. the potential tPl and the components of the relocily
ql" VltPl (10.26)·.
The various differentiation operations· in the two frames·are· related as
follows: . V.  VOl
V' .. Vii
. (10.19) (10.20)
+~.·V
should satisfy the ;nfinityr:ondilions.· Once the potential tPl is determined the pressure at any point is obtained in terms of the measurements made in frame'K1 by.the relation
p{r. t)
Since from (IO.IS)
. and
weobtaiJl
ot ot ot ot 1 ot .
l _.
o Wt1 0 .l
ar1
ot
I
(10.21)
= PI{rl • 'I) == P«>

p[~~:  U(t)· V 1tPi + ~ (Vl'l)']'
(10:27)
 .. 
o 0ot a"
U{t) • VI
.
(10.22)
This is obtained from (10.5). , We observe that in terms of the. description in frame Xl time enters the problem for the potential only through U in the boundary cqndition. We state that the time dependence of the potential comes through the time dependence of the velocity of the body. If the body is translating with a ~onstant vdocity. then the potentiar described in terms of a body fixed reference frame is time independent. The problem for the potential in terms of.the body fixed frame may be iriterpreted as a certain flow past a fixed rigid body. To see this. let us first write the boundary condition (10.25) in the form [ U(t) or
( U(t)
Intcrms of the frame Xl the equation for the potential bc:co~es
+ VltPI] '0== 0
0 on FI(rl)"
VII+. .. 0
The boundary condition takes the form
+ qJ . D =
0
(10.28)
Introduce now a velocity Viand I. potential Cl>1 such that on VI = VI$I == U(t)
(oF
or
ii: 
 ) U· VIFI + VltPl·V IFI = 0
+ ql =
U+; VltPl
Equation (10.24) and the boundary condition (10.25) then take the form
,\"2$1 = 0
(10.29) F(r)
and V$j . D
=
VI'
D
=0
on
=0
lIO.30)
IdealFluid
Aerod~
Unsteady Acyclic: Motion the sphere due to the reaction ofthe fluid. The sphere is or"radi~s a. The velocity of the sphere is time dependent, thal is the sphore is in accelerating motion. W~ choose a bodyfixed reference frame and introdu~ spherical c0ordinates r, 8, qJ with origin at the center of the sphere (Fig. 10.3). The axis
From (10.2&) it follows that'
"(Vl~l  V.),(U)
as r
00
(10.31)
The ft!lw problem represented by (lO.28, 29, 30) is that for flow past a fiXed,rigid ~y, the velocity at jnfinity being U (see Fig. 10.2). It is thai at each' instant the flow field corresponds'to that of steady flow past the fixed body, the uniforln velocity at infinity ~ng the negative of the velocity of the body at the i~stant under consideration. The only difference between a strictly steady flow past a fixed rigid body and that 'of the translating body as described id the bodyfixed frame is that in the latter
seen
'I1(t)
.~
~
".
/
Fla. 1'.1 Flow in the bodyfixed reference frame.
'111\
\
I
'"
I
I ...... _"", /
case the velocity at infinity changes from instant to instant. In this sense the solution for the flow problem for a transiating body and that for the corresponding steady flow past the body is the same. These observations are in line with the considerations in Chapter 9 where it was shown that for an irIotational motion characterized by a singlevalued potential the state of motion at O!le instant is independent of that at another instant. . When the problem for a iranslating rigid body is viewed in terms of the bodyfixed frame as flow past a fixed rigid body, the potential· 4>1 is referred to as the disturb.ance potential. This signifies that 4>1 arises due to the . disturbance by the body of an originally un;form stream. The velocity
"
Fla. 10.3 Coordinate system for • transJatiag spheR.
from which 8 is measured is chosen opposite to the bod>, velocity ~. The bouJ'ldary condition, * from (IO,2S), is on This red uces to ,  a
Ucos8 on ,=a
at/> . '.
a,
(10.32)
ql
s tllen known ,relocity
3S
= Vl4>l
<J)1
qJ;
In light" of this we expect the potential to be independent of the coordinate that is, we expect the motion te;> be axisymmetric. Hence we have
the di.$turbance velocity. The potential
and the
t/>  4>(/, 8, I)
The equation for the potential ttaen takes the form
are called respectively the total potential and the total ,,'elocily or, simply the potential amI the velocity. We note that the disturbance potential and velocity in .the bQdyfixed frame correspond to the actual potential and velocity in the frame K. 10." Translating Sphere A rigid'spl}cre is translating through me fluid. We wish to deterl1)ine
ar
i} r
(2 .'
SID
0i}4»'
aT>
a SID +...... (. v
a8
Ii
04» 00
II
;I
. ! 10,33)
W~ construct the solution by separation of variables ("",,nsult fur instanc:e Hildebrana 1949 or, Sneddon 1957). We assume that
4>(r, 0) = R(T) 0(0)
ot1
(10 J.+)
the. prasure distribution over tbesphere and the force and moment
• We shall omit hereafter the subscript'l to denote thaI the description is with rtspect to the bodyfixed frame. for, this is the only frame we shall use,
111
The general solution of the·(10.36).replaeing k by n(II
+ I}, is of the
form
(10.3S) where k is a separation constant. This equation 'is equivalent to two separate 'equations for R aJidS:
R  C.T" '+ ~ r·+'1
where, C. and D. are consta~t coefficients. SinCe the solution for. should at Ie. ~tbe fi~ as r  00, we set C. equal to zero fIld write
A.(,.. dR) _ kR. dr •
~(sin 8 del + k8 sin 8 dB) . d8
() 0
()0.36)
R
D~ rll+1
CD
(10.43)
The solution for the potential is then of the form (10.37)
pI,.,
These are ordinary differential equatioDlwhose solutions are well known (see for instance Hildebrand). We first consider (10.37). By puttinS:
k  n(1I
~r, 8)  ,,0 r IOt"
.:1
(10.44)
where Ot" is a conslant.'Which in the present problem may bea function of time. Writing (10.44) explicitly we have
(10.45)
+ I}
(10.38) (10.39)
where II is nonnegative, and by .writing Cos8.,.,· (10.31) mav be brougbtinto the form
 (1  p'> _. dp dp
We now determine the eoefficients by application of the boundary CU)ftdition(lO.32): We have
d[
tiS] +
.11(,.
+ I)S()
.
0.0.40)
ar
a~
__ Ott _
rl
2Ot1 cos 8 _ ...
r'
where 0 is expressed as a run~tion of p. this is Legendre's d.ifferentitll equation. The general solution of this equation is ofthe form
The boundary condition requires that the following relation be satisfied
o ::A;,.P..(p) + B;.Q.(p)
(IOAI)
. at/> ate cos 8 U,COS 8 , .  (i'  a)  .  l '  2Ot 1 ~ a al
ar
.

• • •
(10.46)
where p .. and Q" are ugeNire funet/olu and A .. andB" arc constant coefficients. Th: furtetion Q..tP) .becomesinfinitc. on the ~s 8 0 and 8  fr. Hence we set the,coefficienfB.. equal to zero. The functionP..{p) becomes infinite on the axis if n is not an integer. Therefore we restrict n to integral values and write (10.42) 0 A. ..p ..<P)
This ,is satisfied· by' setting
0C0.
and
«a, Ott. • . . == 0
Ua'
ot1
2,
00.47)
We have now obtained the solution for 4> which is
JJ, ,.8 Y'\, ,
1'he fl!Jlction p ..(p) is a certain polynomiaUn p of degee n. The first few
polynomials are
Po(p)
t) _ U(t)a' cos 8 _ ~,aa.u(t). r
2,1
2,a
(10.48)
==
I
PI(p) """ p cos tJ
p.(p) P3<t~)
t(3l" \) ,t(3 cosl8 
1)
== l(5p'  3p)  !(S cost 8 
n 005 8,
We may now determine the streamline pattern for this motion. The motion is axisymmetric and the equation of a streamline in any axial plane is represented by '1'<" 8) ==. constant
288
IdealFluid Aerodynamics
Unsteady Acyclic Motion
2.9
.
drp = r sin OUB dr
Now using (10.48) we obtain
.
To determine this function" we have the reliltion that along a streamline
+ r2 sin OUr dO :::::! 0
(10.49)
10'" sin () u, ==  =Ual  = r ao 2 rl
a,; Va' 2 cos 0 u====r ar 2 r'
With these relations (I0.49) yields

Ual(sinlO d 2 sin () cos (), )' '1 r d() == 0 2 r r,
or
_U;' d( Sin: 0) == 0
tp{r, 0)   
Hence the stream function is given by
Ua' sinl () 2 r
+ a constant
== 0
(lO.SO)
The co~stant may be set equa: to zero, so that " streamhnes are therefore described by the equation sin· 6 , == constant
r
on ()  O. The
(10.51)
The streamline picture given by (10.51) is shown in Fig. 10.4. Note that this picture describes the disturbance velocity field. It is the streamline pattern observed from a spacefixed reference frame that coincides with the bodyfixed frame at the instant of time under consideration. In the preceding considerations we have omitted exhibiting explicitly the time Qependence of" and U in the various relations. . The pressure over the surface of the sphere is given by
p{a. (), t) :: p«>  p[iJ,;  V(t) .. \ iJt
FiI. 10.4 Streamlines
or the
diIturbanc:e field resulting' from a tnlnslating sphere.
We may express (10.52) in the form
~p =(p _ p )
GO
pUI{t)( == 1.2
9.  SID
4
I)
f)
. 
pa dU cos{) 2 dl
V4> + ! (V tfJ)l] at ra 2

==
where
(10.~2)
~Pl
+ ~Pz
9.
(10.54)
p[U1(t) == p.., + 2 4 {9 cos· (J
dU S)  a dr cos 0]
() and
PI
==  2 
pVt(t)(l
4SID10) ,
(10.55)
if the sphere' is moving with a constant velocity the pressure distribution over the sphere is given by put p(a. 0)= p.., + 8 (9 cost 0  5) (10.53)
()p'l.
==   
pa dV
2 dt
cos ()
(10.56)
The variation of tJpI and ()P. over the sphere are shown in t=:ig.'IO.5. It i~
190
IdealFluid Aerodynamics
Unsteady Acyclic Motion
191
seen that while ~Pl is symmetrical with respect to the plane 0 = 11'/2, ~P2 is not symmetrical. We may state that acceleration of the sphere leads to asymmetry in the pressure distribution and consequently to a force on the sphere. We further recognize that the force is entirely due to lJP2 which may be referred to ·as the. pressure distribution due to acceleration.
1.0
0.8 0.6
0.4
experiences a resistance to its motion .. The resisting force is known as the drag force or simply as the drag. If the sphere is movin~ with coos!ant velocity the drag is zere! In other words, on~ the sphere has been set mto a constant velocity motion nO further application of an external force on the sphere is necessary to maintain its motion! The present theoretical result that the force (.In a sphere moving with constant velocity is zero is of course not substantiated by experiment. We shall take up further consideration of this. result later; we note that there is no moment acting 011 the sphere: Let us now ask about the force that must be applied to the sphere by an external agency so as to keep the sphere in motion through the fluid. Denoting by F. such external force we have, according to Newton's second law ·of motion,
d (mU)=F.+F dt
or
0.2
F
0.2
0.4
•
=..!~U)F
dt
(10.58)
0.6 0.8
1.0.
Thi5 states, as we· know, that the external force should balance the inertial force of the body and the resisting force of the fluid. Substituting for F from (10.57) and noting that 11 = e.U we obtain
F. =
:t[(m + ~ )uJ
8p 1I'a
(l0.59)
1.2
e
Fla· lO~ ~ure distribution over the suda.:e of a translating sphere.
on the basis of this relation we may state that the reaction of the fluid on a solid sphere accelerating (in translation) through the fluid is equivalent to increasing the mas~ of the sphere by the amount
m' = i1Ta 1p .
(10.60)
The force on the sphere is given by F ==
J( If po dS =
sphere
pa dU
2
dt" f.f D cos 8a
I
SID
.•
We refer to this mass as the additional apparent or virtual mass. Since the volume of the sphere is (t)1I'a ' it follows from (10.60) that the additional apparent mass is half of the mass of fluid displaced by the sphere.
0 dO dgJ
10.~
Force on a T1'1IDSiating Body of ArbitrJU')' Sbape
We note that F has a nonzero component only in the dir~tion of e%, that is in the direction of the instantaneous velocity of the sphere. We find
F= 
3
~ 1I'a'p dU e
dt
%
(10.57)
We conclude that a solid sphere accelerating through an ideal fluid
The flow field resulting from the translatory motion of a rigid solid body of a more general shape than a sphere may be analyzed on similar line~ ~s those followed in the preceding section. In analyzing such flow fields It IS convenient to em;>lvy coordinates that are especially suited to the body Shapes under consideration. For the solutiofl of flow fields involving the motiGn ofso!idbodies of more general shape than a sphere reference may be made to Lamb (1932... We shall concern our.selves with some general
,191
IdealFluid Aerodynamics Unsteady Acyclic Motion
193
results relating to the force and moment exerted by the fluid on a rigid body of arbitrary shape translating through the fluid. We have seen that in the case of the sphere the fluid exerts no force on the sphere jf the sphere is moving with a constant velocity. A reaction force hy the fluid appears only if the sphere is aca:lerating. We now inquire whethc:( such results hold, for bodies of arbitrary shape also. Itis possible to find the answer without having a detailed knowledge of the flow field under consideration.
With (10.62), the relation (10.61) for the force becomes
F
==
fjp ~~
8
D
dS
+ fjp[q22  U·
8
q}
dS
(10.64)
Note that since the operations in the right member are with respect to tQe bodyfixed frame and since with respect to that frame the surface S is time independent we can write
•
itp acP 0 dS ==0 itPcPn dS . If ot ot If 8 , 8
U )( (0 )( q) = (U • q)o ..;.. (U • o)CJ we write
(10.65)
. The other integral may be shown to vanish if the motion is acyclic or equivalently if the potential is single valued .. First using the relation
j[q2 _U .qJDdS = fj[~1 0 (U ,o)q] dS 8 2 8
U)(
jJ (0 x
8
q)dS
(10.66) (10;67)
FIB. 10.6 Illustrating the detennination of the force on a translating rigid body.
According to the boundary condition (10.25) we have
U
·0
== grad cb • 0 = q • 0
on S
Consider a rigid body of arbitrary shape translating with a velocity
U(t) th~ough the ftuid. Let S denote the surface of the body (Fig. 10.6).
Hence we can write
The flUId force acting on the body is given by
F 
j,.
8
dS
(10.61)
¥2
80
it[fli 0_(U. D)q] dS == j[~ D 8 .
(q.
o)q] dS
Our immediate aim is to express F in terms of the velocity potential. It is Plnvenient, as already pointed out; to solve for the velocity potential by working in the bodyfixed reference frame. If the spacefixed reference frame is chosen such that it coincides with the bodyfixed frame at the time instant under consideration the pressure. according to Eq. (10.27), is given by . p(r, t) = p""  P [ where q = grad cP (10.63) For convenience we shall employ often q for grad!/>. The operations in the right member of (10.62) are with reference to the bodyfixed frarr.~.
Now, for any region'Ro enclosed by a fixed surface So and containing ftuid only, it can be shown that
ff[q; 0 (q. o)qJ dS ==
'
fff
R,
[q
.'Vq  q. Vql dT == 0
iJcP
ot 
U •q
+ qlJ 2'
COplider the region of fluid enclosed between the surface Sand anuther fixed surface 1; enclosing'S (Fig, 10.6). The surface ,l: is of arbitrary shape and located at an arbitrary distance from the body. We then have
(10.62)
ff[q; 0 (q. o)qJ dS '"7{f[;1 0 D)q]
(41'
8
~
dS
(10.68)
,
It follows that the value of this integral should be independent of tlte shape and location of the surface l:. We have seen that for a fimtt: rigId body, q vanJsbes as 1(,1 as 00 (see Section 9.17). The surface element
,+
191
IdeaJ..F1uid Aerodyoamica
00.
Unsteady Acyclic Motion
195
dS grows at' r· as r 
It therefo.rc follows that for a finite rigid ~y
q'dS",'!
and
,.
By means of two planes normal to e cut out a small slice of the solid body (Fig. 10.7) and pick as shown an elemental area D dS on the surface of the slice. Note that the normal D in general need not be norm,l to e. Let e1 denote the direction along the curve of intersection between the body surface and a cutting plane. The unit vectors e1 and e are therefore
If the body is a rigid infinite cylinder we have seen that q vanishes as 1/,' as r  00 if the circulation is' zero. If the circulation is not zero the ,. component of q vanishes as llr·.mit the 6component of q vanishes as llr only (see Section 9.17). The surface element dS of 1: now ~rows as, as ,  00. Putting these results together, we state that for the infinite cylinder if the circulation is zero
ql
dS and (q. D)q dS", 1. as r 
,..
00
and if the circulation is not zero
•
I",ea
a, _ and .... not coplanar.
Fit 10.7 Illustrating that a part or the rOR:eOn the body is related to the .circulation around
th~
ql dS and (q. D)q dS,...,! r
Hence it follows that the integral over 1: in Eq. (10.68) mav be made arbitrarily small, both for a finite body and an infinite cylinder ~ith circulation, by choosing the surface at larger and larger distances from· the body. We conclude that the integral aver 1: should be zero and that consequently the integral over S is zero: Hence we have
(10.69)
•
body.
It follows that the expression (10.64) for the force becomes F=
normal. Also D is normal to e1• We ~enote the e1 edge of the surface clement D dS by dl, ~nd choose the other edge say db of D dS such that
~
ifp~
R
dS  pU(t) )(
if
B
DdS == ..
(D )( q) dS;
)( db
(q • db)dl
(10.72)
(10.70)
Then we have
D d5)(
Consider next the inte~1 I, ==
if •)(
8
q
== (di )( db)
e •D
)(
)( ..
== (q • dl)db 
q dS.· This integral is related in a
It follows that q dS
certain way to the circulation around the body and will' vanish if the circulation is zero. To see this let us consider the component of the vector I in some fixed direction e. We have
== (q • dI)e • db
(10.73)
since e • dl i~ro. Let us write
dh
e •I
==
if
B
== e·db
(10.74)
e •D
)( ..
dS
(10.71)
This is the normal distance between the cutting planes.
296
ldealfluid Aerodynamic:.
Unsteady Acyclic Motion Using. the relations (10.73) and (10.74) we obtain from (10.71) . .
197
e . ~ n x q dS s .
=
f(f dl) =J
AI III
q"
dh
r.(h) dh
(10.75)
10.6 Impulse Let F denote the force applied eJ..ternally to the body to translate it through'the fluid. Then, according to Newton's second laW' of motion we have d (mU) = F,
Al
'If
III
where h is distance measured along the fixed direction e and
d,
+F
(10.78)
where m is the mass of the body. Rewrite the preceding equation as is the circullJtion around the curve of intersection between the body surface and the cutting plane situated at h. The limits hi and hz denote the extremities of the bOtty measured along the direction e. From (10.75) it· follows that if the circulation around any arbitrary circuit drawn :>0 the surface ofthe body is zero, then the component of the integral I in any direction is zero, which means the integral itself is zero. We therefore conclude thatfor motions without circulation or, equil'alently for motions characterized by a singlevalued potential the integral I vanishes:
F = ~(mU)  F
, dt Substituting from (10.77) for F we obtain
(10.79)
F. =
:,(mu  ifpl/lDdS)
8
(10.80)
I
==
if
8
D X
q dS
~0
(10.76)
Note that the integral in (10.77)" is actually a function of time only and consequently a/at in front of it. actutllly ~eans dldt . .Recall tha~ for acyclic irrota.tional motion, that IS f~r mO~lon charactenzed by ~ smgle valued potential,  ~ is equal to the ,".'pulslVe pres~e 1IT (see Sect~on ?9). The integral of the impulsive pressures IS called the Impulse. Denotang It by
We shall concern ourselves at present with acyclic motions only. We . conclude that for acyclic motions the force on the body is given by (10.77) The time dependence of the potential ~ arises only through the time dependence of the velocity U of the body. In fact we have #,.r, t) and
J
we have
J "'" and
if p~ if
dS =
8 8
1IJ1l
dS
(lO.81)
F=dt Equation (10.80) then takes the form
dJ
(10.82)
==
~[r;
F, "'" (mU
U(t)]
This states that
d dt
+ J)
(10.83)
the force applied externally on the body = the time rate orch.ange (if the momentum of the body + the time rate of change of the Impulse applied on the fluid . where o~/au denotes the gradient of ~ in the space of U. On account of ihis we may immediately state that if a finite rigid body is mOl'iftg with a constant velocity through all infinitely extending ideal fluid the force on the body is zero. This result is known. as d'Alembert's paraJox. We shall consider later the implications of this result. .The im~/se is the impulsive force retfU,ired to establish impulsively the state of motion under consideration ('See SectIOn 9.9). 10.7 1be Apparent Mass Tensor Equation 10.83 suggests that the reactio~ of th~ flu.id to a body translating through it is to change in some sense the mertlal force of the
298
IdealFluid Aerodynamics
Unsteady Acyclic Motion
299
oody. The rate of change of the impulse vector, in general, is n(>t in the direction' of the acceleration of the body. This means the external force F. has 10 be applied, in general, in a direction different from that of th.e acceleration of the body through the fluid. To illustrate these ideas and to develop computational formulas for J we now introduce tlte notion of the f!Pparenf mass tensor. Consider, in terms .of'the bodyfixed reference frame, tne matnematical problem for determining~. We have
Va~ =
(10.88) and (10.~9) and the infinity conditions, the potential ~ is obtained from . where cp is given by
~ = UI!PI
+ u,!Pt + Ua!P,.= U .cp
We have
(10.90) (10.91)
= (!PI' !Pa, !Pa) ;.ow we express the. impulse J in terms ofcp and D.
cp
,1
0
=ffp~n dS
s
, a~ grad ~ • n =  = U(t) • D
an
on
S
=ifP{U.
s s
==
cp)n dS
and certain infinity conditions. Let us choose a Cartesian coordinate system XIo xa, X3 and ".Jet elf ea, e, denote the reference unit vectors. We write (10.84) and (10.85) Since the equati':>n and theooundary.condition for solution for ,p in the form
~=~I+~a+~1
~ are
~(ffp!ptD dS )u
8
t
(10.92)
linear we seek a
Let us denote by / , the component of J in the e, the direction. Then we have (10.93) i == 1,2,3 We introduce the symbol m lri with the meaning
where each of the functions ~lo ~, ~ is a solution of the, following problem· VI~i == O· (10.86) a~, ·on S grad ~; • D == an ==
.
. ",n,
m le , ==
ffp!Plen, dS ..
8
(10.94)
where; may be 1,2 or 3. We note that it is convenient to'set ;==1,2,3 (10.87) where !Pi unlike ,pi' are scalar functions. of position only. Time enters through "t. The system (10.86) then takes the form (10.88) grad !Pi· n
Now, using (10.89) we may write the preceding equation in the form
m'le ; =
ffp!p/c ~:i dS.
s
1pt
(10.95) .
1pa
a!Pi = a;; = n
According to Green's fheorem (2.141) if functions the relation .
and
are two harmonic
(10.96)
i
on
S,'
i
==
1,2,3
(10.89)
The derivatives of !Pi and, !Pi should sati!fy ihesantetype of infinity conditions as ~ and its derivatives do. If !Pi are determined according to
• Those familiar with summation convention should note that such~onvention is Qot used here.
If' s..
i1p1 a1pa
a,,.
dS =J(1pa a1p1 dS
If' s.
an
holds for an; surface So enclosing a region in which V21pl and V21p2 are zero. We consider the region enclosed between the surface S ~nd an arbitrary surface l: enclosing S, and apply (10.96) to the functIOns !Pi
300
IdealFluid Aerodynamic:s
Unsteady Acyclic Motion
301
and tpk' Ttten we obtain
or
fftp;.~tpt dS fftpt atp; dS _IC(~i otpt _
s n
8
on
,.y'.
tpt
on
Oft) dS.
on
Equation 10.103 shows that in general the external force F. is not in the direction of the acceleration of the body. Equation 10.103 suggests that the coefficients mit may be thought of as a ditional apparent or virtual masses that need to be ad~ in a suitable way to the mass of the body when determining the force that must be applied on the body so as to translate it through the fluid. The coefficients mit form a set of nine numbers which may be displayed as the array mu mil
mil
We ,now let ~ go to infinity and conclude (on the basis Of the behavior of the l~tegrand as r  (0) that the integral over ~ must vanish. Hence we obtam (10.97) It follows that (10.98) The components of the impulse.! are therefore given by
'; e; •
(
mIl mlS) mu mil
mal mil
.I ! m~t
It
i, k
==
1, 2, 3
(10.99)
whe'e the m;1: are given by (I().94) or (10.95,. We now return t~ the exp~sion (10.83) for the force applied externally to the body and wnte
Since mil: is equal to ml:; there are only six independent additional apparent masses. The set of mil: is known as the additional apparent or virtual mass 'tensor. It depends on the.shape of the body and, ,for a given choice of the body fixed axes, is a constant (or any given body. The expression (mc5 il: + mil:) may. be' referred to as the apparent mass tensor. For any body it is possible to find three mutually perpendicular directions such that
F. 
:'(mu +t(tm~t)ei)
:t(mUi,+tm;~I:)
if ; ~ k if iIi:
With respect to such axes 10:103 becomes
(10.100)
The component of F. in the direction e. IS therefore given by
F,.  (m
(10.101)
+ mii) , dt
, dUi
•
il,2,3
(10.104)
F"  e;.F. ==
Introduce the symbol c5it defined by
c5it
== 0 == 1
(10.102)
When the body moves in the direction of one of such axes it would seem to have onl)r an increased ~ss. The sum m + m';i is known as the apparent mass for t;anslation in the idirection, and the corresponding m as the additional apparent mass. It is thus seen that for a translating rigid body there are at least three additional apparent masses. For bodies of revolution two of them are equal; for a sphere all the three are equal.
ii
where I andk may be 1,2, or 3: Equation (10.101) may then be rewritten 'as
10.8 Kinetic Energy and Impulse
The additional apparent masses may be computed on the basis of (10.94) or (10.95). The computation involves first the determination of the vector q» and then integrations over the body surface of the products of the components of"(p and those of D. It is possible to calculate the mil; by suitable differentiation of ttie kinetic energy of the fluid. We now show how. this may be done.
,
FII =!(mc5i l:
I:
du ' + m~I:
dt
'
(10.103)
For exa~ple the el component is given by
F"

(m
+ mu) dtl + mIl  dt + mIl _ I . dt
Ju
dUI
du
101
IdealFluid Aerodynamics
U!lSfeady
A,cyc~
Motion,
JOJ
Consider a region Ro that is enclosed by a scrface So and that contains only fluid in motion. The kinetic energy T of the fluid in Ro is given by (10.10S) The volume integral may be related .to a certain .surrace integral over So. For this purpose we use Grcen's theorem (2.139). From this theorem it follows that for a harmonic function such as ; we have
The kinetic energy ,as given by (10.107) may impulse J. Using the boundary condition
be readily
related to the
.grad ;
we obtain
0 •

U(t)
0
a c'n S
0
fff(grad ;)1 dt
R.
ff; grad;
s,
0
a dS
! U ofPtf.adS _!U (10.108) 2'· 2 s · This is similar to the general result that for a finite dynami~l system twice the kinetic energy is equal tQ ~e scalar product of the momcnt,uin and the ,velocity. ' Equation (10.108) may be expressed as
T   !Jtp;UoadS 2Jj'
T =
j.
Hence. the kinetic energy of the fluid enclosed by So is given by
! Iu,/,  .! I{Imaut)' u,
2, 2,
t
Iff p;grad;·.dS  lff p;.dt T .. 2
8, ,
.
2
0;
~
(10.106)
I!,
.. 1( . + maulI muur
2
Now consider the fluid in the region bctWccn the surface S of a moving body, and an arbitrarily drawn 'surface 1: enclosing the body. For the kinetic energy of that fluid, we obtain
T
r= 
Jj' 2
Jt e. + dS + J[ e; ~ dS ~ an lJ' 2 on 8 E
1
(10.109)' The kinetic energy of the fluid is thus a quadratic function or the components of the velocity of the body.. The coefficients in this function arc the additional apparent massc5. From (10.108) or (10.109) it readily follows that
' ' + _...... + m.....•.+ 2m~lU. ' _ ..' u. + lmu.....J
where a is directed as shown i~ Fig: 10.6•. We now let t go to infinity. For acyclic motion involvin! a finiie rigid bOdy we have
/, ... aT
and
au,
ii'
(10.110)
,; an dS "';i
oq,
as
I' 
co,
mll  aut
aT) (au, .
(10.111)
Hence we conclude that for such a situation
As an example let us compute the coefticicnts mit for a translating sphere of radius a. According to (10.32) and (10:48) we~ve
Jte;'~dS_O l! 2 on E
T   ~ffp~~dS:It 8
as rco. and
oq, _
on
U(t) coIIJ on S
'
It follows that for acyclic motion involving 'a fi"ile rigid body the kinelic energy is given by .
~ip;srad ;oiadS(10.107)
motioM.
, ~ Hence the kinetic energy is given by
Ma.8.
U(t) ,>   2 ' 8. a cos
where S is the surface of the body. It is easily verified that Ihis rell#t is InII! als%r aCyclic motion involving
T  
eif o.J.., on
2
;
':.I. dS

'lra1pU'J.w
2
0
cos' 8 sin 8 d6
(10.112)
8
an infinite cylinder. '[I U "QI valid/or cyclic . int,olving multivalued potentiills. .
thatis./or molions
_ ."a'pU· .,. 'lra·e.ull
3 3
IdealFluid Aerodynamics
It follows that
mea  0
for all i and k
where R is to be considered fixed. Time differentiation with·mpecl to the spacefixed referenc::e f~me is denoted by dldt. Note that in tM t~rm not equal to' I
mil  ftro'p
This is the
~ult
(IC. II 3)
:, III ~
r
R
pqtlT
had obtained before (see Eq. 10.(0). In arriving at chosen the direction el of tbe coordinate axes in a direction opposite \0 tluot of the velocity of the sphere. Instead, if we choose the axes such' that the velocity ofthc'sphcre is not parallel to any of the axes we obtain the result
thi!result we
had
we
It l.r lID; imp/led tllat R l.r 0 vo/umechiJnging wil'h tim~.On account of_the bouadary condition on S we have ..
(UV·.O
Heoce the expression (10.115) ba:omes
on S
10.9 M....t _ a T....aa. . . Body We shall now obtain an expression for the moment exerted by the fluid on a finite rigid body translating through it. The mom~"t tak~" with r~spect to tM origin 01 coordJntlt~s Is giw" by
:,IIIr x
'R
pfI
tit
+iJr .x Plq • >t tis.
J:
(16:116)
Tbt moment exerted by the fluid on thebody is denoted by M. the.
IDOJDeIDt being, taken about the spacefixed origin. Hence the moment
acthig on the fluid in R is
M
M
jr x
B
padS
(10.114)
jrx,.
'I
tiS
(10.117)
where, as before. S denotes the surface of the body. To express this relation in terms of the velocity potential it is more convenient to proceCd on the basis of the law of the angular momentum of a dynamical system
than to follow the steps used for obtaining the expression for the force on the body. We const1VCt tM requir~d ~xpr~ssio" lor M first with r~spect 10 0 space fixetl r~/erena IrfllM. Later we shall rel~te that expression to the calculatioDs in a body fixed reference frame. Consider the fluid that occupies at some instant the region R between the. surface S of the moving body' and an arbitrarily draWDfix~d surface 1: enclosing S (Fig. 10.6). The angular momentum of that fluid with respect to the spacefixed origin is
Since the rate ()f change of angUlar mornentuni of the fluid that is in R at some instant is equal to the moment 'on that fluid althat instant, We obtaiD,using (iO.1l6) and (l0.117), the fonowing rdation for M:
M 
:'fII
R
r
?C pfI tiT
+§r xp(q. >I tiS +iJr x P" tiS (10.118)
J: J:
Consider ~ integt81 over R. For any ~on Re containing only fluid eDcIoeed by the surface S, we have
Iffr
R
x pfldr
It therefore foll()ws that
. . jr x.". tiS
B,
(10.119)
where as before. denotes the fluid velocity and is u~for convenience instead of grad.. Then according to the considerations given in Section 6.9, the rate of change of angular momentum of fluid under consideration is given by
dt
tI
IIf.
R
. r x pq dT+* yrx p(q .••)fa dS
1
•Ie +ffr x
8
IfI
R
r x pq tlT' 
i~ xp"" tiS +
B
i
.Z
r x p.". dS
(10.120)
When the calculations are done in a spacefixed rcferente frame the pressure is given by
(10.115)
p[(U ':"" q). DJ dS
.p(r, t) 
p~  p(~~ +.~)
IcIea1Fluid Aerodyoamica
It therefore follows that
Vaste.dy Acyclic' Motion
sad
.lt . x ,;. dS _ 11'r E
M  :t§r
·B

~ ICr x Pfa dS _Jtr x dtll' ,11' E E
(q.
pi rrdS (10.121) 2
". jr
B
X·
pfadS
jr x.. dS
•
From Eqi. (10.118, 11.0, 121) we obtain'.
. 1aeIe,. is tile lIDpulsive pressure (SC!C Section 9.9).
I. the momerrt imptdM.
xp4 dS +jr xp[f. 
We With this notation we write
caD"
the imp• • '
ja)q] dS
(10.122)
F.  dt (mU + ,FJ
d M.  ~(L dt F _
d
We now let 1: go to infinity. For acyclic motion involving a finite rigid body the integral over 1: v.nishes as 1/" as ,  ' 00. For acyclic motion involving an infinite'rigid cylinder the'integral vanishes as 1/,· as r  00. It therefore follows that the moment on the body is given by
+".>
tit
d
d',!
M .. d·JIr x pt/IDdS dtll'
B
(10.123)
on the body is given by (10.124)
Md~J·
Thus the force on the body is given by the negative of the rate of change 01 the impulse and the moment by the negative of the rate of change oft. momentimpulse. . We shall now relate dJldt and dJ .Idt which are ~me derivatives in a spacefixed reference frame to the .co~?Onding time derivatives co... sttucted in a bodyfixed reference frame .. Let us denote by XI the bodyfixed frame and by"I and J. the impulse and the momentimpulse '.lith respect to tlie XI frame; .". I OOng the moment with respect to 0 1 the origin of coo~nates in XI~ . We denote the spacefixed frame by K and the origin in XbyO. The frame XI translates with a.velocity U(t) with respect
~L
In a_sinWar way we can show that the ftWQ
lUCCC
F 
dt
dj pfadS
B
Let F. and M.. denote 'the exterIraJ force and monf.1nt applied on the body to translate it through the ftuid. We then have
d .  (mU)  F.
dt
+F
d . (L)  M. + M .dt Where mU is the linear momentum of the body, and L is its angular momentum. It follows that
'
At the instant .under consideration let 0 1 be at the position ~ with respect to 0 (Fig. 10.8). ~ impulse and 'mo~entimpulse computed in frame K
F. 
dt
~(mU) 
F  d (mu dt
ffp4B
dS)
M, _ d L  M _ ~(L dt d"
We introduce the notation
_ICr Jtr B
B
x.p4dS)
, ,  ifp4 dS
ff1lrDdS
o
FIt. 10.1 Illustrating the relation between, the spacefilled and bod)'fixed descriptions of impulse and momentimpulse.
and
are then related to those computed in frame ~l by
We have
We now express ~_ in a form similar to the expression .
(10.~)
for ~.
~~  ,,1, +'~ X ~I
FUrthermore. dcooting the tirne derivatives in K and Kl respecti~ly by
Kd Kid
~
"1
~_  jr xHdS
ir x
B
B
8
p(U. 'P)adS
d.
we observe
and . 
d.
t( j~'Paf x ads)u•.
,,:hcrc. use
~s been ~of
IS
Eq. (10.90). The component of ~III in the
.
(10.129)
direction e~
then given by
Kd I _K'd I __ _ d. ,, d."
+~x
_I
Kd ' Kd 11+~X.1 d." d.
11;4  jP'Pt(r x .)idS
B
We choose 'M two frame" to be coincident, at tM instant, untIer consideration. We sball then have, noting that IdfJd. is equal to U,
Kd _
(10.130)
d'" 
I
_ 
Kid
d." 
I
+, U X
,
I"
It therefore follows that
The coefficients I ••• just like the 'mass coefficient mit. arc constants that only on ,the body shape and the choice of the orientation of the coordInate axes ID the bodyfixed frame. It may be verified that
depe~
F _ KdJtp~d.S _ KldJtptfds'
. d.lf' B
dt lr
(10.l2S)
I",,, I,t
We ~e that to deferm~ne the force and moment on a rigid body tramlatlllg through the flUid we need to compute eighteen coefficients. namely m. t and lit. that arc cbaracteri~tic of the given body. Since Init is equal to m» these are actually fifteen tndependent coefficients.
B·
and
.' M _ Kdif'r x p~ dS  ~ d. B dt
Kldif'
B
r x pt! dS
+ 1](1) x
. if .
B
p~ dS
(10.126)
10.10 Vllif'orm l .......·doa
When a rigid body moves through the fluid with a constant velocity the force on the body is zero but the moment exerted by fluid on the body is not zero. We have
Hencefor•h we shall we only 1M bOdyJixed frame. Consequently 'we shall drop tM supencript KI and also tM subscript 1 on the impulse and momentu1rpuise computed in the bodyJixed frame. With this understanding we
wri~'
.
F _  d,,/  d 
FO
M u)(,,/
dt
ii.
if
R
pt/>n dS
(10.127)
... u xffp4*nclS
and
M_Ux"
d"/,,,
dt
.
,,0
($
s
(10.131)
_.!fJr xp~dS'+
dt s'
U
)(ffp~dS.
8
(10.128)
It follo~s that the body ~i11 tend to rotate; to maintain it in uniform translation through the flUid. an external. moment must be applied to,the body constantly.
JII
ldealFlujd Aerodynamics
Unsteady Acyclic Motion
JII
I
constant velocity through the Guid. Let the velocity vector ma~e an an~e at with the ex!! of the body. We choose the bodyfixed coordmates with the origin at the of the nose, the :I:,axis along the body axis.' and the :l: «aplane as the plane Containing the body axis and the velocity vector 1 (Fig. 10.9).
As an example consider the case of a body of revolution moving with a
tip
10.11 Permanent TransJ8doo From Eq. (10.131) we conclude that if the vectors U and $ are parallel
there is no moment on the body in steady translation. Hence in such a situation the body will remain permanently in steady translation. The· axes along which permanent translation is possible are known 1IS axes of permanent translation. For a body of revolution the axis of revolution and any transverse axis are such axes.
10.12 Remarks
In this chapter we were concerned mainly with the force and moment exerted by the ftuid on a rigid body translating through it. The fluid motion is strictly acyclic. The results obtained are applicable to both a finite body and an infinite cylinder as long as the motion generated is without cifC1Jlalion. Modification of the results is necessary when motions with· nonzero circulation are to be considered. The extension of the present res1,lits to the case of acyclic fluid motion generated by a·translating and rotating rigid body may be carried out by following a procedure similar to that used for the case of the tran$lating body. It is found that for a translating and rotating body there are in general thirtysix additional apparent inertia coefficients such as the mil: and lit. Of these there are actually twentyone independent coefficients. For the treatment of the subject offorces and moments _cting on a body moving through an ideal fluid when the ftuid motion is cyclic and when the bod)' is both transiating and rotating, and for examples of the application of the subject referenCe may be made to Lamb (1932) or to Kochin, Kibei, alid Roze (1964).
11
JIll. JU A body or revolution ·translating Uniformly thrOUgh the .fluid.
Webave
It can be shown that
mil: 
0 for ; " k '
Using'this result and noting that
U  (III' 0, II,)
We obtain
Hence it follows that
, M _ eJ.,maa  mu)lI l",  e.tit(maa  mlJ cos, at sin at
, It is seen that if maa is larser than mll' which usually is t~e case, the, ftuidmoment on the body tends to increase the angle at. In thiS sense t~e translatory motion under consideration is unstable. We note that there IS DG mom"n t on the body.if at is either'zero or ~/~·
Steady AC)dic Motion
and the problem is to determine fJ) as the solution of the equation
JIJ
Chapter 11
VIfJ) 
0 '
(11.~)
Steady, Acyclic Motio",
in the region extellior to the body sUch that the solution satisfies the boundary conditions '
V • gradF grad fJ) • grad F'" 0 on F(r) _ 0
and
(11.4)
gad fJ)
The aim of our investigations is to determine the velocity potential for the steady jJow of an ideal fluid past certain solid bodies of aerodynamic interest. For tins purpose we are to obtain the solution of Laplace's equation UDder c::ertain p~bed boundary conditions. Because Laplace's equation forms the basiJ. for a great deal of mathematical physics, its .solutions have bou investigaW4 CX&cDsively, and, consequently, there exists a ,large body of .nathematical theory about it. There are variou,s methods of constructing its solutions. ' The task of constructing an exact solution directly by satisfying the given boundary conditions. generally proves diflicu1t. In view of this, it is fruitful to approach the solution to our problem from sOme simple k\own solutions of LaPla:"'s equation ~nd t,o , dilCuss their significance in terms ~f fluid flow. SlDce the equation IS linear we can find new solutions by superposing various known solutions. Simple flows may be used for building up the more complicated flow fields that must satisfy prescribed boundary conditions. In this chapter we shan see how the solution for the steady acyclic flow past a fixed body may be built up by superposing certain singular solutions of Lapla~'s equation. Having thus obtained the solutions for a fcw' typi~ bodies we ~hall examme the theoretical results in light of observations. We start With a statement of the ntathematical problem.

V  U at infipity ,
(1I.S)
If, as before, we denote the disturbance or perturbation' velocity and potential by • and ; respectively we have
•  grad;
The problem in terms~of the perturbation potential is
..
VI;_ 0
as r _
00
grad;' 8J'8d F  U· grad F on ,F(r) _ 0
, components of grad ;  0 This is precisely the problem for.a riSid body moving with a velocity _ U through an otherwise undisturbed fluid (seesection 10.3). Once the velocity potential fJ) is detennin~the pressure field is obtained from the steadYlitate Bernoulli equation
p(r)  H 
e. (grad fJ)1 2 '
(1,.6)
!1.1 StateaNat of the PMbleaa
Consider the steady flow past a fixed rigid body. We choose a body
fixed coordinate system'and describe the body surface by
F(r)
== 0
(11.1)
w~ H is a constant that may be determined' once the pressure and velOCIty at one reference point is given. The solutions of (11.3) may be found by the method 0/ ~ParQtio" 0/ varkzhles or by the singularity method. An example of the method of separation ,ofvariables is given in Section 10.4. Many other examples may be ~~und 10. Lamb (I932~. The Singularity method consists of super ' poSitIon of s10gular solutions. We shall use this method. Witli this in mind .we shaU investigate various simple functions that satisfy Laplace's equatIon to see what sort of boundary conditions they fulfill and how such fuactions could be effectively combined to build solutions that interest us.
11.1 SImple Pol)'llOlllial Sol~doas
Let the velocity and pressure of the undisturbed stream at infinity ~ denoted respectively by U and PtD. Let the velocity and the velOCity potential at any point of Lhe flow field be denoted by V and~. We then have (11.2) V  grad~= V~
W~ choose <:a.rtesian coordinates x, y, z and consider some ~imple functJ.ons of posItion. We shall first specJalize them to satisfy the potential equation and then investigate the flow fields represented by them.
Jll
J14
1. Let US then consider tbe expression
~  A%+ By
IdealFluid Aerodynamics
Steady Acyclic Motion
Jl$
+ Cz
(11.7)
where A, B, C arc constants.. This expresSion satisfics the potential equation, so it. may be regarded as a velocity potential. The velocity components arc given by
u ~
Equation (11.12), or equivalently the potential (11.11), represents a velocity field in which there is no velocity component in the zdirec:tion and the flow appears the same in all planes perpendicular to the Zaxis; the motion is therefore twodimensional. The streamlines of the flow arc given by the difterential equation
d% a dy a
Il II
a%
 A
a constant
~.
,,   B a constant
y
ay
.~ w·in
C
a constant
Hence tJle velocity at every point has the same magnitude and direction. Thus the function (11.7) represents aftow that is ~orm in spacc" Cottverselywc can state that the PQtentjal (o~ a ftuid with a uniformVelocity V  (U, Y, W) is
~. ~%,lI,Z) 
3 2 I
~I
X
U%
+ Yll + Wz
(11.8)
2
The streamlines arc aU straipt lincs parallel to V. 1. CQnsider DOW the funCtion
~  Azi
+ JJyI + CzI
(11.9)
Fil. 11.1 Streamlines defined by zy ... constant.
where, as before, A .. B, C ~ constants. Since (11.9) should satisfy the potential .equation. we require dial . .
A +B + C  0
(lI.lO)
which, according to the relations (11.12), hecomes
This can be satisfied iomanyways. Suppose we d~ ~t the potential be.independent.ofthe zco.ordina~..
Then w,'set
d% dy :::::% y
Integrating this equation we obtain
(11.13)
CO
and obtain
A
+B 0
~
or
B  A
(11.11)
xy = const.
(11.14)
In such a case the potential (11.9) becomes
 A(zI  ,.)
Ibc colTC$ponding velocity con:!ponents are given by
Il
= 2A%
(11.12)
"  2Ay
wo
as the equation for the streamlines, each streamline corresponding to a different value df the constant. The streamlines form a family of rectangular hyperbolas. as shown in Fig. ILL We notice that at rhe po!nt z = 0, y = 0, the velocity is zero. Hence the origin is a slagnarivn."oinr. ami the streamline passing through the: stagnation point is usually referred to as the slagnaiion streamline. In the present cas~ it is given by the equation :ry = 0
316
ldealFhaid AerQdynamics
Steady Acyclic Motio9' 
Jl7
It thus follows that the X and Yaxes form the stagnation streamline. , It may be noticed that at the stagnation point the directions. of the four branches of the stagna~on streamline are different from one another. 'Tn this sense the stagnation poi~t exhibits a singular behavior and may ~ thought ,of as a singular point in the flow; it is a saddle. .
presence of the bOdy in t~e originally uniform stream. Since we. have already determined the function that represents a uniform stream, we need to investigate further only functions that iue likely to describe the disturbance field. From what we have said, these functions sholJld be such that the disturbance velocity dies out with distance from the body. This ~eans that such functiol}S should be composed of terms involving I~verse powers of the space coordinates. We shall now investipte a Simple example of such a function. '.__' 11.3 The Source Potential Let us c(msider the function A f(r)  
,
(11.15)
JIll. lU TwcMIimeDsionat low apiDlt •
wau.
The flaw .field described by the
repRlCDt
potenti~l (q.l1) may be thought to
where A is a CODstant and r is the magnitude of the position vector r from a fixed reference point to a point in space. To use this function as a veloci~y poten~ial (of some flow) we first check to see if it satisfieS the potential equation; It tl!rns out that·
ccrtaiit local regions that occur as parts of more general flows.
One example is the ,flOW in the vicinity of the stagnation point for a twodimeiasional flow against a wall, as shown in Fig. 11.2. Another example is the f1aY!'.in the neighborhoodof a rectangul~r corner, as shown in Fig., t 1.3. It ,should be noted that the  potential (lUI) and the velocity (11.12) exhibit singJllar behavior at infinity. In fact, the velocity at infinity is infinite. The functions (11.7) and (11.9) are simple examples of polynomials in the coordinates x, Y. z. We can proceed to consider polynomials of more general form, specialize them to satisfy the potential equation, and then investigate the. particular t\ow fields repreFIao 11.3 Flow inside a Re sented by them. Finally, by a proper selectanplar corner. tion of such functions we may attempt to build a solution or Laplace's' equation that satisfies prescribed boundary conditions. Instead of pursuing this tedious path, we shall now pass on tt' consider some specific functions that will quicily lead us to an analysis ofthe steady flow past a body. In case of stea':1y flow past a fixed bo~y, we know that, atinfinity the velocity of the fluid reduces to that of a uniform stream. This suggests that the flow field may be considered to be built up of two partsone that represents tbe uniform stream and the other the disturbance dee to the
vt(~) == div (grad~)
_
~
That is, the function (11.1 5) satisfies the potential C«uation at all points of space except at the point r  0, where ii has the indeterminate form 0/0. To, ~etermin~ the value of VI(A/r) at the point r  0 we use the integral defimtlon of dlvet"Eence and write
VI(.1)  div (grad~) = lim dT Jf , • n dS .!. Ir(grad~) . r ,
"'''0
AS '
We choose for the arbitrary volume dement dT a sphere' of radtus center at r  O. If er is a unit vector in the direction of r, we have
g rad ""   I er ·
E
with
A
,
A
r
Consequently, we obtain
[VIA] ..
.~
al r~.
== A,O AT hm 
,1 if oyer
opMre
A  er • DdS'
,1
(1l.t6)
Since Ot~' 1M spMN'
• The result may easily be verified. in Ca.rtcsians.'
31.
and
Ideal·Fluid Aerodynamics
Steady Acyclic Motion
J19
=1
.,. 8
A
A
'
a ccnstant
(11.16) reduccsto
[VI~J atrO==00 r
(11.17)
Thus the Laplacian of AIr is zero everywhere except at the 'point r .. 0, where it is infinite. It foilows then that the function AIr may be regarded ''as the velocity potential of a certain flow field as loilg as the point r = 0 is excluded from. our considerations. The fact that r =0 is a peculiar or singular point" should not disco.urage us from further (:onsideration of AIr as a velocity potential. In' (act, as We shall presently learn, this function will play a great part in building up the solution to the problem of flow over a body. 'In regard to the singular point, we try to 'arrange our considerations in such a way that it is outside the region of physical interest or, if this is not possible, we simply learn to rC90gnize it as a peculiar point and expect such ~ point to indicate physicalfy untenable results. During the course of our ,investigations we will meet other functions that arc; singular in some respect or other but we will soon learn to use them with advantage Now, let us look at the flow field represented by the potential
In either case ilie magnitude of the velocity is inversely proportional to rl and as such increases as we approach the point r == 0, where it actually becomes infinite. Conversely, the magnitude of the velocity "ecreases with distance from that point and completely vanishes at infinity. We noti~e again that the point r == 0 is a singular point to be excluded from physical considerations. From the velocity field (11.21) we see that fluid is being continuously created at the point r = O. Such a point is called a source, the corre.sponding potential (11.20) a source potential and the flow field represented by it a source flow. In the case of the velocity field (11.19), fluid is being continuou!.ly annihilated at the point r = O. Suc:} a point is called a sink, the correspon~ing potential (11. 18) a sink potential, and the flow field represented by it a sink flow. Point sources and sinks do not possess any physical reality as such but are important mathematical concepts useful in the 3nalysis of fluid motion. We now determine the quantity of fluid that is continuously appearing at a source or di~*ppearing at a sink. The net outflow of volume· of fluid through any dosed surface S drawn in the flow is given by the surface integral
fJ
8
V • D dS
=
if
S
grad C!> • D dS
C!>(r)
=~
r
o!, equivalently, by the volume in!egral (11.18)
The velocity at any point in space, is given by
IIf
R
div V dT
=
IIf
R
VIC!> dT
V(r) =
grad~ == A!!' r,.a
A!.
, rl
(11.19)
We observe that the direction of the velocity is radial at all points of the field and that its magnitude is Constant over the surfaCe of any cho~n sphere. The streamlines of the flow are straight rays through the point r = O. If the constant A is positive, the fluid flow is directed toward. the point r = O. If instead of A we choose A, that is, if we consider the potential A (11.20) <I>.(r) =  r
where R is the regiog 'enclosed by the surface S . . First consider a closed surface SI enclosing a region R} that does not contain the source. Then the net outflow of fluid through S} is
8.
Rl
since V2C!> = 0 at all points except at the source. For the same reason, if So is a surface endosing the region Ro that contains thp. sourcf', we have
the fluid flow would be directed outward from the point r = 0, since the correspollding velocity is given by
Before we compute the net outflow of fluid through So we observe that if
• We talk about volume ir.~;ead of mass because the density is constant for our ftuid.
V(r) = A!! = A .!.. r' r3
(l1.21)
.J20
IdealAuid Aerodynamics
Steady Acyclic: Motion
311
S. is any other surface enclosing So, and if R. is the region between So and
SI'
fJ
8,
V • D dS 
fJ
8.
V • D dS ....
Iff
R,
V'4» dT .... Q
Consider now the case of a source q (i.e., of strength q) Dot situated at the origin of the chosen coordinate system. In Fig; 11.4 let point 0 represent the Qrigin of the coordinate system, the. point S the location of the sOU(Ce, and P a field pomt. Let
.
~
or
if
8,
V • a dS ....
fJ
8.
• denote the vector OS
.
~
V • D dS
r denote .the positin vecto~ OP
pc",y,.)
where D is an outward normal with respect to the region RI • It follows. therefore, that the outflow of fluid through every surface enclosing the source is the same and is equal to the volume offluid, say q, that is being created per unit time at the source. We thus set
".. fJ
.8.
V· D dS
(11.22)
To evaluate this integral we choose for the arbitrary surface So a sphere of radius ro with its center at the source; substitute for V from (11.21) and set D = er . Then (11.22) becomes q =
if
Ipbere
V • D dS..
"
fi ~
r.
er • er dS = 471' ~
(11.23)
.... UA SoURle not at the origin.
.pbere
and
r~
The quantity q is known as the strength of the source or silDply tb.e source strength. In terms of the source strength, the source potential ""Boy be expressed as (11.14) Similarly, in the case of a sink, the potential may be expl'eSliBll
denote the vector SP
(11.26)
(11.21)
Then the potential at P due to the source situated at S is
q1. ql q 1 cD(r)         =  .... . 471' rl 4.".lr11 471' Ir  sl
as
The fluid velocity at the point P is then given by (see 11.21) VCr) .. .!L .!!. .. .!L (r  s) 471' r l l 4'11' Ir  sll
Let us choose Cartesian coordinates and write r = (%,11, z) and
«P(r) ==
4"71' r
.!L!
where q now represents the strength of the sink (or simply the sink strength) defined asthe 'Volume of fluid that disappears per unit time at the sinL It should be noted that in the expressions (11.24) and (11.25) Tis the magnitude ot r, the position vector drawn from the source, not from the origin of a chosen coordinate system. to a point in. space (socalled field point). However, if the origin of the coordinate system is chosen to be at the source, then r is Identical :with the usual position vector. In such a case we speak of the expression (11.24) (or of 1l.25) as the potential due a source (or 11 sink) situated at the origin.
s
= a, '1,0
Then the Cartesian form of the source potential (11.26) is
<I>(x, y, z) =  4 7l' [(x _ E)2
Q
to
+ (y
_11])2
+ (z _ ,)2)~
(11.28)
If u, v, w denote the Cartesian components of. the corresponding fluid
311
velocity, then!rom Ea
II(Z, 1/, z)  _I) we obtain
IdealFluid Aerodynamics
Steady Acyclic Motion where
313
q zE ~=="" 47r [(z  E)I + (1/  "1)1 + (z  C)I]~
(11.31)
v(z, 1/, z) _
w(z 1/ z)
.!L
47r [(z  E)I
+
1/  7l (1/  "1)1
+(z 
,)I1.~ ,)I]~
(11.29)
For a g;ven value of constant over a circle
%,
the potential and consequently the ,flow field is
1/1
+ Zl =
constant
_.!L
47r [(z  E)I
"
+ (1/ 
z C
'1)1
+ (z 
We shall now look into the usefu~ness of the ..:oncept of source and sink.
11.4 Source in • Uniform Flow (AxIsymmetric Flow oyer • Semiinftgite Body of Revolution) .
A number of physically interesting Jlow fields can be obtained by combining th~' ftow fieldll of suitable distributions of sources and sinks with that of a uniform stream. A simple examlile of such a combination is that of a single source with a uniform ftow. Let the source strength be q and the velocity of the uniform stream be V. We choose Cartesian coordinates such that their origin is at the source and the Xaxis points in the direction of the uniform stream U (Fig. 11.5). We thus set
This means the flow field is symmetrical about the Xaxis. It is an axisymmetric flow. In axisymmetric flows it is convenient to use cylindrical coordinates and occasionally spherical coordinates. Sometimes in the analysis of a single problem it is profitable to go from one set of coordinates to another. In view of this possibility we denote, as shown in Fig. 11.5, by r, rp, x cylindricalcoordinates and by R, 8, rp spherical coordinates. Their relation to Cartesians is then given by
Cylindrical
1/ .... r cos rp
(11.32)
z == r sin rp
Spherical
xRcos8
1/ .... R sin fJcos rp
ViU
For the potential
~
of the combined flow field we have
(11.33)
$(x, 1/, z)  the potential due to the uniform stream + the potential due to the source
 Ux . [Xl
z  R sin 8 sin rp
It should be noted that in this notation R denotes the magnitude of the wual position vector, while r _ [yI + Zl]~  R sin 8 (11.34)
We shall now look at our problem of a source in a uniform stream in terms of spherical coordinates. The potential (11.30) of the combined flow then takes the form
+
A 1/1 + Zl]~
(11.30)
y
;::::::
~
vau
$CR, 8, rp) = UR cos (} If "R; "" we hav~
A R
(11.35)
". denote the components of the. velocity V at any point R, 6, rp
0$
"R 
z
Circle:
x
,,2 + z2 '" cons\.
(JR

== U cos (}

+ . R
(11.36)
.Il
II, 
.

1 (Jcl)
'U' (}
. SIO
R (J(}
II
Fig. 11.5 Coord,"a~e system for the problem of source in a uniform flow.
.,
l(J<1) _0 R sin (J orp
324
IdealFluid Aerodynamics
Steady Acyclic Motion Hence it follows that the streamlines are given by
iURI sin' 0  A tos 0
Equati.ons (1' 1.35) and (11.36) show what we have already noted, that the po.tentJal a~d the flow are independent of the coordinate 'P, that is, ;!re aXisymmetric. Thus the flow field looks the same in allplancs defined by 'P:& const. Therefore to analyze the ~ow field further we need only look at cne of these planes. Such a plane 15 shown in Fig. 11.6.
= constant = C
(11.39)
By assigning different values for the coqstant, we obtain the various streamlines. We now look at the stagnation streamline, that is, the streamline passing through the point R., 0,. Substituting in (11.39) for Rand 0 from (1137), we obtain C==A as the value of the constant for the stagnation streamline. Thus the
equation for the stagnation streamline is
K~~~~~~.iX
lUR' sin' 0  A(I
+ cos 0) =
0
(11.40)
or
R .... [2A 1 + COsOJ~ U sin' 0
Fla. lUi Flow resulting from a lOurce in uniform flow.
.... ,JU cosec~ fA 2
We 1l0te that R is an even function of 0 (as it should be, for the flow is axisymmetric), ~nd thus for 0 JI. 11' thefe are two branches of the streamline. For 0 == 11' all values of R are possible, that is, the whole negative Xaxis is part of the stagnation streamline. This streamline appears r.s shown in Fig. 11.6. In terms of the cylindrical coordinate r, the stagnation streamline .is described by ,
From the. vel~ity field (11.36) we notice that both the vel~ity components vamsh simultaneously at the point and
.0 .. 0..... 11'
R .... R.
A)~ (U
(11.37)
The p.oint (R., OJ is therefore a stagnation point in the flow and is denoted by S m the figure. The streamlines in any axial plane are described by
tp(R, 0) = constant
, ... R sin 0.. [ 2A (1
U + cos ~).JH
(11.41)
If T denotes the point where this streamline intersects the raxis, then
''1'
where" is the streamfunction. Along a streamline we have
== OT ==
J'ii
(11.42)
d",
==,aR dR + a8 dO a" a"
= sin Ou, dR
+ R'sin 0 uR dO
As 0 goes to zero, the stagnation streamline reaches an asymptote for which, say, r = r,. Then ',==' as 0+0
(11.38)
See Eq. (5.49) or (9.106). With 01.36) and (11.38) we obtain
Aj~ =2
(U
(11.43)
d", == d (
UR' sin' 0 ) 2  A cos 0
The stagnation streamline may now be drawn in as shown in Fig. 11.6 by the line KSTL  KST1L1  OS. Other streamlines, computed according to (11.39), appear as shown in the figure.
J26
IdealFluid Aerodynamics
Steady Acyclic Motion
J'7
The flow field in all other axial planes (i.e., containing the Xaxis) appears exactly the same as shown in the figure. We observe that the surfa~ form~ by the stagnation streamlines (Fig. 11.7) obtained by revolving the hne KSTL about the Xaxis, is a surface that divides the whole flow' field jnto two distinct regions..:.one external and one internal to that surface.·. Since the surface is a streairn surface, no fluid flows across it. The internal region, therefore, consists entirely of the fluid emanating from the .source. The source flow is pressed together (by the uniform
passing per unit time to the right through any cross section is equal to the strength q of the source. If the cross section is chosen in the cylindrical portiop of the body, the fluid volume passing thr~ugh it ~r unit time is simply 7r(d'/4)U, since the flow velocity at a considerable distance from the source is equal to U, the velocity of the uniform stream. Therefore
q !! tJlu
.. 4 or
A~.
.!L  dU
411' 16
(11.46)
"""'rd
I I
Fie.H.7 The semiinfinit~ body of revolution geQ~rated by source ;0 a uniform ftow.
stream) '~n~ made t~ flow to the right. The external region consists only of the OrIginally umform stream now displaced. Thus the stream surface formed by the stagnatiop streamlines behaves just like that of a solid body placed in a uniform stream and may be regarded as such. The body appears as a semiinfinite body of revolution with a blunt nose, that is, a round nose. From the preceding considerations we may now conclude that the flow field due to the motion of an originally uniform stream past a semiinfinite body of revolution whose axis is parallel to the uniform stream can be represented by a simple superposition of a single source and the uniform stream. The surface of the body cannot be prescribed arbitrarily but is to be described by an equation of the form (see 11.41)
r
With the source strength known, the potential of the flow field is determined according to (11.35). Finally, it.m.y be noted that using the principle of superposition we . have constructed a soluJ9n of the mathemati~1 problem posed by (11.3) to (11.5), that is, a solution.of Laplace's equation that satisfies prescribed bOundary conditions at infinity and on the bOOy.
11.5 Source.... Sink ID a Uniform Flow (AxIsymmetric Flow oyer a Closed Body of Rnoludon)
The superpo$ition of a single source and a, umform stream'lead us to axisymmetric flow past a semiinfinite ~Y of revoluti~n. To o~tain, such a flow past a closed finite body of revolution, we may situate a smk on the axis of the semiinfinite body at some suitable distance downstream frolJl the source and make the strength of the sink equal to that of the source. In this way the flow produced by the source is completely sucked in by the sink; the flow closes up behind the sink just as it opened out in front of the source; and we obtain the flow over an elongated body ... Ith rounded noSe and tail. Let us theD consider the superposition of a uniform :..:ream U, a source of strength q and a sink of strength q, the line joining the source and the sink being parallel to the uniform stream (Fig. 11.8). The source faces the uniform stream. We choose the origin 0 of ~he coordinate system midway, between the source and sink and the Xaxis parallel to U. The various coordinates are de~ignated as shown in the figure. The velocity potential at the field point P is thea giveri by
~(P)
== const. JI+"~os ()
(l1.44)
At a large distance from the nose, the body is nearly cylindrical. If d denotes the diameter of this part of the body, then (11.44) becomes
d / . r'J_vI +cos()
2 2
(l1.4S)
. The strength of,the source that represents the disturbance due to the body may be determmed from (11.43) (since d = 2r,), or dire<;tly as follows. Consider the flow within the surface of the body. The volume of fluid
• For this reason the stagnation streamline it sometimes referred to as the dividing streamline. .
q 1 . q 1 Ux t'411' Rt 411' Rl
1
J(x  a)2 + r2
== Ux + A[

1
]
(11.47)
J(x + a)1 + r'
31.
where
IdealFluid Aerodynamics
Steady Acyclic: Motion
We observe that
U A
and
a
=0
4%,a _ 0 (%,1 _ a'}l 4%,a
for
z, > a
one half the distan~ between the source and sink
and
The flow field, as before, is axisymmetric, and, we look at the ftow in any planC' tp == constant.
U
+ A.
1M" 0 (~  a J
for
r
%,
<. a
~
11
X~L~~LX~~
______~~________~x
II
11
Source
Sink
FIa.".1
Coordinates (or the IOUI'CIHinIt combination in a UDifonn Itream.
Fla.".'
Axial
Row put a Rankine body.
Ifur , u,,' U. arc the velocity components with respect. to the cylindrical coordinates " tp, %, we have
ur

acJ)
0' .. ..... A{ ' [(x  a)'
+ rIJ'" 
[(z
' + a)1 +} rIJ~
•Hence there are two stagnatiop points on the Xloaxis, one for which z, is less than a and the other for wh1ch :r, is greater than a. They are shown as A and B in Fig. 11.9. In any axial plane along a streamline we have
u
"
0 r Otp
oz'
1()cI)
0'1' d'l' _  dr
(11.48)
or
0'1' +  dz ... ru., dr 
oz
ru r dz
(11.50)
u. == ()cI) ...
U_ A{
z  a [(z  a)1
+ ,..]~
_
+a } I(z + a)1 + r']~
%
[~ (5.50) or (9.107)]. With (11.48) and ('Il.50) we obtain after some simplification '
If z, and" denote the coordinates of a stagnation point in the flow, they arc determined by settinJ the righthand members of (11.48) equal to zero. We thus obtain (l1.49a) '," 0
and find that x, is, to be detennined from the equation
U
d = d(A
'1', ;
:r  a _ A z+a J(z _ a)' + rl J,C:z + .a)' + "
+
ur')
2
Hence it foBows that the streamlines arc giw:n by
A(
z  a _ :r + a , ) J(z 0)1 + rl J(x + a)' + ,2
+ ! U,.s .. C,
2
a constant
(11.51)

A [ x,  a
lx, _
+a ] all ~ 1%, + all
z,
== u.
0
(r .. 0, z,)
(11.49b)
Different streamlines corr.;spond to different values of C. We now look at the stagnation streamline. The value of the constant for this streamline is obtained by putting' = " = 0 and x = x,. It is
,
'
JJO
simply equal to zero. Thus the stagnation streamline isdcscribcd by the equatiof'l
. Steady Acydic: Motion
JJI
A[J(%  a)1a + ,.. ;===:::;:::~
% 
J(%
% + a 1 . ] +  U ,.. .. 0 + a)' + ;.a 2
We note that these Rankine bodies form family of bodies for which only the ratio r,la can be prescribed for a given Y./V.
a
(11 52) .
U.6 LIDe DlsbibatJoa of Soarees . . SIDb In • Uniform Flow: AxIsymmetrIc Flow Ofti' SleDder BodJes of Reyolutloa
As an extension of the above s.imple example of axisymmetric flow over a certain closed body ofrevolution, we may next consider the superposition of a uniform stream and a suitable distribution of sources and sinks."long
To trace the streamline it is convenient to express (11.52) in tenns.of Rand 8 as follows: A(cos 81

cos 8,)
+ !URI sin' 8 2
(l
(11.53)
where 81 and 8, are as designated in Fig. 11.8. From (11.53) it readily follows that the whole Xaxis except tbe part betwccn source and sink forms a part of the stagnation streamline.  The rest of this streamline is a closed curve shown by the line ACBDA in the figure. We observe that, as befo,re, this stagnatiQn streamline is a dividing lin~ between the originally uniform stream and the fluid flowir~ from the soutee to the sink. Thus a surface formed by the stagnation st~mlines of the whole flow behaves just like a body of revolution obtained by revolving the line ACBDA about the Xaxis. C~nsequently, we may conclude that the particular superposition we have consIdered of a source, sink, and uniform stream rcprcscnts axisymmettic flow of an originally uniform stream past a. closed body of revolution whose surface is d:;cribed by an equation such as (1l.52) gr (11.53). The body is a symmetric ovoid whose maximum transverse raCiiuS r is given ' by the relation
11 •
FIa.I1.10 Superpoaition or a unitOnn ,tream and a source distribution along a line parallcl to the stream.
an axis parallel with the uniform stream. In this way, by selecting different
distributions, it is possible to arrive at the axisymmetric flow over bodies of revolution of various shapes. Since we are seeking closed bodies, the sum
r,Va + r.1 l
4a
,~
(11.54)
of the strengths ofal/ the sClurces and sinks, no matter what their distribution, should be zero. That is, for the body to be closed there must be enough sink
strength to suck up all the fluid produced by the sources present. The sources and sinks arc generally distributed continuously so as to obtain smooth bodies although single sources (and sinks) arc also allowable. The method of analysis of the flow field in this case generally follows the lines of the preced~ng two sections. We shit;! now consider briefly the superposition of a continuous line distribution of sources and a unifonn stream, the line containing the sources being parallel to the uniform stream. We choose, as before, cylindrical coordinates r; tp,. % with the Xaxis lying along the line of squrccs and pointing in the direction ofU, the uniform stream (Fig. 11.10). Let
ff(~)
whi~h follows readily from (l1.S2) by se~ng s  Ojn iL Bodies of ~s type are called RIl1rklM botJk.J after Rankine who fint·
sU80"'CSted the idea of forming the type of Bow under consideration (Fig. 1,1.9). Their nose and tail are blUnt. The maximum.speed on the body occun at z  0 and r  re. Denoting by Y.. the maximum speed we obtain from (11.48) flr('e,
V• .. u..('c,
% % 
0) 0
0)  U + 2)( a , IL . '(QI + , Iy~ Eliminating A. by using (11.54) we have . •
...... 1 ( ' '" 1 ~ 1+:..£.
(11:S5)
U
]. a
I I
+ (rcta)'
(11.56)
• For points between the source and sink. 8. = " and 8.  0 and hence \11.52) is not. satisfied; for z < Q and z > 'Q.' 8.  6. J" or zero and hence (11.52) is satisfied.
denote ~he intensity of the source distribution per unit length. Positive valuf!S of f(%) denote sources, negative values sinks. 'Let the source diStribution extend from %  0 to % = I. The condition that the total strength
In this c:oiltext the word "sources" means sources and sidks.
JJZ
of the distribution is zero is expressed by
IdealFluid Aerodynamics
Stc8dy Acyclic Motion
3JJ
1.'
fez) d%  0
At % .. E, consider an element. . . . dE of the distribution. Then, ac:c:ording to Eq. (l1.~), t.he potential at a field pointp(%, ",,) due to the elementary sdurce I(E) dE is
I f(~ dE 1 feE) dE  4" ~ ..  4tr JI=(%===E==)I=+:::::;,I
The potential at P due to all such elementary sources distributed along the Xaxis from 0 to I is then given by Cl»l(%' " 91)  
! (' .
4")0 J(%  E)I + ,1
feE)
dE
(11.57)
During this integration %, r, " ate kept, constant, for they are the c0ordinates of the field point for which the potential is being computed. Now, to obtain the total polential CI» of the flow field resulting from a superposition of the source distribution and.the uniform stream U we add to the potential Cl»l that due to the· uniform stream. The latter is simply U%. Thus the total P.Otential is given by CI»(%, " 9')
ftow. The practical calcUlation of the flow quantities is best carried out by numerical methods. We shall not enter into the details ofthese methods here but shall now pass on to some examples'showing the results of such calculations. In Fig. 11.11 we show four shapes of bodies of revolution that correspond to certain specified source distributions. These are taken from the calculations of Fuhrmann (1911) as reported by Prandt! (1925). In the figures the assumed source distributions are indicated on the· axes. The upper halves of the figures show the streamlines of the disturbance ftow field due to the source distribution, and the lower halves show the streamlines of the combined ftows. We thus sec t~at tile superposition of a continuous line distribution of sources and a uniform stream parallel to that line represents the axisymmetric ftow past a certain body of revolution. The source pOtential (11.20) is the simplest singular solution of Laplace's equation. 'There are other singular solutions which, like the source, can be used to generate ftow past a body. We shall dow describe another such . solution, the socalled doublet singularity.
11.1 "be Doublet Poteatlal
Co .sider the situation where a source. of strength q and a sink of equal strength are situated very close together, say at a small distance I apart. Let u.s denote the sink by 0, the source by S, and a field point by P (Fig. 11.12). Further, let
....:... r~OP
= U% + Cl»1
=ux!i'
41F
0
feE) dE J(% _. E)I + ,.
(11.58)
~
1 OS
The potential at P due to the source and sink is given by
The velocity components "., "r' ".. are expressed by
ac!>
II., II .... r
a% 
U

+ 411' Jo feE) [(% _
I ('
% 
aCI»
ar 
,
411' 0 [(%  E)I
l'
E)I
+ ,a]IA dE
(11.59)
E
J\ E)
+ ,I]~
dE
u.. '"  r
lac!>
a9' == 0
(11.60) Now, if we bring the source and sink together by letting I go to zero, th~n the potential 01.60) will also go to zero. Howe'ver, if at the same time as I goes to zero q is allowed to increase indefinitely in such a way that the product ql remains finite and equal to a constant fl, then the potential
, The ftow field is a:itisy~metric. Once t~e source distribution is specified, the velocity components can be evaluated. We. can. then proceed as before to determine the stagnation points. the streamline~, and, in:pa~icuiar, the stagnation streamlines of the
Steady Acyclic Motion
will not vanish. Instead it will assUme the value given by
JJJ
<l»(P) _ lim .!L (rl 
1"04" .+00
wItb.
rrl
!.)

(11.61)
. .9
~~
~
~
If 8 is the angle between I and r, we notice that as 1_ 0,
(r  rJ + OT  I cos 8
p
.J ..
~
....
:;
'""
il H:!
!i ...
'So
.~
... C!
.. :2
p
i ..
:1.,8
It 1:
.1
.oj
,0
Ji'Ia. 1I.U mustrating'the derivatiOD of the doublet potartial.
and Hence the e.<>tential (11.61)
~. ~
'; a
'S ._ Ii:
1~]
"0
°1~"O_
=.B ~
becomes
~(P) ... ~ It. cos 8
... ....
~
1 8 ;j ..
11 .8 ~ .~
~
":'
4,,"
(11.62)
.8.5
B"
"""= "'l0J
.0 _
.!! .,8 ~:I"
is
"'8 ·t·; !
which is known as the doUblet potenti4l. The particular sinksource combination it rePresents is callCcl a douIiIet or a dipole or a double source . The angle 0 in (11.62) is measured with respect toa particular direction, namely the direction of the line sqment extending from the sink to the ,.source. The direction from the 'Sink t&thesource of a doublet is known as the axis of the doublet. The constant /I is generally called the strength oj the doublet or simply the doublet strength. Since /I originated as the product ql, it is also referred to as the moment of the doublet: We notice that to specify a doublet we need to give two q~ntiti~' magnitude, namely, its strength, and a direction, namely, its axis. Thus we may describe a doublet as.a vector (11.63) f'  ",.e,. where ell is .a unit vector in the direction of the doublet axis.
336
IdealFluid Aerodynamics
Steady Acyclic Motion
337
Introducing the vector fL, we can express the doublet potential (11.62) as
1 fL' r «I>(P) =    ' , 41T
,a
(11.64)
We notice that the doublet potential, and cousequently the flow field. due to a doublet, is axisymmetric about the axis of the doublet.' This is in contrast to the source fiow, which is spherically symmetric. The velocity field of a doublet is readily obtained from the equation V
This equation is useful for conveniently obtaining the form of the doublet potential in term,s of any desired coordinates, It should be noted that in the exP pression (11.64) r is the position vector drawn frolr. the doubletand not from the origin of a chosen coordinate system to a field point. When the origin of coordinates is located at the doublet, then r becomes identical with the usual position vector. In such a case we speak of the expression (11.64) as the potential due to a doublet situated at the origin. Consider now the case of a doublet L:=~,fj Doublet fL situated at 5 from the origin 0 of a FIa.ll.13 Doublet not at the origin. coordinate system (Fig. 11.13). Let R denote the position vector from 0 to a field point P, and r 1 the vector from the doublet to!, Then the poteniial at' P due to the'doublet at 5 ~s given, from (11.64), by
= grad «I> ==  41T grad fL' r 1 ,.1
~~~~~~x o
Fig. 11:14 Coordinates for doubletftow with doublet at origin.
•
[f uR , u9' u., are the velocity components wiihJespect to the coordinates . R, 0, f/J, we have
UR
==
=
oR = ,27r 7
OCI>
14 cos 0
,
U 9
«I>(R)
= «I>(P) =
 
1 fL' r 1 a 41T .Irtl
R00
1
1 i)«lI
.u sin 0 = 41T R8
OCI>
(11.69)
u.,=~=O
1 fL' (R  5) =  41T IR  51 3
(1\.65)
_R sin 0 Of/J
In any axial plane along a streamline we have d", =  R sin Ou. dR
We shall now write down the form of the doublp.t potential in each'ofthe coordinate systems: Cartesian x, y, z; spherical R, 0, f/J; and cylindrical r, f/J, x. We choose the origin of the coordinates at the doublet, the Xaxis in the din~ction of fL, and designate the various coordinates as shown in Fig. 11.14. Then from (11.64) or (11.65) we obtain
/l «I>(x, y, z) =  417' {x2
$(Rj
+ RI sin OUR dO =
a constant
2
d(l!. sin 0) = 0 417' R I
Hence the streamlines are given by sin 2 0 _ C R , (11.70)
+ y2 + Z2}~
x
(11.66)
(11.67)
e, rp) =
 417'
Ii
.u
cos 0
]it
«I>(r, f/J, x) =  j
.rr ',:r
r2 2}~
x
+r
(11.68)
Different values of the constant yie1d different streamlines, The streamlines appear as ,shown in Fig. 11.15. All the streamlines begin and end at the doublet. They proceed, as shown, f; om the source side of the doublet toward the sink side. Near the axis of the doUblet, the streamline directions are the same as that of the axis. This is the motivation for defining tbe doublet axis as proceeding from the sink to the source. The doublet, just like the source, is a singular solution of Laplace's equation. It does not satisfy that equation at the point where the doublet
331
IdealFluid Aerodydamic:a
Steady Acyclic Motion
339
is situated. At that point both the potential and the velocity of the doublet become infinite. The doublet flow field, just like the source field, dIes out with distance from the singularity and vanishes at infinity. In fact, the doublet dies out faster than the source.
where
~,
is th~partial·derivative of~. with respect to dista.nce along the n direction ell' We thus see that the doublet potential can be generated by simply differentiating the source potential with respect to distance in a certain chosen direction. According to (11.72) the negative of this derivative would be the potential of the doublet and the chosen direction would be the axis of the doublet.· In thjg way, by successive spatial differentiation of the SQurce potential, we can build up a number of higher order ~ingular solutions (referred to as sources or poles of higher order) of Laplace's equation. .. suitable superposition of such higher order poles may then })( used in the solution of specific problems. For our purposes the source and doublet singularities are adequate.
0:'
is the potential due to a source of.unit streDgth and
U.8 Doublet in a Uniform Stream: Flow oyer a Spbere
Let us now consider the superposition of 'a doublet .... and a uniform stream U whose direction is parallel to the doublet axis. Since the source side of the doublet should face the uniform stream, its axis opposes the uniform stream. We choose the origin of coordinates at the doublet, the Xaxis in the direction of U, and designate the various eOOrdipates as before (see, for instance, Fig. 11.14). We thus set
UiU
1'11.11.15 Stream!iDel or. doublet Bow.
and
.... == ip
There is a simple and significant relation between the doublet and source potentials. To see this we considtr a. doublet ell of unit strength and
express its potential. following (11.64), by
.~
II.
   e .r41T" a
1
In termso( spherical coordinates R,6, " the potential at a field point due to the uniform stream is Ux:"" UR cos f) and the potential due. to the doublet, according to relation (11.65), is
r
(11.71)
Since
~d
~ = grad(~)
== e· • grad (11) " 41T r
1 cos 6 P 41T
IiI
Consequently, the potential due to the combined flow is given by
.J>(R, 8, V')
the potential (11.71) tak~ the form
.
== ==
U R cos ()
+ ..!!. cos ()
41T R'
(11.73)
==
ell • grad (~,)
(UR + E....!..) cos () 41T R'
o. '" ==  on (",,)
(11.72)
• The negative sign in (l1.72)could have been avoid~d if we defined the axis as tbedirectioo from the source to the sink instead of from the sink to the source.
Steady Acyclic Motion
340
341
w~
IdealFluid Aerodynamics In the axial plane along any streamline have
We notice that the potential, and consequently the ftow field, is axisymmetric. The velocity field is given by the components
tis
dtp
== R sin Ou, dR + R'sin OUR dO
=
0cJ) 
oR
==
(
U  II 
211" RI
1") cos 0
Hence the streamlines (11.74)
== d[(UR
2
a;~
I _
L") sin' oJ = 0 4r.R
==
C, a constant
given by
S . ·2 ()
tI,
== ~::   ( U + :11" ~I) sin 8
1 ==  _ . · .• 0
( URI \ 21rR
L)
(1'1.76)
11
•
R sin O~
The coordinates R" 0, of the stagnation point are obtained by setting the right members of (11.74) equal to zero. We thus have
By substituting the coordinates of tr stagnation point, from (11.75), in the lefthand member of (11.76), we find that the value of C for the stagnation streamline is zero. Hence the stagnation streamline is described by the equation
( .URI    sm i 0 == 0 Jl 21r R .
0, == 0 or 11"
and
1) .
(n.77)
R
R,
==
(Lj 21rU
p
R
or, equivalently, by (11.75) and
Jl 1 UR  211"  == 0 R
I
sin l 0== 0 for any for any
(11.77a) (l1.77b)
We sec, therefore, that there are two stagnatiqn points shown by A. and B in Fig. 11.16.
0
Equation (11.7 a) shows that the whole Xaxis forms a part of this streamline. Equation (11. 77b) shows that the circle
R
==
C'~IJ""' ==
0,
a constant
(11.78)
~~~x
describes the rest Of that streamline (Fig. 11.16). Other streamlines of the ftow appear as shown in the figure. The surface containing all the stagnation streamlines of the whole field is obtained by rotating the circle of radius R == a about the Xaxis and is, therefore, a sphere of radius a. We may thus conclude that .the flow resulting from a particular combination of a doublet and a uniform stream represents the flow ,due to a sphere in a uniform stream, the radius of the sphere being given by (11.78). Conversely, we may also state that the flow field due to the motion of an originally uniform stream of speed U past a sphere or radius a may be represented by the superposition of the uniform stream and a doublet whose axis opposes the stream and whose strength is given by ( 11.79)
Fig. 11.]6 Flow past a sphere.
341
IdealFluid Aerodynamics
Steady Acyclic Motion
343
Substituting (11. 79) into Eq. (11.73) we obtain for the potentiaJ
.(R,
9) ...
u( 1 + ;;.)R cOs 9
(11.80)
Consequently, the velocity compoAents are given by
uB(R,9)
==
u,(R,9) 
u( 1  ~:) cos 9 u( 1 f. 2~') sin 9
(11.81a) (11.81b)
as we have seen, by a superposition of the uniform stream and a doublet with its axis opposing the streaJil. This means that the lateral flow of a uniform stream past a certain body ofrevo1u.tion (namely, the sphere) can be represented by the .superposition of the stream and a. doublet on the axis of revolution of the body with the doublet ax.is normal to the body axis. One can generalize this idea and represent the lateral JIow over a IIonspherical body 0/ revolution by a suitable distcibution 0/ doub~ts 'along the axis o/rel)oIUliim with the (lXes o/the doublets being normal to rife body ax';s; the doublet axes oppose the direction of the undisturbed stream at infinil'y (Fig. 11.17).
The disturbance field is that of a doublet [see also (IO.48lJ·
11.9
u.e DIstrlbUtIoa of Doublets Ia • Ualfona Streua: Lateral .... AxIsymmetric Flow Past • Body or Reyolution
)
Since the sphere has no distinguished axis of revolution, t~e flow field due to. the' motion of a uniform stream past a sphere can be gIven another interesting and useful interpretation. Let Us choose, as before, the origin
Fig. 11.18 Axial flow past a body of revolution by combination of • doublet distribution L..1d a unifonn stream.
A line distribution of doublets can also be. used to genera~· a body' of revolution in axisymmetric flow. Consider a distributicn 8f doublets along a line such that the axes of the doublets point along the lin~ (Fig. 11.18). Such a distributi'!n is equivalent to a continuous one of sources together with a source at one end and a sink at the other. The superposition of axial and lateral flow solutions leads to the flow past a body of revolution at an angle of yaw.
11.10 Flow
rast
Arbitrary Bodies of Rel'olutfoa
1'11
Fla. 11.11 Laltefal ftow past a body of revolution by combination of a doublet distribution and a unironn stream.
of coordinates at the center of t~e sphere and the Xaxis parallel to the uniform stream. If we select the Xaxis as the axis of revolution of the sphere, then the flow field is that due to a ~niform st~ea~ parallel to the axis of revolution. However, if we conSIder any hne tn the plane normal X as the axis of revolution of the sphere, then the flow field is that due to a uniform stream normal to the axis of revolution. Such a flow is generally referred to as lateral or transverse flow over a ~dy of re~olu tion. Whichever point of view we wish to take, the flow field 1S determmed,
"0
We now turn to tbe practically important problem of determining the flow past a given body of revolution. If the. surface of the body is analytic, that is, if it can be described analytically, the solution for tbe flow field can be found by using appropriate curvilinear coordinates and the method of separation of variables. Ellipsoids of revolution .and the sphere are examples of such analytic bodies. Flow past such' bodies has been extensively treated and for the details Lamb (1932) and Munk (1934) may hi consulted. In this connection reference may be made also to' Thwaites (1960)_ where further r;:ferences will be"found. For the flow past a body of prescribed shape the powerful method of slJperposition of simple singular solutions is widely used. The problem now is tbat of determining the distribution of singularities which reprc:sent the flow over a given body of revolution. This direct problem is much more important than the indirect problem we have considered before where
Steady Acyclic Motion
314
us
IdealFluid Aerodynamics
the flow past a body ofrevoJution was g:nerated by superposing a specified line distribution of sources and doublets on a uniform stream. The direct problem appears to'have been treated first by Karman (1927). For axisymmetric motioo: he u$ed a continuous distribution of sources along the axis of the given body and gave a method for computing the distribution. For lateral ftow past the given body he used a continuous distribution of doublets along the axis with the doublet axes opposing the undisturbed lateral ftow. A method was given for computing the doublet
According to this theorem if and ",. are two scalar functions!,f position we h a v e ' .
"'1
if
&
and
("'1 grad ",.  ;. grad ;1)' Ii dS.,
fff(tfol VI4>t &
fa V'th) d.,.
(Jl.82)
where Ro is a region enclo$ed by the surface So. Let us set
1 ;1r
Fla. 11.19 Example of a body of ~olutiOD which caDDOt be fepraent.ed by a distribution of singularities on its axis &lODe.
where r is the distance from a ·fixed point P to another point and; i£ harmor;c function. Then (11.82) becomes .
II
distribution. For the details consult Karman (1927). An approximate analytic solution for the ftow past very slentkr bodies ot revolution is given in Chapter 20: The assumption that the ftow about a given body of revolution can be represented exactly by a distribution of sources 3nd doublets along the axis of the body is valid only if the shape of the body satisfies certain conditions. It is. thus applicable to socalled sle~r bodies only.. It is not applicable for a body with discontinuity in its surface slope, such as the body shown in Fig.l1.l9. It can be shown, however, that the acyclic flow
Suppose that the point P is external to So. Then Vl(l/r) vanishes at all points of Ro and (11.83) reduces to
f§ (~
Se
grad c/>  ", grad;) • n dS = 0
(H.84}
past any arbitrary body may be represented by a distribution .of sources or doublets on the surface of the body.(see next section). Hence the flow past
arbitrary bodies of revoJution may be found by using surface distributions of singularities. Such a treatment was originated by FluggeLotz (1931), who employed a surface c.iistribution of source rings. Smith and Pierce (1958) gave a method of calculating, by means of an electronic compurer, the axisymmetric flow past arbitrary bodies of revolution on the basis of surface distribution' of sources. Hess (1962) gave a similar method' for computing lateral flow past arbitrary bodies of revolution. For the details and specific problems their work should be consulted; for other references see Thwaites (1960). . 11.11 Flow Past an Arbitrary Body
Suppose now that the point is in the region Ro. SinceVl(l/r) becomes infinite at r equal to zero, it is necessary to exclude this point from the . region to which (11.83) applies. We draw a small sphere of radius a with center at P and apply (11.83) to the region between £ and So (Fig. 11.20). We obtain
.pber~.
tf (; ~ + ~)
c/>(P) = 
tiS
+
fj(; grad; So
c/>grad;) ·ndS = 0
£2
Now let £ go to zero. Noting that ciS on the sphere is angle we obtain
time'! the solid
417'
1fj(lr grad; 30
¢ grad
1 .1' .n tiS ,.
'
(11.85)
We now consiaer the r~oblem of acyclic flow past an arbitrary body. Such a motion can ~ represented by a surface distribution of sources or doublets. This can be shown by the U5e of Gre:n's theorem (2.141).
where f(P) is the value of t/> at the point P. Tilis gives the value of,J, tit any pob!t P ;11 terms of the values of ¢ and CJ,p/an on the boundary.
316
IdealFluid Af;r"odynarnica
Steady Acyclic Motion this way we obtatn
341
Now, the source and dOllblet solutions of
V'+
are given respectively by
== 0
1 41Tr a
,~ Jl:(! grad 9> 41T lr r 8
9> grad!) • odS
r
9>. ==
== 4>{P) == 0
if P is in R if P is outside R
(11.86)
Note that 0 is the outward normal to S. The statements made in the preceding paragraph therefore apply also to this ~se. The distribution expressed by (11.86) can, further, be replaced by a surface distribution of sources only or of doublets only. Let 9>1 be harmonic in the regionR I enclosed by the surface S. Say 9> is the disturbance potential under consideration. Now if P is a point in R we have
4>{P)
and
==  ..!... 1t(! grad 9>  9> grad !) • 0
41T It' r
B
r
dS
o == ..!... It' r grad 9>1 Jl:(! 41T
B
9>1 grad !) . n dS
r
F\i. 11.20 Illustrating the application of Green's theorem to show that the flow past an arbitrary body may be rep~nted by surface distribution of singularities.
Adding these two we obtain
and
~"
4>{P)
==  ~ ,t:( ! (grad 9> 41T It' r
B
grad
~1) • D dS
==
ell • grad
9>.
where eI denotes the axis ofthe doublet [see (11.72»). With these relations I· . Eq. (11.85) may be expressed as
+ ~ Jl: (9) 41Tlt'
B
9>1) grad! . 0 d S
r
4>(P)
=
if ~.
8,
grad
4> • 0 dS +
ff
8,
4><0 • grad
4>.) dS
This shows that 9> at P is that due to a surface distribution of sources and doublets. The density of the sources is  grad 4> • D per unit area. The density of the ·doublets is 4> per unit area. The doublet axes are alonE; the inward normals to the surface. The results (11.84) and (Il.8S) can be readily extended to the disturbance potential 4> of theirrotational motion iIi the .region R exterior to a solid b9dy. For this purpose we first apply the results to the region between the surf~·S of the ~y and aD arbitrary surface E enclosing S and then let E to to infinity. The integrals over I: go to zero as ~ goes to infinity. In
This again expresses that 9> at P is that due to a surface distribution of sources and doublets. The density of the sources and doublets is however different from that in (11.86). This means the representation of 4>{P) by a surface distribution of sources and doublets is not unique. This is in line with the uniqueness theorems: According to these theorems we know that 9> is uniquely determined only when either ~ or iJ9>/on is prescribed on the boundary of the region in which 9> is harmonic. The function ~I is to be tjetermined by prescribing either iJ9>I/'iln or ~1 on the su'rface S. Suppose that we require 9>1 = ~ on S. Then the above equation reduces to
~(P) =

..!... Jl: ! (grad 41T It' r s
~
grad
~1)' n dS
(11.87a)
J4I
IdealFluid Aerodynamics
Stca~y
Acyclic Motiori
349
This shows that t/> atP is that due to a surface distribution of sources only of strength (grad. t/>  grad i/lJ . D per unit area. Alternatively, if we require that grad on S we shall have
It is customary to express the pressure, particularly on the surface of a body, by means of a pressure coefficient defined as
C == P  Pao II 1V I
tP •D== grad tPl •n
P
(11.89)
ao
,tP(P) == 1 Jt (~
41T lf'
 ~l) grad! • n dS
r
Hence, from Eq. (11.88), it follows that the pressure coefficient at any point is given by (l1.87b) Thus CII at a stagnation point is l.
CII = I  
This shows that tP(P) is that du:e to a surface distri~ution of doublets only. The representation (11.87) is unique. We thus conclude that the aCyclic flow pastan arbitrary body may be represen~ by a distribution of sources alone or doublets alone on the surface oi the body.. Hess and Smith (1962) .gave a method of computing with· the aid of an electronic computer such a flow using a surface distribution of sourceS alone. For details and specific examples they may be consulted.
Vi Vao I
(11.90)
SpMre. We consider first the case of a sphere of radius a immersed in an originally uniform stream U. The pressure distribution over the surface of thl sphere, according to (ll.90), is given by the coefficient
C..(a, (),9' )
==
1 _ V,(a, 8, 9') I U.
U I +.U I R •
ll.n Pressures
We shall now describe the tlieoretically calculated pressure distribution over a sphere in.a steady flow and over certain slender bodies of revolution in axial flow. Once the velocity potential bas. been determined, the pressure field is given by the Bernoulli equa~on
==1
Ul
where ti, (), 9' are the spherical coordinates of a.point on the sphere. Over the sphere, from (11.81), we have
p
wnere
+ tpVl == H
a constant
and
VI == (grad 41)1 The constant H,can be evaluated by means of the free stream coitt"jtions, that is, the conditions in the undisturbed stream at infinity, or by means of the stagnation conditions, that is, those existing at a stagnation poInt. Thus if p", and V ao denote the free stream pressure and velocity, and if P. denott=s the pressure at the stagnation point, we have
u, = tUsin 8
Consequently, the pressure distribution over the sphere is given by ,
C..(a, (), 9') == 1 
t
sinl 8
(11.91)
H == Pao + tpVaoZ
The pressure at any point is therrgiven by
p == P, 
== P.
!p~'2
or, equiva.lently, by (l1.1S8)
This distribution, in any axial plane 9' = const" is shown in Fig. 11.21, where the angle fl is measured from the forward stagnation point. Since the distribution is symmetrical with respect to fl, the distrihution shown in the figure applies equally to eitht=r the top or bottom half of the sphere. In the same figure some measured pressure distributions are also shown; we shall talk about them a little later. Since the theoretical pressure distribution over an} meridian section of the sphere is svmmetrical about the diameter AB and also about the diameter CD, it follows that the theoretical value for the force on the sphere is 7.ero, Also, sin..:e ali the pressure forces pass through the center
Ideal· Fluid Acrodynamica
Steady Acyclic Motion
Jji
of the sphere, the moment on the sphere,5 zero. 1)ris is' in line with the ,considerations given in Chapter 10, Skillin BotIy 01 Rnolldio.. We "lOW consider the case of a slender body of revolution in axial fto~ obtamed by combining a specifi~ line distribution of sources and sinks with a uniform stream. The pressure distribution over the bodies sJtown in Fig. 11.11 were calculated by Fuhrmann, and his results are reproduced in Fig. 11.22. In the same
Theoretical • Mllsurec! R  4.3 X 105 Musurec!'R. 1.6 x 105 , /
,
/
1.0 1++"~,.....60t44+I
o
30
pp.
60
90
{J (d8lrees)
1
150
180
Cl'.
{J: angle from foIward
, "pu'
_Illation point
PII. 11.11 Pres!i~ distribution over the spnere fixed in an originally uniform stteam.
figure expcrimcbtally measured pressures are als" shown; we shall return to these later. The force can be obtained by integration of the pressure forces over the body. Such calculations show that the force on each of the bodies consid ~rcd is zero.
11.13 Discussion
Let us nofN examine the theoretical results we have obtained in light"of experimen.tal observations. We first consider the case oftbe sphere. In Fig. 11.21 we hate included .the results of pressure measurements made at two values of the Reynolds number pUd/p" where d is the diameter of the sphere and p, is the viscosity of the fluid. We notice ,that the. theoretical distribution differs fl om the mC8SUred one, which itself depends characteristically on the valu of the Reynolds number. Over the front part of~: 'e sphere (i.e., in the n"'t;hborhood of the forward stagnaLi:)n point A) there is a fair agreement, particularly at the larger Reynolds number. between the measured and tIieoretical results, but they differ drastically over the rear of the sphere.
Plate 10 II,m pasl a srl'ere illustrating: (0) laminar separation; (b) turbulent separatinn. l','urh:sy of Professor O. G. Tietjens. Plate 14 of Prandtl and Tietjen$
(1934).
351
IdealFluid Aerodynamics
Steady Acyclic Motion
151
The. theoretical pressure decrea!;Cs from its stagnation value at A to a minimum value at C ~ .... here the velocity is a max.imum) and rises agair. steeply to the stagnatIOn value at the rear stagnation B. The measured pr~ssure doc:s not ~xhibi~ ~uch recovery over the rear of the sphere. ThiS means tnat the Ideal flUid of our theory. in which effects of friction are completely absent. negotiates the (ising pressure over the rear part of the
Cp I
Cp
I
J
(b)
J
(d)
  Theoretical •••• Measured
Fla. lI.n Pressure dislriblltion over bodies described in Fig. 11.11.
sphere, whereas the actual fluid endowed with friction, no matter h"w ')mall, fails to do so. The theoretical flow field looks as shown in Fig. 1116 and does not depend on any par~meter like the Reynolds number. On the other hand, the actual flow field of a real fluid depends on the role of fluid friction and thus on the Reynolds number. It is beyond our intent to g~ into the details of the variation of the actual flow field past a sphere With the Reynolds number. !t is adequate for our purposes to note that at the Reyno!ds numbers that intorest us, the actual flo,," past the sphere separates from the body (see Plate 10). The point of ~epardtion depends at. the natu.re of the boundary layer on th<! forward part of the sphere. WIth a la~l£'ar layer separation occurs in the fmnt rart of the sph..:re. whereas With a turbulent layer it occurs in the rear. This ex.plains ·.... hv the measured pressure di~tribution3 differ b~twer.:n themselves. We thu~ see th;lt the actual flow tield pa,{ the sphere C00SIStS \~f separated flow ill ~ontrast to the unseparated flow predicted hy tl.e theory. Agl~ep.lent
between the two flow fields on the whole is bl:tter at the larger Reynolds number for which the boundary layer is turbulent and separation is removed to the rear end of the sphere. From the measured pressures it follows that the sphere in the actual flow experiences considerable drag in contrast with the zero drag predicted by the theory. On the basis of the foregoing discussion we conclude that the theoretical model we have employed for the flow past a sphere is indeed ;r poor approximation to the real flow. It should be rp.placed by a more reaJistic model based on a physical underst3nding of the actual flow. Further consideration of this matter is outside our present interest. We now turn to the slender body of revQ.\utioJ1 in axial flow. Referring to Fig. 11.22, we see that in this case there is very gcod agreement between the theoretical and measured pressures over the whole body except right at the rear end. At the rear end the theoret;cal pressure steeply rises to the stagnation value. while actually· such a rise does not take place. This deviation at the rear end is aga' n attributable to the effects of friction. From the measured distributio' w( learn that there is a drag for,ce acting on the body but that it is very t nall. This may be interprete~•. to a certain extent, as a confirmation cf the theoretical result that the drag is zero. Ct thus appeal'S that the theoretical model we have adopted for the axial flow past a slender body of revolution i!i: II eood approximatio:1 to the actual flow. From the point of view of the actual flow of a fluid past a body, the sphere and the body of revolution represent two different types of bodies. The sph~ is an example of a bluff body for which separation of the flow is an important feature (sec Section 1.9). Flow separatioll prevent~ the rL,~ o~ pressure one expects, on the basis of ideal ftuid theory, over the rear of the body and gives rise to a considerable drag force on the body. The crag depends on the position of separation and becomes smaller when the separation occurs nearer 'the rear of the body. In C\intrast to the sphere. the body of revolution in a· uniform axir.! Ilow is an example ofa streamlined pody for which flow separation, if it occurs at all. does so very near the rear of the body (sec Section 1.9). In this case the drag of the body is exceedingly small. It should be borne in mind that the drag we have been talking about is that d~e to the normal pressures acting on the body. It is (":tiled pressure or form drag and is quite distinct from the skin ffictil)n drag arising out of frictional stresses acting tangentially 0n the body. Skin fri.:{:on drag does not exist in the ft0W of an ideal fluid. For oluff bodies fdction drag is smal i compared to the pressure drag. whereas for streamlind hodies the pres.'iure drag is small compare~ to the friction dra~
Jj4
IdealFluid Acrodynamics
ODU
Steady Acyclic Motion
Jjj
11.14 Force
Arbitrary Body: d'A1anberi's Paradox
Wc consider, as we have ~n doing, steady acyclicjrrotational motion past a fixed rigid body. On th,.e basis of the considerations given in Chapter. 10 (~ Sections 10.5, 10.9, ~O.IO), the force and moment exerted by the fluid on the body are given by
It is important to bear in mind t~e assumptions underlying.,d' Alembert's result. They are
F=O
M ==
u)( ffp~dS
8
1. The ftuid is ideal, that is, inviscid and incompressible. 2. The fluid is unlimited, and the body is completely immersed in the fluid. 3. The body is moving uniformly, that is, with a constant velocity. 4. The motion is irrotational. 5. The circulation around every closed circuit is zero and consequently the velocity potential is single valued
We shall now look at the significance of some of these assumptions. If the fluid is viscous, friction comes into play and the present theory is out of place. If the fluid is non viscous but compressible, further investigation is necessary to decide about tbe applicability of d'Alembert's result. The theory of compressible inviscid fluid flow tells us that d' Alembert's paradox holds if the motion is completely subsonic and the a~sumptions (2) to (5) are made. D' Alembert's result will not apply if the fluid is limited in extent with either a free surface or a solid boundary too near the uniformly moving body. It is known that a body submerged to an insufficient depth in a fluid with a free surface, such as water, and moving uniformly experiences a drag. This drag (known as wave drag) is a consequence of a system of waves that appear on the free surface and continually remove energy to infinity. When a body moves nonuniforll)ly (i.e., accelerates) through an ideal fluid, the fluid as we had seen exerts a force on the body. D' Alembert's result, which is based on irrotational motion, has no significance for the rotational motion of an ideal fluid. The assumption that the fluid motion is acyclic and the velocity potential is singh:; ..alued has a very important consequence, As·we can surmise from the consideratIons of Section 10.5 it is this assumption that is responsible for the prediction of zero force on a body in a unifo;m stream. We emphasize that d'Alembert's result applies only to the force on the body. Even though the resultant of the pressure forces on the body turns out to be zero, the resultant of their momer.ts (with respect so some reference point) generally need not vanish. Thus an arbitrary closed body executipg a uniform motion under the conditions we are considering usually experiences a moment even thou~h the force on it is zero.
where t/I is the disturbance potential. The result Utat the force on an arbitrary body is zero, as mentioned before, is known as d'A.lembert's Paradox, after d'Alembert (l717178})." The !'esult implies that the socalled drag force on the body due to fluid resistance is zero. D' Alembert was the first man to attack. the problem of fluid resistance by means of a rational theory and then meet with the unforeseen result of zero drag.· He wondered how one could explain by theory the resistance of fluids when a carefully laid out mathematical theory led him to such a paradoxical result. Before the time of flying people were preoccupied with the problem of drag on a body moving uniformly through a fluid that is otherwise undisturbed. At that time d' Alembert's paradox generally meant the theoretical result of zen;) drag. What surprise~ us is not so much the zero drag force furnished by the ideal fluid theory developed so far but the zero lift force it predicts for lifting bodies such as the wings of an airplane. The absence of drag for the type of motion we have considered is understandable. If there is drag, work should be continually done by external forces acting on the body in order to maintain its motion. The work done should then rest in the fluid. It should appear a3 an increase in the internal energy of the fluid or as an increase in the kinetic energy of the fluid. In the latter case, energy should continually flow to infinity. Either of these possibilities is impossible in the motion of an ideal fluid under the conditions we have envisaged so far. We know that the internal energy of an ideal fluid is coilstant,t and we have seen that the disturbance velocity of the fluid due to the body dies out so rapidly with increasing distance from the body that there can be no flow of energy to infinity. With regard to the question of lift. it will be our ma.in concern hereafter to understand the physical circumstances more clearly and develop the theory further so that it can furnish us with useful results. .
• For additional interesting reading 0'1 this and related topics see Durand (19 }4) t In fact, it does not appear in our conslaerations.
11.15 Circulation as tbe Agency for Force
One of the principal aims of the present study of ideal fluid theory is to determine theoretically the fluid fo~ acting on bodies of aerodynamic
356
IdealFluid Aerodynamics Steady Acyclic Motion
J57
interest such as the wings of an airplane. But in light of d' Alembert's par.adox it appears that any attempts to achieve such an aim may prove fruitless. To a<;sure ourselves that the situation is not hopeless we shall now examin\: tbe steps that need to be taken to resolve d'Alembert's paradox anJ ~l:bsequently develop a satisfactory ideal fluid theory for the flow past a body such as an airplane wing. Consider tbe steady flow past a fixed rigid body and a.fsume that the motion is cyclic. This means we assume that the circulation around cirC14its ~nclo,.in:f!. the body is no! zero and that consequenTly the velocity potential IS not slflglecalued. To fix ideas we may suppose that the region exterior to the body when it is finite is somehow made into a doublv connected region. In such a case, on the basis of the considerations giv:'n in Section 10.5, we conclude that the force on the body does not vanish and is in fact given bj" [see (10.70)] .
normal
It therefore follows that
if
s
n x q dS =
~ n x V dS =
8
e
if
S
Y dS
where e is normal to the cross section of the cylinder in the riirection of D x V. Since V does not vary along,. the length of the cylinder we choose a unit length of the cylinder and write
dS = dl x ]
F
F k: pU x
if
11
n x q dS
(11.92)
bo~y.
where q is the disturbance velocity and S, as usual, is the surface of the It was shown in Section 10.5 that if e is any fixed direction [see (J 0.75)], then
e·
if
S
n x qdS
'" = Jr.(h)dh
'"
where is the circulation around a certain circuit drawn on the body. Note that since
r.
Flg.ll.13 Illustrating the relation between the circulation 3"d the force for twodimensional flow past an infinite: cylinder.
if
s
n x U dS = 0
where dl is an element of length along the contour of the cylinder. Denote by C this contour. Then, pir unit length "f the cylinder. we obtain
we
hUll;:'
if
s
n x q dS =
1f
s
e
n x V dS
if.
s
VdS
=e
(11.93)
c
f
V dl =
er
(11.94)
where V is the total flUId velocity given by
where r denotes the circulation around C in the sense of rightha.nd rotation about e. Substituting (11.94) in (11.92). we obtain the force on the cylinder as F = pU )(
er
To il\usirate explicitly the relation between the circulation and the •'!orce. let us consider the twodimensional flow past an infinite cylhder \vlth Its generators norm;" tv the fl ee stream. A plane of the motion is shc'wn in Fig. 11.23. The vcclor~ U, n, and V all.llc,VI lie in such a plulle. Furthtrmt)r~. sine.: V m.'Sl b.:: tangential to the body surface, D and " arc
per unit length. Since IU x el = U, and the direction of U x e is a unit vector e, normal to U and e as shown in the tigure. the furce buomcs
per unit length. We speak of this force as the Iif! nu drag.
Oil
lhe cylinder. There iJ
Jjl
~e
IdealFluid Aerodynamics
thus see that.ifthere is a circulation r around an infinite cylinder of arbltr~ry cros~ section .placed in a uniform strea~ U, the cylinder will expenence a hft force (I 1.95) L= pur per unit length (or span). This r~ult is known as the KuttaJoukowski theorem. We shall return to this later. . We have tHus come to recognize that a nonzero circulation around a bo~y in a uniform .;trea~ is essen~ial if a force were to act on the body. ThIs means that the velocIty potentIal must be multivalued. From what we know so far,. a mul~ivalued potential is possible only if the region exterior t~ the ~y IS lJlul~lply connected. Now the region outside a finite threedlmenslOn~1 ~y IS ce~inly not a multiply connected region. How can . the ~tentlalm that r~glOn ~~e multi valued, and how can we explain !h~ hft force on ~rtam fimte hftmg bodies such as wings 7 Our theory 10 It~ pres~nt state IS of.no h~lp in resolving these difficultiei. Any clues f~r stralghtenmg out the sltuatio~ ~ust co~e from a physical understanding of the ~ature of the Bow ov~r h~ng bo~he~ .. To gain such understanding, a conve~lent .place to s~ IS WIth the IQfiDltely long lifting body, the socalled mfimte or twodImensional wing, in a uniform stream. . In. contr~st wi~h the threedimensional situation, the region outside an mfimte cylinder I~ doubly connected, and the potential, theoretically at lea~t, can be mulhvalued. Then the circulation in any circuit enclosing the cyl~nder ~eed not ~ zero. Now, however, the difficulty is that, given a cylmder m a certam stream, we have no way with our present theory to know whether or not ~her~ is circulation around the body; and if there is, h~w ca~ one determme Its magnitude? Again, clues to resolve these dlfficultles must come· from a physical understanding of the ftow over a lifting body. Our task for the next.few chapters is then clear. First we shall take up the problem offormulatm~ the' twodimensional wing theory and analyzing so~e.ofthe problems assocIated ~ith it. We will then be in an advantageous posltJon to tak~ up the formulatIon of the threedimensional wing theory and the. analYSIS of s~me problems associated with it. As a first step in developmg the. analrtlcal fra~ework for the twodimensional wing theory we shall conSIder 10 the next chapter steady acyclic twodimensional motion.
Chapter 12
Steady TwoDimensional Acyclic Motion
When the motion takes place in.a series of planes parallel to a given plane and is the same in each oftltese planes, we speak of plane or twodimensional motion. The velocity, the pressure, and other quantities related to the . ftow are equal at corresponding points of the planes. They are thu6 independent of a space coordinate measured along an .. , is nor]1lal to the planes. We shall denote this axis by Z and the corresponding coordinate .by z. The velocity compOnent w alon~ Z is zero. The streamlines of the ftow are plane curves and lie in parallel planes normal to the Zaxb. $.teady plane motion is possible rnly in the case of an infi" .. ely long cylindrical body placed in a uniform stream with its generators normal to the stream. Such a body is known as a twodimensional body. Physically there are no exact examples oftwodimensional ftow, only situations where the motion can be considered a good approximation to twodimensional ftow. Studies of twodimensional ftows are, however, important. :rhey contribute to our understanding of th~ nature of ftuid motion. From the mathematicaJ point of view, twodimensional motion involves two independent variables in the governing equations, and this is a great help. It is particularly amenable, as we shall see, to mathematical analysis.
11.1 Recapitulation
We shall gather here for convenience some pertinent relations that have been introduced at several places in the preceding chapters. We choose the z = 0 plane as the representative plane of motion (Fig. 12.1). In analyzing twodimensio~1 ftows we of en use cylindrical polar c{>ordinates besides Cartesians. Hence we shall record both the Cartesian and .polar forms of the relevant terms and equations. In analyzing twodimensional motion we can work in terms of either the velocity potential, or the stream function, 01 both, as will be shown in Chapter 15. In this chapter we shall obtain the stream function and the potential for some simple flows . . First we use Cartesians x. y. z. The velocity V, the 'potential <1>. the stream function "', and all other quantities are functions of x and y only. 359
160
y
IdealFluid Aer<;>dynamics Steady TwoDimensional Acyclic Motion
161
n.'!
Furtber Considerations Relating to tbe Stream Function
W'e'recall that along any streamline '" == C a constant or, equivalently (12.2)
y
~~%~~+X
(12.1)
0/1
K
C2
Fig. 12.1
Coordinates ror twodimensional 110\:.
We thus have
ax
== (u(x, y), vex, == <1>( x, y) '" == ",(x, y) 0<1> == U == a",
v <1>
oy
y), 0]
:
..
•
·o/I=CI
~
dz a
b
•
)oX
Fig. 12.l Illustrating the relation between the stream runction and the flow through an arbitrary curve.
0<1>
oy
== v == _ 0",
ox
o~
In Cartesians this takes the form
ox
V~== +:=0 ox· oy· .
VI",
iJI4l
a", d + a", dy == v dx + u dy =
oy
0
02.3)
==
a·", + a·", 2
ox
= 0
Z,
In polar coordinates we have
: ha.ve
oyl
In terms of the cylindrical coordinates r, 0,
V = [ur(r, ~), u9(r, 0), 0] cI> = <1>(r, 0)
olp dr
or
+ 0", 00
dO == u. dr
+ ru_ dO == 0
v..
(12.4)
'" ==
",(r, 0)
0<1> = u = or r
1 0'1'
r
Consider any two streamlines "'. == C 1 and Ip = C2 and an arbitrary curve AB joining them (Fig. 12.2). Denote by dl an element of length along the curve and by, n the normal to that element. The net out flow of fluid mass through AB, measured per unit thickness normal to the plane of motion, is given b~
B
00
! 0<1> =
r
U
dO
= _ e
a",
or roO\r 00 r 00 r 00
m==pi
.d via AB
V·ndl
(12.5)
Now, if in Cartesians, dl = (dx, dy)
y2<1> =
\2",
! ~(r 0<1» + ! .~(! 0<1»
rot ror, or or
_ 0 
and
n = (a, b)
=! ~(r alP) + 1i(l
Olp) =
0
we then have dl . n = a dx and
+ b dy = 0
362
IdealFluid Aerodynamics
Steady TwoDimensional Acyclic Motion
363
From these equations it follows that
D
=
(dY _ dX)
dl' dl  v dx
(12.6) (12.7)
and
v . D 01 == u dy
Substituting (12.7) into (12.5) we obtain
m
the velocity vector. Also, the direction of the streamline passing through that point is that of the velocity vector. Since grad el> at any point is normal to the equipotential line,. passing through the point, it follows that the equipotential line and the streamline passing through any point are normlll to each other at that point (Fig. 12.3). "We thus see that the equipotential lines (el> == const.) and the streamlines (1p == const.) form an orthogonal net.
==
pfB
A vl&AB
(u dy  v dx)
In terms of 'the stream function this becomes
~
p
= fB(01p dy
A'
oy
+ 01p
ox
dX)
=
J:dlp
C1

= 11'(8)  1p(A)
==
C1
(12.8)
PII. W
Streamlines and equipOtential lines are orthogonal.
This ~ho",:s that the difference between the numerical values of the stream functIOns LS equal to the volume rate offluid flowing between them. Consider now a closed circuit <'C drawn in the region of the flow. The net outflow of fluid mass through <'C is then given by
12.3 Problem In Terms of tbe Stream Functioa Consider steady flow p~t a cylinder. The problem of determining the flow field in terms of the stream function consists of solving the equation
m
==
pL V·
BA
n dl
==pJv d1p
== lim (1p(8) 
1p(A)]vl& <'C
where A and B are adjacent points on rt'. If there are no sources of fluid in ~he region enclosed by the curve, tht: net outflow of mass through the ~urve IS zero, and consequently the integral f d1p around f(/ is also zero. This means 1p is a single valued function in any region in which there are no sources. It will be multivalued along circuits enclosing sources. In the plane of motiQll we can draw two sets of curvesone set described by the equation el> = const., the socalled eqUIpotential lines, and the other set. described by the equation 11' = const., that is, the'streamlines. At any po lOt of the flow field the gradient of the potential points in the direction of
ip the region exterior to the cylinder such that 1p or its derivatives satisfy certain specified conditions on the contour of the body and at infinity. Since the contour of the body is a streamline, it follows that on that contour 1p must be a constant. Since the velocity at infinity must approach the undisturbed vel~ity, it follows that at infinity the spatial derivatives of 1p should assume the corresponding components of the undisturbed velocity. Thus, in Cartesians, the problem consists of detcrrmining 1p{x, y) the solution of the equation
as
such that
1p{x, y)
==
a constant
on
F(x, y) = 0
364
IdealFluid Aerodynamics
Steady TwoDimensional Acyclic Motion
365
and that at infinity
In terms of 'the polar coordinates r.
e.
these equations take the form
0'1'  == U' '1 oy 0'1'
VJ{r,
and
e) == e) ==
Vr cos Ur cos
e+
Ur sin Vr sin
e
(12.11) (12.12)
C1>(r,
12.S Source Flow
ox
== U. j
e+
e
where F(x, y) == 0 describes the contour of the body and U is the free stream. * As. a fir~t step in building up the twodimensional flow fields past certaIn bodIes, we shall obtain the stream function and. the potential for three simple flow fieldsuQ,iform stream, a source, and a doublet. Our procedure will be to take the velocity field as given and then to find the stream function and the potential by integration of the velocity components.
In threedimensional flow the source flow is such that the streamlines are radia11ines in all directions from a point and the veiocity is a fu'nction only of the distance frolr. the source to a field point. A similar flow in two
y
12.4 Uniform Stream
Consider a uniform stream with velocity Q. Let U and V denote its components along the X and Yaxes (Fig. 12.4). Then
a'l' ==
and
ay
U
==
()<!)
ax
It therefore follows that
tp(x, y)
and
==
Vx Ux
+ .uy + Vy
(12.9)
FIg.12.5 Source flow.
C1>(x, y)
(12.10)
dimensions is described when the streamlines in any plane of motion :lre all radial lines from a point in that plane and the velocity is a function' only of the polar coordinate r measured from that point (Fig. 12.5). The velocity field, with respect to r, e, and z of such a flow, is then #I represented by v == u,.er (12.13) where ur = uir) only. We first see whether this velocity field satisfies the incompressibility and irrotationality conditions. Jt is readily verified that the curl of V vanishes everywhere. The incompressibility condition requires that divV With Eq. (12.13) this becomes
== 0 ==
0
Fig.12.4
Uniform flow. ,
 (ru r )
• Such a problem is known as Dirichlet's Problem.
or
o
366
IdealFluid Aerodynamics
Steady TwoDimensional Acyclic Motion
361
or
ur ==
B r
Integration of these equations yields say
where B is a constant. It therefore folloM that the velocity
1p{r, 8) (12.14)
!L 8 + const.
2.,..
(12.17)
V
==
B ur(r)er   er r
To obtain the corresponciing potential we note that
a~
is acceptable at all points except at the point r == 0, ~here it becomes infinite. This point is thus a singular point of the flow. It can further be seen that at this point div V ~ 0 'but is in fact infinite. This means the point is a source. It is often referred to as a plane or twodimensional sourct. Let.q denote the strength of the source, that is, the volume measured per unit thickness normal to the plane of motion of fluid being created at the source Per unit time. The strength q is then also equal to the volume of ftuid ftowing out of any curve enclosing the source. If, for simplicity, we choose a circle of radius e with its center at the source we obtain
q 1

 = ur
ar
211 r
and
Otll  = ru,,= 0 iJ(j
integration of these equations yields
~("
8)
== q
211
log r
+ const.
(12.18)
In Cartesians, the stream and potential functions are
where dl is an element of length along the circle. From (12.14) V • er on the circle is equal to B/e. Therefore
q
V'(x, y} =
.!L tanJ(lf.)
2."
:r
+ constant
==!!.[
eJelrde.
dl .. 211B
(12.15)
~(x, y) = :1.. log (Zl
4."
12.6
+ yl) + constant
Now, in terms of the source strength we can rewrite (12.14) as
Combinati9D of a Sourr..e and' a Sink of £qual Strength
V
_.!L er
211 r
(12.16)
The threedimensional picture of a plane source is simply a doubly infinite line source obtained by a ,uniform distribution of point (i.e., three:"dimensional) sources along a straight line. Hence q is the constant sour~ strength per unit length. If we assume such a distribution alon,g the Zaxis, we can obtain by inte~ation over the distribution the ftow field of a twodimensional source. To obtain the stream function w~ note that
=ru  
Consider a source and a sink each of strength 9 situated at the points .A and B, respectively (Fig. 12.6). Let BC be I\n ~is directed from ttv: sink tl) the source. If P is any field JX'int, let 01 dellotP. the angle P.A C and ()I the angle PBC. The stre~m function at P due to the source at .A is given by
oV'
08
q
when the zero strea:nline is taken as the axis BA.C. The stream function :it P due to the sink at B is given by
, q V'.  2". 8• 
r
211
and
  u, ' 0 or
oV'
when the zero streamline is again taken as the axis BAC.
Steady TwoDirnmsional Acyclic Motion
368
369
IdealFluid Aerodynamics
Il.7 Doublet
The stream function at P due to the combined flow is then given by 'I' = '1'1
+ '1'1
_
=
where 83 is the angle APB.
.!L (0 1
21T
OJ = .!L O.
21T
p
(12.19)
The doublet in two dimensi6ns is defined in the same way as the doublet ii'} three dimensions. Thus if the distance / between a source and a sin ... of equal strength q is allowed to go to zero such that the product ql remains equal to a constant we obtain a doublet of strength lA, The axis of the doublet is directed from the sink to the source,
p,
y
t
p
to'ig. 12.7 Derivation of a doublet.
We obtain thestream function for the doublet as follows. Choose the sink point as the origin of coordinates and set up temporarily Cartesian axes X, Y such that the Xaxis !uns from the sink to the source (Fig. 12.7). Let P be any field point wjth the coordinates r, 0 or, equivalently, x, y. Angles and 1 are :neasured as shown. The stream function at P due to a doublet situated at 0 with its axis in the direction of X is then given by
e
e
Fig. 12,6 Sourcesink combination.
tp(r, 0) =
with Ql~,.
lim 10
21T
.!L (81

0)
(12.21)
The streamlines are then described by the equation 'I' =  Os = com.dnt 21T or, equivalently, by
To evaluate. the limit we observe that for small I we may write
q
0
1
_
e == _1 sin 0
r 
1cos 0
Substituting this in {l2.2n we obtain
03
''"
constant
(12.20)
The latter is the equation of a circle passing through the point P, the source 2.t A and the sink at B. Therefore the streamlines are circles all of which pass ~hrough the sink and the source (Fig. 12.6). The directions of the streamlines are as shown.
IA sin tpr, 0) =   ( 21T r
e
(l2.1V
tp(x. y) = IA Y 21T Xl + yl
Idea:Fluid Aerodynamics The streamlines of the doublet flow are, therefore, described by the
equati~n
Steady Tw~DimensiOnaJ Acyclic Motion
Jl1
~
12.8 SoDRe aDd SiDk of Equal Stnllgtla
a UIIif_ Streua.
'P  ~ sin 6 . 271' r
= C,
a constant
Consider the superposition of a source, a sink each of strength q and a uniform stream U parallel to the direction from the source to the sink; We choose the origin of coordinates midway between the sourte and sink and th~ Xaxis in the dim.:tion from !he source to the sink (fig. 12.9). Thus U == iV, the source is at x = a, the sink is at %  Q.
p
+
u
~~
~~~~~~~~~.~~~~
..... 11.f FlO1lf put • Raultinc oval.
The stream function at any field point P w~thcoordinatt,5. T,8 or, equivalently, %, y is given by
"  Vr sin 8 + !L (8,  8J
2.". Fla. n.B Douclet flow.
This ;s the ~uation of a circle whose center is cn the axis 0 = 71'/2 and· . whose diameter is given by
It can be shown· that there are two stagnation pointsA'and B such that
OA OB ..
.4==sin 6
2.".C
.,
P
. . The st~lines are obtained from the equation stagnation streamline.is described by
j::;. .".V qa
tp
= constant. \The
The streamlines are as shown in Fig. 12.8. The potential for the doublet is given by 4>(, 6) == _ ~ cos 6 ,• 271' r (12.23)
cfI(x, y) =  : 2.". Xl + yl
UR sin 0 + .!L (6.  ( 1)
2.".
== 0
p
X
This shows that the whole Xaxis except the segmC'lllll"etween the sourcc~ and the sink forms a part of the stagnation streamline. The rest of this line is a closed curve as shown in Fig. 12.9. The cur~'e is symmetrical about Xaxis. It is known as an oval.
• The algebraic details are left to the reader as an exercise.
J71.
IdealFluid Aerodynamics
Steady TwoDimensional Acyclic Motion
J73
We thus·~ (as expected) that the superpositIon we are considering represents the twodimensional flow past, an infinitely long symmetrical cylinder. ' This method of sourcesink superposition can be extended to represent the flow past symmetrical cylinders of arbitrary' shape. Cylinders obtained in this way are called Rankine ovals. 11.9 Doublet lit Uniform Stream: Flow O'fer a Circular CyliDder Let us consider now the combination of a d~ublet '" and a uniform stream U with the axis of the doublet opposing the stream. We choose the
p(r. ')
The stagnation points in the flow are obtained by seui,ng these equations to.zero. We thus find that there ,are 'two stagnation points A and Bgiven by .
A:r  J",/21TU, B:r  J ",/21TU,
(12.26)
The streamlines ~ftbeJlow are described by th~ equation 'P == constant. The constant for the stagnation streamline turns out to be zero. Hence this strelllJlline is given by'
( 1    1) , sin 8  0 Il 21TU
,1
(12.27)
or by sin 8 == 0 for any r and (12.28) (12.29)
x
r
== J ",/21TU, a constant for any 8
FIg.12.10
Flow
past a circular cyl~r.
Drigin of coordinates at the doublet and the Xaxis in the direction of the uniform stream (Fig. 12.10). Thu!:
V==iU
f.L == if
Thr: stream fUDction.at any fipld point P(r, 8) due to the combined flow is then given by
Equations (1228, 29) show that the stagnation streamline consists of the whose Xaxis and a circle of radius J iJ/21rU with its center at the doublet (see Fig. 12.10). The surface formed by the stagnation streamlines lying in all the planes of motion is that of an infinitely long circular cylinder with its generators normal to those planes. Thus the superposition we are disclJssing represents the twodimensional flow past a circular cylinder. • It follows that if a circular cylinder of radius a is placed in a uniform stream V, the disturbance field due to the cylinder is represented by a doublet whose axis opposes the stream and whose strength is
'" _ 21rUaz
The stream function and the velocity components are then giveA by
..I'r, 0) = Ur sin 8    ' ~ 21T ,
,
'" dn 8
== U I r
\
The velocity components are
u (r 0)
r'

1
21TU r
!) sin 8
1) cos
f)
l'(r, 6) (12:24)
u( I 
;:) r sin 8
(12,30)
u,(r, 8)  U( 1 JJ U ( 1   
~) cos 8
(12.31)
==  0"" == r of)
1
21TU "
u,(r, 8) a
U( 1 + ;:) sin 8
(12.32)
and
il (r 0)
, '
==  0"" == or
U( 1+
 JJ 21TU rl
1) .
SID"
(12.25)
• The ftow is usually referred to as that over a cin:le. (n twodimensional How similar nomenclature is generally used.
314
IdealFluid Aerodynamics
St~dy
TwoDimensional Acyclic Motion
315
Let US look at thf! pressure distribution over the cylinde~. In ~rm~ of the pressure coefficient, the pressure at any point on the cylinder IS !Iven by
e,.( a, ll) .... 1 v
,.
V1(a. 0) Ul
02.33)
i      Theoretical
1.0
F
  M..sured R  6.7 x 1()5 ~,..  Measured R  1.9 x 1()5

t
\.
':\
o
,
..l. [1
Ll.
"
,
¥
~\
1.0
\\ \ \\ '.\ \\
.. "
\
1"'_
T~
1
! 1
~
, point, but at the rear of the cylinder the diS!=repancies are considerable. Also notice tha! the experimentaf results vary with the Reynolds number. Tht: cylinder, like the sphere, is a bluff body, and at the Reynolds :lumbers that intere:;t us separation plays a major part in the actual flow over the cylinder. (see Section 1.9). Experimentally, the cylinder experiences a considerable drag. The theory so f)ir developed, therefore, cannot Oe used to predict the actual flow of a fluid past the cylinder. In spite o)f this shortcoming the theoretical solution for the circular cylinder, as obtained above, is of considerable interest to us. It will play an important part, as we shall see, in our analysis of a twodimensional lifting body, the socalled infinite wing. In light of this consideration it is instructive to note that the theoretica! result of zero_ force on the cylinder is again related to the absence of any circulation around it. This in turn is a consequence of the iact that the potential of the flow field is singlevalued'
\
\
L IL
J
12.10 FIo" Past an Arbitrary Cylinder
We now turn to the motion of a uniform stream past a cylinder of arbitrary cross sec~i~n. In' this case we can attempt to represent the tlqw field by the superposition of a uniform stream aod a certain distribution of plane sources alone or doublets alone along the crosssectional contour of the cylinder.· The flow field thus obtained is characterized by a singh:~ valued potential and hence zero drculation. Again. the force on the cylil1Jer turns out to be zero. Because our concern is to d~velop the thl"ory so t'tat it can predict usefUl rl"slllts for the forces on a lifting body, we shat! now have tc l<'oK into ftow~ in which the potential is multivaiucd and the circulation is Ot't zero.
• See Smith and Pierce
(l9S~).
2.0
I

.,
\.
~
L
I
I
\
~
3.0 \) 30 60
~
"
.
jL
150 ISO
90 120 (deilrees)
c,~ ''
/I: angle measured from forward
stagnation point
Fit. 11.11 Pressure distribution over the circular cylinder.
On the cylinder,"r
== 0 and ", == 2U sin 0 and therefore VI == 4U' sin l 0
C"(a,,0)
Hence Eq. (12.33) becomes
==
I  4 sinl 0 '
This result is plotted in Fig. 12.11, which also contains measured pressure distributions for two values of the Reynolds number pU~1 JJ, where d is the diameter of the cylinder. As eT.pected, the. theor.etlcal distribution ~s symmetri..:al about the X andY axes and gIves fiSC to zero for~ on th~ cylinder. As in the case of the sptlere, the theoretical and experimental res~lts show some agreelll&Rt in the neighbort,ood of the forward stagnatIOn
CirculatiOll and Lift for an Infinite Wing in Steady Flow
J'f7
Chapter 13
Let ~$' choose the ,center of these CIrcles aa the Origin of coordinates. Then wl,th respect to cylindrical coordinates r, 8 in the plane, tile velocity field is descriQcd by
Circulation and Lift for an Infinite Wing in Steady Flow
v  u,e,
",.w,.
(13.1)
In this chapter we take up the formulation of the theory of lift of a twodimensional lifting body, known as the in.ftnite wing, fixed in a steady flow. For this purpose we first study flows with circulation. <Such flows are called circularory flows. As alreadj pointed oui. circulation is essential for lift. Wewish to emphasize that the theory 4eveloped SO far has no means of telling us whether or riot there will be ,circulation around a body placed in a uniform·stream. It is interesting to note that· around 1900 when mechanical flight was already realized as possible there was no rational theory to explain and predict the aerodyna.ruc lift obtainable from certain bodies that we call wings. We owe to Lanchester, Kutta, and Joukowski the final recognition of the connection between lift of a wing and the circulatory flow around it. Kutta (18671944) and Joukowski (18471921) independently laid ~e foundations for a quatttitative theory of lift of an, infinite wing. For a finite (i.t:., a threedImensional) wing Lanchesier (18781946) seems to have been the first to contemplate the connection between lift and circulation. l{owever, he did not develop a practical mathematical theory. Thiswas done by Pr.andll (18751953).
FiK. 13.1 Flow With circular s~: Bow wjth' constant vorticity.
only. In ~rder for this to re!,rtsent a physically possible tJow field it should satisfy the inco'llpressibility (;ondition (5.28), which now takes ~be furm '
au, =
08
0
Therefore it follows that in (13.1) ", shoWd be a function of r only, th~t is we have ' Let us take or
V= u~(r)ee
", = Cr
V (13.2)
13.1 Circulatory Flow with CoastaDt Vortkity
Our aim is to seek some twodimensional flows in which the circulation r = <p V • dI is not zero. With this in mind let us suppose that there is some tbw in, which the streamlines are closed' curves. Then the circulation taken around a streamline may not vanish', for the direction of the streamline at every point is the directi9n of the velocity, thus making V • dl positive all along that line. This suggests that as an initial step in searching for flOW3 with circulation we should look for flows with closed strearr.lines. As a simple example let us consider a flow in· which the streamlines are concentric circles (Fig. 13.1).
=
(13 3)
Cre,
, wJlere. Cis a I.:OI!stant, It Ca.l be verif.ed that (\3.3) is compatible with the equatIon of ~otion (5;)1). The circulatIon aruund a srreaJnlinc r = conS'. is
r="
iclrcle,
V.dl=J Cr 2 dO=').;a , T .. 7TLr
(134)
We tJhus see that (13.3) represents a physically possible cir(,uldtory ft"'\\i
378
IdealFluid A,erodynamics
Circl.:lation und Lift ljr an h.tlnite Wing in Steady Fiow
379
We now inquire whether or not thi$ flow is irrotatit'nal. The vorticity at any point in the flow is given by n(r, 0)
= k ~ druo == k! dCr'
'r dr r dr
== 2Ck
a constant
This means the v~locity field (13.3) repre3Cnts a rotational flow wit;1 uniform vorticity. Since vorticity is twice the angular velocity, it follows that the whole fluid is rotating like a rigid body with a constant angular velocity (j) == Ct. The flow repreSented by (13.3)· is' accordingly called circulatory flow with constant vorticity or constant rotation. *
(a)
13.2 Cil'C:oIatory Flow witltoat Rotation: Vortex "FInw
Since our interest is in irrQtational motion, we now ask whetht:r or not tllere is a ~irclliatory flow where the streamlines are circles but the flow fieid is irrotational. . Such II flow is possible arid' can be determined as follows. Now, the velocity.field has to satisfy both the incompressibility and irrotationality conditions. From the condition of incompressibility we obtain, as before, that V = u,(r)e, The irrotationality condition then becomes
n. == k! d~.! == II
r
ar
(i3.5)
Fig. 13.2 cdJy.
(c)
It follows therefcre that if we set
Vortex r1.:> ... : III point vorter; (h) ci<cl:lat!l1g flow arollnd a circl'lar ,,ylinJer;
ru, = K
or
a constant
(e)
03.6)
V=K~
r
the motion wouid be irrotational cxcep! poss~bly at th~ point r = 0, where the vorticity, a.:cording to (13.5), hecoll!es indetecmir.ate and the vel.)City, Clccording to (13.6), becomes infinit~ (Fig.' B.la). In a moment we shall cl~termine the value of the vorticity at that poict. The circuiation along any ~tf('.amline 1 = con~t. i!: given b}
We thl1S see that the velocity field given by (13.5) represents a circu!atl)ry flew Ih1t is Irrotational everywhere except pos~ibly at r = O. We call this now a cirr.lllatvryjfow lIilhout r(lration. To evaluate the value of the vorticity at the point r = 0 we use the rdatien between vorticity and .:ircul~tion. Aecordi!1g to .his '~e have
(Ql.,o • k =: (curl V)"
0 •
k
113.~)
r
dl = <t> l(dfl J~tr~le r .J = 271"K a constant .
r =li>
y.
~
= lim _1
6S~O~S
Jc.
~
V • dl
(1:l.7)
• !'10m .that bra!J:1C of the p~oce of vo~idty everywhere, lhe cir.;:lIl~tlOn (13.41 vark., from smlmh1le LO sln:amlme.·
.... here C k IS :l ~i1lJlI dosed curve Iymg in t;,e pl<!ne \ and enclosing the r()~11l r = 0 nnu a ~mall 2rea !lS. Choc~e C~ a!; a str~amline r = ~. Then E'l. (1.3~) bCl.:on,e,
(~L \ . •)
= Illn '.;;o~S
.
r,
Circulation and Lift for
310
~n
Infinite Wing in Steady Flow
MealFluid Aerodynamic:s
3'1
where O'is the magnitude of the vorticity and r~ is the circulation around the streamline r ,.. B. Now, from Eq. (13.7) we see that the circulation has the same C9nstant v.l1lue along allStrtamJjnes. Therefore
The spatial picture of the point vortex i~ a doubly infinite straight line normal to the planes of motion. Such a line is called a vortex line or a vortex filament.
13.3 Circulation as tbe Strr.ogtb of • Vortex Flow
r.) 1m ., ( Uro I' constant . . 48"0 dS
00
(13.9)
We _thus see that t!teflow represented by (13.6) .is irrotational everywhere eXCept at the point r = 0, where the vortiCity is infinite. It is a circrd4.tory'flow with concentrated 'vorticity. It is genel1illy known as vortex flow. Th" center point of the vortex flow is called the point vortex or simply the vortex. it is customary to refer to the velocity at any point of the y6rtex flow as the .velocity inducetl by the vortex. It mUsi: be und~r stood that this is simply a matter of convenience and does not mean that the vortt:X is actually causing the flow, for they just roexist. Because of its significant properties, vortex flo\\ piays an important part in aerooY<1dmics. Vortex llowcan be used to Tepresent approximately certain flows that can be realized physically. When dbing this, howev~r, we should remember that the flow near th~ vortex point ca'lnot be physically true, for at that point "oth the velocity and the vorticity are infinite. To avoid this difficulty there are two possibilities. One pOssibility is: to arrllJlge the vortex point tQ 1M' a fictitious one lying outside the fluid. Fot instance, the region .r'nclosed by any oftbe circular sireamlina of the flow can be considered as the cross section ~f a solid. 'In such a ()JlSC the vort.ex lies within thetJody and thus i~ fictitious. The resulting flow represents irrotational circuJatory motion (lrmma (I Circular cylinder (Fig. 13.2b). Another possibility oocurs when the center of the vortex lies in the fluid. In thi~ east we assume (on the basis of experimental eviderice and the !hoory or_cuI·flow) that there is R ~uid core or nucieus surrounding the center orthe flow and that the core rotates approximately like a solid body. Wi~hin the core we have· circulatory flow with constant vorticity {see Section Ill) and outside the core we have circulatory flow without rotation. Imide the core ", '" r and outside ", '" llr (Fig. 13.2c). Such a combination is known as an eddy 0': a free vortex or simply a vortex.· Th~ central core is called the vortt'x core. The tornado and water spout are examples of such a flow. Also, we ~e shall see later, a lifting body trails b~hind it free vortices.
• If al1 eddy <)(;CUR in a ftuid that is otherwise undisturbed, the spatial >eation of the eedy remains unalterro. Hf)wever, if a 'uniform stream is superposed on t, it will move with the strUm. Thus if a vortex is locate(\ at a point in the fluid wherl IIC velxity is V, the Vf)rte)t (i.e., the con; and the asso•.;ialed vutside circulatorY flow field) will teOld to mov: with t!i.e vdoc~y V. SUCh a vortex is therefore known as a fr~. 'Jor/ex.
Let us now look at some considerations related to the circulation in a vortex flow. Ha~ing alread~ seen that the circulation around every strcamline has the same value, we now seek the circulation arouO(~ an arbitrary circuit. Since the center of the vortex is a singular pomt, It is essential that the Cfnter be excluded from the rest of the flow field. We do this by surrounding the vortcx center with an arbitrary closed curve that may be located as cJ0se as is necessary to the center and agree hot to cross into the region en" closed by the curve. As a consequence the region exterior to the curve is doubly connected.· According t() the considerations of Section 9.14 Fla. 13.3 Symbol for we note· the following: vortex ofstrength r.
..z
The circulation around any reduclOle circuit not enclosing the vortex center is zero. The circulation around any irredl'cible circuit enclosing the vortex center is not zero. The circulation around every irreducible circuit has the same value. We thus see that the circulation around any closed cume enclosing the center of vortex flow is a constant for the whcJe flow field. It is therefore natural to choose the circulation r as the measure of a vortex flow. It is usual to call r the stre,1gth of the vortex flow or simply the strength of the vortex. We thus speak of a vortex of strength r and represent it diagrammatically as shown in Fig. 13.3. From Eqs. (13.6) and (13./, it follows that t:le velocity ofa vortex flow can be expressed in terms of the circulation by
r e, VCr, () ) =  271" r
(13.10)
It should be remembered that the sense of th:: circulation r in this equation is that rightband rotation about the Zaxis. that is, of a counterc1ock wise progressIOn along a circuit in the X Yplane. The value of r needs to ~e sne.:ified.
or
e~.:lude
• When ,mg,,:.H111CS :;uc~ a, vc)rtices :::nd sourus ocrur in the fiow it is necess3ry to th{IT1 from the regIOn of interest. In such a case the rest of the region Will be ITIJi!lp\ connected.
181
JdealFluid Aerodynamics
Circulation and Lift for ~1l Infinite Wing in Steady Flow
381
13.4 Stream and Potential Functions (or a Vortex Flow
The :;tref\m and potential iunctions of a vortex flow are related to its velocity field as fo1l9WS: 1 oV' 0<1>   =  = U =0
r
according to Eq. (13.11), V',(r, fJ) =
217"
.£.. log
(!:)
a
The stream function for the combined flow is therefore given by V'(r, 0)
00
or
r
and
=
V'i
eV' 1 0<1>   =  ~ =
or
r
00
Ufl
= 
r
£:2). f' + 'Pz = U (1:"1 r sm (J + ~ log(r) \ r .. rr a
y
(13.13)
217"r
Integration of these equations yields
r, ,
tp(r. 0) =  217" log,.
and
r
+
a constant
(13.11)
r
<Wr, 0) = ,
r
217"
0 + a constant
(13.12)
ILt~x
We note that the vortex potential is multivalued.
13.5 Uniform Flow Past a Circular Crlinder with Circulation
Let us consider the superposition of a uniform stream U past a circular cylinder of radius a and a circulatory flow around the cylinder. Such & combined flow plays a basic part; as 'we shall see, in .the analysis of the flow past an infinite wing. It is not suggested here that when a circular cylinder is placec in an originally uniform str~am the resulting steady flow will be such that there is a nonzero circulation around.a circuit enclosing the cylinder. All that we are concerned with at present js to examine the cOllsequences if the flow is one with circulation because mathematically !.uch a flow is admissible. Let us choose tile origin of coordinates at the center of the cylinder and the Xaxi!. in the direction of U (Fig. 13.4). We use polar cocrdinates r, O. For the convenience of virm"lization of the dow f..eld, we shall take the sense of the circlIlatQryjlow clockwise and designate the circulation as  r. Note thut ,he value oj l' is not spfcij'ied. The stream functioil for the flow field dUl! to the motion of the uniform stream past the cylinder is given by (12.30).
'l'1(r, 0) =
Fig; 13." Cl)ordi:1ates for How with cin'ulation past a circular cy!inder.
The velocity components of the flow are furnished by
U
r'
lCl1p (r 0) =  . r
oR
=
U(1  a 
2 )'
r2
cos
2 )
e
Sin ()
•
{13.14)
u,(r, 0) =  OVJ 
ar
=Ur
Tj
(a2 1+
r
Ufl
r
J
(13.15)
2rr r
At the stagnation points both given by the equations
and
are· zero. They are therefore
(I ,
U ( "1
a:)coso =
r
0
(1316)
+r2
a
2 )
sin 0 = 
r
211" r
1 
u( i

~)r sin 0
from which we deduce the following results. 1. If r = 0, there are two stagnation points with the coordinates r = a. o = 17" and r = a, (} = O. This result wt: had obtaineJ previously (fig. 2. If r :F 0, the stagnation points should pe locate,] such th,:t sin 0 /\ negati,'e, that is, !it.s betwe~n rr and 2rr
13.5a).
The !>tream fun(;tion for the circulatory Bow ar0und the cylinder is,
e
Circulation and 1.ift for an Infinite Wing in Steaay Flow
J61
3. One solution of the equations is
r=a
and
8 ... sin 1 (
41rUa
r )
;;..
""
C
u
...;
This :shows that in this case r ~ 4"Ua. Thus if r < ~Ua, there are two stagnation points on the surface of the cylinder, astltown in Fig. 13.Sb. If r = 4.".Ua, the two points coincide and only one stagnation point occurs ~ the cylinder at 8 = 3."./2 (Fig. 13.Se) . 4. All" alternate solution of the equations is given by
V
.. ...
::::>
.,
::::>
~
!\
...
~
.,
:;
.t:
u
~ <C
.
u
co
8
z:::
. 2
3.".
....
~ 0..
.51
~
r == _1_. 4.".U
Thi~'
rr ± .Jr'  (4.".ua)~J
c
~
:::I
U
~
~
u:;
0
~
on
~

.......
'
.,
~
.....
'"
~
means r is real only if r ~ 4TrUa. We thus see that if r > 41TUa, there is a stagnation point outside the cylinder such that 8 = 31T/2 and r > a (Fig. 13.5d).· This.case will not concern us in our work. We emphasize that the location of the stagnatil)n' points 0." the body depends crucially on the vdlue of the circulation. In the present case they move downward as the circulation increases, The flow field, which is symmetrical with respect to both X and Yaxes when the circulation is zero, becomes more and 'more unsymmetrical with respect to the Xaxis as the value of the circulation i$ increased. We are concerned lhere ·with the ca8C of r < 4~Ua. The type of flow we are discussing here can be realized with a real viscous imcompres!>ible fluid by rotating a circular cylinder placed in an otherwise uniform stroam.t Pictures obtained experimentally of the flow are shown in Plate II. There is striking similarity between the streamlines determined experimentally and theoretically. Let us now look a! the pressure distribution over the cylinder. It i given by the Bernoulli equ~tion
p(a, 0)
pr~nt
=
H  ipV2(a, 0)
384
• There is a similar stagr.ation j>Oint inside the' c)linder. but this is irrelevant in our context. If r. a"d '. (corresponding to  and + in the above equation) are the, cocrdinates of the inside and outside points. '.', = a'. t Viscosity is the agency for generating the circu:atory motion ~round the rotat:.lg cylinder. N.otation of a circular cylinder in an inviscid stream has no eITe:!.
Circulation and Lift for an Infinite Wing in Steady Flow
387
~!'late H Flow pa~! a rotatil'g r.ylinder; (' is the speed cf the u:ldisturbed uniforrr. Slream, ~ is .he periphC'r<li ~pce<! of the cflinder. CUllnesy uf PrGfe~~(]f O. q. TidJens J>lati:s 7 !c 9 cf Prr.ndtl 31:d Ti\!tj<!Tl,; (1934). Las, two pictures appear aisu as Fig. V of O. G. 'h~tje:ls: Stromungslehre, Vol. i, Springe.Vec1ag. 19W.
J"
On the surface of the cylInder " r have
""
Ideal· Fluid Aerodynamics
Circulation and Lift fOr an Infinite Wing in Steady Flow
319
0 and, therefore, using Eq. (U.15), we
VI(a, 6)  u"a,O)
..
. 4U • SIDI 8
l + 4.,,1 1 +  :..0 q r a 2Ur. ."a
l
pre.venting separation and eddy formation; that is, as a device:. for the 50called boundary layer control, has been the object of several experimental investigations. For some details consult, for example, Goldstein (1933) where other. references may be found. 13.6 Flow with CircUlation P.st
l1li
(13.17)
Arbitrary Cylinder The concept of steady flow with Circulation can be extended to the flow past a cylindec of arbitrary crosssectional shape. The eAtension is based on the definitien of circulatien . .If for the flow past an· arbitrary cylinder we require that the circulation around a circuit enciosing.the cylinder be net zero we speak. of the flow as one with circulation . .It is of cour.~e not implied that if a cylinder of arbitrary crosssectional shape is placed in an originally uniform stream the resulting steady flow will be one with circulation. Whether it is so er net"is a matter forindependeru censideration. We recall that mathematically the regien. exterior to a cylinder is a deubly connected regien, and consequently· the solution for· the flew past the cylinder, given the shape of the cylinder contour and the free stream velocity at infinity, is not unique until the circulation is speci.fied (see Section 9.18). . Since complex irrotational flows satisfying certain prescribed conditions can· be .built lip by superposing several simpler Hows, we may regard the flow with circulation past an arbitrary cylinder as the result of superpesitien ef two flows: (1) a flow with zero circulation past the cylinder and, (2) a purely circulatory motion about the given cylinder. The latter flow may be theught of as resulting from a continueus distributien of vortices along the contour of the cylinder. It can be' verifieJ that the fluid velocity in such a flow at any point of the body contour is tangential ta the contour and that the velocity will vanish at infinity. Furthermere. it can be shown that the velocity is finite on th~ centeur if th:!re are no discon.inuities in the slepe of the centour. At points where the slope is discentinueus the velocity will beceme infinite .. Some ef these features are considered again in Sections 17.7 and 17.d where the disturbance fieid due to. an airfeil in an originally uniform stream is represented as resulting from a centinuous distribution ef vertices en the airfoil surface. 13.7 KuttaJoukowski 1beorem aDd the Problem of the Circulation 1beory of Lift
The pressu~ .on the cylioderis then obtained by 'substituting (13.17) into the Bernoulh equatiOL. The pressure distributiop is syrnm~.ncaJ with mpcct to the Yaxis but unsy~metrical with respect t~ tile Xaxis. This means that tbe cylinder experIences a nonzero force io the Yclirection but a zero force in the Xdirection, that is, there is a lift but no drag on the cylinder. The lift L per unit length of tbe cylinder i~ obtained readily by integratmg the Y_ c:ompenents of the pressure forces over a unit length of the cylinder. Thus
L ==
raw ")0 p sin 6 a dO
2
P =a
Substituting
lD
fl.
o.
. u,l(a, 0) sin 0 dO
this for u.1 from Eq. (13.17) and integrating we ebtain
L
= pur .
(13.18)
This is a significant result in that for the first time we have a nenzero fo~ce on a body. Further, it emphasizes the crucial importance of circulatIOn areund a body as an, agency of lift on the body. The eccurrence o~ lift on the cylinder can be explained in a simple manner: With 1l~ clrculatien, the flow over the cylinder in a uniform stream .IS sy~metncal and the velocities above and below are equal. If a clockWIse CJfc~lato~ flow is now superposed (Fig. 13.5b) the velocity above the cylinder Increases, whereas the' velocity belew it decreases. Cen~t"quently. accerding to Bernoulli's theorem. above the cylincel there IS a lew pressure (socalled "suction"), whereas below it there is a high pressure. This results. in a :lift ,en the cylinder. This· simpl:: physical picture of the cennectlOn bi=tween circulation and lift in the case ef the cylinder was. first given. by Rayleigh. It is interesting to. kD~W that Rayleigh. gave thiS exp!anatlOn to account for the irregular flight of a spinning tennt~ ball. He did net, however, explaif! the origin ef the circulatjrm. ,The lift.:xpcrienccd by retating bodies (e.g., sphere~' lind cylinder~) is usually known as the Magnus effect. The use of the rotating cylinder as a lifting device and as a device fer
We consider again the steady flow with circulatien past an arbitrary cylinder. Cheese Cartesians with the Xaxis in the d:rection of the free stream U. The circulatien around the cylinder, as shown. is taken in the cleckwise directien. As !'hown in Sectien 11.15 tt.e ferce on the cylir,der, per unit span. is given by (13.19) F = pUrj
1'10
IdealFluid Aerodynamics
Circulation and Lift for an Infinite Wmg in Steady Flow
391
or
!
= pur
where I' is the value of the ~irculation, j the dlrectioll of the Yaxis, and L the lift. This result, know/). as t.he KuttaJoukowski theorem, statts ihat if there is a circulation of magnitude r ar?una' the cylinder and if the un· disturbed velocity at infinity is of magniiude U, then a lift exis's, the magnitude of which is pur per unit span. Kutta (1902) and jc'ukowski (1906), independently of each other, arrived at this result. Since the circulation r is not known, this theorem does lIot permit an immediate c!etermination of the lift on the cylinder. It, however. furnishes the foundation for the theory of lift. It shows that such a theory mu::t rest on the pOSSibility of a finite circulation existing around a lifting cylinder and on the possibility of predicting theoretically the vr.lue of that circulation, gil'en th.e shape vf the contour of the cylinder and the free"'itream velocity. To develop the theory of lift we need to consider: I, the circumstances under which the :;teac!y flow past an infinite cylinder is a flew with nonzero circulation, 2. the cri terion that determines the vaiue of the ;::ircuiation when it exists, 3, the physical basis for the existence of a steady flow with circulation past a cylinder when sUl'h a flow exists.
13.8
the airfoil shape and the free stream are given. Theoretic~lIy, on the basis of the mathematical considerations given so far, a solutIOn for t.he flow field past tht. airfoil can be dbtaine~ for an~ value of the ClJ'c.ulatlo~ an~ thus the solution is not unique untIl a deflmte valo~ of the circulation IS specified. The theoretical flow pattern for three different vc:lues of the circulation is shown in Fig. 13.6.
(a)
(b)
Airfoils, Circulation, and the Kulla CODditk'n
(e)
From experimental observ~tions it had been known, from the '"ery beginning 01 flit;ht, that onZv certain bodies that have a rroftft· with a sharp (i,e" point<:d) trailing l'dge are suitahle as lifting bodies or wings, Only wings with a sharp trailing edge appea: to have wellJefined lif: or: them. We can describe a wing roughly as a flat or slightly cambered plate, symmetric with respect to a metiian plane. The thickness of the wing is much smaller than its other dimensions. The cross sections of the wing in planes paralld to the median plane are called air/Oil p.ofiles or simply profiles, A picture of such a profile was shown already in Chapter I. For sometime our concern will he solely wit~ a wing that is an infinite cylillder of invariable pwfile, The geometry of such an infinitl! wing is completely determined by the shapl! or the airfoil profile. For this reason we shall refer to the infinite wing as the airfoil (Fig. 1.10), On the basis of the ~arJy remarh in the preceding paragraph it is natural to expect that only the steady flow past an airfoil.will be a flew with welldefined circulation. This expectation is well born out bye)(periment. Now we face the question of how to determi'1e the circulation al'Ound the dirfc)l!, givcn its shape and the freestream velocity. Experiment som.. s that n",e :low past the airfoil and the lift on it ale uniquely delerminl!d once
retical flow pattern for flow with circulation past an airfoil. Fig. 13.6 Theo
s
The theoretical considef'ltions given So far~ w~ emp~asize, do nGt he.lp if the particular value of the welldefined circulatIOn that mu~t eXist i~/steady flow past an airfo~1. To be a~le to specify th~ clrc~I"tlOn ~e bring in additional consideratIOns. ConSider the flo'v ~. an ,deal flUid 7) past a corner ( see F 'lg . 13 . , It can be verified (by solVing, for the flow field by the me~hvd of separation of variable's) that the velOCity of the flow in the neighborhood ot the (;orner may be described by 'I
u, ., r( ra)/a sin rrO
(J
U '. r(ra )la r
'iTO cos at
where rand () are polar coordinates with the origin at the corner (see (llso
392
IdealFluid Aerodynamics
Section 15.2). We see that at the corner if 0: that is when ttow is inside a corner, and
u,+ 00
Circulation and Lift for an Infinite Wing in
Stcad~
Flow
J9J
u, 0
<."
>."
if. 0:
that is when flow is outsidt: a corner.
 ..:
~
Kutta (1902) and independently bV Joukowski (1906). It is usually known as the Kutra condition. This theoretical condition has been found to agree completely with experimental observations (Section 13.9). It follows that when an airfoil is set mto uniform motion through a fluid such as air or water, a circulatory flow around the airfoil must somehow come about. We shall describe briefly this aspect of the generation of circulation in the next seCtion. The addition of the Kutta. condition to oOr considerations completes the fram~ork necessary for an adequate ideaJftuid theory of the lift on an airfoil in steady flow. The resulting theory is known as the drcu/acion theory of lift. The recogniti,on of the crucial role of circulation in the generation of lift and the determination of a unique value of the circulation by means of the Kutta condition are land marks in the development of modern aerodynamics. Recall the opening remarks of this chapter 13.9 ".lIe Generation of, Circulation It remains to describe the physical basis of the preceding considerations .. For this purpose we examine 1he sequence of events that are observed experimentally when an airfoil is set into uniform motion from rest through a real fluid such as air or water. T.he flow pattern at the first instant of motion, as it appears from a bodyfixed reference frame, is as shown in Fig. 13.80. The flow is actually like the flow without circulation of an ideal fluid. Thus at ·the fira instant of motion the real fluid goes around the sharp trailing edge with a very high velocity. Owing to the action of the viscosity of the ftuid (no matter how small the viscosity) such a motion cannot, however, continue; soon a surface of discontinuity (i.e., a vortex sheet) emanates from the edge and a vortex. begins 10 form near the edge. Such a vortex is known as the starting vortex. As the airfoil proceeds in its motion the starting vortex grows rapidly in intensity while the extent of the vortex sheet increases (Fig. U.Sb). As a reaction to the generation and development of thestilrling vortex, which is a rotation of a part of the fluid, a rotation in a sense opposite to tlttt of the starting vortex is created in the rest of the fluid. In particular this rea~tion appears as a circulatory ftow of the fluid around the airfoil. The growth of the Circulatory flow follows that of the vortex .. The growing cir~ulatory flow modifies continuously, as shown in Fig. L1.8. the flow pattern around the airfoil. As the airfoil proceeds, the strength'ofthe starting vortex and that of the circulation around the airfoil grow simulta~eously until the flow field around the airfoil is such that the ftuid flows off smoothly from the traiting edge as shown i, Fig~ 13.Se. The full development of the starting vortex and of the circulation around the airfoil takes place quite quickly (usually
Fia.13.7 Flow past a comer
,~~cu atlOn r there WIll be in general flow around the trailing edge from one SI e to the oth~r and in all such cases the ('eloeily at th~ edge (which is a
ve OClty at thai edge; in such a case the flow will wat'e the trailin~ edf7e ' 0
,.'
N~w .we
return to. the flow past the airfoil. For arbitrary ('alues of the
hOte~et, there will be no flow around the trailing·edge and hence no infillit;
sharp corner WIth
0:
> .,,)
will be infinite. For one particular ('alue of r
smoO(h~v,
et u~ ar~und an Dlrfotl be of a strength just sufficient to make the flow leave the aufin/ smoothly at the trailing edge. This condition was put forth by
This o?serva~ionmay be made the basis for determining a unique value fO,r the clrcu~atlOn 10 t?e ftow past an airfoil and consequently for determInIng a umque ~ol~tlOn for that flow. We require that the circulation
J91
IdealFluid Aerodynamics
Circulation and Lift for an Infinite Wing in Steady Flow
.J9S
in about the time. i~ takes for the airfoil to move its own chord length). When such a conditIOn h3.s been reached and theairfoil has been in motion fcr a sufficiently long time the starting vortex is left way behind the airfoil. It has then !>ractically no influenCe on the flow around the airfoil. Whenever the condition of smooth flow at the trailing edge is disturl:ied, say by a
results when the airfoil is first accelerated from rest and then immediately . stopped.. We thus see that the experimental obser ..ations amply ro!'port the essential features of the circulation theory of lift described in the preceding section. Furthermore, as we shall see in the later chapters, tht: lift calculated by means of this theory is in fair agreement with experimental observations. We ::an now see, in light of the considerations given in this section, why the exist:nce of circulation around a body or the lack of it C<Juld not be dccidc.d solely on the concept of an· ideal ftuid. The generation of circulation,depends entirely on the'nature of the viscous flow of a ~cal ft~id past certain bodies: Once the role ofvi9COsity in generating the clrculahon has been recognized, we ret.um to the idea of an inviscid fluid and take into account the role of viscosity by assuming that a well"'<1efined circulation exists around the airfoil and that it is determined by the KlItta condition.
13.10 Matbttnatical Problem
We conclude thi!; chapter yrith a statement of the matheptatical problem for the circulation theory of lift. In terms of the ~elocity pOtential CI> the problem is to determine CI> as the solution of the equation
VICll = 0
in the region exterior to the airfoil such that
grad $ . D = 0
grad CI>  U
on the surface of the airfoil at infinity
(C)~
~
and the circulation r satisfies the Kutta condition. In terms of the stream function", the problem is to determine V' as the solution of the equati<.'n
FiR· 13.8 Developme!lt of flow pattern, as shown by experiment, around an airfoil: (a) Flow at the first instant of motion; (b) Flow a little later showing the growth of the starting vortex; ~c) Flow afterwards when t.he growth of the slarting vortex and the associaled circulation aro.md the airfoil is complete so Ihat the flow leav~ the trailing edge smoothly.
V'tp==o
in the rcgion 'cxterior to the airfoil such that 'I:' .. a constant o~ the airfoil surface, the spatial derivatives of tpapproech the correspondlDg components 'of U at inficity, and r sa~fies the Kutta condition. The Kutta condition may be eJt}'ressed in a more e).plicit form. If the ·trailillg edge ~/e is/nite, as s.hown in Fig. 13.9a, the velocity at the trailing edge musf be zero; otherwise a t'e/ocity discontinuity, which cannot be permitted, will result at the edge. That a velocity discontinuity cannot be allowed is seen as follow~. A velocity discontinuity, which can be only a tangential discontinuity, requires that the component of t~e veloc!ty ilormal to the discontinuity must be continuous through the dlscontmulty
Change in the speed of the airfoil or in its angle of attack, a new starting vortex is formed, and a new value of the circulation is established such as to restore smooth flow at the trailing edge. PhotograPhs of the consecutive stages of flow around an airfoil starting from rest are shown in Plate 12. Plate 13 shows the flow as viewed from a spacefixed reference frame. Picture 130 refers to the situation soon after tht: formation of the starting vortex. Piuure 13b refers to the situation that
396
IdealFluid Aerodynamics
Circulation and Lift for an Infinite wing in Steady Flow
397
Plate 12 Consecutive stages of flow past an airfoil starting from rest. Courtesy of Piofesso~ o. G. rietjens. Plates 17 and 18 of Prandtl and Tietjens (1934).
Plate 12 (Continued)
398
IdealFluid Aerodynamic:;
Circulation and Lift for an Infinite Wing in Steady Flow 7). Now at the trailing edge tho velocity C?n the upper_slae of .moil is parallel to tMt side whereas the velocity on the lowl:!' side IS pataller to the lower side (Fig. 13.90). This means, if the velocities are finite; the components normaJ to the ,discontinuity that' must emanate, tromthe trailing edse In, that f?ISC wilJ not be continuous. We conclu~e that the velocities on either side of the airfoil at the trailing edge must be zero;' ill other wotdla trailmg ~~ with a fortt~ ang/~ must be a,&tapation point. Considtr now a taped ~g edge ..OJ shown iri Fig. 1l.9b. Now the vClocities on either side of the ~rfoil at theedgiare in the same direction,
th~ ~ ~pter
(b)
JlI&. 13.' Shape of trailing edge: (a) trailing edge with a finite
u.iling edge.
arigle~ (b) cusped
and according to the Kutta condition we may expect that all that is
necessary is for these velocities to be finite. As cab be readily shown the magnitudes of the velocities must be the same; If their magnitudes are not
the same a discontinuity 'must be assumed to emanate from the edge. Then. because the pressure must be continuous across thedisc0.ntinuity, we should have. according to Bernoulli's relation. at the trailinsi edge
P+ + fu +'.,.p0 =p +fu' 2 2where P 4enotes the pressure. II the velocity, Po the sugnation pressure. and the plus and minus subscripts denote respectively the: upper and lower sides of tbt airfoil. Since p+ is equal to p_, it follows that the velocities on Cither side must have the same magnituae. We c;onc1ude that a cuspet1 trailing edg~ Med not be a stagnatio" poi"t, but at such an edge the [)('!OCity mwt be fi"ite and have the same magnitude 011 eIther side of the airfait. From the preceding considerations it follows that there cannot be a sulface of discontinuity (i.e .• a vortex sheet) emanating from [he trolling edge of an airfoil in steady flow whether the trailing edge angle is finite or zero. Note that this result does not hold for an airfoil in unsteady mOtion or for a finite wing in steady flow. We return to consider the mathematical problem for determining the steady flow past an airfoil. As has been pointed out before. it is in general
Plate 13 Flow past an airfoil as viev.ed from a spacefixed reference _frame, (0) !mmediatcly after starting the airfoil. Note the starting vortex and the formation'of the circulatory flow over the. airf"jL (b) Situation when the airfoil \Vas stopped after tIJe formation of th~ starting vortex, Note the "stopping" ..ortex and the decay of the circulatory flow_ The starting vortex had been detached, Courtesy of Profes.w( 0_ G. 1 ietjens_ Plate 22 of !'I'PIi:ltJ and Tietjcns (1934)_
Circulation and Lift for an Infinite Wing. in Steady Flow
400
401
IdealFluid Aerodynamics we must require that
not possible to obtain in a direct manner an exact solution of the problem for an arbitruy allioil. Approximate c\o,ed form solutions may, however. be constructed for the thin airfoil by the method of superpositIOn of ~imple singular solutions. This method is described in Chapter 17. In constructing the solution for the airfoll problem we are concerned with solving Laplace's equation in a plane. given seme auxiliary conditions on an arbitrary contour in the plane. Difficulties arise when the contour is not a simple one such as a circle or an ellipse. In view of this we may introduce in the plane a new set of coordinal~s in terms of which the contour may be represented as a simple [oml and attempt to solve the problem in terms of the new coordinc.tes. In gerieral, however, when a transformation of coordinates is employed the Jorm of the governing equation for the problem will also change unles, the transformation satisfies certain conditioils. Let us consider this matter in a little more detail. In Cartesians we are to seek the solution of the equation
(13.20)
(~~J + (~~r == (~~J + (~:IJ ~ 0
O'ql oiA olq, ax'
4
olQl _ 0 oy' oy"
+ o'ql ... 0
~ oql +~ql oq == 0 ox ax oyog
It can be shown that the transformation (13.21) which satisfies the above four conditions is given by a relation of the form
(13.24)
where f(x, y) may.be either the velocity potential or the stream functIOn. For the reasons outlined above, let us introduce a ,JIew set of coordinates ql and q. such that ql = ql(X, y)
where i and h denotes an arbitrary function of the definite fUDction (x + iy). As is known,. (:t + iy) and (ql + iql) are complex variables. We thus conclude that the transformation of C()ordinates we are seeking must. be affected by' means of the functio!lS' of a complex vatiable. We are thus naturally led to the study of the functions of a complex variable and of their use in analyzing the twodimension~ irrotational motion of an ideal fluid. We now tum to thissLudy.
!=i.
q. = q.(x, y)
(13.21)
Let g(qlt q.) correspond to f(x, y). It may be verified that (13.20) then takes the form
(13.22)
If we desire that this equation reduce t.> Laplace's equation in ql and qt, that is,
(13.23)
Elements of the Theory of Functions of a Complex Variable
493
Chapter 14
Elements of the Theory of Functions of a Complex Variable
function 'P, the velocity components u and :11, any new independent coordinatts ql' q. which may be employed to relate the potential problem connected with a complex geometry to that connected with a simple geometry. To investigate the general solution of (14.1) we introduce new independent variables E(%, y) and 'J{%, y), which will reduce the equati.on to a form that may be integrated readily. It is found, as lIlay be verified. that if we choose
i(%, y)  %+ N y
,,(%, y)  % ~y
Eq. (14.1) reduces to
In this chapter we shall acquaint (\urselves with the elements of the theory of the functions of a complex variable. We introduce the complex variable through .the general solution of Laplace's "'Iuation in two dimensions. Finding the solutions of Laplace's equation is the essence of our problem no matter how we wish to attack it. Following the introduction of the complex variable, we study the differential and integral calculus of an analytic fwrction of a complex variable. A differentiable function is an analytic function. We shall find that the properties exhibited by the so· called real and imaginary parts of an analytic funytion are identical with those exhibited by the potential and.stream functions ofa twodimensional potential motion of an ideal fluid. Furthermore, analytic functions provide us with the type of transformation of independent variables we were inquiring about in the last chapter. In the next chapter we shall apply the results·we obtain here to the problem of twodimensional potential motion. The theory of functions of a complex variable is an elegant. and rigorous study. Our main concern here is with the role of this theory in twodimensiomll potential motion, and as such we must coptent ourselves in presenting the elements of the theory with only the necessary rigor. The reader should bear this in mind and should refer to the works cited at the end of this book to app;ei:iate the full beauty and scope of the theory. 14.1 Geaenl Solution of Laplace's Equation in Two Dimeaslons: IntrodadloD of ibe Complex Variable ,I o'g olg Vg=+=O where
aE~
(14.2)
~i .. 0
04.3)
g(E, '1)  g[%(E, '1), y(,E, '1)]
The functions %(E, '1) and y(~, '1) are simply the inverses of the transformation relations (14.2). Equatio~ (14.3) readily integrates to
gtE, '1)

J.(E)
+ !..(r,)
Eand 1].
(14.4) .
Ny)
where };(E) and 1.(1]) are any arbitrary functions of the variables It therefore follows that
g(i, y)
= h(%
+ .fIy) + f.(% 
whereh andf. are arbitrary functions of the single variables (% + N y ) and (%  ~y); respectively. These variables are, however, d~nite independent func1ionsof % and y. The functions h and/. are entirely arbitrary· except for the obvious requirement that h. should possess the first and ~nd derivatives with respec! to its own argument (% + Ny) and that f. should possess the first and .second derivatives with respect to its own argument (%  N~).
Let g(%, y) be any function governed by the equation
oy2 With reference to twodimensional irrotational motion, g(x, y) may represent anyone of the quantities: the velocity potential <I>,.the stream
ax'
(14.1)
If a.significant meaning can be attached to the entity the single y ) and (%  ~y) may be Interpreted as simple variables (% + combinations of a pair of real variables or numbers. Then a whole class of functions of the variabl~ (% + N y ) and (%  • ly) be~ome ./ available for constructing solutions of Laplace's equation. Square roots of negative numbers made their appearance early in mathematical pursuits, and a system of analysis involving such· quantities has been developed extensively·t The entity .../ I is cal1 the imaginary unit and is usually .ed
ri
N,
• fa and /1 should be such that together they yield a function g(z. 1/), whkh, being a physical q~tity. should not involve V  1. t They arose Il!' early as the Middle Ages when mathematicians soupt a general solution of quadratic: equations.
402
IdealFluid Aerodynamics denoted by the letter i. A combination of the type x + iy is known as a complex or imaginary number. If x and yare variable, the single variable :r: + iy is said to be a complex variable. A function of the wtriable x + iy is known as a function of a complex variable or simply a complex function. The other independent combination x  iy is known as the complex conjugate ofz + iy. In the same way, x + iy is the complex conjugate of x  iy. We thus see that in seeking the solutions of Laplace's equation in two dimenl.ions we are naturally led to a study, of the theory of functions of the type f(x + iy) and F(x  iy). We shall now go into some of the details of this theory. The theory is developed around the variable x + iii· The role ofth~ conjugate x  iy and its functions will become apparent during the study of the functions of the variable x + iy. 14.1 Nomeadature and ~ebra or Complex Numbers We introduce the imaginary unit i through the relation
is
Elements "of the Theory of Functions of a Complex Variable
Division
(a
40S
+ ib) = (c + id)
=
where c + id # O. Two complex numbers
+ ib)(c  id) (c + id)(e  id) (ae + bd) . be  ad + 1el + ~ (.1 + d'
(a
~
(14.10)
Q'
+ 'ib and c + id are equal only when
a
=c
and
b
=d
14.3 Geometric Interpretation Every complex number a + ib represents an ordered pair of real numbers (a, b).· The order of a number in the pair denotes whether it is the real or imaginary part of the complex number. Now, every ordered pair of real numbers represents a point in a plane. This means complex numbers may be represented geomt!'trica/ly by paints in a plane. For er.ch complex number there corresponds only a single point in tho plane, and, ~onversely, for each point there corresponds a single complex number. Let us set up a Cartesian system of axes to mark out points in a plane or, equivalently, to represent a system of complex numbers. Complex numbers that have only real parts are represented by number pairs of the form (a, 0) and lie on one of the axes. Such an axis is called the real axis. Complex numbers that have only imaginary parts are represented by number pairs of the form (0, b) and lie on the other axis, which is called the imaginary axis. An arbitrary complex number % represented by a number pair (a, b)is then given by a point whose coordinates with respect to the real and imaginary axes are, respe~tivel)', a and b. A plane, the points of which represent a system of complex numbers. is known as :l plane of complex numbers or simply a complex plane.t It is usual to denote the real axis by X, the imaginary axis by Y, and an arbitrary complex number % by the number pair (x, y). Since a number pair (x, y) uniquely specifies a vector in a plane, a complex numb~r may also be interpreted as a vector in a plane. Thus a complex number z = (x + iy), an ordered pair of real Rumbers (x, y), a point in a plane, a vector in a plane are ;]H equivalent. We may thus alternately denote a complex number by % = (x, y) (14.11)
* Inst\!ad of introducing the imaginary unit i we may define a cornpk'x numo:r a~ an ordered pair of real numbers and develop the algebra of complex numbers a~ the alg\!bra of ordered pairs of real numOer~. See for instance, Konrad Knopp (!952), t The names Argand plane and Gaussian number plane are also used.
==
1
(14.5)
and define a complex number as any combination of the form
a
+ ib
where a and b are real numbers. A complex numblW thus represents a pair of real numbers We denote a complex number by a single letter, say z, and wflte
z=,a+ib
(14.6)
The real numbers a and b are known. respectively, as the real and imagiMry parts of a co~plex numbe. z and are denoted by
a = Rez
(14.7)
b= Imz
The c~mpJex number II  ib is ealled the .conjugate of the number z = a + ib and is denoted' by the symbol The algebraic operations of addition, multiplication, and division for complex numbers are defined in the same manner as for real numbers. Thus, recalling (14.5), we have Addition and subtraction
z.
(a
+ ib) ± (c + ld) == (0 ± c) + i(b ± tf) + ib)(c + id) == (ae
 he/)
n4.8)
(14.9)
Multiplication (a
+ i(ad +
be)
IdealFluid Ac:rodyuamic:s'
Ek:menta oftbe Theory. of FunctiODS .of a Complex Variable
. The magnitude of the vector that repu;lICOls a· complex number % is known as the JIlodulus magnitude or a~solute value of the number .:: and is denoted by 1%1. Since %is the number z + iy, ,ve hue
0,
1%1
=!'
../zi + yI
(14.12)
The angle between the real axis and the vectof% is known as the argument of %and denoted by • (14.13)
. The al~br~ o~rations on complex numbers ha.'C simple geometrical lDterpretations. It IS~ seen that ~~ty and addition compJex numbers. corresponds to equality and addition of v~ors in a plane. Multiplication of complex numbers may also be pen geometric meaning. but this we shall do later. Just ~ for vl=Ctors. the question of whether one complex number is greater or less than another does not arise. All we can say is that the complex numbers are either equal or not equal. Of course, we tan always compare the magnitudes of tile cumplex numbers.
or
r
1nwaNr) ail
14.4 Polar and EIpoaeodal Forms of a Complex Number
In polar coordi~ates, a point or ~ position vector in the plane is indicated by the number pair (r, 6), where r IS the length of the position vector while
y
I o+i6
.(.6)
I
I
16
z. ~r, I)
I
I
r
~~~~I~X
1
I
I
1
:I""(r,I)
Fta·14.2 Polar fonns of a complex number and Its conjugate. Complex plane.
ogives it!: direction (see Fig. 14.~. Thus in polar coordinates a complex nUr:Jbet z may be represented by the number pair (r, 0) and denoted as
We
z = (r, 0)
This 'sbowsthai the argument of any given compleT. number is' only
determined to within an added multiple. of 21T. In o.ther words, the arg % is multivalued, with any two su~ive values differing by 21T. The value of the argument corresponding to the range 0 and 21T (or 1T and 1T) is usuaUy known as its principal value. The conjugate i == z  iy has the following properties (see Fig( 14.1) then have
(14.16)
Izl
= r
(1.4.17)
(14.18)
arg z = 0
Iii == Izl
and arg %~ .. .,..arg %
(14.14) (14.15)
Tit"" ~ is the reftection' of the point % in the real axis (see Fig. 14.1).
• Sometimes also known u the amplitude of z.
For any, given z: r h~s a definite single positive value, while 0, as pointed out earher, h,as infinItely many values that differ by multiples of 21T. If ~e conSider the complex number z ill its Cartesian form :r + iy and. substitute x = r cos 8, y = r SID () we obtain Z :': r( co<; f) + i sin 0) (14.19)
IdealFluid Aerodynamics This is known as the polar form of the complex number z. The expression (cos 0 + i sin 0) is a complex number whose magnitude is unity and lrgument is O. It is simply a unit vector in the direction of the complex number z.* Differentiating the expression (cos 0 + i sin 0) with respect to 0, we obtain d( cos 0 + i sin 0) == i dO cos 0 + i sin 0 This integrates to i (J4.20) (cos 0 + i sin 0) == e ' This relation is generally referred to as Euler's formula. Using Eqs. (14.19) and (14.20) we may express a complex number z as
(14.21)
. EI:ments of the Theory of Functions of a Complex Variable
y
109
.c~~~~~~x
Fla.
1~.3
Illustrating the product of complex numbers
This is known as the exponential form of a complex number. Putting together the several forms of z we hav~ z = rei' and
14.5 Function of a Complex Variable
( 14.22) (14.23) (14.24)
==
r(cos 0
+ i sin 0) == x + iy + y2
Y x
1
Izl == r = J Xl
arg Z'= O = tan .
if for each value of a complex variable z a new complex number' is generated by some rule, we say tl1at is a function of the compleX variable z and write , = ,(z)
t
All the different forms of the complex number are found useful in applications. The form that is more suitable depends entirely on the problem on hand. Using the exponential form, we see that the product of two complex numbers %1 = r1ei • and Z2 = r2ei ' . is given by
%lZ2
= r1 r.e i (8,+h)
}
(Other notations such as ,. = f(z) or F(z) may also be employed.) Thus the notion of a complex function is exactly the same as that of ~ realfunctiori. If a single value of , corresponds to each value of z, we say' = '(z) is a singlemlued function; if more than one value of , is produced for each z, we say {(z) is a multiplevalued function. Expressing the complex numbers' and z in terms ·of their (eal and imaginary parts , = (~, 1/) = (~ + i1/)
z = (x, y)
that is, and
1%1%21 = r1r2 = IZ11 1%21
(14.25)
.
arg 1%1Z21
= 01 T
O 2
= arg Z1 + arg %2
We notice that multiplication of any complex number z by a number of the !urm ei'l', where (p" is real, is equivalent to rotating the vector representing %. through an angle rp in the complex plane (Fig. 14.3). This lends gl..omel ;al meaning to the product of complex numbers. Using the exponential form we obtain the result that (14.26) (cos 0 i sin o)n = (cos nO + i sin nO)
we interpret the function ,(z) as equivalent to defining two real numbers (~, 1/) for each value of a pair of rral variables (x, y). 'We thus
~~
.
,=
== (x + iy)
~ ~ ~(x, y)
1/ = 1/(x, y) and
,(z) =
~(x,
+
y)
+ i"1(x, y)
This is known as De Moivre's theorem.
• Recall that any veak>r may be represented as a magnitude times a unit vector.
where ;(x, y) and t](r;y) are called the real and imaginary parts of the complex function ,. In this form the complex function' = ,(z) is simply a combination of a pair of real functions of two real variables. We will
41.
IdealFluid AerodyDamic:s
Elements of the Theory of
Functio~
of a Complex Variable
111
find this in.terprctation of a compl~x funcqon sometimes useful. Generally, ho',vever, important information ,~ults ~y using theunseparated form where a function {(z) is exhibited in terms of the single comolex variable z.
1".6 Aaalytic F1mctioa
Of all the functions of a C<Jmplex variable, the ones. that are of interest
Points of the complex domain z, where the function ,(z) is analyt.it, ani called regular poifJ/S of the function. Points where the function is not analytic are called singular point.! of the function. The rules fOl differentiation of analytic functions are formally the same as for real functions. The proof may be carried out in exact analo&jl. Thus we have the following simple results:
in our applications (ar. in alL physical sciences) are thoSe that are differentiable. Such functions are called analytic functionS. For a compl~x function the concep~ of calculus, such as dffferentiability, continuity, alld limit, are f maUy the same as for a function of a,real variable. Differentiability of a CQIIlplex function ~plie3 its continuity and guarantees, as shall be seen late•. that the flmction may be repeatedly diffe~ntiated or integrated any number of times. We shall concern ourselvt:S with the details ofJonly the condition. of differentiability: A complex function' = ,(z) is sai4~o bedifferentia\>le·at a point z if as %1 approacltes z the limit
d%
~ (f(%) + g(%)] == 1'(%) + g'(%)
~[f(%)g(%)J = f'(z)g(%) + f(z)g'(%)
if g(z) yI: O. If ~(z) is an analytic functiCJn, and if wW is another analytic function, the function ~en by g(z) = "1'(z)] is arso analytic, for we have dg{;:) dw d, =dz d{ dr Thus an analytic functio., of an analytic jultCtion is analytic. From these various rule .. one may be able to obtain a wide class (If analytic funClio..as from a few basic analytic functions., Later we shall consider some examples of :o:uch functions.
~[f(z)l = 1'(1.) g(%)  fez) ,'(z) dz g(z~ g(zt
lim {(zJ  {(%) '1'" Zt  % exists. If it is possibleito construct this limit, we ~ it the derivative of the function {(z) at the point z and denote it by d'itk.or {'(z). Thus we writo
d{ ii5 {'(z)
= lim {(Zt) ~ {(z)
'1'"
dz·
Zt
~
z
or, putti:lg {(Zt)  ,(z) '"'" ~t and,zl  ~  ~,
a(,
= lim ~{
,u+oAz
(14.28)
14.7 Cau<:byRiemana Conditions
The cond.ition for the differentiability of a (;omp}ex function implit! important conditions on its real and imaginarj parts. Since a complex function ,(z) = ~ + iTJ is no more than a pair of real functions, ~(:r, y) and TJ(;r, y), it i<; natural to suppose that tht complex functior. is differentiable if ~ and 7] are differentiable. This. however, is not true. The differentiability of ~(x, y) and 1](r.,) does nC!t in itself ilnply the differentiability of the function {(z) = ~ + itl. For example, as ~an be readily verified. the simple ftlnction ,(z) = 2x + 6yi is not differentiable although ~ and 1] are. Besides differentJaL:ity, the real and imaginary parts of an ar.alytic function mllst fulfill other important conuitions, which we shaH now obtain. Consider the function
~(::;)
dz
Thisidea of differentiability and derivative fOT a complex function il' the same as that for a function of a real variable. There is, however, an important distinction that we should bear in mind. Since %1 and are complex numbers or, equivaiently, points of a plane, there are infinitely many directions or paths along which ~, that is, the point ~" mavlbe chosen. Under such circumstances we should state that for a complex function { = ,(z) to be dtJferentiable at a point z, the limit: lim ~"6.z, or, eq:.tiv
z
AJ;o
alently, the derivative d'ldz should assume the s4me value no matter from what direction ~ approaches zero. This, then, is precisely the cont/.ition for a complex function to be analytiC. Such a requirement does not arise explicitly in case of a function of a real variable, for all values of a real variable lie on a single straight line. •
• For • more rigorous discussion consult· books on the theory of functions or a
=
~(x. y)
Writing
~z = ~x :::..~
+ 11)(X. y) + I:::"y
tI _\/]
oomplex variable.
=
I~~
Elements of the Theory of Functions of a Complex Variable
413
ldealfluid.. Aerpdy.wunics
we express the. derivative.·of
,.as
lim
(d.c+;411l ~o
Ii' =
dz
4,~.O Az
Jim ~C ""
tlE
Ax. + 'i li~l
+ i ~f}
Of the infinitely many directions along. which ~z = ~.:+.i Ay may be taken, we choose specifically the two directiOMcorresponciing to Azt = !l:e and LUI = {~y, that is, parallel to the x and yaxes, respectively (Fig. 14.4). For the derivative along Az1. we have
(14,30)
y
Now, it is natural to expect that similar additional conditions might be to establish that the derivative d~/d: will actually assume a single definite value irrespectiVt: of the direction from which ~z approaches zero. This. however. is not the case, for it can be shown that the CauchyRiemann equations are suJjicient to ensure this independence of the derivative. We thus conclude that the real and imaginary parts of an analytic function are a
r~quired
pair of real junctions that, besides "eing differentiable, satisfy the CauchyRiemann conditions. 14.8 Some Consequences of CauchyRiemann Equations
We shall now consider some significant results that follow directly from the CauchyRiemann conditions. 1. If the real and imaginary parts of a function of a complex variable satisfy the CaUChyRiemann equations, the co'mplex function is an analytic function. 2. From Eqs. (14.32) we obtain
or
(14.33)
S\milarly, we have
\,21)
Fig. 14.4 IlIustr:ating the computation of the derivative of a complex function.
= 2 = +
ox oy2
0217
02 17
0
(14.34)
For the derivative along LU, we have
(~).= lim A~ + ,dz
A.~O
i
jAy
~f} == ~ oe + ~
i
oy
oy
(14.31)·
It follows, therefore, thin jf the derivative of , has to assume the same value from both these directions, we should require that
~=
dz
oe + i ~~ ~ + ! oe ox o~ 01/· i Oy
O~
ThiS ,hows that both the real and imagInary parts of an analytic function satl~fy Laplllcc's equation. They both are thus harmonic functions. J. The real and Imaginary parts are, howe\er, not independe·nt harmonic functIOns. For, If the real part ;(x, y) is specified, the CauchyRiemann equations determine the imaginary part I,(X. y) ·to within an arbitrary addiu\e Clln,tant. In this,ense the real and imaginary parts of an analytic functi,'n 3re \;lId to be conjugate h;Jrmonic junctions. 4. C,llNdcr the tWll famIlies of cunes in the xy plane described by the t'ulictioilS E(x, y) = canst.
I/(X,
ThisequatiOQ redUces to the following pair of partial differential equations. which are known as Ca.uchyRiemann conditions;
Y)
=
(llnst.
Ox
o~
=
of}
iJy
~
where': 3nd " are the p:al and imaginar~ rart~ of an allah tiC functIOn. The angle of Inlcr,c<:lIon hetween the tV.,l families I'; given b:,
(1432)
.Oy
=  
ax
114
ldcalAuid Ae:rodyiwnica
Elements· of .the Theory of FunCtions of • Complex Viuiab1e
115
where
However, by the CauchyRiemann conditions this dot product is. zero. This meansth~ two families of c~es intersect each otheret rightangles. We therefore state that t~ two families. of cUrves defined by the real and imaginary parts of an analytic function are orthog01lQ1 to each other. 5. If the function , := {(z) is analytic and if the derivative dJ' "ii!!! '.2 ~.O
M.19 Some Aaalytk FIII!CtIoas We now consider some exampl~s of analytic functions. Such functions will be used repCat~ly in our later considerations. One of the simplest functions is .a power function z", where n is a positi~ integer. It can be readily seen that it is analytic at all points of the complex plane. It follows, at once, that a s.ocalled polynomial function ao + a 1z + asz' + ... + a"z", wher~ n is a positive integer and the a's' are constants (complex, in general),. is also analytic in the enpre comple~ plane. We further conckde that a rational function
00
"
dz
b o + b1z
+ a z + or' + ... + 0,,2(" + b r t + ... '+ l"z"
1
then the inverse function
z= .Z<O
exists and is analytic with its derivative given by
, dz
is analytic at all points where the deriominator does not vanish. From these considerations we may conclude that a power series
I
«>
a"z"  a.
+ a z + art + ... + a.z" + · ..
1
It
1
{'
1
d{fdz
z
5     
d{
1 hi~ result is of vital significance in the theury and :lpplication of' conformal transformation. The proof of the r~\.llt is left as a probJem for tne r~ der.
14.9 R.emarks
It is appropriate at this stage to recognize the intimate connection between the analytic functions of :l complex variable a:1d the solutions for the problem of tw().(iimensional irrotational motion of an ideal fluid. We realiz: that the.prope,ti«! exhibited by the real and imaginary parts of an analytic function are identical. with the properties exhibited by the veloc;ty potential and st,eam functions of a twodimensional irrotational motion ofll/I ideal fluid and vice versa. The problem of such a motion, consequently reduce~ to that of finding an analytic function whose r~land imaginary parts satisfy certain prescribed ~oundary conditions. Consider now the qu(.Stign of relating the solution for the tw~men sior.a1 potential motion for one geometry with thet for another geometry. For this purpose, as we have seen, we must introduce coordinate transformations that obey certain rules (see Section 13.10). These rules, as we now realize, are identical with those governing th~ real and imaginary parts of an analytic fUilction of a complex variable. We may thus conclude that the theory of twodimensional poteRtial motio!) and the theory of an analytic [unction of a complex variable are id:ntical.
where 11 is an integer anc the a's are constants is an analytic function within ·the rt.gion of convergence of the series.· Conversely, it is possible, as we will indicate later, to express an analytic functionwithin the r~gion it is analyticin terms of a power series. This procedure is very helpful in discussi!1g the theory and app:ications of complex functions. . Now, the exponential function e ' , the trigonometric functions sin z and cos z, and. the hyperbolic functions sinh z and cosh z can all be defined, just as in the case of the corresponding real functions, by means of power series that are convergent :verywhete. Thus all these funt;tions are analytic in the entire complex plane. Another important function is the logarithmic function log %. Let)Js write wnere , = z and 8  arg z. As pointed oct earlier, although r is single valued, 9 has infinitely many values that differ by rr.u\tipl~s of 27T. To show 9 == 8'. + 27Tk this explicitly we set where k :::i: 0, ± I, ± 2, ... , and 9 p is the principal value of arg z, that is, the value of 0 in the range 0 to 27T, We then have
z = re;('·+lrt)
(14.35) log % = log r + :(0. + 27Tk) This shows that the logarithm has infin;tely many values that differ by multiFles of 27Ti. Thus log z is a multivalued function. The various
• For the convergence of series. anti so forth, refere'lce may be maGe to Knct'P's or any other suitable book.
and
416
IdealFluid Aerodynamics
\:i1uo:s of such a funcllon are called brunche~" The princioal value brancn of log: IS (log r + lJp)' Lo:t the pnint ~ descrlhe a closed curve f~om an initial point :0 in the :·plant: (Fig 14,5), C)n'iI(Jcnng tho: curve C I , which docs not enclose the
Eleinents .of the Theoty ,of Functions of a Complex Variable
417
y
~
I
corresponding value of , may be represented by a point in the complex plane of the variable', Henceforth we shall refer to these planes respectively as the z and "planes. We thus see that geometrically the function ,(z) maps or transforms points of the 'lplane into points of the 'plane. In this sense a' function ,('l) may be considered as a transformation or a mapping function. It transforms curves and regions in the 'lplane into curves and regions in the 'plane. Generally it is not convenient to represent the two complex planes of 'l and Csuperposed on each other. Therefore to represent transformations
y
/
/
" e
.plane
(plane
CurveC.
otz
~
Fig. )·1.5
lIIustr;.ting the multivaluedness of log :,
oZ
p,'rrlt: = 0, we observe that log:: returns to its initial value as:: returns to its initiai point on the cune, Considering the cune Co which does enclose the point =, = O. w,e see that log z chang~s by, ± '2"Ti aecording to whether the cunc IS deSCribed In the positive or negative sense of O. We thus conclude that to go from one branch to another 01 log =, we mu"l make a circuit cnclo~rng the pornt : = 0; otherwise we remai~ in the same hranch of the functIOn, The poi;lt : = 0 is called a branch point, For any particular branch llf lvg:, the derivati\e
L~x
F18. 14.6 Notation for mapping,
or mappings we utilize separate representations of the two planes (see, for ex.ample, Fig. 14.6). Writing 'l = x + iy and, = ~ + i1], the points in the 'l and 'planes a.e given by the number pairs (x, y) and (~, 1]), respectively. The meaning of the function '('l) as a transformation is also clear from the point of view of its real and imaginary parts; which are given by
~
1]
. ,I I 1  ')g: = if: ::
= ~(x, y)
= 1](x, y)
i; defined for all f'nlnrs except the point:: = O. Thus each branch of the logarithmic functIOn IS anal:'.lc at all pornh except: = 0
I 4.11 Geometrical Significance of a Complex Function: :'Iiotion of Mapping
This pair of real functions produce a pair of real numbers (~, 1/) for every given pair of real numbers (x, y).
14.12 Some Simple Transformations
We now consider the overall features of some simple transformations. 1. Consider the transformation
ConSider thc funcrion ~ ,= ~(~). According to thh function for c\erv given value l)f :; a \;tlue of ~ is generated. r~ach \;.tlue of:: m'a: be reprcs~nted hy d pornt In the complex pldne of th~ variable ~ and each
• :'ote that although ar.£ ' i~ rnultivalucd. : klg: arl,cs out of the nluillpir<:ily "f arg :,
ihdl "
""ek \ .ilu.:J
•
The !l1Ulllr il "t' of
where b is a complex constant. This represents a simple translation of the entire 'lplane by the vector b (see Fig. 14.7), It follows that under this transformation. the shape of any geometrical figure remains unchanged,
411
t, "
IdealFluid Aerodynamics
Elc:menta of the Theory of Functions of a Complex Variable
419
)
           .. x. ~
Fig. 14.7 Mapping gi~ rise to translatiOl\.
It is thus seen that the vector zis strerch~c1 by the factor a == IAI and then rotated through the anglefJ =:= arg.A. (see Fig. 14:8). Thus the transformation { ::I Az produces a rotation and magnification. Under such a transfoonation the shape (not the size and orientation) of any geometrical figure remains WlChanged. The potnts 00 and 0 are both fixed ~ints of the mapping. 'the'transformatIOn { == oz, where a is real, denotes a pure stretchin!, whereas the transformation { == zeil denotes a pure rotation by the angle
fJ·
3. It follows that the transformation
{('I.) = Az
+b
The mapping' =z is known as identity transformatjon, for it l;;aves every point unchanged. t:le points that map into themselves are called the fixed points of a transformation. 2. Consider next the transformation
, :::II
,4z
where A and b. !tro complex constants, represellts a translation,a rotation, and a magnification. Under such a transformation geometrical figures rem.ain similar. 4. Consider next the transformation 1 {('I.) = : 'I.
where A, in general, may be a complex constant. To discuss the properties of this transformatioll it is convenient to usc the polar form of a complex number. Let us therefore write '[ == rei'
A
Tht function 11'1. is well defined except for the point z == 0 for which it becomes infinite. If we agree to speak of a point at infinity, the function 1/1. may be considered defined throughout the zplane. We use polar
coordinates as before and write the transfcrmation a5
== aeil
We thus have
We then have
Of
I" == a Izl arg { == (J + fJ
y."
and
•
arg ~
alS(l
1[1 1
,
1
:::II
1'1.1
r
==
argz == Ii
The transformation may
be exhibited as
, == (; eif )
e 1i'
== Z1 ~2i'
a
where '1.1 is a point that lies on the same line as '[ and has a magni!ude given by the relation
=o._~_,
_ _ _ ' _ _. _ _ _ _ _
x,
~
Izl=1 Izl
The point
Z1 IS
1
FiR. 14.8
Mapping gIving nse to rotl/tion and magnificati:)n.
said to be th¢ inversion of z with respect to the unit circle
idealFluId Aerodynamics
i'
Elements of the Theory of Functions of a Complex Variable
411
1
d~/dz, and c5z are all complex numbers. To proceed, we assume that at the point z the derivative d~/dz does not vanish:
I
d~ ;c 0
dz We shall later consider the case where d~/dz does vanish. From (14.36) we obtain
~~r+X
Ib~1 = I(~~)J Ibzf
arg
(14.37)
and
t'jg. 14.9
Mapping glvmg rise to inversion and reflection
( 14,38) ,dz , This shows that an elementbz through a point z under the transformation ~ = ~(z) has its magnitude magnified by the factor IdGdz\ and has its
arg
b~ =
(d~) + arg b.:
(se.e Fig. 14.9). The vector ~ is obtained from ZI by reflection in the real aXIs. Th~s t.he transformation ~ = liz denotes an inversion with respect to the UnIt CIrcle and then a reflection in the real axis. As may be verified. generally the shape of geometrical figures is not preserved bv this transformation.
14.13 Conformal Transformation: Transformation
by Analytic FUDctions We shall now go into the rlature of the transformation brought about by an analytic function. For this purpose we study the local changes in a transformation. Let ~ = ~(i) be an analytic function. With this as a transformation points and curves in the zplane are mapped into points and curves in th~ ~plane. Let ~ be the image point in the ~plane of the point z in .. he zplane. Also let curve C, be the image in the ~plane of curve C, in the z· plane (see FIg. 14.10). The direction along C; corresponds to the direction along C,. At z. let tJz be an infinitesimal element of the curve C drawn in the direction Of. the curve. Then. at ~, the infinitesimal elemen; b~ of the curve C, IS the Image of bz The elements c5~ and bz are' then related by
b~
=
d~(z) bz
dz
(14.36)
argument increased by the amount arg (d~/dz), where the derivative d~/d: is evaluated at the point z. In this sense d~/dz fS the local scale factor (complex naturally) of the transformation. Since ~(z) is analytic, dCd: is independent of the direction bz. Therefore all infinitesimal elements passing through the point z are scaled hy the same factor d~/dz. Now. consider another curve C:' passing through the same point: and let C,' be its image passing through the point Let:x be the angie of intersection between the curves C: and C; measured from C, to C;'. and let {3 be the angle of intersection between the curves C, and C;' measured from to (see Fig. 14.10). Denote by (bz), an dement of curve C; through z and by (b~)' its image. We then have the following relations
r
C,
C/
where d'idz is the derivative of ,(z) at the point z.· The qu::.ntities t5~.
• Recall that ,(z) is analytic. and therefore d'idz .curve C •.
doe~
flat depend on the direction of
(l.U9)
121
Elements of ~ Theory
pC
Functions of a Complex Variable
12J
and arg (t5CY = arg
(~:) + arg (6%)'
(14.40)
the critical point of the transformation. Consider first the exampl~ of .a transformation C... z", where 71 is a positive integer. The functIon IS analytic in the entire z.pJane with the derivative
Subtracting (14.38) frolii (14.40) we obtain
8.(g (~n

arg (~ = arg (~z)'  arg (6%»)
(14.41)
(' = d( _
dz
nz1
or
P=at
We have thus proved that under the transfor'nUltion , == C(z), which is analytic, the angle bet wee,. arry two curves passing through any point z at which dC/dz ~ 0 is preserved and also its sense remains unchanged. Such a trarisformatiQo is known as ~ ,:onformal transformation. Furthermore, all infinitesimal elements of length passing through z receive the same magnification of their magnitudes, in fact, the (compl:.x) scale factor of the mapping is the same for all the clements passing thwuglt a given point. Summing up the preceding considerations we may state that the mapping affected by an analytic function is conformal ih a neighborhood of every point at which the derivative of the mapping function does not vanish. Elemental figures at any poi~t z go into elemental figures at the image pomt C Since the transformation is conformal, the figures remain similar. •. Another way of saying it is that when regi0Ds are transformed by an analytic functiofi, geometrical Mmilarity is preserved for the minute structure of the regions. The scale of mapping at any point is giveq by d{(z)/dz, a complex number, and gerlerally this sc'lle factor changes [rom point to point of the zplane. Because of this the regions involved in a transformation may not look alike on the aggregate although their elemental structures are simil~r. In concluding this section we point out the significance of the requirement that the transformation be an analytic function. If it is not an analytic function, the deriv~tive d{/dz may have different values in different directions. In such ..1 case the angle between any two curves passing through a point will not be preserved, that is, the mapping will no long~r be conformal. In ligh~ of these considerations, one could say that the definition of an analytic function could have been based on the requirement that a transformation be conformal. Recalling our inquiry with regard to a transformation th!lt would relate a potential problem with respect to one geometry with that of another geometry, we now see that what we are after is a confortnal transformation.
This is zero at the origin; wh~rcas everywhere else it is different from zero. Therefore the transformation is conf9rmaJ everywhere except perhllps for the one point z  0, which is a critical point.
y
r.
I
"
~~+~x
\
FJa. 14.11 Illustrating the nonconfonnal nature at a critical point of a trusConnation.
Writing z
== rei'
and (
==
pel,., we express the transformation as pel. = r "eitll/
or
p.!:: r"
rp == nO
14.14 Critical Point of • TransfonnatioD
We now consider the natur:: of the transformation at a point where the derivative of the mapping function is zero. Such a point is referred to as
From this we conclude that circles about the 0'Cigin in. the zplane (i.e., curves defined by r _ Izl  const.) transform into .ci.rcl~s about the origin in the Cplane and rays emanating from the o.ngm 10 the z.pl~e (i.e., curves defined by 0 = arg z =const.) transform mto !i!ys emanat~ng from the origin in the {plan·e. See Fig. 14.11, where the lransformatlo.n , .... z. is used. If 0 == 01 and 0,= 0, are any two rays in the z.plane, theIr images in the {.plane are given by rp = nOt and rp == nO.. Therefore t~e angle between the image rays is n times the angle between the rays 10 the z.plane. In other words, the mappin! C  z" is not conformal at
IdealFluid Aer~ynamics
Elements of the Theory of Functions of a Complex Variable
415
the CrItical poir.t t = 0 At thIs point the transformation has the property 01" multiplying angles by n.
If. Instead of the transformation ~ = z" we consider the transformation (z  :;0)" "'e com;lude. cn the same lines as above, that the transformatlDn is conformal everywhere except at the critical point z = zo, where it has the proreny of multiplying angles by n. Now let us consider a transformation in the general form ~ = ,(z) and let us assume that the point z = '::0 is a critic~: point. We wish to know the ndture of the transformation at [hi~ point. The function ~(z) is analytic and tile point ::Q anl' it, neighhorr.ood are regular points. Let ~o be the image of Zoo For pOints ~ in the neighborhood of ~o, using Taylor's expansion. we may write
~(:) =
Let us apply this result to the transformation , = z". At the critical point z = 0, not only the first derivati~e of, is zero but all the first n  I derivatives are also zero. This means the transformation , = z.. is nonconformal at z => 0, where it multiplies jill angles by the factor n.
14.15 Complex Integrals
We now pass or. to the integration of complex functions. In the next sections we shall develop several important results relating to complex integrals. As we shall set;, many of the considerations underlying these
y
~(:) = sVo)
+ a 1(z


2 0)
+ 02(Z

ZO)2
+ an _I(Z
ZO)"I
Ok
+ l',,(z
 zo)"
+ ... + aJo+I(z

zo)"+1
+ ."
(14.42)
where the wefficients
are given by
k
a =
k! CZk '~'o
L~x
1 'd~)
Since 'Zo is a critical point, d~/dz is zero at that point. Consequently, the coefficient (/1 IS Lero. Not knowing the function ~(z) explicitly, we have no kn,wo'1cdg? of its higher derivatives at zoo Let us suppose that the first ('1  I) .ieriratires are ::ero otthe point Z = zoo Then the expansion (14.42) reduces to
( = '0
Fla. 14.11 Defining a line integral.
+ a"(z
 Zo)"[l
+ an+! (z a"
zo)
+ a,,+2 (z
a"
 ZO)2
+ ...
J
This may further he rewritten as (14.43) whl:re fez) is the function given by an times the expression in the square bra,kets. We, thus ~ec that the transformation ~(z) 'lear a critical point ...wy be ·represented ill the form of the equation (14.43). We observe [h'll the derivative ilfl,]., at the pomt z = Zo does not vanish. In other words, the point Zu is not a critical poir.t of the iunctionf(:.:) and this function is a conformal transformation at th.:: point Zo. With this understanding we realize that the PO:lIl z = Zo is a critical pOint of the :msfornlation ~ = (z) only through the function (z  'Zo)n. From tne previous discussion about the propertie'\ of the transformation (i  zo)", we conclude th;: following.
( (0 = (oz  Zo)}'(z)
,=
results are similar to those introduced in Sections 9.13 and 9.14, which deal with circulation and its relation to the convectivity of space. The notions of simply and multiply .connected regions and of reducible and irreducible circuits will again be employed. Consider that a complex. function ~ = ,(z); not necessarily analytic, is defined in a certain region of the complex plane z. In that region let C be any path (or curve) joining two arbitrary pointsPo andP 1 (see Fig. 14.12). The complex numbers denoting Po and Pi are Zo and Z10 respectively. The expression
is known as the (line) integral of ,(z) from. Zo to ZI along the path C. The
actual value of this integral, in general, depends on the path C and the endpoints zoand %1. The foregoing definite mtegral of a function of complex variable z = x ;y may be expressed in terms ofline integrals of the real and imaginary parts of the function. For this purpose we set
The trm:sformatlOn ~ = ~(z) is not conformal at a critical point. At such a pOil1l ii i/(J, 'he property of,null f;lying angles by n. The factor n is the order {tJP dail'alhe d".~!dzn. whicf: first becomes non::ero at the critical paint.
+
or
'(z) = ~(x, y)
+ iTJ(x, y)
116
and obtain
IdealFluid Aerodynamics
Elemenb of the Theory of Function,s of a Complex Variable
IJ7
and
r
Z ',W'(7}
dx
+ edy)
) • • •0
which constitute the real and imaginary parts of the complex integral where (xo. Yo) and (X1o YI) are. respectively, the points Zo and ZI and the integrations are performed along the given curve C. This equation is the basis of further discussion. 14.16 The Cauchy Integral Theorem We are most interested in complex integrale; whose values depend only on the endpoints and not on the particular path that joins them. To investigate such integrals we .think of the complex in·tegrals in terms of the line integrals of real ~nd imaginary parts (as expressed by 14.44) and ask for the conditions under whic~ the values of these integrals depend only on the endpoints and not on the path that joins them. From calculus· we recall the important theorem: If a region is simply connected and in that reKion u(x, y), v(x, y) are differentiable functions, then the line integral
L"(Z) dz
For II. simply connected region and differentiable and 7}, these integrals will be independent of the path if the following conditions are satisfied:
e
07] =
ae
oy
ox
oe '07] ::iJx oy
But these are precisely the CauchyRiemann conditions. If these are satisfied, the function '(z) is analytic. We thus arrive at the important theorem: In a simply connected region the integral
i
P"Z"W')
[u(x, y) dx
+ v(x, y) dy1
if and only if
Polzo.w.)
1:"(Z) dz
is independent of the path of integration if ,(z) is analytic in that region. In such a case ,(z) dz will be given an :xact differential of some function, say F(z). We will then have
will be Independent of the path jo;ning Po and PI
"=
ou ov oy ox
This condition assures us that the differentialu dz + v dy is an exact differential, say of the functiontfo(x. 1/). We then have
J..
fll {(z) dz = f"' d[F(z)1 =
••
F(zl)  F(2:0)
This bJ itself does not imply that ,p{PI)  cM.Po) is the same no matter what path joins the endpoints Po and PI' For cM.P1 )  cM.Po) to be independent of the path" we must require that #..x, y) be single valued., The singlevaluedness of 4> is assured by the requirement that the region under consideration be simply connected. We now apply the above theorem to the line integrals
Consider now the integration of a complex function ,(z) around a circuit (i.e., a closed curve) C (Fig. 14.13). For t'lie direction of integration along a closed curve we adopt the convention that the positive sense is that for which the region enclosed by the curve lies on the left. The integral of {(z) around C is then denoted by
fc
,(z) dz
Let Po = Zo and PI = ZI be any two arbitrary points on the circuit and lC':t C1 and C1 denote its two parts that connect %0 to %1 (see Fig. 14.13). Positive orientation of curves C1 and C. is from %0 to %1' We then have
~~ {(z) dz
• See for instance Courant ,(1934).
==..
f. ",
along
C1
+..
J"
along
C.
418
IdealFluid
Aer~ynamic:a
Elcmcnh of the Theory of Fun.:tior.~ of a ComF~c:. Variable
If the region is simply connected and {(z) is analytic, it follows from the
previous theorem that
or
J'1
•0
al~ng
C1
=f."1
'0
along
C.
==
0
f.>o
_I
along
C•
fa {(z) dz /
Fig. 14.14 functIOn.
Multiply connected region resulting from n0l1dn:tlyticicy of a comple",
L~~~x
Fla. 14.13 Illustrating the line integral around a circuit.
We thus arrive at the famous Cauchy Integral theorem: in a simply connected region, then
If {(z) is analytic'
(14.45)
fc {(z) dz = 0
14.17 Integration in Mnltlply CODJJea ~ Regions
the region ol analyticity of the function /(ZI, we ~Llr~()und the singular poin,s by closed curves ('I ... en and exdude from cOn5ldera~IOn the regions included by these curves. Thi, artifice makes the regIOn ot tntere:;t multiply connected. We consider for simplicity a doubly ccnnected region such as that extericr to a closed curvc 1'6' {see Fig. J 4.15) and a functIOn Sl:) that is analytic in that region. Con!>ider first a redudb/c circuit such as (', (see ngu;e). The region enclosed by C 1 is simply connected and by the Cauchy
a
for ail dosed curves C in that region. The contents of this and the previous theorem are equivalent.·
On the basis of the Cauchy theorem several significant results may be derived for complex integrals in multiply connected regions. Such regions may arise, as we know, because of physical considerations in a problem, or they may arise because of our desire to exclude from a region of consideration points at which a complex .function is singular. For instance, suppose 'that in a certain region bounded by the closed curve C the function f(z) is analytic everywhere except at certain points ofthe region, say at the points z" z, •.. Zn (see 14.14). Ifwe are to concern ourselves only with
Fig.
• The Cauchy theorem may be pro~ for less restrictive conditions than assumed here. The interested reader should refer to the works cited at the end of this book.
Fig. 14.15 lliustrating integration in a doubly connected reg'<Jn.
430
IdealFluid Aerodyilamica
Elements of the Theory of Functions of a. Corttplex .Variable
4,31
theorem we obtain
1.. {(z) dz _ kl
1.
{(z) dz
0
Consider next an irreducible circuit such as Ca. Now we cannot use the Cauchy theorem and assert that the integral
k.
vanishes. It mayor may not be zero. Let C. be another irreducible circui' but re~oncilab/~ with C. (see figure). The region bounded by the curves C. and C. is not simply connected but is, in fact. doubly connected. wi render this region simply connected by introducing the barrier such as ab from a point a on C. to a point b on C. (see figure). We then apply the Cauchy theorem to the integral over the new boundary curve of this region. The new boundary is the closed curve C. + ab  C. + ba. The positive sense of the curVe is determined according to the convention· that the enclosed region should lie on the left. We then obtain
c'
PII. 14.16 mustrating integration in a multiply cOnneCted R£ion.
out in the samt direction. We write
£. '~)dz t.

{(z)dzO
or
1..
k.
{(z) dz ==
1..
k.
{(z) de
(14.46)
The value of the integralovcr C is the .same as over any other irreducible cUrve thitt encloses all· the curves C. '... C.. and lies in the region under consideration:
These results for an analytic fimction {(z) in a doubly comtected region may be summed up as follows: I. The integral t {(.%.) tk over any reducible circuit is zero. 2. The integral over any irreducible circuit may not vnnish ,i1l general. At this stage its value cannot .be evaluated. 3. The integral has the same value for all irreducible circuits. These results may be readily extended to multiply connected region~. Con"ider anply connected region tltatis bounded by the curves C, CI , C•• . . . • Cn. where the interiors of CI .... C .. are contained in the interior of C (see Fig. 14.16). This region may be rendered simply connected by introducing nonintersecting barriers as shown by dotted lines in the figure. If { = C(z) is analytic in that region, wt obtain the following results: 1. The val~e of the integral
3. The value of the integral' over any tworcconcilable but 'irreducible circuits ~1 and ~I; both of which enclose only the curves C1 ••.• 'to C~ and lie in the region under consideration, is the same and is equal to tht sum of the integrals over the curves C1 to Ca.:
1. _1.. .. "+1.. + ... 1.. hi ~. kl ·k. ka
14.18 Some Simple Iutegrals
(14.47)
f
{(z) tk
over any reducible curve is zero.
2. The value of the integral over the circuit C is equal to the sum of the integrals over the circuits C1 ••• C,,' where all the integrations'are carried
• See Section 14.23 for tbis evaluation.
To illustrate the precedingeonsiderations we seek the integral of the power function'(z) == z" .
• The following considerationa apply equally well to the function A~, where A ita complex constant.
131
IdealFluid Aerodynamics
Elements of the Theory or Functions of"a Complex Variable
131
where n is a positive or negative integer. The integral of this fU,nction is of vital significance in the applications and, in fact, in the integratIOn of more general functions. We treat four separate ~ses. 1. Case when n == O. Then the function is simply a constant equal to unity, and we obtain
that includes in its interior the excluded int~riors ,of the.:curves Cl , ••• Crc (see Figs. '14.16 a~d 14.17). J We now evaluate the integral around the, circle: Using the polar form we obtair
,( . zn dz = ,(
IlrcJe R
f
{(z) dz
=f
dz
== 0
(Re")"iRe1e d8 •
1e;rele R
for any curve whatsoever.
== iR n
+1·f e,h.+111 d8
== iRn+: _1_ 1(&+11' n+1 ==0
'Ir
0.
From this and the results given in the previous section it follows that tbe integral of z", n > 0 over any circuit whatsOever lying even in a multiply connected region va.nishes. 3. Case when 11 < 1. Let us write n == ;m,where m is a positive integer greater than 1. We then have
{(n) ==  , ), z'"
,
t
m> 1
Cle"
PII. 14.17 mustrating the integration of z· in a multiply connected region.
and
d,
m dz ==  %.+1
2. Case when
11
> O.
Now the derivative of the function is given by
d{·· ,,1  == nz dz
Since 11 > I, the derivative exists everywhere and the function is analytic everywhere. We may immediately state on the basis of Cauchy's theorem that in a simply connecteO region
It follows that the function is ana~ytic everywhere except at the point z == O. We exclude this point by enclosingit in the interior of a circuit C. If we assume that there are no other excluded regions (which may arise due to physical considerations in [8 problem), the region external to C is a ,doubly'conriected region. Then from the results of the previous section we observe that for any circuit not enclosing tbe point z ... 0,
fz" dz 
0 with
11.> 0
.j %'"
,( 1. dz == 0,
m> 1
for any circuit whatsoever. . . . . . Consider a multiply connected region such. as I~ shown I~ Fig. 14:17. For such a region, on the basis of the results gI~en 10 the pre~lous sec:tlon, it is sufficient only to consider the value of the 1Otegra~ ov~r !rr~collc'l~ble '"educible circuits. Furthermore, the shape of the Circuit IS 10deed Immaterial.Consider then'a circle z l1li R and say it is an irreducible circuit
and thaffor all circuits enclosing the point z == 0 the integral has the same value. We evaluate this value by choosing a circle Izi ... R. We'shall then find that its value is zero. Suppose that in.addition to the singularity at the pointz == 0, there are other excluded regions. THe regi"on then is nply connected instead oT being double connected. Even in such a situation we shall find, just in the same manner as was done above for 11 > 0, that the integral over any circuit whatsoever is zero.
434
IdealFluid Aerodynamics
Elements of the Theory of FunCtions of a Complex Variable
435
(14.49)
We thus conclttde that for any circuit whatsoever we obtain the result
1...!..dz = 0,
rzm
m> 1
2.,
i.. d~ = 0, Jcz
'fc z
m> 1
whether the re&ion under consideration is doubly connected or nply connected. 4. Case when n = 1. We then have
,(z) =
for every circuit C. 3.
1.
dz
=0
= 21'1';
y
(for every circuit not, enclosing z
= 0)
(l4.~)
z
~
(for every circuit enClosing z == 0 once).
aplant
and
The function liz is therefore analytic everywhere except at the point z = O. We exclude this point by enclosirig it in the interior of circuit C. We assume that there are no other excluded regions. Then for any circuit not enclosing the point z = 0,
and for all circuits ertclosing the point z = 0, the integral has the same value. This value is readily found. We have
~~~~X
f,c
dz  = logz 1'1
Z
'1
FfK. 14.11 Illustrating the derivation of the Cauchy integral fannula.
For the function {(z) 1.
where ZI and Zl are coincident points on the circuit C. We know that log z is multivalued and that its value changes by 71'1'i over any positively directed circuit that encloses the point z = 0 once (see Section 14.10). We thus obtain dz  = 21'1';
== (z  zJ" we readily obtain the results:
"
t(Z dz
zo)" dz == 0,
5
Q or
"< ..... 1
(14.51)
f,c z
for every circuit C. 2.!
for every circuit C that contains in its interior the point z = O. The same conclusions are valid, as may be verified, if in addition to the singularity at z = 0 there are other excluded regions and the region is thus "ply connected. The above resultsconceming the integral of the function
j(z  Zo)"
== 0
(for ,every circuit not enclosing z = (for every circuit enclosing z
zJ (14.52)
== 21'1';
== Zo once).
14.19 The Cauchy Integral Formala
Let us consider a simply connected region and seek the integral of the function (14.53) where ,(z) is analytic in that region and Zo is any poi:tt in it. This function (14.53) is analytic everywhere except at z = z~. We exclude this point by a closed circuit C~. as shown in Fig. 14.18. From the results of Section 14.17
,(z) = z"
where n is a positive or negative integer, may In any simply or multiply connected region'
be summed up as follows:
(14.48)
1.
£z"
dz
== 0,
" 50
for every circuit C.
~
436
IdealFluid Aerodynamics
.. or the 'Ibc!ory of Functio.ns of a Coinplex Variable
!
431
It therefore follows that (14.54) reduces to we know that
,( ,(z) dz Jcz  %0
='0
ic zZo
C(z.) 
C(z) dz _ 217'i'(Zo)
(14.57)
where C is any circuit in the region and does not contain %0 in its interior, and that ! C(%) d% =! ,(%) ,d% (14.54) Jc % %0 'fc. % %0 where C is any circuit i~ the region and contains Zo in its interior (see Fig. 14.18). We e~press the righthand side of (14.54) as ,(z) d% _ C(~o'" ~ 'fc .%  zoo "fco % %0
Tbis res~lt constitutes the Cauchy Integral FormulQ: In Q simply connected region in which ,(%) is Qn Qnalytic function
217'1 'fc% Zo
~!
,'(%) dz
(14.58)
where (: i~ Qny circuit in ,the region Q"d Zo ;S Qny point in its interior. If we use the letter %instead of %0 f9t the Jeneral point, we express the formula a~
!
+!
C(%)  '(%0) d% 'fc. z  %0 '(%0) d%
Zo
~14.55).
{(z) _
.L!
'(IX) dlX
%
217'i Jc IX 
V4.59 )
_ 21Ti'(%o)
+! '( %) 'fc.
z 
Recall that all these integrals arc independent of the choice of the. curve Co (or, equivalently, of the curve C) as long as Co lies in the reglo~ of interest and contains in its interior Zo. Let us choose Co as the Circle Iz  z.1 = p and write
where now II is the (dummy) variable of integration. Cauchy's integral formula is a very remarkable statement. It shows·the strong interrelation ,among the various values of an analytic function. The values of the function at every point of a re~on are determined by its values on the boundary of the region. To know the values in the interior we need specify only the values of the boundary. The connection with the properties of the solution of Laplace's equation is readily seen. 14.20 UllliDdted DUrereatiabUity of an Analytic FaaCdoa Using Cauchy's integral formula; we .show that an analytic function C(z) may be ;epeatedly differentiated any number of times. By definition We have _' _ lim 4;\Zo)  ,(z) \dz/.t • .... %0% '
!
'f • e
{(%)  '(%0) d% = ! C(%)  '(%0) dz % %0 Jclrcle ~ % Zo
(14.56)
Now, if ,,(%) ' {(Zo)1 does not become infinite on the circle (which is what we assume), there must be some number E such.that on the circle ,,(%)  '(%0)1
C
<E
Then the absolute value of the integral (14.56) is equal to or less than the product: length the circle times the absolute value of the integrand. Therefore we have
of
I!
vanish:
'(%) Z 
Jclrcle,
%0
~(%o) d% < 217'P ~
'p
I
where z and Ze are any two neighboring points. If the region is simply connected and {(z) is analytic in it, and if C is any curve that lies in the region and contaiRs z and zo,.using (14.59) we obtain '
or
217'£
dC di
. ... hm ', ...... %0  z 2,,;
__ I_lim
1 1f (I '%0  1) '(II) dlX  IX %
t' CIt 
The number E, and consequently the absolute value of the integral. can be made smaller and smaller by choosing the r'adius of the circle smaller and smaller. But tlTe integral does not depend on p. Hence the integral should
!
2,,; ...
... _' ! 1
ic.. (CIt '(or:)
+(IX)
dlX
av)(1X  %)
!
Jrirrle
dlX
{(z)  {(%o) dz = 0
p
2"i j (IX  z)'
% 
%0
4JI
Repeating. this process, we find
IdealFluid Aerodynamics
Elements of the Theory of Functions of a Complex Variable
4J9
d'{ ... l!. 1.. {(at) dat dz· 21ri jc (at ·z)·
d", == .!!.!. 1..
'(at) dat tlz"21riJc (rt z)"
(14.60)
where the number at  z is expre~:;c:d in terms of the numbers at  Zo anp z  zoo Now we write 1 1 1 (at  zo)  (z  ..%0) == at  Zo 1  (~ ..:. %o)/(at  "0)
We thus obtain an integral representation for the derivatives of an·analytic function. The integrals involved may be shown to be finite. Therefore we conclude that a complex function that is on~ differentiable may be differentiated any number of times. The function and all its derivatives are analytic. We repeat this impartant result: An analytic function has derivatives of all orders in the region in which it is analytic.
14.11
If % is in the interior of the circle t, lz  zol < ,Ot and we may use the series
repre~entation
zol
( l_~)l==i(~)" at  Zo at ,,_0 %0
Taylor Series
Which converges. Introducing this series we rewrite Eq. (14.61) as
C(z).
Since a power function z", where n is a positive integer, is an analytic function, a power series 1: A"z", where A" is a ~mplex coefficient, represents in its circle of convergence an analytic function. We now show
~ '(at) dat 21TIJcat  Z
J:
==
~ J:
{(at) [
21TiJc
(z JOt ,,o(at  ~0)"+1
I
%o)".J
 ,,0 [2" (at _'(at) dOt](Z I 27Ti Jo zo)"+1
c
z )"
0
== IA ..(z where
GO
,,0
zo)"
(14.62) (14.63)
L.._ _ _ _ _ _ _ _ _ _
x
seri~.
Using the equation (14.60) of the previous section we observe that
1 , =  ) " n! dz" .. Equation (14.62) then become's
Fig. 14.19 Illustrating the derivation of Taylor
A
(d
n
(14.64)
that" if {(z) is an analytic function, then it may be expressed as a power series in a circle about any point of the regioh of its analyticity. Consider a simply connected region in which the function C(z) is an analytic function. Let z and Zo be any two points in that region and let C be any circle about Zo such that z is contained in the interior of the circle (see Fig. 14.19). According to the Cauchy formula we have
{(z) =
{(z)
1 == Ico nO
n ! dz
(dnt) ~
dz
Zo
(z  Zo)"
.zo
= '(zo)
. + (d() 
(z 
%0)
+ 
(d
Z ')
%0
dz 2
(z  %0)2 ._ + ...
2!
(14.65)
~ "{(Ot)
21T'JoOtZ
dOt
( 14.61)
This is the Taylor series for {(z) about the point zoo The series converges in an'y circle contained in the circle C. It is interesting to observe that in the interior of the circle C the value of the function ,<z) at every point is completely determined by the values of the'functiQaaMf ~aL... the point zoo . .'
Elements of the Theory of Functions of a Complex Variable
#0
IdealFluid Aerodynamics
441
14.11 Laurent Series
The Taylor series is an expansion for an analytic function about a regularpoint lying in a simply connected region in which the.function is analtyic. We now consider a doubly connected region and inquire into the possibility of expanding an analytic function about a point lying in the excblded region. If the excluded regiQn contains singular points of the function, our inquiry is with regard to the expansion of the ftlnctior•.'lbout a singular point.
y
To handle the integral around C. we proceed in a similar manner. Since for IX on C. IIX  zol < Iz  zol we write
;=; == (z Z 
1
1 zo)  (at  zo)
2:0 ,,0 Z Zo
1
== z 
at ( zo, 1  Z 
10)1
2:0
_ _ i(at  zo)" 1 Then the integral over C. in Eq. (14.66) may be written as
 ...!_.1.
2111
where
Jel IX 
'(IX) dlX
Z
==I A_,.(z "1
zt)}n
(14.69)
. A_"
== ~ 1.
2m Je.
{(IX)(IX 
ZO)',1
dat
(14.70)
Combining the results we obtain
{(~)
== I
aD
A,,(z  Zo)"
.~O
+I
00
A_,,(z  Zo)"
AI
(14.71)
Introducing a single notation, this may be rewritten as
~..~x·
Fla. 14.20 Illustrating the derivation of lAurent series.
Consider an annular region bOlJnded by two circles C1 and C. concentric with respect to the point Zo (see Fig: 14.20); In that region let the function { == {(z) be analytic. If z is any point in that region, then according to the . Cauchy integral formula we have {(Z)
{(z)
== I ==
00
A"(z  z,,)"
Ao
+ A (z 1
Zo)
+ A.(z 
ZO)I
+ ... A_I ++
z  Zo
A_2 (z  zo)·
+ ...
.
(14.72)
where
A"
= ~~~
{(IX) dlX 2m J cIX  Z
== 2.,( ( . {(IX) n+1 111 ~ .:x  zo)
1
dlX (n an integer)
(14.73)
(14.66) As in the previous section we can express
(14.67)
The curve C may be taken a~ any circle between CI and C. 01; any equivalent circuit. This series representation, Eq. (l4.71)·or (14.72), for an analytic function in an annular region is known as the 4Jurent expansion. The series converges for z in the; annular region, that is, for every z such that P2 < Iz  zol < PI' where PI and P2 are, respectively, the radii ~f the' outer and inner circles.
14.13 Integration of a Function with Singularities: the Residue Theorem
Consider a region interior to a cjrcuit C. Let ~(z) be a function that is analytic everywhere in the region except at a point zoo From the results in
where (14.68)
441
Section 14.16 we know that theintegtal
IdealFluid Aerodynamics
Elements of the Theory of Functions of a Complex Variable
y
IIJ
f
{(z) dz
vanishes for all circuits lying in the region and not enclosing the p>'liDt· Zo anJ !!!!It it has the same value for all circuits in the region and enclosing the point zoo . TheYAh!~ of the integral could not, however, be determined at that time. Now, using the Laurentexpansion, we can immediately find its value. Let Co be a circuit inJhe interior of C such that the singular point Zo is contained in the interior of Co. Then "according to the Laurent expansion. about the poi~t %0' (14.72), the value of {(z) at any point on Co is given by
{(z)·
o
!A ..(z co
co
%0)"
o
Fig. 14.21
700)"
We therefore obtain
ZIO
J.. {(%) dz == co J.. (z .fA" J JCI
.... 211iA_l
Illustrating the derivation of the residue theorem.
where
(14.74)
A_1(Zj)
==
the residue at the singular point
%
=
%/
since all the integrals except that of (z  zo)l·vanish, and the value of '" (;:  %0)1 dz is simply equal to 211i (see Section 14.18) . .The coefficient A~1 is assumed known from the Laurent expansion. Hence the value of the integral is found. The coeffiCient of (z  z;,)l term· in the Laurent expansion of a functi01l {(z) aboui %0. which is· a.singular point of {(%), is called the residue of {(z) at zoo Equivalently, the residue of {(z) at a singular pq;nt Zo is equal 10
Equation (14.75) COlIst.itutes the socalled residue theorem: the ralue of the integral of a function ,(%) orer an arbitrary closed curre C is equal to the sum of the residues of ,(z) at the Singular points enclosed hy C. This theorem naturally has many applications. In applications. the residue will be known from the Laurent expansion, and hence one may pod the value of the integral.
211i JCa
_1
J..
{(z) dz
where Co is a circuit that (ontairu only Zo in its interior. Equation (14.74) 'states the value of the integral is equai to 2111' times the residue. Consider now an arbitrary. region in which {(z) is analytic except for a finite number of singularities at the points ZI. z••... z" (~ Fig. 14.2'1). Let C1• C•• ... C .. be sufficiently small circles aboutthe respective centers ZI ... Zll' If C is any circuit that contains in its interior only the singularities ZI ... Zt. say, we have
1..
JC
{(z) dz _
J.. ,(z) dz + J.. {(z) dz + ... + J.. ({%) dz JCa JCI Jet
+ A_1(Zs) + ... + A_1(Zt)]
~
... 211i[A..1(z.)
11
 211i!At {Z/)
(14.75)
TwoDimensional Motion and the Complex Variable
445
Chapter 15
The equipotential lines described by <1>(%. y) = const. and the streamlines described by 'Y(x, y) = const. form an orthogonal set of curves. Similar properties are exhibited by the velocity components, for we have
TwoDimensional Motion and the Complex Variable
and
AU au =
ox oy au = au og ax
(condition of incompressibility) (condition of irrotation)
this .chapter we shall learn how to represent and to analyze the twodlme.1slonal potential motion of an ideal fluid in terms of the functions of a complex variable. We begin with the notions of what are called complex potential and complex velocity, which, as we shall see, are the combinations <1> + and ~  iv respectively. The problem reduces to that of find,ing an analytic function that must take certain values on the prescribed boundaries cf the motion. We shall see how this may be done by using the method of conformal mapping. In illustrating the application of the various ideas we consider mainly the problem of steady flow past a fixed arbitrary cylinder.
I~
i':
15.1
Cnmplex Potential and Complex Velocity
Let us recall the relationship between the potential and the stream function and that between the velocity components in twodimensional motion. Denoting. as before, the potential by <1>, the stream function by 'Y, and the velocity components by u and t', we have
Of course, the curves u(x, y) = const. and r(x, y) = const. form an orthogonal net. We thus see that the interrelation~ between the potential and the stream fllnction and those between the u and vcomponents are exactly the same as those existing between the real and imaginary parts of an analytic function. We, rna;. therefore, naturally combine the potential and the stream function into an analytic function of a complex variable or, cont'erse/y. assert that the real and imaginary parts of any analytic function represent, respectit'ely. the pO!f'fltial and the streaM function of a certain twodimensional potentialflow. Equivalently, the u :lOd vcomponents may be interpreted as real and imaginary components of an analytic function and vice versa.· It i~ customary to combine <1> and 'Y into an analytic function and call it the complt:x potential. DeBoting it by F(z) we write
F(z) = <1>(x, y)
+ i'Y(x. y)
(IS. I)
variable~.
<1> = <1>(%, y).,
u = u(x, y),
~
t'
== '¥(~, W
The x. yplane in this context becomes the plane of the complex The derivative of the complex potential is given by
==
L'(X, y)
The functions
ct» and 'Yare related through the differential equations
=u=
~z).= dF = a<t> + j iJ'Y
dz
ax
a:r
0<1>0'1"
ox
oy
0'1"
= ay  , ay = u(x, y)  iv(x. y)
~:
. a<l!
( IS.2)
=V=·
0<1>
oy
ax
They both obey Lal)lace's equation
~2'Y
The righthand side of this equation is the complex conjugate of u + iv, whi~h IS the Yeioclty vector. It has. h.owever. become customary to call the conjugate 14 IV the cumplex oelocity. It is usually denoted by W(z)
• In (.\ct. in twe dimensi.ms the corr'por.ents of an:, vector field who5e divergen.:e and ar~ both ""ril may be cc' nblneJ into an analytic fU'1ctiort.
= o·
<:lori
IdealFluid Aerodynarilics We thus have
TwoDimensional Motion and the Complex Variable
447
We see that 'I' becomes zero on the rays 8  0 and 0 == ex, where ex is given
W(z)
= u(x~1/) 
iv(x, 1/) ..
~
dz
(15.3)
by
The angle ex is restricted to the range 0 to 2.".. This means that the function Az", where A is real and n is greater than one, represents the flow around a
15.1 Flows Represented by Some Simple Fmtctioas For the flows introduced in Chapters 12 and 13 we may readily find the complex potentials and the complex velo 'ities. Here we shall consider some simple anal}tic functions and see what flow fields are represented by them. We investigate the functiori Az", where n.is an integer and A is a comple..<. constant. Let the complex.velociiy be represented by
W(z) .. Az"
Then the 9omp1ex potential is given by
F(z)
A == : z"+1
,...___ }+ constant
1. Case when n
== O.
We have
.,.""'1
u  iv, == W == A == U  iV. where U and V are constants. (15.4)
CI>
+ i'l' == F == Az =y (Ui: +( V1/) + i(U1/ ~
F(z)
Vz)
(15.5)
The function
== Az.
Fig. IS.1 Equipotential lines and streamlines defined by the fur.ction Azt/2, where A is Ii real constant.
therefore represents a uniformftow in an arbitrary direction. 2. Case when n = 1. If A is real, we obtain
u  .iv == W:= Az
==
A(x
CI>
+ i'Y = F
== AZ'
2
=.~ (xl 2
+ i1/) yl) + iAxy
corner of angle oc = 'iT/n, or, conversely, the flow in a corner of angle expressed by the complex potential
'X
is
The equipotential lines and the streamlines form an orthogonal net of rectangular hyperbolas (see Fig. IS.! and 'compare Section 11.2). 3. Case when n > 1. Express the potential F(z) as Az" and .1ssume that ,( i~ r~al. Using the porar form,
CI>
F(z) = Az'/«
Then for .the complex velocity we have
W(z) := A .2! z(~a)/a = A 2: r(·a)/aei(ral/a
<X
(15.6)
( 15.7)
oc
+ i'Y == F(z) '= Az"
= Ar"(cos nO
+ i sin nO)
4. Case II'hen n = I. First we assume that A is real and positive. We then have
LI 
Consider the' stream function given by
iv = W(:)
~~
= Ar(cos 0  i sin 0)
( \5.8)
( \5.9)
'Y :;: Arn sin nO
<P
+ i'Y =
F(z) = A log = = A log r
+ iAO
IdealFluid Aerodynamics We conclude that the flow is that due to a source at the origin. The strength 0f the source is given by q=27rA (15.10) Considei next that A is purely imaginary, equal, say, to ib with b Ne then have
TwoJ;>imensional Motion and the Complex Variable F(z)
A complex constant
449
W(z)
Flow
o
A
No Flow
Uniform flow in an arbitrary direction. Flow around a corner·of angle
IX 
> O.
Az A complex Az" >1 A real
II
u <t>
, .IV = w = I b =
~
'b .cos 0  1510 fJ) .' I r(
(I5.11) (15.12)
.,,/n.
+ io/ =
F
= ib log z = bO + ib log r
A logz A real and> 0 ib logz
A z
Source at origin. Strength q  2."A. Vortex at origin. Circulation r  2."b (positive circulation countelclockwise). Doublet at origin. Axis in xdirection. Strength JJ :z 2."A.
We conclude that the flow is that due to a vortex at the origin. The circulation r of the vortex is given by
r
b>O
= 2TTb
(15.13)
A
/Z
.b
Recall that according tQ our convention the positive direction of r is counterclockwise. We thus observe that the complex potenrials for the source and vortex flows are of the form
z A real and >
~
F(z) = A logz
where A, in general, is a complex coefficient. If A is purely real we obtain a. source flow, whereas if A is purely imaginary we have a vortex flow. In this sense the source and vortex belong to the same class of singular flows. 5. Case when n = 2. We assume that A is real and positive. We then have ( 15.\4)
By the superposition of the complex potentials for elementary flows, such as those described above, we may arrive at the complex potential for More general fto'w fields. For example, consider the combination of a uniform stream and doublet with its axis opposing the stream. Choosing the xaxis in the direction of the stream, we have for the combined potential
F(z)
==
Uz
A +z
== ( Ux + A 0) + c;s
i( Uy _ As:n 0)
(15.17)
<I>
+ io/ =
F = _
~ = _ A cos ()
:; r
+
iA sin ()
r
(\ 5.\ 5)
Hence the stream function of the combined flow js given by
UJ'
We conclude that the flow is that due to a doublet situated at the cmgm. The axis of the doublet is in. the direction 0f x and the strength of t:!C doublet is given by fl = 2rrA ( 15.[6; Prol:eeding in this manner we may obtain the corresponding flow, fur n <, 2. All these flows; as we know, are singular flows with the <;ingului\'1 at the origin. For convenience we list on the following page the llllpcnanr Ilows of the fuol:tion Az", where n is an integer.
T
==
U
A sin 8 U . 0 A sin 0 Y  .    == r sin    r r
We observe that 0/ == 0 on the circle
Thus the potential (15.17) represents uniform flow past a circular cylinder whose radius is equal to cylinder.
J A/V.
There is no circulation around the
450
IdealFluid Aerodynamics
TwoDimensional Motion and the Complex Variable the circuit C, whereas iri the circuit.
451
To obtain the uniform flow past a circular cylinder with circulation around it we add to the potential (15.17) the potential
r
is the total circulation of all the vortices contained
iblog~·
a
15.4 Flow Past an Arbitrary Cylinder
We now consider the steady flow past. a fixed cylinder of arbitrary cross section. Far from the body the velocity is uniform. A priori we do not know whether or not there is a circulation around the body. We assume, however, that there are no sources or vortices or other singularities in the flow field. Our goal, as before, is to obtain a solution for the flow field given the uniform velocity at infinity and the shape of the surface of the body. We choose the origin of the coordinates in the interior of the body. Then in the region exterior to the body the complex velocity may De represented by a Laurent expansion about the origin.· Since at infi· ity the complex veloc.ity should represent the uniform strean!" and near the cylinder a perturbation, we write
for a clockwise circulation, where a is a real constant. Addition of the constant has no effect on the flow field. We then have
F(z) = Uz
+~z
ib log ~ a
( 15.18)
The stream function of this flow is given by
'V = Ur
SIO () 
.
A sin () r

b log a
r
This becomes zero on the circle
r=a=/f, V
Thus the potential (15.18) represents flow past a circular cylinder with circulation. The radius of the' cylinder is again A/V.
W(z) = Ao
A} A2 A, ~ An + Z +  +  + ... = .." Zl Zl 0 z"
\.15.20)
J
whJ:re the coefficients A" are complex in general. The corre ponding complex potential is given as the integral of (15.20), which is
15.3 Circulation and Source Strength
We shall now begin our considerations with respect to the steady flow past any fixed arbitrary cylinder. As a preliminary in this regard we show that the circulation around any closed curve and the outflow of fluid through that circuit may be expressed as the real and imaginary parts 'Jf a complex function. By definition we have
I' =
1: V . ds = J (u dx + t' Jy) = 1: d<t> k J'c Yc 1: I,
~~
According to (12. lOa), the vo{.ume outflow of fluid through the circuit is given by Q = (u dy  r; dx) = I. d\}" Therefore it follows that
11. + a consta.t (15.21) nZ n  1z Our problem is simply fo sJetermine these coefficients so as to satisfy the prescribed conditions, such as the body conditioll. and perhaps the ~utta condition, if the body is of the appropriate shape. The problem, although it could be stated so simply, is not us~aJly as simple to solve! However, we can derive, on the l>asis of the theory of the functions of a complex variable, several important results without actually computing all the coefficie!lts in the expansion (15.20). We can readily remark about the coefficients Ao and A}. If we choose the Xaxis in the direction 01 the uniform stream, we should have
F(z) = AoZ
+ A} log z  !
~
An   ;:i
Ao = V
Now, for any closed circuit C we have
(15.22)
l'
+ iQ =
];.
,t, W(z) liz = F(Zz)  F(Zl)
(15.19)
r + iQ =
fc
W(z)dz
where ;:2 and Zl are coincident points on the curve C. Thus l' + iQ may be expressed as the jump in the complex potentIal around the circuit. If either r or Q is nonzero. the complex potential must be multi valued. The outflow Q is the total strength of all the sources contained in the interior of
= Oif C does not eRclose the body = 211'iAl for all circuits enclosing the body
• A3. there are no
the surface.
~ingularities
(15.23)
on the surface of the body, the expansion would include
452
IdealFluid Aerodynamics
TwoDimensional Motion and the Complex Variable
From (15.30) it follows that
45J
Expressing A1 as we have
'Y(a, 0)
== Va
sin 0
+ bllog a + { J .
. _ GO
<1 5.24)
(15.25) Since the bo~y is closed, the total source strength should be zero. Hence we should set (15.26) As regards b1 , we cannot set it equal to zero, for we afe dealing with a multiply conn'!Cted region and the circulation [or equivalently W(%) dz] mayor may not be zerO (see Sections 9.14 and 14.17). Thus b1 remains as an undetel'mined coefficient. On the basis of Eq. (15.26) the series expansion (15.20) for the complex velocity may be rewritten as Using (15.29) we have
(J
==
. _1__1_1 b" cos(n  1)0  an } ~I n  1 a"1\ sine"  1)6 b 1 10g a
t
bl is undetermined b.. == 0 for n == 2, 3, ... al == Val Q .. == 0 for n == 3, 4, ...
Substituting these values in (15.28), we obtain the complex potential for the flow past a circular cylinder as
W(z)
=
Ao
+ % + ,,_I ;%
GO
ib 1
~ An ~
F(%)
(15.27)
== == ==
V(%
+
aJ +
l )
%
ibllog ~
(15.31)
If the circulation is clockwise and its magnitude is
r, this equation becomes
The comflex potential then takes the form
•
F(%)
1 I
;:1 .
V(%
+a
+
i.£. log:' 21T (i
I 
F(%)
==
AoZ
+ ib 1 10g %  !
,,I
A"  n 
1%
+ a constant
(15,28)
The complex velocity is given by W(%) V J 
15.S Flow Past a Circular Cylinder
To illustrate the direct determination of the coefficients in the expansion (15.27), let us determine the ftow field due to a cylinder of radius a placed in an originally uniform stream U. We choose the direction of the Xaxis as that of U. The coefficients are to be determined according to the surface condition that (15.29) 'Y(r == a, 8) == 0 In view of thi:ioundary condition it is convenient to use the polar form and write the complex potential given by (15.28) in the form <I>
(
. %1
alo) + . r
1 21T %
The circulation naturally remains undetermined.
15.6 Complex Representadon of Forces and Moments Acting on· an Arbitrary Body: Blasius' Relations One of the important results that can be established on the basis of the series expansion (15.27) is the famous Kuttaloukowski theorem ab~ut the
lift on an arbitrary cylinder. To do this, and for the sake of lurther applications, we express in complex notatio~ t~e forces and moments acting on an arbitrary cylinder. Using the prInciples of conservation of linear and angular momentum we can set up, for steady condltlOn.s, formulas for the forces and moments acting on a fixed arbitrary cylInder Immersed iri a twodimensional flow field (see Appendix A). If CO IS any arbitrary circuit that l;ontains the body in its interior, and If X and Y denote, respectively, the x and ycomponents of the iorce on the body, we obtain, as shown in Appendix A,
X
+ j'f" =
F(z) = Ure;'
+ ibl(log r + if}) + (oc + ;(3)
_L
where we have set
~_'_n e nd n  1· rn
GO
+'b
.(,,118
(15.30)
1
Ao
== V
constant = oc I ;(3
Al == an
and
+ ib"
== 1
Y=
e V2 dy  1 pu(u Jy  v dx) Y 2 Yeo Co _ 1 e v 'd x pvC u dy  r ax) ieo 2 • e.
2
f
IdealFluid Aerodynamica The moment M on the body with respect to the origin is given by
TwoDimensional Motion and the Complex Variable where
ISS
Bo
== Ao'
(15.36)
M = 
re"c. VI(X dx + y dx) + pic•(vx ~ L2 i
uy)(u dy  v dX)]
B I  2AtAi B. _ AI' + 2AtA. etc.
with M taken positive for rotation about the zaxis according to the righthand rule. Th~se formulas, when expressed in terms of the complex velocity W(z) = u  'iv and the complex variablt. z, := X + iy, lead to very simple relations for the forces and, moment. One combines the equations for the force components to define a complex force· X  i Y, in analogy to the comple}(potential and eomplex velocity. We thus obtain·
Since we wish to integrate Wl(z) and zW'{z), we really are interested only in the coefficients of I/z~n (15.34) and (15.35).
y
X  iY = i!!! [W(z»)' dz 2 i co
M
(15.32)
=  ~Re{fc.cW(Z)J'z dZ}
~15.33)
X 'j Y
____• .~
u
....,I'+t'~
X
These equations are gen.:rally known as Blasius' relations. These relations may be readily extended to situations where more than one cylinder is immersed in the flow field. The simplicity of the Blasius relations is indeed striking.
Fil. 15.1 Flow with circulation past an arbitrary cylinder.
15.7 Force and Moment on an Arbitrary CylinC:er: KuttaJoukhwski Theorem
Consider the steady flow past a fixed arbitrary cylinder (~ Fig. 15.2). The uniform velocity at infinity is U. We choose the·Xaxis in the direction of U. According to (15.27), the complex velocity for this flow may be ex pressed as
The forte and moment on the body are therefore given by
== i!! ==
2 c.
2
f w,
dz
I.P2 TTl·B1 ~
== where
pTT{ 2A oAI}
Ao = U
AI = ib l =  j
= ipUr (r counterclockwise) == ~ipJJf ({'clOCKwise as shown in Fig. 15,2)
05.37)
1:.
271'
(for positive r)
or X =
(15.34)
We the'n obtain
0, Y = pur. The moment on the body is given by
B 1 · B. () () W 2() = B0+++++··· Z Z z, z· Z4
and
) W2( Z . Z = BoZ
M=  eRet'I w d:} 2 ~'o
Z
2
= _
+
BI + Ba z
+ () z·
+ () z·
+ ...
(15.35)
J
Tl1e reader should do this as a problem.
2 = rrp Re(iB2) = 2rrpU Re (iAt)
e Re (21TiB
2)
(15.38)
156
IdealFluid Aerodynamics
TwoDimensional Motion and the Complex Variable
457
We thus see: as anticipated, that the force on a cylinder of any cross sectIon whatsoever IS equal to only a lift force of magnitude pur per unit span of the cyhnder. Indeed, we do not yet know what the value ofthe circulation ~s, if it .exists. In general, there exists a moment on the cylinder. It is IOterestIng to observe that if we are concerned only with the overall f~rce a~d moment,. all we need to ·know are the coefficients .AI (i.e., clTculatlOn) and A2 10 the series expansion for W(%). For the circular cylinder the reader may verify that the moment is zero. 15.8
]t lapping
the complex potential F(z). We then have
F(z) = 1(0 = Il'{z)]
Therefore it follows that
dF dF d{ ==
dz
d{ dz
or,
of Flows
iF  =dF 1d{ d% dUdz
To rr~~d with the analysis of ftow problems involving complicated boundarIes we employ the technique of mapping. In this section we give ·some general results relating to such mapping. Consider the zyplane, which we refer to as· the zpJane, and a certain flow ~eld within ~ given I(~on. The ftQ.w field is a splution of Laplace's equatIOn and satIsfies certain prescribed conditions on the boundaries of the region. It is characterized by the potential ~z, y), or the stream function 'I'(x, y), or by the complex potential F(%) = 4» + i'l', which is an analytic function. We now employ a transformation E ... E(z, y), 'I = '1(z, y) which will map the region and its boundaries in the z..plane into another desired c~nfigur~tion in .the E,'1plane, which we refer to as the {plane. The dIfferentIal equatIon and the boundary conditions are also transformed correspondingly, and the solutions in the two planes are images of each other exeept perhaps at certain peculiar points. Let us denote by cz, and '1',. respectively, the images of cz, and '1". We have '
It immediately follows that the derivative
ceases to be analytic,
dUdz does not vanish. At the critical points
0/ F( 0 exists at all points where 0/ the trans/ormation F(O
The two flow fields are related as follows. We express the flow in the {plane in terms of the flow in the zplane. The inverse of the transf. "mation
{(z) = E(z, y)
+ i1/{z, y) + iy(E. 1/)
(15.39) (15.40)
is expressed as
%
= z(O =
z(E,1/)
From the results of Section 14.8 we know that if the derivative (d{fdz) ~ 0, then the inverse function exists with its derivative given by dz  = l' d{ dUdz
(15.41)
The complex potential
1(0 is
then obtained from F(z) by replacing z by (15.42)
cz" = cz,~~, 'I) == ~[%(E, 'I); Y(E, 'I») '1', = 'I'{(E, 'I) 'fl%(l, 'I), y(E, 'I» The complex potentiill in the {plane which is an image of F(%) is given by
F(O
:=
%({):
£(0 = £[z(O]
Iknoting the complex velocity in the Cplane by
W(O we have
cz"
+ i'l', + 11/{z, y) forms
W(C) == iF
d{
We know that if the transformation is such that E(z, y) an analytiC function, say ({%), then for all points where
I
d{ 12 = (OE\I ... (OE'f = (~)I t
dz
ox!
.oy!
ox·
(OrN ~ 0
oy!
= dF ~ = W(z)~. == W[z({)] d{
d%
dz (dUdz)dUdz (tS.43)
th~ complex potential F( 0 is analytic. In other words, F( 0 is analytic at all POlOtS where the transformation ({z) = E + i1/ i$ flot critical. .We could also have reached this conclusion fr~m the following consideratIons. Under the transf9rmation ((%) the ftow field in the zplane characterized by F(z) is mapped into a flow field in.the CplanC'characterized by
We see that the complex velocity in the 'plane will become infinite at the critical points 0/ the trans/ormation. There/ore in applications one usually chooses the· ,ram/ormation $Uch that the critical points do not lie in the region 0/ interest. HoweL'lfr,'if/OT onereason or hmI,her they must be allo.l't!d in the
458
IdealFluid Aerodynamics
TwoDimensional Motion and the Complex Variable
459
region of interht (as in the case of an airfoil, as we shall see) one requires that, in order to keep W(,) from becoming infinite, the complex velocity W(z) should be =ero at the critical points. I~ applications the known flow in the zplane is usually that of uniform mohon or that of flow past a circular cylinder or the flow for some other simpl.e geometry., One then seeks' the flow in the {plane over a comparatlv~ly complex geometry. The problem then is simply to determine the mapping function '(z) given a few requirements with regard to the correspondence between the two planes.
15.9 TransfonnatioD of Circulation and Source Strength
zplane and that for the circular cylinder as the 'plane.· We first inquire ,(z) or, equivalently, whether there is a unique transformation z = z(') that will map an arbitrary closed curve in the zplane into a circle in the ,plane and the region exterior to the curve into the region exterior to the circle. In answer to this inquiry it can be shown that there exists such a transformation and that it is uniquely determined if a given point of the zplane (in the region considered) and a given direction through that point are required to go into a given point of the {plane and a given direction through the point. t Consider the zplane, that is, the plane of the arbitrary cylinder (Fig. 15.3). We denote the undisturbed stream by the velocity vector V ... whose
y
,=
When mapping flow fields we shall be interested itt knowing how singular
flow~, such ~s sources and vortices, transform. To investigate this aspect co?sider C, In the zplane. l.et rand Q represent, respectively, the circu
lation around C, and the total strength of the sources included in the interior of C,. The circulation P is the total circulation of the vortices included in ~,. Under the mapping { = {(z), ~et the contour C, go into the contour C, In the {plane and, cQrTespondingIy, the interior of C into the interior of C,. Then for the circulation r, around C, and the tot~1 source strength Q, enclosed by C, we have
"
r  rlz),.
,  z(f)
~
'....,+~x
r, + iQ,=
i W(O
Cc
d{
_!
W(z) _1_ d{ dz
FiI. 15.3 Transformation of flow past an arbitrary cylinder into that past a cil'cular cylinder.
magnitude is V", and whose direction is given by the angie ex measured from . the Xaxis. The. complex velocity at infinity in the zplane is therefore given by V ",eia.. In transforming the flow past the arbitrary cylinder into the flow past the circlet· in the 'plane we require that the velocity at
• Previously we ~rded the zplane as' the plane of the known ftow field and the 'plane as that of the required ftow field. The present interchange of the roles of the symbols z and C should cause no confusion. t This result is a simple extension of Riemann's theorem: Given in the zplane a region A bounded by a single curve (or simple boundarY), and in the Cplane a circle C, there exists an analytic function C  C(z). analytic in A, such that to each point of A there is a corresponding point in the interior of C and, conversely, a point in .the circle corresponds to one and only one point of A. The function C(2:) becomes unique if a given point of A and a direction through that point are chosen to correspond to a given point in the interior of C and a given direction through that point. The boundary of A is tranSformed uniquely and continuously into the cifCl,Jmferenc:e o~ the circle. This theorem is readily extended to exterior regions by using the notion of inversion. We shall not go into the proof Qf the theorem or its extension. t We use circle and circular cylinder synonymously, also arbitrary curve and arbitrary cylinder.
icc' . dC/dz. dz.
(15.44)
! •W(z.) dz ... r + iQ ie
We thus see that the Circulation and the totalsource strength, with respect are equal to those with respect to its iMage Cr. to the contour It may be verified that if a source or a vortex is situated at a critical point of the transformation, the circulation and the source strength are not preserved under the transformation.
C,.
15.10 Transf()nnatiOD of Flow Past an Arbitrary Cylinder into that Past a Circular Cylinder
The Trans/orma,ioll. We now proceed directly to the problem of determining by means of mapping the flow past an atbitrary cylinder. For this purpose we take the flow past a circular cylinder. for which 'the solution is known, as·the basic flow field for the transformation. Following the usual custom. We shall denote the plane of motion for the arbitrary cylinder as the.
°
TwoDimensional Motion and the Complex Variable
460
461
IdealFluid Acrodynarrucs
infinity in the {plane should also be V rr>e i •. This is an obvious requirement, for the perturbation field due to a body must vanish at infinity no matter what the shape of the body. Now, if z = z(O is the mapping function, we have.
Wa}ldz/d' Since we require that at infinity the velocities be equal, it follows that we should ha:ve (15.45)
W(z)
=
series (15.46) as function of the shape of the bo.undary curve of the arbitrary cylinder. We shall ret~rn to this ~ro~lem 10 the. next ch:per, where we shall consider the mappmg of an airfOil IOto a circle. A, present we proceed, on the basis of the seri~s re~resentation (I 5.46), t~ an eval~atlOn of the flow field in the zplane (I.e., 10 the plane of the arbitrary cyh.nder) in terms of the flow field in the ,plane (i.e., in the plane of the circle).
Complex Potelltial aU Complex Velocity for tire FloNl ~ast tire Circle. Let , = fl denote the center of the circle. and a i~s radiUS (Fig. 15.3). The complex function { = !1. + ae" descnbes the Circle ..' Recall t~at the complex potential for the flow past a circular cylinder where the ongm of
'II 'I
or z
=,
fplane
for ,  ex>
1""\(11)
ihis means that the transformation function should be such that the point at infinity in the zplane goes· into the poiJilt at infinity'in the {plane. Since the directions of the velocities at infinity in the two planes are the same, we have required that in the transformation the direction IX through the point % = 00 go into the direction IX through the point { = 00. These requirementS a~ met by setting the scale factor dz/d{' equal to unity at
+
,=

v.
11~+1l
"     4  _ fl

I.:::..f... ~ •
00.
We may then. express the transformation in the series·
z( r) = r
... '<>
F'tI. 15.4 Illustrating the derivation of the complex potential and velocity for the flow
+
C1
,
+ fJ + fJ + ...
,I~""
past the circle.
the coordinates is situated at the center of the circle and the real axis is directed along the direction of the undisturbed stream is given ,by
The.derivative of zW is then given by the series
={+1;.2' ,,I '"
_
rr> C
(15.46)
dz = 1 _ C1
rl{
.
{'
ao
.. 1
{lS.47) . Note that there is no constant term such as Co in the series (I S.46). This, however, is no restriction, for the location of the circle is arbitrary and we can choose the coordinates of the circle in such a way that the constant term in the series expansion vanishts. Presence or absence of a constant term in the series (15.46) has no effect on the series (lS.47). Our problem in the mapping of the flow past an arbitr~ry cylinder into that past. a circle is then to determin.e t~e coefficients C 1 , Ct , ••• , of the
{~+1
= 1 1; n  "
2 C. _ 3 C. _ ... {C C
r
.2 'IT a '<>1 where {I is the complex variable whose plane i~ as shown in Fig .. 15.4 and· r is the "agnitude of the clockwise CirculatIOn. aro~nd the Circle. To obtain the complex potential for the flow past the cl~cle I~ the {plane where the complex velocity at infinity is V rr>e'· and the Circle IS located at , =!1. we use the transformation
F(1)
=
Vrr>{l
+ ilog .! +
r;{
Vrr>
a2
r
(15.48)
{ = {al) = 'Ii + fl
The complex potential ·in the 'plane is then given by
F({) .;,. Vaoa  fl)e«
j
(15.49)
r V""a + i 2 log { ae !1. + r!1. e j.
'IT
l
j.
(15.50)
'<>
• The series in (15.46) and (15.47) arc Laurent expansionS about the point at infinity in the ~plan~.
The circulation, the value of which we do not know a priori, remains unaltered during the transformation.
462
IdealFluid Aerodynauuc:s • . .
TwoDimensional Motion and the Complex Variable
463
For the complex velocity in'the 'plane ~ then obtain
where p«> is the pressure at infinity. For the magnitude of the veloCity at ,any point z = (z, y) we have
VI(Z, y) =
(15.S1) It is convenient to have W({} expres$Cd as a series in 1/'. Fqr this purpose we use the following expansions:
Idz/d" To, obtain the pressure distribution over the surface of the cylinder we set , = ~ + aell in the functions and dz/d~.
IW(z)11 == Iwmll~
Wm
(C 
~)I
1( ==, 1  ~)1
1
~
(C 
=,+r+r+'" ",)I,=.!. + 2e. + 3e. + ...
",I
r
r
e'
Force tuUJ'MolM1It 011 'M ArbitrlUY CylUukr. We ttad seen that the force on any ar~itrary cylinder pla~ in an originally uniferm str~m of speed V «> is simply a lift force whose magnitude i,s p V «> r per unit length of the cylinder (see Section 15.7). 'Since under conformal transfort1\ation the value of the circulation is preserved, we conclude that the lift force on the arbitrary cylinder in the zplane is equal to that on the circular cylinder in the 'plane. The moment on the arbitrary cylinder is given by
Mo  whe~
We may then express the complex velocity as
e Rei W,z)zdz 2 ie.
(15.54)
Wm = Ao + Al + AI + ... ,==
where
,r
_0
f A~ e"
(15.52)
Co is a control circuit enclosing the cylinder and Mo is the ~oment with respect to the origin of coordinates in the zplane and is positive when anticlockwise. The righthand side of (15.54) is readily expressed as an integral in the plane of the circle, for we have
WI(Z) dz = WI(C) _1__ dz de (dz/d{) de 1 ' = Wmde (dz/d{) Equation (15.54) then.becomes
(15.55)
Velocity Field ill 'M PIllM of tlie ArbitrlUY Cylilukr. velocity in the zplane is given by
W(z) = dF(z) ==
The complex
Mo = 
eJ Wlm z(C) de 2ice dz/de
(15.56)
.E F({} _1_
d{ dz/de (15.53)
dz
1
= W({} dz/d,
where Ce is any circuit enclosing the circular cylinder.· . , We immediately proceed to express the moment Mo in te!J7ls of the coefficients of the series expansions for the complex velocity V. l{}'and the mapping function %(0. Using the expansions (15.46), (15.47), and (15.52), we have
(l5,S7)
where W({} and dz/d{ are given, respectively, by Eqs. (15.52) and (15.47).
The PresslUe Field ill tM zpIllM. According to Bernoulli's equation the pressure at any point z, y in the zplane is given by
p(z, y) = p«>
{l5.5P'
with one another.
• Ccneed not be the image of C•• for ..n circuits enclosing the circle ~ reconcilable
+ lpV«>' ( 1 
VI) VZ.
«>
IdealFluid Aerodynamics
TwoDimensioDaI Motion anci the Complex Variable
and
Re 2tri(lB,C1
+ B.J 
Re (i4trC1 VOO1 e'N

2V..,rl'r')
In viftt of this, Eq. (15.62) may be rewritten as
Reace the moment on the cylinder is given by
M.  Lm oos (c)  ex)
+ M,.
{15.65)
M.  Re (pv.. rl"" i2trpV ·C1 r lllr) ..  Re(r.,....... ~ i2trpV.,,IC1e'"
~ L  lift on the cylinder  py.. r
y
R.....b. We have thus IeCnthat the velocity and prcssurc fields in Ibe flow past an arbitrary cylinder may be completely determined by the
(15.59)
L
_tbod of conformal transformation .. The problem reduces to that of "ding the coefficients iB the series expansion for the transformation f\ulction and of specifying the circulation .aro~d the given .cylinder. We bow that welldefined· circulation appears. only on certain types of cylinders that have airfoillike cross sections with sharp trailing edges and Ibcn the c:ircu1at:ion may be fixed by means of the Kutta condition. In the DeXt chaptcrwe shall Consider airfoils and sec ho. to determine analytically the circulation around them. Also, we shall extend the results of this chapter and obtain the aerodynamic properties or.a certain typcof airfoil
.2'
_~~L
_____________
~
_______.Z
.... IS.5 ID~ the fora ud moment on the
u:bitrary cyliDder.
To reduce (15.59) furtberlet us write the complex numbers I' and C1 as
l'
.·PM~
(15.60)
(15.61)
C  rjse*'1 1
with these, Eq. (IS.59) takes the form
(15.62)
The r~rcc system on the arbitrary cyUudcrii shown in Fig. IS.S.·There •• force ~ual to the lift L ~ng at the origin.O and' a moment M. about O. If"  ",.., is a point in the plape of the arbitrary cylinder, the moment ..,.,. such a pOint is given by
M,.  M.  Lm cos (c) . ex)
(15.63)
UIin, (15.'2) _obtain
M,.  2trpV~·~·sin 2("  at)
(15.64)
The Problem of the Airfoil·
Chapter 16
The Problem of the Airfoil
of the airfoil) or simply the chord (see Fig. l~.l). We denotethe length of the chord by I although usually it is denoted by c. We choose the Xaxis along the chord from the leading edge to the trailing edge and the origin of coordinates at some ~nt on. the chord, usually at the ltading edge or at the center of the chord. Then the surface of the airfoil above the Xaxis is known as the. u~per surface and that below the Xaxis is known as the lower surface. We denote them as
The problem of the airfoil may CORcern one or the· other of the following aspec:l$the determiDation of the aerodynamic characteristics. such as circulation, lift, moment, and pressure distribution for a given airfoil;. the design of aD airfoil to exhibit certain specified characteristics; the determination of the effects on the airfoil characteristics of changes in the profile and in the flow configuration. In a direct approach to the problem, one attempts to express the characteristics of the airfoil in terms or the shape. or the airfoil surface, the speed of the uniform stream, and the aititude of the airfoil with respect to the uniform stream. For this purpose Laplace's equatibn is to be solved directly for prescribed boundary conditions. If the method of conformal transformation is employed, the problem reduces to that of 4ietermining the mapping function that transforms an arbitrary airfoil into a circle. This is rather difficult. A procedure for finding such a mapping function has been gi~en by Theodorsen, and we shall describe his methOd in the last section of this chapter. The solution according to this method is not. available in a convenient closed form and cenerally involves considerable computati~n. The problem. of the arb,trary airfoil may, however, be solved approxlmatel>: by assummg that the airfoil is sufficiently thin so that the boundary conditions may be simplified. We shall preseot the simplified the0r:Y of the arbitrary air:oil in the next chapter. In this chapter we shall concern ourselves malDly With the indirect problem, namely, that of transforming a circle into a airfoillike profile and of determining the properties of an airfoil thus
11  '1..(%) for 11  '1,(%) (or if
>0
< 0
Lf~r
E
I
,
PIa. 16.1
Illustrating the airfoil nOmerJC\ature.
The mean line of the airfoil is known as the camber line of the airfoil (see Fig. 16.1). It is defined by 'Ie  i('1"
+ 'I,)
The thickness of the airfoil is defined by
'1, = 1('1..  'I,)
The actual thickness is given by 2'1,. Thus one speaks of the airfoil as compose4 of a thickness envelope wrapped around a camber line. 1f'1. == 0, the airfoil is symmetrical;' . otherwise it is cambered. In theory, we talk of airfoils for which '1,  0, that is, of an airfoil of zero thickness. The angle between the direction of the undisturbed stream at infinity and the direction of chord lin. from the leading edge to the trailing edge is called the angle of attack or the angle of incidence. 16.2 Mapping at the Trailing Edge As the preliminary step in specifying the cirCulation around an airfoil, we conSider the properties that should be exhibited ~y the mapping at the trailing edge of an airfoil. Let %  %T denote the trailing edge, and let the point on the circle corresponding to zT be denoted by 'T' We refer to bOth %T and 'T as the trailingedge· points. Consider the"C"lcmcnts d%l and dzl
generated.
We begin with some nomenclature associated with an airfoil.
1'.1
~o~.tare
There are two pain.ts Land T on an airfc:l such that the straight line LT is greater in length than any ot~r straight line jo~~ing two points o~ the airfoil. These points a~ called the leading and tratll~g edges. respectlv~ty. The leading edge is the one that faceS' tbe oncommg stream: The h~e joining the il!ading ·and trailing edges is known as the chord line (or aXIS
IdealFluid Acrodynarnica
• ~ Problem of the Aidbil
of the airfoil curve paSsing through the points '2' (see Fig. 16.2). Let dC. and dC., be the corresponding image elements at the image point C2" The angle ". between . . and dza is required to be different from w, as shown in the figure, since the trailing edge is to be a sharp edge. The angle between the image elements dC. and .dC. is, however, equal to 'Ir.Thia means that the mapping shoUld not preserVe angl~ at the trailing edge, that is, should not be conformal. This, in turn, means that the point C C2' should be a critical point of the transformation, • Z(O. The transformation will, in general, possess certain numbers of Critical points. These points are given by the solution of the equation (lk/dO  O.
Since at the trailing edJe tb/dCis zero, W(~•• W(C  'C2') is aISOzere. Hence we must set
',.J .will be ·infinite unless
(~6.1)
Thus the KJitI4 ConditiM r.ires.,htltOli 1M circle.IM compl~ Hlocity be zero ·at lite poilu C'" C2" which co"espo"tbto tM trtii1bW «Jge of th~ air/oil. In other words, the point C~'is a staption point. In the Cplanelet .ch~ the real axis along the line from the origin o to' the 'point Cr. The positive direction of the axis is tJW from.l) to CrIf
PIa. 16.1 Mapping conditions at tbe trailing edge. Thus what we require of the mapping is that one of its critical points should lie on the circle while all the rest of them are included in the interior of the circle so as to leave the regions exterior to the circle and the airfoil conformal to each other. The critical point on the circle then goes into th'e sharp trailing edge of the airfoil. . The angle between the elements thl and th. of the airfoil curve at the traUing edge 'I' 'is readily aetermined. If the first (n  1) derivatives of the transformation function are zero at the point C , then we have r
;......~~~t
x
.
__ IU
DcIoiIIJ IOIDe sym\dI for the loW put tile cirde. .
Denote tbe.ClCDter of the circle by the point C  I' and the ndius of the circle by a (.ce Fig. 16.3). If follows tbat . .
a· 1{2' 1'1
If the arpiiMat of tho vector {r' I' is denoted by ~,we' then have
". = n'lr
The angle is defined as the
l' . .
2'1r 
n."
11'(2 _ '!)
t,ailingedg~
angk.
.Cr  I'  af!~ (t6.2) Uiiag (IS.;O aad (16.2) we obtain for tbe CompJel veloc:ity at the point {I'
W{,,) • V.,e iar + i 1:._1  _ V.,d'~ _. 1 2'1r {I'  I' ({I'  1')' = V + I 1:...! _ VaD , ,iar_ GO 2'1ra,4I _ With this, (16.1) takes the fOTm
16.3 K.tta Coadltlon and the Value 01 Circulation We immediately proCeed to determine the circulation around the airfoil by appliCation. of the Kutta condition .. According to this c~ndition we require that the velocity at the trailing edge beJinitc.and c()ntanuous. The complex velocity at the trailing edge is given by
,4.
·1 W(,  '2'> ... Wa ==. '2')' (dz/dC)'..(r
2'1raVGO [1  ~~(""') + ire'(""'. 0
470 From this we obtain
ldealAuid AerodynamiC$
The Problem of the Airfoil
411
r
4."a V., sirl(1X + fJ)
(16.3)
·U
~~~
~~
This equation shows that the, cir.culation depends, in general, on the airfoilshape represented by the parameters a and p, the speed V." and the angle of attack IX.
16.4
1.0
urt
on the AJrfoU
0.1
The lift on the airfoil is given by L
== prIOr ==.4."pV.,·a sin (IX + P)
(16.4)
0.6
For the lift coefficient we obtain
CL

)
~~
I I/( If
V
to
~ IpV.,1

8"'(~) 1
sin (IX. +
P)
(16.5)
For 6mall angles of attack (strictly for small "IX be expressed as
+ p>, the lift coefficient may
0.4
(16.6)
The chord of the' airfoil; as we shall see, is about 40, being equal to it in the limiting case of a socalled fiatplate airfoil. Thus we'maysay that the lift coefficient is about 2." sin (at + p> or 21r{at + P>for Smll/l angles of The lift on the airfoil is zero when the angle of attack is equal to .... p. We call this angle of attack the nolift angle (or .the zerolift angle) an~ denote it by 010. We thus hav~
02
attack.
o
(16.7)
6/4
~ ~
V Ii
2
If J
0 2 4 cr (dtcnIa)
6
a
10
Introduc:ingllto, it is customary to designate at  lito as the effective angle'of attack and denote it by IX.:
FIll. 16.4 Comparison or theoretical and experimental results for the lift or an ainoi.
16.5 MODleIlt on tile AJrfoU
"'. == IX 
lito
== IX + P
(16.8)
The formulas (16.3) to (16.5) and the consequent realization that the lift coeftkient depends linearly OD the angle of attack (for small angles of attack) are fundamental results and agree .satlsfactorily.with experimental observations (see Fig. 16.4). They are landmarks in the theory of lift. Their discovery was of vital importance in the eventual achievement of winged fiight. Before their discovery, that is, until abo~t the .year 1900, the 'outlook for a rational and practically useful theory oChft was extremely pessimistic. •
• For interesting'and iIluminating historical surveys. reference may be made to Durand (1934). Pritchard (19S7). and KarmAn (I9S4).
as that which tends to increase the angle of attack (as the socalled noseup
moment). Henceforth we shall use this convent;of/. Then, according Eq. (15.59), the moment on the airfoil is given by
Mo .... 2",pV.,·~ sin 2("  at)  Lm cos (d  IX)
t()
It is crut01l1QrJ' to dtjine the positive direction of the moment on an airfoil
== M,.  Lm cos (15  IX)
where m, 15,
~,
(l6.9)
and" are defined by the relations (15.60) and (15.61):
== me;' C 1 == ~e"7
p
The Problem or the Airfoil
m
Let \Jl compute. the moment, denoted ,by M, about any point % in the plane of the airfoil Introd~~gthe' vector (see Fig. 16.5) the moment does not change when the angle of attack .is:varied. Such an inquiry is natural in view of the importance of the role of momeDts in the . dynamits of an airfoil or of a winged vehicl~. We answer thcinquiry.u follows. Equation 16.10, which gives the moment about any point, may be ~ written as.
we obtain
M  M.
a/AII"
+ Llfeos (fl 
lit) lit)
 21TPV..,
~[sin 2(" 
+2
a:
y
sin (lit
+(f) cos (fl 
lit)]
(16.10)
M  21TPV..,·ftI{Sin 2(1It 
"l + ~. [ain <P + 9') +
sin (lilt
+ fJ 
9')])
(16.12) From this equation we ."., that if we chooIethe point.  Jauch that
where use has been made of the expressions for M,. and L. This equation shows dult M, like L, depends on the, airfoil shape, the speed V.." and the
_  ftI 1"+17' a
(16.13)
r&~~~~x
then the moment about J becomes independent of the angle of attack.f We thus conclude that there is a point about which the moment remains unaltered as the angle of attack is varied. Such a point we call the aerodynamic center. Its coordinates are given by (16.13). See also Fig. 16.5. The moment about the aerodynamic center obtained· by substituting (16.13) into (16.12) tUrns out to be
21TPV..,rq'I sin 2(fJ
+ ,,)
which indeed is independent of Qt. Furthermore, this is exactly the moment at nolift. We thus write
it Ff8. 16.5 Deftoiag &be symbols uaed in relation to the IDOIDCDt on the airfoil.
engle of attack lit. Furthermore; unlike. the lift; the moment explicitly involves a cOefficient, C1  tlIe'''', of the transformation function.
MotM'" tit Nolift. From (16.9) we see that even when the lift on the airfoil is zero, which happens for lit  ~ (J, there i$ a moment. In such a situation the moment about anyone point is the same as the iDoment about .any other point. We denote this moment at nOlift by M. Then from Eq. (16.9) we obtain
moment at nolift
 moment about the aerodynamic center  2.".pVoo~sin2(JJ
+ y)
(16.14)
This shows that, for given values of P and V"'" &J depends on~y on the shape of the airfoil. The moment about any point z may now be expressed as
M 
if + L Iz.  %'1 cos [arg (z 
z)  a]
+ sin (A
 B).
it 
_2.".pV..,I~ sin 2("
+ (J)
From this it follows that the moment is 'Zero about any point lying on a line
(16.11)
AmHly. . . . CAllln all , . Monw", dolli till AcrotlyMlJlk Cellter. Wo inq~ whether there·is a point i~ the plane. of the airfoil about which
t
• Use sin 'I' = sin ",.and 2 sin A cos B ... sin (A + B) Th~ reader should verify this. .
•
IdealflUid AerodyDamica
Tbo Problem of the Airfoil on the airfoil is given by the Bernoulli equation
411
paralkl to 1M direction of 1M lift and at a perpendiC1llar distance
k
M L
06.15)
. I". p on the alr,OI'1  p"" 
e y""1[1 2·
IW(I)I' on• airfOil]
V""
from 1M aerodynamic center i (see Fig. 16.5).
The intcrsectiDn of this line with the chord line of the airfoil is usually referred to as the center of pres.re, d~notcd by the symbol C.P. This should not be misunderstood as the poil\,t of application of the resultant of the pressure forces acting on the airfoil, for in general there is no such point; there is·only a line of action ·ofthe resultant force. The changein the center of pressure with variation of the angle of attack is an important characteristic of airfoils, particularly for consi<ierations of stability. The moment about any point may also be expressed in nondimensional form'
The magnitude of the velocity on lbe airfoil surface is obtaiDed from (16.18) as
IW(I)I on the airfoil 
2V"" [sin (IX
+ P)
Idl/del on the cirele
1
 sin (IX  6)]
C
M'"
M.
IpV",,'1'
Thus the moment coefJicient for themo~ent about the aerodynamic center is given by
CM

Ip""V""fI
it ,

411' (~, sin 2(" I
eM
This completes the essential results in the theory of lift of an airfoil. It is remarkable that we were able to obtain these results without constructing explicitly the solution of the ftow field. This was pOssible because of the use of the simple but powerful methods aWorded by the theory of functiOns of a complex variable. In the next chapter we shall see hO'tlHhe socalled simplified theory of an arbitrary airfoil may be handled without employing the complexvariable representation. Now, we proceed to consider some details of the mapping between a circle and an airfoil. 16.7 . TransforDlation 01 a Circle into an Airfoil We now. consider the problem of generating airfoillike profiles by choosing a circle and a transformation function. The transformation from a circle to an airfoil may be expressed as .
%== %({)
""C == { + ~ !! 7{"
+ p>
(16.16)
Introducing the coefficients CL and
Jc I
we may rewrite Eq. (l~.IS) as (16.17)
OD
eM
CL
166 Velocity .... Pre de DistribatIou
tile .AlrfoD Smface
(16.19)
We may readily obtain expressions also for the distributiOns of velocity add pressure on the airfoil. The Complex velocity on the airfoil surface is given by
W(I) on the surface
~ [W(O on the circle1 [(dz;dO on the Circle]
Such a function, as we had seen (see Section 15.10), meets the requirements at infinity: that the point at infinity in the %pJane goes into the point at infinity in the {plane and that the complex velocity at infinity be· mapped into itself. Consider the derivative of the transformation (16.19)
Denoting the points on the circle by
CJl+a~ w~ obtain, using'(IS.Sl) and (16.3),
W(e> on the circle  V""e  V""e,(eII'
 i2 V"" [sin (IX
d% d{
==
1. _ C 1 {2
_
2 C2
r:
.:.
3 C,
{'
(16.20)
+ i2V"" sin (IX + p)esin (IX  6)]eII
(16.18)
+ P) 
This becomes infinite at the point { = 0, that is, at the origin of the coordinates. It may become zero at a number of points: Let us say that there are k suchyoints and let us denote them by {I> {2, ... , {k' These are then the solutions of the equation
The velocity di.ltribution on the airfoil is given by multiplying the rightband ·aide of (16.18) by (dl/dO 1 on the ~Ie. The pressure distribution
416
The Problan of the Airfoil
HI
In the expansion of the leftband side of tbiI equation, the coefticientof the term is II
'I,
1"
16.' The Joakowskl Traasfonaatioll We now apply the preceding ideas to obtain the simplest transformation. For this purpose we look for a f~nc:tion %(0 whose derivative has only two zeros. Denoting them by and ,. we should have, ac:cording·to (16.21), 
Ie.
This must equal the coefficient of the expansion given in (16.20). This latter coefficientis~ however, zero. Th~fore we should have
(16.21)
""in
'1
or
'1 + ,.
0
'Now, one of these points should be the trailingedgc point. Let us set This means that the centroid of the critical points of the transformation is the origin of coordinates in the. 'plane. We now have sufficient information to choose the circie a.nd construct the transformation. The procedure is as follows: 1. Choose k points in the Cplane to be the zeros of dz/dC. We denote these zeros, as done above, by the points C, C., ..•• ~. 1 2. The transformation is then given by
(16.22), (16.24)
'.Cl
Then it follows that
C ..:.C2' ...
dz _
(16.2$l
I
Equation (16.22) now takes the form
dC
'(I _t2')(1 + C,.)
CC
.. 1..1:. C•
r
(JA.26)
The integra! Qfthis equation may be expressed as
z_
The transformation function then becomes
{I:
%(0 _ { + C1 + C. + ... + CII
.
cr'
(J6.23) Choosing th(. real axis along the vector {,. we set
(16.27)
wherethecomplexcoeflicientsC1 , C., .• ~, Cl:aredeterminedbyCl""'~' 3~ Choose the centrt>id of the %eros of dz,dC as the origjn 0 of the coordinates. 4. Choose, one of these zero points to be the trailing«lgc point. As before, denote it by C,.. . S. Draw a circle with center at any point but 'such that it encloses allthe zeros of dz/d, except the point ,,. and passes througl1 :C,.. Denote, as before, the center of the circle by the point p. The radius of the circle is thel\ given by aI{,.  pi
{,. c
where C is a real positiL1e number. The transformatk , (16.27) the form
(16.28)
the~
takes
(16.29)
6. Choose the line 0 to {,. as the real axis e The imagina 'j axis fJ is .. then automatically, fixed. Oerioting, as before, by  {J the a gument of ",.  p we bilve ae i , =: {,. _ /.I
7. Choosing different sets. of zeros of dz/d{ and different circles, we may obtain an infinity of different airfoil shapes.
This simple transformation is known as .the Jouko ....ski transformation. It not only leads to the loukowski ai~roils but also plays a basic role in many approximate theories of the airfoil. We shall return to a consideration of the loukowski profiles in the next two sections. Now let us exatnine the basic properties of the transformation (16.29). The points·' = C and, = C map, respectively. into 'the pointt z = 2C and z = 2C (see Fig. 16.6). Consider the circle described by
{ = Ce"
m
IdealFluid Aerodyaamica
The Problem of the Airfoil
Points on the circumference of this circle go into points on the znis (in tbe zplane) along the strip 2C ~ z ~ 2e. As we traverse along the top half of the ~le from Ato B to D. we cover the top of the strip from A to B to D. As we traverse along the bottom half of the circldrom D to E to A. we cover the bottom of the strip from D to E to A. Denote the points in the CpIane by
C,e"
"
We thus conclude that the transf9rmation (16.29) maps circles about the origin in the {plane into a family of confocal ellipses in the zplane and rays through the origin in the {plane into a family of confocal hyperbolas in the z7plane. The foci of the ellipses and hyperbolas are the same and &relocated at z  ±2C. The ellipses and the hyperbolas form 1m orthogonalnet. Consider the mapping of the circles exterior to the circle r = C. As the radius r. of these circles gOes to C, the corresponding ellipses tend to the strip  2C ~ z ~ 2C. Thus we see that the exterior to the circle C is mapped into the exterior to the strip. What about the mapping of the interior to the circle r  c? The interior to the circle is again mappedinto the region exterior to the strip. To see this let us consider the mapping of the two circles and
'"  ro
,... 16.6 Joukowski tranarormadon.
Then the transformation (16.29) may be expressed as
z  (,
+ ~), cos 8
'" _ C' ro where ro is greater than C. Thl s the circle", = ro is exterior to the circle C, and the circle '"  (C'lr J i. interior to the circle C. From (16.31) we conclude that both the circles are m:l.pped into the same ellipse. We thus conclude that the mapping represented by (16.29) is double valued. To avoid this double valuedness, we introduce two sheets of the zplane, one over the other and suitably joined across the strip, and consider that the exterior of the circle C is mapped into one of the sheets and the interior into the other· sheet. The ooints x = ± 2C are then known as branch points, and the strip 2C ~ x ~ 2C is known as a branch cut. The sheets are known as Riemann sheets.
(16.30) 16.9 Joukowski Airfoils Tile Circle. We now apply the Joukowski transformation to a circle. We choose the circle such that it passes through C= CT = C and encloses the point C=,  C =  C. If, as before. J.l denotes the center of the circle T and a its 'radius, we have
(\6,33)
y  (, 
~I)
sin 8
For r equal to a constant ro we obtain, eliminatiD~ 8,
xl
( ro
+ C') + '0'
('0 _CI\I  1
'0'
yl
(16.31)
This equation represents an ellipse in the zplan~ hAving foci at x = ±2C (Me Fig. 16.6). For 8 equal to a constant 80 we obtain from (16.30) on eliminating r, • Xl . Y =1 (16.32) (2C cos Oo)t (2C sin 00 )' This equation represents confocal hyperbolas having foci at x = ±2C.
The quantities m and (3 may be taken as parameters defining the circle. Generally m and 8 are chosen small. The TrailingEdge Angle. This is given by
'T
= 77(2  n)
where n is the order
~
the derivative of .:a) which first becomes nonzero
IdealFluid Aerodynamks
The Problem. of the Airfoil
III
~t tlie trailing edge (see Section 16.2). For the Joukowski transformation It may be readily verified, n is 2. Therefore it follows that '
generated from this circle we .fint observe that the loukowski transformationlDkY be expressed as·
II: 2C := II:
.,.0
We thus conclude that the airfoils generated ~y the Joukowski transformatior: have clISped trailing edges.
+ 2C {t t
('  C)I C)I
(16.34)
Denoting the argument ofa 
~M F/(ItP/(Ite A1rfoi!. Choose the circle with center at the origin and radiUS equal to C. We thus have
II .. 0,
::&
c)/a + C)b/9'(seeFigure 16.7) we obtain arg (%  2C)  arg (z  1<..1  29' (16.35)
a  C,
P =0
We ha.1.! seen that such a circle transforms into a strip on the Xaxis e~te~dlDg from z == 2C to' z +2C. We talk of this as die flatplate alrfotl. Its chord is equal to 4C.
The Circ~/(IrA'c Airfoil. Choose po == me/ria == im. Then p is not zero and a IS equal to C ~ P'(see Fig. 16.7). To interpret the airfoil
rplant
Now, for points on the circle 9' is a constant. Ibis means the corresponding image points should describe a curve such Lhat for t~ose points the lefthand side of (16.35) remains a constant equal to 29'. Such ~ curVe is simply a circular arc passing through the points % ... 2C lind II:  2C and the point D (see Figure 16.7). This point, as may be verified, is given by
OD  lim
We thus conclude that the pres~r.t choice for the circle generates a circular arc airfoil. Naturally it has n I thickness, just like the flat plate . . The chord of the airfoil·is ·,C. The maximum camber is given by the ratio of the maximum ordinate to the chord:
 1 tanp
4C
2m
p is a measure oj the camber of the airfoil TIw SYllUlWtriclll Airfoil. Choose po  ~m . . Then P is equal to zero. Setting (16.36) maC where e < 1
Thus the angle we have
a  m
+C
::&
C(I
+ e) •
(16.37)
(see Fig.
y
zplane
16~8.)
A point on the circle is given by
, == II + ae l ' ==
z == (ae j ,
'
ae"":" m
The corresponding point on the airfoil is given by
D

m)
+ .C "(ae"  2q
t
m)
(16.38)
z+ ___ .
~..c ..
2C
_2T
~>" 2",
• From z
= , + (C"m we have.
:  2C
_2C"'2C
x
and. similarly,
.
= .
{!
+ C'
{
({  C)" = .
,
: + 2C = <C + C)"
Fl•• 16.7 Illustrating the derivation of Ihe circulararc airfoil. Equation 16.34 immediately follows.
,
III
ldealAuid Aerodynamics
The Problen\ of the Airfoil
If on the righthand side of this equation we replace 6 by 6. the lefthand side will become z. This means that the airfoil curve below the real axis is a reflection of the portion of the curve above that axis. In other words, the airfoil is symmetrical with respect to the real axis. It has thus thic1cness but no camber.
The thickness of the airfoil ~t the ctnter is given by
I.
_2,(6  i)  4C8
'max 
"
(plane
The maximum thickness ocx:un at the. point where dy/d6 is zero. This Occurs at (J _ (2.,,/3), which corresponds to quarterchord point from the lead;ng edge. We thus obtain
4CB 4
I max
3Ji
The thickness ratio is defined by
T=chQ.rd
Hence we have
T 
(16.40)
3../3 8 4
1.2998
y
,plane
We thus see that 8 is a measure of the thickness of the airfoil. It is referred to as the ec:centriclty..
Air/oil CtUnbn . . TlrkkMu. Choose the center of the circle at any atbitrary point /I ... where IS 'JfI:. (11/2) or 11 as in the preceding two cases. Let /I. denote the intersection of the vector ar i , with yaxis (see Fig. 16.9). A circle with center at /Iq and radius equal to l,uo  'xl transforms into a circular arc. The arbitrary circle, the points of which are described by , = ac" + /I, goes into an airfoil that includes in its interior
,.,it"
me".
Fig. 16.8
Illustrating the derivation of the symmetric JOl,lkowski airfoil.
The chord of the airfoil is given by
1= 2C
+ /%LI
C
T
where
%L
is the leadingedge point of the airfoil. We have*
%L
C = h +  = C(l + 2E) tL = 2C(1 + 2E2 + ... )
I = 4C( I
l
I + 2E
Therefore we obtain
• Note that "
+ £2 + ...)
( 16.39)
=
(a
+ m) =  C( 1 + 2c).
Fla. 16.9 IIIustratinc the derivation of an arbitrary Joukowski airfoil.
IdealFluid AerodynamiCs
t~e ~ircular an::. We say that the cin::ular an:: forms the skeleton of the ~rfod. The angle {J determines the mean curvature of the profile. The
dlstan.ce
The Problem of the Airfoil
~ ~lm~le gra~hical construction may be given for the arbitrary 10ukowski airfoil. This follows directly from the geometrical meaning of the simple tran!lfo~ti~ns w(0 .. { and w(0 .. (CI/O (sec Section 14.12). Such a construction IS known as the Trafftz graphical construction.
16.10 Properties of JoukowskJ Airfoils
I", _. ,...1. determines the thickness of the profile.
~e
~xpressed lD n·ondlm~nslonal form. To obtain these properti~ we specialIZe the results of ~on. 16.4 and 16.5 by using the 8'!Ometrical properties of !~e 10ukowskl 8lrfods. First we observe that tlte coefIicient C1 7
These ~roperti~, 1ik~ the properties of any airfoil, are conveniently
(16.19) in the general transformation
The corresponding properties for the flat plate, circular arc, and symmetric airfoils arc given in the follov.jng table. Fol' the circular arc the results arc expressed in terms of the maximum camber defined n<?ndimensionally by 2.,,/1 (sec Fig. 16.1). Similarly; for the symmetric profile they are expressed in· terms of the maximum thickness· defined by the parameter T (sec 16.40). We sec that the angle of nolift is directly related to the camber of the airfoil. The tift coefficient is affected by both ~Diber and thickness. It may be verified that a 10 per cent change in 4m/1 causes only a I per cent change in C L' Thus we may say that the cUcuiararc camber has no Significant effect on the lift,' it only shifts the nf!li/t angle. On the other hand, thickness affects the lift J;Dore significantly than the camber. Increase in thickness increases lift. Because of viscouit effects, however, this increase is not
z  z(O.' +!::! + .! + ...
tUnas out to be
_
C
, ,.
C
realized practkally.
Airfoil
Property
FIIIl Plate
Circular Arc
Symmetric
Cl =
~en7
== C.
for the 10ukowski transformation. Hence we have
o
(16.41)
Cr.
tan14m I
o
~2w(1
~  C and ,, 0
2.. siD II
2,,[1 +
(7)1~sin(1I + f1)
+ O.77T) sin at
4
For the general loukowski airfoil with thickness and camber we then obtain the follOwing results:'
ill
C 87r(i) sin (at + {J)
L ..
0 0
1
4
f7' 4 1 + (4m)1~e" 7
m I[
~
(16.42a)
«e  fJ ! _.~ _ ca e"
I I
al
"  sin 2P 4
0
(16.42b) kll (lo.42c) (16.42d) (16.46e)
_! [1 + (~rrWsinsin2p p> . 8 I (<< +
0
eM I
4tr(7J sin 2{J
CL
~_·CM
·Recall that the center J.l of the circle, its radius a, and the angle {J are c.onnected by (16.33), which is
me kJ
= J.l = ae"
 C
·'See Sections 16.4 and 16..5 for the definitions of the quantities involved.
The aerodynamic center is at the quarterchord point for the flat plate and.practically at that point for the symmetric airfoil. For these the line of action o(the lift passes through the aerodynamiC center. We may therefore say that for these airfoils the center of pressure and the aerodynamic center coincide. Joukowski profiles have been investigated extensively both theoretically and experimentally. Examples of calculated and measured pressure distributions over a Joukowski profile are shown in Fig. 16.10. The agreement between them is satisfactory. The differences can be explained as due to the influence of the viscosity of the fluid which, of course, is neglected in the theory. A comparison of the calculated and measured lift on a
IdealFluid Aerodynamics
ne Problem
of the Airfoil
"7
1.4
I
1.2
ep
I
 ....... #1 ~
1
1.0
0.8
I II
I
1i,
II
/ Ii 1/
/
~
~
I
I
/,
f
j
0.4 0.2
.~ I,
1/ /
~
'~/~
I
o
'l
So
0rr__~~
 ..... ,
0.2
.... J
10"
//
o
so
10"
1S0
_til
Fig. 16.11 Com~rison of th~retical and experimental results (or the lift on a Joukowsld airfoil.
16.11 Otber ,Airfoils With the Joukowski airfoils of moderate thickness there is no freedom of varying the camberline appreciably from a circular arc. We have .seen that an increase in the camber of the circular arc does not improve the lift considerably. Also, the maximum camber lies rather close to the center of the chord, whereas a position within the forward portion of the chord is usually preferred. To avoid these disadvantages one may generate airfoil profiles from a circle by means of a transformatien that im::ludes more than two zero~ of the derivative of the mapping function. Airfoils derived in this way are known as Mises Airfoils The procedure for actually <;onstructing them is similar to that for the Joukowski
F•• 16.10 Comparison of theoretical and expen'mental result r th Joukowski airfoil. . 5 lor e pressures over a
10ukowski ~rofile is. shown in Fig. 16.11. The agreement is again satis~act~ry. particularly 10 the region of small angles of attack. The calculated hft IS somewhat lar~er th~n t~e ~easured value, indicating that the calculate~ value of the cI.rculatton IS shghtly larger than th. actual value. This agalO may be explamed by the influence of viSCOSity.
JdeaIFluidAcrodynamics family of airfoils; Their properties may be readily determined. In this connection reference may be made to Mises (1959).
The Pro~ of the Airf'oD
lire KfumlutTnlftt Air/oill. The 10ukowski family and' the Mises family of airfoils all have cusped trailing edges. Naturally it is not possible to construct such edges in practice. , Hence one seeks a transformation that will generate from a circle an airfoil profile whose trailingedge angle T is not zero. A simple transformation of this kind is set up as follows. We had seen that if , .... Co is a critical point (specifically a zero of the first derivative) of the transformation z  1(0, thenwe may express &(0 in a series expansion about the point Co and write
ibldC.
We require tbat the tran.fo~tion should be such. that for ,  00, 1. To satisfy this requirement. on the basiS of (16.47). we
should set'
Zr  IICr
Then the transformation takes the form
(16.48)
,,'  1 Cr' _·C++··· 3 C
or the fonn
Z
(16.49)
I(P 
~
IICr .,<C  Cr)
z(Co)  (C 
C'>f(O
(16.43) Finally, if we set (16.50) as '
• + "Cr
_  ftC
• "'T 
<C + Crt
(C  C),
(16.50)
where II is the order of the first nonzero derivative of z(0 at Co. and f( 0 and its first derivative does no~ become zero or infinite at Co (see Section 14.14). ·At Co angles are not preserved but are multiplied by the factor II. In applicati~n to the airroil problem. the trailingc:dge point' C C" is a zero of dzldC. Hence the transformation may be expressed in the form (16.44) Now, if we choose only the zeros for tk/dC. as done in the case of the 10ukowski transformation, the other zero i. at the point
C" ... C. where C is a real positive number, we may write
;; +'''C . (C + C)'IT{2  II)
(16.51)
This is the simplest transformation that win produce a sharp trailing , edge with a finite angle. The trailingedge angle is given by
T"
In terms of this angle the transformation may be written as
C Cr
because the origin of coordiJiates is 19 be at the centroid of tile zeros. Then the transformation may also be expressed u
Z _ _ _ _ _ :; C
2" 
T
c
a _ e)'b,'/.
,(t + e)'I.,',.
T
T z+C
2". 
(16.52)
zeo  1('1') (()  z(h.4r)
(C + Cr) f{O
(16.45)
If the trailing edge is required to be a cusp, we set reduces to
"
= 0, whence (16.52)
From the relati,ons (16.44) and (16.45)we obtain
:t(C)  2({,,) _ (t  Cr)
z  2C' (C. C)I Z + 2C .... (' e)1
+
,(t +
C,,)
which is simply the Joukowski tral'lsformation as given previously by
(16.34). • r The transformation (16.52) is known as the KarmanTrefftz tranSlormation. It is a si~ple extension of the Joulcowski tran~formation so as to obtain a finit trailing edge. The airfoils generated by. It ~re known, as t~e Karm{mTrefftzfamily. A typical airfoil of this famlly)s shown In Fig. 16.16. The skeleton of the airfoil is made up of two Circular arCS. The
Setting z({r)  Zr and Z(Cr)  ~Zr, we.ha~
Z Z
Zr ' (' '
+ Zr
({
+ {T)
err,
+ ...
'
(16.46)
We may also txpress (16.46) in the series form:
Z
,.!LC + III1C,,'
nCr 3
C,
(16.47)
FIa 16.11 A KarmanTrefrtz airfoil.
490
IdealFluid Aerodynamics
The 'Problem of the Airfoil
where
491
properties of such airfoils may be readily de~ined by suitably specializing the general relations previously developed. In this connection reference may be made to Durand (1934). 16.12 1beoclonen's Method (or the Arbitrary Airf'oD To deal with the problem of an airfoil of arbitrary shape we need to determine the mappi!1g function that transforms the given airfoil profile into a circle. A method suggested by Theodorsen (1932) is here briefly described; full details can be found in Theodorsen. The mapping function is built up in two stages; in the first the airfoil ,profile in the z plane is mapped into a contour in the {' plane by employing the Joukowski transformation
p
=
1
(~r, (L)' 2C 2C
=:
This establishes the function 1p{O). Denote a point'on the image circle in the, plane by
, =:
CelPtet.
Ret.
(16.58)
where is a constant that is yet to be det~rmined. the correspO.nding points on the pseudocircle and the exact circle are related accordmg to (16.54). Tbus we have
"'0
z = {'
+ c· _ C'
Ce""'e(' ~CeIPOe
ffo
=: 
"
'
=:
exp
(~ ~1 {"
C,,)
+ iB" eIn.
R"
(16.59)
(16.53)
setting
If the airfoil is an ellipse. this transformation will map the airfoil into an exact circle. Hence. to have the image contour in the plane as close t,o a circle as possible. it is necessary to position the coordinate axes in the airfoil plane (i.e .• the z plane) such tbat the airfoil profile is as nearly elliptical as possible with respect to tbose axes; The contour in the " plane is referred to as the pseudocircle. The second stage consists of finding the mapping function that transforms the pseudo~circ1e into an exact circle in the {plar..e. Thoedorsen suggests the mapping C (16.54) {'= Cexp ")
C" A. _=
we obtain
+ iB"
=:
An
r
'"
",  "'0
o=:
'e"
.. 1
~ R" (A" cos ntf,
+ Bn sin ntf,)
A" sin ntf,)
(16.60)
tf, :::
.. 1 t R" (B" cos ntf, 
(16.61)
(i
where R
=:
Cert. a constant. We further have (16.62)
1 {"
where the coefficients C... complex in general, have to be determined. Denote a point Oil the pseudocircle by ,
(16.63)
C'
i6
=:
Ce""'et•
(16.55)
(16.64) These equations determine the tbree unknowns A ... and B~. ~o find them, however, we need to kn·ow 1p{tf,) or, equivalently, 8(tf,). ~blch!s what we wish to find. 1t follows that they n~ to be solved by Iteration (see Theodorsen) or by numerioal methods. For the details reference should
Note that Ce describes a circle in the plane which goes into a Joukowski airfoil in the z plane. Thus e1'(8) den,otes the deviation of the image contour in the {' plane from the Joukowski circle. The corresponding points on the airfoil and pseudocircle are related as follows, from (16.53). by x =: 2C cosh 'P cos 8
y
=:
r
"'0.
2C sinh", sin 8
be made to the literature.
(16.56) ( 16.57)
from which we find 2 sin' () =
P+
[pI + (~n~
2 sin' 'P«()
=:
_p + [p2 + (~)]YI
Elements of Thin Airfoil Theory
Chapter 17
In these relations G> .. G>(z, ,) is the velocity potential and F(z, ,) .. 0 describes the airfoil profile. Denoting by . . . q(z, ,) the Jler""bation or disturbance velOCity due to the airfoil, we write V(z, ,) _ V., + q(z, y)
(17.S)
Elements of Th~n Airfoil Theory
where V., as usual is the free stream velocity. Denoting the com~nents of • by '! and v, and the components of V by II, and ,II" ,we have
.II,(Z, , ) . .
Y., cos II
+ u(z, ,)
(17.6) (17.7)
"J.z,,) .. Y.o sin «+ f1(z,,) The problem of calculating the ftow field lpld the aerodynan* properties of any given.arbitrary airfoil with no restrictions as to its thickness, camber, or angle of attack is complex in practice. calculation could be dorie by using Theodorsen's method. This method, however, is not convenient either for a rapid estimation of the velocity or pressure distribution over the airfoil or for designing an airfoil profile that will have a prescribed surface distribution of velQCity or pressure. It thus appears profitable to have an alternative approach that will in some way simplify the mathematical conditions of the problem and enable the construction of a closed form solution (approximate naturally) for the problem. Such aD alterrlativC! approach is provided by the socalled thin airfoil theory. The elements or this theory are presented in thill chapter. Thin airfoil theory had its beginnings in the early days of aerodynamicS, some onbe original attempts being those of Ml,lnk (1922), Birnbaum (1923). and Glauert (1926).
.r
The
~. 6,tv.. •.
" cos.
_11.1 CoontiDatel aDd r.ymbols for the flow put an aidoiL'
17.1 Formulattoa of the Problem
ta Te;tIIS 01 die
Pertarlletloa FJeld
Consider the steady ftow past a fixed airfoil of arbitrary ~pe (Fig. 17.1). The ftowfiela is the solution of the following mathematical probleftt.
where «is the angle of attack (see Fig. 17.1). Introduce ~  t/J(z, y) as the potential for the pertutbation Ydoeit)'. We then have
Dijferentiill equation
...
YW"'
u(z, y) .. ~!
(1Ic1>
(17.8)
(17.9)
azl
iJIq" + .. 0
/JOUlfl}ary condition
a,.
(17.1) f1(z, y) ... (17.2) (17.3) and grad G> ... V.,
it
t/>
grad ~ . grad F .. 0 or F(z, ,) = 0
+ grad t/>
(17.10)
Infinity condition
grad G> .. V., at infinity
G>(z, ,) ... (zV., cos II + 'V... sin ex) +
In terms of the
perturbatio~field, the above mathematical problem
VI.L _O't/> _
'f'
Klllla condition
The circulation around the airfoil is such that the velocity is finite and Continuous at '!be uai1iDa ec1&c (17,4)
takes the following form:
Differential equation
+~ "
azl
iJyl
a ... 0
l
(17.11)
491
494
IdealFluid Aerodynamics
EIements of Thin. Airfoil Theory
We observe that v would be a known linear function of but for the presence of u in the first term of the righthand side of ~1 !.20~. In this sense this equation is said to express the boundary condition m a non.. linear f o r m . . To simplify the equation (17.20), we ~ntroduce ~n ass~,?ptiOn8. We assume that the airfoil is sufficiently thin and elongatrd (that It.IS only II small deviation from its chord line) and that the angle ofattack is~fficiently .
BoflllliDry condition
(V'"
tlg,.
+ grad ,;) . grad F 
0 or F(%, 11)  0
(17.12)
(17.13)
Infinity condition
components of grad ,;  0 at infinity
KUlla condition
The circulatior. around the airfoil should be such that the perturbation velocity q  grad", should be finite and continuous at the trailing edge (17.14) As we knpw, it is'r.ther difficult to solve this direct problem. To make some progress with the construction of the solution we simplify the problem by introducing the approximation of small perturbations. With such an approximation the boundary condition may be redllcedto a simpler form than (17.12) and transfeiTed to the axis (chord line) of the airfoil. Then the problem itself may, be represented as a superposition of three other simpler problems. We now proceed with the development of such a scheme.
small so that the perturbation velocities Uand v are small compared to V",:
II
« V""
0%
v« V.,.
011
The shape of the airfoil is such that
of < of
In other words, the ~ormal to the airfoil curve is practically normal to the chord line. With these assumption~ we set
V", cos
(It ... 
Villi
Vooat
17.1 SimpUftcadon or the BoIIDdary CODdItIoa
Consider the boundary condition as ex~ by (17.12). For convenience let us temporarily work in terms of u and v instead of';" In component form (17.12) appears as
(V", cos IX
Y.., sin (It
and neglect U of/h  u dnltk in comparison with <»th~r terms appearing in the boundary condition. Equation 17.20 then becomes
4'1 v[%, '1(z)]  Yeo 4%  Voo~
(17.21)
. of + u) of + (V", 810 at + v) 011 () %
0 on F(%,.1I) = 0 (17.15)
This simplified relation is sometimes known as the. linearized form of the
bountlary condition.
We immediately obscrvethat the assumption of small perturbations is not valid at and near the stagnation points occ:urrin.l on the airfoil for in such regions the perturbation velocity is of the same order of magnitude as the undisturbed velocity. We further note that ifthe airfoil 'has a rounded leading edge, the assmnption that oFIOz is much less than oFLOJL is violated ~..and near ..that eci.gc. On the basis of these considerations, we conclude that the simplification we made cives rise inherently to certain rqions' where the solution of theresultins problem is not yalid. Fortunately for applications such regions are quite small and the solution of the simpler problem is remarkably useful.
Suppose the airfoil curve is described hy
11  '1(%)  "1u(%) for the upper I.arve  "11(%) for the lower curve
Then we have
F(%,1I} . '1(%)  11
(17.16) (17.17) (17.18) (17.19)
== 0
of
0%
4'1 d%
on 11  "1(%), 0
of
 ... 1 ()y
on substituting (17.1'8) and (17.19) into (17.15), we obtain
(0) (V", cos IX
17.3 Transfer or the Boundary Coaditioa
(17.20)
+ u)...!l. 4%
4
V", sin at
~ %~ I
Although (1.7.21) is linear,lts application is rather inconvenicrpt·(or, the value of the velocity component has to be computed at points on the airfoil. In view of the assumptions we have already made, this inconveni'ence
IdealFluid AerodynamiQl
Elements of Thin Airfoil Thc:ory
49t
may be removed as follows. Assuming that v(x, y) may be expanded in a Taylor Series about points on the chord line, that is on the Xaxis, 0 S; x S; I we write
v(x,1J(x)] = vex, 0)
, ,
,
+ (£!:)
1J(x) iJy z.o
+ .. ",
0 S; x S; I
Fill. 17.2 Flow past the airfoil at zero angle of attack.
Let~l(X, y) and ~I(X, y), respectively, two problems:
Consistent with our previous approximations, one may neglect (iJv/ iJy)., 01J and all the other higher terms. We may therefore set .
v[x,1J{x)] =v(x, 0),
Pc the solutions of the 'following
0 S; x S; I
(17.22)
On our making this approximation ...(11.21) becomes
d1J vex, 0) == Vco   V Gt, ..
_ d'l ( a~l).,.0 Vco dx iJy
ox oy
2.
VI~I
dx
0 S;
X
S; I
(17,23)
iJ~l, O~l + 0 at inn.ity
0~1  ,O~l 
In view of (17.16) we clarify (17.13) by writing it explicitly as
o(x, 0+) = V.. d1Ju  v.. Gt} dx. d1J1 · ~(x, O = Vco dx  VcoGt )
ox ' oy
=0
are finite and continuous at the trailing edge
OS;xS;1
(17.24)
whefe 0+ and 0_, respectively, denote the tGp and bottom sides of the Xaxis io the strip 0 S; x ::;: I. This transfer of the boundary condition as exhibited by (17.24) is possible only i/v(x, y) can be expanded in a Taylor Series about the chord line. We htzve assumed that it is, but ollr assumption can be checked only a posteriori. 17."Fnmc Work of tbe Theory of the Thin Airfoil Introducing the simplified boundary condition and its transfer to the 'axis of the airfoil,we express the mathematical problem governing thc perturbation field as
(17.25) (17.26)(17.27)
o,p.
ox 'oy
O~I + 0 at infinity
The circulation being such that
0"'. ,O~I are finite and continuous at trailing edge ox oy ,
Then. as may be verified, ~l +c/l. is a solution of the problem for~. We thus conclude that the solution for ~ may be obtained as the superposition of the solutions ~l and ~. of the respective problems (l) and (2). Problem 1 represents, as shown in Fig. 17.2, the steadyftow past the given airfoil at zero angle of attack, the speed of the free stream ht:ing V... Problem 2 represents, as shown in Fig. 1'7.3, the steady ftow past a ftat plate at an angle of attack equal to Gt.
The circulation is such. that
a~ o~ Ox an d Oy are fi' and ' mte contmuous at t he tral'1' ed ge ( I7 28) 109 .
• R.call the explicit meaning of this fOml which we use for convenience (see 17.24).
7
~
o
Fig. 17.3 Flow past a flat plate at an angle of attack.
IdealFluid Aerodynam;cs Problem I may further be represented as the superpositior. of two other ptoblems. To show this, let us introduce the socalled camber function defined by , "Ie = i('1u and the thickness function defined by We then have
(17.30)
Elements of Thin Airfoil Theory the plu$ sign corresponding to y = 0+ and tfte minus sign corresponding to y = 0_. The term V~(d7J./dz) represents the boundary condition for the flow past an airfoil that has only camber and no thickness, the airfoil being at zero angie of at,tack. The term ± V ~(d7J,/dz) represents the boundary condition for the flow past an airfoil that haa only thickness and no camber (i.e., past a symmetrical airfoil), the .irfoil being at zero angle of atack. We thus conclude that the solution for the .flow past QII arbitrary thin airfoil at zero QIIgle of attack may be obtained as rite superposition of two solutions, one for a symmetrical airfoil at zero QIIgle of attack and the other for ~ cambered airfoil at zero angle of attack. The profiles of these latter airfoils are related to that of the given airfoil through
+ '11)
(17.29)
, 'lu == 7Je + '1, '11= '7e  '1,
The boundary condition fpr
(17.31)
1>l: thus takes the form
dx dx
for
for
( 01)1\
oy!z..
=v~(d7Je+d7J')
=
y=O+ y = 0_
E;.
v~(d7Je _
dx
d'1l) dx
This we may express in turn as
FIa.17.5 Showing an alternate decomposition of the problem of an arbitrary thin airfoil.
( 01)1 \
OY!e.O±
== V~ d7Je ± +'~ d7J1
dx dx
(17.32)
,
I

v.
(17.29) and (17.30). It thus follows that the solution for the flow past an arbitrary thin airfoil at a nonzero angle of attack may be obtained as the superposition of the solutions for the three simpler problems represented SChematically in Fig. 17.4. Alternatively, as shown in Fig. n.s, we may represe~t the problem as the superyosition of th~ problem for symmetrical airfoil at zero angle C!f attack and that for a camDered airfoil with zero tbickness at an angle of attack. The solution of these problems may be attemp~ed by the method of complex variables. However, we shall not follow this procedure. In the followjng we construct the solutions of the three problems by the 'method of superpositIon of singular solutions or Singularities. Such a method finds application alsO'in the problems of flow past a finite wing and a body of revolution. First, using the approximations of the present simple theory, we obtain a relation for the pressure.
a
+
JIll. 1704 DJustrating the decomposition of the problem of an arbitrary thin airfoil into' ' three simpler problems.
17.5 Pressure Relation in the
S~ple
Theory
The pressure at any P9int is given by the Bernoulli equation
500
In terms of the perturbation q we have VI
IdealFluid Aerodynamics
EIemenB of Thin Airfoil Theory
y
501
= (V", + q) .. (V"" + q) =
=
+ 2V", • q "" q2 V~I + 2(V",ucos ex + V",vsin ex) + u2 + Vi
V", 2 _ IpV
I
(17.33)
The Bernoulli equation then takes the form
p
= p 00
""
00
1(2 V", • q + L) V V
2 2
p = P   V
2 ""
2
(u cos ex + 2 2 V",
""
'"
V
V",
sin «
+ u + Vi) 2
(17.34)
.v""
•
~~~~~~X
V",2
Fig. 17.6 Coordinates for the problem of the symmetric airfoil at zero angle of attack.
Consistent with the approximations made in the simplification of the boundary condition,. we set cos ex = I, sin ex = Cl and neglect those terms in the parenthesis that are smaller than uIV",. We then obtain
p = p""  pV",u
that over the strip 0 ~ x ~ 1 the velocity component v be an odd function ofy:
=
Poo  pVoo
oc/J
ax
The correspondil)g pressure coefficient is given by
p  p"" C" == pV!/2
= 2
u.,. 2 oc/J = .Voo Voo
ax
(17.35)
Equation!! 17.34 or 17 .35 e~press the pressure as a linear function of the velocity component u. This means that when we solve for u by superposition of separate solutions, we may ob~ain the pressure by superposition of the pressures corresponding to the separate solutions. Similarly, the forces and moments on the airfoil may be obtained by superposition. 17.6 Symmetrical AiFfoil at Zero Angle of Attack: Solution by Sourc~ Distribution We first consider the problem of a thin symmetrical airfoil at zero angle of attack (see Fig. 17.6) .. The mathematical problem for the disturbance potential c/J is the following:
These considerations suggest that the disturbance field may be represented as that due to a ~itable distributio.n of sources along the X~axis in the range 0 ~ x ~ I. The velocity field due to such a distribution naturally satisfies Laplace's equation (17.36) andlhe infinity condition (17.38). The problem of the symmetrical airfoil at zero angle of attack thus reduces to that of the superposition of the unifoim stream and a certain source , distribution, as represented in Fig. 17.7. Once the source distribution is determined, all the desired information about the Bow field is readily obtained. Let q = q(x) denote the intensity
y
.
1
v..
)
V1c/J
=
°
at/>
07.36)
(x, O±)
vex, O±) = 
oy
. = ± Voo
 , dx
d"ll
q(:c)
(17.37)
':.
0
Vc/J
=0
C"
(17.38)
~
at infinity
t
(x~)
Since the flow field is symmetrical about the airfoil, there is no circulation around the airfoil. The surface condition, Eq. (17.37), requires
Fig. 17.7 Source distribution for the symmetriC airfoil.
Ideal:Fluid Aerodynamics of lAC source distribution ;,. q is thus the sou~ce strength per unit length. Then the perturbation potential and the perturbation velocity components at the field point (x, t/) arc given by 1 ¢(x, g)  217'
Elements or
Thin
Airfoil Theory
SDJ
strength q(x) tIz. Hence choosing a circuit. C. as shown in Fig. 17,8. we have
q(x) dx 
11
0 0
£
V· D ds·
q(E) log [(x .
E)' +g'1~
dE
(17.39) (17.40) (17.41)
y
_ lim
+w ...... and
ic
i.
V • D ds
u(x, g) 
217'
1 LI X  E 0 q(E).(x _ ·E)":""+gI dE 1
(17.45)
t(x,1/) == 2
17'
Llq(~)( x. E~' + g1 dE
To determine the source distribution q(x) we use the boundary condition  (17.37). Consequently, we require that
C
1+1
11
O~x~l
x
(17.42)
Fil. 17.1 Outflow through a circuit enclosing an element of the soun:e distributioa.
Care is necessary in evaluating tholefthand.)ide of this equation. We observe that for E ~ x, the integral goes to zero as 11 goes to zero. For E  x, the integrand becomes indeterminate. Theaetailcd evaluation of the lefthand side of (17.42) IS given later in Appendix E. As shown there, lV' find that
lim J..L'q(E) , .. o± 217' 0 (x 
Here D is the outward normal and t:!s is an clement of length along the . circuit. Since vex, 0'_) == vex, 0+), Eq. (17.45) becomes
q(x) ±2v(x,0±)
E)' + g.
g
dE _
± q(x)
2
(11.43)
Equations 17.39 to 17.41 giving the perturbation potential and the perturbation velocity components now take the form
¢(x, 1/) ==
On substituting (17.43) into (17.42) we obtain
. q(x)  2v(x, 0+) dfJI ==2'v..,dx
.
...!It
17'
V
Lib
0
(17.44)
dz
_'II (E) Jog [(x 
.
E)' + gl),.. dE
IL
(17.46) (17.47) (11.48)
U(x,1/) _ V.., (' dfJI(E)
17'
This equation determines the source distribution. It shows !hat the source strength at the point x is proportional to the slope of the airfoil at that point. The result expressed by (17.44) could have been obtained by an elementary lI,fgument: Consider an element q(x) dx of the source distribution. We khow that the outflew of fluid through any circuit enclosing that element and passing through its endpoints is equal to the source
• Note that the symbol q was used previously to denote the perturbation velocity. In that context q meant the magnitude of q. It is hoped that the present use of q to denote 90urce strength will cause no confusion.
v( x
,11 
)  V..,
17'
l'
Jo
dx
x  ~ de (x  e)1 + 1/t
dfJ1(e) 11 0 dx (x  e)1
+ 1/1
de
The pressure distribution is $iven by
CII(x, y) == 2
U(~ y)
..,
On the surface of the airfoil, the xcomponent of the perturbation
504
Elements of Thin Airfoil 1lteory IdealFl\lid Aerodynamics Introduce further .Ifte notation
u(O)
(17.49)
50S
velocity and the pressure are given by
u(~, 0) = Vao (' dfJ.(~) _._1_ dE 11' Jo d:r: ~.;... E
C (~ 0 ) = 2 u(i~ 0) P , ± V ao
= u[~{O), OJ
dfJ. (0) = dfJl [~(O)J d~ dz
(l7.50)
The ycomponent of the velocity is, of course, given by the surface condition
v(~,
O±)
dfJ. = ± Vao dz
It readily follows that there is neither a resultant force nor a moment on the symmetrical airfoil. We .thus conclude thafthe thickness (If a thir;i airfoil contributes only to the pressure distribution over the ~irfoil. To obtain the velocity compOnent u and the pressure distribution over the airfoil we need to evaluate the integral on the righthand side of (17.49). We observe that this. integral is an improper integral· in the sense that the integrand becomes infinite at the point E = ~, ~. being in the interval of integration. The value of the in~egtal is given by its socalled principal or Cauchy principal value. The principal value of an improper integral
o ~======:::t=:::::~::::::~=t,~.~ x
where/eEl becomes•. infinite at a point defined by the limit
E= C
in the interval (a, b) is
Fig. 17.9 Introducing the variable e.
J.
iCE) dE =
lim [I.c/(E) dE
...... •
II
+ (t feE) d$] Jc+.
For the numerical colDputationof u(~, 0), from {I 7.49), it is convenient to express the functions u(~, 0) and (~.ldx)(J) as conjugate Fourier Series (see Appendix D). For this purpose we introduce the variable 0 such that I (17.51) ~ =  (1 + cos 0) . 2
(see Fig. 17.9.) Positive values of 0 describe the top of the airfoil, whereas negative values of 0 describe its bottom surface. The variable Emay then be expressed as 1 (17.52) ~ =  (1 + cos rp) 2
• For the notion of improper integrals and for certain elementary properUes. reference may be made to Tom M. Apostol (1961).
~pects
Then, in terms of the variables 0 and rp, (17.49)
u(0) = V,., 11'
~kes
the form
(17.53)
i·
0
dT/t () rp
d~·
sin rp d rp cos rp  cos 0
This is known as Poisson's integral formula. The integrand in the integral on the righthand side becomes infinite, as it should, at rp = O. On the basis of the detailed considerations give'! later in Appendix D, it follows that [u(O)/V",J and (dT}t/dx)(O) are expressible as conjugate Fourier series. We observe that (dT},/dx)(O) is an odd function of 8. Hence we express it as
d'lt(8) = dx
of their
"~l
i
A, sin n8
( 17.54)
r06
IdealFluid Aerodynamics
Elcn;:!nt!o of Thin Airfoil Theory
}'
567
Nhere
A"  ·2 
d'1 '(8) sin n8 d8 .". • dz
i
r
(17.55)
Then [u(9)/VGO] is given by the conjugate Fourier series
1
(17,56)
"(8) Y GO
== 1;A 1
GO
cosn8
It
I
\
I
where the Coe~cients A" are obtained.from (17.55). As required, u(8) is an even function of 8. To treat thin airfoil theory, computational techniques for determining conjugate Fourier series, and also their derivatives and integrals, have been developed, and for these the reader should consult the approp~ate references. • . From.(17.47) and (17.48) we infer, as shall be,shown in detai11ater (see Appendix B), that the perturbation velocity field exhibits the following . features: 1. The velocity components " and v are finite everywhere outside the strip y ,.. 0, 0 ~ ~ ::s; I. They vanish as , .. JzI + 11 goes to infinity. . 2. On the XaxIs, for. z < 0 and ~ > I, the component" is finite while v IS zero. , ,3. Al~ng the strip, excepting the edges, the" component is finite. Further, ItlS contmuous across the strip. At the edges, for any nonzero value of the source strength, " becomes infinite. 4. Along the strip the vcomponent i~ finite at all points where the so~ s~ngth is finite. At any point on the strip, the jump in v across the stnp IS SImply equal to the source strength at that point. Sinc:e at a round ~ge. the source strength become~ infinite, ,the vcomponent becomes mfimte at a round edge, that is, at the leading edge of the airfoil. As .expected, we 'observe that the'solution of the li~earlUd problem is not vahd at and near the leading and trailing edges of the airfoil. .
+J'
v...,
o
~=IJ=c(=")=t======_:::=:..... __==____~.,.
I
X
Fig, 17.10 Coordinates for the problem of the cambered airfoil,
Kufla Condition: Now. the circulatio'n around the airfoil is not zero.
lts value should be such that the perturbation velocity components u = ()4>lox and t· = o4>loy should be finite and continuous at the trailing edge.
These requirements, particularly the surface condition (17.58), suggest that the dist".rbance field may be represented as that due to a suitable clstribution of vortices along the Xaxis in the interval 0 ::s; z ::s; I. We distribute the elementary vortices as shown in Fig. 17.11. the circulation around each of the vortices being designated in the clockwise direction. The velocity field due to such a distribution of vortices of any finite strength automati,;ally satisfies Laplace's equation (17.57) and the infinity
17.7 CamberM Airfoil of Zero 1bkkMss at Zero SoIutioD by Vortex DistributIon
ADele
of Attack:
Now the mathematical problem for the disturbance potential", is the following (see Fig. 17.10): VI", .. 0 (17.57)
.. .,
~
... ~
.~.~
z O~ _ V d'1c , . GO dz '
V", ... 0 at infinity
• See, for inItaDce, Thwaitca (1960), where other retcn:nces may be round
(17.53) (17.59)
508
IdealFluid Aerodynamics
Elements of Thin Airfoil Theory
condition (17.59). The strength of the vortices should, however, be distributed in such a way that the surface condition (17.58) and the Kutta condition are satisfied. This, then, is the problem of the cambered airfoil of zero thickness. Once the vortex distdbution is determined, all the desired information about the flow field is readily obtained. Let y = rex) denote the intensity of the distribution; y is thus the vortex strt.ngth per unit length (see Fig. 17.11). 'We note that y dx is the circulation around the vortices distributed in the element of length th. The perturbation potential and the perturbation velocity components at the field point (x, y) are then given by
Compare with (17.43). Therefore we have
rex) = 2u(x,0+) = u(x,O+)  u(x,O_)
(17.63)
'since u(x, 0_) = u(x,O+). Using the relation (17.63) we may immediately express the pressure distribution over the airfoil and the lift and moment on the airfoil in tctrqas of the vortex st;engthy(x). For the pressure distribution over the airfoil we have
c (x, 0) = ±
p
2 u(x, O±)
V
«>
=~
y(x) V
«>
(17.64)
t/i..x, y) =
u(x, y) =
 1. 211'
1 ("
i'
0
0
y(E) tan1  y  dE
x E
The lift on the airfoil is given by (17.60)
L= f[p(X,O)  p(x,O+)]dx
211' Jo y(E) (x _
~, + y' dE
xE·
(17.61)
=
v(x, y) =  2'17'
Ii'
E V""Zi'[Cp(X,O_)  Cp(x,O+)]dx 2 0
r(E) (x _ E)'+ y'dE
(17.62)
= pV""
'fherpressure at any point is given by where where u(x, y) is to be obtained from Eq. (17.6i). We observe that (17.61) and (17.62) are respectively similar to (17.41) and (11.40), which refer to a source distribution along the,strip y =' 0, o ~ x ~ I. We may, therefore, conclude that the u component of the velocity due to the vortex distribution exhibits the same characteristics as the vcomponent of the velocity due to the source distribution and, that the vcomponen! due to the vortex distribution exhibits· the same characteristics as the ucomponent due to the source distribution. See the preceding section for detailed description. In part~cular, we note that the circulation at any point on the distribution is simply equal to the jump at that point in the ucomponent of the velocity across the vortex sheet. This follows from the result
. u(x,O±) = hm  1
vo±
i
',,(x) dx
(17.65)
=
pVx,r
r
=
on the airfoil with respect to the origin (which here is the leading edge) is given by
M = i'[P(X, 0+)  p(x, O_)]x dx
Jo
f' y(x) dx
is the circulation around the airfoil. The moment
= pV"" i'y(X)X dx
(17.66)
For the lift and moment coefficients we obtain
cL
.If
2 1 = V"" I
i'
0
y(x)dx
(17.67) (17.68)
2'17'
i'
0
c =_1..!.
V""I'Jo'
('v(x):r dx
y(~)
Yz z d~ (x ~) + y
=±,
2
y(x)
We now consider the question of determining y(x). According to the surface condition (1758) we require that
O~x~l
'. EXcept for a ncgatiw: sign.
$10
'dealFluid Aerod~
Elements of Thin Airfoil Theory
511
or
 1
f.' .
In view of (17~74) we conclude that if we choose
v(~,0
211.
·1 ,,(E) dE Z  .E
df'J \ _ V. _ 0
GO
%J
dz '
O:S; z:S; ,. (17.69)
K ,,(6)  sin 6
(17.75)
In evaluating. the lefthaQd side of this equation. we naturally seek the Cauchy principal value of the integral. Equation (17.69) must, of course, be supplemented by the Kutta condition. We find, as is shown in detail later (see Appendix B), that for a .vortex dIStribution such as we are considering, the • velocil)1 component becomes infinite at the edges of the distribution if the vortex, strength althose edges has a nonzero value (similar to the behavior of U due to a so~ distribution). The ucomponent is, however, finite at the edges iftlte vortex strength there is finite. From this· it follows that to satisfy the KUlla condition we should require that the vortex strength at the trailing edge be zero. We thus have y(z l)  0
(17.70)
where K is a constant, the integral I will becGoe zero. In terms of the original variable z, (17.75) becomes
(17.76)
On the basis of these considerations it follows that the solution of (17.69) is indeterminate to the extent of the function represenied by (17.76) or, equivalently, by (17.75). To proceed to the solution of (17.69) we introduce, as before, the variables 6 and tp ~nd rewrite the equation as
The. mathematica~ problem for determining y(z) thus consists of solving the IDtegral equation (17.69) with the condition (17.70). First we show that the solution of (17.69) is indeterminate to the extent of a ce~ain function. To see this and determine that function, consider the integral
I == (' r(E)
:! fW y(.p)
11
0
sin tp . dtp 2VGO cos tp.  cos 6
dfIll (8) _ d'1e [%(6)]
== d'1e (6)
dz
(17.77)
where
J.
2
...!!L
dz
dz
z E
(17.71)
Introducing the variables 8 and tp such that
I z   (1
Equation (17.77) is again Poisson's integral formula (compare with Eq. (17.53». On the basis of the detailed considerations given later in Appendix
+ cos 8) + cos tp)
0, it follows that y(8)/2VGO and (tlrJ./dx)(O) may be expressed as ~onjugate Fourier series.· Since (tlrJ./tk)(8) is an even function. of 6, let us write
E == 2(1
we rewrite. the integ.al as
1
I
2l! (6) =
dx
W
d
J
B
2
+I
GO
.. 1
B" cos n6
(17.78)
where
2 B=" 11
i
0
== 
i
r
r
df'J ccosn8d8 dx
n
==..0, 1,2, ...
07.79)
o
y(tp)
sin tp . dtp costpcos6
(17.12)
With the use ofexpression (17.78), (17.77) becomes
Now, on the basis orthe formula
i
r
. ci
cos ntp d' q;' = cos 9J  cos 0
11
sin n8 sin 0 '
1 (r y( tp) sin tp dtp = J 2V cos tp  cos 8
11
0
<Xl
Bo
+
2
1: Bn cos nO
1
(17.80)
n
== 0, 1,2,. ,.
ICSUll
(17.73)
which wilL be proved later (see Appendix E), we have the
Since this equation is linear in )' and its solution is indeterminate to the extent of the function (17.75), we may e'xpress the solution as
1'(8)
f.
.!.. == 0 o cos q,'  cos 0
dq>
(17.74)
== Yl(8) + )'2(8) + !5(jSID
(17.81)
511
IdealFluid Aerodynamics
Elements of Thin Ailfoil Theory
511
where 1'1 and Y2 are, respectively, the solutions of the equations:
If'y\(IP) 17 0 21/"" cos
Sln({! ({! 
cos 8
({!
d rp = ~ ~B cos n8
1
n
07.82) (I7.83)
1. J yke)
r To
which vanishes for 0 = O. In (17.87) we substitute for K from (17.90) and obtain B 1  co~ () ~ . ) (17.92) y(0)=2V",j. +~B"smnO ( 2 sm 0 1 Using this relation we may express all the desired aerodynamic properties of the airfoil in. terms of the B's, which are the coefficients of the Fourier expansion for (dllc/dx). The pressure ciistribution over the airfoil is given by yeO) C.(O) = C.[x(O).O±] =  V",
sin
({! 
0
2 V" cos
cos 8
d({! = Bo 2
The solution of (17.82) is given by the conjugate Fourier series· (see Appendix D)
/'1(0) = 2 V",
"0
I
00
Bn sin nO
(17.84 )
To obtain the solution of (17.83) we first observe that on the basis of the formula (17.73) we have the result
Bo = 2[ 2
1 cos 0 + ~ B"sm n8J .
sin 0
~
1
(17.93)
!. (r
17
J
cos cos
0
cp 
cp
cos
0
dcp
= I
(17.85)
To obtain the lift and moment coefficients we first rewrite (17.67) and (17.68) in terms of the variable 0:
CL
From this it follows that if we set
= 1
YI(
0) = V B cos 0 "" 0 sm 0 •
V""
·rrY .6) sin 0 d8
.0
(17.94) (17.95)
(17.86 )
eM =
 
(l7.83) is identically satisfied. Thus the function (17.86) is the required solution for Y2' Substituting the expressions (17.84) and (17.86) into (17.81), we obtain
2V..,
r
i
r
y(0)(1
+ cos 0) sin 0 dO
0
Substituting (17.92) into (17.94) and (17.95), and carrying out. the integrations, we obtain (17.96)
C M = B04 'IT 'IT 'IT + B12 + B.4
"" y«J) = 2 V"" I Bn sin n8
1
VB(K + ~   + cos 0 )
sm 8 V""Bo
(17.87)
To determine the constant K we use the Kutta condition (17.70). Accordingly, we require tha't (17.88) 1'(0;= 0) = 0 This leads to the condition
K  1 (  ·+ cos 0) = 0 sin 0 . V",Bo
= !!. (Bo
4
+ B 1 ) + ~ (B1
4
17
4 B 2)
(17.97)
C =  4L + 4 (B1 + B2)
where, to recall,
for
0= 0
(17.89)
This is satisfied by setting
.K = V",Bo
2 B" = 'IT
fr .!k
0
d (8) cos nO dO, dx
n = 0, 1,2, ...
(17.90)
For we have cos 0  1 sin 8
'1 sin 8/2 cos 0/2
sin
":'2 sin 8/2 ' =
2
8 tan2
(17.91 )
• This may be veri tied by substituting the series into (17.28) and notin'g that
1 7T
J"
0
ncp sin cp
cos 4>  cos '"
"'" = cos /If)
The relations (17.96) and (17.97) show that the force system acting on the airfoil may be represented as 'consisting of a lift force acting at the quarterchord point from the leading edge and a moment about that point. We have thus completed, for a cambered airfoil of zero thickness at zero angle of attack, the task of expressin& in terms of the airfoil shape all the desired aerodynamic ~nformation. We now' pass on to cor.Slder the problem of a flat plate at an angle of attack.
114
IdealFluid Aerodynamics
Elements of Thin Airfoil Theory Then (17.98) takes the form
515
17.8 Flat Pilite Airfoil at an Angle of Attack: Solution by Vortex Distribution The mathematical problem for the disturbance potential is expressed as follows (see Fig. \7.12): \24> = 0
v(x,O~) =
!.
1T
J
C·Y(!p)
0
sin If d!p == 2V..,Cl cos II'  cos 8
(17.100)
oc/> oy (x, O±) =
 V",<x
In view of the fact that the solution ofthis equation is indeterminate to the extent of the function K/sin 8 (scc 17.75) and because of the result expressed by (17.85), the solution of (l7.IOO) may be written as
Yc/> = 0 at infinity
· K utla Condilion: T he clrcuI ' at IOn be suc h t hat at the trailing edge.
y
ot!> oc/> a and a are finite
x
y
y(8)
== . SID
K
8
cos 8 2V..,Cl.sm 8
==
2V O! 1 ( K . .., sin 8 2V..,0!
= + cos 8)
(17.101)
To satisfy the condition (17.99), the constant K should be set equal to 2 V.., IX. We then obtain 1  cos 8 (17.102) ,,(8) == 2V..,1X . 8 sm This is the required solution. . ' The pressure distribution over the ftat plate IS then given by
"((.I)
~~~~P~~~O~@H@~Q~@~@~Qr.Q~9,Q~Q~ ~O~..x O
C.(8) == Cp [z(8), 0%)
o
I
== .... 1':8) ..,
... ..,..2«1  cos 8
'sin 8
FiB. 17.11 Coordinates and vortex distribution for the problem of the flat plate.
_ 2« tan ~
(17.103)
Following !limilar considerations as those given in the previous sectiori, we immediately conclude that the: disturhance ·field due to the flat plate may be represented as that due to a suitable vortex distribution y(x) along the Xaxis in the interval 0 :S;.x :S; I and that y(x) is given as a solution of the integral equation
21T
I
2
The lift and moment coefficients arc given by
CL ==  1 V..,
i·
0
1'(8) sin (J d8
(17.104)
l'
0
y(~)  X ~
d~
= V(X,O±)
y(x
==
== 21TCl
VacQ(
(17.98)
with the Kutla condition
I ntroduce. as before, the
C M == _1_ 2 V..,
(17.99) the relations
== I) = 0 variables 0 and cr through
.r
~.
Jo
r· 1'(8)(1 + cos 8) sin (J d~
(17.105)
==
1T IX
2
=
=
I
I
~ (I
+ cos 0)
+ cos r: )
== ...: C L
4
•
~
(I
These relations (17.104) and (1;.105) show that the force system on the flat plate m~y be represented as consisting of only a lift force acting at the
516
IdealFluid Aerodynamics
Elements of Thin Airfoil Theory
517
quarterchord point. This point is thus the aerodynamic center. In addition, it is the center of pressure· at all angles of attack. 17.9 Aerodynamic Characteristics of a 1bio Airfoil We now su.pe.rimpose the results of the previous three sections to obtain the characte~lstJcs of an arbitrary thin airfoil that has both thickness arid ~:ber and IS at an angle of attack to the undisturbed stream. Recalling
d71* dz
=:
and is equal to 211:
dC I • = 211 da.
(17,110)
As we know, tl}e thickness contributes nothing to the lift. The camber fixes the lift at zero angle of attack and the zerolift angle. 'From Eq. (17.109), the zerolift angle is obtained as
ot,,=
d71. dz
±
Bo
+ Bl
2
d71, dz'
'F
(17.111)
we write in terms of the variable 0 [defined by x
d71 (0) dx
(//2)(1
+ cos 0)]
(17 0 .1 6)
The moment coefficient (with respect to the leading edge) is given by
= d71. (0) + dz (0) d71, dz
11 .,. 8 .,. 11
~
~
The top of the. air.foil corresponds to positive values of 0, whereas the bottom of the aIrfoIl corresponds to negative values of 0. Introd ucmg th e . . . F ouner expansIOns already employed we have
d (0) = 2
= _ CL
4
+ :!!. (Bl + B.)
4
(17.112)
d71
. Bo
z
+! B" cos nO + !1 1
co
co
An sin nO
(17.107)
where Bn
2 =  fW d71X• (0) cos nO dO d 11
0
n = 0, 1,2, ...
2 An = ::11
r ' J d'1 (0) sin nO dO dz
w
n = 0, 1,2, ...
As .we know, the thickness contributes nothing to the moment on the airfoil. From (17.112) it follows that the force system acting on the airfoil may be represented as consisting of a lift force acting at the quarterchord point and a moment equal to (11/4)(B1 + BJ about that point. This moment is independent of the angle of attack and is determined solely by the camber of the airfoil. It follows that the quarterchord point is the aerodynamic center. Thus the moment about the aerodynamic center is given by
0
The pressure distribution on the airfoil is given by
C (0) = 2 u(O) .. V co
=
eM = :!!.(Bl + B.) 4
= the moment coefficient due to camber
A$ we know, it is also the moment coefficient at zero lift.
(17.113)
20:tan~ + 2[~Otan~ + ~BnSinnOJ 2~A .. cosnO
=2[(0: 
~o) tan~  ~BnsinnO + ~AnCo~noJ
2110:  (Bo
(17.)08)
The lift coefficient for the airfoil is given by
CL
=
+ B 1)11 = 27T( 0: _
Bo : B!)
(17.109)
The slope of the lift curve with respect to the angle of attack is a constant
Some Features of Flow with Vorticity
519
Chapter 18
Some Features of Flow with Vorticity
Having presented the elements Qfthe theory oflift for the soctllled infinite wing, we now wish to take up the elements of the theory of flow past a finite wing. In the formulation of the theory of the finite wing, the disturbance flow field will be represented as that due to a certain spatial distribution of vorticity. Thus, as a preliminary step in the formulation and analysis of the finitewing problem, we must study some properties of vortex motion and, in particular, must learn how to represent the velocity field. We do so in this chapter. We restrict ourselves only to those features of vortex motion that are pertinent to the study of the flow past a finite wing. For a more detailed study of vortex motion and of the varied problems involving VQrtex motion, the reader should consult other books cited at the end of this book. 18.1 Recapitulatioa
Fla. 18.1 Vortex line.
18.1 Vortex Line, Surface, Tube, and FOameot The field lines of the vorticity field are called vorl,ex lines. Analytically they are described by t~e differential equation
n)( cis =
0
(IS.4)
where cis is an element of a vortex line. A vortex line is ~epr~seh ~ as shown in Fig. IS.1. At any point in the ilow field: the dlrec~lon. of the vorticity vector (or equivalently of the ang~lar vel~lty vector) IS given, by the direction, at that point, of the vortex hne passmg through that pomt. In cartesians, if we write
n
= (0,..0".0.)
dy dz
Eq. (IS.4) becomes
dx
We first recall some of the kinematical notions already introduced in Chapter 9. If V = V(r, I) is the velocity field of a fluid in motion, the angular velOcity w == w(r, I) is given by
0,. = OM = O.
W 1curl V
The vorticity n = O(r, I) is simply the curl of the velocity
(IS.I)
The circulation
rt'=
£.
n=
curl V
(IS.2)
V • cis around any closed curve
~ is related
to the
vorticity by the equation
r ==
v
f
V • cis ,.
==
II II n .
8
8
curl V • adS
DdS
curve~.
(lS.3)
Fig.18.2 Vortex surface.
where S is any (capping) surface whose boundary is the 518
520
IdealFluid Aerodynamics
Some Features of Flow with Vorticity
521
I f. at any instant of time. we draw an arbitrary line in the flow field and draw the vortex lines passing through that line, a surface is formed. Such a surface is called 3 l'Orfex surface and is represented as shown in Fig. IS.2.
surfaces SI and S2 which cut the tube (see Fig. I ~.4). Then, according to (IS.6), the outflow of vorticity through the surface S of the region R vanishes. We therefore write
ff
8.
$1.
0
dS
+
ff
$1.
0
dS
+
ff
8.
$1.
0
dS =
{f
S
$l· n.dS = 0 (IS.7)
81
. Fig, 18.3 Vortex tube.
Ifwe consider a closed curve and draw all the vortex lines passing through it, a tube is formed. Such a tube is called a vortex tube' and is represented as shown in Fig. IS.3. A vortex tube of infinitesimal cros:;sectional area is known asa vortex filament.
18.3 Vorticity Field i!l a Divergeoceless Field
Since the vorticity is the curl of another vector field, we have div n = di~(curl V) = 0 (IS.5)
· . FIg. 184 Illustrating the derivation of the spatial conservation of vorticity.
Here S denotes the surface of the wall of the tube in the portion under conside;ation. On the' wall of the tube, $1 lies inthe slJrface Sw' Hence the integral over SID vanishes
Thus vorticity is a divergenceless field. Consider. at any instant, a' region of space R enclosed by a closed surface S. We then have
fI
$1.
0
$l·ndS = 0
{in.
8
•
n dS =
IfI
R
Consequently, we obtain div $1 dT= 0 (1S.6)
ff
8.
dS
+
If
8.
$1.
0
dS
=0
(\8.8)
This equation states that the (net) outjlow of l'Of/icity through any closed surface is zero. This is true at every instant of time.
18.4 Spatial Conservation of Vorticity: Streogth of a Vortex Tube
I
In this equation 0 is an outward normal, outward with reference to the region R. If we draw the normals on the surfaces SI and S2 in rhe same sense and denote them by 0) and O2 • respectively. Eq. (18.8) may be rewritten as
Consider. at any instant. a vortex tube drawn in the flow field. Denote by R the region space enclosed between the wall of the tube and,any two
If
8,
$1 .
0)
dS =
ff
S.
$1 . n2 dS
( 18.9)
S11
IdealFluid Aerodynamics
Some Features of Flow with Vorticity 18.5 Consequences of the 1beorems 'of Helmholtz and Kelvin
S1J
This states that the flow of vorticity through any crosssectional surface S of a. vortex tube is equal to the flow of vorticity through any other cross~ sectIOnal surface SI of the tube. This is true at every instant of time. If S denotes any crosssectional surface of the vortex tube (18.9), may be expressed as n • D dS = a constant (18.10) s This states t~at the flow of tlorticity through any crosssectional surface of a v~rtex tube IS a constant all along the tube. This is true at every instant of tIme. In view of the intimate relation between circulation and vorticity, the result (18.10) may be expressed equivalently in terms of circulation. Let '6' denote any closed curve that embraces the vortex tube ('6' encloses the tube and lies on its wall): Then, using Eqs. (18.3) and (18.10), we have
If
To the above considerations on the spatial conservation of vorticity we add the theorems of Helmholtz and Kelvin on the permanence of vorticity and circulation. See Sections 9.4 and 9.5. According to these theorems, for an ideal fluid under the action of pOfential body forces, we have
.Q. rw =..£ fJn'D dS = 0 Dt Dt,
S
(18.13)
where S is any surface bounded by a fluid curve reo For an infinitesimal surface element n ds, Eq. (18;13) may be written as

Dr. =
Dr

D ( n • n dS) = 0 Dt
(18.14)
Pw
=
ff
s
n. D dS
= a constant
(18.11)
This states that the circulation around any closed CUrl'e embracing a vortex tube is a, constant all along the tube. This is true at etlery instant. of time. . E~uatlOn (18.11) expresses the spatial conserlVJtion ofvorticity in the sense ImplJed by that equation. For a vortex filament of variable crosssectional area, this equation takes the form
where c is the boundary curve of the surface element. The implication of these theorems for irrotational. motion has been discussed already (see Section 9.6). We now give some important consequences of (18.13 and 18.14) forrotational motion. Consider at any instant of time a vort~x sheet drawn in the fluid. At that instant for every element n dS v . •oe sheet. Choosing an element of the sheet and following that element in its iDotion, we observe that as a consequence of Eq. (18.14), =
Pc
= n· n ds
= a constant
(18.12)
where n ds is any crosssectional area of the filament and c if the ,boundary curve ofu ds. If we take D in the direction of n, Eq. (18.12) reduces to
n'DdS 0
p. =
.a ds =
a constant
This sh~ws that ~he vorticity at any section of a vortex filament is inversely pro~ortlOnal to I~S crosssectional area. An important consequence of the spatial conservatIOn of vorticity is that a vortex tube, and so also a vortex fi.lament or a l'ortex line, cannot begin or end abruptly in a fluid. It should either form a closed ring or end at infinity or at a solid or free surface. The circulation around any closed curve embracing a vorte~ tube. or equivalently the outflow of vorticity through any cross section of the tube, is a characteristic of the tube as a whole and is called the strength of the t'Ortex tube. If we consider a vortex filament of variable crosssectional area and shrin k the area to zero in such a way that the vorticity goes to infinity as the area goes to zero, and the strength. of the filament remains constant we arrive at the conception of a l'ortex filament with concentrated I'Orti~ity.
for all times although nand n dS may change (see Fig. 18.5). This means n • n 'o(anishes at all times for that surface element. We thus conclude that the surface element remains an element of a vortex sheet. From this it follows that a surface which is a vortex sheet at one instant remains a vortex sheet for all times. We further state that fluid particles that are part of a vortex sheet at some instant are part of it for aI1 times. Furthermore, it follows that fluid particles that are part of a vortex tube (or of a vortex filamCTIt or of a vortex line) at some instant are part of it for all times. . Consider a vortex tube and follow it as it moves along. Let t'C be any clOsed curve 'embracing the tube. The curve re mo'tes with the tube and always embraces it. Since re is a fluid CUfYe, according to Eq. (18.13), the circulation around rc remains constant for all times. This means that the circulation around a vortex tube, or equivolently the strength ofa vortex tube, remains a constant for a/l times as the tube floats along, regardless of the changes experienced by the vortex tube.
521
IdealFluid Aerodynamics
Some Features of Flow with Vorfkity in t('rms of the vorticity nCr, I) we need to invert the equation
525
The spatial conservation of vorticity as expressed by Eq. (18.11) and the consequences, as described above, of the theorem on the permanence of vorticity or circulation, are usually referred to as Helmholtz's theorems of vorlex motion. The spatial conservation ofvorticiiy is purely a kinematical property, for it directly follows from the fact that the divergence of any
at t + ~t
n = curl V We do this as follows. Considering an incompressible fluid, we have
divV = 0
(18.15)
(18.16)
D
On the bac;is of this relation, we may express V as the curl of some other vector field, say of A(r, t). Hence we set V = currA (18.17)
Since the curl of any gradient vector is zero, the vector A is indeterminate to the exte:.t of the gradient of a scalar function of position and time. From (18.17) it f0110ws that curl V = curl (curl A)
= grad (div A)  V2A
(18.18) (18.19)
We now stipulate that divA = 0 This is permissible since A is indeterminate to the extent bf a gradient vector. From (18.18), (18.19), and (18.15) we obtain
VIA = curl V =  n
(18.20)
This is Poisson's equation for A. We call A a vector potential. Once' A is determined as a solution of(18.20), the velocity field may be deduced from Eq. (18.17): In Cartesian, if we express
A
Illustrating that a surface that IS a vortex sheet at one instant remains a vortex sheet for all times.
Fig. 18.5
= (A."
= (0."
All' A.)
n
nil' 0.)
(18.20) reduces to three scalar equations
VIA., = 0",
curl vector is zero. The theorem on the permanence of vorticity or circulation is derived with the use of the equation of motion. Consequently, . this theorem and the results that follow from it are applicable in the motion of an ideal fluid (or an inviscid compressible fluid for which th('re is a simple p  p relation) under the action of potential body forces (see Section 9.6). 18.6 Velocity Field Due to Vortex Distribution in an Incompressible Fluid In applications one is concerned with the problem of expressing the velocity field in terms of the vorticity field. To obtain the velocity V(r,t)
VIAll = 0, VIA, = 0,
The solution of (18:20) is express~d as (see Section 18.10).
A(r, t) =
...!.. i~rJ n(5, t) d'T
41T
R
J.
Ir  51
(18.21)
where n(s, t) d'T is an elenient of the vortex distribution situated at the point sand R is the region in which the vorticity is distributed (see Fig. 18.6).· Note that the integration is with respect to the coordinates of the
516
Id~IF1uid
Aerodynamics
Some Features of Flow with Vorticity Since we have and (18.25) may be rewritten as
517
o n·o.dS =r
dl = dl
n
Fla. 11.6 Nomenclature used in the derivation of the velocity re5ulting from a vortex distribution. vortex distribution. The velocity 'field i5 then given by V' = curl A =
r dl I5A(r)=(18.26) 417 Ir  51 The contribution to the velocity at the point r from the element of the filament is then given by r dl t5V(r) = curl.   (18.27) 417 Ir  51 In carrying out the curl operation we keep dl and 5 fixed. Equation (18.27) reduces to t5V(r) = dl x (r  5) (18.28) 417 Ir  siS This is known as the BiolSavarl law. The velocity at r due to the whole
1:.
..!. cUrliIJ n(5, t) d.
411"
. R
Ir  51
(18.22)
n dT situated at 5 and similarly by t5V the contribution to V at r, we have
I5A(r, t) =
If we denote by t5A the contribution to A at r due to thevortex element
.!.. n(l, t) dT 4" Ir  51
1
(18.23) 08.24)
t5V(r, t) = 
4"
. n(s, t) curlr   dT
Ir  II
We include the subscript. on the curl to emphasize that the curl is to be takep with respect to the coordinat~s of the point r. 18.7 Velocity Field of a Vortex FOament: BiotSavart Law ConsIder a vortex filament of strength r. Choose a volume element dT of this filament as the cylinder formed by a crosssectional surface 0 dS and an element of length dl along the filament (see Fig. 18.7). Then the contribution to the vector potential A at a field point r, from the vortex element at 5 is given by t5A(r)
r
=..!. nCI)
417 Ir  51
(0 dS • d/)
(18.25)
FIg.11.7 Nomenclature used in the derivation of the BiotSavart law.
528
IdealFluid Aerodynamics
Some Features of Flow with Vorticity
519
vortex filament is obtained by integration of the ex pression (18.28) over the length of the filament. We thus have V(r) =
r
417
f
Now, let h denote the normal distance froJ,Il the field point r to the filament jlnd let 51 denote the point of intersection with the filament of the normal to it from the field point (Fig. 18.8). We then have
dl x (r  s) If'  .s13
(18.29)
'Sl 
Since r is the strength of the filament, it is a constant and hence appears outside the integral.
= h CoseC (J == h cot 8 ds == hcosec;l'8 dB
'1,
S
18.8 . Simple
A~plications
Two simple examples of the application of BiotSavart's law are given here and can be used later. Consider an infinitely long straight vortex filament of strength r (Fig. 18.8). To calculate the velocity field choose the origin of coot'dinates
o
Fig. 18.' Semiinfinite straight vortex filament.
and
f
Hence (18.30) yields
Fig. 18.8 Velocity field due to an infinite straight vortex filament.
rtJ
rtJ
,sin ds == . 1  8 r11 h
il'·
0
sm 0 dO
2 == h
V(r)e
r
21Th
(18.31)
at some point on the,filament. Then, according to (IS.29), the velocity at a field point is given by V(r) =
L. foo ds x (r ~ s)
417
00'
We readily conclude that the motion is twodimensional, the plane of motion being normal to the vortex filament, and titat the flow field is that of a twodimensional or point vortex. Consider, as a second example, a semiinfi'rite straight vortex filament extending from 0 to 00 (Fig, 18.9). The velocity field is then giv~n by
V(r)
Ir  51' •
where ds is an element of the filament at 5. Denote r  sby r 1 and the direction of d... x r 1 bye (see Fig. IS.8'),; If 0 is the angle measured from ds to r (such that 0 5: 0 5: ...), we have ds x r 1 = er1 sin 0 ds and the above expression for the velocity becomes
V(r)
== er 41T
ieo sin 8 ds
== e r41Th
il' '0
(1
0
r11
sin 0 ds (18.32)
== e (18.30)
r
41Th
+ cos (}o)
0 == erfoo sinI ds 41T
OIl,
r1
since e is a constant.
Thus, if the field point is located in the plane ()o == 1T/2,the magnitude of the velocity is r/41Th. If the field point is located in a plane situated far away from the origin of the vortex filament, the angle 00 tends to zero, and
$JI
IdealFluid Aerodynamics
Some ~eatura of Flow with Vorticity and equal to~. Thus we have
.000... 0.1;
5JI
lim D.{P)£ dS ~ dS . .
(18.34)
FII. 11.18 Finite segment of a straight vortex filament.
the magnitude of the velocity tends to r/211~; the value for an infinite filament. The velocity field due to a finite segment of a straight vortex filament is given by
In this way we shrink the narrow region of vorticity to a single surface on which the vorticity itself is infinite but ~.defined by (18.34) is finite. We call such a surface a vortex sheet. of concentrated vorticity, or simply a vortex sheet, and ~ the strength of the vortex sheet. The strength ~ has the dimensions of vorticity per unit area. We now relate ~ with the velocities on either side of the vortex sheet. According to the definition of a curl vector, we have
$1 dT'  curl V dT' 
if .
18
x V dS
VCr)  e.l:.... ["sin 6 ds
41Th JIi
 e   (cos 61
41Th
Sec Fig. IS.10.
r '.

cos 6.)
(lS.33)
where V denotes the fluid velocity. We choose tIr as the infinitesimal cylinder described before and write.
~ dS ~ Ii~ $1.£ dS == ,"0 Jj' D X V tiS lim t{ .....
18
18.' Vortex Sheet A vortex sheet is defined in a manner similar to that of a vortex filament
with concentrated vorticity (sec Section 18.4). Consider the narrow region of vorticity enclosed between two neighboring surfaces of vorticity SI and SI (Fig. IS.Il). Let P be a point on an intermediate vortex surface S. Choose at P an element D tiS of S and an elemental cylinder with n dS as the cross section through P and with height equ~1 to E the distance at P );)etwccn SI and SI' If n(p) denotes the vorticity at P, we can write
n(p) tiT'
=
_u where dSI and dS. are respectively the two faces of the cylinder and "walt" ~s the wall of the cylinder. The iimit of the sum of the integrals over dSI and dS. is D X (VI  V,) dS, and that of the iatcgra1 over the wall vanishes. Hence we obtain
481 U.
~i~'(ff.. x V dS +
II
D·X
V dS
+
II•• tis)
V
== n(p)£ tiS
or
~ dS
=;:
I! X (VI

V,) tiS
~ = D X (VI  V.)
(18.35)
where tIr is the volume of tbe infinitesimal cylinder. We now wish to let E go to zero and $1 go to infinity in such a way that remains constant
nE
We note that ~,D and (VI  VI) form a righthand system of orthogonal vectors (Fig. 18.12). It follows that the vector (VI  V,) is tangential to the vortex sheet. We see that
VI VI
= ~
X D
(18.36)
_ _     . . . ... 81
!~+s
(VI  VI)' D = 0
or,
VI' a
= VI' D
. (18.37)
_..ll'_ Sa
and
(18.3S)
FII. 1'.11 Illustrating the concept of a vortex sheet.
We therefore state that there is a l'eloeity discontinuity across a l'ortex sheer; the discontinuity occurs on~v in the tangential component of the
531
IdealFluid Aerodynamics
Some Features of Flow· with Vorticity
533
velocity, whereas the normal component remains continuous; the magnitude of the discontinuity is equal to tht: magnitude of the. strength of the sheet. A
vortex sheet is thus. a surface of tangential discontinuity, attd, conversely, a tange!ltial discot:ltinuity is a vortex sheet.
Identify cp of this equation with A.,(r) of{l8.39) and choose tp as the function We then have tp(r) = (lr  rll)l VI = n.,(r) grad cp = grad A", grad
Ip
(IB.40)
+
l
V21p
r  rl s Ir  rll = 0 everywhere except at r = r l where it becomes infinite
=:: 
VIVI
FI&. 18~1Z Relation between vortex· strength vortex sheet.
and the velocity discontinuity across a
Because the point r = r l is a signular point, in the sense that tp, grad 'I', and V 2tp become infinite at that point, we surround that point by a small spher~ of radius P ,:"ith <!enter at r l and apply Green's theorem in the region co~tamed between the sphere and an arbitrarily drawn large surface ~ (FIg. IB.13). To cover all space we remove ~ to infinity and shrink the sphere to the point rl' In this way we obtain
Vortex sheets .and· vortex filaments of concentrated vorticity are not physical possibilities. However, they form suitable analytical apprC?ximations when the vorticity is confined to physically narrow regions. Tbe conceptS' of vortex filaments and sheets are widely applied in wing theory.
IIf
R
n.,(r) dT = lim Ir  rll 'Ia:> _ lim
po
:rr '" Ir rl r I I3
,JI (A
r
~ r + grad A",) • n dS
1
3 rl1
Ir  rII· (18.41)
It' sphere
,JI
(A", Irr p
+ grad A",) . D dS
Ir _ r11
18.10 Solation for the Vector Potential
We wish to construc~ the solution of (18.20)
VIA
=:
nCr)
or, in Cartesians, of the following system of equations
VIA",
= ..,..O",(r)
(18.39)
V2A, = O.(r) VIA, = n,(r)
We consider all space and assume that the field A t'anishes sufficiently strongly at infinity. We shall amplify this vague assu'mption later. It is necessary only to consider 'the solution to one of the equations of the system (18.39). To construct the solution we use Green's theorem in the form given by (2.141)
Jf.f; 1>
\2'1'  'I' \21» 1fT =
if
(1) grad V' 
tp
grad
cpr· ntiS
Fig. 18.13
Illustrating the computation of the \ector potential.
JJI
IdealFluid Aerodynamics
By our assumption that the field dies out sufficiently.strangly at, infinity, we require that A. and grad A. vanish, as we approach infinity, in such a way that the limit of the surface integral over 1; vanishet as 1; g~s to infinity. We thus set equal to zero the limit of the integral over 1;. CQnsider the integral over the sphere and introduce spherical coordinates 'I'f 8, t/> with origin at the point r 1 . 'The integral over the sphere is then alto
Chapter 19
ecru
Elements of Finite Wing Theory
II (A. +
flf
R
p
°o~·) sin 8 dO dt/>
We now come to the problem of ftow past a finite wing. A finite wing is any threedimensional hody which has the distinctive property that' when it is suitably placed in an originally uniform stream the lift on the body is far greater than its drag. By suitably placed we mean that the orientation of the budy with respect to the uniform stream is such that any lifting characteristic it may have can come into play. Thus whether a body is a wing or not depends not only on the shape of the body but also on its orientation. In describing the shape ofa wing the following terminology is common (see Fig. 19.1). Wing Section. In a wing there is a fixed direction ,such that planes normal to that direction cut the wing in ,cross sections of airfoil shape. Any such cross section of. the wing is known as a wing section. Span. The fixed direction itself is k~own as, the spanwise direction. The two wing sections that have the greatest distance between them along the spanwise direction are called the wing tips. This span wise distance between the wing tips is known as the sPan of the win,g and is denoted by h Section Chord. This is simply the chord of a wing section. Leading and Trailing Edges. The line joining the leading ed,ges of all the w~ng sections is called the leading edge of the wing. A trailing edge of the wing is similarly defined. Midchord Line. This is the line joining the midpoints of the chords of the wing se~tion. It extends from tip to tip. Slraight Wing. If the midchord line is a straight line, the wing is called a straight wing. Swept Wing. A wing th'lt is not straight i~ said to be a swept wing. Rectangular Wing. A straight wing for which the chords of all the wing , sections are of the same length is known as the rectangular wing. Tapered Wing. This is a straight wing for which chord length of the section varies along the span. Twisted Wing. This is a wing for which the section chords vary in direction along the span. 535
and the limit of the integral as p goes to zero is equal to 411'4.(rl ) OJr) d.,. = Ir ...... rll
'
41TA.(r~)
(18.42)
The variable of integration is r. Switching the roles of rand r 1 , we write A.(r) =
...!..
41T
flf
R
O,.(rl) d.,. Ir  rll
(18.43)
Now the variable of integration is rl. The solutions for A.(r) and A,lr) are similar to that for Air). From these we conclude that the solution for A(r) is given by A(r) =
41T
...!..
,flf
R
O(rl) d.,. Ir' rll
(18.44)
Compare this with (18.21). We now amplify the behavior of the field at infinity. Let us choose for the arbitrary surface 1; a large sphere of radius R with its center at r t • Introduce spherical coordinates R, 0, t/> with origin at r 1 • The limit of the surface integral over 1; is then equal to
lim
800
ff (A. + ROA.)
oR
sin 0dO dt/>
For this limit to vanish we require that, as R approaches infinity, A. sho:'ld approhch zero and iJA.,IoR vanish more strongly than IIR. Similar behavior is required of A. and A z' This is what is meant by the vague . assumption that the field vanishessuffkiently strongly at infinity.
J36
IdealFluid Aerodynamics
Elements of Finite Wing TheOry
$31
y
(01
(II)
(d)
(b)
Fla. 19.1 Some nomenclature associated with wings: (a) rectangular wing; (b) plan form of a tapered wing; (c) ;Ian form of a swept wing; (d) plan forms of tapered and swept wings; (e) plan form of a delta wing.
(e)
Fig. 19.1 (Continued)
531
IdealFluid Aerodynamics
Elements of Finite Wing Theory must proceed from the wing. Such a sheet is known as a trailing vortex sheet. If we assume that the circulation is uniform along the span and drops to zero at the wing tips, then there wiH be two trailing vortex lines (or vortex filaments with concentrated yorticity) originating at the wing tips. Such a concept of two trailina YOIIe& lines was introduced by Lanchester. Although it is satisfactory for analyzing certain problems, it
Plan Area. This is the projected area. of the wing on a plane parallel to the spanwise direcnon. This area is denoted by S. Asptct· Rat;o. This is the ratio of the square of the span to the plan area. and is denoted by At. We thus ~ave
At = b
t
S
For a rectangular wing the aspect ratio is simply equal to blc, where c is the . . chord of a wing section. Mean Chord: The length defined by the ratio blAt is defined as the mean chord of the wing and is denoted by There is a vast variety of possible shapes for a wing (some examples are shown in ~ig. 19.1). This being the case, any general discussion of the flow past an arbitrary wing is likely to becomplcx and extensive: In addition, for arbitrary wing shapes 're may not even have a good physical insight into the nature of the flow Fast the wing. Furthermore, we arc ~aturaIiy not i~tc!CSted incompletely arbitrary wing shapes. Therefore the development of the finite wing theory took place with reference t~ particular wing shapes. The first mathematical formulation of the theory was made by Prandtl (in 1918) for straight wings oflarge aspect ratios. He was concerned with the oroblem of extending, in a proper manner, the results of the· theory of the infinite wing to develop a theory of the finite' wing, a theory tbat would yield practically applicable results. A presentation of Prandtl's theory is our main concern in this chapter. The ideas underlying Prandtl's theory arc important and have served as the basis tor further developments of the finite wing theory.
c.
.
Fig. 19.2 Vortex sheet trailing behind a wing.
19.1 PnDdtl's 'I1Ieory The following idtas constitute the formulation of Prandtl's model of the flow past a finite wing "f large aspect ratio. I. Associated with the lift on the wing there is circulation' around the wing. At the tips of the wing the lift, and consequent~y the circulation arouo.d the wing, vanishes. Thus theJift and the circulation vary along the span of the wing, their distribution being symmetrical about the midsection of tlie wing. 2.. Since the circulation around any section of ·the wing is equal to the outflow of ¥eRicity through that .section,it follows that the variatioq of circulation along the wing span must be accompanied by the shedding of vorticity from the wing. Vortex filaments with their vorticity predominantly in the direction of tbe undisturbed stream originate fr~ the. wi~g and proceed ·downstrcam from the trailing edge. If the circulation vanes continuously along the wing span, a continuous sheet of tra;~ vorlices
is not appropriate for analyzing the flow close to the wing, which is our present concern. For this purpose, the concept of a continuous trailing vortex sheet is more appropriate. This concept was postulated by PrandtL In the actual case of a finite wing moving through air one can observe such a trailing vortex sheet. At some djstance behind the wing the sheet. rolls into discrete vortices: They are eventually dissipated by the action of viscosity. In the formulation of a theoretical model for the flow past a finite wjn~ in view of the laws of vortex motio:l for an ideal fluid, we have to allowtbe trailing vortices to 'extend to infinity in the downstream direction. It is known that e'yen in an ideal fluid <;uch a vortex sheet under the mutual influence of its various parts will gradually roll up into a pair of vortices (see Fig. 19.2). For studying the Row close to the wing, this rolling up of the trailing Vorte!' ~neet may be ignored. We thus assume that the trailing Yortex sheet extends to infinity without any tendency for rolling up. 3. Thl" wing itself may be replaced, for instance as in the case of the infinite wing, by a cpntinuous distribullon on the wing surface of vortex lines that are in the spanwise direction. One may use a vortex sheet on tbe top and a vortex sheet on the bottom of the wing, Such a vortc.< sheet is called a theet of bound cortices or a bound fortex sheet, "bouad" meaning
51(J
IdealFluid Aerodvnamics
Elements of Finite. Wing Theory
b
511
bound to the wing. For analyzing the lift on the wing, we may replace the wing'by a single bound vortex sheet. A bound vortex sheet differs in one important respect from the usual vorte~ sheet, which may be referred to as a free vortex sheet. Across 1bebound vortex sheet, whiclr represents a wing, a pressure difference 'tll'II!J .st., Across a free vortex shl"et, however, a pressure difference does not exilt. 4. The fl )w past a finite wing may thus be represented as the flow past a certain vortex sheet. Part of this sheet is a bound vortex and the rest is a free vortex sheet.
2"
~
2
Fig. 19.4
Lifting line model for flow past a wing.
FII.19.3 Vortexshcet model for flow past a wing.
We now assume that the velocity resulting from the vortex sheet alone is small in comparison with the velocity of the undisturbed stream. We then assume that both the free and bound parts of the vortex sheet lie in a plane parallel to the undisturbed stream. The resuldng flow model is shown in Fig. 19.3. For a wing of large aspect ratio (i.e., for which c b) the bpund part ~ the vortex sheet may be approximated by a single bound vortex line, naturally of varying strength. Such a line is known as the lifting line. PriiDdtl used the lifting line to represent the wing. Therefore his theory is also known as (Prandtl's) lifting line theory. The flow model for this theo.ry is as shown in Fig. 19.4. This model is the basis for the rest of our considerations. 5. Consider t~e velocity at the lifting line (i.e., at the wing). At any point on this line the velocity is the resultant of the velocity V", of the undisturbed stream and the induced velocity q resulting from the' vortex distribution. We note that q has no component in the direction of V""' it has a' spanwise component (i.e., a component along the lifting line) and a component that is normal to the lifting line and the velocity V'" (~ee Fig. 19.5). WI': find that this normal component is directed downward (as sI»Wn in the figure) all along the lifting line. We refer to 'it as the downwash ,,~and denote it by w.
We now assume that the spanwise component of the induced velocity is very small compared to its downward component. As an immediate consequence of this ass\lmption we ignore tbe spanwise component and take the velocity at any point of the lifting line as equal to the resultant of the duwnwash wand the velocity V 00. We !:Ienote this resultant l'elocity by VR and note that it is normal to the lifting line. We observe that the present assumption can be valid for a wing of sufficiently large aspect ratio and even then only for those sections of the wing that are sufficiently distant from the wing tips. The assumption cannot be regarded as
«
 ~Fig. 19.5
Velocitle~
at the lifting lint.'
542
IdealFluid Aerodynamics
Elements of Finite Wing Theory
543
appropriate near the tips of any wing or for' any section of d low aspect ratio wing. In general, the downwash velocity is not uniform along the lifting line. , ~onsequent1y, the resultant vel~city V R is' also not uniform along tne lifting line. Choosing a Cartesian coC''"diitate system as shown in Fig. 19.5 we write '
equal to VR(Y)' Then, according to the KuttaJoukowski theorem, the force on the slice of the finite wing is given by.
(W(y) = pV R(Y) )( jr(y) dy
(19.2)
w = w(y)
V R = VR(y) = V",  w(y)k
(19.1)
where y is the coordinate along the lifting line or, equivalently, along the span of th'e wing, 'and k is the unit v~tor in the direction of Za;.is. We
I
where j is the direction of the Yaxis. We note that this force is normal to the velocity V R and not to the velocity V",. Both the magnitude and direction of the force may vary with y. 8. According to the airfoil theory, the circulation around an infinite wing is proportional to the speed of the undisturbed stream and for small angles of attack to the angle of attack measured from the zerolift direction
kw(y) I
I I
I
~~~~~~x
Voo
r(y)
v.
Fig. 19.6 Flow representation at any section of the wing. Fig. 19.7 Defining the induced angle of attack.
note that V R is nornlal to the lifting line all along that line. In this sense the velocity at any section of the lifting line or, equivalently, at any section of the wing, is two dimensional. This result which is simply the assumption that we have made, enables us to use the results of the theory for an infinite wing to build a quantitative thwry for the finite \'ling. 6. Denote by r = r(y) the circulation along the lifting li~e. Consider the x,zplane which cuts the lifting line at the distance y. The situation in this plane is the twodimensional one of a bound point vortex of strength rey) in a freestream VR(Y) at the vortex point (see Fig. i9.6). This, of o(·urse, is the vortex representation of the flow past an infinite wing. Thus, according to tbt> present considerations, each section of the finite wing acts exactly as the section of an infinite wing,tllat is placed in ~.n originally undisturbed stream of velocity VR' This expresses the basic idea of Prandtl's quantitative theory of the finite wing. The velocity V R, which is to be calculated according to Eq. (19.1), is not the same as the velocity V 00 of the:' undisturbed stream past the finite wil1g. 7. Consider a slice L~f the finite wing the slice being of thickness dy and situated at y. It follows that the force acting on the slice is equal to the force acting on a similar slice of an infinite wing whose section is the same as the section under consideration of the finite wing, and for which the qrculation is equal to r(y) and the velocity of the undisturbed stream is
(see Section 16.3). Denoting by a.R = a.R(y) the angle between VR(y) and the zerolift direction of the section at y of the finite wing we write (19.3) where K is a constant that depends on the form and size of the wing sec.ion under consideration. Since the form and size of the wing section may, in general vary with y, we have expressed the constant K as K(y). From the airfoil theory we filld that
K(y) = ![dC L (y)JC(Y) 2 da.R
1 =  ao(Y)c(y) 2
(19.4 )
where c(y) is the chord at y of the finite wing and ao(y} is the slop.: of the lift versus angle of attack curve for the corresponding airfoil profile (see Section 16.4). 9. Denote by ~(y) the, angle between the velocity V and the zerolift direction of the wing section at y alld by (1.,(.11) the :Ingle between thevelocitiesVII(y) and V (see Fig. 19.7). We then have
f f.
(195)
IdealFluid Aerodynamics
Elements of Finite Wing Theory
545
The angle ()(, is known as the induced angle of attack. It is given by
x,(y)
Hence for the downwash w(lI) we obtain
=
tan
1
w(y) 
" :\s~umJng .,: is \er~ ,mall compared to Vx we may write approximately
) :x (.II =  , V"
y
'7 ~
I
V
w(y)
== 
1
fbll
b/l
dr
4'IT

dy
1 ('1)   d7J Y  '1
( 19.8)
w(y)
( 196)
RCcall that positive w denotes downwash, that is, a velocity directed in the negative zdirection. We note that the downwash cannot be determined till r(y) is known. But r(y) itself is not known until w(y) and consequently (XR(y) are obtained. 11. To proceed, we combine (19.3), (19.7), and (19.8) and obtain
1 r(y) == K(Y)VR(Y) [ cx(y)   4 ' V
b 2
fbi!
b/l
'IT ""
dr dy ] d ('1)Y Y ~ '1
(19.9)
Since w<Ylis assumed to be much less than V""' we approximate V R(y) by V"". Equation (1 ,)~9) then be¢omes
~~x
r(y) == K(y) V""cx(y)  where, according to Eq. (19.4),
[
dr d7J ] d ('1)4'IT '/1 y Y  '1
b/l
If
(19.10)
K(y)
!.
2
==
la.,(y)c(y)
Fig. 19.8
Illustrating the calculation of downwash at the lifting line.
Equation (19.5) then becomes (XR(Y) = :x(y) _ w(y) ( 19.7) V" This is a fundamental relation of Prandtl's theory: 10. We now eXI)fcss the down wash 1\"(.11) in terms of the clrcuiation 1'('1). Let y = y(y) denot(' the stn:ngth of the trading \llrtn sheet rcr unit length along .II (see Fig. 19.8). Then the downwash at the ,cctll>n 'I of the lifting line is gi\en by (see Section II\X) .
Equation (19.10) is an integrodifferential equation for r(y) and is known as Prandtl's equation of finite wing theory. It is a relation between the circulation around the wing, the geometrical features such as the chord and angle of attack distributions, and the properties of the wing sections. It is thus the basis. for obtainin~ important information about the aerodynamic characteristics of finite wings of sufficiently large. aspect ratios. To determine the circulation r(y), one has to solve (19.10) with the additional condition that the circulation falls to zero at the wing tips:
r(  ~)
=
r(~)
= 0
(19.11)
lhe dlstribution~
;'(y)
and I'(I/l.irc .. Idted h\ the cquJtion (veri!) thl,)
;.( II' 
We shall consider ~ater the problem of solving (19.10). '12. We now give the expressions for calculating the lift and drag.ol a finite wing and the components of the moment acting on the wing. By definition the lift is the component of the force on the wing in the direction norma~ ttl the velocity V 00 and the spanwise direction. and the drag is the component in the direction of V <7)' Accordingly. the lift and drag of a slice dy of the wing are given 'by
dl'
./ i/
dL(y)
==
dF(y) cos (Xi ~ dF(y)
~
(19.12)
dD(y) = dF(y) sin (Xi
dF(y) (X,(Y) = 1J.,(y) bL(y)
(19.13)
546
IdealFluid Aerodynamics
Elements of Finite Wing Theory
547
wher~, as t>efo~e, bF(y) is magnitude of the force ;SF acting on the slic~ of the wmg (see Fig. 19.9). In these relations we substitute pVR(y)r(y) dy for bF(y), (see 13.19), approximate VH(Y) by V ao , and obtain
bL(y)
= p Vao
r(y) dy
(19.14)
~D(y) = (XI(Y) ~L(y)
= pV«> (Xj(y)r(y) dy
(19.15)
= pw(y) r(y) dy
may say that the drag expressed by Eq. (19.17) is the drag that must be counteracted in order to obtain lift. Such a drag, which originates due to the induced velocity field, is known as the induced drag. signify this fact, it is denc:ed by the special symbol D j • The work done against this drag force appears as the kinetic energy of the downwash field that must be created to obtain lift. Equivalently, the induced drag may be interpreted as the force that necessitates the expenditure of work in creating the vortices trailing behind a finite wing.
To
z
y
f
~ ~
Veo
&L
!
2"
&Di
J'~ X
Fig. 19.9
Lift and drag on a slice of the wing.
These e~ uations express the span wise distribution of the lift and drag forces. IntegratIOn of these equations yields the total lift L and the total drag D of the wing. We thus obtain
L=pV", _ozr(y)dy
Fla. 19.10 Defining tbe rolling and yawing moments on the wing.
The moment on the wing with given by
fes~t
to the origin of coordinates is
f
Ol2
(19.16)
M
== ==
f
bll
.
yJ )( (i6Dj
bll
+ k6L)
+
i
fbll
bll
D
=
f
Ol2
011
rxlY) ~L(y)
Ol2
k
f
bll
y 6D,
y 6L
b/l
= pV",
•
f
rxj(y) r(y) dy
w(y) f(y) dy
b12
(19.17)
=p
f
Ol2
(see Fig. 19.10). It is customary to caU as the roiling moment the component of M in the negative xdirection and as the ya'wing moment the component in the negative' zdirection (see figure). Denoting these components by MR and M", respectively, we have
b/2
It is thus seen that there is a nonzero drag Or) a finite wing. It \ani~hes only Ifthedownwash vanishes all along the span. This happens only if the wing IS Infinite. For a finite wing experiencing a nonzero lift, the downwash cannot be zero and consequently the drag cannot vanish. Therefore we
MR = M"
f
bll
bll
Y 6L
==
p~'",
fbll
fbll
b/l
r(y)y dy
(19.18) (19.19)
Olt
==
f
Y lJDi == P
w(y) r(y)y dy
Oil
b/l
511
IdealFluid Aerodynamics
Elements of Finite Wing Theory
549
19.1 'Problems of Interest
We now consider the type of problems that may be solved by means of Prandtl's theory. The basis for our considerations are the equations (19.8) for the downwash w(y) at the lifting line and the integral equation (19.10) for the circulation I'(y). A relatively simple problem is to prescribe the distribution of circulation (or equivalently of the lift) r(y) along the span of the wing and to seek to determine the distributions of the downwash w(y), the angle of attack oc(y), and the chord c(y). Recall that the angle of attack is measured from the total zerolift line that depends on the camber of the particular wing section under consideration. It therefore follows that if the camber varies along the span, the angle of attack will also vary even if the socalled geometrical angle of attack is kept constant all along the wing. span. The distribution of the chord determines the plan form of the wing. The problem we have posed is kuown as the first problem and may be solved by a straightforward calculation. An imporfa!1t example of this problem is that for an elliptic lift distribution and is given in Section I ~.3. A more interesting and more difficult problem is to determine the aerodynamic characteristics of a wing, sur:h as r(y), ~L(y), ~D;(y), when its geometry is prescribed, that is, when the distributions of chord c(y), wing section represented by ao(Y) and the zerolift line, and the angle of attack cx(y) along the span' are given. This problem is known as the second problem. It requires t~e solution of the integral equation for the circulation. A method for its soiuaion is given in Section 19.4. A third problem is the. socalled problem of minimum drag. The problem is how to make induced drag a~. small as possible under given conditions. Some aspects of this question will be discussed in Section 19.6.
the problem of the elliptic hft distribution yields many results of great practical significance. . We first compute the downwash distribution uSing (J 9.8) and (19.20). We have
w(y)
=
I f~i2 df
47T
d
d1}
(1})Y  1}
[
b 2
Y
= _
~ rrb 2
f
O'2
.
1
( . .,
1)
)2lJ
Y.i
_1}_
d1J
(19~1
)
0,2
b
Y 
1}
2
L
b
I{· ~~y
~ 2
Fig. 19.11
Elliptic lift dIstribution.
19.3 Elliptic Lift Distribution: Elliptic Wing
We now consider the first .problem where the spanwise distribution of the circulation r(y) is given.' The distribution of the lift is simply propOl·tional to 'hat of the circulation. The problem become~ particularly simp;e for the socalled elliptic lift d:stribution expressed by
Note that the integral is improper at 1} = y, and hence one should obtain .' t he prmclpaI va I ue of the integral . IntrodUCing a change of variable. through the relation b y=cos() (19.22) 2 we rewrite (19.21) as fo SW cos rp d (1" 23) w()rp ( )27Tb 0 cos ()  cos f{' The value of the integral is 7T (see Eq. 17.73). Hence the downwash is given by
w(y) =  ,
fo 2b.
a constant
( \9.24)
or
f2(y) yl
(19.20)
ro2 + (bI2)' = 1
where roo is a constant. The distribution is represented by the upper half of the ellipse described by this equation (see Fig. 19.11). The constant ro is the ma:.imum circulation at the midsection of the wing. The solution to
We thus conclude that the do .... mvash corresponding to atl ell/pilc lift distribution is a constant all along the span. It foIIO\,;, that ID thl~ ca~e the induced angle of attack ~I is also constant along. the span. ", J Since the down\l'ash is a constant, the distribution o.f the !tluuct:u drag . along the span is al~o eltiptic. We hav.c.
bD.(y) = p ..... (y) f(y) dy
= constant l'(y) dy
550
IdealFluid Aerodynamics
Elements of Finite Wing Theory
551
The drag of the wing is given by
D;
= pf
= wp
bll
w(y) r(y) dy
bi!
biZ
f
r(y) dy = ~ L
V""
biZ
(19.25)
is a parabola with its vertex at the origin and its axis along the axis of the drag coefficient (see Fig. 19.12). It is readily verified that the yawing and rolling moments on such a wing are zero no matter how the chord, the angle of attack, and the wing section are arranged. The spanwise distributi'on of chord. wing section. and angle of attack are related to the lift distribution by the equation (follows from 19.3)
r(y) = ioo(Y)c(y)[V""cc(y)  w(y)i
=rJ.,L
where wand
rJ.,
The integral 19.11. Wefind
I::
are constant.
r(y) dy is the area of the semiellipse shown in Fig.
f
b/2
f(y) dy = 7Tbro
biZ
4
Therefore the lift and drag are given by L
= 4 "" r 0 b ~pV
(19.26)
=  p r 02 8
7T
(19.27)
Fla. 19.11 Polar diagram for a wing with elliptic lift distribution.
The lift and drag coefficients of the wing are given by
For a M'ing with elliptic distribution this becomes (19.28)
Cn
roJI  l;yjb)2 = r(y) =
~ ao(Y)c(y{ V",rJ.(y)  ~;J
(19.30)
==
,
0,
=
2
~(Fo)2
4S V""
·!pV.czS
=C 2
'TT
I S b
L
(I9.29)
=~
. 'TT
I r
2
We see that this relation may be satisfied by many different combinations of oo(Y), c(y), and cc(y). If we choose the same airfoil profil, (i.e., same oo(y) and same camber) all along the span and also keep the same geometrical angle of attack along the span, ao(Y) and rJ.(y) will be constant in (19.30). We may then solve for c(y) and obtain
c(y) = constant r(y)
.f{
where S is the plan f o · f h' . . rm area 0 t e wlr.g and rR IS the aspect ratio of the ;Ing. A curv~Jthat shows the lift coefficient of a wing as a function of the . rag coefficleYlt IS known as the polur diagram. From (19.29) we have CT. = "/7TMC v . This shows that the polar diagram of a wing with an elliptic lift distribution
where the constant is determined by the values of r 0' V"" rJ., b, and ao. This shows thot to obtain an elliptic lift distrihutioll on a (geometrically and aerodynamically) untwisted wing, the span wise distribution of the chord should be elliptic. A wing with such a distribution of chord is known as the elliptic wing. Examples of the plan form of an elliptic wing are shown in Fig. 19.13.
551
Id,.:alFluid A:rodynamics
an ellipt:c wing may be expressed readily in terms of the geometrIcal properties of the WIng. To do this we first write
c(y)
F~r
ro
Elements of Finite Wing Theory
553
=
co.
2Y)= j (b
I 
(19.31 )
slope of the lift curve becomes 00' Denoting by (CL )"" the lift coefficient of the infinite wing we write (19.36) Then (19.34) and (19.35) may be expressed together by the relation (19.37)
where Co is th,e. chord of the midsection of the wing. Substituting the relatlo ,1 (19.31, Into (19.30) and solving for we obtain.
ro
r 0_ 1
2bV",,cx.
+ (4b/a oco)
(19.32)
Introduce the aspect ratio defined by
where S is the plan form area of the wing. For the elliptic wing we have
and
S =  cob 4
'TT
0.5
iR=~
'TTC o
Hence (19.32) may be rewritten as
Fig. 19.13 wings.
Plan forMs of elliptic
2bV""ex (19.33) 1 + ('TTiR/a o) Substituting this relatipn for r 0 in (19.28) for the lift co.efficient we obtain
ro =
o
Flg.19.14 Variation of lift coefficient with aspect ratio for a thi!) elliptic wing.
C
7Tb L = 2S Voo = 1
ro
ao
+ (ao/'TTiR) ex
(19.34)
This expresses the lift coefficient (or a finite (elliptic) wi~g a~ so~e an~le of attack in terms of the lift coeffiCientJor the correspondIng mfiOlte wmg at the same angle of attack.  For a thin wing, the theory of the infinire wing shows that
00 ~
From this it follows that for a given aspect ratio, the lift varies linearly with the angle of attack  The slope of the C L versus the cx curve is given by deL Go a = =   __ dex I + (G O /7TlR)
27T
_ )
I +2J/I{'
(19.38)
Thus for a thin wing (19.37) takes the approximate form (19.35)
r=   21T1x
(CJ.)cc
C,
Cr.
( \9.39)
Recall that Go is the slope of the lift coefficient versus angle of attack curve for an infinite wing that has the same section as the finite wing. This is clear from the (19 ..35). which shows that as the aspect ratio goes to infinity. the
• NlllC that
tin
and :x an: con~tants_
This expresses in a simple way the effect of the aspect r'l:tio on the Iif.t of.a thin elliptic wing. The varia~ion ()f Cr./t C/) < with A~ for such a Wing IS shown in Fig. 19.14. The properties of an elliptic wing. as expre;sed by (19.29) and (19.34). are afair representation also llfthe properties of rectangular. and trapezoIdal wing shapes. Hence the impllrtillll"= 1'1' "lldyIn~ the eHlpll'" WIng for
554
IdealFluid Acrodynamics
Elements of Finitc Wing Theory
555
which simple theoretical formulae may be obtained. It is illuminating to realize that the results expressed by (19.29) and (19.34) were not known even ~s empiric~l facts, before Prandtl's wing theory was developed: Expenments ~arrted out a posteriori have amply confirmed the predictions of the theory. This is really a remarkable feature, particularly in view of the many assumptions made in the theory.. But such is the characteristic of a good theory.
2. Far downstream from the wing the induced flow does not depend on the streamwise coordinate and hence may be regarded as twodimensional in all planes normal to the trailing vortex sheet, that is, to the mainstream direction. This suggests that the downwash, and thus als~ the circulation, at the lifting line may be readily. obtained if the problenHor the twodimensional flow in the cross planes (i.e., the y,zplanes) far downstream is formulated and solved. The distribution of the downwash on the vortex sheet with the
19.4 . Solution for the Arbitrary Wing: Treft'tz's Method
. P~oble.m. We no~ consider the second problem of determining the dlstnbt;tlOn.s ~f the 11ft, drag, and moments for an arbitrary wing of given shape. For thiS purpose we must first obtain the distributions of the downwash w(y) and the circulation r(y), given the distributions o.(y), c(y), and o:(y). We are concerned with the solutions of (19.8) and (19.10), which are w(y) where
z
= 1 fb/t
411
y(y)
b/l
~d7J 7J  Y
(
I'(y)
~~~~r~r_·,__,,__.~x
2w(y)
= _ dr(y)
dy
and
Fla. 1'.15 Streamwisc distribution of the downwash on the vortex sheet.
(l9.lOa)
streamwise coordinate % is shown in Fig. 19.15. We denote the down wash on the trailing vortex sheet far downstream by w a,(Y): wco(y)
r(y) = K(y)[VcoO:(y)  w(y)]
=
where
K(y) [ VcoO:(Y)   I
Jbl! dr d ] (7J)_7Jb12
411
dy
Y  7J
==
w(%, y, 0)
for
% +
00
(19.40)
K(y) = t/2 oo(y)c(y)
The equation for r(y) is to be supplemented by the condition
r(  ~) = r(~} =
The contribution to the dbwnwash woo(y) from the bound vorticity of the lifting line is insignificant and may be neglected. The contribution from the trailing vortex sheet is that due to rectilinear vortex filaments that extend to infinity in both directions. We thus have wco(y) = Y(7J)   d7J = 2w(y) 211 bll7J  Y
1
0
fbll
(19.41)
. Trefftz PlDne. Many methods have been developed to obtain r{y) directly as a solution of the integral equation (19.10). 'We shall not consider these methods. Instead we describe a method of solving for lI(y) and r(y) that does not use the integral equation directly. Such a method was developed by Trefftz and is known as Trefftz's method. This method depends on the following observations, which are valid under the. as~ sumptions of Prandtl's theory:
I. The downwash at any point on the lifting line is half of the down
The situation far downstream in a cross plane (i.e., in a plane normal to the vortex sheet) is shown· in Fig. 19.16. Such a cross plane is known as the Trefftz plane, also sometimes as the wake plane. It is seen that the flow field in the Trefftz plane is that due to the vortex distribution r(y) on the strip z = 0 and b12 ::5:: y ::5:: b12. Thus our prohl m is essentially reduced to that of finding the flow field resulting from a twodimensional vortex sheet and satisfying certain boundary conditions. Relation betwu. 1M jIow in tlte Trefftz plane and w(y) and f'(y). The flow in the Trefftz plane is irrotational everywhere except on the vortex
wash at a corresponding point on the trailing vortex sheet far downstream and
556
IdealFluid Aerodynamics
Elements of Finite Wing Theory
55'"
3 8
e
~y
The velocity field is continuous everywhere except across the vortex strip. Across the strip the normal component ~' is continuous, but the tangential component v' is discontinuous. Atany point y on the strip, the jump in r' is simply equal to the vortex strength y(y) at that point. We have
v'(y 0+) =  y(y)
,
..
2
v'(y, 0) =
y~)
See Fig. I ~.16 for sense of y(y). Jt 'follows that
y(y)
= 2v'(y, 0+) =
2 ocf> (y, 0+)
oy
where 0+ refers to the upper side of the vortex strip and 0 refers to its lowp:r side. Since y(y) is equal to dr/dy, we obtain
dr  =
dy
I'(Y)
.
=
2
ocf> + (y, 0 ) oy
or
Fig. 19.16 Flow situation in the Trefftz plane.
Similarly, we find
r(y) = 2cf>(y, 0)
Thus we have strip. The velocity field may therefore be represented by a potential
r(y)
= 2c/>(y, 0+) = 2cf>(y, 0)
(19.46)
c/> = c/><JJ, %)
Denoting the y and zcomponents ~f the velocity by v' and w' we have
Problem in the Trefftz Plane. Now we have to determine the potential cf>(y, z). It must have the following properties:
v'(y, z)
==~
iJz
1. cf> is a harmonic function, that is, it must satisfy Laplace's equation
(19.42) 2. grad cf> vanishes at infinity. 3. cf> is an odd function of z:
'f\l2cf>
=0
z)
w'(y, z) == oc/> The downwash at the vortex strip is then given by
(19.43)
cf>(y, z)
= cf>(y,
This follows from Eq. (19.46), which shows that (19.44) and (19.45)
~(y) ==
w'(y, 0)
==  ~: (y, 0)
cf>(y, 0+) = .cf>(y, 0) for

b b 2~ y ~ 2 b >2
and for the dowllwash at the lifting line we have
1 . w(y) =   w (y. 0)
2

10c/> ==    (y, 0) 20%
cf>(y, 0) = 0
for
y
b < 2
and
y
558
IdealFluid Aerodynamics
Elements of Finite. Wing Theory'
559
4. In view of (19.46) we require that
y  b/2 go into pointo bj4 and 6 0 to 6 == 'fT, whereas points on the bottom of the strip defin¢ by z  0 and y  bj2 to y =:' b/2 go mto points r == b/4 and 6 = 0 to 6 == 'fT. Forpoints on the strip we have ycosf) 2 Setting
5. From Eqs. (19. lOa), (19.45), and (19.46) it follows that 4>'must satisfy the boundary condition 2c/>(y, 0+) = K(Y>[ VCl)cx(y) where
K(y) =  ao(Y)c(y) 2
1
b
(19;53)
+ ~~ (y, 0)]
1
for

~~y~~
(19.47)
t .... reif
we express the cornplexpotential in'the tpIane by
F(t)  ",,(r,6)
+ i'P,(r,8)
tplane
z
Solution/or the Flow ill the Trefftz PlllIIe. To determine 4> we use the method of the complex variable. Introduce the complex variable' defined by (19.48) '=y+;z
and express the complex potential in the 'plane as
+~y
"2
•
.........
_f<_~+
Fm = 4>(y, z) + i'l'<y, z)
(19.49)
Fla. 19.17 Mapping of the ftow in tJIe Trefftz plane.
where", is the harmonic conjugate of 4> and represents the stream function of the flow. We will not be concerned with it, however. The complex velocity in the 'plane is then given by
v(y, z) iw(y, z)
Then the .complex potential in the '~plane is obtained as
==
W(n  dF (0
d,
(19.50)
F(C) ==.F[t(O]
In terms of this representation the problem reduces to that of determining an analytic such that its real part has all the properties (I) to (5) outlined above. To find the function F(t,,) it is con~enient to transform the problem from the 'plane to the plane of another complex variable I defined by the mapping
The potential", itsc:lf is given by t/J(y,z) .... ""[r(y, z), 6(y, z)] In view of the pro~rties (3) and (4) for t/>(y, z) we require that (19.54)
Fm
",,(r, 6) == ",,(r, 6) for r
and
~~
(19.55)
(19.51)
",,(,,0) = ""(" 'fT) .... 0 for all ,
This maps the strip z = 0, b/2 S; y expressed by
~
> !?
b into a circle in the Iplane
(19.52)
4
In general, the complex potential F(t) may be. assumed to be of the form
'F(t) ==
b " t =  e' 4
ib" +
1
1 ,"
ia" =
t"
ib~ + ia" ,''''
1,'"
See Fig. 19.17. We are concerned with the regions outside the strip and the circle. Points on the top of the strip defined by z = 0+ and y = b/2 to
=
i! [btl cos "6 + a" sin "6) + i(a" cOs n8 
b" sin "6)]
Elements of Finite wing Theory
IdealFluid Aerodynamir.s
561
where a. and b. are real constants. We then have
~,(r, 8)
Substituting the relations (19.60) and (19.61) into the boundary condition (19.47) and rearranging the terms we obtain
== I 1
., 1
r"
(b" cos,,8
+ a. sin ,,8)
tbV., (~r SID"
~ la"
.
8 [ . 0 +ao(Y)c(y)] ao(,1)c(y) () (19.62) SID n 4b 4b « 11 '
In view of the requirement (I9.SS) we set the btl's zero and obtain
~t(r, 8)  fa. sin ,,8
1
We now introduce the notation (19.56)
r"
and
, ., a F(t) == II!! , I t·
A."
(19.S7) and
p
v.:(~)'
4b
(19.63)
The potential #..11, Z}, which in terms of r, 0 we sball express as #..r, 0), is then given by
( 0) == ao[Y(O)] c[y(O)]
(19.64)
#..11, z) = #..r, OJ =
I"a sin nO !! 1 r·
Then (19.62) takes the form
(19.S8)
!A" sin nO[np(O) + sin 0] = p(O)«(O)
1
00
(19.6S)
To determine the coefficients a" we must usc the boundary condition (19.47). However, we must first compute the velocity w'(Y, z)  (a~/a%). Now, the complett velocity in the Cplaneis given by
v'  iw'
iii
W(e> _ dF(C>V_ dF(t) _1_ dC dt d(C/dt)
~n
1
where «(8) == «[y(O)] and 0 ranges from 0 to 11. Thi"s is the equation that determines the constants A. or, equivalently, the constants a". Later we shall discuss the method of obtaining the solution for the A,,'s. This completes the solution of the problem in the Tre1ftz plane.
Sollltioll/or r(y) fIIIIl w(y).
Using ,19.S1) and (19.S7) we find
Introduce the notation reo) !5 r[y(o)]
" 11
"(
,
% 
)
IW
. '(')
"'
11
% 

~
r"+1
a" cos"O  i sin n8 2 sin 0
and
Denoting w'[y(r, 0), z(r, 0)] by w'(r, 0) we have '( w,, 8) = From (19.S9) we obtain
~n
w(O)
(19.S9)
== w[y(8)] .,
1
~
1
a" sin nO ,,,+1,1 sin 8
Then from Eqs. (19.4S), (19.46), (19.60), and (19.61), and the notation (19.63) we have reO)  2bY., IA" sin nO (19.66) (19.61)
::(11.0)  w,(~,8)
(19.60)
w(O) = V.,
f nA" si~ n8 0
1 SID
aDd from (I9.S~'we have,
#.y, 0+)  ~ ( ,8 4
b
)
., . 'I (a .) SID fl', . b·
"1
"(19.61)
Once the A,,'s are determined from Eq. 09.6S),the solutions for the distributions of circulation and downwash for an arbitrary wing onarge aspect ratio are complete. The coefficients A" are functions of the geometrical characteristics of the wing and the freestream speed. Equations (19.66) and (19.67) thus express the spanwisc distributions of the circulation and down wash in terms of the geometry of the wing.
4
IdealFluid AerodynatDics
Elements of FJDite Wmg Theory
We observe that while the lift is determinecllty the coefficient Al only, the cirag.nd the distributi9ns of the lift and drag are determined by all the ~cients A". The dqJendelJce of CL.on'the angle of. attack is through the coef6cient A~ and that of CD, on the angle of attack is thrQugb all the Coefficients A •. To bear this in mind, for a given wing we write
19.5 Forces ad Momeats on
aD
ArbItrary Wing
ct.rn
'Pending the question of determining the coefficients A.. lNe proceed to obtain expressions for. the forces and moments on the wing. The lift and of a slice of the wing are given by (19.14) anti (19.15). Using the relation b 11 =  cosB
Ci'" trAY 1(i)
and
(1'9.7S) (19.76)
2
we rewrite these equations in terms of B. We th}lS have
6L(8) .. pY.,
er(8) sin 8 2
d6
2 Substituting into these equations the expressions (19.66) and (19.67) for r(I) and w(B) we obtain ., (19.68) 6L(8)   p V.,·". I A. sin "e sin 8 d8
1
"D~B)  p ~ w(B) r(8) sin 8 d8
(19.69)
These equations give the Iptmwise diltributio" of lift aM drag. . IntcgrUion of tblle equations Crom 8  tr to 8  0 yields the total lift and drag of tJewinS. We obtain L
where i is.a representative angle of attack for the wing,~, for example, the angle of attack~at the midsection. Recall that the drag coefficient for a wing with an elliptic lift distribution is equal to CL I/trAt. Equation (19.76) describes !he polar curve (i.e., the curve of CL versus CD,) for an arbitrary wing. Since () depends on the angle of incidence, it is not, in general, a constant, for CL~nd CD. vary with the angle of incidence. Therefore the polar curve for an arbitrary Wing is not, in peral, a parabola. This is in contrast to the parabolic polar curve for a wing with an elliptic lift distribution. The rolling arid yawing moments on the wing are expressed by (19.18) . and (19.19). In tennsoCtbe.variable fI they become
MR
.

bli'" pV..  ' . r(8) sin 28 d8
8
L'
l 6L(8)  pV.,lb
i Al
1
M• .... p ~
b (" . ~(8)w(8) sin 28 dB
°
09.77)
J.
(19.78)
(19.70)
Dj =f.°"D,(8)  pV., Ibl:!f"A"
ff .
2
(19.71)
Substituting in these for r(8) and w(B) from (19.66) and (19.6"1) and carrying out the integrations we find
(19.79)
The lift and drag ~fficients fOF the wing are therefore given b
CL E 
L bl .. .,,A1  trAt/ ipV.,ts S
D I . . trAtf"A,,1 ... AtAll[1 ipV., s 1
I
j
(19.72)
'II~ tr M • ....  pV., b .. (2"
8
'1
+ I)A.A H1
(19.80)
CD.
E.
+ I"(~A'f1
I l'
The corresponding mo~t coefficlent ·are given by
. CMR II
(19;73)
!pV.,ISC
J
M .'
==
~
.. tr}RA 11(l where
.
+ 6) _.eL (1 + (5) trAt
tr \I (M, AI 4
l)A .. A,,+l
. l'
(19.81)
(19.82)
CM •. :
(19.74)
M =:  4tr· 1;(2n +  (£)1 ... pV., SC
Bear in mind that c(y), :x(y), and b.
all·t~
coefficients .A. depend on the quantities aoM,
564
IdealFluid Aerodynamics
Elements of Fmite Wing Theory·
565
19.6 Question of the Smalleit Drag
(19.65)•. which· is
We now ask for the conditions under which the smallest value of the induced drag is ob~ained for a given lift coefficient and aspect ratio of the wing, that is,for a given value of the coefficient Al (see .19.75). From (19.76) . for the drag coefficient it follows that since !5 is never negative, in view of (19.74), the drag is smallest when the coefficients A" vanish for all n greater than one, that is, A" = 0 for n> 1 (19.83) When this condition is fulfilled, the distributions of circulation and downwash, according to (19.66) arid (19.67), become'
r(O) = 2bVooAl sin 0
I
where
.
A. sin iI6(np(O)
+ sin 0] 4b
P(O)«(O) sin 0
1.
..(0) s aO<O)c(O)
.t\
and 0 varies from 0 to ff. This is an unusual equation. There are an infinite number of unknowns. and they have to satisfy (l9.65)Jor all values of the continuous variable 0 from 0 to ff. Several methods have been developed to obtain approximate solutions for the problem. We consider here two such methods.
or
r(y) = 2bVoo A l
J
2;I
1
b
and
W[y(O)] = V ooA" a constant
This shows that the distribution of the circulation or of lift is elliptic and that the downwash is uniform along the span. We thus conclude that the
smallest drag occurs for an elliptic lift distribution.
MetluHl 0/ Giant. Suppose that the first N coefficients AI' ... , AN are sufficient to describe the distributions of the cirCulation and downwash to the .desired accuracy (see Eqs. t9.66 and 19.67). We then have only N unknowns. To obtain a determinate system·of N equations for these N unknowns we content ourselves withtiie approximation that it is sufficient to satisfy (19.65) only for N suitably choseri values of O. say 01 , ••• O.\". In this way we obtain a system of Nlinear equations for the N unknowns. This system may be expressed by
N
The geometry of the wing for minimum drag is determined according to Eq. (19.65), which now becomes
Al[p(O)
. !A, sin kO.[kP(8.)
1:1
+ sin 8.1 = 1'(0.)«(0.).
s 1.2•...• N
+' sin OJ = 'p(O)cx(~)
This scheme was developed by Glauert and W.IlS used by him for evaluating the properties of different wings.
Putting aoc/4b for p and reverting to the variable y we rewrite this equation as
ao(y.)c(y) + 4b.j 1  (2!1/b) = .!. = a constant ao(Y)c(y)cx(y) Al
This is identical with (19.30). Note that Al is equal to ro/2bV00' The possible wing shapes that satisfy this .equati?n ha.ve bee~ considered previously in Section 19.3. For an untwisted wmg thiS equation dem~nds that the plan form must be elliptic. We thus conclude that for an untwisted
wing. an elliptiC distribution of the chord yields the smallest mduced drag for a givell lift coeffiCient and aspect ratio.
19.7 Determination of tbe Coefficients A,,: Metbods of Glauert and Lotz According to Prandtl's theory .. the solu.tion .for the probl~m of the arbitrary wing is complete once the coeffiCIents A n are determmed from
MetW 0/1.....'4 Lot:. The above method· of solving for the A,,'s has a serioUs iAconveniencc. Each time one attempts to improve the acc::uracy of the results by tak:ng more terms in the series for the circulation, the whole calculation for the coefficients needs to be repeated. Also, as the number of the coefficients is increased. the construction of the solut.ion becomes increasingly laborious. To avoid these disadvantages Irmgard Lotz devised a method of successive approximations where each new step in the solution is based on the results of the preceding steps. We now describe her method· briefly. Lotz's method is based on the Fourier analysis ofihe distributions of the chord c(y), the angle of attack «(Y), and the sectional lift slope ao(Y). The starting pOint is again (19.65). which is expressed as
• Also,
some other methods that have been deviJed.
.1
!
lID
n~ sin ,,0 + I A" ~ sin n8 .;., «(0) sin D
.1
1ID'0'
p(O)
(19.84)
$66
IdealFluid Aerodynamics
'IT.
Elements of Finite Wing Theory
where 0 varies from 0 to
Recall that
1'<0)
== a,«(J)c(O)
4b .
second approximations to the coefficients A". Then A~lI ar~ the solutions of the system (19.89). The *ond approximation A~" is then the solution, of the system:
iA~'I(I1"_1 (19.85) (19.86)
We flOW intrOduce the following expansions:' sin (J
~   == ~ B, cos .,,0,
11.. +1) + ... + IA!~1(111  11... 1) + A~'I(l1o  111... + n) == at"  IA~I!I(111  Pln~I)
 IA~!.cl1a  ~".+I)  ... ad infinitum (19.90)
1'(0)
,0
0 ~ 0 ~
'IT
. at(O) IiA (J 
,,I
I
co
at" sin n(J,
:The coefficients l1,and at" may be expressed, following the usual m~~, in. terms of the distributions I'«(J) and ot{O), respectively. On substltutnlg th~ expressions (19.85) and (19~86) into (19.84) we obtain
InA" sin n(J+
_I
!
2 ,,1.,0
I I A,,11,[sin (n + ..,,)0 +sin(n + ... + IA_i(PI 
co
.
1')0]
== I
at.. sin n8
(19.87)
"1
. Equating the coefficients of sin nO from both sides o~ this equation we obtain the followir.g system of equations for the CC?CffiCients A,,: iAI<I1..1 ' 11,,+1)
This system can again be solved by iteration. Higher approximatipns for the A,,'s may be constructed by repeating the same procedure. It is seen tb.at the method of Lotz has the great advantage of improving successively the accuracy of the solution for the A,,'s to ariy desired degree. The ideas of Lotzserved as the basis for the developments of ~uriher practical schemes to obtain the spanwi~ distribution of lift ovet an arbi. trary wing. One wellknown scheme is the numerical method ~f Multhopp. For applications of the methods ofGlauert, Lotz, and Multhopp, reference should be made to the literature. •
• In particular, consult the following where further refCI"Cnc:Cs may be f\JUnd: Glaucrt (1947), Miscs (1959), KArman and Burgers (1935), Thwaites (1960).
P._I)
n == 1,2, ..• , 00
+ A,,<I1o ltI." + n) + IA"+I<1~  11.,,+1) + IA,,+.<I1.  11.,,+.) + ... ad infinitum
== ~'"
(.9.88)
An approximate solution for the .coefficients is then obta~ned~ as follows. It is assumed that in the nth equation 'of the system the coeffiCIents of the pms'" ,.. and + . , , te". . .1'1,,+1' . 1 ' 1 "so forth decrease rather rapidly. On the basis of ) such an assumption (which may be verified a posteriori), the system (19.88. is approximated by the following determinate sy~tem:
;1A 1(11_1  (J,,+J '+
... + lA,,_1(111 ....."'{l••I)
n .... 1,2,3,. ..
,00
+ A..<I1o  IP", + n) ==at",
(19.89)
This sy~tem is readily sol~ed by iteration. T~ ~uation for contains only Al and hence may be solved immediately. The .equatlon for ~2 contain~ only AI and A2 and, since,AJ is already known. A2 may be readily obtained. In thi~ manner the coefficients AI' At, and so forth may be determine~ successively. . The accuracy of the coefficichts AI. A 2 , etc., may be Improved by the process of successi'Vt apprqximation. Denote by A~u and A~11 the first and
11
Elements of Theory for Flow Past a Slender Body of Revolution
569
Chapter 20
Elements of the Theory for the Flow Past a Slender Body of Revolution
the axis of the body define a plane. We choose this plane as the x,zplane and dispose of the Y and Zaxes as shown in the figure. In acalyzing the flow past a bod' of revolution it is convenient to use cylindrical coordinates r, 0, x as shown in the figure. Then the surface of the body is described by an equation of the form (20.1) r = R(x)
z
The problem of 50Mng for the flow of an ideal fluid past a body of revolutiop, which may be at an. angle of attack, 'was discussed in Chapter 11 (sec Section 11.10). It was then pointed ()utthat ilis difficu1tto obtain a solution for the exact problem in ~loscd form aDd that an approximate closed form solution may be obtained, however, if the body is a socalled slender body of revolution and the angle ofAttack is small. I~ this chapter we consider the development of such a solution. The corresponding theory is known as the slender body theory for the flow past a body of revolution.' The results of the simplified theory are of practical significan~ only for certain flow conditions. The results, however, give rise to significant concepts th~t form the basis of the socalled slender body theory. This theory is an approximate theory for the flow past a sufficiently slender body of any crosssectional shape. . It is well to bear in mind' the difficulty with the' choice of a proper theoretical model for the actual floW of a real fluid past a bo.dy of revolution at an angle of attack. In the actual case, the flow separates from the ~y and extends from it in ~he form of vortex sheets and vortices. There is no clear in.dication asto hC'IW such a flow field may be represented uniquely by an idealfluid flow fil~ld using distribution of vortex sheets or vorti~ •.
1.0.1 Formul.doD of the Problem iD Terms of the Petba . . . . . Field
o
x
R = r(x)
v~ ~Vaosina
V... cos a
Fig. 10.1
Coordinates for flow past a body of revolution.
This may also be expressed as
F(r, 0, x) = r  R(x) = 0
<1>(r, 0, x), for the flow past the. body is
V'2<1> = 0
(20.2)
The mathematical problem, in terms of the v.elocity potential <1> =
(20.3) (20.4) (20.5)
grad <1> • grad F = V . grad F =.0 on F(r, 0, x) = 0 grad <1> = V co at infinity
We consider the steady flow ~t a body of revolution that is at an angle of attack to the direction of the undisturbed stream (see Fig. 20.1). W.e choose the'origin oftbe coordinate system at the nose (or leading edge) of the body and the Xaxis along the axis of the bOdy, the positive direction of the Xaxis being that from the leading edge to the trailing edge. The directionofdle undisturbed velocity (which we denote·as usual by Yao) and
We now rewrite the problem in terms of tile perturbatIOn field due to the body. Let us denote by q = q{r, 0, x) the perturbation velocity and by cp = cp(r, 0, x) the perturbation potential. We then have grad <1> = V'"
+q=
,\24> = 0
Vex.
+ grad ,p
The differt'ntial equation (20.3) then takes the form
..
(20.6)
(20.7)
510
IdealFluid Aerodyuamics
Elements of Theory for Flow Past a Slender Body of Revolution
The surface condition (lOA) may be rewritten
as
(20,8)
(V..
+ tV,· grad F 
Oon F('. ' ••)  0
separate problems. ~o see this we proceed as follows. From (20.2) we obtain of == 1
The infinity condition (20.S) becOmes
or
grad 4> ....... 0 at infinity
dR of ==(20.9)
0%
d%
(20.16)
To express the equations (20.7) to (20.9) explicidy in terms of the chosen coordinate system we denote the components o( • by u,., U,t u. in the directions ,.~~ e,. ... We then have
The surface condition (2Q.l4) may then be rewritten as
04> [I' == R(%), 6, %] .. u..(R
or
t
6t x)
!t ..."
8r.
The' components of Y..
!,~_ .p ~ _ u " , .at
a.·
== (V.. cos <X + u.) dR
(20.10)
,d%
 V.. sin <X sin 6
are expresiled by
V<0 ~
== (U + uz ) dR
d%
(20.11)
 W sin 6,
0~
%
~I
(20.i7)
Y.. ·" W'_ Sin « siD fl.
In cylindrical coordinates
«c:oi 6, Y.. cos «)
where we have set
Vc:ocos<X:EU
(20.18)
we have
t/J) ",.; div. tr a r) a'
Writing we may express (20.17) as
Vc:o sip
Uz
<X
==
W
"Zl
(20.19)
(20.20)
VI~ 51 div(grad
== uz , +
d .. ;La,. (ru + iJ(j (u,) + a. (ru.)]
_ a"r + ~ +1 ;Ju. + ~
oq, (R, 6, %) == u.(R, 6, %) ,
or
.
a,.
,
I'
06
a.
== (U + uz ,) dR +
l
_ ~ + ! a4> + ! a 4> + a 4> ar' .1' a,. ,.. at]I. azl
l
,
.
d%
(u
Zi
dR 
d%
Wsin
•
6)
(20.21)
(20.12)
Let us write and set
O~l
Equations 20.7 to 20.9 then take the following form:
iJI~ +!a4>+!~ +~';"'O ar' ;. ·ar r"a61 azl
(V.. sin« sin 6
,
(20.13)
or' or
(R, 6, %)
==
(U
dR + uz , ) 
dx
(20.2~)
+ ur)aF + (Veo cos« + u.) OF .. \) a ~ X
on F(r., 6, %) .... 0 (20.14) (20.1 S)
II,. If..are to be ex~ in terms of 4> according to (20.10). The 'solution of the mathematic:al problem repmented by (20.13) to (2O.1S) _1 be aprased as the superposition of t.be solutions of two
•
u,., II,. ". go to zero at infinity
Here 1Ipo
dR . (20.23) WSf:1 0 dx We then readily see t"'at (20.22) represents the surface condition for axisymmetric flow past the body of revolution (see .Fig., 20.2). where the speed of the undisturbed stream is V, and that (20.:':3) represents the surface condition for lateral flow past the same body of revolution (see Fig. 20.3), where the s~ed of the undisturbed· stream is W. In view of this . the nature of the infinity condition and the linearity of the governing differential equation. the mathematical problem represented by the equations (20.13) to (20.15) may be expressed as the superposirion of n~~problems:

oq,z (R, fl, %) == ~
U Z1 
Elements of Theory for Flow Past a Slender BOdy of Revolution
571
S1J
IdealFluid Aerodynamics
z
term ulZ(dR/dx) from (20.17). The surface condition then takes the approximate form '
0</1 ')  (B~ 9, x)  ur(R, 9,x
or .• . dR .. 9 == V.., cos ex   V.., SID ex sm dx "(20.24)
u
x
Fig. 20.2 IIIll stratingaxisymmetric flow past the body of revolution.
(1) the problem representing the axisymmetric flow past the body of revolution, and (2) the problem representing the lateral flow past the same body of revolution. The speed of the undisturbed stream fo~ th~ axisymmetric flow is given by (20.18), whereas that for the lateral flow IS gIVen by (20.19).
z
This is known as the linearized form of the boundary ~ondition for flow past a sle~der body of revolution. I~ is linear i.n t~e sense that (o</l/ih:) (i.e., ulZ ) does not appear in the equatIon expressIng It. We immediately observe that the assumption of small perturbations is not valid at and near the stagnation points on the body. Furthermore, the assumption that the body is slender implies that of/ax is much less than of/or or that dR/dx is much less than unity. This, however, is violated at and near the ends of the body if those ends are rounded. We. shGUld~ therefore, expect that any solution obtained on the basis of the linearized surface condition (10.24) cannot be valid near stagnation points or r~unded ends. These regions where the solution is not vaiid are, however, small. In view of Eq. (20.24), the solution for the flow past a slender body of revolution at an angle of attack may be obtained as the superposition of tile solutions of the following two mathematical problems:
Azisymm~tric
Flow P"st "
S/~_r
Body 0/ Revollltioll
</I  </I(r, x)
(20.25)
1
o
x
where'
V2</1 == 0'</1
or·
+ ! ot/l + 0 ", == 0
r or oz'
0"'1:"'1 .::=..::=.
(20.26)
(20.27)
dR o"'(R or' x)u('R' x)=U dx r lJ
FIa.20.3 Illustrating ,lateral flow past the body of revolution.
== Vao cos !X
(20.21)
V",  0 at infinity Cross Flow or L"tertJI Flow PGSt " Slender Body of Revollltioll
20.2 Boundary Condition for a Slender Body of ReyolutIon
C<:>DsJ(Ier the flow past a: slender body of revolution. By suc.h a. body ~e mean an elongated body of revolution whose crosssechonal radiUS changes slowly with distance along the axis of revolution. We' ssume that the magnitude of the perturbation velocity componen!s is small comp~red to the speed of the undisturbed stream. On the baSIS of th~e consIderations we may neglect the term u:z<oFIih:) from (20.14) or, eqUlvalently, the
= ",(r, 8, x) y2¢> = o'rq, + ~ o¢> +.!. ae", + 02¢> =
cb
(20.29)
or!
, Or
,200 2
o:c2
0
(20.30)
(20.31)
or
o¢> (R, 0, x) = ur(R, 0, x) =  lYsin 0
0 S; x S; I
IdealFluid Aerodynamics where
Elements of Theory for Flow Past a Slender Body of Revolution
575
w V.,sjnlX
Vf,  0 at infinity' (20.32)
To this we add the condition that since the body is closed, the total source strength should be zero:
L'q(%)d%
We note that (20.27) and (20.31.) are the linearized.. boundary conditions for their. r~pective problems. Recalling that in the problem of the thin airfoil we were able to transfer the boundary condition froin the surface of the airfoil to its axis, we may inquire whether such a procedure is also feasible in the problem of the slender'body bf revolution. .As we shall see later, the perturbation velocities in the pr<;»blem of the body of revolution become infinitc"on its axis. This means that they cannot be expand~ in Taylor series about the axis and that one cannot approximate .their values on the surface of the body by their values on the axis. We thus conclude that the boundary condition (2O.24) (or, ~quivalently, the conditions (20.27) and (20.31» carinot be imp<1sed on th~ axis of the body instead of on' its surface. alternative procedure for transferring the boundary condition to the axis in a modified form, hoWever, becomes possible. We. show this in its proper context.
== 0
(20.37)
Explicit determination of q(%) from (20.36) and (20.37) is not possible. Numerical or approximate methods are to be employed to obtain q(%). For
r
An
20.3 Axisymmetric Flo" Past a Slender Body of Reyoludoa: Soludoa by Source Oktribbtioa
x
The mathematical problem is represented by (20.25) to (20.28). The angIe of attack. IX is zero. Hence V., == U. • On the basis of our past experience we know that the disturbance field i!1 such a flow may be rep~nted as that due to a suitable distribution of point sources along the axis of the body in tbe interval 0 ~ % S;; I. Let q == q(%) denote the source .strength per unit le~gth of sltch a distribution (Fig. 20.4). The disturbance potential and 'the disturbance velocity components at ahy field point r, 8, % are then given by .
Fig. 10.4 Illustrating the determination of .the disturbance field for the axisymmetric flow by source distribution.
#.,r, %) ==  1
U
r
( .)
r,
%
ot/J 1 == or == 411'
~ (/:1#
4 11'
u (r
.'
%) 
i' ot/J _ 1. i
I
q(E)'r==::::;:===:; 411' 0 .J(%  e)1 rl
il
de
a very slender body of revolution we may obtain a good approximation to the source distribution by employing the socalled 'slenderbody approximation for the boundary co",dition. To develop this approximation we first seek the value of the integral
Ur
+
r
(20.33) (20.34) (20.35)
oVl
(E)
[(x _ e)1
+ rl)~'.d~ + rt)S., .
at/J 1 [' ==""'8r == 411' Jo 'q(E) [(x _
r
E)I
+ rZp'1 dE
0
q
(E)
[(% 
% Ii"
~).
e
d~
Ii'
for small.values of r, that is, as r . O. It is seen that care is necessary in obtaining this value of the integral, since for r = 0 at ; = x the integrand becomes indeterminate. As shown in Appendix E
q(x) (rur)ro = 
According. to the boundary condition (20.21), the source distribution .  ~.) must s,atisfy the integral equation
 tee) {Zi "
•
21T
(20.38)
~,. !id~ . [(z .,' +). . ....
r}
Rf., == u/r== R(%» == 11 a%
dR
o S;; % S;; I
(20.36)
this result may be obtained by an elementary argument. Consider, as shown in Fig. 20.5, a small cylinderof radius r and length dx, surrounding an element of the source distribution. Then as r . 0, the volume outflow
516
IdealFluid Aerodynamics Elements of Theory for Flow Past a Slender Body of Revolution
577
of ftuM, through the surface of the cylinder is equal to the volume of the ftuid put out by the source element: (21TTU p dz}r_o = q(%) dz
or
(rur)r... O == ' 271 This shows that for small r we have·
ur
'"  ,
q(~)
In this form the poundary condition is imposed on the axis of the body instead of on its surface. Equation (20.40) is known as the slender body approximation for the boundary condition for the axisymmetri~ ftow past a slendet body of revolution. From (20.38) and (20.40) we obtain
1 0 x ~ 1 r and hence U r is infinite on the axis 'of the body in the interval 0 ~ :t ~ I. It appears, therefore, that in determining q(%) from the surface condition
s:
dR q(%) = 271URd%
(20.41)
==
U dS(%);,;;: US'(%) d:i:
JC
where S(x) = 1TR2(X) is the crosssectional area of the body at x. With the source strength thus determined, the perturbation potential and the perturbation velOCity components may be obtained from Eqs. (20.33) to (20.35). We are particularly interested in their values near and on the body. They are obtained ~y constructing the values of the integrals in (20.33) to (20.35) for small \alues of r, that is, for r (I  x). We find that for smail r
«
Fig. 20.5 Illustrating the determination of the source strength by considering the outflow through a cylinder enclosing an clement of the source distribution.
4>(r,
%)
= 2f(x) log ~  [f(O) log %
+ f(/) log (l 
x)
(20.42)
(20.36), we should not attempt to approximate Ur at r = R(x) by Ur at = 0 'in analogy with the transfer of the boundary condition from the surface to the axis of a thin airfoil. It is observed, however, that in the present case rUr is finite as r  0 provided q(%) is finite. This being the case, we may express rU r at r = R(x) as a Taylor series about the axis:
r
rO rO 2 Neglecting the terms involving the derivatives and the powers of R(%) we write (rur)r_RI.,) = (rur)r_O (20.39) ,
+ i'1'(e) log (% uir, x) =
E)
de]
1%
2J (%) log 
,
r
2

[/(0)
%
f(l)  ,
':t)
(rUr),_RCr'
0 = (rur}._O + [ ; (ru r )] fir
R(x)
+ . (ru r )
[a' or
]
R1(x) 
+ ...
+ 1'(0) log % + 1'(1) 10g(1 
+ff"(~)IOg(%  ~)dEJ
and, as we already know,
ur(r, x)
(20.43)
In view of this equation, we express the boundary condition (20.27) u.(r = R(x» = U dx in the modified approximate form
,
dR
== 2  ' r
f(%)
(20.44)
Here we used the following notation
05;x5;1
(20AO)
f(x)
:e q(x) =
41T
}{, S'(x) = U R dR 41T 2 dx
• Note that this is unlike the behavior of L' for y ...... 0 in the case of uniform flow past a Jhin symmetrical airfoil at zero angle of attack.
(20,45)
J.,() ;,;;: df x , etc
dx
57'
IdealFluid Aerodvnamics
Elements of Theory for Row Past a Slender Body of Revolution poteDtial and the velocity respectively, arc, given by
579
We observe that the results (20.42) to (20.44) are not valid near the ends ~f t~e body. If the body is rounded at the ends, 1(0) and I(£) become mfimte. If we choose pointed ends, then R(O) and R(f) are zero, and consequently1(0) and I(£) also vanish. Usually a pointed body is assumed. 20.4 Cross Flo" Past a Slender Body of ReyolutioD: Solution by ~ublet DistributioD
~R) ... _
q(r) _ .....
1.. ~ . (R 411'
IR _ al'
a)
(20.46) (20.47)
.!. grad[M. (R 411"
a)] . IR  all
Now the flow field is no longer axially symmetric. The mathematical problem for the disturbance potential is expressed by Eqs. (20.29) to (20.31)
"Equation (20.46) is (i 1.65). Now, consider the clement e.",(E) dE of the doublet distribution in the present problem. We have s = Ee.,. The potential at .the field point R .. (r, 8,21) due to this element is then given by
z
z
6.1.1 r 8 ) 1 ..1E) r sin 8 dE 'n , ,21 = 411",N\ [(21 _ E)I + rl]~
Therefore the potential at the field point due t6 the whole doublet distribution is given by
'n
;rY
%/
.1.1r 8 )
,
,%
,.
1 1 ~) r sin 8 d~ 4~ o,N\'i [(x  E)I + rl]~ 'i
f. '.
(20 48)
.
For the velocity components we obtain
x
u (r 8 %) 
. "
 at/l = 411" o,N\IE) [(x ar 1...f.'..
3 rl sin 8  411" • P(E) (21 _ E)I + ,.]~ dE
Fig. 20~6
f.'
sin 8 dE E)I + rl]~ (20.49)
Illustrating the determination
or the c:rost ftow by doublet distribution.
8 ) 1 £l u,(r, ,% .. ; a8
u.(r, 8, %) =
i .. 411".
J.' (~)
r
cos 8 d~ • po 'i, (% _ E)I + rl]~"
(20.50)
(20.51)
where we simply set at = 11/2. We shall retain W to denote the undisturbed stream. We represent the disturbance field as that due to a suitable distribution of doublets along the axis of the body, with the axes of the dov~}lets opposing the undisturbed stream (see Fig. 20.6). See Section 11.9. Let '" == "",(:e) denote the' doublet strength per unit length of the distribution. As we know, the flow field due to such a doublet distribution satisfies laplace's equation (20.30), and the infinity condition (20.32). It follows that the distribution ,u(x) s~ould be chosen so as to satisfy the surface condition (20.31). Once ,u(x) is determined. all the desired information about the flow field may be found. To expres~ the perturbation potential and the perturbation velocity components In terms of ",(x) we first recall the pertinent results for a doublet M situated at the point s.from the origin of coordinates (see Section 11.8). If R denotes the position vector from the origin to a field point, the
" aZ.
at/l __ 2.
'p(E) (21 'E)r sin 8 dE 411"J. [(%E)I+rlt'
According to the boundary condition (20.31), the doublet distribution
"'(21) must satisfy the integral equation
{sin 8
(lr J~
~
l'p(E)
;,. ~ [(21 , E + ]
dE}"....
= ur[R(z), 6, z]
R(.)
Wsin8
O~z~;
(20.52) Explicit determination of p(z) from this equation is not possible. Numerical or approximat~ methods to be employed to obtain ",(z). For a.very slender body of revolution we may obtain a good approximation to the doublet strength by employing a socalled slender body approximation for the boundary condition.
an:
5'0
IdealFluid Aerodynamics
Elements of Theory for Flow Past a Slender Body of Revolution and, as we have already seen,
u (r (J %) _ _ 1'(%) sin (J r , , 21T r l
5'1
To d~velop this approximation we first seek the value of u for small ' values of r. that is. as r _ O. We find that
(
r u,
2
)
,_0
= 
J.l(x) .
27T
SID
8 (20.53) In these relations the expression given by (20.56) must be substituted for p.(%). By combining the results of this section and the previous section. we may obtain the perturbation potential and the perturbation velocity components in the flow past a slender body of revolution at a small angle of yaw or attack.' In combining these relations we replace U by V 00 cos IX and W by V00 sin IX.
or
( u)
0
= _ J.l(x) sin (J
"
21T,a
. It!s seen that although u, becomes infinite on the axis of the body. r2u, IS finIte on that axis. Hence we may express ,.2U, on the surface of the body as Taylor series about the axis:
(r
2 u,)r_R(%)
= (r
2 u,)1_0
+ [:
ur
(rIUr)J . R(x)
rO
+ [0:(r2ur)J
or
rO
RI(x)
2
+ '"
20.5 Pressure DistributioD
The pressurt at any point is given by the Bernoulli equation
Neglect~ng the ter~s involving the derivatives. and the powers )f R(x) we may wnte approxImately (r 2ur)r_Rl%) = (r 2ur)r0 (20.54)
~ == Poo + ~ pVIDI[l  ;~IJ
In terms of the perturbation field we have
P
!n view of ~his equation. the boundary condition (20.52) may be expressed ID the modIfied approximate fo. m (rlu,),_o = (rur),_R(z) =  WR2 sin 8
0~x ~I
== P ! pV 2
00
00
1(2 Voo' q +.£) Vi VI
00 00
GO
aD
(20.60)
(20.55)
In th.i~ for.m •. known as the slender body approximation. th~ boundary condItIOn IS Imposed on the'axis of the body instead of on its surface. From Eqs. (20.S3) and (20.55) we obtain
J.l(x) = 27TWR2(x)
The pressurecoefticient is given by
C := P "
.1 V I p GO
PID == 2 V ... q _ L I V I V I
== Ue" V.. ' q = Uu.
Voo
(20.61)
Axild Flo", Put .. Sk_r Ilotly 0/ Rnohltioll. For axial flow we have
(20.56)
= 2WS(x)
whe~e. as before. Sex) is the crosssectional area of the body of revolution. WIth the strength of the doublet distribution ·thus determined the perturbation potential and the perturbation velocity components m~y be o.btained from Eqs. (20.48) to (20.51). Near and on the body they ate gIven. by the values for small r of the integrals in tRese equations. We find that for small r .Jo( 8 ) _ J.l{x) sin 8 ~ r, ,x (20.57) . 27T r
In terms of cylindrical coordinates r,
(J, %
introduced before. we obtain
Ur
C (r
" , ,
(J %)
==
2 u" _
U
.
a+ 1 u% U2
(20.62)
uk, e,
u (r
% ••
x) = J.l(x) cos
e
dx
27T 27T
r
r2
(20.58)
(20.59)
e x) = ..!.. sin 8 dJ.l(x)
To obtain the pressure distribution over the body we evaluate the righthand side of this equation Jor , == R(x) in the interval 0 ~ x ~ I. From Eqs. (20.43) and (20.44) we find that near the body (u,/ U)I is of a magnitude comparable with that of·uz/U. This me~ns that in any attempt at simplifying the pressure relation (20.62) we cannot neglect the term (Ur /U)2. arguing that it involves the square of a perturbation velocity. We may. however, neglect the term (u..lU)! in comparison with the rest. Thus the
S62
IdealFluid Aerody.namics Elements of Theol')"for Flow Past a Slender Body of Revolution
S63
pressu're on the body is given approximately" by
C (R 8 x)
" , ,
==
2 (UII),
U
_
TRIIII
(U I),TRIIII
~
I
pressUre distribution over the body is given approximately by
o~
x ~,
(20.63)
C (R 8 x) = 2
U
II
cos ex
Now, from the surface condition we have
( ur \
U)TRIIII 'i"
" "
~
2 (Ur SIO ~
uri
sin ex
.• 8
~) + u,cos,!
dR.
dx
+ u,'
VI ...
(20.68)
Hence (20.63) may be rewritten as
C,,(R, 8, x) =
_2(U"\
. U )TRIIII
_
(dR\i
dx}
o :$;;
%
:$;; I
(20.64)
have
Lateral Flow Past a Skllikr Body of Rerwilltio,..
For lateral flow we
v..,
= We.
V .... q = WU p sin 8
+ Wu, cos 0
U2+UI+UI
_T_
We see that this pressure distribution cannot' be obtaine.d .by Simp'! ~om bining the pressure distributions of the axial and lateral flows. ThIS IS. so because now (unlike in the case of the thin airfoil) the pressure relatIon involves the squares of the velocity components. . Equation 20.68 may be rewritten in a form that ~i11 ~xpress the pressure on the body as made of two parts, one p~rt whIch IS the ~am~ as the pressure distribution ,in the axial flow and another part whIch IOvolv~s the angle of yaw and the crossflow velocity com~nents only. To do thIS let us first rewrite (20.68):setting cos ex!:::! 1 and SIO ex!:::! ex. We then have
Equation (20.61) for the pressure distribution then takes the form
C,,(r, 8, x) = 
2 W(u T sin 8
C (R 8 x)   2...1! ~ 2 (20.65)
U
_
+ u, cos 8) 
~I
~
"
'.'
Vao
ex. (u r sin 8 Vao
II + . I + u, .cos 8)  Up V' u, .., o ~ x ~ I (20.69)
To obtain the pressure distribution over the body, we first o~serve that, on the basis of Eqs. (20.58), (20.59), (20.53), and (20.56), we may . neglect the term (uJ W)I in (20.65) while retaining the rest. Thus the pressure distribution over the body is give.n ~pproximately by . C"( R, (J ,x) =  2 ( UTSIfi 8 + u,cos (J) _
On the body according to the approximate boundary condition (20.24), we have
dR. ur<R, 8, x)  V"" d%  V..,ex SIO 8,
0 ~
%
~ I
.
(20.70) _
Using this relation in (20.69). we obtain
W

UTI +U,I I
W
o ~ x ~,
(20.66) Now we
have
Flow Past a Skllikr Body of RerwlMtion at an Angk of Yaw.
V.., = V.., (sin
IX
C (R (J %) _ 2 U. " , , V..,
_
U (dR)I,+ exl sin' 8  2ex u, cos 8  (V , 'r d% Y.....,}
•
sin 8er
0~
UII UIII
%
~
I (20.71)
(20.72)
+ sin IX cos Oe, + cos lXe
510
•
)
Now. express the u..compon,?nt as
ll
Equation (20.61) for the pressure l:it any point, therefore, takes the form sin IX C,,(r, 0 ) =  2 u'" cos IX ~ 2   ( Ur ,x V", V.., 8
+u
ll •
+ u,
Uri
cos (J)
where u is the component in the axial flow and ull• is the component in the lateral ft~w. Using (20.72) we rewrite (20.71) as
+ U,I + u",'
V '
C,,(R. 0, x) = C",(R, 0, x)
(20.67) where
+ C".(R, (J, x)
(20.73)
'" Any component of the perturbation velocity is now given by the sum of the
corresponding components in the axial and lateral flows. To obtain the pressure distribution over the body we note that the term (u",IV",)2 in (20.67) may be neglected while retainine the rest. Thus the
~C",(R, 0,:1;) =
C (~ (J x)
2
~I

e:J
2
1X2
O~ x ~ I
II, ('
(20.74)
u '" =  2 2! + "2 ' , V"
sin 8  2ex  cos 0 . V",
U, )2 V' ..,
x
~,
o~
(20.75)
584
Elements of Th~ory for Flow Past a Slender Body of Revolution IdealFluid Aerodynamics
SIS
We see that C P1 is the pressure distribution in the axial flow (see 20.64). The component Cpl repr~sents the effects of the angle of yaw and involves ex and the velocity components of the cross flow only. The expression (20.75) for Cpl may be put into a simpler ~orm. ,Usin, (20.56), (20.58), (20.59) and setting W ~ V..,IX, we reduce (20.75) to the form
ePl(R, 0, x)
z
= ~41X dR sin 0 + IXI(1
dx
00

4 cos' 8),
0
~
x
~
I
(20.76)
x
26:6 Forces
the Body of RHoladdD
On t':le b,asis of the gen~l"al considerations that lead to d' Alembert's . paradox, we assert that the preceding results for the flow past a body of revolution, whether it is at an angle of attack or not, must lead to auro force ori the body. Although the total force on the body is zero, it is of interest to know the distribution orthe force with distance along the axis of the body. W.e shall, therefore, set up the relevan·t expressions for calculating such a distribution. Consider an element 0 dA of the surface of the body (see Fig. 20.7). The element is situated at the point R, 0, x. We have
dA. == RdO tis
where ds is an element of the curve of intersection between the body surface and the plane 0 == constant. The force on the budy is given by
F == 
~ pa dA. == Ji
ff
Ji
paR dO ds
(20.17)
x
where A. denotes the surface of the body. We are usually inte~ted in the components of this force with respect to the x, y, zdirections. Denoting these components respectively by F';' F", F., we .ave
F.,
== 
F" == F. == 
If ff ff
Ji
A
Rp(o· e.,) dO ds
(20.78)
Fig.20.7 Elemental area on the surface of the body of revolution.
Rp(n • ell) dO ds Rp{o· e.) dO ds
(20.79) (20.80)
h d" ate r We then have 'P the angle between n and the direction 0 f t e co or tn .
n = sin tpez
Also, we note that
+ cos 'P cos Oe. + cos 'P sin Oe.
Ji
Drawing the outward normal·
0
as shown in Figure 20.7h, we denote by
ds cos 'P = dx dR ds sin 'P = dR =  dx dx
• Note that in terms of the system r, 8, :z: we have
n = ± II _ (dRfdx)IP! er
1
(dR)

;Z; e.
586
IdealAuid Aerodynamics
Elements of Theory for Flow Past a Slender Body of Revolution
587
On using these relations, Eqs. (2Q.78) to (20.80) become
F.:;::: (' f R dR p dO dx Jo)o dx FII  f.lIrRPCOSOdOdx F. 1r
flow and C PI giyen by (20.76) represents the eff~ of the angle of yaw on the pressure. Prom (20.74) and (20.76), and (20.85) to (20.87) we obtain
dF" == 0 dx dF.,  . == {p",,qlZ) dx dF.
rlx
(20.88 (20.89)
1.1"
d:t
Rp sin 0 dO dx
[I +(R,24)+ (dRD} dS . Vao dx dx
(X
u""
The distributions of the force . . . . F· FII' F• are the n gIVen by . . . components ,
dF. _ R dR
. dx
= 2q
a;>Q(
dS dx
(20.90)
fa.
Jo p
0
iJO
(20.81)
dF _"  8
dx·
dF  ' = R d%
ilr i"
0
pcos
0 dO
(20.82) (20.83)
where uz,(R, x) is known from Eq. (20.43). We see that both the axial and crossflow parts contribute nothing to the distribution dF,,/dx. They both contribute to the distribution of the axial force, the contribution from the crossflow part being qlZ)Q(I(dS/dx). Only the.crossflow part contributes to the distribution of the normal force. 20.1 Moment on' the Body of ReYolutlon Although there is no (resultant) force on the body of revolution, there is, in general, a moment on it. The moment about the origin of coordinates is given by
psin OdfJ
We now rewrite these relations in terms of the pressure coefficient C We have P' where
p
= q..,C. + p..,
'I.., lpV..,*
(20.84)
Using this relation, we rewrite (20.81) to (20.83) as
dF. dRf. .. · dS qaoRd C.dfJ+p..,d% x 0 dx dF ::!.! == q R dx
IZ)
M ffR)( =  ff(R )(
..I.
A
pndA
n)pR dO tis
(20.91)
(20.85)
i
1r
C cos 0 dO
•
0
(20.86)
dF dt = q..,R
fir J C" sin fJ dfJ
0
(20.87)
where R is the position vector to the surface element n dA at the point (R, 0, x). On the basis of this equation we find, as we expect in view of the symmetry of pressure distribution about the x,zplane. that there is no . (rolling) moment about the ~axis and there is no (yawing) moment about Zaxis: (20.92) M.=O, M. 0 The pitching mom~nt Mil is given by
where C p  CiR,O, %), 0 ~ % ~ I. Note that S is ."R'. For the flow past ~ body of revolution at a small angle of yaw the ' pressure coefficient is given by (20.73):
C.iR, 0, x)  Cp,(R, 0, x)
where
CPt
+ epl(R, 0, x)
(20.93)
given by (20.74) represents the pressure for the axial part of the
588
Using (20.87) and (20.90) we obtain
IdealFluid Aerodynamics
Mil = .....
= 2q,,/x.
=
r'(x + R dx dx d~ dR)'dS 2q""cr. r '[x dS + .l.(dS)2] dx Jo dx 27T dx
Jo
' 1(
o
X
+ R dR)dF' dx dx dx
Selected Problenu
(20.94)
2.1
CHAPTER 2 2.2 2.3
2.4
2.5
Neglecting the term (dSjdx)2 we have approximately
(20.95) Since S(I) is zero and the integral 1'S(X) dx is the volume of the body. the pitching moment on the body is given approximately by
2.6.
2.7
2.B
My = 2q""cr.(volume of the body)
It is customary to define a moment coefficient by the relation C